BROMELIACEAE: PROFILE OF AN ADAPTIVE RADIATION This book presents a synthesis of the extensive information available on the biology of Bromeliaceae, a largely Neotropical family of about 2700 described species. Reproductive and vegetative structure and related physiology, ecology and evolution are emphasized, rather than floristics and taxonomy. Guiding questions include: why is this family inordinately successful in arboreal (epiphytic) and other typically stressful habitats and also so important to extensive fauna beyond pollinators and frugivores in the forest canopy? Extraordinary and sometimes novel mechanisms that mediate water balance, tolerance for high and low light exposures, and mutualisms with ants have received much study and allow interesting comparisons among plant taxa, and help to explain why members of this taxon exhibit more adaptive and ecological variety than most other families of flowering plants. This volume concentrates on function and underlying mechanisms, and thus complements a literature that otherwise mostly ignores basic biology in favor of taxonomy and horticulture.    .     is the Robert S. Danforth Professor of Biology at Oberlin College, Ohio, USA. His research career has focused on the biology of epiphytic vegetation, especially bromeliads and orchids. He is author of The Biology of Bromeliads (1980) and Vascular Epiphytes (1990). BROMELIACEAE: PROFILE OF AN ADAPTIVE RADIATION DAV I D H . B E N Z I NG Oberlin College, USA With contributions by B. Bennett, G. Brown, M. Dimmitt, H. Luther, I. Ramírez, R. Terry and W. Till                             The Pitt Building, Trumpington Street, Cambridge, United Kingdom                  The Edinburgh Building, Cambridge, CB2 2RU, UK http://www.cup.cam.ac.uk 40 West 20th Street, New York, NY 10011-4211, USA http://www.cup.org 10 Stamford Road, Oakleigh, Melbourne 3166, Australia Ruiz de Alarcón 13, 28014 Madrid, Spain © Cambridge University Press 2000 This book is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published 2000 Printed in the United Kingdom at the University Press, Cambridge Typeset in 10/13pt Times NR MT in QuarkXPress™ [] A catalogue record for this book is available from the British Library Library of Congress Cataloguing in Publication data Benzing, David H. Bromeliaceae: profile of an adaptive radiation / David H. Benzing; with contributions by B. Bennett . . . [et al.]. p. cm. Includes bibliographical references (p. ) and indexes. ISBN 0 521 43031 3 (hardback) 1. Bromeliaceae. I. Title. QK495.B76B45 2000 5849.85–dc21 99-30141 CIP ISBN 0 521 43031 3 hardback Contents Part one 1 Part two 2 3 4 List of contributors Preface Acknowledgments Glossary Abbreviations Brief overview Introduction D. H. Benzing Basic structure, function, ecology and evolution Vegetative structure D. H. Benzing Habits: general overview Organization for foraging Relationships of the body plans Stems Roots Vascular cells Foliage Trichomes Reproductive structure D. H. Benzing Inflorescences Flowers Fruits, ovules and seeds Pollen grains Carbon and water balance D. H. Benzing Ecophysiological peculiarities The five ecophysiological types Photosynthesis and water economy Crassulacean acid metabolism: basic characteristics Bromeliad CAM: basic characteristics v page ix x xiii xv xviii 1 3 17 19 22 36 42 46 48 50 52 70 79 81 89 98 105 107 110 111 114 115 117 vi Contents Ecological correlates of the carbon fixation syndromes Ecophysiological profiles of the five types of Bromeliaceae Xeromorphy and water relations CAM vs. C3 bromeliads: performances in situ Predictors of photosynthetic capacity (Amax) Hydration CAM reconsidered as an evolutionary response to stress Citric acid: its role in ecophysiology CAM and hydration Additional aspects of light relations 5 Mineral nutrition D. H. Benzing External supply and plant demand Nutritional peculiarities Nutrients in the forest canopy Mechanisms Involvement of foliar trichomes Nitrogen nutrition Architecture and nutritional economy Bromeliads as air quality monitors 6 Reproduction and life history D. H. Benzing, H. Luther and B. Bennett Pollination Floral rewards Fragrances Flowering phenology Breeding systems Synchronization within populations Genetic structure of populations Seed dispersal Seed viability and germination Resource economics and life history The organization of reproductive allocation Demography Asexual reproduction Final comments 7 Ecology D. H. Benzing Frost-tolerance Distribution in forests Roles in succession 120 123 145 151 160 162 168 174 174 176 187 188 197 199 209 229 235 238 240 245 246 264 268 268 276 280 281 284 299 301 305 308 323 326 329 331 339 362 Contents 10 8 9 Part three 10 11 Influences of shoot form on bromeliad distribution Effects of epiphytic bromeliads on trees Effects of bromeliad nutrition on forests Terrestrial Bromeliaceae Bromelia humilis: a case study of terrestrialism Relationships with fauna D. H. Benzing Predators and pathogens Mutualisms Ants and bromeliads Evolution of ant/plant associations Termites Phytotelm bromeliads Bromeliads and the definition of soil History and evolution D. H. Benzing, G. Brown and R. Terry Fossils Phytogeography Chromosomes, hybridization and polyploidy Ancestral habitats Heterochrony Neoteny and tillandsioid radiation Historic relationships between mesophytism and xerophytism in Tillandsioideae Taxonomy: traditional characters Chemical systematics Relationships among subfamilies and Bromeliaceae within Liliopsida Final comments Special topics Neoregelia subgenus Hylaeaicum I. Ramírez Taxonomic problems Ecology and geographic distribution Cytology Vegetative morphology Trichomes Inflorescences Floral morphology Reproductive biology Continuing taxonomic problems Cryptanthus I. Ramírez vii 369 372 382 384 400 405 405 414 421 431 436 437 459 463 464 465 488 493 500 504 509 516 517 521 540 543 545 545 547 547 547 548 548 549 549 550 551 viii Contents 12 Tillandsioideae W. Till Anatomy and morphology Phytogeography and evolution 13 Tillandsia and Racinaea W. Till Evolution Subgeneric treatments of Tillandsia Racinaea 14 Ethnobotany of Bromeliaceae B. Bennett Folk taxonomy of Bromeliaceae Uses of Bromeliaceae Indigenous management of bromeliads 15 Endangered Bromeliaceae M. Dimmitt Factors threatening bromeliad populations In situ conservation Ex situ conservation Conservation laws and their implementation Literature cited Name index Subject index Taxon index 555 559 569 573 575 578 585 587 588 589 607 609 610 615 616 619 621 657 665 675 Contributors David H. Benzing Department of Biology, Oberlin College, Oberlin, Ohio 44074, USA Bradley C. Bennett Department of Biological Sciences, Florida International University, Miami, Florida 33199 and Fairchild Tropical Garden, 11935 Old Cutler Road, Miami, Florida 33156, USA Gregory K. Brown Department of Botany, University of Wyoming, Laramie, Wyoming 820713165, USA Mark A. Dimmitt Arizona–Sonora Desert Museum, 2021 North Kinney Road, Tucson, Arizona 85743, USA Harry E. Luther Marie Selby Botanical Gardens, 900 South Palm Avenue, Sarasota, Florida 33578, USA Ivón M. Ramírez Centro de Investigacion Cientifica de Yucatán, A.C., Mérida, Yucatán, Mexico Randall G. Terry Division of Biological Sciences, University of Montana, Missoula, Montana 59812, USA Walter Till Institut für Botanik der Universität Wien, Rennweg 14, A-1030 Wien, Austria ix Preface Bromeliads enter recorded history with Columbus’s account of Carib Indians cultivating Ananas comosus (pineapple) on the island of Guadeloupe. Within the next hundred years, commercial production began at numerous Old World sites, and by the mid-19th century major European botanical gardens were displaying numerous ornamental Bromeliaceae. Ready access in culture and often novel adaptations for life free of contact with earth soil in turn guaranteed the attentions of early phytogeographers and morphologists. Interest has continued to grow until today more is known about the ecophysiology of the bromeliads than about the members of almost any other family of tropical plants. Major advances in systematics, natural history theory and functional biology over the last two decades have heightened opportunity to reconstruct adaptive radiations and impute the conditions of ancestors and their habitats. Evolutionary relationships inferred from the structure of DNA provide the robust phylogeny necessary to order and date the origins of those aspects of phenotype responsible for current adaptive variety and importance in ecosystems. Molecular, combined with traditional taxonomic, data have already expanded insights on the histories of clades as diverse as the Hawaiian silver swords and stickleback fishes. However, none of the inquiries on plants has considered more than a few of the many traits that shape botanical radiations by dictating growth requirements, mobility, relationships with other biota and ecological tolerances. Enough literature exists for Bromeliaceae to explain in exceptional depth how a sizable taxon of tracheophytes has colonized so many, often unusual, kinds of substrates in varied habitats. Probably no other family exceeds Bromeliaceae for the variety of services provided to dependent biota ranging from detritivores to pollinators, nor does any comparably sized clade employ a more novel array of contrix Preface xi vances to acquire and utilize water and mineral nutrients. Some of the more stress-tolerant bromeliads root in media that exclude most other vascular flora because they lack equivalent capacity to exploit unconventional supplies of moisture and key ions. In short, Bromeliaceae exemplifies botanical radiation in the extreme, hence represents an exceptionally propitious taxon to study related mechanisms and outcomes. This volume is devoted to that challenge. Conversely, it largely ignores taxonomy for its own sake, instead adopting an existing system (Smith and Downs 1974, 1977, 1979) to organize the information more immediate to our purpose. The only deviations involve allusions to certain post-1970s revisions of genera, most of which are identified. Whenever possible, nomenclature follows Luther and Sieff (1996). Originally, this volume was planned as yet another of the familiar collections of contributed chapters published by specialists. However, something closer to single authorship proved to be more conducive to an integrated product – a volume that weaves together the many dimensions of Bromeliaceae that affect where its members occur and how they interact with other biota. For this book to warrant its title, more of the characteristics molded by natural selection than mentioned in the other monographs of families must be described relative to effects on plant performance and cast as products of natural selection, i.e., to the extent possible, set in environmental contexts. Four authorities were asked to assist in the preparation of Chapters 7 and 9. One of these people, along with three additional experts, provided what were retained of the formerly envisioned, more extensive set of specialized topics. Four of these chapters (10–13) give us a snapshot of the more traditional approach to plant systematics and evolution as applied to Bromeliaceae. They illustrate two issues worth the attention of botanists contemplating work yet to do: (1) the extent of current taxonomic ambiguity and nomenclatural confusion in just four of the more than 50 bromeliad genera, and (2) the bases upon which authorities have circumscribed taxa and attempted to infer phylogeny and identify the origins of key features like the absorbing trichome and impounding shoot. This book is divided into three parts beginning with a short overview of Bromeliaceae intended to set the stage for the following core of eight chapters devoted to vegetative and reproductive structure and function, ecology, associations with other organisms, and finally evolution and phylogeny. The third section presents the special topics. Together, these treatments provide a detailed overview of how and why one family of flowering plants, and a truly exceptional one by virtue of adaptive specialization to counter xii Preface drought, came to assume such extraordinary importance in the Neotropics and occupy so many kinds of often demanding ecospace. Clearly, the adaptive history of Bromeliaceae will continue to unfold as inquiry targeting systematics, ecophysiology and ecology proceeds. Material provided in this volume should promote future discovery by highlighting on-going controversies and identifying topics that seem especially ripe for further research. Acknowledgments My interest in Bromeliaceae began more than 45 years ago during a family vacation in southeastern Florida. I recall how a colony of Tillandsia fasciculata bearing bright orange inflorescences and perched high in cypress trees just across the street from our hotel beckoned while the surrounding swamp prevented closer inspection. About a year later, back in Ohio, a friend who had also visited Florida showed me an even stranger sight – an air plant, he reported confidently – fully capable of living on air, hence quite able to flourish tied to its burnished cypress knee suspended from the ceiling of his enclosed porch. My skepticism about the fate of that forlorn T. paucifolia specimen and interest in bromeliads generally were forgotten until a stint as a graduate student assigned to work at the University of Michigan Botanical Gardens reintroduced me to those remarkable monocots. Much of my research since that encounter has been devoted to Bromeliaceae and the phenomenon of epiphytism so prevalent in this family. Interest in Bromeliaceae has taken me through much of tropical America and provided opportunity to work with all kinds of bromeliad enthusiasts including a number of hobbyists who continue to make major contributions to bromeliad taxonomy. I am particularly grateful to members of this dedicated group and the botanists who helped assure my success in field work in remote parts of Brazil (Pedro Nahum, Elton Leme), Ecuador (Calaway Dodson, David Bermudes), Venezuela (Tom Givnish) and Mexico (Germán Carnevali), to name just a few. Members of the Bromeliad Society and the Sociedade Brasileira de Bromelias are greatly appreciated for their warm receptions on many occasions and frequent encouragement of my work. I also wish to acknowledge assistance during the production of this volume provided by Kaelyn Stiles. Toni Renfrow, my research associate for more than 20 years, deserves special recognition for assistance that ranged from the collection of data to xiii xiv Acknowledgments extensive record keeping and editing of most of my publications. I owe more to her for my accomplishments in the pursuit of bromeliad biology than to anyone else. A host of Oberlin College undergraduates worked with me on many projects, more than a dozen sharing authorship of a larger number of publications. Financial support for my work with bromeliads has come from the United States National Science Foundation, the United States Park Service, The National Geographic Society, Oberlin College, and travel grants from a number of public and private sources in this country and abroad. The Marie Selby Botanical Gardens and its staff have been particularly helpful in providing support ranging from lodging to greenhouse and library facilities. Glossary Accidental epiphyte A typically terrestrial species with occasional members that grow to maturity while anchored on trees. Anemochory The dispersal of seeds by wind. Animal-assisted saprophyte A phytotelm bromeliad dependent on litter for nutrients. Ant-fed, ant-house bromeliad A species that produces hollow organs (myrmecodomatia) specifically for housing ant colonies. Ant-nest garden bromeliad A species that regularly to exclusively roots in arboreal ant nests. Atmospheric bromeliad A species directly dependent on the atmosphere for moisture and nutrient ions (member of Type Five). Axenic Applied to trees that by nature do not support epiphytes. Bromelioid An adjective applied to taxa assigned to subfamily Bromelioideae. CAM-cycling A photosynthetic syndrome characterized by diurnal CO2 fixation and nocturnal recapture of respired CO2. CAM-idling A condition of stressed CAM plants that is characterized by continuous closure of the stomata and energy maintenance through internalized CO2 recycling. Capacitance An aspect of water relations related to storage capacity. Relative capacitance is expressed as change in RWC per unit change in C (DRWC/DC). Carton The composite material some ants and termites use to build their nests and runways. Chiropterophily Regular dependence on bats to disperse seeds or pollen. Clade A group of species (lineages) that share a single ancestral lineage. Cleistogamy The condition of a flower that promotes seed production without presenting anthers or stigma to pollinators. Decarboxylase An enzyme that catalyzes the release of CO2 from an organic acid. Diazotroph An organism capable of fixing N2. xv xvi Glossary Domatium (myrmecodomatium) A plant cavity regularly occupied by nesting ants. Earth soil Soil exploited by terrestrial flora (opposite of suspended soil). Entomophily Regular dependence on insects to disperse seeds or pollen. Epiparasite A parasite that taps its host via a fungus. Epiphyll A nonvascular plant that inhabits the surfaces of foliage. Eutroph A plant native to fertile substrates. Everwet forest A forest that receives enough rainfall through the year to support predominantly drought-sensitive vegetation (opposite of seasonal forest). Facultative CAM Carbon fixation via CAM or C3 pathway depending on growing conditions. Facultative epiphyte A species that grows either epiphytically or rooted in earth soil, often emphasizing one or the other habit at a particular location. Frugivore A fruit-eating and often seed-dispersing animal. Genet The product of a single zygote, one shoot with its roots for the nonbranching (monocarpic) bromeliad, or the potentially numerous attached ramets produced by the sympodial (polycarpic) type. Guild A group of co-occurring but not necessarily related species that utilize one or more common resources. Halophyte A plant native to saline habitats. Haustorium The invasive appendage of a parasitic plant. Hemiepiphyte A plant that accesses earth soil with roots for only part of its life and anchors in a tree crown the rest of the time. Homiohydry The condition of maintaining tissue water content relatively independent of ambient humidity (opposite of poikilohydry). Homoptera The taxonomic order of arthropods that includes plantsucking forms exemplified by aphids and scale insects. Hypodermis A subepidermal zone of usually achlorophyllous, thin-walled, collapsible water-storage cells in leaves. Iteroparity The type of reproductive timing of a plant characterized by repeated fruiting over as many seasons (polycarpy) rather than once as in the monocarpic or semilaparous specimen. Lineage The unbroken succession of generations that constitutes the history of a taxon through geologic time. Monocarpy The type of reproductive timing characterized by a single episode of fruiting before the plant dies (semilaparity). Myrmecochory The dispersal of seeds by ants. Myrmecophyte A plant that regularly receives benefit from an associated ant colony. Neoteny A type of heterochrony whereby descendants as adults possess features that were characteristic of the juvenile stages of ancestors. Glossary xvii Nitrogenase The enzyme complex responsible for reducing N2 to organic form. Nutritional piracy The process whereby epiphytes intercept nutrients moving between the supporting tree crown and the forest floor. Oligotroph A plant native to nutrient-deficient substrates. Ornithochory Regular dependence on birds to disperse seeds. Phorophyte A woody plant that supports vascular epiphytes. Phytotelma The cavity formed by a plant to contain a phytotelmata. Foliage plays this role in Bromeliaceae. Phytotelmata A natural plant cavity filled with water often inhabited by aquatic organisms. Phytotelm bromeliad A bromeliad that produces a phytotelm. Pitcairnioid An adjective applied to taxa assigned to subfamily Pitcairnioideae. Poikilohydry The condition of maintaining tissue water content at levels strongly influenced by ambient humidity (opposite of homiohydry). Ramet The individual offshoot or module of a sympodially branched, herbaceous plant. Relative water content (RWC) A measure of water relations calculated as fresh weight – dry weight/turgid weight – dry weight. Reproductive index The proportion of the mature plant body committed to seeds and associated reproductive tissue. Rupestral A plant that grows on rocky soils. Saxicole A plant that grows on rock (a lithophyte). Sciophyte A plant tolerant of deep shade. Sclerophylly The condition describing evergreen foliage that contains much sclerified tissue. Seasonal forest A forest characterized by one or more distinct dry seasons (opposite of everwet forest). Succulent The term describing stems and leaves much thickened to store extraordinary amounts of water. Suspended soil (humus) Soil-like rooting media suspended in the canopies of many tropical and fewer temperate forests (opposite of earth soil). Terrestrial bromeliad A species that typically roots on the ground (in earth soil). Tillandsioid An adjective applied to taxa assigned to subfamily Tillandsioideae. Transpiration ratio (TR) A coefficient produced by dividing the mass of water lost in transpiration by the simultaneous gain in weight attributable to photosynthesis. Trophic myrmecophyte See ant-fed, ant-house bromeliad. Zoochory The dispersal of seeds by fauna. Abbreviations Amax AR chl CAM cpDNA C3 C4 C3–CAM E g H1max MPa MUE PAR PCRC PEPc PPFD PPNUE r RH RuBPc/o RWC TR VAM VPD WUE D DH1 p C ‰ Cleaf maximum photosynthetic capacity acetylene reduction as in the assay for nitrogenase chlorophyll crassulacean acid metabolism chloroplast genome Calvin/Benson photosynthetic pathway Hatch and Slack photosynthetic pathway facultative CAM transpiration diffusive conductance (leaf) maximum acidification for CAM plants megapascal mineral-use efficiency photosynthetically active radiation photosynthetic carbon reductive pathway phosphoenolpyruvate carboxylase photosynthetic photon flux density potential nitrogen-use efficiency Malthusian coefficient relative humidity ribulose bisphosphate carboxylase/oxygenase relative water content transpiration ratio vesicular-arbuscular mycorrhiza vapor pressure deficit water-use efficiency carbon isotope ratio; 13C enrichment in parts per thousand diurnal change in titratable acidity solute potential bulk water potential parts per thousand bulk leaf water potential xviii Part one Brief overview 1 Introduction Lower and middle Cretaceous Magnoliophyta remain too poorly known to warrant definitive statements about many aspects of early angiosperm radiation (Taylor and Hickey 1992). Discovery of a compressed infructescence purportedly from the Late Jurassic of east central Asia has recently expanded its confirmed record (Sun et al. 1998), and raises the specter of more fossils and better resolution ahead. Nevertheless, until this promise is realized, answers to questions as fundamental as the habits (woody vs. herbaceous) of ancestors and the homologies of diagnostic organs (e.g., the gynoecium) will remain speculative. One point germane to bromeliad history is less equivocal: characteristic pollen and macrofossils indicate that Liliopsida had emerged by the middle Cretaceous. However, evidence from several quarters indicates that Bromeliaceae evolved later, and probably not before the Tertiary. Phytogeography also accords with youth that denied Bromeliaceae opportunity to range beyond tropical America except for a single, probably recent dispersal to west Africa (Fig. 1.1). Members of the three subfamilies (sensu Smith and Downs 1974, 1977, 1979) and many of the larger genera (e.g., Neoregelia, Hechtia) further suggest either exceptionally low mobility (unlikely) or too little time to cross barriers breached by many other lineages. Nevertheless, most authorities (e.g., Cronquist 1981; Dahlgren et al. 1985) consider Bromeliaceae phylogenetically isolated among the extant monocots, and a growing body of information on the organization of several sequences of nucleotides within the chloroplast genome (e.g., Ranker et al. 1990; Terry et al. 1997a,b) supports this conclusion. Uncertainty continues over which of the other Liliopsida are most closely related to the bromeliads, particularly which family constitutes the sister group, i.e., shares a common ancestor with Bromeliaceae. Certain gymnosperms and the flowering plants considered primitive 3 4 Introduction Figure 1.1. Geographic distribution of Bromeliaceae. according to the paleoherb hypothesis challenge long-standing notions about the nature of antecedents and the characteristics of the Magnoliophyta that favored its ascent to unparalleled size and ecological dominance among land flora. Hypotheses that zoophilous pollination and certain additional aspects of reproduction drove the angiosperm radiation to unparalleled heights must now accommodate discovery that most of these same attributes occur (albeit in less advanced expressions) elsewhere, especially among the gnetophytes (e.g., Friedman 1992; Kato et al. 1995). Whether inherited as an older, intact suite of characters or derived piecemeal during the initial Lower Cretaceous expansion, these qualities alone cannot fully explain the unprecedented success of the flowering plants. Novel vegetative form and function were also important, as the bromeliads so clearly demonstrate. Rather than the woody archaetype (as exemplified by the ranalean magnoliophytes) posited by the euanthial theory, the angiosperm stock is increasingly envisioned as low-growing shrubs to rhizomatous to scrambling herbs of moist, relatively disturbed (r-selecting), perhaps riverine, habitats (e.g., Taylor and Hickey 1992). Rapid maturation made possible by the combined effects of a novel nutritive tissue (endosperm), much abbreviated (fast maturation) male and female gametophytes, and relaxed Introduction 5 Table 1.1. Plant characteristics presumably responsible for the unprecedented radiation of Magnoliophyta Vegetative (1) Cheap construction (herbaceousness) (2) Rapid growth, potentially short life cycles (3) Exceptionally efficient vascular systems (4) Exceptionally diverse architecture (e.g., vines, herbs, trees) (5) Exceptionally plastic ecophysiology (carbon fixation pathways, H2O balance mechanisms) (6) Exceptionally broad capacity to utilize diverse resource bases (e.g., parasitism, carnivory, and other sources of nutrients unavailable to other flora) (7) Exceptional chemical/mechanical defenses Reproductive (1) The flower as a reproductive organ of unmatched capacity for precise and versatile function (2) Unmatched capacity to manipulate pollinators (3) Inexpensive, short-lived gametophytes (4) Endosperm (5) Devices to routinely screen male gametophytes (pollen tube competition and various pollen recognition systems) (6) Angiospermy and the associated possibilities for packaging seeds for (6) protection and directed dispersal needs for costly mechanical tissue probably account in large measure for global dominance by the flowering plants (Table 1.1). These characteristics, complemented by small size and versatile habits, account for the high densities of species in sites like humid tropical forests. Unmatched capacity to manipulate pollinators and seed dispersers in turn probably spurred the speciation necessary to stock the most densely packed modern communities. To what degree additional uniqueness, like angiospermy, which permits the maternal parent to screen haploid genotypes, and greater physiological variety (e.g., C4, C3 and CAM photosynthesis) influenced outcomes remains more speculative. Even though fossils and the geographic distributions of surviving lineages indicate phylogenetic youth, Bromeliaceae exceed many of the preTertiary clades (e.g., Fagaceae, Platanaceae, Juglandaceae) for number of species and especially for adaptive variety (e.g., diverse habits, habitats). Capacity to produce a simple, cheaply constructed, rapid-cycling body varies among the magnoliophytes, and helps explain why some families (e.g., Asteraceae, Poaceae, Orchidaceae) contribute more extensively to angiosperm diversity than predominantly woody groups. Additional plant 6 Introduction Table 1.2. Plant characteristics that account for the inordinate success of Bromeliaceae in diverse, often demanding, habitats Vegetative (1) Small herbaceous body (2) Rhizomatous habit (3) Propensity for heterochrony/heterophylly (4) Phytotelm shoot (5) Foliar trichome capable of replacing absorptive roots and providing additional services (e.g., light reflectance) (6) Propensity for CAM, succulence and other xeromorphic features Reproductive (1) Less decisive for family success, although pollination and seed dispersal (1) syndromes are diverse to match opportunities in disparate habitats characteristics, such as the tight relationships between numerous orchids and their high-fidelity pollinators and propensity to exploit underutilized ecospace (e.g., forest canopy), in turn account in part for the different sizes of the largely herbaceous clades. Although relatively modest by membership, perhaps because of weaker propensity for speciation, Bromeliaceae exceeds these largest taxa for certain other kinds of biological variety, and most certainly for importance to several kinds of fauna (e.g., mosquitoes). Structure and function itemized in Table 1.1 largely account for the relatively high success of the flowering plants overall, while those traits listed in Table 1.2 represent the finer-scale features that permit Bromeliaceae to surpass most other families on several counts that at least equal species richness as measures of biological importance. This family exhibits an unusually propitious combination of angiospermous qualities and some less pervasive ones conducive to life in widely available, underutilized and often physically demanding ecospace. These more exclusive attributes at once explain how one group of related species can be so ecologically versatile and stress tolerant, and also so often exceed co-occurring flora for impacts in hosting ecosystems. Members tolerate punishing drought as epiphytes and lithophytes; the hardiest terrestrials may not experience rainfall for months and, in the coastal deserts of northern Chile and southern Peru, even for years, surviving solely on more reliable supplies of fog water (e.g., Figs. 1.2, 7.1). Conversely, certain other bromeliads root in alpine bogs and additional kinds of wetlands, and a few populations spend part of each year submerged in flowing water (Fig. 1.4G). Exposures vary from the UV-Benriched irradiance that prevails at .4000 m in the central Andes (Puya) to Introduction 7 Figure 1.2. Bromeliads in situ. (A) Dyckia sp. growing in rocky soil of campos rupestres habitat in Minas Gerais State, Brazil. (B) Brocchinia tatei on marshy soil on Cerro Neblina, Venezuela. (C) Alcantarea regina on granite outcrop in Rio de Janeiro State, Brazil. (D) Large Aechmea angustifolia plant supporting diverse flora in eastern Ecuador. (E) Hohenbergia sp. growing as a terrestrial in Bahia State, Brazil. (F) Guzmania monostachia congregated in the lower crown of Annona glabra in south Florida swamp forest. (G) Vriesea gigantea, a typical phytotelm bromeliad in Espirito Santo State, Brazil. (H) Juvenile of Tillandsia streptophylla growing on the base of Rhizophora mangle in Yucatán State, Mexico. 8 Introduction the much attenuated photon flux under the canopies of evergreen forest (e.g., various species of Cryptanthus, Pitcairnia; Fig. 1.3D). Frost-hardiness adequate for survival at certain temperate latitudes or in tropical alpine habitats characterizes different sets of species. Access to key mineral nutrients runs the gamut from the meager supplies that oblige pronounced oligotrophy (e.g., the Tillandsia that clings to a small twig with its nonabsorptive roots; Fig. 1.3C) to relatively plentiful, for example the quantities provided by symbiotic biota that process the litter intercepted by the phytotelm shoots of hundreds of ‘tank species’ (e.g., Fig. 1.2C,G). Those qualities that grant Bromeliaceae exceptional tolerance for drought and capacity to grow on nutrient-poor substrates required modifications of certain fundamental angiosperm features, but not of others. Bromeliad flowers probably operate with roughly the same mix of breeding systems and attractants for pollinators expressed across Magnoliophyta. Pollen and seed dispersers, while also diverse, again seem unlikely to set records for promoting speciation, ecological variety or dominance for Bromeliaceae compared with other families. In effect, the bromeliads merit special note among flowering plants for the novelty of the vegetative rather than the reproductive characteristics of the most specialized species. What poised ancestors for life in epiphytic, lithic and other sparsely vegetated (underutilized) habitats where more than half of the bromeliads reside today was a body plan conducive to rapid cycling despite growing conditions that limit carbon gain and thus diminish vegetative vigor and reproductive power (Table 1.2; Figs. 2.1, 2.3). A remarkably adaptable leaf and shoot assist resource scavenging (for water and nutrients) and promote stress-tolerance (to drought, high and low exposure). Propensities for neoteny and specialized architectures that foster access to unconventional sources of moisture and nutrients and promote economy during the use of these commodities also encouraged radiation into exceptionally stressful habitats. An ecological taxonomy formulated by German morphologists and biogeographers over a century ago organizes the bromeliads according to often unusual plant features that allow success in widely disparate kinds of habitats (Table 4.2). Most important are aspects of roots, shoot architecture and the foliar trichome, which, depending on the mix of special modifications, favor carbon and water balance and mineral nutrition under relatively conventional to extreme growing conditions. Some suites of characteristics foster epiphytism at relatively humid sites (Types Three and Four), and another (Type Five), use of the same kinds of substrates in drier regions. Five types are recognized in all, and references to specific bromeliads and groups of species hereafter will often employ these designations Introduction 9 Figure 1.3. Bromeliads in situ (continued). (A) Tillandsia recurvata growing on telephone wires in southeastern Mexico. (B) Billbergia porteana growing on the trunk of a palm in Bahia State, Brazil. (C) Tillandsia paucifolia growing on a cypress twig in south Florida. (D) Cryptanthus bromelioides growing in the forest understory in Rio de Janeiro State, Brazil. (E) Feral Ananas comosus in southern Venezuela. (F) Aechmea nudicaulis extending out from a restinga ‘island’ along the coast of Rio de Janeiro State, Brazil. 10 Introduction Figure 1.4. Bromeliads in situ (continued). (A) Granitic dome (inselberg) covered with lithophytic Bromelioideae in Rio de Janeiro State, Brazil. (B) Caatinga with an understory rich in Bromeliaceae in Bahia State, Brazil. (C) Campos rupestres habitat in Minas Gerais State, Brazil. (D) Elfin forest in eastern Puerto Rico. (E) Restinga in Rio de Janeiro State, Brazil. (F) Remnant Atlantic Forest trees covered with bromeliads in pasture in Rio de Janeiro State, Brazil. (G) A riparian colony of Pitcairnia flammea in Rio de Janeiro State, Brazil. (H) Dwarfed cypress forest with bromeliads in south Florida. Introduction 11 (Table 4.2). Familiarity with this scheme is essential to understand bromeliad evolution and functional diversity. Chapter 2 starts the eight-chapter core with a description of how sympodial branching combined with determinant shoots bearing adventitious roots, or none at all, supports the bromeliads as hemiepiphytic vines, alpine cushion or giant rosette plants, myrmecophytes, carnivores or soil-dependent terrestrials among an even longer list of habits (e.g., Figs. 2.2, 5.3B, 6.12D). Modifications of the shoot, and particularly its epidermis, impart exceptional capacity to endure drought and impoverished substrates. Tolerances for the multiple physical constraints that prevail in the most exceptional habitats occupied by members of this family sometimes foster almost exclusive occurrences there (e.g., Figs. 1.2C, 7.1E). Crassulacean acid metabolism (CAM) promotes the water economy that helps many populations survive seasonal drought and avoid photodamage, while anchored on well-exposed bark and rocks. Similarly endowed relatives utilize wetter habitats with the most vigorous individuals often located in the shadiest microsites. In fact, CAM has been recorded in more members of Bromeliaceae than in any other family (Martin 1994). The nearly ubiquitous foliar trichome provides diverse services to Bromeliaceae ranging from protection against potentially injurious insolation and insupportable transpiration through secretion to absorption associated with diverse nutritional modes and moisture supplies (Chapters 4 and 5; e.g., Figs. 2.5, 2.8). Bromeliaceae exceed all other families for variety of sources of nutrients and water (Table 5.6). Foliar impoundments that make litter an option for nutrition in turn assure the so-called phytotelm types importance in communities far beyond what plant numbers or total phytomass usually predict (Fig. 2.4). Dense populations of bromeliads in forest canopies can also markedly influence fundamental system-wide processes and phenomena such as mineral cycling and hydrology. Bromeliad taxonomy remains provisional, and needs substantial improvement ranging from the reordering of species within many genera to the establishment of additional higher taxa to accommodate revelations fostered by accumulating molecular and traditional morphological data. Smith and Downs’s three subfamilies include exceptionally isolated lineages (e.g., Brocchinia, Catopsis, Glomeropitcairnia; Tables 1.3, 1.4) in addition to core taxa, and many clades are almost certainly para- or polyphyletic (e.g., Aechmea, Navia, Vriesea). Pitcairnioideae, while closest to the monocot ground plan by many measures, including the status of the trichome, basic plant architecture and reproductive morphology, is not, as often reported, ancestral to either of the other two subfamilies. 12 Introduction Table 1.3. Bromeliad diversity (number of species) across tropical America Location Colombia Ecuador Peru Rio de Janeiro State, Brazil Costa Rica Florida Venezuela Bromelioideae Pitcairnioideae Tillandsioideae Total 70 56 59 170 27 0 56 125 70 153 17 19 0 188 196 242 199 124 145 17 120 391 368 411 311 191 17 364 Source: From Fontoura et al. (1991) and Holst (1994). Bromelioideae and Tillandsioideae followed parallel evolutionary trajectories to become heavily epiphytic and dependent on foliar impoundments and CAM. Certain other features diverged at least as much (e.g., fruit types, reliance on foliar trichomes). DNA sequences are beginning to help align and redefine the genera (e.g., Tillandsia/Vriesea), and should eventually demonstrate how often, when, and under what conditions important events, like the emergence of the absorbing trichome and CAM, occurred during bromeliad history. Many aspects of vegetative structure and function are homoplasious (e.g., CAM, phytotelm shoot), as are most of the many pollination syndromes recorded for the family. Specialized Bromeliaceae, and some other flora from comparably demanding habitats, inspired inquiry that helped launch the discipline of physiological ecology during the late 19th century. Early functional morphologists and biogeographers, including A. F. W. Schimper, C. Mez and G. F. J. Haberlandt, firmly established the principle that plant function tracks structure, and that both variables reflect growing conditions in situ. Some of the most elegant examples came from experiments performed on advanced Tillandsioideae, specifically those subjects labeled ‘atmospherics’ (Type Five; Table 4.2) because they rely on foliar trichomes to absorb airborne water and nutrients instead of the roots most land flora employ to obtain the same resources from soil. Major contributors since then include C. S. Pittendrigh (1948) who also anticipated some of the discoveries of the current generation of ecophysiologists by postulating how plant habit and aspects of leaves and roots account for the distribution of Trinidad’s bromeliads. His work also helped validate the ecological classification provided in Table 4.2. Bromeliads occupied a prominent place in Leopoldo Coutinho’s efforts in the late 1940s through the mid-1960s (e.g., Coutinho 1963) to demonstrate the 13 Introduction Table 1.4. The bromeliad genera: selected statistics, ecological type and geographic range Genus Number of Ecological speciesa type Acanthostachys Aechmea Alcantarea Ananas Androlepis Araeococcus 2 220 15 7 1 5 I Mostly III Mostly IV II III I and III Ayensua Billbergia Brewcaria Brocchinia Bromelia Canistrum Catopsis Connellia Cottendorfia Cryptanthus Deinacanthon Deuterocohnia Disteganthus Dyckia Encholirium Fascicularia Fernseea 1 62 2 17 49 11 21 5 1 42 1 14 3 120 30 5 2 I III I I and IV I and II III IV I I I II I I–II I I I I 18 I 2 IV Fosterella Glomeropitcairnia Greigia Guzmania Hechtia Hohenbergia Hohenbergiopsis Lindmania Lymania Mezobromelia Navia Neoglaziovia Neoregelia Nidularium Ochagavia Orthophytum Pepinia 28 175 51 47 I and II I and IV I III 1 36 6 9 95 3 95 54 3 26 48 III I III IV I I III III I I I Geographic range East central Brazil Tropical America Southeastern Brazil South America Central America Southeastern Brazil to northern South America Guayanan Shield Tropical America Guayanan Shield Guayanan Shield Tropical America Southeastern Brazil Predominantly Mesoamerica Guayanan Shield Bahia and adjacent states, Brazil Southeastern Brazil Argentina and Paraguay Mostly Bolivia Guianas Southeastern South America Southeastern Brazil Chile Cerro Italia, São Paulo State, Brazil Predominantly west central South America Lesser Antilles, Trinidad and adjacent Venezuela Predominantly Andean Tropical America Predominantly Mexican Mostly Jamaican and southeastern Brazil Mexico and Central America Guayanan Shield Southeastern Brazil Andean Guayanan Shield East central Brazil Southeastern Brazil Southeastern Brazil Chile (San Fernandez island) Southeastern Brazil Predominantly Amazonian 14 Introduction Table 1.4. (cont.) Genus Number of Ecological speciesa type Pitcairnia Portea Pseudaechmea Pseudananas Puya Quesnelia Racinaea Ronnbergia 295 9 1 1 194 15 57 11 I III III II I III IV III Steyerbromelia Tillandsia Ursulaea Vriesea Werauhia Wittrockia 3 518 2 227 64 12 I I, IV, V III I, IV, V IV III Geographic range Tropical America Southeastern Brazil Colombia and Bolivia Southeastern Brazil Predominantly Andean Southeastern Brazil Mostly Andean Panama to Peru, Southeastern Brazil Guayanan Shield Tropical America Mexico Tropical America Primarily Mesoamerica Southeastern Brazil Source: aFrom Luther and Sieff (1996). mechanisms of photosynthesis among Neotropical epiphytes. Bromeliaceae continue to attract investigators seeking more complete answers to questions about carbon, water and nutrient balance, aspects of reproduction, and phylogenetic relationships as detailed in the following eight chapters. Another set of pioneering biologists (e.g., Picado 1911, 1913) chose to study this family because they recognized the importance of the bromeliad phytotelmata to extensive fauna in many tropical American forests. Foliar impoundments reportedly harbor high diversities and abundances of aquatic and soil-type invertebrates, sometimes at densities above those encountered in equivalent volumes of nearby forest soil (e.g., Paoletti et al. 1991; Fig. 8.15). Several more studies provide data on the physical and chemical conditions in these microcosms, and yield insights on why certain bromeliads host so many symbionts. Checklists indicate potentials for litter processing and nutrient release comparable to those that prevail in more conventional rooting media (Table 8.2). Broader perspectives suggest that epiphytic Bromeliaceae, acting with certain other arboreal flora, intercept and release key nutrients in ways that either augment or deprive co-occurring flora depending on conditions at the site (Fig. 7.18). The eight-chapter core that follows these preparatory remarks also considers reproductive morphology, which, along with profiling the vegetative Introduction 15 body, sets the stage to move on to basic life functions. Evolution is reserved for the final installment. A modest third section contains short chapters authored by specialists who treat several genera and the ethnobotany and conservation of Bromeliaceae. As information continues to accumulate, additional, specialized subjects will be able to be included in future volumes, along with updates of the core chapters on basic structure and function, ecology and family history. Part two Basic structure, function, ecology and evolution 2 Vegetative structure All of the impressive functional and ecological variety expressed by some 2700 species of Bromeliaceae is grounded on a single body plan, or what Hallé et al. (1978) might consider one architectural model. Widespread occurrence of this same design among extant monocots and the paleoherb hypothesis (Taylor and Hickey 1992) suggest that early Magnoliophyta possessed much the same basic organization. Except for the occasional monocarp, a somewhat larger group of relatively caulescent species (Fig. 2.1), and another modest-sized assemblage of lateral-¯ owering taxa (Fig. 2.2B), the bromeliads share a distinctly modular bauplan characterized by sympodial branching that leads to series of attached, compact, terminally ¯ owered ramets (Fig. 2.3). Roots, if present beyond the seedling stage, mostly emerge along the lower half of each module. Vegetative form that favors life on arboreal and lithic substrates also imparts substantial horticultural value to many of the bromeliads. Moreover, some of these same features assure exceptional importance in ecosystems, including indispensability to extensive fauna with diverse needs (Chapter 8). Two plant characteristics warrant special note on all three counts: a generally compact, rosulate shoot (the ramet or module) that often impounds moisture and nutrient-rich solids (creates the phytotelma and consequently the phytotelmata; Fig. 2.4) and the usually peltate foliar trichome (Figs. 2.5± 2.9). These attributes, combined with others involving roots and shoots, favor success, including occasional dominance in some of the most exacting kinds of ecospace colonized by vascular ¯ ora in tropical America (e.g., Figs. 1.2C, 7.1). Observations of the kind initiated by some of the most renowned of the pioneering European morphologists constitute much of the vast literature on Bromeliaceae. Contributions dealing with vegetative structure continue and increasingly incorporate more revealing techniques, particularly 19 Cambridge Books Online © Cambridge University Press, 2009 20 Vegetative structure Figure 2.1. Schematic diagram illustrating neoteny in Tillandsioideae whereby an ancestor with mesomorphic foliage organized to maintain a phytotelmata gave rise to descendants that lack phytotelm architecture and extensive root systems and instead exhibit overall miniaturization combined with either reduced or increased numbers of leafy nodes. All scale bars 51 cm. See text for additional details. electron microscopy (e.g., Benzing et al. 1978) and histochemistry (e.g., Owen et al. 1988). Tomlinson (1969) devoted a substantial portion of Volume 3 of the Anatomy of the Monocotyledons to the most recent review of this information. We gratefully acknowledge the importance to our treatment of Tomlinson' s synthesis and the growing body of related, interpretative information being amassed by plant physiologists and ecologists. Our primary concern here is those aspects of vegetative structure that distinguish Bromeliaceae among families (e.g., epiphytism) and foster importance in ecosystems. Featured species showcase adaptive morphology, and, for example, illustrate how shoot architecture in¯ uences access to Cambridge Books Online © Cambridge University Press, 2009 Vegetative structure 21 Figure 2.2. Bromeliad architecture. (A) Dyckia sp. in vegetative state. (B) Dyckia sp. with lateral in¯ orescence. (C) Hemiepiphytic Pitcairnia sp. illustrating heterophylly. (D) Neoregelia abendrothae ramets with only juvenile or juvenile and adult foliage. (E) Brocchinia acuminata, sun (compact) and shade (caulescent) forms. (F) Ronnbergia ecuadoriana illustrating putatively primitive architecture. (G) Cottendorfia florida with leaves cut short to expose thick, ® re-resistant stem. (H) Distichous Dyckia estevesii. Cambridge Books Online © Cambridge University Press, 2009 22 Vegetative structure Figure 2.3. Schematic diagram illustrating three patterns of growth in Bromeliaceae. (A) Sympodial branching with determinant ramets. (B) Monocarpy. (C) Monopodial with axillary ¯ owering. resources that most plants obtain from soil. Finer details of carbon management, water balance and mineral nutrition are deferred to later chapters. Likewise, taxonomy receives scant attention in this chapter except where classi® cation happens to parallel form (e.g., foliar trichomes) that also in¯ uences plant performance. In the ® nal analysis, our subject is how what seems to be the fundamental monocot body plan, combined with often novel arrangements and modi® cations of leaves, permits Bromeliaceae to occur in most of the life zones comprising the American tropics. Habits: general overview Bromeliads range from small plants even by liliopsid standards to some of the most massive-bodied of the monocots. More comparable among the Cambridge Books Online © Cambridge University Press, 2009 Habits: general overview 23 Figure 2.4. Shapes of phytotelm (tank-producing) Bromeliaceae. (A± D) Four arrangements of foliage that produce phytotelmata of different numbers, exposures and depths per shoot. (E) Aechmea veitchii with virtually no impoundment capacity. (F) Carnivorous Brocchinia reducta. (G) Mature shoot of Aechmea bracteata cut open to expose central dry chamber for ants and several older leaf bases con® gured to intercept precipitation and litter. (H) Nidularium burchellii, discolorous foliage arranged in a monolayer. (I) Tillandsia lucida, multilayered shoot. (J) Aechmea brevicollis, distichous phyllotaxis. (K) Billbergia porteana, tubular shoot. (L) Aechmea brassicoides, central leaf forming dry chamber. Cambridge Books Online © Cambridge University Press, 2009 24 Vegetative structure Figure 2.5. Trichomes of Pitcairnioideae. (A± D) Goblet-shaped trichome of carnivorous Brocchinia reducta, illustrated in section (A), view from top (B), and labyrinthine outer wall of a distal cell in the hydrated (C) and dry (D) conditions. (E) Fosterella penduliflora, in section. (F) Shield. (G) Brocchinia tatei, in section. (H) Shield. (I) Brocchinia micrantha, shield. (J) In section. (K) Navia glandulosa, glandular trichome from sepal (left) and ¯ oral bract (right). (L) Uniserrate trichome from juvenile leaf of Navia sp. (M) Trichome shield of Lindmania serrulata. (N) Uniserrate trichome of Lindmania wurdackii. Parts E, F, L, M, N redrawn from Tomlinson (1969). rhizomatous types are the proportions of the individual ramets, or, for the monocarp, just the seedling shoot because these species never branch (Fig. 2.3). The mature seedling and each of its subsequent ramets weighs from a few grams fresh weight (e.g., neotenic Tillandsia and miniaturized Brocchinia species; Fig. 2.1) to thousands of kilograms for those of the Cambridge Books Online © Cambridge University Press, 2009 Habits: general overview 25 Figure 2.6. Trichomes of Bromelioideae. (A) Aechmea penduliflora, in section. (B) Shield. (C) Billbergia brasiliensis, in section. (D) Shield. (E) Canistrum sp., in section. (F) Shield. All parts redrawn from Tomlinson (1969). largest sympodial types. Monocarpic Puya raimondii at maturity exceeds all the other Bromeliaceae in mass and height and probably in the number of years required for its unitary body to achieve ¯ owering size (Fig. 14.2C). The individual bromeliad shoot typically consists of a short stem bearing a few to many, closely placed, alternate, usually spirally arranged, strapshaped to ® liform leaves. Just two organs, one leaf and an enlarged prophyll, constitute each typically rootless ramet of neotenic Tillandsia usneoides (Figs. 2.1, 2.10E). Hundreds of leaves characterize some of its more caulescent relatives (e.g., monocarpic Puya, Tillandsia araujei). Cryptanthus bromelioides and a number of other members of the same genus bear smaller leaves along the rhizome compared with those at its expanded terminus (Fig. 2.11C,D). Certain Bromelia and similarly stoloniferous members of many additional bromelioid genera exhibit even stronger dimorphism, as do many Pitcairnioideae (e.g., Pitcairnia; Figs. 2.2C, 2.12B). Slender juvenile leaves appear on the shoot of Neoregelia abendrothae before the broader utriculate organs that can trap litter and water develop (Figs. 2.2D, 9.12). Heterophylly is less pronounced in Tillandsioideae where, nonetheless, it has provoked more speculation about evolutionary mechanisms (Fig. 2.11B; Chapter 9). Leaf size and proportions, particularly the shape of the base and number per shoot, in¯ uence ecophysiology and accordingly the suitability of speci® c substrates and climates for certain bromeliads. Phyllotactic fractions range from 2/5 to 5/13 and probably go higher among the caulescent species Books Online © Cambridge University Press, 2009 26 Vegetative structure Figure 2.7. Trichomes of Tillandsioideae. (A± B) Trichome of Tillandsia ionantha showing con® guration of shield when dry (A) and wet (B). (C) Rigid trichome shield of Tillandsia bulbosa, abaxial surface. (D) Trichome shield of Tillandsia crocata. (E) Trichome shield of Tillandsia recurvata. (F) Trichome shield of Tillandsia karwinskyana. (G) Trichome of Catopsis nutans, in section. (H) Shield. (I) Leaf of Tillandsia hildae showing banding attributable to presence of trichomes distinguished by the widths of the shields. Parts D, G, H, redrawn from Tomlinson (1969). that bear numerous narrow leaves (e.g., Tillandsia funckiana, T. filifolia; Fig. 2.1). However, even the most congested foliage of this type casts little self-shade except where the blades are imbricate. Overlapped leaves along the tiny shoots of caulescent Tillandsia bryoides ¯ ex outward most while the plant is fully hydrated (Fig. 2.1). Other exceptional taxa exhibit distichous organization (Dyckia estevesii, T. usneoides, Cambridge Books Online © Cambridge University Press, 2009 Habits: general overview 27 Figure 2.8. Abaxial leaf surfaces of representative Bromeliaceae; scanning electron micrographs. (A) Aechmea bracteata (3150). (B) Catopsis nutans (3150). (C) Tillandsia tectorum (3100). (D) Pitcairnia macrochlamys (3150). (E) Tillandsia ionantha (3175). (F) Bromelia sp. (3150). Books Online © Cambridge University Press, 2009 28 Vegetative structure Figure 2.9. Ontogeny of the trichome of Tillandsia usneoides viewed in section (A series) and from top (B series). Redrawn from Billings (1904). T. recurvata; Fig. 2.2H). Occasionally, patterns shift from spiral to distichous along the same shoot (e.g., T. paleacea), or they approach the orthodistichous condition (leaves two-ranked in a slight spiral; e.g., T. myosura). Distichous phyllotaxis is more common among seedlings, especially in Tillandsia subgenus Diaphoranthema, where in the adult it denotes juvenilization. Axillary buds occur along the entire length of the typical bromeliad shoot, but few ¯ ush and, except for the lateral-¯ owered species, those that do produce the standard one or two reiterative ramets (Fig. 6.14). Some of the longer-stemmed saxicoles (Tillandsia diaguitensis) and certain scrambling Pitcairnia species (Fig. 2.12B) branch less predictably, possibly according to physiological status or some external cue like photoperiod. Pitcairnia riparia branches whenever its stolons encounter obstructions that block forward progress. Replacement meristems routinely activate following ¯ oral induction that culminates shoot development with as few as one (Fig. 3.3L) to thousands of ¯ owers arrayed on a well-de® ned in¯ orescence (Figs. 3.2± 3.4). Leaves with armed margins characterize most Bromelioideae, and many Pitcairnioideae, presumably to discourage large herbivores (Figs. 2.12± 2.14). The epiphytes usually display weaker mechanical defenses than the terrestrials, which if native to arid soils (e.g., certain Bromelia, Hechtia) invest most heavily in spines. Most bromeliads can replace a lost apical meristem with an axillary bud, but apparently regeneration proceeds slowly enough and predation is sufficiently high in many habitats to justify high Cambridge Books Online © Cambridge University Press, 2009 Habits: general overview 29 Figure 2.10. Plant architecture and leaf anatomy of representative Tillandsioideae. (A) Tillandsia usneoides, leaf cross-section. (B) T. recurvata, leaf cross-section. (C) T. setacea, leaf cross-section. (D) T. crocata, leaf cross-section. (E) T. usneoides, shoot. (F) T. usneoides, cross-section leaf vein. (G) T. usneoides, cross-section leaf epidermis. (H) T. recurvata, cross-section leaf epidermis. (I) Catopsis floribunda, leaf cross-section. (J) Tillandsia fasciculata, leaf cross-section. (K) T. fasciculata, stomata. (L) T. duratii illustrating leaves capable of holdfast. (M) T. ionantha var. van-hyningii (saxicole). (N) T. ionantha var. zebrina (epiphyte). Parts A± D, F± K redrawn from Tomlinson (1969). Cambridge Books Online © Cambridge University Press, 2009 30 Vegetative structure Figure 2.11. Asexual propagation and related morphology. (A) Tillandsia flexuosa with immature axillary ramet and additional offshoots on spent in¯ orescence. (B) Grass-like basal ramets produced by many soft-leafed Vriesea species. (C± D) Stoloniferous Cryptanthus sp., mature ramet (C) and immature ramet (D). (E) Stoloniferous Nidularium lymanioides growing as a hemiepiphyte. cost to protect the shoot tip. Soft-leafed exceptions include some Cryptanthus species in Bromelioideae and Brocchinia and Fosterella of Pitcairnioideae; according to certain sequences in the chloroplast genome these genera lie beyond the core taxa of their respective subfamilies (Fig. 9.20). Unexpectedly well-defended foliage born by members of some arboreal Bromelioideae (e.g., Aechmea bracteata; Fig. 2.4G) raises the possibility of recent ancestors that rooted on the ground. Sympodial Bromeliaceae branch at different locations along the parent axis depending on the species (Fig. 6.14). Buds inserted at midstem or Cambridge Books Online © Cambridge University Press, 2009 Habits: general overview 31 Figure 2.12. Growth habits, rhizome bracts and foliage of certain Pitcairnioideae. (A) Drought-deciduous Pitcairnia heterophylla. (B) Scandent Pitcairnia riparia. (C) Pitcairnia sp. equipped with rhizome bracts lacking armature below green foliage with expanded blades. (D) Pitcairnia sp. bearing short, spiny basal leaves that progressively give way to smooth-margined, broader and longer photosynthetic types. (E) Single leaf of nonheterophyllic Pitcairnia feliciana. (F) Swollen leaf bases and bulb-like habit of Puya pusilla. somewhat below suffice for most taxa. Two sets of ramets, the ® rst quite small, grass-like and positioned near the base of parent shoots that themselves are still much less than ® nal size, characterize numerous Tillandsioideae (e.g., some Vriesea species; Fig. 2.11). Later, after the mother ramet begins to ¯ ower, one or two more robust offshoots emerge from as many leaf axils midway along the shoot to just below the in¯ orescence. Still other species fail to branch the second time (e.g., Alcantarea Cambridge Books Online © Cambridge University Press, 2009 32 Vegetative structure Figure 2.13. Aspects of leaves of Bromelioideae. (A) Spiny blade margins of Bromelia balansae. (B) Cross-section of blade of Bromelia balansae half way between apex and base illustrating collapsible adaxial hypodermis, stomata and stellate chlorenchyma. (C) Aechmea magdalenae, abaxial epidermis. (D) Hohenbergia urbanianum, cross-section of blade. (E) Portea petropolitana, section revealing nonvascular ® ber bundles. (F) Leaf silhouettes illustrating four patterns of anthocyanin development (dark areas) common in Neoregelia and encountered less often in several other genera. Parts B± E redrawn from Tomlinson (1969). imperialis, Puya dasylirioides; Chapter 6), rendering them essentially monocarpic. Basal ramets in these instances exist primarily to continue the genet should the seedling meristem fail to mature. Exceptional Tillandsioideae and some Orthophytum augment sympodial branching with offshoots where ¯ owers failed to set fruits (Fig. 2.11). Location suggests origin from buds in the axils of ¯ oral bracts that for most Cambridge Books Online © Cambridge University Press, 2009 Habits: general overview 33 Figure 2.14. Aspects of shoots and roots of Bromeliaceae. (A,B) Shoot of Bromelia sp. sectioned and intact illustrating dense masses of trichomes on leaf bases and absence of substantial impoundment capacity. (C) Holdfast roots of Tillandsia edithiae. (D) Abundant apogeotropic roots exposed by removing the leaf bases of caulescent Brocchinia micrantha. (E) Stoloniferous epiphytic Neoregelia sp. (F) Aechmea chantinii illustrating banded distribution of trichomes on abaxial leaf surface. (G) Banded pigmentation marking the leaves of Vriesea fosteriana. (H) Billbergia sp. illustrating irregular spotting on foliage. Cambridge Books Online © Cambridge University Press, 2009 34 Vegetative structure species remain dormant unless activated by injury farther up the in¯ orescence. Monopodial types progressively die from the rear forward at the same time as the shoot apex adds replacements, including a succession of lateral in¯ orescences (e.g., Tillandsia complanata, some Dyckia, Greigia; Figs. 2.2B, 2.3C). Inspection of certain reputed cases of monopody (e.g., Tillandsia multicaulis) reveals the usual sympodial condition that leaf-like bracts obscure when the replacement meristem arises immediately below the spent apex. Exceptional sympodial bromeliads representing all three subfamilies spread via axillary stolons up to several meters long that propagate above or below ground depending on the taxon (e.g., Cryptanthus, Pseudananas; Fig. 2.11D). Quite a few Pitcairnia, certain Cryptanthus and many Tillandsia, among others, possess more upright, leafy, caulescent habits (Fig. 2.12). Some of these plants (e.g., Nidularium lymanioides; Fig. 2.11E) scramble through the lower canopy as hemiepiphytes following establishment on the ground, or they germinate in the canopy and then grow from branch to branch (e.g., Pitcairnia riparia; Fig. 2.12B). Extensive, ® brous root systems characterize all Bromeliaceae except the most diminutive, dry-growing Tillandsioideae. Less typical for Liliopsida, each organ travels basipetally from its point of origin inside the stem through many nodes before emerging to penetrate adjacent substrates (Fig. 2.15). Mycorrhiza occur sporadically, but few reports identify the fungi and none document a plant bene® t (Chapter 5). Sclerotic cortical and stelar parenchyma provide the strength and durability the slow-growing epiphytes and saxicoles require for prolonged suspension. Absorptive capacity probably varies among taxa and within genotypes according to growing conditions, especially the suitability of substrates. Bromeliad leaves develop from basal meristems much like those of most monocots, but mature organs differ among and within genotypes (e.g., Fig. 2.12). Three types of heterophylly, that probably serve as many different purposes as described below, occur through the family. Several Pitcairnioideae rely on synchronized abscission to coordinate leaf displays with the availability of moisture (Fig. 2.12A), and channeled blades with in¯ ated, tightly clasping bases (ligulate leaves) mark the bromeliads with phytotelm shoots (Fig. 2.4). The nonimpounders appear more grass-like, or their leaves possess prominent midribs (Fig. 2.2F). Petiolate foliage similar to that of some dicots can confuse all but the experienced collector (e.g., several Bromelia and Cryptanthus species; Fig. 2.12). Pronounced succulence distinguishes some taxa (e.g., Dyckia, dry-growing Tillandsia; Figs. 2.10A± D, 2.13B), whereas much thinner blades signal accommodation to humid sites (e.g., Catopsis, Ronnbergia; Fig. 2.17A). Cambridge Books Online © Cambridge University Press, 2009 Habits: general overview 35 Figure 2.15. Root and stem structure characteristic of many members of Bromeliaceae. Leaf color is exceptionally vivid and varied among the bromeliads, and trichomes often further ornament foliage. Anthocyanins and chlorophylls in many patterns set off the broad, ¯ at leaves of numerous Type Four species (e.g., Vriesea fosteriana; Fig. 2.17B). Foliar indumenta range from con¯ uent to sparse and from relatively ® ne to coarse textured. Alternating bands of densely and sparsely covered leaf surfaces can produce striking displays (e.g., certain Cryptanthus species, Aechmea chantinii; Figs. 2.14F, 2.18C). Importance to leaf moisture and ion and energy exchange vary according to the amount of leaf area these appendages insulate, certain specializations of their living stalk cells and the shape, mobility and other aspects of the shields (Tables 2.1, 9.1). Cambridge Books Online © Cambridge University Press, 2009 36 Vegetative structure Figure 2.16. Aspects of the leaves of Pitcairnioideae. (A) Puya raimondii, crosssection of blade half way between base and apex illustrating stomata, trichome and details of the mesophyll. (B) Cross-section of blade of Pitcairnia trianae illustrating presence of a palisade. (C) Cross-section of blade of Pitcairnia pungens showing its undifferentiated chlorenchyma. Organization for foraging Evolutionary ecologists and ecophysiologists consider the typical rhizomatous herb as much a collection of closely related, interacting individuals as one plant. Most bromeliads more than a few years old ® t this de® nition because by this age they consist of several to hundreds of meta-individuals, viz. ramets, modules or `sympodial units' which originate, mature and die as subordinated parts of a genetic individual (genet). Modules organized in this fashion act collectively to exploit patchy environments for Cambridge Books Online © Cambridge University Press, 2009 Organization for foraging 37 Figure 2.17. Aspects of the leaves of Tillandsioideae. (A) Catopsis floribunda, crosssection of blade cut half way between apex and base illustrating anatomy of vein and stomatal apparatus. (B) Representative patterns of anthocyanins displayed by many soft-leafed Tillandsioideae and the presence of transverse commissures linking the parallel veins of some species. Part A redrawn from Tomlinson (1969). resources, respond to co-occurring biota, reproduce, and pass genes on to future generations. More than the plant with a more discrete body (e.g., a tree), bromeliads and other ¯ ora with equivalent architecture exemplify how coordinated growth in lieu of animal-like mobility grants capacity to maximize resource capture and avoid certain localized hazards in heterogeneous, basically two-dimensional habitats. Like other clonal herbs, qualities that help determine how effectively the sympodial bromeliad harvests nonrandomly Cambridge Books Online © Cambridge University Press, 2009 38 Vegetative structure Figure 2.18. Leaf pigmentation. (A) Neoregelia sp. with red-tipped foliage. (B) Guzmania lindenii. (C) Cryptanthus sp. with horizontal bands of prominent trichomes. (D) Vriesea erythrodactylon with deeply cyanic leaf bases. distributed commodities include branching angle, the numbers and locations of meristems, ramet life span, and plant capacity to utilize connected modules as integrated physiological sources and sinks. Bell et al. (1979) questioned why Y-type branching and only a few angles of divergence among the much larger number possible characterize so many plants, including Bromeliaceae, with horizontal (plagiotropic) rhizomes and orthotropic (upright), determinant ramets (Fig. 2.3A). Apparently, branch angles above 60° (120° between two daughter axes) allow such ¯ ora to maximize returns on resources committed to foraging, i.e., to theoretically exploit the greatest amount of space per unit of invested biomass (Fig. 2.19 pattern A). Additional species follow either of Cambridge Books Online © Cambridge University Press, 2009 39 Organization for foraging Table 2.1. Aspects of bromeliad trichome structure and related functions Associated qualities of trichome Function Occurrence in the family (1) Retard transpiration and reduce heat load and photoinjury Densely overlapping re¯ ective shields, especially abundant on abaxial surfaces All three subfamilies (2) Absorption of H2O and nutritive ions Stalk cells alive and equipped with dense organelle-rich protoplasts, shields various, wettable shield Primarily Tillandsioideae and Brocchinia, leaf bases in phytotelm Bromelioideae (3) Deterrent to predators and pathogens Structure ranges from the peltate type that protects the underlying softer tissue to the potentially bodypuncturing uniseriate appendages of certain Pitcairnioideae Probably all three subfamilies (4) Attraction of pollinators Peltate types that form dense indumenta that re¯ ect dim light from the in¯ orescence Possibly common among night-¯ owering Tillandsioideae, e.g., T. streptophylla (5) Attraction of seed dispersers As above but densely investing ¯ eshy fruits (Fig. 3.5G) Bat-dispersed Bromelioideae (e.g., Billbergia porteana) but probably uncommon in family (6) Secretion (A) Possibly digestive enzymes to process prey (B) Deterrent to predators Stalk and shield cells equipped with dense organelle-rich protoplasts Uniseriate with glandular distal cell (Fig. 2.5K) Brocchinia reducta and B. hechtioides Navia glandulosa, Ronnbergia petersi two more options that produce different architectures according to Bell and Tomlinson (1980; Fig. 2.19 patterns B and C). Option two features a 45° angle that, should every meristem survive, results in an octagonal instead of a hexagonal grid. Option three (Fig. 2.19 pattern C) promotes linear arrays of ramets. However, before moving on we need to consider why most rhizomatous plants deviate from inherent patterns. Cambridge Books Online © Cambridge University Press, 2009 40 Vegetative structure Figure 2.19. Three patterns of branching that occur among rhizomatous herbs that in¯ uence plant foraging for light, water and mineral nutrients. Two conditions, one fundamental to plant geometry based on the hexagonal grid and the other a characteristic of all but the exceptional twodimensional environment, oblige ¯ ora with creeping, modular bodies to depart from the ideal con® guration. Figure 2.19 (pattern A) illustrates how a rigid branch angle of 60° would cause certain pairs of same-generation ramets to overlap after just three iterations. Unevenly distributed resources further oblige capacity to deviate from a grid pattern to achieve costeffective foraging. Finally, several additional considerations complicate architectural analysis for such plants, especially Bromeliaceae. Bromeliads and other similarly organized herbs determine the sizes and shapes of the spaces they ® ll with leaves and roots by regulating meristem vigor, number and location. Speci® cally, they enhance resource harvest by concentrating appropriate organs where photons, water and nutrient ions occur most abundantly. Effective foraging in heterogeneous space requires opportunistic growth, essentially capacity to selectively invade enriched patches of habitat and avoid or minimize investments in others that offer lesser rewards. Targeting and coordination in turn oblige sensory capacity and communication among connected modules. Some meristems thrive as the genet expands while support for others that occupy less propitious microsites diminishes because of poorly understood, plant-mediated economic analysis and response. Clonal herbs tend to express either `guerrilla' or `phalanx' -type growth depending on the relationship between plant architecture and foraging Cambridge Books Online © Cambridge University Press, 2009 Organization for foraging 41 strategy. Both patterns represent evolutionary adjustments to frequently encountered arrays of resources in situ. The ramets of guerrilla-type plants grow variable distances before they branch (e.g., scandent Pitcairnia riparia; Fig. 2.12B). Conversely, plants with phalanx-type architecture ® ll horizontal space more systematically with ramets that grow as advancing fronts. Greater opportunism characterizes the guerrilla-type plant given its superior capacity to place roots in widely dispersed patches of fertile, moist soil and chlorenchyma in far-¯ ung light gaps. So deployed, biomass yields greater returns in photosynthate, water and key ions than if invested more equitably or randomly among attached ramets. Architectural models tend to distinguish the bromeliads according to the nature of their substrates. The epiphytes, for example, usually produce compact clones (phalanx strategy?) compared with certain terrestrials (e.g., Tillandsia vs. certain Bromelia species), which instead forage more widely across typically more expansive soils. Saxicoles often range over more space than their close epiphytic relatives (Bennett 1991; Chapter 6), but not without exception. Lithophytic Abromeitiella (Deuterocohnia) lorentziana iterates numerous small ramets connected by short stolons to achieve the dense, self-insulating cushions that allow it to exploit protected exposures on the leeward sides of rocks in cold, windswept, south Andean habitats. Similar form under more benign conditions in culture demonstrates genetic control over architecture. Aechmea nudicaulis var. aequalis (Fig. 7.13C) along with some other natives of restinga habitats exhibits a linear array of ramets (guerrilla strategy?), perhaps to assist growth out from under the shade of the shrubs required for establishment. So far, Neoregelia pauciflora alone represents Bromeliaceae among ¯ ora assigned to a recognized model (modi® ed hexagonal; Bell et al. 1979). Bell and Tomlinson' s (1980) three models for clonal herbs lend themselves to numerical analysis, which if considered in additional detail might increase insights on the adaptive architecture of Bromeliaceae. However, ® ndings elsewhere probably need to be modi® ed to interpret conditions among the more leaf-dependent species. At least as important for foraging to these bromeliads is the shape of the shoot and the location there of functions most plants perform with roots. Shoots of Types Three, Four and Five Bromeliaceae exhibit a bewildering variety of sizes and proportions (e.g., Fig. 2.4) that affect foraging in ways accorded fuller coverage below and in Chapters 4 and 5. Further inquiry might also target possible additional relationships between architecture and physiology vs. microtopography and the distributions of light, water and nutrients in space. Resource allocation, including Cambridge Books Online © Cambridge University Press, 2009 42 Vegetative structure Figure 2.20. Schematic diagram illustrating plausible evolutionary relationships among eight derived body plans and the more fundamental rhizomatous architecture characteristic of many monocots. translocation and source/sink relationships, needs to be compared among the attached ramets of clonal Bromeliaceae that exhibit relatively ¯ exible growth and routinely occupy broader expanses of substrate (e.g., many Neoregelia species, Pseudananas) than Abromeitiella and most of the epiphytes. Finally, relationships should be considered between architecture and plant characteristics unrelated to foraging. Body plan complements many aspects of natural history in other ¯ ora (Bell and Tomlinson 1980). For example, species characterized by densely clonal genets more often exhibit self-incompatibility than those with relatively dispersed ramets. Plants with scattered vs. aggregated shoots may attract fewer predators by virtue of lower apparency and so on. Relationships of the body plans Figure 2.20 illustrates how the body plan of a putative, sympodial ancestor with a nonimpounding, rosulate shoot and extensive absorptive roots relates to the eight additional arrangements exhibited among extant Bromeliaceae. Many Pitcairnioideae (e.g., Fosterella, Pitcairnia) conform Cambridge Books Online © Cambridge University Press, 2009 Relationships of the body plans 43 to what this scheme presumes to be the basic family condition. Degrees of caulescence and root development, branching pattern, and many characteristics of foliage differentiate the eight derived arrangements, each of which includes additional, ® ner-scale variation (not shown) among the qualifying species. Roots diminish relative to shoots in ® ve of the eight directions, paralleled by increased plant dependence on absorptive foliage. Stems become more prominent in all three of the remaining directions. Neoteny shaped the outcome in the case of the most specialized of the eight architectures. Aspects of habitats, and especially the nature of the substrate (stability, utility as a source of moisture and nutrients), correlate with body plan. Beginning on the right, sparsely branched, weakly determinant, xeromorphic shoots bearing reduced root systems (e.g., Tillandsia araujei; Fig. 2.1) describe many of the dry-growing lithophytes (Chapters 6 and 9). Even longer-stemmed forms featuring relatively drought-sensitive, often heterophyllic foliage and more roots reside in humid forests as vines and hemiepiphytes (direction two; e.g., Pitcairnia riparia; Figs. 2.2C, 2.12B). Arid, relatively stable rooting media in the high Andes, that limit productivity and permit extended life spans respectively, support bromeliads with two more architectures. In addition to the cushion arrangement (direction three) that grants Abromeitiella insulation from cold, desiccating wind, a tuberous partially subterranean stem affords similar advantage under somewhat less demanding circumstances to members of the Puya tuberosa complex (direction four). Species with similarly swollen stems tolerate ® re in rupestral habitats (Fig. 2.2G). Monocarpic Puya (direction ® ve) illustrate the other, over-represented condition (giant rosette type; Fig. 14.2C) at high elevations in which a single leafy shoot constitutes the entire, long-lived genet. Woolly indumenta, compact shoots and a massive body prevent precipitous cooling as ambient temperatures drop below freezing at night (Figs. 7.2± 7.4). An even more palm-like habit evolved in Brocchinia, speci® cally in B. paniculata and B. micrantha (direction six). Both of these species occupy wet, often densely vegetated sites below 2000 m where a stout, unbranched, tall stem supporting a crown of water-impounding foliage favors gap-phase regeneration (Givnish et al. 1997). Neotenic Tillandsioideae, especially those native to arid and/or cool regions, deviate most from the basic bromeliad bauplan (direction eight). Roots develop sporadically, if at all, on adults. Shoots may be exceptionally leafy (Tillandsia bryoides) or sparsely foliated, just one absorptive leaf and a single enlarged prophyll per ramet in T. usneoides (Fig. 2.1). Cambridge Books Online © Cambridge University Press, 2009 44 Vegetative structure Ancestors probably possessed the rosulate, water and debris-impounding (phytotelm) shoot that continues to characterize most Tillandsioideae (direction seven). Similar shoot form emerged in Bromelioideae and Brocchinia (Pitcairnioideae). Closer inspection of the individual subevents (directions) underlying the family-wide radiation just described more precisesly reveals how Bromeliaceae colonized so many kinds of often stressful habitats and can utilize diverse sources of moisture and mineral nutrients. Speci® cally, these smaller arrays of more closely related body plans illustrate how modi® cations of stems and foliage, and accordingly, the relationships between these two kinds of organs, fostered major shifts in ecology. The sequence within Tillandsioideae mentioned above and illustrated in Figure 2.1 constitutes the most revolutionary among the component radiations because it required fundamental reorganization of the shoot and root systems via heterochrony. Powerful constraints peculiar to epiphytic and other extreme habitats provided the impetus. Pitcairnia and several related genera demonstrate that modi® cation of the vegetative body need not be as fundamental as that experienced by neotenic Tillandsioideae to equip the resultant lineages for life under diverse, and in some instances extreme, growing conditions. In this case, multiple options were realized by stock that probably possessed determinant shoots bearing congested foliage of a fairly generalized monocot type. Indeed, perhaps the prototype for this radiation and that for the entire family were identical. Moreover, this subevent was earth-based in the sense that descendants, like their antecedents, rarely anchor in any but terrestrial substrates and root development remains substantial. Evolution leading to architectures suitable for substrates ranging from bare rock to more accommodating riparian soils and the moisture supplies and irradiances prevailing in deep forest to open, semiarid, hot to cold habitats occurred simply by modifying the form, physiology and longevity of leaves, degrees of foliar dimorphism, and the frequency at which the stems supporting these appendages branch. None of the more extreme modi® cations that elsewhere (directions seven and eight) permit shoots to impound moisture and solids or allow foliage to serve in lieu of absorbing roots emerged here. All but a few species comprising this land-based, pitcairnioid radiation possess the familiar liliaceous, sympodial body plan expressed in the form of serial ramets bearing foliage of various textures and forms that may or may not abscise as the blades senesce (Figs. 2.3A, 2.12). Rhizomes clad with nonphotosynthetic, bract-like leaves characterize these taxa as they do the Cambridge Books Online © Cambridge University Press, 2009 Relationships of the body plans 45 rest of the genus and many other Bromeliaceae. But enough variation occurs despite these shared characteristics to differentiate plant performances and accordingly, oblige correspondingly distinct growing conditions. Rhizome appendages of Pitcairnia andraeana, P. tabuliformis and their kind lack spines as do the more distally inserted appendages designed for photosynthesis rather than protection (Fig. 2.12C). Relatively mesomorphic blades die back from the tip (e.g., P. tabuliformis), or senesce more evenly (e.g., P. andraeana). Additional body plans that emerged during this pitcairnioid radiation feature a few (the short-stemmed taxa) to many (the stoloniferous species) abbreviated, persistent, spiny organs below others equipped with welldeveloped, smooth-margined green blades (Fig. 2.12A,B). A group of short-stemmed, heteroblastic taxa exhibit progressively expanded and smoother-margined appendages between the spiny bract-like and much more expanded, unarmed photosynthetic types (Fig. 2.12D). Populations equipped with short ramets often grow on exposed rocks and cliffs: those with longer shoots sprawl over the ground or climb trunks as hemiepiphytes (e.g., P. riparia; Figs. 2.2C, 2.12B). Deciduous green leaves shed prior to ¯ owering during the dry season distinguish Pitcairnia heterophylla from most members of the short-stemmed group (Fig. 2.12A). Typically short ramets featuring relatively long-lived, thick (often CAMequipped) foliage with marginal spines form the basis for still more habits suitable for exposed, exceptionally arid sites. Pitcairnias that most closely approach this description resemble certain of the generally more xeromorphic members of Dyckia, Hechtia and Encholirium (Fig. 2.2A,B). Somewhat thinner, but still well-armed, leaves that die back rather than abscise during the dry season identify Pitcairnia feliciana as a transitional species that perhaps ® ts more comfortably within this group than the previous one (Fig. 2.12E). Puya takes the putative prototypic body plan into two more adaptive zones. Short ramets with swollen bases essentially render P. pusilla (Fig. 2.12F) and its kind hemigeophytes, while tall, thick, unbranched stems qualify P. raimondii as a giant alpine rosette type (Fig. 14.2C). The ® rst architecture favors reliance on the rooting medium for insulation. Massive stems containing substantial stored moisture and dense crowns of succulent foliage offer the same protection from severe climate to plants that conform to the second body design. Relationships between habit and ecology receive additional consideration below and as parts of the discussions of heterochrony in Chapters 5 and 9, and the operation of the phytotelm shoot in Chapter 7. The balance of this chapter and Chapters 4 and 5 Cambridge Books Online © Cambridge University Press, 2009 46 Vegetative structure and parts of Chapter 7 detail the modi® cations of leaves, roots and stems that underlie the exceptional functional and ecological diversity of the bromeliads. Stems Leafy stems, except for those produced by the most reduced, neotenic species, share similar monocot-type anatomy (Fig. 2.15; Chapter 12). Typically, a narrow cortex-like layer is differentiated into two zones, the outer one occupied by thick-walled, ligni® ed cells. Thinner-walled, sometimes starch-laden parenchyma constitutes the inner region, which lies against the central cylinder and is separated from it by sclerenchyma of various descriptions. Periderm-like (storied) tissue formed by periclinal divisions of cells, derived from the apical meristem rather than a true phellogen, may develop at different depths below the epidermis, sometimes discontinuously, on older axes and under wounds and leaf scars. Bromeliads also lack vascular cambia, and although the exceptional axis thickens considerably (e.g., the larger Puya species), it does so without any unusual mechanism con® rmed so far. Reports (e.g., Harms 1930) that a cambiumlike cylinder occurs in the perennial (multiple crops of ¯ owers; Fig. 3.4J) in¯ orescence axis of Deuterocohnia meziana need con® rmation. Stem vasculature is complex in most Bromeliaceae in part because roots accompany the bundles typically present there. Leaf traces enclosed by suberized `endodermis-like' (Krauss 1948± 49) sheaths intermingle with the still embedded roots, and both entities travel through many nodes before entering a leaf base or, in the case of the root, departing from the stem (Fig. 2.15). Additional vasculature originates beneath each axillary bud. Collateral bundles representing fused series of leaf traces (sympodia), again often with robust sheaths, occur more densely nearer the edge than the center of the central cylinder (e.g., Ananas comosus; Krauss 1948± 49). Tissue vascularizing the stem of the most diminutive Tillandsia, including T. usneoides, forms an almost solid sclerotic core containing scattered phloem strands and a few narrow tracheids probably because CAM and reliance on foliar trichomes rather than roots greatly reduce the need for water transport. Flux mostly occurs over short distances, primarily from wetted leaf surfaces to adjacent parenchyma, and, for structurally less reduced relatives with a differentiated mesophyll, from water stored in the hypodermis to more desiccation-sensitive chlorenchyma during drought (Chapter 4; Fig. 2.13B). Mucilage accumulates in schizolysogenous cavities within the stems and in¯ orescence axes of many bromeliads, particularly Tillandsioideae. Cambridge Books Online © Cambridge University Press, 2009 Stems 47 Exudation accompanies certain kinds of injury and may explain Picado' s (1913) report that digestive secretions elaborated by some Costa Rican natives enhance the utility of phytotelmata for carnivory. His hypothesis remains untested and improbable; more likely, these products deter predators, enhance drought-tolerance, or constitute carbohydrate reserves. Ergastic inclusions in stems include silica bodies in the epidermis and raphide sacs, especially in cortical cells adjacent to the intracauline roots. Bromeliad stems exhibit tropisms consistent with conditions encountered in situ. Shoots produced by phytotelm Bromeliaceae that frequently germinate under branches and along vertical surfaces always grow upright. Sometimes just the in¯ orescence assumes this orientation among the nonimpounding species (Fig. 1.3C). Shoots of `stemless' taxa (e.g., Tillandsia ionantha; Fig. 2.10M,N) lack sufficient length to respond to gravity while not all of the caulescent forms that could, do so (e.g., T. schiedeana). Stem shape also varies among taxa. Seedling axes and those of ramets often become obconical as the apical meristem enlarges during growth. Especially notable are the expanded, corm-like rhizomes of Puya tuberosa and Cottendorfia florida (Fig. 2.2G), and on an even grander scale, the massively upright axes of monocarpic Puya raimondii (Fig. 14.2C). Elongate stolons mark many of the soil-rooted and lithophytic species, especially terrestrial Bromelioideae (e.g., Bromelia, Nidularium, Pseudananas) and the hemiepiphytes (Pitcairnia; Figs. 2.2C, 2.12B), but only the occasional epiphyte (e.g., Vriesea espinosae). Shoots of additional soil-based types sprawl across the ground or grow up into low vegetation (e.g., some Orthophytum, Cryptanthus), as do the offshoots produced on the spent in¯ orescences of several tillandsias (Fig. 2.11). A basal constriction promotes vegetative dispersal by encouraging the ramets of certain Cryptanthus to disarticulate after becoming large enough to grow independently. Mobility fostered by re¯ exed foliage leading to spherical form may encourage the detached ramets of certain species to roll and root well away from parents (e.g., C. acaulis). Ramets typically survive for 3± 4 years, while the seedling shoot (especially those of the monocarps) requires additional time to accumulate the resources necessary to reproduce (Fig. 2.3). One or two seasons pass before the average ramet ¯ owers, and seed crops ripen over additional months to another year. Natives of stressful habitats (e.g., Abromeitiella, Tillandsia hildae) cycle more slowly than plants less tolerant of physical stress (Chapter 4). Daughter ramets receive photosynthate and residual N and P and perhaps other useful constituents from their declining parents. Concentrations of N, P and K among the attached ramets of T. paucifolia collected in south Florida diminished with age, reaching lowest values in Books Online © Cambridge University Press, 2009 48 Vegetative structure senescing, post-fruiting shoots (Benzing and Renfrow 1971a; Benzing and Davidson 1979). Leaching from moribund foliage probably denies daughter ramets equal opportunity to recycle relatively labile K compared with N and P. Roots Bromeliad roots exhibit features consistent with their frequent role as holdfasts. Figure 2.15 illustrates the pronounced scleri® cation responsible for the uncommon durability and mechanical strength of those axes characteristic of many of the epiphytes and saxicoles. Note also the absence of morphological loss; every tissue (primary phloem and xylem, endodermis and pericycle) is present along with a root cap. The cortex is often differentiated with the interzone accounting for most of the volume of the root and much of its strength. Krauss (1948± 49) reported a thick-walled endodermis in pineapple, but this physiologically decisive tissue is less robust elsewhere (e.g., Billbergia sanderiana). Type Five Bromeliaceae (Table 4.2; Chapter 12), including Tillandsia ionantha, lack root hairs or soon lose them; conditions among the other dry-growing forms remain largely undocumented. Roots of the more mesic Bromelioideae and Pitcairnioideae produce abundant hairs as seedlings, as does adult pineapple (Krauss 1948± 49; Fig. 3.8). Roots and shoots grow episodically in culture, faster early in the growing season than later. Harms (1930) reported a progressive loss of vascularity with increasing epiphytism until phloem in strongly neotenic Tillandsioideae is greatly reduced and only tracheids remain to transport water. Cheadle (1955) con® rmed Harms' s ® ndings, noting tracheids exclusively in the roots of several Type Five Tillandsia species and vessel elements with scalariform end plates in all the other examined specimens of this description. Roots originate in the previously mentioned `dictyogenous' zone within about 1 cm of the shoot apex in pineapple (Krauss 1948± 49; Fig. 2.15). Intrusive growth proceeds downward for many more centimeters before the elongating organ emerges from the stem through relatively soft tissue at a node. Cross-sections of older axes reveal darkly pigmented intercauline roots that sometimes occur at sufficient densities to almost ® ll what appears to be a true cortex. Unlike the rot-resistant, cable-like roots of the xerophytes, those of the bromeliads restricted to moist soils exhibit less robust structure and probably more effectively absorb moisture and nutrients (Chapter 5). Unlike a number of aroids, orchids and other ¯ ora that produce one type Cambridge Books Online © Cambridge University Press, 2009 Roots 49 of root for anchorage and another for absorption, roots of the typical bromeliad share the same structure although function may differ according to location. Krauss (1948± 49) recognized `axillary' and `soil' roots depending on where the organ originated along the parent shoot and, more importantly, where it emerges from the stem. Many species with rudimentary phytotelmata (e.g., Type Two Ananas and Bromelia) produce an extensive system of axillary roots to explore adjacent leaf base chambers. Compaction ¯ attens those organs that penetrate the deepest recesses of the shoot. Soil roots in turn extend 1± 2 m beyond the base of a pineapple at depths up to 85 cm when media permit (Krauss 1948± 49). Abundant roots ramify through the remnants of the old leaf bases of arborescent Brocchinia micrantha to intercept nutrient-charged ¯ uids over¯ owing from the debris-laden bases of younger, intact foliage (Fig. 2.14D). Bromeliads native to certain kinds of substrates exhibit matching root growth. Anchorages oblige ageotropism among the Type Five bromeliads, many of which regularly germinate on the undersides of branches after being deposited there by stem ¯ ow. Thigmotropism occurs at least as widely and involves taxa representing all three subfamilies. Brighigna et al. (1990) reported abundant accumulations of a protein/polysaccharide mixture in the caps of the roots that secured some Tillandsia latifolia and T. macdougallii specimens to substrates. Free roots lacked this hydrophilic material as did the exposed sides of the adhering organs, suggesting importance for holdfast. Bromelioideae, and Tillandsioideae even more, rely on foliage for absorption, although anchorage remains essential to all but stoloniferous Spanish moss and scattered relatives with additional alternatives like the curling, twig-grasping leaves of Tillandsia duratii (Fig. 2.10L). Root systems serving neotenic Tillandsioideae diminished more than those of related lineages as ancestors colonized increasingly arid habitats and unyielding substrates (Fig. 2.1). Although the epiphytes and saxicoles alike possess substantially reduced root systems, what remains of these organs suggests different requirements to utilize bark compared with rock. Several polymorphic taxa (e.g., T. ionantha varieties ionantha and vanhyningii; Fig. 2.10M,N) make the most compelling case for substrate-speci® c differentiation. Lithophytic types produce long, scarcely branched roots, usually one each from widely scattered nodes along leafy caulescent shoots. The epiphytes root more profusely, but each ramet remains relatively short except for a few small-bodied species like T. tricholepis and T. capillaris. But even these bromeliads usually root to a single spot, namely where their seeds happen to germinate. Opportunity to promote cost-effective foraging by rooting largely on an Cambridge Books Online © Cambridge University Press, 2009 50 Vegetative structure as-needed basis varies among Bromeliaceae. Extreme emphasis on shoots constrains participation for many Tillandsioideae, especially the essentially rootless members of Type Five as just described. Substrates can have greater in¯ uence elsewhere. Moist, compared with less accommodating, media stimulate root development in a wide variety of cultivated Bromelioideae, particularly Type Three and Four stock that as epiphytes and saxicoles often grow on comparatively barren substrates (Table 4.2). Capacity to strike ® ne roots from rhizomes or along thicker exploratory roots should bene® t populations native to rocks and other substrates that sequester moisture and nutrients in scattered, relatively inaccessible locations. Temporal variations in moisture supply can have similar effects. Media utilized by many Dyckia and Hechtia species, among others, dry out and recharge seasonally (e.g., Fig. 7.1E). But if the rapidly developed `rain roots' of certain cacti and other desert-dwellers also exist in Bromeliaceae, they remain unreported. Vascular cells Cheadle (1953, 1955) included 28 species representing all three subfamilies of Bromeliaceae in his study of tracheary element evolution in Liliopsida. Relatively narrow components, both vessel members and tracheids, occur through the family, but ® ner structure varies among taxa and often differs from one type of organ to another in the same plant (Fig. 2.21). Bromeliaceae were judged moderately advanced and the subfamilies roughly equivalent in evolutionary grade on the basis of tracheary advancement. Water-vascular cells tend to be either similar to or less advanced in stems, in¯ orescence axes and foliage compared with roots, a pattern Cheadle (1953) ascribed to the monocots in general. Cross-sections of the more robust foliar veins reveal many small and usually two broader tracheary cells; fewer elements comprise the lesser veins (Fig. 2.17A). Many species lack protoxylem lacunae in part probably because internodes are short and growth is slow. Closer inspection of the in¯ orescence axes, some of which elongate several centimeters per day and rise through phytotelmata, might reveal exceptions. The generally narrow sieve tube elements look much like phloem parenchyma in transverse view (Fig. 2.17A). Cheadle conducted his survey before certain agencies that in¯ uence the hydraulic architecture of plants were recognized. Aspects of xylem tissue, especially the dimensions of its conductive cells, probably re¯ ect current growing conditions and related plant needs more faithfully than those experienced by ancestors if these conditions differ. Not surprisingly, Pitcairnia Cambridge Books Online © Cambridge University Press, 2009 Vascular cells 51 Figure 2.21. Representative water-vascular cells in the vegetative body of Bromeliaceae. (A) One end of the type of tracheid widely distributed through the family. (B) Vessel element with multiperforate end plate that also occurs in most Bromeliaceae. (C) Vessel element in root of Pitcairnia sp. illustrating the most advanced structure for tracheary function present in the family. Redrawn from Cheadle (1955). alone of the surveyed taxa exhibited vessel elements with transverse, simple perforate plates (Fig. 2.21). Members of this genus feature some of the most mesomorphic foliage described so far for Bromeliaceae, organs that oblige substantial streams of moisture to support relatively vigorous transpiration and photosynthesis (Chapter 4). Narrower cells lacking end walls (tracheids) or vessel elements with multiperforate end plates characterized the rest of the sample (Fig. 2.21). Xylem strands made up exclusively of tracheids, or vessel elements with multiperforate rather than uniperforate end plates, seem best suited for most Bromeliaceae because so many species use water sparingly and succulence is so common (Table 4.1). So what can tracheary anatomy tell us about the evolutionary status of the family and the relationships among its species? Cells that just happen to ® t Cheadle' s de® nition as primitive probably resist cavitation by reason of the same supposedly archaic structure. On the other hand, xylem tensions recorded for Bromeliaceae have always been low, although most records come from succulent types that by virtue Cambridge Books Online © Cambridge University Press, 2009 52 Vegetative structure of this circumstance rarely, if ever, experience sufficient dehydration to rupture water columns in xylem capillaries (Chapter 4; e.g., Fig. 4.18). Clearly, Cheadle' s story remains incomplete. A biophysical critique mindful of the relationship between xylem anatomy and hydrodynamics is needed to explain why tracheary cell structure differs in the root vs. the stem and leaf of the bromeliads and Liliopsida in general. Is Cheadle' s (1953) hypothesis supported, or do the environments of these organs and the demands placed upon them by the rest of the plant differ enough to explain his ® ndings without recourse to evolutionary history? How does tracheary cell morphology among the monocots, and the bromeliads in particular, relate to hydraulic demand and the maintenance of capillary integrity (safety) in different organs of the same plant and among species adapted to different growing conditions? Foliage Like liliopsids generally, all but the exceptional bromeliad emphasizes foliage over stems and roots in the sense that leaves constitute the bulk of the vegetative body. However, Bromeliaceae exaggerate this condition even among the monocots because so many of the functions usually relegated either to shoot or root so often operate together. Accordingly, leaf shapes and textures indicate mode of resource capture and required growing conditions more conspicuously in this family than in most others. Unfortunately, signi® cance was ignored by most of the anatomists who described bromeliad foliage, in part because the tools and perspectives were not available to address them. Foliar trichomes received inordinate attention on those few occasions when inquiry encompassed structure and related function. Low to moderate speci® c weights (leaf mass/unit surface area) and broad and relatively glabrous and ¯ at surfaces usually signal shade-tolerance and drought-sensitivity among the bromeliads. Conversely, thicker, enrolled organs with more water storage and mechanical tissue and a denser indumentum denote life in more exposed, drier ecospace (Figs. 2.10, 2.13, 2.16, 2.17). But some confounding characteristics preclude more precise predictions. For example, McWilliams (1974) reported fresh/dry weight ratios of 5.98 and 17.06 for Puya mariae and Fosterella penduliflora to support his suggested existence of succulent and nonsucculent `strategies' in Pitcairnioideae. In fact, the second species sheds its leaves during the dry season as be® ts a drought-avoider. Among tested Tillandsioideae, species with thin, lightly trichomed foliage (e.g., Guzmania lingulata) and others Cambridge Books Online © Cambridge University Press, 2009 Foliage 53 with unequivocal Type Five status (e.g., Tillandsia usneoides) yielded similarly low ratios. A reading of 13.00 for evergreen, drought-enduring Dyckia brevifolia further indicates how poorly this measure re¯ ects growing conditions in situ. Aspects of leaf margins and indumenta parallel certain taxonomic boundaries, provide information about relationships within subfamilies, and sometimes also indicate the habits (e.g., epiphytic vs. terrestrial) and ecotolerances (e.g., to exposure, to drought) of individual populations. Bromelioideae, especially Bromelia and certain Pitcairnioideae (e.g., Dyckia, Hechtia), feature strongly serrate-margined foliage while Tillandsioideae never do (Figs. 2.2, 2.13A, 2.17). Presence of members equipped with either unarmed (e.g., Brocchinia, Fosterella) or armed foliage (most of the rest of the genera) accords with Pitcairnioideae as the evolutionarily broadest of the three subfamilies without denying that some of its lineages possess what are probably the least derived of the body plans extant within Bromeliaceae. Trichome organization varies most among Pitcairnioideae; Bromelioideae come next. Trichome structure, which is remarkably concentric, and its consistency clearly distinguish Tillandsioideae from the rest of the family (Figs. 2.5± 2.8). However, other patterns are more ¯ uid, such as trichome density and the shapes of the shields that sometimes also vary across the same leaf (Figs. 2.7I, 2.14B,F). Evolutionary polarity is often obscure. Strehl (1983) concluded from her survey of 100 species that adaxial trichomes exhibit more advanced organization than those on the other side of the leaf. Scales on the leaf bases of phytotelm types supposedly feature less `conservative' characteristics than those born nearer the apex, but Strehl provided no compelling rationale to support these conclusions. Closely imbricated, utriculate leaves create the impoundments emblematic of phytotelm Bromeliaceae and lend a uniquely tropical American aspect to the heavily colonized forest (Fig. 1.4F). The same architecture accounts for an important resource base for symbiotic biota dependent on aquatic media and/or moist detritus (Chapter 8). A derived condition endows a group of smaller-bodied relatives (e.g., Tillandsia bulbosa) with dry cavities to house plant-feeding and possibly plant-protecting colonies of ants (Fig. 8.5). Channeled blades often direct precipitation and litter to the axils of the phytotelm types; those of the myrmecophytes roll abaxially to conserve moisture and better expose the green mesophyll to shade-light. Although one or the other condition characterizes many species, a few additional taxa (e.g., Aechmea bracteata, Brocchinia acuminata; Figs. 2.2E, Cambridge Books Online © Cambridge University Press, 2009 54 Vegetative structure 2.4G) provide both kinds of living space, presumably with plant bene® ts expanded accordingly. Long petioles that increase shade-tolerance and perhaps impart other advantages distinguish a variety of understory Bromelia, Cryptanthus, Disteganthus, Ronnbergia and Pitcairnia species (e.g., Fig. 2.12B). Otherwise similar foliage with ensheathing bases displayed as a rosette would intercept less light, i.e., self-shade more. Luther (personal communication) concluded from observations in wet Andean habitats that laminae held well above the stem help a number of terrestrials to avoid overgrowth by bryophytes and other creeping ¯ ora (e.g., Ronnbergia deleonii). Two additional features increase capacity to maintain upright foliage: corrugation adds considerable strength to a strap-shaped leaf according to Krauss (1948± 49; Figs. 2.13, 2.16), as does a channeled compared with a ¯ at blade. Pigmentation Some of the most prized Bromeliaceae owe their popularity to ornamental foliage, which growers have enhanced by selection (e.g., Vriesea fosteriana chestnut; Fig. 2.14G). Several functions probably underlie these displays, and some widely grown Tillandsioideae could be used to test hypotheses (e.g., concolorous and heterochromic V. splendens var. formosa vs. var. splendens; Table 4.11). However, most displays characterize entire populations, for example the horizontal bands of chlorenchyma underlain by similarly con® gured patches of cyanic, abaxial epidermis that mark Vriesea fosteriana and several of its relatives. Guzmania equals Vriesea for ornate markings, including narrow, vertical pin stripes of anthocyanin-laden epidermis extending from leaf base well up into the blade (e.g., Guzmania lingulata). Diverse ¯ ora adapted to the shade cast by evergreen tropical forest (e.g., Vriesea simplex, Nidularium burchellii; Fig. 2.4H) share discolorous foliage purportedly to recycle photons off the `red mirror' provided by an anthocyanin-laden abaxial epidermis and back into the overlying chlorenchyma (Lee et al. 1979). A different explanation probably applies to the intricately marked leaves just described for other Bromeliaceae because these appendages occur in more densely overlapping, self-shading con® gurations (e.g., Fig. 2.18B). Perhaps also signi® cant, these stiff-leafed plants, like all Bromeliaceae, lack capacity to sun-track or reorient in some other manner to avoid overexposure. Carbon budgets and security for cryptic detritivores might be improved as well (see below) by the presence of the irregular zones of chlorophyll-rich mesophyll backed by a cyanic epidermis illustrated by Vriesea fosteriana (Chapter 4). Cambridge Books Online © Cambridge University Press, 2009 Foliage 55 Additional Tillandsioideae display similar jagged arrays of deeper and lighter-pigmented chlorenchyma, but without the red re¯ ectors (e.g., Vriesea hieroglyphica). Both arrangements suggest conditions elsewhere that accompany unusual physiology (e.g., the C4 syndrome in other families, C4± CAM in some Peperomia species). If true, ultrastructure, especially of the deeper green chlorenchyma and associated vascular tissue, should reveal it. Suberized bundle sheaths or unusually large numbers of characteristically aligned plastids and mitochondria in chlorenchyma associated with the commissures that always course through the deepest green zones could indicate a CO2-concentrating mechanism (Fig. 2.17B). Unrecognized variety probably exists in the photosynthetic apparatus of the bromeliads, perhaps more than minor variations on common themes; the ornately pigmented Tillandsioideae appear to be the best candidates for interesting discoveries. Ornamentations beyond those just described fall into several more categories. A deep purple to maroon epidermis on both sides of the leaf bases of many phytotelm types (e.g., Vriesea erythrodactylon, numerous Bromelioideae; Fig. 2.18D) should bene® t the dark-colored detritivores that these plants need to process impounded litter; broken patterns may more effectively obscure the silhouettes of lighter-colored residents. However, not every display prompts an equally plausible explanation. Those irregular interspersions of pigmented and achlorophyllous patches that mark certain Bromelioideae (e.g., Billbergia sanderiana; Fig. 2.14H) occur above the phytotelmata where their presence confers little apparent advantage to tank occupants. Conceivably, herbivores avoid foliage so marked, choosing instead food sources free of the appearance of competing folivores, perhaps leaf miners in this case (Fig. 8.2B). Still another category of Bromeliaceae includes the species with eyecatching, red-tipped foliage (e.g., numerous Neoregelia and fewer Nidularium species; Fig. 2.18A) that warrant the label `bulls eye' bromeliads. Viewed from above, the intensity (density) of the signal increases toward the shoot center where otherwise poorly advertised ¯ owers and fruits reside. Certain Hohenbergia and Wittrockia species along with members of still other bromelioid genera exhibit similar to more expansive patches of red or orange at the same locations. Up to one-third of the distal ends of the leaves of Aechmea pectinata remain bright pink while the dull, strobilate in¯ orescences display similarly inconspicuous ¯ owers. Several Tillandsioideae also bear apically pigmented foliage combined with nondescript ¯ oral bracts that in some cases subtend bat ¯ owers (e.g., Vriesea bituminosa). Cambridge Books Online © Cambridge University Press, 2009 56 Vegetative structure Young leaves that color up or become chlorotic just prior to anthesis characterize a second group of bromelioids also notable for their short in¯ orescences (Fig. 3.2A). Several Brazilian Neoregelia (N. nivea, N. lactea) and some navias attract pollinators by highlighting their ¯ owers with scattered white light. Should leaf bases remain albinistic, seed dispersers may become the targets, which for Brazilian Bromelioideae include birds and ¯ ying and nonvolant mammals (Chapter 6). The same parts of many more species become bright red (e.g., Neoregelia macwilliamsii) prior to anthesis, while the foliage of still another group of taxa (e.g., N. petropolitana) is permanently green (Fig. 2.13F). Experiments and more data on plant visitors, colors during reproduction, and other signals (e.g., Vriesea fosteriana chestnut), and fruit qualities are needed to characterize the reproductive syndromes of bromeliads. Speci® c displays may exist to promote pollination and seed dispersal in some cases and in others encourage the transport of just one of these plant products. Heterophylly Bromeliads exhibit three kinds of heterophylly, each one associated with additional plant characteristics. Leaf form changes but once during ontogeny (typical heteroblasty) in two instances, whereas a third assemblage of species produce leaves of different morphology episodically along stoloniferous shoots (e.g., certain scandent Pitcairnia; Fig. 2.2C). A group of short-stemmed Pitcairnia exemplify the ® rst pattern. Recall that each ramet of P. heterophylla initiates growth with tough, persistent scale-like blades or modi® ed leaf bases armed with a sharp terminal and multiple, marginal spines (Fig. 2.12A). A few transitional organs follow until the vulnerable, linear, deciduous appendages emerge coincident with renewed rainfall. Dormant or leafy, older specimens probably avoid consumption by all but the most determined grazers. Conditions reminiscent of this same two-stage sequence involving short, spiny juvenile foliage and equally persistent, laminate leaves characterize a host of relatives that routinely access more continuous supplies of moisture (Figs. 1.4G, 2.12D). Pitcairnia riparia produces stiff, armed nonphotosynthetic organs until the slender ramet swells preparatory to the emergence of a heteroblastic series of more elongate green leaves (Fig. 2.12B). Contact with some physical impediment seems to stimulate this conversion. Relatively narrow organs then precede broader, more distinctly petiolate foliage as the maturing shoot prepares to ¯ ower. Numerous members of Bromelioideae (e.g., Cambridge Books Online © Cambridge University Press, 2009 Foliage 57 Bromelia, Quesnelia) also bear tightly imbricated, scale-like leaves along all but the apices of stoloniferous ramets. As noted above and illustrated in Figure 2.12, Pitcairnia (sensu lato) exceeds the other bromeliad genera for varied foliage that includes pronounced heterophylly, and, for the adult-type photosynthetic leaf, texture ranging from semisucculence as exempli® ed by species such as P. tabuliformis to the much thinner, and more seasonal, appendages of P. heterophylla (Fig. 2.12A). Pitcairnia feliciana combines drought-deciduousness (blades shrivel simultaneously over a period of a few weeks rather than cleanly abscising as in similarly seasonal P. heterophylla) with unexpectedly thick (expensive) blades armed up to about one-third of the distance from the base with stout marginal spines (Fig. 2.12E). Many phytotelm Tillandsia and Vriesea (Type Four) illustrate the type of heterophylly responsible for inspiring considerable speculation on an important aspect of bromeliad evolution (Chapter 9; Figs. 2.1, 2.11B). Compared with adult-type foliage, leaves born on seedlings and the precocious grass-like offshoots featured by quite a few mesic Tillandsioideae (Table 4.7) exhibit stout morphology and lepidote (trichome-covered) surfaces that promote water storage and economical use in lieu of the phytotelmata that eventually relax requirements for these two attributes (Adams and Martin 1986a,b,c; Reinert and Meirelles 1993). Gas exchange demonstrates more precisely how aspects of water and carbon balance shift as plants mature. Young and mature specimens of Tillandsia deppeana, one of the two heterophyllic species examined, took up CO2 only during the day, indicating no change in photosynthetic pathway during ontogeny. However, responses to drought and water supply varied with the life stage (Adams and Martin 1986a; Fig. 4.9). If we can judge by T. deppeana, the seedlings of tank-bearing Tillandsioideae more closely approach the ecophysiological performances of juveniles and the adults of the more specialized tillandsioids (Type Five) than their own phytotelm-equipped stages. Certain Bromelioideae exhibit similar, although less exaggerated, heterophylly as illustrated by some Nidularium species (Fig. 9.12). Form and function exhibited by the juveniles of heterophyllic Tillandsioideae that also characterize the adults of their Type Five relatives suggest two possibilities. First, Type Five lineages evolved via heterochrony from relatively mesic ancestors like Tillandsia deppeana (Chapter 9; Fig. 2.1). Alternatively, the seedlings of these otherwise mesophytic species simply recapitulate the characteristics of drier-growing antecedents. In either case, heterophylly remains adaptive, i.e., heavily trichomed, succulent Books Online © Cambridge University Press, 2009 58 Vegetative structure juvenile shoots satisfy on-going need to accommodate the drought imposed on epiphytes and saxicoles by high exposure and nonabsorptive substrates, even in humid regions. In essence, habitats wet enough to permit the adult to expend water at the high rate required to grow vigorously nevertheless constitute veritable deserts for the much smaller, hence structurally more constrained (less favorable surface to volume ratio; Fig. 4.17), juvenile. Slow growth obliged by the kind of xeromorphy peculiar to shoot-dependent and predominantly epiphytic Tillandsioideae must precede the opportunity for faster growth that becomes possible only after foliar impoundments develop. Life history Leaf life spans among the magnoliophytes range from just a week or two to many years, more or less according to the economic/evolutionary paradigm used to interpret a variety of plant characteristics (Chapter 4). Limited data suggest that Bromeliaceae conform to the same rule ± that leaf longevity correlates with cost and varies inversely with photosynthetic capacity (Amax). Investments considered in these calculations include expenditures to build and maintain biomass. Thickness and scleri® cation in¯ uence the caloric value of foliage on a surface area basis. Thin, broad and supple, hence relatively cheap, leaves such as those born by many Type Four Tillandsioideae live about a year, while the more expensive, succulent and less productive and consequently more slowly amortized foliage of their dry-growing relatives survives longer. Not surprisingly, speci® c leaf weights (mass/surface area) probably reach family-wide lows in deciduous Pitcairnia and Fosterella. Nitrogen also in¯ uences the evolutionary economics of green tissues, but multiple sources (e.g., NO32 vs. NH41), uneven costs of processing speci® c molecules, and variable plant capacity to recycle tissue N mitigate its utility as currency for economic analysis (Raven 1988). Form and function also distinguish foliage inserted on a single axis. Pineapple illustrates how the oldest and youngest leaves are shorter and narrower than those located in between (Krauss 1948± 49). Position along the phyllotactic spiral also determines when and for how long a leaf remains functional, and how it contributes to the production of seeds and the replacement ramet(s). Organs that develop early usually die before the supporting shoot matures; younger leaves decline more synchronously. Organ cost and payback should vary accordingly, but so far no one has conducted an economic analysis to test this possibility. Cambridge Books Online © Cambridge University Press, 2009 Foliage 59 Pitcairnia heterophylla, and the other members of the small group of drought-deciduous Bromeliaceae, exhibit the most tightly coordinated leaf life histories. Entire complements of green as opposed to the reduced, spiny foliage appear in one brief ¯ ush (Fig. 2.12A). A pre-formed abscission zone, consisting of a band of thin-walled cells located just above a sclerotic hypodermis a few millimeters distil to each leaf sheath, mediates an equally abrupt return to drought-invulnerability. Similar tissues explain why senescing leaves separate about as cleanly among many evergreen Pitcairnia species (Fig. 2.12B). Abscission zones also regulate leaf fall in Ayensua uaipanensis and Brocchinia melanacra, but so far none of the descriptions of these ® re-tolerant Guayanan endemics mentions leaf life spans or phenology. Several weeks pass while the leaves of Fosterella penduliflora gradually die back from tip to base as the more durable corm-like stem prepares for drought. Epidermis Trichomes dominate the literature on the bromeliad leaf, while other aspects of the epidermis and the tissues within receive far less to no coverage. Cuticles range from robust for the xerophytes of warm habitats (e.g., many Bromelia species) to unexpectedly delicate (e.g., Figs. 2.10, 2.13, 2.16). Type Five bromeliads fall across much of this range, even though they share many other features consistent with xerophytism (e.g., CAM, succulence). Tomlinson (1969) labeled the Tillandsia species equipped with exceptionally thin cuticles to match equally delicate epidermal cells `hygromorphic' despite growth habits and habitats that assure only sporadic opportunity to eliminate moisture de® cits. Additional qualities of the cuticle and epidermis in¯ uence phenomena other than water balance. Loose plates of wax that occlude the stomata of Tillandsia deppeana native to northern Mexico may explain why it and certain other Tillandsioideae sometimes grow in drier habitats than characteristic for their ecophysiological type. Particles of the same composition re¯ ect light and help Catopsis berteroniana and Brocchinia reducta trap and retain prey in steep-sided phytotelmata (Fig. 5.3A,C). Additional species with no other indications of carnivory (e.g., Tillandsia heterophylla) probably produce similarly light-scattering cuticles on foliage to avoid photodamage. Prominent ¯ oral bracts bearing the same products presumably help nectar seekers locate nocturnal ¯ owers (e.g., T. heterophylla). Epidermal cells beyond those that comprise the stomata and trichomes vary in shape and alignment; some other, more consistent features probably Cambridge Books Online © Cambridge University Press, 2009 60 Vegetative structure impart important functional properties (e.g., Fig. 2.13). For example, sinuous, interlocking radial walls strengthen the leaves of many bromeliads, as do the inner tangential and radial walls of the xerophytes (e.g., Bromelia) that thicken enough to con® ne the protoplast to less than 10% of the cell volume (e.g., Fig. 2.13C). Uneven radial depths and wedge shapes foster dovetailing that promotes rigidity and helps prevent the separation of the epidermis from the underlying hypodermis (see Krauss 1948± 49; Fig. 2.16A). A rugose, spherical mass of silica further con® nes the space allocated for the protoplast in the epidermal cells of many bromeliads (Figs. 2.13C, 4.23I). Granule size (up to 5 mm) parallels the silica supply for cultivated Ananas comosus (Krauss 1948± 49). Location in the center of a dish-shaped lumina that also re¯ ects photons further underscores the importance of redirecting excess radiance. Bromeliads with stout xeromorphic foliage possess the largest crystals, and soft-leafed species better served by arrangements that enhance light absorption lack them (Tomlinson 1969). Dark inclusions, perhaps tannins, located in the hypodermis below the adaxial epidermis and sometimes in the sheaths of veins (Baumert 1907; Linsbauer 1911), along with the often-abundant anthocyanins, provide additional protection for green cells located deeper in the foliage of many of the heliophiles. Stomata Tomlinson (1969) attempted to reconcile stomatal function with puzzling morphology as several prominent morphologists had done before him. He concurred that certain shapes and arrangements of the cells comprising the stomata and subjacent mesophyll defy interpretation without speculation about physiology. How, for example, can what are often exceptionally thick-walled, achlorophyllous guard cells of certain species continue to regulate gas exchange (e.g., Fig. 2.13B)? Moreover, several diminutive Tillandsia species feature anatomy that seems likely to prevent stomata from closing! Mez (1904) and Billings (1904) deepened the controversy by reporting that the stomata of certain Tillandsia species failed to move during experiments. Features that might allow Bromeliaceae with oddly shaped stomata to nevertheless mediate gas exchange include certain con® gurations of associated hypodermal and subsidiary cells (Figs. 2.10K, 2.13C). Conceivably, movements at these locations suffice in lieu of those usually involving guard cells. Conversely, porosity may not be plant-controlled at all, or at least not Cambridge Books Online © Cambridge University Press, 2009 Foliage 61 in the manner expected for land ¯ ora consistent with some of the other novel attributes of these highly specialized plants. Tomlinson considered the paradoxically thin-walled epidermis and delicate cuticle of those `hygromorphic' Type Five taxa related to such a possibility, but offered an untenable view of water balance and plant nutrition to make his case. Speci® cally, he suggested linkage between thin, porous leaf boundaries and the need to eliminate the excess moisture that these bromeliads supposedly must absorb to obtain enough required ions from exceptionally dilute solutions, primarily precipitation. In fact, whether taken up through roots or foliage, solutes never accumulate in plants by Tomlinson' s implied bulk ® ltration process, nor is another unconventional mechanism or stomatal arrangement required to explain the nutrition of the `atmospheric' bromeliads. High affinities for key ions, combined with low plant demand, allow slow-growing, Type Five Bromeliaceae to subsist on supplies that could not sustain more vigorous ¯ ora (Chapters 4 and 5). Pronounced capacity to scavenge solutes, including certain toxic agents, also underlies the successful deployment of these bromeliads as air quality monitors, as detailed in Chapter 5. Tomlinson (1969) recognized three kinds of stomata based on structure and presumed operation. Related bromeliads possess similar and divergent types indicating capacity for rapid evolution at least in some lineages. Catopsis, Cottendorfia and a variety of similarly mesophytic types, mostly in the same two subfamilies, feature thin-walled guard and subsidiary cells aligned with the rest of the epidermis (Fig. 2.16A). An unobstructed substomatal chamber, which in the two more complex types accommodates extensions of the adjacent epidermal or hypodermal cells, indicates that these bromeliads probably ventilate leaf chlorenchyma in the usual manner. Chloroplasts distinguish the hypodermal cells that line the air passages leading to the mesophyll regardless of the type of stomata present (e.g., Figs. 2.13B, 2.16, 2.17A). Members of all three subfamilies (e.g., at least some Guzmania, Nidularium, Navia) possess the second type of stomatal apparatus, which is marked by thicker-walled guard cells underarched by portions of lateral and terminal subsidiary cells (Fig. 2.17A). Apertures can protrude above the adjacent epidermis, while the rest of the apparatus lies recessed below the leaf surface. An annulus of two to four, thin-walled, U-shaped, green hypodermal cells is located beneath the guard cells (Fig. 2.17A). Lobes from additional cells project into the substomatal chamber, meet, and sometimes even grow up into the stoma. Typical function seems less likely here. Modi® ed subsidiary and neighboring epidermal cells distinguish the Cambridge Books Online © Cambridge University Press, 2009 62 Vegetative structure third type of stomata (Fig. 2.13C). Unlike the second condition, the hypodermal cells in this instance lack conspicuous extensions; however, overlapping protuberances from in¯ ated subsidiary cells sometimes completely occlude the superstomatal cavity. In other examples, adjacent epidermal cells distort the stomatal complex enough to elevate its aperture above the adjacent leaf surface. With few exceptions, certain Bromelioideae and Pitcairnioideae (Ananas, Dyckia, Orthophytum, Puya) possess this most complex of the bromeliad stomata, and these plants always grow under semiarid to drier conditions. Observations made by two pioneering morphologists exemplify the controversy over stomatal function. Lindsbauer (1911) suggested that those protuberances located beneath some guard cells prevent closure even at zero turgor, whereas Haberlandt (1914) considered the same arrangement an impediment to gas exchange. Krauss (1948± 49) described how various cells located in part or wholly under the guard cells swell as turgor mounts and lift and push them apart to promote gas exchange. Tomlinson (1969) adopted a more cautious stance when he wrote that at least those organs with enlarged, lateral subsidiary cells `probably do not function like normal stomata' . More is known about the operation of bromeliad stomata today thanks largely to the efforts of ecophysiologists primarily interested in CAM and water economy among subjects as diverse as Aechmea nudicaulis, Bromelia humilis and Tillandsia usneoides. Foliar conductances were found to routinely oscillate between night and day as expected for CAM plants and nonrhythmically in specimens challenged by abrupt exposures to drier air (Figs. 4.16, 4.19). Only one investigation addressed stomatal mechanics directly. Martin and Peters (1984) applied abscissic acid, a regulator of guard cell turgor in other, better-known ¯ ora, to demonstrate the expected responses in T. usneoides. While these investigators con® rmed plant capacity to regulate diffusive conductance, no one has identi® ed which cells deserve the credit. Another rather exceptional claim made about bromeliad stomata concerns the smallest of the species. Once considered astomatous, the leaves of tiny Tillandsia bryoides (Fig. 2.1) supposedly ventilate through scattered patches lacking, or at most covered with a poorly developed, cuticle (Lindsbauer 1911). Indeed, parallels exist in other land ¯ ora, most notably among the so-called shootless orchids, which depend entirely on CAMequipped green roots without recognizable devices to regulate gas exchange (Benzing and Ott 1981). Instead, these orchids maintain positive carbon balance by coordinating phase three of CAM (the decarboxylation of Cambridge Books Online © Cambridge University Press, 2009 Foliage 63 stored malic acid) in light with capacity to ® x most of the resultant CO2 before it can diffuse to the atmosphere (Cockburn et al. 1985). Evans and Brown (1989b) eventually discovered widely scattered, paracytic stomata (0.07 stomata mm21) located under the trichome shields, but they made no assignment to any of Tomlinson' s three morphological types. Recognition that autotrophic organs perforce need not physically regulate gas exchange to achieve acceptable water and carbon economy might usefully inform future efforts to interpret bromeliad leaf structure and function when conclusions otherwise disagree with impressions of how leaves should operate to conserve moisture. Relationships between stomata and trichomes Trichomes and stomata occur juxtaposed on the foliage of Bromeliaceae other than Tillandsia bryoides. Pattern is most pronounced where both kinds of organs lie in the intercostal grooves of the stouter-leafed types (e.g., Brocchinia paniculata, many Bromelia species and additional Bromelioideae and Pitcairnioideae; e.g., Figs. 2.8F, 2.13A). Tillandsioideae less often exhibit this kind of arrangement. More provocative is the numerical relationship between trichomes and stomata that distinguishes the three subfamilies (Tomlinson 1969). Ratios between stomata and trichomes vary with leaf function and suggest basic conditions that help explain how Bromeliaceae could enter so many, often physically challenging, environments. Stomata and scales co-occur among Bromeliaceae at ratios extending from about 30:1 in Cottendorfia to 5:1 for sampled Pitcairnioideae except Cottendorfia, to 0.5:l for Type Five Tillandsia and just 1.5:1 for the entire subfamily. Bromelioideae fall between these two subfamily averages with 43 sampled species yielding a mean ratio of 3.2:1. Tomlinson' s (1969) implied homology between the trichome and stomata, the possibility of a relatively ® xed number of versatile initials, and the striking ecological differentiation of the subfamilies suggest special signi® cance for this range of numbers. Readily altered ratios of these two organs and the multiple functions possible for the bromeliad trichome, combined with a generally plastic plant architecture, perhaps set the stage for a radiation that, while not exceptional for the numbers of species produced, would exceed most others for adaptive variety and novelty. Leaves born by Type Five Bromeliaceae display dense covers of absorbing, light-re¯ ecting trichomes in lieu of conventional root systems. Conversely, relatively few stomata need be present to CO2-saturate the modest photosynthetic capacities of these dry-growing epiphytes and Cambridge Books Online © Cambridge University Press, 2009 64 Vegetative structure saxicoles (see discussion of optimization theory in Chapter 4). Where liberal supplies of moisture obtained through extensive root systems or capacious phytotelmata assure better-supplied, potentially more productive foliage (higher Amax), stomata must be more numerous, and the trichome functions primarily as an insulator (e.g., many Pitcairnia, Bromelia). Scattered scales with broad, interlocking shields supported by stalks of similarly inexpensive construction (thin walls and vacuolate if still alive) will suffice as exempli® ed by Bromeliaceae with conventional root/shoot apportionments (e.g., Figs. 2.5E,F, 2.8D). In essence, a range of patterns characterized by speci® c ratios and types of stomata vs. trichomes derived from some prototypical condition accommodate extant Bromeliaceae to contrasting kinds of environments. For example, reduction in either the number of organs per unit leaf surface, or simply the size of the trichome shield, occurred during radiation. Other features of shoots and roots also favored emergence of ecological variety and stress-tolerance, for instance when a dense layer of absorbing trichomes and impounding foliage began to complement one another to make several modes of nutrition possible and release the plant from dependence on water in soil (Chapters 4 and 5). Add the material economy granted by the relaxed need for roots to the many additional bene® ts the indumentum and an impounding shoot provide, and the array of substrates and climates accessible to evolving Bromeliaceae became even greater. According to this logic, capacity to allocate proportionally fewer epidermal initials to trichomes among the mesic forms (e.g., Type Four) permitted high stomatal density and sufficient diffusive conductance to support the productivity permitted by relatively abundant supplies of moisture and nutrients. Drier conditions and/or poorly developed root systems in turn mandated proportionally more trichomes of the absorptive type at the affordable cost of fewer stomata per unit leaf surface. Finer adjustments in epidermal structure, also with major consequences for ecophysiology, occurred along additional environmental gradients, at least among Tillandsioideae. For example, stomata serving the soft-leafed residents of humid montane sites (e.g., Guzmania sanguinea, Tillandsia bulbosa) aggregate in the exposed regions of the epidermis beyond the shields of adjacent trichomes in order to permit gas exchange during wet weather (Fig. 4.23F; Table 4.8). Tillandsia with con¯ uent, relatively hydrophilic indumenta necessarily occupy more arid sites to avoid suffocation. Here, far more often than not, the shields ¯ ex upward while dry, simultaneously exposing the stomata and scattering strong, potentially damaging irradiance as described below (Fig. 2.7A,B). Cambridge Books Online © Cambridge University Press, 2009 Foliage 65 Mesophyll Tissues below the epidermis reveal the presence of multiple mechanisms for carbon and water balance; leaves also display additional anatomical detail without recognized signi® cance. At one extreme, exempli® ed by Tillandsia usneoides, an undifferentiated mesophyll surrounds just three to ® ve small vascular bundles (Fig. 2.10A). Little additional strengthening tissue occurs here or in the leaves of many other Type Five species (e.g., T. paucifolia, T. ionantha, T. schiedeana), while some relatives (e.g., T. balbisiana, T. concolor) produce foliage that resists fracture even when folded. Sclerenchyma tends to develop around the veins and immediately below the epidermis depending on the species, and within a single plant, according to growing conditions, especially exposure. Most bromeliads possess a differentiated mesophyll. Multiple layers of collapsible, accordion-pleated water-storage parenchyma lie beneath both epidermal layers of the dry-growers, but depth is greatest on the adaxial side (e.g., Figs. 2.13B, 2.16). Many populations native to everwet sites also possess colorless hydrenchyma tissue, but considerably less than the volume of adjacent green mesophyll (Fig. 2.13E). Occasionally, chlorenchyma lies above and below a central nonphotosynthetic zone (e.g., Aechmea bracteata). Growth in full sun promotes scleri® cation and thicker hypodermal layers and reduces the concentrations of chlorophyll (e.g., Guzmania monostachia; Chapter 4; Table 4.6; Fig. 4.26). Transitions between the often sclerotic outer and inner, more delicate hypodermal zones range from gradual to abrupt. Sharp boundaries typically separate the hypodermis and chlorenchyma (e.g., Acanthostachys, Puya), but seldom is even a poorly differentiated palisade present (Fig. 2.16B,C). Extraordinarily capacious intercostal air lacunae for CAM plants extend the length of the blades of most Bromelia species and many other Type Two Bromeliaceae (Fig. 2.13B). Those of the utriculate leaf traverse the entire organ, including the base where they expand to become arenchyma-like as if required to ventilate tissue denied more immediate access to O2 by the contents of ® lled phytotelmata. Intercostal ducts are sometimes partitioned with diaphragms of abutting stellate-lobed cells much like those characteristic of the spongy mesophyll of many dicots (Fig. 2.13B). The single row of bundles that vascularizes the length of the bromeliad leaf varies in size across that appendage (Figs. 2.13E, 2.16C). Viewed in cross-section, they lie about midway through the thin organ, but closer to the abaxial surface in thicker blades because an adaxial hypodermis Cambridge Books Online © Cambridge University Press, 2009 66 Vegetative structure occupies more of the leaf interior. Extensions of the bundle sheaths usually terminate below the epidermis. Rendered free, the tough vascular bundles of Ananas comosus, Neoglaziovia, Aechmea magdalenae and several additional species of lesser importance support local ® ber industries (Chapter 14). Strands of nonconductive sclerenchyma augment leaf strength for many Bromelioideae (Fig. 2.13E). Those of Portea petropolitana contain septate ® bers, and, as elsewhere, occur in two series within the chlorenchyma, above and below, but never interspersed among the veins. Transverse commissures embedded in narrow septa comprised of compact cells, sometimes densely packed with chloroplasts as already mentioned for some of the variegated leaves, run above and below and cross the air lacunae (Fig. 2.17B). Each junction consists of series of short, narrow water-vascular cells and a thin phloem strand within a continuous sheath of parenchyma. Optical properties Horticulturists rely on several of the more conspicuous aspects of leaf texture, shape and trichome cover already discussed to decide how to cultivate speci® c Bromeliaceae. More subtle structural and chemical indicators provide more precise information about optimum growing conditions. Diagnostic histology, for example the Kranz anatomy of a C4 plant, identi® es the photosynthetic pathway, and accordingly, patterns of carbon and water balance. Speci® c leaf weights, and other indices of succulence and sclerophylly, relate to modes of mineral nutrition, water use, and aspects of leaf natural history, although often in ways that defy simple interpretation (Chapter 5). Finally, even less accessible data on chemical composition and physiology provide additional insights on plant life strategy. Leaf anatomy also reveals which conditions of lighting best suit speci® c Bromeliaceae. Optimum conditions range from high to low photosynthetic photon ¯ ux density (PPFD) and the duration of that exposure from relatively continuous through the day for the lithophyte and savanna-dwelling bromeliad to intermittent for those populations relegated to the forest understory. Still other features of the leaf determine its responses to spectral quality and beam alignment (collimation). Forest-dwelling Bromeliaceae experience diffuse and direct-beam irradiance in different proportions depending on the microsite. All four aspects of the energy supply ultimately determine which combinations of leaf and shoot characteristics favor photosynthesis most in speci® c environments. The presence of at least two photosynthetic pathways and leaf form and Cambridge Books Online © Cambridge University Press, 2009 Foliage 67 light response curves that also differ among closely related species document how rapidly the bromeliads have adapted to different growing conditions (Chapter 4). Distributions of these characteristics through the family also demonstrate that CAM and succulence emerged repeatedly within Pitcairnioideae and Tillandsioideae, just as drier conditions more than once promoted dependence on trichomes in lieu of phytotelmata and adsorptive roots (Table 4.1). However, the bromeliad literature says virtually nothing about the effects of light quality, collimation or sun ¯ ecks on photosynthesis. Surprisingly, none of the considerable text devoted to discussions of the presumed conditions in ancestral habitats (e.g., Bromelioideae; Benzing and Renfrow 1971b; Medina 1974) includes comments about ambient light beyond its intensity in what were supposedly either understory or more open sites. Findings elsewhere can assist interpretations of the functional consequences of leaf form in Bromeliaceae as long as the effects of the light-scattering indumentum are taken into consideration. While the absorptive trichome of a dry-growing Tillandsia promotes nutrition and water balance, it also affects ecotolerance by altering the amount of irradiance available to chloroplasts. Light scattering reduces the threat to vulnerable pigment± protein complexes in some cases, while leaf structure of another sort focuses incident diffuse light and enhances photosynthesis for other species native to lower-energy habitats. Many factors, including cell structure, the relative amounts of certain pigments present, and leaf attitude relative to the source determine how light travels through an epidermis and into the underlying chlorenchyma. Penetration may be enhanced or inhibited by several organ-speci® c characteristics. Raised trichome shields scatter photons from the foliage of drygrowing Tillandsioideae and many other bromeliads (e.g., Fig. 2.8C,E). Conversely, thin-® lm phenomena may help compensate certain understory plants for the differential diminution of shorter-wave radiation in shadelight (Lee et al. 1979). Foliage in such cases appears bluish green under the canopy compared with beyond it. Probably no bromeliad matches the performances of the spectacularly colored, exceptionally shade-tolerant pteridophytes, but several taxa (e.g., Nidularium burchellii; Fig. 2.4H) native to dense rainforest suggest tendencies (fainter bluish tints) in that direction. Timing, speci® cally the capacity of foliage to maximize energy harvest from sun ¯ ecks, as elsewhere, probably depends more on leaf physiology than optics or structure. Anatomy affects leaf transparency beginning at the cuticle, which can be highly re¯ ective to transparent, on down through the mesophyll to the Cambridge Books Online © Cambridge University Press, 2009 68 Vegetative structure opposing epidermis (Vogelmann and Martin 1993). Convex outer tangential walls of the adaxial epidermis of some understory ¯ ora concentrate photons at focal points in underlying chlorenchyma with corresponding diminutions elsewhere, especially if the source is diffuse rather than collimated. Within leaves, the shapes of the mesophyll cells and adjacent air spaces further in¯ uence the paths followed by photons. Organs equipped with a palisade transmit larger proportions of incident, high-angle light than targets comprised of less aligned or more isodiametric cells. Conversely, foliage featuring relatively undifferentiated interiors more readily quenches collimated light, i.e., experiences more intrafoliar shading. Light measured deep in the leaf indicates which kinds of anatomy favor photosynthesis most under speci® c exposures, i.e., which arrangements either screen vulnerable tissues from excess radiation or enhance the capture of marginally adequate photon ¯ ux (Vogelmann and Martin 1993). Presumably, ® ber optic probes could also be employed to determine the suitability of a bromeliad to utilize shade (primarily diffuse) vs. un® ltered (,85% collimated at noon) insolation. Populations (e.g., Ananas comosus; Medina et al. 1991a, 1993) known for their capacity to accommodate speci® c ¯ uences could serve as subjects for tests. Experiments with cloned materials maintained under different light regimens in turn would reveal whether cell shapes and orientations exhibit more or less plasticity in widely tolerant compared with ecologically more constrained genotypes. Unexpected sources sometimes identify interesting subjects for investigations of the light relations of Bromeliaceae. Casual comments on herbarium labels or in the literature may indicate a phenomenon worth consideration. For example, Hallwachs (1983) reported a bluish sheen to the leaves of a forest-dwelling population of Bromelia pinguin in Costa Rica. Usually this wide-ranging terrestrial experiences stronger light, often as a `living fence' where stouter shoots develop shades of yellow-green. Conceivably, its ability to acclimate so broadly rests partly on capacity to alter leaf optics and chemistry to either enhance the capture of shade-light or avoid injury in fuller sun. Conversely, more static arrangements may preclude similar ¯ exibility and, if also insensitive to natural selection, represent a signi® cant constraint on evolution. A constellation of Pitcairnia species (sensu lato) may demonstrate such inertia, in this case related to the organization of the mesophyll (G. S. Varadarajan, personal communication). Few monocotyledons, including Bromeliaceae, possess a palisade comparable to the arrangement that prevails so widely in Magnoliopsida consistent with the linear, liliaceous leaf that so often orients parallel rather Cambridge Books Online © Cambridge University Press, 2009 Foliage 69 than perpendicular to direct-beam irradiance. Most Pitcairnioideae, including all or much of Ayensua, Connellia, Cottendorfia, Lindmania, Navia and Pitcairnia, follow this precedent by producing undifferentiated chlorenchyma (Fig. 2.16C). However, a smaller group of thin-leafed, forest-dwelling taxa exhibit an anticlinally elongated palisade above a basement layer of less densely packed and uniformly shaped components (Fig. 2.16B). Not surprisingly, leaves in this second case exhibit broad, ¯ at laminae held more or less perpendicular to the most intense direct-beam insolation. Additional characters unrelated to shade-tolerance, but indicative of phylogenetic relationship, suggest that these two conditions in¯ uenced evolution within Pitcairnia (sensu lato). Species assigned to closely related Pepinia or Pitcairnia (sensu lato) that share one of the two conditions of the mesophyll also differ relative to the presence or absence of stolons, dimorphic or trimorphic foliage, deciduousness and certain aspects of trichome morphology (G. S. Varadarajan, personal communication). Speci® cally, Pepinia seems to circumscribe two clades that parallel Pitcairnia in being more or less equipped to harvest shade (scattered) light. Taxa with dimorphic chlorenchyma inhabit dark understories (e.g., Pepinia corallina, Pepinia schultzei, Pitcairnia trianae; Fig. 2.16B); the others more often grow as facultative epiphytes or riparian terrestrials (e.g., Pepinia aphelandrifora, Pitcairnia pungens) where they typically receive strong direct-beam irradiance. Similar patterns of leaf anatomy segregate the Puya species by altitude and drought-tolerance as measured by the xericity of their habitats (G. S. Varadarajan, personal communication). Again, shared morphology involving other organs with no direct in¯ uence on light use (e.g., in¯ orescence branched or condensed) indicates the existence of two clades, each disposed by leaf anatomy to colonize either strongly arid or more humid habitats. However, Puya appear less differentiated by light environment because the more mesic species inhabit open wetlands rather than the understories of dense forests. Downs (1974) reported additional species with `palisade-like' layers, and some of the epiphytes among them regularly experience high PPFD as upper-canopy specialists. Aechmea tillandsioides almost always roots in ant carton, a medium that its architects assure will occupy sunny microsites (Chapter 8). Aechmea nudicaulis also tends to encounter full exposure in situ, especially in Brazilian restinga (Fig. 7.13C), across one of the broadest ranges achieved by a bromelioid. Other taxa with similar leaf anatomy include Billbergia macrolepis, Brocchinia acuminata and Vriesea malzinei, indicating convergence on the same arrangement for mesophyll in all three Cambridge Books Online © Cambridge University Press, 2009 70 Vegetative structure subfamilies. Examination of relatives with pronounced propensities for darker or brighter habitats would help sort out which details of leaf anatomy foster success in speci® c kinds of light environments. Trichomes Diverse agencies mediate the properties of plant boundary layers, particularly those covering foliage. The leaf epidermis stands out as a barrier that must at once deter pathogens and predators, regulate gas exchange, help maintain thermal budgets, and either screen excess light or enhance its capture depending on PPFD and plant tolerances. Two additional tasks, water and ion uptake, further complicate function for the bromeliads lacking absorptive roots. Not surprisingly, these are the same species that bear the most specialized foliar indumenta, and have received the greatest attention from investigators. Recognition that the foliar trichome of xerophytic Tillandsioideae performs operations usually associated with roots goes back more than a century, but full appreciation of the consequence of this realignment remains obscure. Recent revelations about the structure and operation of the trichome of carnivorous Brocchinia reducta (Chapter 5) demonstrate the value of more penetrating studies than most of those performed so far. This ® nal section centers on the bromeliad trichome, speci® cally, how it in¯ uences ecology through effects on a variety of basic leaf functions. Chapters 4 and 5 more thoroughly consider foliage as root analog. Almost all of the bromeliads possess some sort of foliar indumentum that also often covers parts of the stem, in¯ orescence and ¯ ower. Glabrous Navia species and others equipped with simple uniseriate hairs (e.g., Cottendorfia) belong exclusively to Pitcairnioideae (Robinson 1969; Fig. 2.5K,L,N). Relatively unspecialized ¯ owers, fruits and seeds, terrestrial habits, and well-developed root systems indicate that of the three subfamilies recognized by Smith and Downs (1974), this one, minus Brocchinia, most closely resembles the extinct, common ancestor, i.e., constitutes the lowest of the evolutionary grades relative to ecology and vegetative structure, including the foliar trichome. However, molecular systematics indicates parallel rather than basal status for Pitcairnioideae (Chapter 9). If so, the uniseriate trichome and glabrous epidermis represent apomorphies, and the peltate hair or scale (e.g., Fig. 2.7), structure and function more reminiscent of earlier times. Perhaps signi® cantly, unbranched hairs also occur on the foliage of pineapple seedlings (Krauss 1948± 49). Strehl and Winkler (1981) and Strehl (1983) assigned the types of tri- Cambridge Books Online © Cambridge University Press, 2009 Trichomes 71 chomes present in each subfamily primitive to advanced status according to the organization of the shield. The peltate scale occupies the basal position in Pitcairnioideae with the uniseriate form considered derived. Aechmea bracteata illustrates the fundamental condition for Bromelioideae, and those organs bearing the less common stellate shield supposedly arose later. Guzmania monostachia receives the same designation in Tillandsioideae, while the relatively reduced trichome of the more mesic forms (e.g., Catopsis nutans; Fig. 2.8B) and those with the most expansive shields (e.g., Tillandsia crocata; Fig. 2.7D) are considered specialized for wetter and drier environments respectively. All three schemes seem premature, but history need not be known to determine functions. About the only trait shared by all bromeliad trichomes is multicellularity. Even the simplest, unbranched appendages (Fig. 2.5K,L,N) consist of adjoined cells, and members of the same genus and several others (e.g., some Fosterella, Navia, Pepinia) possess stellate types (Fig. 2.5F,M). Virtually all other Bromeliaceae bear scales comprised of a two to many-celled stalk topped by a plate-like, single-layered, multicellular shield. Unlike those of the stalk, components of the shield usually die at maturity, but not before their walls develop important optical, hygroscopic and mechanical properties. Aspects of indumenta important to insulation, such as the size and shape of the shield and percent of the leaf surface covered, more closely parallel growing conditions than taxonomic boundaries. Additional sets of characteristics vary to lesser degrees, but impacts on function may be more fundamental. Trichomes of Brocchinia, for example, usually conform to the typical scale architecture (e.g., B. acuminata), the exceptions being those of B. reducta and B. hechtioides that exhibit a goblet shape presumably related to carnivory (Figs. 2.5A, 5.2E,F). On the other hand, every cell of the mature trichome of several, if not all, members of this genus retains its protoplast at maturity, a near novelty in the family (Fig. 5.2G). On a ® ner scale yet, conspeci® cs (e.g., apparent ecotypes of Tillandsia caput-medusae) growing under different exposures and humidities occasionally exhibit indumenta that match those conditions, especially PPFD (Dimmitt 1985). Variables in this case involve both the structure of the individual appendage and density over the leaf surface. Brie¯ y, the widely occurring scale, and particularly the organization of its shield, accounts for most of the recorded structural variety among bromeliad trichomes. Table 2.1 summarizes the more salient of these anatomical peculiarities and identi® es where each type occurs in the family. Information is organized according to function, which ranges from secretion to the attraction of seed dispersers and pollinators. Cambridge Books Online © Cambridge University Press, 2009 72 Vegetative structure Tillandsioideae Billings (1904) described the ontogeny of the trichome of Spanish moss and probably the pattern for the entire subfamily (Fig. 2.9). Development begins with a single initial programmed to produce a uniseriate chain topped by a cell destined to divide into four, equal-sized, wedge-shaped daughters. Each of these four cells in turn divides unequally to yield the central disc to the inside, while the smaller daughters proceed to form characteristic rings of cells and ultimately the wing. The result is a complex device capable of changing shape to effect different functions according to plant needs. Consider the most specialized version of this organ beginning with the central disc. Exceptionally robust, upper tangential walls alternately rise and fall (xerophytic taxa, less so mesophytic types) on ¯ exible radial walls as the adjacent cavities imbibe and lose water as weather dictates (Figs. 2.7A,B, 4.20, 4.21). While dry and collapsed, the upper tangential walls block evaporation through the distal (dome) cell of the subjacent stalk. Elevated, they expose the same cell to ¯ uids that ® ll the hydrated shield. Below, two small foot cells anchor the entire appendage and mediate water and ion ¯ ux from stalk to adjacent mesophyll. Beyond the central disc lie one to four series of rings, each containing twice the number of components as the one just inside (i.e., 8, 16, 32, 64) except for the outermost series that may be incomplete (Fig. 2.7D± F,H). Radial walls, except those of the cells of the central disc, become thickened and rigid. Wings containing more than twice the number of cells present in the outermost ring greatly expand the width of the shield of dry-growing Tillandsioideae. Shields with the widest wings move more than the simple rise and fall of the central disc as leaf surfaces alternately moisten and dry (Fig. 2.7A,B). A soft hinge provided by one or more series of ring cells equipped with pliant upper tangential walls allows the expanded wing to ¯ ex upward upon drying and scatter rather than propagate incident radiation. Warts, ridges and additional irregularities sometimes further enhance re¯ ectivity (e.g., Tillandsia karwinskyana; Figs. 2.7F, 4.23H). Wing development varies with the species, at one extreme endowing the mist-dependent types (e.g., T. crocata, T. tectorum; Figs. 2.7D, 2.8C) with shields that extend far enough above the epidermis to intercept aerosols. Leaves of the more shade-tolerant mesophytes support more widely scattered scales with shorter, nonoverlapping wings that cover less than 5% of the blade (e.g., Catopsis nutans; Cambridge Books Online © Cambridge University Press, 2009 Trichomes 73 Fig. 2.8B). Many-celled stalks (highest in Glomeropitcairnia) supposedly represent the basic condition in Tillandsioideae, but the evidence is thin. Pitcairnioideae The typically peltate trichomes of Pitcairnioideae feature one or fourcelled, small central discs that anchor expansive shields (Figs. 2.5, 2.8D). Stalks tend to be narrow, few-celled, and equipped with less-developed protoplasts than those featured by Tillandsioideae. Shield outline ranges from more or less circular to oblong with entire to deeply incised margins. Stellate form marks the exceptional species (e.g., Fosterella; Fig. 2.5F) as do uniseriate trichomes or none at all (e.g., certain Navia). Concentricity suggestive of Tillandsioideae distinguishes occasional Pitcairnioideae, most notably members of Brocchinia, as described below (Fig. 2.5B,H,I). Navia glandulosa stands out as one of the two bromeliads reported to bear capitate, glandular hairs, in this instance predominantly on the ¯ oral bracts (Fig. 2.5K). Ronnbergia petersi (Bromelioideae) produces similar appendages combined with more typical scales on its sepals (Gross 1991). Trichome structure identi® es Pitcairnioideae allied by similar ecology, but not as faithfully as in Tillandsioideae. Narrow, radially elongated cells often produce a characteristic shield margin among the dry-growers (e.g., Hechtia) and fewer of the more mesic taxa (e.g., some Fosterella, Pepinia). Alpine Puya feature dense layers of sometimes brownish woolly scales (Fig. 7.2). Shields never develop the structure that mediates the valve-like (hydrorectifying) action characteristic of dry-growing (Type Five) Tillandsioideae. Arid-land Pitcairnioideae usually bear trichomes with relatively opaque shields compared with relatives from moister, darker habitats. The pitcairnioid indumentum tends to occur unevenly across the leaf surface, often more densely on the abaxial than on the adaxial side (e.g., Pitcairnia). Scales of many of the dry-growing types anchor between the costa, especially on the abaxial epidermis, where they insulate the stomata (e.g., Dyckia). Relatives from humid sites and those few species that shed foliage preparatory to drought produce scattered trichomes over one or both surfaces. Quite a few of the mesophytes (e.g., Lindmania, Navia) lack adaxial trichomes. Banded indumenta responsible for the ornamented foliage of certain Bromelioideae (e.g., Aechmea chantinii, Billbergia zebrina and especially certain Cryptanthus; Figs. 2.14F, 2.18C) and Tillandsioideae (e.g., Tillandsia flexuosa, T. hildae; Fig. 2.7I) have no parallel in Pitcairnioideae. Cambridge Books Online © Cambridge University Press, 2009 74 Vegetative structure Special mention is due Brocchinia, which, although assigned by Smith and Downs (1974) to Pitcairnioideae, exhibit trichomes that by structure and function (Figs. 2.5B,H,I, 5.2F,G) more closely parallel counterparts in Tillandsioideae. Circular to oval shields contain different numbers of cells, but concentric alignment is common. Relatively elongate components form a rudimentary (e.g., B. reducta) to a better-de® ned (e.g., B. acuminata, B. micrantha) wing peripheral to a less organized group of central cells. Stalks may be uniseriate (e.g., B. acuminata; Fig. 5.2G) or expanded to several cells near the top (e.g., B. reducta; Fig. 2.5A). Well-developed phytotelm architecture and demonstrated absorptive capacity among some members of this enigmatic genus suggest circumstances that also fostered the assumption of many root functions by leaves in Tillandsioideae. A hypothesis offered in Chapter 9 describes how absorptive capacity may have evolved in the trichome of Brocchinia and perhaps elsewhere among primitive Bromeliaceae. Bromelioideae Bromelioideae produce scales organized more like those of Pitcairnioideae than of Tillandsioideae (Fig. 2.6). Speci® cally, a small disc of four cells may occupy the center of the shield (e.g., Billbergia, Cryptanthus, Orthophytum) and those beyond sometimes form rudimentary rings (e.g., Aechmea rosea, Neoregelia pauciflora). One or two of the outermost series of shield cells sometimes constitute a discernible wing (e.g., Canistrum fosterianum; Fig. 2.6F), but alignments never meet the standard set by Tillandsioideae. Modestly lobed to deeply divided margins also parallel those of certain Pitcairnioideae, while pronounced stellate morphology (e.g., some Streptocalyx species) approaches arrangements in this taxon even more. Shields with more or less circular form and entire margins occur widely through many of the largest genera (e.g., some Billbergia, Bromelia, Cryptanthus); those with strongly asymmetric outlines (e.g., Acanthostachys, other Bromelia) comprise a small minority. Two-celled stalks characterize Neoglaziovia, Orthophytum and Pseudananas, three distinguish Greigia, and up to 10 components anchor the shields in additional taxa (e.g., Canistrum, Nidularium; Fig. 2.6). Stalks in all instances lack clear differentiation into a dome and smaller transfer cells as in Pitcairnioideae. Beyond Ananas comosus (Sakai and Sanford 1979), no reports mention ® ne structure that might promote absorption. Scales on the leaf bases of at least some phytotelm Bromelioideae take up nutrients from adjacent phytotelmata (Benzing et al. 1976), but not as Cambridge Books Online © Cambridge University Press, 2009 Trichomes 75 avidly as those of the xerophytic Tillandsioideae tested under the same conditions (Chapter 5). Trichomes of Bromelioideae often occur unevenly across the leaf. Deep intercostal grooves accommodate the scales and stomata of many of the thicker-leafed xerophytes (e.g., Bromelia; Figs. 2.8F, 2.13A) where they probably help conserve moisture. Some Cryptanthus and Aechmea species bear highly ornamental bands of re¯ ective scales (Figs. 2.14F, 2.18C). More often con¯ uent indumenta cover the abaxial surface, whereas the usually astomatous adaxial epidermis features fewer to no scales (e.g., some Cryptanthus and Nidularium). Denser concentrations of trichomes typically invest the leaf bases of the phytotelm forms, including members of Type Two (e.g., Ananas, Bromelia; Fig. 2.14A,B). Shields are generally also wider on the leaf base compared with the blade (e.g., Aechmea bracteata). Trichomes of Bromelioideae have attracted less attention than those of the other two subfamilies, although more than half of the 50-plus bromeliad genera are assigned here. The 800 or so member species occupy a somewhat less extensive array of habitats than colonized by the other two subfamilies (e.g., virtually none in upper montane cloud forests). Trichome functions beyond absorption Bromeliad trichomes perform many of the same functions documented among other Magnoliophyta (Table 2.1). Demonstrated and putative services vary with the species. Spiculate hairs born by certain Pitcairnioidae may impale soft-bodied predators much as similarly shaped organs do elsewhere. Interlocking shields of the peltate types (Fig. 2.8D), in addition to impacts on leaf energy budgets and gas exchange, may shield underlying stomata from invading hyphae, or deter small predators seeking more nutritious mesophyll. Trichomes comprised entirely of living cells (Brocchinia) might contribute digestive secretions to adjacent tank ¯ uids. No bromeliad reportedly elaborates repellents or toxins for protection, although those capitate hairs just described for Navia glandulifera (Holst 1996) could do just that (Fig. 2.5K). Silvery trichomes on the short, compact (nidulate) in¯ orescence of forest-dwelling Neoregelia longisepala produce a strong visual signal for fauna seeking nectar or edible fruit deep in the heavily shaded center of its large rosette. Highly re¯ ective trichomes on the otherwise drab fruits of certain Billbergia seem well disposed to guide fruit-feeding bats (Fig. 3.5G). Multiple, sometimes inherently antagonistic, functions suggest that the Cambridge Books Online © Cambridge University Press, 2009 76 Vegetative structure leaf epidermis constrains ecological options most in Tillandsioideae. Even so, species with similarly dense layers of absorbing trichomes experience diverse microclimates and anchor on widely differing kinds of substrates. Recall that Type Five species, more than 250 in all, share many aspects of foliage in addition to dense indumenta that affect ecophysiology (e.g., CAM, succulence), yet these plants tolerate cool to high temperatures, moisture supplies as distinct as rain, fog and dew, full sun to partial shade and so on. So how can ¯ ora equipped with foliage characterized by the same combination of specializations maintain tolerable thermal, moisture, carbon and nutrient budgets under such disparate growing conditions? Mez (1904) recognized that most dry-growing Tillandsia (Type Five) fail in wet, shaded locations because the foliar epidermis of these plants prohibits survival under conditions suitable for many other ¯ ora. Capacity to endure drought, often while rooted on unyielding substrates as these plants typically do, requires a profound evolutionary trade-off. Hydrophilic scales that so effectively enhance the impact of brief showers on plant hydrature suffocate foliage moistened more frequently because they hold ® lms of moisture over the stomata (Figs. 2.8C,E, 4.11; Table 4.8). A subtly different indumentum mediates another kind of performance for some close relatives of these same obligate xerophytes (Benzing et al. 1978). A few Type Five Tillandsia species (e.g., T. bulbosa; Figs. 4.23F,G, 8.5A) exhibit unexpected tolerance for shade and humidity. Wetted at night, these plants consume CO2 as if still dry (Table 4.8; Fig. 4.11). Exceptional trichome structure and rigidity and distribution relative to the stomata explain this difference. Although organized like those of other Tillandsioideae, the abaxial trichomes born by these species fail to move like their counterparts serving Spanish moss and its kind, and the shield wings lie beyond the stomata (Fig. 4.23F). Permanently ¯ attened against the leaf surface, the scale probably promotes rather than reduces light propagation through the epidermis and, additionally, sheds water. Trichomes on the largely inrolled and astomatous, adaxial surface exhibit the more typical Type Five form and scatter irradiance, although not necessarily to the advantage of T. bulbosa in its shady habitats (Fig. 4.23G). Perspectives on how different features of the indumentum affect heat loads, promote the uptake and retention of moisture, and help expose or shield chlorenchyma from light require biophysical analysis. For example, shields of some of the most drought and heat-tolerant tillandsias, like those of T. bulbosa, move little if at all, but often exhibit elaborate ornamentations (e.g., Tillandsia karwinskyana; Figs. 2.7F, 4.23H). Does the rigid shield displayed by this dry-growing species enhance stress-tolerance by Cambridge Books Online © Cambridge University Press, 2009 Trichomes 77 insulating the stomata? If so, do those accompanying warty walls obviate the need for the light-scattering effect of the ¯ exible trichome? Moving on, does the attenuated wing of the mist-dependent bromeliad (T. tectorum; Fig. 2.8C) promote condensation, and if so, how close does its structure approach the theoretical ideal under ambient conditions? What features re¯ ect the trade-off that has evolved between the needs of these plants to re¯ ect light and obtain and conserve moisture? How much self-shade is cast by various types of indumenta? Might certain of the more transparent trichomes focus scattered photons and enhance photosynthesis in shade-light? Does light-piping ever promote shade-tolerance, for instance for densely trichomed T. pruinosa, which in Florida occupies understories free of the other local Type Five Bromeliaceae? Scale variety among extant Tillandsioideae depicts radiation in miniature ± change in a single organ that assisted colonization of diverse and often demanding habitats by well over 1000 species of bromeliads. Cambridge Books Online © Cambridge University Press, 2009 Cambridge Books Online © Cambridge University Press, 2009 3 Reproductive structure A substantial literature dating back more than a century describes the bromeliad reproductive apparatus. Taxonomists working with dried specimens authored most of the early treatments. Interest continues, but specimen quality has improved allowing analyses to be more comprehensive. For example, Brown and Terry (1992) used liquid-preserved ¯ owers and scanning electron microscopy to determine that the delicate petal scale that ® gures so prominently in the most recent monograph of the family (Smith and Downs 1974, 1977, 1979; Fig. 3.1) circumscribes some genera more convincingly than others. Wet material has also permitted determinations of when certain features appear during ontogeny, and accordingly, their utility for distinguishing taxa of low vs. higher rank. Plant form underlying reproductive phenomena like pollination and seed dispersal and the genetic structure of populations are our primary concern for this review. Unfortunately, few of the hundreds of publications devoted to the reproductive apparatus of Bromeliaceae provide much insight on any of these subjects. Moreover, inquiry on ¯ owers, fruits and seeds continues to be motivated primarily by interests in systematics. The exceptional report that does depart from tradition usually addresses the same question, namely who pollinates which bromeliad? Today, molecular biology is augmenting the morphological data traditionally used to infer bromeliad history. However, cladograms based on nucleotide sequences must be more fully resolved than those illustrated in Chapter 9 to produce the phylogeny necessary to determine where, when and how often decisive features of the reproductive apparatus evolved. Additional information on gross morphology is also needed to address questions such as whether the different conditions of the ¯ ight apparatus of the seeds of Tillandsioideae re¯ ect separate origins (Palací 1997). Speci® cally, do Tillandsia/Vriesea vs. Catopsis or Glomeropitcairnia share homologous or convergent coma morphology? 79 Cambridge Books Online © Cambridge University Press, 2009 80 Reproductive structure Figure 3.1. Aspects of ¯ ower structure among Bromeliaceae. (A) Dyckia ragonesei gynoecium illustrating septal nectaries, placentation and partial fusion of the carpels. (B) Six versions of petal scale morphology in six species representing all three subfamilies. (C) Five recognized stigma types. Part A redrawn from Bernardello et al. (1991); B redrawn from Brown and Terry (1992); C redrawn from Brown and Gilmartin (1989b). This chapter begins with brief surveys of the organization of the bromeliad ¯ ower, fruit, seed and in¯ orescence, emphasizing features that circumscribe genera and subfamilies. Next, we turn to aspects of the reproductive apparatus responsible for distinguishing Bromeliaceae biologically (e.g., heavily epiphytic, extraordinary reliance on birds for pollen dispersal) among angiosperms and shaping plant impacts in communities. Relationships between plant structure and function receive priority throughout. Much additional information on bromeliad reproduction, including more detailed gross structure, appears in Chapters 6 and 9. Cambridge Books Online © Cambridge University Press, 2009 Inflorescences 81 Inflorescences Most bromeliads ¯ ower terminally, either just once if the genet is monocarpic, or repeatedly in turn from each of a potentially in® nite number of ramets produced by sympodial branching (Fig. 2.3A,B). Many fewer taxa produce lateral in¯ orescences, i.e., are monopodial (Figs. 2.2B, 2.3C). Structure usually differentiates the vegetative from the reproductive portion of the shoot in the sense that the ¯ owers and ¯ oriferous branches of the more complex in¯ orescences arise from the axils of bracts rather than from undiminished foliage (but see Cryptanthus; Fig. 3.2H). Racemes and spikes characterize an inordinate number of species; heads typify several more genera (e.g., Neoregelia, Nidularium). Panicles comprised of racemes and spikes occur among the larger-bodied members of all three subfamilies (Figs. 3.2± 3.4). Architecture and peculiarities of development differ, sometimes even among closely related populations, indicating considerable evolutionary plasticity and reason for caution when choosing taxonomic markers. Tillandsioideae illustrate both the simplest and some of the most complex in¯ orescences in the family (Fig. 3.3; Chapter 12). Reduction to a single ¯ ower occurs exclusively in Tillandsia (e.g., Tillandsia usneoides) where miniaturization associated with neoteny precludes more substantial reproductive efforts (Chapter 6; Fig. 2.1). Flowers and branches born on the more elaborate systems associate in either distichous (e.g., Vriesea bituminosa, V. hydrophora, many Tillandsia species; Figs. 3.3A, 3.5A,C) or polystichous arrangements (e.g., Catopsis, Guzmania, Tillandsia imperialis; Fig. 3.3G,H). Pedicles sometimes twist in one direction to produce a secund spike (e.g., Vriesea oligantha; Fig. 3.3E), or they align the ¯ owers in similar fashion along the subdivisions in branched systems (e.g., Tillandsia secunda). Internodes range from elongate (often true of the main axis) to telescoped (more often the ¯ ower-bearing axes), yielding arrays of relatively lax to more congested ¯ owers respectively. Tight heads surrounded by petal-like bracts (Fig. 3.3K) suggest the more familiar pseudanthium of Asteraceae. Core Bromelioideae (e.g., Aechmea, Billbergia, Neoregelia; Fig. 3.2) exhibit similar variety, including many to few-¯ owered forms (e.g., Aechmea pectinata, Neoregelia ampullacea respectively), condensed to spreading types (e.g., various Aechmea, Portea), and species with distichous (e.g., Aechmea tillandsioides) or polystichous (A. bromeliifolia) arrangements of ¯ owers. A few taxa produce pseudanthia (e.g., Canistrum). Much the same variety can be cited for Pitcairnioideae (Fig. 3.4). Architectural Cambridge Books Online © Cambridge University Press, 2009 82 Reproductive structure Figure 3.2. Representative ¯ owers and in¯ orescences of Bromelioideae. (A) Nidulate Neoregelia. (B) Aechmea fulgens. (C) Aechmea bracteata. (D) Aechmea setigera illustrating armed ¯ oral bracts. (E) Greigia sp., whole plant and axillary in¯ orescence. (F) Billbergia amoena showing conspicuous ephemeral scape bracts. (G) Aechmea fasciata. (H) Cryptanthus correia-araujoi, ¯ owers subtended by foliose bracts. Cambridge Books Online © Cambridge University Press, 2009 Inflorescences 83 Figure 3.3. Representative ¯ owers and in¯ orescences of Tillandsioideae. (A) Tillandsia cyanea, prominent ¯ oral bracts. (B) Tillandsia argentea, ¯ oral bracts much reduced. (C) Tillandsia loliacea, miniaturized ¯ owers consistent with neoteny (arrow). (D) Guzmania wittmackii, foliose scape bracts. (E) Vriesea oligantha, secund spike. (F) Tillandsia xiphioides, ® mbriate corolla. (G) Guzmania globosa, head enveloped in mucilage. (H) Catopsis sessiliflora, staminate (below) and pistillate (above) plants, fruit, seed and structure of pistillate ¯ ower. Note vestigial stamens. (I) Tillandsia streptocarpa, mass ¯ owering species. (J) Alcantarea nevaresii. (K) Guzmania lingulata, in¯ orescence forms pseudanthium. (L) Tillandsia albertiana. (M) Tillandsia viridiflora, chiropterophilous. Cambridge Books Online © Cambridge University Press, 2009 84 Reproductive structure Figure 3.4. Representative ¯ owers and in¯ orescences of Pitcairnioideae. (A) Navia caulescens. (B) Navia linearis with isolated pistil. (C) Navia polyglomerata with isolated pistil. (D) Entomophilous Fosterella penduliflora (below) and ornithophious Fosterella spectabilis (above). (E) Pitcairnia with radial ¯ ower and ¯ aring corolla. (F) Pepinia pruinosa. (G) Chiropterophilous Encholirium glasiovii. (H) Chiropterophilous Pitcairnia brongniartiana. (I) Sepal nectary of Dyckia floribunda in section. (J) Perennial in¯ orescence of Deuterocohnia meziana; arrows indicate sites of proliferations for additional ¯ owering after the ® rst season. (K) Pitcairnia flammea with zygomorphic ¯ ower in cyme. (L) Pitcairnia bakeri showing dense spike. (M) Pitcairnia arcuata; stippling on older ¯ oral bracts indicates extent of progressive deliquescence. Cambridge Books Online © Cambridge University Press, 2009 Inflorescences 85 redundancy across all three subfamilies reveals the distinctly homoplasious nature of the organization of the bromeliad in¯ orescence. Other groups of monocots, for example Poaceae, parallel Bromeliaceae for diverse in¯ orescence structure, except that the small, wind-pollinated grass ¯ ower allows more compact arrangements, and the bracts tend to protect and help disperse seeds rather than attract pollinators and frugivores. Imperfect ¯ owers also occur in both families, but the bromeliads more often exhibit related dimorphism. Pistillate plants of dioecious Catopsis and Hechtia feature more abbreviated in¯ orescences with fewer ¯ owers than their male counterparts (Fig. 3.3H). Pistillate Hechtia carlsoniae produces a spike, while its staminate counterpart is di- to tripinnately branched, presumably to enhance male relative to female function (assure high ratios of pollen to ovules). Anthesis usually proceeds acropetally through spikes and cymes (e.g., Encholirium; Figs. 3.4G, 6.2A), and from the outside inward for heads (e.g., Neoregelia; Fig. 3.2A). Exceptions include certain Canistrum species where the buds in the middle of what approaches a capitulum open ® rst. Members of Aechmea section Ortgiesia ¯ ower basipetally, (from the top down), and occasionally from the middle in both directions. Deuterocohnia meziana lacks rivals for its shrubby in¯ orescence that ¯ owers repeatedly for 6± 8 years (Fig. 3.4J). Thick axes warrant closer inspection to con® rm the reputed presence of a vascular cambium. Occasional proliferation of additional ¯ oral primordia around the periphery of spent infructescences allows some of the ramets of several members of Neoregelia subgenus Hylaeacium (e.g., N. eleutheropetala, N. myrmecophila) to reproduce during a second season. Lateral in¯ orescences characterize Greigia and certain members of Dyckia, Encholirium and Hechtia among others (Figs. 2.2B, 3.2E, 6.2B). Tillandsia complanata produces small, multiple spikes on lax scapes from axillary buds on indeterminate shoots; outwardly similar T. multicaulis and T. monstrum in fact remain cryptically sympodial as described in the previous chapter. Lateral ¯ owering occurs sporadically in the exceptional, sympodially branched population, for example the occasional in¯ orescence of Quesnelia lateralis that arises as a neotenic ramet devoid of expanded foliage. Discovery that conspeci® cs ¯ ower from one or the other location obliged the synonymy of Q. centralis with Q. lateralis. Bracts, which sometimes occur in several orders and sizes on the same in¯ orescence, assist reproduction by targeting speci® c pollen and/or seed dispersers. They also protect the developing ¯ owers and later sometimes the ripening fruit (Figs. 3.2± 3.4). Sterile nodes below those bearing ¯ owers Books Online © Cambridge University Press, 2009 86 Reproductive structure usually subtend the largest, most colorful appendages among Bromelioideae (e.g., Billbergia; Fig. 3.2F); similar organs tend to distribute more evenly through the in¯ oresences of Pitcairnioideae and especially Tillandsioideae (Fig. 3.4L,M). A sterile 15± 25 cm extension of the otherwise nidulate in¯ orescence of Guzmania sanguinea var. comosa bears bright orange bracts whose purpose can only be attraction for birds. Bracts may be elaborate and persistent (e.g., Aechmea fasciata; Fig. 3.2G), or much reduced and more ephemeral (e.g., Aechmea fulgens; Fig. 3.2B). Ancillary functions characterize the exceptional species, for example the red bracts that also secrete nectar to attract ant guards to Tillandsia balbisiana (see below). Floral bracts produced by Aechmea setigera bear a sharp terminal spine that seems not to impede either pollination or seed dispersal, but may deter large grazers (Fig. 3.2D). In¯ orescence bracts help identify tillandsioids dependent on certain kinds of pollinators (Chapter 6). Those of fundamentally ornithophilous Tillandsia subgenus Tillandsia both enclose the developing ¯ ower(s) and fruit(s) and provide the major visual signal for nectar-seekers (e.g., Tillandsia punctulata; Fig. 6.1B). The same appendages serve entomophilous and autogamous members of subgenus Anoplophytum and Diaphoranthema as the smaller, often less colorful organs needed primarily to insulate meristems and young ¯ oral buds (Fig. 3.3C,I). Aechmea fulgens (Fig. 3.2B) and various Pitcairnia (e.g., P. bakeri vs. P. flammea; Fig. 3.4K) exemplify the same arrangements in Bromelioideae and Pitcairnioideae respectively. Visual attractants other than anthocyanins brighten the in¯ orescence of some of the other bromeliads. Boat-shaped, foliaceous bracts covered with copious, re¯ ective wax, combined with fragrant nocturnal ¯ owers, probably lure moths to Tillandsia heterophylla. The same appendage born by numerous other Tillandsioideae (e.g., Vriesea cylindrica; Fig. 3.5D) dries out to a light brown before the associated ¯ owers open. Persistent drops of water indicate that V. hydrophora features functional hydrathodes on its ¯ oral bracts (Fig. 3.5A). Appendages subtending the ¯ owers of some Cryptanthus species resemble foliage, as do those of Tillandsia brachycaulos and T. capitata, although the latter color up long enough to help attract pollinators (Fig. 3.2H). Breeding system and other aspects of ¯ owers and the durability of bracts sometimes suggest greater importance for seed dispersal than for pollination. For example, the scape bracts of paniculate and autogamous Aechmea bracteata color up to a bright pink before anthesis and remain undiminished thereafter, apparently to act as fruit ¯ ags. Conversely, the colorful Cambridge Books Online © Cambridge University Press, 2009 Inflorescences 87 Figure 3.5. Flowers and fruits of Bromeliaceae. (A) Guttating ¯ oral bracts of Vriesea hydrophora. (B) Nidulate in¯ orescence of Neoregelia sp. with pollinator. (C) Sticky ¯ oral bracts of Vriesea bituminosa with captured insects. (D) Thief seeking nectar from base of corolla of Vriesea cylindrica. (E) Melaponid wasp visiting ¯ ower of chiropterophilous Vriesea atra at midmorning. (F) Pendant spike of Tillandsia dodsonii. (G) Ripe fruits of Billbergia porteana photographed against the foliage of a shrub at dusk with ¯ ash to illustrate re¯ ective trichomes. (H) Armed fruits of Aechmea angustifolia. Cambridge Books Online © Cambridge University Press, 2009 88 Reproductive structure primary bracts of usually ornithophilous Billbergia fade within days to pale pastels, sometimes even before the youngest ¯ owers open (Fig. 3.2F). Signi® cantly, the sometimes strong-smelling brown to yellow or trichomecovered fruits on pendant spikes attract bats (e.g., B. porteana; Fig. 3.5G). Even more curiously ephemeral, the pink and imbricated bracts of Pepinia fimbriatobracteata soon degrade to a glutinous blackish-brown residue reminiscent of a deliquescent sporocarp of the fungus Coprinus. Several related species (e.g., Pitcairnia arcuata; Fig. 3.4M) engage in less pronounced autolysis. Extra¯ oral nectaries help deter herbivores and perhaps distract nectar thieves for members of at least three genera in two subfamilies. Koptur (1992) photographed ants collecting nectar on the immature bracts of Tillandsia balbisiana, a likely bird-pollinated epiphyte in southern Florida. Galetto and Bernardello (1992) described secretions from glands located on the calyx of nine species of Dyckia and a Deuterocohnia native to northeastern Argentina (Figs. 3.4I, 8.2E). Regular ant-nest users (certain members of Aechmea, Neoregelia) also merit inclusion in this group if they, like so many of the nonbromeliads that share the same media, also produce ant food. Glands reported so far differ from those located in the ovary, conforming instead to the easily overlooked `formless type' (Fig. 3.4I). Orientation, timing, and rates of maturation further distinguish bromeliad in¯ orescences, and help promote relationships with speci® c kinds of pollinators and seed dispersers. Those of the epiphytes often hang below the shoot, while the same organs of the terrestrials almost always stand upright (Figs. 3.5F,G, 6.2A,B). Oddly deviant Pitcairinia corallina produces a bright red spike that, unexpectedly for its color and presumed pollinators, sprawls along the ground seemingly out of range of most nectar-feeding birds. Many months pass before some of the largest and usually monocarpic Tillandsioideae (e.g., Tillandsia grandis, Alcantarea regina) and Pitcairnioideae (e.g., Puya raimondii) with erect, multibranched in¯ orescences exhaust complements of hundreds to tens of thousands of ¯ owers. Certain one to few-¯ owered relatives (neotenic Tillandsia) must set fruit within a few days, but typically dense populations of somewhat asynchronous individuals or attached ramets assure more extended opportunities to set fruit (Chapter 6). Weeks to several months pass between the ® rst visible signs of bolting and the ® rst ¯ ower opening, and growth often accelerates as anthesis approaches. Downs (1974) reported that Billbergia elegans required nearly Cambridge Books Online © Cambridge University Press, 2009 Flowers 89 a month to emerge from the funnelform shoot, but just four more days to expand to ® nal size and present the ® rst ¯ ower. Scapes elongated at rates of up to 10 cm day21. At the other extreme, the short, capitate in¯ orescence of Neoregelia exhibits more conservative and precisely regulated growth, extending just enough to allow the tubular ¯ owers to emerge above the phytotelmata and assure the resulting inferior-ovaried fruits opportunity to develop submerged, perhaps to avoid predators (Figs. 3.2A, 3.5B). Fruits of some species elongate enough at the base to protrude several centimeters above the waterline at ripeness (Fig. 3.6F). In¯ orescences produced by species with phytotelm shoots pose interesting questions about morphogenesis. Speci® cally, does ethylene play a regulatory role similar to that responsible for the measured elongation of the shoots and leaves of certain emergent aquatics (e.g., Nymphaeaceae)? Would the in¯ orescences of the nidulate bromelioids still end up just long enough to position the ¯ owers just above the waterline if they developed while tanks were empty? Agents other than water may also impede gas exchange with different consequences. Copious mucilage, presumably secreted from the ¯ oral bracts or buds, insulates the fruits of Guzmania globosa (Fig. 3.3G) until capsules dehisce after which the plumose seeds require drier conditions to take ¯ ight. Lesser secretions characterize many additional Tillandsioideae, and these products sometimes cause the leaves and ¯ oral bracts to `quill' i.e., stick together (e.g., Vriesea glutinosa). Quite likely, growth factors synthesized in the developing in¯ orescence, or the absence of substances formerly produced by the vegetative apex, activate one or more of the axillary buds programmed to produce ramets. Anthocyanins that temporarily suffuse the shoots of many bird-pollinated types probably owe their synthesis to light and chemical signals originating in the embryonic in¯ orescence. Those scattered taxa (e.g., several Orthophytum, Tillandsia flexuosa, T. paucifolia; Fig. 2.11A) that produce offshoots from buds subtended by ¯ oral bracts raise additional questions about the involvement of hormones in bromeliad reproduction. Flowers Bromeliad ¯ owers range from relatively small to large and inconspicuous to showy. All of the species feature the differentiated perianth characteristic of related taxa like Commelinaceae and Zingiberaceae rather than the tepals of most Liliales. Zoophily prevails except for the occasional autogamous types that require no pollinators (e.g., neotenic Tillandsia; Fig. 2.1). Cambridge Books Online © Cambridge University Press, 2009 90 Reproductive structure Figure 3.6. Fruits and seeds of Bromeliaceae. (A) Fruit of Ananas bracteatus. (B) Germinating seeds removed from berry of Aechmea dactylina. (C) Armed fruits of Ronnbergia deleonii. (D) Fruit and appendaged seeds of Araeococcus micranthus. (E) Cryptanthus bromelioides showing thin-walled fruit and one seed. (F) Elongated ripe fruits of Neoregelia stolonifera extending above the phytotelmata. (G) Billbergia brasiliensis, fruit and seeds. (H) Seeds of Aechmea magdalenae and Aechmea bracteata (right to left). (I) Structure of testa of Billbergia elegans (above) and Glomeropitcairnia penduliflora (below). (J) Seed of Tillandsia germiniflora illustrating coma. (K) Seed of Tillandsia castellanii with multiple embryos. (L) Seeds of Aechmea angustifolia, A. bromeliifolia and A. kuntzeana from left to right. Cambridge Books Online © Cambridge University Press, 2009 Flowers 91 Tiny ¯ owers with exerted styles and anthers born by certain small-bodied navias might be wind pollinated (Fig. 3.4A). Several peculiarities mark the bromeliad ¯ ower, most notably the unpaired sepal that lies above the ¯ oral bract, and the twisting of all three calyx members to the left to cover the margins on the right and produce an involute whorl (Chapter 12). Derived conditions include the imbricated sepals of Brocchinia and the variously overlapped calyces of some Cottendorfia species. Foliaceous texture predominates, but succulent (e.g., Aechmea fulgens; Fig. 3.2B) to membranous types occur, especially the latter condition if a bract encloses the calyx (e.g., many Tillandsia subgenus Tillandsia; Fig. 3.3). Fleshy, colorful calyces augment the attractiveness of some Bromelioideae to frugivores, especially when fruit set is low (Fig. 3.2B). Sepals elsewhere in this subfamily probably assist seed dispersal in additional ways. Those of numerous aechmeas mature to sharp spines, perhaps to encourage frugivores to mash berries and thus release seeds ill equipped to pass through guts undamaged (Figs. 3.5H, 3.6C; Chapter 6). Identical service may be provided by the same three appendages combined to a hard, single sharp projection (Fig. 3.6B). Softer-textured versions of this arrangement grant foraging birds the grip necessary to remove ripe fruits from the ¯ at infructescences of certain nidulate bromelioids (Fig. 3.5B). Corollas are symmetrical except where one petal forms a hood over the stamens as in some Pitcairnia (Fig. 3.4K) and certain Tillandsioideae such as T. imperialis. Petal shape ranges from linear to ovate and the margins from entire through denticulate to ® mbriate (e.g., Tillandsia xiphioides; Fig. 3.3F). Tubular, stiff corollas help de® ne ornithophilous and sphringophilous taxa; those of Tillandsia albertiana seem almost waxy (Fig. 3.3L). Flared and recurved types mark the bee-serviced species (Fig. 3.3A,F,I). Yellow, maroon and scarlet occur less commonly than shades of lavender through white to pale green. Brightly pigmented bracts and foliage dominate the visual signal for many bird-pollinated Bromelioideae and Tillandsioideae. Similar arrangements highlight the ¯ owers of numerous Pitcairnia and Navia, among others. Stamens arise in two whorls of three members each (Fig. 6.1). Antipetalous ® laments commonly join the corolla base in gamopetalous species (e.g., Guzmania), and connation distinguishes additional genera (e.g., Bromelia, Dyckia). Stamen morphology has proven especially informative in relatively well-studied Tillandsia where the disposition of the ® lament, particularly its length and thickness relative to the corolla tube, identi® es some of the subgenera, although not always as faithfully as some Cambridge Books Online © Cambridge University Press, 2009 92 Reproductive structure authorities have presumed (Figs. 3.3, 6.1). Smith (1934a) used the transversely plicate (folded) ® lament (Fig. 6.1C) to circumscribe subgenus Anoplophytum, but Evans and Brown (1989a) disagreed in part because similarly modi® ed stamens characterize certain species assigned to congeneric Allardtia and Tillandsia. Folded stamen ® laments also occur elsewhere in Bromeliaceae (e.g., Nidularium ambiguum, Canistrum lindenii). A complex growth/senescence process recorded by Evans and Brown (1989a) causes plication in Tillandsia. Soft tissues collapse as the ® lament ages, allowing it to bend in response to the torsion effected by the now unconstrained (by turgor) xylem strands. Although the ® lament shortens as folding proceeds, compensatory growth occurs at its base. Hence, plication does not promote self-pollination by pulling the dehisced anthers down against the stigma. Perhaps the pleat-like foldings simply enlarge the tissue mass formed by the six ® laments enough to create a plug-like seal that retards evaporative concentration of deep-seated nectar. Thickened stamen ® laments that occlude the tubular corollas of certain members of Tillandsia subgenus Tillandsia native to dry Mexican habitats probably effect the same outcome (Gardner 1982; Chapter 6; Fig. 6.1A). Three carpels equipped with U-shaped placentas that fuse progressively upward during ontogeny characterize the gynoecia of all Bromeliaceae, but other details vary among taxa (Fig. 3.1; Chapter 12). The orthotropous ovules can be small and numerous (e.g., Fosterella) to few and larger (e.g., some Cryptanthus). Ovary position ranges from superior to inferior in Pitcairnioideae and Tillandsioideae, in the second subfamily contrary to Smith and Downs (1977) who report hypogeny throughout except for halfinferior Glomeropitcairnia. Bromelioideae usually produce fully epigenous ¯ owers that typically precede berries (Fig. 3.6). Pitcairnioideae illustrate greater variety, sometimes substantial variation among close relatives. Receptacle tissue surrounds all to just part of the ovary of many Pitcairnia, and the ¯ owers of Brocchinia range from about half to fully inferior. An epigynous tube obscures the position of the ovary in monotypic Ayensua. Typically hollow styles range from narrowly cylindric and straight to curved according to the conformation of the corolla (e.g., many pitcairnias) to short and stout (Fosterella; Fig. 3.4D). Bromeliaceae exhibit diverse breeding systems, some accompanied by conspicuously modi® ed ¯ oral structure. Imperfect ¯ owers occur in one or more species of Aechmea, Androlepis, Catopsis, Cottendorfia, Cryptanthus, Hechtia and Lindmania. Aechmea marie-reginae, Androlepis skinneri, more than half of Catopsis and all of the Hechtia species are dioecious (Fig. Cambridge Books Online © Cambridge University Press, 2009 Flowers 93 3.3H). Some members of Dyckia warrant closer scrutiny, as do less-studied Cottendorfia and Lindmania; Cryptanthus subgenus Cryptanthus is andromonoecious (Chapter 11). Catopsis species range from perfect-¯ owered to dioecious, and arrangements sometimes vary within the same taxon (Table 6.5). Catopsis pisiformis from Panama illustrates largely hermaphroditic structure that obscures its functional dioecism (Rauh 1983a). Pistillate plants bear outwardly normal anthers ® lled with nonviable pollen. Staminate ¯ owers were not described. Elsewhere, organ structure usually indicates functional competence (e.g., C. floribunda; Fig. 3.3H). Septal nectaries Massive septal nectaries may have predisposed Bromeliaceae to exceptional dependence on vertebrates rather than less demanding insect pollinators. Organ structure conforms to the `labyrinthine common nectarial cavity' type (Schmidt 1985), but details vary with other aspects of the ovary (e.g., Cecchi Fiordi and Palandri 1982; Böhme 1988; Chapter 12). Varadarajan and Brown (1988) examined numerous Pitcairnioideae chosen to represent a range of ¯ oral morphologies and pollination syndromes. They reported that three longitudinal systems of channels, one per septum, always join within the ovary axis (Fig. 3.1A). From there, nectar exits through circular or slit-like ori® ces at locations in¯ uenced by the hypogenous or epigenous condition of the ¯ ower. Triradiate cavities sometimes extend upward to the style along circuitous routes. Glandular tissue occupies different portions of the epithelium lining the collecting system. Superior-ovaried taxa (e.g., Deuterocohnia schreiteri, Dyckia ragonesei) possess three additional secondary nectar channels oriented toward the placenta, while half and fully inferior-ovaried gynoecia (e.g., Pitcairnia heterophylla, Puya harmsii) develop septal channels only. Simple sugars dominate the relatively dilute secretions (Table 6.3). Böhme (1988) examined over 90 species representing all three subfamilies in her attempt to identify features of potential systematic signi® cance. Gynoecial position and the manner in which the three carpels join proved less consistent than the literature claims. She also reported that the amount of nectariferous tissue varied, as did its con® guration and that of the collection system. Auxiliary secretory tissue sometimes extended beyond the gynoecium, occasionally on to the petal bases. Böhme constructed a scheme to depict nectary evolution, and attempted to reconcile gland structure with ¯ oral biology. Cambridge Books Online © Cambridge University Press, 2009 94 Reproductive structure Stigmas Much information useful for taxonomy and interpretations of reproductive function resides in the fresh stigma. Brown and Gilmartin (1984, 1989b) obtained the wet-preserved materials required for the necessary scanning electron micrographs through a network of collectors at botanical gardens and other sites scattered through the Neotropics. In all, they examined material for over 400 species and identi® ed ® ve architectures, viz. conduplicate-spiral, simple-erect, convolute-blade, cupulate and coralliform (Figs. 3.1C, 12.1; Chapter 12). Essentials of the conduplicate-spiral pattern emerge early during ¯ oral ontogeny, not long after the carpels appear. Subsequent growth causes the primordia of many species to change shape, and ultimately correspond to one of the four apparently derived arrangements (Brown and Gilmartin 1988). Speci® c stigma types occur discontinuously through the family (Table 3.1). Surveyed Bromelioideae and Pitcairnioideae mostly exhibited the putatively basic conduplicate-spiral morphology (Fig. 3.1C). Less consistency characterizes Tillandsioideae where every condition occurs in at least one genus. Catopsis, Tillandsia and Vriesea possess stigmas of two or three types, although patterns vary less within subgenera (e.g., all Phytarrhiza examined bore the coralliform type and every available representative of Diaphoranthema, the simple-erect form). Similar polymorphism characterizes Brocchinia and Fosterella within Pitcairnioideae. Stigma morphology paralleled some additional taxonomic boundaries, and challenged the validity of others. Flowers of Guzmania and Mezobromelia, which resemble each other except for the presence or absence respectively of petal scales on fused corollas, also possess distinct stigma types (Chapter 12). Cupulate stigmas, the least common form, appeared only among members of the thecophylloid alliance within Vriesea. Tillandsia subgenus Allardtia, long suspected to be polyphyletic, duly exhibited both conduplicate-spiral and simpleerect stigmas, whereas species (seven) of Gardner' s Group Five recognized within subgenus Tillandsia (Chapter 9; Fig. 6.1) alone demonstrated the simple-erect instead of the conduplicate-spiral form. Tillandsia linearis and T. xiphioides, already recognized as anomalous within subgenus Anoplophytum by their lack of plicate stamen ® laments, displayed a conduplicate-spiral rather than the simple-erect stigma of their supposed closest relatives. Compared with some other aspects of ¯ ower structure (e.g., placentation, pollen nucleation), the morphology of the bromeliad stigma corre- Cambridge Books Online © Cambridge University Press, 2009 95 Flowers Table 3.1. Summary of known distribution of stigma types in the three subfamilies of Bromeliaceae Stigma type Bromelioideae Conduplicate-spiral Simple-erect Convolute-blade Cupulate Coralliform No data Pitcairnioideae Conduplicate-spiral Simple-erect Convolute-blade Cupulate Coralliform No data Tillandsioideae Conduplicate-spiral Simple-erect Convolute-blade Cupulate Coralliform Distinctive type Occurrence 20 genera Cryptanthus, Orthophytum None None None 5 genera 11 genera (Brocchinia and Fosterella, in part) Brewcaria, Cottendorfia, Hechtia, Brocchinia, Fosterella (in part) None None None Connellia Catopsis, Mezobromelia, Tillandsia, Vriesea (all or in part) Catopsis, Guzmania, Tillandsia, Vriesea (in part) Guzmania, Vriesea (in part) Vriesea (in part) Tillandsia (subgenus Phytarrhiza only) Glomeropitcairnia Source: After Brown and Gilmartin (1989b). lates with aridity and reproductive mode. Stigmas belong to the wet type and accordingly, elevate demands for moisture as surface area increases, perhaps enough sometimes to compromise ® tness. Simple-erect stigmas present less of a liability (surface area), and, perhaps not coincidentally, occur primarily among xerophytic Tillandsia. Stigma type also tends to accompany certain other ¯ oral characteristics related to breeding system and the pollen carrier, for example dioecism (Catopsis, Hechtia) and andromonoecy (Cryptanthus), which are often associated with simple-erect form. The cupulate type (Fig. 3.1C) that occurs only among the thecophylloid vrieseas may somehow foster reliance on the bats and birds these largebodied bromeliads routinely attract to set fruit. Varadarajan and Brown (1988) concluded that the degree of compaction of the lobes and papillae of the stigma of Pitcairnioideae more reliably identi® es targeted pollinators than coarser structure. Highly condensed, Cambridge Books Online © Cambridge University Press, 2009 96 Reproductive structure conduplicately folded, spathulate stigma lobes usually signaled ornithophily (e.g., Pitcairnia corallina, P. meridensis), while those bearing semicompact lobes, with or without papillae (e.g., Ayensua uaipanensis, Puya aristeguietae), accompanied ¯ owers that regularly attract bats. Stigmas with ovate to lanceolate, somewhat loosely folded lobes lacking papillae overoccurred among species rendered entomophilous by small, pale, actinomorphic ¯ owers (e.g., Brocchinia steyermarkii, Deuterocohnia longipetala, Pitcairnia brevicalycina) born on lax panicles. Petal scales Smith and Downs (1974) considered the petal scale, also called a ligule, nectar scale or lateral fold, one of the premier indicators of generic-level relationship in Bromeliaceae (Fig. 3.1B). Appendaged corollas characterize about one-third of the family, including at least some members of 14 of the 27 genera comprising Bromelioideae, six more in Pitcairnioideae, and three in Tillandsioideae. However, recent ® ndings drawn from studies of ontogeny question its current broad application as a taxonomic marker, particularly in Tillandsioideae where presence or absence differentiates Vriesea from Tillandsia (Brown and Terry 1992). Even Smith and Downs granted overall similarity greater weight at least once by subsuming Tillandsia pabstiana under Vriesea drepanocarpa. Petal scales clearly lack diagnostic value elsewhere in the family, for example in Pitcairnia and Puya where several species (e.g., Pitcairnia pulverulenta, Puya hofstenii) include individuals with appendaged and naked corollas. The bromeliad petal scale assumes a variety of forms at maturity following a more uniform beginning. Development starts with the emergence of a pair of adaxial excrescences on the base of the expanding petal on either side of the antipetalous stamen ® lament (Fig. 3.1B). Primordia of Pitcairnioideae and some Tillandsioideae fuse into a single, variously bi® d or lobed, tongue-like ¯ ap with an entire margin. Six petal scales characterize the more elaborately appendaged ¯ owers of many Bromelioideae, one on each side of the three stamen ® laments opposed to the same number of corolla members. Final shape may be sac or pouch-like with distal lobations or fringes. Accessory lateral folds of undetermined homology embellish some scales. Conversely, those of typically ant-inhabited Aechmea bracteata remain much simpler, perhaps consistent with small, autogamous ¯ owers and the modest amounts of nectar they produce to encourage outcrossing (Fig. 3.2C). Brown and Terry (1992) discovered that petal scales emerge shortly Cambridge Books Online © Cambridge University Press, 2009 Flowers 97 before the bud expands preparatory to anthesis. Initiation sometimes coincides with microsporogenesis, but usually occurs later along with the postmeiotic, tetrad stage. Most signi® cantly, scales proved to be the last external multicellular structures to form during petal development. So timed, this organ can appear and disappear in the evolutionary sense without effecting fundamental change in corolla structure. As a `terminal ontogenetic character' , these delicate appendages probably represent recent, minor modi® cations of more deeply seated ¯ oral patterns. If so, features determined earlier during ontogeny should provide superior markers for genera and higher taxa. A hybrid between Billbergia nutans (appendaged) and a Cryptanthus (unappendaged) species possessed petals with scales, suggesting genetic dominance at a single locus. Brown and Terry (1992) concluded that petal scales represent synapomorphies in some parts of the family (e.g., possibly several subgenera in Aechmea, Neoregelia subgenus Hylaeaicum) and homoplasies within groups containing more divergent populations (e.g., Tillandsioideae, Pitcairnia, Puya). Most of the speculation about petal scale function has focused on intra¯ oral nectar management. Frequent ornithophily and the occurrence of septal nectaries at the bases of elongated ¯ owers support this contention. Brown and Terry (1992), Ueno (1989) and Böhme (1988) demonstrated that the ducts and pore systems occur in different locations to assure product delivery into capillary rather than noncapillary space located near the base of the corolla. Nectar consistently exits from points on the gynoecium (superior-ovaried species) or the ¯ oor of the hypanthium (inferiorovaried species) opposite the antipetalous stamens. Scales and other ¯ oral parts assume various shapes and juxtapositions to elevate columns of nectar within essentially tubular corollas and stabilize the nectar' s sugar concentration and viscosity by retarding evaporation, or they guide the mouth parts of pollinators. Varadarajan and Brown (1988) obtained information on two Pitcairnia species that bears on scale function. Speci® cally, they noted how this organ appears to diminish without reinforcing selection. Pitcairnia brevicalycina features yellow ¯ owers with or without petal appendages. When present, the scale is simple to match the accompanying modest nectar production. Pitcairnia heterophylla (Fig. 2.12A), on the other hand, produces scarlet, nectar-rich ¯ owers bearing larger scales equipped with elaborate distal modi® cations. Bee pollination seems to explain corolla structure and ¯ oral reward in the ® rst species; birds or large moths with comparably high caloric demands service the second population (Varadarajan and Brown 1988). Although Cambridge Books Online © Cambridge University Press, 2009 98 Reproductive structure petal scales assist nectar presentation in Pitcairnia heterophylla, they appear to be vestigial in Puya brevicalycina. However, nectar scale structure can be deceptive; even a pair of simple, vertical folds with no discernible glandular lining held nectar in the corollas of some Puya floccosa populations. Other modi® cations of the bromeliad ¯ ower, such as the coherent, swollen ® lament bases in some Dyckia, may help maintain reservoirs of nectar in the absence of scales. Fruits, ovules and seeds Fruit type and seed morphology differentiate Bromeliaceae into three subfamilies (sensu Smith and Downs 1974, 1977, 1979; Fig. 3.6), but not as de® nitively as some taxonomic descriptions imply. Dry capsules and naked or double-coated seeds with or without appendages characterize Pitcairnioideae. Seeds equipped with an elaborate ¯ ight apparatus born in capsules indicate Tillandsioideae (Figs. 3.3H, 3.6J; Chapter 12), while the berries produced by most Bromelioideae contain naked seeds equipped with or lacking soft, unbranched appendages (Figs. 3.5G,H, 3.6L). Exceptions include the fruits of Fascicularia, Ochagavia and Orthophytum (Bromelioideae), which tend toward dryness, and those of some Pepinia (Pitcairnioideae) that are just as unexpectedly ¯ eshy. Dehiscence varies among the capsular types, and at least one Ronnbergia species forcibly ejects its ripe seeds. Epigeny sometimes prevails where reports indicate hypogeny (e.g., many Tillandsioideae) and vice versa (e.g., some Pitcairnioideae). Bromeliad seeds range from medium to small by angiosperm standards (e.g., ,0.1 mg for some Pitcairnia), but none approach the proportions of the minute diaspores produced by the orchids and holoparasites. Those of terrestrial Bromelioideae exceed the sizes of the seeds of the related epiphytes if the pattern noted in similarly ¯ eshy-fruited Araceae and Cactaceae also prevails in this subfamily (Madison 1977). McWilliams (1974) determined that the seeds of Tillandsioideae generally weigh less than those of Pitcairnioideae. Dispersal modes probably vary more among the bromelioids than among members of the other two subfamilies (Chapter 6). Mass also varies more among Bromelioideae. Seeds of some taxa (e.g., Acanthostachys, certain Bromelia, and Cryptanthus) exceed all others in size, and accordingly, ripen in smaller numbers. A pliable, sticky appendage probably effects adhesion to substrates and perhaps also to dispersers (Fig. 3.6L). Embryos usually occupy about one-quarter to one-third of the seed Cambridge Books Online © Cambridge University Press, 2009 Fruits, ovules and seeds 99 Figure 3.7. Embryology of Tillandsia usneoides. Redrawn from Billings (1904). volume, with starchy endosperm (and some oil in certain taxa) making up the balance (Fig. 3.7). Development is helobial according to Davis (1966). Billings (1904) described embryology in Tillandsia usneoides as conventional for monocots, but polyembryony occurs in some close relatives (Figs. 3.6K, 3.7). Gross (1985) surveyed 11 species of Tillandsia subgenus Diaphoranthema and discovered one to four embryos in at least the occasional seed of all but T. recurvata. If more than one progeny was present, the largest of the group appeared to be zygotic and the others of undetermined origin and positioned lateral to it. More endosperm remained in seeds bearing one compared with multiple embryos. The outermost layer of the endosperm consists of starch-free, cubical cells containing darkly pigmented, granular materials. Szidat (1922) suggested its identity as an aleurone layer. If so, component proteins, like those of the cereals, probably promote germination by mobilizing food reserves for growing embryos. Thin-walled endosperm tissue deeper in the seed contains abundant starch, usually as lenticular grains. Elongated cotyledons equipped for absorption occupy the distal end of the seed where they remain, rendering germination hypogeal (Figs. 3.7, 3.8). Intercalary growth Cambridge Books Online © Cambridge University Press, 2009 100 Reproductive structure Figure 3.8. Germination. (A) Canistrum lindenii. (B) Pitcairnia flammea. (C) Vriesea scalaris. near the base of the hypocotyl pushes part of that organ and the adjacent radicle through the testa. Seedlings of Tillandsioideae fail to produce roots for weeks to months (Fig. 3.8C). The greatest delays characterize neotenic Tillandsia. Several features describe the ovules and seeds of Bromeliaceae, for example anatropous morphology, two layers of cells comprising each of the two parts of the integument, predominantly starch reserves, and a rel- Cambridge Books Online © Cambridge University Press, 2009 Fruits, ovules and seeds 101 Figure 3.9. Seed types and seed phylogeny in Pitcairnioideae. Redrawn from Varadarajan and Gilmartin (1988b). atively small embryo (Billings 1904; Fig. 3.7). Some endosperm always remains to nourish the young seedling. Mature seeds provide numerous potentially informative, but little-utilized, traits for taxonomy (Gross 1993). Seed morphology varies far more than most of the literature suggests. A closer look at the development of the outer seed coat seems advisable to evaluate several suggestive similarities, including possible homologies between the ¯ ight apparatus of Brocchinia tatei and Tillandsioideae, especially Glomeropitcairnia (Varadarajan and Gilmartin 1988b; Figs. 3.9, 6.1D; Chapter 12). Navia seeds exhibit an interesting parallel with those of Bromelioideae: both lose the outer integument during development, although conditions differ at maturity as described below (Fig. 3.9). The sticky strand of material (funiculus) that helps fasten the seeds of many Aechmea species (e.g., A. angustifolia; Fig. 3.6L) to substrates appears to be derived from the testa. Searches for ever ® ner structure for systematic and functional analysis Cambridge Books Online © Cambridge University Press, 2009 102 Reproductive structure continue. Several classical papers (e.g., Poisson 1877; Szidat 1922; Netolitzky 1926) report modi® cations of the outer testa that in¯ uence seed mobility. However, neither these studies nor the others published since exhaust the possibilities for major revelations about family history, and how aspects of dispersal and seedling establishment favor success on speci® c kinds of substrates. For example, Palací (1997) discovered that the coma of Catopsis (Fig. 3.3H) is not homologous with the ¯ ight apparatus of the other Tillandsioideae, further underscoring the isolation of this genus within its subfamily. Conceivably, Catopsis evolved capacity to disperse among aerial substrates independently. If so, epiphytism and lithophytism, although nearly universal through Tillandsioideae, could well be homoplasious. Bromelioideae Seeds of Bromelioideae reportedly lack appendages, and the outer integument simply degenerates to augment the gelatinous pulp that ® lls much of the often tough husk of the ripe berry (Smith and Downs 1974). Figure 3.6L illustrates some exceptions. Seeds in some cases possess unique qualities that seem likely to encourage transport or adhesion to substrates, perhaps in the second instance acting like the viscin threads of some mistletoes. Seeds of several ant-nest taxa promote myrmecochory with alluring chemicals (Chapters 6 and 8), and closer examination might reveal that some of the material attached to the seed constitutes tissue evolved to serve as ant food. Seeds of Bromelioideae (Ronnbergia) that disperse ballistically remain little studied. Self-fertile Ronnbergia petersii germinates within the pear-shaped, orange (mammal-dispersed?) fruits that contain about 100 seeds, each enveloped in a gelatinous coat. Aechmea dactylina behaves similarly (Fig. 3.6B). Authoritative sources (e.g., Smith and Downs 1979) describe Bromelioideae as baccate, which ® ts reality except for those few, relatively dry-fruited exceptions just mentioned. However, this designation conveys no information about the likely consequences of the diverse colors, sizes, textures and nutritional values of the berries most of these plants produce (Chapter 6; Table 6.7). Shape and seed number per fruit surely in¯ uence appeal and access to vectors and substrates. For example, Madison (1979) suggested that size and form may help some species employ pupal mimicry to support ant-garden status (e.g., Aechmea mertensii). One ¯ at side on the seeds of certain epiphytes (Aechmea bracteata, Billbergia elegans; Fig. 3.6H) may promote sufficient contact with bark to counter gravity. Cambridge Books Online © Cambridge University Press, 2009 Fruits, ovules and seeds 103 Aechmea magdalenae, a widespread terrestrial through Central America south to Ecuador, produces elliptical seeds that seem better suited to germinate on substrates other than bark (Fig. 3.6H). Thick-walled sclerids forming the outer seed coat provide the protection the embryos and endosperm of zoochorous ¯ ora require (e.g., Billbergia elegans, B. rosea; Fig. 3.6I). Inner integuments consist of two layers of heavily scleri® ed cells that vary enough among taxa to distinguish certain genera. Associated differences in hardness and resistance to corrosive gut secretions and grinding crops may reveal specializations for ingestion by mammals vs. avians. Tillandsioideae Seeds of Tillandsioideae feature a plumose coma or ¯ ight apparatus of varied construction and homology (Figs. 3.6J, 6.5D,G; Chapter 12). Development occurs in septacidal capsules of diverse proportions and anatomy depending in part on the position of the ovary. The testa accounts for the ¯ ight apparatus, but differently depending on the genus. Typically, the outer integument closely invests the seed until near maturity when it separates from the inner integument except at the base and becomes a three-layered series of long, hair-like extensions. Both outer layers form the umbrella-like portion of the coma and the inner layer, which remains attached to the seed proper, its handle (Fig. 3.6J). A much less elaborate apical plume contributes some additional buoyancy to most seeds. Catopsis exhibits a ¯ ight apparatus comprised of elongated cells (true hairs with hooked ends) that extend off the seed apex. The short capsule forces the developing coma to remain folded until dehiscence, causing its kinky morphology (Fig. 3.3H). The inner integument is similarly ® brous, but coherent at maturity and sufficiently unique to differentiate groups of species, perhaps what could be members of valid genera. Szidat (1922) and Röhweder (1956) described the tillandsioid coma in great detail, as did Gross (1988a) during her search for seed morphologies valuable for systematic analysis. Walls of the component cells bear potentially signi® cant pits and ornamentations. Hairs of some Guzmania and Tillandsia species exhibit bifurcate cross-walls that may favor adhesion to speci® c kinds of surfaces, perhaps bark vs. rock to match epiphytic or saxicolous habits (Fig. 6.5D). Usually the ¯ ight apparatus extends off the base of the seed with only a short plume at the opposite end (Fig. 3.6J). Sometimes the apical extension consists of no more than a short, membranous hood. However, the apical Cambridge Books Online © Cambridge University Press, 2009 104 Reproductive structure plume in Tillandsia grows much longer and divides into multiple parts; Vriesea subgenus Alcantarea (genus Alcantarea according to Grant 1995a,b) illustrates the opposite condition. Here, relatively short appendages extend from both ends of the fusiform seed, perhaps consistent with strong selection against buoyancy as highly insular saxicoles in southeastern Brazil (Fig. 1.2C). Experiments conducted by McWilliams (1974) and Bennett (1991) indicated that the architecture of the ¯ ight apparatus of Tillandsia, its mass relative to that of the seed proper, and structural details affect seed mobility and possibly securement to speci® c kinds of substrates (Chapter 6). Pitcairnioideae Pitcairnioideae, like Tillandsioideae, produce capsules and dry seeds, but morphology diverges more in the ® rst subfamily, consistent with recognition of roughly twice the number of genera. Conversely, most Pitcairnioideae feature seeds equipped with entire-margined appendages, and no member is particularly well equipped for long-distance dispersal by the form of its propagules (Fig. 3.9). A few species reputedly rely on ants or water to disperse (Chapter 6). A bipartite testa develops except for Navia where only the inner layer remains at maturity. Here, as elsewhere through the subfamily, a scleri® ed inner integument protects the embryo and endosperm. The outer portion forms a hump or wing on the back and apex of the ovules of Puya and Pitcairnia, and sometimes extends over the apex on the ventral side nearly to the micropyle (Fig. 3.9). An almost circumferential wing distinguishes the diaspores of Pepinia, some Puya, and members of several other genera. Seeds of additional taxa (e.g., Cottendorfia, Brocchinia, Fosterella) bear appendages comprised of clusters of sharp-pointed ® bers extending from one or both ends of the seed (Figs. 3.9, 6.1D). Glomeropitcairnia penduliflora (Tillandsioideae) exhibits a ¯ ight apparatus up to 2 cm long, at least outwardly suggestive of affinities between the two capsular subfamilies. Again, seeds contain proportionally massive endosperms and small embryos. Germination exposes the basal part of the cotyledon, which soon becomes green and laminar (Fig. 3.8B). More structural variety exists among the seeds of Pitcairnioideae than among either of the other two subfamilies. Varadarajan and Gilmartin (1988b) divided Pitcairnioideae into six groups according to seed morphology, and hypothesized an ancestral, `unadorned' simpler form (Fig. 3.9). Some modi® cations of the integument correspond to individual genera (the Cambridge Books Online © Cambridge University Press, 2009 Pollen grains 105 `Brocchinia type' and `Navia type' ), while others, for example the `Fosterella type' , likely evolved repeatedly, in this case in Abromeitiella, Ayensua, Connellia, Deuterocohnia, Fosterella, Hechtia and Pitcairnia. Particularly interesting are the seeds of Brocchinia, especially the condition of the basal appendage. Only a simple or digitate projection occurs in the examined terrestrial forms (e.g., B. acuminata, B. reducta); its condition in often epiphytic B. tatei suggests the ¯ ight apparatus of Tillandsioideae (Figs. 3.9, 6.1D). Pollen grains Taxonomists account for virtually every publication on bromeliad pollen, and none of these reports suggest strong correlations between morphology and either related functions (e.g., type of pollinator, breeding system) or taxonomic boundaries (e.g., Halbritter 1988, 1992; Chapter 12). Androlepis and Hohenbergiopsis disperse pollen in tetrads while the other grains separate. Walls vary from smooth to reticulate or foveolate (Fig. 12.2). Three aperturate types occur: porate (exclusively Bromelioideae), inaperturate (various Bromelioideae and Tillandsioideae), and sulcate (Pitcairnioideae and Tillandsioideae), indicating much homoplasy in grain morphology, or insufficient resolution to recognize subgroups within larger, arti® cial categories. Pollen grain morphology remains largely untested for utility in tracing Bromeliaceae through the fossil record (Chapter 9). Cambridge Books Online © Cambridge University Press, 2009 Cambridge Books Online © Cambridge University Press, 2009 4 Carbon and water balance Temperate zone ¯ ora dominate the early literature on plant ecophysiology, but Bromeliaceae account for a disproportionate number of the tropical species studied during this period. Billings (1904) recorded moisture exchange during his inquiry on the biology of Spanish moss. Mez (1904) and several European contemporaries demonstrated how the foliar trichome of this same epiphyte and comparable Tillandsioideae eliminates need for absorptive roots. Harris (1918) contrasted osmotic pressures in the leaves of trees and associated bromeliads and co-occurring vascular epiphytes in Florida and Jamaica. Wherry and Capen (1928) surveyed Tillandsia usneoides growing along Florida highways for its capacity to accumulate nutrients and certain technological metals. Research on pineapple metabolism began in earnest during the late 1930s. Finally, Leopoldo M. Coutinho included numerous Bromelioideae and Tillandsioideae in his pioneering investigations on the pathways responsible for CO2 assimilation by diverse Brazilian ¯ ora (e.g., Coutinho 1963). Since then, a growing number of scientists have been measuring gas exchange, chlorophyll ¯ uorescence and other indicators of plant performance and physiological state to expand the database on Bromeliaceae. Current records of carbon ® xation pathways, many accompanied by data on water balance and light relations, document the ecostrategies of more bromeliads than of members of any other family (Martin 1994; Table 4.1). Moreover, concerns about issues ranging from global change to drought-tolerance assure continuing interest in the ecology and evolution of Bromeliaceae. Ananas comosus, Aechmea magdalenae, Bromelia humilis, Tillandsia usneoides and Guzmania monostachia constitute the best-known species; less complete pro® les of many more taxa further attest to the exceptional variety of growing conditions these plants experience in situ. 107 Cambridge Books Online © Cambridge University Press, 2009 Table 4.1. Details of photosynthesis for representative bromeliads Taxon Photosynthetic pathway D value A Light Light compensation saturation Apparent intensity intensity quantum (mmol m22 s21) (mmol m22 s21) yield Bromelioideae Aechmea magdalenae Aechmea magdalenae Ananas comosus Ananas comosus (high light-grown) Ananas comosus (shade-grown) Ananas comosus Bromelia humilis Bromelia humilis (shade-grown) Bromelia humilis (high light-grown) Bromelia humilis Cryptanthus bromelioides Neoglaziovia variegata Nidularium innocentii Wittrockia campos-portoi Ð CAM CAM Ð C3± CAM? Ð Ð 2.0 mmol m22 s21 216.2 Ð 213.4 Ð Ð 2.2 µmol g21 dry weight s21 223.1 Ð Pitcairnioideae Pitcairnia integrifolia Pitcairnia integrifolia Pitcairnia flammea Puya floccosa Puya copiapina Puya ferruginea Ð Ð C3 C3± CAM? CAM C3 Ð Ð 227.9 222.5 215.4 224.8 Tillandsioideae Catopsis nutans Catopsis nutans Guzmania monostachia Guzmania monostachia Guzmania lingulata Guzmania lingulata Tillandsia usneoides Tillandsia usneoides Tillandsia usneoides Vriesea simplex C3± CAM? Ð C3± CAM Ð Ð _ CAM Ð Ð C3 223.7 Ð Ð Ð 223.7 Ð Ð 0.69 µmol CO2 m22 s21 Ð Ð Ð 1.6 µmol CO2 m22 s21 213.7 Ð Ð Ð Ð 4.8 µmol g21 dry weight s21 224.4 Ð CAM Ð CAM 215.5 Ð Ð 0.3 mmol CO2 m22 s21 213.5216.5 Ð Ð Ð Ð Ð CAM Ð 212.1 Ð Ð Ð Ð Ð Ð Ð Ð 50 Ð 2.2 mmol m22 s21 Ð 26 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Grif® ths and Smith 1983 P® tsch and Smith 1988 Medina et al. 1991a 0.07 Borland and Grif® ths 1989 0.03 Ð Ð Borland and Grif® ths 1989 Borland and Grif® ths 1989 Grif® ths and Smith 1983 10 1000 0.09 Fetene et al. 1990 1000 Ð Ð Ð Ð Ð 0.08 Ð Ð Ð Ð Ð Fetene et al. 1990 Medina et al. 1991b Medina et al. 1991b Medina et al. 1991b McWilliams 1970 Medina et al. 1977 13 300 Ð Ð Ð Ð Ð 0.03 Ð Ð Ð Ð Ð Lüttge et al. 1986a Lüttge et al. 1986a Medina et al. 1977 Medina and Troughton 1974 Medina et al. 1977 Medina et al. 1977 Ð 50 Ð 1000 Ð Ð 600 Ð 250 250 Ð Ð Ð Medina and Troughton 1974 Benzing and Renfrow 1971b Medina and Troughton 1974 Maxwell et al. 1995 Grif® ths et al. 1986 Grif® ths et al. 1986 Medina and Troughton 1974 Martin et al. 1986 Martin et al. 1986 Grif® ths and Smith 1983 Ð Ð Ð Ð Ð 33 Ð Ð Ð Ð Ð Ð Ð 5 µmol CO2 m22 s21 Ð Ð Ð Ð Ð Ð Reference Ð Ð Ð 20 Ð 50 80 Ð Ð Cambridge Books Online © Cambridge University Press, 2009 Ð Ð Ð 0.01 Ð Ð Ð Ð Ð 110 Carbon and water balance On-going inquiry promises to produce additional revelations, perhaps even unrecognized stress-moderating mechanisms, in one or more of the drygrowing species. This chapter considers the data, critiques its interpretations, and identi® es promising avenues for additional research. Ecophysiological peculiarities Capacity to tolerate often hostile substrates and harsh climates in diverse tropical and subtropical American habitats helps distinguish many Bromeliaceae from most of the other ¯ owering plants. Additional ¯ ora also accommodate considerable aridity and impoverished rooting media, but few of these plants draw nutritive ions and moisture from the same unconventional sources. More than half of the bromeliads obtain moisture directly from the atmosphere, or from aquatic impoundments located among overlapping leaf bases rather than soil (Fig. 5.1). And no family comes close to matching Bromeliaceae for the variety of media and the mechanisms and devices members utilize for mineral nutrition. Occasional bromeliads grow in permanent seepages or marshes, or they root in seasonally saturated soils (e.g., some Brazilian Pitcairnia flammea and a number of Bromelioideae). The more exceptional population tolerates periodic submergence in fast-moving streams (Fig. 1.4G). Climatically arid habitats or dry microsites at wetter locations characterize much more of the family. Xerophytes abound and include the over-represented epiphytes and lithophytes, many of which rival the other exceptionally stress-tolerant terrestrial ¯ ora for capacity to survive protracted dry weather solely on moisture stored in succulent foliage (Figs. 2.2A,B, 2.13B). High-altitude Bromeliaceae endure UV-B-enriched insolation and freezing nights assisted by woolly, light-scattering indumenta and dense accumulations of anthocyanins (Figs. 7.2± 7.4). Different combinations of shoot architecture and pigment display enhance light harvest deep in the forest understory in the manner described below. Mineral nutrition remains the least-studied facet of bromeliad ecophysiology. Chapter 5 addresses this subject by considering sources and the plant devices employed to absorb required ions. Additional aspects of nutrition that in¯ uence carbon budgets and water balance following uptake have also received some attention. Complex interactions among plant nitrogen and moisture status and photosynthetic photon ¯ ux density (PPFD) characterized Ananas and Bromelia species monitored in situ and in the laboratory, but impacts on ® tness remain unclear, as we shall see. If exceptionally ¯ exible ecophysiology turns out to be a hallmark of Cambridge Books Online © Cambridge University Press, 2009 The five ecophysiological types 111 Bromeliaceae, it should be no surprise given the family' s frequent dependence on supplies of moisture and nutrients that tend to be scarce, transitory, and unavailable to most other ¯ ora. The literature contains so much information on the carbon and water relations of Bromeliaceae that an exhaustive coverage would likely deter all but the most dedicated reader. Therefore, I have segregated the family into ® ve ecological/functional (ecophysiological) types to facilitate comparisons. Table 4.2 lists the de® ning characteristics of each type, and cites familiar, representative species. Pittendrigh (1948), impressed by observations recorded by Schimper (1884, 1888, 1898), Tietze (1906) and others, re® ned their four-parted scheme based on sources of moisture and nutrients (soil vs. atmosphere) and the plant devices (foliar trichomes vs. roots) used to access them to organize his ® ndings on Bromeliaceae in Trinidad. I follow this arrangement except where important differences warrant dividing Type Three in the old system into two new ones. The five ecophysiological types Pittendrigh' s Type One, the `soil root' group, matches mine (Table 4.2). Members root exclusively in media characterized by plentiful supplies of water (e.g., moist mineral soils, rocks with deep ® ssures) at least through a wet season. Degrees of xeromorphy and photosynthetic pathway vary, obliging requirements for more or less continuous (e.g., mesophytic Pitcairnia and Fosterella; Fig. 2.16B,C) to intermittent (e.g., dry-growing Hechtia and Dyckia; Fig. 1.2A) access to soil moisture. Plant architecture, faithful to the division of labor between shoot and root systems, follows the fundamental monocot pattern, which probably also represents the basic condition for Bromeliaceae as a whole (Fig. 2.20). Foliage routinely performs little if any of the absorptive function that allows the more specialized bromeliads to dispense with roots except for holdfast. Two subfamilies, Bromelioideae and Pitcairnioideae, contribute species to this ® rst and most fundamental of the ® ve ecological types. Type Two corresponds to Pittendrigh' s second, `tank root' designation, and, like his examples, species included here possess somewhat succulent foliage with modestly dilated bases (Fig. 2.14A,B). Enough moisture and nutrient-rich debris collects in the upright shoot to meet plant needs via apogeotropic roots and relatively unspecialized, absorptive trichomes. Roots of cultivated pineapple, which may access soil moisture more effectively than those of some of the other Type Two bromeliads, supplied less water to transpiring shoots than those serving an adjacent stand of Cambridge Books Online © Cambridge University Press, 2009 Table 4.2. The five ecological types: basic characteristics and occurrence in Bromeliaceae Root system Shoot architecture Foliar trichomes Photosynthetic syndrome Habit Type I Absorptive soil roots No phytotelma Nonabsorptive C3 or CAM Terrestrial Most Pitcairnioideae/ many Bromelioideae Type II Absorptive soil and apogeotropic roots Weakly developed phytotelma Absorptive on leaf bases CAM Terrestrial Bromelioideae Type III Mechanical to conditionally absorptive Well-developed phytotelma Absorptive on leaf bases Mostly CAM Terrestrial/ saxicolous/ epiphytic Bromelioideae Type IV Mechanical to conditionally absorptive Well-developed phytotelma Absorptive on leaf bases Mostly C3 Mostly epiphytic Tillandsioideae and a few Brocchinia species Type V Mechanical or missing No phytotelma, often neotenic and miniaturized Absorptive over entire shoot CAM Mostly saxicolous or epiphytic Tillandsioideae Cambridge Books Online © Cambridge University Press, 2009 Taxonomic distribution The five ecophysiological types 113 Bermuda grass in one set of ® eld tests (Ekern 1965). Except perhaps for a few species of Brocchinia (e.g., B. vestita; Fig. 5.3D), Bromelioideae ± most notably certain members of Ananas and Bromelia ± constitute Type Two, and soil and rocks rather than bark usually sustain these plants. My Types Three and Four represent taxon-distinct segregates of Pittendrigh' s Type Three, or what he called the `tank absorbing trichome' group. Members of both new categories produce sizable phytotelma from which the bases of mature foliage and perhaps the entire surfaces of the wholly submerged, younger leaves draw moisture and nutrient ions; roots provide anchorage and conditionally (contingent on the quality of the medium) augment mineral nutrition and water balance (Figs. 1.2E,G, 2.4). Simultaneous feeding through roots and shoots promoted superior growth among some cultivated bromeliads compared with either route alone (e.g., Sieber 1955), but shallow to impenetrable media often oblige nearly complete reliance on phytotelmata in situ. Assignment to Type Three (all Bromelioideae) or Type Four (impounding Tillandsioideae and some Brocchinia) also distinguishes the `phytotelm' or `tank' bromeliads according to the prevailing photosynthetic mechanisms (Type Three, predominantly CAM; Type Four, mostly C3), and whether the foliar trichomes possess low or higher absorption capacity (Type Three, low; Type Four, higher). Pittendrigh' s fourth category, the `atmospheric, absorptive trichome' bromeliads, equals my ® fth type (Figs. 1.3A,C, 2.1). Members lack capacity to impound water or solids (except ant-provisioned materials in the case of the myrmecophytes) in leaf bases, and roots, if present, lack signi® cant absorptive capacity. Instead, these bromeliads depend on dense indumenta of air-exposed foliar scales (Figs. 2.7, 2.8C,E, 4.23E± H). Type Five species represent the ultimate response among Bromeliaceae to the challenges imposed by multiple, physical stresses. However, similar leaf specializations, particularly pronounced succulence, belie the needs of some of these `atmospherics' for frequent irrigations. Many a Type Five bromeliad, for example Spanish moss, holds moisture less tenaciously than certain relatives with nonabsorptive indumenta (e.g., many Dyckia and Hechtia species). Transitional forms abound, especially between Types Two and Three and Types Four and Five. The usefulness of this ® ve-parted classi® cation, essentially appreciation of the ecophysiological variety it organizes, requires some preparatory discussion of photosynthesis and related aspects of water balance. Cambridge Books Online © Cambridge University Press, 2009 114 Carbon and water balance Photosynthesis and water economy Bromeliads face the same dilemma experienced by virtually all land-based ¯ ora: they must obtain CO2 without losing too much moisture. Simply put, transpiration (E) always accompanies photoassimilation (A) unless conducted in a water-saturated atmosphere, a rare event even deep within the everwet tropical forest. Vegetation with more or less continuous access to moisture usually operates with relatively poor water-use efficiency (WUE5 water expended divided by CO2 ® xed). The xerophytes, a different subset of ¯ ora united solely by capacity to counter drought, operate differently. Unlike their less water-constrained, hence more mesic, counterparts, the xerophytes either reduce the ratio of moisture transpired to dry matter accrued (the drought-endurers), or foliage is shed with timing that precludes losses sufficient to cause serious injury (the drought-avoiders). At greatest potential risk of life-threatening desiccation among Bromeliaceae are the members of Type Five (all evergreen, and therefore drought-endurers) that experience frequent high evaporative demand while rooting on naked bark or rock or supported by arid-land soils. More than the rest of the family, these plants must obtain CO2 from air that is often characterized by high vapor pressure de® cits (VPD), and they depend on modest, often temporary, sources to rehydrate. Xerophytic Bromeliaceae, like similarly adapted vegetation belonging to about 25 other families, greatly improve WUE through deployment of a complex mechanism or `syndrome' called crassulacean acid metabolism (CAM). CAM fosters nonautotrophic (dark) CO2 uptake from what tends to be relatively water-saturated night, compared with daytime, air. Severe climates and substrates incapable of storing moisture explain the frequent occurrences of absorptive trichomes, succulence, impounding shoots and CAM among members of Bromeliaceae. Of the approximately 225 species for which records exist (Martin 1994), about two-thirds exhibit some version of CAM. In fact, bromeliads demonstrate both the versatility of this syndrome as an ecophysiological response to multiple physical stresses and the ease of its derivation from more fundamental C3-type photosynthesis. Reversals, perhaps encouraged by historic shifts in climate or the geographic ranges of ¯ ora, mark certain Bromelioideae and perhaps additional lineages in the other two subfamilies (Chapter 9). So far, no compelling evidence indicates C4 photosynthesis anywhere in the family, although some suggestive leaf anatomy makes a case for looking more closely at certain Tillandsioideae. Table 4.1 provides a sample of Bromeliaceae selected to illustrate the Cambridge Books Online © Cambridge University Press, 2009 Crassulacean acid metabolism: basic characteristics 115 family' s mixed ecophysiological composition and the close phylogenetic juxtapositions of C3 and CAM populations in Pitcairnioideae and Tillandsioideae. Bromelioideae appear to be most fundamentally disposed to CAM, with largely montane, wet-growing Greigia (26 species) comprising the largest clade among the exceptions. Other C3 species scatter through small and larger genera (e.g., Nidularium burchellii, N. innocentii), probably also as evolutionary retrogrades. Likely additions (e.g., mesomorphic Ronnbergia) would not alter this tally much considering the documented, near to complete CAM status of all the largest genera (Aechmea, Billbergia, Bromelia, Neoregelia) and most of the medium-sized ones (e.g., Cryptanthus, Quesnelia, Orthophytum; Martin 1994). Ancestors probably operated in the C3 mode prior to evolving the current array of variations on CAM present in descendants, but when, where and how often remain unclear (Chapter 9). Crassulacean acid metabolism: basic characteristics Crassulacean acid metabolism has attracted extraordinary attention because of its pervasiveness and importance in many kinds of habitats. Occurrences in about two dozen families of ¯ owering plants, Gnetophyta, several fern genera, and Isoetes of Microphyllophyta indicate ancient origins and broad utility, or at least compatibility, with other plant characteristics and diverse growing conditions. Performance varies among CAMequipped ¯ ora, and sometimes shifts within the same individual and even the same organ during ontogeny. Lüttge et al. (1986b) recorded differences in several measures of CAM among leaves born by the individual shoots of Aechmea aquilega specimens in Trinidad, and even from one part to another of the same blade! Variation also tracks the seasons and certain less predictable environmental events, particularly those that affect plant moisture status. Little beyond reliance on phosphoenolpyruvate carboxylase (PEPc) to ® x CO2 at night during at least part of the life cycle unites thousands of species under the label CAM-equipped ¯ ora. Included are the obligate (constitutive) types, the `switchers' or facultative CAM plants, the CAMcyclers, and the additional, less-studied forms with even more puzzling patterns of carbon management, gas exchange and related leaf anatomy (e.g., Peperomia; Ting et al. 1985). CAM-cyclers behave like C3 plants by taking up CO2 during the day, but also re® x (recycle) carbon respired at night to malic acid as described below for CAM. All CAM types probably CAMidle, which means that whenever drought reduces stomatal conductance (g) Cambridge Books Online © Cambridge University Press, 2009 116 Carbon and water balance to zero, respired CO2 alone fuels acidi® cation, and during the day allows enough photosynthesis for plant maintenance. CAM plants adjust WUE and CO2 exchange in response to a variety of external cues and metabolic states depending on the genotype and previous and immediate growing conditions. Speci® c determinants include photoperiod, thermoperiod, plant nitrogen status, bulk tissue water potential (Cleaf), salinity, VPD in adjacent air, and light intensity. Enhanced water and carbon economy impart bene® t in arid locations, but other advantages can accrue where challenges differ. Martin' s (1994) demonstration that elevated CO2 (to 430 ppm) doubled peak tissue acidity (H1max) in Tillandsia ionantha supports Knauft and Arditti' s (1969) suggestion that CAM may appreciably improve carbon budgets where abundant respiration (e.g., forest understory) elevates ambient partial pressures at night. Some aquatic macrophytes, including certain Isoetes, deploy CAM in the same way to increase access to dissolved CO2 in soft-water (low carbonate) lakes. Enhanced N economy and perhaps the same for other scarce nutrients and more effective harvest of the radiant energy in sun ¯ ecks probably promote ® tness for bromeliads in a variety of habitats. CAM also helps protect Guzmania monostachia against photodamage (Maxwell et al. 1992, 1994, 1995). CAM-equipped land ¯ ora experience heightened water economy in part because PEPc compared with ribulose bisphosphate carboxylase/oxygenase (RuBPc/o), its functional equivalent in the photosynthetic carbon reductive (PCRC or C3) pathway, exhibits the higher affinity for CO2. This same quality is central to its role as mediator of CAM as the CO2cocentrating enzyme. Most importantly for dry-growing Bromeliaceae, PEPc permits plants to accumulate (concentrate) a carbon supply at night for photosynthesis later. If employed in lieu of some of the RuBPc/o, invested by an otherwise comparable C3 plant, the lower molecular weight of PEPc also improves nitrogen-use efficiency. Yet despite knowledge of these multiple bene® ts and the additional ® ndings on the mechanisms discussed below, questions persist about why CAM plants occur under such a variety of growing conditions. Different CAM plants and the same individuals under different conditions vary in their proportional dependencies on PEPc and RuBPc/o to harvest CO2 from air. Facultative compared with constitutive types track changing environments, alternating between C3 and CAM photosynthesis as the season and other plant and more site-speci® c circumstances change. Additional features that distinguish the different expressions of CAM, some with no documented bene® ts, include the nature of the carbon/energy Cambridge Books Online © Cambridge University Press, 2009 Bromeliad CAM: basic characteristics 117 reserves utilized to generate the PEP, reducing power and adenosine triphosphate (ATP) necessary to trap CO2 from night air, the growing conditions and plant status that promote CAM-idling, and the relative quantities of citrate to malate accumulated during dark CO2 ® xation and metabolized during the day. Bromeliad CAM: basic characteristics Bromeliad CAM conforms to the pattern ® rst recorded for certain Crassulaceae. Although succulence is usually less developed than for much of the other dry-growing ¯ ora with the same photosynthetic pathway (e.g., Agavaceae, Cactaceae), both carboxylases operate in green mesophyll cells. Additionally, the two proteins function mostly at different times of the day, which minimizes futile CO2 recycling. Here too (Table 4.1), 13C in biomass compared with its presence (relative to 12C) in the atmosphere expressed in parts per thousand (½ , or D), indicates the photosynthetic syndrome. Both C4 and CAM plants discriminate less against the heavier carbon isotope 13C (as 13CO2) than do the C3 types. The D values for biomass produced by CAM and C4 ¯ ora range between 28 and 222½ vs. between approximately 223 and 235½ for subjects that depend exclusively on the PCRC pathway (Fig. 4.1). Values at midrange, if not in¯ uenced by a biased source (e.g., decomposing C3 biomass), signal substantial ® xation of CO2 from the atmosphere by both carboxylases (facultative CAM). Certain other phenomena that affect the diffusion of CO2 through the leaf to the chloroplasts also in¯ uence D. Unlike C3 and C4 types which assimilate CO2 exclusively by day, ¯ ora operating in the CAM mode, as exempli® ed by Tillandsia usneoides (Fig. 4.2), do so predominantly at night. Carbohydrates, either soluble sugars or starch and glucans, provide the necessary energy, reductant, and carbon skeletons. Beta carboxylation of PEP and subsequent reduction of the resulting oxaloacetate to malic acid constitute the CO2 capture and storage processes, or what is known as phase one (Fig. 4.2). A relatively brief burst of CO2 uptake around sunrise (phase two) before the stomata close involves ® xation by both carboxylases, primarily PEPc at ® rst with gradual replacement by RuBPc/o. As the day progresses, CO2 generated by the decarboxylation of the malic acid mobilized from the vacuole reaches concentrations several times ambient (the CO2-concentrating mechanism), and reprocessing via the PCRC pathway continues (phase three). Elevated concentrations of CO2 suppress photorespiration and inhibit the accumulation of photo-oxidative Cambridge Books Online © Cambridge University Press, 2009 118 Carbon and water balance Figure 4.1. Distribution of 13C enrichment values (­13C or D) among species representing the three subfamilies of Bromeliaceae (after Medina 1990). Figure 4.2. CO2 exchange by Tillandsia usneoides with the four phases of CAM indicated (modi® ed from Martin and Siedow 1981). Cambridge Books Online © Cambridge University Press, 2009 Bromeliad CAM: basic characteristics 119 reactants that could impair the light-harvesting apparatus in overexposed foliage (Maxwell et al. 1992, 1994, 1995). Well after midday, with the malic acid supply exhausted and respiration now the sole source of CO2 for photosynthesis, stomata reopen allowing PEPc and RuBPc/o access once again to CO2 in the atmosphere (phase four; Fig. 4.2). In one study (Cote et al. 1989) about 50% of the CO2 consumed by wellwatered Ananas comosus from about the middle of phase four to dusk accumulated as malate. Reinert et al. (1995) demonstrated simultaneous involvement of PEPc and RuBPc/o by measuring the isotopic composition of CO2 in air passing over the shoots of Neoregelia cruenta in a Brazilian restinga (Fig. 7.13C± E). Each of the three light environments tested shifted the relative intensities of the four phases, i.e., the proportional involvements of the two carboxylases. Griffiths et al. (1990) and Griffiths (1992) also used on-line mass spectrometry to monitor instantaneous carbonisotope discrimination as Tillandsia utriculata performed CAM (biomass D 517.4½ ). Recycled CO2 accounted for 72% of DH1 (diurnal change in titratable acidity). Both PEPc and RuBPc/o operated during phase two and four, accounting for 4.0 and 22.5% respectively of the carbon gained over the 24-h CAM cycle. Because phase three operates behind closed stomata during the hottest, driest part of the day, CAM plants usually expend less H2O per unit of biomass manufactured than is possible for C3 and C4 plants facing equivalent evaporative demand. Instantaneous transpiration ratios (H 2O/CO2) can drop to about 10:1 (Tables 4.3, 4.4), leading to better performance than typical for C3 ¯ ora. Transpiration rates differ as much or more, but also with some overlap. Ekern (1965) reported that Ananas comosus lost 0.3± 0.5 mg H2O cm22 day21, whereas comparable values for corn (a C4 plant) were 26 and ruderal Xanthium (C3), 43. However, the bene® ts of gaining CO2 primarily after dark exact a cost: like many other CAM types, pineapple grew by far the slowest of these three subjects. The identity of the energy reserves expended to drive phase one and accordingly, where mobilization begins, whether in the cytoplasm or plastids, distinguish at least one bromeliad from many other CAM plants. Ananas comosus and some unrelated taxa (e.g., Aloe) exemplify the more recently described type. About 20± 170 times the reserves consumed over the same period by dark respiration were expended to build up the pool of carbon needed to sustain net photosynthesis during the following photoperiod (Carnal and Black 1989). Starch accounts for much of the CO2 stored as malic acid by most of the nonbromeliads studied, the observed reservoirs of hexose being inadequate. Glucose and fructose fractions changed Cambridge Books Online © Cambridge University Press, 2009 120 Carbon and water balance Table 4.3. Gas exchange characteristics of C3 and CAM bromeliads in Trinidad Transpiration (mmol H2O m22) Net CO2 uptake (mmol m22) Transpiration ratio (H2O/CO2, w/w) C3 species through entire photoperiod Vriesea amazonica Vriesea splitgerberi Vriesea jonghei Tillandsia fendleri 24.2 24.2 31.3 57.6 1.82 10.96 10.33 Ð 31 185 135 46 CAM species during phase four Aechmea lingulata Aechmea aquilega Aechmea nudicaulis 13.3 3.7 4.6 3.33 0.27 0.84 102 30 75 Source: After Grif® ths et al. (1986) enough in pineapple foliage to account for DH1 assuming that pyrophosphate-dependent phosphofructokinase, not ATP-dependent phosphofructokinase, catalyzed the phosphorylation of fructose-6-phosphate. Glucan contributed modestly to acidi® cation in A. comosus. Foliar sucrose also diminished at night, but probably through export rather than metabolism. Tillandsia usneoides monitored in a North Carolina forest illustrates how one CAM bromeliad behaves under a strongly seasonal climate. Subjects gained dry weight and consumed CO2 most vigorously from mid-spring through mid-fall (Martin et al. 1981). Acid ¯ uctuations and net CO2 uptake almost ceased in midwinter. Consistent with the situation in many other CAM types, carbon gain varied with temperature, DH1 peaking in late spring as daytime highs approximated 25± 30 °C. If night air was programmed in a growth chamber to remain at 20 °C, diurnal maxima up to 35 °C had no dampening effect on CO2 consumption; however, as the nights grew warmer, net CO2 uptake fell, beginning with the disappearance of phase four. If the daytime maximum reached 20 °C, phase one could be sustained to a nocturnal low of about 5 °C. Net ® xation ceased below 5 °C, or if day/night temperatures ¯ uctuated by less than 5 °C. Ecological correlates of the carbon fixation syndromes Textbook treatments often imply that vascular ¯ ora segregate into clearly de® ned CAM, C3 and C4 types. Likewise, these authors tend to assign Cambridge Books Online © Cambridge University Press, 2009 Table 4.4. Night-time gas exchange and related phenomena in CAM and C3–CAM bromeliads in Trinidad Species Aechmea aquilega Aechmea aquilega Aechmea nudicaulis Aechmea nudicaulis Aechmea fendleri Aechmea lingulata Tillandsia elongata Tillandsia utriculata Bromelia plumieri Guzmania monostachia Total CO2 uptake Recycled CO2 E Annual precipitation 18.00± 06.00 hours DH1 as % of total 18.00± 06.00 hours Transpiration ratio at study site (mm) (mmol m22) (mol m23) ® xed (mol H2O m22) (H2O/CO2, w/w) 1281 2625 1612 2625 2637 2625 1612 2366 1281 2366 6.5 36.3 46.4 54.8 45.0 47.6 15.9 3.9 0.5 0.6 113 393 301 469 332 309 271 251 72 70 89 83 71 78 56 65 82 95 99 94 Source: After Grif® ths et al. (1986). Books Online © Cambridge University Press, 2009 0.78 0.85 3.08 1.02 4.24 5.86 2.80 2.28 0.92 2.27 49 10 27 8 38 50 72 239 753 422 122 Carbon and water balance speci® c kinds of plant performance and native habitats according to the same, oversimpli® ed paradigm. Bromeliaceae challenge the second generalization except on two counts. First, relative compatibility between C3 metabolism and cool to cold growing conditions probably does explain the near absence of CAM in Bromeliaceae (mostly Puya) above about 3000 m. Second, C4 species appear to be absent, perhaps in part because none of the bromeliads possess certain other characteristics (e.g., short life cycle) usually associated with this photosynthetic syndrome. Reasons why CAM imparts advantage to so many bromeliads in such diverse kinds of habitats, and why one situation favors one compared with another version (e.g., obligate vs. facultative) of this syndrome, remain elusive. For example, many months pass without opportunity to eliminate moisture de® cits for those species native to hyperseasonal sites, while relatives with similar D values sometimes grow where drought may be more frequent but less severe. Leaf anatomy provides no greater insights on ecotolerance, as demonstrated by succulence that varies among CAM Tillandsia anchored on the same trees (e.g., T. paucifolia, dry weight524.5% fresh weight vs. T. balbisiana, same value536.0% in southern Florida). Higher ® ber content in the second species probably re¯ ects mechanical requirements obliged by exceptionally elongated blades. Hypodermal development fails to predict drought-tolerance for commonly sympatric T. paucifolia and T. recurvata (this tissue accounts for 38 and 3% of leaf volume respectively; Loeschen et al. 1993). Other plant characteristics complicate the issue further. Phytotelmata enhance the effectiveness of scarce rainfall for hundreds of epiphytic and lithophytic bromeliads, perhaps enough sometimes to render accompanying CAM more important for other purposes (Skillman and Winter 1997; Skillman et al. 1999). Environmental uncertainty may explain why CAM pervades so much of Bromeliaceae. CAM-equipped ¯ ora remain active during moderate drought, and, if necessary, can enter a quiescent state (CAM-idle) rather than deeper dormancy should stress exceed some threshold. Facultative types capable of shifting into the more productive C3 mode as circumstances allow (e.g., the ® rst rain at the end of a drought) are especially well positioned to thrive on episodic or unpredictable supplies of moisture. Guzmania monostachia demonstrates that plant readiness to exploit ecological opportunity during drought-enforced quiescence includes maintenance of the integrity of the photon-harvesting apparatus as described below. Then again, CAM may not always promote ® tness beyond that possible with an alternative photosynthetic syndrome, but instead represents a sustainable anachronism. Cambridge Books Online © Cambridge University Press, 2009 Ecophysiological profiles of the five types of Bromeliaceae 123 Conditions prevailing where many CAM plants grow, and the selfshading inherent to thick stems such as those of many cacti and the typically compact shoots of many other desert succulents, have promoted a third, erroneous impression that CAM-type ¯ ora routinely require substantially higher exposure than the C3 forms. Heliophily does accord with the heightened supply of ATP (6.5 vs. 3.0/molecule CO2) needed to operate via CAM compared with the PCRC pathway, but additional, undetermined costs for biomass preclude de® nitive comparisons (e.g., Raven 1985). Whatever the price of biomass in energy or any other potentially growthlimiting resource, many CAM bromeliads inhabit the forest understory, and several populations native to more open habitats (e.g., Bromelia humilis; Medina et al. 1986) exhibit greater productivity in partial than in full sun. Numerous studies, including some conducted in situ, indicate how Bromeliaceae, especially the CAM types, respond to speci® c combinations of temperature, nutrient supply, drought and exposure. Diverse methodologies and contradictory results cloud some of the interpretations, but even so the data demonstrate that designation as a CAM type says relatively little about other important aspects of plant biology (e.g., epiphytic vs. terrestrial habit, phytotelmata present or absent, shade-tolerant or intolerant). Our consideration of ecophysiological diversity among the bromeliads begins with a description of the qualities of members of Type Two, the best known of the ® ve categories, and one of the three that contains CAM species exclusively beyond the occasional Brocchinia (C3) that, by shoot form, also belongs here (Table 4.2). Ecophysiological profiles of the five types of Bromeliaceae Type Two species Reports on pineapple greatly outnumber those for any other bromeliad, and information on several ecologically signi® cant phenomena (e.g., effects of photoperiod on CAM) comes exclusively from this cultigen. Unfortunately, domestication precludes uncritical extrapolation to wild types because certain aspects of ecophysiology, in addition to fruit qualities, represent engineered rather than naturally selected characteristics (e.g., Baker and Collins 1939). Manipulations initiated long ago in indigenous South American agroecosystems may account in part for tolerances for certain growing conditions among surviving genotypes, some of which persist untended by humans (Fig. 1.3E). However, the other half a dozen Cambridge Books Online © Cambridge University Press, 2009 124 Carbon and water balance or so Ananas species, closely related Pseudananas, and scattered additional Bromelioideae (e.g., several Bromelia species, Aechmea magdalenae; P® tsch and Smith 1988) share similar ecology, structure and physiology (Table 4.1; Fig. 2.14A,B). Forest understories support the occasional deeply shade-tolerant Type Two bromeliad (e.g., Aechmea magdalenae), while the more heliophilic forms (e.g., Neoglaziovia; Fig. 6.12A) inhabit savannas, open restingas, and of course, for the pineapple, extensive croplands. Opinions vary about basic propensities; some authorities considered Ananas comosus, Bromelia humilis and the rest of their kind fundamentally heliophilic (e.g., Pittendrigh 1948), while other investigators (e.g., Medina et al. 1993) view B. humilis and Bromelioideae overall as shade-tolerant through derivation from forest-dwelling stock. Medina et al. (1991a,b, 1993) further concluded that Type Two bromeliads, and Ananas in particular, are good subjects in which to investigate how lineages descended from relatively light-sensitive antecedents became better performers in full sun. Most ® eld crops, although modi® ed architecturally (e.g., higher harvest index) and shortercycled, ® x carbon no more vigorously on a leaf area basis than their wild progenitors. Pineapple might be an instructive exception. Ananas comosus responded differently to light in separate experiments in part according to the cultivated vs. `wild' status of the subject and preconditioning relative to water, PPFD and N supply. Martin' s (1994) appendix summarizes the extensive data on the light relations and carbon and water balance of pineapple. Studies by Nose et al. (1977, 1981) and Sale and Neales (1980) exemplify a substantial part of that literature. Gas exchange by well-watered specimens at PPFD ranging from 200 to 1500 mmol m22 s21 indicated that photosynthesis light-saturated at 1000± 1500 mmol m22 s21, high enough to warrant designation as heliophiles. Different pretreatments and conditions during other runs produced other values. Nose et al. (1981) recorded only slightly increased CO2 uptake (net over 24 h) as potted subjects responded to exposures ranging from 600 to 1200 mmol m22 s21. Still another group of hydroponically grown plants reacted similarly to 200± 500 mmol m22 s21, but they showed no further increase even when PPFD was doubled. Borland and Griffiths (1989) reported that sufficient time under low or moderate exposure (60 or 600 mmol m22 s21) produced specimens with standard sun/shade characteristics. Low-light plants consumed CO2 less vigorously, and showed lower light compensation intensities and higher apparent quantum yields (Table 4.1). However, some feral subjects demonstrated more fundamentally shade-adapted photosynthetic responses. Cambridge Books Online © Cambridge University Press, 2009 Ecophysiological profiles of the five types of Bromeliaceae 125 Bromelia humilis also grows across shady to fully exposed microsites in coastal strand and inland habitats in northern Venezuela (Medina et al. 1986). Unlike some genotypes of Ananas comosus, high exposure, or perhaps the attendant heating or drought, inhibits growth at some sites to the extent that few seeds and only occasional ramets develop (Chapter 7). Medina et al. (1991b) conducted one of the most ambitious analyses of Type Two Bromeliaceae when they examined cultivated and woodland populations of pineapple and three of its undomesticated congenerics (A. ananassoides, A. paraguazensis, A. lucidus) in northern Venezuela. Aechmea aquilega, Bromelia chrysantha, B. goeldiana, B. humilis and two C3 types, Brocchinia micrantha and Pitcairnia bulbosa, were also included in the survey. All three Ananas species and several of those escaped `varieties' of A. comosus typically grew under taller vegetation. Foliage born by shadegrown individuals contained higher concentrations of N compared with samples collected in the more exposed, rocky sites, supposedly owing to the greater fertility of forest soils and the less scleri® ed nature of shade vs. sungrown foliage. Foliar N, which routinely predicts Amax in C3 plants, also did so for H1max among these Venezuelan bromeliads, but not CO2 consumption during phase four when RuBPc/o assists ® xation. Sun compared with shade-grown foliage produced by the same genotype discriminated less against deuterium (D; presumably fractionated from H2O/D2O in the transpiration stream), indicating heightened reliance on PEPc because fuller exposures promoted CAM (Fig. 4.3). Relatively still air and closer proximity to decomposing C3-type litter may have in¯ uenced D values for the subjects that grew in the understory. Evidence that Medina et al.' s plants discriminated against D according to the prevailing ® xation pathway and incident PPFD was mixed (Fig. 4.3). Contrary to ® ndings elsewhere (e.g., Sternberg et al. 1984), isotopes sometimes failed to distinguish C3 from CAM types. Speci® cally, two C3 species yielded less negative dD values than expected, while records for ® ve CAM types deviated in the other direction. Generally, except for six paired, shade and sun-grown samples, the former gave more positive readings. Presumably, determinants (e.g., variable D/H ratios among the local moisture supplies) beyond the responsible carboxylases must be normalized in more thoroughly controlled experiments. Even so, Medina et al. (1991b) suggested that dD values provide a `very promising' tool to investigate the acclimation of CAM bromeliads and other plants to altered PPFD. Aechmea magdalenae, more than either Ananas comosus or Bromelia humilis, demonstrated that deep shade constitutes a major dimension of the Cambridge Books Online © Cambridge University Press, 2009 126 Carbon and water balance Figure 4.3. Distribution of D and dD values for C3 and CAM species grown in sites characterized by widely divergent light and humidity conditions (after Medina et al. 1991b). realized niche of a Type Two bromeliad (P® tsch and Smith 1988). This extraordinarily robust terrestrial (leaves up to 3 m long) routinely dominates understory sites in humid forests from Costa Rica to Ecuador at abundances up to one spiny shoot per m2. On Barro Colorado island, colonies occupied 25± 100 m2 patches of forest ¯ oor with enough interlocking shoots to largely halt regeneration by local woody ¯ ora (Brokaw 1983). Frequent fruiting and vigorous ramets characterized populations under both closed (.3%) and more open (.35%) canopies. Plants moved into the laboratory received 15 mmol m22 s21 (LL plants) or 300 mmol m22 s21 (HL plants) for 6± 8 weeks prior to measurements of CO2 exchange under high and low PPFD. Brokaw also recorded leaf production by a second set of undisturbed plants in the same Panamanian forest. Low-light-grown subjects responded similarly to weak and stronger PPFD (Fig. 4.4). Amounts of CO2 (net) consumed over complete day/night cycles differed little, and subjects treated either way relied on CAM, but not to the same degree. Eighty to ninety vs. about 60% of the carbon gained under high and low exposures respectively accumulated during phase one, Cambridge Books Online © Cambridge University Press, 2009 Ecophysiological profiles of the five types of Bromeliaceae 127 Figure 4.4. Diurnal course of CO2 exchange by Aechmea magdalenae grown exposed to high (~300 µmol m22 s21; HL) or low (~17 µmol m22 s21; LL) light after being preconditioned for 6± 8 weeks under shade cloth that provided 5 or 35% sunlight (after P® tsch and Smith 1988). with phase four accounting for the balance. Conversely, HL plants failed to maintain positive carbon balances during the LL runs and took up somewhat more CO2, again via CAM, under the HL treatment. Forest plants grew slowly (0.76 g m22 day21), only a small fraction of the growth recorded for some cultivated, exceptionally vigorous CAM plants representing Agavaceae and Cactaceae (Nobel 1991). Aechmea magdalenae specimens growing under relatively open canopy gained weight faster than those in deeper shade purportedly because moisture, not photosynthetically active radiation (PAR) was more abundant there (P® tsch and Smith 1988). Thinner foliage prevailed under dense canopies as did low apparent quantum yields (0.01± 0.001). This forest-dweller Cambridge Books Online © Cambridge University Press, 2009 128 Carbon and water balance surely quali® es as one of the most shade-tolerant bromeliads, and could well rank with the record-holders among CAM plants adapted to lowenergy habitats. It also demonstrated extraordinary capacity for photosynthesis at another shaded location. Koniger et al. (1995) and Skillman et al. (1999) compared Aechmea magdalenae to diverse, co-occurring C3 herbs (e.g., species of Calathea, Dieffenbachia, Piper), also on Barro Colorado island, to explain this bromeliad' s impressive growth as a CAM plant in deep shade. Maximum photosynthetic capacity (17.5 mmol O2 m22 s21) exceeded that for neighboring C3 types (2.1± 6.1 mmol O2 m22 s21), as did nitrogen-use efficiency. Compared on the basis of dry weight and chlorophyll content rather than leaf area, observed rates more closely approached parity owing to the 2± 4-fold thicker foliage of the bromeliad. Calculations using measured Amax further indicated that nitrogen-use efficiency for Aechmea magdalenae equaled 188 mmol O2 mol N s21, twice that of the co-occurring C3 ¯ ora and comparable to values obtained on the same occasion for several of the local trees. Observations conducted in Venezuela also demonstrated how certain metabolites ¯ uctuated in foliage as Ananas comosus and A. ananassoides performed CAM (Medina et al. 1993). Exposures differed enough among sampled populations to induce conspicuous sun and shade plant characteristics. Higher leaf weight/area ratios and greater succulence (water content/leaf area) characterized the HL subjects, as did more vigorous phase one activity. Substantial citric in addition to malic acid accounted for DH1 in leaf sap, although somewhat unevenly among treated populations. Fructose consistently exceeded concentrations of the other assayed sugars, with sucrose yielding the lowest values of all. However, this substrate alone cycled inversely with DH1 suggesting involvement in phase one. Fructose and glucose levels in the sun, but not shade, leaves of several A. comosus cultivars diminished at night, while the proportions of several cations (K1.Ca21.Mg21) changed little. All three nutrients occurred most abundantly in shade-grown specimens. Leaf sap osmolality increased signi® cantly toward dawn only among HL plants. Type Two bromeliads have also demonstrated plant acclimation to temperature, and how this variable in¯ uences CAM. Warm days followed by cooler nights (30/15 °C) maximized CO2 consumption by Ananas comosus, while constant temperatures and inverted oscillations depressed uptake (e.g., Neales et al. 1980). Bartholomew (1982) reported broad day and night optima for phase one. Generally, CO2 consumption diminished during the ® rst and fourth phases of CAM as stress imposed by several agencies increased (e.g., drought in addition to thermal), and decidedly unfavorable Cambridge Books Online © Cambridge University Press, 2009 Ecophysiological profiles of the five types of Bromeliaceae 129 conditions caused plants to lose carbon. Gradual changes impacted carbon budgets less, probably because they more closely parallel circumstances in native habitats. High thermal optima characterized some lowland forms, for example about 27 °C for Aechmea nudicaulis (Type Three) in Trinidad (Smith et al. 1986). Broader surveys of CAM Bromeliaceae would likely con® rm that the most propitious temperatures for carbon gain vary as much as those prevailing in situ. Pineapple indicated that photoperiod affects CAM in at least one bromeliad. Subjects maintained in growth chambers at constant PPFD under different day lengths (daily photon dosages varied) exhibited similar DH1, perhaps because every treatment provided enough irradiance to maintain the carbohydrate reserves needed for robust acidi® cation (Friend and Lydon 1979). However, foliar H1 rose more slowly as day lengths diminished. Short photoperiods strengthened phase one in another set of runs, although plants grown under longer days consumed more CO2 during phase four (Nose et al. 1986). Nitrogen supply also in¯ uences carbon gain according to observations on several Type Two species. Nose et al. (1985) demonstrated that foliar N, which they manipulated in A. comosus by altering the composition of hydroponic media, correlated with nocturnal CO2 uptake. Nitrogen starvation caused plants grown under high light to increase the proportion of DH1 dependent on recycled CO2 relative to supply from the atmosphere (Borland and Griffiths 1989). Low N status also promoted citric in proportion to malic acid synthesis during phase one. Fetene et al. (1990) noted similar responses for Bromelia humilis, and interpreted these results relative to the conditions many Type Two bromeliads experience in situ, as discussed below (Figs. 4.5, 4.6). Still undetermined are the relative contributions of plasticity and genotype to the exceptionally broad ecotolerances of certain Type Two bromeliads. Two Ananas comosus cultivars demonstrated different capacities to adjust across a range of PPFD that may accord with distinct origins in Venezuela (Medina et al. 1991a). Variety Brecheche was probably selected by farmers indigenous to the Orinoco river basin for cultivation in full sun to partial shade in palm swamps, while Spanish Red, the second genotype, provides most of the commercially produced pineapple in that country today. Subjects grown in phytotrons under low and higher light (25± 50 or 325± 400 mmol m22 s21) with ample irrigation and fertilizer developed different ecophysiological pro® les, including responses to shade. Brecheche acclimated to high and low light more successfully than Spanish Red according to DH1, gas exchange and shifts in leaf chemistry. By the end of Cambridge Books Online © Cambridge University Press, 2009 130 Carbon and water balance Figure 4.5. In¯ uence of growth PPFD and nitrogen supply on the nitrogen-use efficiency of Bromelia humilis (after Fetene et al. 1990). the six-week pretreatments, chlorophyll and N contents in LL plants were 2.8 and 1.4 times those of the better-exposed specimens. Type One species Type One Bromeliaceae remain less studied than species representing the other four types in part because they lack commercial value and the novelty of the phytotelm and wholly trichome-dependent species. Even so, enough is known about these plants to report that mechanisms of carbon and water balance vary with the substrate, exposure and climate as with Type Two species (Martin 1994). Certain genera exhibit CAM (e.g., Hechtia, Dyckia, Encholirium) and others C3-type photosynthesis (e.g., Fosterella, Pitcairnia). Views vary on Puya, one of the largest and most ecologically Books Online © Cambridge University Press, 2009 Ecophysiological profiles of the five types of Bromeliaceae 131 Figure 4.6. Light response curves of integrated net CO2 uptake for the dark period by Bromelia humilis expressed per unit of leaf area (after Fetene et al. 1990). diverse of the genera comprising this category (Fig. 14.2C). Medina (1990) assigned the entire taxon C3 status, despite his own contradictory determinations (D5,15± 25½ ; Table 4.1) for several species. Puya copiapina yielded a particularly convincing 15.4½ (Medina et al. 1977). On the other hand, natives of the coldest habitats would deviate from the norm for alpine ¯ ora if equipped for other than C3 photosynthesis. Frost probably exceeds aridity as the primary challenge to these high Andean bromeliads, much as nightly freezes account for the novel physiology and morphology of the giant rosette-forming herbs in other families (Figs. 7.2± 7.4). Facultative CAM probably characterizes a substantial number of Puya species from warmer (lower) sites, and P. floccosa (D522.5½ ) reportedly CAM-cycles (Medina et al. 1977; Smith et al. 1986). Even fewer reports address instantaneous performances. Lüttge et al. Cambridge Books Online © Cambridge University Press, 2009 132 Carbon and water balance (1986a) measured gas exchange by Pitcairnia integrifolia, a pseudolithophyte (Chapter 7) in Trinidad. Carbon gain commenced at dawn, intensi® ed, and then diminished during midday coincident with leaf temperatures that occasionally approached 52 °C. Poor WUE indicated extensive root systems and an abundant moisture supply. Greenhousegrown subjects light-saturated between 200 and 400 mmol m22 s21 and 10± 15 mmol m22 s21 balanced respiration. Quantum yields between 0.02 and 0.03 con® rmed shade-tolerance. Broad shields (Fig. 2.8D) born by the foliar trichomes characteristic of Type One Pitcairnioideae native to relatively dry habitats supposedly reduce transpiration and vulnerability to photoinhibition (Lüttge et al. 1986a). Glabrous, adaxial leaf surfaces produced by architecturally similar P. bifrons re¯ ected just 20% of incident PAR (70° angle to the blade surface), while the densely invested abaxial side scattered 39.4% of the same beam. Restriction of the indumentum and stomata to the abaxial epidermis here and among many relatives suggests greater importance for water conservation than for photoprotection. Such arrangements might provide a third service by repelling pathogens that could otherwise attack susceptible foliage through stomata. The occasional Pitcairnia species, Ayensua and probably some members of Brocchinia shed their green foliage (but not the spiny, reduced achlorophyllous organs in the case of Pitcairnia heterophylla; Fig. 2.12A) to conserve moisture during the driest months of the year. Leaves of some Fosterella species, like many other drought-avoiding monocots (e.g., grasses), lack the same discrete abscission mechanisms, instead gradually withering beginning at the tip. As for many other deciduous plants, the behavior of cultivated, well-watered specimens points to photoperiod as the primary cue for senescence (Chapter 6). Type Three species Types Three and Four Bromeliaceae exaggerate the relatively modest utriculate morphology of the leaves of the Type Two species (Fig. 2.4; Table 4.2). Accordingly, the more capacious leaf axils impound enough moisture and debris to permit abandonment of the substratum for all but physical support. Apogeotropic roots (Fig. 2.14D) occasionally penetrate the phytotelmata, but water and nutrients mostly enter the shoot through foliage. Juveniles remain too small for months to several years to accumulate adequate soil substitutes in leaf axils. Seedlings of Type Three Bromeliaceae (Figs. 3.8A, 6.5A,F) presumably rely on roots to obtain moisture and nutri- Cambridge Books Online © Cambridge University Press, 2009 Ecophysiological profiles of the five types of Bromeliaceae 133 ents. Consequently, occurrences on fundamentally hostile media like bark and rock probably require the presence of mitigating nonvascular ¯ ora that adult morphology (phytotelma) renders unnecessary (Fig. 6.5A,F). Type Four Bromeliaceae, although likewise dependent on phytotelmata later in life, spend the ® rst weeks to months of life root-free, suggesting greater dependence on foliar trichomes and diminished opportunity for substrates to in¯ uence where seeds can succeed (Figs. 3.8C, 6.5G). Surveys in situ and a closer look at trichome function through the plant life cycle would probably con® rm what seems almost certain to be one of the important distinctions between Type Three and Type Four Bromeliaceae. Certain embellishments of the bromeliad shoot provide the phytotelm species several options for nutrition by favoring speci® c kinds of events in phytotelmata (Chapters 5 and 7, and below). Carnivory appears to be limited to Tillandsioideae (Catopsis) and Brocchinia. Little is known about how any of the impounding bromeliads affects events in its phytotelmata. Laessle (1961; Fig. 8.12) noted diurnal rhythms in pH in the solutions impounded by diverse Jamaican bromeliads that Benzing et al. (1972) con® rmed in the laboratory with Aechmea bracteata (Fig. 8.14). Dissolved CO2 accounted for some of the change, but boiling failed to neutralize all of the acidity that developed during the night (McWilliams 1974). Solutes, including organic acids synthesized during phase one, may leak into phytotelmata fast enough to account for the observed acidi® cation. Whether reabsorption from the same unstirred solutions explains the morning rebound is less likely. Presumably, moisture impounded in a bromeliad shoot represents the ® rst in a series of coupled compartments that constitutes a hydraulic continuum, much as soil water is in¯ uenced by the free energy status of the moisture in the adjacent plant. Opportunities to exchange solutes across the same boundaries must be subject to more stringent plant control. Type Four species Tillandsioideae that comprise Type Four mostly possess relatively thin, sparsely trichomed C3-type foliage (Figs. 1.2G, 2.8B; Table 4.2); leaf axils trap substantial amounts of moisture and litter except for the carnivores (Figs. 2.4F, 7.16). The single bromeliad (Guzmania monostachia) with welldocumented facultative CAM also belongs here, suggesting the possibility of additional examples of this and other mixed syndromes as considered below. Light relations probably vary about as much among the Type Four bromeliads as across the entire family. Exceptional members, like Catopsis berteroniana (Fig. 5.3B) and Brocchinia reducta (Fig. 2.4F), require high Books Online © Cambridge University Press, 2009 134 Carbon and water balance exposure to satisfy the inherently high energetic costs of prey-dependence (Givnish et al. 1984). Reliance on ¯ ying, light-seeking rather than nonvolant fauna (primarily ants), as does B. reducta, further restricts fruiting specimens of Catopsis berteroniana to the uppermost perches in supporting canopies. Vertical foliage covered by an extraordinarily re¯ ective cuticle helps protect the three carnivores and certain other Type Four species from potentially injurious irradiance in what are typically well-exposed microsites (Fig. 5.3A). Residence deeper in the canopy, where many other Type Four species occur, mandates more horizontally oriented foliage, that by promoting litter capture may diminish shade-tolerance (Figs. 2.4H, 7.16). Low PPFD limited carbon gain by noncarnivorous, forest-dwelling Tillandsia fendleri and Vriesea jonghei monitored in northern Trinidad (Griffiths et al. 1986). Additional features of foliage, particularly the layering of leaves and their pigmentation (Figs. 2.14G, 2.17B, 2.18B), distinguish other phytotelm Tillandsioideae within Type Four, and in some cases probably promote performance in dim light as Lee et al. (1979) proposed for some other tropical herbs. Quite likely, Type Four bromeliads with lax, sometimes discolorous, monolayered foliage (Fig. 2.4H) rank among the most shade-tolerant members of the family. Leaves of Guzmania lingulata photosaturated below 300 mmol m22 s21, and exhibited compensation intensities less than onetenth as strong whether preconditioned for nine months with high or low (45 vs. 400± 700 mmol m22 s21) exposures (Smith 1989; Fig. 4.7). Smith also noted acclimation, speci® cally that chlorophyll concentration and vulnerability to photoinhibition increased and dark respiration slowed when specimens were transferred from high to low light. Conversely, Amax remained essentially unchanged. Other Type Four species that Pittendrigh (1948) considered shade and exposure-tolerant included Tillandsia monadelpha, T. anceps and Vriesea simplex. Excised foliage of all three species exhibited light responses (low compensation and light saturation intensities) consistent with shade-tolerance (Benzing and Renfrow 1971b). Catopsis nutans and Tillandsia complanata also possess demonstrably malleable light-harvesting systems (see Martin 1994 for summarized data). Type Four bromeliads that bear densely congested canopies featuring unusually thin, uniformly green foliage instead of the bicolored, more robust and monolayered displays of Catopsis nutans and Guzmania lingulata also tolerate heavy shade. And as Pittendrigh surmised (Fig. 7.11), these plants would probably occur more widely if better equipped to counter drought. Grown fully exposed in wet sites, particularly at high alti- Cambridge Books Online © Cambridge University Press, 2009 Ecophysiological profiles of the five types of Bromeliaceae 135 Figure 4.7. Response of net photosynthesis (CO2 uptake) to PPFD by Guzmania lingulata pretreated for nine months in low (average 45 µmol m22 s21) or high (average 250 µmol m22 s21) environments (after Smith 1989). tudes, abundant sun-screening anthocyanins accumulate in the adaxial epidermis. Tillandsia complanata, for example, appears almost chlorophyllfree under such conditions, the light-harvesting pigments masked by a suffuse, brownish-red screen. Substantial shade induces this same wideranging epiphyte to develop deep bluish-green foliage in part owing to more concentrated chlorophylls. It, like many other, routinely more ornamented Tillandsioideae (e.g., Vriesea irazuensis) brought under glass or into shade, or if simply moved from high to lower elevations, develops unremarkable green foliage within weeks. On the other hand, some taxa, or just certain populations within a species, remain brightly colored in sun and shade (e.g., Tillandsia biflora). The apparent incapacity of Catopsis to synthesize anthocyanins may present no serious problems because other compounds serve as well for at Cambridge Books Online © Cambridge University Press, 2009 136 Carbon and water balance least two of the same purposes. Fruits are dry so frugivores need not be attracted, and all of its ,20, probably entomophilous species produce white, green or yellow ¯ owers and associated bracts. The most heliophilic member of the genus, Catopsis berteroniana, features that dense, powdery re¯ ective epicuticle (Fig. 5.3A) over the yellowish foliage just mentioned, as do some related taxa (e.g., C. morreniana). Similar coatings mark the large, distichous bracts of still other Tillandsioideae (e.g., Tillandsia heterophylla), probably to attract nocturnal pollinators. Catopsis species usually con® ned to deep shade (e.g., C. floribunda, C. nitida) lack these conspicuous wax deposits and maintain deep green foliage. Guzmania monostachia, the widest-ranging member of its sizable genus, exhibits exceptional ecological versatility, including tolerance for high to low exposures, assisted in part by facultative CAM (Medina et al. 1977; Maxwell et al. 1992, 1994, 1995). Unstressed, it takes up CO2 mostly by day; while droughted or overexposed, even if well watered, a typical CAM pattern strengthens as described in greater detail below (Fig. 4.8; Table 4.6). Scores for biomass ranged from D5223.7 to 231.5½ . Specimens at the C3 end of the spectrum experienced relatively equable growing conditions, while those with less negative values occupied seasonally drier or relatively exposed sites at wetter locations. Guzmania monostachia may differ regionally on some or all of these counts. Florida' s small population routinely occupies shady habitats, whereas others farther south often experience less diminished sunlight. Structurally comparable Tillandsioideae with somewhat less extensive geographic, but greater altitudinal, ranges (e.g., G. sanguinea) may owe their success to similarly ¯ exible ecophysiology. Some Type Four Bromeliaceae probably CAM-cycle. McWilliams (1970) reported suggestive metabolic rhythms for Vriesea fenestralis, and Medina (1974) identi® ed malic acid as the leaf constituent responsible for these ¯ uctuations in Catopsis nutans, Guzmania mucronata, Tillandsia adpressiflora and Vriesea platynema. Cyclers in other families (e.g., Pereskia of Cactaceae) also lack pronounced xeromorphy, instead achieving water economy as mentioned above by re® xing dark-respired CO2 that would require the expenditure of additional moisture to replace from the atmosphere (Harris and Martin 1991; Martin 1994). Heterochrony appears to explain much of the novel architecture and specialized foliar epidermis of the Type Five bromeliad. If so, evolution began with a more mesic stock and progressed through a series of increasingly abbreviated, xeromorphic and generally miniaturized stages represented today by an array of surviving lineages (Fig. 2.1). Tillandsia subgenus Diaphoranthema culminates this sequence with the most diminutive of Cambridge Books Online © Cambridge University Press, 2009 Ecophysiological profiles of the five types of Bromeliaceae 137 Figure 4.8. Dawn/dusk titratable acidity (DH1) for Guzmania monostachia exposed for 14 days to high light under water-stressed and well-watered conditions (after Maxwell et al. 1994). these putative descendants. Phytotelm Tillandsioideae, as exempli® ed by Tillandsia deppeana and Vriesea geniculata, may resemble evolutionary antecedents according to a long-standing hypothesis (Schulz 1930) and some recent ® ndings on structure and function. Tillandsia deppeana and Vriesea geniculata inhabit relatively humid forests where as adults they maintain substantial, water-tight leaf axil chambers that permit roots to be primarily mechanical (Fig. 4.9). Because the root system lacks appreciable absorption capacity, the shoot must be at least one-quarter full size (i.e., possess foliar impoundments) before opportunities to hydrate and absorb nutrients exceed those brief intervals while leaves remain wet from precipitation. Epidermal structure, in addition to leaf and shoot morphology, changes with plant age, as do related aspects Cambridge Books Online © Cambridge University Press, 2009 Table 4.5. Summary of leaf H2O relations of the bromeliads investigated at six study sites during the dry season in Trinidad. n 5 number of species examined at each site. Number in parentheses indicates mean annual rainfall (in mm) determined from records for 2–6 years Minimum p leaf (MPa) Minimum xylem tension (MPa) Maximum xylem tension (MPa) Point Gourde (1281) n53 Tucker Valley (1612) n55 Areva (2625) n58 Simla (2566) n55 Lalaja (not given) Textel (2637) n55 0.98 0.67 0.82 0.66 0.61 0.92 0.59 0.31 0.56 0.58 0.29 0.58 0.57 0.29 0.46 0.51 0.30 0.47 Source: After Smith et al. (1986). Cambridge Books Online © Cambridge University Press, 2009 Ecophysiological profiles of the five types of Bromeliaceae 139 Figure 4.9. Diurnal patterns of CO2 exchange by juvenile and adult Tillandsia deppeana denied irrigation. Irradiance 650± 900 µmol m22 s21, 25/18 °C day/night leaf temperature, 50± 60/70% relative humidity day/night (after Adams and Martin 1986a). of water relations. Trichomes born by Vriesea geniculata during its ® rst several years of life cover about 80% of the leaf surface, but later that value falls below 5% (Reinert and Meirelles 1993; Table 4.7). Shield morphology, speci® cally its width, accounts for part of the difference in cover value as greater numbers of cell divisions produce the Cambridge Books Online © Cambridge University Press, 2009 Table 4.6. Morphological and physiological characteristics of the central portion of the leaf blade of Guzmania monostachia under three light regimes in situ Microhabitat Thickness (mm) Chlorophyll Succulence Chlorenchyma Air space (mg g21 DH1 DH1 (kg m22) (%) (%) fresh weight) (high PAR) (low PAR) Exposed 0.46(60.02) 0.33(6 0.02) Semiexposed 0.46(6 0.02) 0.32(6 0.03) Shaded 0.34(60.01) 0.21(6 0.04) 25 44 62 8 7 14 228 597 1021 65(6 2.4) 60(6 4.6) 15(6 2.5) Source: After Maxwell et al. (1992). Cambridge Books Online © Cambridge University Press, 2009 Xylem tension (MPa) Dawn Dusk 48(6 5.7) 0.73(60.01) 0.61( 6 0.03) 40(6 4.5) 0.69(60.04) 0.58(6 0.02) Ð 0.50(60.01) 0.45(6 0.02) Ecophysiological profiles of the five types of Bromeliaceae 141 juvenile (418116132164) compared with the adult (418132) pattern (Fig. 2.9). Stomatal densities also rise and chlorophyll concentrations increase on a fresh weight basis to further distinguish adult from juvenile foliage much as these two features and trichome density differentiate mature Type Four and Type Five bromeliads (Benzing and Renfrow 1971b; Martin 1994). Finally, juveniles appear (no numbers yet) to produce relatively succulent foliage, a life-long feature for the Type Five but not the Type Four bromeliad. Despite the relatively xerophytic qualities of the seedling compared with its water-impounding adult, drought probably exacts a high toll on Type Four Bromeliaceae until shoot architecture moderates the threat of serious desiccation. A missing element in this scenario is CAM, which might reduce mortality, especially of juveniles, were its capacity to conserve tissue moisture experienced early in the life cycle. However, D values recorded for Vriesea geniculata seedlings ranged between 26.7 and 27.2½ and for adults between 24.2 and 26.0½ (Reinert and Meirelles 1993), while Tillandsia deppeana exhibited C3-type gas exchange through the entire life cycle (Adams and Martin 1986a,b; Fig. 4.9). Perhaps C3 photosynthesis rather than CAM remains the better strategy because what are often accompanying higher rates of carbon gain shorten the initial, relatively drought-sensitive life stage. Several additional indicators, none directly related to photosynthetic pathway but all signi® cant for survival, revealed an even more important distinction between the early juvenile and older Tillandsia deppeana specimen. Young and more mature subjects alike achieved quite respectable water economies; in fact performances equaled those of some C4 and CAM plants (Fig. 4.10). Gas exchange also demonstrated why growth accelerated. Maximum values for g, E and Amax all increased up to a full order of magnitude on a fresh weight basis as adult replaced juvenile morphology. Such size-related differences in Amax may be characteristic for many other plants, and the phenomenon may be especially accessible to study using epiphytes (Zotz 1997b), although not necessarily heterochronic, Type Four Bromeliaceae. Responses to imposed drought followed by irrigation also supported neoteny as the explanation for the evolution of Type Five Bromeliaceae. Net photosynthesis by initially well-watered juveniles of Tillandsia deppeana maintained in a growth chamber dropped 60% after just one day (Fig. 4.9). Additional days of drying brought further, smaller diminutions, but some carbon gain continued through the run. Adults treated identically responded more dramatically, almost ceasing net uptake by the end of the Cambridge Books Online © Cambridge University Press, 2009 142 Carbon and water balance Figure 4.10. Water-use efficiencies of adult and juvenile Tillandsia deppeana at constant leaf temperatures and relative humidity of 55% (after Adams and Martin 1986a). ® rst photoperiod. Carbon balance became negative sooner, by about noon each day thereafter during the imposed, nine-day drought. Additional mature plants, also drained of their tank contents, moderately desiccated, and then re® lled, failed to resume pretreatment rates of photosynthesis for about a day. Similarly treated juveniles completely recovered shortly after leaves were moistened ± a capacity attributed to the denser cover of absorbing trichomes present. So it seems that adult T. deppeana emphasizes carbon gain, expending much water in the process; juveniles operate with greater economy, the trade-off being slower growth (Adams and Martin 1986a). Cambridge Books Online © Cambridge University Press, 2009 Ecophysiological profiles of the five types of Bromeliaceae 143 Type Five species Of the several hundred Tillandsioideae that satisify Type Five criteria, Tillandsia usneoides provides the fullest opportunity to pro® le a representative species. However, responses to similar treatments have varied and probably for several reasons, including the age of the truss, its composition (number of components, live and dead shoots) and preconditioning. Kluge et al. (1973) demonstrated increasing CO2 consumption as exposure rose from 200 to 1000 mmol m22 s21, but pretreatment at 200 mmol m22 s21 had lasted just six days. Sections of shoots that had previously grown for eight weeks at 30, 300 or 800 mmol m22 s21 photosaturated at about 500 mmol m22 s21 according to O2 exchange (Martin et al. 1989). Another set of plants taken from trees immediately prior to runs under 7± 15% full sunlight (100± 200 mmol m22 s21 PAR at midday) differed less than expected from fully exposed (1500± 1600 mmol m22 s21) controls in those features that usually reveal sun or shade-adaptation (Martin et al. 1985, 1989). Low-light plants contained more concentrated chlorophyll, while chlorophyll a/b ratios remained unaffected. High irradiance enhanced starch deposition, but had no other visible effects on chloroplast structure. Finally, internode length, leaf size, stomatal density, and trichome and guard cell morphology were indifferent to PPFD. Maximum acidity measured about 60% of that recorded for fully irradiated greenhouse subjects. Spanish moss collected from the darkest of three exposures (55, 80, 1000 mmol m22 s21) in a South Carolina forest achieved higher DH1 than specimens maintained under glass (Martin et al. 1986). Performance did not, however, deviate substantially from that of the more exposed subjects at the other two ® eld sites. Stomata were somewhat more numerous and trichome shields only slightly smaller compared with the plants grown in fuller sun. Chlorophyll data paralleled those recorded for the specimens maintained under glass. Additional runs yielded light saturation values between about 125 and 400 mmol m22 s21 using several methods to determine carbon gain. Apparently, T. usneoides in nature and cultivation acclimates across a broad range of PPFD without substantial morphological or metabolic adjustment, and a full day of only about 10 mol m22 can saturate photosynthetic capacity. Evidence that high exposures photoinhibited still other batches of Spanish moss further underscored its shade-adapted character (Martin et al. 1986, 1989). Desiccation typically depresses carbon gain, but Tillandsia usneoides has sometimes exhibited the opposite response to drying and another, more Cambridge Books Online © Cambridge University Press, 2009 144 Carbon and water balance Figure 4.11. Effect of wetting on CO2 exchange by Tillandsia usneoides. Plants were removed from the chamber, brie¯ y dipped in water and returned to the chamber (after Martin and Siedow 1981). consistent reaction to excess moisture. According to Kluge et al. (1973) and Martin and Siedow (1981), partial dehydration stimulated uptake, but the same condition slowed CO2 consumption on another occasion (Martin and Schmitt 1989). Conversely, wetted leaf surfaces always depressed gas exchange, almost eliminating it for a variety of Type Five species (Fig. 4.11). A con¯ uent layer of water-saturated trichome shields apparently slowed the diffusion of gases through the underlying stomata while the foliage was wet (Benzing et al. 1978; Martin et al. 1981; Martin and Siedow 1981; Fig. 2.8C,E). Treated identically, only the leaves of the sparsely trichomed, phytotelm types and myrmecophytic Tillandsia bulbosa behaved as if still surface-dry (Table 4.8; Fig. 2.8B). Tillandsia usneoides does not represent its type particularly well, but then neither do any of the 250± 300 other species characterized by the same combination of absorptive trichomes, largely holdfast roots, succulence and CAM. Nor do these most specialized of the bromeliads occupy similar habitats or exhibit the same water and light relations. Some representatives retain moisture more tenaciously or rehydrate more rapidly than others. Spanish moss, for example, had lost more than half of the moisture held in initially well-watered shoots after just 1± 2 weeks of drought (e.g., Penfound Cambridge Books Online © Cambridge University Press, 2009 145 Xeromorphy and water relations Table 4.7. Characteristics of the leaves of heterophyllic Vriesea geniculata as a juvenile and an adult Trichome density on midblade % covered by (mm22) trichome shields Chlorophyll concentration Density of (mg mg22 stomata (mm22) dry weight) Juvenile Abaxial Adaxial 18.366.8 13.363.4 66 87 19.8 67.7 0 263.8 Adult Abaxial Adaxial 6.864.3 2.961.6 2 0 26.86 5.2 0 423.4 Source: After Reinert and Meirelles (1993). and Deiler 1947; Biebl 1964; Martin and Siedow 1981; Fig. 4.12), in part owing to unfavorable architecture. Quite a few relatives (e.g., T. filifolia, T. gardneri, T. tectorum) possess surface to volume ratios no better suited for dry climates. Some of these same tillandsioids also feature a surprisingly thin epidermis and cuticle (see Tomlinson 1969; Fig. 2.10A± H). Conversely, Tillandsia concolor, T. karwinskyana and T. circinnatoides, among others, resist desiccation more effectively, aided in part by compact habits and stouter epidermal barriers bearing layers of scales with appressed rather than elevated shields (Fig. 2.10K; Table 4.10). Differences in shoots and leaves, especially the indumentum, among Type Five species may relate more to the nature and timing of the moisture supply (dew, cloud droplets or rainfall) and the need to dissipate potentially damaging radiation than to requirements for high WUE. Xeromorphy and water relations Bromeliads counter drought in two ways, neither of which brings to bear the powerful xylem tensions that the so-called euxerophytes generate to maintain adequate supplies of moisture during prolonged dry weather. Survival depends on either high hydraulic capacitance ± enough to qualify as succulents, combined with extraordinary WUE or drought-avoidance in the sense that shoots jettison relatively vulnerable (permeable) foliage before scarcity imposes intolerable moisture de® cits (e.g., Pitcairnia heterophylla; Fig. 2.12A). Leaf characteristics, viz. capacitance and the anatomy responsible for water storage, insure that species comprising the ® rst group Cambridge Books Online © Cambridge University Press, 2009 146 Carbon and water balance Figure 4.12. Water content of clumps of Tillandsia usneoides over the course of three days while sheltered from precipitation (after Penfound and Deiler 1947). experience, at most, modestly depressed Cleaf (bulk leaf water potential) even while substantially desiccated. These bromeliads differ little from the typical mesophyte in the range of Cleaf experienced even during the driest weather, i.e., when relative water contents (RWC) are most depressed. Well-watered ¯ ora typically sequester enough moisture to continue A (and accompanying E as well) for some interval after opportunity for replenishment ceases (e.g., supporting soil reaches critical wilting point). A few additional hours of operation at Emax causes the stomata to close and g to approach zero. Viability may continue, but growth ceases until tissues recharge. Succulents greatly extend the time available for carbon gain under the same conditions through more sparing use, usually aided by CAM, of larger stores (greater capacitance) of water per unit of transpiration surface. Capacitance probably varies many fold among Bromeliaceae as does Emax, and these two properties relate inversely. Zotz and Andrade (1997) report that measured relative capacitance (DRWC/DC) was about 0.70 MPa for Tillandsia fasciculata and 0.30 MPa for Guzmania monosta- Cambridge Books Online © Cambridge University Press, 2009 Xeromorphy and water relations 147 chia. Values for epiphytes in general (ferns low end, cacti high end) range from about 0.15 to 0.70 MPa (Andrade and Nobel 1997). As de® cits mount within the succulent plant, moisture sequestered in its often voluminous, collapsible water-storage tissue or hydrenchyma moves to adjacent, more vulnerable chlorenchyma, a transfer requiring only modest (,1.0 MPa) C gradients across short distances, whether in stems or leaves (Figs. 2.10, 2.13, 2.16). Basically, succulence tends to decouple E (hence A) from the environmental supply of moisture, imparting major advantage where many bromeliads grow. Compared with many other drygrowing plants, bromeliads possess limited capacity to access soil water, i.e., powerful xylem tensions would serve no useful purpose for the epiphytes and saxicoles, or for the terrestrials, many of which possess weakly absorptive or poorly developed root systems. Foliage provides most of the capacitance for the large majority of drought-enduring Bromeliaceae. Stems contribute modestly to shoot volume and water supply except for certain species of Puya that, unlike the rest of the dry-growing bromeliads, experience drought and relief on a daily basis. As alpine endemics, these plants co-occur and look more like some Espeletia (Asteraceae) and the other cool-growing, giant rosette-forming herbs than the balance of Bromeliaceae, including lower-elevation Puya (Fig. 14.2C). Taxonomic affinities aside, the group shares convergent architecture and probably much ecophysiology. Unlike most high-altitude ¯ ora, which diminish in physical stature where ranges exceed the tree line, species of Espeletia, Puya and to a lesser degree Lupinus (Fabaceae) become more massive-bodied (Fig. 14.2C). Compact shoots and in¯ orescences and dense indumenta clearly distinguish those populations adapted for the coldest Andean ranges, in part to promote heat retention as described in Chapter 7 (Figs. 7.2± 7.4). Frigid nights also exacerbate water balance by freezing soil moisture and reducing root conductivity ± in effect by promoting physiological drought. Goldstein et al. (1984) compared seven Venezuelan Espeletia species graded by size and distribution between 3100 and 4200 m. Natives of the coldest and driest habitats produced the tallest, thickest stems containing the largest volumes of water-® lled parenchyma. Usually wetter, lower sites harbored the test subjects with less stem capacitance per unit surface of supplied foliage. Basically, reserves available to the coldest-growing forms are sufficiently large and adequately coupled hydraulically to leaves to permit undiminished transpiration for up to 2.5 h, enough time for the morning sun to thaw soils frozen the previous night and warm chilled root systems. Consequently, photosynthesis can occur through the entire Cambridge Books Online © Cambridge University Press, 2009 148 Carbon and water balance Figure 4.13. Diurnal ¯ uctuations in malate content (DH1) and nocturnal consumption of CO2 for six species of Tillandsia relative to percent of mesophyll occupied by hydrenchyma (succulence) expressed on a dry weight (DW) basis (after Loeschen et al. 1993). photoperiod. Additionally, Cleaf ¯ uctuated least, turgor loss points were lowest, and hydraulic resistance to ¯ ow from soil to foliage was highest in the tallest species, further underscoring the importance of stored water to mitigate short-cycled drought. Similar water relations probably explain the caulescent habit in Puya, its densely packed foliage, thick indumenta, and restriction of the most massive species to paramo and puna rather than more permissive, lower-elevation habitats. The adaxial hypodermis, which sometimes occupies more than half of the leaf interior (e.g., 53% in Tillandsia variabilis; Figs. 2.13B, 4.13), sequesters much of the water that sustains most dry-growing Bromeliaceae during droughts. A second, shallower layer of similarly colorless, thin-walled tissue lies between the chlorenchyma and abaxial epidermis, even in some of the relatively mesic forms, for example Catopsis floribunda, Vriesea incurvata and many Pitcairnia species (Fig. 2.16B,C). Observations indicate that moisture in this compartment is more labile than that residing in green mesophyll. Sideris and Krauss (1955) reported accordion-like shrinkage of the pleated-walled hypodermis of droughted Ananas comosus, while the dimensions of the adjacent chlorenchyma changed little (Fig. 2.13B). Presumably, the low bulk elastic moduli (i.e., high wall elasticity) and turgor-maintenance responsible for similar behavior in Peperomia magnoliaefolia (Schmidt and Kaiser 1987) and Agave deserti (Schulte and Nobel 1989) also prevail in Bromeliaceae. Cambridge Books Online © Cambridge University Press, 2009 Xeromorphy and water relations 149 Stiles and Martin (1996) reported a leaf bulk elastic modulus of 3.3 MPa for Tillandsia utriculata. Values calculated from pressure/volume curves indicated capacity for osmotic adjustment (0.13 MPa from full hydration to the point of turgor loss), and corresponding opportunity to maintain turgor during drought. Stiles and Martin cited these attributes to help explain why tested specimens could continue to gain carbon after 31 days without irrigation. Griffiths et al. (1986) and Lüttge (1987), among others, have also reported osmotic adjustment in desiccating Tillandsia shoots, although they made no attempts to determine the bulk elastic moduli of stressed foliage. Leaf succulence occurs in all three subfamilies and supports species that root on diverse substrates and experience many kinds of climates. Substantial capacitance also characterizes much of the memberships of Types One, Two and Five. Principal sources of moisture for these bromeliads include the contents of modest phytotelmata, soil, rainfall, dew or mist exclusively or two or more together. Plant demands (Emax) also differ (Tables 4.3, 4.4). Less is known about the effects of dehydration on viability. Severely droughted Tillandsia ionantha (Benzing and Dahle 1971) and some relatives had lost more than half of their initial stores (e.g., .64% for T. recurvata and T. usneoides; Biebl 1964), yet shoots fully rehydrated and regained full photosynthetic capacity soon after irrigation. Remarkably, Cleaf never fell below 21.0 MPa (see also Table 4.5). Guzmania monostachia survived losses in excess of 90%, the record for documented desiccation-tolerance (Zotz and Andrade 1997). Poorly understood mechanisms beyond closed stomata that deny access to exogenous CO2 cause photosynthesis to decrease as Cleaf falls below certain values. Taylor and Martin (in Martin 1994) used ethylene glycol solutions to demonstrate reduced O2 evolution and undiminished respiration in sectioned shoots of Tillandsia usneoides as Cmedium was lowered to 24.0 MPa. Although interesting, their ® ndings offer little insight on in vivo performance. Values for Tillandsia utriculata specimens that were so severely desiccated in growth chambers that no CO2 exchange could be detected still averaged about 21.5 MPa (Stiles and Martin 1996). These numbers also represent the lowest readings reported for a bromeliad. A robust foliar epidermis accompanies CAM in Bromeliaceae except for those exceptional `hygromorphic' Tillandsia species (Fig. 2.17). Stout cuticles and thick, inner tangential and radial walls stand out most prominently, the latter often reducing the lumina to a few percent of the total cell volume (Figs. 2.5J, 2.6A, 4.23I). Ananas comosus possesses such gas-tight Cambridge Books Online © Cambridge University Press, 2009 150 Carbon and water balance Figure 4.14. Light re¯ ectance off adaxial leaf surfaces of Catopsis nutans and Tillandsia fasciculata, the second species while wet and dry. Responses for Catopsis nutans were undifferentiated while dry surfaces of Tillandsia fasciculata re¯ ected more light (after Benzing and Renfrow 1971b). foliage that illuminated shoots released O2-enriched (78%) bubbles into hydroponic media through the ends of severed roots (Ekern 1965). Thirteen species representing all three subfamilies and all the ecological types revealed how excised leaf blades with sealed edges resisted desiccation over CaCl2 (Benzing and Burt 1970; Table 4.10). Two of the most decidedly mesomorphic Tillandsioideae examined, Guzmania lingulata and Tillandsia multicaulis (Type Four), proved most vulnerable to drying, losing more than 20% of their initial weights during the ® rst 24 h. Conversely, subjects intermediate in form between Types Four and Five (e.g., T. achyrostachys and T. karwinskyana) held moisture much more tenaciously. Tested Bromelioideae dried to similar degrees. Ananas comosus performed as expected considering Sideris and Krauss' s (1928) demonstration that intact shoots required several months to lose about half of their original mass. Peltate trichomes positioned above the stomata and sometimes colocated within pronounced intercostal grooves (Figs. 2.8F, 2.13A) help conserve water for many Bromeliaceae. A dry, re¯ ective indumentum also scatters considerable sunlight, reducing heat loads and the likelihood of photodamage to underlying chlorenchyma. A large druse crystal positioned in the center of each cell of the adaxial epidermis augments both functions (Figs. Cambridge Books Online © Cambridge University Press, 2009 CAM vs. C3 bromeliads: performances in situ 151 2.13C, 4.23I). Re¯ ectivity shifted with conditions during experiments on several Tillandsioideae. Surface-dry foliage of Tillandsia fasciculata returned up to 45% of the incident radiation between 450 and 600 mm (Fig. 4.14). While wetted, values for the same surfaces fell 25± 50%, to within the range recorded for Catopsis nutans and several other Type Four species that also bear widely scattered rather than overlapping trichome shields (Benzing and Renfrow 1971b; Fig. 2.7B). Baumert (1907) used `thermoneedles' and a galvanometer to demonstrate similar dynamics in some other bromeliads. Another aspect of the leaf epidermis may in¯ uence gas exchange and consequently drought-tolerance for certain Bromeliaceae. Wax deposits occlude the stoma of sparsely trichomed Tillandsia deppeana, and could explain why this wide-ranging, generally mesomorphic epiphyte sometimes accommodates substantial aridity. Perhaps the simplicity of this arrangement, and presumably related ease of modi® cation, favor plant accommodation to long-term shifts in growing conditions better than alterations that require more fundamental change of, for example, stomatal density, succulence or photosynthetic pathway. Pittendrigh' s (1948) survey of Bromeliaceae arrayed across the seven climatic zones he recognized for Trinidad (Fig. 4.15) at once demonstrates the tendency of CAM and C3 types to segregate along regional moisture gradients and the importance of xeromorphy to drought-tolerance. Leaves serving residents of the everwet, seasonal evergreen, and drier deciduous forests across the island differed quantitatively by several characteristics that in¯ uence the accumulation, storage and economical use of water, including blade thickness, percent area covered by indumenta, and the dimensions of the trichome shields (Smith 1989; Table 4.9). Moreover, stomata occurred most densely on the leaves of the least drought-resistant C3 taxa. Many additional data and the optimization theory discussed below help explain these associations and identify which bromeliads should exhibit high or low values for Amax and Emax. CAM vs. C3 bromeliads: performances in situ Bromeliaceae offer exceptional opportunity to evaluate the impacts of the CAM and C3 syndromes on plant ® tness, and compare more immediate measures of plant performance, because related species distinguished by these arrangements occasionally co-occur. Plants of both descriptions monitored in situ in Trinidad and Venezuela exhibited similar, modest rates of gas exchange, relatively low osmotic pressures, and, at most, moderate Cambridge Books Online © Cambridge University Press, 2009 152 Carbon and water balance Figure 4.15. Distribution of epiphytic Bromeliaceae by photosynthetic pathway along a regional humidity gradient in Trinidad. Letters along the top of the ® gure indicate habitat type beginning with (a) deciduous seasonal forest, (b) semievergreen seasonal forest, (c) evergreen seasonal forest, (d) lower montane rainforest, (e) upper montane rainforest and (f) subalpine rainforest (after Smith 1989). xylem tensions (Smith et al. 1985, 1986; Griffiths et al. 1986; Lüttge et al. 1986a,b; Tables 4.1± 4.5). Here and there, a CAM species transpired more rapidly than a sympatric C3 type. Diurnal ¯ uctuations in Cleaf were similar except that lows occurred late in the day and during early morning for members of the two groups respectively. Every bromeliad monitored in these studies consumed CO2 slowly (,3 mmol m22 s21), yet H1max sometimes approached unprecedented highs ± up to 625 mol m23 on a sap volume basis for CAM-equipped Aechmea nudicaulis. However, recycled CO2 accounted for more than half of the total (Table 4.4). Water-use efficiency varied among species, the mean transpiration ratios (TR) averaging 42 for nocturnal gas exchange by ® ve CAM species and 99 for diurnal uptake by four C3 taxa, with some overlap (Table 4.3). Performances over 24 h differed less because WUE diminished during Cambridge Books Online © Cambridge University Press, 2009 Table 4.8. Leaf surface characteristics and the effects of surface wetting on CO2 uptake by nine ecologically diverse Bromeliaceae Trichome density (mm22) Species Wettability of leaf surface Ecological type Adaxial Abaxial Adaxial Abaxial % inhibition of CO2 uptake when wet III I IV IV V V V V V 3.3 3.0 57.6 31.3 63.1 59.5 45.4 75.3 58.5 22.5 14.1 31.6 14.1 32.0 28.2 42.7 47.9 36.3 Low Low Low Low High High High High High Low Low Low Low High High High Low Low 0 0 0 0 Almost complete Almost complete Almost complete 0 ~50% Aechmea bracteata Pitcairnia macrochlamys Guzmania monostachia Catopsis nutans Tillandsia paucifolia Tillandsia ionantha Tillandsia tectorum Tillandsia bulbosa Tillandsia butzii Source: After Benzing et al. (1978). Cambridge Books Online © Cambridge University Press, 2009 154 Carbon and water balance Table 4.9. Aspects of leaf morphology of Bromelioideae distributed across three forest types in Trinidad Lower montane rainforest (n510) Evergreen seasonal forest (n56) Deciduous seasonal forest (n53) 0.4260.14 0.7460.31 0.8660.18 Diameter of foliar trichome (mm) Abaxial 71 Adaxial 36 85 87 124 110 Cover by indumentum (%) Abaxial Adaxial 18 19 95 90 1666 0 1964 0 Leaf thickness (mm) 4 1 Stomatal frequency (1 mm2) Abaxial 24613 Adaxial 0 Source: After Smith (1989). phase four among the CAM species, while the C3 types lost less water overnight (Griffiths et al. 1986; Smith 1989). Specimens sampled in Trinidad also demonstrated the sensitivity of CAM (monitored as DH1) to immediate and previous growing conditions, speci® cally, time since the last rainfall and irradiance received the day before. Griffiths et al. (1986) recorded rapid declines in carbon gain after only one to a few rainless days, a ® nding reminiscent of Adams and Martin' s (1986a) observations on Tillandsia deppeana. During one of the night runs, g fell sharply coincident with the arrival of a warmer, drier air mass, further highlighting the capacity of even the well-watered plant to sense, probably via stomata, conditions that might compromise water balance without pre-emptory adjustment (Fig. 4.16). Most of Griffith et al.' s (1986) subjects were growing in relatively moist habitats in the north of Trinidad. Less permissive conditions prevail southward, and here photosynthetic pathways more consistently predicted plant distributions (Fig. 4.15). Segregation by photosynthetic syndrome sharpened along the steep, generally north/south rainfall gradient until only CAM types, and relatively few of these species, inhabited the driest regions. A more compressed ordering of similar nature characterized Bromeliaceae arrayed from lower to upper perches in the wettest montane forests (Fig. 7.11). Only C3 species with lax foliage tolerate deep shade here, while an assemblage composed of Type Three and Four taxa occurred in the more Cambridge Books Online © Cambridge University Press, 2009 CAM vs. C3 bromeliads: performances in situ 155 Figure 4.16. Effects of an abrupt decrease in ambient relative humidity and higher temperature accompanying a weather change in Trinidad on Aechmea aquilega and Aechmea nudicaulis relative to g (conductance) and CO2 exchange. Data points represent averages for both species. Arrow indicates time of weather change (after Griffiths et al. 1986). fully illuminated space higher in the canopy. Type Five species were absent throughout, probably excluded because everwet conditions even in the most exposed microsites preclude the drying necessary to permit sufficient gas exchange through dense, hydrophilic indumenta (Fig. 4.11; Table 4.8). Conceivably, CAM favors survival for some wet-growing populations by providing needed stress-tolerance during the infrequent, severe drought. Still, restriction of a species like Aechmea aripensis (CAM) to an extremely Books Online © Cambridge University Press, 2009 156 Carbon and water balance humid (.6 m year21) habitat on a single mountain seems paradoxical without imputing some other plant bene® t, or an accident of history such as a founder event. Perhaps CAM simply represents a suboptimal, but sustainable, anachronism for A. aripensis under present circumstances. Aechmea, despite its large size and problematic status as a `good' genus (Chapter 9), appears to be CAM-equipped throughout (e.g., Medina and Troughton 1974; Medina 1990). But exceptional ecophysiological versatility also characterizes its membership, and may contribute to the survival of A. aripensis. Recall that A. magdalenae matches extensive, co-occurring C3 ¯ ora in its capacity to subsist on shade-light in the understory of humid Panamanian forest. Bromelioideae further demonstrate the puzzling juxtaposition of CAM and high humidity with additional taxa like A. bromeliifolia, several Billbergia species, Nidularium procerum and Quesnelia quesneliana, all of which occasionally root in continuously sodden soils in Brazilian restinga, sometimes in standing water during the rainy season (Fig. 7.13B). Yet by leaf texture and general structure, these plants look like many other CAMtype bromelioids. CAM has receded somewhat in at least a few populations native to exceptionally dark, moist habitats (e.g., Nidularium innocentii, D524½ ; Medina et al. 1977). Apparently, greater capacity for C3 photosynthesis than usual for Bromelioideae favors at least some of the species that grow under conditions usually associated with this syndrome. Zotz and Andrade (1997) compared Guzmania monostachia and Tillandsia fasciculata to discover why these two wide-ranging bromeliads partition microsites by high vs. more moderate exposure on pond apple trees on Barro Colorado island in the Republic of Panama. Their unusually comprehensive examination of water relations revealed how many dimensions of plant structure and function beyond photosynthetic pathway determine vulnerability to drought. On ® rst glance, thinner, less conspicuously trichomed foliage and C3± CAM status suggest that Guzmania monostachia should desiccate faster during the approximately four-month dry season that prevails at the study site. Indeed, leaf areabased transpiration rate exceeded that of Tillandsia fasciculata by about 15%. Still, drought-tolerance was impressive for both species, especially adults, because additional plant attributes brought the overall water relations of these two epiphytes into closer conformity. Guzmania monostachia and Tillandsia fasciculata both produce tanks, but ratios of plant water content to impoundment capacity for the second species (,1.7) change little after the diameter of the shoot exceeds several centimeters (after impoundment becomes possible), whereas values for Books Online © Cambridge University Press, 2009 CAM vs. C3 bromeliads: performances in situ 157 Figure 4.17. Relationship between leaf area (LA) and plant water content (PWC) as a function of plant size for Guzmania monostachia. Each data point represents a separate plant (after Zotz and Andrade 1997). Guzmania monostachia peak and even fall below 1.0 during the same, relatively vulnerable early part of the life cycle. Values for adults compare closely with those of Tillandsia fasciculata (1.7). Leaf area to plant water content ratio falls sharply over the same interval, demonstrating in part why seedlings experience stress sooner than adults denied irrigation under the same conditions (Fig. 4.17). Water potential and solute or osmotic potential (p) ¯ uctuate in tandem over the year, whereas DH1 indicates continuously more pronounced CAM in T. fasciculata (Fig. 4.18). Although capacitance is substantially higher for T. fasciculata (0.70 vs. 0.30), Guzmania monostachia recovered following more severe desiccation (60 vs. 90%). In the ® nal analysis, plants denied irrigation lost similar amounts of moisture for the ® rst 4± 5 days, and, following stomatal closure, held the remaining stores of water with about equal success. Leaf C never fell below 20.8 MPa. Just as Smith et al. (1985, 1986), among others, discovered among the bromeliads of Trinidad, photosynthetic pathway did not reconcile with all of the other determinants of water relations, i.e., speci® c mechanisms Cambridge Books Online © Cambridge University Press, 2009 158 Carbon and water balance Figure 4.18. Aspects of water relations of Guzmania monostachia and Tillandsia fasciculata. (A,B) Seasonal shifts in the diurnal ¯ uctuation of DH1. (C,D) Seasonal ¯ uctuations in osmotic pressure and water potential (after Zotz and Andrade 1997). Cambridge Books Online © Cambridge University Press, 2009 CAM vs. C3 bromeliads: performances in situ 159 presumed to foster drought-endurance are not always coincident. Two explanations for this inconsistency come to mind: (1) different combinations of plant features yield comparable results (equivalent drought-tolerance), or (2) CAM in such cases provides different plant bene® ts. Guzmania monostachia, whose more desiccation-tolerant foliage transpires relatively rapidly, may rely on CAM primarily to avoid photodamage rather than conserve moisture. Maxwell et al.' s (1992, 1994, 1995) ® ndings clearly support this contention for G. monostachia, although the same bene® ts may apply for Tillandsia fasciculata. Returning to Zotz' s question about distributions on pond apple trees, Tillandsia fasciculata may be better equipped to grow where exposure (drought-stress) is more severe in Panama because its seedlings (but not adults) maintain a lower ratio of leaf area to total water content than Guzmania monostachia (although the latter species has greater relative impoundment capacity as a juvenile). Consequently, losses relative to internal reserves are signi® cantly lower than those experienced by Guzmania monostachia subjected to the same evaporative demand. Zotz' s (1997b) discovery that Amax (and E) increase with plant size further underscores the need to know more about the factors that affect carbon and water balance than photosynthetic pathway to explain why co-occurring Bromeliaceae often distribute differently along environmental gradients. Finally, Zotz and Thomas (1999) modeled Guzmania monostachia and Tillandsia fasciculata to compare how effectively phytotelmata supply plants in lieu of water-absorbing roots. Speci® cally, they inquired whether impoundments deliver year-round or only seasonal supplies of moisture for plant use, and if one of these two bromeliads or a particular life stage bene® ts more than another. Factors that potentially distinguish these two epiphytes on this basis include the characteristics just mentioned and several more that also affect water balance. Actual measurements (e.g., time required for ® lled tanks to dry out) conducted in situ on Barro Colorado island provided data for model validation. Simulations indicated that seedlings should experience longer periods with empty impoundments than adults growing under the same conditions in seasonal Panamanian forest. Additionally, shoots of adult Guzmania monostachia would lack foliar reservoirs for fewer days during the year (total of about one month) and have to endure no more than 12 successive days of such deprivation compared with Tillandsia fasciculata (about two months and 16 days respectively). Success of phytotelm Bromeliaceae facing drought will also depend on additional species-speci® c parameters not included in the model such as sensitivity of stomata to humidity (Fig. Cambridge Books Online © Cambridge University Press, 2009 160 Carbon and water balance Figure 4.19. Effects of abrupt change in chamber air relative humidity (RH) on CO2 exchange by Tillandsia usneoides. Daytime RH was 55%, night-time RH was 92% up to arrow at which time RH was reduced to 72%. RH was than increased to 95% and maintained at this level for the rest of the night (after Martin and Siedow 1981). 4.19), tolerance to desiccation and how rapidly tissues reach lethal water contents after tanks empty. Presumably, the relative performances of these two bromeliads and their life stages shift along environmental gradients (i.e., at locations featuring different patterns of rainfall and evaporative demand). Predictors of photosynthetic capacity (Amax) Land plants, presumably including Bromeliaceae, combine aspects of leaf structure, chemistry, physiology and life span to optimize photosynthesis relative to available photons, water and key nutrients (Farquhar and Sharkey 1982). Coordination is particularly tight between A and g and N content, the nutrient that inordinately in¯ uences Amax (Chapter 5). The interactive effects of environment and plant on A and water use range from transitory (e.g., minutes for stomata to substantially alter g in response to abruptly drier air; Fig. 4.16), to orders of magnitude slower (e.g., improved N nutrition that requires days to weeks to elevate inherent photosynthetic capacity; Fig. 4.6). Heterophylly and deciduousness in¯ uence patterns of water use and carbon gain over weeks to months (Fig. 2.12A). Cambridge Books Online © Cambridge University Press, 2009 Predictors of photosynthetic capacity (Amax) 161 Coordination of the sort imputed by optimization theory maximizes the utilization of plant potential (Amax), while minimizing unnecessary expenditures of water to produce the currency (photosynthate) crucial to ® tness. Mediation by the plant is largely effected through regulation of guard cell physiology, which in turn affects g, such that immediate plant capacity to ® x CO2 is fully engaged, i.e., water use to gain carbon is optimized within certain genotype-speci® c constraints described below. Consequently, WUE changes less under unstable conditions than a less responsive (less capacity for both feed-forward and feed-back regulation of g, hence the CO2 supply to chloroplasts) system would allow (e.g., Fig. 4.10). Because PPFD and VPD in¯ uence A and E, and like many other aspects of the environment they both ¯ uctuate in situ, stomata must continuously adjust to stabilize WUE. Other factors that affect the same two processes, like supplies of moisture and nutrients, change more slowly and accordingly, so do the plant responses that optimize their use to harvest photons. Occasional events, like the abrupt arrival of that drier air mass in northern Trinidad, temporarily override plant propensity to optimize E and maximize A, for a time denying the mesophyll full utilization of its immediate potential to ® x CO2 (Fig. 4.16). Threatening conditions simply take precedence, reducing short-term gains in favor of plant survival. All ¯ ora possess some capacity to optimize water use as growing conditions change, but ecophysiological performances vary within plant-speci® c limits that constitute adaptations to the extent that they match requirements for survival in native habitats. In essence, Amax and WUE represent set points adopted to accommodate routine growing conditions, especially PPFD and water and N supplies. Leaf structure, function and longevity evolve in tandem to achieve rates of resource use (and determine related demands) and A appropriate for speci® c environmental contexts and important aspects of the plant (e.g., shoot/root ratio, type of life history). Values for A that fall at the low end of the range for all ¯ ora, and the opposite for WUE, indicate that the bromeliads routinely fail to encounter the conditions necessary to sustain more vigorous photosynthesis and meet the accompanying elevated demand for water (Tables 4.1, 4.3, 4.4; Fig. 4.7). The inverse relationship between Amax and WUE has major biological consequence. Land-dwelling ¯ ora achieve either high WUE, an outcome that promotes ® tness in physically stressful (usually droughty) habitats, or substantial vigor, a sounder response to the stiff competition fostered by more resource-rich (humid) ecospace. Bromeliaceae generally ® t the ® rst more closely than the second strategy according to leaf morphology in addition to those modest values for A, g and E. Densities of stomata Cambridge Books Online © Cambridge University Press, 2009 162 Carbon and water balance (generally low) parallel the gas exchange data for most of the examined C3 and CAM bromeliads, i.e., they correspond to the modest demands for CO2 at Amax (see summarized data in Martin 1994). Foliar anatomy of course reveals little about g, or whether components other than the guard cells control gas exchange, as some unusually structured stomata prompted Tomlinson (1969) and several earlier morphologists to suggest (Figs. 2.13B,C, 2.17A). Hydration Foliar trichomes in¯ uence leaf energy budgets, light reception and water retention across Bromeliaceae, and for the more specialized species they also mediate mineral nutrition and rehydration. The importance of roots and their cost relative to the shoot shifts accordingly, diminishing as the foliar epidermis becomes increasingly multifunctional and the body plan deviates from the typical arrangement among monocots (Table 4.2). Brie¯ y, bromeliads assigned to Type One and many members of Type Two, Three and Four produce relatively extensive, presumably fully operational root systems. Even the heavily scleri® ed organs of at least some dry-growing Tillandsioideae (e.g., Tillandsia subgenus Diaphoranthema) contain some well-developed water-vascular cells (Cheadle 1955). At the other end of the plant, absorptive scales line the phytotelma of Type Two, Three and Four Bromeliaceae; those serving Type Five Tillandsioideae perform the same tasks over the entire shoot. A few of these most specialized species lack roots as adults (e.g., Tillandsia usneoides, T. duratii; Fig. 2.10L). Information on root function is easily summarized. Ekern' s (1965) inquiry on pineapple demonstrated substantial contributions from soil roots to water balance, and Burt and Benzing (1969) and Nadkarni and Primack (1989) monitored the movements of radionuclides, presumably via the xylem, from potting media to the foliage of several Type Three and Four species. Sieber (1955) determined that feeding through roots enhanced the growth of several ornamental species over that effected by placing the same solutions in phytotelmata. However, only the ® rst of these reports identi® es the volumes delivered, and none addresses the possible occurrence of mechanisms that insulate the aerial roots from dry air without compromising access to typically intermittent moisture supplies. Brighigna et al. (1990) described unusual cell structure and hydrophilic materials in the root caps of two Type Five Tillandsia species, but they suggested importance for anchorage rather than for water balance. Moisture absorbed through the bromeliad leaf may enter by two routes. Cambridge Books Online © Cambridge University Press, 2009 Hydration 163 Benzing et al. (1976), Owen et al. (1988, 1991) and Owen and Thomson (1988) con® rmed trichome involvement in the uptake of several solutes and presumably water as well (Chapter 5). Sakai and Sanford (1979) reported extensive membranes and other suggestive ultrastructure in the stalk cells of the scales located on the leaf bases of pineapple, as did Dolzmann (1964, 1965) and Brighigna et al. (1988) for those of Tillandsia usneoides. However, water may penetrate less specialized regions of the epidermis adjacent to the phytotelmata where cell walls and cuticle thin out. Type Five bromeliads employ trichomes in the way ® rst envisioned by a number of European botanists about a century ago (e.g., Mez 1904). Considerable study followed (e.g., Benzing and Burt 1970; Table 4.10), but consensus on some important details, for example whether the most leafdependent species obtain signi® cant amounts of moisture from vapor in air, came later. Suggested alternatives (Haberlandt 1914, Dolzmann 1964, 1965) aside, a simple osmomechanical mechanism adequately explains how the trichome of dry-growing Tillandsioideae relieves water de® cits (Fig. 2.7A,B). Early investigators, including Mez (1904) and Haberlandt (1914), con® rmed the dual roles the tillandsioid trichome plays in moisture and nutrient absorption by noting the effects of vital dyes and hypertonic solutions placed on intact leaf surfaces. Stalks and adjacent mesophyll stained and plasmolyzed respectively, whereas neighboring epidermal cells remained unaffected. Dense indumenta proved to be especially well suited to magnify the bene® ts of light showers that only brie¯ y moisten shoots. Rather than beading up as occurs on most foliage, drops rapidly spread to wet much more surface. Each trichome within range immediately imbibes enough water to engorge the four large cells dominating the central disc, which in turn bulges upward causing the attached wing to ¯ ex down against the leaf surface (Fig. 2.7A,B). As the indumentum of many a Type Five Tillandsia dries, the central disc in each shield collapses, ¯ exing the wing upward to re-establish the trichome' s full protective powers. Speci® cally, a stout plug comprised of the much thickened outer tangential walls of the four innermost disc cells effectively seals off the underlying dome cell, preventing moisture from wicking up from the mesophyll along its path of entry. The indumentum' s recti® cation of moisture exchange across the epidermis of these drygrowing bromeliads parallels the operation of the root cortices of certain desert terrestrials (Nobel and Sanderson 1984). Trichomes also moderate stress by scattering incident radiation, which in exposed habitats often exceeds plant needs and may impede photosynthesis by inhibiting Cambridge Books Online © Cambridge University Press, 2009 Table 4.10. Desiccation over CaCl2 of leaf discs (midblade) of 13 bromeliads representing all three subfamilies and the five ecological types Subfamilies and species Mean density of trichomes (mm22) Ecological type Mean dry weight of 9 mm discs (mg) III III II II I 13.3 7.9 11.5 20.5 8.9 19.1 15.2 23.9 12.6 26.9 Pitcairnioideae Pitcairnia undulata Pitcairnia macrochlamys I I 5.8 4.3 Tillandsioideae Catopsis berteroniana Guzmania lingulata Tillandsia achyrostachys Tillandsia karwinskyana Tillandsia multicaulis Vriesea carinata IV IV V V IV IV 4.2 3.8 12.0 14.4 5.1 3.0 Bromelioideae Aechmea bracteata Aechmea tillandsioides Ananas comosus Bromelia balansae Cryptanthus acaulis % surface covered by shields Adaxial Abaxial % H2O de® cit after ® ve days over CaCl2 20.4 8.7 43.5 41.3 36.9 15 35 80 35 60 20 20 90 90 85 20.5 34.5 23.8 19.8 37.2 0 0 34.8 43.4 0 0 95 95 29.0 28.1 16.5 17.4 60.9 82.6 36.9 24.0 18.7 20.9 40.0 63.0 40.5 21.9 2 4 95 95 4 4 2 4 80 60 4 4 41.5 47.9 10.2 18.2 50.6 40.0 Adaxial Abaxial Source: After Benzing and Burt (1970). Cambridge Books Online © Cambridge University Press, 2009 Hydration 165 Figure 4.20. Partial desiccation and rehydration of excised leaves of Tillandsia streptophylla during exposure to water-saturated air and liquid moisture (after Benzing and Pridgeon 1983). photosystem II. In summary, trichomes of the most specialized Bromeliaceae act as one-way hydraulic valves and energy dissipaters, alternately charging the plant with moisture and insulating it against avoidable drying and excessive insolation. Leaf blades of Tillandsia streptophylla (Type Five) maintained for three days over CaCl2 desiccated about 30%, and then gained no weight during a fourth day in a moisture-saturated atmosphere (Fig. 4.20). However, the same samples fully recovered in a water bath within a few more hours. During another run, Catopsis nutans, a soft-leafed, sparsely trichomed Type Four species, lost more weight and failed to rehydrate appreciably while surface-moistened with mist (Fig. 4.21). Additional bromeliads treated identically performed similarly, recharging within 12 h in liquid water only if members of Type Five (e.g., Tillandsia schiedeana; Fig. 4.21; see also Benzing and Burt 1970). Studies in Puerto Rico on Tillandsia recurvata and T. usneoides demonstrated that these two wide-ranging species could also rapidly eliminate substantial de® cits incurred over several weeks of rainless weather (Biebl 1964). More information on moisture exchange, some of it contradictory, exists for Tillandsia usneoides than for any other bromeliad. Penfound and Deiler Cambridge Books Online © Cambridge University Press, 2009 166 Carbon and water balance Figure 4.21. Partial desiccation and rehydration of excised leaves of Tillandsia schiedeana and Catopsis nutans (after Benzing and Burt 1970). (1947) noted sizable and precipitous weight changes in plants monitored in a closed room in southern Louisiana (Fig. 4.12). Humid air promoted substantial rehydration (discussed in more detail later). Martin and Schmitt (1989) examined Spanish moss for an entire year in North Carolina where, during the course of a 20-day drought, trusses lost about three-quarters of the water present at the beginning of that episode. Similar performances by T. ionantha prompted Benzing and Dahle (1971) to suggest that some Type Five Tillandsia approach poikilohydrous status on three counts. Like true resurrection plants, these bromeliads also rehydrate across intact leaf surfaces, lose moisture rather quickly to dry air, and tolerate severe desiccation, although not the 95± 97% de® cits routinely experienced by many bryophytes and some ferns. However, recall that Cleaf remains high and the chlorenchyma turgid, probably even in severely dehydrated specimens, features foreign to poikilohydry, but expected of succulents. Picado (1913) proposed that Type Five bromeliads make unusually effective use of water vapor, a claim subsequently supported and challenged. Trusses of Spanish moss changed weight in concert with shifting relative humidity (RH) during crude experiments performed by Penfound Cambridge Books Online © Cambridge University Press, 2009 Hydration 167 Figure 4.22. Mean changes in tissure water content for ® ve clumps of Tillandsia usneoides held for three days at 30 °C, 50% RH during the photoperiod and 30 °C, 90% RH at night (after Martin and Schmitt 1989). and Deiler (1947; Fig. 4.12). Another set of plants droughted to an average 222% water content had rehydrated to 506% after two days in humid air (.90% RH). According to De Santo et al. (1976), several Type Five bromeliads among the 10 they tested absorbed moisture from drying atmospheres (,35% RH), but weight gains were minor compared with those reported by Penfound and Deiler. Species from relatively arid habitats (e.g., Tillandsia schiedeana) hydrated most during the 12-h runs, while relatives from wetter forest (T. flabellata) weighed no more after treatment than before. De Santo et al. erroneously concluded that `moisture is captured by the dead (shield) cells and taken into the mesophyll through the living stalk cells' . Martin and Schmitt (1989) conducted the most de® nitive study of moisture exchange to end the controversy. Figure 4.22 illustrates how clumps of T. usneoides responded during threeday runs as RH oscillated between 30 and 90%, and air temperature remained at 30 °C (Martin and Schmitt 1989). Fluctuating plant weights accord with the presence of a proportionally (relative to the whole plant) small volume of strongly hygroscopic tissue. Findings by De Santo et al. (1976) had also suggested that the shoots of Spanish moss consist of two compartments that differ in size and affinity for water vapor in adjacent air. Uptake was modest because only the indumentum possesses sufficient hygroscopic power to draw water from a subsaturated atmosphere, whereas the mesophyll accounts for Cambridge Books Online © Cambridge University Press, 2009 168 Carbon and water balance most of the leaf volume. Recall that succulents, including the thick-leafed bromeliads, maintain Cleaf at .21.0 MPa, whereas water at 99% RH and 20± 30 °C in air, for example, exists at values many fold lower. Osmotic potentials determined for diverse Type Four and Five Tillandsioideae never fell much below 21.0 MPa, and most readings were closer to half that value (Harris 1918; Biebl 1964; Smith et al. 1986; Fig. 4.18). Water vapor surely promotes bromeliad water balance, but without condensation only by slowing E as for other land ¯ ora. Withhold irrigation and death inevitably follows. Tillandsia usneoides died after being shielded from rain for four months even while suspended above the ¯ oor of a swamp forest (Garth 1964). Some of Penfound and Deiler' s (1947) results either contradict physical principles or vapor condensed unnoticed on shoots during the runs (Fig. 4.12). Even so, declining weights indicated that longer experiments would likely have ended as Garth' s did. CAM reconsidered as an evolutionary response to stress Data on CAM and related ecology for Bromeliaceae may exceed that available for any other family; nevertheless, a full accounting of its signi® cance to these plants remains elusive. As indicated above, especially vexing is the frequent association of CAM with everwet climate, or, where precipitation is more seasonal, with phytotelmata sufficient to provide plants continuous access to water. Could the bene® ts of CAM for these bromeliads still relate primarily to water balance and only apply during the exceptional 10, 25 or 50-year drought with negligible consequences for ® tness the rest of the time? Perhaps the unpredictability of the moisture and photon supplies in and under the forest canopy account for the overoccurrence of this syndrome among many Bromeliaceae. Maxwell et al. (1992, 1994, 1995) demonstrated that the advantages of avoiding photoinhibition probably explain the presence of facultative CAM in Guzmania monostachia, but what about the importance of this syndrome relative to other dimensions of the light environment? CAM may enhance plant access to energy delivered in sun ¯ ecks by relaxing the requirements for stomatal control that some C3-equipped understory ¯ ora employ to coordinate CO2 supply with brief surges in PPFD. Abruptly ¯ uctuating irradiance might have less impact on WUE if CO2 supplied from stored malic acid allows g to remain continuously low through much of the photoperiod (see also Skillman and Winter 1997). Rehydration is another possibility considered later. Finally, might the mix of CAM-mediated plant responses and the result- Cambridge Books Online © Cambridge University Press, 2009 CAM reconsidered as an evolutionary response to stress 169 ing bene® ts routinely shift over the year for many Bromeliaceae? Below, we review one more set of studies to consider some mechanisms that help certain CAM, and perhaps certain C3, bromeliads coordinate diverse signals from the environment and physiological status to minimize plant stress and maximize resource-use efficiency as growing conditions change. Carbon recycling lies at the heart of this issue. CAM-idling and recycling Current wisdom assigns CAM in its idling mode importance as a stressmitigating, maintenance mechanism ± a state of quiescence rather than dormancy ± that keeps the mesophyll primed for renewed opportunity, viz. renewed water supply. As stomata close while desiccation progresses and E falls, idling ensues and CAM, now totally dependent on respired CO2, hence diminished overall, continues to provide enough energy to avoid death, or the need to lapse into an inactive condition from which recovery would be slow. Idling also maintains a CO2 source to help protect the lightharvesting apparatus (Maxwell et al. 1992, 1994, 1995). Photosynthesis for the CAM-idling plant returns to pre-stress levels within hours to a day or two following return to wet weather, much faster than possible for a drought-avoiding shrub, or a similarly deciduous Pitcairnia that likewise must ® rst regenerate its canopy. One of the most perplexing facets of CAM concerns certain details of carbon management, particularly the continued prominence of CO2 recycling in recovered foliage and sometimes leaves that never experienced severe drought. Why do apparently well-watered subjects so often depend so heavily on recycled CO2 compared with inputs from the atmosphere? According to Martin (1994), CAM-idling in the strictest sense (recycled CO2 accounts for 100% DH1) has never been recorded for a bromeliad (but see Stiles and Martin 1996). However, many plants examined in situ were processing much more carbon than could be accounted for by gas exchange, and often well above what seemed necessary. Griffiths et al. (1986) and Griffiths (1988) reported values ranging from 50 to 99%, sometimes even during fairly wet weather, for the diverse taxa included in their survey in Trinidad (Table 4.4). Thoroughly irrigated subjects exposed to a variety of growing conditions have often behaved similarly. Why were these plants gaining carbon so feebly compared with respiration, even while seemingly unstressed? Before trying to answer this question, we need to revisit the relationship between carbon and water in the context of CAM. Cambridge Books Online © Cambridge University Press, 2009 170 Carbon and water balance The algorithm used to calculate fractions of DH1 attributable to recycled carbon vs. exogenous CO2 employs the stoichiometry of two titratable H1: one malic acid molecule: one CO2 molecule. Internally generated CO2 represents the balance remaining after subtracting from DH1 the amount from outside according to measured gas exchange. Whatever other bene® ts accompany CAM, its capacity to recycle dark-respired CO2, whether in or out of the idling mode, promotes WUE in the same way described earlier for the CAM-cycler (see also Martin et al. 1988). Fetene and Lüttge (1991) proposed using the ratio of moisture saved through recycling to transpiration, which, by substituting A/E for WUE, was reduced to estimate the advantages of CAM to water balance in Bromelia humilis. Bromelia humilis also demonstrated how cues related to growing conditions affect the carbon budget of a CAM bromeliad. More precisely, Fetene and Lüttge illustrated why well-watered plants sometimes rely so heavily on respired compared with exogenous CO2 ± why, despite adequate irrigation, they so readily reduce g. Respiration always assures some recycling, but only enough to account for a small fraction of H1max while CAM remains robust, i.e., while CO2 reaching PEPc from the atmosphere greatly exceeds the supply from mitochondria. The acidity assignable to recycling by nonstressed CAM types should approximate DH1 in an otherwise comparable CAM-cycler because the stomata of the cycler close at night, trapping endogenous CO2 for nonautotrophic re® xation. Recall that Fetene et al. (1990) and Fetene and Lüttge (1991) demonstrated how well-fertilized and N-deprived B. humilis pretreated at two exposures gained different amounts of carbon under the same high and low PPFDs (Figs. 4.5, 4.6). They also manipulated drought-stress, leaf-to-air VPD and night-time temperature to note shifting dependency on recycled vs. exogenous CO2. While these data provide no de® nitive answers, they suggest some intriguing possibilities and underscore the complexity of the CAM syndrome and its sometimes dubious utility as an indicator of ecophysiological status (prevailing plant stress). More importantly, their manipulations demonstrate how disparate aspects of the environment, both historic and immediate, can affect the operation of CAM. Bromelia humilis, at least, shifts toward CAM-idling whether challenged by N-de® ciency, drought, high temperature or steep VPD in night air, i.e., the same plant response reduces vulnerability to several threats to water balance and perhaps other essential processes like light harvest. High temperatures at night (.30 °C) shifted 24-h carbon budgets closer to or into negative territory, and more substantially for 2N than for 1N plants. Relative reliance on recycled CO2 for phase one increased apace Cambridge Books Online © Cambridge University Press, 2009 CAM reconsidered as an evolutionary response to stress 171 until at 35 °C, contributions from respiration about equaled the amounts of CO2 derived from the atmosphere at 20 °C (30 mmol m22 12 h21 for 1N plants and 18 mmol CO2 m22 12 h21 for subjects grown on unamended soil). CO2 uptake responded more sensitively to temperature than DH1, and about equal amounts of acid accumulated at 35 °C and 20 °C, perhaps due to an unusually high temperature coefficient (Q10) for dark respiration as Lüttge and Ball (1987) noted for some other CAM bromeliads. Regardless of pretreatment, dark respiration rose exponentially with temperature, but more in 2N plants than in better-nourished specimens. Temperature coefficients ranged between 2.3 and 3.0 (10± 25 °C) with 2N plants, once again exceeding the responses of better-fertilized subjects. The more elevated of the two applied VPDs (7.46 Pa KPa21 vs. 15.49 Pa KPa21) reduced CO2 uptake to different degrees depending on pretreatment. Reliance on recycling was about 2± 6-fold greater at the higher VPD. Water saved by recycling as equivalents of E (ratio of recycled CO2 to net nocturnal CO2 assimilation) increased from 0.08 to 0.60 at the lower VPD and from 0.80 to 3.0 at the higher one. Thus, at the lower VPD, amounts of water equal to only 8± 60% more than the total transpired remained unexpended, while savings in drier air rose to 80± 300%. Less moisture-saturated air had reduced g to just 5± 23% of that prevailing when the more humid atmosphere had threatened plant water status less. Ten days of drought almost shut down CO2 uptake and reduced H1max to just 30± 40% of prestress levels. However, recycling increased proportionally from about 25± 35% to near 100%. By day 10, the water saved by recycling amounted to 2± 6-fold the quantity that plants had transpired. After 12 days, CO2 uptake almost ceased, indicating that the protection afforded by reducing g developed much faster here than for some nonbromeliads. Agave deserti, Opuntia ficus-idea and Ferrocactus acanthoides required 11± 20 days just to reduce initial rates of CO2 uptake 50% in another study (Nobel 1988). Apparently, CAM serves Bromelia humilis quite well in its highly seasonal, hot and probably often infertile habitats. Rather than an all-or-nothing response, proportional reliance on endogenous CO2 waxes and wanes with ¯ uctuations in several aspects of the environment capable of reducing water economy and growth and, if severe enough, of in¯ icting serious plant injury. Sensitivity, if greater here than usual, would mean that this bromeliad anticipates threatening conditions sooner than some other CAM types. Or modest capacitance may simply permit water-stress to develop faster for Bromelia humilis compared with these other xerophytes when subjected to comparable droughts. Viewed either way, an unusually sensitive response to conditions that can suppress phase one of CAM (e.g., Cambridge Books Online © Cambridge University Press, 2009 172 Carbon and water balance high temperature, poor nutrition) or accelerate E (e.g., high temperature, high VPD) mitigates the liability imposed by the low capacitance (for a CAM plant) of this bromeliad, thus promoting its tolerance to diverse kinds of stress. Sensitivity to VPD allows plants to reduce nonproductive water use, but do responses vary among Bromeliaceae according to other plant characteristics that in¯ uence vulnerability to drought? Dry air reduces g for Tillandsia usneoides (see Lange and Medina 1979; Fig. 4.19), but species with substantial capacity to replace losses from large phytotelmata (e.g., Aechmea nudicaulis, A. aquilega) behave the same way, although perhaps less sensitively. Light constitutes another agency that effects rapid changes in g and accordingly, shifts relative dependence on atmospheric vs. respired CO2 among CAM bromeliads. Tillandsia usneoides recycled proportionally more carbon after transfer to higher PPFD (Martin et al. 1986), whereas Aechmea nudicaulis did so upon relocation into shade (Griffiths et al. 1986). Finally, chronic, pronounced recycling need not seriously limit growth. Cultivated Ananas comosus rivals some C3 crops for the production of dry matter, yet recycling accounted for 45% of DH1 in one analysis involving irrigated specimens (Sale and Neales 1980). So it seems that a variety of chronic and more transitory stresses, including excessive temperature, nutrient scarcity and suboptimal VPD, promote heavy dependence on recycled CO2 in at least some CAM bromeliads. Signi® cantly, all of these challenges from the environment in¯ uence plant water economy and carbon budgets at least indirectly (e.g., N status through its effects on A). Still, drought often appears to act most decisively, although unevenly according to several studies on Type Five Bromeliaceae in the laboratory and ® eld (see Fig. 4.15 for strong circumstantial evidence). Several of these investigations indicate how drought probably affects recycling for the more notably stress-tolerant epiphytes. Whereas less than 50% of the titratable acidity present in the wellwatered shoots of usually arboreal Tillandsia schiedeana (Type Five) at dawn had come from recycled CO2, 30 days without irrigation in a growth chamber boosted that ® gure to about 90% (Martin and Adams 1987). Recycling varied more over the year on an absolute than on a proportional basis in Tillandsia flexuosa growing in one of its semiarid coastal Venezuelan habitats (Griffiths et al. 1989). Recycled carbon accounted for 76 vs. 73% of the total acid synthesized from mid-wet to mid-dry season, although H1max diminished 35% as aridity intensi® ed. Exogenous CO2 accounted for only 1% of the malic acid accumulated by terrestrial Bromelia plumieri in Trinidad (Griffiths et al. 1986; Table 4.4). Neither Books Online © Cambridge University Press, 2009 CAM reconsidered as an evolutionary response to stress 173 moisture-stress nor nutritional de® ciencies were mentioned, but readings date from February and March, two especially arid months (,25 mm precipitation) at this strongly seasonal site. On that occasion, no rain had fallen for several weeks. Thicker-leafed Aechmea fendleri recycled proportionally more CO2 than A. nudicaulis, suggesting that succulence may elevate endogenous CO2 enough to simulate a stress symptom in a relatively well-hydrated subject. However, Griffiths (1988) considered hydrenchyma too inert to account for the difference, and Lüttge and Ball (1987) supported his assessment with data from additional species. While achlorophyllous storage tissue comprised 60± 75% of the mesophyll of Hechtia glomerata, it accounted for only 9.5% of the CO2 available for recycling. The presence of exceptionally active tissue constitutes another possibility that has some support. Several CAM plants endemic to warm habitats, including three bromeliads (Aechmea fasciata, Ananas comosus, Hechtia glomerata), exhibited dark respiration with Q10s that ranged from 2.13 to 4.09 between 10 and 30 °C. Additionally, all CAM plants require ATP to mediate the massive traffic in malate across the tonoplast. Finally, a biochemical peculiarity also in¯ uences how much endogenous CO2 certain CAM plants produce compared with others. Bromeliads may stand out because they consume free hexose, which assures relatively high rates of respiration, rather than the glucans many other CAM plants (e.g., Kalanchoe) employ to drive phase one. Loeschen et al. (1993) used 12 dry-growing Tillandsia species to rule out nongreen tissue as a major source of CO2 for phase one. Recycling did not correlate with leaf anatomy; in fact only one subject, T. schiedeana, deviated substantially from the 1:1 ratio mandated by the stoichiometry of malic acid production during CAM (Fig. 4.13). Tillandsia schiedeana alone acidi® ed beyond what could be explained by the consumption of exogenous CO2. Several other taxa yielded values above one, but less than two. Note that T. schiedeana (30% water-storage tissue) occurs about midway within the range (0± 53%) exhibited by their sampling. Tillandsia usneoides produced a modestly positive value despite its undifferentiated mesophyll (Fig. 2.10A), while T. valenzuelana (53%) synthesized less acid than gas exchange predicted. Different degrees of stress supposedly accounted for the mixed results, and this explanation is plausible given the single pretreatment provided to all 12 of these ecologically diverse species. Additional inquiry might pro® tably focus on the functions of foliage with anatomically uniform vs. dimorphic mesophyll. Perhaps drought depresses A in a less precipitous fashion among subjects with the second compared with the ® rst type of leaf structure, i.e., one or the other kind of Cambridge Books Online © Cambridge University Press, 2009 174 Carbon and water balance plant reduces g sooner as water de® cits develop. Green cells in T. usneoides presumably lose turgor faster when plants are subjected to drought than those of species featuring hydraulic coupling to collapsible, water-storing parenchyma. Moreover, architectural constraints may oblige a speci® c type of leaf anatomy even if another option would grant superior drought-performance. Perhaps the leaves of Spanish moss are simply too small to support a division of labor between water storage and photosynthesis. Finally, what other physiological or structural peculiarities co-occur with an anatomically undifferentiated mesophyll? And what about those exceptionally thin epidermal layers and delicate cuticles illustrated in Fig. 2.10? Citric acid: its role in ecophysiology Citric acid affects a variety of cellular processes central to the ecology of some (e.g., Clusia species; Franco et al. 1992) and perhaps many CAM plants. Substantial amounts of this tricarboxylic acid augment malic acid to account for DH1 in certain species ± 30± 50% of H1max depending on N supply and PPFD in a study using Bromelia humilis (Lee et al. 1989). These two acids share several and differ in other qualities that in¯ uence carbon and water balance and stress-tolerance. Malic, but not citric, acid synthesis can effect carbon gain and heighten WUE (Lüttge 1988; Franco et al. 1992). However, citric acid synthesis does recycle carbon, and, like malic acid, it increases p (although only one-half as much) if starch or some other polymer constitutes the substrate (Lüttge 1987, 1988). Citric exceeds malic acid for capacity to supply CO2 to chlorenchyma (three vs. one CO2 per molecule) during phase three when overexposure and stomatal closure maximize potential for photoinhibition (see below). Citric compared with malic acid synthesis also yields substantially more reducing power (Lüttge 1988). Finally, citric acid endows vacuoles with high buffering capacity, a fact that may explain why some Bromeliaceae (e.g., Aechmea nudicaulis, as discussed below) exhibit such vigorous CAM despite distinctly nonsucculent foliage. CAM and hydration Bromeliads account for some of the highest DH1 values recorded, but rarely were the responsible solutes (citric vs. malic acid) identi® ed (Table 4.4), obscuring potential consequences for photoprotection, energetics and hydration. On this last point, CAM purportedly allows succulent Senecio medley-woodii (Asteraceae) to access additional moisture from desert soils Cambridge Books Online © Cambridge University Press, 2009 CAM and hydration 175 (Ruess and Ellers 1985). Martin (1994) sought evidence of coupling between hydration and Cleaf in Tillandsia ionantha, as did Schmidt and Blank (see Lüttge 1987, 1988) using T. recurvata, neither attempt having notable success. Findings on other species make a somewhat better case. Aechmea nudicaulis in Trinidad exhibited a mean nocturnal depression of 0.52 MPa associated with an H1max of up to 625 mol m23. Xylem tension followed, as did hydration, both increasing as phase one progressed. Aechmea aquilega provided the most compelling evidence for CAMdriven hydration. Xylem tension had diminished from its maximum night value of 20.54 to 20.24 MPa at dawn, dew having formed on foliage between these two records. However, C3 Vriesea amazonica exhibited a similar response under the same conditions. Interpretation is further complicated by a second reality: both species possess phytotelmata that storms probably topped up shortly before the readings were taken (Griffiths et al. 1986). Finally, lower densities of trichomes on the blades compared with the bases indicate less capacity for hydration by the ® rst compared with the second route. Osmotic potentials that peak early in the day should enhance water balance most for the Type Five species, and under certain circumstances perhaps prove decisive for survival. Tillandsia paleacea, T. purpurea and T. werdermannii native to the hyperarid (except in El Niño years) coastal deserts of Peru seem especially well situated to bene® t from daily ¯ uctuations in p driven by CAM. Heavy nightly mists (`garuas' ) off the Paci® c Ocean probably insure adequate hydration between May and October. The rest of the year, dew, which quickly evaporates in the early morning sun, must suffice. Conditions elsewhere may grant CAM similar importance and account for some of that paradoxical anatomy among generally xeromorphic Type Five bromeliads. Finely dissected shoots and attenuated trichome shields (Fig. 2.8C) could make contacts with moisture that pass too quickly to sustain less specialized ¯ ora adequate to support T. tectorum and its kind. In quite another ecological context, Type One bromeliads with deeperrooted neighbors may bene® t from CAM-dependent p if water drawn up to shallower soil horizons nightly by hydraulic lift arrives in timely fashion. Species equipped with large phytotelma, and those that root in continuously moist soil, should bene® t less from the osmotic consequences of DH1. Cambridge Books Online © Cambridge University Press, 2009 176 Carbon and water balance Additional aspects of light relations We need to revisit the subject of how bromeliads respond to high and low light to round out this discussion of ecophysiology. Undiminished PPFDs from sea level to above 4000 m, to the much attenuated ¯ ux that penetrates to the ¯ oor of dense evergreen forest, sustain numerous Bromeliaceae. Aspects of leaf and shoot structure, pigments and physiology, and leaf life span parallel these contrasting energy supplies. Species with the thickest foliage and most compact rosettes, i.e., those that operate with the highest leaf area indices (e.g., Hechtia, Dyckia; Fig. 2.2B) typically experience the highest exposures and quite often also the potentially complicating effects of drought. Indeed, these plants often grow more vigorously in partial shade. Overexposed, the more sensitive types become chlorotic, grow feebly and may not reproduce. Xeromorphic Bromelia humilis reportedly exists in a chronically photoinhibited state in the sunniest microsites within certain habitats in northern Venezuela (Medina et al. 1986). Exceptionally ¯ exible Guzmania monostachia, with its much more lightly constructed foliage, demonstrated signi® cant inhibition in Trinidad, but escaped the more serious, longer-term impairment of the light-harvesting apparatus that drought-stress combined with high exposure can in¯ ict (Maxwell et al. 1992, 1994, 1995). In fact, this species, more than any other bromeliad, has demonstrated how plants can reduce photodamage that would diminish carbon gain in environments characterized by shifting PPFD. Plant features that promote carbon gain at either low or high ¯ uences fall into two categories: structural (relatively static) and chemical/physiological (more dynamic). Widely tolerant Spanish moss showed little adjustment by either route as noted above. Other taxa exhibited expected patterns, sometimes with additional responses peculiar to Bromeliaceae. According to Dimmitt (1985), Tillandsia caput-medusae from a highly exposed, Sonoran Desert site bears a denser and more re¯ ective indumentum than a second set of plants collected from wetter, forested locations. Cultivated side by side for several years, the sun-adapted stock continued to produce its better-shielded foliage. High and low-light phenotypes rather than ecotypes of many more species stand out at a glance. In addition to a more glabrous surface, thinner, laxer blades characterize the foliage of specimens acclimated to shade. Typically broader leaves with shorter, but stiffer more upright, blades probably enhance drought-tolerance in addition to reducing exposure to direct-beam light by increasing tank capacity relative to shoot Books Online © Cambridge University Press, 2009 Additional aspects of light relations 177 volume for the sun-grown phytotelm bromeliad like Tillandsia utriculata (Fig. 4.23B,C). Adjustments to PPFD can also involve reorganization of the chloroplast. Bromelia humilis cultivated under 20± 30 compared with 700± 800 mmol m22 s21 PPFD developed thicker granal stacks in addition to higher concentrations of chlorophyll and exhibited lower light compensation intensities. However, light response curves proved less sensitive to preconditioning, especially for those subjects provided supplemental N. Zeaxanthin, a component of the xanthophyll cycle, occurred more abundantly in the foliage of the high-light plants (Fetene et al. 1990). Aechmea magdalenae indicated that Bromeliaceae bene® t from the same mechanism mediated by carotenoids that reduces photo-oxidative damage in other plants stressed by supersaturating PPFD (Koniger et al. 1995; see also Skillman and Winter 1997). The size of the xanthophyll-cycle pool (267.2 mmol mol21 chlorophyll) approximated those determined for the 12 C3-type species assayed in the same understory habitat in Panama. Several-fold higher ratios of carotenoids to chlorophyll in the foliage of adjacent gapphase and canopy trees indicated weaker capacity to accommodate high PPFD and likely shade-plant status for this Type Two bromeliad and the cooccurring herbs. However, Aechmea magdalenae alone among the compared understory species deviated from the usual correlation between Amax and xanthophyll pool size because of its capacity to ® x CO2 at twofold and higher rates on a leaf area basis compared with those companion C3 plants. Zeaxanthin (X) typically accumulates in light-saturated foliage at the expense of violaxanthin (V) and antheroxanthin in a partially characterized mechanism involving the transthylakoid pH gradient that dissipates excess excitation energy. Subsequent darkening allows zeaxanthin-epoxidation and organ recovery, although more slowly in Aechmea magdalenae than violaxanthin-de-epoxidation according to assays conducted on the same plants examined for Amax. Expoxidation state (EPS) (the % V to X conversion) indicated the extent to which Koniger et al.' s subjects were exposed to excess irradiance and could engage the xanthophyll cycle to avoid photoinhibition. Values of the ratio of variable to maximum ¯ uorescence (Fv/Fm) consistently above 0.84 further indicated no sustained impairment among the A. magdalenae specimens monitored over several sunny days. Many of their readings followed exposure to prolonged sun ¯ ecks that constitute the primary source of energy for this understory species. Consistent with performances recorded elsewhere (e.g., Medina et al. 1977), Guzmania monostachia responded differently on several counts at different points across steep light gradients (Maxwell et al. 1992, 1994, Cambridge Books Online © Cambridge University Press, 2009 178 Carbon and water balance Figure 4.23. Plant responses to water and light. (A) Unidenti® ed Tillandsioideae on fallen tree illustrating leaf injury resulting from the consequent abrupt change in exposure to light. (B) Shade-grown Tillandsia utriculata. (C) Tillandsia utriculata at the same site exposed to stronger light. (D) Tillandsia kurt-horstii on granite outcrop in Bahia State, Brazil. (E) Tillandsia kurt-horstii illustrating substantial surface area available to intercept fog water provided by its dissected shoot and dense indumentum of trichomes with elongated shields. (F) Abaxial leaf surface of Tillandsia bulbosa (3150). (G) Adaxial leaf surface of Tillandsia bulbosa (3150). (H) Portion of trichome shield of Tillandsia karwinskyana illustrating its rough light-re¯ ecting texture (3225). (I) Adaxial leaf surface of Brocchinia micrantha illustrating the large crystal present in each epidermal cell (3150). Cambridge Books Online © Cambridge University Press, 2009 Additional aspects of light relations 179 1995). Pronounced nocturnal acidi® cation characterized the more light and drought-stressed plants perched in the lea¯ ess crowns of seasonally deciduous trees, but not so some other, more shaded specimens rooted on nearby evergreen supports (Fig. 4.8). Foliage also differed by chlorophyll content, a/b ratio, blade thickness, proportional volumes of chlorenchyma vs. colorless hypodermis, and some revealing physiology (Table 4.6). Darkadapted (10 min) chlorophyll ¯ uorescence and O2 evolution relative to PPFD provided measures of the photochemical efficiency of photosystem II and an important response to high light. Quantum yields routinely diminished through the day as did Fv/Fm (0.70± 0.42) in fully exposed plants compared with individuals maintained in 40% shade (Fig. 4.24). Recovery for both treatment groups began after midday as PPFD started to fall, and ® nished by late afternoon, indicating no damage of the type that requires more time to repair (Long et al. 1994). CAM that persisted in well-hydrated, unscreened subjects, but disappeared in shade, further demonstrated the probable photoprotective function of a mechanism that supplies CO2 to irradiated mesophyll while stomata are closed. Too little capacitance or impoundment capacity in leaf bases prevails even if coupled to high WUE and a typically (for Bromeliaceae) modest Amax (Table 4.1) to allow G. monostachia growing in either sun or partial shade to prolong carbon gain through the dry season. Maxwell et al. (1992, 1994, 1995) concluded that a dual mechanism helps maintain the integrity of the photon-harvesting system of G. monostachia. In addition to light or drought-induced CAM, sufficient photochemical capacity mediated by the xanthophyll cycle exists to down-regulate photosystem II activity. In one set of runs, CO2 regenerated from malic acid synthesized the previous night provided 24% of the carbon reprocessed via RuBPc/o; the same source dominated in droughted specimens subjected to the same high PPFD because net CO2 uptake had diminished 87%. Paralleling CAM was exceptional (compared with many C3 species) capacity for radiation-less dissipation of excess excitation energy at potentially incapacitating exposures. Figure 4.25 illustrates how de-epoxidation of the xanthophyll pool constituents proceeded over the course of two days of contrasting integrated PPFD during the rainy season in Trinidad. Note that maximum conversion prevailed at midday, was more pronounced in unshaded specimens, and reached highest values during the brighter of the two days. Plants moved between sites featuring 30 or 100% exposure rapidly adjusted chlorophyll contents (Fig. 4.26), suggesting that acclimatization also involved the loss and gain of photosynthetic units. After just ® ve days in full sun, chlorophyll Cambridge Books Online © Cambridge University Press, 2009 180 Carbon and water balance Figure 4.24. Diurnal variation in photosystem ¯ uorescence characteristics (FV/Fm) of Guzmania monostachia under natural conditions in Trinidad. (A) Plants growing in exposed microsites under contrasting daily PAR (s, 43.7 mol photons m22; d, 24.3 mol photons m22). (B) Semiexposed plants under contrasting daily PAR (s, 26.2 mol photons m22; d,14.6 mol photons m22). CAM activity preceding the dark periods averaged 65 and 48 mol H1 m23 (after Maxwell et al. 1992). Cambridge Books Online © Cambridge University Press, 2009 Additional aspects of light relations 181 Figure 4.25. Zeaxanthin content as a percentage of violaxanthin, antheraxanthin and xeaxanthin present for Guzmania monostachia at intervals over two contrasting days during the rainy season. (A) On the ® rst day total PPFD was 17.4 and 7.1 mol photons m22 day21 for the exposed and semiexposed populations respectively. Values for the second day (B) were 33.3 (exposed) and 15.7 (semiexposed) mol photons m22 day21. Open circles indicate plants that received more photons; ® lled circles are plants that received fewer photons (after Maxwell et al. 1995). Figure 4.26. Total chlorophyll content over ® ve days following transfer of Guzmania monostachia plants between 100 and 30% PPFD (after Maxwell et al. 1995). Cambridge Books Online © Cambridge University Press, 2009 182 Carbon and water balance content had fallen to less than 50% of its former value. The opposite treatment indicated that this process is reversible. In summary, seasonal variation in PPFD rapidly promotes two contrasting conditions of the photosynthetic apparatus of Guzmania monostachia that together enhance carbon gain by long-lived foliage with generally low inherent capacity to ® x CO2. Sink strength is maintained relatively high throughout the lower-energy photoperiods of the wet season. More depressed (down-regulated) photosystem II efficiency coupled with more active CAM in turn compensates for higher PPFD and greater droughtstress (reduced access to CO2) during the drier months. However, Fig. 4.27 illustrates that exposed plants still gained more carbon during the dry season if partially shaded, whereas fully and semiexposed plants continuously perform at about the same level. Additional species with comparable habits and habitats probably operate like Guzmania monostachia did in situ and Tillandsia deppeana in the laboratory (Adams and Martin 1986a). Epiphytic Vriesea platynema (Type Four) exhibited suggestive behavior during extreme drought ± essentially no gas exchange ± less than half way through the dry season on unshaded perches in a northern Venezuela cloud forest (personal observation). Shoots were performing C3-type photosynthesis 2± 3 days after their previously dry phytotelma had been re® lled. A second group of nonphotosynthetic pigments plays a more conspicuous role in the light relations of bromeliads. High exposure routinely induces extraordinary accumulations of anthocyanins, primarily in the epidermis, presumably to screen solar radiation in the absence of a con¯ uent, photon-scattering indumentum. Permanently red to maroon or green foliage differentiates other populations regardless of PPFD (e.g., certain forms of Tillandsia capitata, T. flabellata). These same pigments provide additional, better-known services as attractants for pollinators and seed dispersers. Involvements in mineral nutrition and shade-tolerance constitute additional possibilities worth consideration. Intricately dissected, ® xed patterns of chlorophyll and anthocyanins displayed by some Type Four species (e.g., Guzmania zahnii, Vriesea fosteriana; Figs. 2.14G, 2.17B, 2.18B) suggest several functions, possibly including enhancements of photosynthesis. Leaves in some cases feature species-speci® c arrays of chlorophyll-rich zones (shutters) and chlorophyll-poor regions (windows or fenestrae), sometimes with densely cyanic epidermis located below each shutter. Nitrogen and phosphorus contents that also distinguish shutters and windows, like the differences in pigmentation, diminish as the blade matures (Benzing and Friedman 1981; Fig. 2.14G). Cambridge Books Online © Cambridge University Press, 2009 Additional aspects of light relations 183 Figure 4.27. Diurnal patterns of CO2 assimilation during two days representative of the rainy and dry seasons. Measurements were performed on the same days for both exposed (A) and semiexposed (B) plants. Integrated PPFD was 24.9 and 34.9 mol photons m22 day21 for the rainy and dry-season days respectively (after Maxwell et al. 1995). Shutters ® xed CO2 more vigorously on a surface area basis than adjacent windows until the two shades of green converged with age, after which uptake occurred at about the same modest rate across the blade. Species with suggestive coloration warrant closer inspection for the possible presence of an accompanying CO2-concentrating mechanism. Cambridge Books Online © Cambridge University Press, 2009 184 Carbon and water balance Paraffin-embedded and sectioned blades revealed no anatomical differentiation between the chlorophyll-laden and adjacent, paler, mesophyll cells. However, the commissures that often join the more robust parallel veins of these Type Four species routinely traverse only the shutters (Fig. 2.17B). Further inquiry might reveal the nonuniform distributions of key enzymes, uneven wall thickenings, and concentrations of plasmodesmata that elsewhere indicate the presence of the gas-tight barrier needed for C4 photosynthesis. Other taxa (e.g., Ronnbergia; Fig. 2.2F) with similarly thin foliage exhibit different arrangements of pigments suggesting still other possibilities. Fenestration unevenly partitions the photosynthetic machinery in foliage, but bene® ts from the resulting visual effect may exceed any advantages related to carbon gain. Nutrition is a distinct possibility for plants reliant on intercepted litter and the biota needed to process it. A fenestrated leaf can also be larger at reduced cost per unit area than an otherwise comparable concolorous organ by distributing unevenly the same amount of N and P allocated to energy capture. The resulting larger leaf area would increase impoundment capacity, hence plant access to moisture and nutrients, without requiring additional investment in these two scarce commodities. Perhaps resource-rich zones distributed in a matrix of less nutritious tissue also discourage herbivory. Finally, could the nonuniformly pigmented bromeliad leaf deter gravid folivores by appearing already occupied by larvae? Because Type Three and Four bromeliads need detritivores to extract nutrients from litter, selection should promote characteristics that favor plant attraction of these fauna (Fig. 2.18D). Some circumstantial evidence supports this possibility as it applies to pigmentation. Horizontal variegations (e.g., Guzmania zahnii; Fig 2.18B) and parallel stripes (e.g., G. lingulata) tend to be most pronounced on the bases of leaves where capacity to obscure the outlines of tank occupants would be most bene® cial to the plant. Importance as a signal to remind resident pollinators of a sporadic food source (as suggested for the red/orange markings on certain Gesneriaceae) seems less likely because some of the most elaborately marked bromeliads (e.g., Vriesea fosteriana, V. hieroglyphica, Guzmania zahnii) ¯ ower at night and disperse wind-carried seeds. The plausible explanation must also account for those occasional species (e.g., Guzmania bismarckii) that produce ornamented foliage as juveniles only to become concolorous later. Light harvest might improve with certain ornamentations. Conceivably, a compact shoot captures more incident PAR if its congested leaves possess Cambridge Books Online © Cambridge University Press, 2009 Additional aspects of light relations 185 Figure 4.28. Transparency to PAR at noon on a clear day of the leaves of three bromeliads distinguished by leaf form and pigmentation (after Benzing and Friedman 1981). both translucent, relatively inactive and opaque, more vigorously photosynthetic zones. Should windows diminish the quantum yields of the upper leaves compared with what uniformly green organs could achieve, overall more incident PPFD might still be utilized by a multilayered shoot. Vriesea fosteriana and its kind could be especially well suited to harvest sun ¯ ecks or operate where the upper layers of the shoot become photoinhibited during the brightest part of the day. A number of deeply shade-tolerant Type Three and Four bromeliads (e.g., Aechmea miniata, Lymania smithii, Tillandsia viridiflora, some Vriesea splendens) display discolorous foliage, which elsewhere purportedly enhances the utility of shade-light (Lee et al. 1979). Speci® cally, the maroon to red adaxial epidermis scatters unabsorbed irradiance back up into the mesophyll. Its co-occurrence with horizontal, monolayered leaf displays (e.g., Nidularium burchellii, Aechmea fulgens, Vriesea simplex) supports this contention (Fig. 2.4H). Densely congested, self-shading shoots characterize many Bromeliaceae with affinities for high-energy sites (Fig. 1.2G) or, if such a plant is shadetolerant, its foliage is exceptionally thin by family standards (e.g., Tillandsia Cambridge Books Online © Cambridge University Press, 2009 186 Carbon and water balance Table 4.11. Arrangements and conditions that should prevail among certain Bromeliaceae if the interpretations of leaf form and pigmentation detailed in the text are valid (1) Leaves bearing a cyanic abaxial epidermis should occur in monolayers and be relatively long-lived. Leaf structure, composition and physiology should indicate shade-tolerance. (2) Bromeliads equipped with fenestrated foliage should require relatively high exposure. A densely inhabited phytotelmata would be consistent with protective function. (3) Leaves with darkly pigmented bases are most likely to characterize shoots that grow singly rather than form compact clusters. Exceptionally rich tank faunas would also be signi® cant here. (4) Species with multilayered shoots habitually encountered in moderate shade should have translucent, uniformly green, inexpensive (thin, low N/unit area) foliage. (5) A densely cyanic adaxial epidermis should characterize specimens exposed (5) to strong illumination. leiboldiana, T. complanata). Ecologically ¯ exible Guzmania monostachia exempli® es what could be a less vulnerable alternative (i.e., more leaves to lose to predators) with no corresponding trade-off in shade-tolerance. Three or more of its concolorous but translucent leaves must lie directly over one another under full sun to deny the lowest organ adequate light to balance respiration (Benzing and Friedman 1981; Fig. 4.28). Conversely, little PAR penetrated the foliage of Nidularium burchellii or the similarly monolayered, but concolorous, leaves of shade-tolerant Catopsis nutans. Optical enhancements for the capture of shade-light may also involve light-focusing protuberances on the cells of the adaxial epidermis and thin® lm effects. Suggestively felt-like textures and bluish-green re¯ ectance characterize some deep forest natives like Nidularium burchellii and shade-grown Bromelia pinguin. Arrangements of mesophyll cells that probably affect light propagation through the interiors of the foliage of shadetolerant bromeliads are described in Chapter 2. More study is warranted to determine whether certain mechanisms and ecological correlates prevail. If the logic offered above to explain the purposes of the unusual combinations of pigments, shoot architectures and leaf texture apply, additional research should con® rm the predictions summarized in Table 4.11. The following chapter completes the coverage of bromeliad ecophysiology with a discussion of mineral nutrition. Cambridge Books Online © Cambridge University Press, 2009 5 Mineral nutrition Arid climates and harsh substrates explain why certain Bromeliaceae figure so prominently in studies of drought-tolerance and CAM. Mineral nutrition has drawn sufficient attention to dispel misconceptions about how and from where the most specialized species secure essential ions, but certainly less interest than warranted by the presence of additional, even more exceptional mechanisms. Contrary to appearances, none of the epiphytes invades host vasculature, nor does anchorage on bark or rock necessarily impose nutritional stress given the frequent access these plants have to fertile alternatives like decomposing litter, ant carton and prey (Figs. 5.1–5.3). Nitrogen-fixers and plantfeeding ants assist still other Bromeliaceae (Fig. 5.1). On balance, only a small fraction of the family, namely certain dry-growing Tillandsioideae (Fig. 1.3A,C), rely exclusively on precipitation and dry deposition for nutrition, hence deserve the loosely applied label ‘air plant’. Leaf chemistry indicates that Bromeliaceae accumulate the expected six macronutrients and nine trace elements in the usual proportions (Table 5.1). Uptake also includes additional ions that support the same and other functions elsewhere. For example, Si, which in grasses helps deter grazers and stiffens the Equisetum stem, contributes to the light-reflecting granules in the epidermis of bromeliads native to sunny exposures (Fig. 4.23I). CAM types probably utilize Na like other similarly equipped xerophytes. Now and then, certain required elements concentrate far beyond metabolic needs; others accumulate for no recognized purpose, although they convey useful information about environments. Type Five Tillandsioideae exhibit sufficiently high affinities for certain ‘technological’ metals (e.g., Cr, V, Zn) and S to serve as inexpensive alternatives to the mechanical devices usually employed to monitor air quality (Tables 5.2, 5.3). Much of the literature germane to bromeliad nutrition deals primarily 187 188 Mineral nutrition Figure 5.1. Schematic diagram illustrating the major sources of mineral nutrients for Bromeliaceae dependent largely on shoots for uptake (i.e., primarily the epiphytes and lithophytes). with systematics and comparative morphology. Other reports worth reviewing for this chapter emphasize air pollution or the fertility of precipitation and rooting media in tropical forests. Data on the growth of certain ornamental species in hydroponic and aseptic culture mostly appear in horticultural journals and publications for hobbyists. Except for one preliminary survey (Benzing and Renfrow 1974a), no treatment compares Bromeliaceae among vascular flora relative to sources, needs and tolerances for shortages and oversupplies of mineral ions. Our purpose here is to update this summary insofar as the still meager database for bromeliads and extrapolation from other, better-known taxa permit. External supply and plant demand Everwet forests that harbor adaptively diverse bromeliad species demonstrate how family members partition nutrient capital in shared ecosystems. Resolution is uncommonly high for reasons related to plant architecture, physiology and capacity to grow on the ground and up through the canopy to its most exposed and hostile perches (e.g., Figs. 5.1, 7.11, 7.12). External supply and plant demand 189 Figure 5.2. Aspects of Bromeliaceae related to mineral nutrition. (A) Tillandsia utriculata illustrating litter impounded by a phytotelm shoot. (B) Autoradiograph of a medially sectioned Tillandsia caput-medusae shoot after being fed 45Ca through the surface of a single leaf base. (C) Small arboreal ant carton in southern Venezuela supporting seedlings most of which appear to be Codonanthe sp. (Gesneriaceae). (D) Phytotelmata of Vriesea gigantea containing a drowned insect in Espirito Santo State, Brazil. (E) Trichome of Brocchinia reducta, no label present (control). (F) Autoradiograph of Brocchinia reducta trichome exposed to 3H-leucine. (G) Autoradiograph of Brocchinia acuminata trichome exposed to 3H-leucine. (H) Autoradiograph of Tillandsia streptophylla trichome exposed to 3H-leucine. 190 Mineral nutrition Figure 5.3. Aspects of Bromeliaceae related to mineral nutrition (continued). (A) Loose epicuticle on foliage of Catopsis berteroniana. (B) Catopsis berteroniana growing as an epiphyte in Bahia State, Brazil. (C) Atrophied trichome on the upper part of the leaf of Brocchinia reducta above the phytotelmata illustrating the fibrillar nature of the loose epicuticle (3250). (D) Brocchinia vestita growing on boggy soil among carnivorous Heliamphora sp. on Cerro Neblina, Venezuela. External supply and plant demand 191 Table 5.1. Mineral nutrients present in the foliage of a typical eutroph (cereal crop) and of Tillandsia paucifolia on nutrient-stressed and betternourished cypress trees in Florida Generalized minimum requirement for eutrophic vegetation Nutrient N P K Ca Mg S Mn Fe B Cl Cu Zn Mo % dry weight ppm 1.50 0.10 1.00 0.50 0.20 0.10 Concentration in T. paucifolia growing on nutrient-stressed cypress % dry weight ppm 0.36 0.072 0.33 0.66 0.17 0.05 50 100 20 100 6 20 0.10 Concentration in T. paucifolia growing on relatively vigorous cypress % dry weight ppm 0.35 0.085 0.54 0.98 0.23 0.097 27.5 154.8 15.2 — 9.2 35.5 1.43 22.5 195.8 18.3 — 10.0 41.8 1.60 Source: After Benzing (1990). Substrates, degrees of dependence on roots vs. shoots for ion absorption, and engagements with nutrition-enhancing mutualists often differentiate bromeliads that share the same kinds of microsites. Appreciation of this variety requires familiarity with phenomena that operate at the level of the hosting ecosystems, to the individual plant, and on down to its subcellular components. A modest literature devoted to plant-feeding ants, diazotrophic microbes, impounded litter and canopy leachates describes the sources of nutrients available to specific kinds of bromeliads. These reports also indicate the existence of numerous nutritional modes (Table 5.4). Other publications describe the uptake of certain solutes, particularly the absorptive qualities of the foliar trichome. Two concepts never mentioned in any of these treatments allow more fundamental comparisons of bromeliads, viz. nutritional sufficiency, the state of being adequately nourished, and mineral-use efficiency (MUE), a metric that expresses plant performance relative to the deployment of acquired nutrients. Considerations on both counts enhance prospectives for this overview. Table 5.2. Concentrations of metals (ppm) in the shoots of Tillandsia usneoides collected in the Big Thicket National Preserve in southeastern Texas. Data represent contents of five collections. An asterisk indicates concentrations in ash. Values for the other elements are based on dry weight Sample 1 2 3 4 5 Ash (% dry weight) Ag* Cd Cr Cu* Mg Mn* Na Ni Pb* V* Zn Zr* 3.59 4.03 3.71 3.48 6.04 2.0 5.0 ,1.0 ,1.0 ,1.0 0.32 0.22 0.23 0.24 0.23 2.27 0.69 0.97 1.08 1.41 150 70 100 70 50 1713 2180 1293 1526 2730 5000 7000 7000 5000 3000 3617 1433 511.3 321.0 1460 3.97 2.70 2.72 1.60 2.50 700 300 300 700 300 150 30 70 30 70 68.54 26.40 23.17 36.99 30.59 70 70 70 70 200 Source: After Benzing (1989). Table 5.3. Mineral uptake efficiencies of Tillandsia paucifolia for 11 elements expressed as percentages of initial levels. Whole plants were immersed in aerated treatment solutions for 0.5 h per day Treatment series 2: Micronutrients at 5 31028 M in all runs, macronutrients at five concentrations Treatment series 1: All elements equimolar Element and initial concentration in shoots (mg kg21 dry weight) 60 days 120 days 60 days for all runs 1025 M 1026 M 1027 M 53 1028 M 33 1024 M 83 1025 M 2 31025 M 1 3 1025 M 5 31026 M N (4600) P (680) K (4000) Ca (6800) Mg (2200) N/C 1132 N/C N/C N/C N/C 129 N/C N/C N/C N/C N/C N/C N/C N/C N/C N/C 136 N/C 127 151 241 118 N/C 186 115 134 115 119 174 N/C 113 113 N/C N/C N/C 113 N/C N/C N/C N/C 115 N/C N/C N/C Mn (32.5) Fe (124) B (13.3) Cu (16.2) Zn (36.8) Mo (1.40) 1454 1218 N/C 1235 1405 1341 287 121 N/C 707 372 136 267 143 138 663 295 N/C 324 N/C 125 453 322 N/C N/C N/C 135 160 200 123 162 N/C 132 193 211 116 196 N/C N/C 241 179 N/C 168 N/C 113 256 169 N/C 210 N/C 120 235 176 N/C Source: After Benzing and Renfrow (1980). Note: N/C, final level less than 110% of initial concentration. 194 Mineral nutrition Table 5.4. The nutritional modes of Bromeliaceae Type Occurrence in family Remarks (1) Root-dependent terrestrial Most Pitcairnioideae, many Bromelioideae, those few Tillandsioideae with well-developed root systems Shoot contributes little to nutrient uptake, root system well developed A few Brocchinia species and many Bromelioideae and Tillandsioideae Detritivores and saprophytes process litter impounded in phytotelmata for plant use Modifications present to trap and process prey are relatively unspecialized Aerobic N2-fixers active in phytotelmata Several tubular species likely candidates (e.g., Billbergia porteana) (2) Tank-based (A) Animal-assisted saprophyte (B) Carnivore (C) Diazotrophassisted (D) Vertebrateassisted (3) Myrmecotrophs (A) Ant-house (B) Ant-nest garden (4) Atmospherics Two Brocchinia species and Catopsis berteroniana Undetermined number of phytotelm species Undetermined number of phytotelm species Brocchinia acuminata, many Bromelioideae like Aechmea bracteata, unknown number of bulb-forming Tillandsioideae and additional phytotelm species with dry leaf axils Mostly Bromelioideae (e.g., Aechmea, Neoregelia) Tillandsioideae exclusively Importance to plant nutrition largely unproven Ant-dispersed and routinely rooted in ant cartons Dense indumentum of absorbing hairs extracts nutritive ions from passing fluids. Substrates serve primarily for anchorage External supply and plant demand 195 Concentrations of key elements in some reference organ, usually a leaf of a certain age or location on the plant, indicate enough about nutritional status to formulate fertilizer regimens for many crops. Benzing and Renfrow (1971a) and Benzing and Davidson (1979) surveyed Tillandsia paucifolia on this basis to investigate the nutritional sufficiency of an epiphytic bromeliad relative to its hosts in southern Florida. Specifically, leaf composition was compared among the epiphytes and these values in turn with those obtained from the foliage of the supporting trees (Figs. 7.8, 7.9). Trends were parallel, i.e., stressed trees supported the most deficient bromeliads. Comparisons involving two or more species of Bromeliaceae or any other flora present a greater challenge. Nutrition peculiar to specific patterns of natural history (e.g., that typical of the representative annual weed vs. a perennial herb or tree) influences which concentrations of what key elements represent nutritional sufficiency for certain flora. Requirements in each case parallel important nutrient-related performances like Amax and fecundity. A bromeliad equipped to draw on a nutrient-enriched phytotelmata or a relative that instead roots in fertile soil must accumulate the relatively high concentrations of N and P required to achieve the requisite vigorous photosynthesis to compete with co-occurring vegetation drawing on the same abundant resources. The more stress-tolerant, Type Five relative obliged by architecture and impoverished substrates to subsist on scarcer, transitory supplies, primarily precipitation and canopy washes, operates with lower demands (Table 5.5; Fig. 5.1). Biomass produced per unit of incorporated nutrient, a number that quantifies MUE, also parallels adaptation, and reflects growing conditions experienced in situ by tested subjects. Energy returns on investments of the elements whose supplies limit growth to the greatest degree vary on two time scales and accordingly, yield two values for MUE. Instantaneous mineral-use efficiency, essentially the maximum output (Amax) possible per unit of invested (in foliage) N or P, expresses shortterm return, and by this measure the typical phytotelm bromeliad substantially outperforms its Type Five relatives. Bromeliads probably assort differently on the basis of lifetime (leaf life span) yields relative to investments of key nutrients. Field and Mooney (1986) proposed a more precise coefficient to compare plants using photosynthate accrued from investments of one nutrient in particular. Their index, known as potential photosynthetic nitrogen-use efficiency (PPNUE), recognizes the tight correspondence between the N allocated for the construction of energy-harvesting organs and Amax. 196 Mineral nutrition Table 5.5. Mineral elements in leaf blades (% dry weight) representing the five ecological types in Bromeliaceae Species Ecological type/ location Ca Bromelia karatas Type II/Jamaica 1.19 Pitcairnia bromeliifolia Type I/Jamaica 1.06 Catopsis floribunda Type IV/Jamaica 1.15 Aechmea nudicaulis Type III/Jamaica 0.38 Guzmania lingulata Type IV/Trinidad 0.44 Tillandsia balbisiana Type V/ Florida 0.83 Tillandsia usneoides Type V/ Florida 0.67 K Mg 2.48 1.02 1.59 1.94 1.59 0.34 0.50 0.14 0.14 0.18 0.20 0.25 0.14 0.29 N Na P 0.67 0.064 0.068 1.37 0.036 0.084 1.36 0.32 0.091 0.57 0.18 0.057 0.88 0.19 0.063 0.36 0.41 0.035 0.82 0.55 0.012 Source: After Benzing and Renfrow (1974a); Table 4.2. Vegetation constrained by N-poor foliage – species with low critical concentrations of this key nutrient (the concentration needed to realize growth potential, i.e., to achieve Amax) – gain carbon slowly, and thus express inherently poor instantaneous PPNUE. In essence, leaf longevity and Amax vary inversely because investments in green tissue must be amortized via photosynthesis plus a marginal gain to support growth, reproduction and plant maintenance. Slow payback and additional time for profit mandate comparably durable biomass made relatively expensive by obligatory investments in defensive chemistry, strengthening polymers, and other compounds beyond the pigments, enzymes and additional plant constituents required to harvest photons. Similar leaf characteristics (e.g., high durability, high cost in carbon/unit leaf surface area) equip plants for drought and infertile substrates, and both conditions also constrain Amax. Therefore, PPNUE should be modest for most Bromeliaceae, and foliage relatively long-lived and well defended. Preliminary data obtained with a portable photosynthesis analyzer in wet pre-montane forest at Rio Palenque, Ecuador and in a northern Venezuelan cloud forest support one of these suppositions. Ecuadorian Guzmania monostachia fell closer to findings for woody evergreens than deciduous trees, shrubs and annuals on the basis of Amax as a function of N content on a leaf area basis (Field and Mooney 1986). Table 5.6 provides values for two epiphytic ferns, co-occurring phytotelm Vriesea platynema, and foliage from the supporting guava tree at the just mentioned site in northern Venezuela. However, both sets of numbers would indicate more about local growing conditions and plant access to Nutritional peculiarities 197 nutrients and water if referenced on the longer of the two time scales – if expressed as integrated rather than instantaneous MUE. Foliage that generates photosynthate slowly over an extended life span signals scarce resources. Shorter-lived, less expensive (in carbon) leaves capable of fixing CO2 faster on a leaf area basis represent the superior strategy in more equable habitats, which as already indicated tend to favor the superior competitor. Nutritional peculiarities Plants reflect the fertility of their habitats in additional ways. Eutrophic types, those species endemic to nutrient-rich substrates, exhibit critical concentrations of N and P that fall at the upper end for vegetation generally. High shoot/root ratios, short life cycles as well as foliage that turns over rapidly further identify these taxa. Moreover, fecundity is exceptionally elastic: it tracks available nutrients as illustrated by annual weeds and their derived cultigens, the consummate eutrophs. Nutritionally deprived specimens grow poorly and, faced with extreme scarcity, fail to reproduce. Sufficiently fertilized, they do quite the opposite. Another response favors plant life on habitually deficient media. Dilute supplies of nutrients never seriously stress the inherently slow-growing oligotroph, even though concentrations of key elements may fall well below levels in the comparable tissues of the better-supplied members of the same stock (e.g., Tillandsia paucifolia; Table 5.1). Shoot/root ratios tend to be low, foliage evergreen, and the life cycle of the plant protracted. Salt-tolerators Additional peculiarities related to mineral nutrition and ion balance adapt many land plants for still other chemical environments. Sensitivities to heavy metals (Tables 5.2, 5.3) and NaCl vary among taxa and may distinguish populations within species. Diverse Bromeliaceae occupy saline habitats, some as narrow endemics like Pitcairnia halophila on rock outcrops just above high tide in the southern portion of Costa Rica’s Puntarenas Province (Grant 1994a). Pitcairnia integrifolia occurs under similar conditions in Trinidad. However, low osmotic pressures (,0.91 MPa) in foliage indicate either salt-exclusion or avoidance of saline media (Lüttge et al. 1986a). Certain members of Bromelioideae and Tillandsioideae also inhabit maritime sites, and possess architecture that obliges contact with salt. 198 Mineral nutrition Tubular Aechmea nudicaulis, which ranges from Mexico southward, approaches the intertidal zone in southeastern Brazil where it, along with the many other bromeliads of the restinga formation (Fig. 7.13C), often grows close enough to the surf to regularly intercept salt spray. Bromelia humilis extends from several hundred meters above on to low-lying, saltladen soils along the south Caribbean coast, also without measurable effects on tissue chemistry, as described below. Several sets of data indicate tolerance and perhaps even benefit from access to sea salt among Type Five Tillandsioideae (Fig. 1.2H). Molar concentrations of Na exceeded those of K in the shoots of Tillandsia flexuosa native to Venezuela’s northern coast (Griffiths et al. 1989). Leaves of Tillandsia paucifolia growing on red mangrove (Rhizophora mangle) in south Florida contained Na at up to several percent of dry weight (Benzing and Renfrow 1971a; Benzing and Davidson 1979). Additional populations farther inland bore lower salt burdens, but K/Na ratios remained above unity. Surveys of additional Type Five Tillandsia (e.g., Shacklette and Connor 1973) further suggest benefits from Na where scarce supplies of K (e.g., telephone wires, treetops; Fig. 1.3A) or poor root development favor substitutions. Gómez and Winkler (1991) used AgNO3 and sectioned leaves to determine whether members of a suite of wide-ranging Mesoamerican bromeliads (Tillandsia dasyliriifolia, T. caput-medusae, T. baileyi, T. festucoides, T. schiedeana, Catopsis sp. and Aechmea sp.) growing on red mangrove accumulate salt at a site along the Pacific coast of Guatemala. Plants were generally more robust and their foliage thicker compared with relatives in other kinds of nearby forest. Staining indicated that Tillandsia dasyliriifolia and the unspecified Catopsis contained the most Cl2. Highest concentrations occurred in the mesophyll, specifically in cells surrounding the vasculature and those located below the trichomes. Co-occurring Bromeliaceae and Orchidaceae sometimes develop different chemical profiles in the same tree crowns, perhaps in part owing to their distinct body plans. Encyclia tampensis (Orchidaceae) foliage contained substantially more K than Na compared with Tillandsia paucifolia (Benzing and Renfrow 1974b; Benzing 1978b). Other bromeliads operate more like the orchid. Foliage that resisted penetration by 36NaCl in one experiment helps Pitcairnia integrifolia remain largely unaffected by sea water in coastal habitats in Trinidad (Lüttge et al. 1986a). Correlations between leaf chemistry and relative dependence on roots vs. shoots for uptake underscores the need to better understand trichome function to discover how high Na status develops. Nutrients in the forest canopy 199 Oligotrophs and other extreme strategists Among the most oligotrophic of the bromeliads judging by substrates, plant structure and growth rates are the pulse-supplied forms, those species without phytotelmata, ant nests or soil roots to provide more continuous streams of essential ions (Fig. 1.3A,C). Aerosols, precipitation and canopy washes suffice instead. In fact, nutrient requirements for Type Five Tillandsia may rank among the most modest of all for vascular flora. Predictably, these plants grow slowly and produce durable foliage that sometimes contains exceptionally low concentrations of N, P and K compared with phytotelm and soil-rooted species (Tables 5.1, 5.5). Because drought and habitually scarce supplies of key elements promote similar leaf morphology and oblige low vigor, Brocchinia species native to the moist, base-poor savanna habitats of the Guayanan highlands of northern South America probably demonstrate bromeliad oligotrophy in its purest form. Bromeliaceae illustrate several additional nutritional modes. Among the family’s membership are several carnivores, at least 50 ant-fed species of two types, and another, much larger and more important group labeled the animal-assisted saprophytes (Table 5.4). Before moving on to describe these relatively novel arrangements, we need some information about the fertility of several kinds of habitats. Supplies of nutrients available to Bromeliaceae more or less fall into two categories distinguished by whether or not mutualists are involved. Few records beyond those for cultivated pineapple describe the media that support terrestrial Bromeliaceae, but given the diverse kinds of sites these plants occupy, species dependent on absorptive roots probably experience soils of widely varying qualities. Enough is known about sources for the epiphytes to warrant a brief survey. Nutrients in the forest canopy Nutrient stocks accessible to arboreal flora originate from the atmosphere and the soil, in the second case following movement up the transpiration streams of supporting trees. Availability to resident Bromeliaceae depends largely on the capacity of these plants – which varies among species – to utilize leachates, litter, certain animal products or prey. Opportunities for the individual bromeliad may be limited, but no other family exceeds this one for nutritional variety – either for the diversity of the sources utilized or for the number and novelty of the devices employed to tap them. Most of this variety is expressed among the epiphytes. Table 5.6. Net photosynthesis and N and P present in the leaves of five epiphytes growing on the trunk of a single guava tree and the foliage of that phorophyte at Rancho Grande, Venezuela, 7–10 January 1988 Net photosynthesis (mmol CO2 m22 s21) Species and photosynthetic pathway Psidium sp. (C3) Vriesea platynema (C3) Stelis sp. a (CAM) Peperomia sp. a (?) Microgramma lycopodioides (C3) (drought-deciduous fern) Pleopeltis astrolepis (C3) (a resurrection fern) Unirrigated After 1 day After 2 days After 3 days 27.369.6 0 0 0 3.70 61.7 0 29.9612.5 — 0 10.0262.60 0 0 0 0 — — 7.15 60.82 — — 6.96 62.11 0 0 — 6.5360.98 Source: After Benzing (1990). a Gas exchange monitored day and night. Note: The dry mat of suspended humus supporting these plants was soaked with water after the first measurements. Leaf N (g m22) N P 0.64 0.26 — — 0.142 0.104 0.079 0.027 — — 0.0214 0.0056 Nutrients in the forest canopy 201 Figure 5.4. Wet and dry-season inputs from the atmosphere for five elements in a cloud forest in northern Costa Rica (after Nadkarni 1984). Inputs from the atmosphere Nutrients from the atmosphere arrive as dry deposition, including vapor, and in solution. Vapor provides S and N, in the second case mostly as NH3, while NH41 and NO32 predominate in precipitation. Magnesium, K and P, among the other elements required in relatively large quantities by plants, also arrive primarily dissolved in rain. Still others, including certain potentially troublesome metals, more often concentrate in dry deposition. Rates of delivery vary many fold depending on the site, with annual inputs of decisive nutrients, for example P, ranging from 0.07 to 1.70 kg ha21 year21 in just two studies (Nadkarni 1984; Newman 1995; Fig. 5.4). Certain elements exceed others in the particulates that lodge on plants, and chemical signatures often indicate specific origins, for instance whether derived from land, sea or biota (e.g., Clarkson et al. 1986). Composition varied among the shoots of Tillandsia usneoides sampled across the southeastern United States (Shacklette and Connor 1973; Connor and Shacklette 1984). Shifting proportions of Al, Ba, Ca, Ga, Fe, Sn and Y indicated multiple, site-specific sources. Concentrations of technological metals also changed with the sample, but differently as described below. Should the indumentum contribute to the remarkable scavenging capacities of the ‘atmospheric’ bromeliads by enhancing capacity to trap 202 Mineral nutrition nutrient-laden aerosols, then another service can be added to the list of benefits provided by the bromeliad trichome (Table 2.1). Precipitation contains all of the plant nutrients in at least minute amounts to concentrations exceeding those in many soil solutions (Table 5.7). Local and regional geology (e.g., desert, proximity to ocean), type of vehicle (e.g., rain, cloud water), season, and a variety of other natural and anthropogenic factors determine the compositions of these solutions. Global change is also an increasingly important player. Mounting inputs reflect changing land use, especially the burning of biomass and the consumption of fossil fuels (Moffat 1998). Impacts on oligotrophic flora and many Bromeliaceae and other plants dependent on foliage and the atmosphere to acquire mineral nutrients seem likely whether burdens of reactive N continue to increase or level off. Acidified precipitation could exacerbate the problem given the weakly buffered nature of the rooting media and phytotelmata of many of the epiphytes. A rising CO2 level further challenges attempts to predict plant responses to elevated delivery of H1, P and N. Inputs from the atmosphere relative to other sources vary in importance with the nutrient and the affected flora. Sulfur deposition often exceeds local requirements, whereas inputs of N, P and K fall well short of the needs of all but the primarily rain-fed communities like raised ombrotrophic Sphagnum bogs and heathlands (Clarkson et al. 1986). Epiphytes in forests over impoverished soils can substantially influence system-wide nutritional dynamics (Chapter 7). Several surveys (e.g., Nadkarni 1984; Clark et al. 1998; Table 5.11) illustrate how much of the stocks of several key elements reside in suspended phytomass. Table 5.8 demonstrates why phytotelm bromeliads are so well suited to subsist on impounded litter, and at high densities in infertile ecosystems collectively immobilize significant quantities of local nutrient capital. Data collected by Nadkarni (1984, 1986) in lower montane, Costa Rican rainforest are useful for this discussion because the habitat supported considerable numbers of Bromeliaceae, and she specified the times of delivery and the vehicles. Precipitation tended to be more nutritive during the dry (N 0.28 ppm, P 0.95 ppm) compared with the wet (N 0.05 ppm, P 0.11 ppm) months. However, the greater volume of wet-season rainfall considerably diminished dry-season influences on total inputs (Fig. 5.4), although not necessarily the welfare of impacted Bromeliaceae if uptake parallels concentrations in sources. Nitrate far exceeded NH41 as a nitrogen supply for the local epiphytes. Clouds and mist (occult precipitation) also delivered substantial inputs, in part because concentrations were several fold higher than those in rainfall (e.g., Clarkson et al. 1986; Coxson and Nadkarni 1995). Table 5.7. Chemical composition (ppm) of rainwater, throughfall and stem flow in the forest canopies of central Amazonia, eastern Panama, Haiti and South Florida Central Amazoniaa Eastern Panamab Haitib Stem flow Nutrient Rainwater Throughfall Stem flow Rainwater Throughfall Na K Ca Mg Mn N (NH41) N (NO32) N (NO22) N (total) P (PO432) S Fe Zn 0.12 0.10 0.07 0.02 0.27 1.24 0.25 0.19 2.11 6.58 1.72 0.97 0.02–1.4 0.4–3.6 0.04–0.64 0.003–0.68 2.4–5.6 2.0–2.8 0.44–0.80 0.010–0.036 3.00 1.00 4.30 trace 1.23 0.17 0.11 0.002 0.41 0.003 0.05 0.56 0.01 9.20 0.27 0.02 0.151 0.095 0.033–0.075 0.024–0.068 0.053–0.340 0.017–0.079 0.036–0.416 0.029–0.062 0.15 0.17 0.41 Source: aFrom Junk and Furch (1985). bFrom Benzing and Renfrow (1980). South Floridab Stem flow 0.25–1.30 3.04–9.60 0.52–0.61 0.40–0.76 0.017 Mineral nutrition 204 Table 5.8. Distribution of nutritive elements in the shoots and phytotelmata of Guzmania monostachia specimens growing in a swamp forest in south Florida. Values are averages for two specimens Mineral content (mg) Element Vegetative Mature organs infructescence N P K Ca Mg Na Mn Fe B Cu Zn 151.2 18.8 399.5 189.6 116.3 48.4 1.02 1.41 0.27 0.035 0.042 50.6 10.7 90.4 21.1 18.3 2.56 0.24 0.37 0.068 0.017 0.16 Phytotelmata 197.8 11.3 17.0 288.0 24.0 4.4 0.45 2.91 0.55 0.031 0.28 Percentage of total plant pool replaceable from foliar impoundments 98.0 38.3 3.5 136.7 17.8 8.6 35.7 163.5 162.7 59.6 271.8 Source: After Benzing and Renfrow (1974a). Bromeliaceae experience shared habitats differently depending on the mix of species present. For example, had the two bromeliads illustrated in Fig. 5.1 resided in the dry Honduran forest (1200 mm year21) Kellman et al. (1982) examined, they would have encountered half or more of the annual inputs of important nutrients arriving via precipitation during as few as 1–10 rainy days. Contact with nutrients would have been brief for the Type Five bromeliad (assuming it harbored no plant-feeding ants) compared with its tank-equipped relative, which maintains more continuous contact with sources represented by impounded litter and possibly nitrogen-fixers. A variety of phenomena in addition to storm frequency and duration affect the timing of the delivery of nutrients to arboreal Bromeliaceae. Ions scrubbed from the atmosphere during a storm enrich early more than laterarriving precipitation. Subsequent contact with coating and exchange sites within the canopy further enrichs or depletes solutions after rain becomes throughfall and stem flow. Concentrations of Ca21 and K1 usually increase at this stage, while the abundances of others change less predictably (Table 5.7). Nutrients in the forest canopy 205 Table 5.9. Mineral element content in stem flow and the outer bark and foliage of dwarfed and relatively vigorous Taxodium distichum hosting Encyclia tampensis and Tillandsia paucifolia in south Florida Element (mg 5 g21 dry weight) Status of host (1) Bark Water extract N P K Ca Mg Na Dwarfed Vigorous 0.22 0.004 0.11 0.38 0.004 0.37 HCl extract (0.01 N) Dwarfed Vigorous 0.13 0.014 0.14 33.9 0.33 0.24 0.34 0.096 0.74 107.0 1.11 0.20 Nutrients remaining after both extractions (bark wet-digested) Dwarfed Vigorous 5.31 0.063 0.07 6.46 0.251 0.33 (2) Foliage Dwarfed Vigorous (3) Stem flowa Dwarfed Vigorous 53.0 72.0 2.80 21.5 6.50 33.5 0.28 0.02 0.24 6.22 0.08 0.13 42.7 0.12 0.03 47.6 0.53 0.02 103.5 4.20 1.95 165.5 7.50 3.75 0.76 0.017 0.25 0.40 0.017 1.30 3.04 0.52 0.60 9.60 0.61 0.64 Source: After Benzing and Renfrow (1974b). Note: aStem flow mineral element content is expressed in ppm. Bark, suspended soils, and other aerial media So far, discussion has focused on nutrients in aerosols, precipitation and canopy washes, and, for the phytotelm bromeliad, on inputs from symbiotic diazotrophs and litter impounded in phytotelmata. Except for the wholly shoot-reliant species, the properties of ant carton and other animal products, bark, suspended mats of humus, and rotting wood that contact roots also influence the nutritional welfare of epiphytic Bromeliaceae. Several publications report the fertility of canopy substrates for these plants. Dilute HCl stripped only modest quantities of N, P and K (Table 5.9) from the outer bark of dwarfed cypress trees that supported exceptionally nutrient-deficient Tillandsia paucifolia in Florida (Chapter 7; Benzing and Renfrow 1974b; Table 5.1). Preparations from more vigorous trees hosting relatively robust specimens yielded K and P at 5–7-fold and N at about threefold higher concentrations. Substrates for epiphytic aroids, bromeliads, gesneriads and ferns in an abandoned Theobroma cacao plantation at Rio Palenque, Ecuador also differed in important chemical properties (Table 5.10). Compared with subjacent soil, all five types of media tested Table 5.10. Chemical characteristics of mineral soil and suspended media (one sample each) in wet forest at Rio Palenque, Ecuador Description of material pH % base saturation Outer bark of large Theobroma branches with associated debris and nonvascular plants Outer bark of Theobroma twigs Rotten wood of Theobroma Fern root ball Carton of ant-nest garden Earth soil 6.2 6.7 7.1 5.2 6.3 6.3 79.1 85.8 90.2 56.4 78.4 55.3 Source: After Benzing (1990). Cation exchange capacity 123.5 137.4 163.3 135.1 115.3 31.1 meq 100 g21 K Ca 20.0 49.7 18.7 67.4 4.6 112.3 7.5 57.3 20.1 56.2 0.5 14.0 ppm Mg H Na N P K 25.5 31.5 30.1 11.1 12.2 2.5 25.8 19.5 16.0 58.9 24.9 13.9 2.6 0.3 0.3 0.4 1.9 0.2 3.0 2.2 1.5 1.8 2.9 0.3 0.34 0.22 0.09 0.10 0.39 — 0.67 0.71 0.18 0.25 0.79 — Nutrients in the forest canopy 207 Table 5.11. Mineral nutrient capital in the crowns of two dwarfed Quercus virginiana hosts growing in a coastal strand community in south Florida: percentage of total found in the epiphyte load N P K Ca Mg Na Mn Fe B Cu Zn Mo Specimen 1 35.4 53.4 50.2 41.4 76.4 60.8 44.0 77.1 36.2 55.2 62.4 62.1 Specimen 2 35.9 33.9 57.2 43.4 43.9 69.9 55.8 28.6 39.0 49.4 60.5 50.2 Source: After Benzing and Seemann (1978). exhibited superior cation exchange capacity, higher base saturation values, and a preponderance of N over P and K. Neutral to moderately acid pH prevailed, but may not be typical: suspended humus collected in pluvial forest in northwestern Ecuador produced readings down to 3.8 (Bermudes and Benzing 1989). Lesica and Antibus (1991) discovered that the epiphytes, including many Bromeliaceae, in humid lowland Costa Rican forest at La Selva root in substrates at least as fertile as those available to co-occurring terrestrial flora. However, the much larger volumes of soil on the ground probably assure greater total supplies for plants. Uneven rates of mineralization (higher below; e.g., Vance and Nadkarni 1990) further distinguish rooting media in the same forests, as does soil reaction. Whether this chemical mosaic contributes significantly, as Lesica and Antibus suggested, to the high diversity of local epiphytes, especially relatively root-dependent Bromeliaceae, remains to be seen. More extensive sampling than at either Rio Palenque or La Selva allowed Nadkarni (1984) to determine that humic soils (histosols) suspended within the canopy of a lower montane rainforest in Costa Rica contained large fractions of the total on-site pools of several essential ions. A similar pattern prevailed in upper montane cloud forest in Colombia (Hofstede et al. 1993) and in biomass largely attributable to Tillandsia recurvata in the crowns of dwarfed Quercus virginiana in a coastal strand community in southwest Florida (Benzing and Seemann 1978; Table 5.11). Nadkarni also examined key processes that influence the mineralization of suspended humus, specifically the transformation of complexed N into plant-usable forms. Less nitrification occurred in suspended compared with forest-floor litter, although microbial biomass was about the same in both compartments at Nadkarni’s Costa Rican site. Cellulose discs embedded in canopy debris lost less weight than those worked into litter on the ground at the 208 Mineral nutrition same location (Nadkarni 1986; Vance and Nadkarni 1990). Terrestrial samples weighed 23–45% less after eight weeks, while those incubated within suspended humus over the same interval changed little. Epiphytederived soils also harbored fewer detritivores, had lower water but higher fiber content, had a higher carbon/nitrogen ratio, and seemed to be dissipating polyphenols more slowly than phytomass decomposing on the ground. Densities of mites, adult beetles, holometabolous insect larvae, Collembola, amphipods and isopods averaged 2.6 times higher in earth compared with canopy soils. Only ants occurred at about equal densities in both media. Nadkarni and Matelson (1991) further concluded that the histosols suspended there largely develop in place. Except for the modest amounts of material intercepted in the shoots of phytotelm bromeliads, shed plant parts mostly fall to the ground. Contrary to Nadkarni’s findings for mats of epiphytes and suspended humus in Costa Rica, Paoletti et al. (1991) documented conditions favorable for rapid litter breakdown in the canopy of cloud forest at two sites in northern Venezuela. Up to fivefold greater detritivore densities (number of animals per unit volume of impounded humus) occurred in the shoots of resident phytotelm Bromeliaceae as on the forest floor (Fig. 8.15). Moreover, dried Psidium foliage placed in nylon mesh bags and incubated in the leaf axils of these epiphytes on average weighed 21–27% less after three months – about the same rate of loss recorded for samples buried in earth soil under the host trees. Additional surveys could help determine whether the soil fauna observed in phytotelm bromeliads also attack debris elsewhere in the canopy. Identifying the sources of nitrogen Processes similar to those that fractionate the stable isotopes of carbon and hydrogen during photosynthesis also provide opportunity to track another important element through ecosystems. According to Schulze et al. (1991), fractionation during transfers between trophic levels enriches the 15N content of biomass 3–5‰. Midgiey and Stock (1998) took advantage of this phenomenon to demonstrate carnivory in Roridula gorgonias, and the same approach could help determine inputs from prey and ants to phytotelm and myrmecophytic bromeliads. Stewart et al. (1995) concluded from the isotopic makeup of the N distributed among forest flora at two sites in Brazil that the assayed epiphytes more than the supporting trees rely on the atmosphere, perhaps partly via nitrogen fixation, for this key nutrient. Specifically, heavy N was relatively depleted in most of the assayed epiphy- Mechanisms 209 tes (cacti, ferns, orchids and Peperomia in addition to the bromeliads) relative to foliage born by the phorophytes. Bromeliads yielded the lowest 15N values (x̄ 525.2 and 24.9‰), even lower than those of the other assayed epiphytes (all means5.23.0‰), while the trees exhibited positive readings (x̄ 52.6 and 3.1‰), as expected for plants rooted in soil, a typically more 15N-enriched medium. Conversely, the major N sources in the atmosphere (fixed N2, NO32 and NH41) contain proportionally less 15N (,23.0‰). Unfortunately, Stewart et al. failed to identify the bromeliads to species so their claim that plant habit (architecture) had no effect on 15N/14N ratios in biomass (i.e., utilization of specific N sources) cannot be confirmed for this family. Type Five species should yield lower values than those equipped with phytotelmata, unless little of the N present in tree litter makes its way into the impounding shoot. Clearly, Stewart et al.’s and Midgiey and Stock’s approach holds great promise for more penetrating studies of bromeliad nutrition. Mechanisms Assistance from microbes Certain fungi and the nitrogen-fixers promote plant nutrition in different ways depending on the types of participants and certain other variables. Diazotrophs convert dinitrogen to forms available to themselves and eventually other biota. Leaky exchanges characterize paired, free-living organisms compared with the traffic between partners in the most intimate, coevolved mutualisms (e.g., the legume–Rhizobium association). Mycorrhizal fungi assist hosting flora by promoting the sorption of P and several other mineral nutrients and sometimes water. Nutritional enhancements effected by vesicular-arbuscular mycorrhizae (VAM) accrue largely through enhanced geometry. Much finer than the narrowest rootlet, hyphae simply represent more cost-effective extensions of the plant to explore substrates. Uptake of soil-immobile elements, like P, by VAM more or less occurs in direct proportion to the amount of medium contacted, whether by roots or by associated hyphae. Some of the fungi involved in the less familiar types of mycorrhizae (e.g., ericaceous, orchidaceous) attack soil humus to obtain essential ions (e.g., N), including some for plant use. They may also impart tolerance for certain toxins. Enhanced disease resistance probably exceeds nutritional benefits among the strongly mycorrhizal plants with extensively branched root systems (Newsham et al. 1995). 210 Mineral nutrition Several publications report fungi in the roots of bromeliads, but details vary. Pittendrigh (1948) failed to document his contention that terrestrial Bromelia humilis maintains mycorrhizae in northern Trinidad. Vascular epiphytes at Rio Palenque and in wetter forest between 800 and 1800 m in northwestern Ecuador bore extensive infections, and demonstrated the difficulty of determining consequences for the host flora (Bermudes and Benzing 1989). Roots of arboreal Pitcairnia pungens, the only bromeliad examined, lacked spores and hyphae, but samples from another member of the same genus native to lower montane forest at Monteverde, Costa Rica supported light infections by an unidentified fungus (Lesica and Antibus 1990). Allen et al. (1993) examined the roots of three epiphytic bromeliads (Catopsis nutans, Tillandsia bartramii, T. balbisiana) in seasonal woodland near Chamela, Jalisco, Mexico and found no VAM, although septate hyphae ramified through parts of every sample. Infected plant tissue free of browning and fluorescence suggested something other than a pathogenic relationship. If Bromeliaceae form mycorrhizae, VAM is most likely (Janos 1993), and indeed Rabatin et al. (1993) reported Glomus tenue infecting Vriesea platynema at the more arid of the two montane forest sites they sampled at Rancho Grande, Venezuela. Extrarhizal hyphae extended from the roots outward into the surrounding organic soil much as this hypomycete colonizes similar, relatively undegraded, desiccation-prone substrates at some terrestrial sites. Spores and other diagnostic structures permitted additional identifications. Roots of Aechmea lasseri and Vriesea splendens contained auxiliary cells produced by Gigaspora and Scutellospora at the second, wetter location, but neither fungus appeared in soil from the forest floor. Rodents and invertebrates, especially those partial to the debris impounded in and around bromeliad shoots (Paoletti et al. 1991; Table 8.2), may transport inocula, much as their ground-based counterparts do among terrestrial flora. Experimental inoculation enhanced the vigor of the single bromeliad tested so far. According to Aziz et al. (1990), pot-cultured Ananas comosus grew more vigorously on P-deficient media following infection with VAM. Pittendrigh’s Bromelia humilis and additional terrestrials less closely related to pineapple probably also support mycorrhizae, perhaps obligatorily like so many other soil-based plants. Then again, growing conditions in the canopy may constrain photosynthesis too severely to render the inescapable costs of routine, compared with facultative, mycotrophy supportable for the epiphytes (Janos 1993). Frequent occurrences of several largespored dictyostelid slime molds in canopy soils around the bases of Mechanisms 211 Guzmania berteroniana and Vriesea macrostachya specimens in wet forest in eastern Puerto Rico (Stephenson and Landolt 1998) indicate that heavy propagules alone should not restrict the incidence of VAM among epiphytic Bromeliaceae. Antibus and Lesica (1990) discovered surface-bound acid phosphatases produced either by roots or by adhering micro-organisms associated with 22 epiphytes, including an unidentified phytotelm Aechmea species and myrmecophytic Tillandsia bulbosa, at La Selva, Costa Rica. Assays based on fresh weight placed the two bromeliads within the point scatter depicting the rest of the collection. Both species grew on bare branches, and, like the several anthuriums and orchids and two ferns with similarly exposed, relatively robust roots, they yielded lower readings than those recorded for the epiphytes removed from the moist mats of bryophytes and humus covering the trunks and largest limbs. These more soil-like media possibly harbored a richer microflora; they certainly encouraged finer branching leading to more root surface relative to mass. Acid phosphatases sometimes increase access to organic P in soil, but whether similar benefits accrue to the epiphytes remains unclear, as does the significance of the phytases and acid phosphatases noted in the tank fluids of the sampled Aechmea specimens. Aerobic N2 fixation attributed to cyanobacteria appeared in both terrestrial and arboreal substrates supporting bromeliads in eastern Ecuador (Bermudes and Benzing 1991). Scrapings from Theobroma branches to which a variety of bromeliads and other vascular epiphytes rooted at Rio Palenque exhibited acetylene reduction (AR) rates equivalent to 5.4–17.7 ng g21 sample h21 (Table 5.12). Parallel sampling at the same sites shortly after a heavy rain yielded higher values, ranging from 8.5 to 110.0 ng g21 sample h21. Brighigna (1992) examined 12 Mexican Tillandsia species representing either Type Five or taxa equipped with weakly developed phytotelma for evidence that resident epiphyllae benefit hosts in a manner similar to what certain microflora provide for some vascular epiphytes in India (Sengupta et al. 1981). Freshly excised leaf segments and controls yielded comparable AR activity, but bacteria plated on N-free media from eight of the sampled species reduced significant amounts of acetylene. Isolates from relatively dry-growing T. bartramii, T. circinnatoides and T. schiedeana outyielded all the others. Bacillus and Pseudomonas species most often appeared in the cultures; those of Aeromonas, Rahnella and Vibrio that more typically inhabit wetter environments developed less often. Supposedly, transpiration provides the moisture needed to sustain the more drought-sensitive of 212 Mineral nutrition Table 5.12. Acetylene reduction activity associated with epiphytic bromeliads in the canopies of Ecuadorian forests. Values expressed as N equivalents assume 1 mole of acetylene reduced 5 0.25 moles of N fixed. Standard deviations are shown in parentheses Species Site 1 (Rio Palenque) Aechmea angustifolia A. angustifolia A. angustifolia A. angustifolia A. angustifolia A. zebrina A. zebrina A. zebrina Vriesea ringens Site 2 (Imbabura) Guzmania melinonis Guzmania sp. Tillandsia asplundii Unidentified species (Bromelioideae) Site 3 (Esmereldas) Guzmania sp. (1) Guzmania sp. (2) Material sampled ng N ha21 equivalent of C2H4 Tank fluid Submerged epiphylls Seepage zone (wet) Seepage zone (dry) Lichenized area above bromeliad Tank fluid Submerged epiphylls Seepage zone (wet) Tank fluid 0.0a 0.0a 255b (19.8) 15.3b (10.8) 0.0c 0.0c 43300d 2600d 0.0b 1.12a (0.54) 0.0a 7.6b (1.5) 0.0a 0.0d 15.4c 0.0c 456d 0.0c Submerged epiphylls Submerged epiphylls Submerged epiphylls 5.0a (0.57) 6.4a (3.7) 23.3a (0.43) 29.7c 13.8c 186c Submerged epiphylls 30.9a (9.4) 123c Submerged epiphylls Submerged epiphylls 11.3a (4.9) 3.7a (1.12) 135c 32c Source: After Bermudes and Benzing (1991). Notes: aPer leaf. bPer cm2. cPer plant. dPer seepage zone. these fixers, but considering the impressive water economy of the hosting bromeliads, humidification must be minimal. Tillandsia circinnatoides, for example, endures lengthy dry seasons in Mexican thorn forests by sparing use of the moisture sequestered in its succulent foliage. Puente and Basham (1994) concurred about the improbability of significant N2 fixation on the surfaces of Type Five bromeliads at the same time as they reported a potentially beneficial endophyte in one of these same plants. Their case rests on the recovery of Pseudomonas stutzeri from surface-sterilized pieces of Tillandsia recurvata leaves plated on media lacking combined N. Of the many additional bacteria also recovered, this bacterium alone exhibited nitrogenase activity in AR assays. Tested plants had grown on cacti and telephone wires in Baja California, raising the inter- Mechanisms 213 esting question of how a microbe usually found in wet substrates infects this exceptionally dry-growing bromeliad (Fig. 1.3A). An arrangement noted in certain Poaceae may also benefit other monocots. Organisms that synthesize defensive chemicals reside within the leaf sheaths of the grasses in question, having established there following passage from the previous plant generation via contaminated seeds. More germane to Puente and Basham’s claim, Acetobacter diazotrophicus living in root, stem and leaf tissue meets up to 80% of the N requirement for some Brazilian populations of sugarcane. But in the final analysis, in situ fixation must be demonstrated to determine whether Pseudomonas stutzeri or any other endophyte significantly augments the N budget of a bromeliad. Bermudes and Benzing (1991) demonstrated a likely relationship between cyanobacteria and phytotelm bromeliads in Ecuador. Incubated whole plants and the scrapings from the seepage zones on adjacent bark sometimes promoted substantial AR (Table 5.12). Assays of impounded fluids and adjacent plant parts indicated that the diazotrophs resided among the submerged epiphyllae. However, these colonies never grew as abundantly as those responsible for the gelatinous masses that sometimes clog the central tanks of Brocchinia tatei in eastern Venezuela (Givnish et al. 1984). A broad variety of sometimes heterocystic taxa were recovered from these Venezuelan specimens. A phytotelm bromeliad may favor N2 fixation well beyond simply providing a convenient vessel when its foliage absorbs the NH41, amino acids and other potentially autoinhibitory, low molecular weight nitrogenous compounds that some diazotrophs release into N-poor media (Benzing 1970b; Fogg et al. 1973). Plant-encouraged inputs of certain other required ions (e.g., K, P) from decomposing litter may further enhance diazotrophy. Should the absorptive foliar trichomes featured by these epiphytes parallel the transfer cells of certain other flora involved in more intimate exchanges (e.g., Anabaena and the water-fern Azolla), the case for mutualism rather than a fortuitous relationship becomes even stronger. Location of a nitrogen-generating system comparable to a biological chemostat in leaf impoundments could also help explain why many Type Three and Four Bromeliaceae grow so vigorously and provide such high-quality habitat for diverse canopy-based biota (Chapter 8). Feeding by ants Plant-feeding ants benefit certain Bromeliaceae through mechanisms that divide the myrmecotrophic members into two categories (Madison 1979; 214 Mineral nutrition Huxley 1980; Benzing 1991). One arrangement involves facultative to obligatory use of ant-provided rooting media (i.e., carton; Fig. 8.1C). Bromeliads rank among the better studied of these so-called ant-nest garden types, but even so reports provide at best only sketchy perspectives on a complex phenomenon that also involves numerous species of ants and diverse plants (also species of Araceae, Gesneriaceae, Moraceae and Piperaceae, among other families). Nest-gardens warrant closer scrutiny to determine how they operate. For example, do the tending ants, beyond providing substrates for the plants, deter, ignore or encourage herbivores (e.g., Homoptera; Fig. 8.2D)? Roots certainly reinforce the often brittle cartons, and the results of an experiment conducted by Yu (1994) suggest a second, potentially more decisive influence. Vegetated nests in Amazonian Peru shorn of their foliage, but otherwise left intact, incurred greater damage from heavy rain than unaltered controls, suggesting a sump pump-like action driven by transpiration. Within months, cartons deprived of their leafy extensions collapsed, forcing the ants to relocate. Other outcomes are more precisely documented. For example, ants responding to alluring fragrances, and sometimes edible appendages on seeds, assure plant dispersal from established to developing cartons (Chapter 6). Central to plant welfare, and our principal concern here, is the antconstructed rooting medium, specifically its chemistry, water-holding capacity and durability. Carton represents a complex maché with physical and nutritive attributes (Table 5.10) determined by the behavior of the architects and supplies of local building materials. Soil or feces sometimes receive high priority, whereas other Formicidae prefer fiber and similarly inert materials less accommodating to seeds (Davidson and Epstein 1989). Honeydew and the antibiotic secretions of the mandibular and other ant glands add complexity and potential insect-specific qualities to carton (Maschwitz and Holldobler 1970). However, the mycelia of at least one fungus, Cladosporium myrmecophilum, regularly permeate certain arboreal ant nests without obviously harming either the resident animals or the plants. Ants manipulate materials in tree crowns much as many of their relatives and termites do on the ground (Lobry De Bruyn and Conacher 1990). Certain arboreal Formicidae add to the debris accumulated by the impounding shoots of epiphytic Platycerium and Drynaria ferns in the process of improving these plants as nest sites (Koptur 1992). A diverse collection of opportunistic species colonize the Neotropical epiphytes, including many bromeliads (Chapter 7). Carton galleries crisscross much of the Mechanisms 215 bark in some Amazonian forests, allowing extensive arboreal flora without myrmecochores or special cavities for founding queens to also utilize antprovided substrates. Longino (1986) proposed that the ant-nest garden syndrome simply represents the most conspicuous manifestation of a widespread and general use of carton by epiphytic vegetation. Associations between termites and terrestrial and canopy-based Bromeliaceae come up again in Chapter 8. Members of the second group of bromeliads entice ants to supply nutrients to absorptive foliage rather than roots. Like the nest-garden flora, these ant-fed, ant-house bromeliads and counterparts in other families usually grow as epiphytes. The exceptions (e.g., Aechmea phanerophlebia, Brocchinia acuminata; Figs. 2.2E, 2.4G, 8.1D) anchor on rocks or infertile soil. In all, one or more species in at least seven families (Asclepiadaceae, Bromeliaceae, Melastomataceae, Nepenthaceae, Orchidaceae, Piperaceae and Polypodiaceae) reportedly produce myrmecodomatia (special hollow organs or cavities within more conventional plant parts to accommodate plant-feeding symbionts; Huxley 1980). But, unlike some of the other species engaged in this same kind of mutualism, ant-fed, ant-house bromeliads offer no extrafloral nectar or solid food primarily to support their associates. Low-cost housing represents the ant-house bromeliad’s single contribution to the welfare of its zoobiont, except where inflorescences and perhaps other vulnerable organs provide convenient substrates to farm Homoptera (Fig. 8.2D). In return, the myrmecotroph definitely obtains nutrients, perhaps with a modicum of protection included, but almost certainly less than received by certain better-known terrestrials (e.g., Acacia, Cecropia). Trees and shrubs of this second description shelter massive populations of pugnacious ants in probably nonabsorptive thorns and stems, primarily to deter herbivores. Unlike the relatively docile mutualists that inhabit bromeliads and some other epiphytes, those defending these small trees need not leave their hosts to search for additional food. Ant-house and ant-nest garden ants, by contrast, forage widely, acting as proxies for root systems incapable of exploring as much space as workers scour to feed their nest mates. Exchanges between ants and the bromeliads they inhabit or supply with rooting media remain little studied beyond some crude experiments. Calcium (45Ca), a phloem-immobile element applied to leaf bases, moved throughout the shoots of cultivated ant-house Tillandsia caput-medusae (Benzing 1970a; Fig. 5.2B). Ant-deposited materials extracted from chambers within the bulbs of this same species in Costa Rica, and provided as Mineral nutrition 216 Table 5.13. Quantities of elements intercepted over a 10-month period by sample bottles suspended in the crowns of Quercus virginiana near Tampa, Florida. Values are mg per bottle Sample number Element N P K Ca Mg Na 1 27.2 3.23 3.54 5.36 0.81 2.43 2 7.19 0.41 1.34 5.74 1.09 0.84 3 15.6 1.46 1.76 11.7 2.72 3.68 4 3.20 0.28 — 2.00 0.20 0.30 Source: After Benzing and Renfrow (1974a). an amendment to aseptic media, supported considerable growth by Aechmea bracteata seedlings. More comprehensive inquiries could characterize important aspects of bromeliad myrmecotrophy, specifically, ant contributions to nutrient budgets, consequences for plant fitness, and any peculiarities of ion sorption or N metabolism. Non-nutritional aspects of the ant/bromeliad symbioses receive more attention in Chapters 6 and 8. Nutrition that requires a phytotelma Plant reliance on phytotelmata fashioned from foliage (the phytotelma), although widely homoplasious among the ferns and angiosperms, nowhere exhibits as many interesting dimensions as in Bromeliaceae. Hundreds of species intercept moisture and nutrient-rich solids in cistern-like shoots, while the roots of these plants serve primarily for anchorage to bark or rock (Figs. 2.4, 5.1). Urine sample bottles set out as crude simulators for nearly a year in live oak (Quercus virginiana) trees in central Florida accumulated substantial nutrients, mostly derived from litter (Table 5.13). A short stem bearing numerous, channeled leaves, each with an expanded water-tight base, accomplishes the same outcome more effectively (e.g., Table 5.8). Solids recovered from a mature specimen of Guzmania monostachia growing in a Florida swamp forest yielded larger quantities of several essential elements than present in the tissues of the impounding shoot (Table 5.8). Humus present in the leaf axils of Vriesea platynema specimens early in the dry season at Rancho Grande, Venezuela contained N, P and Mechanisms 217 K at concentrations above those reported in soil under the same trees (Paoletti et al. 1991). Dissolved P occurred in the tanks of several Jamaican bromeliads between 0.1 and 0.51 ppm (Janetzky and Vareschi 1993; Fig. 8.13). However, no effort was made to determine how much of this reserve was immediately available (ionic) to the plant or still required processing by tank biota. Important aspects of nutrition differ among the phytotelm bromeliads according to the kinds of materials intercepted, and who prepares these inputs for plant use. In all instances, animals provide assistance, but the nature of that involvement varies, as does the fate of the participating fauna. Greater plant specialization and energetic cost accompany the utilization of prey compared with litter, so carnivory warrants coverage first. Carnivory Carnivorous plants produce traps often equipped with lures and digestive secretions in order to supplement the typically meager supplies of mineral nutrients present in soils where these plants grow. Impoverished substrates probably favored Darwinian modification of foliage to secure inorganic nutrients in addition to photons, but this change brought complications. As the costs of construction and operation increased while the leaf evolved the qualities necessary for carnivory, its photosynthetic capacity diminished. Consequently, the power to amortize investments in an organ that now provides two vital, but not particularly compatible, plant services also diminished (Givnish et al. 1984). The Givnish et al. model essentially explains the evolution of carnivory using the same economic paradigm described earlier to rationalize the relationships among leaf longevity, chemical makeup and cost, and capacity for photosynthesis. Two additional environmental factors influence the economics of botanical carnivory. Substantial drought or shade, both of which constrain photosynthesis, also render habitats unsuitable for the sundews, pitcher plants and other obvious prey-users as the model contends (Thompson 1981; Givnish et al. 1984). Protocarnivores, those flora that capture fauna with less expensive foliage than the more specialized leaf traps, should exhibit less stringent growth requirements. How many Bromeliaceae experience conditions that favor either condition remains obscure, but the numbers and distributions of the unequivocal prey-users suggest limited opportunity. Claims that a particular bromeliad is carnivorous sometimes rest on weak foundations. Wheeler (1921), in his classic writings on Neotropical 218 Mineral nutrition ants, said of Tillandsia species with hollow bulbous bases (i.e., ant-house species; Figs. 8.1D, 8.5) that ants ‘make fatal incursions into H2O-containing chambers’. Wheeler’s trapping sequence could not be corroborated using either Mesoamerican Tillandsia butzii or T. caput-medusae (Benzing 1970a). Dissected shoots displayed dry axils teeming with brood and adult ants, and repeated attempts to flood intact bulbs by immersion or spraying failed. Also present were those previously mentioned ant-deposited nutrients and the absorptive trichomes needed to exploit them. Picado (1911, 1913) was the first authority to provide experimental results to support claims about bromeliad carnivory. Amino acids added to the phytotelmata of several Costa Rican Tillandsioideae vanished as if absorbed through adjacent leaf surfaces. Picado also discovered proteolytic enzymes associated with mucilage, presumably the same product released by the damaged foliage and stems of many Type Four Tillandsioideae. No glands were apparent, nor did Picado demonstrate that these proteins came from the bromeliad rather than from co-occurring microbes. In more closely monitored studies, two typical phytotelm forms, Aechmea bracteata and a Nidularium hybrid, appeared to take up amino acids and perhaps bovine serum albumin from tank fluids (Benzing 1970b). However, the absorption of organic molecules similar or identical to those that degrading tissues release, although consistent with prey use, is not proof of this activity. Rees and Roe (1980) reported that giant Andean Puya raimondii, the largest of all the bromeliads and a colonizer of some of the harshest of the high Andean habitats supporting vascular flora (Chapter 7; Fig 14.2C), utilizes nutrients released from the carcasses of decomposing birds impaled on its well-armed foliage. Forced to use these exceptionally tall bromeliads for lack of other perches in barren landscapes, local avians often fall victim to the sharp, recurved marginal spines supposedly evolved to harvest them. Givnish et al. (1984) disagreed, suggesting instead that the responsible armature probably evolved its present character to thwart Andean bears whose descendants continue to eat many of the emerging inflorescences of other local Puya species, as do indigenous Homo sapiens. Additional bromeliads produce comparable spines (e.g., Hechtia, Bromelia, other Puya; Fig. 2.13A) that repel large herbivores seemingly without threatening birds. Heliophilic, phytotelm Catopsis berteroniana (Fig. 5.3A,B) ranges from southern Florida to southeastern Brazil, reputedly depending more on animal than plant biomass for nutrients (Fish 1976). Its shoots differ from those of most of the other Type Four species by a more upright stature, yellower color, and the presence of a copious, friable, epicuticle (Fig. 5.3A). Mechanisms 219 Leaf bases bear heavier coats of wax than the blades, and exposed sites at the tops of trees support the densest populations of vigorous specimens. Tanks contained relatively more animal remains and less litter than those of the other co-occurring phytotelm bromeliads located in similarly exposed microsites in Florida. Fish reasoned that UV light reflected from the cuticle helps trap prey, much as some pitcher leaves (e.g., Heliamphora; Fig. 5.3D) improve their catch aided by the same material. Supposedly, flying insects become confused as they cue on sky light according to the usual practice used to negotiate canopy obstructions. After colliding with a poorly outlined shoot, fauna tumble into its impounded fluids and drown, unable to escape because that same light-scattering wax also denies sound footing. Victims eventually decompose, releasing nutrients that enter the shoot via its absorptive trichomes. Plants produced no detectable digestive secretions, relying instead on symbionts to degrade prey. Catopsis berteroniana captured more insects than several other tank bromeliads while combinations of test plants were interspersed on fence posts during an experiment in Florida (Fish 1976; Frank and O’Meara 1984), but the suggested importance of UV reflectance to that outcome requires confirmation. Surveys of nutrients in plants and tanks and rates of interception are also needed to establish whether processed prey contribute significantly to plant welfare. If nutrients provided by nonsymbiotic fauna routinely eliminate or greatly reduce the requirement for litter, then designation as a low-grade carnivore seems reasonable. However, terrestrial Brocchinia reducta (Fig. 2.4F), and probably also closely related B. hechtioides, exhibit the greatest investments and specializations for prey use in Bromeliaceae. Brocchinia reducta ranges more extensively through the Guayanan highlands than most of its fewer than 20, mostly ground-based congeners (Chapter 9; Givnish et al. 1997). The largest populations inhabit moist savannas situated over porous, impoverished soils derived from Precambrian sediments of the Roraima Formation. Brocchinia hechtioides, a somewhat larger version of B. reducta, grows largely confined to the cooler, humid summits of the scattered, ancient table mountains (tepuis) of the Pantui known for their extraordinarily relictual and endemic biotas. Environments at both kinds of locations favor botanical carnivory according to economics (Givnish et al. 1984) and the composition of the co-occurring flora. Many communities inhabited by Brocchinia reducta and B. hechtioides also support one or more species of Drosera, Genlesia, Heliamphora and Utricularia. Collectively, these taxa constitute variety 220 Mineral nutrition unequaled elsewhere among the carvivores except possibly on similarly low-quality substrates in the most humid regions of southwestern Australia. Brocchinia reducta belongs to a group of carnivores defined by several aspects of prey use and a sometimes narrow food base. It and four cooccurring South American pitcher plants (all Heliamphora) mostly trap ants (but see below) attracted by color, fragrance and, in some cases, nectar. Sarracenin, an enoldiacetal monoterpene first identified as the fragrance lure for related Sarracenia flava, also assists Heliamphora, perhaps as one of a suite of fragrances that vary with the taxon and possibly among geographically distinct biotypes of the same species. A similar, perhaps identical, pleasant-smelling product characterizes B. reducta, although this plant does not utilize nectar like that presented on the distal appendage of the more brightly colored, red and green trap leaves of Heliamphora. Hallmarks also include a tall, tubular, uniformly yellow shoot densely covered on adaxial surfaces by a loose cuticular powder that impedes the escape of prey from the single, steep-sided phytotelma (Fig. 2.4F). Brocchinia reducta shoots examined by Givnish et al. (1984) in the Gran Sabana, Venezuela contained abundant, and unusually diverse, degraded prey. Exoskeletons representing 31 families in six orders filled the lowest 1–2 cm of each tank, but ants belonging to eight genera, all potential nectar-seekers, still constituted about 90% of the total catch. Mosquito larvae (Runchmyia and Wyeomyia) frequented shoots with impunity, fully tolerant of acidities (pH around 2.8–3.0) well below those recorded for any other bromeliad except B. tatei. Most indicative of carnivory among the many traits present, according to Givnish et al., is the fragrance. Association of the same odor with the crushed foliage of several noncarnivorous Brocchinia species suggested another, earlier role unrelated to the attraction of insects. Jaffe et al. (1992) detected proteolytic enzymes in the fluids contained in the young, unopened and older pitcher leaves of Heliamphora tatei, but not in the traps of its four relatives, the phytotelmata of co-occurring Brocchinia tatei, or an unidentified local Tillandsia species. However, pH was sometimes much lower in the phytotelmata of the bromeliads (3.4–5.7) than in those of the heliamphoras (.4.9). Gonzales et al. (1991) surmised that Brocchinia reducta competes for its primary prey with Heliamphora nutans on Kukenan-tepuy in southeastern Venezuela. Solenopsis sp., the only ant recorded at the study site, widely exceeded all other fauna in the traps of both plants (Table 5.14). Gonzales et al. (1991) assigned Brocchinia reducta generalist feeder status Mechanisms 221 Table 5.14. Percentage of plants containing arthropods in a patch of Brocchinia reducta and Heliamphora nutans in Kukenan-tepuy. Individuals per leaf tank (mean diameter) is given in parentheses Brocchinia reducta Inhabitants Diptera (larvae) Culicidae Wyeomyia sp. Chironomidae Metriocnemus sp. Prey Aranea Diplopoda Acarina Collembola Homoptera Lepidoptera Coleoptera Diptera Ceratopogonidae Culicoidel spp. (adults) Sciaridae Others Hymenoptera Formicidae Solenopsis sp. Chalcidoidea Large plants .40 cm (n55) Small plants ,25 cm (n56) Heliamphora nutans (x̄512 cm; n515) 0 (0) 0 (0) 47 (1.9)a 80 (14.2) 100 (33.2)a 67 (23.5) 80 (1.8) 0 (0) 20 (0.2) 0 (0) 60 (0.6) 0 (0) 80 (1.2) 67 (1.2) 0 (0) 50 (4.0) 100 (36.2)a 83(2.3)a 17(2.3) 0 (0)a 13 (0.2) 7 (0.1) 0 (0) 0 (0)a 0 (0)a 0 (0) 0 (0) 0 (0) 0 (0) 40 (1.2) 83 (11.2)a 100 (7.4)a 50 (0.8) 27 (3.6) 13 (0.5) 33 (0.4) 100 (28.8) 0 (0) 100 (144.7)a 100 (2.8)a 100 (159.0) 0 (0)a Source: After Gonzalez et al. (1991). Note: aIndicates statistically different samples (Mann–Whitney U-test) P,0.05; between small and large B. reducta (in second column) or between H. nutans and all B. reducta (in third column). because it captured the greatest variety of sometimes unexpected fauna. Among the partially digested arthropods in many sampled phytotelmata was a parasitic chalcidoid wasp lured for unknown reasons only to preflowering shoots. Phytotelmata of Brocchinia reducta and Heliamphora nutans both nurtured symbionts, primarily larvae of the potentially preyprocessing midge Metriocnemus sp. Young shoots of the bromeliad produced the strongest fragrance and contained the most abundant exoskeletal fragments. Substantial quantities of plant debris, including frass, suggested 222 Mineral nutrition that each Brocchinia reducta shoot becomes increasingly arthropod-fed after an initial carnivorous stage. Long-established genets bear several cohorts of interconnected ramets so may benefit from both mechanisms simultaneously. Joel (1988) considered Brocchinia reducta a likely Batesian mimic, not of flowers, but of sympatric and rewarding Heliamphora hederodoxa, which it resembles in size, shape, color, smell and nutritional mode, but not nectar secretion. A Batesian designation requires that the bromeliad offer no reward to the operator (potential prey), which it apparently does not, but still attract fauna seeking food. Insects that successfully visit H. hederodoxa pitchers may learn that objects of this general description offer sustenance, and, from that experience, end up even better primed to become prey for Brocchinia reducta. All other cases of mimicry among carnivorous plants involve, according to Joel, Müllerian mimicry. They and co-occurring species that bear nectar-producing flowers utilize a convergent strategy by offering rewards, but to different ends of course. These insect-pollinated angiosperms and the carnivores that also use nectar lures reinforce one another to the extent that pollinators drawn to traps survive that experience. Additional information indicates that the plants that utilize prey qualify as Batesian rather than Müllerian mimics. No evidence suggests that insects visit pitchers by mistake and no specific models seem to exist for any of the carnivores. Finally, recognized prey-dependent flora usually occur too abundantly at the exceptional locations that favor this life style to succeed as deceptive mimics. But then Brocchinia reducta sometimes vastly outnumbers Heliamphora hederodoxa where the two taxa grow interspersed in the Venezuelan Gran Sabana (personal observation). Obviously, the intricacies of Brocchinia reducta nutritional biology exceed the theory and data currently available to explain them. Carnivory represents but one way fauna enhance bromeliad nutrition, and the other mechanisms cost the plant less, occur more broadly through the family, and incorporate mutualisms rather than predation. Feeding by live ants takes plant economy a substantial step beyond the use of prey because nothing need be digested. Thompson (1981) suggested that more epiphytes engage in myrmecotrophy than carnivory because cost is low, and shade and drought so often prevail in tree crowns. Additionally, organs modified to house ant colonies tend to be durable, thickened stems that simultaneously provide mechanical and vascular support and even water storage, a need the epiphytes more often satisfy with succulent foliage. Finally, housing constitutes a scarce commodity for many tropical ants, Mechanisms 223 and that fact in turn probably added impetus to the emergence of plant cavities suitable as nest sites. Animal-assisted saprophytes Litter-dependent bromeliads employ the least complicated and arguably the cheapest arrangement in the family to use animals to assist nutrition (Figs. 2.4, 5.1, 5.2A). Simply put, moist habitat within a funnelform shoot readily sustains a rich microflora, abundant detritivores, and additional fauna that variously hide, feed or oviposit in comparative safety; in return, the host obtains substantial required ions in addition to a relatively constant supply of water. Given the ubiquity and ease of maintenance of biota capable of mineralizing phytomass, plant costs remain low. No myrmecochores or other ant foods are needed, nor must the bromeliad tolerate cultivated Homoptera. Whether vulnerability to herbivory exceeds that of the ant-house, and particularly the nest-garden, species remains unclear. Givnish et al. (1984) highlighted the importance of humus to the approximately one-third of all Bromeliaceae dependent on this medium by applying the label saprophyte, which, although accurate, fails to credit all of the players. So how does what I prefer to call animal-assisted saprophytism compare with botanical carnivory? Phytotelm bromeliads dependent on humus and the pitfall carnivores (pitcher plants) share some, but not all, features related to nutrition. Both groups rely on absorptive foliage and support symbiotic fauna (e.g., Okahara 1932; Plummer and Kethley 1964), but the relationships between plants and attracted animals differ. Most notably, the pitcher plants, at least those that secrete digestive enzymes (e.g., Nepenthes, some Sarraceniaceae), differ from litter-dependent Bromeliaceae in the relatively commensalistic vs. mutualistic nature of their interactions with symbionts. These two arrangements differ in complexity and probably also in importance to the plant. The often diverse collections of invertebrates that colonize bromeliad shoots augment and benefit from the activities of the microbes that decompose phytomass. Specifically, they shred litter, consuming some, and, to varying extents, feed on each other (Chapter 8). Microbes affect mineralization through the entire sequence, and probably promote the digestibility of the material ingested by the detritivores (Cummins et al. 1989). Bradshaw (1983) reported that arthropod larvae and a collection of lower organisms also hasten the consumption of drowned prey in Nepenthes and Sarracenia traps. But are these scavengers 224 Mineral nutrition really necessary for plant welfare, or would enzymes from the plant or associated microbes be adequate for that task? The occurrence of just two or three prey-users within a clade of about 3000 species poses another interesting question. Every other family with carnivorous members, some comprised of dozens of species distributed among two or more genera (e.g., Lentibulariaceae), include no noncarnivorous populations. So why does a device, the phytotelm shoot, that occurs so widely through Bromeliaceae so rarely exhibit the necessary embellishments for prey use? Why have just those two lineages, one in Pitcairnioideae (two species), the other in Tillandsioideae, independently become carnivorous? Constraints related to the costs of lures, enzymes and related devices provide a plausible argument for the general rarity of botanical carnivory, but can economics also explain why so few phytotelm bromeliads utilize animal rather than plant tissues? Botanical carnivory, while consistent in some respects (e.g., all participating plants occur on exposed, infertile and humid substrates), varies on two important counts. First, extrafloral nectaries, fragrances and digestive glands are optional; several combinations of plant characters, some lacking one or more of these three features, suffice for prey use. Second, these three and the other characteristics involved in the direct use of fauna for plant nutrition evolved repeatedly through redeployments of widely available genetic potential (Benzing 1987b). Modifications for carnivory in Bromeliaceae are notably simple compared with those of the more specialized prey-users (e.g., Dionea, Utricularia), and the component features provide unrelated services to close relatives. Many fully autotrophic bromeliads produce floral fragrances and friable cuticles, and their foliage is organized into a steeply tubular phytotelma much like Brocchinia reducta and Catopsis berteroniana (Figs. 2.4F,K, 5.3B). So why are only three bromeliads carnivorous when hundreds of additional species also maintain phytotelmata and grow on infertile substrates, in many cases with limited access to litter? Or maybe we should reverse the question: why do the fauna attracted to bromeliad phytotelmata ever promote plant welfare as prey rather than litter processors? Perhaps phytotelm Bromeliaceae engage fauna for nutrition in ways beyond carnivory and animal-assisted saprophytism as scattered reports suggest. We could be dealing with a collection of mechanisms that manipulate animals in different ways for the same plant benefits. Much could be learned about bromeliad nutrition by examining plants with different shoot architectures, and watching the animals that visit them. Two questions could guide these inquiries. First, are species beyond the three recognized Mechanisms 225 carnivores always animal-assisted litter processors as just described? Second, how much variety exists among the taxa truly dependent on phytomass in the ways they use this resource? An important consideration concerns the nutritional value and digestibility of the two substrates in question, viz. animal and plant biomass. Litter processing with plant-produced enzymes is untenable. Although vegetable, rather than animal, tissue constitutes the most abundant source of nutrient ions in the forest canopy, the former’s recalcitrant nature (e.g., abundant cellulose, lignin) and low quality (nutrient content) prohibit direct recycling from litter. Unsustainable investments in plant protein would be necessary to extract the same amounts of nutrient contained in the much smaller volumes of material secured by the carnivore. Therefore, reliance on mutualists accords with the ubiquity of that biota, the refractory nature of plant cell walls, and the massive amount of spent phytomass that settles cost-free on bromeliad shoots in so many habitats. Nevertheless, access to animal-assisted saprophytism poses problems for some Bromeliaceae, enough perhaps to favor uses of fauna in ways other than as prey or detritivores. Plant investments required to encourage mutualists to release nutrients from impounded phytomass are comparatively low, but yield probably varies with the element. Whereas unusually mobile elements such as K readily diffuse from dead vegetable and animal tissue, others remain more tightly bound in plant remains. Nitrogen, in particular, may mineralize too slowly to allow more than marginal harvest before the impounding shoot dies, according to assays of Guzmania monostachia in Florida (Table 5.8). Nitrogen to potassium ratios in the materials taken from shoots ranged from about 5:1 to 39:1 compared with about 1:1 to 2:1 for functioning tree foliage (Benzing and Renfrow 1974a). Much of the K formerly incorporated in that phytomass had either been recovered prior to abscission, entered the bromeliad, or been flushed from the phytotelmata before uptake could occur. Much more of the N apparently remained immobilized, perhaps significantly depriving the bromeliad. Shoot architecture, anchorages and climate favor specific kinds of phytotelm nutrition. Plants with spreading foliage (Fig. 2.4H) operate like filter-feeders, readily intercepting settling plant debris. Like their sessile, invertebrate counterparts (e.g., bryozoans, corals) in marine habitats, inputs vary according to body form and exposure, which on land influence the types of phytomass intercepted, and the processors the bromeliad will likely host. Laessle (1961) and Frank (1983) described broad differences in the animal and vegetable contents of tanks depending on the microsite. 226 Mineral nutrition Laessle labeled the associated nutritional modes ‘dendrophilous’ or ‘anemophilous’. Primary sources for anemophilous types remain obscure, but probably include more than wind-blown debris. High exposure and a spreading shoot promote an autotrophic community in phytotelmata, and shade, whether self-imposed or from without, encourages heterotrophy. Photosynthetic types might compete with the bromeliad for key ions (excepting N if diazotrophic). Shoot morphology and exposure also influence plant nutrition by affecting washout. Those broad shoots with channeled leaves that so effectively intercept litter (e.g., Fig. 2.4) sometimes overflow, although not necessarily to the detriment of the bromeliad, according to one experiment. A nonabsorbent dye placed in the water-filled axils of immature Aechmea bracteata plants indicated substantial, but incomplete, flushing during heavy showers (Benzing et al. 1972; Table 5.15). Solids and litter, and the biota needed to process it, presumably require even greater turbulence to dislodge. Shoots of other configurations would probably behave differently under the same conditions. Aspects of space and time also influence options for nutrition, and shed additional light on why litter rather than prey use prevails among phytotelm Bromeliaceae. On average, the Sarracenia purpurea leaf (Fish and Hall 1978), and presumably the traps produced by other pitcher plants, survive no more than one year, too little time to build up the complex communities needed to degrade abundant phytomass. Individual leaves of the phytotelm bromeliad may live no longer (Fish 1983), but they develop close together, and important fauna need move only short distances to migrate with their somewhat mobile habitats (Fig. 2.4). Closely connected, sympodial shoots, compared with the more widely separated trap leaves of many of the carnivores, further reduce needs to rebuild communities of detritivores. Climate that influences plant form, which in turn affects litter supply, predisposes certain bromeliads for useful contacts with drought-sensitive vertebrates rather than detritivores. Billbergia porteana (Fig. 2.4K), for example, illustrates how exposure and drought favor a shoot that accommodates neither the litter-dependent nor the carnivorous condition, but instead encourages a third arrangement with beneficial fauna. Upright stature and slender shape minimize exposure to direct-beam insolation and insulate the water supply enough to account for the over-representation of species with this form in strongly seasonal habitats (e.g., Brazilian caatinga; Fig. 1.4B). Tubular form that reduces capacity to intercept litter grants Billbergia porteana and its kind extraordinary opportunity to use amphibians that in Mechanisms 227 Table 5.15. Impoundment capacity of an Aechmea bracteata specimen, the recharge of that same shoot after being emptied, and the dilution of a dye solution in the same tanks by rain showers Percentage of dye Tank number by Percentage of tank remaining in the position from the shoot Capacity of the capacities filled by same tanks after center outward tanks (ml) 2.6 cm rainfall 2.1 cm rainfall Center 1 2 3 4 5 6 7 93 61 43 49 20 15 46 8 81 75 100 89 100 100 100 63 39.5 16.0 59.5 35.0 62.0 66.0 46.5 67.0 Source: After Benzing et al. (1972). some cases already bear evidence of extended and intimate associations with bromeliad shoots. Particularly striking are the exceptionally flattened heads that certain frogs employ to close off the narrow phytotelmata characteristic of several Brazilian species (Fig. 8.4E,F). Prolonged occupancy during dry weather virtually assures the inhabited bromeliad a supply of nutritive excrements. Statistics on visitations and animal outputs relative to plant requirements would help evaluate the biological significance of these relationships. Additional discussion of how the architecture of the bromeliad shoot influences where these plants grow appears in Chapter 7. Tank soil-root bromeliads Certain Ananas and Bromelia species (Fig. 2.14A,B) bear mention because they probably resemble family ancestors in aspects of structure, function and ecology. Pittendrigh (1948; Table 4.2) reserved his Type Two designation, the tank-root type, for these Bromelioideae because roots often ramify more extensively among adjacent leaf axils than through underlying soil. Trichome-covered leaf bases augment uptake for Ananas comosus (Sakai and Sanford 1979), although the amounts of impounded debris often seem inadequate to totally satisfy plant needs. Ecophysiology further distinguishes these plants. In addition to moderate succulence, recessed stomata, and thick cuticles – in effect a basically xeromorphic character – crassulacean acid metabolism (CAM) assures the water economy needed 228 Mineral nutrition to survive lengthy dry seasons in typically warm, lowland habitats (Fig. 2.13B). Studies of Ananas comosus and several Bromelia species have revealed complex relationships among plant vigor, light response, N status and relative emphasis on CAM vs. C3 metabolism (e.g., Fetene et al. 1990; Medina et al. 1991a,b; Chapter 4). Controversy about how phenotype reflects growing conditions now as compared with the past continues. Responses in situ and in the laboratory suggest that members of both genera, if not the entire subfamily, share a decidedly shade-tolerant stock. Whatever the nature of the ancestral habitats, extant wild types utilize high photosynthetic photon flux density (PPFD) less effectively than some other CAM plants (e.g., certain Agave, Opuntia; Nobel 1991) that, along with similarly heliophilic pineapple cultigens, match the productivities of several C3 and C4 crops, in some cases exceeding them in water-use efficiency (WUE). Chapter 4 provides additional details on photosynthesis and water balance in Ananas and Bromelia. Briefly, N invested in foliage – that reliable predictor of Amax mentioned earlier – diminished in situ in Venezuela at high PPFD whether expressed on a leaf area or weight basis. Medina et al. (1986) and Medina et al. (1991a,b) suggested two potential causes: greater soil fertility in understory habitats and more structural carbon in sun vs. shade leaves, i.e., a dilution effect (see also Maxwell et al. 1995). Water and temperature stresses purportedly further curtailed growth in fully exposed microsites, a claim supported in most comparisons by less negative D values in biomass, which indicate greater reliance on CAM. Medina et al. (1991a,b) made no mention of a third agency that probably helps explain why foliar N diminishes with exposure. Strong fluence photoinhibited some Bromelia humilis specimens enough to require several hours in shade for recovery (Medina et al. 1986); more severe stress may chronically reduce N concentrations. Sufficiently overexposed foliage requires weeks to months to replace the labile D1 proteins associated with photosystem II and restore quantum efficiency (Long et al. 1994). Maxwell et al. (1995) reported dramatic reductions in the chlorophyll content in Guzmania monostachia moved from partial to full sunlight (Fig. 4.26). Perhaps a better-protected light-harvesting apparatus, irrespective of N content, accounts for the relatively heliophilic nature of certain Ananas comosus cultivars. Uneven capacity to dissipate excess light energy via xanthophyll-cycle activity may also explain why specific Bromeliaceae perform differently in full sun. Involvement of foliar trichomes 229 Involvement of foliar trichomes Absorbing foliar trichomes operate under a variety of conditions, and structure and function varies to match specific circumstances. Those appendages lining the leaf bases of the phytotelm bromeliad remain continuously bathed in nutritive fluids. At the opposite extreme, no degrading biomass, ant products or earth soil help sustain the hundreds of nonmyrmecotrophic Type Five, mostly epiphytic and lithophytic Tillandsioideae (Fig. 5.1). Uptake occurs solely when precipitation bathes otherwise dry shoot surface equipped with dense indumenta; these same foliar organs mediate additional services during dry weather. Roots, when present, provide holdfast only (Figs. 2.1, 2.10). Our concern here is the uptake of water and nutrient ions by trichomes, primarily absorption of the second of these two resources. Water relations Chapters 2 and 4 describe the foliar scale of Tillandsioideae as a peltateshaped organ comprised of a shield or plate of usually dead cells anchored to the epidermal basement by a living stalk (Fig. 2.7A,B). Briefly, four large, equal-sized central cells dominate the center of the shield and secure it to the dome cell, which constitutes the distal member of that subtending, uniserrate stalk. Extraordinarily thick tangential walls of the central disc alternately rise and fall on flexible radial walls as precipitation and evaporation alternately fill and empty the underlying lumina. Several additional rings of cells, each made up of twice as many members as the one within, surround the four central cells. An outermost, asymmetrical wing contains many more and much more elongated cells than those within (Fig. 2.7D). Absorption occurs while ion-charged fluids contact those parts of the shoot bearing trichomes. Ultrastructure reveals that the dome cell is well equipped to mediate ion uptake. Moreover, its boundary lies just microns away from the nutrientcharged fluids that periodically to continuously engorge the central disc. Diverse organelles, especially rough endoplasmic reticulum, dictyosomes, microtubules and mitochondria, densely fill the protoplast (Dolzmann 1964, 1965). An elaborately folded plasmalemma characteristic of plant transfer cells assures intimate contact with abundant electron-dense material located just within the cell wall (Brighigna et al. 1988). Concentrations of plasmodesmata at every junction along the stalk allow extensive communication with the mesophyll. A parallel apoplastic conduit 230 Mineral nutrition also seems likely, given the absence between cells of the cuticle that invests the outer walls of the stalk. Several workers, including Haberlandt (1914) and Mez (1904), demonstrated that hypertonic salt solutions applied to intact leaf surfaces plasmolyze the mesophyll cells adjacent to the bases of the affected trichome stalks. Vital stains followed the same route, but more compelling evidence of the involvement of the foliar indumentum in nutrition would require the more sophisticated techniques that would not become available for many more decades (e.g., Benzing et al. 1976; Ehler 1977; Owen et al. 1988). Water relations would also attract continuing attention. For example, Brighigna et al. (1988) noted little difference in the ultrastructure of the stalk cells of Tillandsia usneoides whether leaves had been fixed following incubation for eight days at 80% relative humidity or desiccated up to 23% over silica gel. Foot cells of the better-hydrated samples contained larger vacuoles. These investigators also reported that the shields of mature trichomes sometimes retain their protoplasts, a rarely reported condition that challenges the classic explanation of how the foliar scale operates as a oneway valve. Rather than beading up, a drop of moisture placed on the leaf of a Type Five Tillandsia spreads from one trichome shield to the next, in turn, causing the upper walls of the central discs to rise and the wings to flatten as the lumina fill with water (Fig. 2.7A,B). Mez (1904) imputed an accompanying suction mechanism (hence his term ‘trichomepump’) during engorgement without demonstrating that force. Moisture subsequently fluxes from the charged shield into the dome cell and on to the mesophyll until either the water potentials inside and out equilibrate or the indumentum dries. In the second case, the lumina of the central disc collapse, restoring the barrier provided by the thickened walls of the central disc. So configured, the shield prevents water from wicking out the leaf along the path of entry. Reflexed upward, the wing again scatters light, and the silvery, rough texture that highlights the shoots of Type Five Bromeliaceae returns. In effect, trichomes of the type that invest the foliage of Type Five Tillandsioideae serve as one-way valves and energy dissipaters, alternately hydrating the plant and insulating it against water loss, photoinjury and excess heating. Controversy continued for many years over whether trichomes appreciably amend water deficits from adjacent moist air (Chapter 4). Plants do gain moisture by this route, but not enough to replace the need for contact with liquid moisture (Garth 1964; De Santo et al. 1976; Benzing and Pridgeon 1983; Martin and Schmitt 1989; Figs. 4.12, 4.22). Involvement of foliar trichomes 231 Sorption of solutes Bromeliads with essentially mechanical or no root systems accumulate nutrients primarily while precipitation, canopy washes or impounded solutions bathe the foliar trichomes. Potentially beneficial gases (e.g., NOx, NH3, SO2) may also enter foliage through stomata, and perhaps in sufficient quantities for the slowest growers to significantly reduce dependence on the indumentum (e.g., Ziereis and Arnold 1986). Vapor accounts for mercury contamination and perhaps some of the burdens of certain additional ‘technological metals’ discussed below. However, experiments conducted to date dealt exclusively with absorption from prepared solutions over intervals that extended from less than an hour to several months. Individual cells to whole plants were targeted. Trichomes of Pitcairnioideae exclusive of one genus possess relatively low-grade organization and exhibited little, if any, capacity to absorb ions, according to autoradiographs obtained from the few Pitcairnia species examined so far (e.g., Benzing et al. 1976; Fig. 2.5). Brocchinia constitutes the exception, perhaps reflecting an erroneous taxonomic assignment, or, in some of its species, habits and ecology conducive to this kind of trichome involvement that occur nowhere else in the subfamily. Novel trichome structure, which nevertheless varies substantially among the fewer than 20 described species, clearly distinguishes Brocchinia from the balance of Pitcairnioideae (Fig. 2.5). Absorptive trichomes accompany carnivory, myrmecotrophy and litter use in Brocchinia (Figs. 5.2F,G). Trichomes born by Type One B. prismatica also accumulated label in assays using 3H-leucine, while those of litterimpounding B. tatei and B. micrantha were inactive by comparison. Perhaps significantly, trichomes occur at the highest densities and cover the highest percentages of the leaf surface where they assist uptake by carnivorous B. reducta (Givnish et al. 1997; Chapter 9; Table 9.1). Still, the presence of absorbing hairs on nonimpounding relatives like B. prismatica and Steyerbromelia diffusa (also Type One) suggest that closer study would reveal some capacity for trichome-mediated nutrition in all of the Brocchinia species equipped with phytotelma. Ananas comosus provides the most complete picture of how the trichome serves at least some Bromelioideae. Sakai and Sanford (1979) reported that its leaf bases bear overlapping scales, each comprised of a living, two-celled stalk topped by a multicellular, dead shield. Numerous plasmodesmata connected the dome cell with the underlying mesophyll as in Brocchinia and Tillandsioideae. Additional shared structure changed with conditions. 232 Mineral nutrition Similar to Dolzmann’s (1964, 1965) observations on Tillandsia usneoides, leaves fixed while surface-dry featured layers of electron-dense material between the plasmalemma and walls and abundant granules around the periphery that periodic acid–Schiff base reaction indicated was a polysaccharide. Neither substance was apparent in samples that had been presoaked in water for 12 h. Larger numbers of smaller mitochondria with swollen cristae further suggested altered cell status likely related to absorptive function. Dictyosomes and spherosomes remained scattered as before. Trichomes lining the phytotelmata of at least some Type Three Bromelioideae promote nutrition, according to several studies. Resolution varied with the experiment. Amino acids and bovine serum albumin added to the tanks of Aechmea bracteata diminished over time as indicated above (Benzing 1970b). Microbes probably metabolized some of these supplements, but could not explain higher leaf N at the end of the runs. Two studies (Burt and Benzing 1969; Nadkarni and Primack 1989) designed to compare the absorption of several radionuclides provided as inorganic salts by roots vs. shoots confirmed foliar involvement, although not necessarily via trichomes. Several ornamental taxa fertilized in similar fashion to determine optimum culture (Sieber 1955) proved about as competent to feed by either route. Autoradiography demonstrated 3H from labeled leucine in the trichome stalks of Neoregelia sp., although several Tillandsioideae treated at the same time exhibited more impressive accumulations (Benzing et al. 1976). Brocchinia reducta, the best-known member of its ecologically diverse genus, provides the most thorough appraisal of the structure and function of any bromeliad trichome (Owen et al. 1988, 1991; Owen and Thomson 1988). Cells comprising the distal portion of the atypically goblet-shaped organ, about 30 in a radial array, remain alive following maturation if bathed by impounded fluid (Figs. 2.5A,B, 5.2F). A labyrinthine array of electron-dense and more translucent zones mark the outer tangential and radial boundaries of these distal-most cells, especially those that interface with the phytotelmata (Fig. 2.5C,D). Canaliculate spaces represented by the translucent regions disappeared if leaf surfaces were dry prior to embedding in plastic. Studies conducted to determine the permeabilities of the Brocchinia reducta trichome proved especially enlightening. Dextrans conjugated with fluorescent dyes confirmed the presence of pores in hydrated cap cell walls with Stokes diameters of at least 6.6 nm, sufficient clearance for small to medium-sized proteins or the partially hydrolyzed products of larger molecules. Plant secretions, including enzymes, might pass in the opposite direc- Involvement of foliar trichomes 233 tion through the same channels. Vigorous protein synthesis would explain the numerous polysomes in the adjacent protoplasts (Owen et al. 1988). However, final assessment of trichome involvement in carnivory in B. reducta requires determination of whether the plant or its symbiotic microbes digest the prey. Diffusing lanthanum confirmed the existence of an apoplastic conduit extending from the phytotelma through the trichome to the underlying mesophyll. Lucifer yellow, a fluorescent tracer, revealed the parallel symplastic pathway and, for the first time, endocytotic uptake of a nontoxic substance by a walled, whole cell (Owen et al. 1991). Again, entry occurred exclusively via the distal stalk cells after passage through gaps in the thin cuticle (Fig. 2.5C). After traversing the canaliculi, label accumulated in the adjacent plasma membrane, specifically in coated invaginations, prior to migration into the cytosol. Owen et al. (1991) also detected Ca-precipitated pigment in coated and partially coated vesicles, either free in the cytosol or associated with the dictyosomes. Occasionally, accumulations occurred in tubular and swollen elements of the smooth endoplasmic reticulum. Fluorescence in the lumina of the dictyosomes accorded with the passage of label through the Golgi apparatus. Tested amino acids performed differently (Owen and Thomson 1988). Leucine moved most freely through the system, eventually emerging as a visible, insoluble osmium/leucine complex within the cristae of mitochondria, on the surface of lipid bodies, and less often in spaces associated with the tubular invaginations of the plastid interenvelope. Glycine only penetrated to the matrix of the mitochondria in some cells. Arginine-treated trichomes contained complexed deposits in the labyrinthine channels alone, perhaps because the plasmalemma lacks capacity to transport this more structurally complex metabolite. Nyman et al. (1987) demonstrated metabolic involvement in the accumulation of numerous amino acids through the intact leaves of Tillandsia paucifolia, presumably via trichomes. More extended studies illustrated that this same Type Five epiphyte, minus its few roots, can concentrate a variety of inorganic ions (Benzing and Renfrow 1980). Daily 0.5-h immersion in nutrient solutions brought about a 20-fold increase in P content within 120 days. Levels of N and K also rose, but not as much. Identical contact with equimolar solutions (1025–1027 M for trace elements) killed every individual within 60–90 days. Post hoc examination revealed Cu concentrations (dry weight) up to 20-fold above those in controls; Zn and Mo contents also increased substantially (Table 5.3). Concentrations of B, Fe and several macronutrients changed little or not at all. 234 Mineral nutrition Figure 5.5. Uptake of three plant nutrients by leaf discs of five bromeliads over 3 h. Numbers over the bars depicting P accumulation represent the combined densities (mm22) of trichomes on both leaf surfaces (after Benzing and Pridgeon 1983). Additional experiments using labeled inorganic nutrients compared more than 20 Bromeliaceae representing every ecological type and all three subfamilies (Benzing and Burt 1970; Benzing and Pridgeon 1983). Autoradiography differentiated another set of subjects by subfamily and architecture within Tillandsioideae (Benzing et al. 1976). In the first instance, segments of leaf blades of several Type Five Tillandsia accumulated more 45Ca, 32P, 35S and 65Zn from treatment solutions during 3–12-h runs than did comparable samples excised from sparsely trichomed, tankforming Catopsis nutans (Benzing and Pridgeon 1983; Fig. 5.5). Tested Bromelioideae and Pitcairnioideae also exhibited lower affinities for the ions provided. Following exposure to 3H-leucine for 0.5 h, the trichome stalks of all treated Tillandsioideae contained abundant 3H, whereas the adjacent epidermal cells and nonliving shields remained unaffected (Fig. 5.2H). Trichomes of the included Bromelioideae and Pitcairnioideae took up much less or no label (except for several Brocchinia species as indicated above). Without question, the foliar trichome has played a decisive role in bromeliad radiation. Why this appendage acquired its capacity for absorption remains unclear and controversial. Conceivably, nutritional insufficiency rather than drought was responsible for the evolution of the absorptive capacity that allows extant Bromelioideae and Tillandsioideae to play so many important roles and anchor on diverse kinds of often unyielding sub- Nitrogen nutrition 235 strates. Pittendrigh (1948) suggested arid habitats, but access to nutritive ions in phytotelmata may have been more influential (Benzing et al. 1985; Chapter 9). A reconstructed phylogeny would help determine how carnivory, myrmecotrophy and animal-assisted saprophytism relate to one another and to the absorbing trichome, and when and how often each of these features evolved. Nitrogen nutrition Nitrate, sometimes augmented by substantial NH41, constitutes the bulk of the N supply for most land flora. However, low temperature retards mineralization enough in certain Arctic ecosystems to permit at least one tundra species to meet about half of its requirement from organic sources (Chapin 1993). Nonmycorrhizal Eriophorum vaginatum (Cyperaceae) took up several amino acids during experiments designed to test relative availability. Growth was superior on the organic supplements compared with either NO32 or NH41, in part owing to more favorable absorption kinetics and only modest capacity to assimilate nitrate. Combined supplies of the most abundant amino acids in native soils exceeded concentrations of the prevailing inorganic species (2–8 vs. 0.5–1.1 mg Ng21). Myrmecotrophy, carnivory and the other variations on tank nutrition, and humic rooting media similar to the substrates that support many boreal plants suggest that certain Bromeliaceae might also possess special capacities to utilize organic N. Facility to use nitrate vs. ammoniacal N may also differ among species depending on the pH of substrates in native habitats. Decomposing litter, excrement from ants and other symbionts, and digesting prey all increase the variety of N-containing compounds adjacent to root systems and foliar trichomes. Those amino acids Picado (1913) added to the tanks of some Costa Rican bromeliads, the similar demonstration by Benzing (1970b), and Owen et al.’s (1988) findings on Brocchinia reducta trichomes all indicate permeability and physiology consistent with unusual N nutrition. Certain Bromeliaceae possess distinct capacities to utilize oxidized vs. reduced inorganic N in aseptic culture, but these findings must be considered relative to conditions in situ to assess possible biological significance (Benzing 1970c). Ammoniacal nitrogen usually prevails at low pH, while nitrate predominates under more neutral conditions. Humus-rich substrates support many of the terrestrials (e.g., Cryptanthus, Orthophytum), and the bryophyte-based mats at cooler, wetter montane sites harbor a relatively sparse microflora in addition to sometimes abundant Tillandsioideae Mineral nutrition 236 Table 5.16. Growth by three bromeliads representing three ecological types on diverse sources of N plus all other required nutrients expressed as percent dry weights relative to those of 2N controls. The N sources were provided at 0.002 M nitrogen equivalent concentrations Aechmea bracteata (Type III) Vriesea jonghii (Type IV) Pitcairnia andreana (Type I) 251 231 266 83 253 245 177 167 244 80 164 151 217 215 170 138 169 43 152 117 110 53 161 44 106 63 105 140 244 224 249 78 129 83 91 89 252 94 162 86 76 181 (NH4)2SO4 KNO3 (NH4)2SO4 1KNO3 Alanine Asparagine Aspartic acid Citrulline Glycine Glutamine Glutamic acid Ornithine Tyrosine 15 amino acids Urea Source: After Benzing (1970c). (Vance and Nadkarni 1990). Phytotelm types may experience preponderances of NH41 because the fluids they impound typically range between about pH 3.5 and 6.0. Finally, NH41 frequently exceeds NO32 in precipitation (e.g., Coxson and Nadkarni 1995). Different responses of the seedlings of five bromeliads representing Types One (Pitcairnia andreana, Puya mirabilis) and Three and Four (Aechmea bracteata, A. recurvata, Vriesea jonghei) to aseptic media containing amino acids, amides, urea and NH41 and NO32 salts as sole sources of N may reflect the compositions of the supplies available to these plants in situ (Benzing 1970c; Table 5.16). Phytotelm types developed rapidly on a greater variety of organic sources than soil-rooted forms. Glutamine consistently promoted growth compared with N-free controls, and usually about as effectively as NH41 and NO32 alone or combined. Other amino acids (e.g., alanine, glutamic acid) were inhibitory. Aechmea bracteata, which regularly hosts ant colonies and collects litter in phytotelmata, thrived on the greatest number of organic species (Fig. 2.4G). Aechmea recurvata and Vriesea jonghei responded somewhat similarly and the two Type One Pitcairnioideae proved least able to substitute organic for inorganic N. Nitrogen nutrition 237 Figure 5.6. Course of nitrate reductase induction in the roots and foliage of Vriesea hieroglyphica fed Ca(NO3)2 through the same organs (after Nievola and Mercier 1996). Mercier (1993) recorded a more extensive array of plant responses to N supply using glutamine and urea, vs. reduced and oxidized inorganic sources in sterile culture. Seedlings of Pitcairnia flammea and Vriesea philippo-coburgii represented Types One and Four respectively, while Tillandsia pohliana (Type Five) added one of the most leaf-dependent bromeliads for comparison. Together, NH41 and NO32 always promoted vigorous growth, as did glutamine. Concentrations of phenolics, free NH41, and various growth factors in tissues also varied among the three subjects depending on the N source. Root formation in Vriesea philippo-coburgii and Tillandsia pohliana was delayed by (NH4)2SO4. Conversely, Ca(NO3)2 and NH4NO3 stimulated growth, perhaps consistent with measured changes in the relative concentrations of endogenous cytokinins. Whether or not the biota in phytotelmata generate more or less NH41 compared with NO32, Vriesea hieroglyphica specimens fed the second source either through the root system or through the shoot developed considerable nitrate reductase activity (Nievola and Mercier 1996; Fig. 5.6). 238 Mineral nutrition However, induction occurred more slowly (4 vs. 2 h) in roots than in foliage. The type of supply (e.g., Ca(NO3)2 vs. (NH4)2SO4) also influenced the activities of additional N-processing enzymes, namely glutamate dehydrogenase (GDH) and aspartate aminotransferase (AAT) in Pitcairnia flammea, Vriesea philippo-coburgii and Tillandsia pohliana. Differences consequent to feeding were most pronounced in the electrophoretic mobilities of GDH. Urea inhibited AAT in terrestrial Pitcairnia flammea but promoted activity in epiphytic Tillandsia pohliana. While somewhat difficult to relate to plant habits and conditions in nature, Mercier’s findings demonstrate distinct N relations among three ecologically divergent bromeliads. Architecture and nutritional economy Additional aspects of morphology beyond the inflated leaf bases, that when closed favor myrmecotrophy and when open the capture of litter and its processing by symbionts, also promote bromeliad nutrition. Benefit from these other arrangements accrues through increased mineral-use economy. Once again, the most conspicuous examples come from Tillandsioideae. Especially noteworthy are the Type Five species, which collectively constitute an evolutionary grade whose members probably emerged repeatedly from more mesic stock characterized by phytotelm architecture (Gilmartin and Brown 1986; Chapters 9, 12 and 13). Tillandsioideae include species that possess well-developed to nearly nonexistent root systems well suited respectively for relatively equable to harsh anchorages. Shoot organization parallels this trend with progressive miniaturization accompanied by either reduced numbers of vegetative nodes per ramet or greater caulescence and expanded leafiness (Fig. 2.1). Tillandsia usneoides exhibits the most abbreviated of all the derived bauplans with its nearly root-free ramet, which at maturity bears just one leaf and a single prophyll prior to the appearance of a solitary, terminal flower (Figs. 2.1, 2.10E). Structural reduction illustrated by Tillandsia subgenus Diaphoranthema, which includes T. usneoides, and T. bryoides (a notably caulescent form; Fig. 2.1), prompted Tomlinson (1970), among others, to invoke heterochrony without comment about causes, mechanisms or consequences. Improved material economy offers a plausible explanation for this phenomenon because it accords with two unrelated growing conditions, and the fundamental capacity of vascular flora to emphasize root or shoot development according to plant needs. Moreover, current growing conditions indicate that these bromeliads are ideal candidates for the kind of hetero- Architecture and nutritional economy 239 chrony that explains the array of body plans exhibited among extant Tillandsioideae. Two benefits potentially accrue to any vascular plant that, by virtue of morphology and habitat, requires just one rather than two organ systems to conduct basic vegetative functions, viz. absorb water and essential ions and harvest photons and CO2. Plants so organized can deploy biomass in patterns unavailable to more conventionally organized flora. Plants with shoots, but few or no roots, also potentially require less time to mature than others obliged to allocate resources more evenly between green and nonproductive (root) tissue owing to the principle of compounding interest. A shortened life cycle in turn favors rapid population growth, which, combined with the enhanced inputs for propagules made possible by rootlessness, increases capacity to exploit habitats that impose relatively high rates of mortality on juveniles (e.g., Cole 1954; Benzing 1981a). More succinctly, economies of time and material inherent to the body plan of the Type Five bromeliad should promote fitness wherever drought or infertility slow carbon gain and scattered safe sites, disturbance or any other agency that kills large numbers of pre-reproductive individuals mandate compensatory (elevated) fecundity (Benzing 1978a). Dry-growing Tillandsia experience some of the most stressful (nonproductive) of all the environments colonized by land flora, and they often root on unstable media that mandate rapid plant cycling. Small, mobile seeds, a consistent characteristic of Tillandsioideae, also accord with the scattered and ephemeral nature of bark, which is the substrate these plants use more frequently than any other. The multifunctional shoot combined with a largely vestigial root system not only help explain why Bromeliaceae dominate so many arboreal and lithic floras in the American tropics, but also abrogate a basic principle of morphology. Plant success on land beginning some 450 million years ago required diverse novelties (e.g., vascular tissue, cuticle), and that the sporophyte be reorganized to accommodate two rather than the single ancestral aquatic medium. Part of the plant body now occupied an energy-rich, but drying, environment, namely the atmosphere, while the other portion was, and continues to be for most modern flora, relegated to dark, moister and more nutritive space. Exceptions include the fully exposed epiphyte and lithophyte that once again inhabit relatively uniform ecospace where the longstanding need for shoot/root system differentiation no longer applies. Similar opportunities and corresponding conditions mark the more retrograde flora (e.g., rootless Ceratophyllaceae), and with fewer functional trade-offs given the more forgiving nature of water vs. the atmosphere as a growth medium. 240 Mineral nutrition Whatever advantages prompted evolution from Type Four to Type Five status, ecological latitude diminished in certain other respects as stress-tolerance increased. Absorptive trichomes with large, hydrophilic shields complicate life in shady habitats and preclude existence in overly humid ones (Benzing et al. 1978; Fig. 4.11; Table 4.8). Additionally, rootlessness may exacerbate certain metabolic problems, particularly ion balance, specifically the need to dispose of excess of either H1 or OH2, depending on the N source (oxidized or reduced). Many plants avoid damaging buildups by dumping potentially toxic inorganic ions through extensive root systems (Raven 1985, 1988). Type Five Tillandsioideae may avoid this complication by growing slowly or perhaps they render these products harmless by some other mechanism. Bromeliads as air quality monitors Instruments provide most of the data on air quality; flora constitute a cheaper alternative, but are not without disadvantages. Plants differ in their avidities for certain naturally occurring and anthropogenic substances depending on the species, the physiological status of the specimen, growing conditions and a host of additional variables. For example, assays of contaminated foliage reflect microrelief and electrical attractions between plant surfaces and charged aerosols. Coatings on foliage and other organs continuously change as surfaces saturate independent of the mix of substances moving around them. At best, the composition of plant tissue offers a biased, time-averaged record of the chemical environment of that organism; it does not mirror the performance of the mechanical device designed to strip air of all of its contaminants. Nevertheless, absolute and relative concentrations of certain components in ashed tissue represent useful data. For example, abnormally elevated levels of certain technological metals indicate anthropogenic contamination. Ratios of Na to Al or to some other element abundant in inland soils signal marine influence. Elements that fluctuate less in concentration relative to tissue dry weight than to percent ash, a value particularly sensitive to the thickness and composition of coatings on plant surfaces, are probably nutrients. Factor and principal component analyses more precisely define associations of elements that distinguish anthropogenic from natural sources, and differentiate constituents essential for life from those potentially harmful to organisms. A substantial literature indicates that lichens exceed most vascular flora as worthwhile devices for air quality surveillance, and reasonably so. Most Bromeliads as air quality monitors 241 higher plants root in soil, and hence contact a source of many of the same elements that concern the environmentalist. Additionally, lichens often respond more sensitively to the gases and metals that pollute ecosystems and threaten public health. Industrial and automotive emissions already account for the collapse of formerly dense populations of lichens near many industrial and urban sources. Various conifers also exhibit exceptional susceptibility to certain air pollutants (e.g., O3), but being essentially nontransportable as adults, are useless for many applications. Unlike most vascular flora, the epiphytes are small and easily relocated. They also grow above ground and subsist on nutrients from aerial rather than soil-based sources. Furthermore, pollutants impact some of these plants as much as they do the lichens, but with broader possibilities to quantify the effects. Stomatal conductance and associated physiology unique to vascular flora offer special opportunity for inexpensive and nondestructive diagnosis of sublethal injury and subsequent plant recovery. Spanish moss was the first bromeliad to provide information on atmospheric chemistry, and it and several congeners continue to stock most of the surveys that employ vascular epiphytes. Wherry and Capen (1928) examined ashed shoots for signs of contamination in Florida. Martinez et al. (1971) and Robinson et al. (1973) used the same approach and bromeliad to determine that lead from auto exhaust occurred at extraordinary concentrations in a number of roadside collections; T. usneoides also revealed Ni contamination via aerosols near a battery plant in South Carolina (Carcuccio et al. 1975). Schrimpff (1981) employed T. recurvata to map polluting metals, pesticides and aromatic hydrocarbons at two locations in Colombia, South America. Burdens of several metals in Spanish moss increased toward a petrochemical complex in southeast Texas (Benzing 1989). Total S in the foliage of Tillandsia balbisiana, T. paucifolia and T. utriculata reached highest concentrations in collections taken closest to the urbanized coast of southeastern Florida (Benzing and Bermudes 1992). Spanish moss revealed alarming contamination from Hg vapor emanating from a gold refining operation in Brazilian Amazonia (Calasans and Malm 1994). Exposures of a few months within and near processing sheds elevated concentrations from less than one up to 60 ppm! Contaminated clothing worn by the employees who boil off metallic Hg spread the risk off-site. Calasans and Malm (1994) also placed baskets of T. usneoides inside the electrolysis chambers of a chlorine-soda facility in Brazil for 15–68 days and recovered Hg up to 13000-fold (30–35 mm g21 dry weight) above levels in control plants. Other samples maintained outside the factory contained 5–175 times ambient concentrations. 242 Mineral nutrition Experiments and horticultural practice indicate extraordinary responses among Bromeliaceae to a variety of toxic substances. Arboriculturalists formerly sprayed lead arsenate to selectively kill T. usneoides and T. recurvata infesting shade and orchard trees in Florida. Copper salts continue to serve the same purpose. For example, two applications during one growing season of Cu(OH)2 at 35g l21 eliminated T. recurvata on crape myrtle in southern Louisiana (Holcomb 1995). More costly Cu-based fungicides have been employed for many years to control ballmoss in Texas (Shubert 1990). Caldiz and Beltrano (1989) and Bartoli et al. (1993) achieved 100% kills of T. recurvata and T. aeranthos with little damage to Argentinian hosts using the herbicides atrazine and simazine. Experiments such as those of Benzing and Renfrow (1980) demonstrated susceptibility to overloads of several metallic micronutrients (Table 5.3). Tolerances to corrosive gases varied according to the treatment. Tillandsia balbisiana, T. paucifolia, T. recurvata and T. utriculata survived 6-h exposures to O3 (0.15–0.45 ppm) and SO2 (0.3–1.2 ppm), applied alone and combined, in continuously stirred tank reactor exposure chambers (Benzing et al. 1992). Neither DH1 nor foliar conductance diminished during or immediately following these comparatively short nocturnal runs. Many of the test subjects flowered the following year in greenhouse culture, indicating no significant delayed injury. Remarkably low diffusive conductances even for a CAM plant probably reduced exposure of the mesophyll enough to avoid damage over such short runs. Fumigated at 3.8–4.0 ppm SO2 for several days, T. aeranthos exhibited considerable leaf necrosis (Arndt and Strehl 1989). Shacklette and Connor (1973) and Connor and Shacklette (1984) conducted a study that exceeded all the others using bromeliads in its geographic and chemical dimensions. Briefly, factor analysis identified three series of elements that accounted for almost three-quarters of the variation in log concentrations in the ash of 123 samples of Spanish moss collected across the southeastern United States. Aluminum, Ba, Ca, Co, Ga, Fe, Mn, Ti, Yb and Zr constituted a pedological association present in samples from all sites. Three more elements (Ca, Na and Sr) that occur abundantly in less widely distributed substrates were associated with, but somewhat distinct from, the other 10. Inconsistent proportions between the concentrations of members of the first series of 14 elements and total plant ash suggested enrichments by wind-deposited particles. Plant absorption contributed far less. Distributions of several of the elements in this first series among the samples indicated different terrestrial sources. Limestone dust from Bromeliads as air quality monitors 243 roadbed fill and quarrying operations probably produced the frequent, proportionally high values for Ca and Sr relative to Al. Collections with the highest Na:Al ratios usually came from Florida and the Gulf and Atlantic coast sites (Shacklette and Connor 1973) to the west and north, indicating strong marine influences. Presumably coatings on leaves near the ocean contained more sea salt and marine limestone dust, whereas materials derived from terrestrial soils predominated farther inland. Nine elements (B, Cr, Cd, Cu, Li, Ni, Pb, V and Zn) formed a second series made up principally of the technological metals. Generally low logarithmic correlations of these constituents with log Al in ash suggested that sources other than soil accounted for their occurrence in the samples. Lead concentrations peaked near heavily traveled highways, as noted in some of the other studies with T. usneoides (e.g., Martinez et al. 1971). Occurrences of Zn, Cu and Cd in the same vector fan of the Connor and Shacklette varimax model further indicated that these plant constituents originated from vehicles, most likely from lubricants, tires and abraded body parts, if not exhaust. The same explanation probably applied to Cr and Ni. Boron, Li and V emerged at the other end of the technological metals vector fan. Residual oil used for heating and power generation, in addition to some soils, represent potentially major sources of V in the southeastern United States (Schroeder 1970; Zoller et al. 1973). Glass manufacture employs considerable Ba and Li, and Cd, Cr, Ni, V and Zn indicate other industrial processes. Coal-burning volatilizes many metals. Connor and Shacklette concluded that, unlike the soil–element association, anthropogenic aerosols rather than absorbed natural particulates accounted for the extraordinary accumulations of technological metals. A geographic dimension further differentiated the series one and series two metals. Substratum-related elements exhibited broad regional (e.g., landward vs. coastal) trends in their proportional concentrations in Spanish moss, while the technological metals showed more localized occurrences reflecting nearby discrete sources. Sheline and Winchester (1976) also identified characteristic combinations of elements in samples of Spanish moss in northern Florida, which they considered consistent with plant uptake of aerosol particles. Magnesium, P and K formed a third varimax series in Connor and Shacklette’s study, distinguished from the others by pattern of occurrence and status as plant macronutrients. Log concentrations of all three elements varied independently of the other 23 surveyed and percent ash, reputedly because the bromeliad controlled their accumulation according to need. 244 Mineral nutrition Two inherent characteristics probably account for the remarkable affinities Type Five bromeliads exhibit for diverse substances: dependence on the atmosphere rather than soil for nutrients, and a relatively nondiscriminating organ (the foliar trichome) for absorption. Capacity to scavenge often scarce ions is essential for these plants. However, the same mechanisms apparently promote toxic accumulations of required (e.g., Cu) and other substances if supplies become too enriched. Consequently, oligotrophic members of Tillandsia, probably more than most vascular flora, offer exceptional opportunity to monitor air quality. However, as slowgrowing CAM plants, these bromeliads do not match the sensitivity of much other vegetation (e.g., certain crops) to chronic exposures to several common corrosive gases. 6 Reproduction and life history D. H. B E NZ ING, H. LUT HE R AN D B. BEN N ETT Reproduction is central to the evolutionary theme of this volume, but an uneven literature precludes a balanced treatment. Numerous reports deal with issues like pollination, while a host of other subjects (e.g., breeding systems) remain largely ignored. Reasons for the disparity range from botanical tradition to differences in the ease and costs of pursuing speci® c kinds of inquiry. Publications on ¯ ower, fruit and seed structure exceed all others on bromeliad reproduction to the extent that even a reasonable overview of this material requires a separate chapter (Chapter 3). Issues such as seed dispersal and gene ¯ ow and demography, along with the other phenomena that in¯ uence reproductive success and evolution, receive ® rst priority here. A single, fundamental bauplan and life under often stressful growing conditions help explain why most Bromeliaceae share several de® ning aspects of natural history. Life cycles usually proceed slowly and several years to decades pass before the typical bromeliad produces the ® rst of usually multiple seed crops (iteroparity) from branched, determinant shoots (Fig. 2.3). About half a dozen genera in Pitcairnioideae and Tillandsioideae also include one or more members that fruit just once (the monocarps) after some 15 to many additional years devoted solely to resource accumulation (Figs. 2.3B, 14.2C). Genets of the iteroparous types that survive long enough fragment into multiple, autonomous units, and especially successful genets of the most aggressive terrestrials dominate extensive habitat (e.g., Brokaw 1983; Murawski and Hamrick 1990). Numerous less routine aspects of reproduction and life history accommodate speci® c Bromeliaceae to speci® c kinds of ecospace and growing conditions. Fruit and seed morphology indicate dispersal by wind, perhaps ¯ owing water, several groups of vertebrates, ants and possibly some other invertebrates. Breeding systems and pollinators also distinguish 245 Cambridge Books Online © Cambridge University Press, 2009 246 Reproduction and life history Table 6.1. Representative Bromeliaceae and their putative pollinators Species Aechmea magdalenae Aechmea distichantha Alcantarea regina Ayensua uaipanensis Billbergia horrida Billbergia porteana Dyckia ferox Encholirium glaziovii Fascicularia bicolor Guzmania lingulata var. minor Hohenbergia blanchetii Nidularium procerum Pitcairnia corallina Pitcairnia loki-schmidtiae Pitcairnia brevicalycina Puya alpestris Puya ferruginea Tillandsia duratii Tillandsia argentea Tillandsia utriculata Vriesea carinata Vriesea in¯ ata Vriesea atra Werauhia gladioli¯ ora Pollinator Reference Hummingbirds Hummingbirds Bats Bats Hummingbirds and bees Hummingbirds Hummingbirds Bats Hummingbirds Hummingbirds Hummingbirds Hummingbirds Hummingbirds Bats Insects Hummingbirds Bats Moths Small moths Hummingbirds and small moths Hummingbirds Hummingbirds Bats Bats Murawski and Hamrick 1990 Bernardello et al. 1991 Vogel 1969 Varadarajan and Brown 1988 Ruschi 1949 Ruschi 1949 Bernardello et al. 1991 Sazima et al. 1989 Mez 1896 Stiles 1978 Ruschi 1949 Ruschi 1949 Varadarajan and Brown 1988 Vogel 1969 Varadarajan and Brown 1988 Johow 1810 Vogel 1969 Bernardello et al. 1991 Gardner 1986a Gardner 1986a Ruschi 1949 Ruschi 1949 Vogel 1969 Vogel 1969 populations, sometimes even conspeci® cs. Nectar-seeking birds service many Bromeliaceae, and moths and bats transport pollen for some of the others, as do additional kinds of insects and here and there a nonvolant mammal. At another extreme, the occasional population routinely self-pollinates, sometimes via cleistogamy. Interest in asexual reproduction has grown in recent years, and several preliminary reports indicate greater variety than expected for plants with such a uniform body plan. Pollination Diverse fauna, but predominantly birds, set the fruits of Bromeliaceae (Table 6.1). However, few accounts provide the additional data necessary to determine how pollinators affect the structure of plant populations, in¯ uence the quality of the resulting offspring, or affect the evolution of the Cambridge Books Online © Cambridge University Press, 2009 Pollination 247 ¯ ower. Gardner (1984, 1986a,b) conducted one of the most provocative inquiries, drawing on her extensive knowledge of Mexican Tillandsia. Other authorities choose various Pitcairnioideae, while Bromelioideae remain least studied of the three subfamilies. Gardner' s descriptions and analyses provide a starting point to consider the breeding mechanisms of Bromeliaceae, but ® rst we need some background on this remarkable genus and its subfamily (Chapters 12 and 13). Tillandsioideae Tillandsia (sensu Smith and Downs 1977) contains over 500 described species according to the latest count (Luther and Sieff 1996; Chapter 13). This tally will surely grow, although not the size of Tillandsia per se if the views of several authorities prevail. Smith and Downs (1977) listed just 410 species in seven subgenera and one of these segregates, Pseudocatopsis, has already been elevated to Racinaea by Smith and Spencer (1992). Additional components (e.g., subgenus Pseudalcantarea; Beaman and Judd 1996) will likely also prove untenable as currently conceived, in this case owing to convergence on a similar chiropterophilous ¯ oral syndrome (Fig. 3.3M). The paraphyletic status of Vriesea and the affinities of several other taxa (e.g., Catopsis and Glomeropitcairnia) that stand well removed from core Tillandsioideae further underscore the need to better resolve Smith and Downs' s organization of this subfamily (see Chapters 12 and 13). Vriesea (two subgenera, .225 species) closely parallel Tillandsia in architecture, geography and ecology, with formal assignment to one or the other taxon based wholly on the presence or absence of petal scales (Fig. 3.1B). Even Smith and Downs (1977) occasionally challenge the utility of their key character (e.g., recognition that T. pabstiana5V. drepanocarpa despite the absence of scales). But whatever the taxonomic fate of Tillandsia vs. Vriesea, populations currently assigned to these two genera and perhaps several others collectively constitute one of the largest assemblages of closely related bromeliads. Moreover, parts of this clade exhibit signs of continuing, active radiation. Plants representing many different species have been hybridized in culture, and additional combinations are spontaneous (Table 6.2). Frequent sympatry and substantial ecological equivalence, particularly among the epiphytes, further suggest evolutionary youth. Guzmania (.150 species) populate everwet forests of the Colombian Chóco with about 40 described species, many sympatric and scarcely distinguishable by vegetative characteristics or substrates. Most important for our purposes, Cambridge Books Online © Cambridge University Press, 2009 248 Reproduction and life history Table 6.2. Some hybrids involving Tillandsia subgenus Tillandsia Tillandsia brachycaulos3T. bulbosa T. brachycaulos3T. balbisiana T. brachycaulos3T. capitata T. brachycaulos3T. caput-medusae T. brachycaulos3T. ionantha T. brachycaulos3T. foliosa T. brachycaulos3T. mirabilis T. fasciculata3T. foliosa T. fasciculata3T. lieboldiana T. ¯ abellata3Vriesea incurvata T. ¯ abellata3T. tricolor T. ionantha3T. schiedeana T. punctulata3T. krukof® ana T. jalisco-monticola3T. xerographica Source: After Gardner (1984). better-known Tillandsia, especially subgenus Tillandsia, provide exceptional opportunity to consider the in¯ uences of pollinators on cladogenesis, the characteristics of ¯ owers, and the integrity of closely related populations. Subgenus Tillandsia, a primarily Mesoamerican assemblage of .150, mostly epiphytic and often markedly drought-tolerant species, makes up the second largest (after Allardtia) of the formally recognized segregates comprising genus Tillandsia. Flowers with slender, tubular, regular to somewhat zygomorphic corollas, at most ¯ aring modestly, characterize the entire subgenus (Fig. 6.1A). Nevertheless, Gardner (1986b) was able to employ shared ¯ oral characteristics to differentiate 85 of its member species, plus a few similar taxa from Allardtia, into ® ve groups preparatory to more extensive study. At issue were reproductive biology and systematics, and especially what appears to be an exceptionally high incidence of multivalent pollination syndromes among members of Group One. Prominent ¯ oral bracts that enclose ¯ ower buds and young fruits, the just mentioned narrow petals rolled into a tube, well-insulated, deeply placed nectar, and exerted sexual organs suggest fundamental ornithophily for Group One and perhaps the entire subgenus. Different arrangements prevail elsewhere, especially in much smaller Group Two, which indicate other primary pollinators (Fig. 6.1A). Overall, as many architectures make up what appear to be basic ¯ oral themes for the ® ve groups comprising most of subgenus Tillandsia. Variations on the ¯ oral pattern expressed by Books Online © Cambridge University Press, 2009 Pollination 249 Figure 6.1. Aspects of ¯ owers and seeds of Bromeliaceae. (A) Flower structure characteristic of Gardner' s (1986b) ® ve groups of species recognized mostly within Tillandsia subgenus Tillandsia. (B) Flower of Tillandsia punctata demonstrating light-colored petal tips. (C) Plication of stamen ® lament illustrated from left to right by Tillandsia gardneri, T. stricta and T. aequatorialis. (D) Seed morphology among species of Brocchinia. Cambridge Books Online © Cambridge University Press, 2009 250 Reproduction and life history all members of Group One demonstrate continuing capacity to evolve as changing environments reorder the advantages of relying on one vs. other kinds of fauna, or so it seems. Certain Tillandsia also demonstrate the in¯ uence of rooting medium and plant size on breeding system and ¯ ower morphology. The most diminutive forms (e.g., T. capillaris, T. recurvata), those species that tend to colonize twigs consistent with their small stature, also often display much reduced, autogamous ¯ owers (Fig. 3.3C). These species consistently set self-seeds, sometimes by cleistogamy (T. capillaris; Gilmartin and Brown 1985), perhaps because they lack capacity to entice fauna to fertilize enough of what are already reduced numbers of ovules in miniaturized capsules. Anthers in several cases form a hood above the stigma that at once prevents outcrossing and assures fruit set (Till 1992a). Additional members of this highly neotenic subgenus produce showy ¯ owers that emit powerful perfumes (e.g., T. crocata, T. myosura). Sel® ng also occurs in subgenus Tillandsia, and most conspicuously where monocarpy rather than small size or ephemeral substrates mandates that most ovules become seeds (e.g., T. utriculata in Group Two; Fig. 6.1A). However, variations on the basic ¯ oral plan of Tillandsia subgenus Tillandsia Group One offer superior opportunity to learn about the evolution of reproductive biology because several of Gardner' s Mexican subjects illustrate recent or on-going change in this system (Gardner 1982, 1986a). Sometimes geographic distributions and ecology indicate what may be related adaptation involving aspects of subjects other than their ¯ owers. Circumstantial evidence suggests that pollinators currently isolate many co-occurring populations, and that they also fostered much speciation within oversized Group One. Tillandsia andrieuxii (lavender corolla, diurnal anthesis) and T. erubescens (chartreuse, nocturnal), for example, exhibit such close overall similarity that Mez (1934± 35) considered them varieties of the second taxon. Tillandsia parryi illustrates a similar pattern with accompanying changes in ecology that indicate additional change, perhaps even cladogenesis. Epiphytes quite similar to T. parryi collected near Monterey, Mexico and described as T. sueae (Ehlers 1991), and similar plants growing south of Xilitla in San Luis Potosi State, ¯ ower just once (monocarp) and display lavender corollas that open in midmorning (Gardner 1982). However, specimens east of the city of San Luis Potosi occur as iteroparous lithophytes equipped with chartreuse petals that separate at dusk. Winter vs. summer ¯ owering further suggests dependence on different kinds of pollen carriers. Certain members of subgenus Tillandsia attract the same kinds of polli- Cambridge Books Online © Cambridge University Press, 2009 Pollination 251 nators using different variations on the same basic ¯ oral syndrome. Populations comprising Group One serviced by nocturnal visitors lack fragrances, whereas species elsewhere in the same taxon (Group Three), especially night-¯ owering T. heterophylla, produce powerful perfumes. Appropriate timing and color clearly suffice for fruit set in Group One, perhaps rendering osmophores functionally redundant and an unnecessary investment. Lavender petals among certain members of Group One become more re¯ ective, hence visible in dim light, simply by accumulating less anthocyanin (e.g., T. seleriana). Tillandsia streptophylla uses the same basically ornithophilous syndrome (large ¯ oral bracts, exerted sexual appendages, tubular corolla, extensive nectar production) to signal crepuscular and night ¯ iers with densely lepidote, light pink ¯ oral bracts. Deep purple-¯ owered T. punctulata may do the same even more subtly by displaying a lightly pigmented, exerted style with matching petal tips (Fig. 6.1B). Little impetus may exist to augment with odors or other major investments a possibly minor backup arrangement needed only to commit the few gynoecia overlooked by the diurnal pollinators (birds for T. punctata) that these species target more expensively and conspicuously (large, bright red and green ¯ oral bracts). A widely shared feature of ¯ oral development may predispose many Tillandsia subgenus Tillandsia species to high fruit set and mixed-paternity progeny. Diurnal ¯ owers typically last about 48 h, extending access to fauna active after sundown. Similarly, ¯ owers that open during the night tend to remain turgid into the following day. Related embellishments to attract night or day ¯ iers vary with the example. Tillandsia roland-gosselinii represents one extreme by relying on a brilliant, parrot-like combination of a large, bright red scape and slick green ¯ oral bracts to promote seed set. For good measure, and normally as a prelude to lavender ¯ owers in similarly colored relatives, the entire shoot becomes scarlet. Finally, and incongruously, emerging petals add a relatively faint, pale chartreuse signal just before sunrise. Gardner' s use of ¯ oral characters to segregate 85 species into ® ve groups revealed ecological correlates, some of which may constrain ¯ oral evolution. Most members of Group One occupy arid habitats, i.e., belong to ecological Type Five, or, if equipped with thinner leaves that impound moisture (Type Four), constitute relatively xeromorphic members of that assemblage. Soft, green, essentially glabrous foliage more consistent with conditions in everwet forests prevails through Groups Two and Three. However, most of these plants exhibit nocturnal or diurnal anthesis respectively. No comparable information exists for Groups Four or Five (only one Books Online © Cambridge University Press, 2009 252 Reproduction and life history or two species in each), nor are enough data available to speak with authority about timing for primarily ornithophilous and diurnal Group One. Factors other than pollinators in¯ uence the evolution of the ¯ ower, and possibly did so in Tillandsia. Aridity and extended anthesis in addition to frequent dependence on birds may explain some of the distinguishing ¯ oral characteristics shared by members of Group One. All of these plants possess distally broadened and ¯ attened stamen ® laments that, combined with an apically narrowed corolla, may deter all but legitimate pollinators ± those with long mouth parts like hummingbirds (Fig. 6.1). Alternatively, aridity, speci® cally its capacity to concentrate nectar enough to impede extraction, rather than gate-keeping explains the same morphology. Protogyny prevails in Group One as it does through most of the rest of the subgenus. Anthers fail to reach the exerted and precocious stigma except in some autogamous populations where mature organs of both types extend the same distance beyond the corolla (Fig. 6.1A). Occasionally, the two-tiered con® guration lasts only a few hours as if to encourage allogamy after which elongating ® laments brush self-pollen against any stigma that remains receptive. Fewer than every ¯ ower favors autogamy by this mechanism in still other species (e.g., T. achyrostachys, T. concolor, T. capitata, T. matudae), perhaps to relieve plants unable to mature every potential capsule. Benzing and Davidson (1979) determined that specimens of T. paucifolia bearing the largest numbers of fruits in Florida invested exceptionally large proportions of their N and P there, enough to slow the growth of the next ramet compared with subjects with some barren ¯ owers. Mixed ¯ oral syndromes may help account for the relatively frequent spontaneous hybridizations among some members in Group One (Table 6.2). Tillandsia punctulata, with its white-tipped, purple corolla, noctural anthesis and bird-attracting bracts, often crosses with diurnal, green-¯ owered (prominent green ® laments) T. krukoffiana in the highlands north of Puebla, Mexico. However, conclusions about the importance of ¯ owers and pollinators vs. agencies more remote to this outcome are best drawn within a broader context. Intensive agriculture in Mexico and Central America beginning about 4000± 5000  may have encouraged gene exchange among Tillandsia subgenus Tillandsia populations through the activities of pollinators that foraged more selectively in pre-agrarian habitats (Gardner 1984). Uniform chromosome numbers indicate a minor role for polyploidy during the history of Tillandsia beyond subgenus Diaphoranthema (Chapter 9). Genetic analyses (see below) of several populations of Cambridge Books Online © Cambridge University Press, 2009 Pollination 253 Mexican Tillandsia ionantha (Group One) and T. recurvata (subgenus Diaphoranthema), except for a single triallelic locus in the latter (Soltis et al. 1987), and Kress et al.' s (1990) less comprehensive analysis (three enzymes) of Florida T. recurvata, T. usneoides and T. recurvata, support this hypothesis. Distributions in many instances (e.g., T. fasciculata, T. utriculata) across Mesoamerica into northern South America, and included ranges of numerous close, more insular relatives (see below), accord with comparative youth and recent colonizations of separated habitats. Capacity to readily adopt different pollinators to service often self-compatible ¯ owers and ¯ oral morphology conducive to spontaneous autogamy (e.g., T. recurvata) probably assisted the exceptional radiation demonstrated by the size of Tillandsia subgenus Tillandsia. Outlying populations of several species (e.g., T. balbisiana, T. ¯ exuosa in Florida) regularly set self-fruit, perhaps as founders did to establish populations. Breeding systems even shift across short distances. Southernmost Florida Tillandsia balbisiana, for example, produces bright red ¯ oral bracts, whereas members of isolated outlying colonies farther north at about mid-peninsula develop little color, yet mostly ripen abundant seeds. Floral syndromes that unambiguously target insects also characterize Tillandsia subgenus Tillandsia. Tillandsia utriculata (Group Two) initiates anthesis after dark with ¯ owers bearing large, creamy petals further distinguished by an apical twist (Fig. 6.1A). The lateral aperture exposes the style and six stamens with circular, uniformly slender ® laments. Anthers attach in versatile rather than basi® xed fashion, supposedly to promote sphingophily (moth pollination; Vogel 1969). Self-compatibility probably describes all T. utriculata, and sometimes this bromeliad requires no assistance to reproduce. Certain populations in northeastern Mexico mature relatively low percentages of gynoecia (average 33% at seven locations; Gardner 1982, 1984), while plants in Florida with similar ¯ owers, but paler bracts, set nearly every fruit. Breeding systems in these outlying populations may re¯ ect depauperate faunas, or, again, bottlenecks effected by autogamous founders. Like those of Group One, members of Group Three possess large primary bracts, perhaps owing to a bird-serviced ancestry. Pigmentation usually follows suit (e.g., T. imperialis, T. ponderosa, T. deppeana), but not ¯ ower structure, which better matches another group of visitors. Characteristically basi® xed anthers exceed the lengths of those presented by members of the other four groups, and more pollen is produced. Diurnal anthesis and corolla shape also signal melittophily (bee pollination). Firm lavender petals curve gently or roll back to provide a credible landing site for medium-sized visitors (Fig. 6.1A). Cambridge Books Online © Cambridge University Press, 2009 254 Reproduction and life history Petals of zygomorphic-¯ owered T. multicaulis, rather than curving downward to expose the anthers, twist apically to provide access to nectar and pollen along one side of the corolla tube much like many Pitcairnia species (Fig. 3.4K). One petal rolls down the side of the T. deppeana ¯ ower to again furnish a landing site, presumably for Hymenoptera. Tillandsia heterophylla (Group Three) alone in subgenus Tillandsia stands out for its large, leafy-green, glaucous ¯ oral bracts that presumably help guide moths to the pale, spreading corolla. A spicy, sweet fragrance complements this sphingophilous arrangement. Gardner also attributed phalaenophily (pollination by moths) to Tillandsia tortilis and T. lepidosepala, in part because small, densely lepidote (light-scattering) shoots characterize both species. Imbricate ¯ oral bracts project a dull, rose-pink hue to highlight the protruding, re¯ exed, moss-green petals surrounding the uniquely included stigma and conspicuous yellow anthers (Fig. 6.1A). Abundant pollen produced precociously indicates protandry, a second novelty for subgenus Tillandsia. A long ¯ exible scape (T. tortilis), or characteristic orientation on rocky substrates for shorter-stemmed T. lepidosepala, positions ¯ owers downward. Floral variety exceeding that present in subgenus Tillandsia occurs elsewhere in Tillandsia and the rest of Tillandsioideae. Unequivocal chiropterophily in at least two versions and entomophily and ornithophily, much as previously described, appear repeatedly, as do additional mixed and more exclusive syndromes for insects and birds (Fig. 3.3). Powerful fragrances, included rather than exerted sexual appendages, widely ¯ ared white, yellow and lavender corollas, small, dull bracts, and continuously green shoots characterize most of the allogamous Anoplophytum, Diaphoranthema and Phytarrhiza ± species also known as the fragrant tillandsias (Fig. 3.3A,F,I). Catopsis (.20 species) seems to lack capacity to produce anthocyanins, and its usually modest-sized, white to yellow ¯ owers (Fig. 3.3H) often emit pleasant fragrances during the day (e.g., C. paniculata) or night (e.g., C. nutans). Many Guzmania and Vriesea species ® t the ornithophilous syndrome, as does Mezobromelia (four species). Van Sluys and Stotz (1995) provided one of the most comprehensive accounts of ornithophily involving Tillandsioideae by observing Vriesea neoglutinosa in an open habitat within the Reserva Forestal de Linhares of Espirito Santo State, Brazil. Records were kept for large and smaller clumps of plants over a ® ve-day interval during the approximately onemonth period that local plants ¯ owered. Tubular, odorless ¯ owers subtended by red bracts opened before 06.00 hours and secreted nectar most copiously in the morning and again later in the day before the corollas with- Cambridge Books Online © Cambridge University Press, 2009 Pollination 255 ered. Four species of birds exploited this resource, with greater attention accorded by territorial Amazilia ® mbriata and Polytmus guainumbi than trap-lining Chlorostilbon aureoventris and Phaethornis idaliae. Visitation peaked during the morning, but continued through the afternoon. Patches offering the fewest ¯ owers received the fewest visits (range 1± 7 per day). Small compared with large patches also experienced signi® cantly fewer visits per in¯ orescence (1.33 vs. 1.79). Capsules ripened by plants in small patches contained fewer seeds than those obtained from the larger clumps (152.6 vs. 180.5). Similar values obtained by Snow and Snow (1986) for epiphytic V. incurva and V. jonghei in Atlantic Forest led them to conclude that pollen supply limited local seed production. Utley (1983) and Vogel (1969) studied Central American thecophylloid Vriesea species (part of section Xiphion), many notable for bat-attracting, unusually large, pale, wide-mouthed ¯ owers featuring hood-like arrangements of stamens with oversized anthers (see Fig. 3.5E for a Brazilian Xiphion). Heavy nocturnal odors help advertise for pollinators, while the large green bracts primarily protect buds and developing fruits. According to Utley, derived forms (e.g., V. vietoris, V. leucophylla, V. hainesiorum) abandoned bats, adopting instead brightly colored in¯ orescences and diurnal anthesis to utilize birds. Tubular corollas and symmetrical androecia further differentiate ornithophilous from chiropterophilous forms. Grant (1995a,b) erected genus Werauhia to recognize the close relationship and distinctness of these unusually large-bodied Tillandsioideae. Vriesea of the Organ mountains of southeastern Brazil constitute another exceptionally broad radiation within Tillandsioideae far south of the ranges of most Tillandsia subgenus Tillandsia. While ornithophily prevails in this second group as well (e.g., V. carinata, V. erythrodactylon with red/orange bracts and green to yellow petals; Table 6.1), other species lure unrecorded fauna with fragrant day or night ¯ owers often subtended by green, deep carmine or dry, brown ¯ oral bracts. Mixed systems characterize several of the more ornamental species as in Mesoamerica. Vriesea philippo-coburgii bears reddish bracts and yellow day ¯ owers with pleasant aromas, suggesting versatile syndromes designed for bees and birds as in some Tillandsia (e.g., T. imperialis). Night-¯ owering taxa lack bright pigments, producing instead pale to dark corollas associated with odd scents never reported for Tillandsia. Flowers of Vriesea longiscapa open at dusk and emit a yeasty grease aroma; those of nocturnal V. regnellii display even larger, 3± 5 cm, ¯ ared, pale corollas sprinkled with wine-red dots accompanied by another disagreeable odor. Nocturnal Vriesea unilateralis seems oddly disadvantaged if it has to Cambridge Books Online © Cambridge University Press, 2009 256 Reproduction and life history compete for the same fauna given its lack of re¯ ective bracts, a ¯ ared corolla or a strong scent. Sazima et al. (1995) con® rmed chiropterophily for six Vriesea species native to Brazil' s Atlantic Forest. Flower color ranges from cream (V. gigantea), through yellow (e.g., V. sazimae), to brownish red (V. bituminosa). Stiff, tubular corollas always ¯ are more than those of their bird-pollinated relatives, open at dusk and begin to collapse by midmorning (Fig. 3.5E). Anthers bend to the lower side except for V. gigantea where the display remains radial. Flowers distichously inserted on spikes or branched in¯ orescences with divergent or secund orientations emit nectar tainted with garlic-like odors. Subtending bracts exhibit shades of green, with or without dark spots, to deep carmine. All three of the species with opposite-¯ owered spikes secrete abundant mucilage, perhaps to deter nectar thieves. Two species of long-tongued, small glossophagine bats visited one to all six of these bromeliads. Hummingbirds sometimes harvested residual nectar from withered ¯ owers after dawn. Additional vrieseas representing section Xiphion and recently resurrected Alcantarea with similar ¯ oral syndromes indicate even wider use of bats through this complex of primarily rock and bark-dwelling Tillandsioideae (Fig. 3.3J). Some larger-¯ owered Alcantarea attract larger bats, including the stenodermatine frugivore Artibeus lituratus. Andean Guzmania rival Mesoamerican Tillandsia and Brazilian Vriesea as subjects to investigate interesting aspects of bromeliad pollination. Ecuadorian Guzmania alcantareoides parallels certain other bat-pollinated Tillandsioideae, including Alcantarea (Fig. 3.3J) and Tillandsia subgenus Pseudalcantarea (e.g., T. viridi¯ ora; Fig. 3.3M; Beaman and Judd 1996). Large white ¯ owers that open at night further advertise by smelling like slightly spoiled cabbage. Stamens with outsized anthers, that along with the petals droop limply by morning, distinguish this species from homoplasious relatives. Bat-attracting Guzmania present either smaller, widely ¯ ared, white, cream or pale green corollas with a conventionally arranged androecium (e.g., G. coriostachys, G. fosteriana) or, like G. mucronata, feature a larger, campanulate, green corolla enclosing anthers arrayed somewhat like those of thecophylloid Vriesea (Luther 1993). Other species clearly attract other kinds of pollinators, or they defy assignments to any of the recognized categories. North Andean Guzmania wittmackii, a day-¯ owering, close relative of G. alcantareoides, produces a brilliantly colored in¯ orescence to catch the attention of birds, demonstrating once again the plasticity of the ¯ oral syndrome in Tillandsioideae, as does wide-ranging Guzmania monostachia (autogamous in Florida, mostly Cambridge Books Online © Cambridge University Press, 2009 Pollination 257 allogamous and more brightly colored beyond). Additionally, some Guzmania species exhibit combinations of ¯ oral characteristics about as incongruous as those presented by certain Tillandsia and Vriesea. For example, white ¯ owers that open around midnight and close soon after dawn accompany bright orange-rose ¯ oral bracts in Ecuadorian Guzmania kentii. Summarizing brie¯ y, Tillandsioideae, especially as demonstrated by the Tillandsia/Vriesea complex and Guzmania, repeatedly co-opted widely available pollen carriers that require speci® c plant form and function to manipulate to set seeds. And where monocarpy, small plant size or ephemeral substrates mandate substantial fecundity, lineages sometimes abandoned pollinators entirely. However, the evolutionary pathways that link the ¯ oral syndromes remain largely unexplored. Reconstructed phylogenies would increase insights on historical events and underlying determinants, for example the extent to which past conditions of ¯ owers and in¯ orescences limited options for pollen dispersal later. At this point, nonrandom distributions of ¯ oral syndromes among related lineages suggest that the ¯ oral biology of ancestors in¯ uenced outcomes in descendants, i.e., operated as phylogenetic constraints. Vriesea psittacina, the type species for its genus, attracts birds, as do many of what appear to be its closest relatives (e.g., V. carinata), whereas a second group of conspeci® cs that includes many Vriesea section Xiphion species (now Werauhia) depend primarily on bats. Many members of a third, natural assemblage of about 15 saxicolous species ± former Vriesea species (genus Alcantarea) also native to southeastern Brazil ± mostly attract day-¯ ying insects with large, perfumed, yellow-petaled ¯ owers (Fig. 3.3J). To what degree do these patterns re¯ ect inherent barriers to arrangements that would promote fruit set by other kinds of pollinators? Pitcairnioideae Pitcairnioideae rival Tillandsioideae for ¯ oral variety and kinds of pollinators attracted. Also, tendencies to deviate from basic designs characterize some clades more than others. Reproductive structure and pigmentation suggest near to complete dependence on insects for Brocchinia, Cottendor® a, Deuterocohnia, Dyckia, Encholirium, Fosterella, Lindmania and Hechtia. Conversely, in¯ orescence shape (massive cylindrical to loose paniculate), diverse ¯ ower colors, radial to zygomorphic corollas, hypogenous to epigenous architecture and observations in situ indicate that more varied fauna service Pitcairnia and Puya. Pronouncements about Navia Cambridge Books Online © Cambridge University Press, 2009 258 Reproduction and life history (probably many syndromes, but no entries in Table 6.1) and some of the other Guayanan endemics would be premature. At this point, Pitcairnia more than any of the other pitcairnioid genera compares with Guzmania, Tillandsia and Vriesea for ¯ oral variety. Members of Pitcairnia (sensu lato), which is the largest genus (.250 species) within its subfamily, mostly produce elongate tubular ¯ owers with actinomorphic to moderately zygomorphic corollas, which in the second case open on one side below the intertwisted petal tips to expose anthers and stigma (Fig. 3.4F,H,K,L,M). Further embellishments heighten appeal to birds (red corolla, copious nectar, diurnal presentation; e.g., P. corallina, P. nubigena), bees (white, yellow to green petals, lesser amounts of nectar, diurnal opening; e.g., P. brevicalycina, P. albi¯ os), moths (white corolla, strong odor, abundant nectar, crepuscular/nocturnal anthesis; e.g., P. ¯ ammea var. pallida, P. unilateralis) and bats (pale corolla, unpleasant night odor, abundant, exposed nectar; e.g., P. loki-schmidtiae, P. palmoides). Pitcairnia ® mbriato-bracteata surely ranks among the most exceptional species relative to ¯ oral biology. Reddish ¯ owers born on an equally garish, sinuous in¯ orescence extend from what by anthesis has become a glutinous mantle of autodigested, brownish, overlapping ¯ oral bracts. More modest deliquescence characterizes additional species such as P. arcuata (Fig. 3.4M). Pitcairnia rubro-nigri¯ ora holds the record for striking ¯ ower color with an almost black-purple corolla contrasting with the bright red calyx. A comparably red in¯ orescence produced by Pitcairnia corollina snakes along the ground, exposing foraging birds to terrestrial predators, unless of course typically precipitous substrates, often cliff sides, reduce this threat. Bat-serviced Pitcairnioideae, like comparable Tillandsioideae, manipulate pollinators with scents, abundant nectar, and ¯ owers and in¯ orescences organized along two patterns. Chiropterphilous Pitcairnia and Puya display large, well-separated and exposed ¯ owers that open sequentially to present extended, tubular perianths enclosing often clustered stamens (Fig. 3.4H). Pale to greenish petals that form a more or less wide-mouthed gullet and musty, nocturnal fragrances further distinguish these taxa from ornithophilous relatives. Conversely, Encholirium glaziovii, the bromeliad with the most thoroughly documented dependence on bats (Sazima et al. 1989), like much of the rest of its genus, produces a cylindrical brush-type, many¯ owered (.200) spike 1.5± 1.8 m tall (Figs. 3.4G, 6.2A). For about 10 days, relatively small, protogynous ¯ owers, each with a wide mouth, persistent (several days) perigon and stiff, spreading stamens and style, bloom in a wide acropetal belt (Fig. 3.4G). Enough dilute nectar (4.6% solids) issues Cambridge Books Online © Cambridge University Press, 2009 Pollination 259 Figure 6.2. Habits and seeds of certain Bromeliaceae. (A) Unidenti® ed Encholirium in Bahia State, Brazil. (B) Hechtia schottii in Yucatán State, Mexico. (C) Fruiting monocarpic Tillandsia utriculata in south Florida. (D) Seeds of Tillandsia paucifolia glued in groups of four to the bark of Taxodium distichum in south Florida in the manner employed to test germination. Cambridge Books Online © Cambridge University Press, 2009 260 Reproduction and life history from the many simultaneously active gynoecia to ¯ ow down the grooved in¯ orescence axis. While chiropterophilous Tillandsioideae typically inhabit dense, humid montane forests as epiphytes and attract diverse phyllostomids, terrestrial Encholirium glaziovii more closely parallels similarly pollinated Mexican Agavaceae. Populations occupy open, semiarid, rocky scrub communities characteristic of the `campos rupestres' of interior southeastern Brazil (Fig. 1.4C). Its single bat visitor at the study site, trap-lining Lonchophylla bokermanni, hovered to collect nectar beginning about 30± 60 min after dusk. Visits at 5± 40-min intervals consisted of several wide loops around a spike interrupted by nectar collections that lasted less than a second, yet long enough to brush pollen on and off the animal' s snout. Several other ¯ ower-dependent bats in the same area ignored E. glaziovii, apparently preferring less exposed food plants in nearby gallery forest and cerrado. Sphingids and some other moths sporadically visited the same E. glaziovii ¯ owers. Exclusively Guayanan Navia (,100 species) contains many narrowly distributed lithophytes. The small ¯ owers born by many taxa are probably inadequate to satisfy the caloric needs of vertebrates. During anthesis brightly pigmented foliage highlights sessile, somewhat larger ¯ owers born in the typically capitate sessile in¯ orescences of one group of relatively robust species. Just the proximal portions of the younger leaves color up to deep orange/red (e.g., N. arida), or they bleach to brilliant white (e.g., N. jauaensis). Foliar pigmentation changes little in other instances, that role falling to the ¯ oral bracts (e.g., N. splendens). Densely congested ¯ owers with somewhat oversized stigmas and anthers presented above the shoot suggest anemophily or dependence on small insects in another part of the genus (Fig. 3.4A± C). Even fewer records address Connellia (® ve species) from the same poorly collected, remote upland habitats. Corollas are showy, rose-pink in C. smithiana, and location among the leaf-like bracts suggests pollination by Hymenoptera. Most of the remaining larger genera (Cottendor® a, Dyckia, Hechtia, Fosterella), and several lesser ones (e.g., Brewcaria, Steyerbromelia), exhibit principally entomophilous, relatively small, open ¯ owers in a variety of mostly pale pastels. Fosterella spectabilis alone in its otherwise white to cream-¯ owered genus probably attracts birds with a coral red, predictably more elongated corolla (Luther 1997; Fig. 3.4D). Members of Puya, the second largest of the nearly 20 pitcairnioid genera, also utilize diverse pollinators through mostly Andean ranges. Ortiz-Crespo (1973) observed specimens in the botanical gardens at the Cambridge Books Online © Cambridge University Press, 2009 Pollination 261 University of California at Berkeley and in situ in Ecuador. Plants at both locations produced concentrated nectar from relatively large, often bluish to green, showy ¯ owers rendered additionally conspicuous by bright orange anthers. Tubular corollas, unaccompanied by scent, the production of sticky pollen and a simple stigma further implicated birds as the primary vectors. Colibri coruscans, and a few, much larger Patagona gigas, maintained near continuous presence, while a colony of Puya aequatorialis ¯ owered for about six weeks at a site approximately 20 km north of Quito, Ecuador. Small, agile Colibri coruscans, the most aggressive of the lot, often denied two or three additional hummingbirds access to nectar. In¯ orescences had been under its surveillance for several days before the ® rst ¯ owers opened. Other Puya species reportedly serviced by hummingbirds include P. chilensis (yellow ¯ owers), P. berteroniana and P. venusta (most consistently by Patagona gigas). Flocks of hungry Austral blackbirds (Curaeus curaeus) also visited P. chilensis. In¯ orescence and ¯ ower structure either encouraged or denied use by additional ® nches and ¯ ycatchers according to ¯ ight and feeding behaviors (Johow 1910). Species that lack sterile extensions on the lateral axes of what are usually dense panicles also possess ¯ owers with deep corolla tubes (subgenus Puyopsis; e.g., P. venusta) to discourage all but the hover ¯ iers. Those taxa (subgenus Puya; e.g., P. chilensis) able to accommodate perching birds also offer shorter ¯ owers accessible to ¯ ying and stationary feeders alike. The diurnal moth Castnia eudesmia feeds both as a larva and as an adult on some of the same Chilean Puya species. Its reputed ability to drive similarly disposed insects and even birds from favored food sources warrants further study. Abundant nectar disposes the larger-bodied Puya species for ornithophily; requirements to set fruit, especially the monocarps, and often hyperdispersed populations may permit no alternatives. Few other trap-liners range into paramo and puna formations where Puya often dominate otherwise sparse ¯ oras (Fig. 14.2C). Certain high-elevation species strengthen the case for obligate dependence on powerful ¯ iers to the extent that their self-incompatibility characterizes the other alpine Puya. Puya mirabilis (self-compatible) and P. ferruginea demonstrate the feasibility of chiropterophily at lower elevations. Insufficient capacity to support ¯ ower visitors with high caloric demands may help explain the absence of many additional Pitcairnioideae and more than a modest contingent of Tillandsioideae at these same cold, barren sites. The importance of plant size to frost-tolerance in tropical alpine habitats may further limit the success of Bromeliaceae above the tree line (Chapters 4 and 7). Cambridge Books Online © Cambridge University Press, 2009 262 Reproduction and life history Varadarajan and Brown (1988) considered stigma morphology diagnostic for the primary pollinators of certain Pitcairnioideae. For example, ornithophilous Pitcairnia usually possess compact, conduplicately folded stigmas (Figs. 3.1C, 12.1) bearing spathulate lobes covered with densely packed papillae. Organs characterized by less condensed parts, with or without papillae, accompany large, white, actinomorphic corollas, strong fragrances, and nocturnal presentation, perhaps to round out an attractive combination of traits for bats (e.g., Ayensua uaipanensis, Puya aristeguietae). Those taxa equipped with stigmas bearing lanceolate, still more loosely folded lobes free of papillae supposedly also produce small, diurnal, white, yellow to green ¯ owers attractive to bees (e.g., Brocchinia steyermarkii, Lindmania guianensis, Deuterocohnia longipetala, Pitcairnia brevicalycina). Varadarajan and Brown further proposed that the degree of lobe compaction and the disposition of the papillae reveal ¯ oral syndromes more reliably than does gross stigma shape. Petal scales and septal nectaries also help identify targeted fauna by providing information on the quantities of nectar produced and its mode of presentation (Fig. 3.1A,B). Bernardello et al. (1991) examined a variety of Argentinian Pitcairnioideae to identify the plant characteristics responsible for the attentions of certain kinds of ¯ ower visitors. Nectary structure and the composition of secretions were emphasized. Their conclusions, several con® rmed in nature, sometimes contradicted those of Varadarajan and Brown, as indicated below. Bromelioideae Approximately the same array of fauna pollinate Bromelioideae as pollinate Pitcairnioideae and Tillandsioideae. Likewise, ornithophily probably predominates, according to reports and ¯ oral syndromes displayed among the larger genera (e.g., Aechmea, Billbergia, Neoregelia, Quesnelia; Table 6.1). Flexibility, like that documented by visits by hummingbirds and butter¯ ies (Eurema diara and Phoebus spp.) to Costa Rican Bromelia pinguin and B. karatas (Hallwachs 1983), may be especially common in this subfamily. Unusual morphology complicates some interpretations, for example the nearly submerged, capitulate in¯ orescence of most Neoregelia species (Fig. 3.2A). Flowers equipped with lavender to pink or white corollas barely extend above the surface as if to deter predators seeking buds and developing fruits (Fig. 3.5B). Nevertheless, insects and hummingbirds readily access nectar located deep within the tubular corollas. Members of the Neoregelia/Nidularium complex and the other nidulate Cambridge Books Online © Cambridge University Press, 2009 Pollination 263 Bromelioideae usually exhibit one of two ornamentations, which in one and perhaps both instances attract fauna (Fig. 2.13F; Leme 1997, 1998a,b). Immature foliage and the lower parts of older leaves often accumulate bright red through orange to purple pigments (occasionally albinistic as in certain Navia species, e.g., Navia ocellata (formerly N. lactea)) following ¯ oral induction, and sometimes these displays persist long enough to in¯ uence seed dispersal. Rather than conspicuously pigmented centers, other Neoregelia and a few Hohenbergia and Nidularium species and Wittrockia superba feature red to purple leaf tips, although not necessarily for the same purposes (Fig. 2.13F; Chapter 11). Parallels exist in Tillandsioideae (e.g., Vriesea platynema, V. minuta). Guzmania sanguinea much more closely resembles the ® rst group in the way its ¯ owers extend just above the surface of the phytotelmata maintained by utriculate foliage. Sweet fragrances emanating from small ¯ owers on often dull in¯ orescences signal entomophily in exceptional Aechmea (e.g., Aechmea purpureorosea, A. lingulata), the largest (.150 species) and among the most arti® cial of the bromelioid genera. Corollas of Aechmea fasciata change color from powder blue early in the day to deep rose-red by late afternoon whether pollinated or not (Fig. 3.2G). Most Billbergia produce pendant, rapidly elongating in¯ orescences bearing large, ephemeral pastel primary bracts and hummingbird-serviced ¯ owers (Fig. 3.2F); exceptional entomophilous types (e.g., fragrant B. horrida) stand upright and present ¯ owers subtended by vestigial, equally pale bracts. Billbergia robert-readii departs farthest from the norm with its odd-smelling, night-blooming, upturned grayish chiropterophilous ¯ owers born on a lax spike. Reproductive organs generally suggest that fewer Bromelioideae than Pitcairnioideae or Tillandsioideae depend on bats to produce seeds. Ant-inhabited Aechmea bracteata ripens full crops of purple-black berries on densely ¯ owered panicles in the greenhouse following displays of small yellow ¯ owers that probably attract insects in situ (Fig. 3.2C). Its deep pink in¯ orescence bracts seem to advertise fruits rather than ¯ oral nectar. Unlike the similarly pigmented appendages born by many Billbergia that sometimes begin to fade even before the last ¯ ower opens, those of Aechmea bracteata remain turgid and bright, withering only after the comparatively inconspicuous berries begin to shrivel. Characteristics of ¯ owers, in¯ orescences and foliage further indicate that insects pollinate most members of several of the larger, predominantly terrestrial genera and also the large majority of Hohenbergia species. Cryptanthus, which deviate from other Bromelioideae by chromosome numbers and gender expression (subgenus Cryptanthus; Chapter 11), all Cambridge Books Online © Cambridge University Press, 2009 264 Reproduction and life history qualify. Andromonoecy (staminate and perfect ¯ owers on the same plant) prevails except for members of exclusively perfect-¯ owered subgenus Hoplocryptanthus. Corollas range from white to pink, and the exceptional species (e.g., C. exaltatus, C. odoratissimus in Hoplocryptanthus) emit powerful, spicy perfumes. Closely related Orthophytum includes populations with conspicuous red bracts (e.g., O. saxicola), while more of its membership exhibit entomophily (e.g., O. humile). Short, globose, axillary in¯ orescences obscured by dense foliage mark Greigia species as candidates for some of the more unusual pollination and seed dispersal syndromes among Bromeliaceae (Fig. 3.2E). Flowers range from drab (e.g., Mexican G. oaxacana) and obscured by subtending, foliaceous bracts and adjacent foliage to quite colorful (e.g., Chilean G. sphacelata) due to red to pink corollas and bracts. Some members of similarly overlooked Fascicularia, Fernseea and Ochagavia display showy, birdattracting in¯ orescences. Floral rewards Most Bromeliaceae reward pollinators with abundant nectar produced by glandular tissue located in gynoecial septa (Fig. 3.1A). Pollen may augment these secretions, and perhaps largely replace it for the exceptional, small¯ owered species (e.g., Fosterella penduli¯ ora; Fig. 3.4D). No evidence suggests that visitors collect resins, fragrances or any other nontrophic rewards, nor is ¯ oral deception a recognized strategy for fruit set by any Bromeliaceae. Nectar chemistry has become a relatively popular subject in recent years. Prior to Bernardello et al.' s (1991) study, readings were available for fewer than a dozen species (Percival 1961; Scogin and Freeman 1984; Freeman et al. 1985). Septal nectaries throughout Bromeliaceae ® t Fahn' s (1979) `structural' type, because they contain nectariferous parenchyma and a distinct epithelium (Fig. 3.1A). Sucrose, fructose and glucose occurred in every sample, whereas alkaloids, lipids, phenolics and protein did not, although antioxidant activity of undetermined origin characterized three species (Bernardello et al. 1991). Secretions collected from 12 of 20 taxa contained amino acids at concentrations (,7.5 mg ml21) lower than those detected in the extra¯ oral secretions of several of the same species. Refractometer readings indicated that dissolved solids ranged from about 16% in Dyckia ferox to 48% in one of the several samples obtained from Deuterocohnia longipetala. Disaccharides and monosaccharides among tested Pitcairnioideae agreed more closely, although the ratios of Cambridge Books Online © Cambridge University Press, 2009 Floral rewards 265 constituent sugars varied substantially among species. Hexose-rich products occurred exclusively in Dyckia ragonesei and D. velascana, while sucrose predominated in the nectars collected from the other taxa. Both Bromelioideae assayed, Aechmea distichantha and Bromelia serra, yielded hexose-dominated nectars, but the ratios of fructose to glucose were high and low respectively. Among surveyed Tillandsioideae, all of which belonged to either Tillandsia or Vriesea, sucrose again dominated, while glucose exceeded fructose. Con® rmed ornithophils presented sugars at concentrations up to twice that considered typical for bird-serviced ¯ owers in other families (Stiles and Freeman 1993). Ratios of the major constituents also deviated somewhat from expectations. Secretions of hummingbird-pollinated ¯ owers usually contain two to four times as much sucrose as hexose, and fructose and glucose approached parity, or the former predominated (Freeman et al. 1985). Puya demonstrates disparities between species pollinated by hummingbirds vs. passerines that Baker et al. (1998) suggest re¯ ect evolutionary responses to these distinct groups of avians. Sucrose dominated the sugars in nectar produced by four species (mean 57.0%) reliant on hummingbirds; hexose predominated in the secretions of three others regularly visited by passerine birds. Other values reported for Bromeliaceae generally agree with those obtained by Bernardello et al. (1991). Percival (1961) also encountered hexose-rich nectar in Aechmea bracteata, indicating that pollinators probably continue to visit this spontaneously autogamous species, while ornithophilous and self-incompatible Billbergia nutans differed with its sucrose-rich product. Freeman et al. (1985) noted high sucrose to hexose ratios for bird-serviced Tillandsia macdougallii. Puya spathacea, the only species examined by different investigators, yielded higher sucrose to hexose ratios for Scogin and Freeman (1984). Total solids also varied within species, perhaps re¯ ecting unequal amounts of evaporation. No relationship seems to exist between the structure of a nectary and the chemistry of its product. Martinelli (1994) determined the speci® c gravity of numerous ¯ oral nectars during his study of the pollination biology of 35 Bromeliaceae native to Atlantic Forest in Rio de Janeiro State, Brazil (Table 6.3). Readings ranged from 12.4% for Alcantarea regina to 30.6% for Billbergia amoena var. amoena. Mean sugar content for 27 taxa was 18.5%. Nectar volume averaged 11.7 ml for Quesnelia lateralis and 218 ml for Alcantarea regina, but higher sugar content in the ® rst instance (30.5 vs. 12.4% respectively) reduced the difference in calories per ¯ ower. Cambridge Books Online © Cambridge University Press, 2009 Table 6.3. Mean volume per ¯ ower and sugar concentration (range in parentheses) in the ¯ oral nectar of some Brazilian Bromeliaceae. Primary pollinators and breeding system are also provided where available Species Aechmea fasciata var. fasciata Alcantarea regina Billbergia amoena var. amoena Neoregelia marmorata Pitcairnia ¯ ammea var. pallida Pitcairnia ¯ ammea var. ¯ ammea Quesnelia lateralis Vriesea atra var. atra Vriesea haematina Vriesea hydrophora Vriesea neoglutinosa Volume (ml) 15.6 218 17.6 17.9 60.8 30.7 11.7 209 137 90.8 63.3 Sugar content (%) 28.0 12.4 30.6 26.0 12.7 17.4 30.5 13.3 21.1 13.0 19.8 (22± 32) (8± 20) (23± 32) (24± 30) (8± 19) (11± 25) (24± 36) (9± 19) (19± 23) (7± 15) (16± 28) Pollinator Breeding systema Hummingbirds SI Hummingbirds Hummingbirds Hawkmoth? Hummingbirds Hummingbirds? Bats Hummingbirds Bat; hawkmoth? Hummingbirds SC ? SC SC SI SC SC SC SC Source: After Martinelli (1994). Note: a SI, self-incompatible; SC, self-compatible. Cambridge Books Online © Cambridge University Press, 2009 Floral rewards 267 Less information is available on how Bromeliaceae reward nectarseeking bats than on birds. Sazima et al. (1995) assayed products obtained from those six Brazilian chiropterophilous vrieseas observed in situ. Dissolved solids ranged between 17.6 and 19.6%, meaning that concentrations were somewhat higher than the average for the other bat-serviced ¯ ora tested. Relatively modest nectar volumes per ¯ ower (,0.25 ml), and mostly distichous spikes that bear one or two open ¯ owers at a time, may oblige this premium (Fig. 3.5E). Conversely, enough of the dilute secretion (4.65%) offered by Encholirium glaziovii from the many ¯ owers open on a given day on its brush-type in¯ orescence accumulates to ¯ ow down the scape axis (Sazima et al. 1989; Fig. 3.4G). Nectar produced by three of the four night-¯ owering Vriesea species studied by Martinelli (Table 6.3) contained less sugar than the average for species in what is largely an ornithophilous group. Field work con® rmed ornithophily for several of the bromeliads surveyed for nectar chemistry. Chlorostilbon aureoventris and Sappho sparganura visited Dyckia ¯ oribunda, D. velascana and Puya spathacea ¯ owers. Chlorostilbon aureoventris foraged on Deuterocohnia longipetala, Dyckia ragonesei and Tillandsia lorentziana. Colibri coruscans cropped a population of Vriesea friburgensis. At least some ¯ oral syndromes lacked exclusivity. The butter¯ y Papilio thoas collected nectar offered by a cluster of Dyckia ¯ oribunda. Apis mellifera harvested pollen from these same Pitcairnioideae and Dyckia velascana, as presumably could native bees. Floral characteristics alone persuaded Bernardello et al. (1991) that Dyckia ferox, Deuterocohnia haumanii and Aechmea distichantha utilize birds. Tillandsia duratii, one of the most fragrant of all Tillandsioideae, also appeared to be psychophilous (pollinated by butter¯ ies), while its white¯ owered and comparably fragrant relative T. xiphioides probably attracts sphingids. Bernardello et al. (1991) disagreed with Varadarajan and Brown' s (1988) contended entomophily for Dyckia and Deuterocohnia. Dioecious Hechtia attracts diverse insects, according to Mitchell (1974) who recorded Bombus sp., Apis mellifera and an unidenti® ed beetle foraging mostly on male, `nutty' -smelling in¯ orescences of H. scariosa. Coleoptera by the dozen crawled over the ¯ owering shoots of a strongsmelling Hechtia sp. in central Mexico (Benzing, personal observation). Honey bees replaced a small wasp as the most frequent visitors to the male ¯ owers of H. schottii as the day progressed in a scrub forest in Yucatán State, Mexico (Fig. 6.2B). John Utley (personal communication) agrees that some members of this genus at least approach cantharophily (beetle pollination) according to ¯ oral characteristics (e.g., aminoid odor). Cambridge Books Online © Cambridge University Press, 2009 268 Reproduction and life history Fragrances Anecdotes about the qualities of ¯ oral fragrances among the bromeliads abound, but few of these publications (e.g., Hegnauer 1963, 1986; Chapter 13) include chemical determinations. Adjectives that range from sweet and pleasant (triterpenes such as citronellol, geraniol and nerol in some tillansdias; Hegnauer 1963) to musty, garlic-like, greasy, bituminous, as in coal gas, and reminiscent of spoiled cabbage indicate varied chemistry and diverse targets. Associations with speci® c ¯ ower form, color and timing indicate two groups of chemicals, one attractive to the common insect pollinators and the other to bats. Closely allied lineages tend to produce similar odors (e.g., Tillandsia, particularly subgenera Anoplophytum, Diaphoranthema and Phytarrhiza, and Catopsis sweet-smelling and Vriesea section Xiphion unpleasant types). Lures may be convergent with those produced by other taxa. Knudsen and Tollsten (1995) reported reliance on the same and related S-containing compounds by bat-serviced ¯ owers representing six families (no Bromeliaceae). Odors that attract prey to Brocchinia reducta shoots belong to the ® rst category (Chapter 5). Timing also distinguishes the fragrant-¯ owered bromeliads. Chiropterophilous types smell strongest at dusk or later during the night. Quite a few sweet-scented species do the same, or their emissions peak after sundown following more modest activity that day. Pale corollas, sometimes with ® mbriate margins (T. xiphioides; Fig. 3.3F), probably indicate moth pollination. Till (1992a) reported that several species of Tillandsia subgenus Diaphoranthema (e.g., T. aizoides, T. virescens) developed strong fragrances at about dawn, became odorless between about 10.00 and 17.00 hours, and then resumed advertisement through the night after which the corolla withered. Sazima et al. (1989) reported only a faint scent from the in¯ orescences of Encholirium glaziovii, which otherwise is well equipped for bat pollination. Knudsen and Tollsten (1995) interpreted this de® ciency as evidence of derivation from ornithophilous stock. If true, one ¯ oral character changed less than several others as ancestors abandoned one group of vertebrates for another. Flowering phenology Influences of pollinators Like most other ¯ ora, bromeliads use cues from the environment to coordinate fruit set and seed dispersal with local conditions that affect repro- Cambridge Books Online © Cambridge University Press, 2009 Flowering phenology 269 ductive success. Stimuli responsible for inducing speci® c phenomena parallel those that trigger the same processes in other tropical plants. Likewise, shoot and in¯ orescence architecture, ¯ ower physiology, and additional agencies constitute the inherent factors that determine for how long the individual bromeliad and its population engage in speci® c reproductive activities. Pollination demonstrates the con¯ uence of environment, plant and population in setting the schedule for one stage of the reproductive cycle. Cause and effect relative to who pollinates which population of Bromeliaceae and for how long each year seems straightforward enough at ® rst glance. Dependence on speci® c fauna indeed obliges temporal (e.g., diurnal vs. nocturnal anthesis) precision as illustrated by the ¯ oral syndromes previously described in Tillandsia. Pollinators further in¯ uence plant timing according to other aspects of their natural history, for instance the resident periods of migrant birds and the schedule of emergence and life spans of ¯ ower-visiting insects. Additional characteristics of the environment, like the seasonality of rainfall, affect the number, quality and success of seeds. In effect, phenology de® es explanation to the extent that environmental cues, animal biology and plant characteristics interact to set tolerances for plant schedules. Mechanisms that operate at the level of the ¯ ower, the plant, its population and the hosting community all affect the timing and direction of pollen exchange, and accordingly, in¯ uence the numbers and genotypes of progeny. Answers to a variety of questions would illuminate important determinates for speci® c Bromeliaceae. For example, how might ® tness be affected by the number of ¯ owers produced by a single ramet and the order in which they open? Do certain displays encourage or discourage visits by speci® c kinds of pollinators thereby affecting, for example, the proportions of single vs. biparental seeds or mate selection for allogamous parents? Do certain arrays of ¯ owers manipulate speci® c kinds of pollinators more effectively than others to promote fecundity and bene® cial combinations of genes? How should a population be arrayed in space, and the phenology of its membership synchronized, to maximize reproductive success? Again, Tillandsioideae provide some informative examples, and underscore the challenges of interpreting bromeliad reproduction in the context of adaptation. Depending on the species, a bromeliad (one genet, variable numbers of ramets) displays one to many pollen-receptive and/or pollen-donating ¯ owers on a given day, and a few days to many months suffice to complete anthesis. Impediments to, and tolerances for, adequate fruit set vary Books Online © Cambridge University Press, 2009 270 Reproduction and life history accordingly. For example, co-occurring individuals with massive, paniculate in¯ orescences (e.g., Tillandsia grandis, Alcantarea regina) need less synchronization to exchange genes than smaller, sparser-¯ owered types. On the other hand, each Tillandsia macdougallii specimen expends its modest (,15) complement of ¯ owers within a few days, and smaller-bodied species (e.g., T. ionantha, T. caerulea) must exchange pollen during even briefer intervals. Gardner (1984) reported that synchronized ¯ owering by Mexican T. andrieuxii and T. erubescens (,10 ¯ owers/shoot, 2± 3 open/day) assured that local populations completed anthesis within 2± 3 weeks. Some bromeliads follow more episodic schedules characterized by ¯ oriferous days interspersed among barren ones (e.g., T. schiedeana, Vriesea splendens). Fragrant, night-blooming Tillandsia dodsonii exhibits an especially curious pattern better known in some Neotropical orchids. Thousands of plants simultaneously, but irregularly, each produce one to four fragrant, nocturnal ¯ owers with large, white to creamy corollas per distichous, pendant spike (Fig. 3.5F). Cool nights following late afternoon showers supposedly promote this kind of behavior in Sobralia in some of the same Ecuadorian habitats. However, stimuli with the same effect operate elsewhere. Three colonies of T. dodsonii maintained under glass at the Marie Selby Botanical Gardens in Sarasota displayed 12 ¯ owers on eight spikes on 18 December 1996 (Benzing, personal observation). Nothing followed until ® ve nights later when 12 more buds opened on the same in¯ orescences. Several factors promote reproductive success among the more diminutive species that outcross; Tillandsia albertiana advertises its single, unusually heavy-textured, bright scarlet ¯ ower, one per shoot, longer than the usual 1± 2 days (Fig. 3.3L). It, like most of the other neotenic species, also grows gregariously. Well-established populations comprised of millions of less than fully synchronized shoots potentially exchange genes over extended intervals, about two months for T. usneoides in parts of the southeastern United States. Jaramillo and Cavelier (1998) recorded Tillandsia complanata in ¯ ower through nine consecutive months and anthesis among T. turneri specimens at the same Colombian site limited to two 3-month intervals. Kubisch (1965) reported that 27 Tillandsia species distributed across Mexico each ¯ owered over 3± 4-month intervals. Summer schedules predominated, but no month lacked reproducing populations. Two polymorphic taxa engaged pollinators twice at different times of the year at the same locations with different results. Tillandsia carlsoniae remained in anthesis from October to November and March to April (Wül® nghoff 1967), but no fruit developed Cambridge Books Online © Cambridge University Press, 2009 Flowering phenology 271 following the fall event. Tillandsia plumosa behaved similarly, except that phenology divided the population more evenly, and fruits developed on both occasions. However, Kubisch' s data say little about synchronization among co-occurring bromeliads, or the timing of ¯ owering of the same species at different locations. Martinelli (1994) examined the phenology of 35, mostly ornithophilous, bromeliads native to Atlantic Forest in Rio de Janeiro State and identi® ed two patterns described by Gentry (1974). Most common was the `steady state' arrangement (22 species). Plants in this category open 2± 4 ¯ owers each day or night over periods of three or more weeks. Another 13 subjects showed `cornucopia' -type phenology because they displayed 3± 8 receptive ¯ owers each day over just 3± 10 days. Billbergia pyramidalis var. pyramidalis and B. amoena var. amoena exhibited the second pattern, but populations showed less synchrony and ¯ owers were available for 8± 12 days. Martinelli also recorded a strong seasonal bias for these bromeliads, with 73% of the species ¯ owering between November and February, which happen to be the wettest and warmest months. Only seven species (23%) ¯ owered during the much drier, coolest months of April to September, perhaps in part because many of the local hummingbirds migrate to lower elevations to avoid the frequent misty days that also characterize midwinter. Coordinated reproduction can bene® t mixed ¯ oras and the pollinators they share for different reasons. Sometimes co-occurring populations reduce competition by ¯ owering asynchronously, while convergent schedules may bene® t any participant unable to attract enough attention without the assistance of one or more populations that share its pollinators. The ® rst arrangement may also set the stage for speciation leading to arrays of sympatric lineages like those interfertile Mexican Tillandsia species mentioned earlier. However, no pairs of bromeliads unequivocally rely on reproductive phenology to coexist, although one report describes a suggestive pattern. Ecological sorting of predisposed populations, or simple chance instead of concerted evolution, probably accounts for the staggered phenology of the several bromeliads (Aechmea nudicaulis, Guzmania monostachia, G. nicaraguensis) that constitute part of a mixed guild of bird-serviced species in a humid Costa Rican forest. Whether it was fortuitous, sorted or evolved in situ, Stiles (1978) demonstrated sequential ¯ owering that may bene® t participating ¯ ora through maintenance of a group of shared pollinators (Fig. 6.3). By offering nectar continuously rather than ¯ owering on more overlapped schedules, and then only for part of the year, these plants favor a well-fed avifauna and reduce competition for their services as pollen Cambridge Books Online © Cambridge University Press, 2009 272 Reproduction and life history Figure 6.3. Flowering phenologies over four years of three bromeliads and two nonbromeliads comprising part of a guild of ornithophilous herbs in a wet Costa Rican forest (after Stiles 1978). vectors. Incidentally, none of the participating bromeliads are sufficiently related to conclude that disruptive gene ¯ ow has in¯ uenced the guild' s ¯ owering schedule. Martinelli (1994) failed to identify mutually complementary phenology during his analysis of ¯ oral biology among 35 primarily ornithophilous Atlantic Forest Bromeliaceae (Fig. 6.4). Moreover, high rates of fruit set despite protandry that often obliges outcrossing demonstrated little or no competition for pollinators. Another report bears on the question of whether Bromeliaceae may in fact possess exceptional capacity to adjust anthesis to track changing environments. Wright and Calderon (1995) noted that ¯ owering among a 17-member bromeliad ¯ ora on Barro Colorado island, Panama (® ve during the dry season and the other 12 during wetter months) indicated no major phylogenetic or weather-related constraints on scheduling compared with some other local epiphytes (e.g., the more numerous orchid species). Several communities in southeastern Brazil suggest that co-occurring Bromelioideae partition the services of the local seed dispersers (Fischer and Araujo 1995). Photoperiodism An unknown number of Bromeliaceae cue on photoperiod to coordinate important plant activities, like ¯ owering and branching. Cultivated materials underscore the pervasiveness of day length as a ¯ owering stimulus. Cambridge Books Online © Cambridge University Press, 2009 Flowering phenology 273 Figure 6.4. Flowering phenology of 15 bromeliads native to wet Atlantic Forest at Macae de Cima (900± 1400 m), Rio de Janeiro State, Brazil distinguished by pollination syndrome. Stippled blocks represent intervals of maximum ¯ owering of night-¯ owering species. Black blocks indicate bird-serviced, day-¯ owering species (after Martinelli 1994). Raack (1985) recorded the dates of ® rst ¯ owering for more than 100 species and hybrids representing 12 genera growing under glass in southern Ohio. Just three subjects initiated anthesis during February while 15, the record number, did so during November. Timing and precision varied with the subject. Aechmea warasii began to ¯ ower between the ® rst and tenth day of January over four consecutive years. Guzmania zahnii routinely did so between the ® rst and middle of June. Exceptional species behaved less consistently, for example Guzmania sanguinea var. brevipedicellata, which entered the reproductive phase during March, July, December and November in as many years. Conspeci® c varieties (e.g., Aechmea fulgens) sometimes followed distinct schedules, and Vriesea simplex 3`Mariae' ¯ owered on dates falling roughly between those of its parents. Table 6.4 illustrates the mostly consistent ¯ owering exhibited by seven Billbergia taxa in central Florida. Mastalerz (1957) conducted greenhouse experiments to demonstrate Cambridge Books Online © Cambridge University Press, 2009 Table 6.4. Dates of maximum ¯ owering for plants representing clones of eight species of Billbergia under cultivation in central Florida. Note that winter ¯ owering is characteristic and phenology tends to be consistent for speci® c clones Species B. amoena var. amoena B. distachia var. distachia B. euphemiae var. euphemiae B. horrida var. tigrina B. nutans B. pyramidalis var. concolor B. saundersii var. debilis B. vittata clone A B. vittata clone B B. vittata clone C 1988 1989 1990 1991 1992 Ð 11 19 Feb Ð 11 30 Mar 3 Mar 4 Jan 8 Nov Ð 11 17 Jan 18 Jan 2 Feb 2 Feb 10 Oct 28 Apr 6 Mar Ð 11 18 Nov 6 Jan 22 Jan 3 Jan 10 Feb 10 Feb 30 Mar 17 Mar 17 Mar 21 Feb 10 Nov 21 Jan 21 Feb Ð 11 21 Jan 16 Mar 30 Mar 10 Apr 2 Apr 25 Dec 16 Oct 18 Dec 16 Mar Ð 11 Ð 11 1 Jan Ð 11 Ð 11 Ð 11 Ð 11 27 Jan Ð 11 Ð 11 Ð 11 Source: Data provided by D. Beadle of Venice, Florida. Cambridge Books Online © Cambridge University Press, 2009 1993 Ð 11 Ð 11 Ð 11 26 May Ð 11 Ð 11 Ð 11 Ð 11 Ð 11 Ð 11 1994 1995 14 Jan Ð 11 11 Mar Ð 11 1 Mar Ð 11 Ð 11 14 Jan Ð 11 14 Jan 12 Mar Ð 11 22 Mar Ð 11 Ð 11 12 Mar Ð 11 Ð 11 Ð 11 Ð 11 Flowering phenology 275 photoperiodic ¯ owering in Billbergia nutans. Day lengths extended with arti® cial light delayed the onset of ¯ owering from mid to late January to about 1 April in one run. Another group of plants covered with black cloth to effect short days ¯ owered four days before the fully exposed controls. Subjects maintained in a vegetative state with a simulated summer regimen, nevertheless, produced ramets, thus engaging in what is usually a post-¯ owering event. Photoperiodism may be especially important for B. nutans because it ranges farther poleward in Brazil than the other members of its genus. Downs (1974) investigated photoperiodism among a selection of Bromeliaceae with `not very conclusive' results except for deciduous Pitcairnia heterophylla (Fig. 2.12A). After each treated shoot had generated about 16 of the unarmed green leaves under 12-h or shorter days, the scalelike, spiny, nongreen foliage and bulbous base that usually presage ¯ owering began to appear. Fourteen-hour days prevented this transition, but not the formation of additional green leaves. Shoots induced to swell under short days reverted to the production of green leaves as if some additional requirement for ¯ owering remained unsatis® ed. Neither simulated drought followed by heavy irrigation nor defoliation induced in¯ orescences to appear. However, after producing additional linear leaves, and without altering the photoperiod, plants bulbed and ¯ owered. Apparently, ¯ oral induction requires short days, and, judging by the failure of short-day subjects with less foliage to reproduce, also the presence of more than 16 green leaves. Inadvertent smoke-induced ¯ owering by pineapple plants in Hawaii ® rst alerted investigators to a pervasive chemical sensitivity (Downs 1974). Subsequent inquiry demonstrated that a variety of synthetic auxins (e.g., 2,4-dichlorophenoxyacetic acid, 1-naphthalene acetic acid, indole acetic acid), ethylene and related compounds stimulate diverse Bromeliaceae including Ananas. Beta-hydroxyethylhydrazine (BOH), an ethylene-generating compound, continues in widespread commercial use, but the underlying mechanisms remain poorly understood. Ethylene and acetylene promote auxin synthesis during several other growth responses (e.g., hypocotyl unbending), and probably act similarly in responsive bromeliads, or these agents somehow sensitize shoot meristems to endogenous auxins and possibly other native growth factors. Hydroperiod, broadly de® ned as changes in the height of the water table and the arrival of dry or wet weather, coordinates the activities of many tropical plants. Phenology related to the second stimulus remains undocumented in Bromeliaceae, but certain behaviors are suspicious, for example Cambridge Books Online © Cambridge University Press, 2009 276 Reproduction and life history the weather-related altered ¯ owering schedules exhibited by some of those phytotelm species comprising the described Costa Rican guild of ornithophils (Stiles 1978). A closer look at ¯ owering, bud break and, for the deciduous types, leaf turnover relative to moisture supply rather than some accompanying seasonal cue like photoperiod could prove rewarding. Breeding systems Plants observed in situ and in cultivation, ¯ oral morphology and several experiments con® rm the existence of diverse breeding systems among the bromeliads. Speci® c mating systems demonstrate expected correlations with certain aspects of natural history and ecology; occurrences often follow taxonomic boundaries (Table 6.3). For example, many watch-spring Billbergia species (subgenus Helicodia) regularly set self-seeds, whereas members of putatively more primitive subgenus Billbergia generally do not. The known exceptions among subgenus Billbergia occupy ranges that usually extend beyond the geographic center for the subgenus in southeastern Brazil, suggesting derivation from what may have been autogamous founders in a largely self-incompatible clade (e.g., B. amoena var. amoena, B. pyramidalis var. pyramidalis). Additional features that in¯ uence the proportions of uniparental vs. outcrossed progeny characterize many bromeliads. Dichogamy (asynchronous maturation of sex organs) and herkogamy (juxtapositions of sex organs that limit sel® ng) promote allogamy for many of the self-compatible types, and for the self-incompatible plant may help diminish stigma clogging. Protogyny occurs almost without exception among the .150 members of Tillandsia subgenus Tillandsia (Gardner 1982). Martinelli (1994) reported protandry for all 17 of the bat or bird-pollinated Vriesea species (not all closely related) he studied in Brazil. Dichogamy or weak protandry characterized the four Quesnelia species also included in his survey. Plant characteristics in addition to those already mentioned and substrates also seem to in¯ uence the breeding mechanisms of Bromeliaceae. For example, species cultivated by ants (e.g., Aechmea mertensii, A. tillandsioides) regularly set self-seeds (Madison 1979), while Bush and Beach (1995) suggested that epiphytism generally favors autogamy. All of the monocarpic Bromeliaceae can probably set self-seed, some routinely, yet many of the same species invest substantial resources to attract pollinators. Puya raimondii sets it own fruits with up to 8000± 10000 large, brightly pigmented ¯ owers well provisioned with nectar. Considerably smaller Brocchinia tatei performs similarly with hundreds of yellow, diurnal Cambridge Books Online © Cambridge University Press, 2009 Breeding systems 277 ¯ owers. Tillandsia prodigiosa and T. eizii display large primary bracts colored bright rose to soft pink and green to light purple from which tubular corollas protrude. Spectacular Alcantarea regina produces showy yellow corollas in addition to red bracts. Allogamy, if basic to Bromeliaceae as McWilliams (1974) suggested, has repeatedly given way to arrangements that favor or oblige (e.g., via cleistogamy) offspring sired by a single parent. Closely placed, developmentally synchronized stigmas and anthers probably reinforce self-compatibility to account for the high rates of fruit set characteristic of monocarpic Bromeliaceae. Maturing stamens reorient to contact any still receptive stigmas in some populations of Tillandsia utriculata (Gardner 1982). Whatever the fundamental condition for the family, biparental reproduction enforced by self-incompatibility (SI) occurs widely in Bromelioideae and Tillandsioideae and, although mostly uncon® rmed, probably in Pitcairnioideae as well. The only genetically con® rmed case of SI in Bromeliaceae involves Ananas comosus, and the mechanism is homomorphic gametophytic SI; Ananas ananassoides and A. bracteatus can self (Brewbaker and Gorrez 1987). Distributions across Magnoliophyta suggest that neither homomorphic sporophytic SI nor heteromorphic SI is likely to occur in Bromeliaceae, while the still poorly understood, late-acting SI systems cannot be ruled out. Martinelli' s (1994) discovery that tubes produced by allogamous pollen on the stigmas of certain apparently self-incompatible Vriesea species native to southeastern Brazil grew faster than those from self-pollen supports McWilliams' s claim and underscores the subtly mixed nature of at least some of the sexual systems of hermaphroditic Bromeliaceae. Self-sterility is more difficult to demonstrate, especially in situ, where a variety of environmental factors also reduce fruit set. Consistently barren in¯ orescences among stock protected from predators and pollinators more reliably signal SI, although the behaviors of individual plants may not re¯ ect that of populations and certainly not widespread species. Specimens collected from a particularly ornamental colony of Tillandsia caputmedusae near San José, Costa Rica routinely fail to set fruit in the Oberlin College greenhouse, whereas stock originating from another population in southern Mexico characterized by a much duller, pink in¯ orescence do so every year in the same enclosure. Numerous additional Tillandsioideae (e.g., T. bulbosa, T. polystachia, T. stricta, Guzmania monostachia) exhibit similar color polymorphisms and sometimes unusual mechanisms to insure some reproduction even if the Cambridge Books Online © Cambridge University Press, 2009 278 Reproduction and life history sexual process fails. Wide-ranging Tillandsia paucifolia sets self-fruit in Florida, while certain apparently self-incompatible relatives in South America proliferate offshoots on otherwise barren in¯ orescences (Fig. 2.11A). Tillandsia dasyliriifolia operates the same way in the tintales (low inundated forests) of Yucatán State, Mexico (Fig. 6.5C). Martinelli (1994) conducted self and cross-pollinations and concluded that at least 20 (mostly Tillandsioideae) of the 35 Atlantic Forest bromeliads he manipulated can produce self-seeds. Fluorescence microscopy indicated masses of tubes from selfed grains, many extending into the ovules. Pollen from seven more species, all Bromelioideae, also germinated on the stigmas of the donors, but their tubes barely penetrated into the style. No obvious aspects of life history or ecology distinguished these plants in ways that might explain why they possess different sexual systems. Mechanisms other than SI oblige outcrossing for many bromeliads (Fig. 3.3H). Dioecy occurs in every subfamily, but not extensively (few species), and close relatives sometimes breed by different mechanisms. Most of the large genera exhibit hermaphroditism throughout, Aechmea mariaereginae being the single major exception. Predominantly dioecious clades concentrate in Central America. Wholly dioecious Hechtia (,50 species) occupies a primarily Mexican range (center of diversity in Chiapas State), with no species reported south of Nicaragua. Sexually mixed Catopsis occur from Mexico to Panama, although occasional perfect-¯ owered populations extend to the east and south (C. berteroniana from Florida to southeastern Brazil; Table 6.5). Aechmea mariae-reginae and related monotypic Androlepis range from Costa Rica into Colombia and Venezuela. Should Cryptanthus, and perhaps also Dyckia, contain dioecious members, then conditions favoring this arrangement for Bromeliaceae must also exist south of the Equator. Smith' s (1986) brief allusion to his discovery of a unisexual ¯ ower ± gender not provided ± on a herbarium specimen of a Cottendor® a demonstrates our ignorance about even the most basic aspects of reproduction among the more horticulturally obscure components of this family. Flower structure suggests that Catopsis exceeds all the other bromeliad genera for varied and sometimes labile gender expression. What appear to be consistently dioecious or hermaphroditic species, about six of each (Palací 1997; Table 6.5), comprise about two-thirds of the genus. The remaining members for which we have information appear to be either predominantly dioecious (e.g., C. morreniana) or perfect-¯ owered (e.g., C. berteroniana, C. wangerinii). Catopsis nutans follows a site-speci® c pattern with perfect-¯ owered populations in Florida and dioecious forms in Mexico, El Cambridge Books Online © Cambridge University Press, 2009 Breeding systems 279 Figure 6.5. Aspects of bromeliad reproduction. (A) Seedlings of an unidenti® ed epiphytic member of Bromelioideae growing on a moss-covered branch in Atlantic Forest in Rio de Janeiro State, Brazil. (B) Seedling of an unidenti® ed Encholirium sp. characteristically growing against a rock removed from a campos rupestres habitat in Minas Gerais State, Brazil. (C) Offshoots propagating off the in¯ orescence of Tillandsia dasyliriifolia in Yucatán State, Mexico. (D) Coma hairs of Tillandsia balbisiana (3450) showing morphology that may increase adhesiveness to suitable substrates. (E) Approximately two-year-old seedlings of Tillandsia paucifolia that as seeds had been glued to this Taxodium distichum branch (about actual size). (F) Seedlings of Wittrockia superba on a moss-covered rock in Atlantic Forest in Rio de Janeiro State, Brazil. (G) Approximately three-month-old seedling of Tillandsia balbisiana (3125). Books Online © Cambridge University Press, 2009 280 Reproduction and life history Table 6.5. Gender expression in Catopsis C. berteroniana C. compacta C. delicatula C. ¯ oribunda C. hahnii C. juncifolia C. mexicana C. micrantha C. minimi¯ ora C. montana C. morreniana C. nitida C. nutans C. paniculata C. pisiformis C. sessili¯ ora C. subulata C. wangerinii C. wawranea C. werckleana Flowers perfect or plants rarely functionally unisexual (southern Mexico) Dioecious Status of taxon unclear Flowers perfect Dioecious Flowers perfect Status of taxon unclear Dioecious Status of taxon unclear Dioecious Mostly dioecious ¯ owers, sometimes perfect (Costa Rica) Flowers perfect Flowers perfect or plants dioecious (Mexico and Central America) Status of taxon unclear Flowers perfect Flowers perfect or plants dioecious (Mexico and Central America) Dioecious Flowers perfect, or plants rarely dioecious Status of taxon unclear Status of taxon unclear Source: After Palací (1997). Salvador and Guatemala. Panamanian Catopsis pisiformis produces one type of ¯ ower with functional gynoecia, but the anthers contain no effective pollen (Rauh 1983a). Monomorphic, functionally unisexual ¯ owers also characterize certain populations of C. morreniana. Insights on questions such as how often and why dioecism evolved in Catopsis should emerge as knowledge of phylogeny increases. Meanwhile, sex ratios and other potentially revealing demographic data need to be collected before more of the wild populations disappear. Several colonies of Hechtia schottii contained from 1.7 to 2.0 pistillate to staminate plants in Yucatán State, Mexico, indicating that ratios among dioecious Bromeliaceae can deviate from 1:1 (I. Ramírez, personal communication). Synchronization within populations Data for Bromeliaceae permit some remarks about reproductive coordination within populations. Hechtia schottii devotes enough axillary meristems to in¯ orescence to support ¯ owering through much of the year in Yucatán Cambridge Books Online © Cambridge University Press, 2009 Genetic structure of populations 281 State, Mexico (I. Ramírez, personal communication; Fig. 6.2B). Large plants often bear two in¯ orescences at different stages of maturity, unlike many of the other members of the same genus. Likewise, except for the occasional errant ramet, populations of the sympodial bromeliads usually ¯ ower during the same short intervals each year. Plant longevity, minimum size at ¯ ower induction and year-to-year differences in seedling recruitment in¯ uence what fraction of a population of a monocarpic bromeliad ¯ owers during a given season, as documented for Puya dasylirioides below. Candidates for mass ¯ owering include Tillandsia imperialis, Alcantarea imperialis and at least one Brocchinia. Not one specimen of giant Brocchinia micrantha growing along the highway ascending the escalara to the Gran Sabana in eastern Venezuela bore fruits or ¯ owers among the dozens of individuals large enough to do so (Benzing, personal observation). Perhaps too few of these exceptionally long-lived bromeliads typically co-occur to more than rarely encounter simultaneous ¯ owering at a single location. Moreover, just one individual of smaller-bodied and presumably shorter-cycled Brocchinia tatei in the same locality was reproducing that year. Longer-term studies indicate that some years can be more fruitful than others for slow-growing monocarps. Numbers of ¯ owering specimens of effectively semilaparous Puya dasylirioides varied almost 10fold during the four-year study conducted by Augspurger (1985) and discussed below. Specimens of shorter-cycled Tillandsia utriculata fruit every year in Florida, but no counts are available (Benzing, personal observation). Genetic structure of populations Soluble proteins (allozymes) provide markers to determine the genetic structures of populations and patterns of gene ¯ ow. Findings for Bromeliaceae accord with those described for other ¯ ora relative to ¯ oral biology and certain additional aspects of life history. For example, Soltis et al. (1987) chose two bromeliads to characterize the effects of different breeding systems on the genetics of arboreal vs. terrestrial types. Specimens representing Central American Tillandsia ionantha (subgenus Tillandsia, Group One) came from Mexico where brightly colored foliage during anthesis, a stiff, tubular corolla, and uneven seed crops indicate ornithophily (Fig. 2.10M,N). Plants often failed to set fruit in closed greenhouses. Tillandsia recurvata (subgenus Diaphoranthema), the bromeliad exceeded only by T. usneoides for greatest distribution, grew in the same general area. Consistent with experience across its geographic range, fruits routinely Cambridge Books Online © Cambridge University Press, 2009 282 Reproduction and life history developed following displays of small, poorly advertised ¯ owers. Comose, wind-carried seeds (Fig. 3.6J) characterize both populations, although mobilities may differ, as described below. Allozymes coded by 19 loci revealed pronounced inbreeding for Tillandsia recurvata (ballmoss), while T. ionantha (16 loci examined) more closely matched the genetic pro® le of an outcrosser. Speci® cally, values for the proportions of polymorphic vs. homomorphic loci, mean heterozygosity, mean number of alleles per locus, and observed heterozygosity all exceeded those for T. ionantha. Like other autogamous species, T. recurvata exhibited less genetic variety locally, but substantially more diversity overall. Birds apparently mix the alleles of neighboring Tillandsia ionantha. Inbreeding coefficients (F values) approached those expected at Hardy± Weinberg equilibrium, whereas numbers calculated for T. recurvata indicated almost no heterozygotes. However, even as an outcrosser, T. ionantha demonstrated less overall genetic similarity than most allogamous plants, in part because so many of the characterized loci code for more than one allozyme, i.e., are polymorphic. Findings also indicated diploidy or diploidized, polyploid genomes with the single exception of the locus responsible for phosphoglucomutase, which in T. recurvata and closely related T. usneoides coded three isoforms. Clearly ballmoss and Tillandsia ionantha organize genetic variety in different patterns with potentially different consequences. Limited gene exchange may promote capacity to accommodate diverse, site-speci® c growing conditions for T. recurvata. Similar behavior allows many a cosmopolitan weed to tolerate diverse soils, climates and land uses far beyond native habitats. Adventiveness in ballmoss, speci® cally its ability to rapidly colonize so many kinds of substrates, including telephone wires, may rest on a similar genetic foundation (Fig. 1.3A). However, T. usneoides, even more than T. ionantha, amply demonstrates that allogamy, in this case mediated by insects visiting fragrant nocturnal ¯ owers, need not impede success across broad, ecologically heterogeneous ranges. Kress et al. (1990) examined four enzyme systems (isocitrate dehydrogenase, 6-phosphogluconate dehydrogenase, phosphoglucose isomerase and phosphoglucomutase) to conclude that sampled populations of Tillandsia recurvata, T. usneoides and T. utriculata in central Florida exhibit distinct genetic structures related to their disparate breeding systems. Autogamous Tillandsia utriculata was monomorphic for two of the three readable enzymes and polymorphic for the third. Tillandsia recurvata was polymorphic only for 6-phosphogluconate dehydrogenase. Every subpopulation of Cambridge Books Online © Cambridge University Press, 2009 Genetic structure of populations 283 Spanish moss possessed more than one allele for all three enzymes except for a single instance of monomorphy for phosphoglucomutase. Izquierdo (1995) mapped genetic variety in representative populations, or in one case the only known population, of four species of Aechmea subgenus Podaechmea in Mexico. Aechmea mexicana and A. lueddemanniana occur widely compared with A. macavughii and A. tuitensis, which occupy highly insular ranges in the west central part of that country. Izquierdo' s results only partially matched predictions from geography. Aechmea mexicana yielded the lowest values for observed heterozygosity (0.054) and mean number of alleles (2.4) per polymorphic locus. Almost half of the loci examined were polymorphic, whereas the two narrow endemics exhibited 2.5 and 2.6 alleles respectively per polymorphic locus. Most of the populations sampled for each species included high frequencies of homozygotes and evidence of pronounced genetic drift. Overall, Izquierdo' s results indicate constrained gene ¯ ow among all four species probably fostered by extensive inbreeding and low seed mobility. Clonal growth also strongly in¯ uenced genetic structure. For example, the most proli® c genet of A. tuitensis at the single site sampled accounted for 32.1% of all the local ramets. Murawski and Hamrick (1990) examined nine colonies of terrestrial Aechmea magdalenae, seven located on Barro Colorado island and two on the adjacent mainland, to characterize the genetic structure of a plant chosen primarily to represent tropical understory herbs. Of 18 loci that regulate 10 enzyme systems, six were polymorphic for at least one of those colonies with the effective number of alleles ranging from 1.09 to 1.25 (xÅ 51.21), more than recorded for either Tillandsia ionantha or T. recurvata, but below values obtained for the Mexican Aechmea species. Mean percent polymorphic loci within colonies was 24.1%, placing A. magdalenae below average among sampled herbaceous clone-formers. Highest genetic diversities characterized those colonies featuring the largest numbers of ramets, and mean heterozygosities exceeded values expected at Hardy± Weinberg equilibrium. Murawski and Hamrick concluded that small effective population sizes (number of genets) probably initiated by at most a few founders, and subsequently augmented by outcrossed progeny in most of the nine cases, accounted for the large amongpopulation components of genetic diversity. Frequent genetic identity between neighbors (within 10 m) demonstrated reliance on stolons for colony development. Population structure paralleled reproductive biology here as in the Tillandsia species with distinct breeding systems and those four self-fertile, Cambridge Books Online © Cambridge University Press, 2009 284 Reproduction and life history architecturally varied members of Mexican Aechmea (e.g., A. mexicana with short vs. A. tuitensis with long ramets). However, Aechmea magdalenae represents a special case because it rarely fruits, just three events in as many years of observation. Each successful pollination observed during the study involved a single shoot, always within one of the smaller colonies. All of the other ramets branched after achieving some threshold size, perhaps following spontaneous abortions of the still vegetative meristems. Trap-lining Phaethoris superciliosis alone visited the few ¯ owers produced to set the multiseeded, large, ¯ eshy berries that probably account for the multiple genotypes within colonies. Aechmea magdalenae also exhibited a more structured gene pool than the co-occurring trees, probably owing to its patchy distribution, large numbers of densely packed ramets, and shorter-range seeds. In the ® nal analysis, A. magdalenae was a poor choice for a wild type if its periodic cultivation for ® ber by indigenous people has altered the reproductive performances and related genetic structures of the populations Murawski and Hamrick assayed in Panama. Seed dispersal Morphology described in Chapter 3 con® rms the dispersal of bromeliad seeds by diverse fauna, wind, gravity and, probably for the exceptional species, ¯ owing water. We need to review this information before dealing with issues of seed transport and consequences for plant distribution and evolution. Again, as with pollination and breeding systems, an uneven literature precludes equal treatments of the subfamilies. Tillandsioideae Tillandsioideae disperse among arboreal and lithic substrates via small, wind-transported seeds structured according to a single aerodynamic design (Fig. 3.6J; Chapters 12 and 13). Most notable is the coma, which consists of numerous hairs extending from both ends of the integument (Alcantarea, Catopsis, Glomeropitcairnia), or just its base (all the other taxa). If hygroscopic movements like those that assist targeting by some other anemochorous taxa (e.g., Salix) also bene® t Bromeliaceae, they remain unreported. Subtle embellishments of a different type probably do increase seed success in another way. Hairs comprising the ¯ ight apparatus of certain Catopsis species possess barb-like ends and these, like the joints and projections featured in some other taxa (e.g., Tillandsia balbisiana; Fig. 6.5D), seem likely to promote adhesion to substrates and perhaps especially Cambridge Books Online © Cambridge University Press, 2009 285 Seed dispersal Table 6.6. Distance traveled by the seeds of 10 bromeliads native to Florida with coma present and removed. Note that the coma promotes airworthiness to different degrees for different species Distance traveled (cm) Species Catopsis berteroniana Catopsis ¯ oribunda Catopsis nutans Guzmania monostachia Tillandsia fasciculata Tillandsia paucifolia Tillandsia pruinosa Tillandsia setacea Tillandsia usneoides Tillandsia utriculata With coma Coma removed 115 73 125 80 130 110 62 145 105 132 26 46 53 55 60 55 50 30 50 25 Source: After Bennett (1992c). hospitable ones. Experiments indicate that buoyancy and mobility correlate with the apportionment of mass between the coma and the seed proper. Bennett (1992c) investigated seed dispersal using four Tillandsia species chosen to represent closely related bromeliads distinguished by polycarpy vs. monocarpy and occurrences on bark or rock or both media interchangeably. Peruvian T. sphaerocephala grows as a lithophyte in semiarid habitats; T. ionochroma occurs in the same general area with populations distributed either on small trees and shrubs or on rock outcrops. Nothing is known about the genetic structure of this second taxon, hence the possibility of substrate-speci® c ecotypes. Tillandsia fasciculata and T. utriculata usually anchor on trees, the latter sometimes on rock, through much of Mesoamerica and parts of northern South America. Both populations Bennett tested root exclusively upon, or at least begin life anchored to, trees in southern Florida. Of Bennett' s four subjects, only T. utriculata fruits just once and perhaps routinely so only in Florida. He recorded three features of seeds related to their mobility: terminal velocity, distance traveled in a wind tunnel, and the morphology responsible for these performances. Sedimentation rates differed substantially (Table 6.6). Seeds of T. utriculata settled slowest (0.21 m s21), and those of T. sphaerocephala half again as fast in still air Cambridge Books Online © Cambridge University Press, 2009 286 Reproduction and life history (0.33 m s21). Distances traveled ranked the two taxa in opposite order (110.9 vs. 88.2 cm respectively). More broadly, the epiphytes achieved relatively low terminal velocities at least in part because they allocate proportionally more biomass to the coma vs. the seed proper. Percentages (60.0± 61.7%) of the aggregate seed mass represented by the ¯ ight apparatus grouped the obligate (T. utriculata and T. fasciculata) and facultative (T. ionochroma) epiphytes together, with saxicolous T. sphaerocephala (41%) as the outlier. In short, the more consistently bark-dependent the taxon, the greater the relative cost of the coma, the more buoyant its seeds, and the greater the dispersal range. In addition to providing buoyancy, the coma must secure the seed after impact until germination occurs and roots replace hairs for holdfast. Successful dispersal measured by adhesiveness varied inversely with terminal velocity and plant habit. Finally, comas responded unevenly to different kinds of targets. Seeds of all four species blown against wood, masonite or even less bark-like concrete blocks in the wind tunnel more often stuck to the ® rst two media, and differences continued after that. Impacted propagules of T. ionochroma remained most securely attached to all three substrates despite its identity as a lithophyte, whereas those of T. utriculata proved to have the weakest grip. An expanded set of subjects might have performed less consistently had it contained certain additional taxa. For example, Alcantarea (15 species) occupy granitic outcrops in southeastern Brazil, sometimes as narrow endemics (Fig. 1.2C). Quite likely seeds equipped with the exceptionally short appendages illustrated by these bromeliads (e.g., apical and basal comas approximately one-third and one-half relative to the length of the whole propagule respectively for A. nevaresii; Leme and Marigo 1993) re¯ ect the island-like distributions of their habitats, and perhaps also the surface textures of the local rock outcrops (Fig. 7.1G). Garcia-Franco and Rico-Gray (1988) conducted the ® rst in situ study of bromeliad dispersal using a small patch of cloud forest in the Parque Ecologico, near Xalapa, Mexico. Tillandsia deppeana, a wide-ranging, montane, phytotelm epiphyte that grows predominantly (58% of adults) on local Liquidambar styraci¯ ua, inspired the experimental design and provided the test material. Preliminary observations indicated that the surveyed phorophytes supported on average 4.2 adult T. deppeana specimens, 36% of which occurred between 3 and 11 m above the ground. Mean distance between occupied trees was 14.8 m and the two-dimensional crosssection of an occupied crown averaged about 88 m2. Fecundity equaled 3839 seeds per ramet. Cambridge Books Online © Cambridge University Press, 2009 Seed dispersal 287 A single tray containing 8500 freshly collected seeds dusted with a UV¯ uorescent powder and placed on a branch 6.8 m off the ground in the `origin tree' served as the parent bromeliad. Five days later, the authors scanned the six down-wind traps, each of which consisted of a suspended series of ® ve rectangular strips of sticky tape exposing a total of 3.4 m2 of adhesive surface. More than half (4449) of the initial stock remained on the tray, and of those missing just 171 had impacted the tapes and then always between 3.3 and 10.6 m above ground. Sixty-two percent traveled between 8.8 and 15.0 m and 28% between 24.5 and 28.8 m from the source. The most mobile seed ¯ oated 38.0 m with just a 0.037% likelihood of reaching any of the six targets. Garcia-Franco and Rico-Gray used the observed outputs of the local bromeliads, population density (4.2 fruiting plants per tree) and the mean area (cross-section) of the local Liquidambar styraci¯ ua crowns to calculate that 4415 seeds encounter, but not necessarily adhere to, each phorophyte each year. If true, pre- and post-germination mortality reduces this number of offspring dramatically and obliges the uncommon T. deppeana in the Parque Ecologico that does achieve maturity to potentially disperse multiple crops of seeds to sustain the population, i.e., to be polycarpic. We return to the circumstances that favor iteroparity vs. monocarpy at the end of this chapter. Although impressive given the challenge, Garcia-Franco and RicoGray' s experiment failed to faithfully simulate natural conditions. First, they made no effort to assess recruitment where success probably occurs most frequently ± on the origin tree. Most of the one and two-year-old T. paucifolia seedlings surveyed in Florida shared cypress tree crowns with probable maternal parents (Benzing 1981a). Second, sticky tape may intercept airborne seeds more or less effectively than twigs and branches that exhibit more varied surface textures and effects on passing air streams. Finally, Liquidambar crowns represent three- rather than two-dimensional targets. Additional problems also probably compromised their results, but the willingness of these investigators to seek explanations for complex phenomena in situ deserves high praise. Pitcairnioideae Septicidal capsules derived from fully superior to inferior ovaries and medium-sized to small seeds suggest relatively uncomplicated, mechanical dispersal for most Pitcairnioideae. Membranous or hair-like extensions of the outer integument of the seed often promote buoyancy, except for Navia Cambridge Books Online © Cambridge University Press, 2009 288 Reproduction and life history that alone lacks a two-layered testa (Fig. 3.9). Occasional arrangements suggest extraordinary functions. Gross (1993) assigned the spongy, textured wing of Pitcairnia aphelandri¯ ora, a bushy species described from along the Napo River in Ecuador, and Pepinia punicea importance for water transport ± a reasonable supposition that warrants testing. Zoochory seems less likely among Pitcairnioideae, although cryptic elaiosomes like those attached to the dry seeds of other myrmecochorous ¯ ora may turn up yet. Rodents present another possibility if wild relatives share the partiality domesticated guinea pigs exhibit for certain Pitcairnia and Puya seeds. Diverse shapes and sizes, which exceed those of Tillandsioideae and most Bromelioideae (although delivery in a berry complicates comparisons on this second count), suggest multiple dispersal mechanisms and diverse rooting media. Whatever its biological basis, seed form circumscribes several genera enough to serve as a taxonomic marker. Pitcairnia seeds, for example, possess hair-like projections from both ends of the testa, while those of closely related Pepinia bear a wing more reminiscent of Puya (Fig. 3.9). The occasional epiphyte (all facultative; e.g., Pitcairnia heterophylla, Brocchinia tatei) possesses arguably the most airworthy seeds in the subfamily. Conversely, low seed mobility elsewhere helps explain the often clumped dispersions and fragmented populations of certain terrestrials (e.g., Andean Puya species; Fig. 9.2). However, rather ordinary morphology need not preclude long-range dispersal. Pitcairnia, for example, ranks among the most widely distributed of the bromeliad genera, and P. feliciana alone illustrates the successful outcome of a transoceanic dispersal (Fig. 1.1). Insularity, mostly on tepuis, like isolation on oceanic islands, possibly favored the unappendaged and probably short-ranged seeds of Navia. Although basically terrestrial, Pitcairnioideae exploit a variety of kinds of soils and, quite often, less accommodating substrates like precipitous rocky cliffs (e.g., various Hechtia, Navia, Pitcairnia; Fig. 7.1B). Perhaps the myriad shapes and sizes (six major classes; Fig. 3.9) of the seeds that characterize this subfamily partly re¯ ect the microtopography of rooting media, which for the saxicoles may parallel the lithology of colonized outcrops. Brocchinia (Fig. 6.1D) offers extraordinary opportunity to match bromeliad seeds with speci® c kinds of substrates because its fewer than 20 species colonize bark, soil and rocks alone or interchangeably. Data on seed longevity and requirements for germination (e.g., light) would complement those ® ndings. Cambridge Books Online © Cambridge University Press, 2009 Seed dispersal 289 Bromelioideae Mostly baccate fruits assure that seed dispersal among the bromeliads reaches its greatest complexity in Bromelioideae (Fig. 3.6A± H,L). Several members of this subfamily employ ballistic mechanisms, and many more of the zoochorous types produce appendaged seeds perhaps better suited to adhere to feathers, beaks or fur than pass through a gut (Fig. 3.6L). Fruits of Ronnbergia explodens burst at ripeness to release numerous sticky seeds, whereas a light touch induces R. deleonii to eject similar projectiles several meters. A third set of taxa (e.g., several Aechmea species) encourage ants to carry seeds laced with pheromone-like chemicals (Davidson and Epstein 1989; Seidel et al. 1990). Fruits of additional taxa (e.g., Fascicularia, certain Orthophytum) more closely approach capsular than ¯ eshy status. However, most Bromelioideae utilize avians, which means that the extensive literature on ornithochory based on ® ndings for other, better-known plants may also provide insights on how many hundreds of bromeliads manipulate similar fauna to disperse their seeds. Almost certainly Bromelioideae employ some of the same mechanisms described in other families. Unquestionably, reproductive variety within Bromelioideae exceeds that implied by the typically terse statements offered by taxonomists to the effect that many-seeded, inferiorovaried, ¯ eshy fruits characterize the entire taxon. Devices and mechanisms that mediate zoophilous pollination are more amenable to selection by ¯ ower visitors than aspects of fruits and seeds by frugivores because plant bene® t increases as pollen moves among rather than away from members of the same species. The probability that a seed will succeed rises as distance from conspeci® cs, whether the parent or some less closely related individual, increases. Not surprisingly, bromeliad ¯ owers and in¯ orescences, more than berries and infructescences, display familiar dispersal syndromes. Then again, a closer look at Bromelioideae may reduce this disparity. Frugivorous birds favor certain colors and food item sizes, and select among available fruits to meet species-speci® c dietary requirements (e.g., Levey 1987; Martinez del Rio et al. 1988; Stiles 1993). Some species seek primarily lipid-rich meals, while others prefer items much richer in carbohydrates (e.g., thrushes vs. tanagers). Some avifauna possess remarkable capacities to detect differences in the concentrations of both nutrients (Martinez del Rio et al. 1988; Stiles 1993). Many of the mammals that disperse seeds locate food by smell, but they often lack the visual acuity of birds and so on. Furthermore, like pollinators, seed dispersers vary in kind Cambridge Books Online © Cambridge University Press, 2009 290 Reproduction and life history and local abundance through tropical America, increasing the advantage of plant specialization to use speci® c fauna. Quite likely much of the variety exhibited by the bromelioid fruit and infructescence parallels the needs and sensory capacities of these diverse frugivores. Quests for unifying principles in the 1960s and 1970s prompted the hypothesis that ornithochorous ¯ ora belong to two categories distinguished by fruit size and chemical composition and the characteristics of the birds that disperse their seeds. According to McKey (1975), a modestsized, but inordinately important, collection of tropical trees exempli® ed by many Lauraceae, Arecaceae and Burseraceae depends on large-bodied, obligate frugivores, while the balance of the bird-users rely on avifauna with broader diets. Bellbirds, hornbills and quetzals, among the other specialists serving the ® rst group of plants, exhibit narrow, plant-based diets that oblige meals of protein and oil-rich pulp. Bills are specialized to manipulate and discard rather than help swallow large seeds that consequently need no hard testa. Because dispersal is exceptionally reliable, these animals permit their food plants to produce relatively few, unusually wellprovisioned embryos (massive seeds). Plants comprising the much larger second group of tropical ornithochores produce smaller-seeded fruits that provide their dispersers with less expensive, leaner meals, mostly just carbohydrates, inorganic compounds and water. Protein and lipids come from other kinds of food. Hard envelopes, either a scleri® ed integument or an endocarp, protect the seeds of the generalist against often muscular guts and grinding crops. Differences in the numbers of seeds produced per plant, the cost of the individual fruit, and rate of seed success prompted McKey to consider members of the ® rst category more K-selected than the species constituting the less specialized second type. Small plant size, patchiness and frequent disturbance (substrates) in tree crowns (Benzing 1981b) favor adoption of the typically many-seeded, cheap fruits and generalist-type dispersers that serve epiphytic Bromelioideae. Super® cially these plants fall into McKey' s poorly resolved second category, which in reality probably includes multiple syndromes shared with other ¯ ora. For example, Snow and Snow (1971) identi® ed a mixed assemblage of epiphytes, including several bromeliads, utilizing the same avifauna in tropical America. Arboreal Anthurium species and one Ripsalis, all characterized by small berries, in addition to local Loranthaceae, comprise this guild in Trinidad. Less information exists for the aroids and cacti, but the mistletoes produce moderately large, naked Cambridge Books Online © Cambridge University Press, 2009 Seed dispersal 291 seeds, 1± 2 per fruit, enveloped in viscin that hastens passage through guts and helps secure the voided seed to bark. Certain Bromelioideae, particularly some Aechmea species and those Anthurium and cacti mentioned above, may parallel American Loranthaceae for features that favor safe passage through the honey creepers and tanagers most often responsible for their dissemination in tropical America. However, a second feature in addition to the presence of a hard seed coat fails to satisfy the mistletoe model; no bromelioid reportedly also produces the seed toxins that the branch parasites employ to distinguish predator from potential disperser. Rather than being endozoochoric like the mistletoes, some arboreal Bromelioideae may use birds and possibly other large fauna to disperse seeds without ® rst swallowing them as discussed below. Fruits and seeds of terrestrial Bromelioideae vary even more than those of the epiphytes, perhaps in some cases to attract fauna absent in the forest canopy. McKey' s bipartite scheme essentially consolidates Bromelioideae by dispersal syndrome, whereas the more recent dichotomy formulated by Levey (1987) imputes importance to another aspect of the fruits these plants produce. According to Levey, the ease with which seeds separate from the husk, the morphology of the berry, and aspects of the associated bracts suggest important facts about dispersers and matches between speci® c frugivores and bromeliads. Feeding behavior differentiates the avians that consume small-seeded, soft-walled fruits. The `gulpers' swallow food items intact, while the `mashers' only ingest the pulp, increasing the likelihood that any delicate seed appendage or enveloping mucilage present will remain intact to promote holdfast until anchoring roots develop (Fig. 3.6L). Fruit form suggests involvement of both types of dispersers for the bromeliads, while chemical composition is less informative. Fruit chemistry indicated no exceptional nutritional qualities among a sizable sample of Bromelioideae (Table 6.7). Reducing sugars predominate, ranging from ,80 to less than 15% on a dry weight basis. Lipids in pulp never exceeded 3%, nor did protein (not shown). Also important for the frugivore is the size of the potential meal, which, in addition to the mass of the individual berry, varies on a per plant basis with ripening sequence on the individual infructescence and in communities according to the density of fruiting ramets. A single specimen of Bromelia balansae or Aechmea bracteata left unvisited until its entire crop ripens could satiate several sizable dispersers. Another taxon equipped with exceptionally small fruits (e.g., Cambridge Books Online © Cambridge University Press, 2009 Table 6.7. Aspects of fruits and seeds of representative Bromelioideae. Most of the values for fruit chemistry are averages for two determinations of two samples, each of which was comprised of many ripe berries from one to several plants. Low seed counts (*) are probably due to poor pollination in the greenhouse Species Aechmea penduli¯ ora Aechmea tillandsioides Bromelia balansae Billbergia brasiliensis Neoregelia pascoaliana Neoregelia stolonifera Pseudananas sagenarius Quesnelia testudo Fruit color Fruit armed Bract/leaf color at fruit ripeness Fruit dry weight (g) Likely seed disperser Number of seeds per fruit Seed dry weight (mg) Soluble Seed carbohydrate Lipids appendaged (% dry weight) (% dry weight) Blue Yes Dull red bracts 0.34 Bird 113.5 0.56 Yes 38.9 3.5 Blue Yes Red bracts 0.37 Bird/ants 22.7 5.67 Yes 14.2 5.3 Orange No None 9.93 51.8 No 29.3 1.13 No None 2.95 Nonvolant mammal? Bat 2.5 Orange 109.4 11.6 No 62.1 1.36 White Yes None 0.22 Bird 238.6 0.81 No 80.4 3.1 White/ blue Yellow/ brown Yes None 0.22 Bird/ants 27 2.27 6.51 None nonvolant mammal Ð * 54.8 2.95 White/ purple Yes None 212.2 (multiple fruit) 0.24 Yes (both ends) No 36.7 No 25.5 2.85 Bird 2.0* Ð 1.25 Source: After K. Stiles et al. (unpublished). Cambridge Books Online © Cambridge University Press, 2009 Ð Seed dispersal 293 Figure 6.6. Fruit ripening phenology of co-occurring Bromelioideae in coastal habitats in São Paulo State, Brazil, illustrating degrees of overlap among species dependent on the same kinds of dispersers. (A) Bird-dispersed species. (B) Mammal-dispersed species. Heavy lines indicate periods of greatest fruit production (after Fischer and Araujo 1995). Araeococcus micranthus) may mitigate that disadvantage by occurring at high densities or attracting smaller fauna. Seed size, numbers per fruit, ratio of seed to pulp mass, and ripening schedule range widely among Bromelioideae (Figs. 3.6A± H, 6.6; Table 6.7). How these and other qualities favor the use of speci® c kinds of rooting media, or determine importance as food for particular frugivores, remains obscure. Among berries generally, those of Bromelioideae range from small (e.g., Aechmea, Neoregelia, Nidularium) to moderate-sized (e.g., Billbergia, various Bromelia). Groups of related species often exhibit similar fruits and Cambridge Books Online © Cambridge University Press, 2009 294 Reproduction and life history seeds. For example, members of Cryptanthus subgenus Cryptanthus (e.g., C. beuckeri) produce modest-sized berries even by subfamily standards, each containing just one to a few large seeds. Berries ripened by members of subgenus Hoplocryptanthus (e.g., C. pseudoscaposus) package many more, smaller seeds. Finer points of reproductive morphology conceivably also match bromeliads, dispersers and substrates, for example aspects of seed shape and surface texture respectively to certain kinds of rooting media or digestive systems. A ¯ at side probably helps secure the ovoid seeds of Aechmea bracteata (Fig. 3.6H) against gravity, and the same snug contact may promote imbibition from moist bark. Dejean and Olmsted (1997) report that seed shape accounts for the regular occurrence of this epiphyte in the crotches of rough-barked hosts in the seasonally inundated forests along Mexico' s northern Yucatán coast (Fig. 8.1B). Additional important detail distinguishes tissues around the seeds (Fig. 3.6I). Berry color ranges from drab (e.g., some Billbergia species), to orange (e.g., Bromelia), to white, lavender and blue (e.g., Aechmea, Neoregelia). Those sticky seed appendages (Fig. 3.6L) probably also enhance water balance, and by encouraging bill-wiping promote transfers of adhering seeds from feeding birds to bark. Most bromelioid fruits lack fragrances consistent with bird use; some of the exceptions emit pungent, pleasant odors (e.g., Ananas, Bromelia) or suggestions of rotting fruit (e.g., Billbergia zebrina). Seeds of Billbergia elegans germinated following passage through a bat to provide the only published documentation of endozoochory in Bromeliaceae (Abendroth 1957). Presentation also varies. Those pleasantly aromatic berries of Bromelia balansae nestle amid congested, armed, foliage well positioned to defy any large frugivore. Small, nonvolant mammals probably harvest most of the largely obscured, brownish, globose fruits presented by certain Greigia species. Foliose bracts mostly hide the dull orange, melon-scented fruits of an unidenti® ed Wittrockia. Additional, more subtle aspects of fruit display and release may enhance dispersal for certain Bromelioideae. Remarkably little pressure, probably well within capacity for a small frugivore, need be administered to separate the ripe Aechmea bracteata berry from its pedicle and extrude several of the sticky seeds (Fig. 3.6H). Spontaneous discharges following detachment (e.g., certain Billbergia) suggest the condition antecedent to ballistic dispersal. Foraging behaviors may help explain some otherwise puzzling reproductive morphology. For example, Abendroth (1965) noted that Tachyphonus coronatus checks for ripeness by pulling at the oversized calyx of the berries born on the capitulate infructescence of Neoregelia concen- Cambridge Books Online © Cambridge University Press, 2009 Seed dispersal 295 trica (Fig. 3.5B). Several seeds routinely remained in the discarded leathery husks. Pedicles sometimes (e.g., Neoregelia stolonifera) undergo intercalary growth to elevate the mature berry above the surface of the phytotelmata (Fig. 3.6F). A broader survey might reveal suites of plant characters speci® c to certain dispersers, or the types of animal behaviors needed to successfully manipulate one or another kind of fruits and seeds. For example, are the armed berries and ¯ oral bracts that seem so well disposed to discourage fruit gulping (Figs. 3.2D, 3.5H) associated with seeds that rely on digestible appendages for anchorage? Indeed, can the delicate attachments that characterize so many bromelioid seeds survive passage through the guts of frugivores? Why do seeds such as those of Billbergia lack equivalent embellishments (Fig. 3.6G)? Finally, which fauna are targeted by certain Aechmea subgenus Chevaliera species that produce berries with hard, sharp apices stubbornly embedded in equally resistant stem tissue? Myrmecochorus Bromeliaceae comprise part of a guild of convergent, ant-dependent taxa that includes representatives of about 10 families, but primarily Araceae, Gesneriaceae, Piperaceae, Moraceae and Orchidaceae (Davidson 1988; Benzing 1991). Occasional Tillandsioideae participate (e.g., T. fasciculata; Catling 1995), but too sporadically to warrant inclusion among the nest-garden ¯ ora. Bromelioideae and some other plants restricted to ant-provided substrates share methyl-6-methylsalicylate, benzothiazole and additional 6-substituted phenyl derivatives as volatile seed constituents (Davidson and Epstein 1989; Seidel et al. 1990). Other aromatics that also proved attractive in cafeteria-style tests with nest-gardening Camponotus femoratus, Crematogaster linata and Azteca spp. in Peru included ortho-vanillyl alcohol and limonene. Ant food, often an oily seed appendage, characterizes many of the carton-users, but none of the bromeliads. Pulp containing more concentrated lipids and proteins than necessary to attract bats or birds may encourage certain ants to mine seeds from the berries of the bromeliads they farm. Madison (1979) noted a ¯ eshy funiculus on seeds produced by ant-gardened Aechmea brevicollis, A. mertensii and A. angustifolia in Brazil. Aechmea tillandsioides, yet another nest specialist, produces the same carbohydrate-laden fruit recorded for Bromeliaceae without known relationships with ants (Table 6.7). Perhaps something else, neither food nor fragrance, prompts collection and incorporation into ant carton. Ule (1906) and Madison (1979) suggested pupal mimicry based on seed shape and size and possibly odor as the mechanism some bromeliads use to colonize one nest from another. Perhaps birds Cambridge Books Online © Cambridge University Press, 2009 296 Reproduction and life history remain the primary vectors of these bromeliads, consuming intact berries after which the ants recover and use the voided seeds much as bats and ants act in tandem to help disperse certain epiphytic ® gs. Lower vertebrates, and perhaps even some macroinvertebrates, disperse the occasional ¯ eshy-fruited bromeliad. Hyla truncata, a frog that sometimes consumes fruit almost exclusively according to gut contents, and terrestrial Neoregelia cruenta (Fig. 7.13E) reportedly mediate the establishment of a dominant shrub (Erythroxylon ovalifolia) in certain Brazilian restingas (Fialho 1990). Seeds defecated in moist leaf bases more often succeed than those deposited elsewhere. Another documented disperser of certain nonbromeliads (e.g., Pandanus species), the omnivorous land crab Gecarcinus lateralis, decimated the abundant fruits of Bromelia pinguin in a coastal dry forest located in central Vera Cruz, Mexico. These crustaceans were so intent on feeding that they risked predation by climbing the infructescences, but only fragments of integuments turned up in dropping. Nevertheless, Garcia-Franco et al. (1991) suspected that the rare seed that germinates may endure for hundreds of years via less vulnerable ramets. Table 6.8 summarizes characteristics of infructescences, fruits and seeds that distinguish ® ve putative seed dispersal syndromes in Bromelioideae. Note that the persistent, often stiff sharp calyx is assigned importance as a deterrent to the gulper (Fig. 3.5H). Conversely, its presence may help the masher detach berries from dense infructescences. Timing represents yet another variable, which is not shown in Table 6.8. Coordinated ¯ owering is not the only reproductive activity that simultaneously promotes a continuous supply of food for required fauna and relaxes the need for plants to compete for dispersal services. Fischer and Araujo (1995) reported evidence that shared seed dispersers in¯ uence phenology among Bromelioideae native to four lowland coastal habitats in southeastern Brazil, three covered by forest and the fourth situated along an open, rocky seashore. Subjects included members of Aechmea, Quesnelia and Nidularium, some in two, three or all of the surveyed communities (Table 6.9). They also examined local Tillandsioideae, all of which depend on wind currents to disperse. Unspeci® ed passeriforms and, according to feeding trials, two mammals ± Proechimys iheringi, a rodent, and Philander opossum, a marsupial ± carried seeds for the sampled Bromelioideae. All three Quesnelia species, the two populations representing Nidularium, and one of the ® ve local Aechmea species attracted mammals rather than birds to usually less colorful berries than those presented by the other four Cambridge Books Online © Cambridge University Press, 2009 Table 6.8. Putative seed dispersal syndromes of Bromelioideae Additional remarks Type Fruit color Fruit presentation Fruit size (length) Fruit fragrance Birds Bright colors: red, blue, white, purple Presented on elongate, spreading infructescence Small (,15 mm) No odor Fruits armed or unarmed perhaps depending on reliance on gulpers or mashers. Presence of seed appendage may re¯ ect the same dichotomy Bats Dull colors, sometimes densely covered with re¯ ective trichomes Presented on exposed, elongate infructescence Medium (15± 20 mm) Odor of rotten fruit Fruits unarmed, seeds unappendaged Dull colors Sometimes hidden by ¯ oral bracts on short, compact infructescence (e.g., Neoregelia) Small to large Odor present Plants terrestrial or low- growing epiphytes Bright colors as in bird-dispersed species Presented exposed (e.g., certain Aechmea species) or relatively hidden (e.g., Neoregelia) Small (,15 mm) Volatile, antattracting compounds associated with seeds Ants may disperse seeds from feces of larger animals Nonvolant mammals Ants Cambridge Books Online © Cambridge University Press, 2009 298 Reproduction and life history Table 6.9. Dispersal syndrome, shade-tolerance and location of substrates among 10 animal-dispersed Bromelioideae native to coastal Atlantic Forest habitats in São Paulo State, Brazil Species Habit Aechmea organensis Aechmea gamosepala Aechmea nudicaulis Aechmea pectinata Aechmea distichantha Quesnelia arvensis Quesnelia humilis Quesnelia testudo Nidularium antoineanum Nidularium innocentii Facultative Facultative Facultative Facultative Terrestrial Terrestrial Epiphyte Epiphyte Epiphyte Facultative Location of substrates Seed Shade- (average height above disperser tolerant ground in m) Birds Birds Birds ? ? Mammals Mammals Mammals Mammals Mammals No No No No No No Yes Yes Yes Yes 1.0 2.6 6.6 3.2 0.0 0.0 3.6 4.6 2.7 0.5 Source: After Fischer and Araujo (1995). bromeliads (Table 6.9). Characteristic short, foliose infructescences positioned close to the centers of leafy shoots further obscured the fruits of Nidularium. Fischer and Araujo mentioned no odors, but the marsupial may feed by olfaction and sight. Ornithochorous Aechmea presented more exposed and more colorful berries on inclined or horizontal in¯ orescences readily accessible to small perching birds. Only Aechmea pectinata of the mammal-dispersed taxa deviated from precedent by also bearing bright red fruits on a vertical infructescence. Light requirements and rooting media usually paralleled dispersal mode. All of the observed Tillandsioideae, eight Vriesea species and Tillandsia stricta, anchored in the canopy exclusively, six of the nine only in wellexposed microsites. Conversely, the surveyed Bromelioideae rooted in soil or were facultative epiphytes at proscribed elevations where forests occurred (Table 6.9). Distributions further indicated that four of the berry-producing subjects tolerated deeper shade and relied primarily on mammals to disperse progeny. Bird-sown Aechmea occurred most abundantly in the restinga formation and in the other, even more sparsely vegetated seashore communities. They also anchored higher in tall vegetation than relatives dependent on mammals. Just two taxa, Aechmea distichantha and Quesnelia testudo, combined shade-intolerance and dispersal by mammals. Phenology consistent with dispersal mode characterized all 19 subjects, but only the zoochoric bromeliads illustrated patterns reminiscent of the Cambridge Books Online © Cambridge University Press, 2009 Seed viability and germination 299 temporally structured guild Stiles (1978) reported for those bird-pollinated species in Costa Rica (Fig. 6.3). Each of the three bird-dispersed Aechmea species (A. gamosepala, A. organensis, A. nudicaulis) fruited over 2.5± 5.0month intervals, yet activity overlapped remarkably little anywhere (Fig. 6.6). Schedules, whether coincidental or modi® ed through selection to match local circumstances, assured nearly continuous supplies of berries for the dependent fauna. The four less consistently co-occurring mammal-disseminated species (Quesnelia testudo, Q. humilis, Nidularium antoineanum, N. innocentii) in the two tall forest habitats also provided more or less constant supplies of fruit with only modest phenologic redundancy (Fig. 6.6). Schedules for terrestrial Quesnelia arvensis and Aechmea distichantha overlapped much more at the rocky shore and restinga scrub locations. All nine Tillandsioideae shed seeds during the dry season, and even then rainmatted comas prevented many propagules from leaving the dehisced capsules. Site-speci® c timing characterized Nidularium innocentii, perhaps to heighten the attention of the local mammals needed to disperse its seeds (Fig. 6.6). Individuals that advertised edible fruit from March to May occurred exclusively in the restinga community, whereas plants in the same condition from December to early February invariably occupied either the riparian or dense forest sites. Were ripe fruit also a March to May phenomenon in the second and third communities, N. innocentii might have to compete with three other bromeliads for the same frugivores. A different pattern prevailed in restinga, where N. innocentii alone requires mammals and, apparently for unrelated reasons, fruits later in the year. Seed viability and germination Seed viability and germination rank among the least-studied aspects of bromeliad reproduction. Downs (1963) and colleagues tested diverse conditions and reported a variety of sometimes puzzling responses (Table 6.10). Longevity and temperature optima varied, and most subjects responded to light (were photoblastic). Seeds of Billbergia elegans remained viable for at least 72 weeks and about half of those of Neoregelia concentrica germinated 76 weeks after harvest. An unidenti® ed Puya continued to respond for 30 months (Vasak 1969), con® rming potential for seed banks, although no other evidence supports this possibility. One brief exposure to light induced Billbergia elegans to germinate, whereas a single, much longer pulse or several shorter ones over as many Cambridge Books Online © Cambridge University Press, 2009 Table 6.10. Effect of various frequencies and durations of light exposure on the germination of seeds of bromeliads Species Pitcairnia sp. Pitcairnia ¯ ammea Vriesea haematina Vriesea scalaris Nidularium fulgens Aechmea coelestis Puya berteroniana Percent germination Duration of experiment (days) Continuously dark 8 h light: 24 h 1 h light: 24 h 26 26 18 18 10 10 17 0 0 0 6 0 0 0 10 91 96 100 98 83 100 0 0 96 100 98 65 0 Source: After Downs (1963). Books Online © Cambridge University Press, 2009 1/4 h light: 12 h Ð 0 Ð 100 Ð Ð 100 1/12 h light: 2h 92 Ð 96 Ð 98 91 Ð Resource economics and life history 301 days sufficed for some relatives (e.g., Alcantarea regina, Puya berteroniana). A linear dose response leading to 70 and 100% success after ® ve days characterized Aechmea nudicaulis and Wittrockia superba. Downs' s (1963) list of 33 species representing all three subfamilies indicated that only Tillandsia stricta germinated about as well in light as in darkness. Five additional Tillandsioideae exhibited some germination in covered containers, but exposure usually increased yields. Experiments identify phytochrome as the pigment responsible for light sensitivity, but behavior varied with the subject. Exposure to far red light completely reversed the effects of red light for some species, while the seeds of others responded inconsistently. Temperatures prevailing in situ predicted behavior in the laboratory. For example, seeds of natives of lowland, tropical and subtropical regions (e.g., Aechmea nudicaulis, A. coelestis, Wittrockia superba, Vriesea scalaris) failed to germinate at 15 °C and responded best between 20 and 30 °C. Conversely, yields for Puya berteroniana, a high Andean terrestrial, began to decline as temperature rose above 15 °C, and only a few percent would germinate at 25 °C. Resource economics and life history Economists employ tools that can also be used to analyze the growth and body form of organisms in terms of evolution and ecology. According to this approach, plants resemble factories because they also acquire raw materials, in this case CO2, water, light and essential ions, to fabricate valueadded `products' , speci® cally progeny. Like any enterprise using the same materials that other factories require to manufacture the same products, cooccurring plants compete. Just how severely they interfere with one another depends in large part on the distinctness of the strategies used to obtain mutually required resources. Green plants vary in the ways they amass and process the resources necessary for photosynthesis, and then how the resulting photosynthate is allocated among different parts of the body to support growth and reproduction. More to the point, plants differ in ways that determine ® tness under speci® c growing conditions, including those attributable to the presence of other plants in the same habitats. Interferences among co-occurring species diminish as patterns of resource use diverge or become complementary in some way as populations assemble to establish communities during succession and over longer-term evolutionary time. Ultimately, organization and compatibility emerge that result from and bene® t plants with Cambridge Books Online © Cambridge University Press, 2009 302 Reproduction and life history distinct strategies growing together (e.g., a small, light-sensitive herb growing in the shade of a more heliophilic tree to cite a simple example). According to the economic/evolutionary interpretation of plant growth, structure and function, the vascular plant apportions photosynthate among roots, stems and leaves, and then reproduces within constraints peculiar to its natural history. That natural history or ecological strategy in turn allows the plant to accommodate speci® c growing conditions (kinds of habitats). For example, frequent disturbance mandates that the weedy annual invest an extraordinary proportion of its biomass in many small, long-lived, photoblastic seeds. On a coarser scale, ruderals also allocate more dry matter to shoots than roots. Longer-lived plants adapted to more stable substrates favor vegetative organs to achieve architectures more conducive to competition and resource conservation for extended life and repeated reproduction (e.g., most Bromeliaceae). Additional variety distinguishes the architectures of the herbaceous perennials, much of it dictated by the differential availability of resources in speci® c kinds of habitats. For example, compared with vegetation native to moist forest, desert plants invest more biomass in roots than in shoots because survival is challenged more by the acquisition of adequate moisture than by that of light, which is the more abundant of the two resources relative to plant needs. On a ® ner scale, photon-trapping and processing molecules receive biosynthetic priority over those responsible for the dark reactions of photosynthesis in shade compared with sun leaves. The developmental program does recognize that conditions vary in the same environments. While life-history type proscribes fundamental structure and function, opportunity for modest adjustment (phenoplasticity) remains. Although environment affects the expression of all developmental programs, plasticity varies with the ecological strategy. Plants with relatively short life cycles native to habitually heterogeneous habitats, and those types forced to accommodate often changing conditions, possess the greatest malleability. Features like root/shoot ratios shift dramatically among weeds depending on the prevailing supplies of photons, water and key nutrients ± much more so than for the slow-growing, stress-tolerant perennial. As longlived plants more or less con® ned to limited ecospace (unlike a vine), bromeliads, or at least the individual ramet, exhibit relatively modest plasticity. In effect, their architectures (dense rosettes of foliage incapable of sun tracking) represent time-averaged, relatively ® xed responses to past selection and current conditions. Reproductive allocation (RA), the ratio of phytobiomass or calories committed to reproductive vs. vegetative tissue by the individual plant, Cambridge Books Online © Cambridge University Press, 2009 Resource economics and life history 303 serves as a crude measure of ecological strategy, but multiple functions (e.g., green in¯ orescence bracts) and other complexities preclude precise cost-accounting. Iterative growth (iteroparity) causes additional problems. Much of the investment in the perennial in¯ orescence of Deuterocohnia schreiteri (Fig. 3.4J), for example, performs repeated duty by ripening successive crops of seeds over an equal number of years. Nevertheless, no other metric has been employed as often or can be so easily obtained to infer ecological strategy and related requirements for growth. According to the analogy between ¯ ora and industry, RA reveals how plants deploy photosynthate to maintain ® tness under speci® c growing conditions. Finer details of form (e.g., leaf anatomy) were accepted long ago as indicators of plant performances and conditions in habitats (e.g., water-use efficiency and water supply respectively); architectural analysis simply extends this practice to a higher level of plant organization. Bennett (1991) applied regression statistics to determine whether RA increases with epiphytism among the same bromeliads he used to compare seed mobility. Speci® cally, he determined whether the slope of the allocation regression would sort these species in the following sequence from low to high: T. sphaerocephala, T. ionochroma (saxicolous then epiphytic), T. fasciculata, and ® nally T. utriculata. Four of the ® ve populations yielded signi® cant RA to V (vegetative dry mass) regressions. Vegetative tissues accounted for 28± 79% of the variability in RA. Regression slopes for the iteroparous species ranged from 0.145 (T. fasciculata) to 0.213 (T. sphaerocephala), with no statistical differentiation. Tillandsia utriculata yielded a much higher value at 0.466. Every y-intercept deviated signi® cantly from zero, except for epiphytic T. ionochroma, which produced a value of 0.356. If immature ramets of the iteroparous taxa contribute resources to attached ramets as they ¯ ower, then the intercept may be positive. Because the generally more caulescent saxicoles also branch more frequently than the epiphytes, their intercepts should be greater, and those for the monocarpic populations of T. utriculata zero or negative. Benzing and Davidson (1979) determined how T. paucifolia allocates mineral nutrient capital for reproduction in Florida habitats distinguished by the numbers and sizes of this epiphyte present (Figs. 1.3C, 6.7). Previous examination of the same species (Benzing and Renfrow 1971a; Figs. 7.8, 7.9) had revealed correspondences between N, P and K concentrations in shoots and whole-plant mass at maturity and number of fruits. They also found that the most fecund individuals subsequently produced the smallest ramets, suggesting greater inherent emphasis on seed production than the Cambridge Books Online © Cambridge University Press, 2009 304 Reproduction and life history Figure 6.7. Ontogeny of Tillandsia paucifolia. A± I correspond to the nine size/age classes used for the demographic analysis described in the text. A± G represent pre¯ owering stages, H, adults fruiting for the ® rst time, and I, plants with ramets (after Benzing 1981a). Cambridge Books Online © Cambridge University Press, 2009 The organization of reproductive allocation 305 continuance of, or at least the robustness of, established genets. However, adults always reserve enough resources to support additional fruiting later, as Garcia-Franco and Rico-Gray (1988) noted for T. deppeana in central Mexico. The organization of reproductive allocation Inferences about ecological strategy based solely on RA lack the resolution possible with additional data on seed size, number and packaging. Variations in the ® rst two, if not all three, of these parameters correlate with other features of plants indicating adaptive value. But how tightly coupled are plant and environment at this level of detail? Certain bromeliads produce ovules in numerous gynoecia, while others ripen comparable crops of seeds in many fewer ovaries. Do these arrangements decisively in¯ uence reproductive success? Might hyperovulate gynoecia compensate for infrequent pollination as in many Orchidaceae (Benzing 1987a)? Conversely, could large numbers of gynoecia, each containing fewer ovules, spread the risk of predation? More ¯ owers and fruits can extend the time available to attract pollinators and disperse seeds, but the same higher numbers and longer displays also increase plant apparency for searching herbivores. Perhaps some more fundamental aspect of plant architecture, possibly the basic organization of the in¯ orescence (e.g., the short, dense head of most Neoregelia species), constrains fruit number. Members of Tillandsia vary enough to consider how seed packaging relates to life history among a group of closely related, but ecologically diverse, bromeliads. Those populations of Tillandsia ionochroma and T. utriculata observed by Bennett (1991) produced approximately the same number of seeds per infructescence (about 11 200 vs. 13300), but in 74 compared with 56 capsules. Tillandsia utriculata matured 237 seeds per fruit, while T. ionochroma averaged just 153. Are these distinct loadings adaptive, selectively neutral, or mandated by inherent design (phylogenetic) constraints? At least two hypotheses provide plausible explanations for the same condition. First, perhaps pollen supply historically limited fruit set for T. ionochroma, whereas shoot size has had the greatest in¯ uence on fecundity for self-pollinating T. utriculata. Alternatively, the sample year was an exceptionally good one for T. ionochroma and its seed crop larger than usual. In fact, relatively few T. utriculata specimens bore fruit, just 5.3% of the surveyed near-adults (close to or just short of ¯ owering size), and capsules per mature plant varied several fold. Demographics confuse the comparison even further; Tillandsia ionochroma outreproduced Cambridge Books Online © Cambridge University Press, 2009 306 Reproduction and life history T. utriculata on a population-wide basis because more of its shoots ¯ owered that year. Conceivably, breeding systems and associated patterns of resource allocation help explain why these two taxa package seeds differently. Material economy accrues from producing more ovules in fewer gynoecia, assuming no density-independent predation. If seed mass also remains constant (untrue among Tillandsia species), investments in capsule walls and ancillary non-green tissues that also support reproduction decrease on a per progeny basis as seeds per fruit rises. Savings may be especially signi® cant for the monocarp with its heavier dependence on seeds to maintain populations. Case histories Bennett (1991) cited aspects of branching, seeds and demography to support his contention that the durability and areal extent of rooting media have helped differentiate the life histories and architectures of certain Tillandsia species. Saxicoles compared with the examined epiphytes consistently exhibited lower fecundity and seed mobility. Rock-dwelling T. sphaerocephala and T. bi¯ ora bore more ramets but dispersed fewer seeds than the epiphytes (e.g., 2362 to 70 681 seeds per infructescence for T. sphaerocephala and T. utriculata respectively), and populations included fewer juveniles per adult. Post-¯ owering shoots of Tillandsia bi¯ ora averaged more than 20 ramets and T. sphaerocephala seven compared with fewer than two offshoots for the epiphytes and facultative T. ionochroma. Shoots that branched before ¯ owering suggested diminished apical dominance among the saxicoles. In all, six characteristics distinguished Bennett' s bark-users from his rock-users: higher ratios of seed weight to coma length, longer comas, more proli® c seed production, a larger in¯ orescence, greater overall plant mass, and stronger apical dominance (less branching). One of Bennett' s seven subjects demonstrated the effects of substrates on evolution more impressively than the other six. Tillandsia utriculata exceeds most members of its genus by number of intraspeci® c taxa, many of which appear to represent the products of ® ne-tuning to local growing conditions, especially the stability of rock vs. bark as a rooting medium. Plant architecture and habit (epiphyte vs. saxicole) shift together across Mesoamerica, more or less according to climate and available substrates. Recall that Florida populations examined by Benzing and Davidson (1979) and Bennett (1991) grow exclusively in tree crowns under relatively humid conditions, while those in Mexico often experience greater drought and fre- Cambridge Books Online © Cambridge University Press, 2009 The organization of reproductive allocation 307 quently anchor on rocks. Capacity to produce ramets and display bright colors to promote outcrossing also shift with location. Average fruit set among plants representing a series of Mexican populations observed by Gardner (1982), whether anchored on rocks or on bark, ranged from 16 to 50%. Florida specimens achieved greater success, perhaps because stamens contact the style as ¯ owers age, a feature not reported elsewhere for this species. North American subjects also lack the same bright pigmentation and routinely die without activating a single axillary bud, sometimes even after injury destroys an immature in¯ orescence. Relatives in Mexico vary on both counts by routine production of deeper pink ¯ oral bracts and frequent branching. Some other features of reproduction, including nocturnal anthesis and pale white to greenish, modestly spreading corollas, prevail everywhere. Tillandsia utriculata var. utriculata ranges through the same part of northeastern Mexico occupied by T. utriculata var. pringlei (formerly T. pringlei, but synonymized by Mez in 1896), where they differ in size, aspects of leaf, in¯ orescence and ¯ oral morphology, capacity to produce ramets, and substrates. Gardner (1982) considered the second variety a diminutive, relatively xeromorphic, epilithic derivative of the more broadly distributed, predominantly arboreal T. utriculata var. utriculata. A third, closely related and according to some authorities conspeci® c saxicole, T. karwinskyana, constitutes an even more stress-adapted lithophyte. Also Mexican, it differs from T. utriculata var. pringlei by still shorter stature (,60 cm vs. up to 1.25 m), an even more abbreviated in¯ orescence, a broader, more ornamented trichome shield, precocious ramets, and unvarying saxicoly. All three taxa hybridize freely, and in some combinations cross with additional, more distinctly differentiated taxa (e.g., T. makoyana; Gardner 1984). These three taxa suggest that rocky substrates encouraged lithophytic populations to diverge from epiphytic stock suited for more equable conditions. Outcomes varied, enough change occurring in some instances to justify varietal designations. Smaller stature and condensed (unbranched or sparingly branched vs. bi- to tripinnately compound) in¯ orescence within this complex may re¯ ect the more meager supplies of litter on rocky substrates compared with bark, and greater exposure to desiccation. Caulescence and abundant sympodial branching accords with greater opportunity for extended survival on rock compared with bark (Fig. 2.10M,N). Flowers and pollination biology remain relatively unchanged compared with shoot architecture and tendency to clone. Evidence of similar transitions from bark to rock, and perhaps back to epiphytism, exists elsewhere in the genus (e.g., T. fasciculata). Cambridge Books Online © Cambridge University Press, 2009 308 Reproduction and life history Saxicolous and arboreal habits and plant architecture distinguish closely allied genotypes elsewhere in Bromeliaceae. Saxicolous populations within some Quesnelia species also exhibit longer-stemmed shoots than the otherwise similar epiphytes. Moreover, in some instances (e.g., Q. testudo), plants ¯ ower less regularly if rock-dwelling rather than epiphytic. However, Bennett' s example and the Tillandsia utriculata complex illustrate just one of the two modi® cations for saxicoly among the bromeliads. Stable media also favor monocarpy and extraordinary large size as illustrated by Tillandsia grandis and several Alcantarea species (Figs. 1.2C, 7.1D). Demography Several accounts, in addition to those provided by Gardner (1982) and Bennett (1991), describe aspects of the demography of one or more populations of bromeliads, and some of these reports include data on recruitment and survivorship. Epiphytes predominate, with Tillandsia paucifolia heading the list (Figs. 1.3C, 6.7). Figure 6.8 illustrates statistics for colonies supported by Taxodium distichum in the Big Cypress National Preserve of southern Florida. Benzing (1978b, 1981a) also observed germination in culture and in situ and the fates of seedlings over four successive seasons. Hurricane Andrew provided an exceptional opportunity in 1993 to observe the impact of a major tropical storm on this same epiphyte and several of its relatives in the southern part of the same state. Recruitment Seeds of Tillandsia paucifolia arti® cially secured to the bark of trees in Florida performed much as naturally dispersed progeny do. Test patterns consisted of 12 groups of four seeds, comas intertwined, affixed with glue in two parallel rows (Fig. 6.2D). Rainfall soon effected the intimate contact with substrates that seeds need to germinate. A small subset of supports accounted for most of the year-old survivors, which totaled less than 4% of the more than 6000 seeds sown each spring between 1978 and 1981 (Table 6.11). Trees with few or no spontaneously occurring epiphytes failed to accommodate more than the occasional tested seed. Consistent success or failure on speci® c parts of the most suitable trees indicated that hospitality also varied on the best hosts, probably owing to ® ne-grained differences in exposure to light and precipitation. Taxodium distichum proved to be the most favorable tree of the tested species for T. paucifolia, consistent with its heavy use by most of the other Cambridge Books Online © Cambridge University Press, 2009 Demography 309 Figure 6.8. Size/age structure of colonies of Tillandsia paucifolia in the crowns of dwarfed cypress trees in south Florida censused over three consecutive years. See caption for Fig. 6.7 for descriptions of the nine size/age classes (after Benzing 1981a). Cambridge Books Online © Cambridge University Press, 2009 Table 6.11. The fate of 6000 seeds of Tillandsia paucifolia attached to the bark of diverse trees in south Florida after one year Phorophyte(s) Number of patterns exposed Number of patterns shaded Number of seeds tested Number of seedlings produced Sanibel Island live oak forest Quercus virginiana Forestiera segregata Bursera simaruba Myrsine guianensis 0 0 0 2 5 3 1 2 500 300 100 400 0 0 0 6 (on 2 exposed trunks) Sanibel Island mixed mangrove forest Avicennia germinans Conocarpus erecta Ficus aurea Rhizophora mangle 0 0 0 0 3 4 1 5 300 400 100 500 0 0 0 0 Avicennia germinans Bursera simaruba Conocarpus erecta Ficus aurea Rhizophora mangle 4 2 3 0 0 2 1 2 1 5 600 300 500 100 500 57 (on 2 exposed trunks) 0 11 (on 1 exposed trunk) 0 0 Bursera simaruba Forestiera segregata 1 4 0 0 100 400 0 1 Taxodium distichum 9 0 900 110 (on 8 trunks) Habitat Sanibel Island mixed mangrove/ shell mound community Sanibel Island coastal strand shrub community Dwarfed cypress forest on mainland Source: After Benzing (1978b). Cambridge Books Online © Cambridge University Press, 2009 311 Demography Table 6.12. Percent germination of Tillandsia paucifolia seeds after 14 weeks under various misting regimens while attached to cut limbs of four supports. Each sample group comprised 100 seeds Misting regimen (30-min application) Support One/day One/2 days One/4 days One/6 days Bursera simaruba Conocarpus erecta Rhizophora mangle Taxodium distichum 32 33 22 23 35 17 24 21 25 21 26 18 7 6 20 9 Source: After Benzing (1990). local bromeliads and epiphytic orchids (Table 6.11). Conversely, Ficus aurea failed to nurture even one of the hundreds of seeds glued to its smooth, stable bark over the four-year survey, perhaps for the same reason that extracts of this tree inhibited germination of the orchid Encyclia tampensis in another study (Frei and Dodson 1972). Bursera simaruba, at best an occasional substrate for Florida Bromeliaceae, and then only in cracks and knotholes, regularly shed bark in small fragments, often with test subjects attached (Fig. 7.7F). Both Bursera and Ficus retain considerable foliage most winters in Florida, rendering their crowns darker and therefore even less suitable for heliophilic Tillandsia paucifolia than those of fully deciduous cypress. Controls affixed to cedar lathe and maintained under a daily greenhouse mist regimen germinated at .90% during each of the four years. Tillandsia paucifolia seeds performed much as they had in situ while attached to 6± 9 cm30.5 m sections of limbs following the technique utilized for securement to trees in Florida. Examined supports included Taxodium, two occasional phorophytes (Rhizophora mangle and Conocarpus erecta) and Bursera simaruba. Timers activated a misting system for 30 min every 1, 2, 4 and 6 days, after which bark surfaces bearing seeds air-dried within 3± 4 h and even sooner on sunny days. Between 6 and 35% of the 100 seeds representing each treatment germinated within 14 days (Table 6.12). Except for those on Rhizophora, which experienced fairly consistent success, seeds performed best under the three wettest regimens. Growth following germination also measures performance, and dryness depressed epiphyte vigor on all four of the tested phorophytes. Subjects irrigated just once every sixth day grew to only 10± 20% the size of those Cambridge Books Online © Cambridge University Press, 2009 312 Reproduction and life history Figure 6.9. Survivorship among four cohorts of Tillandsia paucifolia seedlings that had germinated while glued to the bark of Taxodium distichum in south Florida (after Benzing 1990). individuals subjected to the two wettest treatments. Seedlings misted every fourth day grew somewhat faster, but only at about half the rate exhibited by subjects moistened each or every other day. Persistent seed coats and intertwining coma hairs precluded more precise quanti® cation of seedling size (Fig. 6.5G). Survivorship Two techniques served to document the life history of T. paucifolia in Florida: continued surveillance of the year-old survivors of the seed germination exercise, and censuses of naturally established colonies occupying other Taxodium distichum specimens in the same region (Benzing 1978b, 1981a). Results from one site in the Big Cypress National Preserve exemplify those for the entire sample (Fig. 6.9). Each spring from 1978 through 1981, 480 seeds had been sown on the trunks or large limbs of the same 10 trees. Revisits the following year revealed survivors in every cohort, although 1979 brought the greatest success (Fig. 6.5E). Here, as at all the other sites that witnessed some germinations, survival increased dramatically following heavy mortality during the ® rst few years of life. Dried remains de® ed determinations of causes, which were probably drought, frost or pathogens. Other seedlings simply vanished. No plants ¯ owered during the experiment, although several eight-year-olds appeared close to Cambridge Books Online © Cambridge University Press, 2009 Demography 313 maturity on the last visit. Uneven growth, presumably related to site quality and competition among survivors in the same test patch, caused the sizes of equal-aged plants to differ several fold. Tillandsia paucifolia anchored on randomly felled, 85± 200-year-old dwarfed cypress trees in a single, mixed Pinus/Taxodium forest provided the life table data illustrated in Fig. 6.8. Values recorded for each year represent plants distributed among 10± 15 cypress crowns harvested during each of three consecutive winters. Mean numbers of residents per tree differed among years in part because epiphyte density varied systematically across the sampled forest. However, apportionments among the age/size classes remained fairly constant. Phorophyte age determined by growth rings failed to predict the demographic structures of the resident colonies of bromeliads. Categories A (de® nite ® rst-year seedlings) and H± I (adults), more agediverse groups, usually contained fewer individuals than those between A and H (Fig. 6.8). When numbers exceeded 10 per phorophyte (Table 6.13; tree number 6), most young of the year clustered within a meter or so of a putative maternal parent ± a plant that had fruited the previous season. However, the presence of a plant (or plants) with a spent infructescence in a crown one winter did not assure the occurrence of yearlings there the next. During the 1979/80 and 1980/81 seasons, 99 of 116 new recruits on 20 trees possibly originated from one or more post-fruiting adults sharing the same support. Those 17 others must have been fugitives, up to ® ve on a single tree, that arrived from parents in other crowns. Two trees that harbored fruiting individuals in 1978 or 1979 supported no one-year-old seedlings the next season, whereas other trees with the same history bore 1± 36 such juveniles. Plant structure that indicates how many fruits had been produced on an infructescence deteriorates too quickly to estimate the size of a seed source, which in the area studied could be just one capsule per plant to about 20. The largest specimens harvested for the survey were ripening up to 12 capsules, with less than three as the average. During 1981, the only year that seeds were counted, 11 fruiting specimens yielded 15 capsules containing in all about 1500 seeds. Category B juveniles, second-season young perhaps augmented by the occasional slow-growing third-year seedling, remained aggregated and numerically diminished compared with those of category A. Little clustering remained among specimens large enough to qualify for category C. Here, larger numbers and more uniform occurrences among hosting crowns re¯ ected additional convergence by plant size involving parts of four to ® ve, perhaps even more, successive cohorts. Reduced numbers of Cambridge Books Online © Cambridge University Press, 2009 Table 6.13. Age structure of colonies of Tillandsia paucifolia on 10 dwarfed Taxodium distichum trees determined in January 1980 Age/size category (mm) A 0± 3 (young of the year) B 4± 5 C 6± 10 D 11± 15 E 16± 20 F 21± 30 G 31± 50 H 501 Nonfruiting adults Fruiting adultsb Total juveniles Total adults Total epiphytes Host age (years) Host number 1 2 3 3 16 29 10 7 7 9 1 1 2 82 3 85 Ð 4a 32 25 14 3 3 3 3 3 2 87 5 92 75 2 5 10 4 5 5 7 1 0 2 39 2 41 49 4 0a 18 43 12 4 7 7 4 1 1 92 2 97 117 5 6 0 2 21 8 3 9 5 0 0 1 48 1 49 Ð 36a 3 36 3 2 6 7 4 4 0 94 4 98 67 7 8 9 10 Mean6standard error 1 18 15 12 5 9 4 4 4 0 66 4 70 131 5 7 35 21 7 10 13 2 2 4 102 6 108 65 5 8 21 34 22 18 15 0 0 2 125 2 127 140 0 72 43 13 7 8 3 0 0 1 149 1 150 160 5.66 3.4 18.1 6 6.71 27.8 6 3.61 13.1 6 2.81 6.5 6 1.8 8.2 6 1.3 7.3 6 1.3 2.16 0.4 1.560.5 1.560.4 83.46 12.6 3.0 6 0.5 91.76 10.5 Source: After Benzing (1981a). Notes: aTree bore one or more fruiting epiphytes the year previous to census. bIncludes both seedling adults and asexual adults in fruit. Cambridge Books Online © Cambridge University Press, 2009 Demography 315 individuals overall, and additional dispersion through categories above C, re¯ ected age-related, accelerated growth in addition to continuing attrition. Category C' s numerical superiority could, of course, signal extraordinary reproductive success between 1970 and 1975. Trees examined in the winters of 1979/80 and 1980/81 supported on average 2.8 adults, including 1.3 in fruit. Fourteen of the 26 specimens bearing capsules showed no evidence of prior reproduction. The remaining 12 bore vestiges of spent ramets suggesting that mortality, while highest in young plants, continues to be substantial thereafter. Because about one out of every four fruiting subjects had not reproduced before, life expectancy following maturation appeared to average about three more seasons. Bennett (1986a) reported 4± 16% attrition, often caused by detachment, among individuals comprising populations of Guzmania monostachia and three Catopsis species in Taxodium/hardwood forest in south Florida. Losses for Catopsis berteroniana approached 30% for the full year, enough theoretically to require complete replacement in three to four times that interval should all the individuals be actuarial equivalents. In fact, most of the casualties had not fruited. Clearly, some exceptionally long-lived fraction of the local colony constituted the effective population, or parents in adjacent habitats were supplementing the local seed supply. Hietz (1997) conducted a two-year study of site and age-related mortality among Catopsis and Tillandsia populations in humid montane forest in Mexico. His ® ndings suggest that certain aspects of substrates in¯ uence the probability of survival to reproduction. Consistent with ® ndings for the other Bromeliaceae studied, survivorship increases with plant size. Individuals less than 2 cm in length experienced 0.33 survivorship, while that number for plants longer than 15 cm had diminished to 0.06. Mortality caused by factors other than branch fall, the leading cause of death, increased with branch diameter, possibly re¯ ecting reduced access to light toward the center of occupied crowns. Perhaps growth to maturity requires that seeds germinate on twigs sturdy enough to support the resulting adult, but not so thick as to virtually assure local environments that kill attached bromeliads before they can reproduce. Although slow-growing, Tillandsia paucifolia colonizes dwarfed cypress fairly rapidly, after which populations mostly accrete from adults anchored in the same crowns. Upward limits on plant densities may exist, but if so their nature remains obscure. Conceivably, most of the unoccupied substrate that outwardly seems comparable to utilized bark in the crown of a dwarfed cypress tree in south Florida is in fact hostile as the germination Books Online © Cambridge University Press, 2009 316 Reproduction and life history tests suggested. Or perhaps disturbance, periodic freezes and ® res in addition to modest reproductive power hold numbers well below carrying capacities, as Benzing (1981b) suggested by imputing operation of a lottery-type mechanism. Findings on T. paucifolia in Florida underscore why so few additional arboreal plants share the canopies of dry, tropical American forests with Type Five Bromeliaceae. Most members of this family, like vascular epiphytes in general, simply lack the necessary stress-tolerance to survive even on the most accommodating substrates there. Colonists of these sites obviously possess exceptional hardiness, but drought still probably kills more seedlings than any other agency, with resistance increasing markedly with body mass. Conversely, disturbance exacts a more random toll because substrates of all ages and conditions, including those near failure, intercept seeds and then end the lives of all the epiphytes they eventually carry to the ground. Tillandsia paucifolia further illustrates what may be exceptional behavior for a perennial xerophyte. According to data collected during the 1978± 81 seasons, rates of establishment are modest, but consistent compared with some similarly slow-growing desert ¯ ora. For example, certain Agave species recruit seedlings less than one year in 10 (e.g., Turner et al. 1969), as do some cacti. However, once a juvenile reaches threshold size, a feat contingent on close proximity to a nurse plant, life usually continues to maturity. Other bromeliads (e.g., Vriesea neoglutinosa and Neoregelia cruenta in certain Brazilian restingas) may more closely parallel these terrestrials (McWilliams 1974), as does Puya dasylirioides and Andean P. clava-herculis (see below). Grubb et al. (1963) observed that 70± 80% of the arboreal bromeliads in humid montane sites in Ecuador were immature, while percentages in much drier forest fell well below these values. McWilliams (1974) noted similar disparities along moisture gradients in southeastern Brazil. Results from Augspurger' s (1985) multiyear survey of Puya dasylirioides and the information on Tillandsia paucifolia provide opportunity to compare the consequences of polycarpy for a terrestrial vs. an epiphytic bromeliad. More signi® cantly, these two species illustrate certain variations on that life cycle speci® c to widely different circumstances, and, presumably, associated constraints on plant success. Rather than cypress tree crowns in seasonal, subtropical forest, Puya dasylirioides inhabits sphagnum bogs, and, less frequently, rocky slopes at 2100± 3000 m in Costa Rica' s highest mountains of the Talamanca range. Although both species produce axillary offshoots from determinant ramets and incur greatest mortality Cambridge Books Online © Cambridge University Press, 2009 Demography 317 Figure 6.10. Numbers of ¯ owering individuals within a single population of Puya dasylirioides in Costa Rica over eight consecutive years (after Augspurger 1985). during the ® rst year of life, they differ in important aspects of recruitment and survivorship and the reproductive performances of genets. Unlike the 7± 10 years a Tillandsia paucifolia specimen needs to mature, seedling shoots of P. dasylirioides ¯ ower on average after 36 years, and do so with near certainty once a plant survives through the 12 seasons required to achieve or exceed the minimum shoot diameter of about 11 cm (Fig. 6.11). Populations also behave differently. Rather than the relatively steady ¯ owering exhibited by colonies of Tillandsia paucifolia in Florida, Puya dasylirioides ¯ owered less regularly, 38± 323 plants per year, each specimen bearing 50± 1224 capsules (Fig. 6.10). Rosettes that became sexual over a range of sizes (rosettes 20± 70 cm wide), and seed crops that increased exponentially with that metric, largely accounted for the uneven fecundity among fruiting individuals (Fig. 6.11). Most telling of the features that distinguish the reproductive behavior of these two bromeliads is the timing of branching relative to ¯ owering. Contrary to the pattern seen in Tillandsia paucifolia and most of the other bromeliads native to the forest canopy (Fig. 6.7), Puya dasylirioides produces ramets earlier, before rosettes reach 10 cm, about one-half the minimum diameter required to bolt. However, many individuals, especially those on rocky substrates, either fail to branch, or overtopped ramets die Cambridge Books Online © Cambridge University Press, 2009 318 Reproduction and life history Figure 6.11. Probability of survival of Puya dasylirioides plants during a four-year period relative to rosette diameter at the beginning of that interval. Note that mortality approaches zero after shoot diameter reaches 11± 20 cm. Black bars indicate plants that were already large enough to ¯ ower when the study was initiated (after Augspurger 1985). rendering the genet de facto monocarpic. On ® rst consideration, exceptionally high juvenile mortality and far less vulnerable adults suggest that polycarpy rather than monocarpy constitutes the superior reproductive strategy for this species. However, some additional plant characteristics that in¯ uence seed production also need to be considered. Augspurger speculated that P. dasylirioides combines certain advantages of both mechanisms to match local growing conditions. Its long-lived ramets produce the massive seed crops that only an extended life cycle permits on the infertile, often sodden, cold substrates these plants utilize. By also branching precociously, the average genet is more likely to produce the requisite large seed crop than could a more typically polycarpic bromeliad constrained by the same harsh growing conditions. For example, if Tillandsia paucifolia were to operate under the same environmental con- Cambridge Books Online © Cambridge University Press, 2009 Demography 319 straints, fecundity would fall because its renewal meristems remain dormant until the parent module bolts. Ramets produced by Puya dasylirioides provide different service that requires earlier appearance; they act as replacements for failed meristems rather than devices to permit the successful genet to produce repeated modest crops of seeds over successive seasons. After reaching threshold size, surviving shoots eliminate the weaker ramets, which at this point no longer serve a purpose anyway. One interesting question remains unanswered. Ramet production appeared to be site-speci® c. Perhaps this tendency indicates polymorphism or plasticity to match growing conditions, particularly substrates, that impose different rates of mortality on this bromeliad across its heterogeneous montane habitats. Precocious branching is the appropriate response where seedling shoots often die before becoming too large to allow ¯ ushed axillary buds opportunity to survive long enough to ¯ ower instead. The same logic offered to explain the reproductive biology of P. dasylirioides may explain the early production of tiny offshoots by predominantly monocarpic Bromeliaceae native to rocky substrates (e.g., lithophytic Alcantarea regina, Tillandsia rauhii, T. clavigrea, T. krukoffiana; Figs. 1.2C, 2.11B). Catastrophic mortality Certain members of Cryptanthus, Dyckia and Encholirium native to the `campos rupestres' of southeastern Brazil, Ayensua and Brocchinia melanacra of similar habitats in the Guayanan highlands, Hechtia in Mexico, and some Andean Puya, among others, resist ® re with thick mantles of persistent, insulating leaf bases (Figs. 2.2G, 6.12C± E, 14.3B). Termite carton regularly provides additional shielding in some regions (Fig. 8.1E). None of the arboreal types possess similar accouterments, with devastating consequences in certain formerly less ® re-prone habitats. Today, for example, little evidence remains of Catopsis berteroniana and the four to ® ve Tillandsia species that densely colonized cypress forest in southwest Florida prior to the installation of a drainage grid designed to lower the water table. Plant collectors bear some responsibility, but careless smokers, deliberate arson and the US Army Corps of Engineers assured the ® nal, devastating outcome. Findings by Rocha et al. (1996) and Alves et al. (1996) in Brazil underscore the vulnerability, at least over the short term, of terrestrial Bromeliaceae native to generally ® re-free ecosystems. Only about 0.1% of the shoots of Vriesea neoglutinosa comprising dense colonies Cambridge Books Online © Cambridge University Press, 2009 320 Reproduction and life history Figure 6.12. Fire and Bromeliaceae. (A) Colony of Neoglaziovia variegata following ® re in Bahia State, Brazil. (B) Caatinga, a ® re-prone community that hosts bromeliads in Minas Gerais State, Brazil. (C) Unidenti® ed Encholirium sp. that survived ® re at the same site illustrated in B. (D) Hechtia schottii displaying how ® re burns away the dead leaves without killing the thick stem in Yucatán State, Mexico. (E) Unidenti® ed Encholirium sp. following ® re in Minas Gerais State, Brazil. (12493 ramets ha21) in a 210 ha patch of restinga surrounded by Atlantic Forest appeared to be alive following a burn set by local farmers. Fifteen months later, regeneration had elevated the number of viable shoots to 3483 ha21, about a 28% recovery (Alves et al. 1996). No mention was made of the tank occupants (arthropods) whose numbers in the same bromeliads scorched by ® re had been reduced about 67%. Many Aechmea bromeliifolia and A. phanerophlebia specimens are seriously injured or killed by ground ® res that may be more common today in certain campos rupestres sites (personal observation). Patchy distributions that locate plants away from the dead grasses and other ¯ ammable phytomass scattered through these thinly vegetated sites further suggest recurrent burns. Figure 6.5B illustrates how seedlings of Bromeliaceae and certain Cambridge Books Online © Cambridge University Press, 2009 Demography 321 other taxa take advantage of the insulation and perhaps protected supplies of moisture provided by rocks in Brazil' s campos rupestres. Bromeliads through much of Mesoamerica probably experience catastrophic storms more often than killing ® res. Between 1871 and 1964, 0± 11 hurricanes per year crossed the Caribbean (Walker 1991), sometimes in¯ icting damage visible for decades. Studies underway in Puerto Rico and elsewhere aim to document succession following several recent events, but plans target woody rather than epiphytic ¯ ora. Information for Bromeliaceae consists largely of an account of the fate of a small sample of T. paucifolia in the Everglades National Park plus observations on some relatives on nearby taller cypress, and populations elsewhere in Florida (Lowman and Linnerooth 1995). Craighead (1963) considered hurricanes an important threat to that state' s epiphytes after witnessing storms like Hurricane Donna in 1960 that killed an estimated 90% of the bromeliads at certain coastal sites. Conversely, T. usneoides may owe its unusually extensive range through Mesoamerica largely to the powerful winds generated by these massive tropical disturbances. Hurricane Andrew, a Class Four storm that crossed extreme south Florida in August 1992, in¯ icted considerable damage on local Bromeliaceae. One impacted colony of Tillandsia paucifolia contained ,75 mapped plants anchored on 28 dwarfed cypress trees located about 40 km inside the eastern border of the Everglades National Park. Each epiphyte was identi® ed by a metal tag either secured around a nearby branch or nailed to the trunk. All of these plants belonged to age/size class H (prefruiting adults) or fruiting adults and by then had experienced attrition from the initial sample of 100 individuals. The third annual visit in December 1992 to record mortality and reproduction revealed no uprooted hosts, but many broken limbs and smaller axes. Only 40 plants or their anchorages could be relocated. Of the 18 surviving bromeliads, eight had been fruiting at the time and of these only four continued to bear one or more ripening capsules (Fig. 6.13B). Larger (,8± 10 m) Taxodium distichum specimens comprising a cypress ª domeº community on deeper, organic soil near the site just described sustained greater damage than the more dispersed, better-anchored trees bearing the tagged epiphytes. Many trunks stood askew, but still supported viable Tillandsia fasciculata, T. balbisiana and T.3smalliana (Fig. 6.13A). Here as well, stiff foliage remained largely intact despite winds that exceeded 240 km h21 at a coastal monitoring site some tens of kilometers to the east. Shoots of about 10, thinner-leafed, phytotelm Catopsis berteroniana comprising a colony anchored on tall shrubs approximately 15 km Cambridge Books Online © Cambridge University Press, 2009 322 Reproduction and life history Figure 6.13. Hurricane damage in south Florida. (A) Damage caused by Hurricane Andrew in August 1993. Note that many clumps of Tillandsia fasciculata remain intact in the crowns of still upright Taxodium distichum. (B) Damage, including aborted capsules, incurred by a tagged Tillandsia paucifolia specimen still anchored on a dwarfed Taxodium distichum tree near the cypress `head' illustrated in A. east of the cypress sites incurred extensive damage, and several moribund specimens dangled upside down on broken twigs. However, here too survivors can probably rebuild the population. Oberbauer et al. (1996) also assessed the damage in¯ icted on Florida Bromeliaceae by Hurricane Andrew. Demographic surveys of the epiphytes, including ® ve species of Tillandsia, residing in three cypress dome forests were conducted about 10 months after the storm. Several factors, including the type of substrate and the identity of the epiphyte, particularly its rooting characteristics and size, determined survival. Mortality also varied among domes, with the largest stands offering the greatest protection to resident arboreal ¯ ora. Tillandsia balbisiana and T. usneoides proved most vulnerable, but still lost relatively few members of their comparatively dense populations. Preference for smaller twigs in the ® rst instance, and the absence of a root system in the second, probably contributed to these outcomes. Post-storm densities (all ages) for the other three species (T. paucifolia, T. recurvata, T. utriculata) sometimes exceeded those recorded before Andrew' s arrival (different transects were used). Young plants, possibly encouraged by the additional light that penetrated the damaged canopy, supposedly accounted for the difference. However, failure to distinguish between newly germinated and surviving seedlings in the assessments of affected populations lessens opportunity to appraise Overbauer et al.' s results and conclusions. Loope et al. (1994) also surveyed vegetation along the path of Hurricane Andrew, and noted that the vascular epiphytes incurred greater mortality Cambridge Books Online © Cambridge University Press, 2009 Asexual reproduction 323 than plants of any other habit. Up to 90% of the population had succumbed at the most impacted sites, but even here enough stock remained for recovery. Surviving bromeliads appeared to be more susceptible to sun scalding than co-occurring Orchidaceae. Canopies opened by high wind will both exacerbate the effects of frost and favor those species that require high exposure (e.g., Catopsis berteroniana, many Tillandsia species). Mounting evidence suggests that while powerful hurricanes sometimes fell entire stands of trees, especially on ridge lines and along marine coasts, recovery begins immediately, mostly by trunk sprouts (Walker 1991). If representative, observations on several species in south Florida indicate a similar dynamic for epiphytic Bromeliaceae ± that populations mostly regenerate from survivors, largely obviating needs for immigrants or seed banks. Asexual reproduction Although Bromeliaceae probably arose from ancestors equipped with the same modular design featured by most extant lineages (e.g., Figs. 2.2D, 2.3, 6.7), propensities to branch range from nil for the few monocarps to prodigious (e.g., Abromeitiella, some saxicolous Tillandsia; Fig. 2.20). Relative emphasis on asexual vs. sexual reproduction among the iteroparous types re¯ ects genetic program and environment. Certain Aechmea magdalenae populations may rarely ¯ ower because indigenous farmers selected exceptionally proli® c, subsexual genotypes to promote ® ber production. Excess shade suppresses sexual reproduction more than vegetative growth in relatively heliophilic genotypes like Catopsis berteroniana. Likewise, Bromelia humilis seldom ¯ owers, and most of its ramets abort if located in certain overexposed Venezuelan coastal habitats (Lee et al. 1989). Several saxicoles (e.g., Aechmea wittmackiana, Quesnelia testudo, Vriesea philippo-coburgii) remain predominantly asexual everywhere, perhaps for the same reasons imputed earlier as adaptive behavior for certain rock-dwelling Tillandsia. Patterns vary within species. Recall that Tillandsia utriculata is monocarpic in Florida, but iteroparous in Mexico. Tillandsia secunda produces offshoots on infructescences and from leaf axils in northern Ecuador, but farther south in the same country relies exclusively on seeds. Seedling shoots of all but the monocarps typically produce one or more ramets, usually after ¯ owering, each of which replaces itself in the same fashion and so on ad in® nitum (Figs. 2.3, 2.11, 6.14). Stands of aggressively vegetative species like Deuterocohnia haumanii cover hundreds of hectares of semiarid Argentinian habitat possibly with genets up to centuries old. Asexual activity imposes considerable structure on populations but, except Cambridge Books Online © Cambridge University Press, 2009 324 Reproduction and life history Figure 6.14. Five patterns of asexual reproduction (branching) among Bromeliaceae. (1) Monocarpic/monopodial; (2) precocious basal: `grass' offshoots only; (3) midregion axillary ramet(s) that develop following ¯ oral induction; (4) combined features of 2 and 3; (5) axillary ramet(s) in upper region only following ¯ oral induction; (6) midregion axillary offshoots develop as plantlets propagate from spent in¯ orescence. Books Online © Cambridge University Press, 2009 Asexual reproduction 325 for Aechmea magdalenae on Barro Colorado island and those several Mexican relatives also subjected to genetic analysis, without much documentation. Stolon length and number appear to be especially decisive (Fig. 2.11). Those of Pseudananas sagenarius grow several meters, while Abromeitiella illustrates the opposite arrangement required for the cushion habit so common in severe alpine habitats (Fig. 2.20). Phenotypic plasticity further in¯ uences the shapes and sizes of genets, but to what degree remains undetermined. Ramets characteristic of Bromeliaceae fall into three categories distinguished by size relative to the parent shoot, site of origin along the subtending axis, and the timing of development (Figs. 2.11, 6.14). Tiny, almost grass-like offshoots initiated long before the parent ¯ owers describe many Tillandsia and Vriesea species (Fig. 2.11B). Ramets whose appearance is delayed develop more robustly, usually around the time the in¯ orescence emerges (Fig. 2.11A). Adventive offshoots characterize a far smaller number of mostly self-incompatible taxa, speci® cally those capable of producing such plantlets in lieu of fruits (e.g., some populations of Tillandsia paucifolia and several Orthophytum species; Fig. 2.11A). Tillandsia species that proliferate from the in¯ orescence often scramble over the ground, and sometimes grow into low shrubs, as does facultatively epiphytic T. ¯ exuosa native to semiarid coastal strand habitats in Venezuela (Fig. 2.11A). Two varieties of Tillandsia latifolia dominate vast expanses of treeless Peruvian coastal desert according to a somewhat different arrangement. Tillandsia latifolia var. major forms relatively large rosettes, alone or in small clusters, scattered across loose sand. Proliferative in¯ orescences produced by its smaller relative, T. latifolia var. minor, bend forward under their own weight to produce successive rows of progeny oriented into the on-shore stream of life-sustaining, mist-laden sea air. Tillandsia paleacea marches up-wind in similar fashion, unassisted by the in¯ orescence, but leaving behind the same trails of lifeless, desiccated shoots. Ramets of many Cryptanthus species readily disarticulate where they constrict at the base. Those of C. acaulis detach during dry weather as shrinking tissues and recurving foliage create sufficient tension. Poorly developed root systems and somewhat spherical shape may combine to foster tumbleweed-like mobility that helps disperse the abscised offspring. Several members of Orthophytum (e.g., O. benzingii) may rely on axillary in¯ orescences that regularly terminate as plantlets to colonize patches of soil on typically rocky substrates like those illustrated in Fig. 7.1. Elongate, initially upright shoots tend to bend or twist downward reminiscent of stoloniferous Rubus and similar plants notable for spreading clonal growth. Cambridge Books Online © Cambridge University Press, 2009 326 Reproduction and life history Final comments Many aspects of bromeliad reproduction match important aspects of environments like the stability and resource richness of their diverse, often unusual substrates. Vigorous ramets and comparatively modest seed production often accompany occurrences on rocks and stable soils like those that support many of the saxicoles and Aechmea magdalenae respectively. The generally more ephemeral and widely dispersed anchorages of the epiphytes oblige greater plant mobility than ramets provide, so seed production predominates and genets remain relatively compact. Malthusian coefficients (r), viz. expressions of the mathematical function that describes unconstrained population growth, provide more precise indicators of the selective pressures that shape life history and plant architecture in situ. Values for r vary among populations; so does l (realized growth rate) among identical populations in different circumstances. Many plant qualities in¯ uence r, including seed size and reproductive allocation. Numerous additional factors affected by environment, such as the probability of fruit set and the fate of progeny, in turn establish l below r. Bromeliaceae vary substantially on many of the plant-based factors, one of the major exceptions being reproductive schedule, which for the vast majority of species is iteroparous. Another explanation beyond the descriptive one provided earlier to explain this bias reveals more precisely why polycarpy greatly exceeds the incidence of monocarpy through the family. If the members of two populations share the same inherent features that in¯ uence r except for time required to mature, the more precocious population compared with that slower to fruit will potentially grow faster even though individuals in the second case can become larger and invest more biomass in seeds (Cole 1954; Stearns 1976). Likewise, populations of monocarps potentially expand faster than those of otherwise identical plants programmed for sequential reproductive efforts because the polycarp must retain signi® cant reserves to continue the genet. Environment comes into play in these comparisons as well. Time is needed to produce seeds from basic resources, but the interval (juvenile stage) required to accomplish this transformation varies with the organism and its circumstances. Habits (e.g., roots present or absent), photosynthetic pathway as it affects Amax and water balance, and the availability of key resources, among other factors, also affect outcomes by in¯ uencing rates at which raw materials like CO2 and moisture are acquired and converted to phytomass and eventually offspring. Two species help illustrate why reproductive schedule is strongly biased toward iteroparity among the bromeliads. Florida Tillandsia utriculata rou- Books Online © Cambridge University Press, 2009 Final comments 327 tinely sets self-seed, and its phytotelm shoots collect enough moisture and litter to support the obligatory massive seed crops (.10000) to sustain populations as a monocarp. Polycarpic Tillandsia paucifolia co-occurs with T. utriculata, but subsists on a more modest resource base consistent with the greater stress-tolerance required for Type Five status. Similarly exhaustive fruiting would disadvantage T. paucifolia even if reproduction continued to occur in about one-half to one-third of the 15± 20 years T. utriculata requires to ¯ ower. Seed crops would remain modest, probably less than 1000 units (in up to 6± 8 capsules), and far below what is needed to maintain populations such as those monitored in Florida. Much more than 15± 20 years would be needed for Tillandsia paucifolia (no phytotelma) to marshal the reserves a tank bromeliad like T. utriculata (phytotelm present) commits to its single massive seed crop. Recall that attrition, while highest in the ® rst few years of life, continues after maturity. Dislodgements and other lethal events related to the relatively ephemeral nature of bark would probably doom just about any population of epiphytes whose members needed, as do some of the terrestrial monocarps, decades to mature (e.g., Puya dasylirioides). Schaffer and Gadgil' s (1975) model demonstrates the relative advantages of semilaparous (monocarpic) vs. iteroparous reproduction by otherwise identical bromeliads. Equation 6.1 expresses the rate lm at which a population of monocarpic (m) individuals ± here exempli® ed by T. utriculata ± multiplies: lm 5CBm. (6.1) The notation C enters the probability that the average seed will survive to reproduce; B is the mean number of seeds ripened annually by each population. For polycarpic T. paucifolia, the corresponding l is determined by Equation 6.2. lp 5CBp 1P. (6.2) Here, P represents the probability that an adult will survive from one reproductive season to the next to potentially achieve multiple seed crops. Again, B denotes the size of the population' s annual seed crop. Because seeds succeed with equal frequency in this comparison, C has the same value in both cases. Equating rate expressions 6.1 and 6.2 and dividing through by C produces Equation 6.3: P Bm 5Bp 1 . C Cambridge Books Online © Cambridge University Press, 2009 (6.3) 328 Reproduction and life history According to Schaffer and Gagdil, in order for the population of monocarps to multiply as fast as another assemblage representing an otherwise identical polycarp, Bm must exceed Bp. Performance in Florida indicates how much larger seed crops would have to become to accommodate such an adjustment by T. paucifolia. Few juveniles survive to fruiting there (C5 ,0.001, whereas P is probably .0.7). A mature specimen ripens between 100 and 300 seeds (1± 3 capsules) each reproductive season, whereas Bennett' s ® gure for T. utriculata tops 13000 for its single, suicidal, reproductive effort. Should circumstances change such that adult survivorship diminishes ± that is, if multiple fruiting threatens the viability of the population (P becomes smaller) more than the failure of juveniles ± the bene® ts of monocarpy, and hence the likelihood of its evolution, increase. Single bouts of seed production become sustainable only if P/C falls to what among Bromeliaceae must be an extraordinarily small value. Several plant characteristics, including seed weight ± about 25% higher for T. utriculata than for T. paucifolia, vitiate our hypothetical comparison of these two bromeliads. A better-provisioned embryo potentially improves survivorship, in this case granting T. utriculata corresponding greater facility (suitability) for monocarpy. Mobility may further compromise our model according to Bennett' s (1991) discovery that T. utriculata seeds possess unusually long comas. Evidence on hand points to the sizable resource base and related capacity to produce large crops of relatively expensive, mobile seeds as requisites for monocarpy in Tillandsia; superior ability to counter agencies that depress C, principally aridity, seems less likely to affect reproductive success as much. Tillandsia utriculata also exceeds T. paucifolia in shade-tolerance and probably range of acceptable substrates. Values for P and B shift most dramatically where extensive clonal growth mitigates the need for abundant, sexually produced offspring. Widely available, durable substrates encourage spreading, iterative growth most of all (e.g., wetland Typha and Spartina). Numerous small, vagile seeds characterize much of this ¯ ora, but they do little to help maintain the established population. Some Bromeliaceae ® t this pattern as illustrated by saxicolous vs. arboreal Tillandsia, particularly the strongly asexual, caulescent types. Certain Aechmea, Bromelia, Neoglaziovia, Neoregelia and Quesnelia of rocky, coastal and other expansive and stable habitats also exhibit life histories and architectures that most closely approach the cat-tail type, and probably for the same reasons. Cambridge Books Online © Cambridge University Press, 2009 7 Ecology Considerations of the relationships between Bromeliaceae and climate and substratum emerge repeatedly in the chapters devoted to plant structure, physiology and reproduction. Nevertheless, many aspects of ecology either go unmentioned or warrant greater attention in a monograph that claims adaptive radiation as its central theme. This chapter and the following one address this de® ciency by revisiting the diverse and often demanding growing conditions experienced by the bromeliads through tropical America. It also raises the less familiar issue of how hosting ecosystems owe many of their important attributes to the presence of these often keystone species. Chapter 8 highlights many of the most intimate associates of the bromeliads, namely their pathogens and predators, and especially the mutualists. Several facts in addition to our focus on evolution oblige the emphasis on ecology. First, dense populations of epiphytic Bromeliaceae and companion ¯ ora demonstrably in¯ uence the structure, economy and carrying capacity of many Neotropical forests. The terrestrials in turn sometimes constitute much of the understory, and where trees are scattered or absent they may dominate entire ecosystems. A number of saxicoles achieve near monoculture on precipitous outcrops (Figs. 1.2C, 7.1). Finally, tankforming and bulb-producing Bromeliaceae engage in bene® cial exchanges with a variety of nonpollinating and/or seed-dispersing invertebrates and higher animals. Services rendered to the extensive fauna that use these plants surely exceed in kind, if not also in abundance, those provided by members of just about any of the other families of herbs present in the same communities. We begin by considering how climates and substrates in¯ uence the distributions of Bromeliaceae in space and time. Readers should consult Chapters 4 and 5 for details about carbon, water and ion balance mechanisms (ecophysiology), because all of these phenomena mirror growing 329 Cambridge Books Online © Cambridge University Press, 2009 330 Ecology Figure 7.1. Bromeliaceae on rocky substrates. (A) Tillandsia recurvata growing as an epiphyte on a lithophytic shrub while nearby Tillandsia kurt-horstii is anchored just as exclusively on the surrounding rock in south Bahia State, Brazil. (B) Unidenti® ed Encholirium sp. featured in E dislodged to expose its extensive super® cial root system. (C) Tillandsia tenuifolia forming a characteristic debris-collecting colony on Cambridge Books Online © Cambridge University Press, 2009 Frost-tolerance 331 conditions in situ. Next, the bromeliads are treated as components of communities, with an examination of how they affect co-occurring vegetation and interact with neighboring fauna. Our review concludes with effects on integrative processes in ecosystems, primarily mineral cycling and energetics. Once again the epiphytes receive top billing, but only because less is known about the terrestrials. For the moment, cultivated Ananas comosus and related Bromelia humilis, plus a scattering of additional, soil-rooted species, must represent this more broadly varied of the two ecologically distinct groups of bromeliads. Frost-tolerance Bromeliaceae range higher into montane habitats ± to above 4000 m, and farther poleward ± than members of several other sizable, predominantly tropical families (e.g., Arecaceae). However, relatively few populations, just a handful including members of Tillandsia, Greigia, Chilean Ochagavia and closely related Fascicularia, Puya (Fig. 14.2C) and a scattering of other Pitcairnioideae, regularly experience subfreezing temperatures. Tolerance for more protracted frost at higher latitudes is even less common. More precisely, geographic ranges, plant morphology and mostly undescribed ecophysiology divide frost-tolerant Bromeliaceae into two categories. Several species of Puya, and similarly long-lived `giant rosette' herbs in several other families (e.g., Asteraceae, Lobeliaceae), comprise a convergent ¯ ora native to tropical alpine habitats around the globe (e.g., also Africa, Hawaii). Frost and intense UV-B-enriched radiation oblige congested, highly re¯ ective foliage often invested with a dense indumentum of woolly trichomes (Fig. 7.2). Sunrise ends the nightly cooling cycle, which if much extended would overwhelm the well-insulated but otherwise vulnerable shoot meristem. Small size increases risk to the extent that only the rare seedling survives beyond the ® rst year or two of life. Miller and Silander (1991) reported that virtually every specimen of large-bodied Ecuadorian Puya clava-herculis to reach a certain modest size eventually ¯ owers, an achievement that requires a nurse plant, usually a tussock grass rather than a shrub or high cushion or mat-forming type. Figure 7.1. (cont.) the side of a granite outcrop in Bahia State, Brazil. (D) Alcantarea sp. on a granite outcrop in Rio de Janeiro State, Brazil. (E) A near monoculture of a lithophytic Encholirium sp. in Bahia State, Brazil. (F) Orthophytum sp. on granite in northern Minas Gerais State, Brazil. (G) Seedlings of lithophytic Tillandsia araujei and an unidenti® ed Vriesea sp. illustrating the rough texture of this highly stable medium. Cambridge Books Online © Cambridge University Press, 2009 332 Ecology Figure 7.2. Trichome development and the relative condensations of the in¯ orescences of four Puya species along an elevational gradient in Ecuador. Puya aequatorialis experiences a mean annual temperature of 16.5 °C at 1980 m while P. clava-herculis at 3962 m experiences 6.6 °C (® gure assembled from data provided by Miller 1986). Litter provided by companion ¯ ora further promotes survival among Puya clava-herculis juveniles by insulating the soil enough to retard the formation of the ice crystals that injure delicate roots. Taller vegetation excludes this same bromeliad at lower elevations, although conditionally depending on ® re, which enhances its competitive performance. More sensitive Puya species respond less favorably to the same challenge. For example, few P. raimondii specimens survive to reproduce today owing to the increasingly common custom of setting a® re the massive skirts of spent foliage before seeds can ripen. None of the alpine bromeliads has received as much attention as some similarly stress-adapted herbs in several other families. Presumably the same growing conditions and matching plant tolerances account for the Cambridge Books Online © Cambridge University Press, 2009 Frost-tolerance 333 Figure 7.3. Coupling of the temperatures of ¯ owers with that of air for Puya hamata and Puya aequatorialis in early July after subjects representing the second species had been relocated to the paramo habitat of higher-altitude Puya hamata (after Miller 1986). high-elevation, tropical distributions of all of the giant rosette species. Information on the water relations of Espeletia (Asteraceae) and Puya clava-herculis appears in Chapter 4. Miller (1986) demonstrated the importance of the shape and bulk of the in¯ orescence and the investing trichomes for thermal regulation among four Puya species native to different elevations in Ecuador (Figs. 7.3, 7.4). Lowermost Puya aequatorialis var. aequatorialis (1900± 2100 m) produces a tall, spreading in¯ orescence bearing no pubescence, whereas P. hamata and P. clava-herculis (.3400 m) feature much more condensed and insulated arrangements (Fig. 7.2). Dense layers of woolly light brown to white scales cover the ¯ oral bracts, almost obscuring the ¯ ower buds and developing capsules. Puya aff. vestita, consistent with its mid-elevational (3200 m) range, exhibits in¯ orescence and trichome development between these two extremes. Costs to plants estimated by dividing the dry weight of the indumentum by that of the supporting ¯ oral bract ranged from 0 to 57.8%, indicating the likelihood of substantial plant bene® t for the most heavily endowed species. Measurements in situ demonstrated that the two Puya species with the thickest cylindrical spikes and densest pubescence maintained their ¯ ower buds up to 2.4 °C warmer than adjacent night air. Additionally, thermal equilibration after sundown took 4± 9 times longer for these high-altitude types compared with relatives from lower elevations (Figs. 7.3, 7.4). Daytime readings that also exceeded ambient indicated capacity for the Cambridge Books Online © Cambridge University Press, 2009 334 Ecology Figure 7.4. Exponential heat decay curves and time constants for ¯ owers of four Puya species. Time constants were determined where the heat decay curves crossed 1/e5T1 20.632 (T1 2T2). This is the time required for a 63% decrease in the total difference between organism (T1) and air temperature (T2). See Miller (1986) for additional details. rapid heat gain that probably favors pollen and ovule development. Differential success between capsules on illuminated vs. shaded sides of the same in¯ orescences (mean seed number 301.9 vs. 225.1 respectively) indicated close thermal tolerances and the importance of exposure for high-elevation P. clava-herculis. Bromeliaceae with extratropical distributions occur in both hemispheres, but unlike the giant rosette types, none differs architecturally from related taxa or their own populations native to warmer habitats. Hardiness has emerged repeatedly in different parts of the family and at widely scattered locations. Ochagavia extends poleward to 35.5° on the Chilean mainland and the San Fernandez islands where temperatures sometimes fall to several degrees below zero. However, Tillandsia usneoides exceeds all contenders by a substantial margin: northernmost populations reach the 38th parallel along the Atlantic coast of the USA, and even stronger marine in¯ uences allow a 6° deeper penetration into central Chile. Sensitivity to the same degrees of frost that some Puya and other bromeliads must tolerate for only a few hours at a time in tropical alpine habitats precludes success in all but the mildest parts of temperate zones. Still, Spanish moss occasionally freezes solid in its coldest North American habitats. Cambridge Books Online © Cambridge University Press, 2009 Frost-tolerance 335 Exceptional rather than routine events probably act most decisively to set the geographic limits of extratropical Bromeliaceae. Freezes sufficient to boost coffee prices are fairly common in southeastern Brazil where several large genera (e.g., Billbergia) reach farthest poleward. Few records describe performances in situ, but cultivated materials suggest major differences among species. For example, Wurthman (1984) reported that Vriesea (e.g., V. carinata, V. corcovadensis, V. guttata) tolerate frost better than members of the same genus from the deeper tropics in southern Mexico, Central America, and northern South America. A quick glance at the range of Vriesea carinata mapped in Fig. 7.5 suggests why. Distribution south through coastal Santa Catarina State takes this popular ornamental species into latitudes comparable to those in central Florida. Jenkins (1999) summarized the literature on frost-tolerance and cold sensitivity among Bromeliaceae, and concluded that neither altitude nor latitude in home ranges consistently predict plant responses in culture. Brazil' s `campos de altitude' (high montane grasslands) subject indigenous Bromeliaceae to substantial freezes during winter inland from the Atlantic coast far north of the southernmost penetration of the family. Fernseea itatiaiae on Pico Itatiania (2787 m in São Paulo State) tolerates at least 26 °C, but lacks the characteristic compact morphology of similarly stress-tolerant members of Puya. Plants at substantially lower elevations even farther north cool enough now and then to incur visible leaf burn (Leme and Marigo 1993). Occasional frosts damage the shoots of the large rosettes of Alcantarea spp. that grow fully exposed to heat loss on rocks near Teresopolis at elevations somewhat below 1000 m (Leme, personal communication). Comparable dynamics prevail at similar elevations hundreds of kilometers farther north in Minas Gerais and Espirito Santo states. Largely Bolivian Abromeitiella (5 Deuterocohnia) probably also tolerate potentially lethal temperatures, and together with the taxa just mentioned and others like them constitute a much larger collection of frost-tolerant bromeliad ¯ ora than present in North America (Fig. 7.6). However, more is known about the effects of low temperatures on Bromeliaceae of the southeastern United States. All of Florida' s native bromeliads experience at least occasional freezes, and some of the outcomes are recorded. Hall' s (1958) brief account describes how several Tillandsia species fared during an unusually cold spell during 12 and 13 December 1957. Minimum temperatures that night remained below zero for several hours across all but the southernmost tip of the state. Mortality varied according to the region, circumstances at Cambridge Books Online © Cambridge University Press, 2009 336 Ecology Figure 7.5. Distributions of four Brazilian Vriesea species reveal why some South American Bromeliaceae exhibit extraordinary cold-tolerance in cultivation. Certain Tillandsia species and members of Bromelioideae (e.g., Fascicularia) and Pitcairnioideae (e.g., Deuterocohnia) range even farther poleward as indicated on the map of the continent. speci® c sites, and the identity of the bromeliad. Again, geographic distributions often predicted outcomes. For example, Tillandsia usneoides and T. setacea near Jacksonville survived unscathed and T. recurvata incurred only minor damage fully exposed in nearby habitat. Populations with more pronounced tropical affinities (e.g., T. utriculata, T. fasciculata) often fared worse hundreds of kilometers south. However, results elsewhere were more capricious, for instance the death of an entire Books Online © Cambridge University Press, 2009 Frost-tolerance 337 Figure 7.6. Distribution of eight of Florida' s Bromeliaceae based on herbarium vouchers. Note that Tillandsia bartramii alone exhibits a predominantly northern occurrence in the state. Four of these species are restricted to the lowest three tiers of counties. Cambridge Books Online © Cambridge University Press, 2009 338 Ecology colony of T. setacea near Sebring, which is some 300 km below Jacksonville. Co-occurring species sometimes performed differently. Near Palm Beach where temperatures remained below freezing for several hours, T. flexuosa died to the last specimen, while a small population of T. polystachia, a natural hybrid (T.3smallii) with a comparable range in that state, survived largely intact. Delaney' s (1994) report from the Archbold Biological Station in Florida' s central pinelands describes how in 1983 one of the worst freezes on record killed almost all of the local bromeliads and orchids beyond scattered individuals located along permanent waterways. A similar event just six years later nearly eliminated the survivors, but also provided an opportunity to observe recovery from what must be normal, although not necessarily as frequent, setbacks for sensitive ¯ ora in central Florida. Today, the few remaining butter¯ y orchids (Encyclia tampensis) persist in refuges near deep impoundments that provide supplemental humidity and heat. At the same time, three formerly abundant bromeliads, Tillandsia setacea, T. simulata and T. variabilis, are rebuilding populations elsewhere in Highland County. However, recent experience suggests an uncertain future if current land use practice continues. Long-time residents blame forest clearing and extensive arti® cial drainage for what appear to be increasingly severe droughts and devastating winter kills. Seed dispersal can restore a contracted range, or foster expansion into new territory as conditions, including frost, permit. However, distinguishing an expanding from a shrinking population requires analysis. Occasional colonies of Tillandsia fasciculata in southern Georgia, hundreds of kilometers north of their nearest counterparts in Florida, pose this question. Are these outliers remnants of a formerly more contiguous population that existed during the warmer mid-Holocene, or infrequent progeny wafted northward by tropical storms? Allozymes might reveal evidence of parentage ± perhaps polymorphism suggesting broader genetic bases than one or a few founders per site could provide. Breeding systems, ¯ oral advertisements and pollinators provide additional opportunities for critical analysis. Aspects of climate other than frost probably also in¯ uence range boundaries, particularly for certain Tillandsioideae sensitive to high humidity. Moisture and temperature acting together help shape the distribution of one exceptionally mobile species. Edward McWilliams (personal communication) has gathered substantial evidence suggesting that the northern and eastern range limits of Tillandsia recurvata in Texas have shifted over the last several decades. According to experiments, rainfall during the Cambridge Books Online © Cambridge University Press, 2009 Distribution in forests 339 coldest months, through effects on survivorship in marginal habitats, seems to determine range most decisively. For example, chilling damaged moistened more than surface-dry plants during runs conducted in growth chambers. Tillandsia recurvata is the subject of another experiment farther east in even more humid territory on the campus of Louisiana State University at Baton Rouge. First reported there in 1971, it now infests 36 woody endemic and exotic species over about 40 ha of managed landscape (Holcomb 1995). Migration has been slow, with descendants located at most about 1.2 km from the site of the ® rst record, probably on or near the contaminated nursery stock that accounted for the introduction. Humidity seems to be slowing additional expansion at the moment, perhaps by inhibiting dispersal. Matted comas often persist on spent in¯ orescences, and many of the seeds that do escape from capsules move no further than the trees that support their maternal parents. Gilmartin (1973) demonstrated how cool temperatures favor certain Bromeliaceae by mitigating drought. More than 300 species occur in Ecuador, many as epiphytes at proscribed elevations in the Andes. Seventeen taxa distribute across both the eastern and western slopes, and do so according to growing conditions that oblige different elevations on the two exposures. Habitats receiving maximum precipitation occur between 800 and 1200 m on the eastern and 1200 and 1600 m on the western slopes of the Ecuadorian Andes. Most importantly, less than 5 cm of rain falls per month for at least four successive months on the Paci® c side, while no month is equally arid to the east. Consequently, trans-Andean species (e.g., Tillandsia cernua, T. confinis, T. emergens) almost always occur at lower elevations, often more than 1000 m lower, on the Amazonian than on the Paci® c slopes where the accompanying cooler temperatures diminish the evaporative power of dry-season air. Distribution in forests Host specificity The typical epiphytic bromeliad anchors on a number of kinds of trees like most arboreal ¯ ora. Even inanimate objects, including telephone wires, accommodate certain Type Five Tillandsioideae (Fig. 1.3A). Several of the smaller Catopsis species occasionally do the same, somehow maintaining water-® lled shoots upright. Those few Bromeliaceae reported to utilize just a single species of tree tend to occupy narrow geographic ranges, or if more Cambridge Books Online © Cambridge University Press, 2009 340 Ecology broadly distributed reside in low-diversity woodlands (e.g., Pinus/Quercus forests in Central America and Mexico). Figure 1.3B illustrates one of the exceptions, in this case Billbergia porteana rooted on the palm that provides most of its substrates through parts of Bahia State, Brazil. Elsewhere, for example in caatinga (Fig. 1.4B), the same plant roots on diverse, mostly broad-leafed trees. Valdivia (1977) recorded the distributions of 153 vascular epiphytes, including 33 bromeliads, on 45 woody taxa in east central Mexico. No bromeliad rooted exclusively on a single kind of tree, nor did the crowns of every support utilized by another local epiphyte also accommodate members of this family. Hospitality varied, with one tree species harboring up to 107 different epiphytes, including many bromeliads. Conversely, Acacia cornigera, a myrmecophyte aggressively defended by its resident populations of Pseudomyrmex sp. against other insects and encroaching vines, supported no arboreal ¯ ora. Perhaps the ants remove seedlings here much as certain Crematogaster species preen the Leptospermum specimens that support their nests in Malaysia (Weir and Kiew 1986). A reversed relationship, in which the ants quartered in orchids and bromeliads defend some Mexican phorophytes against leaf-harvesting ants and a folivorous beetle, is described in the next chapter. Garcia-Franco and Peters (1987) reported nonrandom use of available trees by six Tillandsia species in four habitats along an altitudinal gradient in Chiapas State, Mexico. Individuals aggregated on certain parts of the crowns (trunks and large, medium and smallest branches) of acceptable species, leaving the others largely unoccupied. Hietz and Hietz-Seifert (1995a) used nearest-neighbor analysis in Vera Cruz State, Mexico to demonstrate clustering by several bromeliads whose seeds often succeed close to parents. The same authors (1995b) noted that tree identity had `practically no in¯ uence' on which epiphytes (66 species of which 25 were bromeliads) grew in speci® c crowns among six sites arrayed between 720 and 2370 m. Fontoura (1995) questioned whether phylogenetic relationships among co-occurring trees paralleled usage by an assemblage of Atlantic Forest Bromeliaceae richer in Bromelioideae. She mapped every adult present through the crowns of 122 specimens representing 46 species in 20 families. Twenty, 10310 m plots that totaled 0.2 ha contained bromeliads belonging to ® ve genera. Only 17% of the surveyed supports (.2.5 cm diameter at breast height) hosted one or more of these epiphytes, and occurrences indeed paralleled family affiliations. Myrtaceous ¯ ora supported the largest number of species; local Bromeliaceae also overutilized Rubiaceae, Melastomataceae and Cambridge Books Online © Cambridge University Press, 2009 Distribution in forests 341 Monimiaceae according to the relative availability of member trees large enough to be included in the survey. Girth (age) usually failed to predict epiphyte loads, contrary to the observations made by Yeaton and Gladstone (1982) and Ibisch (1996) in Costa Rica and Bolivia respectively. A more evenly apportioned array of trees would have increased the statistical rigor of Fontoura' s inquiry, but no special effort was required to identify the most pronounced associations between woody and epiphytic ¯ ora anyway. For example, the lone Coussopoa microcarpa specimen (Moraceae) included in the quadrats supported bromeliads belonging to all ® ve genera represented at the site. Undetermined characteristics of the censused bromeliads promoted unevenness in another dimension. Local Billbergia species, and to a lesser extent those belonging to Quesnelia, proved relatively `selective' , whereas co-occurring tillandsias rooted too diffusely through the canopy to discern patterns if they existed. Billbergia species grew almost exclusively on Alchornea triplinervia. Surveyed Vriesea species ranged most widely among anchorages, and because they shared so many trees with the other taxa also contributed inordinately to the generally gregarious nature of the local bromeliad ¯ ora. Fontoura' s study also demonstrated how phylogenetic constraints affecting ecophysiology can in¯ uence how bromeliads partition shared habitat. Nidularium populations consistently grew in deep shade on the lower sections of tree trunks in the fashion characteristic of this genus through its relatively con® ned range in southeastern Brazil. Billbergia also preferred the understory, but pronounced heliophiles elsewhere (e.g., B. porteana; Fig. 1.3B) occupy drier, better-illuminated sites. Vriesea and Tillandsia species included in the survey mostly colonized the upper canopy, but relatives tolerate shade equal to that experienced by Nidularium through diverse habitats across tropical America. Brazilian Tillandsia and Vriesea belong to much larger clades that probably incorporate substantially more ecological variety than is present in either Billbergia or Nidularium. Three species of Catopsis and four more populations representing Tillandsia and Guzmania monostachia utilized a variety of trees irrespective of several characteristics that affect crown opacity in hardwood/Taxodium forest in south Florida (Bennett 1984, 1986a, 1987). Instead, the number of stems representing each support, often Fraxinus caroliniana, most consistently predicted epiphyte abundance. Stem thickness had no discernible effect on recruitment. Two additional phenomena came to light during Bennett' s investigation. Members of acceptable tree species often bore no Cambridge Books Online © Cambridge University Press, 2009 342 Ecology bromeliads, while those with colonists usually supported multiple individuals reminiscent of Fontoura' s conclusion about gregariousness in Brazil. Second, apparent and realized niches were distinguished for the ® rst time for bromeliads. Speci® cally, some of the subjects (e.g., Guzmania monostachia, Tillandsia fasciculata) owed their presence in the study plots to mass effect, i.e., they grow but rarely reproduce there. All three of these studies demonstrate that many bromeliads recruit anchorages according to a variety of often subtle, but probably widely shared, tree characteristics. Conversely, other Bromeliaceae, sometimes species with similar habits, partition anchorages that seem equally accessible to the seeds of all potential users. For example, Aechmea bracteata grows abundantly on several kinds of phorophytes in semievergreen forest in the Sian Ka' an Reserve in the Yucatán (Olmsted and Dejean 1987; Fig. 8.1B). Tillandsia balbisiana, a wider-ranging species that roots on more than a dozen tree and shrub species just in south Florida, occurred at even higher frequency at the surveyed Mexican site, but only in the crowns of Bucida spinosa, a locally common tree invariably free of Aechmea bracteata. Closely related Tillandsia dasyliriifolia shared several types of supports with Aechmea bracteata, and also densely infested local Bucida crowns. Dejean and Olmsted (1997) looked more closely at Aechmea bracteata in the inundated forests and adjacent woodlands of Sian Ka' an to determine why it colonizes some kinds of trees but not others. First, only 10.3% of the individuals representing just ® ve of the 13 species available to intercept seeds supported one or more individuals of this epiphyte. Utilization was even lower on hammocks and nearby semievergreen forests. Overall, only 27 (35%) of the woody species surveyed hosted A. bracteata, presumably because they lack certain characteristics, perhaps the necessary architecture and/or suitable bark. Of the 145 genets of A. bracteata encountered in the three types of forest, more than 70% rooted where a large branch forked or a sizable limb joined the trunk. Most of the rest of the sample perched on nearly horizontal branches or at the base of trunks. Dejean and Olmsted concluded that Bucida spinosa is well suited by its smooth, thin twigs to intercept comose seeds (Tillandsioideae), but not those produced by baccate Aechmea bracteata (Fig. 3.6H,J). However, failure to encounter even one specimen on 103 trees suggests additional, more decisive deterrents. Bromeliads and co-occurring epiphytes sometimes partition common habitats, as four species of Tillandsia, including T. balbisiana and T. dasyliriifolia, and Aechmea bracteata plus some Orchidaceae demonstrated at a second study site near Cancun, Mexico (Zimmerman and Olmsted 1992). Cambridge Books Online © Cambridge University Press, 2009 Distribution in forests 343 In all, 10 species of orchids, along with the bromeliads, exhibited the usual interchangeable use of the local woody ¯ ora by utilizing 15 of the 19 kinds of potential supports. Although these 15 tree species tended to be either poor or superior substrates for members of both families, the bromeliads (except Aechmea bracteata) mostly rooted on axes less than 5 cm in diameter. Conversely, the orchids congregated exclusively on stouter stock. Hietz (1997) encountered the same pattern involving several species each of Catopsis and Tillandsia vs. several orchid species in a more humid Mexican montane forest. Still another report (Rudolph et al. 1998) documented that Tillandsia complanata occupied branches of all ages and the orchid Epidendrum marsupiale only twigs 3 cm thick or less provided by the same phorophytes in an Ecuadorian rainforest. Decisive factors: bark quality and the nature of the relationship Tests have helped demonstrate how epiphytic Bromeliaceae fail to use every substrate within dispersal range. Benzing (1978b) glued the seeds of Tillandsia paucifolia to a variety of trees that support or exclude this species in south Florida (Fig. 6.2D). Germination was also tested in the greenhouse using sections of branches cut from four of the same phorophytes (Table 6.12). Results in both instances paralleled those determined for the same bromeliad in situ. Physical characteristics of the barks, particularly stability and wettability, appeared to be most in¯ uential. Contrary to ® ndings on the co-occurring orchid Encyclia tampensis (Frei and Dodson 1972), phytotoxicity (allelopathy) had no measurable effects on host suitability for Tillandsia paucifolia. Seeds germinated on all four kinds of tested trees if adequate misting was provided in the greenhouse. Exceptionally smooth bark that readily sheds seeds and precipitation probably explains why Ficus aurea remains almost epiphyte-free (axenous) in south Florida. Figures 6.2D and 7.7F illustrate why certain physical attributes of the barks of Bursera simaruba and Taxodium distichum help differentiate the ® rst species as an infrequent phorophyte and the second tree as one of the leading substrates for Bromeliaceae in this state. Few of the many factors and circumstances that affect epiphyte/tree pairings probably apply broadly enough to warrant generalizations. Parasitism and the need for a fungus to germinate may explain the speci® city reported for the exceptional mistletoe and orchid respectively (Benzing 1990). Conversely, the much reduced, primarily mechanical root systems that preclude comparably intimate interactions between tree and epiphyte more likely grant the relatively shoot-dependent bromeliads unusual latitude. Cambridge Books Online © Cambridge University Press, 2009 344 Ecology Figure 7.7. Bromeliads and substrates. (A) Tillandsia tenuifolia as an epiphyte in Rio de Janeiro State, Brazil. (B) Tillandsia tenuifolia on rock about 10 m from the plant shown in A. (C) Quercus virginiana in poor condition densely colonized by Tillandsia usneoides in central Florida. (F) Dwarfed Quercus virginiana supporting abundant Tillandsia recurvata in coastal strand habitat about 10 km north of Naples, south Florida. (E) Robust Quercus virginiana almost free of epiphytic Bromeliaceae in central Florida. (F) Exfoliating bark of Bursera simaruba with a single young Tillandsia recurvata plant attached. Cambridge Books Online © Cambridge University Press, 2009 Distribution in forests 345 Hietz and Hietz-Seifert (1995b) cited this possibility to explain why bromeliads exceed local orchids in the proportional use of Pinus in certain habitats in Vera Cruz State, Mexico. But whatever the underlying mechanisms, the hospitality of a speci® c substrate for a designated epiphyte is best viewed as conditional rather than the all-or-nothing phenomenon so often implied in the literature. Even the heavily infested phorophyte probably fails to support colonists on every surface exposed to dispersing seeds. Moreover, epiphyte loads shift among phorophytes of the same identity depending on location as if hospitality depends on site-speci® c variables, particularly humidity and light. Age-related factors also in¯ uence the utility of bark, as does the size of the seed supply. Rhizophora mangle in Florida supports no canopy ¯ ora where other kinds of trees do. Yet the same or related bromeliads thrive in its crowns elsewhere, for example at the other side of the Gulf of Mexico along the Yucatán peninsula (e.g., Tillandsia brachycaulos, T. streptophylla; Fig. 1.2H). Occupied vs. adjacent, barren microsites on the same phorophytes warrant closer scrutiny to determine whether Rhizophora and some other trees only sporadically host epiphytes because local seed rains seldom reveal the few potential anchorages that exist in what are largely uninhabitable crowns. Conceivably, only the areas represented by older or appropriately oriented bark, water courses or colonies of lichens or bryophytes provide acceptable seed beds. Bennett (1986a, 1987) reported that several Bromeliaceae concentrated on bark embellished with mosses in the Florida swamp mentioned above. Layers of humus and the nonvascular ¯ ora responsible for its production also seem to sustain the seedlings of several Bromelioideae in southeastern Brazil (Fig. 6.5A). Ordering among species in the same successional sequence may re¯ ect different requirements for unaltered vs. conditioned bark as described below. Similar dynamics may apply to certain saxicoles (Figs. 6.5F, 9.12). Once again, Tillandsia paucifolia helps explain epiphyte distribution, in this instance why a Type Five bromeliad occupies so little of the substrate present in the forests it inhabits in southern Florida. Seedlings resulting from the germination tests not only survived lengthy drought during the greenhouse runs just described, they actually required it. Within weeks more frequently irrigated subjects died, apparently from suffocation or overgrowth by microbes. Mist applied for half an hour once every 1± 2 days promoted the highest rates of germination and subsequent growth (Table 6.12), indicating that the regenerative niche for this species is quite arid. Another similarly drought-tolerant relative exhibited an even more Cambridge Books Online © Cambridge University Press, 2009 346 Ecology narrowly moisture-de® ned distribution in certain Mexican habitats than T. paucifolia does in Florida. Tillandsia recurvata colonizes but one host in parts of Baja California, Mexico, in this case Idras, and then only on western exposures in what is generally low-growing scrub forest and desert (Barry 1953). Heavy nightly fogs moving on-shore off the nearby Paci® c Ocean account for the nonrandom distributions of this trichome-dependent bromeliad. Little rain falls to encourage germination on the other exposures, essentially none during exceptionally dry years. Stature in addition to bark quality likely promote ® delity to this single support. Mature Idras extend well above the co-occurring shrubs, thus favoring the conversion of mist into a life-sustaining moisture supply. Surfaces especially well suited to intercept seeds and condense aerosols may further enhance host quality, a possibility ripe for testing with real and arti® cial substrates. Aspects of seed dispersal and the stems in the paths of wind-born propagules may exceed the hospitality of bark as primary determinants of where many Tillandsioideae occur. Speci® c crown shapes and certain contours of trunks, limbs and twigs probably favor or discourage seed impaction and retention (Table 6.6). Con® gurations that cause wind streams to eddy or reduce the velocity of suspended seeds, or cause them to wobble or spin, may account for the heavier utilizations of certain trees by members of this subfamily compared with other potential targets located in the same currents. `Snag effects' plausibly explain the disproportionate occurrence of Tillandsia flexuosa on Bumelia celastrina on Big Pine Key in south Florida (H. Luther, personal communication). Involvement of seed dispersers Diverse fauna also in¯ uence where epiphytic Bromelioideae grow, but more is known about how certain of the other ¯ eshy-fruited plants disperse (Chapter 6). Several strangling ® gs congregate in the crowns of speci® c kinds of trees on Panamanian Barro Colorado island in part because abundant stem spines (e.g., Hura crepitans) or other surface features provide secure anchorages for seeds carried there by frugivorous birds (Todzia 1986). Platypodium elegans crowns remain mostly uncolonized despite apparently hospitable bark, at least in part because its wind-dispersed seeds attract no birds or bats seeking variety following earlier meals of ripe ® gs. None of the local berry-producing bromeliads were mentioned. Bats that roost by day and pass seeds among the dead leaves suspended below the crowns of palms may account for the narrow occurrence of Billbergia porteana illustrated in Fig. 1.3B. Cambridge Books Online © Cambridge University Press, 2009 Distribution in forests 347 Nest-garden ¯ ora, including the bromeliads (Madison 1979; Davidson 1988), present an even more complicated pattern to unravel. Here, ants choose the phorophyte, and by also creating the rooting medium simultaneously reduce their own and the epiphyte' s dependence on the tree. Complex animal needs affect the choice of phorophyte; Davidson and Epstein (1989) report preferences for Peruvian trees with extra¯ oral nectaries (e.g., Inga), suitability for Homoptera (e.g., Calyptranthes), spiny surfaces that prevent the ascent of predatory ants (e.g., Tococa), or aromatic oils. The same supports growing in deep shade or weakened by disease remain largely nest-free, presumably because insufficient resources exist in their crowns to satisfy the high caloric and additional extraordinary demands of gardening (carton-producing) ants. Ants regularly cultivate several Aechmea, Araeococcus, Neoregelia and Streptocalyx (now Aechmea) species along with members of about 10 additional families in certain lowland forests (Fig. 8.1C). Several nonbromeliads constitute the most frequent nest-users at Cocha Cashu, Peru, where more than three-quarters of all the surveyed cartons supported one or more of these plant species. Restriction to ant-provided substrates often exceeded 95% of the hundreds of sampled gardens. A less common Neoregelia sp. and Streptocalyx longifolius rooted in 1.8 and 3.2% respectively of more than 800 examined nests. An impressive array of fragrances and diverse food rewards promote the dispersal of different ant-garden ¯ ora. Seeds of 9 of the 10 taxa tested, including the two bromeliads, contained volatile methyl-6-methylsalicylate (6-MMS). Aechmea longifolius, along with seven other species, further encouraged ant carriage with benzothiazole (Davidson and Epstein 1989; Seidel et al. 1990). Arti® cial seeds impregnated with 6-MMS attracted considerable attention from nest-building Camponotus femoratus, but elicited little interest from nongardening congeners in the same habitats. Ule (1906) and Madison (1979) proposed pupal mimicry as a possible incentive for myrmecochory among a guild of Amazonian nest-garden ¯ ora, including Aechmea mertensii, that share similarly suggestive seed sizes and shapes. Nest size, exposure to sunlight, characteristics of the local soils, and ant identity all in¯ uence garden ¯ oristics, complexity and vigor in western Amazonia (Davidson 1988). Anthurium gracile occurred above statistical expectation on small nests, whereas Codonanthe uleana uniformly occupied cartons regardless of size. Every one of the relatively common ant-nest garden species except Neoregelia sp. more often co-occurred with other guild members than grew alone. Lower densities of plants on deteriorating substrates, and their absence on all but ant-provided media, indicated near obligacy for many of the participating plants. Phytotelm bromeliads, Cambridge Books Online © Cambridge University Press, 2009 348 Ecology probably because they impound moist, nutrient-rich debris in leaf axils, respond least adversely to the carton erosion that invariably follows abandonment by the ants. Additionally, nests occupied exclusively by bromeliads compared with those hosting diverse ¯ ora more often occurred on trees rooted in relatively impoverished soil. Ant identity and behavior distinguished these systems even further. The simplest gardens were usually tended by taxa (e.g., Azteca, Hypoclinea) with solitary rather than multiple queens. Davidson' s suggestion that trees on low-quality substrates lack capacity to produce the substantial quantities of food that polygynous ants need to sustain their more elaborate nests and gardens warrants testing. Quite possibly, the bromeliad-centered arrangements she noted may not constitute gardens in the described sense at all, but simply represent carton embellishments constructed to enlarge the plant-provided chamber a foundress chose to establish her colony. Bromeliaceae and other arboreal ¯ ora can suppress phorophyte vigor and probably also shorten the lives of trees through mechanisms discussed below, but do they ever reduce ® tness enough to promote the evolution of axeny? Some of the plant characteristics that determine whether a tree is also a phorophyte must be heritable, hence amenable to natural selection. However, axeny sometimes accompanies other, unrelated conditions. Pioneers (e.g., Cecropia) typically host few, if any, bromeliads or other epiphytes, most likely because they maintain smooth barks and grow too fast. Heavy loads of epiphytes require decades to develop even in everwet forests. However, longer-lived, potentially more vulnerable forest dominants also support little or no canopy ¯ ora, and these are the trees that could shed epiphytes to advantage. Role of light and nutrients Arboreal Bromeliaceae obtain nutrients from diverse media, some richer and more continuously available (e.g., decaying litter) than others (e.g., precipitation; Fig. 5.1). Evidence from several Tillandsia species suggests that the identity and condition of the phorophyte can be important for the bromeliad, and perhaps especially so for the type unassisted by impoundments or access to earth soil. Certain trees leak ions or more readily intercept nutrientcharged aerosols than others (Tukey 1970). Epiphyte welfare is further sitedependent to the extent that the chemistry of washes changes while coursing through the canopy. Concentrations of some dissolved plant nutrients tend to fall (e.g., N), while the abundances of others (e.g., K) rise. Additionally, well-provisioned compared with de® cient organs often yield richer leachates. Cambridge Books Online © Cambridge University Press, 2009 Distribution in forests 349 Spanish moss ranges widely through the more humid, relatively frost-free parts of the southeastern United States. Another set of constraints in¯ uences its distribution locally. Schlesinger and Marks (1977) cited nutrient supply as one of the more decisive reasons why only some of the many accessible trees become heavily used. Assays indicated that certain dense colonies of this exceptionally proli® c bromeliad on Florida cypress bene® t from relatively fertile washes compared with less vigorous colonies located in the crowns of nearby Pinus. They also noted that Tillandsia usneoides overoccurs in certain kinds of mixed hardwood communities in southern Georgia and Florida. Principal components analyses of the concentrations of nine elements in foliage obtained from 24 locations grouped all but one collection by origin from one of the three sampled types of forest. Certain soil-born elements (e.g., P) in¯ uenced the ordinations more than others (e.g., Mg, Na) that primarily enter the canopy from the atmosphere. Performances in growth chambers demonstrated that supplemental P enhanced plant vigor, while additional Mg did not. Loose bark was assigned minor responsibility for the relatively poor development of T. usneoides on Pinus. Adults showed no effects in tests for allelopathy, although seedlings might have responded differently to the same prepared leachates. Contrary to Spanish moss, less shade-tolerant T. paucifolia reaches peak densities in Florida on trees that produce unusually dilute canopy washes. Unmatched abundances of this epiphyte infest the crowns of scrub cypress dwarfed in part because the underlying soils are so thin and sandy (Figs. 1.4H, 7.8, 7.9). Media that deny trees adequate supplies of required ions in turn probably explain why the T. paucifolia anchored there remain so small, yet occur in such high numbers. Better-nourished, and consequently more densely foliated, trees rooted in deeper media in Florida produce more ionrich bark and stem ¯ ow, but sustain many fewer, albeit individually larger and better-provisioned, bromeliads (Benzing and Renfrow 1971a; Benzing and Davidson 1979; Figs. 7.8, 7.9; Table 5.1). Scarce supplies of N, P and K depress the vigor of T. paucifolia on dwarfed cypress, whereas plants accumulate enough Mg and a number of the additional essential elements to avoid similar de® ciencies at all of the seven sites surveyed (Figs. 7.8, 7.9). Nevertheless, shortages of key ions, although growth-retarding at the depressed levels provided by dwarfed cypress, constrain this bromeliad less as a reproducing population than the denser shade cast by the same host located on higher-quality soils. Presumably these smaller bromeliads collectively produce more seeds than the colonies of the same epiphyte in denser forests, even though the more Cambridge Books Online © Cambridge University Press, 2009 350 Ecology Figure 7.8. Relationship between P concentration (% dry weight) in mature (fullsized) ramets of Tillandsia paucifolia in south Florida growing on diverse kinds of trees (r50.63, P .0.001; after Benzing and Renfrow 1971a). Figure 7.9. Relationship between Mg concentration (% dry weight) in mature (fullsized) ramets of Tillandsia paucifolia in south Florida growing on diverse kinds of trees (r50.13; after Benzing and Renfrow 1971a). Cambridge Books Online © Cambridge University Press, 2009 Distribution in forests 351 scattered plants at the latter sites individually produce larger numbers of capsules. Martin et al. (1985, 1986) con® rmed shade-tolerance for Spanish moss growing in a North Carolina forest by examining plants in situ and in the laboratory. However, several ® ndings remain puzzling. Samples grown at 7± 15% of full sunlight (100± 200 mmol m22 s21 photosynthetically active radiation (PAR) at midday) differed little from fully irradiated (1500± 1600 mmol m22 s21) controls in some of the features that usually indicate adaptation to low light in land plants. Members of the former group contained more chlorophyll in foliage by weight, but chlorophyll a/b ratios remained undifferentiated. High photosynthetic photon ¯ ux density (PPFD) promoted starch accumulation but had little effect on chloroplast structure, internode length, leaf size, stomatal density, or the morphology of the trichome shields or guard cells. Nocturnal acidi® cation, a reasonable index of daily photosynthesis for a CAM plant, measured about 60% of that recorded for fully irradiated greenhouse subjects. In the ® nal analysis Martin et al. (1986) concluded that Spanish moss acclimates across diverse exposures without substantial morphological or physiological adjustment. Fully conditioned by growth in the forest understory, a full day of about 10 mol m22 PAR saturates photosynthetic capacity (Amax). Finally, fertility appears to compare more closely with exposure as a determinant of site quality for T. usneoides than for more heliophilic T. paucifolia. Vertical stratification Schimper (1884, 1888, 1898) speculated that uneven responses to light and moisture supply explain why Neotropical epiphytes stratify in dense forests. More recent reports cite this pattern for Bromeliaceae, and a substantial literature deals with underlying mechanisms (Chapter 4). Bennett (1987) determined that Catopsis berteroniana, C. floribunda, C. nutans and Guzmania monostachia occur at different heights above ground in swamp forest in southern Florida (Fig. 7.10). Catopsis berteroniana regularly experienced the fullest exposures, while Guzmania monostachia seems to require more light through much of the rest of its extensive range in middle and northern South America. The two other Catopsis species also tolerate substantial shade at this study site. All four of Bennett' s subjects possess broad-leafed, phytotelm shoots, suggesting that distinct physiology and other qualities of foliage account for their capacity to partition shared habitats. Extraordinary costs related Cambridge Books Online © Cambridge University Press, 2009 352 Ecology Figure 7.10. Vertical distribution of four bromeliads in a swamp forest in south Florida showing the range, standard error and mean elevation for each species (after Bennett 1987). to the heavily wax-covered, re¯ ective shoot and additional aspects of carnivory may oblige heliophily in Catopsis berteroniana (Fig. 5.3A). Conversely, deeper green leaves accord with the darker microsites occupied by its two relatives. Catopsis nutans alone displays a monolayered canopy that probably further enhances light capture in low-energy environments (Fig. 4.28). The atypical shade-tolerance of Guzmania monostachia in southern Florida may be a legacy of those sporadic freezes mentioned earlier if they favored genotypes able to avoid frost injury by growing in the heat-trapping forest understory. Pittendrigh (1948) surveyed the Bromeliaceae of Trinidad to determine how the availability of moisture and light permit its members to partition everwet forests of the northern mountains and range across the island' s ® ve progressively drier life zones farther south (Figs. 4.15, 7.11). He found that species segregate through three strata in communities dense enough to feature steep gradients of humidity and PPFD (Fig. 7.11). Pittendrigh' s socalled `exposure' bromeliads occupied the uppermost perches at these humid locations; some of the same species also grew through the entire canopies of certain more sparsely foliated seasonal woodlands. Considerable succulence and dense indumenta or large impoundments combined with Cambridge Books Online © Cambridge University Press, 2009 Distribution in forests 353 Figure 7.11. Schematic diagram illustrating the vertical strati® cation of Bromeliaceae in wet montane forest in northern Trinidad according to Pittendrigh (1948). Three ecophysiological types are recognized based on differences in shade and moisture-tolerance and related shoot architecture. sparser covers of absorbing trichomes characterized these most heliophilic taxa (Figs. 2.4, 2.8). Pittendrigh' s `sun' group, most of which feature broader, relatively shallow phytotelm shoots, inhabit intermediate heights in dense humid forest, and elsewhere tend to experience the same moderate PPFD. His `shade-tolerant' populations congregate even lower in humid forest. They also exceed all other Trinidad Bromeliaceae for thin foliage and sensitivity to drought in part because shallow phytotelma more effectively impound litter than moisture (Fig. 2.4A± D). No effort was made to determine whether any of these epiphytes utilized speci® c kinds of trees more often than others. Pittendrigh' s survey inspired investigations designed to detect aspects of carbon and water balance that underlie his three light-related categories (e.g., Benzing and Renfrow 1971b; Griffiths and Smith 1983; Smith et al. 1985, 1986; Griffiths et al. 1986; Lüttge et al. 1986a). A guiding question concerned why members of Group Three range more widely through dense Cambridge Books Online © Cambridge University Press, 2009 354 Ecology humid forest than upper or middle canopy specialists. Pittendrigh' s claim that shade-tolerant Bromeliaceae would occur in full sun if less sensitive to drought re¯ ected his conviction that these species, like their extinct terrestrial antecedents, are fundamentally heliophilic. Subsequent data would con® rm that the shade-tolerant bromeliads require abundant moisture, but contrary to his views about ecophysiology unchanged from earlier times spent in exposed habitats on the ground, extant forms readily photosaturate and achieve high quantum yields in shade-light (Martin 1994; Fig. 4.7). Shade-tolerant Tillandsioideae (all of Pittendrigh' s designates in Trinidad belong to this subfamily) can perform like sciophytes and acclimate to higher PPFD, but they never escape the limited capacity to conserve moisture imposed by mesomorphic foliage equipped for C3 photosynthesis. Pittendrigh noted that `shade-tolerant' types ¯ ourish in full sun if a nearby stream or comparable source of humidity lowers evaporative demand in adjacent air. Anthocyanins synthesized to screen the mesophyll of highly exposed specimens constitute the most conspicuous adjustment to high exposure under these conditions. Guzmania monostachia responds to potentially injurious PPFD by adjusting leaf structure, chemistry and physiology (Maxwell et al. 1992, 1994, 1995; Figs. 4.24± 4.27; Table 4.6). Photosynthetic pathways tend to predict ecophysiological performances and related growing conditions, but perhaps less reliably among Bromeliaceae than in some other ¯ ora (Chapter 4). Consistent with patterns elsewhere, Pitcairnioideae and Tillandsioideae native to the lower canopy and forest understory typically exhibit C3 photosynthesis (Table 4.1). CAM types provide a less consistent picture. Most Bromelioideae native to dark, humid forests ® x CO2 primarily at night, although not necessarily in the manner displayed by similarly equipped taxa native to more exposed and drier habitats or to gain the same bene® ts. Aechmea aripensis, A. downsiana and A. fendleri possess large phytotelmata and inhabit some of Trinidad' s wettest northern montane habitats, yet they employ CAM (Griffiths and Smith 1983). Terrestrial and CAMequipped Bromelia humilis, and several Ananas species, including a number of feral selections of cultivated A. comosus (Medina et al. 1986), grow more vigorously in partial than in full exposure. Soil-rooted Aechmea magdalenae spreads by ramets to form dense colonies (.1 large rosette m22) in heavy shade (,5% full sunlight) in moist Panamanian forest (Brokaw 1983; P® tsch and Smith 1988), further obscuring the advantages of CAM for some Bromeliaceae. Cambridge Books Online © Cambridge University Press, 2009 Distribution in forests 355 Perhaps some obscure plant advantage granted by CAM explains what only appears to be an ecological paradox in Trinidad' s Bromeliaceae. Possibly only those rare, but decisively, dry years account for the existence of this water-conserving mechanism in populations native to what are generally wet zones. Alternatively, considerable latitude may characterize processes as fundamental to niche de® nition even as carbon and water balance in the climatically permissive humid tropical forest. Suboptimal strategies characterize many other plants in the sense that many populations operate closer to marginal than to ideal growing conditions. Relegations of populations of bromeliads to suboptimal habitats may be common. Peculiarities of dispersal, a local pathogen, some competitor, or in the case of certain CAM bromeliads in northern Trinidad, a trait unrelated to energy, water or nitrogen relations, may account for what appear to be poorly matched plants and growing conditions. Introduced ¯ ora that exploit alien more broadly than native habitats (e.g., Schinus terebinthifolius in south Florida) provide impressive examples in other families. So far, no bromeliad, despite widespread use in managed landscapes, has escaped to naturalize far beyond its home range to demonstrate comparable behavior (exceptions may include Billbergia pyramidalis, a Dyckia species, and Portea petropolitana in parts of Florida, as well as Fascicularia pitcairniifolia on the Isles of Scilly and in parts of western France (Nelson and Zizka 1997); Chapter 15). Bromeliads native to Brazil' s Atlantic Forest also segregate into discrete exposure classes, although perhaps not as precisely as those Pittendrigh noted in Trinidad. Veloso (1952) and Veloso and Klein (1957) determined that of 54 species observed at diverse sites, 20 are heliophilic, 13 others moderately tolerant of shade and dryness, and the remaining 21 taxa mostly root on the bases of trees in dense shade. Reitz (1959) recognized four strata in the tallest forests in Santa Catarina State, each populated by generally nonoverlapping assemblages of species. Tank-producing Nidularium innocentii var. paxianum and N. procerum var. procerum dominated the lowest part of the canopy and sometimes densely enough to appreciably humidify the understory. Vriesea incurvata, V. ensiformis, V. carinata and additional, somewhat more drought-tolerant species anchored on tree trunks from 2 to about 8 m off the ground. Species Reitz labeled `indifferent' , i.e., moderately shade and drought-tolerant types, constitute the third and most numerous and taxonomically varied (e.g., Aechmea nudicaulis, Canistrum lindenii, Vriesea jonghei, Wittrockia superba) of his four groups. Bromeliads restricted to the upper canopy (e.g., Cambridge Books Online © Cambridge University Press, 2009 356 Ecology Vriesea rodigasiana, Tillandsia spp.) require the highest exposures and endure greater aridity than the others. Curiously, several of these heliophiles (e.g., Vriesea rodigasiana) possess lightly trichomed, water-impounding foliage about as thin and broad as that of relatives relegated to deep shade. Presumably, distinct water relations and capacities to harmlessly dissipate excess irradiance underlie the disparate ecology. Epiphytic Bromeliaceae, particularly dry-growing Tillandsioideae, often show little indication that they partition shared canopies. For example, Tillandsia caerulea, T. didisticha, T. latifolia var. divaricata, T. floribunda, T. straminea and Vriesea espinosae seem to anchor interchangeably through the crowns of primarily legume hosts in thorn forest in Loja Province, Ecuador (H. Luther, personal communication). Tillandsia paucifolia, T. balbisiana and T. recurvata behave similarly on dwarfed cypress (Fig. 1.4H) in southern Florida, while T. fasciculata, owing to its greater mass, never reaches maturity unless supported by an axis stout enough to bear considerable weight. Bromeliaceae differentiated by photosynthetic pathway often occur in overlapping arrays on shared trees in wetter forests than those two just mentioned, demonstrating once again (Chapter 4) the failure of these multidimensional syndromes to narrowly proscribe growing conditions. Zotz (1997a) examined the distributions of Guzmania monostachia, Tillandsia fasciculata and Werauhia sanguinolenta on Annona glabra in Panama. All three species possess phytotelm shoots (but utilize different photosynthetic pathways: C3± CAM, CAM, C3 respectively), which could explain why they tended to share space except that Werauhia sanguinolenta more than the other two species concentrated in the upper, presumably driest, most exposed parts of sampled crowns. Zotz also demonstrated that substrates suitable for adults served as well for the more drought-vulnerable seedling stage. Then again, perhaps adjacency to Lake Gatun precludes environmental gradients, which at drier sites might affect distributions more decisively. Light and humidity gradients partition resident bromeliads to different degrees in dense forests. Failure to segregate as extensively in one region as in another sometimes re¯ ects the absence of certain kinds of stock. For example, seven bromeliads, six probably drought-vulnerable types equipped with phytotelm shoots, and ant-house Tillandsia bulbosa concentrated around mid-level at three locations in wet Guayanan lowland forest (Ter Steege and Cornelissen 1989). Deepest shade excluded the entire family, much as Gentry and Dodson (1987) reported for pre-montane rainforest at Rio Palenque, Ecuador, where eight aroids, four dicots, eleven Cambridge Books Online © Cambridge University Press, 2009 Distribution in forests 357 orchids and seven ferns accounted for all of the local, compact sciophytes. Abundant secondary hemiepiphytes representing Araceae, Cyclanthaceae and the ferns share this same darkest space. Failure of all of the many bromeliads perched higher in the canopy to also colonize the deepest portion of the same two forests in Guayana and Ecuador seems odd considering the situation farther south (e.g., Veloso 1952; Reitz 1959). Figure 1.3D illustrates Cryptanthus bromelioides growing on rocks and earth soil in mature Brazilian Atlantic rainforest. Members of Lymania, Nidularium and Neoregelia, among additional genera largely restricted to or endemic to southeastern Brazil, also demonstrate equivalent shade-tolerance. Were subfamily Bromelioideae to range northward in greater force, the family might be far better represented (ignoring Pitcairnia) in the understory north of the Equator. Uneven occurrences of the major taxa making up Bromeliaceae in¯ uence the family' s role and importance in communities in different parts of tropical America to the extent that certain lineages possess propensities for shade-tolerance in addition to other important attributes like epiphytism (Table 1.3). Bromelioideae and Tillandsioideae contribute about equally to the arboreal and lithophytic ¯ oras of the Atlantic forests of Brazil (e.g., Veloso and Klein 1957), while Tillandsioideae predominate farther west and north. Tillandsioideae, speci® cally Catopsis, Guzmania, Tillandsia and Vriesea, accounted for 20 of the 22 epiphytes censused in lower montane rainforest at Monteverde, Costa Rica (Ingram et al. 1996). Pitcairnia contributed the two additional arboreal species. Guzmania and Pitcairnia reach exceptionally high diversities in the everwet montane forests of the Colombian Chocó, the second genus almost entirely as terrestrials and hemiepiphytes. Bias toward Tillandsioideae remains strong at humid locations through Ecuador, Panama and into Mexico. Pitcairnioideae, particularly Puya, along with Tillandsia, predominate at high Andean elevations. Pitcairnioideae account for much of the substantial bromeliad ¯ ora of the Guayanan Shield, mostly as terrestrials adapted to infertile, moist substrates. Kelly (1985) demonstrated the exceptional ecoversatility of at least the occasional bromeliad during his study of a 26± 28-m-tall Jamaican lower montane rainforest. Nine orchids, ® ve bromeliads, three ferns and one Anthurium constituted the compact (nonvining) vascular ¯ ora growing at least 12 m above the ground. A similarly mixed assemblage of seven orchids, two bromeliads, three ferns, two dicots and the same Anthurium occupied the mid-canopy (4± 12 m). Below 4 m resided three ferns and Cambridge Books Online © Cambridge University Press, 2009 358 Ecology phytotelm Hohenbergia pendulaflora, which anchored as well on all but the highest branches. This large-bodied bromeliad also colonized 79.2% of the surveyed trees, more than any of the local compact epiphytes. Numerous hemiepiphytes grew through the same canopy, but as climbers equipped with heterophyllous foliage and adventitious roots these plants may be especially well suited to grow across environmental boundaries that de® ne accessible space for the more compact arboreal ¯ ora. Vining Pitcairnia may share this extraordinary opportunity for functional specialization along a single genet (Fig. 2.2C). Additional patterns Exposure, moisture and nutrient supplies, the behaviors of dispersers, and aspects of substrates in¯ uence local distributions of epiphytic Bromeliaceae. Plant-based agencies operate as well. Tillandsia paucifolia demonstrated that propensity to establish near the sources of seeds contributes to its gregariousness in cypress forests in southern Florida (Benzing 1978b, 1981a; Table 6.13). Kernan and Fowler (1995) determined that disturbance and growth requirements combine to order in space members of a guild of arboreal, relatively shade-tolerant Araceae and Bromeliaceae (Aechmea pubescens, Tillandsia anceps, Vriesea heliconioides, Vriesea sp.) in humid (5000 mm year21), primary, pre-montane forest in the Corcovado Basin on the Osa peninsula of southwestern Costa Rica. Tree falls and lesser disturbances at Kernan and Fowler' s study site force a rotation involving all seven epiphytes, but participation is uneven. Each of ® ve successional stages was mapped to determine its representation within the forest mosaic. Mature forest (recovery stage four) was further divided into forest interior and edge, the latter being the 5 m zone bordering any of the three recovery stages. Measurements obtained with a forester' s cruising prism indicated the amount of bark surface available to epiphytes between 0 and 15 m off the ground within each of the ® ve types of space (Kernan 1994). Observed utilizations (numbers of plants present) and those expected on the basis of available substrate provided the residuals required to determine the relative dependencies of the seven epiphytes on speci® c stages of the forest cycle. Few adult bromeliads inhabited early or mid-recovery forest. More individuals were present by the late-recovery stage, and numbers peaked in fully mature habitat. However, guild members on average were 1.333 more abundant than expected in the early stage, 1.283 more common in midrecovery, 4.043 that number in the late-recovery stage and only 0.483 as Cambridge Books Online © Cambridge University Press, 2009 Distribution in forests 359 frequent as predicted by available substrates in mature forest. Of the individuals anchored in mature forest, 1.353 the expected number occurred in edge space, while just 0.513 as many plants occupied interior regions as the bark present there could theoretically support. Combinations of species differed across the successional mosaic, suggesting that guild members vary in their responses to growing conditions that change during the forest cycle. Gaps favored the bromeliads compared with the aroids, i.e., all the bromeliads exhibited relatively large positive residuals in mature forest. All three aroids exhibited the same, although somewhat muted, pattern. Generally, low densities (total epiphytes) through the forest indicated little likelihood of substantial interspeci® c competition, and accordingly, no need for ecological differentiation to allow the seven populations to co-occur. Nonuniform capacities to colonize young trees in gaps and persist through canopy closure could explain the uneven occurrences of the three aroids and four bromeliads among age-graded habitats. Distinct responses (Vriesea heliconioides least and Philodendron saggitifolia most shade-tolerant) to closed canopy revealed light and possibly drought as key determinants of guild structure. Additional factors attending forest regeneration, including shifts in the relative availabilities of speci® c kinds of trees and bark exposures, might also in¯ uence outcomes at Corcovado. Interpretation was further complicated by the need to rely on adults to infer the behaviors of what in effect are unidenti® able juveniles. Conceivably, the regenerative niches of these epiphytes overlap more than the adults suggest by their distributions. Differential growth and mortality and varied capacities to establish on young vs. older substrates could account for the observed ordering of guild members through the forest mosaic. On the other hand, recall that Zotz' s (1997a) survey in semievergreen forest in Panama indicated that the same kinds of microsites served the seedlings and adults of three bromeliads about equally well. Kernan and Fowler (1995) looked more closely at the substrates these seven epiphytes utilize at Corcovado National Park for signs of differential use irrespective of tree identity. Several characteristics of the epiphytes, speci® cally seed and root morphology, could also affect plant distributions. Vriesea heliconioides, for example, produces relatively short organs and accordingly, clings less securely to the same thick axes that the other guild members surround with roots. Height above ground, inclination relative to gravity, surface texture, and many additional characteristics further differentiated the local barks as potential seed beds. Reliance on frugivores to disperse essentially unappendaged seeds (Aechmea pubescens) vs. wind Cambridge Books Online © Cambridge University Press, 2009 360 Ecology currents (Tillandsia and Vriesea) probably also in¯ uences which of the many local substrates best accommodate speci® c bromeliads. Although the amount of bark surface arrayed between 0 and 15 m above ground was relatively constant across all the sampled sites, the entire guild except for Aechmea pubescens and Anthurium hacumense grew most abundantly at about 5 m elevation. Trunks and branches with diameters between 5 and 20 cm also occurred at about the same frequency through the understory, but all seven epiphytes tended to congregate on axes less than 10 cm thick. Thinner (,5 cm) supports were most favored by Araceae, while the bromeliads, particularly Vriesea heliconioides, rooted on relatively robust substrates (5± 20 cm). Among the bromeliads, V. heliconioides rooted more often than predicted on horizontal and Tillandsia anceps on vertical surfaces. Fewer than expected numbers of epiphytes anchored on axes inclined between 50 and 90°. Bark texture sometimes unexpectedly biased occurrences (e.g., Vriesea heliconioides over-represented on smooth compared with rougher surfaces). Kernan and Fowler imputed mechanisms that foster coexistence and persistence to explain their ® ndings. Coexistence supposedly requires frequency-dependent negative-feedback regulation mediated by heterogeneous bark surfaces and corresponding capacities among guild members to colonize speci® c types of surfaces. Disturbance, including lethal drought during El Niño events in addition to the more continuous background of tree falls, provides the basis for a frequency-independent persistence mechanism (frequency-independent mortality) that prevents the local extirpation of those guild members most vulnerable to competitive exclusion. However, no such mechanism is likely to operate at the study site until populations expand substantially beyond their current low abundances. Additional investigators have sought evidence that co-occurring bromeliads interact to structure communities, and that co-occurring species belong to ecologically de® ned groups (guilds). Hazen (1966) conducted a computer-assisted analysis to determine if spacing along the branches of trees supporting dense colonies of mostly Guzmania monostachia and Tillandsia leiboldiana at a site in Costa Rica indicated competition. No pattern was found. Catling et al. (1986) and Catling and Lefkovitch (1989) used statistical techniques to circumscribe what they considered natural assemblages containing bromeliads, orchids and ferns on cultivated Citrus in Belize, as discussed below. Pittendrigh' s (1948) discovery that Bromeliaceae of Trinidad' s northern wet montane forests belong to three ecologically distinct groups inspired Cambridge Books Online © Cambridge University Press, 2009 Distribution in forests 361 Figure 7.12. Hypothetical phorophyte illustrating the common pattern of epiphyte occurrence through tree crowns in humid forest (after Johansson 1975). Epiphytes are differentiated according to the nature of the substrates and other growing conditions they require (after Benzing 1990). additional research. Several investigators applied his paradigm to other members of the same family elsewhere (e.g., Veloso 1952), or they sought parallels in other arboreal ¯ ora. For example, Johansson (1975) recognized ® ve life zones, each distinguished by its resident orchids, in West African trees, three at different depths in the crown, plus one on the upper and another on the lower trunk (Fig. 7.12). Species sometimes occupied adjacent zones, but usually no more than two. Presumably, drought Cambridge Books Online © Cambridge University Press, 2009 362 Ecology challenges survival on the outermost twigs, whereas insufficient light probably limits epiphyte success deeper in the canopy. Gentry and Dodson (1987) and Catling et al. (1986) noted similar apportionments of bromeliads, ferns and orchids (mid-canopy most species-rich) in Ecuador and Central America respectively. Greater niche overlap recorded for Tillandsia imperialis and additional nonbromeliads by Hietz and Hietz-Seifert (1995c) probably re¯ ected wetter conditions in the Mexican cloud forests they surveyed. Bromeliaceae in addition to Tillandsia recurvata in Baja California (Barry 1953) sometimes occur asymmetrically around vertically oriented stems. Yeaton and Gladstone (1982) noted no over-represented compass orientations on sampled trees, contrary to Bennett' s (1984, 1986a) ® ndings on three Catopsis species and Guzmania monostachia in south Florida where eastern exposure for G. monostachia and northern sides of trunks for Catopsis nutans hosted the fewest plants. Tillandsia pruinosa occurred above expectation on the east sides of supports in a second survey (Bennett 1984), whereas T. flexuosa about as often faced east or west. Although frost helps establish the northern extensions of much of Florida' s ¯ ora, it seems less likely to account for the more ® nely resolved distributions of resident epiphytes. Coldest winds generally blow from the northwest, but the small-diametered axes (,2 cm) that supported many of Bennett' s bromeliads offer little protection in any quadrant. Exposure to precipitation, sun and the other agencies that affect water balance more likely impose site-speci® c mortality, especially for seedlings, which in Florida begin life at least eight months before the lowest temperatures and driest months of the year. Distributions in tree crowns might also re¯ ect movements of the wind currents that deliver comose seeds. Roles in succession Colonization and revegetation of new and denuded, established substrates respectively tends to be orderly and predictable. Several reports indicate that Bromeliaceae participate in succession in the forest canopy, and community structure sometimes suggests similar dynamics occur on the ground. Particularly impressive are the bromeliad-rich restingas of Brazil (Fig. 7.13C± E). Many of these systems resemble the dune sequences of the Great Lakes of North America that prompted some of the initial insights on the mechanisms responsible for plant succession. Claims concerning arboreal Bromeliaceae come from side-by-side observations rather than time-course analyses, in part because succession in tree crowns proceeds so slowly. Similar change involving the lithophytes probably takes even longer. Cambridge Books Online © Cambridge University Press, 2009 Roles in succession 363 Figure 7.13. Bromeliaceae in restinga. (A) Tillandsia stricta growing beneath a nurse shrub suggesting its need to establish on bark prior to anchorage on soil. (B) Unidenti® ed Hohenbergia sp. perched just above the high waterline in a seasonally inundated swale in Bahia State, Brazil. (C) Aechmea nudicaulis in restinga in Rio de Janeiro State, Brazil. Shaded compared with fuller-exposed ramets tend to be greener and the foliage more spreading. (D) Shrub island with an understory and apron of Neoregelia cruenta in Rio de Janeiro State, Brazil. (E) Neoregelia cruenta around another restinga island. Cambridge Books Online © Cambridge University Press, 2009 364 Ecology Figure 7.14. Colonization of Alnus acuminata trees in Bolivia by bromeliads, ferns and orchids relative to the thickness and presumed age of the supporting stems. Note that some bromeliads and orchids established on the youngest exposures. Delayed arrival by participating ferns (and Ericaceae and Piperaceae which are not shown) indicated that many of the local epiphytes require conditioned substrates. Whereas bromeliad diversity soon leveled off, additions of orchids continued until more than twice the number of species comprised this contingent (after Ibisch 1996). Ibisch (1996) reported that certain bromeliads (e.g., Tillandsia adpressa) and orchids rapidly establish on 1± 2-year-old horizontal branches and freshly exposed sites on the older axes of Alnus acuminata in some Bolivian montane rainforests (Fig. 7.14). Growth rings indicated that the later-arriving epiphytes (e.g., ferns, Ericaceae) displace some of these pioneers. Axes supported 6± 8-year-old ¯ owering bromeliads by year 10. After another 5± 10 years, large tank bromeliads and cushion-forming orchids dominated what had become densely occupied surfaces. Mature communities prevail after 20± 25 years. Ericaceae, ferns and Peperomia require site preparation by nonvascular plants, bromeliads and orchids. Freiberg (1996) compared the relative contributions of 77 local epiphytes, including ® ve Bromeliaceae, to arboreal ¯ ora supported by mature specimens of three species of canopy-emergent trees in a Costa Rican moist forest. Epiphyte cover on Hura crepitans, Cebia peltandra and Couratari stellata was 51, 58 and 81% respectively. Values (18.0, 13.7, 18.6%) just for Vriesea amazonica, which occupied more space than any of the other participating species, indicated that Bromeliaceae were dominant and persis- Cambridge Books Online © Cambridge University Press, 2009 Roles in succession 365 tent. Precisely when this exceptionally proli® c tank-forming bromeliad arrives in a maturing tree crown may in turn in¯ uence how extensively the less substrate-neutral ¯ ora representing families like Rubiaceae and Orchidaceae also become part of these suspended communities. Bromeliads depend less on site conditioning than many other arboreal plants, but pioneer status in canopies and on rocks requires con® rmation of the type provided by Ibisch (1996). Competence probably varies with the subject, the mesophytic forms and particularly members of Bromelioideae (Fig. 6.5A,F) requiring more accommodating media than the totally trichome-dependent, xerophytic species, as noted above for Tillandsia paucifolia (Fig. 6.5E). Ant-dispersed Bromeliaceae utilize more complex substrates, often obligately. As such, the nest-garden types represent a subset of epiphytes and accordingly, deserve separate recognition, as do the phytotelm forms that create habitat for ¯ ora and fauna, including many more kinds of ants than create seed beds for the epiphytes. Yeaton and Gladstone (1982) sought evidence that a mixed arboreal ¯ ora containing several bromeliads interacts while colonizing the crowns of Crescentia alata in Guanacaste Province, Costa Rica. The trees in question formed a small, unevenly aged plantation within 100 m of dry, deciduous thorn forest inhabited by the same collection of epiphytes, including the several unidenti® ed Tillandsia species. None of the bromeliads arrived ® rst on young bark, contrary to the situation recorded in Bolivian montane forest and perhaps also on remnant trees following forest clearing at a Mexican site (Hietz-Seifert et al. 1996). Of the nine epiphytes present, up to seven occupied a tree simultaneously, speci® cally the largest and presumably the oldest specimen at the study site. Orchids increased in relative abundance as the supports grew larger except for Encyclia cordigera whose dominance on young trees subsequently diminished. No resident consistently rooted at speci® c heights off the ground or in any other way distinguished itself by anchorage on Crescentia. The orchids tended to be their own nearest neighbors and only Brassalvola nodosa associated with a second species, Encyclia cordigera. Weberocereus glaber, a humiphic cactus, alone required older trees, probably because its seedlings need a more thoroughly conditioned, moisture-retaining bark. Although slower to appear on new substrates, the bromeliads equaled the orchids as colonizers of unmodi® ed bark. Except for the single cactus, seed mobility and plant fecundity probably in¯ uenced arrival time most. Hundreds of thousands of microsperms ripen in a single orchid capsule, assuring a far denser seed rain than possible for any of the bromeliads (100± 300 seeds per capsule typical for Tillandsia) or the cactus. The heavier, Cambridge Books Online © Cambridge University Press, 2009 366 Ecology plumose Tillandsia seed and need to attract frugivores by Weberocereus may further delay the appearance of the nonorchids. Much apparently hospitable bark that nonetheless remained largely free of attached plants, even on older trees, further indicated little opportunity for interactions among co-occurring epiphytes. Certain epiphytic bromeliads co-occur with other arboreal ¯ ora regularly enough to indicate substantial ecological similarity. Catling and Lefkovitch (1989) identi® ed four regular associations of epiphytes located between 0.3 and 5 m above ground in a 2 ha plot of Guatemalan cloud forest. Age (thickness) of the substrates predicted compositions that ranged from two to ® ve species. Of the two types of groups, one early and one later in developing, the former contained fewer ferns and orchids. Participants in the more diverse associations that included two unidenti® ed Tillandsia species were larger and engaged in more seasonal than continuous ¯ owering. Complete life histories might distinguish the identities of these plants as pioneers or later arrivals, and reveal any rules that govern how these groups assemble. The ant-nest sequence Ant-dispersed Bromeliaceae warrant separate mention, as do their relatives that produce habitable cavities for other plants and animals. At least some garden-tending ants cultivate plants selectively, planting the seeds of preferred species while removing the adventive seedlings of others. Plant competition further determines garden composition later. Davidson and Epstein (1989) reported how a series of more light-demanding `nest parasites' replace pioneering Peperomia macrostachya at a site in Amazonian Peru. A similar assemblage of slower-growing aroids, bromeliads and woody epiphytes eventually eliminates more precocious and heliophilic Codonanthe uleana in the same region. Fewer of these later arrivals offer ant food, and some of them reduce nest quality for the ants by clogging carton galleries with roots (e.g., Aechmea angustifolia in Ecuador; Fig. 8.1C). Bromeliads as substrates for flora Communities that resemble those assembled by ants on cartons develop in association with certain phytotelm bromeliads (e.g., Aechmea in Ecuador). A variety of aroids, ferns, gesneriads (e.g., Columnea) and peperomias, and even an occasional Clusia or Ficus (Fig. 1.2D), sometimes nearly obscure Cambridge Books Online © Cambridge University Press, 2009 Roles in succession 367 the hosting epiphyte. If succession or ants or any other seed dispersers help nurture these communities, their involvements remain undocumented. Hietz and Hietz-Seifert (1995a) made no mention of assisting fauna in their report that epiphytic Tillandsia punctulata serves as a nurse plant for Peperomia in Vera Cruz State, Mexico. Less diverse plants colonize the shoots of bird' s-nest Anthurium and the many ferns that also impound litter, but hold less moisture than phytotelm Bromeliaceae. Nonbromeliads also serve as nuclei for ant-garden development, but later arrivals, including a bromeliad, can end up sharing space or even eliminate the pioneer. Epidendrum immatophyllum provides the inducement Azteca sp. needs to initiate nest-building at a site in Belize (Catling 1995). Its massive, foundress-attracting root system dominated the smaller, presumably younger carton on Citrus. Nest size and garden diversity, including the appearance of eventually codominant Aechmea tillandsioides var. kienastii, proceed apace. In all, one or more of 13 vascular taxa rooted in 282 of 288 scored cartons. However, this bromeliad, Coryanthes speciosum, Codonanthe macrodenia and Polypodium polypodioides, in addition to Epidendrum immatophyllum, occupied at least 10% of all the sampled cartons, the bromeliad and orchid closer to 55%. Epidendrum immatophyllum most often occurred without companion ¯ ora. Despite its unmatched frequency in nest gardens, of the ® ve participants this orchid most often (5.5%) also grew elsewhere. Aechmea tillandsioides var. kienastii exceeded statistical expectation in its use of local cartons, and contributed more to total plant cover on larger nests secured to older, thicker limbs. Certain terrestrial Bromeliaceae also provide co-occurring ¯ ora scarce resources ± water in some cases and drier substrates in at least one other situation. Philodendron leal-acostae counters drought in Bahia State, Brazil by extending its roots into the moist leaf axils of adjacent terrestrial Bromeliaceae (Mayo and Barroso 1979). Contact occurred often enough to prompt its discoverers to suggest that the relationship `could be essential' to the aroid. Scarano et al. (1997) identi® ed several large phytotelm Bromelioideae that act as nurse plants for three Clusiaceae that contribute to the woody overstory in freshwater swamps of the Atlantic Forest of southeastern Brazil. A shrubby Erythroxylon owes its dominance in some Brazilian restingas to the accommodating shoots of terrestrial Neoregelia cruenta. Several carnivorous Utricularia spend entire life cycles con® ned to the aquatic microcosms provided by certain Bromeliaceae (Fig. 8.4B). Phytotelm bromeliads promote biotic change by producing broadly hospitable ecospace and providing other resources for additional life forms, both animals and plants. Sometimes the same qualities responsible Books Online © Cambridge University Press, 2009 368 Ecology for self-sufficiency and utility for other biota restrict distributions to robust anchorages (e.g., massive species of Alcantarea and Tillandsia). Heavy specimens that lose their grip on stouter supports may create the opportunity pioneer epiphytes require to recolonize surfaces formerly closed to their propagules by the presence of established vegetation (Ibisch 1996). Perhaps timing distinguishes plant succession on bark vs. soil most of all. Phorophytes die and, prior to that, shed branches, twigs and bark (selfprune, exfoliate), causing substrates to cycle faster than is characteristic for many terrestrial settings, especially rocky exposures. Disturbance of this magnitude probably selects for traits that shorten the life cycle of the barkuser within the limits set by difficult growing conditions. Aridity, by impeding photosynthesis, simultaneously constrains opportunity for succession and the evolution of abbreviated life histories; it may also limit plant capacity to saturate living space or achieve the more structured organization exhibited by communities characteristic of more durable substrates. Bark and twig specialists, especially the Type Five bromeliads, because they require years to reach ¯ owering size, probably escape density-related mortality more often than their faster-growing relatives dependent on richer resource bases. Nevertheless, something other than drought limits the abundances of at least some dry-growing Tillandsia species. For example, colonies of T. paucifolia exhibited similar demographics and dispersions on 50± 200-year-old trees (Figs. 1.4H, 6.8). Much more of the space in these same crowns appeared equally habitable yet remained unoccupied, suggesting little opportunity for interference among co-occurring populations, much as Kernan and Fowler (1995) discovered in wetter Costa Rican forest. If, in fact, community status requires interaction among co-occurring ¯ ora, as does plant succession, then assemblages comprised of bromeliads, like wide-ranging and stress-tolerant T. paucifolia, may not qualify. In summary, parallels and differences characterize succession as textbooks describe this process for terrestrial ¯ ora compared with similar dynamics in the forest canopy involving epiphytic bromeliads. Progressions in both situations begin with the arrival of relatively stress-tolerant recruits capable of growing unaccompanied, and diversi® cation continues once conditions also suit previously excluded, more exacting vegetation. Superior fecundity and mobility and equivalent stress-tolerance allow certain species of orchids to accompany or precede the bromeliads. But whether bromeliads initiate a sequence or join it later, plant size and architecture often assure eventual dominance, especially on relatively harsh substrates. Books Online © Cambridge University Press, 2009 Influences of shoot form on bromeliad distribution 369 Influences of shoot form on bromeliad distribution Shoot morphology and the nature of the foliar indumentum determine whether a bromeliad can access nutrients in substrates like litter and prey (Chapter 5). Likewise, shoot form in¯ uences suitability of the phytotelm types for speci® c climates, especially conditions of humidity and irradiance. A tubular shoot (e.g., Billbergia porteana; Fig. 2.4K) formed by upright, tightly overlapped leaves reduces exposure to direct-beam PPFD, and the resulting deep phytotelm provides substantial insulation for impounded moisture. At the same time, this arrangement curtails capacity to intercept litter and precipitation compared with the ¯ at, spreading rosette characteristic of the more shade-adapted species (Fig. 2.4H). Additional con® gurations that ® t neither of these models require different explanations. For example, the vase-like shoots of Aechmea bracteata allow precipitation and humus to collect in the bases of moderately aged leaves and house ants in a central, totally enclosed chamber (Fig. 2.4G). No arrangement is absolutely ® xed of course, as Tillandsia utriculata illustrates under high and low exposures (Fig. 4.23B,C). Sugden (1981) demonstrated how wind speed, exposure and humidity sort a collection of co-occurring bromeliads by shoot architecture. At issue were eight Tillandsioideae growing along a minor ridge-valley sequence on the Serrania de Macuria in northern Colombia. Rugged topography and constant wind direction maintain canopy height between about 1 and 10 m. Moisture arrives solely as mist for all but about two months during the year. Figure 7.15 illustrates the order these bromeliads follow as they distribute by ecotolerance across two adjacent ridges beginning with the relatively cloud-free leeward slope on the right to ridge top and beyond. Heliophilic Guzmania monostachia, semibulbous Vriesea heterandra and a few succulent Tillandsia bulbosa specimens that feature onion-like shoots regularly utilized by ant colonies at drier, warmer sites grow at the lowest elevations in the thinnest canopies (Figs. 7.15, 8.5). Approaching the summit, the ® rst and last species virtually disappear, replaced by more hygrophilic taxa with lax, soft rosettes and shallow tanks (Guzmania lingulata, G. sanguinea and Vriesea splendens) that reach maximum densities at or near ridge top. More generally distributed V. heterandra also becomes commoner in the especially thick ridge-margin forest. Greater cloud ¯ ux, denser and taller canopies, and coalesced rain drops carried over from the next windward ridge combine to create wetter conditions along sheltered ridge margins compared with the leeward slopes. Epiphytes on windward inclines intercept the most moisture of all, but Cambridge Books Online © Cambridge University Press, 2009 370 Ecology Figure 7.15. Shoot architecture and occurrence of eight bromeliads along a ridgevalley system in a northern Colombian cloud forest (after Sugden 1981). during drier periods the same air currents impose high evaporative demand. Bromeliads with the shallowest tanks (¯ attest rosettes) rarely grow here. Scattered individuals of V. heterandra, which is easily identi® ed by its upright, rigid leaves and capacious reservoir, occur instead. Taller trees in the gully below support delicate, shade-tolerant Guzmania lingulata and G. sanguinea. Guzmania monostachia begins to reappear and Vriesea heterandra continues as a fairly common epiphyte. Although cloud contact diminishes at this point, a dense canopy and relatively still air substantially reduce drought-stress. Similar morphology over extensive ranges indicates that none of these eight bromeliads underwent gross structural change to accommodate conditions speci® c to the Serrania de Macuria ridge system. Selection operating at the Colombian site simply arrayed pre-adapted types to match the multiple local microclimates. Another bromeliad ¯ ora exhibits the same general pattern in the central cordillera of Costa Rica (Burt-Utley and Cambridge Books Online © Cambridge University Press, 2009 Influences of shoot form on bromeliad distribution 371 Utley 1980). In this instance, strong winds sweeping up from the Atlantic coastal plain permit only the most xeromorphic of the local stock, those species with modest phytotelmata and dense indumenta (e.g., Vriesea incurvata, V. chontalensis, Tillandsia adpressa var. tonduziana) to occupy the most demanding exposures. Residents with deeper tanks and softer, more glabrous foliage grow on the less breezy, leeward sides of trees or hills in patches of sheltering forest (e.g., Werauhia attenuata, Vriesea comata). Reliance on leaves to secure the resources that most plants acquire with roots further assures that the size and shape of the shoot inordinately in¯ uence geographic distributions for many Bromeliaceae. Form and function at this scale affect species through a much wider range of growing conditions than those prevailing in the mountains of northern Colombia and Costa Rica. Figure 7.16 plots four models representing Bromeliaceae equipped with carnivorous, humus-based, tubular (vertebrate-fed?) and myrmecotrophic shoots according to their putative capacities to maintain adequate ion, water and carbon balance along gradients of light, moisture and litter supply. The carnivorous habit, here exempli® ed by terrestrial Brocchinia reducta (Fig. 2.4F), requires intense PPFD and a liberal supply of moisture to compensate for the self-shade cast by tight, overlapping leaves and the costs of the fragrances and abundant leaf coating needed to lure and trap prey (Givnish et al. 1984; Chapter 5). Next by niche breadth come the tubular species represented by Billbergia porteana (Fig. 2.4K). Although similar in shape to the carnivores, these plants lack the costly lures and thick, powdery cuticle Brocchinia reducta and Catopsis berteroniana employ to capture prey. Nidularium burchellii intercepts more direct-beam sunlight than plants faithful to any of the other three models, but its monolayered, relatively drought-sensitive foliage and shallow leaf bases restrict occurrence to humid, usually understory habitats. Demand for litter as a nutrient source further mandates anchorages deep in Atlantic rainforest along with numerous comparably vulnerable relatives (e.g., many Canistrum, Nidularium and Wittrockia species). Ant-fed, ant-house types (model four) provide the greatest ¯ exibility of all (Fig. 8.5A). Dry leaf base chambers attract plantfeeding ants and leaf succulence serves in lieu of a phytotelma. Mostly nonoverlapping leaf blades and transparent, immobile and appressed (light-focusing?) trichome shields (Fig. 4.23F) permit survival to the low ends of light gradients. This admittedly facile speculation on comparative functional morphology barely begins to unravel the complex, niche-de® ning interplay that Cambridge Books Online © Cambridge University Press, 2009 372 Ecology prevails between moisture, nutrient and energy supplies and the architecturally mandated growing requirements that encumber speci® c bromeliads. For example, no consideration is assigned to differences in the longevities of shoots that affect mineral-use efficiency, hence nutritional requirements. And no weight was granted to photosynthetic pathway despite signi® cant consequences for carbon and water budgets. Inherent differences in plant vigor and related needs for nutrients like N were also ignored. Should a root system provide more than mechanical support, the inferences depicted in Fig. 7.16 become even more tenuous. Economic models purported to predict where (what kind of environment) speci® c plants should achieve highest performance must incorporate external and inherent constraints on growth and reproduction. Moreover, a fuller understanding of how shoot form in¯ uences shade and drought-tolerance and the utility of certain substrates as soil substitutes will yield additional insights on plant adaptation and evolution. Probably no other radiation within the boundaries of a single family equals that of Bromeliaceae for variety or extremes in the types of habitats colonized, the kinds of substrates utilized, and the manner in which resources are captured and retained. Effects of epiphytic bromeliads on trees The case for parasitism Textbooks routinely dismiss the autotrophic epiphytes as unimportant to hosts, and they say nothing about the roles these plants might play within communities and ecosystems. Abercrombie et al. (1970) de® ne the epiphyte in their Dictionary of Biology simply as `a plant attached to another plant, not growing parasitically upon it but merely using it for support' . Commensalism is the standard explanation except when heavy infestations shade out foliage or break overburdened branches. Casual observers throughout tropical America continue to believe that bromeliads, particularly Tillandsia usneoides (Fig. 7.7C) and T. recurvata (Fig. 7.7D), parasitize trees. Accumulating data and some thoughts about how the nonhaustorial epiphytes obtain nutrients suggest that these claims lack merit, but raise interesting issues of another sort. Frequent occurrences on dead branches and rocks, and even the occasional telephone wire, demonstrate the dispensability of a living host for epiphytic Bromeliaceae (Fig. 1.3A). Self-sufficiency is further indicated by roots, which although they enter crevices in bark, never penetrate function- Books Online © Cambridge University Press, 2009 Effects of epiphytic bromeliads on trees 373 Figure 7.16. Two graphic models that use economic considerations to predict the occurrence of shoot-dependent (holdfast roots only) bromeliads relative to the availability of light and moisture and litter and ant products as sources of nutrients. ing vasculature. In fact, roots of the species most often considered parasites lack signi® cant absorptive capacity or the adult (e.g., T. usneoides, T. capillaris, T. duratii; Fig. 2.10L) rarely, if ever, produces them. In the ® nal analysis, claims for parasitism by direct or indirect (epiparasitism) mechanisms lack substance. Still, appearances call for an explanation. Billings (1904) commented on the premature decline of trees supporting Cambridge Books Online © Cambridge University Press, 2009 374 Ecology Spanish moss in the southeastern United States (Fig. 7.7C), but he ventured no further than to say that experiments conducted over many years would probably be necessary to identify the cause. A pervasive condition involving arboreal ¯ ora and some data on distributions in tree crowns provide a glimpse of what seems to take place between at least some epiphytic Bromeliaceae and their hosts. Knowledge of tree architecture and ontogeny aid interpretations of some of the more provocative observations. Despite morphology that precludes bromeliad parasitism, trees densely colonized by essentially rootless Tillandsia sometimes show distress seemingly imposed by these epiphytes. Heavily encumbered Quercus virginiana in parts of central Florida tend to feature fewer and smaller leaves than usual and inordinately large numbers of dead and dying twigs and larger branches. Figure 7.7C illustrates a live oak festooned with T. usneoides and scattered T. recurvata colonies, whereas Fig. 7.7E shows the much denser crown of a relatively epiphyte-free live oak about 300 m distant. Did the bromeliads on the ® rst oak arrive after or before it began to deteriorate? If before, then how might the bromeliads contribute to this decline? Landscapers in central Florida continue to use herbicides to remove Tillandsia recurvata and T. usneoides from shade and orchard trees to reduce what Ruinen (1953) labeled `epiphytosis' . Treated phorophytes usually escape injury, and, more revealing to us, they often produce fuller crowns within a season or two. Several more authors reported similar signs of distress in trees supporting abundant orchids (e.g., Cook 1926; Johansson 1977). The offending ¯ ora in these cases supposedly acted by girdling and epiparasitism respectively. Epiphytic Bromeliaceae are only occasionally mycorrhizal (Chapter 5), and examination of T. recurvata on a variety of hosts in Florida revealed no girdling roots (Benzing and Seemann 1978). Epiphytes often aggregate on dead branches, supposedly illustrating parasitic cause and effect. `Shootless' African Microcoelia (Orchidaceae) served as Johansson' s (1977) example; Tillandsia recurvata exhibits the same association in Florida (Benzing 1979). Approximately 80% of the mature ballmoss colonies scored through the middle to lower parts of crowns of 10 mature live oaks with normal-appearing foliage occurred on slender twigs (,5 mm). The remaining adults anchored on larger branches and trunks. Fully 70% of those smaller axes that supported mature bromeliads were dead, but so were many others free of epiphytes. Here at least, tree development provided a more convincing explanation for the overoccurrence of ballmoss on lifeless twigs than parasitism. Trees, despite their larger size, like many herbs exhibit determinate Cambridge Books Online © Cambridge University Press, 2009 Effects of epiphytic bromeliads on trees 375 Figure 7.17. The progression of events during phorophyte ontogeny that can produce the illusion of parasitism by resident epiphytes. The crown area enclosed in dashed lines represents the self-shaded region where branches will normally die whether or not epiphytes are anchored there (after Benzing 1979). growth. Beginning with a single trunk, the crown expands in programmed fashion after which it dies, usually in stages. In fact, for much of the life cycle new branches arise while many of the older ones succumb. Three factors determine how long each component (stem) of the crown survives: its place in space and time within the ontogenetic sequence, the model-speci® c form of the species (Hallé et al. 1978), and local circumstances that affect the plant' s ability to conform to its architectural model. Cambridge Books Online © Cambridge University Press, 2009 376 Ecology Crown growth begins with the formation of a number of primary lateral axes (Fig. 7.17). Years later all or most of the lowest members have died and fallen away, leaving those above to proliferate. Surviving second-order axes undergo a third rami® cation and so on as the crown continues to enlarge. Repeated subdivisions, usually no more than ® ve or six in all, yield a sequence of progressively smaller branch complexes arrayed in patterns dictated by the developmental programs associated with the more than 20 currently recognized tree models (Hallé et al. 1978). At some point, the crown periphery where most of the remaining active meristems occur reaches maximum size and lapses into a kind of dynamic equilibrium. Over additional seasons, terminal shoots elongate slowly while generating leaves and reproductive organs, much as herbs (e.g., Solidago) regenerate from perennial bases each year in an old ® eld community in the central United States. For a time, new branchlets replace those lost through attrition and the outline and density of the crown remain little changed. However, wholesale senescence eventually ensues and new meristems no longer replace those spent, and the crown begins to thin. Eventually, whole branches die, and ® nally wind, pathogens or predators dispatch the entire plant. Viewed ontogenetically, a tree constitutes an ordered mosaic of semiautonomous parts, each programmed for a ® nite life span. Every component of the crown falls into one of a series of distinct and successive, but temporally overlapping, populations of shoots. Major axes ± the trunk and its ® rst-order branches ± live longest according to a simple rule: whatever their size, each component must remain autotrophic, the larger ones through the vitality of attached, higher-order shoots. Death follows exposure below the light compensation intensity of attached foliage whether through self-shading or overgrowth by competing ¯ ora. Relatively determinant lateral shoots that serve primarily to provide photosynthate for expanding leaders fall away as the crown expands. Displacement inward insures that the durability of such anchorages for epiphytes typically falls well short of the life of the whole tree (Fig. 7.17). Whether growth proceeds normally or not, plants attached to these axes assume the appearance of parasites, i.e., they appear to have starved the stems that now can provide only mechanical support. If one of the branches making up a tree crown remains exposed longer than usual as the model unfolds, the life of that axis is prolonged, and accordingly, the likelihood of its colonization by epiphytes increases. At some point, this exceptional shoot dies, but only after similarly programmed parts of the same crown fall away on schedule. The longer oppor- Cambridge Books Online © Cambridge University Press, 2009 Effects of epiphytic bromeliads on trees 377 tunity for autotrophy is extended, the more abundant and larger the affected epiphytes become. In essence, as the durability of the exceptional branch increases, so does the chance that the epiphytes located there will appear responsible for its postponed death. Little is known about the light requirements or life histories of the bromeliads that mimic parasites. Most of them probably need several years to mature and additional time to become robust adults. Assuming at least moderate vulnerability to the shade and rain shadows cast by heavy foliage, the highest densities of the largest specimens should occur on those substrates most conducive to photosynthesis. Such sites concentrate near, but not at, the center of the crown, exactly where several studies (e.g., Johansson 1975; Catling et al. 1986) documented the highest abundances of epiphytes (Fig. 7.12). Substrates closer to the trunk become shaded too quickly during tree ontogeny to support comparable densities of epiphytes, while perches at the crown margin subject attached plants to greater evaporative demand and more intense light (Fig. 7.17). So it seems that the overoccurrences of epiphytes on dead and dying branches need not signal parasitism. More likely, these associations occur because the branches involved are relics granted extended life by extraordinary exposure or, if smaller, like those dead oak twigs bearing adult T. recurvata in Florida, simply died after a normal, brief life span. Mechanisms unrelated to nutrition Bromeliads sometimes injure the trees they utilize in ways that have nothing to do with mineral nutrition. Holcomb (1995) reported that the naturalized population of Tillandsia recurvata mentioned earlier is killing the lower branches of crape myrtle trees. Growth ceases after clumps of the epiphyte encircle the tips of affected shoots. Calver et al. (1983) attributed the rapid spread of Tillandsia aeranthos and T. recurvata in and around La Plata, Argentina to allelopathy. Infested conifers and broad-leafed trees alike lose vigor and leaf area, which probably bene® ts these heliophiles much as the stressed canopies of cypress favor dense colonies of T. paucifolia in south Florida. No evidence was provided for the existence of the alleged inhibitor beyond demonstrating the phytotoxicity of unfractionated leachates in crude bioassays. Results of another assay using similarly prepared extracts of T. recurvata and tobacco callus were negative (Holcomb 1995). Disease is another possibility. Fungi (e.g., Rhizoctonia spp.) harbored in the protocorms and roots of certain orchids become pathogenic in more susceptible ¯ ora (Hadley Cambridge Books Online © Cambridge University Press, 2009 378 Ecology 1982). Those vesicular-arbuscular mycorrhizal fungi reported in certain arboreal Bromeliaceae and described in Chapter 5 should pose no problems for hosts, but additional, unidenti® ed mycelia infecting some of the same roots and those of other epiphytes may attack adjacent stems. Large branches occasionally break under the weight of attached bromeliads, and possibly smaller shoots succumb to constricted vasculature. If orchid roots indeed girdled those Citrus branches that Cook (1926) inspected in Puerto Rico, then Bromeliaceae with wiry holdfasts should pose an even greater threat to similarly vulnerable hosts. In any case, dismissal of this possibility requires more than one set of observations on ballmoss in Florida. Competition for light can be a problem for the heavily infested tree, for example those previously described live oaks and cypress in the southeastern United States (Fig. 7.7C,D). Leguminous Cercidium praecox may be exceptionally vulnerable owing to its reliance on green bark while lea¯ ess during extended dry seasons in central Mexico. Montaña et al. (1997) employed a multinomial model to demonstrate that dense colonization by Tillandsia recurvata reduces the vitality of this small tree branch by branch. Speci® cally, heavily shaded axes produced fewer lateral shoots than less encumbered branches in the same crowns, yielding a pattern similar to that just described for associations involving this same bromeliad and Quercus virginiana in Florida. Curiously, trees that support massive trusses of Spanish moss and high densities of Tillandsia recurvata colonies sometimes exhibit considerable die-back and not just the loss of those small, subordinate branches mentioned in the discussion of tree architecture and ontogeny. Undersized, chlorotic foliage further suggests a systemic rather than a localized cause. Benzing and Seemann (1978) coined the term `nutritional piracy' to describe a proposed mechanism. Nutritional piracy A tree need not be a host to parasites to nourish plants anchored in its crown. Arboreal Bromeliaceae possess several options that grant them noninvasive access to the nutrient capital present in living supports. In each case, the epiphyte ipso facto functions as an indirect parasite, or what might be more descriptively labeled a nutrient pirate. Rather than invading xylem or phloem, the resident bromeliad in effect pirates nutrients as they ¯ ux between the tree and its rooting medium during routine biogeochemical cycling. Simply put, essential ions pass from phorophytes to associated Cambridge Books Online © Cambridge University Press, 2009 Effects of epiphytic bromeliads on trees 379 Figure 7.18. Schematic representation of input of nutrients from the atmosphere and mineral cycling in a tropical humid forest depicting the pirating activities of resident epiphytes. arboreal ¯ ora as part of a pervasive ecosystem-level process, effectively rendering mechanically dependent ¯ ora parasites of a remote kind. Elements like K and P, unless immobilized in wood or some other similarly inert, relatively durable tissue, move through a tree rather rapidly, returning to the soil within months to a few years. Litter delivers nutrients to the understory for subsequent release by mineralization; precipitation leaches additional quantities, especially of K, directly from living tissues. Returned to the root zone by either route, recapture and redeployment follow (Fig. 7.18). Porous, infertile substrates oblige efficient recycling to retain sufficient nutrient capital within the ecosystem, much of which at any time resides in phytomass (Jordan 1985). Failure to recycle scarce ions shed in litter and leachates could stress trees on impoverished sites, perhaps as much as if they supported substantial masses of parasites. `Parasitism' describes associations that involve direct transfers; nutrients ¯ ow from one individual comprising the combination into the other. Mistletoes and the root parasites penetrate host vasculature with haustoria, whereas the epiparasites (e.g., Monotropa) employ a fungus, obviating the Books Online © Cambridge University Press, 2009 380 Ecology need for plant-to-plant union. Lacking invasive organs and probably the appropriate fungi, the epiphytic bromeliad functions neither as a parasite of either type nor as a competitor for nutrients in soil, the most extensive plant-accessible source in a forest. Nevertheless, phorophytes lose essential ions to resident epiphytes to the extent that this strategically positioned ¯ ora taps the cycle that permits long-lived vegetation to remain adequately nourished through extended life spans. Nonimpounding Bromeliaceae extract ions from rainfall, leachates and dry deposition, whereas those with phytotelm shoots also utilize litter, which incidentally usually falls through the canopy rather than augmenting the suspended, soil-like substrates so essential to the more drought-sensitive epiphytes (e.g., arboreal bryophytes; Nadkarni and Matelson 1991). Moreover, slow growth, long life and modest litter production (no leaf abscission) combine to insure that the nutrients co-opted by epiphytic vegetation remain sequestered above ground unavailable to the tree for extended periods. So it seems that capacity to scavenge ions from canopy ¯ uids and shed phytomass, durable foliage, and a propitious site in the biogeochemical cycle place arboreal Bromeliaceae in an extraordinary position to in¯ uence the welfare of trees and affect broader community-wide phenomena that also involve plant nutrients. Data from two sites in Florida demonstrate the magnitude of the pirating activities of local Bromeliaceae and perhaps the consequent effects on the growth of co-occurring Quercus virginiana (Benzing and Seemann 1978; Fig. 7.7C,E). Crowns of the oak trees located several kilometers north of Naples in coastal strand vegetation supported dense Spanish moss and ballmoss populations and scattered individuals of Tillandsia balbisiana, T. fasciculata and T. utriculata (Fig. 7.7D). Fruticose lichens, poikilohydrous Selaginella arenicola and scattered forbs provided a sparse, patchy cover over impoverished, acidic, sandy soil (Table 7.1). Sampling demonstrated de® ciencies in the trees dependent on these relatively sterile media. Tillandsia biomass contained 35± 57% of the total N, P and K present in the crowns (foliage, subtending twigs, and attached bromeliads) of two representative oaks at the more impoverished of the two sampled locations. Oak leaf chemistry also con® rmed the relatively oligotrophic quality of this habitat for epiphytes compared with that supporting more vigorous phorophytes near Tampa, Florida (Table 7.2). These data pose several worthwhile questions: how much dwar® ng would prevail today among the woody ¯ ora at the coastal strand site had no bromeliads been present during its development? Would Quercus virginiana be more robust, i.e., Cambridge Books Online © Cambridge University Press, 2009 Table 7.1. Soil fertility data (mean ppm and standard error) and pH (mean and standard error) for two sites in Florida Site 1: Dwarfed oak Site 2: Vigorous oak Available Ca Exchangeable K Available Mg Total N Available P pH 3226221 4776124 23.360.8 54.565.7 41.863.41 63.3610.8 227064281 939262234 12.760.3 25.867.8 5.0360.16 4.8260.27 Source: After Benzing and Seemann (1978). Table 7.2. Mineral nutrient concentrations in leaves of dwarfed and vigorous oaks (Quercus virginiana) and the shoots of their respective Tillandsia usneoides colonists % dry weight Leaves from canopies of dwarfed oaks Leaves from canopies of vigorous oaks Shoots of T. usneoides on dwarfed oaks Shoots of T. usneoides on vigorous oaks ppm N P K Ca Mg Na Mn Fe B Cu Zn 1.43 1.88 0.95 1.19 0.236 0.286 0.140 0.133 0.826 0.846 0.463 0.520 0.752 0.629 0.700 0.587 0.205 0.234 0.197 0.153 0.051 0.043 0.130 0.130 128.1 204.7 57.3 114.0 145.5 116.0 471.3 457.3 20.0 21.2 16.3 17.0 13.8 16.2 10.3 10.0 37.0 1.30 38.2 1.36 27.0 1.66 57.0 1.40 Source: After Benzing and Seemann (1978). Cambridge Books Online © Cambridge University Press, 2009 Mo 382 Ecology better nourished, in the epiphyte-free system? How much of the nutrient capital currently sequestered in the local bromeliads would instead be supporting growth elsewhere in the ecosystem, for example the terrestrial herbs? Bromeliaceae may also signi® cantly in¯ uence how key nutrients are apportioned in more typical (fertile) ecosystems. For example, Nadkarni (1984) reported relatively nutrient-rich throughfall (Ca, K, Mg, N, P enrichments) beneath Clusia alata branches bearing heavy growths of epiphytes, including bromeliads, in a Costa Rican cloud forest using pair-wise tests. Contrary to the wetter months, dry-season rainfall lost ionic strength while moving over the same phytomass. Extrapolations to whole systems will require ® ner-grained sampling because forests and even individual tree crowns constitute successional mosaics. Limbs that support colonies of epiphytes younger or older than those sampled by Nadkarni probably gain or lose solutes accordingly, as do entire ecosystems depending on their state of maturity. Effects of bromeliad nutrition on forests Epiphytic Bromeliaceae may not be parasitic, but they capture nutrients that otherwise would be available to supporting trees for an initial interval (following arrival from the atmosphere) or for second or more successive terms of service (recycling ions). Consequences for phorophytes and ecosystems vary depending on the context. Determining factors range from the abundance and maturity of the resident arboreal ¯ ora (not just Bromeliaceae) to the fertility of the local soils. Expanding populations of epiphytes accumulate nutrients, but later they probably reach equilibrium, matching gains with losses, i.e., they act more like charged capacitors than sinks as at Nadkarni' s Costa Rican site. A second issue concerns nutrient use. Does the N and P sequestered in epiphyte biomass support enough photosynthesis to compensate for the diminished outputs by hosts and other plants deprived of these same resources? Would, in fact, nutrient capital sequestered in epiphyte tissue be deployed elsewhere in the ecosystem had arboreal ¯ ora failed to colonize the site? Few inventories of tropical forests identify the contributions that resident epiphytes make to standing phytomass, or to total nutrient capital; none of them report contributions to overall forest productivity (e.g., Fittkau and Klinge 1973; Edwards and Grubb 1977; Golley et al. 1978; Tanner 1980; Grubb and Edwards 1982). Vascular epiphytes never account for more than a few percent of the total above-ground phytomass, too little Cambridge Books Online © Cambridge University Press, 2009 Effects of bromeliad nutrition on forests 383 at ® rst glance to imagine substantial in¯ uences. However, perspectives change when the contribution of the epiphytes, which are predominantly green tissue, is compared in terms of impacts on the energy budgets of hosting ecosystems. Edwards and Grubb (1977) determined that the resident epiphytes weighed about half as much as the rest of the foliage comprising the canopy of a New Guinea lower montane rainforest, and Tanner (1977) obtained values up to 35% in bromeliad-rich, montane Jamaican sites. Combined nonvascular and vascular epiphytes also accounted for much of the green phytomass in a Costa Rican el® n forest (Nadkarni 1984). Moreover, this compartment contained up to 45% of the totals of several key elements present in tree leaves and epiphytes combined. Contributions of the epiphytes to whole-system photosynthesis probably also vary depending on growing conditions, particularly water supply. Bromeliads and the other epiphytes affect forest productivity and mineral-use efficiency (MUE) according to local climate through its effects on energy returns from investments in green tissues. Trees, owing to the relatively continuous supply of moisture in soil, probably outperform associated arboreal ¯ ora on leaf area and weight bases in all but the wettest locations. The same relationship probably applies for N and P committed to support photosynthesis. Better adapted for drought, the epiphytes should achieve higher water-use efficiency, but lower instantaneous returns on committed nutrients. C3-type compared with CAM foliage, including that produced by some Bromeliaceae (Table 5.5), contains more concentrated N, and usually supports higher rates of light-saturated photosynthesis (Larcher 1980). However, leaf life span, which is usually longer for CAM plants, grants greater parity relative to integrated MUE (Chapter 4). Table 5.5 compares photosynthesis on a leaf area basis for mixed epiphytic plants, including Vriesea platynema, anchored on the lower trunk of a single Psidium specimen in a seasonal, lower montane forest in northern Venezuela. Water-balance mechanisms and moisture sources accompany each entry. The tree performed best on an instantaneous basis in terms of photosynthetic rate and N and P-use efficiency, but the well-watered C3 epiphytes followed close behind. However, integrated values would have reduced the disparities because some of the foliage sampled abscises (Microgramma lycopodioides) or becomes sufficiently moisture-stressed to curtail carbon gain during dry weather. Note that only one of the epiphytes exhibited net photosynthesis prior to irrigation. Vriesea platynema remained inactive until recharged, its sizable phytotelmata having dried out before the run. Adams and Martin Cambridge Books Online © Cambridge University Press, 2009 384 Ecology (1986a) reported similar behavior in Tillandsia deppeana during comparisons of drought-tolerance between adults and juveniles (Fig. 4.9). Both sets of data help explain the absence of Type Three and Four bromeliads in strongly seasonal habitats. Species equipped with phytotelma require at least 1500 mm precipitation year21 in Jamaica (Laessle 1961), and Gilmartin (1973) recorded a similar relationship in Ecuador. A sizable mass of vegetation suspended in a forest canopy, particularly if dominated by phytotelm Bromeliaceae, traps substantial nutrient capital and promotes important, less recognized biological variety as described in the next chapter. Those Venezuelan epiphytes on the guava tree, for example, clearly augmented biodiversity by expanding the list of local species, but more importantly they increased the variety of important goods and services (e.g., food and habitat) for other canopy-based biota. Arboreal ¯ oras incorporating the highest variety, those typically native to pre-montane to mid-montane forests, probably augment supporting ecosystems the most. Conceivably, the complex, multilayered systems (sensu Pittendrigh 1948) that these plants help elaborate utilize resources more efficiently (e.g., calories unit N21 time21), or increase productivity and N2 ® xation beyond levels possible in their absence. Conversely, the far less diverse bromeliad ¯ oras, mostly Tillandsia species, that nevertheless sometimes densely populate certain drier forests, probably favor different outcomes. Here, diversion of scarce ions co-opted from trees to support the more modest photosynthetic outputs of Type Five Bromeliaceae more likely diminish system-wide productivity and reduce instantaneous mineral-use efficiency. Terrestrial Bromeliaceae Substrates receive inordinate emphasis in the literature devoted to bromeliad ecology. Authors often allude to a subject' s epiphytic or terrestrial status as if capacity to root on one or the other kind of medium indicates fundamentally distinct biology. Not so: arboreal and soil-based species frequently share similar life histories, architecture and ecophysiology. Moreover, additional Bromeliaceae root interchangeably on trees and on the ground. We turn under this ® nal heading to the taxonomic correlates of these two life styles, followed by some thoughts about mechanisms insofar as they differentiate, or in the case of the facultative types fail to distinguish, populations with terrestrial from those with arboreal habits. Substrates further subdivide ground-based Bromeliaceae according to anchorage on soil or rock, and within the ® rst group by the type of habitat (e.g., restinga, Cambridge Books Online © Cambridge University Press, 2009 Terrestrial Bromeliaceae 385 alpine). Lithophytes differ in their relationship to supporting media depending on the functions of roots vs. shoots. Bromeliads constitute such a conspicuous presence in the canopies of so many Neotropical forests that the terrestrials, fully half of the nearly 3000 species, often go unmentioned in texts except for the familiar pineapple. Conversations about other important family characteristics frequently re¯ ect the same bias. For example, published descriptions of the phytotelm shoot and absorbing trichome often allude to importance in aerial environments, ignoring comparable bene® ts for large numbers of identically equipped relatives that root in soil or on rock. Systematics suggest that success in tree crowns is recent, and probably required features that emerged in terrestrial stock and continue to serve those descendants still growing on the ground. At least 90% of the arboreal bromeliads belong to fewer than one-third of the approximately 60 genera. Except for several comparably sized collections of species in Pitcairnioideae (e.g., Puya, Pitcairnia, Navia), epiphytism pervades the largest groups of closely related populations (e.g., Tillandsia/Vriesea complex), some showing signs of continuing rapid evolution (e.g., Tillandsia, several subgenera; Chapter 9). Rocky outcrops accommodate additional species in the same alliances indicating similarity between epiphytism and saxicoly. Some of the epiphytes also have close, soil-rooted relatives, especially in Bromelioideae, the subfamily in which substrate use is most ¯ uid. Overall, Bromeliaceae offer unparalleled opportunity to identify aspects of fruits, seeds and dispersal vs. those of plant form and ecophysiology that most powerfully dictate matches between plants and rooting media. Plant features that favor the use of unconventional substrates emerged repeatedly as Bromeliaceae colonized diverse, often demanding habitats (Figs. 9.24, 9.25). Extant Tillandsioideae exhibit the most pervasive dependence on bark and rocks. Some rupicolous Alcantarea andVriesea suggest the likely ancestral habit with shoots that more closely resemble those of certain unspecialized terrestrials (e.g., Fig. 2.2F; Chapter 9). Unfortunately, plant bene® ts (natural selection) responsible for the origins and the timings of the appearances of CAM, phytotelm architecture and absorbing trichomes, and the other modi® cations that enable unusual ecology, remain unresolved despite considerable inquiry and spirited dialogue (e.g., Pittendrigh 1948; Medina 1974; Benzing et al. 1985; Gilmartin and Brown 1986). Members of the largest bromelioid genera (e.g., Aechmea, Billbergia), along with species assigned to many of the smaller ones (e.g., Araeococcus, Cambridge Books Online © Cambridge University Press, 2009 386 Ecology Acanthostachys), seldom root on the ground. Additional clades feature less consistent habits, and many members utilize bark and rocks interchangeably (e.g., Nidularium, Quesnelia); occasional epiphytes stand out in otherwise terrestrial taxa such as Fascicularia (e.g., F. bicolor). Still other fairly sizable assemblages (Cryptanthus, Greigia) never root above ground, at most scrambling over low rocks as rhizomatous clone-formers. Requirements for moisture among members of the strictly earth-bound taxa run the gamut from low to high (e.g., Cryptanthus), and scattered species (e.g., certain Greigia, Puya species) routinely tolerate standing water or saturated soils as discussed below. Except for parts of Tillandsia and Alcantarea, which favor rocky substrata, Tillandsioideae grow as epiphytes. Primarily ground-based Pitcairnioideae includes the occasional hemiepiphyte (Fig. 2.12B) and some true epiphytes, but quite often species of both descriptions qualify for facultative status, as described below. Exceptional members (e.g., Navia tenaculata, Pitcairnia heterophylla) occasionally root on trunks and large branches in humid forests, but most of the membership colonize rock as do many species in several other genera. Hechtia, Navia and Puya mostly grow as lithophytes, or rupestrals, or root on impoverished wet or arid soils, quite possibly as did family ancestors (Fig. 7.1). One of the Guayanan endemics exhibits diversity that mirrors much of the ecological variety expressed across the entire family. Brocchinia tatei closely resembles phytotelm Tillandsioideae, and requires similar humid conditions as an epiphyte or saxicole (Fig. 1.2B). Other members of this enigmatic genus survive periodic partial inundation in hyperseasonal savannas (e.g., B. prismatica), or achieve the shoot morphology necessary to house ant colonies (B. acuminata; Fig. 2.2E) or capture prey (e.g., B. reducta; Fig. 2.4F), or they grow on unembellished rock without bene® t of impoundments (e.g., B. maguirei). Brocchinia micrantha simulates some Puya species at lower altitudes as a giant, palmlike terrestrial. An undescribed saxicole ranks among the smallest of all bromeliads (Holst, personal communication) Finally, B. acuminata assumes a near vining habit in deep forest like some of the more caulescent Pitcairnia and Guzmania (Fig. 2.2E). Conspicuously absent are members equivalent to the succulent xerophytes (e.g., Hechtia, Dyckia) and the essentially rootless, wholly trichome-dependent Type Five Tillandsioideae. Facultative types Certain members of all three subfamilies, but mostly Tillandsioideae and Bromelioideae (e.g., Nidularium, Quesnelia) root on bark and soil. These Cambridge Books Online © Cambridge University Press, 2009 Terrestrial Bromeliaceae 387 facultative types occur in greatest numbers in cool, humid, montane forests, particularly those featuring thick, sodden mats of bryophytes, lichens and vascular plants in the canopy and covering the ground (Fig. 1.2B). Another, smaller group of dry-land inhabitants, primarily members of Tillandsia, succeed about as well on bare rock or arid soils as on nearby shrubs and cacti. Adjacent aerial and terrestrial substrates elsewhere support more distinct ¯ oras probably re¯ ecting their more disparate qualities relative to root aeration and moisture supply. A smaller number of additional taxa (e.g., predominantly terrestrial Bromelia and Ananas) exhibit less, but still enough, ¯ exibility to qualify as accidental or occasional epiphytes. Occasional populations responding to local conditions constitute still another category along the terrestrial/arboreal continuum. Zimmerman and Olmsted (1992) reported that Tillandsia dasyliriifolia grows on bark and then soil as plants mature at certain seasonally inundated forest sites in Yucatán State, Mexico (Fig. 6.5C). Juveniles consistently occurred on small twigs, mostly axes less than 5 cm in diameter. Older, much more massive adults with sizable phytotelmata consistently ¯ owered on the ground, apparently re-established there after falling from overburdened, weak perches. This wide-ranging species grows exclusively on bark in many upland habitats. Inundation during the wet season probably eliminates all of the relatively moisture-sensitive seedlings of T. dasyliriifolia except those attached to a host, whereas larger plants tolerate the same annual ¯ ooding. Less explicable is the failure of this proli® c bromeliad to colonize thicker supports capable of bearing its full adult weight. Wide-ranging Tillandsia stricta sometimes performs similarly, forming almost con¯ uent carpets on the sand beneath certain shrubs forming restinga `islands' northeast of Rio de Janeiro, Brazil (Fig. 7.13D). Phytotelm bromeliads displaced from canopies at the same sites usually die because they fail to maintain the upright orientation necessary to intercept adequate moisture and sunlight. Mechanisms responsible for arboreal compared with terrestrial status in the facultative bromeliad remain obscure, having so far attracted little interest. Obligate and facultative epiphytes alike readily complete life cycles in ordinary pots containing appropriate soils. Various Tillandsioideae, especially the dry-growing species, whether epiphytic or lithophytic, present the greatest challenge to horticulture. Most of these bromeliads respond poorly to excess moisture, including the mesic types as seedlings. Subjects with well-developed phytotelma (e.g., Catopsis, Guzmania, most Vriesea species) tolerate high humidity better than those with denser layers of trichomes (Table 4.8). So far, nothing de® nitive from greenhouse Cambridge Books Online © Cambridge University Press, 2009 388 Ecology experience or nature indicates which plant characteristics dictate rooting media more than others, or why versatility differs among often closely related taxa. Observations in situ, including those just mentioned on T. dasyliriifolia and T. stricta, plus a single experiment provide some grounds for speculation on the basis of obligate epiphytism. Older plants of routinely arboreal species once dislodged from the canopy may survive if litter, or some other kind of porous material, prevents undue contact with soil that remains moist longer than nearby bark. At one extreme, the heavily trichomed Tillandsia specimen perishes within a few months regardless of the type of underlying medium, probably because these plants so readily suffocate in the humid conditions beneath compared with within the canopy (Fig. 4.11; Table 4.8). Bromelioideae generally fare better on alternative media. Light poses an additional threat, as does altered orientation. Loosened epiphytes that remain suspended upside down or accompany a toppled support to the ground also usually die. Bleached foliage, especially among downed cloud forest bromeliads, testi® es to the devastation imposed by abruptly elevated exposure (Fig. 4.23A). Matelson et al. (1993) documented the fates of dislodged epiphytes, including 23 Type Four Tillandsioideae, some of which had fallen in gaps and others below closed canopy in lower montane rainforest at Monteverde, Costa Rica. They re-examined each shoot or cluster of attached ramets monthly for a year and once again on day 637. Mortality continued unabated for specimens assigned to all eight of the categories established for members of six angiosperm families, the pteridophytes and nonvascular ¯ ora. One year later, only 27% of the vascular types remained alive and just 7% survived two full seasons. Bromeliads lived longer in gaps than in deep shade, probably because exposures in the ® rst instance more closely duplicated the drier conditions of the canopy. Excess humidity, insufficient aeration, or diseases encouraged by both agencies probably accounted for most of the deaths, although predation cannot be discounted. Jaramillo and Cavelier (1998) attributed lower rates of ¯ owering among fallen compared with epiphytic specimens of Tillandsia turneri (Type Four) they surveyed in a Colombian montane forest to the relative inhospitability of soil. Amorim de Freitas and Scarano (1998) compared the incidence of epiphytic vs. terrestrial specimens of Nidularium innocentii and N. procerum in a 0.25 ha plot of lowland Atlantic Forest in Rio de Janeiro State, Brazil that included habitats differentiated by hydroperiod (soils continuously, seasonally or never ¯ ooded; Fig. 7.13B) and exposure to sun. They also harvested ramets from terrestrial specimens for attach- Cambridge Books Online © Cambridge University Press, 2009 Terrestrial Bromeliaceae 389 ment to trees that were already supporting these same relatively shade-tolerant phytotelm bromeliads. Soil-rooted ramets greatly outnumbered those secured to bark, and N. procerum occurred exclusively in permanently ¯ ooded and N. innocentii only in periodically inundated sites, perhaps re¯ ecting differences in photosynthetic pathway (C3 vs. CAM) and prevailing degrees of shade (Scarano et al. 1999). No seedlings occurred on soil or bark, suggesting that sexual reproduction is rare or limited to favorable years for these populations. Moreover, the few epiphytic specimens (,2% of the totals for both species) owed their epiphytic status to growth up trunks from older stock that had established on the ground. Severed ramets arti® cally attached to trees rooted, grew and ¯ owered; however, every new shoot produced during the three-year experiment inexplicably failed to orient upright, and, lacking capacity to impound water, eventually died. Amorim de Freitas and Scarano concluded that space for attachment (less for bark than for soil), lack of competitors, and `rosette stability' accounted for the tendency of the studied populations to colonize continuously ¯ ooded to seasonally wet sites rather than the drier locations. Presumably, conditions where these two versatile (e.g., range from 0 to 1300 m) bromeliads regularly occur as trunk epiphytes differ enough to favor that habit over terrestrialism. Amorim de Freitas and Scarano' s ® ndings point the way for additional questions including designs for observations and experiments that could employ these two bromeliads for more de® nitive inquiry into the basis of facultative epiphytism. Exclusively arboreal and soil-rooted species often coexist, but the former usually prevail in dense forest. Terrestrial bromeliads achieve highest cover values in woodlands with lower, more transparent canopies, for example in restingas and deciduous forests (Fig. 7.13C± E). Several other genera exhibit proclivities to grow either as low epiphytes or as terrestrials in heavy shade (e.g., Disteganthus, Nidularium, many Pitcairnia, Ronnbergia). Aechmea magdalenae, which was discussed at some length in Chapter 4, ranks among the most proli® c (by branching) of the deep forest terrestrials. On Panamanian Barro Colorado island, this spiny, up to 2-m-tall plant occurs densely enough to almost completely suppress forest regeneration (Brokaw 1983). Large rodents, and possibly peccaries prior to their elimination from the island, probably helped contain the spread of this vigorous clone-former. Cambridge Books Online © Cambridge University Press, 2009 390 Ecology Ecological variety Terrestrial Bromeliaceae accommodate diverse growing conditions that often include substantial drought and, less frequently, ® re (Figs. 2.2G, 6.12). Opportunity for extended life cycles on durable substrates like the Precambrian outcrops (inselbergs) of the Brazilian Shield (Fig. 1.4A) and the nightly freezes characteristic of tropical alpine habitats explain certain other peculiarities of land-based Bromeliaceae. Cool, semiarid conditions favor still other terrestrials (e.g., Ochagavia and Fascicularia in Chile), as do boggy substrates at high elevations (e.g., some Brocchinia, and Puya, and most Greigia). Stands of Brocchinia tatei specimens over 1 m tall dominate many hectares of windswept, oligotrophic mires on the ¯ anks and summits of numerous Guayanan tepuis (Fig. 1.2B). Certain lithophytic bromeliads rank among the most tenacious of all vascular ¯ ora. Species representing all three subfamilies manage to exploit sheer rock, sometimes virtually unaccompanied by additional higher plants. Figures 1.2C and 7.1G illustrate Alcantarea regina and Tillandsia araujei growing almost alone on a granite dome a few kilometers north of the city of Rio de Janeiro, Brazil. Figure 7.1E shows an unidenti® ed Encholirium sp. anchored on a somewhat less precipitous, but equally barren, igneous outcrop in southern Bahia State. Figure 7.1G illustrating immature Tillandsia araujei, and another, unidenti® ed Tillandsioideae, also reveals the coarsely textured surface that assists the securement of seeds and adhesive roots. Success on these substrates requires the attached plant to either accumulate a soil-like medium, or rely totally on episodic contact with precipitation for moisture and nutrients. Rupestrals Pitcairnioideae and Bromelioideae have radiated extensively on rocky soils distributed from Mexico to Argentina. Speciation among the rupestrals has been especially pronounced on the ancient substrates derived from South America' s two oldest geological formations. The better known of the two corresponding bromeliad ¯ oras occupies the upland (to 2000 m) rocky ® eld habitats called `campos rupestres' located over the Brazilian Shield, mostly in the states of Minas Gerais and Bahia. Endemics abound, many displaying adaptations peculiar to the growing conditions that characterize these thinly vegetated (except for the gallery forests) but ¯ oristically mature regions. Shoots of rupestral Dyckia, Hohenbergia and Orthophytum, among Cambridge Books Online © Cambridge University Press, 2009 Terrestrial Bromeliaceae 391 others, exhibit striking convergence as if multiple, co-occurring lineages adopted a limited number of acceptable architectures. Homoplasy includes additional families, for example Eriocaulaceae (e.g., Paepalanthus bromelioides), that closely resemble co-occurring thin-leafed, phytotelm Tillandsioideae. Similar leaf morphology and shoot form further complicate the already substantial taxonomic challenge for students of Dyckia and Encholirium. Tendencies of some of the local nonsucculent bromeliads (e.g., Alcantarea hatscbachii, A. duarteana) to produce narrow, upright foliage incapable of impounding reservoirs comparable to those of the more typical mesic tillandsioids may re¯ ect long histories of competition for light in these same grassland communities (Chapter 9). Thin, stony soils overlying predominantly quartzitic bedrock (Figs. 1.2A,E, 1.4C) and seasonal climate also fostered the exceptional endemism, including many Bromeliaceae (e.g., Cryptanthus leopoldo-horstii, Neoregelia diamantinensis, Vriesea oligantha), characteristic of the semiarid highlands (the Chapada Dimontina) of south central Brazil. Many of the campos rupestres plants in this region exhibit qualities suggestive of evolutionary stasis comparable to that of Bromeliaceae con® ned to the Guayanan sand savannas and ¯ at-top mountains far to the north where South America exposes that other part of its ancient granitic core. Numerous nonbromeliads of the campos rupestres add to the area' s botanical novelty, and reveal its extraordinary growing conditions and perhaps island-like insularity as indicated by the presence of woody representatives of typically nonarborescent families (e.g., Asteraceae, Eriocaulaceae, Velloziaceae). One especially powerful force probably accounts disproportionately for the unusual structure of much of the native ¯ ora. Ground ® res fed by sparse fuel punctuate dry seasons that extend through early to late winter (May/June to August). Frequent restriction of many local Bromeliaceae to cracks and depressions in the ubiquitous exposed bedrock testi® es to the protection this arrangement affords seedlings and probably many adults (Figs. 6.5B, 7.1F). Relatively thick, sometimes bulbous stems bearing persistent coriaceous foliage increase heat-tolerance for those species more routinely exposed to ¯ ames (e.g., Cottendorfia florida, Cryptanthus schwackeanus, Dyckia dissitiflora, Encholirium spp.; Figs. 2.2G, 6.12C± E). Similar morphology characterizes additional Pitcairnioideae at other locations, for example in the Guayanan highlands where monotypic Ayensua uaipanensis resembles ® re-tolerant Vellozia enough to account for its former assignment to that family. Less insulated Tillandsia (e.g., lithophytic populations of T. arhiza and Cambridge Books Online © Cambridge University Press, 2009 392 Ecology closely related T. streptocarpa) occur exclusively on naked rock beyond the reach of ® re, while the ubiquitous local termite cartons protect other bromeliads nestled amid ¯ ammable grasses and forbs (Fig. 8.1E). Burns may induce ¯ owering in some Pitcairnioideae as occurs for certain orchids and some other herbaceous perennials in other ® re-prone communities (Leme and Marigo 1993). Although the rupestral bromeliads of northern South America and southeastern Brazil represent different parts of the family (no species and few genera overlap), they share too many structural characteristics to imagine anything other than parallel ecological histories. Aquatics Although Bromeliaceae concentrate more in arid than in humid habitats, a substantial number of the terrestrials constitute wetland ¯ ora, by virtue of either propensities for occurrences in seepages or media subject to seasonal inundation (Figs. 1.4G, 7.13B). Epiphytes native to pluvial mossy forests like those of Colombia' s Chocó also qualify as hydrophytic because rooting media, whether suspended or on the ground, usually exist at or near ® eld capacity. Several reportedly rheophytic Guzmania and Pitcairnia species unequivocally qualify as aquatics. Modi® cations for immersion in swift, ¯ owing water include ® rmly rooted, long slender shoots and lax, linear foliage. Guzmania acorifolia, true to its name, looks more like a slender-bodied, emergent aroid than a typical bromeliad along the Nembi River of Colombia. None of the rheophytes has been examined for features that promote photosynthesis in submerged foliage, or resist shear in turbulent water. Many more taxa exhibit riverine-type ecology, and resemble typical members of their genera. Pepinia punicea regularly inhabits rocky streamside sites in southern Mexico and Belize. Occasional populations tolerate brief submergence, while others occur well above the high water mark. A white-¯ owered form of the unusually polymorphic Pitcairnia flammea grows more as an amphibious than a submerged aquatic along the low banks of mountain streams near Teresopolis, Brazil (Fig. 1.4G). Additional bromeliads with no obvious adaptations for ¯ ood-tolerance experience partial inundation for months each year in coastal swales and swamp forests as described below. Lithophytes Lithophytic Bromeliaceae exhibit considerable variety in the ways they meet basic needs despite anchorage on demanding to essentially unyielding Cambridge Books Online © Cambridge University Press, 2009 Terrestrial Bromeliaceae 393 substrates. Exposed outcrops free of deep ® ssures or pockets of soil deter all but those populations capable of amassing soil substitutes (Figs. 1.2C, 2.4), or subsisting on the moisture and ions that foliar trichomes scavenge during transitory contacts with precipitation and related washes (Fig. 4.23E). Some of these same true lithophytes share space with their simulators on more highly weathered or fractured facies that provide more continuous supplies of moisture and nutrients to ¯ ora equipped with extensive absorptive roots (e.g., Navia, Pitcairnia). Capacity to rely on foliage rather than roots for absorption accounts for the inordinate contribution Bromeliaceae make to the lithophytic ¯ ora of tropical America. Propensity for saxicoly and the often hyperdispersed nature of rocky habitat in turn explain much of the narrow endemism that distinguishes this family (Fig. 1.4A). Dozens of Tillandsioideae (e.g., Tillandsia neglecta, T. thiekenii, T. sucrei) grow exclusively on rock, some restricted to one or a few formations in Minas Gerais and contiguous states in southeastern Brazil. Additional Andean populations, particularly in Peru, also cling to rock, and like their relatives on the inselbergs of Brazil, enough divergence has occurred to justify recognition of dozens of narrowly insular species (e.g., T. ecarinata). Certain Type One succulents (e.g., Hechtia, Dyckia, Encholirium; Fig. 2.2A,B) native to rocky outcrops fall somewhere between the true lithophytes and their simulators. Water-storing shoots recharge during the wet season through shallow, highly branched root systems rather than absorptive foliage (Fig. 7.1B,E). Members of several predominantly soil-based genera (e.g., Bromelia) tap leaf axils, and perhaps also deep-seated supplies in fractured rock. Beyond the requisite dispersal mechanisms and capacity to establish on precipitous surfaces, the simulators, and probably many of the intermediate forms, possess no obvious additional qualities for saxicoly, and indeed many of them grow about as well in soil or have fully terrestrial close relatives (e.g., many Pitcairnia). Lithophytic bromeliads evolved repeatedly where suitable substrates and appropriate stock permitted. Taxa like Tillandsia calcicola, which is endemic to the limestone outcrops of western Jamaica' s Cockpit Country, represent recent derivatives from wider-ranging stock, probably something similar to typical T. fasciculata. Additional rock-dwelling relatives of this same robust, mostly epiphytic bromeliad occur scattered through Mesoamerica. Tillandsia utriculata provides a parallel through much of the Caribbean into Mexico. Extensive rocky habitat in Mexico, the Andes, southeastern Brazil and the Guayanan highlands fostered far greater radiations in Bromeliaceae Cambridge Books Online © Cambridge University Press, 2009 394 Ecology than occurred in Jamaica, some of these events accounting for more than 100 surviving lineages. Guayanan Pitcairnioideae provide the best example, apparently because extended time, deeply dissected topography, and geology combined to create exceptionally propitious conditions for vicariance. Some 140 of the approximately 750 species comprising this subfamily and seven of the genera occur exclusively in the Guayanan highlands, many con® ned to narrow ranges. A shield of proterozoic igneous and metamorphic rock (granites, porpyries, gneisses and schists) overlain by up to 3000 m of weathered sandstone known as the Roraima Formation underlies the `Pantepui' where much of Pitcairnioideae apparently differentiated and remains as an admixture of relic and more advanced lineages. Local species representing the other subfamilies (e.g., Aechmea brevicollis, Catopsis berteroniana, Tillandsia complanata) generally range well beyond this region. Low, dome-shaped, lava intrusions known as `lajas' and the much taller ¯ at-topped, steep-sided `tepuis' occur nonuniformly across the Pantepui (Fig. 9.1). Their summits, routinely shrouded by clouds, support uniquely oligotrophic and boggy vegetation on acidic, highly degraded quartzites and sandstones (Fig. 1.2B). Carnivorous plants occur in variety equaled only on the similarly ancient and impoverished soils of southwestern Australia (Givnish et al. 1984). Brocchinia reducta and B. acuminata rank among the most widespread of the Guayanan endemics, ranging through low and high savannas and up on to some of the table mountains. The balance of Brocchinia (e.g., B. cowanii, B. bernardii, B. cryptantha) and members of the other specialized pitcairnioid genera of the region exhibit much greater insularity, sometimes inhabiting only one or a cluster of tepuis (e.g., Steyerbromelia and Brewcaria on Duida and neighboring Marahuaca, Ayensua unipanensis on Auyan Tepui and nearby Uaipan Tepui). Numerous of the approximately 90 described Navia species (e.g., N. saxicola on Cerro Yapacana) grow on just one of these steep-sided formations where they often co-mingle on thin, azonal soil or shear rock with various Brocchinia, Connellia and Lindmania. Like epiphytism, saxicoly remains poorly understood. Few lineages in other families match the most stress-tolerant bromeliads and orchids for capacity to cling to impenetrable rock, and of those that do, few provide equal opportunity to study underlying mechanisms. Populations of Vriesea ensiformis in the state of Santa Catarina, Brazil grow on rocky outcrops, trees, cacti and even soil, as does Vriesea cereicola in central Peru. However, the typical bromeliad remains faithful to a single kind of substratum. For example, an assemblage of dry-growing Tillandsia (e.g., T. brachyphylla, T. Cambridge Books Online © Cambridge University Press, 2009 Terrestrial Bromeliaceae 395 grazielae, T. reclinata) native to southeastern Brazil anchor on rocks, but never the trees that also lie within range of their wind-dispersed seeds. What could account for this variety of behaviors? Redeployment of absorptive function from roots to foliage has not rendered these bromeliads indifferent to supporting media as an agent of Darwinian selection. But how rock vs. bark has in¯ uenced evolution remains unclear. Related epiphytes and saxicoles share what seem to be comparably specialized mechanisms for carbon, mineral and water balance, so something else must relegate speci® c populations to one or the other medium. Potentially decisive features, including succulence, CAM, absorbing trichomes and primarily mechanical roots, characterize all drygrowing Tillandsia. The phytotelm forms share another combination of structure and function irrespective of the rooting medium. However, body plan and certain aspects of reproduction among the ® rst group of species more closely match the type of substrate. Type Five Tillandsia obliged to root on rock routinely exhibit more pronounced caulescence and ¯ ower less frequently than their epiphytic relatives (Fig. 2.10M,N; Chapter 6). Roots develop sparingly, and although individually quite long usually branch less than those employed to grip the more ephemeral surfaces provided by bark. Even stoloniferous Tillandsia usneoides ® ts expectations if its rocky perches at certain Andean locations account for body form more than the trees utilized in the vast majority of its other modern habitats. Conversely, most of the consistently epiphytic Tillandsia species feature relatively compact shoots, root more profusely, and generate short ramets after regular, often annual, ¯ owering. Not surprisingly, the weightiest Tillandsioideae, those with the largest phytotelma (e.g., Alcantarea imperialis, Tillandsia grandis), also colonize rocks, as do most of the other monocarps because these plants require so many years to amass the resources needed to produce the single, necessarily large crop of seeds (Fig. 1.2C). Life on loosely consolidated media also seems to favor caulescence, but the effect on root development has been different. Tillandsia latifolia, T. purpurea and T. paleacea native to the treeless coastal deserts of Peru produce extensive polsters comprised of thousands of elongated ramets, many initiated on spent in¯ orescences by T. latifolia var. vivipara. Roots are few, perhaps re¯ ecting relaxed requirements for conventional anchorage, or the impossibility of securement on such unstable media. The occasional epiphyte, like Tillandsia duratii, exhibits a similarly vestigial root system, in this instance perhaps because foliage that tightly curls around nearby objects provides adequate suspension (Fig. 2.10L). Cambridge Books Online © Cambridge University Press, 2009 396 Ecology Bennett (1991) demonstrated that demography differentiates several species of Tillandsia according to the substratum. Exclusively saxicolous T. sphaerocephala features a lower seedling to adult ratio than facultative T. ionochroma and several other, consistently arboreal taxa (Chapter 6). Additionally, more branching and less ¯ owering characterized the cliffdwellers compared with the epiphytes. Populations of T. ionochroma differed by the type of anchorage, although less so than the wholly epiphytic compared with epilithic species. Bennett concluded that the durability of rock for plant anchorage encourages asexual reproduction (persistence of established genets) over recruitment by seeds. Conversely, greater fecundity and mobility better match the epiphytes with their shorter-lived substrates. High reproductive power leading to frequent recruitment ® gures prominently in the intermediate disturbance model Benzing (1981b) cited to explain the co-occurrence of Tillandsia species on many of the same substrates in Florida. Tillandsia streptocarpa and closely related T. arhiza exhibit two suggestive combinations of architectures and substrates in Minas Gerais State, Brazil. Populations of the ® rst species grow as either lithophytes or epiphytes, or root on both trees and rocks at a single location. The only colony of Tillandsia arhiza encountered occupied exclusively rocky exposures just outside the city of Diamontina. None of the hundreds of unusually robust individuals showed signs of ever fruiting. Tillandsia streptocarpa, in contrast, bore capsules at numerous sites, and typically from ramets with many fewer leafy nodes. Tillandsia arhiza appears to be a lithophytic derivative of wider-ranging T. streptocarpa, differing from it in aspects of vegetative form and reproduction that again appear to re¯ ect the durability of rock compared with bark. Lithic facies of diverse types support Bromeliaceae, and none of these that by virtue of toxic constituents (e.g., ultrama® cs) might exclude colonists reportedly does so. Texture, exposure and fracturing/weathering probably in¯ uence hospitality most. Bark varies in many more dimensions that potentially affect plant ® tness, and indeed account for niche partitionment by narrowly de® ned (twig, knothole, humus and other kinds of substrate specialists; Fig. 7.12) orchids (Benzing 1990). Rocks may differ in important, still unrecognized ways for use by lithophytic Bromeliaceae. Tillandsia tectorum densely colonizes fewer than all the strata exposed at some Peruvian locations, but elevation and associated fog belts may in¯ uence occurrence more than local geology (Luther, personal observation). Cambridge Books Online © Cambridge University Press, 2009 Terrestrial Bromeliaceae 397 Restingas Bromeliaceae occur at exceptionally high densities in certain strand communities along marine coasts, especially in the relatively well-studied `restingas' of southeastern Brazil (Figs. 1.4E, 7.13C± E). Restinga ¯ ora exhibit characteristic zonation beginning with wind-sculpted to prostrate, salt-tolerant vegetation (e.g., Clusiaceae, Arecaceae) just above highest tides. Progressively taller shrubs and trees, including distinctive vegetation `islands' , extend up to several kilometers inland (Fig. 7.13D). Shallow brackish lagoons separate successive ridges at many locations, and some of the higher depressions ¯ ood with fresh water, converting resident Bromeliaceae (e.g., Aechmea bromeliifolia) into emergent aquatics for weeks to months each wet season (Fig. 7.13B). Predominantly terrestrial taxa representing Aechmea, Billbergia, Bromelia, Hohenbergia, Neoregelia and Quesnelia constitute much of the restinga understory. Aprons dominated by one or two species often extend outward for several meters across open sand. Shoots change from spreading to more tubular as full sun replaces shade (Fig. 7.13C). Several Tillandsia and Vriesea species grow as epiphytes and terrestrials, and in that order if they require a nurse shrub until large enough to survive on the ground (e.g., Tillandsia stricta; Fig. 7.13A). Bromeliads that range inland to higher elevations in Atlantic Forest and beyond outnumber the endemics (e.g., Neoregelia cruenta), consistent with recent stocking from older (pre-Holocene) formations. Bromeliaceae of the restinga of Rio de Janeiro State distribute across the beach ridge system according to qualities of the substrate that probably re¯ ect stability and age, and plant exposure to wind and sun and perhaps salt spray (Fig. 1.4E). Occasional natives like Neoregelia cruenta populate the entire system, maintaining high densities in scattered colonies by switching from terrestrial to epiphytic habits as open area near the beach gives way to more continuous woody cover inland (Lacerda and Hay 1982). About 30 km northeast of the city of Rio de Janeiro in the coastal sand dune ecosystem known as the Barra de Marica, Dyckia pseudococcinea occupies open, relatively ¯ at habitat dominated by forbs and low shrubs several hundred meters behind the forward sand ridges. Scattered Aechmea nudicaulis, Neoregelia cruenta and Vriesea neoglutinosa co-occur, but become more abundant seaward as nonvegetated substrate expands. Tillandsia stricta follows a less continuous distribution, reaching highest densities on the ground beneath the nurse shrubs required for its establishment (Fig. 7.13A). Cambridge Books Online © Cambridge University Press, 2009 398 Ecology Densely branched shrubs (e.g., Clusia spp.) account for most of the restinga biomass at the Barra de Marica. These same species probably help enrich and consolidate the sandy soil for more demanding vegetation. Aechmea nudicaulis, Neoregelia cruenta and Vriesea neoglutinosa form most of the fringing aprons (Fig. 7.13C± E). Like Bromelia humilis described below in coastal Venezuela, shade promotes greener in addition to more lax foliage. Unscreened irradiance appears to severely stress the ramets that represent extensions of genets that began life under the more equable conditions provided by taller vegetation. Continued proliferation despite the more severe conditions imposed by open habitat indicates either substantial translocation of photosynthate among attached ramets or greater capacity for photosynthesis in full sun than demonstrated by Bromelia humilis in Venezuela (Lee et al. 1989). Island remnants in the form of senescing and dead shrubs, columnar cacti, and stranded bromeliads suggest age or storm-related regenerative cycles perhaps comparable to the mangrove dynamics along storm tracks in the Caribbean. Skeletons of woody ¯ ora and interspersed, equally rotresistant stolons of bromeliads mark the loci of former restinga islands. Bromeliad importance measured by biomass and cover values peaks around mid-cycle. According to Hay et al. (1981) and Lacerda and Hay (1982), Neoregelia cruenta and co-occurring phytotelm bromeliads add substantial amounts of organic matter to underlying soil, heightening its fertility and cation exchange capacity. Speci® cally, soil sampled from beneath N. cruenta colonies vs. a few meters distant in open habitat tested at 1.15 vs. 0.39% for soil organic matter and 3.96 vs. 0.77 for meq 100 g21 soil for cation exchange capacity. Soil reaction compared more closely (pH 5.3 vs. 5.4). A relatively spreading shoot also assures N. cruenta importance as a recruitment site for certain shrubs (Fialho 1990), and a source of fresh water for fauna during droughts (Fig. 8.4D). Frogs that frequent these and the more tubular, moist refuges provided by co-occurring Bromeliaceae probably lack alternatives during the dry season (Fig. 8.4). Coastal Bromeliaceae range northward from the restingas of Brazil into Mexico to reappear along Florida' s increasingly urbanized southwest coast and on through the Bahamian islands. Species, subfamilies and plant habits shift en route. Bromelia maintains a more or less continuous presence in strand communities northward into Mexico, but no further. Predominantly terrestrial Bromelioideae prevail in the southern hemisphere giving way completely to epiphytic Tillandsioideae north of the Caribbean. Tillandsia contributes virtually everywhere, and accounts for all of the up to seven Cambridge Books Online © Cambridge University Press, 2009 Terrestrial Bromeliaceae 399 species that represent the family in the nearly extirpated coastal landscapes located between Naples and Tampa Bay (Fig. 7.7D). Local Tillandsia balbisiana, T. fasciculata and T. utriculata occasionally grow on sandy soils in the fashion of T. stricta in Brazil. Tillandsia dasyliriifolia growing as a low epiphyte represents Bromeliaceae along the coast north of Merida (Yucatán State), Mexico. Closely related T. utriculata occupies the same kind of microsite in south Florida. Of the seven species of Bromeliaceae native to the west coast of Florida, Tillandsia recurvata reaches densities great enough at some locations to almost obscure the foliage of low-growing Quercus virginiana (Benzing and Seemann 1978; Fig. 7.7D). Night fogs that regularly move on-shore through the drier winter months probably encourage this extraordinary abundance. Paci® c coast Bromeliaceae, except for the dense populations of several Tillandsia species that grow nearly unaccompanied by other vascular ¯ ora through parts of the Atacama region of Chile and similarly hyperarid regions of Peru, remain poorly documented (Chapter 9). Recall that Tillandsia recurvata survives exclusively on fog water at similar latitudes in Baja California (Barry 1953). Alpine species Alpine, as distinct from montane, Bromeliaceae range from Mexico to Argentina, Bolivia and Chile, but diversity peaks in the northern and central Andes. One subfamily, and primarily Puya, accounts for most of the family' s representation above 3000 m. As the tallest plants in many of these stark, thinly vegetated landscapes, members of this genus often attract inordinate attention from local vertebrates, particularly birds (Fig. 14.2C). Perches, shelters, nesting sites and, during anthesis, abundant nectar assure high importance to these demanding animals. Suitability for cool, dry sites permits several of the largest puyas to dominate some of the highest communities in South America. A moderate number of tillandsias, some additional Pitcairnioideae and fewer Bromelioideae co-occur with alpine Puya, but except for Greigia these bromeliads represent exceptions within clades with more fundamental affinities for warmer habitats. If tolerances for drought, frost and high irradiance existed in the ancestors of Puya, they probably intensi® ed as northwestern South America began its on-going orogeny. Whatever the case, Puya almost certainly underwent much of its diversi® cation during the past 3± 5 million years in concert with and probably in response to the combined effects of substantial climate change and mountain-building. Cambridge Books Online © Cambridge University Press, 2009 400 Ecology Current distributions and paleoclimatology suggest that much of the impetus for speciation leading to approximately 185 species was provided by mean temperatures that oscillated through the Plio-Pleistocene, repeatedly expanding and contracting paramo and subparamo habitats and fragmenting the ranges of resident ¯ ora. Taxon-speci® c structure and function suggest characteristics that differentially in¯ uenced the magnitudes of alpine radiations, and continue to affect the vulnerabilities of surviving lineages. Dry fruits that release seeds poorly designed for wind carriage probably help explain the exceptional endemism in Puya (Figs. 3.9, 9.2). Fires set by farmers threaten a number of taxa represented today by scattered small populations. Puya compacta, P. nutans and P. sodiroana, among others, persist as isolated stands, each composed of fewer than 50 individuals occupying a handful of paramo habitats in just one of the 11 discrete centers of diversity arrayed from Colombia to Chile (Fig. 9.2; Varadarajan 1990). Relatively few lineages (e.g., P. floccosa) occur in two or more of these regions. Puya dasylirioides ranks among the more exceptional taxa as an outlier with a disjunct range that reaches central Costa Rica. The other alpine bromeliads (more southerly Abromeitiella, now Deuterocohnia) require further study to determine how closely their histories parallel the pattern illustrated by Puya. Bromelia humilis: a case study of terrestrialism Cultivated Ananas comosus excepted, the most extensive literature on the ecophysiology of a terrestrial bromeliad concerns closely related Bromelia humilis. This robust plant ranges through low to moderate elevation (,1500 m), humid to semiarid habitats across the southern Caribbean, including the Windward Islands and the ¯ oristically mature coastal communities situated along the northern rim of South America. Pittendrigh (1948) used this heavily armed, stress-tolerant species to exemplify Type Two, the tank-root type, in his four-part system designed to showcase the major adaptive modes expressed among the Bromeliaceae of Trinidad (Table 4.2). Wet humus impounded among the shallow leaf axils of tank-root Bromeliaceae supposedly satis® es much of the plant' s needs for moisture and nutrients. Although Pittendrigh inferred much information about bromeliad ecology and physiology from gross form, in this instance he probably overemphasized the importance of apogeotropic roots. Foliar trichomes located on the leaf bases of Ananas comosus exhibit ultrastructure consistent with absorptive function (Sakai and Sanford 1979), and Cambridge Books Online © Cambridge University Press, 2009 Bromelia humilis: a case study of terrestrialism 401 these same appendages probably provide similar service for all of the other Type Two species. Little to no rooting into substrates further underscores the importance of tanks, and opportunity for sensitive species to avoid salt on saline soils as described below. Lee et al. (1989) investigated Bromelia humilis in alluvial plain and beach habitats near the mouths of the Tocuyo and Tucurere rivers in northwest Venezuela. Previously thought to be a salt-tolerant, mangrove-style landbuilder here, its distribution and generalized ecology indicate a secondary role in community dynamics. Salt-avoidance associated with and perhaps promoted by versatile shoot form and function foster different performances along steep gradients of exposure, fertility and salinity. Plants extensively occupied both realized (specimens fruitful) and apparent (specimens subreproductive) niches at the study sites. Qualities manifested in favorable ecospace suggested parallels with highly productive CAM plants in other families (Nobel 1991). However, B. humilis failed to turn in comparable performances in Venezuela. Bromelia humilis illustrates some noteworthy inconsistencies between structure and function. Its unusually well-ventilated mesophyll (mean intercellular space 9.7% vs. 2.8± 4.9% for many other CAM taxa) seems unnecessary for a plant capable of only modest rates of photosynthesis. Substantial succulence and high leaf area indices combined with rosulate form further suggest equivalence with more vigorous Ananas comosus and Agave deserti (Nobel 1991). Ecology also fails to accord with ® ndings obtained from individual plants (e.g., Lee et al. 1989). Impenetrable thickets, created by nearly monospeci® c, con¯ uent understories of B. humilis along the Caribbean coast below Rancho Grande, Venezuela, clearly document this sturdy xerophyte' s capacity to dominate open habitats. However, such stands could represent many seasons of slow growth and demonstrate the advantages of ® re-retardant foliage bearing ¯ esh-rending armature over less protected ¯ ora rather than the results of high vigor. Records from Venezuela demonstrate how moderate compared with full insolation promotes growth and DH1 and elevates chlorophyll and N concentrations in B. humilis. Photoinhibition experienced by plants subjected to high exposures required several hours to dissipate in shade, even among individuals acclimated to full sun. Such plants had developed more compact shoots, and possessed xeromorphic foliage well provisioned with the xanthophyll-cycle intermediates that help prevent photodamage in overexposed leaves (Figs. 4.24± 4.27). Leaves subjected to unscreened irradiance also senesced sooner (one year) than foliage produced by the deepest green specimens protected by Cambridge Books Online © Cambridge University Press, 2009 402 Ecology Table 7.3. Some distinguishing characteristics of the foliage of three phenotypes of Bromelia humilis encountered in a coastal habitat in northern Venezuela. All measurements except for pigments (season unspecified) were recorded during the dry season Shaded Net CO2 uptake 22.2 (mmol m22 day21) DH1 calculated as 173 malate (mol m23) CO2 recycling 56.0 (as % of CO2 ® xed) Respiration (mmol h21 g21 fresh weight) 5.4 Succulence (kg m22) 0.98 Dry weight/fresh weight 0.186 ratio Total chlorophyll 204.9 (mg g21 fresh weight) Carotenoid/chlorophyll 0.31 ratio Cell sap p (MPa) 1.14 (dawn) 1.05 (dusk) Xylem tension (MPa) 0.54 (dawn) 0.49 (dusk) N content (% dry weight) 0.68 (dawn) 0.62 (dusk) Exposed green Exposed yellow 43.4 205 7.2 160 37.6 5.6 1.21 0.186 74.0 0.46 87.0 6.2 1.18 0.219 55.8 0.91 1.17 (dawn) 1.06 (dawn) 0.86 (dusk) 0.94 (dusk) 0.59 (dawn) 0.62 (dawn) 0.47 (dusk) 0.57 (dusk) 0.08 (dawn)a 0.05 (dusk)a Source: After Lee et al. (1989). Note: aOne set of values for pooled green and yellow exposed plants. taller shrubs. Individuals in full sun expressed two additional phenotypes marked by bright yellow or light green leaves and corresponding chemical compositions. Other symptoms of stress, including diminished CO2 consumption and greater reliance on recycled (respired) carbon rather than carbon from the atmosphere to supply CAM, paralleled chlorosis (Table 7.3). Failure of all but an occasional sun-bleached specimen to ¯ ower or produce a ramet demonstrated how severely overexposure depressed ® tness. Photosynthesis inhibited by strong light that also hastens leaf turnover denies Bromelia humilis the vigor necessary for land-building in the estuarine setting Lee et al. (1989) investigated in coastal Venezuela. Instead, other, more proli® c, sun and possibly salt-tolerant vegetation traps most of the sediments destined to form island basements. Presumably the bromeli- Cambridge Books Online © Cambridge University Press, 2009 Bromelia humilis: a case study of terrestrialism 403 ads situated in full sun at the study sites had either grown out from under the taller woody vegetation that dominated the islands, or arrived as detached shoots from the same, more favorable habitat. Rather than classing it as a pioneer, Lee et al. (1989) declared B. humilis a subordinate species that requires shade to achieve its modest importance in communities dominated by taller ¯ ora. Moisture and N supply, and perhaps high temperature, but not salinity affect its vegetative vigor and ® tness. Roots supposedly avoid salt by invading the leaf axils rather than the underlying tidal muds. Low chlorine levels in foliage indicated that rainfall either ¯ ushes the upright shoots often enough, or the trichomes and roots exclude any contaminating sea salts. Conditions most favorable for B. humilis at the coastal sites differ from those conducive to higher yields from some other CAM plants. Wellwatered and fertilized, an Agave and an Opuntia species achieved Amax in near to full sun as do numerous other desert succulents (Nobel 1991). Some pineapple varieties also outperformed Bromelia humilis in undiminished sunlight, proof that nothing inherent to their shared architecture or family affiliation precludes vigorous photosynthesis. Ecotypes may account for certain inconsistencies in the literature, for example a report that drought more than overexposure suppressed growth in a second, nonsaline forest site (Medina et al. 1986). However, severe climate imposes multiple stresses, and in Venezuela complicated attempts to identify reasons for speci® c plant performances (Chapter 4). High exposure heightens water de® cits and raises leaf temperatures in addition to damaging the light-harvesting apparatus. Nitrogen plays a complex and still poorly understood role in carbon balance according to performances recorded for Bromelia humilis and similar bromeliads, in terms of both its availability in situ and its fate following absorption. In addition to B. humilis, three Ananas species, including feral populations of A. comosus in northern Venezuela (Fetene et al. 1990; Fetene and Lüttge 1991; Medina et al. 1991b) and Panamanian Aechmea magdalenae (P® tsch and Smith 1988), produce more N-rich foliage capable of higher rates of net photosynthesis in shade than in full sun (Figs. 4.5, 4.6). Medina et al. (1986) suggested that moisture-stress exacerbated by intense insolation and infertile soil (less impoverished in forests) accounted for the diminished yields recorded for unshaded subjects. Instead, overexposure may assure relatively low N content in stressed foliage, and accordingly, lower photosynthetic capacity independent of the environmental supply. Half or more of the total N in a typical green cell Cambridge Books Online © Cambridge University Press, 2009 404 Ecology supports photosynthesis as a constituent of the necessary enzymes, cytochromes, pigments and coupling factors, all of which accumulate in speci® c proportions in¯ uenced by N nutrition and exposure. Many relatives exhibit similar tendencies to bleach (e.g., Fig. 4.26) that elevated anthocyanins often mask in full sun. Many more Bromeliaceae than currently recognized may respond similarly to high PPFD, possibly re¯ ecting underlying mechanisms that promote highest performance in diffuse light (Chapter 4). Many of the data on interactions among temperature, exposure, moisture and nutrient supply as they affect bromeliad CAM come from Venezuelan Ananas ananassoides, A. comosus and A. paraguazensis and Bromelia humilis. Medina et al. (1991a,b, 1993), who collected much of this information, concluded that Ananas originated in humid, lowland understory sites in northern South America where it remains most diverse. Although fundamentally suited for understory habitats, capacity to produce relatively drought-adapted foliage, i.e., greater xeromorphy and increased reliance on CAM, improves performance under higher exposures, although more for some species than for others (A. ananassoides vs. A. paraguazensis). Cambridge Books Online © Cambridge University Press, 2009 8 Relationships with fauna Vascular plants literally energize every major land-based ecosystem. They also furnish co-occurring biota additional resources and services unrelated to nutrition. Bromeliaceae stand out on this second count by providing an extraordinary array of bene® ts to a diverse, incompletely inventoried fauna (e.g., Figs. 8.1± 8.4) and even some ¯ ora (Fig. 8.4B). For example, many of the phytotelm types achieve keystone status more as animal habitat than as food (Fig. 2.4). These primarily arboreal bromeliads also lend a distinctive visual aspect to forests throughout the Americas while they contribute inordinately to the often exceptional biocomplexity of the same speciesrich communities (Fig. 1.4F). Visits to bromeliads by animals to collect pollen, nectar and fruit are considered in Chapter 6; carnivory, mycorrhizae, symbiotic diazotrophs, and certain aspects of ant-house and ant-gardened mutualisms receive due treatment in Chapters 4 and 5. Conversely, enemies and the ants that deter predators for many Bromeliaceae have largely gone unmentioned. Evidence of how extensively family members serve still another set of organisms, many of which occupy and release nutrients from litter impounded in phytotelmata, consists mostly of checklists of surveyed taxa. As we shall see, even this preliminary literature indicates that the bromeliads in¯ uence events in many Neotropical ecosystems far beyond what relatively small body size and usually modest contributions to aggregate phytomass would predict. Predators and pathogens No comprehensive records of herbivores or pathogens exist for Bromeliaceae, nor does the available information suggest extraordinary susceptibility. If anything, immunity to certain wide-ranging plant-users 405 Cambridge Books Online © Cambridge University Press, 2009 406 Relationships with fauna like the leaf-cutting ants and larger grazers may be relatively well developed. Bromeliads regularly sustain damage in situ, but usually no more than scattered necrotic spots and some chewed holes in leaves and reproductive organs (Figs. 8.1F, 8.2). Massive defoliation rarely occurs, perhaps owing in part to the presence of strong defenses obliged by relatively unfavorable architecture. Because bromeliads are essentially stemless plants with even less opportunity to store reserves in often vestigial root systems, severe herbivory probably compromises regenerative capacity more for them than for most woody plants and many other herbs. Scattered reports indicate that Bromeliaceae possess potential chemical deterrents, and may be exceptionally well provisioned with these compounds. Chedier and Kaplan (1996) demonstrated seasonal, perhaps drought-related, ¯ uctuations in the amounts of waxes, condensed tannins, and total phenols present in the foliage of Nidularium procerum, N. innocentii and Quesnelia quesneliana. Triterpenoids and steroids occur in Florida Spanish moss (Atallah and Nicholas 1971) and several other tillandsias (Arslanian et al. 1986). Hegnauer (1963) reported a steroid fraction with oestrogenous activity in the cuticles of several Tillandsioideae. Williams (1978) encountered ¯ avonoids in uncommon variety, but mostly of unknown signi® cance, distributed among species representing all three subfamilies (Chapter 9). Sclerophylly, trichomes and succulence probably also reduce the palatability and nutritional quality of much bromeliad foliage compared with nearby, often shorter-lived, softer forage, for example the trees hosting the epiphytes. On the other hand, certain bromeliads attract devastating attention from animals immune to most counter-measures. Large carnivores, and particularly primates seeking prey in overlapping leaf bases of the phytotelm types, in¯ ict damage that neither chemicals nor most of the mechanical impediments available to the bromeliads can discourage. What can be said at this time about predation pressure on Bromeliaceae vs. other ¯ ora? Annual rates of leaf consumption for a variety of kinds of tropical evergreen forests range into the low double digits. Now and then, a few trees experience losses equivalent to most of their canopy (Lowman 1995). The same observer and others (Lowman et al. 1996) reported depredations caused by an unseen dietary specialist ± up to 25% (mean leaf area missing 10.4%) ± of the foliage of a small colony of Peruvian Aechmea nallyi. Only once have I witnessed a more severe event. On that occasion, an unusually abundant grasshopper had grazed at least 25% of the leaf area serving the adults of a Guzmania monostachia population located in the lower canopy of a swamp forest in southern Florida. Two years later new Cambridge Books Online © Cambridge University Press, 2009 Predators and pathogens 407 Figure 8.1. Associations between ants and termites and Bromeliaceae. (A) Lithophytic Aechmea phanerophlebia in Minas Gerais State, Brazil with an associated termite trail (arrow). (B) Aechmea bracteata in Yucatán State, Mexico with a termite carton partially exposed on the shoot (arrow). (C) Ant-nest garden dominated by Aechmea angustifolia in Ecuador. Lower portion cut open to expose abundance of roots. (D) Shoot of Aechmea phanerophlebia in Minas Gerais State, Brazil cut open to reveal ant carton. (E) Lithophytic Encholirium sp. in Minas Gerais State, Brazil associated with termite carton. (F) Colony of lithophytic Aechmea phanerophlebia in Minas Gerais State, Brazil showing insect damage despite the ants that inhabit most shoots. (G) Dyckia specimen uprooted in Minas Gerais State, Brazil to expose a cluster of associated termites (arrow). Cambridge Books Online © Cambridge University Press, 2009 408 Relationships with fauna ramets had fully restored the colony to its usual condition. Several additional bromeliads may be exceptionally susceptible to herbivory, for example wide-ranging Aechmea bromeliifolia in Bahia and Minas Gerais states in southeastern Brazil (personal observation). Equally severe events sometimes impact reproductive organs, but how often remains unclear and difficult to determine because damage to seeds is often cryptic. Garcia-Franco and Rico-Gray (1991) reported heavy predation on the in¯ orescences of Tillandsia deppeana in a deciduous forest near Xalapa, Vera Cruz, Mexico, enough to preclude seed production by some individuals. Elsewhere, capsules of tillandsias often exhibit little or no insect damage (e.g., T. balbisiana in southern Florida; personal observation). Unlike those voracious orthopterans in Florida, many of the invertebrate fauna that attack Bromeliaceae are relatively oligophagous, some exclusive to a few genera and occasionally a single species. Documentation includes the specialized behavior of Dynastor napolean, a butter¯ y native to southeastern Brazil. Larvae neatly trim and roll up the tip of each partially eaten leaf as if to confuse its own searching predators. A moth combines bromeliad tissue in an otherwise unexpected diet. The usually solitary Castine phalanis that survives its cannibalistic brood mates bores into the center of the hosting shoot, ultimately destroying it. Strymon basilides, another lepidopteran, feeds on developing bromeliad ¯ owers and fruits, sometimes in sufficient numbers to abort the entire in¯ orescence and occasionally challenge pineapple culture in Brazil. Beutelspacher (1972) and DeVries (1997) cite a variety of Tillandsioideae as hosts for the larvae of Riodinidae butter¯ ies (see also Chapter 12). Figure 8.2B illustrates the characteristic burrows of a leaf miner exclusive to the leaves of Aechmea bromeliifolia and an unidenti® ed Hohenbergia in southern Bahia State, Brazil in December 1996 (personal observation). Stranger still, the crab Metasesarma rubripes crops the ¯ owers of several bromeliads in some coastal habitats in southeastern Brazil (Fischer et al. 1997). Less conspicuous invertebrates also attack Bromeliaceae. Members of the symphylid genus Hanseniella cause problems for pineapple farmers in northern Australia (Paroz 1981). Symptoms include a characteristic witches' broom-type branching following destruction of the root tip. Many of the standard greenhouse-inhabiting Homoptera thrive on a variety of bromeliads, and a constellation of scale insects (e.g., Diaspis bromeliae, Hemiberlesia palmae, Saissetta hemisphaerica), and mealy bugs (e.g., Pseudococcus brevipes, Rhizoecus falcifer) can kill untreated specimens. The Cambridge Books Online © Cambridge University Press, 2009 Predators and pathogens 409 Figure 8.2. Herbivores and Bromeliaceae. (A) Bromelia sp. in Minas Gerais State, Brazil showing the impact of an unidenti® ed trichome-grazer. (B) Characteristic damage caused by a leaf miner on Hohenbergia sp. in Bahia State, Brazil. (C) Insect damage to spikelets of an unidenti® ed Hohenbergia sp. in Bahia State, Brazil. (D) Aphids on the buds of a cultivated Dyckia sp. (E) Extra¯ oral nectar on the calyx of the same Dyckia sp. featured in D. Cambridge Books Online © Cambridge University Press, 2009 410 Relationships with fauna Table 8.1. Partial list of pests that were not currently present or widely distributed in the United States intercepted on imported bromeliads by the United States Department of Agriculture Plant Quarantine Division inspectors during 1965–67 Plant pest Acutaspis tingi McKenzie Acutaspis umbonifera (Newstead) Arphnus melanotylus Ashlock Asterolecanium epidendri (Bouche) Atta mexicana (F. Smith) Cimolus vitticeps Stal Crophinus costatus (Distant) Dendrocoris variegatus Nelson Diabrotica porracea Harold Dysdercus mimulus Hussey Dysmicoccus probrevipes (Morrison) Exptochiomera albomaculata (Distant) Helicina zephyrina Duclos Melanaspis odontoglossi (Cockerell) Metamasius hemipterus hemipterus (Linnaeus) Metriona trisignata Boheman Mimosestes dominicanus (Jekel) Mormidea collaris Dallas Ochrimnus vittiscutis (Stal) Ogdoecosta biannularis (Boheman) Oplomus rutilus Dallas Paroecantus aztecus Saussure Procyrta intectus (Fowler) Statira denticulata Champion Tnethecoris distinctus Hsiao & Sailer Xenochalepus omogerus (Crotch) Family Country of origin Diaspididae Diaspididae Lygaeidae Coccidae Formicidae Coreidae Lygaeidae Pentatomidae Chrysomelidae Pyrrhocoridae Pseudococcida Lygaeidae Helicinidae Diaspididae Curculionidae Mexico El Salvador Honduras Costa Rica Colombia Mexico Mexico Mexico Honduras Mexico Mexico Mexico Mexico Jamaica Jamaica Chrysomelidae Bruchidae Pentatomidae Lygaeidae Chrysomelidae Pentatomidae Gryllidae Membracidae Lagriidae Miridae Chrysomelidae Guatemala Mexico Mexico Guatemala Mexico Honduras Mexico Guatemala Honduras Mexico Honduras Source: From Davidson (1969). nematode Tylenchocriconema alleni uses its piercing proboscis to feed on the mesophyll of young leaves while sheltered in the phytotelmata of Tillandsia flabellata (Lehman 1987). A mollusk apparently favors the trichome shields that densely invest the abaxial surfaces of the foliage of an unidenti® ed Bromelia in Minas Gerais State, Brazil (Fig. 8.2A). A more nitrogen-de® cient forage is hard to imagine. On balance, patterns of predation indicate diverse fauna, and preferences for speci® c organs and tissues in addition to narrow host ranges. Table 8.1 lists some known pests recorded on imported specimens; Table 8.2 itemizes tank inhabitants, many of which cannot be bromeliad feeders, Cambridge Books Online © Cambridge University Press, 2009 411 Predators and pathogens Table 8.2. Invertebrate taxa present in the tanks of bromeliads and on the forest floor at Rancho Grande, Venezuela Acari, Mesostigmata Acari, Orbatei Acari, Prostigmata Acari, Trombidiidae Araneida Blattodea Carabidae Carabidae, larvae Catopidae Chelonethida Chilopoda Coccinellidae Coleoptera, phytophagous larvae Collembola Curculionidae Dascyllidae, adults Dascyllidae, larvae Dermaptera Diplopoda Diplura, Campodeidae Diplura, Japygidae Diptera Brachycera larvae Diptera, Drosophilidae Diptera, Nematocera larvae Dytiscidae, larvae Enchytraeidae Formicidae Gastropoda Glossoscolecidae Glossoscolecidae, cocoons Gryllidae Hemiptera Hirudinea Histeridae Liodidae Lucanidae Nitidulidae Oniscoidea Pedipalpida Phalangida Pselaphidae Psocoptera Ptilidae Scydmaenidae Staphylinidae Symphyla Bromeliad Forest ¯ oor 4 0 0 0 12 15 4 0 0 0 13 2 3 2 0 0 66 0 68 0 0 1 0 37 1 0 456 2 12 14 1 29 4 0 0 0 0 57 0 1 2 0 3 0 5 0 7 4 2 1 17 4 1 1 5 7 24 0 10 2 1 1 0 3 46 3 6 4 2 1 0 2 64 1 0 2 2 8 0 2 4 1 1 32 2 3 23 1 1 15 26 11 Source: From M. C. Paoletti, personal communication; see also Paoletti et al. (1991). Cambridge Books Online © Cambridge University Press, 2009 412 Relationships with fauna recorded in situ. One migrant that fails to show up on the ® rst list warrants special note. A predator from Mexico is raising serious concern in Florida (Frank and Thomas 1994). Initial sightings occurred in Broward County in 1989, but USDA inspection records suggest that Metamasium callizona (Coleoptera) arrived at least a decade earlier, presumably aboard a shipment of horticultural stock from Mexico. Additional members of this weevil genus feed on Ananas and other bromeliads in tropical America, whereas more oligophagous M. callizona exhibits greater partiality for Tillandsia, particularly T. utriculata in the state of Vera Cruz, its type locality. The same behavior is expressed in the newly established population, and densities of affected T. utriculata in some parts of Florida now exceed those Frank recorded in Mexico. Gravid females chose relatively mature shoots equipped with a caudex at least 2 cm in diameter. Younger specimens are either too small to provide sufficient food, or they fail to protect the developing larvae. Small species may be wholly ignored for the same reason, but not so large specimens of Catopsis and Guzmania that have already come under attack in Florida (Frank and Thomas 1994). Slits made in the leaf bases prepare the host for egg-laying. Larvae require 8± 10 weeks to feed, after which pupation occurs over 18± 24 additional days. Florida' s strictly monocarpic T. utriculata may fare worse than its counterpart in Mexico where longer contact with this insect may account in part for the regular production of ramets (see Chapter 7 for additional reasons). In fact, infestations of T. utriculata and T. fasciculata had become severe enough by 1996 to prompt listings among Florida' s official endangered ¯ ora. Metamasium callizona continues to spread into central and southwestern Florida. As of September 1996, con® rmed reports included 10 counties. Although members of at least 12 more genera can be infected (Frank and Thomas 1994), collections of cultivated Bromeliaceae have so far remained mostly beetle-free, sometimes even where nearby trees host native Tillandsia utriculata harboring con® rmed infestations. Attempts are underway to introduce a parasitoid of Metamasium callizona to Florida. An undescribed Admontia sp. (Diptera) looks promising, but so far has proven difficult to rear in sufficient numbers to conduct the federally mandated, pre-release determination of its host range (Cave 1997). Bromeliads also support a number of dis® guring, usually sublethal, rusts (Lineham 1987) and additional, potentially devastating fungi. Commercial growers rely on spray programs that incorporate broad-spectrum fungicides to head off additional problems. For example, Puccinia tillandsiae Cambridge Books Online © Cambridge University Press, 2009 Predators and pathogens 413 grows on Tillandsia punctulata, and this same microbe parasitizes at least three relatives in Florida and Mexico. Puccinia pitcairniae attacks two and maybe more Pitcairnia species, and Uredo nidulaari infects Nidularium longiflorum. Helminthosporium rostratum acts as a leaf pathogen on Aechmea fasciata (Marlatt and Krauss 1974), which is the most widely marketed of the ornamental bromeliads. Phytophthora causes sufficient heart rot to require chemical control in some pineapple plantations. Anthropogenic in¯ uences may be intensifying certain natural challenges to some wild populations. Fusarium infects a number of bromeliads, including pineapple, and the same or another member of this genus may have contributed to the massive die-off of Tillandsia usneoides through much of central Florida during the early 1970s. Air pollution is probably slowing recovery in parts of that state, and could be contributing to on-going declines elsewhere (e.g., New Orleans area). Effluents from the large concentration of petrochemical facilities, possibly exacerbated by microbes and other agencies, may explain the almost total extirpation of Spanish moss to the west along the Gulf coast and north into the Big Thicket National Preserve (Chapter 5). Bromeliad viruses other than those that oblige meristeming to obtain uninfected pineapple stock receive no mention in technical print. Undoubtedly, many more, if not all, Bromeliaceae host these agents, but so far horticulturists seem mostly unconcerned. A rather lively and still unresolved debate (e.g., Mason 1976) once swirled around the presumed viral involvement in some of the more common leaf variegations, particularly the longitudinal striping that accounts for the popularity of numerous cultigens (e.g., Aechmea fasciata), and occurs sporadically through populations of certain wild types (e.g., Guzmania monostachia in Florida). Alternating horizontal bands of lighter and more deeply chlorophyllous tissue (e.g., Vriesea fosteriana, V. hieroglyphica) comprise another category of leaf ornamentations (Figs. 2.14G, 2.17B). These stable patterns may promote plant bene® ts by enhancing the hospitality of phytotelmata, and accordingly, the effectiveness of phytotelm architecture for plant nutrition, although other services are possible (Chapter 4). Less regularly distributed spots of variable dimensions and colors elsewhere (e.g., many Billbergia species; Fig. 2.14H) more closely resemble viral lesions. Small patches of chlorotic tissue on Billbergia lietzei and B. saundersii foliage, among other taxa, sometimes become suggestively necrotic. Undoubtedly, the enemies of the bromeliads far exceed those mentioned in this brief account. Today, we know too little about this subject to even attempt to address broad questions such as relative vulnerabilities to Books Online © Cambridge University Press, 2009 414 Relationships with fauna different kinds of pathogens and herbivores, and the nature and effectiveness of the corresponding plant defenses. A better test of the uniqueness of Bromeliaceae relative to use by predators and pathogens requires comparisons with other families (e.g., Agavaceae) of monocots that share similar habits and propensities for stressful habitats. Mutualisms Bromeliad-users beyond the herbivores and pathogens differ in degrees of dependence and the kinds of bene® ts received from and provided to their hosts. Most intimate are those fauna obliged to spend at least part of their life cycle associated with Bromeliaceae, sometimes speci® c taxa. A second, less dependent type need bromeliads for shelter or to feed, but not to reproduce. Still less plant-dependent are the facultative users. If aquatic, these organisms also inhabit phytotelmata provided by other vegetation; if birds they can nest elsewhere. Least bromeliad-dependent are the casual visitors, a group that includes the pollinators and seed dispersers, all of which forage for the same products and services across broad ¯ oras. We begin with the mammals and work down the phylogenetic scale to the exceptionally important ants and ® nally the litter-processing communities that develop in bromeliad phytotelmata. Mammals Several monkeys, marmosets and tamarinds regularly dismember vegetation to locate food in forest canopies. Several epiphytic Bromeliaceae are important enough for these animals to warrant special consideration in rescue efforts. For example, Leontopithecus rosalia, the golden lion tamarind of Brazil' s Atlantic coast forest, forages widely among the local phytotelm species to support its omnivorous diet. Several arboreal relatives also possess exceptionally long ® ngers, adapted perhaps to provide unusual access to desirable articles positioned among tightly overlapping leaf bases (E. Leme, personal communication). Cages provisioned with bromeliads collected from the same forests set aside as preserves in Rio de Janeiro State help condition reared animals for successful foraging following release (personal observation). Callithrix geoffroyi, the white-faced marmoset, uses Bromeliaceae in much the same way in Espirito Santo State (Leme and Marigo 1993). Other mammals that help regulate populations by eating the meristematic center of the shoot and dispersing seeds of the bromelioids include tree opossums and squir- Cambridge Books Online © Cambridge University Press, 2009 Mutualisms 415 rels, and other arboreal rodents (e.g., Fischer and Araujo 1995; Fig. 6.6). Freeze and Oppenheimer (1981) report that a number of monkeys feed on young in¯ orescences and destroy plants by eating the center of the shoot. Birds Nadkarni and Matelson (1989) recorded close ties between avifauna at Monteverde, Costa Rica and arboreal ¯ ora in both primary lower montane cloud forest and the epiphytes inhabiting the isolated trees left standing in cleared pastures. Bromeliaceae received more visits than even co-occurring Ericaceae, Gesneriaceae, Loranthaceae and Marcgraviaceae, which also offered especially attractive nectar or fruit, but fewer reasons overall for birds to spend time there. Bryophytes stood out as well, providing nesting material, and, judging by bird behavior, food, presumably sheltered invertebrates. Overall, 15% of the visits to tree crowns by 81 bird taxa recorded during 289 h of observations involved some type of epiphyte use. Perching occurred in 67% of all the encounters with canopy ¯ ora; feeding in 48%; vocalizing in 22%; and the gathering of nest materials, drinking and bathing in 2%. Nine bird species, including hummingbirds, ¯ ycatchers and the common bush tanager, paid some attention to epiphytes, often bromeliads, during more than half of their time in the canopy. Those users that showed the greatest interest rarely or never visited isolated trees despite the usually dense colonies of resident epiphytes high light favors on such supports (Fig. 1.4F). Birds with broader foraging patterns appeared about as often in dense forest as in degraded habitat. Sillett (1994) looked more closely at epiphyte use by eight insectivorous birds at a series of Costa Rican sites located between 2800 and 3100 m in the Cordillera de Talamanca. Tree crowns at these altitudes support the heavy loads of bryophytes characteristic of upper montane wet tropical forests, while phytotelm Tillandsioideae constitute most of the less abundant local vascular epiphytes. Feeding behaviors among the eight subjects varied along a continuum, with some species tied to a speci® c kind of substrate and others to certain prey that by distribution obliged wider searches through the canopy. All of these birds spent at least one-third of their time seeking food among the epiphytes, two (Margarornis rubiginosus, Pseudocolaptes lawrencii) more so than the others. Pseudocolaptes lawrencii devoted 74% of its time during Sillett' s observations to tank bromeliads, moving from plant to plant pulling out and tossing aside impounded debris to expose cockroaches and other invertebrates. Cambridge Books Online © Cambridge University Press, 2009 416 Relationships with fauna Troglodytes ochraceus alone worked the humus around the bases of these plants (37% of observations). Pizo (1994) surveyed the associations between birds and arboreal Bromeliaceae during several visits to Atlantic Forest at 600± 850 m in the Paranapiacaba mountains of São Paulo State, Brazil. Twenty-three species provided one or more kinds of resources, many identical to those Nadkarni and Matelson enumerated in Costa Rica, but selectivity often changed with the location. Eight birds foraged only through the understory (,10 m), six others consistently remained above this zone (10± 25 m), and 10 less fastidious species utilized Bromeliaceae throughout the forest. Cacicus haemorrhous (the red-rumped cacique) and Platypsaris rufus constructed nests largely of Tillandsia usneoides, while Thraupis cyanoptera and T. ornata used a variety of materials to raise young exclusively in the space located between a phytotelm shoot and the adjacent tree trunk. Epiphytes belonging to other families provided lesser services and accounted for fewer bird visits. Remsen (1985) also indicated that bromeliads are inordinately important to the organization of bird communities in the high forest of Bolivia. Brazilian Rhopornis ardesiaca may be the most plant-dependent of the birds observed using bromeliads so far (Leme and Marigo 1993). Large clumps of terrestrials that harbor abundant arthropods and provide favorable nesting sites accounted for most of its time in the liana forests of the southern part of Bahia State. Geranospiza caevulescens, the crane hawk, owes its common name to an extraordinary anatomical feature that facilitates bromeliad use (Bokermann 1978). This bird ranges from Mexico to Argentina aided in its searches for frogs and other tank fauna by long legs equipped with exceptionally mobile tarsal joints. Bromeliads sometimes belong to mixed assemblages of plants that share the same pollinators. Stiles (1978) described the phenology over four years of such a guild of ornithophilous plants in lowland humid forest at Finca La Selva, Costa Rica (Fig. 6.3). More than 40 participating taxa ¯ owered more or less sequentially to produce a near continuous supply of nectar for the local hummingbirds. Such arrangements could be coincidental, evolved in situ via character displacement, or established by ecological sortings of suitable genotypes among those already available (Chapter 6). Some of the bromeliads responded to weather in a way that promoted the welfare of local pollinators during drought. Rather than producing less nectar during the driest of the four years like the rest of the guild, Aechmea nudicaulis ¯ owered more profusely. In effect, the more stress-tolerant, if not stress-stimulated, taxa together maintained a relatively perturbation- Cambridge Books Online © Cambridge University Press, 2009 Mutualisms 417 Figure 8.3. Syncope antenori, a tadpole of a bromeliad-adapted frog from Peru (redrawn from Krügel and Richter 1995). resistant food source capable of sustaining a needed population of resident pollinators. Fischer and Araujo (1995) describe arrangements in coastal Brazil where sympatric Bromelioideae offer ripe fruit according to schedules that manipulate fauna to disperse seeds (Fig. 6.6). Anurans Plant-compatible morphology and reproductive strategy characterize the more bromeliad-dependent vertebrates and invertebrates. Bromeliocolous anurans conform by exhibiting one of two patterns: they produce either sedentary or free-swimming larvae. Gastrotheca fissipes incubates young on its back until legs develop. Tadpoles of several of the free-foraging species exhibit exceptionally slender bodies to negotiate the cramped quarters imposed by closely overlapped leaf bases (Rivero 1984, 1989). Other shapes exist to counter the effects of stagnation and limited food supplies. Larvae of Jamaican Hyla brunnea feature laterally narrow tails compared with nonbromeliocolous relatives, purportedly to encourage gas exchange in media largely stripped of oxygen by the decaying remains of its egg masses (Nobel 1929; Fig. 8.3). Vestigial mouths and guts accompany dependence on vitelli in lieu of feeding. Reproductive efforts and nursery volumes match, with fewer than 20 eggs per clutch in the more extreme species. Breeding space provided by Bromeliaceae varies in importance with the location. Fully seven of Puerto Rico' s 18 natives regularly use one or more phytotelm species to reproduce. Three members of Eleuthrodactylus, including E. jasperi, which is just 20 mm long and the only live-bearer in the western hemisphere, exhibit complete plant dependence. Billbergia zebrina of Brazil' s cerrado, and a less frequent epiphyte or lithophyte elsewhere, provides unusually favorable shelter for Hyla venulosa. Sticky latexlike secretions from its head allow this animal to block off the narrow phytotelma and further frustrate pursuers and avoid desiccation during the dry season. Cambridge Books Online © Cambridge University Press, 2009 418 Relationships with fauna Aparasphenodon brunoi and similar frogs employ a bony cranial carapace to put a variety of Brazilian Bromeliaceae to similar use. Protruding eyes retract into the oral cavity when danger threatens. Figure 8.4E,F illustrates a mature specimen of one of these animals resting in the central tank of an unidenti® ed Hohenbergia in Bahia State. Removed and placed on one of the leaf blades, this frog exhibited remarkable strength in its unusually short limbs while repeatedly trying to crawl back to its original refuge rather than jump on to nearby vegetation. The Costa Rican `poison arrow frog' Dendrobates pumilio demonstrates one of the most elaborate behaviors recorded for a bromeliad-user (Young 1979). Although terrestrial most of its life, this brilliantly colored animal relies on exceptionally complex behavior and impressive powers of navigation to reproduce exclusively in the phytotelmata of epiphytic Bromeliaceae. One by one, tadpoles hatched on the ground dutifully climb aboard their mother' s back for carriage to one of a series of nursery plants. Each deposit consists of a single larva and an unfertilized egg provided as food. Rearing involves repeated visits to the same leaf axils and additional sacri® cial eggs. The mother' s persistence is exceeded only by her uncanny ability to relocate each developing offspring. Peixoto (1977) documented visits by and breeding strategies of 26 hylid species representing eight genera in Atlantic Forest. Some patterns of reproduction (e.g., Gastrotheca fissilis) accord with constraints imposed by small volumes of water at breeding sites. Life cycles in other cases incorporate longer juvenile stages and corresponding requirements for extensive foraging and larger phytotelmata. Larval characteristics beyond the outsized tails that aid gas exchange include a depressed body, unusual dentition, and the absence of papillae on the anterior part of the upper lip. An undetermined number of frogs center their entire existence on Bromeliaceae, and at least a few of them remain faithful to a single plant (Abendroth 1971). Fritziana goeldii routinely breeds in Aechmea and Billbergia shoots in the Organ mountains of southeastern Brazil, where its larvae require just a few days to develop limbs. Tubular types (e.g., Aechmea nudicaulis, Billbergia vittata) rather than species with more spreading, shallow shoots (e.g., many local Vriesea species) attract the heaviest use. Several adults may spend the entire day virtually motionless in as many leaf axils in the same shoot. Each evening they depart to forage through the surrounding canopy. Less is known about the lizards (e.g., Abronia) and snakes (e.g., Bothrops schlegeli) that often visit bromeliads. None of these reptiles appears to be modi® ed for bromeliad use, nor do they spend as much time there as the more desiccation-prone amphibians. Cambridge Books Online © Cambridge University Press, 2009 Mutualisms 419 Figure 8.4. Biota in bromeliad tanks. (A) Unidenti® ed frog hiding by day in the leaf base of an unidenti® ed Billbergia sp. in Rio de Janeiro State, Brazil. (B) Trapbearing rhizomes of Utricularia humboldtii in the tank of Brocchinia tatei on Cerro Neblina, Venezuela. (C) Damsel¯ y recently emerged from the phytotelmata of an unidenti® ed Aechmea sp. in Espirito Santo State, Brazil. (D) Method used to remove the biota residing in the phytotelmata of Bromeliaceae. (E) Unidenti® ed frog in Bahia State, Brazil using its bony cranial plate to block off the central tank of an unidenti® ed Hohenbergia sp. specimen. (F) The same frog featured in E. Cambridge Books Online © Cambridge University Press, 2009 420 Relationships with fauna Salamanders Some salamanders rival the bromeliocolous frogs for dependence on phytotelmata, and here too plant shape matches the body form and reproductive biology of the animal. In fact, Bromeliaceae appear to have played a decisive role in the history of this group of unusually drought-vulnerable vertebrates. Distributions leave little doubt that bromeliad-provided cavities ® lled with moist, decomposing debris and abundant invertebrates helped this fundamentally Laurasian group make its single substantial incursion into the tropics (Wake 1987). About 80% of the more than 140 species in 11 genera of plethodontids occur in Mesoamerica and northwest South America where many of these animals remain con® ned with their supporting montane bromeliads to narrow altitudinal ranges. Diversity peaks in mid-elevation (750± 2500 m) cloud forest, most notably in Costa Rica. One transect extending from near sea level to 4000 m in southern Mexico yielded 15, mostly discrete, salamander populations, many exclusive to large Tillandsia and Vriesea shoots. Plethodontidae equipped to utilize Bromeliaceae differ from their more northerly counterparts by direct development (no larval stage) and exceptional drought-tolerance, in addition to the narrower distributions. Salamanders most often reported in bromeliad shoots belong to Dendrotriton, Nototriton and Chiropterotriton. Behaviors and life cycles distinguish these animals from those adapted to other substrates, but perhaps somewhat less so than some of the frogs. Nevertheless, shapes appropriate for negotiating the narrow spaces between overlapping leaf bases stand out at a glance. Adults typically possess small (,50 mm in length) trunks equipped with long prehensile tails, elongate limbs with widely separated digits, and frontally directed eyes. Phylogenetic studies indicate substantial homoplasy in those features associated with bromeliad use, possibly related to a limited number of niches for these animals in tropical America. Relatives that prefer suspended mats of mosses, higher plants with other shapes, and debris instead of phytotelma possess shorter appendages and trunks similar to those of the land-based forms. Little is known about the behavior and mobility of the bromeliad-dependent species. Gregariousness increases during dry weather; on one occasion up to 34 Dendrotriton xolocalcae specimens occupied a single bromeliad (Wake 1987). As many as one in two of the larger shoots sampled housed at least one salamander. Cambridge Books Online © Cambridge University Press, 2009 Ants and bromeliads 421 Invertebrates Bromeliaceae probably ranks highest in importance among the nonwoody families that promote the welfare of canopy-based fauna in tropical America, if only because so many of its members provide such extensive shelter from predators, drought, wind and torrential rains. Arboreal bromeliads also elevate carrying capacities simply by humidifying the canopy atmosphere and expanding and elaborating its surface (i.e., increasing its `fractile universe' ). Numbers of individuals and biomass identify the invertebrates as the major bene® ciaries. Phytotelm bromeliads, among other suspended ¯ ora, and the humus they impound promote conditions so broadly equable that earthworms and diverse other desiccation-prone soil fauna sometimes ascend trunks to visit or, if capable of reproducing there, to permanently reside above ground. In effect, soil-dwelling invertebrates representing numerous phyla, along with their often closely related, exclusively arboreal counterparts, utilize substantial portions of the forest canopy as an extension of what temperate biologists tend to consider habitat con® ned to the ground. Of all of these animals, the ants, owing to their abundance, social organization and diverse activities and diets, in¯ uence the greatest variety of events above and below the soil surface. Not surprisingly, more is known about the associations of these insects with bromeliads than about any of the many other relationships except those involving mosquitoes. Ants and bromeliads Bromeliads that encourage prolonged contact with useful invertebrates fall into four nonexclusive categories, the ® rst three of which involve ants. Participating plants can be considered ant-guarded, ant-gardened, antfed/ant-house providers, or the phytotelm types. Relationships of the second and possibly the third kinds also promote plant dispersal. A few preliminary remarks about ant biology prior to, and some additional comments after, describing these interactions will help explain why Formicidae alone among the insects engage in so many kinds of mutualistic, nonpollination relationships with Bromeliaceae. Eusociality and often broad diets grant the ants unparalleled ecological versatility and propensities to interact with vegetation in complex and diverse ways. Some of these partnerships exhibit sufficient intimacy to suggest coevolution. Arboreal ants native to the lowland humid forests of tropical America achieve remarkable densities despite possession of Cambridge Books Online © Cambridge University Press, 2009 422 Relationships with fauna powerful, piercing mandibles, high mobility (and accompanying energy requirements), and additional indications that prey rather than plant products constitute the primary nutritional base. Sometimes, ant abundance simply becomes too high and the masses of their bodies too great to impute strict carnivorous diets. In fact, vegetation, including many of the bromeliads, and diverse Formicidae associate more closely in some food webs than either mouth parts, behavior or trophic pyramids would predict, and for two reasons. Ants that farm Homoptera (Fig. 8.2D) do so to access phloem sap which is one of the richest and most easily digested of the edible plant products. While these arrangements can impose signi® cant burdens on supporting ¯ ora, net bene® t accrues to the host if the ants tending its aphids or scale insects also deter enough additional, more voracious herbivores. Many woody ¯ ora and some Bromeliaceae utilize similarly pugnacious, but less destructive, guards with rewards of extra¯ oral nectar or solid food on stems, leaves and developing ¯ owers and fruits (Fig. 8.2E). Less common, but no less revealing of the value ants represent as plant resources, are the myrmecodomatia produced to encourage nesting by potential defenders, and for the epiphytes, surrogates for diminished root systems (Fig. 8.5). Ant-guarded species A signi® cant number of the diverse epiphytic ¯ ora that routinely root in ant cartons produce extra¯ oral nectar or pearl bodies (lipid-rich, particulate food). However, no reports (e.g., Madison 1979) that describe these rewards mention a bromeliad, although important products may be seasonal and so far overlooked. Ants densely covered the young in¯ orescence of Aechmea angustifolia specimens that dominated large nests at Rio Palenque, Ecuador, but I was unable to get close enough during the brief opportunity to determine why. Some terrestrial, nonmyrmecophytic bromeliads de® nitely protect vulnerable organs with seasonally provided ant food elaborated speci® cally for that purpose. Galetto and Bernardello (1992) examined 20 Argentinian bromeliads in two genera and discovered that nearly half, speci® cally eight Dyckia species and a Deuterocohnia, attract ants with secretions vented on the abaxial surfaces of the sepals. Sugar was abundant, but contrary to most extra¯ oral nectars a single constituent, in this case sucrose, accounted for .97% of the total. Relatively high concentrations of amino acids (121± 975 mg ml21) accorded more with ant guards than pollinators as the targeted fauna. Nectaries in all nine cases belonged to the `formless' type and secreted Cambridge Books Online © Cambridge University Press, 2009 Ants and bromeliads 423 Figure 8.5. Four species of Tillandsia offering different amounts of space for nesting ants in the Sian Ka' an Biosphere Reserve (Quintana Roo State, Mexico). Section indicates how much of the lower shoot is available for occupancy. Percent values indicate frequencies that sampled plants housed ant colonies (after Dejean et al. 1995). sugary product through adjacent stomata rather than through special ducts like those featured in their much more elaborate and productive counterparts located in the gynoecial septa (Fig. 3.4I). Organs occurred more or less randomly, with one to ® ve droplets appearing on a single perianth member (Fig. 8.2E). Broader surveys will almost certainly reveal additional Cambridge Books Online © Cambridge University Press, 2009 424 Relationships with fauna ant-guarded relatives with similar arrangements. Koptur' s (1992) discovery of ants feeding on nectar presented on the primary bracts of an immature in¯ orescence of Tillandsia balbisiana in south Florida may document a pervasive phenomenon. Ant-nest garden species Descriptions of the gardens cultivated by carton-constructing ants appear elsewhere in this volume with regard to establishment, succession, plant nutrition and reproduction (Chapters 5 and 6; Fig. 8.1C). Nest-garden Bromeliaceae mostly belong to Bromelioideae, with just an occasional, possibly accidental, tillandsioid listed in enumerations of the plants that engage in these mutualisms. Werauhia gladioliflora roots in carton in southeastern Ecuador now and then (H. Luther, personal communication) as does Tillandsia fasciculata in Belize (Catling 1995). Features of the participating ants and the plants may militate against nest-garden status for members of Tillandsioideae. Dry fruits and plumose seeds generally offer no incentives for myrmecochory, although cartonassociated Dischidia species (Asclepiadaceae) prove that wind and ant carriage sometimes occur sequentially in the Old World tropics (Benzing 1990). Shoots that exceed roots as absorptive organs probably further predisposed Tillandsioideae to ant-house over ant-garden mutualisms. Warm, humid forests composed of many small trees on fertile soil support the highest densities of Neotropical nest gardens. For example, 34 discrete cartons, each sustaining a sizable ¯ ora tended by thousands of ants, occurred on a single 10 310 m plot in an Amazonian caatinga (Madison 1979), many inhabited by Aechmea mertensii, A. brevicollis and A. longifolius. Aechmea angustifolia predominates among ant-cultivated bromeliads at Rio Palenque, Ecuador (Fig. 8.1C); Aechmea tillandsioides var. kienastii achieves comparable status in parts of Central America and in Trinidad (Pittendrigh 1948; Catling 1995, 1997). Aechmea tillandsioides warrants special mention for its exceptional attractiveness to ants even in the greenhouse. Interest in its infructescence, perhaps primarily as a site for farmed Homoptera, sometimes culminates in the production of protective cartons that cover many of the junctions between adjacent ¯ oral bracts (Rivero and Barard 1983). Pittendrigh noted in his classic study of bromeliad ecology (Pittendrigh 1948) that Citrus trees in a northern Trinidad grove each supported 3± 10 discrete ant-nest gardens, many occupied by Aechmea mertensii. Ant-nest gardens associated with a discrete colony, or fragments of larger, polygy- Cambridge Books Online © Cambridge University Press, 2009 Ants and bromeliads 425 nous populations of Camponotus femoratus, occurred on 16± 39% of the trees in ® ve Peruvian sites supporting different kinds of forest (Davidson 1988). Highest densities characterized seasonally ¯ ooded, open communities. Exposed microsites receive the greatest attention from ants either because the host trees there produce relatively abundant ant food ± a possibility explored more fully below (Kleinfeldt 1978; Davidson and Epstein 1989) ± or because the ants avoid locations too wet or cool to raise their brood. Nest-garden builders include members of Anochetus, Azteca, Camponotus, Crematogaster and Dolichoderus. All of these ants probably consume substantial honeydew to obtain the energy needed to construct extensive cartons. Even so, nest occupancy is often shared (parabiotic; a poorly understood phenomenon), with up to three species partitioning common living space. Extremely aggressive Camponotus femoratus and much smaller Crematogaster linata parabiotica usually share Amazonian gardens in eastern Peru (Davidson 1988). Both species contribute to nest construction. The latter builds the characteristically thin layers of carton over runways, nest sites in crevices, and long-term food sources such as extra¯ oral nectaries and Homoptera colonies. The former enlarges some of these shelters with decaying leaves and other detritus improving the site for roots. Any third species is usually smaller still (e.g., Solenopsis spp.) and likely parasitic on its larger-bodied house mates (Wheeler 1921, 1942). Camponotus femoratus alone accounted for the thousands of seeds encountered in the Peruvian nests it occupied with Crematogaster linata parabiotica. Densities and behaviors better indicate the importance of the nest-gardening ants than does the abundance of their constructions. Ants known to create nest gardens dominated the arboreal fauna at Tambopata, Peru (Wilson 1987). Individual tree crowns fogged with insecticide in four types of lowland forest yielded impressive numbers of taxa ± 43 species representing 26 genera in one sample alone, more than the entire ant fauna native to the British Isles! Crematogaster linata parabiotica, here the nest companion of Camponotus femoratus and Monacid debilis, occurred in almost half of the 513 samples. Wilson attributed the great success of the ant-nest garden species at Tambopata to their ability to produce capacious nests aided by symbiotic epiphytes, including several nest-dependent bromeliads. Catling' s (1995) inquiries on ant-nest gardens in the Stann Creek District (tropical moist forest) of Belize merit special attention because he addressed some new questions and a bromeliad ranked as one of the major players. Citrus orchard was chosen for the survey because of its simplicity Cambridge Books Online © Cambridge University Press, 2009 426 Relationships with fauna compared with wild-type forest. Catling addressed three issues: the ® delity of nest-garden taxa to ant carton, plant succession on that substrate, and the possible in¯ uences of the nest-builder, which was an unidenti® ed Azteca species, on the distributions of the recognized ant-dependent and other local arboreal ¯ ora. A total of 288 nests located in the crowns of 73 trees comprised one set of samples. Five of the twelve (of which three were bromeliads) local nestgarden species infrequently (,5%) or never rooted on any other than antprovided substrates. Of these ® ve populations, just two, the orchid Epidendrum immatophyllum and Aechmea tillandsioides var. kienastii, occupied more (,60%) of the censused gardens than the others (,40%). Although entirely con® ned to cartons, A. tillandsioides var. kienastii seemed to require site preparation by Epidendrum immatophyllum, or more speci® cally its extensive root system, for establishment. Later, as the ant nest expands (ages), the frequency of the orchid' s presence and its contribution to total plant cover on a carton decrease relative to that of the bromeliad. Several other nest-garden species (e.g., Coryanthes speciosa and Codonanthe macrodenia) exhibited similar patterns of occurrence apparently also related to requirements for speci® c exposures. Like Epidendrum immatophyllum, they yielded space to the bromeliad as shared nests grew. Five sites provided the circumstances Catling needed to pursue the third question ± whether Azteca sp. in¯ uences the distribution of arboreal ¯ ora in the sampled orchards. Eighty trees with and 175 more without ant nests supported one or more of the surveyed epiphytes. Yates chi-square and analysis of variance indicated higher plant diversity in the crowns of antoccupied trees, but only because ® ve of the commonest epiphytes rooted exclusively in nest gardens. Ants appeared to in¯ uence the welfare of only two of the remaining 27 epiphytes. Ionopsis satyrioides and I. utricularioides, both short-cycled twig orchids, occurred less commonly in crowns with compared with those without resident ant colonies. Catling made no attempt to determine whether these ants, like some others that inhabit the shoots of several Mexican bromeliads (e.g., Aechmea bracteata), reduce herbivory on the tree or for the associated epiphytes (Dejean et al. 1992). Except possibly for the two twig orchids, this Azteca species, unlike an arboreal Crematogaster species in Malaysia (Weir and Kiew 1986), does not remove epiphytes from the trees that support its nests. Ant-fed, ant-house species Ant-fed, ant-house Bromeliaceae outnumber those that root in cartons. Moreover, the variety of zoobionts involved is also greater in accordance Cambridge Books Online © Cambridge University Press, 2009 Ants and bromeliads 427 with the greater complexity and higher costs of the second form of myrmecotrophy for the ants and possibly the plants (Benzing 1991; Figs. 2.2E, 8.5). Farming involves elaborate social behavior in addition to carton construction, which in turn oblige relatively rich diets. Many of the same fauna that nest in myrmecodomatia just as readily accept hollow twigs, cavities under loose bark, or rolled-up dead leaves in addition to the plant organs apparently evolved to attract them. However, enticements vary, and no records suggest that any of the bromeliads equals certain other ant-house ¯ ora (e.g., Myrmecodia; Eshbaugh 1987; Davidson and Epstein 1989; Jebb 1991) for specialization for ant use. Nor do they associate as regularly with speci® c zoobionts. On the other hand, more species of bromeliads than members of any other family may pro® t by accommodating nesting ants. Genera (e.g., Hoya, Hydnophytum) with recognized myrmecotrophic members typically contain few species, or if larger include only the exceptional population unequivocally modi® ed to host nesting ants (Huxley 1980). Bromeliaceae follow suit. Conspicuously ant-adapted Tillandsia bulbosa, T. butzii, T. caput-medusae, T. seleriana and some similarly con® gured relatives (Fig. 8.5) constitute but a small fraction of this largest of all the bromeliad genera. Many additional Bromelioideae and Tillandsioideae and at least one Pitcairnioideae (Brocchinia acuminata; Fig. 2.2E) also house ant colonies, but at unknown frequencies and with undetermined consequences for plant welfare. Several bromelioids (e.g., Aechmea brassicoides, A. bracteata, A. setigera, A. melinonii; Fig. 2.4G,L) possess unusual, bulbous phytotelm shoots in which ants often raise brood. Figure 8.1D illustrates predominantly lithophytic Aechmea phanerophlebia in Minas Gerais State, Brazil partially dissected to expose pupae and adults among the upright leaves. Virtually all of the Amazonian Bromelioideae harbor ant colonies along with impounded moisture in much the same fashion, perhaps re¯ ecting high densities of ants that render dry nest space an especially scarce and heavily exploited commodity. Dejean and Olmsted (1997) identi® ed four kinds of living space provided by the older ramets of Mexican Aechmea bracteata (Figs. 2.4G, 8.1B). Young specimens of Aechmea bracteata develop water-® lled leaf axils ® rst, after which the possibilities for housing multiply. Flowering is preceded by production of a single inrolled leaf whose amphora-like shape excludes precipitation and probably most falling litter (Fig. 2.4G). Dead, post-fruiting ramets open up still another kind of space until plants that bear shoots representing every stage of development offer a variety of kinds of chambers to fauna with different needs for moisture, food and securement. Bases of the outermost foliage of mature, living shoots Books Online © Cambridge University Press, 2009 428 Relationships with fauna intercept some precipitation but soon dry out. Better-insulated, somewhat younger leaves within the shoot maintain deeper, more permanent phytotelmata. Chambers created by still younger and upright organs sequester small amounts of moisture and debris, and the central cavity remains completely dry. Fine-grained distributions within forests suggest that the ant-fed, anthouse bromeliads are constrained less by energy requirements than are their nest-dwelling relatives. Tillandsia bulbosa and T. butzii sometimes reproduce in considerable shade, equipped in part for dark habitats by ¯ at, rigid trichome shields that, uncharacteristically for a Type Five bromeliad, transmit instead of scatter photons whether wet or dry (Benzing et al. 1978; Fig. 4.23F,G). These same appendages also shed rather than imbibe moisture, reducing the threat of suffocation which excludes the more typical Type Five Tillandsioideae from overly humid ecospace (Fig. 4.11; Table 4.8). Questions worth pursuing about these myrmecotrophic bromeliads concern possible parallels with more specialized, better-known systems. For example, do plant vigor and certain qualities of the microsite in¯ uence which species of ants occupy speci® c bromeliads, and whether certain occupants provide better protection or more abundant plant nutrients than others? Jebb (1991) reported 19 ant species belonging to 14 genera nesting in Papua New Guinea Hydnophytinae. Weaker-growing specimens in shade harbored the greatest diversity of mostly timid ants, whereas betterexposed specimens usually supported comparatively aggressive Iridomyrmex species. Different ant behaviors and plant bene® ts may also provide multiple options and mediate site-speci® c outcomes in tropical America. Seventeen species of ants representing 11 genera in four subfamilies utilized the leaf base chambers of Tillandsia bulbosa in Quintana Roo State, Mexico (Olmsted and Dejean 1987). No mention was made of any correlations between plant exposure and occupancy by speci® c symbionts. Rates of occupancy indicate importance to the ants and plants that engage in myrmecotrophic relationships, and accordingly, considerable impetus for certain ¯ ora to evolve myrmecodomatia and other enticements for ants. Well over half of the Tillandsia butzii and T. caput-medusae specimens observed in Mexico and Costa Rica contained brood (Benzing 1970a), and the pseudobulbs of every Schomburgkia tibicinis specimen cut open by Rico-Gray and Thien (1989) housed an active colony. No bromeliads produce seeds equipped with recognized elaiosomes, nor do any of the ant-fed species supply an alternative source of nutrients dedicated to symbiotic fauna. However, pulp that adheres to the seeds of myrmecotrophic Cambridge Books Online © Cambridge University Press, 2009 Ants and bromeliads 429 Bromelioideae may still bene® t the ants that nest in these species or cultivate them in cartons as discussed below. Sometimes a phorophyte bene® ts by hosting as few as one ant-house bromeliad. Occupied trees in seasonally inundated and more elevated semievergreen woodland in the Sian Ka' an Biosphere Reserve in Quintana Roo State, Mexico exhibited little damage attributable to the local leaf-cutting ants (Atta spp.) compared with uncolonized conspeci® cs in the same forests (Dejean et al. 1992). Atta cephalotes more heavily cropped Bursera simaruba trees free of ant-inhabited Aechmea bracteata or Schomburgkia tibicinis specimens, enough in some instances to reduce the total leaf area by more than half. Hypoclinea bispinosa most often occupied these epiphytes, but additional, less common ants (e.g., Azteca sp., Neoponera villosa) also probably deter leaf cutters. Other colonies relegated to dead branches in crowns devoid of bromeliads and orchids provided similar protection to trees representing seven species, not including Bursera simaruba. Additional myrmecotrophs, especially Tillandsia balbisiana, T. bulbosa, and T. streptophylla, also appeared to be protecting their supports against the potentially devastating attacks of a common chrysomelid beetle. Much of the canopy of one phorophyte appeared to owe its undamaged condition to a colony of a small, unidenti® ed ant housed in the shoot of a single bromeliad. Outcomes were impressive enough to prompt Dejean et al. (1992) to propose the use of transplanted ant-inhabited epiphytes to protect orchards that can be more attractive to leaf cutters than the average tree in mixed forest. At the same time they cautioned against the possibility that ant-tended Homoptera on the same bromeliads and orchids might spread pathogens (e.g., viruses). Aechmea bracteata warrants special note among the ant-assisted Mexican epiphytes for its similarity to Aechmea phanerophlebia, particularly the vase-shaped shoot (Figs. 2.4G, 8.1D), and the exceptional opportunity offered to symbionts by the four kinds of living space. Diverse, often aggressive, ants number among the usual occupants (Dejean et al. 1995; Dejean and Olmsted 1997). Surveyed plants mostly sheltered ant colonies, some comprised of thousands of workers. Pachycondyla villosa and Hypoclinea bispinosa predominated, but six other species in four more genera occasionally also quartered there. An even richer ant fauna inhabited the spaces among the outer, dead leaves of 97% of the dissected plants. Bene® ts to trees gained from the presence of ants that discourage herbivores as destructive as the leaf cutters would exceed just about any imaginable liability imposed by the presence of the few epiphytes necessary to make this service possible. Cambridge Books Online © Cambridge University Press, 2009 430 Relationships with fauna Dejean and Olmsted (1997) documented succession and related partitionment of Aechmea bracteata by fauna on the Yucatán peninsula. In all, 91.5% of 248 plants harbored one or more species of ants, and many additional symbionts were also present. Larvae of Odonata and Diptera occupied green shoots only, whereas the other large invertebrates there were phytophagous (e.g., caterpillars, Tettigonidae). Dead post-fruiting ramets supported mostly detritivores (e.g., Ascaridea, Diplopoda, Isopoda, Thysanura, Collembola termites, Coleoptera) and their predators (e.g., Aranea, Phalangidae, Chilopoda, certain ants). Occupancy varied with the type of habitat (three were surveyed), especially among the ants. Ants also segregated among the spaces provided by the epiphyte according to body size, diet, competitiveness and tolerances for neighbors. Central cavities in green ramets usually sheltered relatively large aggressive ants (mostly Pachycondyla villosa at ® rst and then Dolichoderus bispinosus as plants aged). Azteca spp. and Dolichoderus spp. built carton partitions there, and Pachycondyla villosa put rough plant debris to the same use. Fewer species of ants (14 vs. 16) occupied the narrower, older leaf bases and did so less consistently (31.3 vs. 91.5%). Inhabitants also tended to be smaller (e.g., Monomorium ebeninum). Somewhat fewer ants (21 vs. 25 species) occupied dry ramets, but these included some fungus-cultivating types (e.g., Cyphomyrmex minutus) that apparently use scraps provided by other ants to grow mycelia. Predators specialized for speci® c prey (e.g., Leptogeny spp. for isopods, Odontomachus bruneus for termites) were also present, along with many generalists (e.g., Crematogaster sp., Tetramorium simillinum, Wasmania auropunctata). Certain species tended to co-occur (e.g., Monomorium ebeninum and Camponotus abdominalis), and did so intimately enough to suggest parabiosis comparable to that seen in certain ant gardens (e.g., Wheeler 1921; Davidson 1988). Dejean (1990) uncovered intriguing evidence that chemotaxis and molecular imprinting operate during the selection of either Aechmea bracteata or Schomburgkia tibicinis by Pachycondyla villosa in the Sian Ka' an Biosphere Reserve, Mexico. Experiments involved batches of workers and winged females reared in vessels provisioned with leaf tissue harvested from either of these two sympatric epiphytes. Females obliged to imprint on one or the other host as larvae, upon maturation sought nesting sites in tubes containing tissue of the same bromeliad or orchid. Animals raised in the absence of reference plants attempted to found colonies irrespective of the kind of leaf tissue they encountered in the arti® cial myrmecodomatia. Cambridge Books Online © Cambridge University Press, 2009 Evolution of ant/plant associations 431 Evolution of ant/plant associations In addition to ¯ exible diets and sociality, potential year-round activity and the division of labor between reproductive and sterile (worker) castes insured that the ants above all other insects would evolve the most specialized and multifaceted associations with plants (Davidson and Epstein 1989). At this point, we still know too little about the life histories and the other biological phenomena that underlie ant/plant mutualisms to infer how the many kinds of partnerships arose. But no single explanation could suffice for every example anyway, even among the myrmecophytic bromeliads. Almost certainly, different circumstances and pathways account for the diverse combinations of participants and the varied materials and services exchanged as consequences of these intimacies. Substantial homoplasy characterizes the ways that ants and plants utilize one another, but partnerships seldom exhibit the taxonomic speci® city that often signals prolonged interdependence. Davidson and McKey (1993) cite frequent shifts among the ants and the ¯ ora predisposed, or already modi® ed, to support them to explain the unexpectedly low incidence of coevolution and cocladogenesis among pairs of interacting ants and plants. Outcomes for the individual ant-house epiphyte can be conditional, ranging from positive to negative depending on the behavior of its zoobionts (e.g., do they farm Homoptera but simultaneously deter more costly predators?). More remote variables, like the quality of the microsite, also in¯ uence the net effects of symbiotic ants on plant welfare (Davidson and Epstein 1989). Several facts concerning ants and plants should guide the formulation of hypotheses that address the less obvious aspects of myrmecophytism, especially origins and selective advantages. For example, nest-garden and anthouse status almost certainly evolved repeatedly among Bromeliaceae, and probably also within some of its genera, but not with equal frequency through the family. Members of Bromelioideae and perhaps a few Tillandsioideae root in cartons, while populations representing all three subfamilies engage in ant-house arrangements. Because bromeliads are small, slow-growing herbs unable to offer high caloric rewards, the ants that inhabit them tend to be opportunistic nesters, thus relatively indifferent to host identity. Whereas the simple presence of an ant colony assures nutrients for the ant-house bromeliad and the occupant of a nest garden, protection from herbivores is not ipso facto an accompanying plant bene® t. Ant/plant mutualisms succeed in part because much of the challenge to the long-term viability of the insect colony, and accordingly the bene® ts Cambridge Books Online © Cambridge University Press, 2009 432 Relationships with fauna provided to associated ¯ ora, is countered by an ample supply of replaceable workers. Thus the inclusive group, most importantly its defended queen(s), achieves considerable immunity from predators and potentially a long life ± often as long as, or longer than, that of the plants they utilize and nurture. Evolutionary access to this kind of mutualism probably depended on a variety of predisposing ant and plant characteristics, some obvious and others more difficult to recognize. Ancestors of the ant-house bromeliads probably also produced cavities among imbricated leaf bases, as do all but Type One, Two and some Type Five species; antecedents of the nest-garden types likely offered attractive seeds, although not necessarily typical myrmecochores with edible appendages (Chapter 6). Rooting media sufficiently impoverished to assure that ant-provided nutrients would enhance growth were probably also necessary to promote ant-house status involving elaborate myrmecodomatia and abundant, plant-provided ant food. The same stock had less to gain from adopting ant dispersal as discussed below. Evidence provided by plant morphology and chemistry, ant behavior and the obligatory nature of some of the alliances, particularly the nest-garden systems, suggests diffuse coevolution between certain Formicidae and the bromeliads they feed and sometimes protect. Tighter alliances, i.e., species-speci® c combinations, are improbable, and more so for the ants than for the plants. Camponotus femoratus, one of the commonest Amazonian ant-nest garden builders and a member of a fundamentally arboreal genus, seldom if ever occurs unaccompanied by cultivated plants. But which members of the co-occurring nest-garden ¯ ora it chooses seems to make no difference (Davidson 1988). Securement of a nest by roots, and the elimination of excess moisture by transpiration (Yu 1994), may occur irrespective of the makeup of the garden. Less ¯ exibility characterizes the other side of the relationship; certain nest ¯ ora, including some bromeliads, root nowhere but on cartons. Facultative combinations suggest beginnings for ant-house status ± for example those between a variety of bromeliads lacking elaborate modi® cations to accommodate brood and primitive Formicidae with nonspeci® c housing needs (e.g., Pachycondyla and Odontomachus). Frequent associations between opportunistic, docile occupants and only marginally bulbous tillandsias indicate broad proclivity among arboreal species to utilize plant cavities. For example, mostly feral ants occupied 10± 15% of the Tillandsia paucifolia shoots surveyed in Florida (Fig. 6.7; Benzing and Renfrow 1971a). A group of structurally distinct relatives illustrates an even more persuasive sequence in Quintana Roo State, Mexico. Cambridge Books Online © Cambridge University Press, 2009 Evolution of ant/plant associations 433 Olmsted and Dejean (1987) listed ® ve co-occurring bromeliads (four of the ® ve are illustrated in Fig. 8.5) in order of water-tightness and the volumes of space comprising the bulb chambers (numbers in parentheses denote frequency of ant occupancy): Tillandsia flexuosa (0%); T. baileyi (30%); T. balbisiana (42%); T. bulbosa (41%); and T. streptophylla (53%). Aechmea bracteata, A. phanerophlebia (Figs. 2.4G, 8.1D) and those numerous, similarly constructed, largely Amazonian (e.g., A. bromeliifolia), water-impounding and ant-accommodating Bromelioideae constitute end products of a similar evolutionary progression based on the same fundamental rosulate architecture, mutual bene® ts, and environmental context (Fig. 2.4). Ants offer myrmecophytic Bromeliaceae up to three direct services; how many apply in each case is difficult to determine. Corresponding plant requirements also range from one to three depending on the nature of the interaction and the environment; speci® c enhancements probably also vary somewhat among individuals comprising a population, especially those bromeliads engaged in the ant-house syndrome as already discussed. Only the nest-garden forms require ant assistance for seed dispersal. Whether the two other services (feeding and protection) follow must be demonstrated. Ant-house and nest-garden bromeliads seem to perform well enough in conventional culture, in effect growing at normal rates without assisting ants. De® nitive tests for predator deterrence should be less challenging than attempts to demonstrate any peculiarity of nutrition resulting from longterm association with ant-provided substrates. Massive defoliation following removal of especially pugnacious ants demonstrated the primary bene® t of myrmecophytism for the bullthorn acacias (Janzen 1966). Antnest gardens could also be fumigated, and the bromeliads hosting ant colonies similarly deprived of any ant-mediated protection from herbivores. Recall that Yu (1994; Chapter 4) noted the physical collapse of cartons after clipping eliminated the capacity of associated ¯ ora to dissipate excess moisture by transpiration. Myrmecochory enhances bromeliad ® tness in the forest canopy to the extent that the ants also provide the required rooting medium. Likewise, ¯ ora that root more pervasively should be poorly positioned to bene® t from the attentions of seed-dispersing ants. Compared with the nest-garden types, bromeliads that provide ant housing and anchor on bark can broadcast their offspring more widely owing to the less exacting nature of their mutualists, none of which produce extensive cartons (rooting media). Not surprisingly, Formicidae that colonize cavities through the forest canopy, Cambridge Books Online © Cambridge University Press, 2009 434 Relationships with fauna including those provided by a variety of nonmyrmecophytic plants, usually ignored the seeds of several ant-nest epiphytes during feeding trials in Peru (Davidson et al. 1990). The existence of so many ecologically similar, ant-assisted and nonmyrmecophytic Bromeliaceae raises questions about the value of ant services. Why, for example, are so few species taking advantage of the pervasive availability of potential mutualists? Recall that close relatives of many of the ant-house bromeliads typically co-occur (Chapters 5 and 7). Major differences in vigor or fecundity, hence nutritional requirements, appear unlikely to distinguish members of the two groups. Conceivably, closed rather than open leaf axils enhance plant success for other reasons, perhaps by increasing drought-tolerance. Then again, closer inspection of the ants that regularly exploit these cavities might reveal as yet unnoticed services such as the removal of mites or fungal spores. We could just as reasonably question why only a dozen or so phytotelm Bromeliaceae beyond Aechmea bracteata combine ant occupancy with capacity to process impounded litter for nutrients (Fig. 2.4G). More attention should be devoted to the ants that engage in long-term mutualisms with bromeliads. Diet, aggressiveness, vigorous colony growth, carton manufacture, and propensities to disperse seeds probably predisposed a subset of arboreal ants to nest in or cultivate certain bromeliads and the other nest-garden ¯ ora (Davidson and Epstein 1989). However, none of these attributes represents more than embellishments of more pervasive conditions. Characteristics that determine whether a given species farms Bromeliaceae on carton or more casually adopts interfoliar cavities to raise young are probably based on social organization and physiology that foster suitability for manipulation by ¯ ora. Experiments performed by Orivel et al. (1998) demonstrated that nest gardening may constitute a more fundamental feature of Formicidae than previously assumed; certainly it is not exclusive to the primarily arboreal taxa (Dolichoderinae, Formicinae, Myrmicinae) characterized by welldifferentiated caste structure, polygyny and colonies served by large numbers of exceptionally aggressive foragers. Pachycondyla goeldii and Odontomachus mayi (Ponerinae), two relatively primitive species (monogynous, monomorphic worker caste, relatively small colonies) thought to be no more than opportunistic occupants of abandoned gardens (Davidson 1988; Davidson and Epstein 1989), in fact harvested provided Aechmea mertensii and Anthurium gracile seeds and planted them in carton constructed by the same ants at a site in French Guiana. Nest gardens resulted. Tests designed to demonstrate preferences for the seeds of these two epi- Books Online © Cambridge University Press, 2009 Evolution of ant/plant associations 435 phytes over those of a variety of other species that never root in carton proved inconclusive. As the consummate chemical communicators, some of the more advanced ants may require precursors from plants to produce certain pheromones or other bioactive compounds. Possibly the same end products occur in both the ants and associated ¯ ora. Either way, the stage would be set for a plant to exploit the susceptible ant. Discovery that the seeds and sometimes the fruits of 10 nest-garden species native to Peruvian Amazonia contain high concentrations of the same aromatic compounds raises a number of questions that bear on the nature and origins of the nestgarden phenomenon (Davidson and Epstein 1989; Davidson et al. 1990; Seidel et al. 1990). The most likely candidate for chemical cueing for myrmecochory is methyl-6-methylsalicylate (6-MMS), a tenaciously held constituent that potentially in¯ uences ant and plant success after the seed separates from the rest of the fruit. Seeds of Aechmea longifolius continued to read positive for bioactive aromatics after repeated washing with pentane. Especially provocative was the discovery that the mandibular glands of male Camponotus femoratus contain the same compounds, most notably 6-MMS, as the seeds of the tested ant-garden ¯ ora, and that these components were probably endogenous rather than plant-derived. At least two of the ® ve compounds exhibit fungistatic activity indicating potential to protect brood. Interestingly, Camponotus lacks metapleural glands, which in some other taxa produce useful antiseptic secretions. Chemicals associated with seeds that today help maintain sterile cartons might have acted earlier to counter pathogens among the abundant micro¯ ora present in the moist humus that many epiphytes utilize for roots. Prolonged occupancy of a carton by its builders may oblige the use of some external source of an inhibitory compound, perhaps impregnated seeds, to avoid brood-destroying fungi. An alternative hypothesis posits mimicry as the mechanism that moves seeds from one ant-provided substrate to the next. Seed size and color suggested to Ule (1906) and Madison (1979) that physical resemblance helps disperse some of the Amazonian bromeliads. Volatile chemicals may enhance this relationship or even mediate the deception unaided by visual or nutritional cues (Davidson and Epstein 1989; Davidson et al. 1990; Seidel et al. 1990). Insects that parasitize ant colonies gain entrance by manipulating ant behavior with fragrances so why not plants, including the nest-garden bromeliads? Cambridge Books Online © Cambridge University Press, 2009 436 Relationships with fauna Termites Two additional groups of social insects associate with the bromeliads, one quite unexpectedly, the other more casually although perhaps to advantage if the adjacent plants also harbor colonies of pugnacious ants. Wasps often nest near vegetated cartons in Ecuador, where together with the ants they pose an exceptionally formidable challenge to marauding vertebrates, including curious biologists. Stranger still, termites often seek out established bromeliads, both on the ground and in trees (Fig. 8.1A,B,E). Termite nests and covered trails typically support no rooted ¯ ora, consistent with the termite diets and the hardness of their cartons (soil and partially digested wood) compared with those manufactured by ants. Thorne et al. (1996) noted that many of the trees on Guana island in the British West Indies supported covered trails engineered by Nasutitermes acajutlae leading to the leaf axils of large Tillandsia utriculata specimens. Younger plants not yet able to impound precipitation escaped attention. Covered pathways mapped on 97% of 115 trees intersected at least one adult bromeliad. Tunnel-building occurred at night, and progressed from the epiphyte to the ground rather than the reverse, apparently because the plant offered a superior supply of the moisture required to construct carton. Bene® ts to the termites are obvious, but not so those for the plant. Also unclear was the consistency of the association from year to year. Rainfall during the 24 months preceding Thorne et al.' s observations had been unusually light on Guana island, and the resulting aridity perhaps sufficient to force the local termites to seek unusual sources. Moisture is an especially scarce commodity where precipitation quickly percolates into the porous limestone beneath the island' s shallow soils. Although Nasutitermes acajutlae exceeds many of its relatives for water economy, diminished populations indicated considerable vulnerability to extreme drought. Revisits during wetter seasons and observations elsewhere would help settle the issue of regularity, as would closer inspection of the bromeliads on the question of plant welfare. While thirsty insects probably consume little of the hundreds of milliliters of moisture impounded in a typical shoot, termite earthworks may pose a greater threat as occasional trails were moist up to 30± 35 cm beyond the contacted shoots. Other termite constructions covered rather than drew water from leaf axils, possibly insulating the supply for these plants. Unidenti® ed termites also regularly associate with terrestrial and lithophytic Bromeliaceae and other low-growing ¯ ora in the campos rupestres habitats of Minas Gerais State in southeastern Brazil. Figure 8.1A illus- Cambridge Books Online © Cambridge University Press, 2009 Phytotelm bromeliads 437 trates Aechmea phanerophlebia anchored to rock below which a termite trail extends to the ground. Mounds of inhabited carton occurred more commonly around soil-rooted Dyckia and Encholirium specimens (Fig. 8.1D,G), among other local perennial herbs (e.g., Velloziaceae). Circumstances here also suggested opportunistic behavior effected by termites seeking the structural strength offered by an established plant, or, in the case of the spiny-leafed Pitcairnioideae, perhaps a deterrent to local ant-eaters. Bene® ts for Bromeliaceae could include nutrients and moisture on relatively rocky, well-drained soils, or insulation against the ® res that regularly sweep across these hyperseasonal habitats. Conversely, vulnerability could be high where litter rather than wood-feeding species are involved. Termites were abundant among the bases of the shoots of an unidenti® ed Alcantarea population growing on otherwise bare rock in Bahia State, Brazil (personal observation). Depending on diet, these insects could either help dislodge these unusually heavy, long-lived plants, or reduce the threat of pathogens by preventing the accumulation of too much dead leaf tissue. Dejean and Olmsted (1997) provide the most detailed account of how termites interact with a bromeliad, in this case with Aechmea bracteata (Fig. 8.1B). Populations native to inundated forests, and to lesser extents nearby upland sites in certain areas of the north coast of the Yucatán peninsula, afford termites, in addition to the ants already mentioned, space to raise brood in living and spent shoots. Green ramets sheltered only Nasutitermes sp., while the dead ones no longer capable of retaining water housed several Rhinotermitinae and Nasutitermitinae, perhaps to access edible leaf tissue as much as to obtain dry nesting space. Nasutitermes sp. even outcompetes certain ants for the use of this large epiphyte, but plants are shared with some other taxa, and at a price. Carnivorous species (e.g., Anochetus emarginatus, Odontomachus bruneus) routinely cohabit with Nasutitermes sp. in different parts of the same bromeliads, where they probably treat the termite workers as a ready source of ant food. Phytotelm bromeliads Bromeliaceae provide high-quality living space in the form of impounded humus and moisture for diverse biota. Volumes per unit area range from substantial at locations humid enough to support high densities of phytotelm species (Figs. 1.2C, 2.4) to negligible where aridity mandates water storage in succulent foliage rather than open reservoirs. Arthropods also hide and forage among the enclosed dry leaf axils of certain Type Five Cambridge Books Online © Cambridge University Press, 2009 438 Relationships with fauna Tillandsioideae (Fig. 8.5), but the contributions these plants make to the carrying capacities of habitats must be minor. Fish (1983) probably identi® ed the upper end of the range of potential impacts with his calculated 50000 l of moisture suspended per hectare based on the densities of resident Bromeliaceae that Sugden and Robins (1979) reported in a Colombian cloud forest. One tank bromeliad can provide expansive and varied accommodations for aquatic and terrestrial biota; a mature Glomeropitcairnia erectiflora specimen, for example, offers both a central reservoir that impounds several liters of water and, for the more desiccation-resistant organisms, equal or greater volumes of progressively drier debris packed among older, no longer water-tight leaf axils. Brocchinia micrantha produces even more of the same kinds of ecospace, and maintains them longer as a giant terrestrial in eastern Venezuela and Guayana. Lithophytic, Brazilian Alcantarea regina currently holds the record for water-tight volume, at 45 liters (Zahl 1975; Fig. 1.2C). Theoretical considerations Picado (1911, 1913) envisioned phytotelm Bromeliaceae as the physical basis for a fragmented, aerial swamp distributed through the humid forests of tropical America and occupied by abundant fauna seeking shelter, breeding space and food. Extensive biota and no putrefaction despite substantial impounded litter persuaded Picado to conduct noteworthy experiments designed to demonstrate how plants and their symbionts interact to mutual advantage. While not carnivorous as he suspected, the tested bromeliads nevertheless depend on phytotelmata for moisture and nutritive ions through a mechanism labeled animal-assisted saprophytism (Chapter 5). Resident biota in turn bene® t from several services provided by the hosting plant. Cool conditions may have accounted for Picado' s failure to detect putrefaction because unpleasant odors sometimes reveal substantial anaerobiosis at warmer sites (e.g., Aechmea nudicaulis, Neoregelia cruenta in restinga; Fig. 7.13C,E). Conversely, subsequent study (e.g., Laessle 1961; Maguire 1971; Frank and Lounibos 1987; Paoletti et al. 1991) would corroborate his observations that bromeliad phytotelmata favor exceptionally diverse and densely packed biota. Frank and Lounibos (1987) addressed Picado' s speculations concerning phytotelmata as swamp analogs according to island biogeographic theory (MacArthur and Wilson 1967). Biota determine whether a phytotelm bromeliad acts more like a swamp Cambridge Books Online © Cambridge University Press, 2009 Phytotelm bromeliads 439 or an island (Frank and Lounibos 1987). Colonizers of phytotelmata swamps should arrive at the earliest opportunity and more or less as complete assemblages comparable to those occupying similar kinds of ecospace nearby. True symbionts, as opposed to the casual visitor, would be plantcavity specialists. Once established, the resulting tank-based communities should exhibit high diversity and relatively stable composition. Were phytotelmata island the better descriptor, residents would tend to arrive according to their vagility, with the most mobile forms appearing ® rst. Intense interactions and con® ned spaces would promote signi® cant species turnover and limit biodiversity. Various arthropods, including a number of mosquitoes and some ostracods, identify phytotelm bromeliads as islands by colonizing them more or less exclusively. However, many more lower organisms (e.g., algae, rotifers, other small crustaceans) arrive passively, and also occupy a variety of other types of nearby wet habitats. The specialists distribute unevenly among bromeliad shoots according on one hand to their requirements and on the other to certain site-speci® c qualities of the plants, particularly exposure and whether litter or algae form the trophic base for the community. In essence, bromeliads represent swamp fragments and islands depending on the identity of the user and certain local circumstances that in¯ uence living conditions in water-® lled plant cavities. Resident microflora, flora and invertebrates Numerous surveys (e.g., Picado 1911, 1913; Laessle 1961; Maguire 1971; Fish 1976; Frank 1983; Paoletti et al. 1991; Table 8.2) document the broad hospitality of the bromeliad shoot for micro¯ ora, invertebrates and even some vascular plants ± namely, carnivorous Utricularia humboldtii in Brocchinia tatei (Fig. 8.4B) on Cerro Neblina, Venezuela and Utricularia reniformis and U. nelumbifolia in several Brazilian Alcantarea and Vriesea species and Brocchinia micrantha in Guayana. These last two bladderworts discriminate among certain Vriesea species in southeast Brazil. For example, Utricularia nelumbifolia regularly inhabits the shoots of intermixed populations of Vriesea crassa and V. atra anchored on the steep ¯ anks of at least one granitic dome near Rio de Janeiro, but not those of co-occurring Alcantarea imperialis. Utricularia reniformis colonizes various Tillandsioideae and nearby wet, moss-covered rocks. The aquatic moss Philaphyllum tenuifolium reportedly ¯ ourishes in the shoots of an unidenti® ed Vriesea or Alcantarea at Alto da Serra, São Paulo, Brazil (Hoehne 1951). Cambridge Books Online © Cambridge University Press, 2009 440 Relationships with fauna Bermudes and Benzing (1991) encountered Chlorophyta in addition to those cyanobacteria described in Chapter 5 in the Ecuadorian bromeliads they examined. Numerous algae, mostly diatoms, turned up in additional surveys (e.g., Lyra 1971). Still other reports (e.g., Laessle 1961) list protozoans and fungi, but no bromeliads have been screened for microbial activities important to N cycling, beyond the presence of nitrogenase (Table 5.12), or to the mineralization of impounded plant debris. Janetzky and Vareschi (1993) noted correlation between the litter present and bacterial counts in the phytotelmata maintained by several Jamaican bromeliads. Laessle (1961) recovered 60 kinds of invertebrates from the shoots of bromeliads in Jamaica; Picado (1913) encountered 130 in Costa Rica. Frank (1983) reported that about half of the 470 identi® ed arthropod species collected in bromeliad phytotelmata were mosquitoes. Some 400 species in 15 genera representing at least 20% of those residing in the Neotropics breed there at least occasionally (Fish 1983). Total fauna far exceed these numbers even ignoring the microinvertebrates. A recent survey of several Venezuelan Bromeliaceae revealed an impressive array of taxa (Paoletti et al. 1991; Table 8.2), some new to science. Among the tank-residing macroinvertebrates (.3.0 mm), no fewer than 80 collections represented as many undescribed species. Preliminary examinations further indicated the possibility of at least three new genera. Oliveira et al. (1994) provided data that for the ® rst time included apportionments of biomass among the fauna sampled in bromeliad shoots (Fig. 8.6). Phytotelm bromeliads pose signi® cant public health hazards in parts of tropical America. Tank-dwelling arthropods include blood-feeding midges (e.g., Ceratopogonidae), horse¯ ies and mosquitoes known to carry yellow and dengue fevers, certain strains of equine encephalitis, some ® larial worms, and a variety of lesser diseases (e.g., Klein 1967; Zavortink 1973). Several Anopheles species (e.g., A. homunculus, A. neivai) capable of transmitting malaria favor phytotelmata for egg-laying, and continue to encourage tree cutting and heavy herbicide use to eliminate the offending bromeliads in parts of Amazonian Brazil (Reitz 1983). Linkage between malaria and bromeliads received intense study during the Second World War in Trinidad where natives like Aechmea aquilega supported Anopheles bellator (Pittendrigh 1946, 1948). Occasional relationships bring commercial bene® ts, for instance to the owners of some Central American cacao plantations where midges (Ceratopogonidae and Ceridomyiidae) that Privat (1979) considered important for fruit set by this self-incompatible crop use the local bromeliads to reproduce. Books Online © Cambridge University Press, 2009 Phytotelm bromeliads 441 Figure 8.6. Biomass of invertebrates (.2.0 mm) and vertebrates present in the phytotelmata of Neoregelia cruenta in a Brazilian restinga (modi® ed from Oliveira et al. 1994). Community dynamics Activities within and around the bromeliad phytotelma remain too poorly studied to justify more than a few tentative remarks. Those few reports that exceed mere checklists of inhabitants mostly deal with the biology of a few frogs and a somewhat larger number of invertebrates, primarily mosquitoes, particularly species of Wyeomyia, the chemistry of the impounded ¯ uids, and the structure of the included food webs. More is known about Cambridge Books Online © Cambridge University Press, 2009 442 Relationships with fauna certain aspects of tree holes and the water-retaining ¯ oral bracts of Heliconia (e.g., Maguire 1971). Naeem (1990) demonstrated how litter in¯ uences the structure of communities in certain aquatic systems. Bromeliads with their unusually large, structurally more varied and enduring shoots probably harbor correspondingly more complex microcosms (Chapter 5). Laessle (1961) demonstrated that local conditions and plant shape and size determine whether a bromeliad-based community builds upon autotrophic or heterotrophic foundations in Jamaica. Exposed, spreading rosettes sometimes supported considerable algae; Ecuadorian specimens tested by Bermudes and Benzing (1991) also hosted cyanobacteria judging by assays for nitrogenase (Table 5.12). At shadier sites, trophic pyramids build on accumulated litter (Laessle 1961; Frank 1983). Biota in the catchments provided by Neoregelia cruenta and Aechmea nudicaulis (Figs. 8.7, 8.8) in a Brazilian restinga con® rmed and expanded the list of characteristics that differentiate phytotelmata located in the sun vs. the shade (beyond and under shrubs). Temperatures that exceeded 36 °C appeared to exclude some shade-dependent fauna from exposed shoots (Oliveira et al. 1994). Odonata preferred sun, while exceptionally tolerant Ostracoda ¯ ourished in both settings. Colonizations of cleaned tanks and to lesser degrees glass vessels varied by taxon, being particularly rapid for Ostracoda and Diptera (mosquitoes). Some strictly aquatic forms like the ostracod Elpidum bromeliarium travel among plants by clamping on the skin of migrating frogs (L. C. S. Lopez, personal communication). Tank fauna also varied depending on the shape of the shoot in comparisons of funnelform Aechmea nudicaulis and more spreading Neoregelia cruenta (Oliveira et al. 1994; Fig. 8.7). Subjects grow intermixed through a restinga near Rio de Janeiro where the adults of both species impounded about 80 ml of water either in one central, deep tank (Aechmea nudicaulis) or in several shallower, more exposed leaf axils. Phytotelmata maintained by Aechmea nudicaulis remained relatively full through the dry season compared with those of Neoregelia cruenta, which shrank to a few percent of capacity. Associated fauna also varied, with the ostracods predominating in the shoots of Aechmea nudicaulis, and some copepods assuming that status in Neoregelia cruenta. Rotifers and insect larvae exhibited more even apportionments between the two hosts. Miller (1971) noted that three groups of Diptera (Cullidae, Ceratopogonidae, Chironomidae) partitioned the bromeliad ¯ ora on St John and Anegada in the Caribbean Virgin Islands according to moisture supply, although the mosquitoes, occasionally predatory types, occupied Cambridge Books Online © Cambridge University Press, 2009 Phytotelm bromeliads 443 Figure 8.7. Recolonization of the phytotelmata of Aechmea nudicaulis and Neoregelia cruenta by copepods, ostracods and Culicidae during winter and summer in a Brazilian restinga. Recollections occurred at 4± 16-day intervals after the initial census (modi® ed from Madeira et al. 1995). some of the sampled plants. Cullidae predominated in the shoots of montane Aechmea lingulata, which remain continuously water-® lled on St John. Along the considerably drier coast of Anegada, Tillandsia utriculata hosted fewer mosquito larvae, and instead harbored abundant ceratopogonid midges. At lower elevations on St John where humidity usually Cambridge Books Online © Cambridge University Press, 2009 444 Relationships with fauna Figure 8.8. Data for copepods only for the plants considered in Fig. 8.7. registers somewhere between the other two collection sites, representatives of all three families colonized the same plants, although midges of both taxa dominated. Apparently these insects tolerate drier shoots, faring less well as the length of the rainy season increases. Fidelity and adaptation Dependence on phytotelmata varies among the users. Lower forms (e.g., algae, protozoans, rotifers), compared with many of the higher organisms reported in bromeliad shoots, occupy a variety of other kinds of cavities. Conversely, Laessle (1961) reported three ostracods (Metacypris) as bromeliad endemics. Reitz (1956) imputed obligatory status to related Elpidium bromeliarum, which inhabits numerous bromeliads native to Santa Catarina State, Brazil in addition to the two restinga species investigated by Lopez et al. (1993). Mosquitoes rely on sophisticated behavior, small size and keen senses to use bromeliads. Some species favor plants that belong to a single genus (e.g., Frank and Curtis 1981a,b), while others exploit more diverse ¯ ora, including nonbromeliads. For example, Frank and O' Meara (1985) demonstrated Cambridge Books Online © Cambridge University Press, 2009 Phytotelm bromeliads 445 that ovipositing Wyeomyia vanduzeei preferred Tillandsia utriculata over Catopsis berteroniana in Florida. Wyeomyia mitchellii was less in¯ uenced by plant identity than by aspects of the macrohabitat, judging by the greater numbers of eggs and larvae in shaded compared with fully exposed shoots. Several Costa Rican bromeliads consistently failed to support the same Paramecium species that reach substantial densities in nearby Heliconia bracts. Germer (1982) and Frank and Lounibos (1987) reported the same outcome for algae comparing water-® lled beakers and Billbergia pyramidalis shoots during an experiment in Florida. In the ® rst instance, runs employing untreated plants and others emptied and re® lled with clean water, and arti® cial microcosms concocted with ® ltered and un® ltered tank ¯ uids, failed to yield conclusive results. Toxins offer one possibility, but higher losses in un® ltered tank water and untreated shoots suggested a greater role for predators. Some gravid tree hole-users cue on chemical and physical traits that relate less to the identity of the tree than to the probable duration of a cavity' s moisture supply (Bradshaw and Holzapel 1984). Wyeomyia smithii seeks relatively young Sarracenia purpurea leaves for egg-laying by following a food stimulus (Bradshaw 1983). Signals emanating from both the phytotelmata and the containing shoot seem to prompt bromeliad utilization. Frank (1985, 1986) used colored receptacles to demonstrate that Wyeomyia vanduzeei prefers dark green targets, while W. mitchellii favored the deep red type. Anopheles aegypti consistently choose the darkest containers when offered the same set of options augmented with additional vessels in several lighter colors. Preferences for sunny vs. shaded phytotelmata could explain these results as mosquitoes vary substantially on this basis, some assiduously avoiding exposed spaces where plants exhibit brighter silhouettes. Signals that indicate plant maturity and health provide additional useful information, especially to insects with extended larval stages. The presence of an in¯ orescence or necrotic foliage reveals shoot age and condition respectively and accordingly, the likelihood of holding water long enough to complete larval development (Frank 1985, 1986). Discrimination also occurs in space, sometimes over distances measured in centimeters (Frank and Curtis 1977b). Lateral, as opposed to central, tanks attracted more egg-laying in some trials, perhaps because heavy rain disturbs the contents of the younger leaf bases more than the older ones displaced below and therefore better insulated by overhanging foliage. Egg rafts laid by Wyeomyia vanduzeei proved vulnerable to washout from Cambridge Books Online © Cambridge University Press, 2009 446 Relationships with fauna Figure 8.9. Colonization of arti® cial bromeliads by Diptera. Animals present were counted every 30 days (after Krügel 1993). Tillandsia utriculata shoots in Florida compared with larvae or pupae. Should losses be great enough, compensatory egg-laying seems likely to evolve. Seventy-four mature Guzmania weberbaueri specimens interspersed among 96 ersatz bromeliads (funeral vases) for 13 months in Peruvian Amazonia attracted 26 species of ¯ atworms, nematodes, annelids, insects and vertebrates representing about 2800 individuals (Krügel 1993; Fig. 8.9). Monthly counts proved too infrequent to fully document colonization. Two Culex species predominated among the immigrants, with nematodes constituting the second best represented taxon. If selective, the mosquitoes tended to favor the bromeliads. Odonata and Salatoria, the top invertebrate predators, consistently avoided the wide-mouthed, 250 cm3 grave vases. Microhydid Syncope antenori proved less fastidious by using real and simulated phytotelmata at about the same rate for egg-laying. Nematodes apparently arrived by migrating up from the ground. Neither copepods nor ostracods colonized either type of vessel, although both groups usually appear in the checklists of bromeliad fauna. Following discovery of a suitable medium, some bromeliad-users bring Books Online © Cambridge University Press, 2009 Phytotelm bromeliads 447 Figure 8.10. Total number (eggs, larvae, pupae) of Wyeomyia vanduzeei and Wyeomyia medioalbipes demonstrating relationships between numbers of immature mosquitoes present and the size of the hosting phytotelmata of Tillandsia utriculata in south Florida (after Frank et al. 1977). additional adaptive behavior into play. Gravid Wyeomyia vanduzeei and W. medioalbipes gauge the relative utility of phytotelmata provided by Florida Tillandsia utriculata according to shoot volume (Frank and Curtis 1977a; Frank et al. 1977; Figs. 8.10, 8.11). Numbers of eggs laid through the season tracked tank size rather than the amounts of impounded organic debris available for feeding. Values for larvae also increased with ¯ uid volume, and those inhabiting the more heavily populated shoots matured Cambridge Books Online © Cambridge University Press, 2009 448 Relationships with fauna Figure 8.11. Number of instar II Wyeomyia vanduzeei relative to impoundment capacity of Tillandsia utriculata in south Florida (after Frank et al. 1977). slower than their counterparts at lower densities elsewhere. In effect, uncharacterized, density-dependent cues help regulate populations by moderating larval demands for ® nite, plant-provided resources. Life cycles match substrates in a second, time-related dimension. Progress from egg to pupae requires at least two weeks rather than the mere 4± 5 days for species adapted to ¯ ood waters and similarly ephemeral breeding media. Cambridge Books Online © Cambridge University Press, 2009 Phytotelm bromeliads 449 Apparently, bromeliad shoots rank among the more durable of the many kinds of breeding sites utilized by Neotropical Culicidae (Frank 1983). Coevolution probably played no more than a minor role in shaping the relationships between phytotelm Bromeliaceae and the mutualistic invertebrates they accommodate. When change did occur, it was mostly asymmetrical as in ¯ owers that mimic their pollinators or certain food sources of those animals (Orchidaceae). Dependent fauna changed more in the interaction with the bromeliad shoot and the affected characters involved body form, life cycle and behavior. Ant-house Bromeliaceae represent exceptions to this rule assuming that the bulbous shoot (Fig. 8.5) indeed re¯ ects a response to plant-feeding by ants, yet none of these animals possesses features evolved primarily to assist plant nutrition. A third party like Toxorhynchites haemorrhoidalis can exaggerate the impact of plant form on certain co-occurring tank occupants. Larvae of this predatory mosquito heavily cropped other immature dipterans inhabiting the same single, deep pool provided by the tubular shoot of Venezuelan Aechmea nudicaulis (Frank et al. 1984; Lounibos et al. 1987; Fig. 7.13C). Aechmea aquilega harboring this same relatively immobile insect provided safer refuge to potential prey simply by offering multiple, relatively isolated leaf axils. Vulnerability to foraging birds or reptiles could follow quite different, bromeliad-speci® c patterns. Morphology suggests that bromeliads have in¯ uenced the evolution of fauna representing diverse taxa. Several dragon¯ ies (Calvert and Calvert 1917), crane ¯ ies (Alexander 1912) and syrphids (Knab 1912) bene® t from body shapes that long histories of tank use could explain. The damsel¯ y Leptogrion perlongum needs its unusually long abdomen to oviposit in the water sequestered deep in bromeliad shoots (Fig. 8.4C). Pronounced ¯ attening favors penetration and greater mobility among appressed leaf bases for some bromeliad-inhabiting lumbricoids and isopods (M. G. Paoletti, personal communication). Certain earthworms may be exceptionally well adapted by a variety of characteristics to live in and around epiphytic bromeliads, as indicated below. Land-dwelling species of Aratus and Metopaulias represent notable exceptions within the predominantly marine crab family Grapsidae by possessing unusually long, slender legs and narrow carapaces. Both modi® cations favor arboreal life generally and, for several species, movements in and around the bromeliad shoots used for shelter and reproduction (Laessle 1961; Read 1969; Abele and Means 1977). McWilliams (1969) encountered Sesarma miersii among the leaf bases of a variety of Bromelioideae at two sites in the state of São Paulo, Brazil, although Cambridge Books Online © Cambridge University Press, 2009 450 Relationships with fauna Table 8.3. Effects of conditioning by egg-laying Metopaulias depressus on conditions in bromeliad leaf axils in Jamaica Unconditioned tanks Conditioned tanks Dissolved O2 pH Ca21 Declines to as little as 15% at night Median value 4.8 1 mg in 240 ml leaf axil (average-sized nursery) Usually remains above 35% Median value 6.8 3.12 mg in leaf axils of about the same volume Source: After Diesel and Schuh (1993) inhabited plants occurred near brackish water suggesting transitory rather than long-term occupancy. Metopaulias depressus removes damsel¯ y nymphs to protect its larvae in some Jamaican bromeliads, a behavior that presumably required extended tank use to evolve (Diesel 1992). Metopaulias depressus takes parental care well beyond the protection and feeding of progeny by also mitigating the harsh physicochemical conditions that so often prevail in bromeliad phytotelmata. Diesel and Schuh (1993) con® rmed that unmanipulated media in the axils of Jamaican bromeliads tend to be too acidic, anoxic and Ca-depleted to meet the needs of young crabs before they begin to feed and become less vulnerable to such unfavorable limnology. Attending adults remedy all three de® ciencies by removing litter, which promotes oxygenation, and by adding fragments of snail shells preparatory to egg-laying. Calcium ions released from the gastropod remains promote carbonate buffering and facilitate chitin synthesis at molting. Table 8.3 illustrates the contrast between unmodi® ed leaf axils and those conditioned by adult crabs. Molecular systematics indicates that M. depressus and the other six arboreal crabs of Jamaica share a common ancestor that lived no more than four million years ago (Schubart et al. 1998). Bromeliaceae probably played a key role in this radiation. Plant-provided benefits to resident fauna Much of the phytotelm bromeliads' impact on associated biota depends on conditions in the impoundments located among its overlapping leaf bases. Destruction awaits prey lured by fragrances and color to the lubricated, steep-walled phytotelma of Brocchinia reducta (Fig. 2.4F). Although frogs and some arthropods thrive in the ¯ uids accumulated by this carnivore, Cambridge Books Online © Cambridge University Press, 2009 Phytotelm bromeliads 451 phytotelm Bromeliaceae that depend on litter rather than prey support a much more extensive biota. However, experiments using arti® cial vessels provisioned with the same contents as nearby bromeliad shoots indicate that some colonists discriminate among different kinds of plant cavities (e.g., Krügel 1993). What makes the space produced by the shoot of a humus-based bromeliad so accommodating for such a variety of associates, yet apparently inimical to some others? With sufficient time, any catchment ® lled with moisture and litter will attract some occupants, but not those that respond to or require additional, plant-speci® c characteristics. Perhaps the bromeliad embellishes the already rich mix of resources in its phytotelma with oxygen from adjacent leaf tissue, while simultaneously discouraging certain other biota. Observations conducted in situ indicate multiple in¯ uences dictated by the bromeliad and others by its immediate environment. Laessle' s (1961) observations in Jamaica on some 75 specimens representing species of Aechmea, Hohenbergia, Guzmania, Tillandsia, and Vriesea revealed that the angle of the leaf axil in¯ uenced the welfare of resident fauna through its effect on the chemistry of the impounded ¯ uids. Water surface to volume ratios (e.g., higher in the relatively lax shoots of Aechmea paniculigera than in the more upright Vriesea species; Fig. 2.4A± D) and within a rosette (greater in older than younger leaves) generally correlated with higher O2 and lower CO2. Acidity, O2, and CO2 often ¯ uctuated from day to night although unevenly. Diurnal oscillations varied among the leaf bases in a single shoot (Table 8.4), and from one rosette to another, with the most pronounced changes occurring in the most exposed plants. Concentrations of both gases usually re¯ ected the kinds and numbers of organisms present rather than any identi® able features of the hosting plant. Janetzky and Vareschi (1993), working in Jamaica on some of the same species examined by Laessle, recorded similar diurnal ¯ uctuations in O2 and higher readings in the phytotelma of exposed compared with shaded specimens (Aechmea paniculigera and Hohenbergia sp.). They also noted that O2 concentrations diminished from the surface downward (Fig. 8.12). Low O2 tensions and muted ¯ uctuations characterized tanks containing the most detritus. Bacteria counts paralleled the modest values recorded for the better-known black water systems that likewise owe their chemical peculiarities to degrading vegetation. High concentrations of dissolved N and P underscored the value of the phytotelma for plant nutrition (Fig. 8.13). Krügel (1993) working with terrestrial Guzmania weberbaueri in Peruvian Amazonia also reported how pH varied with plant exposure and tank Books Online © Cambridge University Press, 2009 452 Relationships with fauna Table 8.4. Water chemistry in seven leaf axils of a single Aechmea paniculigera shoot. This specimen was growing in full sunlight. Leaf axil number 7 was uppermost and received the most light. Leaf axil number 1 was the oldest on the shoot that still held water. Daytime readings were taken between 09.45 hours and 11.30 hours and those after dark between 20.30 hours and 22.30 hours CO2 (ppm) O2 (ppm) pH Leaf axil Day Night Day Night Day Night 1 (oldest leaf) 2 3 4 5 6 7 (central tank) 12.0 5.5 4.0 6.0 9.0 8.0 11.0 12.0 12.0 21.0 16.0 16.0 18.0 44.0 0.5 6.6 6.0 7.6 5.6 5.4 7.8 1.6 1.4 1.6 2.6 1.6 1.6 0.4 5.4 6.3 6.7 5.8 5.6 5.2 4.9 5.2 6.7 6.8 5.9 5.3 5.2 4.7 Source: After Laessle (1961). contents, perhaps in the second instance re¯ ecting respiration by resident saprophytes. Still another study conducted on immature Aechmea bracteata under controlled conditions provided a record of change in the ¯ uids sequestered in the central and one lateral cavity of a phytotelm bromeliad (Benzing et al. 1972). Cleaned leaf axils of one set of specimens were re® lled with distilled water only; those of another group also contained rotting leaf debris, green algae or damsel¯ y nymphs. Small glass beakers treated identically served as controls. Photon ¯ ux density equaled about 12% of full sunlight during the 12-h photoperiods. The results, some of which are depicted in Fig. 8.14, indicate that the resident organisms in¯ uenced the chemistry of impounded water most of all. Illuminated or darkened, plant chambers and beakers ® lled with distilled water never exceeded 60% O2 saturation. During the same runs, CO2 became more concentrated in the tanks, probably because some of the epiphyllae remained despite judicious cleaning. Simultaneously high CO2 and low O2 required the presence of animals or decomposing biomass. Only illuminated algae elevated O2 to saturation. Fluids remained almost continuously acidic, sometimes more so at night than during the day whether or not other materials were present or the exposure was high or low; beakers containing illuminated algae alone produced pH readings near 7.0. Cambridge Books Online © Cambridge University Press, 2009 Phytotelm bromeliads 453 Figure 8.12. Oxygen saturation (%) in bromeliad phytotelmata in Jamaica. (A) Sunexposed specimen with algae. (B) Shaded specimen with litter. Measurements were taken at different depths in the morning, at midday and in the evening (after Janetzky and Vareschi 1993). Books Online © Cambridge University Press, 2009 454 Relationships with fauna Figure 8.13. Concentration of total phosphorus present in 16 phytotelmata of Aechmea paniculigera in Jamaica (after Janetzky and Vareschi 1993). Substantial evidence indicates that the bromeliad does not actively in¯ uence the abundance or diversity of the fauna that use its phytotelma. Oxygen and CO2 exchanges between water-® lled leaf axils and the atmosphere proceed as if the same contents resided in comparable arti® cial receptacles. Shoot size and shape and environment to the degree that photosynthetic photon ¯ ux density (PPFD) in¯ uences sun vs. shade morphology (Fig. 4.23B,C) largely determine the volume and con® guration of plant-created habitat and consequently the kinds and quantities of the resources likely to accumulate there. Characteristics like leaf coloration in¯ uence carrying capacity for certain residents to the extent that they reduce predation (Figs. 2.14G, 2.17B). The availability of potential stock sometimes also shapes the composition of the bromeliad-based community. For example, despite substantial native Odonata, none of the local species uses bromeliads to reproduce in Florida (Frank 1983). Virtually nothing is known about the optical properties of the bromeliad shoot beyond those concerned with photosynthesis (Chapter 4). Closer inspection might con® rm the presence of some other leaf functions. For example, do the darkly pigmented leaf bases of certain species (e.g., Vriesea erythrodactylon; Fig. 2.18D) help hide like-colored residents? Might the uneven distributions of chlorophyll and anthocyanins of some relatives (e.g., V. fosteriana; Fig. 2.14G) blur the outlines of lighter-colored prey? Padro Nauum (personal communication) examined about 100 plants each Cambridge Books Online © Cambridge University Press, 2009 Phytotelm bromeliads 455 Figure 8.14. Diurnal ¯ uctuations in the concentration of dissolved CO2 in the central tank of juvenile Aechmea bracteata and beakers containing the same materials (after Benzing et al. 1972). Cambridge Books Online © Cambridge University Press, 2009 456 Relationships with fauna of a variegated and a concolorous form of an Aechmea hybrid being propagated for sale in his nursery near Rio de Janeiro. A single unidenti® ed species of frog, one per plant, inhabited about 10% of the former and none of the latter specimens. However, the same fertilization used to promote leaf color also caused the variegated plants to develop a more compact and perhaps more alluring shoot. Several other growers in Brazil also report that bromeliocolous frogs prefer plants with variegated foliage. Sharp spines and compact shapes appear to promote plant use as well. Certain ornamentations may persuade fauna to pursue other kinds of relationships with Bromeliaceae. Conspicuously marked shoots of some taxa (e.g., red leaf tips on a number of Neoregelia species; Figs. 2.13F, 2.18A) could parallel those of the Neotropical gesneriads (e.g., Dalbergia) whose orange and red-blotched foliage supposedly reminds the local pollinators of their presence between ¯ owering seasons. However, many of the most impressively variegated bromeliads (e.g., Vriesea fosteriana and V. hieroglyphica; Fig. 2.14G) ¯ ower at night, attract bats, and subsequently ripen dry fruits and seeds. Whether or not these ornamentations affect plant/animal relationships, they certainly impact photosynthesis (Benzing and Friedman 1981). Transitory coloration associated with pollen and seed dispersal is discussed in Chapter 6. Litter processing Those detritivores that Paoletti et al. (1991) reported at higher densities in the shoots of certain Bromeliaceae than in subjacent soils contribute to a process crucial for plant success (Fig. 8.15). In less than the 2± 3-year lifetime of the typical ramet, resident biota transform impounded litter into ® ne-textured, humic soil. Psidium sp. foliage enclosed in nylon mesh bags, and incubated for just three months in the axils of Aechmea filicaulis and Vriesea splendens in the same cloud forest in northern Venezuela, lost 21± 27% of its initial weight, as did a second set of samples buried in the ground beneath the host trees (Paoletti et al. 1991). Surveys (e.g., Naeem 1990) conducted on larger-scale, better-known aquatic systems suggest how most phytotelm bromeliads rely on processors housed in leaf axils to accomplish this transformation and extract nutrients from intercepted litter. Three groups of organisms assist one another when feeding on the litter that constitutes the trophic base for heterotrophic forest stream communities (e.g., Cummins et al. 1989). Microbes, particularly hypomycetes and saprophytic bacteria, initiate the breakdown process by chemically altering structural polymers and weakening the integrity of waterlogged foliage and Cambridge Books Online © Cambridge University Press, 2009 Phytotelm bromeliads 457 Figure 8.15. Densities of extractable microfauna in contents of epiphytic bromeliad shoots, suspended humus, and in rotten wood and soil/litter on the ground. The wet site is cloud forest located along a ridge line (~1000 m) at Rancho Grande, northern Venezuela; the dry site is in seasonal woodland several hundred meters over the leeward side of the same ridge (based on data collected by Paoletti et al. 1991). wood. A second group of arthropods known as the `shredders' (primarily amphipods, isopods, caddis¯ y, may¯ y, midge, and certain Diptera and beetle larva) produce even ® ner fragments (mostly feces and uningested fragments ,1 mm) from the bioconditioned, larger particles. A third group of mostly ® lter-feeding microinvertebrate `collectors' that includes many mosquito larvae subsist on the resulting particulates. Breakdown rates vary with the source. Especially recalcitrant foliage incubated in mesh bags lost less than 0.01% of its initial dry weight per day; more labile (digestible) foliage disappeared at .0.15% per day (Cummins et al. 1989). Unfortunately, checklists of invertebrates enumerated in the surveys of bromeliad shoots lack the resolution reported for the betterstudied lotic ecosystems. However, judging by the numerous nonpredatory Books Online © Cambridge University Press, 2009 458 Relationships with fauna Cullidae and direct litter feeders reported in bromeliad tanks, saprophytes and shredders probably operate there as well. The time required to mobilize plant nutrients in the phytomass impounded in a bromeliad shoot appears to vary according to the element, characteristics of the source, and a host of environmental circumstances, most notably the identities of the processors. Climate also plays a crucial role; leaf axils dry out for months each year among the relatively drygrowing forms (e.g., Tillandsia flexuosa in south Florida), but others remain ® lled permanently like those of Guzmania weberbaueri monitored by Krügel (1993) in Amazonian Peru. Clearly, possession of phytotelm architecture does not guarantee continuous as opposed to pulse-supplied status for the bromeliad with a strictly mechanical root system (Chapters 4 and 5; Zotz and Thomas 1999). Guzmania monostachia growing in a Florida swamp forest indicated that at least the ramets of this bromeliad at one site cycle too quickly to permit plant access to large fractions of the more refractory elements sequestered in impounded litter (Table 5.8). Humus collected from its leaf bases contained substantially more Kjeldahl N than K compared with the usual concentrations of these elements in mature, living tree foliage (between 1:1 and 1:2). Phosphorus merits special attention as an element that often occurs in even shorter supply for tropical forest ¯ ora. Agencies beyond the life span of the hosting ramet help determine how much of the nutrient capital delivered in impounded litter ends up supporting plant growth instead of other organisms residing in the same phytotelmata or exits the system altogether. An emigrating detritivore deprives the plant of the nutrients incorporated in its body, as does the animal taken by a visiting predator. However, fauna that spend even part of their lives using the bromeliad release useful nutrients during that interval. Impact on the balance sheet just happens to be less favorable for the plant when the animal is removed before what would have been its more timely voluntary departure. Resident predators contribute somewhat differently to plant nutrition depending on their diets and foraging patterns. In addition to the frogs, centipedes, scorpions and other relatively mobile carnivores that associate with bromeliads, additional, more obscure and slower movers also import nutrients for plant use. Gastropods that hide by day in the moist cavities provided by the phytotelm bromeliad emerge at dusk to feed on nearby vegetation. Those cyanobacteria noted in Ecuadorian bromeliads (Bermudes and Benzing 1991) bene® t the system as scavengers of another type. Conversely, denitri® ers and certain other mediators of the N cycle might diminish the fertility of a phytotelmata. Heavy Cambridge Books Online © Cambridge University Press, 2009 Bromeliads and the definition of soil 459 rains almost certainly affect the ledger by washing out nutrients before the bromeliad can absorb them (Table 5.15). A ® nal point concerns global change. Tropical forests generate nitrogen oxides that molecule for molecule exceed CO2 many fold as radiatively active (greenhouse) gases. Resident Bromeliaceae probably represent too small a component of such systems to warrant concern as major contributors, but if present, important processes might be more practically studied in these relatively self-contained microcosms than on broader scales. Phytotelm bromeliads are de® nitely useful for surveillance. Monitoring could determine whether changes in the atmosphere or local climate are altering the phytotelm community in ways that portend impacts with broader consequences (Lugo and Scatina 1992). Enhancing the utility of these microcosms as biosensors is the exceptional vulnerability of the hosting bromeliads, particularly the epiphytes.. Canopy-anchored ¯ ora in general react with unusual sensitivity to humidity, PPFD and temperature along both horizontal and vertical gradients ± sometimes within the same tree crown (Figs. 7.11, 7.12, 7.15). Aspects of carbon and water relations (e.g., CAM) obliged by relatively harsh substrates and frequent drought in arboreal habitats reduce niche breadth for many epiphytic plants (Benzing 1998; Chapter 4). Moisture impounded in leaf bases moderates climate somewhat for the phytotelm bromeliad, but judging by the narrow altitudinal distributions of many of the montane species, not substantially so. Indumenta that in¯ uence gas exchange differently depending on the moisture regimen further reduce the ecotolerances of Type Five bromeliads (Fig. 4.11). Bromeliads and the definition of soil Temperate zone ecologists tend to de® ne soil in rather parochial terms because so often they fail to appreciate how similar media build up in the canopies of humid tropical forests. By physical and biological characteristics, these suspended substrates represent arboreal extensions of the upper, organic horizons of the `earth soil' located below (Nadkarni 1981, 1984; Lavelle and Kohlmann 1984; Paoletti et al. 1991). Both resources contain abundant plant nutrients, and support active roots originating from the supporting trees and their epiphytes. Important soil phenomena like N2 ® xation and mineralization also characterize both compartments. Shared biota, including an undetermined number of fauna that move between them, make the case for a continuum even stronger. Litter-feeding arthropods abundantly inhabit suspended humus, and Cambridge Books Online © Cambridge University Press, 2009 460 Relationships with fauna some groups occur at even greater densities in the shoots of epiphytic Bromeliaceae. Palacio-Vargas (1982), working with mites from 22 families in a Mexican forest, observed certain species exclusively in the canopy, others only in earth soil, and a third group seemingly at home at both locations. Shoots of an unidenti® ed Tillandsia harbored 18 Collembola taxa, some only during the wet season. Thirteen more species never left the ground. Factors underlying the greater tolerances of the bromeliad-inhabiting populations were not obvious, and no attempts were made to identify them. Paoletti et al.' s (1991) survey in northern Venezuela demonstrated that the diversity of soil-type invertebrates in bromeliad shoots can equal, and sometimes exceed, that in subjacent earth soil (Fig. 8.15). Moreover, detritivores and scavengers occurred in about the same proportions in both compartments, with only slightly higher values for predators on the ground. Density peaked in the canopy, sometimes reaching 10-fold those values recorded for comparable volumes of earth soil. In essence, the local bromeliads represented `hot spots' of invertebrate diversity and abundance. Plentiful detritus and high humidity at both locations probably account for the similarities between these two communities. Similar media on the ground and in the canopy raise questions about resident fauna, and whether Bromeliaceae sometimes in¯ uenced the evolutionary histories of some of these symbionts. Speci® cally, how many of the detritivores that inhabit suspended soils or bromeliad shoots acquired dietary, life cycle or other de® ning characteristics above ground? More fundamentally, what kinds of features distinguish related arboreal and terrestrial biota differentiated for life in one kind of space exclusive of the other? Are any of the similarities convergent, or do fauna share certain features through common origins? Geology and paleoclimate undoubtedly in¯ uenced the amount of exchange that occurred between the residents of these two compartments. Long before the late Tertiary, humid tropical conditions prevailed more widely than today. How much of this history is re¯ ected in extant taxa? How many of these populations span the two compartments and routinely migrate between them? To what extent do the bromeliads help blur the distinctions between canopy and earth soils today, and have they in¯ uenced those characteristics that differentiate the more substrate-speci® c fauna? Fragoso and Rojas-Fernández' s (1996) study of earthworms illustrates just one of the many possibilities. Fragoso and Rojas-Fernández encountered four species of earthworms in the shoots of Androlepis skinneri and Aechmea mexicana in the canopies Cambridge Books Online © Cambridge University Press, 2009 Bromeliads and the definition of soil 461 Figure 8.16. Abundance of the earthworm Eutrigaster sporadonephra in epiphytic Aechmea mexicana and Androlepis skinneri in tropical rainforest in southeastern Mexico (after Fragoso and Rojas-Fernández 1996). of seasonal rainforest in southern Mexico. Of these, only Eutrigaster sporadonephra was absent on the ground except for the occasional individual located in a large rotten log. Both bromeliads were used unevenly, with scattered shoots accounting for inordinate numbers of worms in accordance with the capacity of these detritivores to migrate in search of mates and superior habitat (Fig. 8.16). Large bromeliads probably attract the most animals because they contain litter that remains continuously moist. Unlike most of the earthworms native to tropical forests and savannas, which tend to reproduce only during the wet season, Eutrigaster sporadonephra does so continuously, perhaps aided by the everwet conditions afforded by phytotelm bromeliads. Similar densities of animals during the rainy and dry seasons further indicated that arboreal Bromeliaceae provide permanent refuges rather than temporary heavens during dry weather as reported for the arthropod fauna of some Tillandsia species in central Mexico (Murillo et al. 1983). Cambridge Books Online © Cambridge University Press, 2009 462 Relationships with fauna Fragoso and Rojas-Fernández proposed two, somewhat overlapping hypotheses to explain the near complete restriction of Eutrigaster to arboreal bromeliads in southern Mexico. As epigeic species (residents of the upper layers of soil), members of this genus mostly occupy continuously humid habitats where the well-aerated, moist space these animals require is always available on the ground. Seasonal habitats present problems because litter tends to accumulate during the driest months when these worms can least use it, while wetter periods see much of that forage disappearing. However, the bromeliad shoot provides moisture and food on a more dependable schedule, and hence constitutes a superior resource that the worms prefer over soil in marginally humid habitats. The second hypothesis factors in two additional provisos: recent arrival in North America across the current land bridge with South America and litter-dependence as a phylogenetic constraint. Eutrigaster has clear Gondwanian affinities, and north of Costa Rica most of its terrestrial representatives inhabit cool pine± oak forests where they thrive in the soil litter layer like all of the other epigeic earthworms. Below about 1000 m, populations consistently inhabit bromeliad tanks or large rotten logs, as if obliged to do so by conditions that routinely render the understory too hostile (dry) for such vulnerable detritivores. This pattern suggests that Eutrigaster spp. dispersed northward over the last 3± 4 million years through cool montane forests, and subsequently invaded lowland communities by way of the bromeliads that provide substrates resembling those on the ground at higher elevations. Finally, Eutrigaster sporadonephra remains a bromeliad specialist in tropical Mexican forest by virtue of its basic epigeic nature that to date has precluded accommodation to seasonal drought. Books Online © Cambridge University Press, 2009 9 History and evolution D. H. B ENZ ING, G. B ROWN AN D R. TERRY Although evolution is the theme that ties this volume together, until now adaptive aspects of phenotype such as CAM, the phytotelm shoot and the absorbing trichome have dominated discussions. Times, places and why these features emerged during Bromeliaceae radiation have received far less attention. At this point, we switch emphasis to beginnings and circumstances that in¯ uenced the adoption of those characteristics that de® ne much of the family as exceptional for adaptive novelty and importance in communities. Fossils, ontogeny, phytogeography, paleoclimate, cytogenetics and the structure of the genome provide insight on the geologic history and phylogeny of Bromeliaceae and point out directions for additional inquiry. Phylogenetic analysis informed by the molecular structure of key segments of DNA and a fuller understanding of the morphology and adaptive biology of representative species will eventually reveal the identities and dates of the major evolutionary events responsible for the distinctness of the more advanced Bromeliaceae among Magnoliophyta. Conditions in primordial habitat(s) head the list of enduring questions: were these sites dark and humid like the forest understory or exposed and dry? More fundamentally, did scarcities of mineral nutrients or drought play the more decisive roles in the evolution of the foliar indumentum and phytotelm shoot? Heterochrony has also ® gured prominently in speculations about bromeliad radiation, but without much thought given to the incentives (plant bene® ts) responsible for this process. A framework that arrays extant lineages in evolutionary space and geologic time (cladogram) will also resolve long-standing disagreements about taxonomy. For example, should Brocchinia in one instance, and Catopsis and Glomeropitcairnia in another, remain aligned within two of the three currently recognized subfamilies (sensu Smith and Downs 1974, 1977, 463 Books Online © Cambridge University Press, 2009 464 History and evolution 1979)? On another note, just a handful of characters, many of questionable value, distinguish the core bromelioid genera (e.g., Aechmea, Neoregelia, Nidularium). Similar problems prevail in Pitcairnioideae (e.g., the boundary between Pitcairnia and Puya and the likely arti® cial nature of Navia) and Tillandsioideae (e.g., how many genera should be recognized within Vriesea and Tillandsia?). More broadly, should the three subfamilies of Smith and Downs be replaced by a larger number of taxa of similarly high rank? Presumably, taxonomic revisions consistent with cladograms constructed with data from multiple DNA sequences will more accurately demonstrate evolutionary relationships within Bromeliaceae than are depicted in any of the existing systems. Likewise, these frameworks should reveal which one of the several candidates already proposed is indeed the sister group of the family, and point out where the bromeliads belong within Liliopsida. More immediate to our goal, an evolutionary taxonomy will permit meaningful organization of the information on plant structure, function and related ecology summarized in the preceding eight chapters. Fossils Fragments of foliage, one ¯ ower and some pollen comprise the reported geologic record for Bromeliaceae (e.g., Gómez 1972; Smith and Downs 1974). Some of these assignments cannot be taken seriously. Poorly preserved leaves described as Bromelia tenuifolia from the Dakota Formation in Kansas constitute one of the two described New World macrofossils, and venation suggests closer affinity to the cycads. Four more binomials were established for organs discovered in Europe. Among these, Bromeliaceophyllum rhenanthum and B. oligovaenicum from Upper Oligocene (,20 million years ) brown coal deposits in Germany lack the distinctive epidermal features needed to con® rm family affiliation. Bromelianthus heuflerianus from the Eocene (,40 million years ) of Italy bears four rather than three-merous ¯ owers (an occasional condition in Dyckia), and it belongs to a poorly preserved ¯ ora containing additional remains no more persuasively assigned to Orchidaceae (Schmid and Schmid 1977). Gómez (1972) described foliage with more convincing morphology from 36 million-year-old sediments in the San Ramon region of Alajuela Province, Costa Rica. Karatophyllum bromelioides bears marginal spines and prominent impressions of peltate trichomes. However, microfossils offer the greatest opportunity to track Bromeliaceae through geologic time Cambridge Books Online © Cambridge University Press, 2009 Phytogeography 465 because most of the family' s lowland members grow in the forest canopy, or, if terrestrial, occupy nonsedimentary environments. A. Graham (personal communication) reported two types of Tillandsia-like pollen from the late Eocene Gatuncillo formation of Panama. Authorities familiar with extant material should examine his ® ndings and also seek signs of Bromeliaceae in the growing number of samples coming on-line from other Neotropical sites. An exceptionally poor fossil record combined with the almost complete restriction of ,2700 species, many with wind-dispersed or endozoochoric seeds, to tropical America argues for a youthful Bromeliaceae. Quite likely the family became distinguishable during the early mid-Tertiary ± sometime between 40 and 65 million years . But if this is true, then we face the question of why Pitcairnia feliciana exists as the sole transatlantic vicariant (Fig. 1.1). Pitcairnia glaziovii with similar ¯ owers occurs at about the same latitude in eastern Brazil (Leme and Marigo 1993). However, foliage with pronounced armature (Fig. 2.12E) and a generally xeromorphic character suggests closer affinities between P. feliciana and certain Caribbean populations. More vexing than questions about relationships among extant Bromeliaceae is why, if just a single member colonized Africa, does it belong to one of the more ill equipped of the .50 genera to travel so far? Additionally, why did this event take so long to occur, and then happen just once, or at least involve just one lineage? Perhaps dispersability was less decisive in shaping family insularity than some other aspects of natural history. Infrequent naturalizations (Chapter 7; Nelson and Zizka 1997) despite widespread use in anthropogenic landscapes suggest that Bromeliaceae generally lack the invasiveness of many other plants. In any case, molecular systematics could help determine whether Atlantic sea¯ oor spreading stranded P. feliciana, or con® rm the more probable explanation that this taxon holds the family record for what appears to be a rare accomplishment, namely colonization across a wide oceanic barrier. Phytogeography Plant distributions often suggest where evolutionary radiations occurred, and which conditions of land form and geologic history and soil and climate favored speci® c events. Explanations for why one clade proliferated more or less than others in a given region require additional information on reproductive biology and ecology (e.g., the types of breeding systems present, predisposition for cladogenesis via possession of propitious Cambridge Books Online © Cambridge University Press, 2009 466 History and evolution features like CAM in arid regions). Smith (1934a) recognized three centers of what he called `family development' and some secondary sites where a genus or a cluster of allied genera supposedly achieved current diversities and distinguishing characteristics. Tillandsioideae, and to a lesser extent Pitcairnioideae, exhibit highest densities of genera and species, including some of their reputedly more primitive lineages, in the geologically young northern Andes (Tables 1.3, 1.4). Dozens of closely related and often ecologically similar members of Guzmania, Pitcairnia and Tillandsia, among other lineages, co-occur in the everwet, low montane forests of southwestern Colombia and northern Ecuador. Bromelioideae demonstrate comparable clustering through similar and drier ecosystems in southeastern Brazil, mostly on soils derived from some of South America' s oldest rocks (e.g., Fig. 1.4C). More genera of Pitcairnioideae occur in the Guayanan highlands than in any other region, although species more densely pack considerably smaller areas elsewhere (e.g., Pitcairnia in Colombia and Ecuador). Pitcairnioideae Although some 275 pitcairnioid species and nearly half of the genera reside in the Guayanan highlands, six nowhere else, little evidence besides the molecular data that place Brocchinia at the base of Bromeliaceae indicates that Andean habitats were stocked from this direction (Steyermark et al. 1995; Terry et al. 1997a; Fig. 9.1). Most Pitcairnioideae endemic to these ancient, highly weathered substrates resemble evolutionary relics (e.g., Brewcaria, Connellia). The larger taxa (e.g., Navia) also show signs of protracted stasis, but additionally much recent, low-level radiation fostered by impoverished substrates and deeply dissected topography within a Tertiary refugium (see below). Distinct populations sometimes occupy one or a few isolated tepuis as if the products of infrequent seed dispersals among widely scattered land islands. Puya suggest that little exchange has occurred in the opposite direction. Those few species with Guayanan representatives number among the few wide-ranging types (e.g., P. floccosa). Local Bromelioideae (e.g., Aechmea nudicaulis, A. tillandsioides) and Tillandsioideae (e.g., Vriesea platynema, Tillandsia complanata) mostly also occur well beyond the Guayanan Shield. Other Pitcairnioideae exceed the Guayanan endemics for ¯ oral specialization (e.g., dioecious Hechtia), and a larger number of species at least match them for lithophytism, leaf succulence and the presence of CAM (Table 4.2). Predominantly Mexican Hechtia far to the north, and Cambridge Books Online © Cambridge University Press, 2009 Phytogeography 467 Figure 9.1. Distribution of tepuis and additional highland habitats at lower elevations across the Guayanan Shield. The Rio Caura separates the eastern from the western regions. Encholirium, Dyckia and allied, smaller genera in southeastern Brazil and Bolivia (e.g., Abromeitiella, now Deuterocohnia), constitute secondary radiations far from what Smith (1934a) presumed to be the primary ancestral habitats. Similar structure and frequent interfertility within Dyckia and Encholirium suggest recent speciation encouraged by conditions peculiar to certain elevated habitats (e.g., campos rupestres; Fig 1.4C; Chapter 7) and ¯ uctuating Plio-Pleistocene climates that displaced and fragmented life zones. Pitcairnia range most widely among the genera owing to an unmatched ability to accommodate hot, humid climates and shade and perhaps disperse long distances via rivers. Less varied Puya feature a more generalized ¯ ower, and the genus' s mostly well-differentiated 185 or so species usually occupy cool, arid to boggy habitats. Puya illustrate evidence of vicariance about as persuasively as any bromeliad genus. Adjectives that describe the dispositions of the genus' s species in addition to Andean include widely disjunct and narrowly endemic. Varadarajan (1990) proposed 11 centers of diversity, each containing 5± 40 species arrayed from northern Colombia to north central Cambridge Books Online © Cambridge University Press, 2009 468 History and evolution Figure 9.2. Geographic locations of centers of diversity of Puya (after Varadarajan 1990). Argentina (Fig. 9.2). Moderate to high-elevation habitat (800± 5000 m) applies in all but one instance. Exceptional center number 11 nestles in central Chile and features mostly coastal desert habitats except for some sites in the adjacent highlands that rise to approximately 2000 m. About 75% of the Puya species occur in one or more of these 11 locations. Monophyletic clusters of taxa, like the seven-member P. tuberosa complex, typically occupy widely separated ranges in several of Varadarajan' s diversity centers. Sympatric species tend to be paraphyletic or polyphyletic, i.e., represent different radiations within the genus. Related arrays of species also indicate that thermal constraints limited vertical speciation unevenly. A large group of allied lineages native to paramo lack Cambridge Books Online © Cambridge University Press, 2009 Phytogeography 469 close relatives at lower elevations, whereas the ancestors of species that reside in puna habitats, primarily in Peru, Bolivia and Argentina, also spawned descendants adapted for cool, tropical sites. Elevation separates members of several pairs of sister species within this second group (e.g., P. harmsii, 3000± 3600 m vs. P. lilloi, 800± 2000 m; P. weberbaueri, 2800± 4000 m vs. P. lasiopoda, 500± 2300 m). Additional information on Puya radiation can be found in Chapter 7. Varadarajan and Gilmartin (1988a) concluded from plant geography and cladograms based primarily on morphological characters that sufficient divergence and coherence among groups of taxa exist to warrant recognition of three tribes among the genera of Pitcairnioideae (including Pepinia, sensu Smith and Downs 1974; Fig. 9.3). Brocchinia was resolved at the base of the entire assemblage as a `distinct lineage that diverged early in the evolution of the subfamily' . Although arguably the most ecologically and morphologically diverse of all the bromeliad genera, the exceptional mix of characters responsible for this distinction occurs among fewer than 20 species all of which range exclusively through habitats associated with the Guayanan Shield (Fig. 9.1). Varadarajan and Gilmartin offered no comment on the suggested close relationship of Brocchinia to Tillandsioideae (Benzing et al. 1985). Pitcairnieae and Puyeae share several apomorphies along with novel suites of identifying features. Only Puyeae ® ts Smith' s (1934a) proposed Andean and pitcairnioid origin for Bromeliaceae as described later. Molecular data derived from the chloroplast gene ndhF also place family origin in the Guayanan region as indicated above and described in greater detail below. According to Varadarajan and Gilmartin (1988a), all three pitcairnioid tribes differentiated in the Guayanan highlands where Brocchinia and much of Pitcairnieae (Ayensua, Connellia, Lindmania, Navia, Steyerbromelia) and Brewcaria of Puyeae remain con® ned today. Later, deteriorating global climate and Andean orogeny would foster conditions that promoted continued evolution among stocks of Pitcairnieae and Puyeae that had migrated westward. Although many lineages would result according to the pattern described for evolving Puya, ¯ owers and fruits would not diverge enough to warrant recognition of more than a few additional genera. Speciation at lower elevations produced the densest concentrations of closely allied populations, for example about 50 Pitcairnia species in Ecuador alone. Current distributions suggest that much smaller (,30 spp.) Fosterella radiated to accommodate mostly drier, low to mid-elevation (,2000 m) habitats arrayed through west central South America (Ibisch et al. 1997). Cambridge Books Online © Cambridge University Press, 2009 470 History and evolution Figure 9.3. A 149-step cladogram based on morphological data that resolves the major groups of genera in Pitcairnioideae into three groups. A± O represent hypothetical extinct ancestors of extant genera (after Varadarajan and Gilmartin 1988a). Cambridge Books Online © Cambridge University Press, 2009 Phytogeography 471 Diversity in this instance peaks in inter-Andean valleys centered in southern Peru where about half of the genus resides. The remaining species occupy similarly modest, but usually nonoverlapping ranges. Central American F. penduliflora constitutes the major outlier some 2000 km north of its nearest relatives. Relatively uniform, primarily entomophilous ¯ owers further differentiate Fosterella from Pitcairnia/Pepinia (Fig. 3.4D,F± H,K± M). Failure to adopt a comparably plastic shoot architecture probably also accounts for the disparate sizes of these two groups. Seed characteristics that affect dispersability and capacity to accommodate seasonal or continuous supplies of moisture differ less. More genera (Deuterocohnia, Dyckia, Encholirium, Hechtia, Puya) represent Puyeae than Pitcairnieae beyond the proposed source area in the Guayanan highlands. Ancestors migrating from northern South America positioned Hechtia farthest poleward in the northern hemisphere, largely in Mexico. Similar differentiations in the central Andes produced several more genera, some narrowly endemic (Abromeitiella, now Deuterocohnia) and others (Encholirium, Deuterocohnia, Dyckia) more dispersed and even farther removed from ancestral habitats in southeastern Brazil and adjacent northeastern Argentina. Some taxa, like Abromeitiella, probably arrived via the Andean cordilleras already equipped by ecophysiology and a miniaturized Puya-like habit for windswept high altitudes and membership among the cushion ¯ ora of that region (Fig. 2.20). Amazonia perhaps provided a second corridor for the migration of lowland types, especially if drier, cooler intervals beginning during the late Tertiary indeed periodically favored savanna over dense forest. Pitcairnia sensu lato, which includes over 250 species in sites from Mexico to Argentina, represents the product of the largest of the radiations in Pitcairnioideae. Paradoxically, relatively few of these populations contribute to the massive bromeliad ¯ ora of southeastern Brazil, whereas Dyckia and Encholirium radiated extensively there. Brocchinia Brocchinia warrants special consideration among Pitcairnioideae owing to its phylogenetic isolation and geographic con® nement to one of South America' s two oldest land surfaces. Extensive data on distribution, morphology, physiology and ecology for what until recently was one of the least-studied bromeliad lineages now provide unmatched opportunity to reconstruct the history of a clade that in important respects parallels the radiation of the entire family (Givnish et al. 1997). Speci® cally, members of Cambridge Books Online © Cambridge University Press, 2009 472 History and evolution Brocchinia possess the same two characteristics (the absorptive trichome and phytotelm shoot) that fostered far more extensive ecological revolutions (more resultant species and adaptive strategies) in Bromelioideae and Tillandsioideae. Epiphytism also occurs in at least two Brocchinia species, aided there, as in the other arboreal Bromeliaceae, by one or both of these same two adaptations. Crucial to the following analysis of Brocchinia is the chloroplast DNA (cpDNA) restriction site map that provides a historic framework to order the pertinent information on Brocchinia structure, function and ecology (Fig. 9.4). Fundamental to Givnish et al.' s (1997) scheme is the putative existence of four monophyletic segregates within the genus minus B. serrata, which probably constitutes a monotypic genus (Holst, personal communication; Fig. 9.5). Future additions (including an undescribed population of ,5 cm adults and reassignment of Ayensua; Holst, personal communication) to Brocchinia and more DNA sequence data could alter this pattern somewhat. Geographic/geologic information germane to bromeliad evolution beyond the legendary con® nement of certain Brocchinia to one or a few of the isolated, ancient, Guayanan sandstone towers includes the conditions of rooting media (Fig. 9.1). Tepuis that rise above about 800 m subject the vegetation growing on their summits to cool wet climates, which along with base-poor parent rock assure impoverished, acid soils. However, the tepuis located in the east, primarily in the Gran Sabana of Venezuela and adjacent Guayana, differ from those farther west in being horizontally bedded, and consequently more deeply weathered and infertile, and thus especially conducive to the evolution of specialized modes of plant nutrition. Accordingly, Brocchinia includes members capable of supplementing meager supplies in soil with key ions from prey, nests of plant-feeding ants, and detritus impounded in phytotelm shoots that may also host substantial colonies of N2-® xing cyanobacteria (Givnish et al. 1984; Figs. 1.2B, 2.2E, 2.4F). The other species, in fact more than half of the total, continue to rely on root systems and soil (Figs. 1.2B, 5.3D), as do certain members of Fosterella, which judged by the restriction site data is the genus closest to Brocchinia. Considerable data suggest that the stock antecedent to Brocchinia ranged through what today is southwestern Venezuela and adjacent parts of Colombia and Brazil that lie over the same Precambrian basement (Fig. 9.1). However, habitats were lower and growth was unassisted by unusual ecophysiology, including carnivory. Instead, architecture paralleled that of the least specialized of the modern descendants (e.g., root/shoot ratios Cambridge Books Online © Cambridge University Press, 2009 Phytogeography 473 Figure 9.4. Resolution of four clades of Brocchinia species according to cpDNA restriction site data (after Givnish et al. 1997). Cambridge Books Online © Cambridge University Press, 2009 474 History and evolution Figure 9.5. Reconstruction of ecological evolution in Brocchinia in relation to the presence of phytotelm shoot (gray shaded bars), key landscape of eastern/central tepuis (bar), and rise to ecological dominance and widespread geographic distribution (dashed rectangle). Uncertainty as to whether the phytotelm shoot arose independently in B. paniculata and the ancestor of the micrantha/reducta clade, or in their common ancestor, is indicated by a lighter shade of gray. Three out of four species equipped with specialized mechanisms to capture nutrients also exhibit ecological dominance and widespread geographic distribution (after Givnish et al. 1997). comparable to those of many other rhizomatous monocots, no phytotelma). Trichomes may have possessed absorption capacity based on the evidence and according to the logic described below. Subsequent radiations, four in all according to the restriction site map (Figs. 9.4, 9.5), took several directions ± in three cases toward specialized nutrition to compensate for infertile soil. Presumably the clade represented exclusively by B. prismatica and the entire B. melanacra cluster (also B. amazonica, B. cowanii, B. maguirei, B. paniculata and B. vestita), except for tall, palm-like and tank-forming B. paniculata, diverged less from the antecedent architecture than members of the clades represented by B. micrantha and B. reducta. Several conditions, including the associated phytotelm and arborescent morphologies, evolved repeatedly, and then less frequently disappeared (e.g., low-statured B. steyermarkii; Fig. 9.5). The more ecologically conservative, saxicolous/rupestral Brocchinia Books Online © Cambridge University Press, 2009 Phytogeography 475 species (e.g., B. delicatula) tend to occupy restricted ranges, often at low elevations, perhaps con® ned there by competition with many other plants, including some Pitcairnioideae (e.g., Navia), adapted to the same climates and rocky substrates. Gigantism in B. micrantha and B. paniculata and some populations of B. tatei stands in stark contrast to the miniaturization illustrated by otherwise unspecialized B. cataractarum, B. delicatula and especially the undescribed population just mentioned. Myrmecophily, as well as carnivory and the other modes of soil-free nutrition facilitated by phytotelm shoots, apparently emerged after antecedents penetrated higher elevations where temperatures and rainfall selected for the predisposing shoot architecture, viz. impoundments such as those featured by B. reducta (Figs. 2.4F, 9.5). Nutrient-poor substrates, especially in the east, would promote the evolution of dense indumenta comprised of large absorbing trichomes to complement the shoot morphology associated with carnivory and ant-feeding in Brocchinia (Chapter 5; Figs. 2.5A± D, 5.2E± G). Following establishment of capacity to access supplemental sources of ions, Brocchinia acuminata, B. hechtioides, B. reducta and B. tatei probably colonized additional tepuis, and the most versatile species (B. acuminata, B. reducta) also migrated to lower elevations across the Pantui to become the most wide-ranging members of the genus. Epiphytism (facultative) would also require wet forest as it still does (B. hitchcockii, B. tatei) and the ¯ ared shoot with channeled foliage suited to utilize impounded litter (Fig. 1.2B). Gigantism was probably encouraged by competition for light in forest gaps, a challenge similarly addressed by many palms with their tall unbranched shoots equipped with spreading crowns of relatively inexpensive pseudobranches. Brocchinia melanacra demonstrates that the occasional relative without a tank may also occupy a broad range, in this case across much of the region occupied by the genus, possibly aided at some locations by structure that promotes immunity to wild ® re. Little imagination is required to visualize the likely origin of the phytotelm shoot assuming some rosulate stock, or the importance of abundant rainfall to allow such a device to replace absorptive roots. Speculation about plausible pathways to an epidermal trichome that mediates nutrition from a foliar phytotelmata requires some thought about plant surfaces. Epidermal appendages perform many tasks across Tracheophyta, including Bromeliaceae, depending on location on the plant, morphology, density and physiology (Table 2.1). Moreover, functions within lineages shift as other phenomena that impact the plant also change. Somewhere in the history of Brocchinia, ancestors, like certain extant lineages (e.g., B. maguirei), possessed trichomes that provided service other than absorption. Books Online © Cambridge University Press, 2009 476 History and evolution However, if that other function required living cells, then absorption was an additional option, which as adopted in the state prevailing in Tillandsioideae and specialized Brocchinia obscures evidence of the preceding service. Many plant trichomes, including those with multiseriate, capitate organization (including Navia glandulosa; Fig. 2.5K), perform tasks that require living cells. Secretion tops the list, the products including digestive enzymes (some carnivores) and excess salt (some halophytes), but far more often lubricating mucilages and toxic metabolites and adhesives to repel or neutralize small-bodied predators. Additionally, these activities usually peak on young, relatively vulnerable surfaces (organs), consistent with visibly precocious trichome maturation (e.g., recall the Syringa shoot tip illustrated in most introductory botany texts). Later, these organs often atrophy as need diminishes, not unlike the sequence reported for Brocchinia reducta (Givnish et al. 1984; Fig. 5.2E± G). On older parts of blades, well above the phytotelmata, the absorptive hairs of this carnivore shrivel, but left intact could provide no further nutritional bene® t anyway (Fig. 5.3C). More importantly, cells comprising the secretory trichome necessarily possess porous walls that perforce permit diffusable substances to penetrate in addition to exit the leaf symplast. Should appendages with these characteristics also occur where shoot architecture promotes contact with nutrient-charged ¯ uids and rooting media are impoverished, evolutionary opportunity and economic impetus exist to improve plant nutrition by drawing on this alternative source. Absorptive scales among Bromeliaceae, including certain Brocchinia, may have followed such a route, and quite plausibly beginning before rather than after the phytotelm shoot evolved. Almost certainly pitfall carnivory accounts for the extraordinarily high densities of unusually large trichomes present on the foliage of B. reducta compared with its myrmecotrophic and nonimpounding relatives (B. reducta has 11.8% of adaxial leaf area occupied by trichomes, B. acuminata 6.7%, B. prismatica 3.8%, B. steyermarkii 0.4%; Givnish et al. 1997; Fig. 5.3E,F). However, absorptive capacity may have preceded the use of prey and plant-feeding ants and quite possibly even the presence of a phytotelma (Table 9.1). Why do certain nonimpounders (e.g., B. prismatica, B. vestita; Fig. 5.3D) possess trichomes capable of accumulating solutes when the architecture of the associated shoot affords the plant no obvious opportunity to exploit this capacity for signi® cant nutritional bene® t (Table 9.1)? Two possibilities come to mind. First, B. steyermarkii and certain other Brocchinia species with similarly conventional architecture retain absorptive trichomes Cambridge Books Online © Cambridge University Press, 2009 477 Phytogeography Table 9.1. Trichome area, density and affinity for 3H-leucine in relation to ecological habit in Brocchinia Species Density (mm22) Leaf surface occupied by trichome stalks (%) Carnivores B. reducta B. hechtioides 194 236 11.8 5.9 Heavy Moderate Myrmecophyte B. acuminata 88 6.7 Heavy Tank epiphyte (humus-based) B. tatei 141 3.8 Little to none Impounding tree (humus-based) B. micrantha 32 1.8 Little to none Nonimpounding terrestrials B. prismatica B. steyermarkii B. cowanii B. maguirei B. melanacra 141 32 18 56 32 3.8 0.4 0.8 1.4 0.6 Heavy Heavy Not tested Light Not tested Ecological habit Labeling with 3 H-leucine Source: Data for all but trichome function after Givnish et al. (1997). despite the loss of tanks formerly present in ancestors (Givnish et al. 1997; Fig. 9.5). Second, trichomes displayed by these bromeliads play no signi® cant role in plant nutrition now, nor did they do so at some earlier time. Perhaps the foliar trichomes of Brocchinia prismatica and B. steyermarkii accumulate 3H-leucine and presumably additional solutes simply because they happen to develop at a propitious location on the plant. Possibly these trichomes soon die, their capacity to reduce thermal loads, photoinhibition and transpiration unimpaired, if not enhanced, by this secondary condition, as for many other Pitcairnioideae (Fig. 2.8D). However, while immature (and alive), near the basal meristem that produced them, these appendages can absorb simple metabolites like amino acids that in situ would never accumulate in sufficient quantities in leaf axils to promote plant welfare. In effect, the assay with tritiated leucine revealed an artifact ± a phenomenon without ecological signi® cance for these plants. According to the second possibility, this sort of activity would become important as ancestors gave rise to increasingly leaf-dependent Bromeliaceae, speci® cally certain nutritionally specialized Brocchinia, phytotelm Bromelioideae, and especially Tillandsioideae. Typical monocots, by virtue of the way their leaves develop, are better Books Online © Cambridge University Press, 2009 478 History and evolution disposed than the typical dicot to evolve modes of trichome-assisted, tankbased nutrition. However, just two Brocchinia and one Catopsis species among Liliopsida have exploited this potential to the extent of evolving carnivory. Members of additional families (e.g., Commelinaceae, Liliaceae, Rapateaceae) parallel many more of the bromeliads in maintaining substantial supplies of moisture and detritus in in¯ ated leaf axils (i.e., they too produce phytotelma and perhaps utilize litter as a source of nutrients). Astelia species (Liliaceae) reputedly also possess absorbing trichomes with peltate organization (Oliver 1930). Occurrences on substrates as impoverished as those supporting many Brocchinia might have promoted nutritional specializations in one or more of these nonbromeliad lineages as well. Clearly, much work needs to be done to reconstruct trichome evolution in Bromeliaceae, including a more comprehensive survey of the structure and function of those organs serving primitive and more advanced lineages in Brocchinia. Bromelioideae Bromelioideae concentrate in Brazil, particularly in the southeast, where a majority of the ,800 species occur, many nowhere else (Leme 1997; Table 1.4). Rugged topography, diverse climates and ancient granitic and other kinds of substrates suited for saxicolous and rupestral habits have favored substantial divergence, often involving localized populations (e.g., Cryptanthus; Fig. 11.1). Atlantic Forest and the adjacent drier habitats have been especially conducive to expansion, and they remain the homes of entire genera (e.g., Canistrum, Cryptanthus, Nidularium, Orthophytum, Quesnelia). Much larger Neoregelia (.90 species) exhibit the same insularity except for Amazonian subgenus Hylaeaicum (e.g., N. myrmecophila, N. longisepala), which Ramírez (1994; Chapter 10) considers more closely allied to Aechmea subgenus Lamprococcus. Other genera (e.g., Aechmea, Billbergia) extend through much of frostfree America, but more often owing to the high mobilities of a few exceptional species (e.g., Aechmea nudicaulis) than to massive migrations or secondary expansions remote from the center of subfamily diversity. Terrestrial Bromelia (,50 spp.) probably owes its exceptional dispersal among bromelioid genera to drought-tolerance and adaptation for low, often coastal habitats. Ancestors that would initiate sizable (e.g., Greigia) or smaller (e.g., Disteganthus, Fascicularia, Ochagavia) genera did migrate west and northward, while some possible returnees (e.g., Hohenbergia, Ronnbergia) modestly augmented diversity in ancestral territory. However, Cambridge Books Online © Cambridge University Press, 2009 Phytogeography 479 poorly resolved taxonomy complicates attempts to track the migrations of Bromelioideae. Several bromelioid genera exhibit geographic disjunctions that suggest erroneous systematics rather than ancient or more recent long-range dispersals. Hohenbergia (,40 spp.), for example, occurs about evenly divided between southeastern Brazil (subgenus Hohenbergia) and Mesoamerica (subgenus Wittmackiopsis). More than a dozen members grow exclusively in Jamaica, mostly on isolated, modest elevations in the western part of the island. At least one more species also ranges into the mountains of Cuba. Ronnbergia, a genus otherwise con® ned to northwestern South and Central America where it probably originated, contains two southern outliers of dubious taxonomic assignment in Bahia State in Brazil. Araeococcus and Streptocalyx exhibit similar dispositions, and the Brazilian and perhaps all the members of the latter genus probably belong to Aechmea (Smith and Kress 1989; Smith and Spencer 1992). Smith' s (1934a, 1962) suggested Amazonian beginning for Bromelioideae accords with the high incidence of taxa in similarly warm, humid habitats farther south, but little else. Whether or not he is correct, conditions in Brazil' s southern coastal states, particularly Rio de Janeiro, have promoted far more speciation and ecological variety. Several clades (e.g., Neoregelia subgenus Neoregelia, Nidularium) populate Atlantic Forest with numerous, little-differentiated and interfertile, often understory species. Many members of Cryptanthus, Encholirium, Orthophytum and Dyckia exhibit similar insularity as members of nearby rupestral ® eld communities. In fact, circumstances through this part of South America favor unmatched bromeliad diversity ± in all, about 40% of the species representing three-quarters of the genera. Endemism involving all three subfamilies further indicates how conducive conditions in the grasslands and rupiculous ® elds of Bahia, Minas Gerais and adjacent states have been for family expansion. Floras located on still other kinds of sites (e.g., campos do altitudes, restingas and campinas, the white sand communities of the Amazon Basin) contain far fewer, but sometimes locally abundant, species (e.g., Aechmea nudicaulis, Neoregelia cruenta; Fig. 7.13C,E), many recruited from nearby, ¯ oristically older and more biodiverse forests. Closely related, if not ancestral, populations continue to inhabit some of these woodlands. Migrations purportedly also explain the ranges and composition of Brazil' s southernmost bromeliad ¯ ora. Topography, plant distributions and endemism persuaded Winkler (1980) that numerous Bromelioideae representing Aechmea, Billbergia, Cambridge Books Online © Cambridge University Press, 2009 480 History and evolution Canistrum, Nidularium and Wittrockia accompanied members of Dyckia, Tillandsia and Vriesea to colonize, or more likely repopulate, Rio Grande do Sul State. Another set of Tillandsioideae with Andean affinities (representatives of subgenera Anoplophytum, Diaphoranthema, Phytarrhiza) invaded southeastern Brazil from the west, whereas still other Tillandsia and Vriesea species arrived through Winkler' s `coastal gate' , in effect the maritime corridor that extends more or less southwest to northeast along the Atlantic coast. Some of these arrivals probably date from the early Holocene shortly after rebounding global temperature diminished the threat of frost. Few Bromelioideae (e.g., Fernseea, Greigia) occupy habitats as cool as those supporting more extensively montane Pitcairnioideae (e.g., Puya) and Tillandsioideae (Tillandsia, Vriesea), perhaps re¯ ecting conditions experienced by ancestors in the Andes compared with those prevailing in the lower, older topography of eastern South America and Amazonia. Rossi et al. (1997) noted this differentiation in Costa Rica where native epiphytic Bromelioideae (Aechmea) compared with more diverse Tillandsioideae, mostly inhabits lowland, moist habitats. Bromeliad xerophytism exhibits a similar pattern. Despite near complete reliance on CAM, Bromelioideae, except for the more succulent types (e.g., certain Bromelia, Neoglaziovia) may fall short for drought-tolerance compared with numerous species of Hechtia, Dyckia, Puya, additional Pitcairnioideae and many Type Five Tillandsia. Conversely, no other part of the family except certain Tillandsia so often colonizes rocks and tolerates salinity. Finally, several Bromelioideae (e.g., some Cryptanthus; Neoglaziovia) and more Pitcairnioideae (e.g., certain Dyckia and Encholirium; Figs. 2.2G, 6.12A,C± E) of the campos rupestres and similar hyperseasonal ecosystems endure ® re better than any Tillandsioideae. Tillandsioideae Tillandsioideae range more widely than members of either of the other two subfamilies (Fig. 1.1). Tillandsia usneoides holds the record for area colonized and long-range dispersal except for Pitcairnia feliciana on the second count. Racinaea insularis grows exclusively on the oceanic Galapagos Islands, and a small population of T. usneoides may be native to Bermuda. Relatives such as Tillandsia fasciculata and T. paucifolia (members of predominantly Mesoamerican subgenus Tillandsia) that occur on many of the Caribbean islands and deep into South America (e.g., T. juncea in southeastern Brazil) further underscore the mobility of the comose seed, and Cambridge Books Online © Cambridge University Press, 2009 Phytogeography 481 suggest that Bromeliaceae indeed emerged later than most of the other sizable, primarily tropical angiosperm families. Smith (1934a) chose Tillandsia to make his most detailed case that geographic distributions indicate the ages of clades still young enough to qualify as genera, or monophyletic groups of component species (e.g., subgenera, sections). The total area occupied by such an alliance divided by the smallest area necessary to include parts of the ranges of every component population yields its so-called `cohesion ratio' . Use of the cohesion ratio to reconstruct plant history requires a rather ® xed notion of how clades evolve and eventually disappear (see Wiley (1988) for an update on centers of origin and biogeography theory). Basically, as a clade matures, the area occupied by its expanding membership presumably enlarges, and accordingly, so does the accompanying cohesion ratio. Barring extinctions, this number increases as new populations continue to emerge farther from the center of origin, and the ranges of the older ones remain static or contract. However, other dynamics produce similar patterns. At issue here is Smith' s contention that biogeography re¯ ects the evolutionary history of certain bromeliad taxa more or less independent of their environments and certain biological properties of the constituent populations. Clades expand to different sizes, and then erode at uneven rates for poorly understood reasons, but the causes are numerous, interactive and involve the inherent properties of the plants considered and their growing conditions. Likewise, dispersability, ecotolerances and the distributions of favorable habitat surely affect the dimensions of the areas populations occupy through time. Small size (few species) may signal a relatively youthful clade ± an incipient radiation ± or extended stasis, just as a con® ned range could re¯ ect recent origin, sedentary propagules, stringent requirements for growth or impending extirpation. Glomeropitcairnia contains just two species, and exhibits no evidence of ever having exceeded its current narrow range in the Lesser Antilles and a few sites in adjacent northeastern Venezuela. However, novel ¯ oral, fruit, seed and trichome morphology and the cpDNA data described below suggest relictual status (no close extant relatives) whether or not the taxon was ever larger. Greater similarity to core Tillandsioideae or some other component of Bromeliaceae would accord with more recent origin. Findings during the past decade that indicate probable paraphylesis for several genera, including Tillandsia and some of its segregates (entire subgenera to inclusive clusters of species), further challenge Smith' s use of phytogeography to help reconstruct the evolutionary history of Bromeliaceae. Cambridge Books Online © Cambridge University Press, 2009 482 History and evolution Cohesion ratios can be useful to formulate questions about evolutionary history and in¯ uential plant characteristics. For example, the least common overlap for Tillandsia subgenus Diaphoranthema constitutes only a tiny fraction of the total area occupied, primarily because T. usneoides and T. recurvata range so widely, the ® rst species probably because of its tendency to release buoyant fragments (Fig. 7.7C). Abundant self-set seeds and extraordinary capacity to grow on diverse substrates may underlie the nearly as expansive distribution of T. recurvata. However, a number of close relatives with similar ecology and propensity for autogamy remain quite insular, probably for reasons related to additional aspects of life history peculiar to these populations. Overall, the genera of Tillandsioideae probably exhibit less cohesion than those of Pitcairnioideae in part because they produce wind-carried rather than unappendaged seeds. Total and the least common areas for Puya converge most in the southern Andes, reputedly because its members share greater genetic identities there (Smith 1934a). However, differences in the availability of habitats that can support groups of related species (e.g., expansive, cold plateau farther south) may be more in¯ uential. Plant influences on geographic range Ecotolerance and other aspects of plant biology affect the phytogeography of certain Bromeliaceae. Compared with fundamentally cool-growing clades like Puya, most members of Pitcairnia utilize more pervasive, warm and humid ecospace that also features numerous seed-transporting waterways. Seed morphology and overoccurrence along streams further suggest that hydrochory in¯ uenced dispersal by members of this genus. Climate affects the distributions of many Tillandsioideae through constraints related to plant architecture in addition to ecophysiology. Considerable predominantly soft-leafed, tank-dependent Tillandsioideae occupy con® ned ranges in montane regions consistent with their relatively stringent requirements for high humidity and perhaps moderate temperatures (Chapter 4; Rossi et al. 1997). Conversely, Type Five Tillandsia occur at high to low altitudes under widely varying humidities. The lowland natives (e.g., T. balbisiana, T. schiedeana), which tolerate severe evaporative demand, range more broadly, in part again because this kind of habitat occurs so extensively through tropical America. Species achieve high densities locally in both hot and cold deserts, for example in Mexican thorn forests and in higher (.2000 m) Andean sites. Cloud-dependent types (e.g., T. tectorum), although often Cambridge Books Online © Cambridge University Press, 2009 Phytogeography 483 common in favorable sites, exhibit greater insularity consistent with the vertically compressed (cloud zones) and horizontally dissected nature of their montane habitats. Excessive warmth and seasonal drought probably explain the relative poverty of Amazonian Bromeliaceae compared with the adjacent highlands, but causes may vary with the taxon. Simple explanations such as the absence of the cool nights supposedly necessary to foster vigorous CAM lack currency (e.g., McWilliams 1974; Chapter 4). CAM characterizes extensive Bromelioideae native to equally tropical sites elsewhere (e.g., Atlantic Forest). Similarly, CAM-equipped Tillandsia range from sea level to above 3000 m in sites arrayed from Mexico to Argentina. Conditions in the drier and most bromeliad-de® cient parts of Amazonia (central and eastern regions) favor neither phytotelm (C3) nor driergrowing Tillandsioideae. Not many Bromelioideae or Pitcairnioideae occur here either. Equally pronounced dry seasons at higher (cooler) elevations to the west constrain the family far less, and certain Tillandsioideae (Guzmania) and Pitcairnioideae (Puya, Pitcairnia) achieve unparalleled diversities in pluvial Andean habitats adjacent to western Amazonia. Perhaps Amazonia limits the possibilities for Bromeliaceae, and especially Tillandsioideae, more than most other regions inhabited by the family because it offers fewer combinations of acceptable growing conditions. Most of the local bromelioids are epiphytes, and many possess bulbous shoots that feature a well-protected phytotelmata and accommodate ant colonies in the drier recesses of the younger foliage (Fig. 2.4G). Other species root in arboreal ant nests (e.g., Aechmea mertensii, Neoregelia myrmecophila; Fig. 8.1C). Abundant moisture and heat during the wettest months probably exclude Type Five Tillandsia for reasons described in Chapter 4. Finally, monotonously low, compared with the more dissected Andean and Guayanan, topography further militates against diverse arboreal ¯ oras, but this constraint overlaps with the previous one. A recent assessment (Ibisch et al. 1996) of Peruvian Bromeliaceae and co-occurring families addresses three of the same questions just considered, namely: has epiphytism inordinately favored speciation; what can phytogeography tell us about the evolutionary history of a clade and whether characteristics of the membership in¯ uenced its size and range; and why are the Amazonian bromeliads so few and similar in adaptive type? Ibisch et al. concluded that epiphytism does not promote extraordinary rates of speciation (`abnormal evolutionary activity' ), at least not in Peru. They also noted that epiphytes sometimes contribute substantial Cambridge Books Online © Cambridge University Press, 2009 484 History and evolution Figure 9.6. Occurrence of species representing different habits in dry, moist and wet Ecuadorian forest (after Gentry and Dodson 1987). diversity to local ¯ oras (up to 35%), but importance to species richness diminishes as spatial scale increases. Ibisch et al.' s conclusion about arboreal vs. terrestrial habits and speciation rests on evidence from plant geography, speci® cally that the obligate bromeliaceous epiphytes of Peru (28.8% of the 452 species) range more widely than the terrestrials (true also for Araceae, Orchidaceae and Piperaceae). Endemism (i.e., con® nement of a species within the political boundaries of Peru) is just 10.1% for the epiphytes of Bromeliaceae and 19.7% for those of Tillandsia, the largest of the Peruvian genera representing this family. Corresponding ® gures for the terrestrials are 76.6 and 87.5%. Because the authors equated insularity with evolutionary youth (reminiscent of the logic behind the cohesion ratio), lineages with terrestrial habits were deemed more recently derived on average than those of the related epiphytes. Ibisch et al. consider the epiphytes to possess greater `ecophysiological plasticity' and accordingly, less propensity to speciate in response to growing conditions that diverge as ranges expand. Related terrestrials, they say, instead `tend to ® ner niche-tuning' , which involves more `genetic separation and subsequent speciation' . In fact, regional patterns of epiphyte diversity, including Bromeliaceae, show arboreal ¯ ora to be far more sensitive to climate than co-occurring soil-rooted shrubs, vines and trees (Gentry and Dodson 1987; Benzing 1990; Fig. 9.6). Cambridge Books Online © Cambridge University Press, 2009 Phytogeography 485 Figure 9.7. Map of Peru indicating major life zones and associated diversities of vascular epiphytes. Bromeliaceae represent about 8.5% of Peruvian epiphytes and concentrate in lower montane rainforest as do their counterparts in Orchidaceae, Araceae and Peperomia which constitute 78.0, 5.2 and 5.0% of the arboreal species respectively (after Ibisch et al. 1996). Conceivably, epiphytic Bromeliaceae exhibit relatively low endemism in Peru because most of them reside (many sympatrically) in the everwet, lower montane life zone that continues unbroken along the eastern slopes of the Andes into Ecuador and Colombia at its north end and into Bolivia to the south (Fig. 9.7). Species/area curves that demonstrate decreasing importance to overall species richness for the epiphytes relative to the Cambridge Books Online © Cambridge University Press, 2009 486 History and evolution Figure 9.8. Generalized species/area curves of epiphytes and terrestrials in epiphyterich forest based on observations in Peru. Arboreal ¯ ora lose relative importance as the size of the area considered increases (after Ibisch et al. 1996). terrestrials accord with differences in average plant size and the related tendency for arboreal ¯ ora to occur at higher densities than soil-rooted types (Fig. 9.8). Rather than habit-speci® c differences in `ecological plasticity' , other plant characteristics and statistics biased by Puya and the inordinate occurrence of lithophytic Tillandsioideae in Andean habitats may further account for Ibisch et al.' s conclusion about plant habits and patterns of speciation. Seed mobility more plausibly affects phytogeography and varies greatly among Bromeliaceae, and especially among the natives of Peru. Peruvian Puya (73 wholly terrestrial species) exhibits 89% endemism, which is not surprising given its frequent co-occurrence with only one to a few closely related species and immobile seeds compared with those of anemochorous Tillandsioideae and largely zoochorous Bromelioideae. High endemism in Puya is also consistent with the same elevational trend expressed by two (Orchidaceae, Piperaceae) of the other three families they surveyed. Insularity should increase as habitats become more alpine and, ipso facto, dissected by deep valleys. Bromelioideae differ in mobility and exhibit high insularity in another part of South America. Strictly Brazilian and unfailingly terrestrial Cambridge Books Online © Cambridge University Press, 2009 Phytogeography 487 Orthophytum and Cryptanthus (Fig. 1.3D) bear relatively dry fruits (Fig. 3.6E) compared with predominantly arboreal and wider-ranging relatives (e.g., Aechmea, Billbergia; Fig. 3.6C,D,G). Much the same can be said of certain Tillandsioideae even though the entire subfamily disperses via comose seeds. Exclusively lithophytic Alcantarea produce less airworthy propagules than largely epiphytic Vriesea and Guzmania, probably re¯ ecting ranges that often include but one island-like inselberg (Fig. 1.4A). Bennett (1992c; Table 6.6) demonstrated similar differences among seeds that co-vary with bark or rock as the substrate among members of a group of Tillandsia species. Compared with the typical terrestrial bromeliad, the epiphytes indeed may be less fastidious about substrates (recall the low host speci® city described in Chapter 7), as Ibisch et al. suggest. Moreover, Amazonian Bromeliaceae may range widely (and remain depauperate) in part because alternating hot, dry weather and hyperhumidity characterize vast regions also largely free of topographic barriers to plant dispersal. Hyperdispersed populations probably assured that competition would not appreciably in¯ uence the number of Amazonian bromeliads despite their considerable similarity (e.g., frequent shared dependence on phytotelmata and plantfeeding ants). Speci® cally, plant interference would not promote character displacement leading to speciation. For whatever reason, a much underutilized (empty) living space has not fostered the species-rich arboreal ¯ ora that a lottery-type mechanism may favor (Benzing 1981b) elsewhere, including those communities in adjacent regions (e.g., Peru, Colombia) characterized by continuously humid montane habitat (and often many, ecologically similar, populations). In our view, inquiry on the importance of plant habit to speciation in Bromeliaceae should be pursued by ® rst determining how many members of the largest groups of related, ecologically mixed species are epiphytic or terrestrial. A second step requires determinations of the degrees of interrelatedness of the arboreal and soil-dependent populations within these groups. Recent, active speciation should yield clusters of genetically similar descendants, and a slower version of the same process, the opposite outcome. Essential to this kind of assessment is a reliable measure of evolutionary (genetic) identity, probably an index based on multiple DNA sequences. Current taxonomic boundaries are not reliable indicators of relationship, and phytogeography alone lacks the power to reconstruct phylogeny, and hence determine whether the size of a constellation of species parallels its age. Ibisch et al. speak of combining taxonomic, ¯ oristic and life-form Cambridge Books Online © Cambridge University Press, 2009 488 History and evolution analysis to investigate the relationship between propensity to speciate and plant habit ± in this case epiphytism vs. terrestrialism. We applaud this suggestion and emphasize the importance of also recognizing the many additional plant characteristics and aspects of habitats (e.g., extent, continuity in time and space, amenability to partitionment by epiphytes) that affect the geographic distributions of ¯ ora. Answers to questions as fundamental as the contribution of one habit over another to the expansion of a clade and its range require an approach commensurate with the complexity of the phenomenon under scrutiny. Chromosomes, hybridization and polyploidy Bromeliaceae have experienced some polyploidy, probably including one particularly notable event, and timing suggests that consequences for evolution were greater early rather than more recently during family history (Brown and Gilmartin 1986, 1989a). Unlike the grasses, Rosaceae and the other families that immediately come to mind as exemplary of extensive polyploidy and hybridization, the bromeliads possess a relatively constant high base number. Here and there a cluster of species exhibits twice and occasionally threefold this number of chromosomes. Ananas comosus is triploid. However, counts exist for only about 1 in 10 binomials, and sampling has been uneven. No reports exist for a substantial portion of the smaller and mediumsized genera, several of which (e.g., Brocchinia, Fascicularia, Greigia, Navia) combine other characteristics that suggest considerable distance from the genera that constitute the cores of their respective subfamilies (sensu Smith and Downs 1974). Tillandsioideae and Bromelioideae distinguish Bromeliaceae among ¯ owering plants less by polyploidy than by their displays of heteromorphic karyotypes (Fig. 9.9) and discordant counts in root tips vs. microsporocytes. Small chromosomes, aneuploidy (e.g., Tillandsia complanata, n520 or 22) and occasionally phenotypically undifferentiated polyploid races further impede attempts to reconstruct the origin of the bromeliad karyotype and identify possible incidences of reticulate evolution. Lowest common denominator derivatives (e.g., base number of 2n516, x58) inspired early hypotheses concerning base chromosome numbers in Bromeliaceae (Billings 1904; Lindschau 1933; Weiss 1965; Marchant 1967; Sharma and Ghosh 1971; McWilliams 1974). Brown and Gilmartin (1989a) modeled karyotype evolution using phylogenetic evidence that favored the then popular notion that Bromeliaceae and Velloziaceae (x58) Cambridge Books Online © Cambridge University Press, 2009 Chromosomes, hybridization and polyploidy 489 Figure 9.9. Microsporocyte from Vriesea schwackeana demonstrating bimodal karyotype. are sister taxa (e.g., Huber 1977; Dahlgren and Rasmussen 1983; Dahlgren et al. 1985; Gilmartin and Brown 1987; Ranker et al. 1990). Their scheme, which also accords with Dahlgren' s Bromelii¯ orae, imputes the origin of a dibasic x517 lineage via hybridization between x58 and x59 parents, followed by a second cross with another x58 lineage to yield x525 (Fig. 9.10). However, ® ndings from rbcL sequences (Clark et al. 1993; Duvall et al. 1993) that identify Rapateaceae as the sister family for Bromeliaceae contradict Brown and Gilmartin' s model. Unfortunately, only one count (2n522), for African Maschalocephalus, represents Rapateaceae in the literature. Moreover, the occurrence of the other 15 genera in South America Cambridge Books Online © Cambridge University Press, 2009 490 History and evolution Figure 9.10. Proposed model for chromosome base number evolution in Bromeliales. The extant base number n525 is synapomorphic for Bromeliaceae and derived by hybridization and polyploidy involving a paleodiploid (n58) and paleotetraploid (n517). The dibasic paleotetraploid developed from hybridization and polyploidy involving paleodiploids n58 and n59 (after Brown and Gilmartin 1989a). Cambridge Books Online © Cambridge University Press, 2009 Chromosomes, hybridization and polyploidy 491 renders Sharma and Ghosh' s (1971) consequent assignment of x511 for the entire family especially tenuous. Although chromosome numbers suggest identity between Cryptanthus, the only bromeliad genus with x517, and the ancient paleotetraploid, descending aneuploidy offers a second, and according to Brown and Gilmartin (1989a), Brown and Palací (1997) and the ndhF data, more plausible explanation. Higher than expected amounts of nuclear DNA in cells with only 34 chromosomes (e.g., C. acaulis, C. beuckeri) also suggest aneuploidy, i.e., condensations of formerly discrete chromosomes (Ramírez 1996; Chapter 11). Less dramatic reduction beginning with the x525 condition occurs elsewhere (e.g., Aechmea tillandsioides, n521; Tillandsia umbellata, n518; T. leiboldiana, n519). Brown and Gilmartin further considered Cryptanthus too specialized for paleotetraploidy by virtue of its polygamous breeding system (subgenus Cryptanthus only), which has no equivalent elsewhere in Bromeliaceae. They also cited interfertility between Cryptanthus beuckeri and C. bahianus and Billbergia nutans as additional evidence of a more contemporary than relictual status. Comprehensive sampling of Cryptanthus and closely allied Orthophytum, which lack a single chromosome count, might help reveal how the number 25 originated and its relationship to 17. Arti® cially produced, and probably also the natural, hybrids in Tillandsioideae outnumber those in the other two subfamilies even though many Bromelioideae also exhibit interspeci® c fertility (Table 6.2). Most of these intermediates appear to be diploids. Two species of Bromelia, and one each in Ananas, Nidularium and Pseudananas, reportedly possess duplicated sets of chromosomes (Brown and Gilmartin 1986). Polyploids also exist in Dyckia and Fosterella, and one Guzmania specimen yielded a suspiciously high chromosome count. Tillandsia, especially subgenus Diaphoranthema, which includes the most miniaturized (heterochronic) species in the genus, tops the list of exceptions. Of the 20 populations examined, 12 produced tetraploid ® gures, and another (T. capillaris; Till 1992a; Chapter 13) a hexaploid number. Only eight taxa remain diploid and diminutive Tillandsia loliacea includes diploid and tetraploid races. Polyploidy also accompanies ¯ oral morphology that promotes autogamy through much of subgenus Diaphoranthema. Flowers of Tillandsia capillaris f. hieronymi fail to open, yet fruits usually result (Gilmartin and Brown 1985). Several relatives approach this condition. Till (1992a) reported additional cases of cleistogamy (e.g., T. angulosa, T. castellanii, T. landbeckii subsp. landbeckii) and autogamy sometimes associated with larger, fragrant ¯ owers (e.g., T. myosura, T. virescens). Facultative apomixis Cambridge Books Online © Cambridge University Press, 2009 492 History and evolution and routine inbreeding, combined with polyploidy and variable karyotypes (4n584± 96) may underlie the exceptional polymorphism exhibited by members of the T. capillaris complex, and perhaps also their site-speci® c propensities to root exclusively on bark or rock. Miniaturization associated with paedomorphosis probably obliged autogamy as Diaphoranthema radiated and plants became too small to support ¯ owers large enough to attract most pollinators (Till 1992a; Chapter 6). Conceivably, cleistogamy is conserving gene combinations well suited for the stringent growing conditions encountered by vascular epiphytes and saxicoles of such vulnerable sizes. Cleistogamy, unusual architecture fostered by heterochrony and polyploidy all affected evolution, but how in this instance and whether synergistically remains unclear. Closer examinations of the breeding systems and cytology of subgenus Diaphoranthema and less specialized and probably paraphyletic Phytarrhiza should be informative. Bimodal karyotypes exceed the incidence of polyploidy in Bromeliaceae, and exceptionally small chromosomes characterize the entire family (Fig. 9.9). Palací (1991) discovered an even more unusual condition among populations comprising the Tillandsia friesii complex (subgenus Anoplophytum). These plants possess chromosomes graded into four classes: relatively large ones, those of medium size with a satellite at the end of one arm, others of similar proportions but with no satellites, and a ® nal category for the small, `dot-like' members. However, claims (e.g., McWilliams 1974) that the heteromorphic karyotype consistently cooccurs with extraordinary morphology and propensities for unusual substrates (absent in Pitcairnioideae, maximal in Tillandsioideae) lack foundation. Also perplexing are the lower counts obtained from root tips compared with pollen mother cells from the same epiphytes and lithophytes. According to Brown and Gilmartin (1986), the primarily mechanical root may tolerate levels of aneuploidy unsustainable in organs required to perform more exacting physiological functions. Certainly the sclerenchymatous nature of the roots of Tillandsioideae and some Bromelioideae make counting difficult enough to favor false readings (Fig. 2.15). Karyotypic asymmetry is no less real, but whether its association with specialized structure and function indicates importance as a mechanism to perpetuate speci® c genotypes for demanding habitats requires con® rmation. Cambridge Books Online © Cambridge University Press, 2009 Ancestral habitats 493 Ancestral habitats A. F. W. Schimper' s (1884, 1888, 1898) declaration that the vascular epiphytes evolved from stocks native to the understories of humid tropical forests continues to provoke conversation and inquiry. Pittendrigh (1948) rejected this judgment for the bromeliads following extensive observations in Trinidad. He reported instead that arboreal Bromelioideae and Tillandsioideae arose from more light-demanding, arid-land ancestors. Colonization of the lower canopy by Tillandsioideae supposedly occurred from the top down as the well-developed indumentum still featured by the presumed plesiomorphic Type Five Tillandsia (Fig. 2.8C,E) diminished to what today serves Type Four bromeliads (Fig. 2.8B). At the same time, progressively in¯ ated leaf axils provided the continuous supply necessary to meet the growing demand for water obliged by increasingly more droughtsensitive and shade-adapted foliage (Fig. 1.2G). Pittendrigh proposed that Bromelioideae changed less during the same transition. Most of its epiphytic members supposedly remain impressively stress-tolerant, and they anchor on arid and sun-exposed substrates in tree crowns much as ancestors did and some modern forms continue to do on the ground (Fig. 1.3E). Many arboreal and terrestrial types, like the majority of Tillandsioideae, also rely on phytotelmata, in this case elaborated from the smaller impoundments featured by antecedents with more conventional Type Two architecture (Fig. 2.14A,B). Terrestrial Bromelia humilis and relatives with comparable morphology (Type Two; Table 4.2), Pittendrigh reported, exemplify the primitive bauplan. Life in the canopy was also presaged by the presence of foliar trichomes, that, if as competent as those of Ananas comosus (Sakai and Sanford 1979), already supplemented the nutritive ions and moisture that ancestors and some extant members of this subfamily still obtain from soil (Fig. 1.3E). According to Pittendrigh, the argument that Tillandsioideae colonized drier parts of the forest canopy ® rst and wetter sites later rests on two points. First, Tillandsia retains what taxonomists consider the least specialized of the ¯ owers present in the subfamily (Smith and Downs 1977). Second, most dry-growing Tillandsioideae belong to this same genus. In essence, ¯ oral morphology and several vegetative characteristics that affect water relations supposedly evolved in concert. Like Tietze (1906), Pittendrigh also considered the foliar trichome unlikely to have replaced the root as the primary absorptive organ under humid conditions, (i.e., in Schimper' s understory habitats). Findings since 1948 have persuaded several subsequent authors to take opposing views on the evolutionary Cambridge Books Online © Cambridge University Press, 2009 494 History and evolution status of bromeliad xerophytism and the origins of speci® c features related to water balance like the phytotelm shoot and absorbing scale. Pittendrigh' s decision to extrapolate the ecology of ancestors from that of extant descendants drew less criticism ® ve decades ago than it should today. Also troublesome is the evidentiary basis of his judgment about optimum conditions for speci® c taxa, viz. plant distributions relative to prevailing growing conditions in lieu of the more immediate measures of plant performance possible today. As it turns out, several of Pittendrigh' s shade-tolerant but supposedly heliophilic subjects, and certain other thinleafed Tillandsioideae, exhibit low light compensation and saturation intensities (e.g., Benzing and Renfrow 1971b; Smith 1989; Fig. 4.7), although not as low as those recorded for some other understory plants considered shade specialists. Moreover, much vegetation, including Bromeliaceae, acclimates across steep light gradients, and closely related genotypes often thrive under widely divergent exposures. Several authorities since Pittendrigh have also adopted rigid notions about the ecological consequences of speci® c plant characteristics, and how rapidly some of these features can evolve (e.g., Ortlieb and Winkler 1977; Smith 1989; Fig. 9.11). Recent ® ndings on the ecophysiology of diverse tropical ¯ ora exemplify the ® rst problem. Certain extant CAM plants, including several Bromeliaceae, tolerate unexpectedly deep shade (e.g., Aechmea magdalenae; Figs. 4.4, 4.5) in humid habitats, and Type Two Ananas and Bromelia so often perform best in the forest understory that Medina et al. (1986, 1991a) concluded that they probably originated there. On the second issue, mixed responses among a collection of Ananas comosus cultigens observed by the same investigators illustrated how selection by indigenous farmers has altered light relations among closely allied genotypes. Guzmania monostachia with its facultative CAM and sun and shade phenotypes illustrates similar ecophysiological variety of natural origin (Table 4.6; Chapter 4). Current ecophysiology may reveal fundamentals that constrained evolutionary options for clusters of closely related lineages (e.g., occurrence of all Nidularium in dark, moist habitats). But do the bromeliads collectively exhibit sufficiently consistent light relations to justify conclusions about conditions in ancestral habitats? Fully one-third of the bromelioid genera (e.g., Disteganthus, Lymania, Ronnbergia, Nidularium) contain a preponderance of species largely relegated to the forest understory; many more (e.g., Aechmea, Bromelia, Canistrum, Cryptanthus) exhibit mixed tolerances for exposure. Conceivably, deeper understanding of the structure of the photosynthetic apparatus will help reconstruct the historic responses of Cambridge Books Online © Cambridge University Press, 2009 Figure 9.11. Postulated phylogenetic relationships within Bromeliaceae based in part on the taxonomic distribution of CAM and C3 photosynthesis at the level of genera. According to this scheme, the epiphytic habit and CAM arose more than once. A progressive loss of CAM is indicated for Bromelioideae (after Smith 1989). Cambridge Books Online © Cambridge University Press, 2009 496 History and evolution individual lineages to light, just as the capacity to sequence DNA already provides a way to quantify relationships among genes of interest (and indirectly the plants that possess them). More can be said about ¯ exibility among closely related genotypes. Within Tillandsioideae, Tillandsia excels for ecological versatility; its phytotelm forms, especially those with discolorous and sparsely trichomed leaves displayed in minimally overlapping arrays (e.g., T. monadelpha), rank among the most shade-tolerant of all the bromeliads. Close relatives (same subgenus) occupy desert and lithic habitats that subject resident ¯ ora to undiminished sunlight (e.g., Brazilian T. kurt-horstii; Fig. 4.23D,E). Quite likely, options for the descendants of early Bromeliaceae exceeded those articulated by Pittendrigh enough to allow dark and moist and brighter and drier sites in the canopy and on the ground to be colonized by close relatives. Pittendrigh' s notion about concerted evolution fails to recognize that ¯ oral characters need not change synchronously with those that dictate vegetative function, nor does reproductive morphology always parallel phylogenetic relationship (e.g., Benzing and Renfrow 1971c; Brown and Terry 1992; Till 1992a). Mosaic evolution applies as much to Bromeliaceae as to the rest of the tracheophytes. No less problematic is the issue of which growing conditions fostered speci® c kinds of evolutionary change. Selection by an agency other than drought, or if by drought then by a mechanism involving heterochrony, may explain why the foliar trichome achieved certain attributes central to plant survival. However, Tietze and Pittendrigh' s notion about xerophytic ancestry and why the bromeliad trichome became a root analog represents just one of several possibilities, and not necessarily the most compelling one. Medina (1974) chose a more pervasive plant condition on which to base his version of bromeliad ancestry. Antecedents were supposedly similar to certain extant Guayanan Pitcairnioideae (e.g., Brocchinia, Navia) native to humid, sunny, and hence relatively permissive, habitats. Exceptional drought-tolerance, assisted by CAM, like that expressed by modern Dyckia, Hechtia, many Tillandsia and most Bromelioideae, came later as evolving lineages entered drier zones, including the forest canopy. Tillandsioideae supposedly adopted its tolerance for diverse growing conditions along two routes that began with a common semixerophytic stock at or around the time epiphytism emerged. Thin-leafed, shade-growing types capable of high quantum yields at low photosynthetic photon ¯ ux density (PPFD) and poorly suited to resist desiccation represent one end point (Fig. 4.7). The other progression, which would culminate in Type Five Cambridge Books Online © Cambridge University Press, 2009 Ancestral habitats 497 Tillandsioideae, evolved to accommodate the dryness and exposure experienced in tree crowns and on rocks and desert soils (Fig. 2.1). Smith (1989) associated himself with Medina, likewise rejecting Schimper' s hypothesized ancestral habitats in the forest understory. He offered no comment about Benzing and Renfrow' s (1971b,c) evidence that several of Pittendrigh' s shade-tolerant bromeliads perform well in lowenergy environments, and, more importantly, provide no persuasive evidence of earlier accommodations to higher PPFD. Smith' s scheme (Fig. 9.11) imputes monophyly for the family, a common mesophytic (C3) stock, and the multiple origins of CAM suggested by Medina. It aligns Bromelioideae and Pitcairnioideae as sister taxa that shared a C3-type ancestor after Tillandsioideae branched off the same lineage. Smith and Medina joined Pittendrigh and Tietze in citing aridity to explain the absorptive capacity of the more specialized bromeliad trichome. Foliar indumenta serving Type Five Tillandsioideae probably do possess unparalleled capacity among homiohydrous ¯ ora to mediate water balance in the absence of absorptive roots (Figs. 4.20, 4.21). Moreover, impact on water relations is so striking that it obscures the trichome' s less conspicuous role in mineral nutrition, and accordingly, perhaps why this organ became a root substitute in the ® rst place. If the growing conditions experienced today by the Guayanan bromeliads Medina considered similar to family ancestors also prevailed in ancestral habitats, services provided by the foliar indumentum may have been less comprehensive then than now. Quite possibly the multifunctional trichome of dry-growing Tillandsioideae owes its debut as a replacement for the absorptive root to needs equivalent to those provided for carnivorous Brocchinia reducta, which involve mineral nutrition more than water balance. This species regularly grows in humid habitats, often rooting in seasonally water-saturated soils. If the trichome of ancestral Tillandsioideae (perhaps the stock for Brocchinia as well) was similar in evolutionary grade to that of B. reducta, drought would encourage its further elaboration to provide the greater variety of services Type Five Tillandsioideae require of the foliar indumentum. A third scenario, involving heterochrony, explains how conditions in humid forest could bring about the same outcome. If nothing else, Brocchinia illustrates how rapidly the structure and function of the bromeliad trichome can shift as ecology, shoot architecture and nutritional mode change (Fig. 2.5A,B,G± J). According to Smith (1989), Schimper' s ideas also fail to explain the ecological history of Tillandsioideae because no members with mesic shoots and soil roots routinely reside in the understory of the humid tropical Books Online © Cambridge University Press, 2009 498 History and evolution forest, at least not in Trinidad. But elsewhere they do. Numerous caulescent forms without overlapping leaf bases (e.g., Tillandsia insignis, Guzmania caricifolia, G. graminifolia, and additional species assigned to Guzmania subgenus Sodiroa) inhabit cloud forest where their roots interchangeably anchor shoots to soil and bark. Relatives native to more demanding sites provide no better support for the Tietze/Pittendrigh hypothesis considering that they deviate even more than the mesic forms from the profusely rooted monocotyledonous stereotype (Fig. 2.20). Members of several bromelioid genera (e.g., Aechmea, Ronnbergia; Figs. 2.2F, 2.4E) native to similarly moist habitats exhibit equally generalized monocot architecture. Many more taxa illustrate how readily colonists from the understory can move to the forest canopy, or as lithophytes also become independent of soil. Tillandsioideae native to the rupestral ® elds of southeastern Brazil support Medina' s proposed parallel between the evolution of Brocchinia and more extensively epiphytic Tillandsioideae. Alcantarea farneyi, A. hatscbachii and A. duarteana native to the stony grasslands of Minas Gerais and neighboring Bahia states, along with Brocchinia on comparably ancient terrestrial media, suggest how readily mesic lineages can exchange arboreal and lithic for more conventional terrestrial substrates (Figs. 1.2B, 1.4C). These three Alcantarea species resemble the grass-like pitcairnias more than their phytotelm relatives; they also impound no more moisture or debris in leaf bases than Pittendrigh' s Type Two Bromelioideae. By contrast, Alcantarea edmundoi, A. imperialis and A. regina, among others, possess massive shoots that accumulate many liters of material at maturity (Figs. 1.2C, 7.1D). Recall that Brocchinia also includes savanna endemics without appreciable interfoliar chambers (e.g., B. prismatica, B. vestita; Fig. 5.3D) and others (B. tatei; Fig. 1.2B) with phytotelma rivaling those of hundreds of utriculate-leafed species assigned to the other two subfamilies (Fig. 2.4). Clearly the evolutionary distance separating a widely occurring monocot architecture from the arrangement that allows many Bromeliaceae to create a soil substitute is modest. Narrow-leafed members of Brocchinia and the previously mentioned species of Alcantarea with the same architecture may owe their shared upright stature to the conditions Medina (1974) envisioned for his ancestral Bromeliaceae. Graminoids and the other herbs with upright, narrow foliage responsible for the shallow, but dense, canopies of grasslands overtop co-occurring ¯ ora equipped with relatively ¯ at rosettes. The rosulate bromeliad competing for light in such communities may well have responded by evolving similar upright leaves, and consequently, modest impoundment capacity. But even if true, is this condition in Brocchinia and Cambridge Books Online © Cambridge University Press, 2009 Ancestral habitats 499 Tillandsioideae and also certain Bromelioideae basic or derived (Fig. 2.4E)? Then again, perhaps emphasis on the characteristics of the adult is misplaced ± at least so far as explaining the relationship between Type Four and Type Five Tillandsioideae is concerned. The evolutionary pathway suggested by two heterophyllous members of Tillandsioideae also presumes a mesic stock, and perhaps epiphytism prior to the emergence of the more stress-tolerant, neotenic forms (Fig. 2.1). This pathway also imputes the existence of an ancestor that was already reliant on absorptive foliage and a phytotelma in place of a root system that had formerly provided access to soil. Contrary to several other authorities (e.g., Schulz 1930; Tomlinson 1970; Benzing et al. 1985), Medina (1974) rejected heterochrony to explain the xeromorphic character of the seedlings of phytotelm Tillandsioideae. Instead, these features were dismissed as vestigial, simply recapitulations consistent with antecedents that were generally more xeromorphic than their descendants. Findings reported by Adams and Martin (1986a), Reinert and Meirelles (1993) and Zotz and Andrade (1997) challenge Medina' s hypothesis that juveniles and adults of soft-leafed taxa share important aspects of water balance because they experience the same climate. His view ignores certain size vs. form-related determinants of ecoperformance that change during ontogeny (Figs. 4.9, 4.17). Speci® cally, it overlooks the coupling of plant morphology, especially surface to volume ratio, and rates of CO2 and H2O exchange. Adams and Martin and Reinert and Meirelles based their interpretation of heterophylly on current plant function vis-à-vis prevailing growing conditions. They chose not to assign relictual status to form that appears to grant the juvenile greater drought-tolerance than if it were constructed more like the adult. Tested seedlings of heterophyllic Tillandsia deppeana desiccated more slowly and continued to photosynthesize longer without irrigation than the adults in accordance with expectation based on needs in native habitats (Fig. 4.9). Tanks sustain the mature specimen through dry weather, whereas the juvenile routinely faces drought unassisted by access to a comparable reservoir. Episodes beyond a few rainless days challenge water balance until the seedling, being effectively rootless, achieves capacity to impound substantial moisture, a process that requires many months to years. Until then, the phytotelm bromeliad necessarily operates in the Type Five mode, surviving as its more stress-tolerant relatives do by virtue of their specialized indumentum and underlying water-storage tissues. Reinert and Meirelles (1993) projected the same kind of performance for Vriesea geniculata for the same reasons. Cambridge Books Online © Cambridge University Press, 2009 500 History and evolution According to the hypothesis involving heterochrony, the trichome of Tillandsioideae initially became a root analog to serve the relatively vulnerable seedling more than the better-supplied adult. That event in turn set the stage for the neotenic derivation of more broadly drought-tolerant (Type Five) descendants (Fig. 2.1). Conceivably, an ancestor already equipped with absorbing scales to complement its other features as a semixerophyte could have produced multiple epiphytic stock that, as Medina envisioned, subsequently radiated into both more humid and drier parts of the canopy. Absorptive function may also have appeared without great immediate consequence in the manner described above for Brocchinia. Whatever the pathway, the trichome shield would evolve its most elaborate form when environments favored an indumentum, that in addition to absorbing water and essential ions slows transpiration and scatters excess photons off dry leaf surfaces. Whatever the nature of the early bromeliads or their habitats, much homoplasy involving bauplan, important aspects of organs like the foliar trichome, and ecophysiology in¯ uenced the dimensions and directions of the ensuing radiation. CAM, xeromorphy and phytotelm shoots all emerged repeatedly, and occasionally derived conditions returned to former states. Such a reversal appears to have allowed Nidularium to penetrate exceptionally dark, moist habitats in Atlantic rainforest as terrestrials and trunk epiphytes. Presumably, C3 Greigia evolved from comparable ancestry to accommodate cooler, moister montane habitats. Brocchinia features a set of homoplasies that involve the same characters that have been most instrumental for the success of Bromeliaceae in diverse, often demanding habitats, especially the forest canopy (Fig. 9.5). Of all the genera, this one comes closest to paralleling the adaptive history of the entire family. Heterochrony Heterochrony was just invoked to suggest how dry-growing (Type Five) Tillandsioideae evolved from stocks comparable in form and function to extant lineages characterized by phytotelma and thin, sparsely trichomed foliage (Type Four; Fig. 2.1). Miniaturization distinguishes members of these two evolutionary grades as does architectural abbreviation in a manner consistent with similar phenomena in many other groups of plants and animals (Hanken and Wake 1993). Presumably bene® ts related to anchorage on bark and rock as opposed to more resource-rich soil favored this transition in Bromeliaceae. Cambridge Books Online © Cambridge University Press, 2009 Heterochrony 501 Before moving on to greater detail, we need to consider how heterochrony can alter plant form and function enough to propel a lineage from one adaptive zone or evolutionary grade to another. Examples elsewhere indicate that Bromeliaceae ® t a familiar pattern on this basis, although among paedomorphic ¯ ora the combination of affected plant characteristics is quite novel. Moreover, change ranges from nonexistent in some members of the family to revolutionary for others. Genes expressed early during ontogeny are better timed to affect the phenotype of the adult than those expressed later. Re-program a morphogenetic cascade and the shapes, numbers and relationships among organs may shift with potential consequences for function, life history and ultimately ecology. Organ redundancy may occur, for example the leaves on a shoot may increase or diminish in number, or fail to develop altogether as when thalloid Lemnaceae emerged from some aroid-like stock. Here, as in Tillandsioideae, miniaturization paralleled the emergence of a simpler bauplan. Also like Type Five Tillandsioideae, the duckweeds occupy specialized environments ill suited for larger plants equipped with conventionally organized shoot and root systems. Several invertebrate phyla (e.g., Gnathostomulida, Loricifera) exhibit comparable reductions and structural reorganizations to accommodate life within the interstitial spaces of suitably textured soils (Hanken and Wake 1993). Comparative morphologists recognized the interconnectedness of ontogeny, evolution and phylogeny more than a century ago, and since then have employed this paradigm to reconstruct evolution, primarily morphological change. Heterochrony explains the polarities of certain graded character states among related organisms, ranging from ® ne details such as ovary position or ¯ ower type (e.g., cleistogamous vs. chasmogamous ¯ owers of certain Impatiens spp.) to the total transformation of a bauplan as in the duckweeds. Bromeliaceae include a more complete array of transitional forms than Lemnaceae, and these survivors demonstrate how the numbers, shapes and sizes of body parts, and ultimately the design of the entire organism, evolved as the rates and timing of discrete morphogenetic events changed. By de® nition, heterochrony includes developmental and phylogenetic components. However, like all Darwinian processes, effects on ® tness in¯ uence magnitude and direction. Edwardo Morren, the Belgian horticulturist, formally recognized heteroblasty, if not heterochrony, when he described Tillandsia heterophylla more than a century ago. Subsequent authors noted similar distinctions between the early and later life stages of additional Type Four Tillandsioideae (e.g., Schulz 1930; Tomlinson 1970). Seedlings representing many of the Cambridge Books Online © Cambridge University Press, 2009 502 History and evolution Figure 9.12. An unidenti® ed Nidularium sp. seedling growing on rock in Rio de Janeiro State, Brazil. Were adult foliage also present, heterophylly would be apparent. phytotelm species remain relatively succulent and compact for several years until adult foliage emerges. Bromelioideae may provide a more muted and so far less-studied parallel. Some of its most extraordinary members, like heterophyllic Neoregelia abendrothae, produce shoots comprised at ® rst of narrow leaves and later of others with water-tight, in¯ ated axils and broader, channeled blades (Fig. 2.2D). Figure 9.12 illustrates the seedling of an unidenti® ed Nidularium growing on a mossy rock in a Brazilian wet forest. Note that this plant lacks the capacity to impound water like the adult. Nevertheless, the ® liform leaves already possess the serrated margins characteristic of its subfamily. Schulz (1930) commented on the relatively xeromorphic character of the seedlings of certain Type Four Tillandsioideae. More detailed studies using Tillandsia deppeana (Adams and Martin 1986a,b,c) and Vriesea geniculata (Reinert and Meirelles 1993) con® rmed his impressions and further dem- Cambridge Books Online © Cambridge University Press, 2009 Heterochrony 503 onstrated correlated shifts in water relations as described above and in Chapter 4. Features of the epidermis that affect gas exchange and energy balance also change as the shoot matures and develops foliar impoundments (Table 4.7). Brie¯ y, trichomes of the juveniles more densely (80% for saxicolous Vriesea geniculata) invest the blades and possess broader shields than those of the adults. Densities of stomata also increase over the blade, but not the leaf base, as plants age. On the other hand, CO2 exchange and D values indicated life-long dependence on C3 photosynthesis. Of the two routes to juvenilization (paedomorphosis), neoteny rather than progenesis ® ts Tillandsioideae. Both pathways require that vegetative ontogeny be decoupled from sexual maturation, but in different ways. Neoteny prevails when the adult of the derived genotype more closely resembles the juvenile than the adult stage of the evolutionary antecedent. Relatively few nodes need to develop before the neotenic, determinate shoot ¯ owers (e.g., Tillandsia usneoides; Fig. 2.1). Consequently, reproduction is precocious in terms of development, if not also in real time. Whether or not life cycling speeds up, less biomass need be invested in vegetative tissue prior to ¯ owering than was possible when shoots and root systems were more elaborate. Precocious sexuality marks the progenic descendant, for example the specimen that ¯ owers from one or more, formerly sterile (pre-reproductive) nodes. Some Type Five Tillandsia deviate from the abbreviated bauplan exempli® ed by Spanish moss and a number of its relatives and initiate ¯ owers at the tips of shoots comprised of dozens of leafy nodes (e.g., T. bryoides; Fig. 2.1). However, miniaturization prevails, and few roots develop as above. Mexican and Central American Tillandsia xerographica demonstrates how still other Type Five Tillandsioideae lack the gross structural characteristics of the neotenic forms, yet exhibit similar tolerances and capacities attributable to an indumentum and ecophysiology comparable to their more diminutive relatives. Despite its relatively large size, broad succulent foliage bearing con¯ uent layers of absorbing trichomes and relatively large, if primarily mechanical, root system, this epiphyte typically inhabits dry scrub forests with more than a dozen markedly heterochronic members of the same genus. Tillandsia duratii further demonstrates that Tillandsioideae can tolerate exceptionally stressful habitats unassisted by the revolutionary morphological reduction (except for the loss of roots) fostered by neoteny (Fig. 2.10L). Heterochrony in¯ uences reproductive capacity depending on how it affects certain aspects of plant organization and related performance, and whether the impact is measured in the affected individual or in its inclusive Cambridge Books Online © Cambridge University Press, 2009 504 History and evolution population. If the neotenic plant matures in less time than its predecessor, then a population of such plants, if otherwise equal (e.g., same seed size, Amax) to that of the ancestor, should expand faster owing to an enhanced Malthusian coefficient (Chapter 6; Benzing 1978a). Conversely, the same change reduces the reproductive power of the individual genet to the degree that precocity reduces the amount of photosynthetic tissue available to ripen seeds. However, the absorbing trichome relaxes this trade-off for Tillandsioideae compared with ¯ ora without a similar root analog, i.e., an equally multifunctional shoot. Theoretically, a population of Type Five compared with Type Four Tillandsioideae could grow faster if all factors in addition to body plan that also in¯ uence fecundity were equal because its members devote proportionally more resources to reproduction at the expense of vegetative tissue. However, Amax is substantially lower in the CAM-equipped Type Five bromeliad, and for this reason neoteny provided ancestors a mechanism to enter an adaptive zone colonized by few other families (Benzing 1978a). In effect, exposure to drought on relatively ephemeral substrates (bark), that respectively constrain photosynthesis and heighten plant mortality, magni® ed the bene® ts of heterochrony (in this case resource-use efficiency for reproduction) as Tillandsioideae radiated from humid to drier parts of the forest canopy. Neoteny and tillandsioid radiation Extraordinary stress-tolerance and an economical bauplan only partly explain why Tillandsia (sensu Smith and Downs 1977) ranges so broadly, and perhaps why it contains so many, often ecologically similar species. Several bursts of speciation occurred in different parts of tropical America, each under somewhat similar circumstances. Common potentials fostered by close relationships assured parallel evolutionary responses (homoplasy) to the same environmental challenges in widely scattered regions. Subgenus Tillandsia represents one of the most proli® c of these radiations. Resultant lineages in this case remain overwhelmingly Mexican with sizable numbers of additional species distributed through adjacent Mesoamerica (Chapter 6). A few populations range beyond northern South America (e.g., T. juncea into southeastern Brazil). Gardner (1982, 1986a,b) provided insights on how pollinators and substrates (bark vs. rock) probably in¯ uenced speciation in special cases. However, relatives distributed primarily below the Equator better illustrate how geography, climate and various features of ancestors, including capacity to undergo Cambridge Books Online © Cambridge University Press, 2009 Neoteny and tillandsioid radiation 505 neoteny, helped Tillandsia, broadly de® ned, achieve its impressive size and exceptional vegetative structure and function. Much of Tillandsia (e.g., subgenera Allardtia, Pseudalcantarea, Pseudocatopsis) shows little evidence of paedomorphosis. Instead, these plants retain the bauplan that incorporates determinant, leafy modules (ramets) produced in seriatum by sympodial branching (Fig. 2.3A). Subgenus Tillandsia exhibits a greater range of architectures, whereas the remaining three, largely South American subgenera, viz. Anoplophytum, Diaphoranthema and Phytarrhiza, illustrate structural simpli® cation and reductions in size unmatched elsewhere in the family. Homoplasy, which is common within groups of paedomorphically miniaturized fauna and ¯ ora (Hanken and Wake 1993), also makes an impressive showing in this part of Tillandsia. Subgenera Phytarrhiza and Diaphoranthema probably constitute a clade positioned close to, or perhaps inclusive of, Allardtia and Anoplophytum. Evidence includes phytogeography, which for both subgenera accords with climate change during the late Pleistocene (Smith 1934a; Till 1992a). If Phytarrhiza and Diaphoranthema represent evolutionary grades rather than clades, then the more advanced of the two architectures emerged repeatedly. Lineages in Phytarrhiza considered closest to certain Diaphoranthema (e.g., T. streptocarpa) feature relatively large, polystichous shoots and sizable in¯ orescences with several to many, often fragrant, chasmogamous ¯ owers (Fig. 3.3I). Those of Diaphoranthema possess fewer, often distichously arranged leaves on more miniaturized axes. One to several, often autogamous ¯ owers per shoot typify the most diminutive Diaphoranthema (Figs. 2.1, 3.3C). Compared with Diaphoranthema, species of Phytarrhiza generally occupy warmer, moister sites, and those with the broadest distributions show considerable polymorphism (e.g., T. streptocarpa). Conversely, members of the ® rst subgenus tend to resemble microspecies by phenotype and range (Till 1992a). Structurally uniform populations mostly occupy con® ned areas in the southern Andes to central Argentina (except T. recurvata and T. usneoides) and often set self-seed. Interestingly, the few, similarly insular Phytarrhiza (e.g., T. reichenbachii, T. peiranoi) of northern Argentina remain well north of many of the Diaphoranthema species (e.g., T. pedicellata, T. erecta, T. aizoides). Till' s claim that the cooler, drier conditions under which many Diaphoranthema grow impede photosynthesis too much to allow the more massive-bodied Type Five members of Phytarrhiza to mature often enough to sustain populations is consistent with the views expressed here on the Cambridge Books Online © Cambridge University Press, 2009 506 History and evolution Table 9.2. Character polarities in Tillandsia subgenera Phytarrhiza and Diaphoranthema Ancestral Derived Roots well developed Roots strongly reduced or absent Leaves in a rosette Leaves on more elongate stem Leaves wide and ¯ at, not succulent Leaves narrow and terete, succulent Leaves polystichous Leaves distichous Scape of in¯ orescence well developed Scape of in¯ orescence abbreviated or lacking In¯ orescence compound, many¯ owered In¯ orescence simple, few to one-¯ owered Floral bracts glabrous outside Floral bracts densely lepidote outside Sepals glabrous, free, obtuse Sepals lepidote, especially posteriorly (5adaxially) connate, acute Petals conspicuous, blades enlarged Petals inconspicuous, blades narrowed Style and ® laments rather long, stigma and anthers reaching the throat of the corolla or exserted Style and ® laments strongly abbreviated, stigma and anthers deeply included in the corolla Flowers without fragrance Flowers fragrant Source: After Till (1992a). advantages of heterochrony in Tillandsioideae. If our hypothesis is correct, neoteny combined with ¯ oral biology that maximizes fecundity provided ancestral forms the opportunity to colonize isolated, arid montane habitats and set the stage for what Till (1992a,b) considered to be at least six episodes of speciation leading to as many monophyletic clusters of four to seven species in Diaphoranthema. Origins for these six lineages supposedly lie in subgenus Phytarrhiza, and the `ancestral' and `derived' characters listed in Table 9.2 describe their parallel evolutionary histories. Tillandsia usneoides demonstrates how neoteny granted exceptional ecological latitude to a single lineage (Tomlinson 1970). Rather than sharing the compact, juvenilized morphology displayed by most other Diaphoranthema, Spanish moss continued to evolve, perhaps under the impetus of two powerful advantages. Plant characteristics that foster these dual bene® ts include much elongated internodes, ageotropism and sufficiently suppressed apical dominance to allow every leaf to subtend a developed branch (Figs. 2.1, 2.10E). Resultant capacity to fragment in turn Cambridge Books Online © Cambridge University Press, 2009 Neoteny and tillandsioid radiation 507 set the stage for dispersal that no other Bromeliaceae would match. Genets that expand to form thick trusses of delicate pendant shoots further promote success to the extent that self-shade and the humid air trapped within ameliorate the harsher conditions that prevail outside (Fig. 7.7C). Information on chromosome numbers, reproductive biology and alpha systematics places modest-sized (,30 spp.) Diaphoranthema among the better-known subgenera of Tillandsia. Nevertheless, major questions remain unanswered, for instance why this and related clades speciated so proli® cally in the southern Andes. As luck would have it, data are available for some primarily Argentinian populations in Anoplophytum (34 species), that third subgenus which remains largely con® ned to southern South America. Palací (1991) determined that 10 populations representing four species in Anoplophytum (T. alberi, T. friesii, T. muhriae, T. zecheri var. cafayatensis) and numerous members of subgenus Diaphoranthema exhibit similar structural, geographic and demographic attributes, especially in the province of Salta, Argentina and adjacent Bolivia. In short, it appears that these Anoplophytum and the codistributed species of the Phytarrhiza/ Diaphoranthema complex shared comparable histories while genus Tillandsia underwent multiple radiations around the margins of its range in the southern Andes and points eastward, primarily on the inselbergs of southeastern Brazil. Relatively large size and polystichous leafy shoots suggest that neoteny contributed little to the evolution of Anoplophytum. Expansion also occurred without deviation from the basic diploid (n525) condition. Floral biology remains little studied, although several species produce conspicuous, fragrant ¯ owers (e.g., T. xiphioides), and bright red to pink ¯ oral bracts indicate ornithophily for many others (e.g., T. aeranthos, T. stricta, T. tenuifolia). Plants (T. alberi, T. friesii, T. muhriae, T. zecheri var. cafayatensis) that Palací (1991) examined occur at high altitudes (mostly .2500 m), often as widely dispersed populations (across .500 km on a north/south axis) in extreme northwest Argentina and just over the border into neighboring Bolivia. All four taxa share similar morphology. Proteins encoded by 14 loci revealed comparatively minor genetic differentiation, perhaps due to mobile seeds or trap-lining pollinators. Genetic diversity observed in T. friesii (P533.8%) exceeded that recorded for epiphytic T. ionantha (subgenus Tillandsia) and T. recurvata (Soltis et al. 1987), but more closely approached values obtained for terrestrial Aechmea magdalenae (Murawski and Hamrick 1990) in Panama (Chapter 6). A Gst reading of 0.228 indicated considerable outcrossing, as Cambridge Books Online © Cambridge University Press, 2009 508 History and evolution did the relatively high proportions of polymorphic loci, average number of alleles per locus, and observed heterozygosity. Isozymes indicated just two copies per locus. Conversely, close genetic identities characterized all of the comparisons involving T. alberi, T. muhriae and T. zecheri var. cafayatensis, values high enough to warrant combining three populations under T. cafayatensis, which in turn Palací considered closely related to, but not conspeci® c with, T. friesii. Occurrence of only about one-fourth of the measured, total genetic variation among, compared with within, its disjunct populations, owing partly to autogamy and insularity, accords with Palací' s decision to recognize forms rather than species. In summary, segregates comprising Palací' s group of closely allied populations of lithophytic Anoplophytum exhibited genetic identity values ranging between 0.96 and 0.99, too close to unity to recognize species. Narrow distributions, including occasional restrictions to a single gorge over similarly extensive north/south ranges, also describe many related, well-de® ned species of Diaphoranthema and Anoplophytum that once probably also possessed little-differentiated gene pools like those recorded for the T. friesii complex. Palací' s data and much additional information suggest that fragmented habitats and founder events encouraged the recent massive radiation of Tillandsia in subtropical South America. Inherent factors were also important. Mobile seeds, frost and drought-tolerance (annual rainfall often below 300 mm), neoteny in some lineages, and capacity to maintain small populations through autogamous reproduction also rank high among the factors that helped Tillandsia, especially the membership of Diaphoranthema, achieve its current status in southern South America. Before leaving the subjects of adaptation, radiation and heterochrony, mention is due Gilmartin' s (1983) and Gilmartin and Brown' s (1986) attempt to infer the phylogenetic juxtapositions of mesophytism and xerophytism in Tillandsioideae, particularly in Tillandsia subgenus Phytarrhiza. Arguably, heightened stress-tolerance and reproductive power fostered by neoteny combined with dissected topography and oscillating climate favored radiation among dry-growing Tillandsioideae in many parts of tropical America. But how often and where in the subfamily did these transitions occur, and was change consistent in direction? Except for large clusters of species within the Tillandsia/Vriesea complex, mesophytism pervades Tillandsioideae. The apparent youth and the mix of mesophytic and xerophytic taxa in these two genera present extraordinary opportunity to determine the evolutionary polarities of climate-sensitive characters. Cambridge Books Online © Cambridge University Press, 2009 509 Al su ca bg nt . ar ea Historic relationships between mesophytism and xerophytism Figure 9.13. Evolutionary tree illustrating the relationships of the seven Tillandsia subgenera and two of Vriesea (sensu Smith and Downs 1977) based on the Wagner unrooted tree method (after Gilmartin 1983). Historic relationships between mesophytism and xerophytism in Tillandsioideae Gilmartin (1983) and Gilmartin and Brown (1986) employed phenetic (overall resemblance) and phylogenetic (cladistics) approaches to infer historic alignments among the mesophytic, semixerophytic and xerophytic habits exempli® ed by members of Tillandsia and Vriesea. They also constructed phenograms and phylogenetic trees for the major segregates of Tillandsia and 36 of the species assigned to subgenus Phytarrhiza. Of the seven subgenera of Tillandsia and the two comprising Vriesea according to Smith and Downs (1977), all but Anoplophytum and Diaphoranthema (both Tillandsia) include xeric and semixeric or mesic members according to leaf succulence, the development of the indumentum and the identity of the primary source of moisture (e.g., atmosphere vs. phytotelmata). Categories of drought-tolerance/sensitivity were limited to just these three degrees by the absence of adequate ecophysiological data. Weather records provided additional resolution. Gilmartin' s (1973) xerophytes experience annual rainfall between 300 and 1000 mm in Ecuador, whereas semixeric types require at least 1000 mm year21. Species labeled mesic occupy even more humid (.2000 mm) zones where no less than 4% of the annual input occurs during the driest month. Fifteen characters considered unrelated to water relations provided the basis for the decisions about phylogeny. Unrooted Wagner networks (Fig. 9.13) placed Diaphoranthema, Phytarrhiza and Pseudocatopsis near one end of the network in agreement with some earlier suppositions about intrageneric relationships (e.g., Smith 1934a). Anoplophytum emerged on the same terminal branch at a point just Cambridge Books Online © Cambridge University Press, 2009 510 History and evolution Figure 9.14. Evolutionary tree for subgenera of Tillandsia and Vriesea based on the character compatibility method (after Gilmartin 1983). below these same three taxa. Subgenera Allardtia and Tillandsia lie at the other end of the network. Every branch except those leading to Anoplophytum and Diaphoranthema respectively includes both mesic and xeric species. The same analysis without Vriesea produced a tree with similar topology. Figure 9.14 illustrates the results of a character compatibility analysis supported by a seven-character clique. Wholly mesic subgenus Allardtia at the base of the tree subtends three main lateral branches: Vriesea (both subgenera), Tillandsia with Pseudalcantarea, and the remaining four subgenera. Overall resemblance based on 15 characters separated the two subgenera of Vriesea, whereas most similar were subgenera Allardtia and Tillandsia and Diaphoranthema and Anoplophytum. Without Vriesea, alignments changed as obliged by the hierarchical nature of the algorithm; the absence of that genus altered resemblance values among the remaining taxa. A 43-character matrix yielded the phenogram reproduced in Fig. 9.15 that purportedly depicts relationships among 36 species of subgenus Phytarrhiza. Six mesic types from Ecuador and northern Peru form Group One, while ® ve of the semixeric taxa constitute Group Two. Nine semixeric species with simple in¯ orescences clustered with T. straminea to yield Cambridge Books Online © Cambridge University Press, 2009 Historic relationships between mesophytism and xerophytism 511 Figure 9.15. Phenogram resulting from cluster analysis (UPGMA of Sokal' s average distance) of species of Tillandsia subgenus Phytarrhiza. Roman numerals designate clusters. Habits are coded: m, mesic; s, semixeric; x, xeric (after Gilmartin 1983). Cambridge Books Online © Cambridge University Press, 2009 512 History and evolution Figure 9.16. Evolutionary tree of the species of Tillandsia subgenus Phytarrhiza based on the Wagner network method. Habits are coded: m, mesic; s, semixeric; x, xeric (after Gilmartin 1983). Group Three. Fourteen xerophytic taxa with relatively southern distributions make up Group Four. At the bottom, more massive-bodied and mesic T. dyeriana joins the larger cluster representing the entire subgenus. The Wagner network aligns the same 36 species somewhat differently (Fig. 9.16). Six semixeric Ecuadorian members of Group Three share a branch with xeric T. cacticola, T. straminea and T. schunkei above the ® ve semixeric Ecuadorian taxa with simple in¯ orescences comprising Group Two. Species of both groups qualify for Type Five status and produce spikes mostly exceeding 2 cm in width except for the three lineages that share the same branch with Group Three. At the center of the network lie ® ve, fully mesic species with mostly compounded in¯ orescences and broad leaves (Group One, and T. dodsonii of Group Three, and T. dyeriana). Group Four resides on the two left-hand branches and shows paedomorphosis, including simple spikes and the expected association with pronounced xeromorphism. Lineages with mesic vs. xeric habits never occupy the same branch anywhere in the network. Gilmartin and Brown' s (1986) second set of analyses concentrated on two closely related Tillandsia subgenera that contain a particularly propitious array of mesophytes and xerophytes to determine the relationships of Cambridge Books Online © Cambridge University Press, 2009 Historic relationships between mesophytism and xerophytism 513 drought-tolerance and vulnerability. Unlike exclusively xerophytic and highly neotenic Diaphoranthema, Phytarrhiza encompasses the entire range of bauplans and related water-balance mechanisms. Habits include those characterized by thin-leafed, sparsely trichomed, phytotelm shoots supported by relatively substantial root systems (e.g., T. dyeriana, T. wagneriana), to the more miniaturized and stress-tolerant types illustrated by T. caerulea and T. paleacea. Semixeric taxa like T. cornuta and T. dodsonii, which impound at most modest amounts of moisture to supplement that stored in semisucculent foliage, bridge these extremes. Gilmartin and Brown (1986) used 11 `alliances' , three represented by individual species and eight more putatively monophyletic clusters of up to eight species, to resolve relationships among mesic, semixeric and xeric habits in Tillandsia subgenus Phytarrhiza. Flowers provided most of the 15 ecophysiologically neutral characters used to reconstruct an unbiased phylogeny. However, several of these choices, including trichome type (expanded or minute shield; Fig. 2.8B,C,E), stem length, presence or absence of roots, ¯ oral bracts with or without trichomes, and leaf shape (thin and ¯ at or terete) may be more predictive of drought performance than the authors assumed. Two sets of equal-length trees incorporating 24 character state changes indicated paraphylesis for Diaphoranthema and Phytarrhiza consistent with Till' s (1992b) conclusion, and sister taxon status for Pseudocatopsis and Phytarrhiza, or some subset of its member species. Tree topology further indicated that changes from character states considered plesiomorphic to apomorphic prior to the analysis outnumbered the reversals. For example, stems usually shifted from short to longer, much as apparently occurred as populations comprising several Tillandsia species adopted rock as the exclusive substrate (Fig. 2.10M,N). Consistent with radiations involving miniaturization elsewhere (Hanken and Wake 1993), trees here also depicted repeated, parallel changes, mostly from mesic to xeric and mesic to semixeric, but never semixeric to xeric habits. Gilmartin (1983) envisioned a montane, moist forest origin for all of the subgenera of Tillandsia except Anoplophytum and Diaphoranthema, which she believed descended from semixeric or more xeric antecedents within, or closely allied to, one of the other segregates, probably Phytarrhiza or Pseudocatopsis. Smith (1934a) proposed that Diaphoranthema arose from stock embedded in Phytarrhiza because both subgenera exhibit the greatest densities of species near the south Andean junction of Argentina, Bolivia and Paraguay. Oscillating arid and pluvial conditions through the Plio-Pleistocene Cambridge Books Online © Cambridge University Press, 2009 514 History and evolution supposedly explain the existence of so many species, both dry-growing and more drought-sensitive types (e.g., Prance 1973, 1982). Speci® cally, during arid phases the stress-tolerant forms migrated into newly expanded dry habitat, while the correspondingly contracted and fragmented populations of their mesic relatives differentiated in moist refugia. Periodic warming reversed this pattern, replacing moist with dry refuges that encouraged divergence among lineages con® ned by ecophysiology to arid landscapes. Relegation of the more drought-tolerant populations to the tips of the branches of phylogenetic trees accords with derived status for xerophytism. Precisely how global change affected radiation within Tillandsioideae and shaped the current distributions of dry and wetter-growing taxa remains obscure, particularly in light of the controversy about the impacts of Pleistocene refugia and Andean orogeny (Gentry 1982) on Neotropical phytodiversity. Aridi® cation de® nitely triggered cladogenesis in some other families at other locations over the past 3± 5 million years. For example, predominantly southwest African Mesembryanthemaceae expanded to ,2000 species mostly in developing xeroscapes (Ihlenfeldt 1994), while Bromeliaceae apparently proliferated on the other side of the Atlantic Ocean. Heterochrony, including miniaturization and reduced numbers of leaves, marked this Old World event as it did the expansion of Tillandsioideae. Chloroplast DNA data described below that challenge Smith and Downs' s Tillandsioideae as a clade also disagree with Gilmartin and Brown' s phylogenetic trees. Nevertheless, the latter two investigators' decision to consider paleoclimate in the attempt to relate the mesophytic, semixerophytic and xerophytic habits in this subfamily recognizes the signi® cance of past selection to interpretations of current plant adaptations and geographic distributions. Reconstructing the circumstances responsible for a major radiation is no less worthwhile than determining the relationships among the lineages spawned by that event. Shifting climate since the Miocene surely affected ¯ ora across tropical America and perhaps certain Bromeliaceae more than most other vegetation. The substantial literature on ecophysiology reviewed in Chapter 4 indicates how affected lineages probably responded. Tillandsia native to extreme western Peru (primarily subgenus Anoplophytum) must rank among the most heavily impacted by shifting Plio-Pleistocene climate of the indigenous ¯ ora. Natives today occupy cold-air, rain-shadow deserts maintained by the interactions of geologic, marine and meteorological in¯ uences. During all but El Niño events, cold, offshore currents virtually eliminate winter precipitation that would other- Cambridge Books Online © Cambridge University Press, 2009 Historic relationships between mesophytism and xerophytism 515 wise arrive on-shore from the Paci® c Ocean. Much of the rest of the year, the eastern slopes of the Andes, which began to rise above relatively modest heights only a few million years ago, desiccate ascending air moving along the tropical interconvergence zone during its southerly migration each summer. Few vascular plants beyond some dew-dependent Tillandsia occupy the Atacama west of the deeply eroded coastal range. Substantial canyons (,1000 m) that dissect these mountains date from the Miocene when much more rainfall arrived across a then less obstructed eastern approach and over a warmer ocean from the opposite direction during the winter. Drier conditions developed during the Pliocene through the Pleistocene, but not continuously or synchronously with those hypothesized changes in the Amazon Basin (Damuth and Fairbridge 1970) which dominate the secondary literature purporting to describe the effects of past climate on plant evolution and biodiversity. Information on paleoclimate must be applied more precisely than it has been so far to reconstruct the adaptive histories of surviving Bromeliaceae. For example, Pleistocene glaciations supposedly brought cooler, drier times and much reduced forests east of the Andes, but quite different conditions probably prevailed west of that barrier. Climates reversed during the much shorter interglacials. Damuth and Fairbridge (1970) and Arroyo et al. (1988), among others, proposed that a low-pressure focus over Antarctica moved northward to the west of South America during the glacials to enhance humidity in the Atacama region. A weaker Humboldt current that no longer penetrated as far north further assured moister air masses over that region during winter. Plant responses to modern ecophysiological challenges (e.g., see Table 4.8) help identify some of the past events that shaped current geographic ranges. Torrential rainfall during El Niño events periodically kill off much of the spectacular near monocultures of soil-bound Tillandsia that occupy extensive areas of treeless Atacama desert. Presumably, more prolonged episodes of comparable humidity during earlier times similarly impacted populations equipped with the same dense, hydrophilic indumenta (Figs. 2.8C,E, 4.11). Paleoclimate has alternately favored different ecophysiological types at given locations according to their vulnerability to drought or excess moisture. Populations probably expanded and contracted, and perhaps fragmented and speciated, according to prevailing humidity and the availability of speci® c substrates. Pluvial phases must have devastated Type Five Bromeliaceae except where topography lessened rainfall, or bark and rock maintained an adequate supply of arid microsites (Tables 4.8, Cambridge Books Online © Cambridge University Press, 2009 516 History and evolution 4.9). Conversely, dependence on phytotelmata and poorly insulated foliage imperiled Type Four Tillandsioideae during the dry phases, and indeed no members of this group currently inhabit the Atacama region. Taxonomy: traditional characters Successive monographers emphasized different characters to systematize Bromeliaceae into genera and higher taxa. One to a few attributes of sometimes doubtful utility, like the presence or absence of petal scales, typically distinguish putative clades (e.g., Tillandsia from Vriesea, Cryptanthus from Orthophytum and Navia from Brewcaria). No consensus developed on which parts of the plant body most reliably indicate evolutionary relationships. For example, the in¯ orescence ® gures prominently in Baker' s (1889) treatment of Bromeliaceae, while Mez (1934± 35) favored ¯ oral characters and pollen morphology. Smith and Downs (1974, 1977, 1979) assigned substantial weight to the petal scales that Brown and Terry' s (1992) developmental studies suggest represent recently derived, unstable features, at least among Tillandsioideae (Fig. 3.1B). Numbers highlight the problem: Baker (1889) recognized 19 genera, while Harms (1930) described 34. Smith and Downs (1974, 1977, 1979) accommodated a substantially larger number of species in just 27 genera. Today the number exceeds 55 (Luther and Sieff 1996) and promises to keep growing. Several taxa remain especially problematic. For example, Mez (1896) initially treated Glomeropitcairnia as a subgenus of Pitcairnia, and later (1935) granted its two species tribal status (Glomeropitcairniaeae) in Tillandsioideae. Several features both distinguish and obscure the affinities of this likely relictual clade, most conspicuously its semi-inferior ovary, seed with apical and basal appendages, the many-celled trichome stalk, and unusual mode of capsule dehiscence. Gilmartin et al. (1989) attributed the co-occurrence of an apomorphic ovary position and plesiomorphic seed and trichome morphology to mosaic evolution. Seed structure seems to place Glomeropitcairnia somewhere between Pitcairnioideae and Tillandsioideae, and indeed, Smith and Downs (1974) considered its reproductive morphology derived from something more like conditions in certain Pitcairnioideae. Nucleic acid sequences also suggest outlier status, but closer to core Tillandsioideae than Pitcairnioideae as we will see. Bromelioideae contains the largest number of questionable genera owing to its greater ¯ oral variety and the frequent reliance of authorities on poorly studied taxonomic characters. De® nitions changed with each monographer. For example, Quesnelia according to Baker (1889) included Cambridge Books Online © Cambridge University Press, 2009 Chemical systematics 517 species that Smith and Downs (1979) segregated into Aechmea, Andrea, Ronnbergia and even Pitcairnia! Harms (1930) and Mez (1934± 35) maintained discordant views on several additional taxa, especially Bromelia and Streptocalyx. Smith and Kress (1989) elevated or resurrected all of the Aechmea subgenera to full-blown genera. Comparisons of the systems of Baker, Mez, and Smith and Downs indicate how markedly disagreements about a few characters have affected bromeliad taxonomy. Moreover, authorities continue to employ petal scales, sepal and in¯ orescence morphology to reshuffle species with no end in sight. Authors will serve the interests of science and bromeliad enthusiasts by defferring further revisions until judgments can bene® t from more data obtained with the most powerful analyses currently available. Two decisions (e.g., Smith and Kress 1989; Smith and Spencer 1992) exemplify the ongoing churning of an already cluttered nomenclature. Both initiatives employed the same traditional characters to revisit long-standing problems. More importantly, neither proposal was informed by the emerging molecular and better-resolved traditional data. The restoration of Aechmea subgenus Chevaliera to generic status and the re-establishment of monotypic Deinacanthon from Bromelia shed no new light on bromeliad phylogeny. Problematic taxa surely need more attention, but formal revision should wait until a fuller understanding of bromeliad history justi® es such action. Chemical systematics Secondary metabolites Systematists working with chemicals rather than morphology operate closer to the genome, and accordingly, should be able to avoid some of the ambiguities inherent to higher-level biological complexity. However, classes of compounds differ in their proximity to the basis of relationship, which is genetic information encrypted in the structure of DNA. Time and cost also vary. Relative ease of isolation and identi® cation assured that the ¯ avonoids would become useful before the more conservative, complex and character-rich proteins and nucleic acids. While not immediate gene products nor immune to homoplasy, the ¯ avonoids and other convenient `secondary' metabolites can provide estimates of relationships among closely allied taxa, characterize patterns of gene ¯ ow, and help identify the parents of hybrids. Ashtakala (1975) reported substantial similarity between the ¯ avonoids, anthocyanins and phenolic acid derivatives extracted from the foliage of one member each of Aechmea Cambridge Books Online © Cambridge University Press, 2009 518 History and evolution and Billbergia. Subjects were distinguished primarily by the presence of 7-apiosylglucoside in Aechmea glomerata and 6-hydroxyapiginin glycoside in Billbergia vittata. Subsequent investigations (e.g., Scogin and Freeman 1984) revealed a variety of additional constituents, some for the ® rst time, in samples representing diverse genera, including Bromelia, Puya and Tillandsia. Scogin (1985) also identi® ed some novel anthocyanins in the corollas of Puya. None of these investigators devoted much attention to biological functions or implications for plant systematics. Williams' s (1978) survey exceeds all the others devoted so far to the secondary metabolites of Bromeliaceae. Her inventory, compiled from 61 species representing all three subfamilies, lists a remarkably wide collection of ¯ avonoid types including ¯ avones, ¯ avonols, C-glycosylated ¯ avones and 6-oxygenated ¯ avones and ¯ avonols. Such immense chemical diversity and the extra hydroxylation or methylation at C-6 of the ¯ avones and ¯ avonols suggests uniqueness and exceptional phylogenetic isolation for Bromeliaceae among Liliopsida. Flavonols (present in 43% of the species) and ¯ avones (just 13% of the total) occurred widely through the samples, whereas members of only one or two subfamilies produced constituents representing the less pervasive classes of related products. For example, 6-hydroxy¯ avones characterized Pitcairnioideae and Tillandsioideae, but only certain members of the second taxon yielded patuletin, gossypetin and methylated 6-hydroxymyricetin derivatives. Several genera demonstrated novel ¯ avonoid pro® les (e.g., Pitcairnia, Tillandsia). A new ¯ avonol, 6,39,59-trimethoxy -3,5,7-49-tetrahydroxy¯ avone, occurred as the 3-glucoside in Tillandsia usneoides, and Alcantarea regina contained the exceptional glycoside patuletin 3-rhamnoside. Williams' s ® ndings run counter to traditional views about evolutionary grades in Bromeliaceae in the sense that the molecular complexity noted increases beginning with Bromelioideae through Pitcairnioideae to Tillandsioideae. Macromolecules Nucleic acids constitute the richest source of chemical characters for evolutionary analysis. Phylogenies based on DNA structure indicate where, and ± depending on the reliability of the referenced `molecular clock' ± when, speciations occurred within clades. Overlain with data on structure, function and ecology, family trees can reveal the same kinds of information about speci® c plant adaptations (e.g., Brocchinia as treated above; Fig. 9.5). Moreover, DNA ampli® cation using polymerase chain reaction (PCR) Cambridge Books Online © Cambridge University Press, 2009 Chemical systematics 519 technology grants access to nucleotide sequences without jeopardizing rare biota. It also obviates the need to clone DNA fragments of interest, a laborious process that previously made molecular systematics much more expensive. In effect, tools originally developed to probe the workings of cells are no less crucial to the reconstruction of a comprehensive Bromeliaceae genealogy. Several features of the DNA molecule account for its extraordinary utility to address questions about systematics and evolution. First, nucleotide sequences provide an essentially inexhaustible source of relatively unambiguous characters to reconstruct phylogenies de novo and test congruence with hypotheses implicit in traditional taxonomies. Genealogies inferred from DNA structure in turn provide frameworks to identify change in individual and suites of plant characteristics through geologic time, and determine the sequence of important evolutionary events like the emergences of absorbing trichomes and phytotelmata. Molecular data typically take one of three forms: maps of restriction sites, extensive nucleotide sequences, and substitutions (insertions and deletions) and inversions of segments of DNA. Commercially available endonucleases hydrolyze nucleic acid molecules at speci® c sites (nucleotide sequences), after which gel electrophoresis resolves the resulting fragments according to size and net charge. The ultimate product, a map locating the restriction sites along intact molecules, for example the circular chromosome of the chloroplast (cpDNA), provides a basis to compare the relatedness of genotypes (Fig. 9.17). More extensive nucleotide sequencing yields correspondingly larger numbers of characters. The capacity of speci® c molecular data to differentiate genotypes with given degrees of identity depends on the genome (chloroplast, nuclear or mitochondrial) under study and the speci® c locus targeted. The highly conserved chloroplast gene rbcL (ribulose bisphosphate carboxylase large subunit), for example, yields insights on relationships among families within orders of ¯ owering plants. Comparisons involving more closely related organisms, those distinguished by less molecular divergence overall, require a more variable locus such as parts of the ndhF plastid gene described below. Most of the molecular data utilized so far by plant systematists, and without exception employed to compare bromeliads, come from cpDNA, which has the following general characteristics (Fig. 9.17). A typical chloroplast contains 20± 200 copies of a covalently closed, double-stranded chromosome that, unlike nuclear DNA, lacks associated structural protein Cambridge Books Online © Cambridge University Press, 2009 520 History and evolution Figure 9.17. Structural aspects of the chloroplast genome from Nicotiana tabacum (tobacco). Inverted repeats are indicated by bold regions along the circle. Large and small single-copy regions are delimited by the inverted repeats. Genes listed on the outside and inside of the circle are encoded on the A strand and B strand respectively. Split genes are indicated by asterisks (modi® ed from Grierson and Covey 1988). (histones), and also differs in the ratios of nucleotides present and the absence of 5-methyl cytosine. Plastid genomes present in most angiosperms resemble each other in size, conformation, structure and gene content. Mostly, they range from 120 to 160 kilobases (kb) and consist of three parts: (1) the asymmetrically positioned, duplicate sequences of approximately 25 kb each, known as the inverted repeats, (2) a unique, single-copy sequence usually varying from 20 to 30 kb in size, and (3) another, larger single-copy region of 75± 100 kb. Cambridge Books Online © Cambridge University Press, 2009 Relationships among subfamilies and Bromeliaceae within Liliopsida 521 Figure 9.18. Consensus tree of nine equally parsimonious trees resulting from the restriction site study of Ranker et al. (1990). Length522 steps; consistency index5 0.86. Numbers along branches refer to characters (mutations) supporting that clade or taxon. Asterisks indicate homoplasious steps (modi® ed from Ranker et al. 1990). Chloroplast genomes of higher plants typically contain about 120 densely aligned genes whose products fall into two categories, viz. (1) those that function in chloroplast protein synthesis, including ribosomal and transfer RNAs, ribosomal proteins, elongation and initiation factors, and RNA polymerase subunits, and (2) those loci involved with photosynthesis, e.g., rbcL, thylakoid polypeptide components of photosystems I and II, proteins of the cytochrome b/f complex, and components of the ATP synthase complex. Chloroplast genes exhibit some prokaryotic features (e.g., they possess `bacterial' transcription initiation sequences, i.e., transcription and translation are concurrent), consistent with the endosymbiotic theory of eukaryotic cell origin. Relationships among subfamilies and Bromeliaceae within Liliopsida Ranker et al. (1990) published the ® rst DNA-based phylogeny for Bromeliaceae using 16 endonucleases and 10 species ± ® ve Tillandsioideae, three Pitcairnioideae and two Bromelioideae ± to compare cpDNA restriction site polymorphism (Fig. 9.18). The resulting maps revealed colinearity between the bromeliad chloroplast genome and those for the majority of Cambridge Books Online © Cambridge University Press, 2009 522 History and evolution other land ¯ ora. Genome sizes ranged from 150.6 to 152.9 kb, of which the inverted repeat accounted for approximately 21.9 kb. Eleven restriction site mutations and one 1.8 kb length mutation distinguished the bromeliad subfamilies and sometimes members within a subfamily. Most mutations mapped to that part of the large single-copy region adjacent to one of the inverted repeats where much variability characterizes the chloroplast genome. Parsimony analysis produced a consensus tree comprised of 22 steps with a consistency index (CI) of 0.86 (Fig. 9.18). That tree prompted three assertions. First, Tillandsioideae, excluding Glomeropitcairnia, occupies the basal position in the family. Second, Bromelioideae and Pitcairnioideae are sister groups and the latter may be paraphyletic. Third, Glomeropitcairnia lies beyond the balance of Tillandsioideae and possibly warrants higher taxonomic status as either a separate tribe or subfamily. Small sample size ± just 10 bromeliads and only 12 phylogenetically informative characters ± and the absence of several key genera (e.g., Brocchinia, Catopsis) in the analysis precluded broader conclusions. Clark et al. (1993) used the rbcL gene to assess familial, ordinal and superordinal relationships among Bromeliaceae and additional monocots that taxonomists often associate in phylogenies. Sequences obtained from members of 20 families, seven species in the case of Bromeliaceae, including all three of its subfamilies, were considered (Fig. 9.19). Three conclusions emerged. First, Bromelii¯ orae sensu Dahlgren et al. (1985), which encompasses Bromeliales, Velloziales, Philydrales, Haemodorales, Pontederiales and Typhales, is paraphyletic. Conversely, Bromelii¯ orae, Zingiberi¯ orae and Commelini¯ orae (sensu Dahlgren et al. 1985) comprise a clade. Third, Bromeliaceae share a closer relationship with Rapateaceae than with Velloziaceae. Smith (1934a) turned to phytogeography to support his case for sistergroup relationship between Bromeliaceae and Rapateaceae, while the results of other studies using morphological characters (e.g., Huber 1977; Gilmartin and Brown 1987) favor Velloziaceae in that position. Clark and Clegg (1990) also used rbcL sequences and two methods of data analysis to try to resolve subfamily relationships in Bromeliaceae. Maximum likelihood analysis supported Bromelioideae and Tillandsioideae respectively as basal in the family, while parsimony analysis resolved Pitcairnioideae in this position. Terry et al. (1997a) investigated phylogeny among the major taxa of Bromeliaceae using the chloroplast gene ndhF to test hypotheses about relationships and the evolution of several characters, including some that Cambridge Books Online © Cambridge University Press, 2009 Relationships among subfamilies and Bromeliaceae within Liliopsida 523 Figure 9.19. Family-level relationships within the Bromelii¯ orae/Commelini¯ orae/ Zingiberi¯ orae complex according to unrooted parsimony analysis of a 52-taxon data set of monocots. The strict consensus of 198 trees generated by Fitch analysis; length51665 (modi® ed from Clark et al. 1993). played pivotal roles in family radiation. The gene ndhF consists of approximately 2200 base pairs and encodes a chlororespiratory peptide. Its rate of change approaches twice that of relatively conservative rbcL. Variable positions among the tested bromeliads occur most abundantly in the 39 half of the gene, with 55% of all mutations located in the 39-most 41% of the sequence region (Fig. 9.20). Of 308 unstable positions, 71 (or 23%) proved informative. Phylogenetic analysis of ndhF sequences produced 120 Cambridge Books Online © Cambridge University Press, 2009 524 History and evolution Figure 9.20. Map of the chloroplast gene ndhF from Vriesea espinosae with the relative position of priming sites indicated. Coding strand (forward) and complementary (reverse) primers are given above and below the line respectively. Numbers indicate the 59-most position of the primer relative to the start (position 1 on the coding strand) in tobacco. Primer sequences are shown. most-parsimonious trees of 406 steps (CI50.58). Figure 9.21 presents the resulting majority-rule consensus tree. Intergeneric divergence values based on rbcL sequences are lower in Bromeliaceae than for most other families of ¯ owering plants, ranging from 2.5% for Billbergia macrolepis vs. Vriesea malzinei and Orthophytum gurkenii vs. Tillandsia complanata to 0.1% for Neoregelia pineliana vs. Nidularium selloanum. Relatively high genetic identities combined with the extraordinarily broad interfertilities among some family members accord with recent, rapid evolution and fossils no older than the mid-Tertiary. The same plastid gene also provides insight on a broader relationship, viz. a maximum divergence value of 8.1 between Araeococcus pectinatus and Stegolepis hitchcockii (Rapateaceae). Still, the identity of the sister group should be considered unresolved, and the position of Bromeliaceae within Liliopsida uncertain (Simpson 1988; Duvall et al. 1993; Davis 1995). Restriction site data for diverse monocot families indicate that a clade containing Typhaceae exceeds Rapateaceae as the contender for sistergroup status, although Catopsis nutans provided the single reading for Bromeliaceae (Davis 1995). Timing is equally obscure at this juncture. Synonymous substitutions that distinguish Bromeliaceae and Rapateaceae and published mutation rates for ndhF (Wolf 1991) suggest shared ancestry only about 41 million years ago. A common Eocene stock does match the primarily Neotropical distributions of both families and those other signs of relative youth for Bromeliaceae. However, substitution rates in ndhF vary among families, rendering this gene unreliable to calibrate evolution. Cambridge Books Online © Cambridge University Press, 2009 Relationships among subfamilies and Bromeliaceae within Liliopsida 525 Figure 9.21. The majority-rule consensus tree for the ndhF subfamily analysis. The number of supporting character state transformations is given below the branch. Trees are rooted with Stegolepis (Rapateaceae). Relationships among the bromeliad subfamilies suggested by the ndhF gene largely agree with those based on Ranker et al.' s (1990) restriction site data. Both studies reconcile with close affinity between Bromelioideae and Pitcairnioideae and place Tillandsioideae nearer the base of the family. Chloroplast DNA sequences further support Pitcairnioideae (sensu Smith Cambridge Books Online © Cambridge University Press, 2009 526 History and evolution and Downs 1974) as a paraphyletic grade within which a monophyletic Bromelioideae arose. Sampling differences probably account for the major disagreements between these two studies, namely the dissimilar placements of Brocchinia (omitted from the Ranker et al. study) and Glomeropitcairnia, in the resulting trees. The ndhF-derived subfamily phylogeny resolves Brocchinia at the base of Bromeliaceae, well removed from the remainder of Pitcairnioideae, a locus consistent with the exceptional morphology and ecophysiology described above, and its recognition by Varadarajan and Gilmartin (1988a) as a monogeneric tribe within Pitcairnioideae. Brocchinia' s relegation to a remote, somewhat parallel position relative to the main bromeliad radiation (Fig. 9.21) raises crucial questions about family history considering the varied form and diverse ecophysiology expressed among members of this remarkable genus. Recall that here too, combinations of phytotelm and nonimpounding shoots and absorbing trichomes underlie capacity to use a variety of substrates for anchorage and nutrition. Occasional epiphytism (B. tatei, B. hitchcockii) heightens importance for historical perspectives even more given the same habit in occasional Rapateaceae (Epiphyton) and the much higher incidence of this life style through Bromelioideae and especially Tillandsioideae, with which Brocchinia shares several ecologically decisive characters (e.g., C3 photosynthesis, phytotelma, absorptive trichomes with radial construction, Fig. 2.5B,H,I). Perhaps the common ancestor for Bromeliaceae and Rapateaceae, or whatever other lineage turns out to be the sister group, shared a tendency to anchor in tree crowns or was already strongly epiphytic and more comparable to tank-equipped Brocchinia and Tillandsioideae than members of the other two subfamilies. Inspection of Rapateaceae for absorbing trichomes and other features (several members also produce phytotelma) that underlie arborealism in Bromeliaceae might prove rewarding. However, some additional conditions will be needed to fully reconstruct the adaptive history of Bromeliaceae. Characters grounded in the structure of DNA should eventually provide the framework needed to polarize those aspects of the bromeliads (e.g., phytotelm shoots, CAM) important to adaptive radiation. However, discordance within and between these two kinds of data may oblige assessments of molecular homoplasy and dissections of phenotypes into components with discrete genetic foundations. In the ® rst instance, uneven rates of change among the referenced DNA sequences and differences between these and the tempo at which the characters of interest evolved may complicate attempts to assign dates to ecologically decisive events. No Cambridge Books Online © Cambridge University Press, 2009 Relationships among subfamilies and Bromeliaceae within Liliopsida 527 less troublesome, the daunting complexity of certain key characters relative to others (e.g., shoot architecture vs. seeds appendaged or not) militates against determinations of how synchronously these features and the plant performances they in¯ uence changed over time. Neotenic Tillandsia illustrates this second problem. Within subgenus Diaphoranthema, the most miniaturized and shootdependent populations may be leafy and polystichous (e.g., T. bryoides), fewer leafed and distichous (e.g., T. usneoides), or leafy and distichous (e.g., T. capillaris; Fig. 2.1). Heterochrony has not progressed uniformly or affected certain aspects of development evenly despite the close phylogenetic relationships. All three bauplans incorporate combinations of two states of each of the same two characters (phyllotaxis and shoot length or, better yet, number of leafy nodes). Speci® cally, organogenesis was altered in different ways depending on the lineage and perhaps its environment, while descendants in every case have become quite small. Our point is this: the power of a DNA-based phylogeny to impute when speci® c states of speci® c characters arose and track the assembly and fates of character suites over geologic time is real and will improve with the number of nucleotide sequences used to construct the necessary genealogy (Givnish et al. 1997). However, phenotype must also be understood in sufficient detail to determine how and when associated performances (e.g., degrees of drought-resistance, capacity for soil-free existence) emerged during bromeliad history. In the case of Tillandsia subgenus Diaphoranthema, does each of the diverse architectures illustrated by one or more of its extant lineages represent a response to a distinct kind of selection or equally acceptable accommodations to life under similar conditions (Fig. 2.1)? Conceivably, discrete components of a common, juvenilized development program were decoupled and reassociated to produce speci® c architectures that enhance ® tness under different sets of constraints on plant success (e.g., different kinds of substrates). Tillandsioideae Figure 9.22 presents the strict consensus phylogeny based on 48 mostparsimonious trees of 367 steps that Terry et al. (1997b) obtained using ndhF sequences. Consistent with the broader survey, it supports a monophyletic Tillandsioideae comprised of ® ve primary lineages, viz. Catopsis, Glomeropitcairnia, a clade containing representatives of Vriesea subgenus Vriesea, section Xiphion, Vriesea splendens (subgenus Vriesea section Vriesea) and members of section Xiphion and others assigned to Vriesea Cambridge Books Online © Cambridge University Press, 2009 528 History and evolution Figure 9.22. The strict consensus tree resulting from the ndhF study of phylogenetic relationships in Tillandsioideae. Numbers above the branches are bootstrap values, while those below (or in parentheses) are decay values. Trees are rooted with Stegolepis (Rapateaceae). Subfamilial, subgeneric and sectional designations accord with Smith and Downs (1974, 1977, 1979). subgenus Vriesea plus the sampled Guzmania and Tillandsia species. Few of the generic relationships proposed by Gilmartin et al. (1989) receive support except for the sister-group relationship between some elements of Tillandsia and Vriesea. By presuming monophyly for Smith and Downs' s (1977) genera, these authors precluded assessments of paraphyly in Tillandsioideae assuring that their ® ndings would differ from those of Terry et al. ndhF structure places Catopsis and Glomeropitcairnia within Tillandsioideae, but suggests early divergences, perhaps before those line- Cambridge Books Online © Cambridge University Press, 2009 Relationships among subfamilies and Bromeliaceae within Liliopsida 529 ages (Guzmania, Mezobromelia, Tillandsia and Vriesea) forming the subfamily core differentiated. Substantial isolation for Catopsis seems even more likely in light of seed structure that differs from that of the other Tillandsioideae (Palací 1997). Of the six genera recognized by Smith and Downs (1977), Tillandsia and Vriesea emerge as paraphyletic, Catopsis is monophyletic according to ndhf, while the status of Glomeropitcairnia and Mezobromelia remain uncertain pending a larger sample. The strict consensus tree is uninformative regarding the monophyly of Guzmania, while the majority-rule tree suggests that some of its members lie closer to elements of Tillandsia than to congeners (Fig. 9.23). The ndhF gene also corroborates sister-group status for Tillandsia subgenera Pseudocatopsis and Phytarrhiza, although its mapped sequences do not support monophyly for the latter taxon as suggested by Gilmartin and Brown (1986). Sampling was inadequate to address the same issue for either Phytarrhiza or Diaphoranthema. However, ndhF data suggest that some component of Anoplophytum is the sister group to Diaphoranthema (Terry et al. 1997b). A shared, xeric habit parallels the close relationship between Tillandsia subgenera Anoplophytum and Diaphoranthema. Gilmartin (1983) proposed some relatively xeric element positioned within the groups represented by subgenera Phytarrhiza or Pseudocatopsis as ancestral to Anoplophytum and Diaphoranthema. Most treatments of Tillandsia imply close relationship between subgenera Allardtia and Anoplophytum. Smith and Downs (1977) distinguished these two taxa using relative stamen length and the presence of plicate ® laments (Fig. 6.1C). Evans and Brown (1989a) examined plication in species representing subgenera Anoplophytum, Allardtia and Tillandsia and rejected its utility to circumscribe Allardtia. Distribution of this condition also persuaded them to question the legitimacy of Allardtia. Subgenus Allardtia probably also includes components of subgenera Anoplophytum and Tillandsia and some species previously assigned to Vriesea. Parsimony analysis of the ndhF sequences resolves a group of nested relationships between Tillandsia complanata, T. geminiflora and T. bergeri that denote close phylogenetic affinities for their subgenera (Allardtia and Anoplophytum). Moreover, subgenera Allardtia and Anoplophytum emerge as paraphyletic (Fig. 9.22). Too little information exists at this time to determine whether any of the lineages comprising these two taxa share a most recent common ancestor, and if so, what characters might circumscribe such a clade. Tillandsia bergeri (subgenus Anoplophytum) exhibits variable ® lament plication and shares a most recent common ancestor with T. tricholepis (subgenus Diaphoranthema) in the ndhF-based phylogeny. Cambridge Books Online © Cambridge University Press, 2009 530 History and evolution Figure 9.23. The majority-rule consensus tree from the ndhF study of relationships in Tillandsioideae showing other compatible groupings. Generic and subgeneric grouping are as in Fig. 9.22. Plication characterizes the stamen ® laments of T. geminiflora (Smith and Downs 1977), but not T. secunda. ndhF sequences also resolve a well-supported monophyletic group at the base of Tillandsioideae, exclusive of Catopsis and Glomeropitcairnia, consisting solely of members of Vriesea section Xiphion (Smith and Downs 1977; Fig. 9.22). Most taxonomies recognize the distinctness of the xiphion vrieseas owing to their usually dull-colored, nocturnal ¯ owers with included sexual appendages (Fig. 3.5E). Utley (1983) proposed the theco- Cambridge Books Online © Cambridge University Press, 2009 Relationships among subfamilies and Bromeliaceae within Liliopsida 531 phylloid vrieseas as a subgroup notable for secund in¯ orescences bearing enlarged primary, mostly green bracts subtending ¯ owers with asymmetrically positioned stamens like those of Vriesea atra (Fig. 3.5E). Of those species traditionally assigned to Vriesea section Xiphion included in the ndhF analysis, only V. malzinei conclusively emerged elsewhere in the phylogeny, in this case as the sister group of Tillandsia funckiana. Signi® cantly, of the 65 Vriesea species examined, only two, including Vriesea malzinei, possess the simple-erect type stigma (Brown and Gilmartin 1989a; Fig. 3.1C). No member of Vriesea section Xiphion inspected so far exhibits this morphology. Other notable associations depicted in the ndhF trees include close affinity between Tillandsia and Guzmania (Figs. 9.22, 9.23). Should this assignment be correct, the frequently disparaged petal appendage (Fig. 3.1B; Chapter 3) would regain some lost currency as one of the more useful among the traditional characters to separate Guzmania, Mezobromelia, Tillandsia and Vriesea. According to the ndhF-derived phylogeny, this sometimes labile ¯ oral structure (Brown and Terry 1992) distinguishes Guzmania and Tillandsia from the other genera with only occasional exceptions (e.g., Vriesea espinosae, V. malzinei). Additionally, conglutinate petal claws and polystichous in¯ orescences would become important to differentiate clades (e.g., Guzmania from Tillandsia and possibly Mezobromelia from some components of Vriesea). Recognition of additional genera seems assured, but revisions should not be published before supplementing the ndhF ® ndings with sequences from other parts of the genome and additional information on certain aspects of reproductive structure. None of the ® ve subgeneric or sectional circumscriptions examined (i.e., those for which more than one taxon was examined) quali® es as a clade, a view variously expressed by other authorities who analyzed the group using more traditional characters (e.g., Gilmartin 1983; Utley 1983; Gardner 1986b; Gilmartin and Brown 1986; Evans and Brown 1989a). Figure 9.24 illustrates the ® rst attempt to use molecular data to determine growing conditions in ancestral habitats relative to the distributions of related, adaptive features in extant lineages. All of the taxa compared for ndhF sequences were scored as mesic, semixeric, or xeric types and mapped over the molecular phylogeny depicted in Fig. 9.23. Again, plant habits, this time often corroborated by ® eld notes on local climate, provided the basis for assigning ecostatus. Note that mesic lineages comprise most of the clades near the base of the subfamily, while some Tillandsia and Vriesea species characterized by semixeric or xeric pro® les occur distally. Cambridge Books Online © Cambridge University Press, 2009 532 History and evolution Figure 9.24. The mapping of water-balance status over the ndhF majority-rule consensus tree for Tillandsioideae. All unresolved nodes are treated as hard polytomies in determining tree length. The equivocal scorings near the base of the family and extending to the base of Tillandsioideae re¯ ect uncertainty with respect to mesic or semixeric scorings. The equivocal scoring of the node supporting some Guzmania, Tillandsia and Vriesea re¯ ects uncertainty with respect to mesic, semixeric, or xeric status. Distributions of lineages representing both habits through the tree demonstrate homoplasy, although change consistently proceeded from less to greater drought-tolerance much as Benzing and Renfrow (1971b,c), Gilmartin (1983) and Gilmartin and Brown (1986), and Schimper before them, proposed for Tillandsioideae. A second overlay addresses the origin of epiphytism as indicated by the distributions of the necessary habits, namely the putative phylogenetic ordering of Pittendrigh' s four ecological/structural types. Figure 9.25 illustrates how the bromeliads with these characteristics distribute across the ndhF-based phylogeny. Phytotelm forms (Type Three here) appear in every subfamily, almost certainly independently, and provided the self-sufficiency required to anchor on more impoverished media like bark and rock. The Cambridge Books Online © Cambridge University Press, 2009 Relationships among subfamilies and Bromeliaceae within Liliopsida 533 Figure 9.25. The mapping of Pittendrigh' s (1948) four ecological types over the majority-rule consensus tree (Fig. 9.23) for the ndhF gene. nonimpounding, essentially rootless and trichome-dependent habit exists only in Tillandsioideae where it probably evolved repeatedly, possibly via heterochrony as just described. However, a broader, more fundamental propensity for epiphytism pervades the family. Some Pitcairnia illustrate this condition by rooting in the canopy unassisted by either absorbing trichomes or foliar impoundments. Unfortunately, molecular phylogenies lack the capacity to track epiphytism when this condition occurs embedded in predominantly terrestrial lineages and unassociated with the features that permit the more specialized forms to grow in the more arid parts of the canopy habitat. Figure 9.26 illustrates three plausible relationships between lineages equipped with Types One, Two and Three habits, all of which, along with Type Four, must be considered evolutionary grades rather than successive stages in a single historical progression. Cambridge Books Online © Cambridge University Press, 2009 534 History and evolution Figure 9.26. Three different hypotheses addressing the evolution of Type III Bromelioideae from Type II ancestors using Pittendrigh' s (1948) four types. (A) Type II Ananas are more closely related to Type III Bromelioideae than to other congeners. (B) Ananas is ancestrally Type II. The cross-hatched branch designates the evolution of the Type II ecological type from a Type I progenitor. The Type II ancestor subsequently gives rise to epiphytic Bromelioideae in one direction and to Type I and Type II Ananas in the other. This hypothesis predicts that the distribution of ecological types in Ananas will exist along phylogenetic lines (i.e., the Type I condition will be synapomorphic in Ananas). (C) The ancestor of Type III Bromelioideae and Ananas is polymorphic (a population consisting of Type I and Type II ecological types). The cross-hatched branch designates the origination of the polymorphic condition from a monomorphic (Type I) progenitor. This ancestor gives rise to polymorphic populations in both directions. In one descendant population, Type II give rise to Type III descendants. Type I and Type II ecological types persist in the lineage that ultimately gives rise to Ananas. This hypothesis predicts that the distribution of ecological types in Ananas will not necessarily exist along phylogenetic lines (i.e., polymorphism is symplesiomorphic in Ananas). Triangles designate cladogenesis within lineages. Cambridge Books Online © Cambridge University Press, 2009 Relationships among subfamilies and Bromeliaceae within Liliopsida 535 Figure 9.27. The mapping of ovary position over the majority-rule consensus tree for the ndhF gene. The ndhF phylogeny also imputes homoplasy for a variety of morphological characters often used as taxonomic landmarks, such as ovary position and seed morphology. Events affecting different organs sometimes occurred in tandem. For example, several lineages characterized by shifting ovary position also adopted different fruit types, or seed morphology (appendaged vs. unappendaged), or both. The derivation of Bromelioideae from within Pitcairnioideae was accompanied by changes in each of these characters, all apparently occurring along a single branch, suggesting signi® cant character linkage. Assuming the same probability of change for each of the plant features mapped along any branch in the subfamily majority-rule tree (Figs. 9.27± 9.29), the probability that all of them occurred on a single axis by chance is less than 1%, according to the concentrated changes test of Maddison and Maddison (1992). Cambridge Books Online © Cambridge University Press, 2009 536 History and evolution Figure 9.28. The mapping of fruit type over the majority-rule consensus tree for the ndhF gene. Bromelioideae Data derived from the ndhF gene also provide the ® rst explicit phylogenetic hypothesis for Bromelioideae built on DNA sequences, and in this case rooted on genera representing Pitcairnioideae as the sister taxon (Fig. 9.30). Unfortunately, several facts reduce the value of this tree for sweeping conclusions about intrafamilial relationships. First, only 15 of the 29 genera account for the results, and some interesting outliers (e.g., Greigia, Fascicularia) are absent. Moreover, just one species (Aechmea haltonii of subgenus Podaechmea) represents the largest and most controversial of the bromelioid genera. Finally, Bromelioideae clearly exceeds Tillandsioideae, and possibly also Pitcairnioideae, for varied reproductive morphology. Nevertheless, Fig. 9.30 illustrates several noteworthy features, most importantly the resolution of three major clades based on the strict consensus of 136 shortest trees of 157 steps. An unresolved core group contains 11 genera. Among its component taxa, Nidularium, Neoregelia and Wittrockia probably relate more closely to one another than to any of the Books Online © Cambridge University Press, 2009 Relationships among subfamilies and Bromeliaceae within Liliopsida 537 Figure 9.29. The mapping of seed morphology over the majority-rule consensus tree for the ndhF gene. other core genera. However, affinities among members of these three taxa, most of which possess nidulate in¯ orescences (Fig. 3.2A), remain obscure (Leme 1997), as are the relationships between the entire, apparently monophyletic assemblage and the other core Bromelioideae. Leme (1998a,b) is currently reorganizing the nidularoid complex employing more characters ± still all morphological ± to ® t a taxonomy expanded to include at least two new genera (Canistropsis, Edmundoa). Ananas, Cryptanthus and Orthophytum form still another unresolved clade positioned basal to the core. Weaker evidence also supports Ananas as the sister taxon to core Bromelioideae and Cryptanthus and Orthophytum as sister taxa and basal to the Ananas-core Bromelioideae clade. Bromelia and Puya form the third and basalmost clade, also unresolved, in the strict consensus tree. Removing Puya (not shown) results in a strict consensus (120 trees, 142 steps) with identical topology. Puya is an especially pivotal taxon because of its potential to reveal how the features that distinguish Bromelioideae among bromeliads arose from less Cambridge Books Online © Cambridge University Press, 2009 538 History and evolution Figure 9.30. A strict consensus of 136 shortest trees of 157 steps for Bromelioideae based on the ndhF gene and rooted with sister taxon sequences (Dyckia and Encholirium). Three major clades are identi® ed. An unresolved core group of 11 genera includes Nidularium, Neoregelia and Wittrockia that appear to be more closely related to one another than to any of the other eight genera. specialized conditions that continue to characterize Pitcairnioideae (e.g., inferior ovary modi® ed for zoochory from a dry capsule). Pitcairnioideae Several studies utilizing DNA structure to infer bromeliad phylogeny have included one or more pitcairnioid species (e.g., Givnish et al. 1997; Terry et Cambridge Books Online © Cambridge University Press, 2009 Relationships among subfamilies and Bromeliaceae within Liliopsida 539 al. 1997a,b). Seed and ¯ ower morphology and additional traditional characters drawn from other reproductive and some vegetative organs have received more attention in attempts to identify relationships within Pitcairnioideae (e.g., Varadarajan and Gilmartin 1988a,b). Just one report deals comprehensively with this subfamily and uses molecular characters ± although in this case less to address bromeliad taxonomy than to investigate historical shifts in ecophysiology, a goal especially well designed to target ecophysiologically diverse Bromeliaceae. Crayn et al. (1999) examined 36 species in 11 pitcairnioid genera to help determine patterns of change leading from C3 to CAM metabolism in a clade chosen because these two syndromes occur among closely related lineages (Chapter 4), sometimes even among members of the same genus. Successful pursuit of questions of this nature requires knowledge of phylogeny as does any attempt to cast an adaptive radiation in geologic time and evolutionary space. Speci® cally, Crayn et al. sought to determine whether Pitcairnioideae is monophyletic, and likewise its component genera, and the evolutionary juxtapositions of these same lineages. An 851nucleotide sequence located near the rapidly evolving 59 end of the matK locus (1520 base pairs) in the chloroplast genome provided the characters needed to construct a molecular phylogeny. While their ® ndings fell short of those required to determine how often and where in the subfamily CAM types emerged from C3 stock, the data are sufficient to support or contradict certain existing hypotheses in addition to providing some new insights including directions for further inquiry. Four of the 10 genera represented in the survey by more than one species (Brocchinia, Fosterella, Pepinia and Puya) exhibited sequence homologies consistent with monophyly. Conversely, Navia (two species from morphologically distinct parts of the genus; N. arida and N. phelpsiae) failed to form a clade. Additionally, Hechtia (four species assessed) may be arti® cial, and Deuterocohnia meziana fell nearer to closely allied Dyckia and Encholirium than the other members of its genus also used for the comparisons. Pitcairnia heterophylla failed to cluster with the other ® ve congeners analyzed, and Pepinia, although similar in vegetative structure and ecology to many pitcairnias, may not have arisen from within this genus. Givnish et al. (1997) determined that Brocchinia serrata probably belongs in another genus, a ® nding which when juxtaposed to the Crayn et al. revelations illustrates how choice and number of samples can in¯ uence answers to questions about phylogeny. matK sequence homologies among the sampled Pitcairnioideae also shed light on the broader issues of where Bromeliaceae belongs within Liliopsida, how its subfamilies relate, and what the genetic divergences Cambridge Books Online © Cambridge University Press, 2009 540 History and evolution among species may tell us about the age of the family and how recently its component lineages multiplied. Of the 851 sequence positions examined, 171 or 38% varied just within Bromeliaceae (the analysis also included one member each of Ananas, Guzmania, Tillandsia and Vriesea), and 72 (8%) of these sites were cladistically informative. More interesting, maximum pairwise divergence was just 5.45% between Fosterella penduliflora and Guzmania monostachia. Similar ® ndings by other investigators using other genomic loci (e.g., Terry et al. 1997a,b and the ndhF locus) indicate that the unusually close intrafamilial relationships among bromeliads indicated by this measure are family rather than gene speci® c. Low mutation rates may account for these extraordinarily low values compared with those separating the members of other groups of ¯ owering plants; however, they could also signal rapid speciation. The second possibility, especially if it accompanied recent emergence of family-distinguishing characters, would also explain the nearly exclusive Neotropical distribution of Bromeliaceae and the absence of a credible pre-Eocene family record. Final comments Despite much continuing ambiguity, consensus is developing on several points concerning origins, status among the monocots, and the phylogenetic juxtapositions of the higher bromeliad taxa. Bromeliaceae is relatively isolated within Liliopsida and probably monophyletic as are Tillandsioideae and Bromeliaceae. Tillandsioideae seems to have branched off from the rest of the family early in its history leaving Pitcairnioideae and Bromelioideae more closely related. Brocchinia, or most of its recognized species, possesses an extraordinary mix of distinguishing characteristics, unique enough perhaps to justify subfamily status. Features responsible for the exceptional stress-tolerance and unparalleled capacity to utilize unconventional substrates (e.g., CAM, phytotelma, absorbing trichomes) exhibited by so many Bromeliaceae are sufficiently homoplasious to seriously challenge attempts at systematic inference. Still equivocal are the closest relatives of Bromeliaceae, which when identi® ed should allow polarization of important characters and recognition of evolutionary directions. Examined genomic sequences identify Rapataceae as the likely sister group of Bromeliaceae, but other candidates exist, and some of those suggested (e.g., aquatic Mayacaceae; Givnish et al. 1998) by their own unusual ecophysiology greatly expand the potential selective pressures that set in motion the modi® cations of structure and function that underlie current bromeliad success in so many, often demanding hab- Cambridge Books Online © Cambridge University Press, 2009 Final comments 541 itats, and permit the more specialized lineages to deploy carnivory and trophic myrmecophytism and serve as major resources for extensive forest canopy biota. Securement of a sufficiently resolved phylogeny will require analysis of additional parts of the genome chosen to match rates of evolution with the hierarchical levels of the taxa being compared. Sampling must also include more lineages, especially those in problematic groups or apparently rooted near the base of the family tree (e.g., Brocchinia and other Guayanan Pitcairnioideae, Catopsis, Glomeropitcairnia) and the putative bromeliad relatives. Progress has been uneven and the work yet to do is no less so. Clearly, Bromelioideae presently holds the record for being least known of the bromeliad subfamilies on many counts. Although several apomorphies (e.g., berry fruits, gelatinous outer ovular integument) de® ne this clade among Bromeliaceae, its members encompass considerable, poorly characterized ecological diversity and perhaps more varied vegetative and reproductive organs than present in either of the other two (three?) subfamilies. Success is also contingent on improved understanding of higher levels of plant structure and function and especially development. Heterochrony warrants greater attention than it has been accorded so far ± more species compared as seedlings and adults relative to morphology and ecophysiology. Additionally, morphometric analysis could more precisely indicate how faithfully adult architecture matches certain stages of the developmental programs of putative ancestral types. The litany could go much farther. In fact, progress toward a truly comprehensive synthesis of bromeliad history is contingent on further advances in virtually all of the subjects considered in the nine chapters just concluded. Cambridge Books Online © Cambridge University Press, 2009 Cambridge Books Online © Cambridge University Press, 2009 Part three Special topics 10 Neoregelia subgenus Hylaeaicum I. RAMÍREZ Neoregelia, with about 100 species (Luther and Sieff 1996), belongs to subfamily Bromelioideae, and consists of two subgenera with largely nonoverlapping ranges: Neoregelia with about 90 species and subgenus Hylaeaicum with 10 species. Subgenus Neoregelia is confined to eastern Brazil except for one species each in northern Venezuela (N. cathcartii) and Amazonian Peru (N. johnsoniae). Subgenus Hylaeaicum is entirely Amazonian in parts of Colombia, Venezuela, Peru, Ecuador and Brazil. Neoregelia is distinguished from the other bromelioid genera with nidular inflorescences (Canistrum, Nidularium and Wittrockia) by its asymmetric sepals and lack of petal appendages (Leme 1998a,b). However, more recent studies (Ramírez 1991, 1994) have determined that petal appendages occur in members of subgenus Hylaeaicum, and Leme (1997) reported these same organs in subgenus Neoregelia (N. carolinae), indicating need to re-evaluate the taxonomic utility of this character. Taxonomic problems Nidularium eleutheropetalum and N. myrmecophilum were successively assigned to different sections of Nidularium, and the genera Karatas and Aregelia until in 1890 Lindman placed them in genus Regelia, which he created by elevating the status of Nidularium subgenus Regelia Lemaire. In 1891 Kuntze had proposed the name Aregelia as a nomen novum for Nidularium, so that its typification must be identical to that of Nidularium. Therefore Mez’s decision to use Aregelia for a genus segregated from typical Nidularium is invalid. Genus Regelia was named after the German botanist A. von Regel, who served as superintendent of the Imperial Botanic Gardens in St Petersburg, Russia. Because the name Regelia had already been assigned to three 545 546 Neoregelia subgenus Hylaeaicum species of Myrtaceae, Smith (1934b) created the name Neoregelia, considering Regelia Lindman and Aregelia Mez, 1896 non Kuntze, 1891, to be synonyms. Smith and Downs (1979) combined Nidularium subgenus Regelia, Karatas section Regelia, and Aregelia subgenus Eu-Aregelia under Neoregelia. Subgenus Hylaeaicum is distinguished from subgenus Neoregelia by its pedicellate flowers, but this delimitation is problematic. My investigation indicates that additional features including position, length and origin of the stolons, bract phyllotaxis (distichous vs. spiral), degree of connation of the sepals, presence of petal scales, and stigma shape also help circumscribe the two subgenera. Smith (1967) erected Neoregelia subgenus Amazonicae, citing Nidularium eleutheropetalum as the type, to include those species of Neoregelia with free petals and a distinct pedicel and restriction to Amazonia. In 1976 Smith resurrected the name Hylaeaicum proposed by Ule in 1907 for Nidularium eleutheropetalum and N. myrmecophilum. Smith and Downs (1979) placed Nidularium subgenus Hylaeaicum Ule and Aregelia subgenus Hylaeaicum (Ule) Mez in synonymy with Neoregelia subgenus Hylaeaicum. Moreover, they noted that Neoregelia aculeatosepala alone possesses a distinct pedicel, eliminating one of the best taxonomic markers for subgenus Neoregelia. More recently, Leme (1997) transferred N. aculeatosepala to Aechmea (A. aculeatosepala) based on the presence of erect petals, serrate sepals, apically obtuse-cucculate, triporate pollen, creased leaves between sheath and blade, and an unbroken crescent or a V-shaped fold clearly visible on the abaxial surface. Also characteristic is the fleshier fruit containing an abundant, sweet gelatinous matrix. Ramírez’s (1991, 1994) studies on vegetative and floral morphology suggest that subgenera Neoregelia and Hylaeaicum may not be closely related and that Neoregelia is polyphyletic or paraphyletic. H. A. Luther (personal communication) has indicated that subgenus Hylaeaicum may share ancestry with members of Aechmea subgenus Aechmea. Leme (1997) also considers Hylaeaicum distinct from subgenus Neoregelia and closer to Aechmea and Canistrum (because of the bluish fruits). I believe that subgenus Hylaeaicum could be related to Aechmea subgenus Lamprococcus, with which it shares petal scales, apical placentation, stoloniferous habit, and Amazonian distribution. Similarities, especially the nidulate inflorescence, with subgenus Neoregelia are probably homoplasious. The systematics of the bromelioid genera with nidulate inflorescences warrant further study to determine whether this condition evolved more than once within Bromelioideae. Molecular data will be especially informative. Vegetative morphology 547 Ecology and geographic distribution Species of subgenus Hylaeaicum mostly inhabit Amazonian rainforests except for a few members that range through pre-montane rain or cloud forests on the eastern slopes of the Andes. About 80% of the species occur between 50 and 600 m. Neoregelia aculeatosepala alone inhabits cloud forests above 1000 m, consistent with Leme’s decision (1997) to transfer it to Aechmea. Distributions suggest that subgenus Hylaeaicum originated in Amazonian Peru and Ecuador. Species of Hylaeaicum belong to Pittendrigh’s (1948) ecological Type Two (Table 4.2), although epiphytism prevails except for the occasional terrestrial or humicolous population on well-drained soil. Finer details of shoot architecture follow two patterns. Shoots in one case feature inflated leaf sheaths forming a ‘multitank’ (sensu Benzing 1980) that impounds water and litter. Trichomes occur sparsely over the leaf sheath surfaces, and most species attract nesting ants (Davidson 1988; personal observation and notes on herbarium labels). Members of the second group inhabit rain or cloud forests at moderately high elevations (800–1600 m). Here, the leaf sheaths form small impoundments that hold almost no water (‘single tank’ sensu Benzing 1980), and the leaves are thinner and covered by a dense mantle of trichomes distinct from those of the first group. Members are always stoloniferous epiphytes. Cytology Published counts for Neoregelia match the basic chromosome number (2n 550) of the family (N. spectabilis, N. concentrica, N. cyanea and two unidentified species; Marchant 1967). Only one number exists for Hylaeaicum (Ramírez 1991; Neoregelia peruviana, 2n550). Vegetative morphology Leaf sheaths that vary in morphology and disposition cover the often long stolons. Typically nidulate inflorescences are usually surrounded by brightly colored young leaves. Most species form colonies of several to many ramets (up to 300–400 for N. pendula var. pendula and N. tarapotoensis), suspended from branches or the trunks of trees, or over rocks or cliffs. Species endemic to Andean foothill forests in Ecuador and Peru tend to form compact, small ramets. Stolons are exceptionally slender and elongate (e.g., N. tarapotoensis). Natives of lowland rainforests tend to form larger 548 Neoregelia subgenus Hylaeaicum funnelform shoots with the foliar sheaths abruptly broadened to form a large phytotelma. Colonies usually consist of a few ramets interconnected by relatively short, stout stolons. Leaf blades are narrow-triangular or ligulate, rarely whip-like, and the margins spiny; color is mostly green, but some red forms of N. eleutheropetala occur in Colombia (Leticia) and Ecuador (Morona-Santiago). Particularly ornamental stock are growing at the Marie Selby Botanical Gardens in Florida, preparatory to flowering for identification. Trichomes The structure of the trichomes on the leaf sheath suggests the presence of two groups of species in the subgenus (Fig. 2.6). Group One, which I describe as the ‘pliable type’, mostly includes species native to low elevations. Shields are radial, containing eight, usually quadrangular ring cells that sometimes feature irregular shapes and asymmetrical arrangements. Wings are scarcely developed in N. eleutheropetala, N. leviana, N. myrmecophila, N. rosea and N. stolonifera, but well defined in N. wurdackii. Densities vary without discernible patterns. The second group, my ‘rigid type’, mostly includes species from the Andean foothills. Trichomes lack shields with the degree of organization just described. Instead, component cells differ in size and shape. Abundant trichomes on both leaf surfaces produce a highly reflective indumentum (e.g., N. peruviana, which forms small rosettes throughout its range in Peru at 250–550 m). Inflorescences Members of subgenus Hylaeaicum produce terminal, usually simple inflorescences on determinate shoots (Chapter 2; Fig. 3.5B). Some species undergo additional branching on the periphery of spent inflorescences, resulting in multiseason fruiting by the same ramet (e.g., both varieties of N. eleutheropetala, N. myrmecophila and probably N. leviana, N. margaretae; Leme 1997). Species of Neoregelia subgenus Hylaeaicum advertise by developing colorful foliage (Fig. 2.13F) at anthesis, specifically: bright red, dark red or dark pink (e.g., N. margaretae, N. rosea, N. pendula var. brevifolia, N. eleutheropetala var. bicolor, N. myrmecophila), purple (e.g., N. leviana, N. stolonifera, N. pendula var. pendula) or green inner leaves (e.g., N. eleutheropetala var. eleutheropetala, N. mooreana, N. peruviana, N. tarapotoensis). Sometimes the whole ramet or just the inner leaves become red Reproductive biology 549 depending on exposure (N. eleutheropetala; H. Luther, personal communication). Floral morphology Certain members of subgenus Hylaeaicum possess petal scales that occur in two forms (Ramírez 1994). The scale apex is either acute or fimbriate, or comprised of finger-like projections (only N. pendula var. brevifolia of all the species studied). Species of subgenus Neoregelia studied by Brown and Gilmartin (1989b) possessed conduplicate-spiral stigmas (Figs. 3.1C, 12.1). Mez (1934–35) reported densely capitulate, contorted stigmas for Hylaeaicum, Canistrum and Nidularium. Three stigma architectures (Brown and Gilmartin 1989b) have been reported for subgenus Hylaeaicum (Ramírez 1994). The simpleerect type was observed in Neoregelia leviana, N. myrmecophila, and N. pendula var. brevifolia, and the conduplicate-spiral form in N. tarapotoensis, N. mooreana, and N. margaretae. Neoregelia stolonifera and N. eleutheropetala var. eleutheropetala exhibit convolute-blade morphology (but see Leme (1997) who failed to record this condition among these species). Stigma type does not parallel other taxonomic characters within subgenus Hylaeaicum. Reproductive biology Flowers bear white petals and green sepals except for N. pendula var. brevifolia, which has white sepals and N. eleutheropetala var. bicolor with its redtinged sepals and floral bracts. Anthesis usually occurs in early morning, and flowers remain open for one or rarely two days. Stamens lie close to the stigma, and both are included, although the stigma lies 1–2 mm below the stamens in some species. Spreading corollas typically protrude above the characteristic flat inflorescence. High fruit set in closed greenhouses suggests that most of the taxa (7 out of 10 studied) are either autogamous or agamospermous, in contrast to subgenus Neoregelia where allogamy prevails (Smith and Downs 1979). Close contact between stamens and stigma in most subgenus Hylaeaicum species accords with autogamy. Number of offshoots per parent ramet varies with the species. One is typical for many populations (e.g., both varieties of N. eleutheropetala), but N. pendula var. brevifolia regularly produces five. Neoregelia subgenus Hylaeaicum illustrates at least three distinct architectures, each associated with a specific pattern of sympodial branching. 550 Neoregelia subgenus Hylaeaicum The first model features small ramets, one to five in number, derived from each parent ramet. Pendulous ramets hang by long, slender, wiry stolons covered by polystichous, spineless, usually dry, brown bracts (e.g., N. pendula, N. tarapotoensis, N. peruviana). Ranges are narrow and confined to the eastern slopes of the Ecuadorian and Peruvian Andes. The second model is characterized by somewhat pendulous, short, stout stolons covered by mostly green, rarely brown, spiny, polystichous or distichous bracts (e.g., N. stolonifera, N. mooreana, N. eleutheropetala and N. myrmecophila). Each parent ramet of Neoregelia stolonifera produces a pair of offshoots on straight, elongate, but stout stolons leading to an expanding regular, dichotomous arrangement. The third model features ramets that branch just once and remain connected by short, stout stolons covered by strongly distichous, green, spiny sheaths bearing small, but conspicuous, rigid and pungent foliar lamina. Interconnected ramets are slightly pendulous to scandent (e.g., N. margaretae, N. leviana, N. wurdackii and N. rosea). Ramets of some cultivated species wither after one or two years, and may serve as seed beds for undispersed progeny (e.g., N. leviana). Like many other Bromelioideae, morphology and fruit color and leaf display suggest that members of subgenus Hylaeaicum rely on birds to disperse. Inner leaves are tinged with bright red, pink or purple, and the sepals become blue or violet. Ovaries are typically white or green, and the fruit protrudes above the infructescence heightening its visibility (Fig. 3.6F). Ripe berries readily detach. Davidson (1988) reported dispersal by ants for N. eleutheropetala whose seeds contain attractive monoterpenes and aromatics (Seidel et al. 1990; Chapter 6). Methyl-6-methylsalicylate, one of these monoterpenes, is present in the mandibular glands of Camponotus femoratus, consistent with Ule’s (1906) hypothesis that nest-garden Bromelioideae promote seed carriage by mimicking ant brood (Davidson 1988). Continuing taxonomic problems Clearly the affinities of members of Neoregelia subgenus Hylaeaicum remain poorly resolved, and will remain so without additional inquiry. Closer scrutiny of morphology would help, but molecular data are essential not only to identify where subgenus Hylaeaicum belongs in its subfamily, but also to sort out the relationships of large, artificial bromelioid genera like Aechmea. At this point, we can say with fair confidence that the 10 members of subgenus Hylaeaicum do not belong in genus Neoregelia, but whether assignment to some other genus or a new one is warranted remains less certain. 11 Cryptanthus I. RAMÍREZ Cryptanthus (Bromelioideae) is endemic to southeastern Brazil, strictly terrestrial, and ranges from Paraíba to Rio de Janeiro states, with one species in Goiás State. Its 41 species inhabit wet forests, restingas, caatingas and campos rupestres, occurring from sea level to about 2000 m. This genus differs from all others in the family by its consistently low chromosome number (n517) and often andromonoecious flowers with simple-erect stigmas (Fig. 3.1C). Ramírez (1996) recently supported the case for the recognition of two subgenera proposed by Mez (1896) with geographical, morphological and ecological data. Characters that distinguish the two taxa are summarized in Table 11.1. Members of subgenus Cryptanthus are subdivided into five sections, members of which occur in Rio de Janeiro, Espirito Santo, Minas Gerais, Bahia, Sergipe, Pernambuco, Paraíba and Goiás states (Fig. 11.1). Leaves of some of these species possess a mid-longitudinal region thickened by a succulent, many-layered hypodermis. Reproduction tends to occur primarily by stolons capable of expanding individual genets to form large colonies (Figs. 2.11C,D, 2.18C). Members of subgenus Hoplocryptanthus occur in Espirito Santo State, except for four species native to Minas Gerais. Fruit set is more common in this group compared with subgenus Cryptanthus (evident on herbarium collections and among cultivated plants). Ramírez (1996) tested Brown and Gilmartin’s (1989a) hypothesized chromosome number evolution and related implications for phylogenetic relationships within Bromeliaceae (Chapter 9). Brown and Gilmartin proposed that the low number of chromosomes in Cryptanthus (n517 or 18) could have originated by descending aneuploidy or constitutes the ancient tetraploid condition according to the scheme illustrated in Fig. 9.10. Determinations of nuclear DNA content for species with n517 or 18 551 552 Cryptanthus Table 11.1. Characters that distinguish the two subgenera of Cryptanthus Characteristic Habit Breeding system Flower fragrance Corolla shape Petal blade shape Stigma lobes Pollen surface Seed number Seed size Habitat Elevation Subgenus Cryptanthus Subgenus Hoplocryptanthus Mostly acaulate Andromonoecy Usually absent Open, reflexed petals Oblong or narrow elliptic Spreading Reticulate Relatively low (c. 8) Relatively large Restingas, caatingas lowland rainforest 0–700 m Mostly caulescent Hermaphroditism Usually present Campanulate or flat Orbicular or wide elliptic Connate Smooth or finely reticulate Relatively high (c. 40) Relatively small High-altitude grasslands, montane forests 800–2000 m (Cryptanthus) and others characterized by n525 (other Bromelioideae) using flow cytometry accorded with descending aneuploidy (Ramírez 1996). Additional studies on seeds (Gross 1988a) and septal nectaries (Böhme 1988) support primitiveness for Cryptanthus, but the presence of andromonoecy, fragrant flowers, a simple-erect stigma and CAM metabolism (e.g., Medina 1990) suggest otherwise. Should Brown and Gilmartin’s (1989a) proposal that the low chromosome number in Cryptanthus reflects ancient tetraploid status (hypothesis two) be accepted, the genus should be elevated to subfamily status (Cryptanthoideae). I propose retention of Cryptanthus in Bromelioideae as an exceptionally derived genus with Orthophytum as its sister group according to cladistic analysis (Ramírez, in preparation) and some DNA sequence data (Terry et al. 1997a). Additional molecular data should help confirm Cryptanthus as a clade within Bromelioideae. Cryptanthus 553 Figure 11.1. Distributions of the two subgenera of Cryptanthus. w, subgenus Cryptanthus; d, subgenus Hoplocryptanthus. 12 Tillandsioideae W. T I L L Tillandsioideae are mostly rosulate herbs characterized by entire leaf margins, radially organized peltate trichomes (Fig. 2.7), usually superior (or nearly so) ovaries, and three-parted capsules that contain plumoseappendaged seeds (Fig. 3.6J; Wittmack 1888; Baker 1889; Mez 1894, 1896, 1934–35; Harms 1930; Smith and Downs 1977; Rauh 1990). Stigma morphology varies more than in the other two subfamilies, with at least five different types present (Brown and Gilmartin 1989b; Gortan 1991; Figs. 3.1C, 12.1). Pollen morphology is similarly variable (Fig. 12.2). Grains are predominantly sulcate with a distal germination region, and represent the diffuse sulcus, insulae-type, operculum-type and Vriesea imperialis-type (Halbritter 1992). Catopsis is exceptional with its simple sulcus, while inaperturate pollen occurs in some Guzmania species. This subfamily comprises the genera Alcantarea (16 spp.), Catopsis (21 spp.), Glomeropitcairnia (2 spp.), Guzmania (176 spp.), Mezobromelia (9 spp.), Racinaea (56 spp.), Tillandsia (551 spp.), Vriesea (187 spp.) and Werauhia (73 spp.) (Smith and Downs 1977; Utley 1983; Till 1984, 1992b, 1995; Kiff 1991; Grant 1993a,b, 1994a,b, 1995a,b; Spencer and Smith 1993; Luther and Sieff 1996; Till et al. 1997). According to Smith and Downs (1977), Spencer and Smith (1993) and Grant (1995a), the genera, subgenera and sections of Tillandsioideae can be distinguished using the following key: 1 Ovary only half superior; seeds equally plumose-appendaged at both ends; flowers polystichous; petals free, bearing two scales at base .................................................................................Glomeropitcairnia 1* Ovary nearly or quite superior; seeds plumose at the base or apex or largely on the base and only slightly on the apex; petals free or conglutinated, naked or with basal scales ......................................................2 555 556 Tillandsioideae Figure 12.1. Stigmas of selected Tillandsioideae (courtesy of Gunter Gortan). All scale bars50.5 mm. (A) Catopsis nutans, simple-erect. (B) Guzmania sanguinea, convolute-blade. (C) Tillandsia bergeri, simple-erect. (D) T. lindenii, coralliform, view from above. (E) T. lindenii, view from below. (F) T. castellanii, simple-erect. (G) Tillandsia 3 polita, conduplicate-spiral. (H) Vriesea simplex (section Vriesea), convolute-blade. (I) Werauhia gigantea, cupulate. Tillandsioideae 2 557 Appendage of the seed largely apical, folded at maturity; sepals strongly asymmetric in most species; ovules with an apical tuft of cellular strands; flowers in at least slightly more than two ranks, unisexual in some species; leaves usually cretaceous-coated ........................Catopsis 2* Appendage of the seed wholly or largely basal, straight at maturity; sepals symmetric or if asymmetric, then the ovules unappendaged and obtuse; flowers always hermaphroditic................................................3 3 Petal claws conglutinated into a tube, equaling the sepals or rarely the petals entirely included in the sepals; spikes mostly with polystichous organization; ovules cylindric, obtuse; seed coma usually of various shades of brown .................................................................................4 3* Petal claws free or with a very short tube exceeded by the sepals; flowers distichous in most species; ovules usually with an apical appendage; seed coma usually white......................................................................5 4 Petal claws naked ..................................................................Guzmania 4* Petal claws bearing scales on the inside...........................Mezobromelia 5 Petals bearing scales on the inside.......................................................6 5* Petals naked........................................................................................9 6 Stamens exserted and stigma of the conduplicate-spiral type morphology; ovules appendaged .........................Tillandsia subgenus Tillandsia 6* If stamens exserted then the stigma not of the conduplicate-spiral type morphology ........................................................................................7 7 Seed with the apical appendage divided into a short coma; petals linear, usually 10–15 times longer than wide, soon flaccid and drooping; ovules appendaged .........................................................................Alcantarea 7* Seed with the apical appendage minute and undivided; petals shorter, elliptical, only 5–10 times longer than wide, firm and remaining more or less erect after anthesis ...................................................................8 8 Flowers with brilliant coloration in most species, bright yellow, orange or red, rarely dulled to white, light yellow or light orange; the adaxial petal pair arranged basally with respect to the abaxial; petal appendages tongue-shaped; stigma usually with the convolute-blade type 558 Tillandsioideae morphology; ovules appendaged ...............................................Vriesea a Stamens included; floral bracts mostly dull green or brownish ................................................................................section Xiphion a* Stamens exserted; floral bracts mostly brightly colored with red, orange or yellow ......................................................section Vriesea 8* Flowers generally dull in color, white, greenish white, light green, yellowish green, yellow or light orange; the adaxial petal pair arranged apically with respect to the abaxial; petal appendages dactyloid with 1–5 fingers of varying length; stigma with the cupulate type morphology; ovules not or barely appendaged............................................Werauhia a Inflorescence simple or compound; when compound, the inflorescence is classically bipinnate or tripinnate, and the lateral branches exceed the primary bracts .....................................section Werauhia a* Inflorescence compound, though appearing to be simple in some species (a pair of flowers indicates a reduced lateral branch), those that appear simple have only pedicellate flowers; lateral branches usually shorter than, but in some species exceeding, the subtending primary bracts..........................................................section Jutleya 9 Sepals asymmetric, free or nearly so, broadest near the apex, not over 12 mm long; ovules obtuse .....................................................Racinaea 9* Sepals symmetric or if slightly asymmetric, then ovate or lanceolate and broadest below the middle, free or variously connate, usually longer than 12 mm ...........................................................................Tillandsia a Stamens included in the corolla, equaling the petals or shorter ......................................................................................................b a* Stamens exserted from the corolla, exceeding the petals ................e b Style slender, much longer than the ovary; filaments from equaling the claw to equaling the entire petal ..............................................c b* Style short and stout; stamens deeply included in the corolla........d c Stamens exceeding the claw of the petal; filaments straight; ovules usually appendaged............................................subgenus Allardtia c* Stamens about equaling the claw of the petal; filaments strongly plicate in many species; ovules appendaged or obtuse ..................................................................subgenus Anoplophytum d Petal blades broad, conspicuous; leaf blades flat or terete, green or cinereous-lepidote, petal color mainly blue-violet; ovules obtuse (mesic members) or short appendaged (xeric members) ......................................................................subgenus Phytarrhiza Anatomy and morphology 559 d* Petal blades narrow, inconspicuous; leaf blades terete, densely cinereous-lepidote, petal color mainly yellowish or brown; ovules short appendaged...................................subgenus Diaphoranthema e Petals erect at anthesis or nearly so, relatively firm; leaf blades narrowly triangular in most species; ovules distinctly appendaged .........................................................................subgenus Tillandsia e* Petals subspreading at anthesis and rapidly becoming flaccid; leaf blades lingulate to narrowly triangular, ample; ovules appendaged or obtuse ................................................subgenus Pseudalcantarea Anatomy and morphology Roots Roots are usually present, simple or branched, and often reduced to holdfasts (Tomlinson 1969; Downs 1974) that sometimes secrete a ‘brown mastic’ (Schimper 1884; Wittmack 1888) or gum-like (Chodat and Visher 1916; Rauh 1990) substance. According to Brighigna et al. (1990), the adhering roots secrete lipo-polysaccharides and bear unicellular hairs. Adult Tillandsia usneoides usually lacks roots. Generally, the extent of rooting is inversely related to trichome density, leaf succulence and impoundment capacity (Benzing 1980). Roots occasionally develop on inflorescences (Tomlinson 1969), especially in certain tillandsias (e.g., T. latifolia, T. secunda, T. somnians). The primary root, if present, is quite short and covered by an endostome cap (Downs 1974; Gross 1988a,b). A crown of root hairs develops in a few species (e.g., Vriesea fosteriana; Gross 1988b). Adventitious roots appear early, and unlike the primary root bear typical caps (Fig. 2.15) and possess a thin rhizodermis. Roots of the more mesic species produce hairs. The hypodermis (5exodermis of Tomlinson 1969) consists of a multiseriate layer of suberized cells (Schimper 1884; Tomlinson 1969). The outermost cortical layers immediately within the hypodermis form a wide sclerotic cylinder which can be more peripheral (e.g., Tillandsia baileyi, T. geminiflora, Vriesea fenestralis), or central as in Catopsis morreniana, Guzmania monostachia, G. nicaraguensis, G. zahnii, Tillandsia aeranthos, T. cyanea and T. tricolor (Strehl and Winkler 1983). The central cortex is parenchymatic and sometimes starch-filled with the inner cortex lacunose. The endodermis is uniseriate and composed of uniformly thickened cells (Tillandsia, Vriesea; Meyer 1940). Pericycles are one or two-layered and thin-walled, rarely with tertiary thickenings in some Tillandsia species (Meyer 1940). The primary 560 Tillandsioideae vascular system is an exarch protostele (Downs 1974) with the number of arches reduced in the epiphytes (Meyer 1940). Roots originate in the meristematic layer (pericambium5pericycle) between the stem cortex and central cylinder (Tomlinson 1969) and remain within the cortex of the stem for some distance before emerging (Schindler 1957; Tomlinson 1969; Rauh 1990; Chapter 2; Fig. 2.15). Intracauline roots strengthen the stem (Borchert 1966), and may largely displace the cortical parenchyma (Tomlinson 1969). Vascular tissue tends to be reduced in roots specialized for anchorage on bark and rock (Harms 1930). Phloem is reduced more than the xylem tissue in the most stress-tolerant Tillandsia species (Downs 1974). Vessel element perforations are occasionally simple, but usually scalariform (Cheadle 1955; Goldberg 1989; Fig. 2.21). Stems Leafy stems are usually short, the foliage congested, and growth is typically negatively geotropic (Fig. 2.10). Less often shoots are more caulescent, but the leaves are just as densely packed (Fig. 2.1). Growth of the elongated types can be ageotropic (Tomlinson 1969). A small group of reduced Tillandsia species exhibits distichous phyllotaxis (mainly subgenera Diaphoranthema and Phytarrhiza; T. albertiana). Stolons equipped with scale leaves are produced by several species (e.g., Tillandsia juncea, T. espinosa). Stolons, rhizomes and leafy stems exhibit similar anatomy. Stems are consistently differentiated into a cortex and a central cylinder (Fig. 2.15; Boresch 1908; Meyer 1929; Tomlinson 1969; Downs 1974). Vascular cambia are absent, but a periderm (cork tissue) may develop from secondary meristematic layers in older parts of stems and around leaf scars and wounds (Tomlinson 1969; Chapter 2). A thin cuticle covers a uniform epidermis made up of cells containing a prominent silica body (Fig. 4.23I). Stomata are absent and the peltate hairs are restricted to regions immediately above the leaf insertions (Tomlinson 1969). The cortex is narrow and tends to become lignified and further rigidified by leaf traces and intercauline roots (Fig. 2.15; Tomlinson 1969; Downs 1974; Benzing 1980). Gum and mucilage occur throughout the stem interior. Shoots, and occasionally the entire plant, are mostly determinant with apical meristems finishing growth by producing an inflorescence (Fig. 2.3). The exceptional monocarpic species (e.g., Tillandsia makoyana) fruit without producing axillary ramets. Most species produce successive determinate shoots by sympodial branching (Rauh 1990; Fig. 2.3A). Occasional individuals (depending on the population) of some of the monocarpic species produce ramets (e.g., Tillandsia utriculata). Anatomy and morphology 561 Inflorescence axes The inflorescence is usually elevated above the leaves on a stem that bears bracts or less reduced leaf-like appendages (bracts); rarely the entire structure is sessile or multiple (e.g., Tillandsia complanata, T. multicaulis). Most axes are straight or flexuose (rarely sinuose), terete or angled, or rarely excavated. Cortex and central cylinder are clearly differentiated. Vascular bundles are more regularly distributed through cross-sections than in leafy stems. The two types of inflorescences are distinguished by bundles that are either isolated and scattered, with those near the periphery narrower, but surrounded by more massive fibrous sheaths compared with the others occupying the central region (e.g., Guzmania), or the bundles at the periphery of the central cylinder are embedded in a sclerotic cylinder that delimits the cortex, while those deeper in the axis possess a complete fibrous sheath (e.g., Catopsis, Tillandsia, Vriesea). Several species of Vriesea possess vascular bundles without sclerenchyma sheaths. Species less easily assigned to either type occur in Tillandsia subgenera Anoplophytum, Diaphoranthema and Tillandsia, and Guzmania (Tomlinson 1969). Trichomes densely invest some inflorescences, especially those of the xerophytes. Leaves Leaves are always entire, and, except for the most reduced forms, usually differentiated into a sheath and a blade (e.g., Fig. 2.7I). An intercalary meristem located just above the base generates the blade (Benzing 1980). Distinct midribs are absent. Leaves of seedlings and juveniles are narrowly triangular to filiform with typically an acute (Tillandsia) or rounded (Vriesea) apex, while adults tend to produce more lingulate foliage. Heterophylly can be pronounced (Mez 1896; Harms 1930; Smith and Downs 1977; Adams and Martin 1986a,b,c; Fig. 4.9). The foliage of mesic taxa is thin and the mesophyll made up of relatively small, densely cytoplasmic cells. Blades are well developed and green or cyanic. Sheaths typically contain enough tannins to impart dark color (Tomlinson 1969). Trichomes are often restricted to the adaxial surfaces of the sheaths of the phytotelm forms. Foliage of the xerophytic species is thickened by the presence of considerable colorless water storage tissue (Horres 1995) and large, green vacuolate cells equipped for CAM photosynthesis (Fig. 2.10). Blades are often narrowed above the pale sheaths. The indumentum of these drygrowing taxa is more or less confluent and uniform over the entire leaf (Type Five). Rarely, spirally inrolled leaves provide holdfast (e.g., Tillandsia 562 Tillandsioideae duratii; Fig. 2.10L). Bulbs (composed of succulent leaf bases) characterize a few species (e.g., T. andreana, T. fuchsii). Pseudobulbs (composed of thinner, inflated leaf bases) with cavities between adjacent leaf bases sometimes harbor ant colonies (Benzing 1970a; Longino 1986; Eshbaugh 1987; Fig. 8.5). Leaf anatomy varies greatly depending in large part on the habit of the subject (Chapter 2). Xeromorphy consisting of thickened cuticles and epidermal cell walls and large vacuolate CAM-type cells sometimes combined with water-storing hypodermal layers characterize many succulent Tillandsia and a few Vriesea species. The more mesophytic types possess thinner foliage with the mesophyll differentiated into a central chlorenchyma between relatively thin hypodermal tissues. Cuticles are thin and smooth (Harms 1930) or granular in Tillandsia ionantha (Tomlinson 1969) or folded in Tillandsia hamaleana and Werauhia gladioliflora (Barthlott and Ehler 1977), and sometimes augmented by epicuticular waxes (Catopsis, Fig. 5.3A; Guzmania, Tillandsia; Tomlinson 1969; Barthlott and Ehler 1977; Frölich and Barthlott 1988) that at least sometimes contain a steroid fraction with oestrogenous activity (Hegnauer 1963). Tillandsioid trichomes (Fig. 2.7) consist of two foot cells, a uniseriate stalk dominated by the distal dome cell, and a shield comprised of a central disk and a peripheral wing with the component cells regularly arranged according to the formula 418132 in Catopsis, Glomeropitcairnia, Guzmania, Tillandsia subgenera Allardtia and Vriesea and genus Werauhia, 418164 in Glomeropitcairnia, Guzmania, Tillandsia subgenus Phytarrhiza (mesic), Vriesea and Werauhia, 418116132 in Catopsis and Guzmania, 4 18132164 in Catopsis, 418116164 in Tillandsia and Vriesea section Xiphion, and 418116132164 in Guzmania and Tillandsia subgenus Tillandsia (Tomlinson 1969; Strehl and Winkler 1981). These numbers demonstrate that Glomeropitcairnia exhibits the most structurally elaborate trichomes, while some Guzmania and Tillandsia subgenus Tillandsia produce the simplest appendages (Chapter 2). Stomata are restricted to abaxial surfaces (except in Catopsis berteroniana) and occur along the intercostae. Simple, unspecialized stomata with guard cells located at about the same level as the surrounding epidermal cells occur in Catopsis and Glomeropitcairnia (Fig. 2.17A). Stomata with modified substomatal cells (lobes extend from their polar ends) and sunken guard cells characterize Guzmania, Tillandsia and Vriesea. Guard cells have supposedly lost their mobility and foliar conductivity is now controlled by substomatal cells (Downs 1974), but functional guard cells continue to serve Tillandsia usneoides (Martin and Peters 1984; Chapters 2 and 4). Each unspecialized epi- Anatomy and morphology 563 dermal cell contains a prominent silica body as in the stem (Tomlinson 1969; Huber 1991; Fig. 4.23I). The hypodermis is derived from the mesophyll, and is usually adaxially differentiated into a peripheral mechanical region (sclerenchyma) and an inner water-storage layer of colorless thin-walled cells (Tomlinson 1969; Fig. 2.10). According to Saunders (1964, cited in Downs 1974), in Tillandsia malzinei the entire mesophyll contains chloroplasts, whereas Tillandsia anceps and Guzmania lingulata possess chlorenchyma embedded within water-storage tissue. The adaxial hypodermis occupies much of the leaf interior of some of the xerophytes. A typical palisade layer is absent (Downs 1974; Tillandsia usneoides; Fig. 2.10A). Intercostal chlorenchyma of the thinner-leafed forms sometimes contains large air lacunae (Tomlinson 1969). Vascular bundles (Fig. 2.17A) are always embedded in chlorenchyma, and their bulk relative to that of the rest of the leaf decreases with increasing xeromorphism. Most Tillandsioideae lack extrafascicular strands; bands of sclerenchyma have been reported in Guzmania and Vriesea just below the epidermis where they remain distinct from the veins (Tomlinson 1969). The inner boundary of the bundle sheath is represented by a distinct ligno-suberized layer, independently surrounding and thus separating the xylem from the phloem. Walls may be thickened (e.g., Catopsis, Guzmania, Tillandsia) to an extent exclusive to Tillandsioideae. Vascular tissues are uniform throughout the family (Tomlinson 1969), and consist of bundle sheath fibers, water-vascular tissue containing tracheids and vessel elements (Cheadle 1955), and sieve tube elements with companion cells (Fig. 2.21). Smaller veins contain narrower tracheal elements, especially among xeromorphic Tillandsia (Tomlinson 1969). Flavonols (Arslanian et al. 1986), triterpenes and steroids (Atallah and Nicholas 1971) characterize the leaves of Tillandsia purpurea and T. usneoides. The hypodermis contains p-coumaric acid and a ligninic substance composed of p-oxy-benzyl residues (Hegnauer 1963), ferulic acid and sinapine acid (Hegnauer 1986). Flavonoids (aglycones, e.g., 6-hydroxykaempferol 3,6,7,49-tetramethyl ether, cirsilineol and jaceosidin from Tillandsia utriculata), flavonols (e.g., gossypetin from Catopsis, patuletin from Tillandsia and Vriesea, 6,39,59-trimethoxy-3,5,7,49-tetrahydroxyflavone and 6-hydroxy-myricetinmethylester from Tillandsia) and flavones (e.g., apigenin, 6-hydroxy-luteolin, luteolin, scutellarein and 3,6,39,59tetramethoxy-5,7,49-trihydroxyflavon) have been isolated from leaves (Ulubelen and Mabry 1982; Hegnauer 1986). Flavonoids have been reported in leaf and stem exudates of Tillandsia usneoides (Wollenweber 1990; Wollenweber and Mann 1992) which suggest that chemotypes exist 564 Tillandsioideae in this species. The putatively isolated position of Bromeliaceae within Liliopsida accords with its flavonoid spectrum (Williams 1978; Chapter 9). Little is known about Tillandsioideae as substrates for phytophagous insects (Chapter 8). DeVries (1997) mentions some members as hosts for Riodinidae butterflies: Guzmania and Werauhia for Napaea eucharila, Werauhia for N. theages theages, and Werauhia for Hermathena candida. According to Beutelspacher (1972), caterpillars of Caria domitianus domitianus feed on Tillandsia. Inflorescences Inflorescences are usually compound (up to several orders), more rarely simple or even single-flowered (Gschneidner 1989; Rauh 1990; Fig. 3.3). Pedicellate flowers are typical for panicles (e.g., Guzmania diffusa); branch reduction results in racemes (e.g., Tillandsia ixioides). Sessile flowers are common on compound spike-racemes (e.g., Tillandsia multiflora, Catopsis spp.). Branch reduction in this type of inflorescence leads to simple spikes (e.g., Tillandsia incurva) or strobils (e.g., Guzmania nicaraguensis). Abbreviation of the main axis of the spike-raceme yields a digitate inflorescence (e.g., Tillandsia fasciculata) or a head (e.g., Tillandsia capitata, Guzmania glomerata). Condensation of the lateral branches to head-panicles or head-racemes occurs in the subfamily (e.g., Glomeropitcairnia). A special situation prevails in Tillandsia complanata and T. multicaulis where these heads represent compound stipitate spikes that are perfoliated to simulate several lateral inflorescences. A compound inflorescence with manyflowered branches is considered ancestral in Tillandsia (Gschneidner 1989), and probably for the whole subfamily. Branching in compound inflorescences is axillary and the expanded display often showy (e.g., several Guzmania and Vriesea species, Tillandsia subgenus Anoplophytum in part; Napp-Zinn et al. 1978). Primary bracts and lateral axes often bear an adossate prophyll and one to several basal sterile bracts. Accessory shoots originating from the same primary bract as in the pitcairnioid Cottendorfia florida occur in Racinaea commixa (Till, unpublished). The subterminal flower is always much reduced and sterile. Flowers are distichously (especially in Alcantarea, Racinaea, Tillandsia, Vriesea and Werauhia) or spirally arranged (especially in Catopsis, Glomeropitcairnia, Guzmania and Mezobromelia). Anatomy and morphology 565 Flowers Tillandsioid flowers are always trimerous, pentacyclic and heterochlamydeous, with contorted aestivation. Opening usually occurs acropetally along an axis, and protandry is common (Melchior 1964). Dioecism is restricted to one taxon (Catopsis subgenus Tridynandra). They can be pedicellate or sessile, and symmetry is radial or rarely zygomorphic (e.g., Tillandsia argentea, T. funckiana, T. paraensis, Werauhia). Flowers are sometimes secund (e.g., Alcantarea, Mezobromelia, Racinaea, Tillandsia secunda, Vriesea, Werauhia; Fig. 3.3E). Diverse fragrances attract bats (Vogel 1969; Rauh 1990), butterflies, moths and other visitors (Till 1984, 1992b; Gardner 1986a; Rauh 1990; Chapter 6). Lures include triterpenes like citronellol, geraniol and nerol (Hegnauer 1963). Many ornithophilous Tillandsia and Vriesea species lack scents (Gardner 1986a; Arizmendi and Ornelas 1990). Sepals overlap with the left margins, and are symmetric except for a few taxa (e.g., Catopsis, Racinaea). Calyx members remain separate or become variously connated. Most are greenish or pale, with brightly pigmented organs, especially in Guzmania and Vriesea, being the most notable exceptions. Adaxial members are often keeled. Sepals rarely exceed the petals (e.g., Guzmania musaica) and are membranaceous, especially toward the margin; more rarely they are coriaceous and firm (e.g., Alcantarea, Vriesea and Werauhia). Raphides occur in sepals more often than elsewhere in the flower. Petals are usually free or conglutinated into a short (Vriesea) or longer (Guzmania, Mezobromelia) tube (Fig. 6.1A). They overlap with the right margins, and are free or conglutinated. Most are concolorous (in the apical part), rarely bicolored (predominantly in Vriesea). Shapes tend to be lingulate with no distinction between the claw and the blade. Blades can be quite expanded (e.g., Tillandsia subgenus Phytarrhiza), the margins usually entire to rarely crenulate (e.g., Vriesea heterandra). Two basal scales of diverse sizes and shapes (Brown and Terry 1992) distinguish the corollas of Alcantarea, Glomeropitcairnia, Mezobromelia, Tillandsia subgenus Tillandsia, Vriesea and Werauhia from those of the rest of the subfamily (Fig. 3.1B). Petal scales become visible just prior to the expansion of the corolla (Brown and Terry 1992; Chapter 3). Folded cuticula sculptures characterize Tillandsia hamaleana and Werauhia gladioliflora (Barthlott and Ehler 1977). Raphide-containing idioblasts have been reported in the petal scales of Werauhia tarmaensis (Rauh 1983b). Stamens occur in two series (diplostemonous) of equal or unequal 566 Tillandsioideae length, but are usually shorter than the corolla except for Tillandsia subgenus Tillandsia and Vriesea section Vriesea where they exceed the petals (Fig. 6.1A,B). Filaments are flat or elliptic in cross-section, thin or succulent, erect or twisted and seldom plicated (e.g., Tillandsia subgenus Anoplophytum; Fig. 6.1A–C), linear to triangular, free or very rarely connate to a tube (e.g., Tillandsia monadelpha). Those of Guzmania and Mezobromelia are agglutinated to the petals. The inner series of stamens in Catopsis is connate with the petals for about half of the length of the filaments (e.g., Catopsis sessiliflora). The plication of the filaments in Tillandsia subgenus Anoplophytum becomes visible during or soon after anthesis (Evans and Brown 1989a). Anthers are usually elongated, stiff or versatile, usually yellow, more rarely whitish, orange or blackish with the attachment of the filaments basifixed or dorsifixed. The connective is usually elongated into a short apical hump that in Werauhia is massive. Pollen examined as acetolyzed or fresh material with the light microscope (Erdtman and Praglowski 1974; Wanderly 1984) or dried samples examined with the scanning electron microscope (Ehler and Schill 1973; Wanderly 1984) failed to reveal important features of the apertures. Fresh pollen fixed in glutaraldehyde proved more informative (Halbritter 1988, 1992, 1995). Tillandsioid pollen grains (Fig. 12.2) are 20–60 mm long and usually sulcate (e.g., Catopsis), or exhibit a distal germination area (all other genera, except Alcantarea where the aperture margin is solid). Several Guzmania species produce inaperturate grains (Halbritter 1988, 1992). Catopsis is distinguished from all the other tillandsioid genera by its entire sulcus margin (Halbritter 1988, 1992), which resembles that of Hechtia and Puya (Pitcairnioideae). Mature pollen grains are two-celled at release, containing a fusiform generative cell and a vegetative cell with a deeply lobed nucleus. Raphides and crystals occur in some pollen grains (Kugler 1942; Halbritter 1988). Microsporogenesis has been described for Racinaea pallidoflavens (Hess 1991). None of the organelles de-differentiates during meiosis. Plastids are amoeboid, exhibit complex internal structure, and gradually begin to accumulate polysaccharides from meiotic prophase I onward. These observations contradict reports (Vijayaraghavan and Bhatia 1985) for other taxa. Mitochondria and dictyosome structure is typical, but the endoplasmic reticulum (ER) is extensive and stains poorly using standard methods. Ribosomes associate with the ER, or sometimes occur in dense clusters as cytoplasmic polyribosomes (Hess 1991). The gynoecium is superior except in Glomeropitcairnia where the pistil is Anatomy and morphology 567 Figure 12.2. Pollen grains of selected Tillandsioideae (courtesy of Heidelmarie Halritter). All scale bars510 mm. (A) Tillandsia straminea, distally and equatorially illustrating sulcus covered by operculum. (B) T. bryoides, distally illustrating pronounced reticulum of subgenus Diaphoranthema. (C) T. fasciculata, distally illustrating the diffuse sulcus with irregularly ruptured exine characteristic of subgenus Tillandsia. (D) Vriesea bituminosa (section Xiphion), distally illustrating sulcus covered by solitary exine elements. (E) Guzmania monostachia, inaperturate pollen. (F) Catopsis sessiliflora, distally illustrating simple furrow and sharply cut aperture margins. 568 Tillandsioideae about half inferior if its ovule-bearing region is considered basal. Were the septal nectaries included, all tillandsioid gynoecia would be partly inferior (Fig. 3.1A; Cecchi Fiordi and Palandri 1982; Böhme 1988; Ueno 1989). Septal nectaries lead into a horizontal channel at the inner base of the perianth (Böhme 1988; Fig. 3.1A). Nectar channels of Catopsis arise directly from above the middle of the septal nectary, while in Guzmania, Mezobromelia, Tillandsia and Vriesea they originate near the top of this glandular tissue. Copious sucrose-rich nectar (Bernardello et al. 1991) is produced by most tillandsioids. Ovules are attached in several rows along the entire length of the ovary cavity (Fig. 3.1A), or only in the basal half (Goldberg 1989). Ovules are bitegmic, crassinucellar and anatropous, and contain an embryo sac with three small antipodial cells. Endosperm is helobial (Billings 1904; Schürhoff 1926; Johri et al. 1992), at first free nuclear and later cellular (Goldberg 1989). The embryo is lateral to the endosperm or rarely embedded within it (Fig. 3.7). According to Martin (1946), the tillandsioid embryo belongs to the ‘Linear group’. The endosperm is copious, but reduced in Tillandsia usneoides (Martin 1946), and starchy (Huber 1991). Polyembryony has been observed in Tillandsia subgenus Diaphoranthema (Subils 1973; Fig. 3.6K) and subgenus Tillandsia (Suessenguth 1921). At anthesis the outer integument is shorter than the inner one, but it elongates after anthesis to form the typical tillandsioid seed appendage (coma) as pseudohairs develop by longitudinal splitting of the outer and the central tissue layer of the outer integument (Figs. 3.6J, 3.7). A chalazal appendage produced from the funicle is also present in many genera (Gross 1988a). Contrary to the above, in Catopsis multicellular strands grow from individual epidermal projections of the chalaza to form a folded coma (Gross 1988a; Palací 1997). Glomeropitcairnia seeds possess a pappiform coma at both ends (Smith and Downs 1977). Comas vary from pure white (most Tillandsia species) to ochraceous and brown (many Vriesea and Guzmania species). The asymplicate zone of the gynoecium forms the style and stigma, while the septal nectaries are formed by the symplicate zone. Styles are elongated in most cases, and vary from shorter to longer than the anthers, and when reduced remain short (e.g., Tillandsia subgenus Diaphoranthema). The three carpels are fully connate or rarely free in the apical region (at the transition into the stigma lobes). Stigma morphology (Figs. 3.1C, 12.1) is diverse within Tillandsioideae and all of the recognized types – conduplicate-spiral, convolute-blade, coralliform, cupulate and simple-erect (Brown and Gilmartin 1984, 1988, 1989b; Schill et al. 1988; Gortan 1991; Fig. 3.1C) Phytogeography and evolution 569 – are represented. Stigma lobe margins are usually papillate (e.g., ‘low to medium papillae in wet stigmas’ in Vriesea; Heslop-Harrison and Shivanna 1977), Glomeropitcairnia included (Till et al. 1997). Werauhia with cupulate stigmas lacks stigma papillae. Tillandsioid capsules are septicidal and of diverse shapes and sizes. Those of Guzmania, Mezobromelia, Racinaea and Tillandsia are usually more cylindric compared with the relatively conical fruits of Alcantarea, Vriesea and Werauhia and the ovoid Catopsis capsule. Dehiscence is apical in the semi-inferior ovaries of Glomeropitcairnia (Smith and Downs 1977). The pericarp is uniform (e.g., several Vriesea and Werauhia spp.) or divided into a stramineous exocarp and a dark castaneous, smooth and lustrous endocarp (e.g., Tillandsia). Fruits require several weeks to a year to mature. Phytogeography and evolution Tillandsioideae range more widely than either of the other two bromeliad subfamilies. Tillandsia pedicellata extends to 44° latitude in central Argentina (Smith 1934a; Till 1984), and Tillandsia usneoides occurs northward to latitude 37° in the eastern United States (Garth 1964). Distribution maps (Smith and Downs 1977) indicate that the centers of greatest diversity lie in the northern Andes and the Antilles, with several secondary centers defined by subgenera of Tillandsia in Mexico and South America (Chapter 9). Glomeropitcairnia is restricted to the Lesser Antilles and northeasternmost Venezuela, while Mezobromelia extends from the northern Andes to the Guayanas and the Greater Antilles. All of the taxa considered that have been cited as ancestral (e.g., Glomeropitcairnia, Catopsis, Tillandsia subgenera Allardtia and Pseudalcantarea; Ranker et al. 1990; Winkler 1990; Beaman and Judd 1996; Terry et al. 1997a) occur in the Antilles, which represent the remnants of the most recent land bridge between the two American subcontinents. Secondary centers of species richness are located in Central America (e.g., Werauhia) and southeastern Brazil (e.g., Alcantarea, Tillandsia subgenus Anoplophytum, part of Vriesea). The joining of the two American subcontinents to produce a Middle American land bridge during the Pliocene probably encouraged speciation, as did immigrations during Pleistocene interglacials (Smith 1962). The massive radiation of Tillandsioideae in northern Peru to Colombia probably began or was accelerated by the uplift of the Andes during the Pliocene and more recent times (Gentry 1982). Alternating episodes of warmer and wetter and colder and drier climate in this region shifted the elevations of life zones (Lauer 1986), prompting speciation, 570 Tillandsioideae particularly in montane taxa such as Guzmania and Puya (Pitcairnioideae; Fig. 9.2; Chapter 9). Amazonian refuges (Prance 1973, 1982) primarily affected lowland flora and probably had less impact on Bromeliaceae, including Tillandsioideae, which in South America is most diverse at higher elevations. The geologically ancient Guayanan highlands, while the home of many primitive pitcairnioid lineages, is lightly populated with Tillandsioideae, and its influence on evolution in this subfamily remains little understood (Chapter 9). Contrary to Pittendrigh (1948) who postulated evolution of mesophytes from xerophytic ancestors, more recent authors see the same progression headed in the opposite direction (e.g., Benzing and Renfrow 1971b,c; Benzing 1980; Benzing et al. 1985; Adams and Martin 1986c; Medina 1990; Winkler 1990; Chapter 9). Read this way, the tillandsioid precursor was mesophytic, possessed absorbing roots, broad, thin foliage, moderate-sized leaf impoundments and foliar trichomes with narrow shields. Inflorescences were probably compound and spirostichous throughout. Flowers were relatively unspecialized and equipped with a polypetalous perianth and a superior ovary. Derived lines are progressively heterochronic and xeromorphic (Benzing and Renfrow 1971b,c; Fig. 2.1); Pittendrigh (1948) considered these same traits plesiomorphic. Very likely none of the extant tillandsioids is ancestral, and extant Tillandsioideae represent intermediate or end points in a massive radiation characterized by much mosaic evolution. Guzmania is entirely mesic, but its flowers with conglutinated petals, agglutinated filaments and tendencies to autogamy appear to be substantially derived. Catopsis is distinguished from the rest of Tillandsioideae by foliar trichomes with cells arranged in the sequence 418132, unspecialized stomata that may also occur on the adaxial leaf surface (e.g., Catopsis berteroniana), sometimes imperfect flowers, or an entire sulcus margin in pollen grains (Halbritter 1988, 1992), a nectar channel that arises directly from above the middle of the septal nectary (Böhme 1988) and a folded seed coma comprised of unicellular hairs of chalazal origin (Gross 1988a; Palací 1997). Harms (1930) was sufficiently impressed by these characters to recognize Catopsis as tribe Catopsideae. However, Terry and Brown (1991) and Terry et al. (1997b) discovered that all of the tillandsioids except Glomeropitcairnia share a common set of chloroplast DNA restriction sites, a finding consistent with Ranker et al.’s (1990) belief that Glomeropitcairnia lies outside core Tillandsioideae. A more recent study using the plastid locus ndhF (Terry et al. 1997b; Chapter 9) places Glomeropitcairnia within Tillandsioideae with the notion that it quickly Phytogeography and evolution 571 diverged from the rest of the subfamily. Vriesea possesses petal scales whose development is terminal during ontogeny (Brown and Terry 1992; Chapter 3). Although most of this genus is mesic, flower morphology suggests relatively advanced status. Tillandsia exhibits the least specialized flowers, but its Type Five members possess exceptionally derived vegetative structure and function. The mesophytic phytotelm types probably more closely parallel the subfamily ancestors in overall character (Chapters 4 and 9). Brown and Gilmartin (1986, 1989a), Brown and Palací (1997) and Till (1984) reported a nearly consistent basic chromosome number of x525 for Tillandsioideae. Such uniformity offers little insight on phylogeny save for the polyploid nature of Tillandsia subgenus Diaphoranthema, a clade also marked by pronounced neoteny and extreme xerophytism (Till 1992b; Chapter 9). Cladistic analyses (Gilmartin 1983; Gilmartin and Brown 1987; Gilmartin et al. 1989; Ranker et al. 1990) have been conducted for several taxa, but with only partially satisfactory results owing to insufficient data. Glomeropitcairnia consistently occupied a position well removed from the rest of the subfamily in contrast to the results of Terry et al. (1997a,b), and xerophytism almost certainly evolved several times within the Tillandsia/Vriesea complex (Chapter 9). Gilmartin et al.’s (1989) trees place Tillandsia and Vriesea in a basal position within Tillandsioideae and support the opinion of Benzing (1980). Tillandsioideae appears to be monophyletic (Ranker et al. 1990) having perhaps originated from some pitcairnioid-like stock (Gilmartin and Brown 1987); however, Ranker et al. (1990) state that ‘Pitcairnioideae are not basal in the family’, a view supported by Terry et al. (1997a,b). 13 Tillandsia and Racinaea W. T I L L Spencer and Smith’s (1993) recognition of genus Racinaea (formerly subgenus Pseudocatopsis of Tillandsia) reflects growing appreciation that Tillandsia sensu Smith and Downs (1977), along with Vriesea, constitute a large complex of closely related species needing major taxonomic reorganization. This chapter deals with Racinaea and the remaining six subgenera of Tillandsia, many of which are likely to be redefined along with similarly paraphyletic Vriesea. Racinaea comprises 56 mainly epiphytic species; Tillandsia includes about 550 species of terrestrial, epiphytic or lithophytic herbs of highly variable architecture ranging from phytotelm forms more than 1 m in diameter (e.g., T. grandis) to dwarf, moss-like epiphytes (e.g., T. bryoides) of c. 3 cm height (Fig. 2.1). Mesophytic taxa are usually rosulate, or possess more elongate stems if saxicoles (e.g., T. australis). Type Five species (the atmospherics) are more often leafy caulescent (e.g., T. cauligera), and lack substantial interfoliar impoundments (Fig. 2.4). Phyllotaxis is spiral or rarely distichous (e.g., T. capillaris), and the leaves are lingulate (e.g., T. fendleri) to narrowly triangular (e.g., T. fasciculata) or linear (e.g., T. setacea), green or densely cinereously lepidote with centrally symmetric scales. Blades are flat and thin (e.g., R. seemannii) or succulent (e.g., T. aizoides). Scapes tend to be distinct and equipped with foliaceous bracts that may decrease in size and shape toward the top of the scape (e.g., T. polystachia), or abruptly change to vaginiform bracts (e.g., T. fuchsii). Inflorescences are usually compounded to form a panicle (e.g., T. marnier-lapostollei), a raceme (e.g., T. ixioides), a spike-raceme of distichous (e.g., T. clavigera) or polystichous spikes (e.g., T. spiraliflora), a digitate inflorescence (e.g., T. carlsoniae), a head (e.g., T. capitata), a simple distichous (e.g., T. xiphioides) or a simple polystichous spike (e.g., T. stricta) or rarely a single flower (e.g., T. albertiana; Fig. 3.3L). The inflorescence of 573 574 Tillandsia and Racinaea T. complanata and T. multicaulis is perfoliated and apparently multiple from the axils of several leaves. Adossate prophylls may be absent. Offshoots sometimes develop on the inflorescences (e.g., T. aizoides, T. denudata var. vivipara, T. flexuosa, T. latifolia, T. mima var. chiletensis, T. paucifolia subsp. schubertii, T. propagulifera, T. pyramidata, T. secunda var. vivipara, T. somnians, T. utriculata subsp. utriculata; Fig. 2.11A). Floral bracts are usually conspicuous, rarely minute (e.g., R. tetrantha var. tetrantha). Flowers tend to be short pedicellate, more rarely sessile, bisexual, usually odorless, or more rarely strongly fragrant. Raphide sacs occur in all the floral parts. Sepals are convolute, symmetric or nearly so in Tillandsia, but asymmetric in Racinaea, all free or equally joined, or just the adaxial pair highly connate. Petals are lingulate to spathulate, free, usually without basal appendages except for a group of xeric species formerly included in Vriesea section Vriesea (Grant 1993b, 1994b), usually uniformly violet, blue, yellow, white, green or red distally. Petal blades can be two-colored (e.g., T. cacticola, T. punctulata, T. purpurea, T. tectorum, and some of those xeric species previously assigned to Vriesea section Vriesea; Fig. 6.1B). Stamens are arranged in two series of equal or unequal lengths and free from the petals and from each other except where rarely connate into a filament tube (e.g., T. monadelpha, T. narthecioides). Anthers are included in or exserted from the corolla. Filaments are usually straight, rarely plicated (e.g., subgenera Allardtia and Anoplophytum in part; Fig. 6.1C) or twisted (e.g., some species of subgenus Tillandsia), flat to somewhat succulent (e.g., Racinaea) and linear to narrowly triangular. Anthers are elongated to subglobose, yellow, orange or blackish, versatile, dorsifixed or basifixed. Pollen is usually yellow to whitish, but orange in subgenera Diaphoranthema, Phytarrhiza and Racinaea. Exines are reticulate to foveolate and the aperture diffuse and of the island-type, Vriesea imperialis-type or operculumtype (Halbritter 1988, 1992). Ovaries are superior except for the septal nectaries, and ovoid to conical. Styles are usually elongated, rarely strongly abbreviated (e.g., Tillandsia subgenus Diaphoranthema, Racinaea), usually colored like the ovary or the filaments and included in or exserted from the corolla. Stigma lobes are conduplicate-spiral or simple-erect in subgenus Phytarrhiza. Coralliform lobes have been reported (Brown and Gilmartin 1989b; Figs. 3.1C, 12.1). However, it should be stressed that the kind of simple-erect stigma in subgenus Allardtia differs from that in subgenus Diaphoranthema, which more closely resembles a reduction of the coralliform type characteristic of T. narthecioides. Ovules are numerous and usually caudate. Fruits are septicidal capsules usually divided into a stramineous exocarp and a castaneous endocarp and longer in the epiphytic Evolution 575 compared with the epilithic species. Seeds are fusiform, the plumose, usually white coma is developed at the micropylar end and is straight, the stick-hairs often breaking off at the exostom to form a second ‘parachute’ (Gross 1988a; Fig. 3.6J). The chalaza is usually elongated to a tongue-like projection (except for Racinaea, mesic members of subgenus Phytarrhiza, and some members of subgenus Anoplophytum), but is not further developed during seed maturation. Evolution Schimper (1888) Mez (1896) and Smith (1934a), to mention only a few authorities, considered the mesic tillandsias (including Racinaea) ancestral to the more xerophytic types. However, Pittendrigh (1948) cited ecological traits as evidence that evolution progressed in the opposite direction, i.e., that Tillandsioideae or perhaps Bromeliaceae as a whole had a dry-land, terrestrial ancestry. Recent ecophysiological studies (Benzing and Renfrow 1971b,c; Benzing 1978a; Benzing et al. 1978; Benzing and Ott 1981; Adams and Martin 1986b,c; Winkler 1986; Medina 1990, Ranker et al. 1990; Chapter 9) support the first view. Mesomorphy pervades Tillandsia subgenera Allardtia, Phytarrhiza, Pseudalcantarea and Tillandsia, and genus Racinaea. Vegetative architecture suggests that heterochrony played a major role during the emergence of drought-tolerance in subgenera Anoplophytum, Diaphoranthema and Phytarrhiza. Studies of stigma form and development provide additional insights on evolution in Tillandsia (Brown and Gilmartin 1988, 1989b; Schill et al. 1988; Gortan 1991). Mesophytic Phytarrhiza exhibit highly specialized coralliform stigmas in most cases, and Racinaea contains several taxa with simple-erect organs that further indicate derived status. Moreover, members of Racinaea possess floral characteristics much too distinct to be basic for genus Tillandsia. The remaining three subgenera, Allardtia, Pseudalcantarea and Tillandsia, appear to be relatively closely related. Only Pseudalcantarea is exclusively mesophytic, and its flowers are equipped with simple-erect to conduplicate-spiral stigmas (Beaman and Judd 1996). Also noteworthy are its flaccid and drooping petals that attract nocturnal pollinators. The conduplicate-spiral stigma (Fig. 3.1C) occurs in all three subfamilies, and predominates in subgenera Allardtia and Tillandsia, a distribution that prompted Brown and Gilmartin (1988) to consider this morphology basic to all the others. Additionally, this stigma type seems to have evolved repeatedly in the family, perhaps to accommodate bird pollination. 576 Tillandsia and Racinaea Subgenus Tillandsia is predominantly semi- to fully xerophytic. Flowers exhibit pronounced adaptation for birds; stigmas are almost exclusively conduplicate-spiral. Members of Allardtia are mostly mesophytic, and their similarly unspecialized flowers can remain largely closed apparently to promote self-pollination. Stigmas are mainly conduplicate-spiral, but the simple-erect type occurs in at least one-third of the membership. Vegetative and floral morphology point to a single prototype for all three subgenera (Allardtia, Pseudalcantarea and Tillandsia) with mesic, but not necessarily phytotelm, foliage and unspecialized allogamous flowers. Its distribution was probably Andean. An antecedent of this description would most closely conform to subgenus Allardtia, but almost certainly the current concepts (circumscriptions) of several Tillandsia subgenera, including Allardtia, are inaccurate. No fossils can be unequivocally assigned to Tillandsia or even to Tillandsioideae, and too little is known about the paleoclimatology and forest refuges of tropical America to draw conclusions about effects on bromeliad evolution (Chapter 9). However, Graham (1997) mentions three paleophysiographic provinces which coincide with taxa of Tillandsioideae or even of Tillandsia: central Mexico through northern Central America with Tillandsia subgenus Tillandsia, southern Central America with Werauhia, and northern South America with Tillandsia subgenus Allardtia, Mezobromelia and the xeric members of Vriesea section Vriesea that Grant (1993b) reassigned to Tillandsia subgenus Tillandsia. Winkler (1986) turned to the foliar epidermis to explain the evolution of Tillandsia. Essential to his scheme was a presumed progressive decrease in the ratio of stomata to trichomes and paleogeographic data from Weyl (1964; Chapter 2). Winkler considered subgenus Allardtia the most primitive clade within Tillandsia. More recently (Winkler 1990) he substituted subgenus Pseudalcantarea as the basic taxon (a view shared in part by Beaman and Judd 1996), although reasons for Winkler’s rejection of subgenus Allardtia were simply the disjunct distribution in Central America and the Antilles and the pronounced positive ratio of stomata to trichomes that also occurs in subgenus Tillandsia. He drew the following conclusions: ‘In the grey species with narrow leaves, radiating evolution took place in Central America and in the southern Andean parts of South America. The species with broad green foliage show the strongest speciation in Ecuador, Colombia and Peru. A special center of highest species diversity is Ecuador. Evolution of species of the genus Tillandsia took place with ancestors from the Andean region probably at the beginning of the Evolution 577 Tertiary. In the outer Andean South America region, evolution is mainly influenced by refugia of forests during the Pleistocene, whereas in Central America and in the Caribbean region continental island formation since the early Tertiary is of overriding importance’. It must be emphasized, however, that epidermal characteristics even when combined with his paleogeographic interpretations seem inadequate to reconstruct the evolution of a genus as complex as Tillandsia. Cytological data discussed in Chapter 9 indicate a generic base number of n525. A few dysploids exist (Brown and Gilmartin 1989a), for example T. complanata from Costa Rica with n522, T. leiboldiana from Mexico with n519, T. scaligera from Ecuador with n52512210 fragments, and T. polystachia from Colombia with n5251B-chromosomes. Counts are still too scarce to justify phylogenetic interpretations. Polyploidy appears to be extensive in subgenus Diaphoranthema (Till 1984), and perhaps helps explain the pronounced heterochrony and the strong tendencies for autogamy and cleistogamy (e.g., T. angulosa, T. castellanii, T. erecta and T. retorta) through this assemblage (Till 1992b). Autogamy, cleistogamy and polyploidy characterize many founder populations and island floras. Autogamous, polyploid species of subgenus Diaphoranthema behave like founders in certain arid regions characterized by scattered suitable habitat (Till 1992b), and scarce pollinators. However, cleistogamic species also occur in subgenus Allardtia unaccompanied by a single demonstrated case of polyploidy (e.g., T. huarazensis, T. selleana, T. walteri). Little is known about the breeding systems and genetic structures of Tillandsia populations except for the allozymic comparison (Soltis et al. 1987; Chapter 6) of allogamous T. ionantha (subgenus Tillandsia) and autogamous T. recurvata (subgenus Diaphoranthema). Predicted patterns prevailed: fewer alleles overall and more individual heterozygosity in the outcrosser and more alleles per locus, but less variation, among near neighbors in T. recurvata. Benzing (1978a) linked the incidence of autogamy and high seed set with extreme epiphytism. Till’s (1984, 1992b) results, which demonstrated frequent autogamy and polyploidy in subgenus Diaphoranthema and the success of many of its members in colonizing extreme habitats, accord with his proposition. Clearly relationships within Tillandsia remain obscure, and very likely closely related Tillandsia and Vriesea are paraphyletic. Recent revisions are available for subgenera Phytarrhiza (Gilmartin and Brown 1986), Diaphoranthema (Till 1984, 1989a,b. 1991, 1992a), Tillandsia (Gardner 578 Tillandsia and Racinaea 1982, 1986b) and Pseudalcantarea (Beaman and Judd 1996) and for Racinaea (Spencer and Smith 1993), but not for subgenera Allardtia and Anoplophytum. Vriesea has been partly revised by Grant (1995a,b), and its xeric members of section Vriesea transferred to Tillandsia subgenus Tillandsia (Grant 1993b, 1994b). Subgeneric treatments of Tillandsia Tillandsia is currently divided into six subgenera (Smith and Downs 1977) and the recently segregated Racinaea (the former subgenus Pseudocatopsis of Tillandsia; Spencer and Smith 1993). Species nomenclature follows Grant (1993b, 1994b), Kiff (1991) and Spencer and Smith (1993). Subgenus Allardtia (,200 species) Phyllotaxis in this group is spiral except for distichous T. albertiana, and flowers are usually odorless, but intensively fragrant in T. diaguitensis, T. xiphioides and T. yuncharaensis. Sepals tend to be symmetric, or rarely asymmetric, and ovate or lanceolate, free or connate, and the adaxial members usually carinated. Petals normally exhibit hues of lilac or violet, but rarely they are white, crimson, yellow or green. Blades are usually distinct yet form a tubular corolla. Petal tips typically spread, and the corolla throat is open in allogamous forms, or rarely the petal tips are cucullate and incurved and the corolla throat is then closed in supposed autogamous forms (Fig. 3.3). Stamens are always included within the corolla, equaling the petals or slightly shorter; the anthers are yellow or cream, the pollen grains heterogeneous, 30–45 mm long, ellipsoidic and equipped with diffuse apertures such as those of the island-type or Vriesea imperialis-type. Exines are reticulate or foveolate (Halbritter 1988; Fig. 12.2). Filaments are usually straight, but plicated in T. cardenasii, T. cauligera, T. churinensis, T. cochabambae, T. latifolia, T. muhrii, T. pseudocardenasii, T. pseudomacbrideana, T. pseudomicans, T. tectorum, T. truxillana and T. zecheri (Fig. 6.1C). Styles are slender, much longer than the ovary, and included within the corolla except for T. secunda (Gilmartin 1972), which might be misplaced here. Stigma lobes are conduplicate-spiral or simple-erect (Schill et al. 1988; Brown and Gilmartin 1989b; Gortan 1991; Fig. 12.1). Ovules are usually appendaged at the chalaza, the resulting seeds usually conforming to the ‘wagneriana’ and ‘incarnata’ types, or less often to the ‘juncea’, ‘utriculata’, ‘funckiana’ or ‘cotagaitensis’ types (Gross 1988a,b). Subgeneric treatments of Tillandsia 579 Geographic distribution. Mainly Andean and Mesoamerican, Greater Antilles, and southeastern Venezuela. The transfers of T. duidae to Vriesea, of T. undulatobracteata to Guzmania, and of T. hutchisonii to Mezobromelia demonstrate that many species are imperfectly known, and without doubt additional taxa will be reassigned to other subgenera or even genera. The distinction between subgenera Allardtia and Anoplophytum is weak, and their separation may not be justified, at least not as proposed by Smith and Downs (1977). The Andean group of Anoplophytum (as classified in Smith and Downs 1977) is here included in subgenus Allardtia (e.g., T. friesii, T. lorentziana, and T. xiphioides alliances). Subgenus Anoplophytum (,60 species) Phyllotaxis is spiral and flowers are odorless. Sepals are symmetric or nearly so, free or connate and the adaxial members usually carinate. Petals are erect and form a tubular corolla that is most often blue, rose or white, rarely yellow or green. Blades are usually distinct with spreading tips. Petal blades are enlarged in T. sucrei. Stamens are included within the corolla, about equaling the petal claw; the anthers are linear, yellow or cream. Pollen grains are 25–45 mm long, subglobose and equipped with apertures of the operculum-type or island-type, but intermediate conditions are common (Fig. 12.2). Less often the apertures are diffuse and in T. geminiflora belong to the Vriesea imperialis-type. Exines are reticulate to foveolate or rarely nearly smooth (Halbritter 1988). Filaments tend to be plicate, but straight in T. candida, T. eltoniana and T. ixioides. Styles are slender and longer than the ovary and more or less included within the corolla. The stigmas are simple-erect (Schill et al. 1988; Brown and Gilmartin 1989b; Gortan 1991; Figs. 3.1C, 12.1). Ovules are obtuse or short appendaged at the chalaza. Seeds belong to the ‘cotagaitensis’ type (Gross 1988a). Geographic distribution. Eastern and southern Brazil, T. gardneri and T. tenuifolia extending into northern South America and the Antilles. Tillandsia pohliana reaches Peru, while T. ixioides, T. jucunda and T. tenuifolia extend into Bolivia, Argentina and Uruguay. Anoplophytum in Smith and Downs (1977) comprises two morphological groups. The Andean group (T. arequitae, T. argentina, T. bagua-grandensis, T. bermejoensis, T. buchlohii, T. camargoensis, T. caulescens, T. chiletensis, 580 Tillandsia and Racinaea T. colganii, T. comarapaensis, T. diaguitensis, T. didisticha, T. dorotheae, T. friesii, T. geissei, T. genseri, T. guelzii (including T. pucaraensis), T. hasei, T. koehresiana, T. lorentziana, T. lotteae, T. muhriae (including T. alberi and T. guasamayensis), T. oropezana, T. pfeufferi, T. ramellae, T. vernicosa, T. walter-richteri, T. xiphioides and T. yuncharaensis) consists of strongly xeromorphic species that almost always possess dense, distichously flowered spikes. Corollas are usually white or less often pale violet or blue-violet, rarely rose. Only T. lotteae and T. xiphioides var. lutea have yellow-green or yellow petals respectively. This group is transferred to subgenus Allardtia in the present treatment. Anoplophytum is here restricted to the Brazilian group (the second one), which contains T. aeranthos, T. araujei, T. bergeri, T. brachyphylla, T. burlemarxii, T. candida, T. carminea, T. chapeuensis, T. eltoniana, T. gardneri, T. geminiflora, T. globosa, T. grazielae, T. heubergeri, T. horstii, T. ixioides, T. jucunda, T. kautskyi, T. leonamiana, T. milagrensis, T. montana, T. neglecta, T. nuptialis, T. organensis, T. pohliana, T. polzii, T. pseudomontana, T. reclinata, T. recurvifolia, T. roseiflora, T. seideliana, T. sprengeliana, T. stricta, T. sucrei, T. tenuifolia, T. thiekenii and T. toropiensis. This assemblage contains both pronounced xeromorphic and rather mesomorphic species. Inflorescences are compounded of distichous spikes, or are polystichously flowered, and dark rose or blue petals prevail. White petals are less frequent (e.g., T. araujei) as are yellow (e.g., T. ixioides) and green corollas (e.g., T. jucunda, depending on the variety). Tillandsia esseriana and T. linearis are assigned to subgenus Phytarrhiza. The length of the stamens relative to the petal claw and the plication of the filaments have been used to separate subgenera Allardtia and Anoplophytum (Smith and Downs 1977). Filament plication, however, occurs late in ontogeny (Evans and Brown 1989a), and may be inappropriate for distinguishing subgenera. In the cases of T. candida vs. T. tenuifolia, and T. chiletensis vs. T. lorentziana, this single characteristic would assign members of each pair to different subgenera despite the obvious close relationships. Quite likely subgenera Allardtia and Anoplophytum constitute a clade (except for several misplaced species), and Anoplophytum is the morphologically more specialized of the two groups. Subgenus Phytarrhiza (37 species – 19 mesophytic, 18 xerophytic) Phyllotaxis is spiral or rarely distichous in some xerophytic species and flowers are odorless or fragrant. Sepals are symmetric or nearly so, free or shallowly connate, the adaxial ones usually ecarinate or only slightly carinate; more rarely they are distinctly carinate. Petal blades are broad, con- Subgeneric treatments of Tillandsia 581 spicuous and form a flat ‘disc’, and only the claws form a tube. Petal color is usually blue or white in the mesophytic species, but yellowish in T. triglochinoides and occasionally T. monadelpha. Blue prevails in the xerophytic taxa, the exceptions being white in T. peiranoi and T. streptocarpa, yellow in T. aurea, T. crocata, occasional T. duratii and T. streptocarpa, brown with yellow in T. humilis, or cream with purple tips in T. cacticola, T. purpurea and T. straminea. Stamens are shorter than the petal claw, equaling to exceeding the pistil; filaments are straight and usually free, but half connate in T. monadelpha. Anthers are orange at least in the xeromorphic species, and contain rather uniform, subglobose, about 20 mm-long pollen grains. Apertures are usually rather wide and of the operculum-type, although transitions occur to diffuse sulci and the island-type (Halbritter 1988; Fig. 12.2). Exines are reticulate. Styles are short and stout and deeply included within the corolla. Stigmas are conduplicate-spiral in xerophytic or coralliform in mesophytic species and in xeric T. humilis, or simple-erect in T. duratii and T. reichenbachii (Schill et al. 1988; Brown and Gilmartin 1989b; Gortan 1991; Fig. 3.1). Ovules are obtuse and slender in mesic members vs. short appendaged at the chalaza and of ovoid shape in the xeric species. Seeds belong to the ‘narthecioides’ and ‘wagneriana’ types in mesic members vs. to the ‘mauryana’, ‘incarnata’, ‘cotagaitensis’ and ‘bryoides’ types in the xeric species (Gross 1988a). Geographic distribution. A group of mesophytic taxa occurs in central Andean to northern South American ranges with a few species extending into Guatemala and Belize. The other group, largely a xerophytic assemblage, extends from the southern Andes to northeastern and southern Brazil. Subgenus Phytarrhiza is clearly distinct from the first two subgenera, but resembles subgenus Diaphoranthema particularly by floral characteristics. Distinctions between the mesic and xeric species of Phytarrhiza exceed those between xeric Phytarrhiza and Diaphoranthema. Gilmartin and Brown (1986) suggested that Phytarrhiza as currently defined is paraphyletic. Tillandsia esseriana and T. linearis, formerly assigned to subgenus Anoplophytum (Smith and Downs 1977), are better placed here. Subgenus Diaphoranthema (,30 species) Phyllotaxis is spiral or distichous and flowers are odorless or fragrant. Sepals are symmetric, free, or adaxially connate and more or less carinate. Petals are lingulate, the claws forming a tubular corolla. Blades are usually 582 Tillandsia and Racinaea narrow and inconspicuous with tips that curve or roll outward leaving the corolla throat open. Sometimes the petal tips are cucullate and the corolla throat is completely closed (e.g., T. retorta). Petal color is usually yellow to brownish, occasionally dark brown-violet to dark coffee-brown, or rarely violet. That of T. usneoides is typically green. Stamens are shorter than the petal claw, deeply included within the corolla and equal to slightly exceeding the style; filaments are straight and free. Anthers are linear and orange, the pollen grains within uniform, subglobose, and 15–20 mm long and equipped with narrow apertures that vary between the diffuse sulcus and the island-type (Halbritter 1988). Exines are pronounced reticulate with free columellae and copious pollenkitt. Styles are short, stout and bear simple-erect stigmas (Schill et al. 1988; Brown and Gilmartin 1989b; Gortan 1991). The tips of the lobes of the ‘simple-erect’ stigmas of this subgenus are often emarginate. Stigma form in Diaphoranthema seems to be reduced from the coralliform condition. Ovules are short appendaged at the chalaza to obtuse. Seeds belong to the ‘cotagaitensis’ and ‘bryoides’ types (Gross 1988a). Geographic distribution. Mainly south Andean extending to northeastern Brazil and Uruguay. Two wide-ranging species, T. recurvata and T. usneoides, range through Mesoamerica and the Antilles into the southern United States. Subgenus Diaphoranthema may be monophyletic (Gilmartin and Brown 1986), although it differs from subgenus Phytarrhiza primarily by generally reduced habits and frequent distichous phyllotaxis. The boundary between these two taxa is as weak as that between subgenera Allardtia and Anoplophytum. In fact, Diaphoranthema exhibits evidence of neotenic derivation from Phytarrhiza-like stock. Moreover, mesophytic and xerophytic Phytarrhiza may constitute natural groups (e.g., the stigma, ovule and seed types differ). If so, the second of the two should probably be united with subgenus Diaphoranthema. Subgenus Tillandsia (,200 species) Phyllotaxis is spiral, with T. pentasticha exhibiting just five orthostiches. Flowers are odorless, their sepals symmetric, free to adaxially connate, and usually adaxially carinate. Petals are naked, except in the xeric species formerly assigned to Vriesea section Vriesea (Grant 1993b, 1994b) where two basal appendages occur. Tubular actinomorphic corollas prevail with only Subgeneric treatments of Tillandsia 583 the tips of the petals curved outward if at all (yet sometimes enough to produce zygomorphy; Fig. 6.1A). Color is usually violet or lavender, rarely green, yellow, white, rose or red. Petals are rarely constricted at the height of the style base (e.g., T. fuchsii). Stamens in two series of equal or unequal lengths surpass the corolla; filaments are straight or spirally twisted, but never plicated, flat and widest near the base or slightly succulent and widest near the top. Anthers are linear to ellipsoidic and yellow to dark brown. Pollen grains are yellow, rarely cream and uniformly about 1.5–2.0 times as long (45–70 mm) as wide. Apertures are diffuse and occasionally exhibit transitions toward the island-type (Halbritter 1988). Exines are reticulate with usually different sizes of meshes, and they feature abundant free columellae. Styles are elongated and usually exceed the anthers. Stigmas are conduplicate-spiral or rarely simple-erect (e.g., T. imperialis, T. plagiotropica; Schill et al. 1988; Brown and Gilmartin 1989b; Gortan 1991). Ovules are distinctly appendaged at the chalaza. Seeds mostly belong to the ‘juncea’ type, with the ‘narthecioides’, ‘wagneriana’, ‘utriculata’, ‘funckiana’ or ‘mauryana’ types accounting for the rest of the membership (Gross 1988a). Geographic distribution. Mesoamerica to northern South America extending into the southern United States and the Antilles. A few species range into Bolivia and coastal Brazil. The xeric species formerly included in Vriesea section Vriesea are centered in the northern Andes. Gardner (1986b) divided this subgenus into five groups and Group One into eight subgroups according to floral structure (Fig. 6.1A). Although preliminary and informal, her work indicates that additional alignments are possible in some subgenera. Gardner also reassigned several species with included anthers and style to subgenus Tillandsia (e.g., Group Five with T. atroviridipetala, T. ignesiae, T. lepidosepala, T. mauryana, T. plumosa and T. tortilis) mostly from subgenus Allardtia (Smith and Downs 1977). However, stigma morphology among species of Group Five just mentioned conforms to the convolute-blade type (Figs. 3.1C, 12.1). These species are therefore provisionally treated here as a distinct group within subgenus Allardtia. Gardner’s Group Four, to which only T. didisticha and T. filifolia belong, seems intermediate between subgenera Tillandsia and Allardtia, but the diagnostic floral characteristics may be convergent. Of special interest are a number of large, broad-leafed species native to northern Peru (T. carnosa, T. ecarinata, T. ferreyrae, T. lymanii, T. platyphylla, T. rauhii, T. spiraliflora and T. teres). Stamens and styles in these cases are only slightly exserted, if at all, and inflorescences are large and 584 Tillandsia and Racinaea extensively branched. Habits are terrestrial or lithophytic and distributions occur within the region containing the highest density of Tillandsia species. Of these eight species, only T. spiraliflora had been assigned to Group Two and T. rauhii to Group Three. The remaining six are of ‘incertae sedis’ (Gardner 1986b). Vegetative morphology and floral structure suggest that these species may be similar to the ancestors of both subgenus Allardtia and Tillandsia. Subgenus Pseudalcantarea (5 species) Phyllotaxis is spiral, and no flower fragrances have been reported. Flowers are arcuate and slightly zygomorphic and completely to almost sessile. Sepals are symmetric or nearly so, slightly connate, ecarinate, or the adaxial ones are carinate in T. baliophylla. Sepals, petals, filaments and the base of the ovary fuse together to form a hypanthium (Beaman and Judd 1996). Petals are erect, but become flaccid and droop at or following anthesis (T. heterophylla), white or pale green, rarely green, and equipped with narrow blades. Stamens exceed the petals (except in T. heterophylla); the versatile anthers reach 13 mm in length. Styles exceed the petals (except in T. heterophylla), and stigmas are simple-erect to conduplicate-spiral (Brown and Gilmartin 1989b; Beaman and Judd 1996). Ovules are long (T. grandis, T. paniculata) or short (T. baliophylla, T. heterophylla and T. viridiflora) and appendaged at the chalaza. Seeds belong to the ‘wagneriana’ and ‘viridiflora’ types (Gross 1988a,b). Geographic distribution. From the Gulf region of Mexico to Nicaragua (T. grandis, T. heterophylla and T. viridiflora), and Hispaniola (T. baliophylla, T. paniculata). Subgenus Pseudalcantarea constitutes a relatively little-studied group that Winkler (1990) considered ancestral to genus Tillandsia. Smith and Downs (1977) assigned Tillandsia heterophylla to subgenus Allardtia, but the uniquely flaccid and drooping petals suggest closer relationship to subgenus Pseudalcantarea in spite of the included stamens. Beaman and Judd (1996) transferred T. grandis and T. paniculata to subgenus Tillandsia. They further concluded from cladistic analysis that T. baliophylla and T. viridiflora, which they retained in subgenus Pseudalcantarea, may represent a basal clade within Tillandsia. Racinaea 585 Racinaea (56 species) Members are mesic to semixeric herbs with spiral phyllotaxis. Inflorescences are usually bi- to quadripinnate, rarely simple bearing usually small, often fragrant flowers. Floral bracts are usually small and dull-colored, and rarely large and/or bright (e.g., R. multiflora, R. pendulispica, R. seemannii, R. tetrantha, R. undulifolia). Pedicels are strongly succulent, causing the flowers to appear sessile. Sepals are strongly asymmetric, free or nearly so, broadest near the apex, not exceeding 12 mm with the adaxial ones ecarinate to carinate. Petals are lingulate, equaling or exceeding the sepals, broadest near the base, often succulent, and with tips that often curve outward. Corollas are campanulate, white to cream or yellow, the petals conglutinated in R. tetrantha. Stamens are shorter than the petals and the often succulent, but apically thin, filaments are arranged in two series of equal length. During early anthesis, the succulent filaments are stiff and prevent the anthers from contacting the stigma. Later, the filaments begin to soften and bend inward until the anthers contact the stigma (facultative self-pollination). Anthers are ovoid and yellow or orange. Pollen grains are subglobose, 15–25 mm long, with apertures that conform to the diffuse or the operculum-type. Exines are reticulate to foveolate (Halbritter 1988; Fig. 12.2). Styles are short, surpassed by the anthers, and equipped with conduplicate-spiral or simple-erect stigmas (Schill et al. 1988; Brown and Gilmartin 1989b; Gortan 1991). Ovules are obtuse. Seeds belong to the ‘narthecioides’, ‘viridiflora’ and ‘juncea’ types (Gross 1988a,b). Geographic distribution. Mainly Andean, but extending into Mesoamerica and southeastern Venezuela. A few species occur in the Antilles and southeastern Brazil. Most species inhabit montane rain and cloud forests. Racinaea, the former subgenus Pseudocatopsis of Tillandsia, is readily distinguishable from the remaining Tillandsia subgenera by its small campanulate flowers that resemble those of Catopsis, and its strongly asymmetric sepals. These distinctions have been deemed sufficient by Spencer and Smith (1993) to warrant generic status. 14 Ethnobotany of Bromeliaceae B. B E NNE T T Beyond its numerous, valuable ornamentals, Bromeliaceae contains relatively few widely used species, pineapple and Spanish moss being the two notable exceptions. Pineapple, Ananas comosus, ranks among the most popular of the tropical fruits (Cobley 1976). Spanish moss (Tillandsia usneoides) was once an important source of low-grade ® ber in the southeastern United States, with annual production of up to 5000 tons. This wideranging species also has important medicinal uses in several regions, and an undocumented amount of material continues to sell for ¯ oral arrangements in the United States. Hortus Third (L. H. Bailey Hortorium 1976) describes nearly 250 selections distributed among 30 genera. Hybrids, some between members of different genera, substantially augment the hundreds of species in cultivation. Although ¯ owers tend to be small and ephemeral, unusual vegetative forms, ornamented leaves (e.g., Figs. 2.17B, 2.18B) and brightly pigmented ¯ oral bracts assure horticultural interest. Red, orange or yellow in¯ orescences of the many bird-pollinated bromeliads often signal from impressive distances. Sizes ranging from diminutive Spanish moss to giant Puya raimondii further entice hobbyists, and shade-tolerance suits many taxa for indoor cultivation. Frequent capacity to grow on a variety of substrates, including drift wood, cork slabs and fern roots, further enhances the popularity of bromeliads. Bromeliaceae ® gure prominently in several additional contexts including cameo appearances in Star Trek movies and Star Trek: The Next Generation. Tillandsia usneoides and other epiphytic bromeliads adorn the sets of Tarzan movies, belying the ® lms' purported African setting. Recent appearances include Medicine Man, where Sean Connery' s character ® nally identi® es Tillandsia punctulata, or rather the ants nesting within, as the source of a cancer cure. The family reaches its cultural zenith in the coastal 587 Cambridge Books Online © Cambridge University Press, 2009 588 Ethnobotany of Bromeliaceae plain of the southeastern United States. Spanish moss, together with live oaks and magnolias, is a quintessential botanical symbol of the old American South. Yet Bromeliaceae is much more than a source of pineapples, house plants and ® bers for traditional cultures, a fact I began to appreciate in the highlands of southern Peru. Bromeliads are signi® cant elements of the mostly treeless landscape above 3000 m (Bennett 1988, 1990, 1991). Resident Quechua distinguish folk taxa of Bromeliaceae, some of which provide fuel, animal forage, medicine and ritual ornamentation. Children make whistles from bromeliad leaves, and their parents decorate weddings and funerals with certain species. The discovery of folk usage and nomenclature has continued in my work with the Shuar, Quichua and Chachi in Ecuador and the Seminole in Florida. Each of these indigenous peoples has a folk taxonomy tailored for local Bromeliaceae. While not essential for survival, members of this family furnish food, ® ber and inspiration for folklore for many traditional Neotropical cultures. Below, I discuss indigenous taxonomy, nonhorticultural uses and indigenous management of Bromeliaceae based on data from my ® eld notes and publications (Bennett 1986b, 1990, 1992a, 1995, 1997a,b, Bennett et al. 1999) and other relevant literature. This review identi® es the most widely used species, and describes the range of derived products and applications. Folk taxonomy of Bromeliaceae Designations include such contrasts as attractive/ugly, friendly/hostile, edible/poisonous. Although members of modern societies make these distinctions daily, classi® cation is even more important for indigenous people, and it extends beyond simple dichotomies. The Tzeltal ethnobotanical study of Berlin et al. (1974) presents the most comprehensive analysis of bromeliad folk classi® cation. The Tzeltal ?ecÈ' refers to members of several bromeliad genera, including Aechmea, Catopsis, Pitcairnia, Tillandsia and Vriesea. The Tzeltal name for Tillandsia usneoides is ?icib, which is unrelated linguistically to ?ecÈ' . A similar pattern occurs in Huastec and Quichua nomenclature. K' ok' om is the Huastec name for Aechmea bracteata and several Tillandsia species, but k' uthay is used for T. usneoides (Alcorn 1984). Hicundo is the Quechua term for most Tillandsioideae, while qaka sunka refers to T. usneoides and other diminutive saxicoles (Bennett 1990). The Seminole in Florida differ in their classi® cation. Ashome, a generic name for epiphyte, refers speci® cally to T. usneoides. T. Cambridge Books Online © Cambridge University Press, 2009 Uses of Bromeliaceae 589 Table 14.1. Common indigenous names for the pineapple (Ananas comosus) in northwest South America Group Chachi Cofan Quichua Quechua Tikuna Shuar Common name Source chilla chiviya chihuilla chihuy chi-ná chiu Bennett, ® eld notes Borman 1976 Orr and Wrisely 1981 Soukup 1970 Schultes and Raffauf 1990 Bennett et al. 1999 fasciculata, T. utriculata and other large arboreal bromeliads are called ashome chobe, the latter term meaning large (Bennett 1997a,b). Pineapple is the most widely used bromeliad, and many of its common names in northwestern South America are related linguistically (Table 14.1). Among the indigenous names are chilla (Chachi), chiviya (Cofan), chihuilla (Quichua), chihuy (Quechua), chiu (Shuar) and chi-ná (Tikuna) (Soukup 1970; Borman 1976; Orr and Wrisely 1981; Schultes and Raffauf 1990; Bennett 1992a). Piñuela, a name derived from the Spanish word for pineapple (piña), is commonly applied to bromeliads in Central America and parts of South America. Huicunto and cognates commonly denote epiphytic and certain terrestrial bromeliads through the Andes and western Amazonia. Related names including huicundo, huiccontoi, huacontoi, huaycontoy, guicundo and hicundo (Soukup 1970; Joyal 1987; Bennett 1990) are employed as far south as Chile and Argentina. Achupalla is also used commonly, but more often for terrestrial species. The Shuar name for epiphytic and some terrestrial bromeliads is kuish (Bennett et al. 1999). Uses of Bromeliaceae Nine, nonexclusive categories of plant uses apply to Bromeliaceae (® ber, food, forage, fuel, medicine, ornamental, ritual/mythical, miscellaneous, commercial; based on Bennett et al. 1999). These categories re¯ ect local applications and perceptions, and may not always coincide with Western notions of utility. For example, indigenous people consider fruits eaten by monkeys, or trees inhabited by spirits, to be useful species. The ® ber category includes bromeliads that provide clothing, thread, rope and paper. Food and forage groups contain plants consumed by Cambridge Books Online © Cambridge University Press, 2009 590 Ethnobotany of Bromeliaceae humans and animals, respectively. This category also accounts for plants used to prepare foods and beverages. Fuel types provide material suitable for cooking and heating ® res. The medical category is self-explanatory. Indigenous and rural peoples also cultivate plants simply for their aesthetic value, whereas ritual/mythical ¯ ora have a place in shamanistic or religious ceremonies. The miscellaneous category encompasses all other applications, including those associated with hunting, ® shing, personal matters and crafts. Commercial plants are those valued outside the local community. Useful bromeliads are listed in Table 14.2. At least 90 species have nonhorticultural utility. Many possess medicinal properties, and two are poisonous. Other species yield fuel, despite the absence of woody tissues. The ornamental, ritual/mythical, food and medicine categories exceed the others in size (Table 14.3). No bromeliad provides material for construction although ® ber extracted from the foliage of several species may be used for lashing. No dye plants have been reported. Fiber At least 13 species yield useful ® ber. Indeed, it and foods are the principal products derived from Bromeliaceae. Aechmea magdalenae, Ananas comosus, Neoglaziovia variegata and Tillandsia usneoides top the list for importance. Indigenous people weave hammocks from Aechmea bracteata, A. magdalenae and Ananas comosus ® ber (Brücher 1989; Bennett 1992b). Hammocks are more comfortable than beds in the lowland tropics, and portability makes them ideal for nomadic life. Quichua women in lowland Ecuador fashion hammocks from Aechmea magdalenae ® ber by ® rst removing the spines from leaf margins. Leaves are then rubbed across the thigh or against a post to loosen the ® bers. Leaves tied to a smooth log are scraped in turn with a knife to remove all non® brous tissue prior to being soaked in water and then dried in the sun. Women fashion a strong twine by rolling three ® bers across their thighs with one hand while using the other to braid the loose ends (Fig. 14.1). Suitability for hammocks and net bags (shigras) to carry fruit, food and game is high (Fig. 14.2A). Puya chilensis leaves yield a rot-resistant ® ber employed in ® shing nets (Mabberley 1987). Bromelia laciniosa ® bers support a small industry in Brazil (Benzing 1980). Bromeliad ® bers also yield string, twine, rope and thread for sewing leather (Mabberley 1987). Philippine natives fashion a ® ne cloth from Ananas comosus ® bers. Brazilian Neoglaziovia variegata is Cambridge Books Online © Cambridge University Press, 2009 Table 14.2. Bromeliad species utilized by humans, their distribution, sources of data, and human uses Species Aechmea bracteata (Sw.) Griseb. Aechmea magdalenae (André) André ex Baker Aechmea nudicaulis (L.) Griseb. Aechmea tessmannii Harms Aechmea tillandsioides (Mart. ex Schultes f.) Baker Aechmea zebrina L.B. Smith Ananas ananassoides (Baker) L.B. Smith Ananas bracteatus (Lindley) Schultes f. in Roemer & Schultes Ananas comosus (L.) Merr. Ananas lucidus Miller Ananas paraguazensis Camargo & L.B. Smith Bromelia alsodes St. John Bromelia chrysantha Jacq. Bromelia hemisphaerica Lam. Bromelia laciniosa Martius ex Schultes in R. & S. Bromelia nidus-puellae (André) André ex Mez Bromelia pinguin L. Bromelia plumieri (E. Morren) L.B. Smith Bromelia serra Griseb. Bromelia urbaniana (Mez) L.B. Smith Catopsis hahnii Baker Catopsis morreniana Mez Catopsis sessili¯ ora (Ruíz & Pavón) Mez Catopsis subulata L.B. Smith Distributiona Referenceb Usesc MEX to COL & VEN MEX to VEN & ECU MEX ECU, COL, PER MEX to BRA 2, 15 12, 16, 21, 27 11 x 26, x FI, FO, ME CM, FI, FO, MI OR, RM FR, OR FR, OR ECU, COL BRA, ARG, PAR COL, BRA, PAR, ARG 1, x 12 12 FR, ME, OR FO FO Pantropical WI, n SA to AB COL to SUR & BRA MEX to NIC COL, VEN, TRI MEX to COS BRA 2, 3, 24, 25, 27, 29, 31 12 12 2, 36 24, 36 36 10 CM, FI, FO, ME, MI, RM FI, FO FO FO, ME, OR FO, ME FO CM, FI COL MEX, WI to GUI & ECU MEX, WI to BRA & ECU BOL, BRA, PAR, ARG PAR, ARG MEX MEX MEX, WI, to n SA & ECU MEX 16 3, 12, 16, 21, 25, 26, 36 2, 12, 22, 24, 35 21 Ð 11 11 11 11 FI, FO, MI CM, FI, FO, ME, MI FI, FO, ME FI, CM FI OR, RM OR, RM OR, RM OR, RM Cambridge Books Online © Cambridge University Press, 2009 Table 14.2. (cont.) Species Catopsis wangerinii Mez & Wercklé Greigia sodiroana Mez Greigia sphacelata (Ruíz & Pavón) Regel Guzmania acuminata L.B. Smith Guzmania eduardii André ex Mez Guzmania melinonis Regel Guzmania monostachia (L.) Rusby ex Mez Guzmania musaica (Linden & André) Mez Guzmania sanguinea (André) André ex Mez Neoglaziovia variegata (Arruda) Mez Pitcairnia angustifolia Aiton Pitcairnia breedlovei L.B. Smith Pitcairnia heterophylla Beer Pitcairnia integrifolia Ker-Gawl. Pitcairnia maidifolia (C. Morren) Decasine in Planchon Pitcairnia (Pepinia) pulchella Mez Pitcairnia pungens H.B.K. Pitcairnia spicata (Lam.) Mez Pseudananas sagenarius (Arruda) Camargo Puya chilensis Molina Puya ferruginea (Ruíz & Pavón) L.B. Smith Puya ¯ occosa (Linden) E. Morren ex Mez Puya gigas André Puya hamata L.B. Smith Puya lasiopoda L.B. Smith Distributiona Referenceb Usesc MEX to PAN ECU CHI COL, ECU COL, ECU n SA, AB to BOL ECU COL COS, COL, ECU, TRI, TOB BRA LA MEX MEX to VEN & PER VEN, TRI HON to COL & SUR 11 36 21 x 6, 8 6, 7, 8, x 6, 7, 8, x 26 26 15, 21, 27 3 11 24 3 26 OR, RM CM, FO CM, FO FR FR, OR FR, OR, MI FR, ME, OR OR OR CM, FI ME OR, RM ME ME OR ECU COL to PER MAR ECU, BOL, BRA, PAR, ARG CHI ECU, PER, BOL COS, COL, VEN & BRA COL COL, ECU, PER PER 35 36 3 12 14, 21 17 35 26 10, 35 x FR ME ME FO FI, ME FR ME OR, MI FO FR Cambridge Books Online © Cambridge University Press, 2009 Puya longistyla Mez PER Puya medica L.B. Smith PER Puya oxyantha Mez PER Puya pyramidata (R. & P.) Schultes in PER Roemer & Schultes Puya raimondii Harms BOL, PER Puya sodiroana Mez ECU Puya weberbaueri Mez PER, BOL Puya sp. PER Streptocalyx (Aechmea) longifolius (Rudge) Baker COL to BRA & BOL Tillandsia benthamiana Klotzsch ex Baker MEX Tillandsia bi¯ ora Ruíz & Pavón COS to nw SA Tillandsia capillaris Ruíz & Pavón PER, BOL, CHI, ARG Tillandsia carlsoniae L.B. Smith MEX Tillandsia chartacea L.B. Smith COL, PER Tillandsia complanata Benth. GA, COS, BOL, BRA Tillandsia dasyliriifolia L.B. Smith MEX to CA Tillandsia erubescens Schlect. MEX Tillandsia fasciculata Sw. US, MEX, CA, WI to n SA Tillandsia gilliesii Baker PER, BOL, ARG Tillandsia guatemalensis L.B. Smith MEX to CA Tillandsia incarnata H.B.K. COL, ECU Tillandsia ionochroma André ex Mez ECU, PER Tillandsia juncea (Ruíz & Pavón) Poiret MEX, GA to BOL Tillandsia lampropoda L.B. Smith MEX to COS Tillandsia maculata Ruíz & Pavón ECU, PER Tillandsia maxima Lillo & Hauman BOL, ARG Tillandsia orogenes Standley & L.O. Williams MEX to NIC Tillandsia oroyensis Mez PER Tillandsia ponderosa L.B. Smith MEX, GUA, SAL Tillandsia purpurea Ruíz & Pavón ECU, PER 31 36 x 35 FU ME FR FR x 10, 35, 36 17, x 14, 31 x 35 5, 6, 7 5, 17 11 30 35 11 20 11, 36 33 11 26 5, 7 11 11 33 10 11 17 11 18 FR, FU FO, FR, ME FR, FU, RM ME FO ME OR, RM ME, MI OR, RM FR FO OR, RM FO FR, OR, RM RM OR, RM CM, OR, RM FR, OR, MI, RM OR, RM OR, RM RM FO OR, RM FO OR, RM RM, MI Cambridge Books Online © Cambridge University Press, 2009 Table 14.2. (cont.) Species Distributiona Tillandsia recurvata (L.) L. Tillandsia rodrigueziana Mez Tillandsia rubella Baker Tillandsia schiedeana Steudel Tillandsia seleriana Mez Tillandsia sphaerocephala Baker Tillandsia streptophylla Scheid. ex Morren Tillandsia usneoides (L.) L. Tillandsia utriculata (L.) L. Tillandsia violacea Baker Tillandsia xiphioides Ker-Gawler Tillandsia spp. Vriesea werckleana Mez Referenceb US to ARG MEX to NIC ECU, PER, BOL MEX to WI, COL, VEN MEX to HON PER, BOL, ARG MEX to HON US to ARG 5, 17, 22, 24, 25, 26 11 10 2 11 5, 7 10 2, 4, 5, 7, 9, 11, 13, 14, 17, 19, 23, 25, 31 US, WI, MEX to VEN x MEX, GUA 11 BOL, BRA, PAR, URU, ARG 34 PER, ECU 14, 28, 32, x MEX to COS 11 Usesc FO, FR, ME, OR OR, RM FO ME OR, RM FR, OR, RM FR CM, FI, ME, OR, RM, MI FO OR, RM ME FU, ME, OR, RM, CM OR, RM Notes: aCountries are shown by the ® rst three letters of their name. Geographical regions are shown by the following two-letter codes: AB, Amazon Basin; CA, Central America; GA, Greater Antille; LA, Lesser Antilles; MAR, Martinique; SA, South America; SAL, Salvador; WI, West Indies. b The numbers correspond to the following references: 1Alarcón 1988; 2Alcorn 1984; 3Ayensu 1981; 4Bennett 1986b; 5Bennett 1990; 6 Bennett 1992a; 7Bennett 1995; 8Bennett et al. 1999; 9R. B. Bennett 1954; 10Benzing 1980; 11Berlin et al. 1974; 12Brücher 1989; 13Burlage 1968; 14Chávez Velásquez 1977; 15Clark 1965; 16Duke 1986; 17Franquemont et al. 1990; 18Goodspeed 1961; 19Hayward 1947; 20Laferriere et al. 1991; 21Mabberley 1987; 22McVaugh 1989; 23Moerman 1986; 24Morton 1981; 25Núñez Meléndez 1982; 26Pérez-Arbeláez 1956; 27 Purseglove 1972; 28Rowe 1963; 29Schultes and Raffauf 1990; 30Smith and Downs 1977; 31Soukup 1970; 32Steele 1964; 33Towle 1961; 34 Usher 1974; 35von Reis Altchul 1973; 36von Reis and Lipp 1982. x, unpublished notes. c CM, commercial; FI, ® ber; FO, food; FR, forage; FU, fuel; ME, medicine; MI, miscellaneous; OR, ornamental; RM, ritual/mythical. Cambridge Books Online © Cambridge University Press, 2009 Uses of Bromeliaceae 595 Table 14.3. Major use categories and the number of species found in each Use category Number of species Fiber Food Forage Fuel Medicine Ornamental Ritual/mythical Miscellaneous Commercial 13 25 21 4 25 37 30 10 10 one of the most important ® ber-producing bromeliads. Its leaves yield a commercial-grade product, called caroá, suitable for manufacturing cordage, coarse fabric, mats and reinforced paper. Each shoot bears about 30 leaves, but only two to four are suitable for processing at each harvest. Fiber content is 12± 14% (Clark 1965; Mors and Rizzini 1966; Purseglove 1972). Fibrous tissues from Tillandsia usneoides, speci® cally the heavily scleri® ed stele, once served as a horsehair substitute in upholstery and mattresses, and as packing material (Hayward 1947; R. B. Bennett 1954; Pérez-Arbeláez 1956; B. C. Bennett 1986b; Mabberley 1987). Thirty-® ve processing plants once operated in Florida alone (Jensen 1982). Preparation was crude, but inexpensive. Fresh material, usually collected in cypress swamps, was wetted, then placed in pits for 6± 8 months to allow the soft tissues to rot away. Final processing occurred off-site where the ® bers were cleaned, sorted and baled (Jensen 1982). Native North Americans weaved clothes from this rough material (Burlage 1968; Wilson 1989). Food Although Ananas comosus (pineapple) is by far the most widely utilized bromeliad, at least 25 additional species provide edible fruits, leaves or meristems. The ® rst European record of the pineapple dates from Columbus' s second voyage. In a letter describing his arrival on Guadeloupe on 4 November 1493, the explorer writes: Books Online © Cambridge University Press, 2009 596 Ethnobotany of Bromeliaceae Figure 14.1. Bromeliad uses. (A) A Quijos Quichua woman extracting ® bers from Aechmea magdalenae leaves. (B) A Quijos Quichua man combing Aechmea magdalenae ® bers and sitting on a hammock made from the plant. Books Online © Cambridge University Press, 2009 Uses of Bromeliaceae 597 Figure 14.2. Bromeliad uses (continued). (A) Shigra (cloth bag) made from Ananas comosus ® bers. (B) In¯ orescence of Aechmea magdalenae. The mature infructescence is edible. (C) Puya raimondii growing in southern Peru. The leaf bases are collected for fuel. (D) Tillandsia usneoides and other Tillandsia species (in bucket) being sold for Christmas decorations in Quito, Ecuador. Cambridge Books Online © Cambridge University Press, 2009 598 Ethnobotany of Bromeliaceae `There were some [fruits] like artichoke plants, but four times as tall, which gave a fruit in the shape of a pine cone, twice as big, which fruit is excellent, and it can be cut with a knife like a turnip and it seems to be wholesome.' (Purseglove 1972) Pineapple also impressed other European explorers. Oveido wrote in his Historia General y Naturales de las Indias, of 1535, `[t]here are no other fruits in the whole world to equal them for their beauty of appearance, delicate fragrance [and] excellent ¯ avor' (Purseglove 1972). As with other New World domesticates, Spanish explorers carried pineapples to all corners of the tropics. Today, we associate the fruit with Hawaii, but the Hawaiian pineapple, like the Irish potato, originated in South America. Hawaii did not receive its ® rst pineapples until the early 1800s. Pineapple probably originated in the Paraná± Paraguay River drainage system (Purseglove 1972). Indigenous peoples were already cultivating the plant throughout the New World tropics by the time Columbus arrived. Resemblance to a pine cone (piña in Spanish) prompted the English name pineapple. Ananas is derived from the Tupí-Guraní language, an idiom still spoken in Paraguay and southern Brazil. Ananas ananassoides, A. bracteatus, A. lucidus, A. paraguazensis and Pseudananas sagenarius, which also yield edible fruits, are all possible ancestors of the pineapple (Brücher 1989). Certain species of Aechmea, Bromelia and Greigia also bear edible fruits (Fig. 14.2B). Today, pineapple ranks among the most widely cultivated tropical fruits, growing best between 25° N and S latitude where rainfall ranges from 1000 to 1500 mm. However, crops can survive wherever annual precipitation falls between 635 and 2500 mm and frost is absent (Purseglove 1972). Fruits are often eaten fresh, but much of the commercial harvest is canned. A fermented beverage is made from pineapple fruits in Panama (Duke 1986) and throughout much of Amazonia. Fermented and nonfermented drinks are also made from Aechmea magdalenae, Bromelia alsodes, B. hemisphaerica, B. nidus-puellae, B. pinguin and B. plumieri fruits (von Reis and Lipp 1982; Alcorn 1984; Duke 1986; Mabberley 1987). Some indigenous Americans consume bromeliad leaves and meristems. The tender leaf bases of Puya hamata are eaten in salads and ground into ¯ our (von Reis Altchul 1973). A sweet drink is concocted from young in¯ orescences and soft leaf tissues in southern Ecuador (Benzing 1980). Leaf bases of P. sodiroana are edible, and Tillandsia complanata leaves are used to wrap tamales (von Reis Altchul 1973). The Pima of Mexico occasionally eat T. erubescens and T. recurvata in¯ orescences, apparently attracted by the high sugar content Cambridge Books Online © Cambridge University Press, 2009 Uses of Bromeliaceae 599 (Laferriere et al. 1991). Shoot apices of T. maxima and T. rubella are consumed in Bolivia and Argentina (Benzing 1980). Highland Quechua drink water trapped in the phytotelmata of Tillandsia oroyensis (Franquemont et al. 1990) much as the Seminole of southern Florida once used Tillandsia utriculata (Bennett, ® eld notes). Forage At least 21 species produce forage suitable for domesticated or wild animals. Monkeys eat the young in¯ orescences and drink water impounded by numerous species including Aechmea tessmannii, A. tillandsioides, A. zebrina, Guzmania acuminata, G. eduardii, G. melinonis and G. monostachia, and many forest people in turn eat these primates (Fig. 14.3A). Hunting is more successful where forest canopies host abundant phytotelm Bromeliaceae. Other animals also depend on these plants and some of their terrestrial relatives. Puya sodiroana is a `favorite food of bear' (von Reis Altchul 1973; von Reis and Lipp 1982), probably the rare Andean spectacled bear (Tremarctos ornatus). Dendrobatid frogs inhabit the tanks of some epiphytic bromeliads. Most indigenous people tip poison darts with phytotoxins (often species of Loganiaceae and Menispermaceae), but inhabitants of Colombia' s Chocó use toxic skin secretions produced by these animals to arm blowgun darts. One dendrobatid species is so potent that a single individual contains enough poison to kill 100 people (Pennisi 1992). Knowledge of which bromeliads host frogs assures the hunter a continuous supply of curare substitutes. Scores of other animals consume bromeliads given the opportunity. Mules eat Pepinia pulchella leaves (von Reis Altchul 1973), and goats consume Tillandsia recurvata shoots (Morton 1981). Native Andean people collect the foliage of Puya ferruginea, P. pyramidata, P. sodiroana and P. weberbaueri (von Reis Altchul 1973; Franquemont et al. 1990) and the seeds of P. lasiopoda, P. oxyantha and P. weberbaueri (Fig. 14.4C; Bennett, ® eld notes) to help raise domesticated guinea pigs. Cattle, sheep and llamas feed on bromeliads including Puya sodiroana, P. weberbaueri, Tillandsia bi¯ ora, T. chartacea, T. fasciculata, T. ionochroma, T. sphaerocephala and T. streptophylla among others (von Reis Altchul 1973; Smith and Downs 1977; Benzing 1980; von Reis and Lipp 1982; Bennett 1990; 1995; ® eld notes; Franquemont et al. 1990). Cambridge Books Online © Cambridge University Press, 2009 600 Ethnobotany of Bromeliaceae Figure 14.3. Bromeliad uses (continued). (A) Bromeliads are one of the preferred foods of the woolly monkey (Lagothrix lagothicha), shown here foraging for young in¯ orescences in a patch of terrestrial specimens. (B) Leaf bases of Puya raimondii collected for fuel in Peru. Cambridge Books Online © Cambridge University Press, 2009 Uses of Bromeliaceae 601 Figure 14.4. Bromeliad uses (continued). (A) Tillandsia sphaerocephala, known as aya huicunto in Quechua, is used as a funeral decoration in the highlands of Peru. (B) Tillandsia ionochroma, known as huicunto in Quechua, is used as a wedding decoration in the highlands of Peru. (C) Puya weberbaueri seeds are collected for guinea pig (cui) feed. (D) Huayruro, a young Quechua boy, with a whistle made from a Tillandsia ionochroma leaf. Cambridge Books Online © Cambridge University Press, 2009 602 Ethnobotany of Bromeliaceae Fuel Several species each in Tillandsia and Puya provide fuel in the high Andes and coastal deserts of Peru and Ecuador (Bennett, ® eld notes; Rowe 1963; Soukup 1970). Puya weberbaueri, a dominant species on valley slopes in the southern Andes, assumes exceptional importance for this reason. Leaf bases and stems burn long and hot. Villagers near Cuyo-Cuyo, Peru set ® res to enhance the growth of this terrestrial bromeliad. Like Serenoa repens in the southeastern United States, P. weberbaueri ¯ ourishes in the resulting open habitat. Burning destroys foliage and eliminates competing species, but the rhizomes survive. Fire also promotes rapid regrowth, which may be more palatable to domesticated animals. Charred leaf bases are easier to collect after their spiny leaves have been removed, and this material readily ignites. The highland Quechua of southern Peru use the leaf bases of P. raimondii in similar fashion (Figs. 14.2C, 14.3B). Medicine At least 25 bromeliad species possess purported therapeutic properties, not a surprising number considering that indigenous cultures tend to use larger portions of local ¯ oras for medicines than for any other purpose (e.g., Bennett et al. 1999). Subfamily Bromelioideae contains the largest number of medicinal species. The Huastecs of Mexico drink water drawn from the phytotelmata of Aechmea bracteata to treat fever. The same ¯ uids are employed for headaches, dizziness, eye trouble and upset stomachs caused from `eating too many chiles' (Alcorn 1984). Lowland Quichua in Ecuador allow children to inhale vapors from boiling A. zebrina leaves to relieve insomnia (Alarcón 1988). Ananas comosus has medical utility attributable to the presence of bromelain, a proteolytic enzyme similar to papain from Carica papaya. Bromelain is currently marketed by William H. Rorer, Inc. under the name Ananase to treat in¯ ammation and related pain. Topical applications also promote wound healing (Physician' s Desk Reference 1984; Hensyl 1989). Rural people apply pineapple juice for similar purposes (Núñez Meléndez 1982). A poultice of young leaves is applied to sprains (Ayensu 1981). Pineapple is also used to treat jaundice, ulcers, intestinal ailments, sore throats, urinary problems and dyspeptic ¯ atulence. Sweetened juice promotes digestion and prevents sea sickness according to some users in Venezuela. Juice expressed from the vegetative buds is ingested to treat respiratory ailments, and that from unripe pineapples supposedly works as a Cambridge Books Online © Cambridge University Press, 2009 Uses of Bromeliaceae 603 powerful abortifacient. Serotonin, a vasoconstrictor, further characterizes this fruit. Antihelmintic (vermifuge), antiscorbutic (treatment for scurvy), cholagogic (promotes ¯ ow of bile), decongestant, diaphoretic (causes perspiration), diuretic, ecbolic (smooth muscle stimulant), emmenagogic (agent that increases menstrual ¯ ow), purgative and refrigerant (relieving fever or producing cooling sensation) activities are also attributed to pineapple fruit (Soukup 1970; Ayensu 1981; Morton 1981; Alcorn 1984; Duke 1986; Schultes and Raffauf 1990; Bennett et al. 1999). Steroids from pineapple foliage possess oestrogenic activity. Boiled, young Bromelia alsodes leaves serve as a poultice to treat trauma and broken bones. Huastecs use the same plant to combat fungal infections in domesticated animals (Alcorn 1984). Basal portions of B. chrysantha leaves are used to disinfect wounds (von Reis and Lipp 1982), and a diuretic and vermifuge are prepared from its fruits (Morton 1981). Bromelia nidus-puellae berries possess antiscorbutic properties (Duke 1986). Bromelia pinguin, like pineapple, contains bromelain with effects similar to that of Carica papaya for treating oedema, in¯ ammation, fever, coughs, bronchitis, lung congestion, intestinal parasites and rheumatism (Ayensu 1981; Morton 1981; Núñez Meléndez 1982). Fruits are reported to be antiscorbutic and diuretic, and capable of causing blood to ¯ ow more freely to relieve cramps (Duke 1986). Trichomes from Bromelia plumieri leaves are applied to burns (von Reis Altchul 1973; Morton 1981). Two pitcairnioid genera, Pitcairnia and Puya, contain several medicinal species. The white foliar trichomes of Pitcairnia angustifolia and Pitcairnia spicata are mixed with honey to treat thrush and to heal cut umbilical cords (Ayensu 1981). An infusion of young Pitcairnia heterophylla leaves reportedly helps control dysentery (Morton 1981). Waxy powder (probably trichome shields) from the abaxial surfaces of Pitcairnia integrifolia leaves is rubbed on venereal lesions in males (Ayensu 1981). An infusion made from ground Pitcairnia pungens root is used for kidney and liver ailments, and the cooked mixture works as a diuretic (von Reis and Lipp 1982). A preparation from Puya chilensis is used as a hemostatic, and an extract made from ¯ owers is applied to hernias (Chávez Velásquez 1977). A leaf decoction of Puya ¯ occosa has purgative properties (von Reis Altchul 1973). Von Reis and Lipp (1982) note the treatment of pneumonia with a stem decoction from the aptly named Puya medica. Some rural Ecuadorians consume Puya sodiroana shoots to treat kidney ailments (Benzing 1980). Peruvian highlanders alleviate throat in¯ ammation, earaches, nose bleeds and swellings with juice from the ¯ owers of an undetermined Puya (Soukup 1970; Chávez Velásquez 1977). Cambridge Books Online © Cambridge University Press, 2009 604 Ethnobotany of Bromeliaceae Tillandsioideae also include important medicinal species. The Chachi of Ecuador treat earaches with juice from Guzmania monostachia in¯ orescences (Bennett, ® eld notes). An infusion of Tillandsia benthamiana mixed with alcohol is taken each morning to cure anemia or kidney troubles (von Reis Altchul 1973). Highland Quechua speakers in Peru treat coughs with a medication derived from T. capillaris (Franquemont et al. 1990). One of the more widely employed Tillandsioideae is T. recurvata; its utility includes treatments of hemorrhoids, gallbladder afflictions and menstrual irregularities (Morton 1981; Núñez Meléndez 1982). The Huastecs drink a cold water infusion of crushed T. schiedeana leaves, and apply the foliage to the head to relieve fever and headaches (Alcorn 1984). Tillandsia usneoides is the most widely distributed bromeliad, hence, not surprisingly, the species with the most varied medicinal utility. Its uses include treatments for coughs, fever, hemorrhoids, hernias, measles, mouth sores, rheumatic arthritis, sores, and ailments of the lung, liver, kidney and heart (Burlage 1968; Soukup 1970; Chávez Velásquez 1977; Núñez Meléndez 1982; Alcorn 1984; Moerman 1986). The highland Quechua of Peru treat dandruff with a rinse made from this plant (Bennett 1990). Huastec women drink a decoction made from a handful of T. usneoides as a contraceptive, but consider this practice dangerous (Alcorn 1984). A decoction of Tillandsia xiphioides ¯ owers reportedly diminishes chest pain (Usher 1974), and hot baths made from several Tillandsia species reduce discomfort associated with neuritis (Steele 1964). Ornamental and ritual/ mythical Many of the bromeliads employed in religious festivals possess ornamental and ritual value. Columbus reported that some Native Americans hung crowns of pineapple leaves over the entrances to their houses as a sign of hospitality, a tradition that continues in Europe and North America. Pineapple motifs are carved on doorways and gateposts as welcome signs (Haughton 1978). Tzeltal speakers in Mexico suspend specimens of Aechmea, Catopsis, Pitcairnia, Tillandsia and Vriesia on doorways of the homes of religious officials and alongside religious shrines in their own households (Berlin et al. 1974). Immigrants from the sierra and indigenous Quichua place Aechmea tessmannii, A. tillandsioides and A. zebrina on trees around their lowland Ecuadorian homes (Bennett, ® eld notes). The Ecuadorian Shuar do the same with Guzmania eduardii, G. melinonis and G. monostachia (Bennett et al., 1999). Huastecs decorate their ® esta sites with Bromelia alsodes (Alcorn Cambridge Books Online © Cambridge University Press, 2009 Uses of Bromeliaceae 605 1984). Méluzin (1997, 1998) describes the uses of several Catopsis and Tillandsia species by the Lenca of Honduras. She attributes utility for maize-planting rituals to the resemblance between the immature in¯ orescences and young corn stalks and to the exceptional longevity of dislodged epiphytes in dry environments such as an altar. Several Tillandsia species serve contrasting purposes in southern Peru. Highland Quechua speakers adorn wedding sites with Tillandsia bi¯ ora and T. ionochroma (Bennett 1990). Both species are called huicunto in the indigenous idiom, and display shiny red or green leaves. The Quechua place ash-colored Tillandsia sphaerocephala shoots on caskets at funerals, perhaps to signify death. Its Quechua name, aya huicunto, means death or soul bromeliad (Fig. 14.4A,B). According to the archaeological record, the funereal use of bromeliads is long-standing. Tillandsia gilliesii and T. maculata were used to wrap some Peruvian mummies (Towle 1961). Tillandsia usneoides serves as a Christmas ornament throughout Latin America (Fig. 14.2D), often in nativity scenes as the bed for the Christ ® gure (Berlin et al. 1974; Alcorn 1984; Bennett 1990, 1995; Franquemont et al. 1990). Highland Quichua speakers sell T. usneoides, T. incarnata and several other species during the Christmas season in Quito, Ecuador (Bennett, ® eld notes; Fig. 14.2D). Tzeltal speakers hang T. usneoides on doorways during celebrations (Berlin et al. 1974). Most important, perhaps, is its cultural signi® cance in the southern United States: columnist James Kilpatrick describes T. usneoides as `an indigenous, and indestructible part of the Southern character; it blurs, conceals, softens, wraps the hard limbs of hard times in a fringed shawl' (Bell and Wilson 1989). Miscellaneous Additional uses ® t none of the categories described so far. For example, Carib warriors in the West Indies produced an arrow poison from decaying pineapple fruits (Haughton 1978). Llipta, a substance obtained from the ash of many bromeliads, including dried Puya weberbaueri ¯ owers (Franquemont et al. 1990), yields the alkalinity needed to release the active constituents from chewed coca leaves (Erythroxylum coca). Living fences constructed of Aechmea magdalenae, Ananas comosus, Bromelia niduspuellae, B. plumieri and Puya gigas protect gardens from animals, and houses from marauders (Bennett, ® eld notes; Pérez-Arbeláez 1956; Morton 1981; Duke 1986). Any associated food or ® ber production constitutes a bonus. Quechua children make whistles from the leaves of Tillandsia ionochroma Cambridge Books Online © Cambridge University Press, 2009 606 Ethnobotany of Bromeliaceae (Fig. 14.4D; Bennett 1990, 1995). Huastecs use T. usneoides as nesting material in chicken houses (Franquemont et al. 1990). Latin Americans often erect a monument where someone has died along a trail or road. In coastal Peru, mourners sometimes spell the deceased person' s name with transplanted Tillandsia purpurea. Forlorn travelers used the same material to inscribe `in® ernillo' (little hell) at one remote outpost (Goodspeed 1961). The lowland Quichua of Ecuador practice the most beguiling use of a bromeliad. A tea prepared from the yellow ¯ owers of Guzmania melinonis supposedly contains a powerful aphrodisiac (Bennett, ® eld notes). Commercial At least 10 bromeliad species have commercial importance beyond horticulture (Table 14.3). Annual pineapple production exceeds 107 metric tons, more than half of which came from Thailand (FAO Production Yearbook for 1992 and 1993, cited in Simpson and Connor-Ogorzaly 1995). Local industries rely on a wider range of taxa. Tourists pay up to US$60 for quality hammocks and US$5 for shigras made from Aechmea magdalenae and Ananas comosus ® ber by indigenous people. Bromelia laciniosa, B. pinguin and B. plumieri also yield commercial ® bers, and the latter produces marketable fruit (Brücher 1989; McVaugh 1989). Greigia sodiroana fruits are sold in Ecuador. Bromelia serra provides caraguata or chaguar ® ber used to manufacture sacks, cordage and, some day perhaps, paper (Mabberley 1987). Neoglaziovia variegata forms the basis of a substantial ® ber industry in Brazil, with 10 000 to 15000 tons of ® ber annually. Known as caroá, this material is well suited for nets, and offers promise as a component of arti® cial silk (Clark 1965; Mabberley 1987). Quichua speakers in Ecuador sell Tillandsia incarnata, T. usneoides and other Tillandsia species for Christmas decorations for up to US$1 per plant (Bennett, ® eld notes). A 4-oz bag of Spanish moss retails for US$3.29, but no market remains for its ® bers in the United States. Bromeliad phytotelmata sometimes harbor fauna more valuable than the plant. John Daly discovered an anesthetic in a skin extract of the poison dart frog Epipedobates tricolor (Pennisi 1992). Epibatidine, the active constituent, operates through a novel mechanism. Alkaloids from bromeliaddwelling frogs have much medical potential (Pennisi 1992). Cambridge Books Online © Cambridge University Press, 2009 607 Indigenous management of bromeliads Table 14.4. Commonly cultivated bromeliads and those with commercial value Species Commercial value Aechmea bracteata Aechmea magdalenae Ananas bracteatus Ananas comosus Bromelia laciniosa Bromelia pinguin Bromelia plumieri Bromelia serra Greigia sodiroana Neoglaziovia variegata Tillandsia incarnata Tillandsia usneoides x x x x x x x x x x x Cultivated Usesa x x x x FI, FO, ME FI, FO, MI FO FI, FO, ME, RM, MI FI FI, FO, ME FI, FO, ME, MI FO FO FI OR, RM FI, ME, OR, RM, MI x x x x Note: aFor explanation of abbreviations, see Table 14.2. Indigenous management of bromeliads The distinction between cultivated and wild ¯ ora is often less pronounced in tropical than in temperate latitudes. Likewise, boundaries between ® elds, fallows and forests are relatively ambiguous. Useful plants, including bromeliads, may be planted, protected or collected in natural vegetation (Bennett 1992b). Ananas comosus grows only as a cultigen in ® elds and gardens, although it may persist in abandoned agricultural plots. Other commonly cultivated bromeliads are listed in Table 14.4. Aechmea bracteata (Central America) and Ananas bracteatus (Paraguay) were farmed for their edible fruits before Ananas comosus was introduced (Brücher 1989). Some unimproved genotypes, such as Aechmea tillandsioides, receive little care following transplantation. Many epiphytic ornamentals fall into this subcategory. A second management technique consists of preserving desired material as forests and ® elds are converted for human use. Higher yields are encouraged by removing competitors, a practice often employed to maintain Aechmea magdalenae in old fallows. Puya raimondii and P. weberbaueri are not sown, but they are often spared when clearing occurs. Prescribed ® res enhance plant utility as fuel in addition to eliminating weeds. Some bromeliads are neither planted, nor are serious efforts made to conserve them. Tillandsia usneoides, a common saxicole and epiphyte in parts of the Andes, is generally ignored except at Christmas. No protection seems Cambridge Books Online © Cambridge University Press, 2009 608 Ethnobotany of Bromeliaceae necessary to maintain supplies in much of South America. Overcollection for ® ber and disease eliminated some populations, prompting attempts to cultivate the survivors. Tillandsia streptophylla (Benzing 1980) and T. sphaerocephala (Bennett 1990, 1995) are dislodged from rock faces so animals may eat them; harvest of the latter species is limited. Most individuals grow on inaccessible rock faces and thus are naturally protected. Humans have expanded the ranges of many useful bromeliads. Ananas comosus occurred throughout tropical America and the Caribbean by the time Europeans arrived (Brücher 1989). Aechmea magdalenae was introduced to Trinidad in 1924 (Williams 1951). Portea petropolitana has recently naturalized in Hugh Taylor Birch State Recreation Area (Fort Lauderdale, Florida). Billbergia pyramidalis and Dyckia brevifolia have established elsewhere in the state (Wunderlin 1998). Indigenous cultures and modern societies alike use many bromeliad species, and we must consider this fact when developing conservation strategies. Further linguistic analyses and ethnobotanical study will identify additional bromeliads, and determine the efficacy of applications in traditional medicine. Cambridge Books Online © Cambridge University Press, 2009 15 Endangered Bromeliaceae M. DIMMIT T The rapid and accelerating destruction of the world’s tropical forests is widely known. Less appreciated are two additional facts: half of the estimated 300 000 species of higher plants occur in these biomes, and epiphytes make up to half of the vascular floras of certain tropical forests (Benzing 1990). Moreover, bromeliads comprise most of the biomass of arboreal vegetation at many wet montane tropical American sites. Because members of this family significantly influence important forest processes and provide substrates and other resources for much canopy-based fauna, their preservation is vital to broader conservation efforts. Terrestrial Bromeliaceae, about half of the family, sometimes dominate communities where climates are harsh (e.g., cool or hot and dry) or substrates (e.g., rock) mandate unusual plant adaptations to obtain nutrients and water (Chapters 4 and 5). Too little information is available in the literature on the sizes and genetic structures of bromeliad populations to determine if more than just a handful of taxa are truly endangered (Chapter 6), and thus the following discussion draws heavily on unpublished observations and findings on other taxa. Bromeliads tend to be locally abundant if not as diverse as co-occurring, ecologically similar flora such as Orchidaceae. Individual phorophytes or rock faces routinely support hundreds to millions of adults. One of the many populations of Tillandsia purpurea in the coastal desert of Peru (PanAmerican highway, kilometer 348 north) covered approximately six square kilometers at a density of ,50 ramets per square meter to total some 300 million ramets (Dimmitt 1989a). Ranges of species vary from exceptionally broad to as narrow as any reported for other vascular flora. Tillandsia recurvata occurs discontinuously from the southern United States to southern Argentina, an area almost as great as that occupied by the entire family. By contrast, T. albertiana, T. grazielae and T. sucrei, among other members 609 610 Endangered Bromeliaceae of this same group of specialized xerophytes, inhabit one or a few known localized sites. Greatest insularity involving the epiphytes occurs in northwest Central America and in the Andes where precipitous topography and exceptionally favorable climate promote exuberant speciation. Isolated granitic outcrops in southeastern Brazil and equally ancient substrates in the Guayanan highlands have also encouraged radiations that again reflect propensities for unusual rooting media in Bromeliaceae (Chapter 9). About 38% of the bromeliad species described in Flora Neotropica (Smith and Downs 1977) are known only from the type locality. The true figure is probably higher, because most of the several hundred species described since Smith and Downs’s monograph was published appear to be narrow endemics. Otherwise, they would have been discovered sooner. The purported rarity of some bromeliads may reflect accessibility to collectors more than true ranges. Several species first reported as highly insular eventually turned up elsewhere (e.g., Tillandsia chiapensis in Mexico and T. dexteri in Costa Rica). Despite the efforts of hundreds of amateur, commercial and scientific collectors willing to travel great distances through remote and difficult terrain, the ranges of many species remain poorly known. Factors threatening bromeliad populations Habitat destruction presents the greatest threat to endangered biota, including vulnerable Bromeliaceae. Growing demand for food and natural resources assure continued forest destruction throughout tropical America. About 45% of all tropical forests are already gone (Koopowitz 1992), and the figure for the Neotropics is ,19% (Koopowitz et al. 1994). About 38000 acres (15300 hectares) a day, or 22000 square miles (56 000 square kilometers) per year, disappear in Latin America. Annual losses are estimated to be 4.6% in Paraguay and 3.9% in Costa Rica, and rates are increasing almost everywhere (World Resources Institute 1986). More than half of the forests in Costa Rica, Ecuador, El Salvador, Honduras and Nicaragua have been replaced (Koopowitz et al. 1994). In Ecuador west of the Andes, only 0.1% of the lowland tropical rainforest remains intact (Gentry and Parker 1992). Unknown, but sizable, portions of the surviving woodlands are severely degraded (Redford 1992). Stochastic models indicate that mass extinction is well underway. More than 5000 species of orchids (20% of the family) may be recently extirpated, and about 50 more experience the same fate every year (Koopowitz 1992). Factors threatening bromeliad populations 611 Bromeliads have fared better because proportionally more New World tropical habitat remains relatively undisturbed. Assuming 15% current deforestation and an annual clearing rate of 0.7–1.0%, the model indicates that 150 species of Bromeliaceae (6% of the family) are already lost and 8–12 more members will disappear during each of the next few years (H. Koopowitz and A. Thornhill unpublished data). Habitat destruction probably accounts for a substantial portion of the bromeliads lost to date. No specimens of Tillandsia klausii have been found since its type locality in Chiapas, Mexico burned (H. Luther, personal communication). Losses short of extirpation are eroding genetic diversity in some of the more widespread species. Agriculture has displaced entire populations of Guzmania blassii and Aechmea magdalenae in Costa Rica (Skotak 1989). The northernmost population of Tillandsia exserta in Sonora, Mexico, which differs from those farther south by its dense, silvery indumentum, will soon be eliminated by resort development (personal observation). Natural events occasionally rival human activity for destructive power. That Tillandsia purpurea population described above and many others in the same Peruvian desert habitats nearly vanished during the El Niño year of 1982 (P. Isley, personal communication), probably suffocated by the several meters of precipitation that fell in this usually rainless region (Chapters 4 and 9). Seedlings were evident in some localities during 1988, however (P. Koide, personal communication). Bromeliad populations have been depleted by collection and a variety of other willful acts. Peruvians set fire to most of the inflorescences of Puya raimondii before the seed can mature. Vriesea hieroglyphica and V. fosteriana have been nearly eliminated in nature to stock urban landscapes in southeastern Brazil (Leme 1984; H. Luther, personal communication). However, heavy utilization does not necessarily deplete populations. Each year tens of thousands of Tillandsia tectorum and relatives with similar silvery foliage end up simulating snow in Christmas displays in Ecuador (H. Luther, personal communication). Despite heavy local consumption and additional collection for sale, this species remains abundant. Some authorities believe that commercial collectors are threatening numerous bromeliad populations (Leme 1984; Read 1989; Rauh 1992), but scientific confirmation is scarce. Importation records offer little help because they generally fail to identify plant material beyond family. At greatest potential risk are the xerophytic species, particularly members of Tillandsia, because these plants grow more slowly than the other desirable bromeliads. Long life cycles also increase incentive to collect 612 Endangered Bromeliaceae wild specimens for direct sale. Market demand and regulations and enforcement prevailing in the countries of origin further determine which species face high risk. Amateur as opposed to commercial collectors are too few and the hobbyist market too small to create significant demand for any but the most appealing of the insular species. Of the approximately 4000 people in the United States who belong to bromeliad societies, most grow only the relatively common tillandsias (personal observations; H. Luther, personal communication). About 200 ambitious Tillandsia collectors reside in Germany, but of these only 15–20 individuals maintain rare taxa (R. Ehlers, personal communication). A growing mass market presents greater challenge than the hobbyist. Several tillandsias have become popular as ephemeral decorations, especially in Europe where tens of millions of plants are purchased annually. About half of the flower shops and 70% of the supermarket garden centers in southwestern Germany offer one or more of these species (TRAFFIC Germany 1988). Tillandsia species are also popular in parts of the United States, for example southern California where nearly every garden store and nursery stocks them. Import volume is telling in this instance. About 18 million plants (not all Tillandsia) were imported into the United States in 1989 (estimated from TRAFFIC USA 1992; Table 15.3). Buyers in Germany and the Netherlands imported a combined 120 000 kg consisting of at least 13 million specimens in 1988 (TRAFFIC Europe, 1992; http://www.traffic.org). These figures increase nearly every year. Losses assure that the number of bromeliads collected greatly exceeds that eventually sold. Schmidt (1992) estimated that only 7% of collected Tillandsia plants remain exportable; weather, insects and unskilled collectors (the Indians who are usually hired to collect tend to mishandle the plants) eliminate the others. Many specimens also die during shipping and storage in dark warehouses. Few Tillandsia species are traded extensively. The majority of the ,550 Tillandsia species listed by Kiff (1991) and the closely related vrieseas (hundreds more) are either phytotelm types such as T. ferreyrae and T. tetrantha or less attractive nonimpounding forms (e.g., T. chaetophylla, T. schiedeana, T. narthecioides). More than 200 taxa are available in the United States, but fewer than 15 are regularly sold in lots of tens of thousands or more (P. Isley, personal communication; Table 15.1). About 130 taxa are merchandized in the United Kingdom, but just eight dominate the market (Blakesley and Powell 1992; Table 15.2). Likewise, 15 of the ,160 taxa available in Germany comprise its mass market (TRAFFIC Germany 1988). Factors threatening bromeliad populations 613 Table 15.1. Most heavily traded Tillandsia species in the United States T. aeranthos (all nursery-grown) T. bergeri (all nursery-grown) T. brachycaulos T. bulbosa T. butzii T. caput-medusae T. fuchsii (5argentea) T. ionantha T. juncea T. kolbii (5ionantha scaposa) T. magnusiana T. tectorum (? not more than 15 000 per year) T. tricolor T. xerographica Irresponsible collectors are jeopardizing the future of the occasional undistinguished, but very rare, species. Tillandsia brachyphylla, T. grazielae and several other taxa are known only from one rock outcrop in Brazil. A major fraction of the population constituting the first species was taken in 1991 for sale in Germany where most of the plants died from neglect (H. Luther, personal communication). Rock-climbing equipment was used to obtain most of the known individuals of the second species. Most of the bromeliads offered for sale originate from one of two sources. Fifty-five percent of the bromeliads imported to the United States in 1989 came from Europe, where they were presumably nursery-grown. All but 5% of the rest originated in two countries: 27% from Guatemala (4.9 million plants) and 13% from Mexico (2.3 million plants; USDA-APHIS data compiled by TRAFFIC USA 1992; Table 15.3). Seventy percent of German bromeliad imports come from Guatemala and Mexico (TRAFFIC Germany 1988). Few reports indicate the ratios of imported to propagated plants. Large nurseries in Guatemala produce an undetermined fraction of that country’s exports, but Mexico has no comparable facilities. The largest Guatemalan exporters claim that nearly all of their plants come from culture, although at least occasionally these same growers receive large shipments of wild material (W. Rauh, personal communication). Much of the Tillandsia stock sold in southern California is propagated by two major local growers. Nursery-grown Tillandsia are more attractive than collected material, but production costs oblige higher prices. 614 Endangered Bromeliaceae Table 15.2. Heavily traded Tillandsia species in the United Kingdom T. argentea (5fuchsii) T. baileyi (presumably pseudobaileyi) T. brachycaulos T. butzii T. caput-medusae T. ionantha (80% of total sales) T. ionantha var. scaposa (5kolbii) T. juncea T. oaxacana Source: From Blakesley and Powell (1992). Table 15.3. Imports of bromeliads to the United States in 1989a Exporting country Countries without wild bromeliads Mexico Guatemala Other Latin American countries Totals Plants imported (millions) Kilograms imported Est. plants (50 plants kg21) (millions)b Total plants (millions) 2.114 154 600 7.73 9.84 1.112 1.386 0.455 22948 71041 9 868 1.15 3.55 0.49 2.26 4.94 0.95 5.067 258 457 12.92 17.99 Notes: aUSDA-APHIS data compiled by TRAFFIC USA (1992) and condensed by author. bAn average weight of 20g was assumed, based on author’s measurements of the smallest and commonest (T. ionantha at 10g) and the heaviest (T. caput-medusae at 100 g) species in the trade. Studies of the impacts of collectors should focus on Guatemala and Mexico. If only 10% of all wild-collected Tillandsia specimens passed muster for export, then 72 million plants were taken in these two countries in 1989 alone, surely enough to represent sizable fractions of several relatively abundant taxa. Indeed, collectors have pushed the populations of several Guatemalan tillandsias, particularly T. xerographica, close to extinction (R. Ehlers and L. Kiff, personal communication). The capacity of bromeliad populations to tolerate sustained commercial collection remains controversial, but vulnerability surely varies among In situ conservation 615 taxa. Resilience depends on density, range and potential to regenerate impacted populations. Most of the heavily marketed species are locally abundant, wide-ranging and supposedly amenable to sustained heavy harvest (H. Luther and L. Kiff, personal communication); others disagree (Read 1989; Rauh 1992). Many dry-growing tillandsias (e.g., T. stricta, T. recurvifolia, T. streptophylla, T. bulbosa, T. caput-medusae) require on average 4–6 years to flower from seed in cultivation, and for the exceptionally slow growing types (e.g., T. xiphioides, T. xerographica) this number probably more closely approaches a full decade (Dimmitt 1984, 1990, unpublished data). Presumably, poorer nutrition and less continuous water supplies slow this process in situ. In situ conservation Expanding human numbers and activities virtually assure that all but the most abundant and widely distributed bromeliads will eventually require protection. Dim prospects for most of tropical America are apparent almost everywhere (e.g., Ecuador west of the Andes). Parks and other reserves often fail to protect resident taxa (Stuart 1992). Those Tillandsia brachyphylla plants described above were stolen from a metropolitan park in Rio de Janeiro. Losses also occur on a grander scale. For example, Costa Rica’s logging laws lapsed in the late 1980s, and for six weeks protected lands were extensively timbered (H. Luther, personal communication). Illegal logging occurred in Mexico’s Lagos de Monte Azul National Park during 1988 within view of the superintendent (D. Hadley, personal communication). Such corruption is common, but seldom reported for fear of retaliation. Government cooperation does not always guarantee security. Consider Geohintonia mexicana, a monotypic genus published in 1992. Despite immediate listing under CITES Appendix I and efforts by the government to prevent illegal collecting and export, this cactus and sympatric Aztekium hintonii were commercially available worldwide by 1996. This loss occurred even though similar events had prompted the authors to withhold the type locality from the taxonomic description. Other losses occur without obvious causes. Tillandsia hirtzii is disappearing from its fragmenting Ecuadorian habitats (H. Luther, personal communication), and drought imposed by heightened air circulation and lower humidity may explain why. Replacement of the formerly rich bromeliad flora by a much smaller number of weedy species, including cleistogamous Guzmania nicaraguensis, also suggests deterioration of the remaining 616 Endangered Bromeliaceae forest. Seedlings of mesophytic bromeliads and orchids no longer inhabit some of the isolated patches of tropical woodland in Chiapas, Mexico (R. Ehlers, personal communication) and Ecuador (J. Kent and A. Hirtz, personal communication). Continuing recruitment of xerophytic tillandsias often contraindicates depletion by collectors given the greater commercial demand for these more stress-tolerant taxa. Animals also disappear from dissected habitats at rates inversely proportional to the size of the fragments (Jones and Diamond 1976; Harris 1984). Establishment of ecological reserves by private foundations provides some basis for encouragement. The Jatun Sacha Ecological Reserve near Tena, Ecuador has acquired several thousand hectares of primary forest with funds from the debt swap program, Conservation International and the Missouri Botanical Gardens, and a loan from The Nature Conservancy. Visitors are encouraged to attend classes that emphasize the values of forests and their products. An adjacent resort for ecotourists is intended to generate funds to purchase and restore more land. The Los Cedros Project of the Centro de Investigaciones de Bosques Tropicales is acquiring remnants of upland rainforest in northwestern Ecuador. The Sierra de Alamos, a large block of deciduous tropical forest in southern Sonora, Mexico, was designated an international biosphere reserve in 1997. Protecting these private lands and public forests as development spreads and humans grow more numerous may prove more difficult than the establishment of such preserves, but success after the fact is no less crucial to preserve tropical biodiversity (Mlot 1989). Ex situ conservation Ex situ (off-site, in cultivation) compared with in situ conservation probably offers greater promise for endangered Bromeliaceae, but only if priorities change. Spectacular rescues of a few charismatic megafauna such as the California condor, peregrine falcon and Arabian oryx attract most of the public and private funds designated for endangered species. Meanwhile, tens of thousands of less glamorous taxa approach extinction or disappear unnoticed. At least 3000 Neotropical plant species have vanished since 1950, and about 100 more become extinct each year (Koopowitz et al. 1994). Even developed countries tend to be indifferent to their endangered biota. About 4400 threatened plant taxa occur in the United States, 800 of which face extinction within a decade (Center for Plant Conservation, 1991; http://www.mobot.org/CPC). Only about 300 are protected by the Ex situ conservation 617 Endangered Species Act (Federal Register, 15 March 1992), and many of these listings were obliged by a lawsuit filed by a consortium of conservation agencies that finally overcame years of political resistance. The Endangered Species Act itself, the only federal vehicle for the protection of rare life forms, is challenged more vigorously at each reauthorization. The Center for Plant Conservation (CPC) is devoted to ex situ conservation in the United States, and currently funds botanical gardens to maintain 549 rare taxa. Preservation of all listed flora will require several times this much money. Most biota reside in the developing world where governments are least able to mount effective conservation efforts. The World Conservation Monitoring Centre of the World Conservation Union (IUCN) lists 20000 species of threatened higher plants, and estimates that better information from the tropics would triple this number (Heywood 1992). The Arizona–Sonora Desert Museum has identified a score of narrowly endemic species in Sonora, Mexico that remain little known to the international conservation community. The Botanic Gardens Conservation International database of rare plants in conservatories indicates that dozens of rare tropical taxa are maintained by five or fewer institutions. Cultivated representatives of most taxa number only a few individuals (Dimmitt 1989b). Mostly North American and European addresses mitigate against large frost-sensitive collections in the majority of botanical gardens. Ex situ bromeliad conservation efforts have also been inadequate. The Marie Selby Botanical Gardens is one of a few institutions specializing in epiphytes, and its bromeliad collection is one of the three best in the world, with about 500 species comprising approximately 20% of the family. Limited space and funds oblige this institution to restrict its holdings to three to five clones for most taxa – too little genetic diversity to restore extirpated wild populations. Terrestrial bromeliads are even more poorly represented in botanic gardens and private collections. If current ex situ conservation efforts are barely adequate to protect a few thousand species, then the preservation of tens of thousands more within the next few years seems unlikely. If a third of the world’s approximately 1500 botanic gardens would accept responsibility for 60000 rare taxa, and if three replicates consisting of 10–50 genets of each taxon are needed for security, then each garden would have to maintain 360 species totaling 3600 to 18 000 specimens. Such an effort would require resources unavailable to most botanic gardens. Moreover, adequate public or private funding is unlikely given the relatively low popularity of botanical gardens among 618 Endangered Bromeliaceae public institutions. Finally, the basic strategy of most conservation programs – specifically, their emphasis on individual species – is critically flawed (Hancocks 1994). Botanic gardens have rarely been stable enough to maintain collections for more than a few decades; ex situ conservation requires commitments that simply do not exist. Less than half of the 776 gardens that responded to a global survey (D. Rae, personal communication) rated conservation among their top three priorities, and fewer than a third of the gardens in the United States are working with endangered native plants (B. Ryder, unpublished survey of 147 gardens). Collections typically change as staff turns over (Skotak 1989; personal observation). While the bromeliad collection at Marie Selby is expanding, the aroid collection has diminished to about a quarter of its former size (H. Luther, personal communication), and a once extensive collection of epiphytic Peperomia species is now represented by fewer than two dozen species. Severe overcrowding is eliminating taxa from the superb bromeliad collection assembled by Warmer Rauh at the University of Heidelberg (R. Ehlers, personal communication). Unavoidable mishaps in addition to deliberate actions take their toll on the best-maintained collections. The Arizona–Sonora Desert Museum lost at least four nonendangered taxa when an irrigation system broke down, and from disease and animal depredation despite provisions for comprehensive surveillance (Dimmitt, unpublished data). Hurricanes devastated numerous botanic gardens and private collections in Florida and Hawaii in 1992; Fairchild Tropical Garden lost 70% of its famous palm collection. Stable economies and governments exceed even the importance of effective institutional stewardship for successful plant conservation. During the Victorian era, English gardens contained over 200 species of tropical rhododendrons (section Vireya). No more than 12 survived the First World War (Adams 1981). Undoubtedly, many other cultivated plants have been lost during wars, economic depressions and other social disruptions. Even labor strikes have exacted a significant toll (Adams 1981). Recall also that botanic gardens are among the first public institutions to experience budget cuts during hard times. Economic globalization increases the chances that belt tightenings will be simultaneous and pervasive. Endangered taxa could be inexpensively preserved at low temperature, but at this point not without risk. Cryopreservation is still experimental, and few taxa have been tested after long-term storage. Genetic drift characteristic of some seed banks may reduce opportunity to maintain viable populations with this technology (Hamilton 1992). Conservation laws and their implementation 619 Ex situ conservation will also benefit from contributions from commercial growers and private collectors, some of whom already deserve recognition for major successes. Ginkgo and the dawn redwood, for example, no longer exist in nature, and we owe their continuance in cultivation to hobbyists. Similarly, a number of Somalian succulents such as Whitesloania crassa (Asclepiadaceae) and Euphorbia horwoodii (Euphorbiaceae) are probably extinct except for cultured stock obtained by collectors. Problems pollinating wild types in cultivation sometimes hinder propagation; again, it is often the dedicated amateur who develops the required techniques. Botanical gardens and professionals should actively enlist amateurs to assist conservation projects, perhaps even support collections, some of which contain better-documented and maintained material than those located at some of the most prestigious conservatories. About 200 notable collections of bromeliads exist in the United States of which perhaps two dozen are also well documented (H. Luther, personal communication). Some of their owners also value diversity, and grow many nonornamental taxa. Global treasures of such high rank could be tied into a conservation network like those being created by the National Council for the Conservation of Plants and Gardens (Great Britain) and the North American Plant Preservation Council, and provisions made for long-term care as recommended by the Bromeliad Society Code of Conduct (Dimmitt 1987). Other collectors should be encouraged to improve their documentation and join the same network. Conservation laws and their implementation CITES (Convention on International Trade in Endangered Species) is the primary tool for regulating international trade in threatened biota (see Akeroyd et al. (1994) for a concise summary). Good in theory, the administration of this treaty has created serious impediments for researchers and conservationists (McMahon 1987; Skotak 1989; see Balistrieri (1993) for a comprehensive analysis). Attempts by professionals and amateurs to obtain export permits from most Latin American countries are routinely thwarted, while unscrupulous business people easily bribe officials to export large numbers of some of the rarest plants. Legitimate permits, when obtainable, are too expensive (e.g., US$800 in Mexico in 1991) for most graduate students and many other investigators. Problems fostered by CITES and other protective regulations originate from the highest diplomatic levels down to the practices of the individual inspector. Germany and Austria proposed to list genus Tillandsia on 620 Endangered Bromeliaceae Table 15.4. Bromeliad species listed in Appendix II of CITES, 1992 Tillandsia harrisii T. kammii T. kautskyi T. mauryana T. sprengeliana T. sucrei T. xerographica Appendix II of CITES at the 1992 convention in Japan; seven taxa were eventually listed (Table 15.4). Appendix II requires the exporter to obtain permits from the country of origin testifying that his or her collection did not threaten the survival of any of the target species. Listing such a large taxon in its entirety ignores the fact that 95% of Bromeliaceae need no protection. Horticultural and scientific collection of living material must be encouraged to save endangered taxa. Impediments created by CITES and similar statutes, however unintended, surely cause losses that could be avoided by conducting salvage operations during forest clearing. CITES and many national laws should be amended to ease problems for qualified collectors. No international statute will protect plants like Vriesea hieroglyphica and the other Brazilian species cited above from overcollection or destruction. Although these bromeliads are mass propagated in nurseries elsewhere (Mercier and Kerbauy 1992), wild material is less costly to obtain in Rio de Janeiro. No less than national laws in concert with a strong conservation ethic will mitigate pressures on local endangered flora. There is much work to be done. Name index Abele, L. G. 449 Abendroth, A. 294, 418 Abercrombie, M. 372 Adams, M. 618 Adams, W. W. 57, 139, 141, 142, 154, 172, 182, 383, 499, 502, 561, 570, 575 Akeroyd, J. 619 Alarcón, R. 594, 602 Alcorn, J. B. 588, 594, 598, 602, 603, 604, 605 Alexander, C. P. 449 Allen, M. F. 210 Alves, M. A. 319, 320 Amorim de Freitas, C. 388, 389 Andrade, J. L. 146, 147, 149, 156, 157, 158, 499 Antibus, R. K. 207, 210, 211 Araujo, A. C. 272, 293, 296, 298, 415, 417 Arditti, J. 116 Arizmendi, M. C. 565 Arndt, U. 242 Arnold, F. 231 Arroyo, M. T. K. 515 Arslanian, R. L. 406, 563 Ashtakala, S. S. 517 Atallah, A. M. 406, 563 Augspurger, C. K. 281, 316, 318 Ayensu, E. S. 594, 602, 603 Aziz, T. 210 Bailey, L. H. 587 Baker, H. G. 265 Baker, J. G. 516, 517, 555 Baker, K. 123 Ball, E. 171, 173 Ballistrieri, C. A. 619 Barard, M. A. 424 Barroso, G. M. 367 Barry, D. 346, 362, 399 Barthlott, W. 562, 565 Bartholomew, D. P. 128 Bartioli, C. C. 242 Basham, Y. 212, 213 Baumert, K. 60, 151 Beach, J. H. 276 Beadle, D. 274 Beaman, R. S. 247, 256, 569, 575, 576, 578, 584 Bell, A. D. 38, 39, 41, 42 Bell, C. R. 605 Beltrano, J. 242 Bennett, B. C. 104, 245, 285, 303, 306, 308, 315, 328, 341, 345, 351, 352, 362, 487, 587, 588, 589, 590, 595, 599, 602, 603, 604, 605, 606, 607, 608 Bennett, R. B. 594 Berlin, B. 588, 594, 604, 605 Bermudes, D. 207, 210, 211, 212, 213, 241, 440, 442, 458 Bernardello, L. M. 80, 88, 246, 262, 264, 265, 267, 422, 568 Beutelspacher, C. R. 408, 564 Bhatia, K. 566 Biebl, R. 145, 149, 165, 168 Billings, F. H. 28, 60, 72, 99, 101, 107, 373, 488, 568 Black, C. C. 119 Blakesley, D. 612, 614 Böhme, S. 93, 97, 552, 568, 570 Bokermann, W. C. A. 416 Borchert, R. 560 Boresch, K. 560 Borland, A. M. 108, 124, 129 Borman, M. B. 589 Bradshaw, W. E. 223, 445 Brewbaker, J. C. 277 Brighigna, L. 49, 162, 163, 211, 229, 230, 559 Brokaw, N. V. L. 126, 245, 354, 389 657 658 Name index Brown, G. K. 63, 79, 80, 92, 93, 94, 95, 96, 97, 238, 246, 350, 262, 267, 385, 436, 488, 489, 490, 491, 492, 496, 508, 509, 512, 513, 514, 516, 522, 529, 531, 532, 549, 551, 552, 555, 565, 566, 568, 570, 571, 574, 575, 577, 578, 579, 580, 581, 582, 583, 585 Brücher, H. 590, 594, 598, 606, 607, 608 Burlage, H. M. 594, 595, 604 Burt, K. 150, 162, 163, 164, 165, 166, 232, 234 Burt-Utley, K. 370 Bush, S. P. 276 Calasans, C. F. 241 Calderon, O. 272 Caldiz, D. O. 242 Calver, F. K. 377 Calvert, A. M. 449 Calvert, P. P. 449 Capen, R. G. 107, 241 Carcuccio, F. T. 241 Carnal, N. W. 119 Catling, P. M. 295, 360, 362, 366, 367, 377, 424, 425, 426 Cave, R. D. 412 Cavelier, J. 270, 388 Cecchi Fiordi, A. 93, 568 Chapin, F. S. 235 Chávez Velásquez, N. A. 594, 603, 604 Cheadle, V. I. 48, 50, 51, 52, 162, 560, 563 Chedier, L. M. 406 Chodat, R. 559 Clark, K. L. 202 Clark, L. G. 489, 522, 523 Clark, T. F. 594, 595, 606 Clark, W. D. 522 Clarkson, D. T. 201, 202 Clegg, M. T. 522 Cobley, L. S. 587 Cockburn, W. 63 Cole, L. C. 239, 326 Collins, J. L. 123 Conacher, A. J. 214 Connor, J. J. 198, 201, 242, 243 Connor-Ogorzaly, M. 606 Cook, M. T. 374, 378 Cornelissen, J. H. C. 356 Cote, F. X. 119 Coutinho, L. M. 12, 107 Covey, S. N. 520 Coxson, D. S. 202, 236 Craighead, F. E. 321 Crayn, D. M. 539 Cronquist, A. 3 Cummins, K. W. 223, 456, 457 Curtis, G. A. 444, 445, 447 Dahle, C. E. 149, 166 Dahlgren, R. 3, 489, 522 Daly, J. 606 Damuth, J. E. 515 Davidson, D. W. 48, 194, 198, 214, 252, 289, 295, 303, 306, 347, 348, 349, 366, 425, 427, 430, 431, 432, 434, 435, 547, 550 Davidson, E. H. 410, 412 Davis, G. L. 99 Davis, J. I. 524 De Santo, A. V. 167, 230 Deiler, F. G. 145, 146, 165, 167, 168 Dejean, A. 294, 342, 423, 426, 427, 428, 429, 430, 433, 437 Delaney, K. 338 DeVries, P. J. 408, 564 Diamond, J. M. 616 Diesel, R. 450 Dimmitt, M. A. 71, 176, 609, 615, 617, 618, 619 Dodson, C. H. 311, 343, 356, 362, 484 Dolzmann, P. 163, 229, 232 Downs, R. J. xi, 3, 11, 69, 70, 74, 79, 88, 92, 96, 98, 102, 247, 275, 299, 300, 301, 463, 464, 469, 488, 493, 504, 509, 514, 516, 517, 526, 528, 529, 530, 546, 549, 555, 559, 560, 561, 562, 563, 568, 569, 573, 579, 580, 581, 583, 584, 594, 599, 610 Duke, J. A. 594, 598, 603, 605 Duvall, M. R. 489, 524 Edwards, P. J. 382, 383 Ehler, N. 230, 562, 565, 566 Ehlers, R. 250, 612, 614, 616, 618 Ekern, P. C. 113, 119, 150, 162 Ellers, B. M. 175 Epstein, W. W. 214, 289, 295, 347, 366, 425, 427, 431, 434, 435 Erdtman, G. 566 Eshbaugh, W. H. 427, 562 Evans, T. M. 63, 92, 529, 531, 566, 580 Fahn, A. 264 Fairbridge, R. W. 515 Farquhar, C. D. 160 Fetene, M. 108, 129, 130, 131, 170, 177, 228, 403 Fialho, R. F. 296, 398 Field, C. 195 Fischer, E. A. 272, 293, 296, 298, 408, 415, 417 Fish, D. 218, 219, 226, 438, 439, 440 Fittkau, E. J. 382 Fogg, G. E. 213 Fontoura, T. 12, 340, 341, 342 Fowler, N. 358, 359, 360, 368 Fragoso, C. 460, 461, 462 Name index Franco, A. C. 174 Frank, J. H. 219, 225, 412, 438, 439, 440, 442, 444, 445, 447, 448, 449, 454 Franquemont, C. 594, 599, 604, 605, 606 Freeman, C. E. 264, 265, 518 Freeze, C. H. 415 Frei, Sister J. K. 311, 343 Freiberg, M. 364 Friend, D. J. C. 129 Friedman, W. E. 4, 182, 185, 186, 456 Frölich, D. 562 Furch, K. 203 Gadgil, M. D. 327, 328 Galetto, L. 88, 422 Garcia-Franco, J. G. 286, 287, 296, 305, 340, 408 Gardner, C. S. 92, 246, 247, 248, 249, 250, 251, 252, 253, 254, 270, 276, 277, 307, 308, 504, 531, 565, 577, 583, 584 Garth, R. E. 168, 230, 569 Gentry, A. H. 271, 356, 362, 484, 514, 569, 610 Ghosh, I. 488, 491 Gilmartin, A. J. 80, 94, 95, 101, 104, 238, 250, 339, 384, 385, 469, 470, 488, 489, 490, 491, 492, 508, 509, 510, 511, 512, 513, 514, 516, 522, 528, 529, 531, 532, 539, 549, 551, 552, 555, 568, 571, 574, 575, 577, 578, 579, 581, 582, 583, 584, 585 Givnish, T. J. 43, 134, 213, 217, 218, 219, 220, 223, 231, 371, 394, 471, 472, 473, 474, 476, 477, 527, 538, 539, 540 Gladstone, D. E. 341, 362, 365 Goldberg, A. 560, 568 Goldstein, G. 147 Golley, F. B. 382 Gómez, L. D. 464 Gómez, M. A. 198 Gonzales, J. M. 220, 221 Goodspeed, T. H. 594, 606 Gorrez, D. D. 277 Gortan, G. 555, 556, 568, 575, 578, 579, 581, 582, 583, 585 Graham, A. 465, 576 Grant, J. R. 104, 255, 555, 574, 576, 578, 582 Grierson, D. 520 Griffiths, H. 108, 109, 119, 120, 121, 124, 129, 134, 149, 152, 154, 155, 169, 172, 173, 175, 198, 353, 354 Griffiths, N. M. 119 Gross, E. 73, 99, 101, 103, 288, 552, 559, 568, 570, 575, 578, 579, 581, 582, 583, 584, 585 Grubb, P. J. 316, 382, 383 Gschneidner, M. 564 659 Haberlandt, G. F. J. 12, 62, 163, 230 Hadley, G. 377, 615 Halbritter, H. 105, 555, 566, 570, 574, 578, 579, 581, 582, 583, 585 Hall, D. W. 226 Hallé, F. 19, 375, 376 Hallwachs, W. 68, 262 Hamilton, M. B. 618 Hamrick, J. L. 245, 246, 283, 284, 507 Hancocks, D. 618 Hanken, J. 500, 501, 505, 513 Harms, H. 46, 48, 516, 517, 560, 561, 562, 570 Harris, F. S. 136 Harris, J. A. 107, 168 Harris, L. D. 616 Haughton, C. S. 604, 605 Hay, J. D. 397, 398 Hayward, W. 594, 595 Hazen, W. E. 360 Hegnauer, R. 268, 406, 562, 563, 565 Hensyl, W. R. 602 Heslop-Harrison, Y. 569 Hess, M. 566 Heywood, V. H. 617 Hickey, L. J. 3, 4, 19 Hietz, P. 315, 340, 343, 345, 362, 367 Heitz-Siefert, U. 340, 345, 362, 365, 367 Hirtz, A. 616 Hoehne, F. C. 439 Hofstede, G. M. 207 Holcomb, G. E. 242, 339, 377 Holldobler, B. 214 Holst, B. K. 12, 75, 472 Holzapel, C. M. 445 Horres, R. 561 Huber, H. 489, 522, 563, 568 Huxley, C. R. 214, 215, 427 Ibisch, P. L. 341, 364, 365, 368, 469, 483, 484, 486, 487 Ihlenfeldt, H. D. 514 Isley, P. 611 Izquierdo, L. Y. 283 Jaffe, K. 220 Janetzsky, W. 217, 440, 451, 453, 454 Janos, D. P. 210 Janzen, D. H. 433 Jaramillo, M. A. 270, 388 Jebb, M. 427, 428, Jenkins, D. W. 335 Jensen, A. S. 595 Joel, D. M. 222 Johansson, D. R. 361, 374, 377 Johow, F. 246, 261 Johri, B. M. 568 660 Name index Jones, H. L. 616 Jordan, C. F. 379 Joyal, E. 589 Judd, W. S. 247, 256, 569, 575, 576, 578, 584 Junk, W. J. 203 Kaiser, W. M. 148 Kaplan, M. A. C. 406 Kato, M. 4 Kellman, M. 204 Kelly, D. L. 357 Kent, J. 616 Kerbauy, G. B. 620 Kernan, C. 358, 359, 360, 368 Kethley, J. B. 223 Kiew, R. 340, 462 Kiff, L. F. 555, 578, 612, 614, 615 Kilpatrick, J. 605 Klein, R. M. 335, 357, 440 Kleinfeldt, S. E. 425 Klinge, H. 382 Kluge, M. 143, 144 Knab, F. 449 Knauft, R. L. 116 Knudsen, J. T. 268 Kohlmann, B. 459 Koide, P. 611 Koniger, M. 128, 177 Koopowitz, H. 610, 611, 616 Koptur, S. 88, 214, 424 Krauss, B. H. 46, 48, 49, 54, 58, 60, 62, 70, 148, 150 Krauss, J. F. 413 Kress, W. J. 253, 282, 479, 517 Krügel, P. 417, 446, 451, 458 Kubisch, F. 270, 271 Kugler, H. 566 Kuntze, C. E. O. 545, 546 Lacerda, L. D. 397, 398 Laessle, A. M. 133, 225, 226, 384, 438, 439, 440, 442, 444, 449, 451, 452 Laferriere, J. E. 594, 599 Landolt, J. C. 211 Lange, O. L. 172 Larcher, W. 383 Lauer, W. 569 Lavelle, P. 459 Lee, D. W. 54, 67, 134, 174, 185, 323, 398, 401, 402, 403 Lefkovitch, L. P. 360, 366 Lehman, P. S. 410 Leme, E. 263, 286, 335, 392, 414, 416, 465, 478, 537, 545, 546, 547, 548, 549, 611 Lesica, P. 207, 210, 211 Levey, D. J. 289, 291 Lindman, C. A. M. 545, 546 Lindsbauer, K. 60, 62 Lindschau, M. 488 Lineham, T. U. 412 Linnerooth, W. 321 Lipp, F. J. 594, 598, 599, 603 Lobry De Bruyn, L. A. 214 Loeschen, V. S. 122, 148, 173 Long, S. P. 179, 228 Longino, J. T. 215, 562 Loope, L. 322 Lopez, L. C. S. 442, 444 Lounibos, L. P. 438, 439, 445, 449 Lowman, M. 321, 406 Lugo, A. E. 459 Luther, H. xi, 14, 54, 245, 247, 256, 260, 346, 356, 396, 424, 516, 545, 546, 549, 555, 611, 612, 613, 615, 618 Lüttge, U. 108, 115, 131, 132, 149, 152, 170, 171, 173, 174, 175, 197, 198, 353, 403 Lydon, J. 129 Lyra, L. T. 440 Mabberley, D. J. 590, 594, 598, 606 Mabry, T. J. 563 MacArthur, R. H. 438 McKey, D. 290, 291, 431 McMahon, L. 619 McVaugh, R. 594, 606 McWilliams, E. L. 52, 98, 104, 108, 133, 136, 277, 316, 338, 449, 483, 492 Maddison, D. R. 535 Maddison, W. P. 535 Madeira, J. A. 443 Madison, M. 98, 102, 213, 276, 295, 347, 422, 424, 435 Maguire, B. 438, 439, 442 Malm, O. 241 Mann, K. 563 Marchant, C. 448, 547 Marigo, L. C. 286, 335, 392, 414, 416, 465 Marks, P. L. 349 Marlatt, R. 413 Martin, A. C. 568 Martin, C. A. 144, 145, 160, 170 Martin, C. E. 11, 57, 62, 107, 109, 114, 115, 116, 118, 120, 124, 130, 134, 136, 139, 141, 142, 143, 144, 149, 154, 162, 166, 167, 169, 172, 175, 182, 230, 351, 354, 383, 499, 502, 561, 562, 570, 575 Martin, G. 68 Martinelli, G. 265, 266, 267, 271, 272, 273, 276, 277, 278 Martinez, J. D. 241, 243 Martinez del Rio, C. 289 Maschwitz, J. 214 Mason, L. 413 Mastalerz, J. 273 Name index Mateson, T. J. 208, 380, 388, 415, 416 Maxwell, C. 109, 116, 119, 136, 137, 140, 159, 168, 169, 176, 177, 179, 180, 181, 183, 228, 354 Mayo, S. J. 367 Means, D. B. 449 Medina, E. 67, 68, 108, 109, 118, 123, 124, 125, 126, 128, 129, 131, 136, 156, 172, 176, 177, 228, 354, 385, 403, 404, 494, 496, 497, 498, 499, 500, 552, 570, 575 Meirelles, S. T. 57, 139, 141, 145, 499, 502 Melchior, H. 565 Méluzin, S. 605 Mercier, H. 237, 238, 620 Meyer, L. 559, 560 Mez, C. 12, 60, 76, 107, 163, 230, 246, 250, 307, 516, 517, 546, 549, 551, 555, 561, 575 Midgiey, J. J. 208, 209 Miller, A. C. 442 Miller, G. A. 331, 332, 333, 334 Mitchell, P. 267 Mlot, C. 616 Moerman, D. E. 594, 604 Moffat, A. S 202 Montaña, C. 378 Mooney, H. A. 195 Morren, E. 501 Mors, W. B. 595 Morton, J. F. 594, 599, 603, 604, 605 Murawski, D. A. 245, 246, 283, 284, 507 Murillo, R. 461 Nadkarni, N. M. 162, 201, 202, 207, 208, 232, 236, 380, 382, 383, 415, 416, 459 Naeem, S. 442, 456 Napp-Zinn, K. 564 Neales, T. F. 124, 128, 172 Nelson, E. C. 355, 465 Netolitzky, F. 102 Newman, E. I. 201 Newsham, K. K. 209 Nicholas, H. J. 406, 563 Nievola, C. C. 237 Nobel, G. K. 417 Nobel, P. S. 127, 147, 148, 163, 171, 228, 401 Nose, A. 124, 129 Núnez Meléñdez, E. 594, 602, 603, 604 Nyman, L. P. 233 Oberbauer, F. S. 322 Okahara, K. 223 Oliveira, M. G. N. 440, 441, 442 Oliver, W. R. B. 478 Olmsted, I. C. 294, 342, 387, 427, 428, 429, 430, 433, 437 O’Meara, G. F. 219, 444 661 Oppenheimer, J. R. 415 Orivel, J. 434 Ornelas, J. F. 565 Orr, C. 589 Ortiz-Crespo, F. I. 260 Ortlieb, U. 494 Ott, D. 62, 575 Oveido, 598 Owen, T. P. 20, 163, 230, 232, 233, 235 Palací, C. A. 79, 102, 278, 280, 491, 492, 507, 508, 529, 568, 570, 571 Palacio-Vargas, J. G. 460 Palandri, M. R. 93, 568 Paoletti, M. G. 14, 208, 210, 217, 411, 438, 439, 440, 449, 456, 457, 459, 460 Parker, T. 610 Paroz, P. R. 408 Peixoto, O. L. 418 Penfound, W. T. 144, 146, 165, 166, 167, 168 Pennisi, E. 599, 606 Percival, M. S. 264, 265 Pérez-Albeláez, E. 594, 595, 605 Peters, C. M. 62, 340, 562 Pfitsch, W. A. 108, 124, 126, 127, 354, 403 Picado, C. 14, 46, 166, 218, 235, 438, 439, 440 Pittendrigh, C. S. 12, 111, 113, 124, 134, 151, 210, 227, 235, 352, 353, 354, 355, 360, 384, 385, 400, 424, 440, 493, 494, 496, 497, 498, 532, 533, 534, 547, 570, 575 Pizo, M. A. 416 Plummer, G. L. 223 Poisson, M. J. 102 Powell, D. 612, 614 Praglowski, J. 566 Prance, G. T. 514, 570 Pridgeon, A. 165, 230, 234 Primack, R. B. 162, 232 Privat, F. 440 Puente, M. 212, 213 Purseglove, J. W. 594, 595, 598 Raack, J. 273 Rabatin, S. C. 210 Rae, D. 618 Raffauf, R. F. 589, 594, 603 Ramírez, I. 280, 281, 478, 491, 545, 546, 547, 549, 551, 552 Ranker, T. A. 3, 521, 525, 526, 569, 570, 571, 575 Rasmussen, F. N. 489 Rauh, W. 93, 280, 559, 560, 564, 565, 611, 613, 615, 618 Raven, J. A. 123, 240 Read, M. 611, 615 Read, R. W. 449 662 Name index Redford, K. H. 610 Rees, W. E. 218 Regel, A. von 545 Reinert, F. 57, 119, 139, 141, 145, 499, 502 Reitz, P. R. 355, 357, 444 Reitz, R. 440 Remsen, J. V. 416 Renfrow, A. 47, 67, 109, 134, 141, 150, 151, 188, 193, 194, 198, 203, 204, 205, 216, 225, 233, 242, 303, 349, 350, 353, 432, 494, 496, 497, 532, 570, 575 Richter, P. 417 Rico-Gray, V. 286, 287, 305, 408, 428 Rivero, J. A. 417, 424 Rizzini, C. T. 595 Robins, R. J. 438 Robinson, H. 70 Robinson, J. W. 241 Rocha, C. F. D. 319 Roe, N. A. 218 Röhweder, O. 103 Rojas-Fernández, P. 460, 461, 462 Rossi, M. R. 480, 482 Rowe, J. H. 602 Rudolf, D. 343 Ruess, B. R. 175 Ruinen, J. 374 Ruschi, A. 246 Ryder, B. 618 Sakai, W. S. 74, 163, 227, 231, 400, 493 Sale, P. J. M. 124, 172 Sanford, W. G. 74, 163, 227, 231, 400, 493 Sanderson, J. 163 Sazima, I. 246, 258, 267, 268 Sazima, M. 256 Scarano, F. R. 367, 388, 389 Scatina, F. N. 459 Schaffer, W. M. 327, 328 Schill, R. 566, 568, 575, 578, 579, 581, 582, 583, 585 Schimper, A. F. W. 12, 111, 351, 493, 497, 532, 559, 575 Schindler, R. 560 Schlesinger, W. J. 349 Schmid, M. J. 464 Schmid, R. 464 Schmidt, A. K. 144 Schmidt, C. 612 Schmidt, J. 148 Schmidt, R. 93 Schmitt, A. K. 166, 167, 230 Schrimpff, E. 241 Schroeder, H. A. 243 Schubart, C. D. 450 Schuh, M. 450 Schulte, P. L. 148 Schultes, R. E. 589, 594, 603 Schulz, E. 137, 499, 501, 502 Schulze, E. D. 208 Schürhoff, P. N. 568 Scogin, R. 264, 265, 518 Seemann, J. 207, 374, 378, 380, 381, 399 Seidel, J. L. 289, 295, 347, 435, 550 Sengupta, B. 211 Schacklette, H. T. 198, 201, 242, 243 Sharkey, T. D. 160 Sharma, A. K. 488, 491 Sheline, J. R. 243 Shivanna, K. R. 569 Shubert, T. S. 242 Sideris, C. P. 148, 150 Sieber, J. 113, 162, 232 Siedow, J. N. 118, 144, 145, 160 Sieff, E. xi, 14, 247, 516, 545, 555 Silander, J. A. 331 Sillett, T. S. 415 Simpson, B. B. 606 Simpson, M. G. 524 Skotak, C. 611, 618, 619 Smith, A. P. 126, 127 Smith, J. A. C. 108, 109, 124, 129, 131, 134, 135, 138, 151, 152, 154, 155, 157, 168, 353, 354, 494, 495, 497 Smith, L. B. xi, 3, 11, 70, 74, 79, 92, 96, 98, 102, 247, 278, 403, 463, 464, 466, 467, 469, 479, 481, 482, 488, 493, 504, 505, 509, 514, 516, 517, 522, 525, 528, 529, 530, 546, 555, 561, 568, 569, 573, 575, 578, 579, 580, 581, 583, 584, 585, 594, 599, 610 Snow, B. K. 255, 290 Snow, D. W. 255, 290 Soltis, D. E. 253, 281, 507, 577 Soukup, J. 589, 594, 602, 603, 604 Spencer, M. A. 247, 479, 517, 555, 573, 578, 585 Stearns, S. C. 326 Steele, A. R. 594, 604 Stephenson, S. C. 211 Sternberg, L. 125 Stewart, G. R. 208, 209 Steyermark, J. A. 466 Stiles, E. W. 289 Stiles, F. G. 246, 265, 271, 272, 276, 299, 416 Stiles, K. C. 149, 169, 292 Stock, W. D. 208, 209 Stotz, D. F. 254 Strehl, T. 70, 559, 562 Strehl, V. T. 53, 242 Stuart, G. E. 615 Subils, R. 568 Suessenguth, K. 568 Sugden, A. M. 369, 370, 438 Name index Sun, G. 3 Szidat, L. 99, 102, 103 Tanner, E. V. J. 382, 383 Taylor, D. W. 3, 4, 19, 149 Ter Steege, H. 356 Terry, R. G. 3, 79, 80, 96, 97, 463, 466, 496, 516, 522, 527, 528, 529, 531, 538, 540, 552, 569, 570, 571 Thien, L. B. 428 Thomas, V. 159, 412, 458 Thompson, J. N. 217, 222 Thomson, W. W. 163, 232, 233 Thorne, B. L. 436 Thornhill, A. 611 Tietze, M. 111, 493, 496, 497, 498 Till, W. 250, 491, 492, 496, 505, 506, 555, 564, 565, 569, 571, 573, 577 Ting, I. P. 115 Todzia, C. 346 Tollsten, L. 268 Tomlinson, P. B. 20, 24, 25, 26, 29, 32, 37, 39, 41, 42, 59, 60, 61, 62, 63, 145, 162, 238, 499, 501, 506, 559, 560, 561, 562, 563 Towle, M. A. 594, 605 Troughton, S. 109, 156 Tukey, H. B. 348 Turner, R. M. 316 Ueno, C. 97, 568 Ule, E. 295, 347, 435, 546, 550 Ulubelen, A. 563 Usher, G. 594, 604 Utley, J. F. 255, 267, 371, 530, 531, 555 Valdivia, P. E. 340 Van Sluys, M. 254 Vance, E. D. 207, 208, 236 Varadarajan, G. S. 68, 69, 93, 95, 97, 101, 104, 246, 262, 267, 400, 467, 468, 469, 470, 526, 539 Vareschi, E. 217, 440, 451, 453, 454 Vasak, V. 299 663 Veloso, H. P. 355, 357, 361 Vijayaraghavan, M. R. 566 Visher, W. 559 Vogel, S. 246, 253, 255, 565 Vogelman, T. C. 68 von Reis, S. 594, 598, 599, 603 von Reis Altchul, S. 594, 598, 599, 603, 604 Wake, D. B. 420, 500, 501, 505, 513 Walker, L. R. 321, 323 Wanderly, M. 566 Weir, J. S. 340, 426 Weiss, H. E. 488 Weyl, R. 576 Wheeler, W. M. 217, 218, 425, 430 Wherry, E. T. 107, 241 Wiley, E. O. 481 Williams, C. A. 406, 518, 564 Williams, R. O. 608 Wilson, E. O. 425, 438, 595, 605 Winchester, J. W. 243 Winkler, S. 70, 198, 479, 494, 559, 562, 569, 570, 575, 576, 584 Wittmack, L. 555, 559 Wolf, K. H. 524 Wollenweber, E. 563 Wright, J. S. 272 Wrisley, B. 589 Wülfinghoff, R. 270 Wunderlin, R. P. 608 Wurthmann, E. 335 Yeaton, R. I. 341, 362, 365 Young, A. M. 418 Yu, W. 432, 433 Zahl, P. A. 438 Zavortink, T. J. 440 Ziereis, H. 231 Zimmerman, J. K. 342, 387 Zizka, G. 355, 465 Zoller, W. H. 243 Zotz, G. 141, 146, 149, 156, 157, 158, 159, 356, 359, 458, 499 Subject index Abscission (see Foliage; Water relations) Acetylene 275 Africa 1, 331, 361, 465 Air quality monitors 61, 107, 187–188, 192, 201, 231, 240–244, 459 Air pollution (see also Global change) 188, 413 Alajuela Province 464 Algae 439–442, 444–445 Allelopathy 343, 349, 377 Allozymes 281–285, 338, 577 Alpine bromeliads (see also Andes) 6–8, 11, 41, 45, 73, 110, 131, 147–148, 331–334, 390, 399–400, 595, 600–601 Amazonia 241, 339, 347, 427, 435, 440, 446, 451, 458, 471, 479–480, 483, 485, 487, 515, 545–547, 598 Ancestral habitats (see also Heterochrony) 228, 493–500 Andes 6, 41, 43, 54, 218, 256, 301, 316, 319, 339, 357, 393, 395, 399–400, 466–471, 480–486, 505, 548, 550, 569, 577, 581, 585, 599, 602, 607, 610, 615 Animal-assisted saprophytism (see also Mineral nutrition; Phytotelmata) 53, 184, 199, 223–227, 235, 438 Antarctica 515 Antilles 569, 582–583, 585 Anthocyanins 254 attract fauna 86, 89 enhance carbon gain 185–186 protect tank fauna 54–55, 413, 454–456 sunscreen 60, 134, 182, 404 Ants (see also Myrmecodomatium) 413, 421–435 ant gardens 86, 88, 53–54, 102, 113, 187, 189, 191, 199–200, 205, 208, 213–216, 222, 231, 235–236, 238, 276, 295, 347, 361, 407, 424–426, 432–435 ant guards 86, 340, 422–424, 433 ant-house bromeliads 21, 23, 53, 96, 187, 189, 199, 213–216, 218, 222, 231, 357, 369, 371, 407, 423, 474–477, 562 carton 206, 366–367, 407, 424–425 defense of trees hosting ant-house bromeliads 340, 429 evolution of ant/plant associations 431–435 myrmecochory 102, 104, 200, 214, 223, 245, 289, 295–296, 347–348, 365–366, 433, 550 prey of Brocchinia reducta 220 removing bromeliads from trees 340 Apomixis 491 Aquatic Bromeliaceae 110, 388–389, 392, 397 Archbold Biological Station 338 Argentina 242, 262, 390, 399, 416, 468–471, 483, 505, 513, 579, 589, 599, 609 Arizona–Sonora Desert Museum 618 Aroids 48, 98, 205, 211, 357–360, 366, 367, 484 Asexual reproduction 30–34 from spent infructescences 30, 279, 323, 325, 574 types/locations of offshoots 324–325 Atacama 325, 395, 399, 515–516, 602, 609 Atlantic Forest 10, 255, 265, 271–273, 279, 320, 340, 355, 357, 367, 371, 388, 397, 416, 418, 478–479 “Atmospherics” 12 Australia 220, 408 Austria 619 Bahamas 398 Bahia State 7, 9–10, 320, 330–331, 340, 363, 367, 390, 408–409, 416, 419, 479, 551 Baja California 212, 246, 362, 399 Barra de Marica restinga 397–398 Barro Colorado island 159, 272, 283, 325, 389 665 666 Subject index Belize 360, 367, 392, 424–425, 581 Benzothiazole 295, 347 Bermuda 480 Beta-hydroxyethyldrazine 275 Big Cypress National Preserve 308, 312 Big Pine Key 346 Big Thicket National Preserve 413 Biogeochemical cycling 378–384 Brazil 7, 9–10, 12, 14, 157, 178, 198, 208, 213, 218, 241, 254–257, 259, 265–266, 272–279, 284, 293, 296, 316, 319–321, 330–331, 335–336, 344, 363, 367, 387–392, 394–395, 398–399, 410, 416, 436, 456, 465–466, 478, 486, 504, 545, 569, 579, 582, 585, 598, 606 Brazilian Shield 390 Breeding systems 86, 245, 253, 276–280, 306, 338 allogamy 250, 253, 256, 282, 508, 549 andromonoecy 264, 491, 552 autogamy 86, 89, 97, 250, 265, 276–277, 283, 549, 577 cleistogamy 246, 250, 277, 491–492, 577, 615 dioecy 83, 85, 92–93, 95, 278–280 self-incompatibility 265, 276–277 protandry 276, 565 protogyny 252, 276 Bolivia 14, 335, 341, 399, 469, 485, 507, 513, 579, 583, 599 British West Indies 436–437 Bromelain 602 Bromeliad Society 619 Bryophytes 166, 345, 380, 414, 439 Caatinga 10, 226, 320, 551 Cacti 50, 98, 290–291, 316, 365, 395, 615 California 612–613, 616 Campos de altitude 335, 479 Campos rupestres 7, 10, 260, 299, 319–321, 351, 390–391, 436, 467, 479–480 Caribbean 321, 393, 398, 414 Virgin Islands 442–444, 577, 608 Carnivory 11, 23, 46–47, 59, 75, 133–134, 187–190, 199, 208, 217–226, 231, 233, 235, 367, 371, 373, 395, 438, 450, 474–478, 497 Brocchinia reducta 219–222 Catopsis berteroniana 218–219 identity of prey 220–221 Puya raimondii 218 Central America 278–280, 335, 340, 440, 569, 576, 607, 610 Cerrado 264–267 Cerro Neblina 7, 419, 439, 467 Chapada Dimontina 391 Chemical control of epiphytic Bromeliaceae 242 Chemical defenses 196, 406 Chiapas State 278, 340, 611, 616 Chile 6, 331, 334, 390, 399, 468, 589 Chloroplast chlorophyll content in foliage relative to exposure 140, 176, 181 chlorophyll fluorescence 107, 177–181 genome 1, 30, 519–521 ndhF gene 522–524 rbcL gene 522 structure relative to shade adaptation 177 Chóco 247, 356, 392, 599 Chromosomes 252, 263, 488–492, 507, 547, 551–552, 571, 577 aneuploidy/polyploidy 252–253, 282, 488–492, 552, 577 Cladogenesis 8, 248, 250 effects of climate 514–516 reticulate evolution 488–492 role of pollinators 246–264 role of substrates 306–308, 484–488 vicariance 366, 369–371, 383, 394, 399–400, 465–482 Clonal growth (see also Genetic structure of populations) 283, 323, 389, 547–548, 551 Cloud forest 75, 182, 207, 438, 585 Cohesion ratio 481–482, 484 Colombia 12, 207, 247, 278, 356, 369–370, 388, 392, 396, 416, 438, 466–467, 485, 487 Commensalism 223 Competition, competitive exclusion, concurrence 313, 359–360, 378, 380, 389, 391 Corcovado Basin 358–360 Corm-like rhizomes 47 Costa Rica 12, 69, 126, 201–202, 207–208, 210–211, 235, 262, 271–272, 276–278, 299, 316–319, 340, 358, 360, 364–365, 368, 370–371, 382, 388, 415–416, 418, 420, 445, 462, 480, 610, 615 Crassulacean acid metabolism (CAM) 6, 11–12, 45–46, 62–63, 65, 67, 108–109, 112–123, 227 adaptive response to stress 168–169 anatomy of CAM foliage 401 CAM cycling 115, 131, 136, 170 CO2 recycling/CAM idling 115, 117, 119, 122, 131, 169–174 evolution of CAM from C3 ancestry 115, 136, 494–500, 538–540 facultative CAM 121–122, 136–137, 156, 356 involvement in hydration 174–175 nature of reserves used to energize dark acidification 173 performance of CAM vs. C3–CAM types in situ 121, 228 Subject index performance of C3 vs. CAM types in situ 120, 151–160, 354–355, 383, 389 relative to exposure to SO2 and O3 244 role of citric acid 174 temperature optima 120, 128–129, 170–171 Cuticle 149, 174, 190, 218–219, 224, 230, 352, 371 Cyanobacteria 211–213, 440, 442 Demography 245, 308–319, 368, 396 catastrophic mortality 319–323 recruitment 308–312, 316–317 survivorship 312–319, 368 Denitrification 458 Detritivores (see also Animal-assisted saprophytism) 55, 207–208, 430 Dichogamy 276 2,4-dichlorophenoxyacetic acid 275 Distribution (in space) 465–466 Brocchinia 471–479 Bromelioideae 479–480 effects of climate 107–186, 482–485 influence of plant characteristics 151, 369–373, 482–488 in tree crowns 341, 361 other patterns for epiphytes 358–360 Pitcairnioideae 466–479 Puya 468 regional patterns 357 Tillandsioideae 480–482 vertical stratification in forests 341, 351–358, 360–362 Deuterium 125–126 Ecological types (Pittendrigh’s) 8–11, 227–228 comparisons with Benzing’s five types, 111–113 Type One 130–132 Type Two 123–130, 227–228 Type Three 132–133 Type Four 133–142 Type Five 143–145 Ecuador 7, 12, 205–206, 210–212, 316, 323, 331–334, 339, 343, 356–357, 364, 384, 407, 422, 424, 466, 484–487, 509–510, 547–548, 550, 590, 602, 606, 610, 615–616 El Niño 175, 360, 514–515 El Salvador 278–280, 610 Endangered bromeliads conservation laws 619–620 Convention on International Trade in Endangered Species (CITES) 619–620 Endangered Species Act 617 ex situ conservation 616–619 667 factors threatening populations 610–615, 620 in situ conservation 615–616 model for rate of extirpation 610–611 most heavily traded species 613–614 species listed in Appendix II of CITES 620 World Conservation Monitoring Center of the World Conservation Union (IUCN) 617 Epiparasitism 373, 379 Epiphytism 7–8, 14, 21, 28, 41, 43, 47, 49, 58, 63, 102, 384, 387–389 accidental types 387 bark vs. rock as substrates 305–308, 328, 344, 385, 395 effects on phorophytes 344, 372–382 facultative types 386–389 host specificity 339–351 inducement for speciation 482–488 life history of Tillandsia paucifolia 308–317 occurrence in family 385–386 partitionment of tree crowns 356–362, 365–366 relative to breeding system 276 role in succession 362–368 survival in hurricanes 321–323 vertical distribution in forests 341, 351–358 Epiphytosis 374 Eocene 464, 524 Espirito Santo State 7, 254, 335, 414, 419, 551 Ethylene 275 Europe 612, 617 Everglades National Park 321 Evolutionary relationships (see also Ancestral habitats; Habits; Nectaries; Trichomes) Brocchinia 471–475 Bromeliaceae vs. other monocots 2, 522–527, 540 Bromelioideae 536–538 fruit types 536 habits among Bromeliaceae 42–48, 513 mesic vs. xeric habits in Tillandsioideae 509–516, 531–532, 570 ovary positions 535 Pitcairnioideae 538–540 Pittendrigh’s four ecological types 532–534 seed types 537 subfamilies within Bromeliaceae 521–527 Tillandsioideae 247, 257, 527–535, 575–578 trichomes 70–72, 77, 475–478, 493–500 668 Subject index Fairchild Tropical Gardens 618 Fauna (see also Ants; Frugivores; Pollination; Seed dispersal; Tables 8.1, 8.2) amphipods 208, 457 Aranea 430 Ascaridea 430 bats 75, 88, 256, 260, 456 Chilopoda 430, 458 Coleoptera 208, 340, 429, 457 Collembolla 208, 430, 460 copepods 443–444 crabs 408, 449–450 detritivores 223, 226, 456–462 Diplopoda 430 Diptera (unspecified) 430, 446, 457 earthworms 421, 446, 449, 460–462 fauna in phytotelmata of Venezuelan bromeliads 411 frogs 227, 296, 398, 417, 419, 446, 450, 456, 599, 606 guinea pig 288, 601 Homoptera 214–215, 223, 408–409, 422, 424, 429, 431 hummingbirds 252, 255–256, 261, 267, 415–416 isopods 208, 430, 449, 457 leaf miners 409 Lepidoptera 262, 408, 430, 564 lizards 418 marsupials 296, 415 mayflies 457 midges 440, 442–444, 457 mites 208, 460 mollusks 410, 458 mosquitoes 421, 439, 441–449, 458 nematodes 410 nonhummingbird birds 261, 265, 271, 284, 290, 399, 415 Odonata 419, 430, 442, 446, 449–450, 452, 454 Orthoptera 406 ostracods 439, 442, 444 peccaries 389 pests on imported bromeliads and countries of origin 410 Phalangidae 430 primates 406, 414–415, 599–600 protozoa 444 rodents 228, 296, 389, 415, 601 rotifers 442–444 salamanders 420–421 Salatoria 446 scorpions 458 snakes 418 spectacled bear 218, 599 symphylids 408 syrphids 449 termites 214, 319, 407–408, 430, 436–437 Thysanura 430 wasps 221, 267 Ferns 166, 208, 211, 214, 357, 360, 364–367, 383 Fiber production 323 Fire: tolerance and effects 21, 43, 319–321, 390–392, 401, 475, 480, 600, 602, 611 Florida 7, 9–10, 12, 107, 195, 203, 207, 218, 241–243, 253, 259, 273–274, 278–279, 282, 285, 303, 307–318, 321–323, 326, 328, 335–338, 343–346, 349–356, 362, 377–382, 396, 408, 412, 424, 432, 447–448, 454, 588, 595, 599, 608, 618 Flowers and flowering (see also Inflorescence; Nectaries; Pollination) 31, 47, 89–98, 249, 549 androecium 91–93, 94, 96–97, 249, 253, 529, 552, 566, 574 calyx 87, 90–91, 294–297 corolla (see also Flowers) 91, 97 evolution (see Cladogenesis) fragrances 224, 251, 254–255, 263–264, 268, 552, 565, 574 Gardner’s five floral types in Tillandsia subgenus Tillandsia 249, 251–254, 583–584 gynoecium 91, 94, 574 induction of flowering 272–276 petal scales 94, 96–98, 247, 262, 516, 545, 549, 565, 571, 574 phenology 267–276 Racinaea 585 rewards (see also Nectaries and nectar) 264–298 stigma morphology 80, 92, 94–96, 262, 556, 568, 574–576 synchronization 269–273, 280–281, 316–317 Tillandsia subgenus Allardtia 578 Tillandsia subgenus Anoplophytum 579 Tillandsia subgenus Diaphoranthema 581–582 Tillandsia subgenus Phytarrhiza 580–581 Tillandsia subgenus Pseudoalcantarea 584 Tillandsia subgenus Tillandsia 582–583 Foliage absorptive function 49 deciduousness 31, 45, 52, 57, 59, 132, 160, 169 life history 58–59, 176 morphology/anatomy 29, 52–54, 59–75, 184, 561–564 optical properties 66–70 organization of mesophyll 65–70 Subject index pigmentation 32–33, 35, 37–38, 54–56, 60, 182–186 Foraging for resources through differential growth 36–42 Fossil bromeliads 1, 3, 5, 464–465, 576 France 355 French Guinea 434 Frost-hardiness 8, 13, 43, 110, 131, 147–148, 261, 316, 331–339, 352, 390 effects on geographic ranges 334–339, 362 Fruits composition of fleshy types 291–292 fruit flags 86, 263 fruit set 268 phenology of ripening 293 structure 87–90, 98–105, 292 Frugivores 91, 136, 288–289, 293–299, 346–348, 359, 366 Fungi (see also Mycorrhizas) 88, 214, 377, 379–380, 430, 435 Galapagos Islands 480 Gallery forest 260, 390 Genetic structure of populations 253, 281–284, 507–508, 609 Germany 612–613, 619 Germination 49, 99, 104, 259, 294, 301, 311, 315, 345–346 Geologic history of Bromeliaceae (see also Fossils) 463 Georgia 338 Geotropisms roots 132, 400, 560 stems 560 Gesneriads 205, 366, 426, 456 Glaciation 515 Global change 107, 202, 459 Gnetophytes 4, 115 Goiás State 551 Gran Sabana 220, 222, 281 Great Lakes of North America 362 Guadeloupe 595 Guana Island 436 Guanacaste Province 365 Guatemala 198, 280, 366, 581, 613, 615 Guayanan Shield (Highlands), Guayana 6, 14, 190, 199, 219, 258, 260, 288, 319, 357, 386, 390–391, 393–394, 438–439, 466–478, 483, 570 Guttation 87 Habits, vegetative (see also Clonal growth; Ecological types; Heterochrony) 19–35, 42–48, 133–134, 238–240 economic analysis 217, 301–305, 371–373 evolution 42–48, 513 organization for foraging 36–42 669 Haiti 203 Halophytes (salt-tolerance) 7, 197–198, 401–404 Hawaii 331, 598, 618 Hemiepiphyte 11, 30–31, 34, 45, 47, 353, 386 Herbivory/herbivores (see also Fauna) 28, 55, 70, 75, 88–89, 184, 215, 223, 405–414, 422, 429 Herkogamy 276 Heterochrony (see also Ancestral habitats) 6, 20, 24–25, 28, 43–46, 48, 57, 81, 83, 85, 89, 136–142, 238–240, 250, 238–239, 463, 492, 500–508, 527, 571 Heterophylly 6, 8, 21, 25, 31, 34, 44–45, 56–58, 136–142, 160, 358, 499, 501–502 Hispanola 584 Holocene 338, 397, 480 Homoplasy 67, 79, 85, 97, 102, 391, 500, 504–506, 514, 517, 526, 535, 571, 575 Honduras 204, 605 Honeydew 214, 425 Host decline (epiphytosis) 372–382 Host preferences for epiphytes 308–312, 339–351 effects of bark 343–346 effects of seed dispersers 346–348 roles of nutrients and light 348–351 Human disease 440 Humboldt current 515 Hurricanes 308, 321–323, 619 Hybridization 247–248, 252, 488–492, 587 Indigenous people 588–590 indigenous management 607–608 indigenous taxonomy 588–589 Indole acetic acid 275 Inflorescence/infructescence anatomy of axis 561 development 88–89 frost-tolerance 332–334 organization 81–89, 257–259, 263–264, 267, 305, 548–549, 564 order of flowering 270–271 perennial types 303, 548 pseudanthium 81, 83 Inselberg 10, 286, 390, 393, 487, 507 Iteroparity 245–245, 250 Jamaica 107, 196, 357, 384, 393–394, 417, 440, 442, 450–454 Jatun Sacha Ecological Reserve 616 Jurassic 1 Kukenan-tepui 221 Lagos de Monte Azul National Park 615 Lake Gatun 356 670 Subject index La Selva 207, 211, 416 Lichens 345 Life history analysis (see also Demography; Reproductive allocation) 301–319 monocarpy vs. polycarpy 326–328 saxicoles vs. epiphytes 305–307, 328, 344, 492 Limonene 295 Lithophytes (see Saxicoles) Loja Province 356 Louisiana 166, 242, 339, 413 Malaysia 340, 426 Malthusian coefficient 326, 504 Mangroves 7, 198, 345 Marie Selby Botanical Gardens 270, 548, 617–618 Mesoamerica 14, 253, 255–256, 285, 393, 420, 479, 583, 585 Methyl-5-substituted phenyl derivatives 295 Methyl-6-methylsalicylate (6-MMS) 295, 347, 435, 550 Mexico 7, 9, 14, 198, 250, 252–253, 264, 270, 276, 278, 283, 286, 296, 305–306, 315, 335, 340–342, 357, 365, 367, 378, 387, 390, 393, 398, 412, 420, 423, 427, 460–462, 471, 483, 584, 598, 604, 610–611, 613–614 Microbes 209–213, 223–224, 232, 235, 345, 439, 451–452, 456–458 Mimicry 222, 347, 376, 435, 550 Minas Gerais State 7, 279, 320, 331, 335–336, 390, 393, 396, 407–410, 427, 436, 479, 498, 551 Mineral nutrition (see also Ants) 64, 110, 160 absorption 193 additions of nutrients to soil 398 assistance from microbes 209–213 critical concentrations of nutrients 196 effects of epiphytes on the biogeochemistry of hosting ecosystems 207, 223, 382–384 involving phytotelmata (see also Animalassisted saprophytism) 111–118, 189, 216–228 mineral-use efficiency (MUE) 128, 191, 195–197, 238, 383–384 mobilization from spent ramets 47–48 modes of nutrition in Brocchinia 472 myrmecotrophy (see also Ants) 230, 235, 238 nitrate reductase 237–238 nitrification 207 nitrogen as a tracer 208–209 nitrogen fixation (nitrogenase) 187, 200, 205, 208–209, 211–213, 226, 440, 442, 459 nitrogen in leaves relative to sun exposure 125, 228, 349–350, 401–404 nitrogen nutrition 58, 110, 116, 191, 195–196, 205, 208–209, 211–213, 226, 235–238, 385, 401 nitrogen-use efficiency (PPNUE) 128, 130–131, 195–197, 372 nutrients from the atmosphere 201–205, 208, 348–351, 377–384 nutrients in bark and arboreal rooting media 205–207, 345, 380 nutrients in the phytotelma of Guzmania monostachia 204 nutrients in terrestrial soils 380–382, 391, 394 nutrients in tissues 192–193, 195, 228, 232, 240–244, 348–350, 380 nutritional modes of Bromeliaceae 200 oligotrophs vs. eutrophs 8, 197, 199, 202, 244 plant architecture as it relates to nutrient economy 238–240 seedling nutrition 236–238 supply vs. demand 188–197 uptake via roots vs. shoots 162, 232 Miocene 515 Mistletoes 291–291, 343 Molecular systematics 79 Molecular clocks 518 Monocarpy 24–25, 43, 47, 81, 88, 245, 250, 257, 259, 261, 276–277, 281, 285, 287, 308, 319, 323, 395, 412, 560 monocarpy vs. polycarpy (iteroparity) 316–319, 326–329 size of shoot at flowering 317–318 Monteverde 388, 415 Mosaic evolution 516 Mutualisms and other relationships (see also Ants; Fauna; Mycorrhizas; Phytotelmata; Pollination; Seed dispersal) ants 421–435 birds 415–417 frogs 417–419 mammals 414–415 salamanders 420 termites 436–437 to assist nutrition 223–227 Mycorrhizas 34, 209–211, 374 Myrmecochory (see Ants) Myrmecodomatium 422 Myrmecotrophy (see Ants) 1-naphthalene acetic acid 275 Naturalization 355, 465 Nectaries and nectar (see also Ants; Pollination) 97, 408 Subject index evolution of the floral types 93 extrafloral nectaries 84, 86, 88, 224, 228, 233, 235–238, 408, 422–424 nectar composition 262, 264–267, 422, 568 timing of secretion 254–255 Neoteny (see Heterochrony) Netherlands 612 Nicaragua 278 Nitrate reductase (see Mineral nutrition) Nitrogen fixation (see Mineral nutrition) North Carolina 612 Nutritional piracy 378–382 Ohio 273, 277 Orchids 48, 208, 211, 270, 323, 338, 341–343, 357, 360, 364–368, 374, 377, 396, 426, 428–429, 449, 485, 610 Organ mountains 255 Ortho-vanillyl alcohol 295 Osa peninsula 358 Osmoregulation, osmotic adjustment 107, 149, 157–158, 175 Ovules 92, 99–100, 104, 568, 574–575 Pacific Ocean 346 Paleoclimate 400, 460, 467, 513–516, 569–570, 576 Panama 126, 128, 156, 177, 203, 278, 280, 284, 354, 357, 359, 389, 507, 598 Pantepui 219, 394 Papua New Guinea 428 Parabiosis 430 Paraguay 513, 598, 607, 610 Paramo (see Alpine bromeliads; Andes) Paraná State 336, 598 Pathogens (see also Human disease) 70, 132, 312, 376–377, 405–414, 429, 437 Parasitism 187, 343, 373–380, 435 Pearl bodies 422 Periderm 46 Peru 6, 12, 14, 174, 214, 295, 325, 347, 365, 393–396, 406, 416, 424, 435, 446, 451, 458, 469, 471, 483–488, 510, 514, 545, 547–548, 550, 569, 579, 583, 588, 600–601, 605, 611 Phloem 48, 50, 66, 378, 560, 609 Photoinhibition (see Photosynthesis) Photoperiodism 28, 128, 272–276 Photorespiration (see Photosynthesis) Photosynthesis (see also Anthocyanins; Mineral nutrition; Water relations) C4 pathway 55, 114, 117, 119–120, 122, 141, 184, 228 distribution of C3 and CAM taxa in Bromeliaceae 108–109, 118 ecological correlates of the three photosynthetic pathways 120–123 671 evolutionary considerations (see also Ancestral habitats) 493–500 exposure to high light incl. photoinhibition, photorespiration, and photoprotection 66–70, 116–119, 150, 159, 168, 176–186, 228, 398, 401–404 light quality 67 optical properties of mesophyll 66–70 pathway relative to life stage 57, 136, 142 photosynthetic capacity relative to N supply 170–172, 195 Pittendrigh’s three exposure types 352–354 quantum yields 133, 179, 185, 354, 497 relative to dispersal mode 298, 351–359, 401–404 sun and shade tolerances and acclimatization 21, 52, 54, 64, 66–70, 72, 76, 123–130, 132, 134–137, 143, 154, 176–186, 228, 328, 351–358, 388 sun flecks 67, 116, 168, 185 Phyllotaxis (see also Habits) 25–28 Phylogenetic constraints 257 Phylogenies and cladograms Brocchinia 473–474 Bromeliaceae relative to other monocots 523 Bromelioideae 538 CAM vs. C3 lineages in Bromeliaceae 495 fruit types in Bromeliaceae 536 ovary positions in Bromeliaceae 535 Pitcairnioideae 470 Pittendrigh’s four ecological types 533–534 seed morphology in Bromeliaceae 537 subfamilies within Bromeliaceae 521, 525 subgenera of Tillandsia/Vriesea 509–510 Tillandsia subgenus Phytarrhiza 511–512 Tillandsioideae 528, 530 xeric vs. mesic habits in Tillandsia subgenus Phytarrhiza 511–512 xeric vs. mesic habits in Tillandsioideae 532 Phytogeography 1–4, 5, 12–14, 465–488, 609–610 Brocchinia 471–478 Bromelioideae 478–480 Pitcairnioideae 466–478 Racinaea 585 Tillandsia subgenus Allardtia 579 Tillandsia subgenus Anoplophytum 579 Tillandsia subgenus Diaphoranthema 582 Tillandsia subgenus Phytarrhiza 581 Tillandsia subgenus Pseudoalcantarea 584 Tillandsia subgenus Tillandsia 583 Tillandsioideae 480–482, 569–571, 576–577 672 Subject index Phytotelmata 14, 19, 49, 53, 55–56, 67, 74, 90, 110, 122–123, 132–133, 149, 186, 220, 232, 354 adaptations and fidelities of residents 444–450 litter processing 207–208, 223, 456–459 nutrient supply for plant 113, 204, 219–229, 438 plant-provided benefits to symbiotic biota 450–456 protective (for symbionts) leaf color 54–55, 182–186, 413, 454–456 resource for mutualists 384, 439–459 shapes of impoundments 23 structure and dynamics of resident communities 225–227, 441–450 substrates for flora 366–368 swamps or islands? 438–439 theoretical considerations 438–439 water chemistry 133, 451–456 water supply for plant 157–160, 172, 227, 383–384 Pleistocene 505, 515 Pliocene 515, 569 Plio-Pleistocene 400, 466, 513 Pollen 105, 289, 465, 555, 566–567, 570, 574 Pollination (see also Flowers) 55–57, 75, 244, 246–268 anemophily 91, 260 Bromelioideae 262–264 cantharophily 267 chiropterophily 83–84, 87, 96, 246–247, 254–256, 258–260, 263, 268, 565 entomophily 84, 86–87, 96, 246, 252, 254–255, 257, 260, 263 melittophily 91, 97, 253, 255, 258–259, 262, 267 ornithophily 84, 86, 88, 91, 96–97, 246, 248, 251, 254–255, 258, 260–262, 265–267, 587 phalaenophily 252, 258, 260–261, 268 Pitcairnioideae 257–262 psychophily 267 relative to phenology 268–276 sphingophily 91, 97, 253–254, 260, 267 Tillandsioideae 247–257 trap-liners 261, 284, 507 Pollution 231 Polyembryony 90, 99 Polyploidy (see Chromosomes) Predators 376 Protocarnivores 217 Puerto Rico 10, 165, 211, 379, 417 Quintana Roo State 423, 428–429, 432 Rainforest 202, 357, 364, 388, 547, 610 Rancho Grande 210, 411, 457 Refugia 514, 570, 576–577 Reproductive allocation (see also Life history analysis) 302, 305–306, 326 Respiration (dark) 173 Restinga 9, 199, 299, 316, 363, 367, 387, 438, 441–444, 551 Rheophytes (see also Aquatic Bromeliaceae) 392 Rio de Janeiro State 7, 9–10, 279, 331, 336, 344, 363, 387–389, 397, 414, 419, 551, 620 Rio Grande du Sol State 336, 480 Rio Palenque 196, 205–207, 211, 422, 424 Riverine bromeliads 392 Roots absorptive function 34, 48–50, 70, 111–113, 137, 162, 237–238 development 35, 100, 237, 560 holdfasts 229 morphology and extent of evolutionary reduction 33–35, 43–44, 48–50, 100, 229, 238–240, 343, 395, 559–560 tropisms 49, 132, 400 Roraima Formation 219, 394 Salta Province 507 San Fernandez Islands 334 San Louis Potosi State 250 Santa Catarina State 335–336, 355, 394, 444 São Paulo State 336, 416 Savannas 66, 199, 219, 386, 391, 461 Saxicoles 6–7, 8–9, 41, 49, 58, 63, 66, 102, 104, 250, 285–286, 288, 303–308, 323, 326, 329, 330, 478 Secondary metabolites 517–518, 562–564 Seed dispersal (see also Ants) 8, 55–56, 75, 91, 98, 103–104, 245, 264, 268–269, 284–299 ballistic 102, 289 bats 294, 296, 346 birds 289–299, 550 Bromelioideae 289–299 crabs 296 dispersal syndromes in Bromeliaceae 297 frogs 296 nonvolant mammals 102, 294–299 Pitcairnioideae 287–288 Tillandsioideae 284–287 water 104, 245, 288 wind 296, 346, 359–360, 362 Seedlings (see also Heterochrony) 28, 47–48, 57, 99–100, 132–133, 139, 159, 279, 287, 322, 345, 356, 362, 365, 387, 389–390, 499, 611, 616 Seeds 98–105 Subject index development 99 elaiosomes 288, 294 evolution 79, 101–102, 104–105 germination 99, 104, 259, 289, 299–301 morphology 90, 98–105, 249, 284–299, 499, 568, 570 viability 299–301 Sergipe State 551 Serrania de Macuria 369–370 Sierra de Alamos 616 Sister group of Bromeliaceae 464, 489, 522, 524, 526 Soil (see also Mineral nutrition) definition of suspended soils 459–462 terrestrial soils, fertility 381, 472–475 Sonora 611, 616 South America 256, 285, 335–336, 390–392, 399–400, 420, 469, 487, 489, 507–508, 515, 570, 576, 583, 589, 598 South Carolina 143 Stems anatomy 46, 560 for water storage 147–148 Sian Ka’an Biosphere Reserve 342, 423 Stomata 59–64, 75–77, 132, 141, 143, 145–146, 153–155, 161–162, 168, 179, 231, 423, 560 humidity sensors 62, 155, 159–160, 172 types in Bromeliaceae 61–63 Succession fauna in shoots of Aechmea bracteata 430 in forest canopy and restingas 358–368 on ant carton 365–366 substrates for other flora 365–368 Talamanca (Cordillera de) 316, 415 Tambopota 425 Taxonomy Brocchinia 471–478 Bromelioideae 536–538 chemical systematics 517–540 Cryptanthus 552–553 current status of Bromeliaceae 11–12 Neoregelia subgenus Hylaeaicum 545–546, 550 Pitcairnioideae 538–540 plesiomorphic character states for Bromeliaceae 5–6, 70 Tillandsia and Racinaea 573–575, 577–578 Tillandsioideae 527–535, 555–559 traditional characters 94, 516–517, 555–559 Tepuis (see Guayanan Shield) Terrestrial Bromeliaceae (see also Alpine bromeliade; Saxicoles) 384–404 facultative types 384–390 673 lithophytes 392–396 profile of Bromelia humilis 400–404 restinga inhabitants 397–399 rupestrals 390–392 Tertiary 5, 460, 465, 577 Texas 192, 241 Trichomes (see also Ancestral habitats; Mineral nutrition; Water relations) development 28, 72 effects on gas exchange 144, 153 evolution 70–72, 77, 475–478, 493–500 evolution of absorptive capacity 475–478 functional variety 39, 46, 64, 67, 72–73, 75, 111–114, 132, 162, 400–401 hydration from moist air 163, 165–168, 230 nutrient uptake 163, 189, 219, 230–235, 477 reflectance 150–151, 163–165, 176, 178, 230, 428, 501 roots vs. shoots 232 structure 6, 8, 11–12, 24–28, 33, 35–36, 38, 52, 63–64, 70–75, 229–230, 400, 549, 562, 570 thermal insulation 331–334 water absorption 162–166, 229–230 Trinidad 12, 17, 115, 132, 155, 175–176, 352–355, 360, 400, 424, 440, 498 Tropisms (see also Roots) 47 United Kingdom 612, 614 Uruguay 579, 582 USA 334–335, 349, 569, 582–583, 587–588, 606, 609, 612, 614, 616, 618 Uses of bromeliads 589–607 commercial 606–607 fiber 67, 590, 595–597 food 595–597 forage 599 fuel 600, 602 medicine 602–604 miscellaneous 605–606 ornamental and ritual and mythical 601, 604–605 UV-B radiation 6, 110, 219, 331 Vascular cambium 85 Venezuela 7, 12, 125, 128–130, 147–148, 176, 182, 196, 198, 208, 210, 213, 219–222, 278, 281, 323, 383–384, 398, 401–404, 438, 456, 460, 481, 545, 569, 585, 602 Vera Cruz State 340, 345, 367, 408, 412 Viruses 429 Water relations (see also CAM; Foliage) elastic modulus 149 674 Subject index Water relations (cont.) fog dependence 178, 325, 346, 395, 399, 482–483, 515 hydraulic lift 175 optimization of use 160–162 poikilohydry 166 retention during drought 146, 149–150, 164–167 seedlings 245–246 storage (capacitance/succulence) 32, 36, 51–53, 65, 67, 76, 111, 114, 122, 145–151, 172–174 tracheary cells 46, 48, 50–52, 66, 162, 378, 560 transpiration 51, 61–63, 147 uptake/hydration 50, 61, 64, 162–168, 174–175 water-use efficiency (WUE, transpiration ratio) 114–123, 132, 142, 161, 174, 228, 383 xylem tensions 138, 140, 147, 175, 402 Windward Islands 400 World Resources Institute 610 Xanthophyll cycle (see also Photosynthesis) 177–179, 181, 228, 401 Xeromorphy (see also Water relations) 58, 111, 145–151, 154, 176, 227, 401, 404, 562–563 Xylem (see Water relations) Yucatán State 7, 59, 259, 278–281, 294, 319–320, 342, 345, 387, 399, 407, 430, 456 Taxon index Abromeitiella (= Deuterocohnia) 42, 43, 47, 105, 323, 325, 335, 400, 467, 470, 471, 495 Abromeitiella (Deuterocohnia) lorentziana 41 Abronia 418 Acacia 215 A. cornigera 340 Acanthostachys 13, 65, 74, 98, 386 Acari 411 Acarina 221 Acetobacter diazotrophicus 213 Acutaspis tingii 410 A. umbonifera 410 Admontia 412 Aechmea 11, 13, 75, 81, 88, 92, 97, 101, 115, 198, 200, 211, 262, 263, 283, 284, 289, 291, 293, 294, 296, 297, 298, 299, 328, 347, 385, 397, 418, 419, 451, 456, 464, 478, 479, 480, 487, 494, 498, 517, 521, 525, 533, 535, 536, 537, 538, 546, 547, 550, 588, 598, 604 A. aculeatosepala 546 A. subgenus Aechmea 546 A. angustifolia 7, 87, 90, 101, 212, 295, 366, 407, 422, 424 A. aquilega 115, 120, 121, 125, 155, 172, 175, 440, 449 A. aripensis 155, 156, 354 A. bracteata 23, 27, 30, 53, 65, 71, 75, 82, 86, 90, 96, 102, 133, 153, 164, 200, 216, 218, 226, 232, 236, 263, 265, 291, 294, 342, 343, 369, 407, 426, 427, 429, 430, 433, 434, 437, 452, 455, 588, 590, 591, 602, 607 A. bracteatus 607 A. brassicoides 23, 427 A. brevicollis 23, 295, 394, 424 A. bromeliifolia 81, 90, 156, 320, 397, 408, 433 A. chantinii 33, 35, 73 A. subgenus Chevaliera 295, 517 A. coelestis 300, 301 A. dactylina 90, 102 A. distichantha 246, 265, 267, 298, 299 A. downsiana 354 A. fasciata 82, 86, 173, 263, 413 var. fasciata 266 A. fendleri 121, 173, 354 A. filicaulis 456 A. fulgens 82, 86, 91, 185, 273 A. gamosepala 293, 298, 299 A. glomerata 518 A. haltonii 536 A. kuntzeana 90 A. subgenus Lamprococcus 478, 546 A. lasseri 210 A. lingulata 120, 121, 263, 443 A. longifolius 347, 424, 435, 593 A. lueddemanniana 283 A. macavughii 283 A. magdalenae 32, 66, 90, 103, 107, 108, 124, 125, 127, 128, 156, 177, 246, 283, 284, 323, 325, 326, 354, 389, 403, 494, 507, 590, 591, 596, 597, 598, 605, 606, 607, 608, 611 A. marie-reginae 92, 278 A. meliononii 427 A. mertensii 102, 276, 295, 347, 424, 434, 483 A. mexicana 283, 284, 460, 461 A. miniata 185 A. nallyi 406 A. nudicaulis 9, 62, 120, 121, 129, 152, 155, 172, 173, 174, 175, 194, 198, 271, 272, 298, 299, 301, 355, 363, 397, 398, 416, 418, 438, 442, 443, 444, 449, 466, 478, 479, 591 var. aequalis 41 A. organensis 293, 298, 299 A. section Ortgiesia 85 675 676 Taxon index A. paniculigera 451, 452, 454 A. pectinata 55, 81, 298 A. penduliflora 25, 292 A. phanerophlebia 215, 320, 407, 427, 429, 433, 437 A. subgenus Podaechmea 283, 536 A. pubescens 358, 359, 360 A. purpureoresea 263 A. recurvata 236 A. rosea 74 A. setigera 82, 427 A. tessmannii 591, 599, 604 A. tillandsioides 69, 81, 164, 276, 292, 295, 424, 466, 491, 591, 599, 604, 607 var. kienastii 367, 424, 426 A. tuitensis 283, 284 A. veitchii 23 A. warasii 273 A. wittmackiana 323 A. zebrina 212, 591, 599, 602, 604 Aectumea setigera 86 Aeromonas 211 Agavaceae 117, 127, 260, 414 Agave 228, 316, 403 Agave desertii 148, 171, 401 Alcantarea 13, 284, 286, 308, 331, 335, 368, 385, 386, 437, 439, 487, 498, 555, 557, 564, 565, 569 A. duarteana 391 A. edmundoi 498 A. farneyi 498 A. hatscbachii 391, 498 A. imperialis 31, 32, 281, 395, 439, 498 A. nevaresii 83, 286 A. regina 7, 88, 246, 265, 266, 270, 277, 301, 319, 390, 438, 498, 518 Alchornea triplinervia 341 Alloplectus 272 Aloe 119 Alnus acuminata 364 Amazilia fimbriata 255 Anabaena 213 Ananas 13, 49, 62, 75, 110, 113, 124, 125, 227, 228, 275, 294, 354, 387, 403, 404, 412, 491, 494, 525, 533, 534, 535, 536, 537, 538, 540 A. ananassoides 125, 128, 277, 404, 591, 598 A. bracteatus 90, 277, 591, 598, 607 A. comosus x, 9, 46, 60, 66, 68, 74, 107, 108, 119, 120, 124, 125, 128, 129, 148, 149, 150, 164, 172, 173, 210, 227, 228, 277, 331, 354, 385, 400, 401, 403, 404, 413, 488, 493, 494, 587, 589, 590, 591, 595, 598, 602, 605, 606, 607, 608 var. Brecheche 129 var. Spanish Red 129 A. lucidus 125, 591, 598 A. paraguazensis 125, 404, 591, 598 Andrea 517 Androlepis 13, 92, 105, 278 A. skinneri 92, 460, 461 Annona glabra 7, 356 Anochetus 425 A. emarginatus 437 Anopheles 440 A. aegypti 445 A. bellator 440 A. homunculus 440 A. neivai 440 Anthurium 290, 291, 357, 367 A. gracile 347, 434 A. hacumense 360 Aparasphenodon brunoi 418 Apis mellifera 267 Araceae 98, 214, 290, 295, 356, 358, 360, 484, 485 Araeococcus 13, 347, 385, 479, 525, 533, 535, 536, 537, 538 A. micranthus 90, 293 A. pectinatus 524 Aranea 221, 430 Araneida 411 Aratus 449 Arecaceae 331, 397, 523 Aregelia 545; see also Regelia A. subgenus Eu-Aregelia 546 Arphnus melanotylus 410 Artibeus lituratus 256 Ascaridea 430 Asclepiadaceae 215, 424, 619 Astelia 478 Asteraceae 5, 147, 331, 333, 391 Asterolecanium epidendri 410 Atta 429 A. cephalotes 429 A. mexicana 410 Avicennia germinans 310 Ayensua 13, 69, 92, 105, 132, 319, 469, 470, 472 A. uaipanensis 59, 96, 246, 262, 391, 394 Azolla 213 Azteca 295, 348, 367, 425, 426, 429, 430 Aztekium hintonii 615 Bacillus 211 Billbergia 13, 33, 74, 75, 81, 86, 88, 115, 156, 262, 263, 273, 274, 276, 293, 294, 295, 335, 341, 385, 397, 413, 418, 419, 478, 479, 487, 518, 525, 533, 535, 536, 537, 538 B. amoena 82 var. amoena 265, 266, 271, 273, 274, 276 B. subgenus Billbergia 276 B. brasiliensis 25, 292 Taxon index B. distachia var. distachia 274 B. elegans 88, 90, 102, 103, 294, 299 B. euphemiae var. euphemiae 274 B. subgenus Helicodia 276 B. horrida 246, 263 var. tigrina 274 B. lietzei 413 B. macrolepis 69, 524 B. nutans 97, 265, 274, 275, 491 B. porteana 9, 23, 39, 87, 88, 200, 226, 246, 340, 341, 346, 369, 371 B. pyramidalis 355, 445, 608 var. concolor 274 var. pyramidalis 271, 273, 276 B. robert-readii 263 B. rosea 103 B. sanderiana 48, 55 B. saundersii 413 var. debilis 274 B. vittata 274, 418, 518 B. zebrina 73, 294, 417 Blattodea 411 Bombus 267 Bothrops schlegeli 418 Brachycera 411 Brassalvola nodosa 365 Brewcaria 13, 95, 260, 394, 466, 469, 470, 516 B. reflexa 473 Brocchinia 11, 13, 24, 30, 39, 43, 44, 53, 70, 71, 73, 74, 75, 91, 94, 95, 101, 104, 105, 112, 113, 123, 132, 133, 199, 200, 220, 231, 281, 288, 390, 394, 463, 466, 469, 470, 471–8, 488, 495, 496, 497, 498, 500, 518, 522, 525, 526, 530, 533, 535, 536, 537, 539, 540, 541 B. acuminata 21, 53, 69, 71, 74, 105, 189, 200, 215, 234, 249, 257, 386, 394, 427, 473, 474, 475, 476, 477, 528, 532 B. amazonica 473, 474 B. bernardii 394 B. cataractacum 475 B. cowanii 394, 473, 474, 477 B. cryptantha 394 B. deliculata 249, 475 B. gilmartinii 473 B. hechtioides 39, 71, 219, 473, 474, 477 B. hitchcockii 475, 526 B. maguirei 473, 474, 475, 477 B. maguleri 249, 386 B. melanacra 59, 249, 319, 473, 474, 475, 477 B. micrantha 24, 33, 43, 49, 74, 125, 178, 231, 249, 281, 386, 438, 439, 473, 474, 475, 477 B. paniculata 43, 63, 473, 474, 475 B. prismatica 231, 386, 473, 474, 476, 477, 498 677 B. reducta 23, 24, 39, 59, 70, 71, 74, 105, 133, 134, 189, 190, 219, 220, 221, 222, 224, 231, 232, 235, 268, 371, 386, 394, 450, 473, 474, 475, 476, 477, 497 B. serrata 249, 472, 473, 539 B. steyermarkii 96, 262, 473, 474, 476, 477 B. tatei 7, 24, 101, 105, 213, 220, 231, 276, 281, 288, 386, 390, 419, 439, 473, 474, 475, 477, 498, 526 B. vestita 113, 190, 249, 473, 474, 476, 498 Bromelia 13, 25, 27, 28, 33, 34, 41, 47, 49, 54, 57, 59, 60, 63, 64, 65, 74, 75, 91, 98, 110, 113, 115, 124, 218, 227, 228, 293, 294, 328, 387, 393, 397, 398, 409, 410, 478, 480, 491, 494, 517, 518, 525, 528, 530, 532, 533, 534, 535, 536, 537, 538, 598 B. alsodes 591, 598, 603, 604 B. balansae 32, 164, 291, 292, 294 B. chrysantha 125, 591, 603 B. goeldiana 125 B. hemisphaerica 591, 598 B. humilis 62, 107, 108, 123, 124, 125, 129, 130, 131, 170, 171, 174, 176, 177, 198, 210, 228, 323, 331, 354, 398, 400–4, 493 B. karatas 194, 262 B. laciniosa 590, 591, 606, 607 B. nidus-puellae 591, 598, 603, 605 B. pinguin 68, 186, 262, 296, 591, 598, 603, 606, 607 B. plumieri 121, 172, 591, 598, 603, 605, 606, 607 B. serra 265, 591, 607 B. tenuifolia 464 Bromeliaceophyllum oligovaenicum 464 B. rhenanthum 464 B. urbaniana 591 Bromeliales 490, 522 Bromelianthus heuflerianus 464 Bromeliiflorae 522, 523 Bromelioideae 10, 12, 25, 28, 30, 39, 44, 47, 48, 49, 50, 53, 55, 56, 57, 62, 63, 66, 67, 71, 73, 74–5, 81, 82, 86, 91, 92, 94, 95, 96, 98, 101, 102–3, 105, 107, 108, 110, 111, 112, 113, 114, 118, 124, 150, 154, 164, 197, 200, 212, 231, 232, 234, 247, 262–84, 288, 289–328, 336, 340, 345, 346, 357, 365, 367, 385, 386, 388, 398, 399, 417, 424, 427, 429, 431, 433, 449, 466, 472, 477, 478–80, 483, 486, 488, 491, 492, 493, 495, 496, 497, 498, 499, 502, 516, 521, 522, 525, 526, 528, 533, 534, 535, 536, 537, 538, 540, 541, 545, 546, 550, 551, 552, 602 Bruchidae 410 Bucida spinosa 342 Bumelia celastrina 346 678 Taxon index Bursera simaruba 310, 311, 343, 344, 429 Burseraceae 290 Cacicus haemorrhous 416 Cactaceae 98, 117, 127, 136, 291 Calathea 128 Callithrix geoffroyi 414 Calyptranthes 347 Campodeidae 411 Camponotus 425, 435 C. abdominalis 430 C. femoratus 295, 347, 425, 432, 435, 550 Canistropsis 537 Canistrum 13, 25, 74, 81, 85, 371, 478, 480, 494, 525, 533, 535, 536, 537, 538, 545, 546, 549 C. fosterianum 74 C. lindenii 92, 100, 355 Carabidae 411 Caria domitianus domitianus 564 Carica papaya 602, 603 Castine phalanis 408 Castnia eudesmia 261 Catopsidae 411, 570 Catopsis 11, 13, 34, 61, 79, 81, 85, 92, 93, 94, 95, 102, 103, 133, 135, 136, 198, 247, 254, 268, 278, 280, 284, 315, 339, 341, 343, 351, 357, 362, 387, 412, 463, 478, 495, 522, 527, 528, 529, 530, 541, 555, 557, 561, 562, 563, 564, 565, 566, 568, 569, 570, 585, 588, 605 C. berteroniana 59, 133, 134, 136, 164, 190, 200, 218, 219, 224, 278, 280, 285, 315, 319, 321, 323, 337, 351, 352, 371, 394, 445, 562, 570 C. compacta 280 C. delicatula 280 C. floribunda 29, 37, 93, 136, 148, 194, 280, 285, 351 C. hahnii 280, 591 C. juncifolia 280 C. mexicana 280 C. micrantha 280 C. minimiflora 280 C. montana 280, 525, 528, 530, 532, 533, 535, 536, 537 C. morreniana 136, 278, 280, 559, 591 C. nitida 136, 280 C. nutans 26, 27, 71, 72, 109, 134, 136, 150, 151, 152, 153, 165, 166, 186, 210, 234, 254, 278, 280, 285, 337, 351, 352, 362, 524, 556 C. paniculata 254, 280 C. pisiformis 93, 280 C. sessiliflora 83, 280, 566, 567, 591 C. subulata 280, 591 C. subgenus Tridynandra 565 C. wangerinii 278, 280, 592 C. wawranea 280, 525, 528, 530, 532, 533, 535, 536, 537 C. werckleana 280 Cebia peltandra 364 Cecropia 215, 348 Ceratophyllaceae 239 Ceratopogonidae 221, 440, 442 Cercidium praecox 378 Cerridomyiidae 440 Chalcicoidea 221 Chelonethida 441 Chilopoda 411, 430 Chironomidae 221, 442 Chiropterotriton 420 Chlorophyta 440 Chlorostilbon aureoventris 225, 267 Chrysomelidae 410 Cimolus vitticeps 410 Citrus 360, 367, 378, 424, 425 Cladosporium myrmecophilum 214 Clusia 174, 366, 398 C. alata 382 Clusiaceae 367, 397 Coccidae 410 Coccinellidae 411 Codonanthe 189 C. macrodenia 367, 426 C. uleana 347, 366 Collembola 221, 411, 430, 460 Coleoptera 221, 411, 412, 430 Colibri coruscans 261, 267 Columnea 272, 366 Commelinaceae 89, 478, 523 Commeliniflorae 522, 523 Connellia 13, 69, 95, 105, 260, 394, 466, 469, 470 C. smithiana 260 Conocarpus erecta 310, 311 Coprinus 88 Coreidae 410 Coryanthes speciosa 426 C. speciosum 367 Costaceae 523 Cottendorfia 13, 61, 63, 69, 70, 91, 92, 93, 95, 104, 257, 260, 278, 470, 495 C. florida 21, 47, 391, 564 Couratari stellata 364 Coussopoa microcarpa 341 Crematogaster 340, 425, 426, 430 C. linata 295 C. linata parabiotica 425 Crescentia alata 365 Crophinus costatus 410 Cryptanthoideae 552 Cryptanthus 8, 13, 30, 34, 35, 38, 47, 54, 73, 74, 75, 81, 86, 92, 95, 97, 98, 115, 235, Taxon index 263, 278, 319, 325, 386, 478, 479, 480, 487, 491, 494, 516, 525, 533, 534, 535, 536, 537, 538, 551–3 C. acaulis 47, 164, 325, 491 C. bahianus 491 C. beuckeri 294, 491 C. bromelioides 9, 25, 90, 108, 357 C. correia-araujoi 82 C. subgenus Cryptanthus 93, 263, 294, 491, 551, 552, 553 C. exaltatus 264 C. subgenus Hoplocryptanthus 264, 294, 551, 552, 553 C. leopoldo-horstii 391 C. odoratissimus 264 C. pseudoscaposus 294 C. schwackeanus 391 Culex 446 Culicidae 221, 443, 449 Culicoidel 221 Cullidae 442, 458 Curaeus curaeus 261 Curculionidae 410, 411 Cyclanthaceae 357 Cyperaceae 235, 523 Cyphomyrmex minutus 430 Dalbergia 456 Dascyllidae 411 Deinacanthon 13, 517 Dendrobates pumilio 418 Dendrocoris variegatus 410 Dendrotriton 420 D. xolocalcae 420 Dermaptera 411 Deuterocohnia 13, 88, 105, 257, 267, 335, 400, 422, 467, 470, 471, 495 D. haumanii 267, 323 D. longipetala 96, 262, 264, 267 D. lorentziana (see Abromeitiella) D. meziana 46, 84, 85, 539 D. schreiteri 93, 303 Diabrotica porracea 410 Diaspidae 410 Diaspsis bromeliae 408 Dieffenbachia 128 Dionea 224 Diplopoda 221, 411, 430 Diplura 411 Diptera 221, 411, 430, 442, 446, 457 Dischidia 424 Disteganthus 13, 54, 389, 478, 494 Dolichoderinae 434 Dolichoderus 425, 430 D. bispinosus 430 Drosera 219 679 Drosophilidae 411 Drynaria 214 Dyckia 7, 13, 21, 34, 45, 50, 53, 62, 73, 85, 88, 91, 93, 98, 101, 111, 113, 130, 176, 257, 260, 267, 319, 355, 386, 390, 391, 393, 407, 409, 422, 437, 464, 467, 470, 471, 479, 480, 491, 495, 496, 521, 525, 533, 535, 536, 537, 538, 539 D. brevifolia 53, 608 D. dissitiflora 391 D. ferox 264, 267 D. floribunda 84, 267 D. extevesii 21, 26 D. ferox 246 D. pseudococcinea 397 D. ragonesei 80, 93, 265, 267 D. velascana 265, 267 Dynastor napolean 408 Dysdercus mimulus 410 Dysmimococcus probrevipes 410 Edmundoa 537 Elpidum bromeliarium 442, 444 Eleuthrodactylus 417 E. jasperi 417 Encholirium 13, 45, 85, 130, 257, 259, 279, 319, 320, 330, 331, 390, 391, 393, 407, 437, 467, 470, 471, 479, 480, 495, 525, 533, 535, 536, 537, 538, 539 E. glaziovii 84, 246, 258, 260, 267, 268 Enchytraeidae 411 Encyclia cordigera 365 E. tampensis 198, 205, 311, 338, 343 Epipedobates tricolor 605 Epidendrum immatophyllum 367, 426 E. marsupiale 343 Epiphyton 526 Equisetum 187 Ericaceae 364, 415 Eriocaulaceae 391 Eriophorum vaginatum 235 Erythroxylon 367 E. ovalifolia 296 Erythoxylum coca 605 Espeletia 147, 333 Euphorbia horwoodii 619 Euphorbiaceae 619 Eurema diara 262 Eutrigaster sporadonephra 461, 462 Exptochiomera albomaculata 410 Fabaceae 147 Fagaceae 5 Fascicularia 13, 98, 264, 289, 331, 336, 386, 390, 478, 488, 536 F. bicolor 246, 386 F. pitcairniifolia 355 680 Taxon index Fernseea 13, 264, 480 F. itatiaiae 335 Ferrocactus acanthoides 171 Ficus 366 F. aurea 310, 311, 343 Flagellariaceae 523 Forestiera segregata 310 Formicidae 221, 410, 411, 422, 432 Formicinae 434 Fosterella 13, 30, 42, 53, 58, 71, 73, 92, 94, 95, 101, 104, 105, 111, 130, 132, 257, 260, 469, 470, 472, 491, 525, 533, 535, 536, 537, 539 F. penduliflora 24, 52, 59, 84, 264, 471, 540 F. spectabilis 84, 260 Fraxinus caroliniana 341 Fritziana goeldii 418 Fusarium 413 Gastropoda 411 Gastrotheca fissilis 418 G. fissipes 417 Gecarcinus lateralis 296 Genlesia 219 Geohintonia mexicana 615 Geranospiza caevulescens 416 Gesneriaceae 184, 189, 214, 295, 415 Gigaspora 210 Ginkgo 619 Glomeropitcairnia 11, 13, 73, 79, 92, 95, 101, 247, 284, 463, 470, 481, 495, 516, 521, 522, 526, 527, 528, 529, 530, 532, 533, 535, 536, 537, 541, 555, 562, 564, 566, 568, 569, 570, 571 G. erectiflora 438 G. penduliflora 90, 104 Glomeropitcairniaeae 516, 525 Glomus tenue 210 Glossoscolecidae 411 Gnathostomulida 501 Gramineae 498 Grapsidae 449 Greigia 13, 34, 74, 82, 85, 115, 264, 294, 331, 390, 399, 478, 480, 488, 500, 536, 598 G. oaxacana 264 G. sodiroana 592, 606, 607 G. sphacelata 264, 592 Gryllidae 410, 411 guava 196 Guzmania 13, 54, 61, 81, 91, 94, 95, 103, 212, 247, 254, 256, 257, 258, 357, 386, 387, 392, 412, 451, 466, 483, 491, 495, 521, 525, 528, 529, 531, 532, 533, 535, 536, 537, 540, 555, 557, 561, 562, 563, 564, 565, 566, 568, 569, 570, 579 G. acorifolia 392 G. acuminata 592, 599 G. alcantareoides 256 G. berteroniana 211 G. bismarckii 184 G. blassii 611 G. caricifolia 498 G. coriostachys 256 G. cylindrica 370 G. diffusa 564 G. eduardii 592, 599, 604 G. fosteriana 256 G. globosa 83, 89 G. glomerata 564 G. kentii 257 G. lindenii 38 G. lingulata 52, 54, 38, 109, 134, 135, 150, 164, 184, 194, 369, 370, 563 var. minor 246 G. melinonis 212, 592, 599, 604, 606 G. monostachia 7, 65, 71, 107, 109, 116, 121, 122, 133, 136, 137, 140, 146, 149, 153, 156, 157, 158, 159, 168, 176, 177, 179, 180, 181, 182, 186, 195, 204, 216, 225, 228, 256, 271, 272, 277, 285, 315, 337, 341, 342, 351, 352, 354, 356, 360, 362, 369, 370, 406, 413, 458, 494, 528, 530, 532, 540, 559, 567, 592, 599, 604 G. mucronata 136, 256 G. musaica 565, 592 G. nicaraguensis 271, 528, 530, 532, 559, 564, 615 G. plicatifolia 528, 530, 532 G. rhonhofiana 528, 530, 532 G. sanguinea 64, 86, 136, 263, 369, 370, 528, 532, 556, 592 var. brevipedicellata 273 G. scherzeriana 272 G. subgenus Sodiroa 498 G. spectabilis 528, 530, 532 G. weberbaueri 446, 451, 458 G. wittmackii 83, 256, 528, 530, 532 G. zahnii 182, 184, 273, 559 Haemodoraceae 523 Haemodorales 522 Hanseniella 408 Hechtia 3, 13, 28, 45, 50, 53, 73, 85, 92, 95, 105, 111, 113, 130, 176, 218, 257, 260, 267, 278, 288, 319, 386, 393, 466, 470, 471, 480, 496, 521, 539, 566 H. carlsoniae 85 H. glomerata 173 H. scariosa 267 H. schottii 267, 280, 320 Heliamphora 190, 219, 220 H. hederodoxa 222 H. nutans 220, 221 H. tatei 220 Taxon index Helicina zephyrina 410 Helicinidae 410 Heliconia 442, 445 Heliconiaceae 523 Helminthosporium rostratum 413 Hemiberlesia palmae 408 Hemiptera 411 Hermanthena candida 564 Hirudinae 411 Histeridae 411 Hohenbergia 7, 13, 55, 263, 363, 390, 397, 408, 409, 418, 419, 451, 478, 479, 525, 533, 536, 537, 538 H. blanchetti 246 H. subgenus Hohenbergia 479 H. pendulaflora 358 H. urbanianum 32 H. subgenus Wittmackiopsis 479 Hohenbergiopsis 13, 105 Homo sapiens 218 Homoptera 221, 223, 347, 408, 422, 425, 429, 431 Hoya 427 Hura crepitans 346, 364 Hydnophytum 427 Hyla brunnea 417 H. truncata 296 H. venulosa 417 Hymenoptera 221 Hypoclinea 348 H. bispinosa 429 Idras 346 Impatiens 501 Inga 347 Ionantha scaposa (see Tillandsia kolbii) Ionopsis satyrioides 426 I. utricularioides 426 Iridomyrmex 428 Isoetes 116 Isopoda 430 Japygidae 411 Juglandaceae 5 Juncaceae 523 Kalanchoe 173 Karatas 545 K. section Regelia 546 Karatophyllum bromelioides 464 Lagothrix lagothicha 600 Lagriidae 410 Lauraceae 290 Lemnaceae 501 Lentibulariaceae 224 Leontopithecus rosalia 414 681 Lepidoptera 221 Leptogeny 430 Leptogrion perlongum 449 Leptospermum 340 Liliales 90 Liliopsida 3, 22, 34, 50, 52, 464, 478, 518, 521–7, 539, 540, 564 Lindmania 13, 69, 73, 92, 93, 257, 394, 469 L. guianensis 262 L. longipes 473 L. serrulata 24 L. wurdackii 24 Liodidae 411 Liquidambar styraciflua 286, 287 Lobeliaceae 331 Loganiaceae 599 Lonchophylla bokermanni 260 Loranthaceae 290, 291, 415 Loricifera 501 Lowiaceae 523 Lucanidae 411 Lupinus 147 Lygaeidae 410 Lymania 13, 357, 494, 525, 533, 535, 536, 537, 538 L. smithii 185 Magnoliophyta 4, 5, 8, 19, 75, 277, 463 Magnoliopsida 68 Marantaceae 523 Marcgraviaceae 415 Margarornis rubiginosus 415 Maschalocephalus 489 Mayacaceae 540 Melanapsis odontoglossi 410 Melastomataceae 215, 340 Membracidae 410 Menispermaceae 599 Mesostigmata 411 Metacypris 444 Metamasium callizona 412 Metamasius hemipterus hemipterus 410 Metasesarma rubripes 408 Metopaulias 449 M. depressus 450 Metriocnemus 221 Metriona trisignata 410 Mezobromelia 13, 94, 95, 254, 525, 528, 529, 530, 531, 532, 533, 535, 536, 537, 555, 557, 564, 565, 566, 568, 569, 576, 579 Microcoelia 374 Microgramma lycopodoides 196, 383 Mimosestes dominicanus 410 Miridae 410 Monacid debilis 425 Monimiaceae 341 Monomorium ebeninum 430 682 Taxon index Monotropa 379 Moraceae 214, 295, 341 Mormidea collaris 410 Musaceae 523 Myrmecodia 427 Myrmicinae 434 Myrsine guianensis 310 Myrtaceae 546 Napea eucharila 564 N. theages theages 564 Nasutitermes 437 N. acajutlae 436 Nasutitermitinae 437 Navia 11, 13, 24, 61, 69, 70, 71, 73, 91, 101, 104, 105, 257, 260, 263, 287, 288, 385, 386, 393, 394, 464, 466, 469, 470, 475, 488, 495, 496, 516, 525, 533, 535, 536, 537, 539 N. arida 260, 539 N. caulescens 84 N. glandulifera 75 N. glandulosa 24, 39, 73, 476 N. igneosicola 539 N. jauaensis 260 N. lactea (see N. ocellata) N. linearis 84 N. ocellata 263 N. phelpsiae 539 N. polyglomerata 84 N. saxicola 394 N. splendens 260, 473, 528, 532 N. tentaculata 386 Nematocera 411 Neoglaziovia 13, 66, 74, 124, 328, 480 N. variegata 108, 320, 590, 592, 606, 607 Neoponera villosa 429 Neoregelia 3, 13, 32, 33, 38, 42, 55, 56, 81, 82, 85, 87, 88, 89, 115, 200, 262, 263, 293, 294, 297, 305, 328, 347, 357, 397, 456, 464, 478, 525, 533, 535, 536, 537, 538, 545, 546, 547, 549, 550 N. abendrothae 21, 25, 502 N. aculeatosepala 546, 547 N. subgenus Amazonicae 546 N. ampullacea 81 N. carolinae 545 N. cathcartii 545 N. concentrica 294–5, 299, 547 N. cruenta 119, 296, 316, 363, 367, 397, 398, 438, 441, 442, 444, 479 N. cyanea 547 N. diamantinensis 391 N. eleutheropetala 85, 548, 549, 550 var. bicolor 548, 549 var. eleutheropetala 548, 549 N. subgenus Hylaeaicum 85, 97, 478, 545–50 N. lactea 56 N. leviana 548, 549, 550 N. longisepala 75, 478 N. macwilliamsii 56 N. margaretae 548, 549, 550 N. marmorata 266 N. mooreana 548, 549, 550 N. myrmecophila 85, 478, 483, 548, 549, 550 N. subgenus Neoregelia 479, 545, 546, 549 N. nivea 56 N. pascoalina 292 N. pauciflora 41, 74 N. pendula 550 var. brevifolia 548, 549 var. pendula 547, 548 N. peruviana 547, 548, 550 N. petropolitana 56 N. pineliana 524 N. rosea 548, 550 N. spectabilis 547 N. stolonifera 90, 292, 295, 548, 549, 550 N. tarapotoensis 547, 548, 549, 550 N. wurdackii 548, 550 Nepenthaceae 215 Nepenthes 223 Nicotiana tabacum 520 Nidularium 13, 47, 55, 57, 61, 74, 75, 81, 218, 262, 263, 293, 296, 298, 341, 357, 371, 386, 389, 464, 478, 479, 480, 491, 494, 500, 502, 525, 533, 535, 536, 537, 538, 545, 549 N. ambiguum 92 N. antoineanum 293, 298, 299 N. burchellii 23, 54, 67, 115, 185, 186, 371 N. eleutheropetalum 545, 546 N. fulgens 300 N. innocentii 108, 115, 156, 293, 298, 299, 388, 389, 406 var. paxianum 355 N. longiflorum 413 N. lymanioides 30, 34 N. myrmecophilum 545, 546 N. procerum 156, 246, 388, 389, 406 var. procerum 355 N. subgenus Regelia 545, 546 N. selloanum 524 Nitidulidae 411 Nototriton 420 Nymphaeaceae 89 Ochagavia 13, 98, 264, 331, 334, 390, 478 Ochrimnus vittiscutis 410 Odonata 430, 446 Odontomachus bruneus 430, 432, 437 O. mayi 434 Ogdoecosta biannularis 410 Oniscoidea 411 Taxon index Oplomus rutilus 410 Opuntia 228, 403 O. ficus-idea 171 Orbatei 411 Orchidaceae 5, 198, 215, 295, 305, 323, 338, 340, 342, 343, 364, 365, 374, 377, 392, 394, 429, 430, 449, 464, 484, 485, 486 Orthophytum 13, 32, 47, 62, 74, 89, 95, 98, 115, 235, 264, 289, 325, 331, 390, 478, 479, 487, 491, 516, 521, 525, 533, 534, 535, 536, 537, 538, 552 O. benzingii 325 O. gurkenii 524 O. humile 26 O. saxicola 264 Ostracoda 442 Pachycondyla goeldii 434 P. villosa 429, 430, 432 Paepalanthus bromelioides 391 Pandanus 296 Papilio thoas 267 Paramecium 445 Paroecantus aztecus 410 Patagonia gigas 261 Pedipalpida 411 Pentatomidae 410 Peperomia 55, 115, 196, 364, 367, 484, 618 P. macrostachya 366 P. magnoliaefolia 148 Pepinia 13, 69, 71, 73, 98, 104, 288, 469, 470, 471, 525, 533, 536, 537, 539 P. aphelandrifora 69 P. corallina 69 P. fimbriatobracteata 88 P. pruinosa 84 P. pulchella 592, 599 P. punicea 288, 392 P. schultzei 69 Pereskia 136 Phaethornis idaliae 255 P. superciliosis 284 Phalangida 411 Phalangidae 430 Philander opossum 296 Philaphyllum tenuifolium 439 Philodendron leal-acostae 367 P. saggitifolia 359 Philydraceae 523 Philydrales 522 Phoebus 262 Phytophthora 413 pineapple (see Ananas comosus) Pinus 313, 340, 345, 349 Piper 128 Piperaceae 214, 215, 295, 364, 484, 486 Pitcairnieae 469, 471 683 Pitcairnia 8, 14, 21, 25, 28, 31, 34, 42, 44, 47, 50, 51, 54, 56, 57, 58, 59, 64, 68, 69, 73, 84, 86, 91, 92, 96, 97, 98, 104, 105, 111, 130, 132, 148, 169, 231, 254, 257, 258, 262, 288, 300, 357, 358, 385, 386, 389, 392, 393, 413, 464, 466, 467, 469, 470, 471, 482, 483, 495, 516, 517, 518, 525, 533, 535, 536, 537, 588, 603, 604 P. albiflos 258 P. andraeana 45, 236 P. angustifolia 592, 603 P. aphelandriflora 288 P. arcuata 84, 88, 258 P. bakeri 84, 86 P. bifrons 132 P. breedlovei 592 P. brevicalycina 96, 97, 98, 246, 258, 262 P. bromeliifolia 194 P. brongniartiana 84 P. bulbosa 125 P. corallina 88, 96, 246, 258 P. feliciana 31, 45, 57, 288, 465, 480 P. fimbriato-bracteata 258 P. flammea 10, 84, 86, 100, 109, 110, 237, 238, 300, 392 var. flammea 266 var. pallida 258, 266 P. glaziovii 465 P. halophila 197 P. heterophylla 31, 45, 56, 57, 59, 93, 97, 98, 132, 145, 275, 288, 386, 539, 592, 603 P. integrifolia 108, 132, 197, 198, 592, 603 P. loki-schmidtiae 246, 258 P. macrochlamys 27, 153, 164 P. maidifolia 592 P. meridensis 96 P. nubigena 258 P. palmoides 258 P. pulverulenta 96 P. pungens 36, 69, 210, 592, 603 P. riparia 28, 31, 34, 41, 43, 45, 56 P. rubro-nigriflora 258 P. spicata 592, 603 P. tabuliformis 45, 57 P. trianae 36, 69 P. undulata 164 P. unilateralis 258 Pitcairnioideae 11, 25, 28, 30, 31, 34, 36, 39, 42, 44, 48, 52, 53, 62, 63, 67, 69, 70, 71, 73, 74, 75, 81, 84, 86, 88, 92, 93, 94, 95, 96, 98, 101, 104–5, 108, 111, 112, 115, 118, 132, 164, 200, 231, 234, 245, 247, 257–62, 263, 264, 267, 277, 287–8, 331, 336, 354, 357, 385, 386, 390, 391, 392, 393, 394, 399, 427, 437, 464, 466–78, 480, 482, 483, 492, 495, 496, 497, 516, 684 Taxon index Pitcairnioideae (cont.) 518, 521, 522, 525, 526, 528, 533, 535, 536, 537, 538, 539, 540, 541, 566, 570, 571 Platanaceae 5 Platycerium 214 Platypodium elegans 346 Platypsaris rufus 416 Pleopeltis astrolepis 196 Plethodontidae 420 Poaceae 5, 85, 213, 523 Polypodiaceae 215 Polypodium polypodioides 367 Polytmus guainumbi 255 Ponerinae 434 Pontderiaceae 523 Pontederiales 522 Portea 14, 81 P. petropolitana 32, 66, 355 Procyrta intectus 410 Proechimys iheringi 296 Prostigmata 411 Pselaphidae 411 Pseudaechmea 14 Pseudananas 14, 34, 47, 74, 124, 491 P. sagenarius 292, 325, 592, 598 Pseudococcida 410 Pseudococcus brevipes 408 Pseudocolaptes lawrencii 415 Pseudomonas 211 P. stutzeri 212, 213 Pseudomyrmex 340 Psidium 196, 208, 383, 456 Psocoptera 411 Ptilidae 411 Puccinia pitcairniae 413 P. tillandsiae 413 Puya 6, 14, 25, 43, 45, 46, 62, 65, 69, 73, 96, 97, 101, 104, 122, 130, 131, 147, 148, 218, 257, 258, 260, 261, 265, 288, 299, 319, 331, 332, 333, 334, 335, 357, 385, 386, 390, 399, 400, 464, 466, 467, 468, 469, 470, 471, 480, 482, 483, 486, 495, 518, 521, 525, 530, 533, 535, 536, 537, 538, 539, 566, 570, 593, 602, 603 P. aequatorialis 261, 332, 333, 334, 528, 532 var. aequatorialis 333 P. alpestris 246 P. aristeguietae 96, 262 P. berteroniana 261, 300, 301 P. chilensis 261, 590, 592, 603 P. clava-herculis 316, 331, 332, 333, 334 P. compacta 400 P. copiapina 109, 131 P. dasylirioides 32, 281, 316, 317, 318, 319, 327, 400 P. ferruginea 101, 109, 246, 261, 592, 599 P. floccosa 98, 109, 131, 400, 466, 592, 603 P. gigas 592, 605 P. hamata 333, 334, 592, 598 P. harmsii 93, 469 P. hofstenii 96 P. lasiopoda 469, 592, 599 P. lilloi 469 P. longistyla 593 P. mariae 52 P. medica 593, 603 P. mirabilis 236, 261 P. nutans 400 P. oxyantha 593, 599 P. petropolitana 608 P. pusilla 31, 45 P. subgenus Puya 261 P. subgenus Puyopsis 261 P. pyramidata 593, 599 P. raimondii 25, 36, 45, 47, 88, 218, 276, 332, 587, 593, 597, 600, 602, 607, 611 P. sodiroana 400, 593, 598, 599, 603 P. spathacea 265, 267 P. tuberosa 43, 47, 468 P. venusta 261 P. aff. vestita 333, 334 P. weberbaueri 469, 593, 599, 601, 602, 605, 607 Puyeae 469, 471 Pyrrhocoridae 410 Quercus 340 Q. virginiana 207, 216, 310, 344, 374, 378, 380, 381, 399 Quesnelia 14, 57, 115, 262, 276, 296, 308, 328, 341, 386, 397, 478, 516, 525, 533, 535, 536, 537, 538 Q. arvensis 298, 299 Q. centralis 85 Q. humilis 293, 298, 299 Q. lateralis 85, 265, 266, 273 Q. liboniana 273 Q. quesneliana 156, 406 Q. testudo 292, 293, 298, 299, 308, 323 Racinaea 14, 247, 555, 558, 564, 565, 569, 573, 574, 585; see also Tillandsia subgenus Pseudocatopsis R. commixa 564 R. insularis 480 R. multiflora 585 R. pallidoflavens 566 R. pendulispica 585 R. seemannii 573, 585 R. tetrantha 585 var. tetrantha 574 R. undulifolia 585 Taxon index Rahnella 211 Rapateaceae 478, 489, 522, 523, 524, 525, 526, 528 Regelia 545, 546; see also Aregelia Restionaceae 523 Rhinotermitinae 437 Rhizobium 209 Rhizoctonia 377 Rhizoecus falcifer 408 Rhizophora mangle 7, 198, 310, 311, 345 Rhopornis ardesiaca 416 Riodinidae 408, 564 Ripsalis 290 Ronnbergia 14, 34, 54, 98, 102, 115, 184, 389, 478, 479, 494, 517, 525, 533, 535, 536, 537, 538 R. deleonii 54, 90, 289 R. ecuadoriana 21 R. explodens 289 R. petersii 39, 73, 102 Rosaceae 488 Roridula gorgonias 208 Rubiaceae 340, 365 Rubus 325 Runchmia 220 Saissetta hemisphaerica 408 Salatoria 446 Salix 284 Sappho sparganura 267 Sarraceniaceae 223 Sarracenia flava 220 S. purpurea 226, 445 Schinus terebinthifolius 355 Schomburgkia tibicinis 428, 429, 430 Sciaridae 221 Scutellospora 210 Scydmaenidae 411 Selaginella arenicola 380 Senecio medley-woodii 174 Serenoa repens 602 Sesarma miersii 449 Sobralia 270 Solenopsis 220, 221, 425 Solidago 376 Spanish moss (see Tillandsia usneoides) Spartina 328 Sphagnum 202 Staphylinidae 411 Statira denticulata 410 Stegolepis 525, 528, 530, 532, 533, 535, 536, 537 S. hitchcockii 524 Stelis 196 Steyerbromelia 14, 260, 394, 469, 470 S. diffusa 231 Streliziaceae 523 685 Streptocalyx 74, 479, 517 S. longifolius 347, 424, 435, 593 Streptocarpus 88 Strymon basilides 408 Symphyla 411 Syncope antenori 446 Syringa 476 Tachyphonus coronatus 294 Taxodium distichum 205, 259, 308, 310, 311, 312, 313, 314, 315, 321, 322, 341, 343 Tetramorium simillinum 430 Tettigonidae 430 Theobroma 211 T. cacao 205 Thraupis cyanoptera 416 T. ornata 416 Thysanura 430 Tillandsia 8, 12, 14, 24, 34, 41, 46, 48, 57, 60, 63, 64, 67, 76, 79, 81, 88, 89, 92, 94, 95, 96, 100, 103, 104, 122, 148, 149, 162, 163, 166, 173, 198, 199, 211, 218, 230, 234, 239, 244, 247, 248, 250, 252, 254, 255, 256, 257, 258, 265, 268, 269, 270, 271, 283, 285, 305, 306, 315, 319, 322, 323, 325, 328, 331, 335, 336, 340, 341, 342, 343, 348, 356, 357, 360, 365, 366, 368, 374, 380, 384, 385, 386, 387, 388, 391, 394, 395, 396, 397, 398, 399, 412, 420, 423, 451, 460, 461, 464, 465, 466, 481, 482, 483, 484, 487, 491, 493, 495, 496, 503, 504, 505, 507, 508, 509, 510, 512, 513, 515, 516, 518, 525, 527, 528, 529, 531, 532, 533, 535, 536, 537, 540, 555, 558, 559, 560, 561, 562, 563, 564, 565, 568, 569, 571, 573–85, 588, 594, 597, 602, 604, 605, 606, 611, 612, 613, 614, 619 T. achyrostachys 150, 164, 252 T. acosta-solisii 511, 512 T. adpressa 364 T. adpressiflora 136 var. tonduziana 371 T. aequatorialis 249 T. aeranthos 242, 377, 507, 559, 580, 613 T. aizoides 268, 505, 573, 574 T. alberi 507, 508, 580 T. albertiana 83, 91, 270, 560, 573, 578, 609 T. subgenus Allardtia 92, 94, 248, 505, 509, 510, 528, 529, 558, 562, 569, 574, 575, 576, 577, 578–9, 580, 582, 583, 584 T. anceps 134, 358, 360, 370, 511, 512, 563 T. andreana 562 T. andrieuxii 250, 270 T. angulosa 491, 577 686 Taxon index T. subgenus Anoplophytum 86, 92, 94, 254, 268, 480, 492, 505, 507, 508, 509, 510, 513, 514, 528, 529, 558, 561, 564, 566, 569, 574, 575, 578, 579–80, 581, 582 T. araujei 20, 25, 43, 331, 390, 580 T. arequitae 579 T. argentea 83, 246, 565, 614; see also T. fuchsii T. argentina 579 T. arhiza 391, 396, 511, 512 T. asplundii 212 T. atroviridipetala 583 T. aurea 511, 512, 581 T. australis 573 T. badensis 511, 512 T. bagua-grandensis 579 T. balbisiana 65, 86, 88, 122, 194, 210, 241, 242, 253, 279, 284, 321, 342, 356, 380, 399, 408, 424, 429, 433, 482 T. baileyi 198, 433, 559, 614 T. baliophylla 584 T. bartramii 210, 211, 337 T. benthamiana 593, 604 T. bergeri 528, 529, 530, 532, 556, 580, 613 T. bermejoensis 579 T. biflora 135, 306, 593, 599, 605 T. brachycaulos 86, 345, 613, 614 3 T. balbisiana 248 3 T. bulbosa 248 3 T. capitata 248 3 T. caput-medusae 248 3 T. foliosa 248 3 T. ionantha 248 3 T. mirabilis 248 T. brachyphylla 394, 580, 613, 615 T. bryoides 20, 26, 43, 62, 63, 238, 503, 527, 567, 573, 581, 582 T. buchlohii 579 T. bulbosa 26, 53, 64, 76, 144, 153, 178, 211, 277, 356, 369, 370, 423, 427, 428, 429, 433, 613, 615 T. burlemarxii 580 T. butzii 153, 218, 427, 428, 613, 614 T. cacticola 511, 512, 574, 581 T. caerulea 270, 356, 511, 512, 513 T. calcicola 393 T. camargoensis 579 T. candida 579, 580 T. capillaris 49, 250, 373, 491, 492, 527, 573, 593, 604 f. hieronymi 491 T. capitata 86, 182, 252, 564, 573 T. caput-medusae 71, 176, 189, 198, 215, 218, 277, 427, 428, 613, 614, 615 T. cardenasii 578 T. carlsoniae 270, 573, 593 T. carminea 580 T. carnosa 583 T. castellanii 90, 491, 556, 577 T. caulescens 579 T. cauligera 573, 578 T. cernua 339 T. chaetophylla 612 T. chapeuensis 580 T. chartacea 593, 599 T. chiapensis 610 T. chiletensis 579, 580 T. churinensis 578 T. circinnatoides 145, 211, 212 T. clavigrea 319, 573 T. cochabambae 578 T. colganii 580 T. comarapaensis 580 T. complanata 34, 85, 134, 135, 186, 270, 343, 394, 466, 488, 524, 528, 529, 530, 532, 561, 564, 574, 577, 593, 598 T. concolor 20, 65, 145, 252 T. confinis 339 T. cornuta 511, 512, 513 T. cotagaitensis 578, 579, 581, 582 T. crocata 26, 29, 71, 72, 250, 511, 512, 581 T. cyanea 83, 511, 512, 559 T. dasyliriifolia 198, 278, 279, 342, 387, 388, 399, 593 T. denudata var. vivipara 574 T. deppeana 57, 59, 137, 139, 141, 142, 151, 154, 182, 253, 254, 286, 287, 305, 384, 408, 499, 502 T. dexteri 610 T. diaguitensis 28, 578, 580 T. subgenus Diaphoranthema 28, 86, 94, 99, 136, 162, 238, 252, 253, 254, 268, 281, 480, 482, 491, 492, 505, 506, 507, 508, 509, 510, 513, 527, 528, 529, 559, 560, 561, 567, 568, 571, 574, 575, 577, 581–2 T. didisticha 356, 580, 583 T. dodsonii 87, 270, 511, 512, 513, 528, 530, 532 T. dorotheae 580 T. duidae 579 T. duratii 29, 49, 162, 246, 267, 373, 503, 511, 512, 561–2, 581 T. dyeriana 511, 512, 513 T. ecarinata 583 T. edithiae 33 T. elongata 121 T. eltoniana 579, 580 T. emergens 339 T. erecta 505, 577 T. erinata 393 T. erubescens 250, 270, 593, 598 T. espinosa 560 T. esseriana 580, 581 T. exserta 611 Taxon index T. fasciculata xiii, 29, 146, 150, 151, 156, 157, 158, 159, 253, 285, 286, 295, 303, 307, 321, 322, 336, 337, 342, 356, 380, 393, 399, 412, 424, 480, 564, 567, 573, 589, 593, 599 3 T. foliosa 248 3 T. lieboldiana 248 T. fendleri 120, 134, 573 T. ferreyrae 583, 612 T. festucoides 198 T. filifolia 26, 145, 583 T. flabellata 167, 182, 410 3 T. tricolor 3 Vriesea incurvata 248 T. flexuosa 30, 73, 89, 198, 253, 325, 337, 338, 346, 362, 423, 433, 458, 574 T. floribunda 356 T. fraseri 528, 530, 532 T. friesii 492, 507, 508, 579, 580 T. fuchsii 562, 573, 583, 613 T. funckiana 26, 528, 530, 531, 532, 565, 578, 583 T. gardneri 145, 249, 579, 580 T. geissei 580 T. genseri 580 T. germiniflora 90, 528, 529, 530, 532, 559, 579, 580 T. gilliesii 593, 605 T. globosa 580 T. grandis 88, 270, 308, 395, 573, 584 T. grazielae 395, 580, 609, 613 T. guasamayensis 580 T. guatemalensis 593 T. guelzii 580 T. hamaleana 511, 512, 521, 562, 565 T. harrisii 620 T. hasei 580 T. heterophylla 59, 86, 136, 251, 254, 501, 584 T. heubergeri 580 T. hildae 26, 47, 73 T. hirtzii 615 T. horstii 580 T. huarazensis 577 T. humilis 511, 512, 581 T. hutchisonii 579 T. ignesiae 583 T. imperialis 81, 91, 253, 255, 281, 362, 583 T. incarnata 578, 581, 593, 605, 606, 607 T. incurva 564 T. insignis 498 T. ionochroma 285, 286, 303, 305, 306, 396, 593, 599, 601, 605 T. ionantha 20, 26, 27, 29, 47, 48, 65, 116, 149, 153, 166, 175, 253, 270, 281, 282, 283, 507, 562, 577, 613, 614 687 var. ionantha 49 3 T. schiedeana 248 var. scaposa 614 var. van-hyningii 49 var. zebrina 29 T. ixioides 564, 579, 580 T. jalisco-monticola 3 T. xerographica 248 T. jucunda 580 T. juncea 480, 504, 560, 578, 583, 585, 593, 613, 614 T. juncunda 579, 580 T. kammii 620 T. karwinskyana 26, 72, 76, 145, 150, 164, 178, 307 T. kautskyi 580, 620 T. kirchhoffiana 252 T. klausii 611 T. koehresiana 580 T. kolbii 613, 614 T. krukoffiana 248, 252, 319 T. kurt-horstii 178, 330, 496 T. lampropoda 593 T. landbeckii 491 T. latifolia 49, 325, 395, 559, 574, 578 var. divaricata 356 var. major 325 var. minor 325 var. vivipara 395 T. laxissima 511, 512 T. leiboldiana 185–6, 360, 491, 577 T. leonamiana 580 T. lepidosepala 254, 583 T. lindenii 511, 512, 556 T. linearis 94, 580, 581 T. loliacea 83, 491 T. lorentziana 267, 579, 580 T. lotteae 580 T. lucida 23 T. lymanii 583 T. macdougallii 49, 265, 270 T. maculata 593, 605 T. magnusiana 612 T. makoyana 307, 560 T. malzinei 563 T. marnier-lapostollei 573 T. matudae 252 T. mauryana 581, 583, 620 T. maxima 593, 599 T. mellemonti 511, 512 T. milagrensis 580 T. mima var. chiletensis 574 T. monadelpha 134, 496, 511, 512, 566, 574, 581 T. monstrum 85 T. montana 580 T. muhriae 507, 508, 580 T. muhrii 578 688 Taxon index T. multicaulis 34, 85, 150, 164, 254, 561, 564, 574 T. multiflora 564 T. myosura 28, 250, 491 T. narthecioides 511, 512, 574, 581, 583, 585, 612 T. neglecta 393, 580 T. nubis 511, 512 T. nuptialis 580 T. oaxacana 614 T. organensis 580 T. orogenes 593 T. oropezana 580 T. oroyensis 593, 599 T. pabstiana (5 V. drepanocarpa) 96, 247 T. paleacea 28, 175, 325, 395, 511, 512, 513 T. paniculata 584 T. paraensis 565 T. parryi 250 T. paucifolia xiii, 9, 20, 47, 65, 89, 122, 153, 191, 193, 194, 197, 198, 205, 233, 241, 242, 252, 259, 278, 279, 285, 287, 303, 304, 308, 309, 310, 311, 312, 313, 314, 315, 316, 317, 318, 321, 322, 325, 327, 328, 343, 345, 346, 349, 350, 351, 356, 358, 365, 368, 377, 432, 480 subsp. schubertii 574 T. pedicellata 505, 569 T. peiranoi 505, 511, 512, 580 T. pentasticha 582 T. pfeufferi 580 T. subgenus Phytarrhiza 94, 95, 254, 268, 480, 492, 505, 506, 507, 508, 509, 510, 511, 512, 513, 528, 529, 558, 560, 562, 565, 574, 575, 577, 580–1, 582 T. plagiotropica 583 T. platyphylla 583 T. platyrhachis 511, 512 T. plumosa 271, 583 T. pohliana 237, 238, 579, 580 T. 3 polita 556 T. polystachia 277, 338, 573, 577 T. ponderosa 253, 593 T. pretiosa 511, 512 T. pringlei (see T. utriculata var. pringlei) T. prodigiosa 277 T. propagulifera 574 T. pruinosa 77, 285, 337, 362 T. subgenus Pseudalcantarea 247, 256, 505, 509, 510, 559, 569, 575, 576, 578, 584 T. pseudobaileyi (see T. baileyi) T. pseudocardenasii 578 T. subgenus Pseudocatopsis (5 Racinaea) 505, 509, 510, 513, 528, 529, 573, 578, 585 T. pseudomacbrideana 578 T. pseudomicans 578 T. pseudomontana 580 T. pucaraensis 580 T. punctata 249, 251 T. punctulata 86, 251, 252, 367, 412, 574, 587 3 T. kirchoffiana 248 T. purpurea 175, 395, 511, 512, 563, 574, 581, 593, 606, 609, 611 T. pyramidata 574 T. ramellae 580 T. rauhii 319, 583, 584 T. reclinata 395, 580 T. recurvata 9, 20, 28, 29, 99, 122, 149, 165, 175, 207, 212, 241, 242, 250, 253, 281, 282, 283, 322, 330, 336, 337, 338, 339, 344, 346, 356, 362, 372, 374, 377, 378, 399, 482, 505, 577, 582, 594, 598, 599, 604, 609 T. recurvifolia 580 T. reichenbachii 505, 511, 512, 581 T. retorta 577, 582 T. rodrigueziana 594 T. roland-gosselinii 251 T. roseiflora 580 T. rubella 594, 599 T. rupicola 511, 512 T. scaligera 511, 512, 577 T. schiedeana 47, 65, 166, 167, 172, 173, 198, 211, 270, 482, 594, 604, 612 T. secunda 81, 323, 528, 530, 532, 559, 565, 578 var. vivipara 574 T. seideliana 580 T. seleriana 251, 427, 594 T. selleana 577 T. setacea 29, 285, 336, 338, 573 T. simulata 338 T. skunkei 511, 512 T. 3 smalliana 321 T. 3 smallii 338 T. somnians 559, 574 T. sphaerocephala 285, 286, 303, 306, 396, 594, 599, 601, 605, 608 T. spiraliflora 573, 583, 584 T. sprengeliana 580, 620 T. straminea 356, 510, 511, 512, 567, 581 T. streptocarpa 83, 392, 396, 505, 511, 512, 580, 581 T. streptophylla 7, 39, 165, 189, 251, 345, 423, 429, 433, 594, 599, 608, 615 T. stricta 249, 277, 298, 301, 363, 387, 388, 397, 399, 507, 573, 580, 615 T. sucrei 393, 579, 580, 609, 620 T. sueae 250 T. tectorum 27, 72, 77, 145, 153, 175, 396, 482, 574, 578, 611, 613 T. tenuifolia 330, 344, 507, 579, 580 Taxon index T. teres 583 T. tetrantha 612 T. thiekenii 393, 580 T. subgenus Tillandsia 86, 91, 92, 248, 249, 251, 252, 253, 254, 255, 276, 281, 504, 505, 509, 510, 529, 557, 559, 561, 562, 565, 566, 567, 568, 574, 575, 576, 577, 578, 582–4 T. toropiensis 580 T. tortilis 254, 583 T. tricholepsis 49, 528, 529, 530, 532 T. tricolor 559, 613 T. triglochinoides 511, 512, 581 T. truxillana 578 T. turneri 270, 388 T. umbellata 491, 511, 512 T. undulatobracteata 579 T. usneoides 20, 25, 26, 28, 29, 43, 46, 53, 62, 65, 81, 99, 107, 109, 117, 118, 120, 143, 144, 146, 149, 160, 162, 163, 165, 167, 168, 172, 173, 174, 192, 194, 201, 230, 232, 238, 241, 242, 243, 253, 270, 282, 285, 321, 322, 334, 336, 344, 349, 351, 372, 373, 374, 381, 395, 413, 416, 480, 482, 503, 505, 506, 518, 527, 559, 562, 563, 568, 569, 582, 587, 588, 590, 594, 595, 597, 604, 605, 606, 607 T. utriculata 119, 121, 149, 177, 178, 189, 241, 242, 246, 250, 253, 259, 277, 281, 282, 285, 286, 303, 305, 306, 308, 322, 323, 326, 327, 328, 336, 369, 380, 393, 399, 412, 436, 443, 445, 446, 447, 448, 528, 530, 532, 560, 563, 578, 583, 589, 594, 599 var. pringlei 307 var. utriculata 307, 574 T. valenzuelana 173 T. variabilis 148, 338 T. venusta 511, 512 T. vernicosa 580 T. violacea 594 T. virescens 268, 491 T. viridiflora 83, 185, 256, 521, 584, 585 T. wagneriana 511, 512, 513, 578, 581, 583, 584 T. walter-richteri 580 T. walteri 577 T. werdermanii 175 T. xerographica 503, 613, 614, 615, 620 T. xiphioides 83, 91, 94, 267, 268, 507, 573, 578, 579, 580, 594, 604, 615 var. lutea 580 T. yuncharaensis 578, 580 T. zecheri 578 var. cafayatensis 507, 508 Tillandsioideae 12, 20, 29, 31, 32, 34, 37, 39, 43, 44, 46, 48, 49, 50, 52, 53, 54, 55, 57, 689 58, 63, 64, 67, 70, 71, 72, 73, 74, 75, 76, 77, 79, 81, 83, 86, 88, 89, 91, 92, 94, 96, 97, 98, 100, 101, 102, 103–4, 105, 107, 109, 112, 113, 114, 115, 118, 133, 134, 135, 136, 137, 143, 150, 151, 162, 163, 164, 168, 178, 187, 197, 198, 200, 218, 224, 229, 230, 231, 232, 234, 235, 238, 239, 240, 245, 247–57, 260, 261, 262, 263, 265, 267, 269, 277, 278, 284–7, 288, 295, 296, 298, 299, 301, 338, 339, 346, 354, 356, 357, 369, 385, 386, 387, 388, 390, 391, 393, 395, 398, 406, 408, 415, 424, 427, 428, 431, 438, 464, 466, 469, 470, 472, 476, 477, 480–2, 483, 486, 487, 488, 491, 492, 493, 494, 495, 496, 497, 498, 499, 500, 501, 502, 503, 504, 506, 508, 509, 514, 516, 518, 521, 522, 525, 526, 527–36, 540, 555–71, 575, 576, 588, 604 Tnethecoris distinctus 410 Tococa 347 Toxorhynchites haemorrhoidalis 449 Troglodytes ochraceus 416 Tracheophyta 475 Tremarctos ornatus 599 Trombidiidae 411 Tylenchocriconema alleni 410 Typha 328 Typhaceae 523, 524 Typhales 522 Uredo nidulaari 413 Ursulaea 14 Utricularia 219, 224, 367 U. humboldtii 419, 439 U. nelumbifolia 439 U. reniformis 439 Vellozia 391, 521 Velloziaceae 391, 437, 488, 522 Velloziales 522 Vibrio 211 Vireya 618 Vriesea 11, 12, 14, 30, 31, 54, 57, 79, 94, 95, 96, 247, 254, 255, 256, 257, 258, 265, 267, 276, 277, 298, 325, 331, 335, 336, 341, 357, 358, 360, 385, 387, 397, 418, 420, 439, 451, 464, 480, 487, 495, 508, 509, 510, 516, 521, 525, 528, 529, 531, 532, 533, 535, 536, 537, 540, 555, 558, 559, 561, 562, 563, 564, 565, 568, 569, 571, 573, 577, 578, 579, 588, 604 V. subgenus Alcantarea 104, 256, 257, 509, 510 V. amazonica 120, 175, 364 V. atra 87, 246, 439, 531 var. atra 266 690 Taxon index V. attomacaensis 273 V. bituminosa 55, 81, 87, 256, 567 var. bituminosa 273 V. carinata 164, 246, 255, 257, 335, 336, 355 V. cereicola 394 V. chontalensis 371 V. comata 371 V. corcovadensis 335 V. crassa 439 V. cylindrica 86, 87 V. drepanocarpa 96, 247 V. ensiformis 355, 394 V. erythrodactylon 38, 55, 255, 454 V. espinosae 47, 356, 524, 528, 530, 531, 532 V. fenestralis 136, 559 V. fosteriana 33, 35, 54, 182, 184, 185, 413, 454, 456, 559, 611 V. fosteriana chestnut 54, 56 V. friburgensis 267 V. geniculata 137, 139, 141, 145, 499, 502, 503 V. gigantea 7, 189, 256 V. gladioliflora 528, 530, 532 V. glutinosa 89, 528, 530, 532 V. guttata 335 V. haematina 266, 273, 300 V. hainesiorum 255 V. heliconioides 358, 359, 360 V. heterandra 369, 370, 565 V. heterostachys 273 V. hieroglyphica 184, 237, 413, 456, 611, 620 V. hydrophora 81, 86, 87, 266, 273 V. imperialis 555, 574, 578, 579 V. incurvata 148, 255, 355, 371 V. inflata 246, 336 V. irazuensis 135 V. jonghei 120, 134, 236, 255, 355 V. leucophylla 255 V. longicaulis 273, 336 V. longiscapa 255, 273 V. macrostachya 211 V. malzinei 69, 524, 528, 530, 531, 532 V. minuta 263 V. neoglutinosa 254, 266, 316, 319, 397, 398 V. oligantha 81, 83, 391 V. paralbica 273 V. philippo-coburgii 237, 238, 255, 323 V. platynema 136, 182, 195, 196, 210, 216, 263, 383, 466 V. psittacina 257 V. regnellii 255 V. ringens 212 V. rodigasiana 356 V. sazimae 256 V. scalaris 100, 300, 310 V. schwackeana 489 V. simplex 54, 109, 134, 185, 556 3 ‘Mariae’ 273 V. sparsiflora var. sparsiflora 273 V. splendens 185, 210, 270, 369, 370, 456, 527, 528, 530, 532 var. formosa 54 var. splendens 54 V. splitgerberi 120 V. triligulata 528, 530, 532 V. unilateralis 255 V. vagans 273, 336 V. vietoris 255 V. viridiflora 528, 530, 532 V. vittata 528, 530, 532 V. subgenus Vriesea 509, 510, 527, 528, 556, 558, 566, 574, 576, 578, 582, 583 V. werckleana 594 V. section Xiphion 255, 256, 257, 268, 527, 528, 530, 531, 558, 562, 567; see also Werauhia Wasmania auropunctata 430 Weberocereus glaber 365, 366 Werauhia 14, 255, 257, 555, 558, 562, 564, 565, 566, 569, 576 W. attenuata 371 W. gigantea 556 W. gladioliflora 246, 424, 562, 565 W. section Jutleya 558 W. sanguinolenta 356 W. tarmaensis 565 W. section Werauhia 558 Whitesloania crassa 619 Wittrockia 14, 55, 294, 371, 480, 525, 533, 536, 537, 538, 545 W. campos-portoi 108 W. superba 263, 279, 301, 355 Wyeomyia 220, 221, 441 W. medioalbipes 447 W. mitchellii 445 W. smithii 445 W. vanduzeei 445, 447 Xanthium 119 Xenochalepus omogerus 410 Zingiberaceae 89, 523 Zingiberiflorae 522, 523 Literature cited Abele, L. G. and Means, D. B. 1977. Sesarma jarvisi and Sesarma cookei: montane, terrestrial grapsid crabs in Jamaica (Decapoda). Crustaceana 32: 91–93. Abendroth, A. 1957. Billbergia elegans. Bromeliad Soc. Bull. 7: 38–39. Abendroth, A. 1965. Bromeliads and birds in our garden. Bromeliad Soc. Bull. 15: 107–108. Abendroth, A. 1971. Bromeliads and frogs. J. Bromeliad Soc. 21: 83–84. Abercrombie, M., Hickman, C. J. and Johnson, M. L. 1970. A Dictionary of Biology. Harmondsworth: Penguin Books. Adams, M. 1981. Vireyas return. Pacific Hort. 42: 34–39. Adams, W. W., III and Martin, C. E. 1986a. Physiological consequences of changes in life form of the Mexican epiphyte Tillandsia deppeana (Bromeliaceae). Oecologia 70: 298–304. Adams, W. W., III and Martin, C. E. 1986b. Morphological changes accompanying the transition from juvenile (atmospheric) to adult (tank) forms in the Mexican epiphyte Tillandsia deppeana (Bromeliaceae). Amer. J. Bot. 73: 1207–1214. Adams, W. W., III and Martin, C. E. 1986c. Heterophylly and its relevance to evolution within the Tillandsioideae. Selbyana 9: 121–125. Akeroyd, J., McGough, N. and Wyse Jackson, P. (compilers) 1994. A CITES Manual for Botanic Gardens. Kew, UK: Botanic Gardens Conservation International. Alarcón, R. 1988. Etnobotánica de los Quichuas de la Amazonian Ecuatoriana. Miscelánea Antropológica Ecuatoriana Serie Monograph 7. Guayaquil: Museos del Banco Central del Ecuador. Alcorn, J. B. 1984. Huastec Mayan Ethnobotany. Austin, TX: University of Texas Press. Alexander, C. P. 1912. A bromeliad-inhabiting crane-fly (Tipulidae, Diptera). Entomol. News 23: 415–417. Allen, M. F., Rincon, E., Allen, E. B., Huante, P. and Dunn, J. J. 1993. Observations of canopy bromeliad roots compared with plants rooted in soils of a seasonal tropical forest, Chamela, Jalisco, Mexico. Mycorrhiza 4: 27–28. Alves, M. A. S., Rocha, C. F. D. and Van Sluys, M. 1996. Population recovery rates in Vriesea neoglutinosa 15 months after a fire. Bromélia 4: 3–8. Amorim de Freitas, C. and Scarano, F. R. 1998. Habitat choice in two facultative epiphytes of the genus Nidularium (Bromeliaceae). Selbyana 19: 236–239. 621 622 Literature cited Andrade, J. L. and Nobel, P. S. 1997. Microhabitats and water relations of epiphytic cacti and ferns in a lowland Neotropical forest. Biotropica 29: 261–270. Antibus, R. K. and Lesica, P. 1990. Root surface acid phosphatase activities of vascular epiphytes of a Costa Rican rain forest. Plant Soil 128: 233–240. Arizmendi, M. C. and Ornelas, J. F. 1990. Hummingbirds and their floral resources in a tropical dry forest in Mexico. Biotropica 22: 172–180. Arndt, U. and Strehl, T. 1989. Begasungsexperimente mit SO2 an Tillandsien zur Entwicklung eines Bioindikators. Angew. Bot. 63: 43–54. Arroyo, M. T. K., Squeo, F. A., Armesto, J. J. and Villagrân, C. 1988. Effects of aridity on plant diversity in the northern Chilean Andes: results of a natural experiment. Ann. Missouri Bot. Gard. 75: 55–78. Arslanian, R. L., Stermitz, F. R. and Castedo, L. 1986. 3-methoxy-5hydroxyflavonols from Tillandsia purpurea. J. Natural Prod. (Lloydia) 49: 1177–1178. Ashtakala, S. S. 1975. Flavonoid composition of Aechmea and Billbergia: two closely allied ornamental bromeliads. Science 100: 546–551. Atallah, A. M. and Nicholas, H. J. 1971. Triterpenoids and steroids constituents of Florida Spanish moss. Phytochemistry 10: 3139–3145. Augspurger, C. K. 1985. Demography and life history variation of Puya dasylirioides, a long-lived rosette in tropical subalpine bogs. Oikos 45: 341–352. Ayensu, E. S. 1981. Medicinal Plants of the West Indies. Algonac, MI: Reference Publications, Inc. Aziz, T., Yuen, J. E. and Habte, M. 1990. Responses of pineapple to mycorrhizal inoculation and Fosetyl-al treatment. Commun. Soil Sci. Plant Anal. 21: 2309–2317. Baker, H. G., Baker, I. and Hodges, S. A. 1998. Sugar composition of nectars and fruits consumed by birds and bats in the tropics and subtropics. Biotropica 30: 559–588. Baker, J. G. 1889. Handbook of the Bromeliaceae. London: George Bell and Sons. Baker, K. and Collins, J. L. 1939. Notes on the distribution and ecology of Ananas and Pseudananas in South America. Amer. J. Bot. 26: 697–702. Balistrieri, C. A. 1993. CITES at 20 [in five parts]. Amer. Soc. Bull. 62: 730–733, 819–821, 930–932, 1040–1044, 1172–1175. Barry, D. 1953. Bromeliads in a Mexican desert. Bromeliad Soc. Bull. 1–2: 8. Barthlott, W. and Ehler, N. 1977. Raster-Elektroenmikroskopie der EpidermisOberflächen von Spermatophyten. Trop. subtrop. Pflanzenwelt 19: 1–105. Bartholomew, D. P. 1982. Environmental control of carbon assimilation and dry matter production by pineapple. In Crassulacean Acid Metabolism, eds. I. P. Ting and M. Gibbs, pp. 278–294. Rockville, MD: American Society of Plant Physiology. Bartoli, C. G., Beltrano, J., Fernandez, L. V. and Caldiz, D. O. 1993. Control of the epiphytic weeds Tillandsia recurvata and Tillandsia aeranthos with different herbicides. For. Ecol. Manage. 59: 289–294. Baumert, K. 1907. Experimentelle Untersuchungen über Lichtschutzeinrichtungen an grünen Blättern. Beitr. Biol. Pflanzen 9: 117–134. Beaman, R. S. and Judd, W. S. 1996. Systematics of Tillandsia subgenus Pseudalcantarea (Bromeliaceae). Brittonia 48: 1–19. Bell, A. D. and Tomlinson, P. B. 1980. Adaptive architecture in rhizomatous plants. Bot. J. Linn. Soc. 80: 125–160. Literature cited 623 Bell, A. D., Roberts, D. and Smith, A. 1979. Branching patterns: the simulations of plant architecture. J. Theor. Biol. 81: 351–375. Bell, C. R. and Wilson, C. R. 1989. Spanish moss. In Encyclopedia of Southern Culture, eds. C. R. Wilson and W. Ferris, p. 394. Chapel Hill, NC: University of North Carolina Press. Bennett, B. C. 1984. A comparison of the spatial distribution of Tillandsia flexuosa and T. pruinosa. Florida Sci. 47: 141–144. Bennett, B. C. 1986a. Patchiness, diversity, and abundance relationships of vascular epiphytes. Selbyana 9: 70–75. Bennett, B. C. 1986b. The Florida bromeliads: Tillandsia usneoides. J. Bromeliad Soc. 36: 149–151, 159–160. Bennett, B. C. 1987. Spatial distribution of Catopsis and Guzmania (Bromeliaceae) in southern Florida. Bull. Torrey Bot. Club 114: 265–271. Bennett, B. C. 1988. A comparison of life history traits in selected epiphytic and saxicolous species of Tillandsia (Bromeliaceae). PhD thesis. Chapel Hill, NC: University of North Carolina. Bennett, B. C. 1990. The ethnobotany of bromeliads: the use of Tillandsia species in the highlands of southern Peru. J. Bromeliad Soc. 40: 64–69. Bennett, B. C. 1991. Comparative biology of Neotropical epiphytic and saxicolous Tillandsia species: population structure. J. Trop. Ecol. 7: 361–371. Bennett, B. C. 1992a. Uses of epiphytes, lianas, and parasites by the Shuar people of Amazonian Ecuador. Selbyana 13: 99–114. Bennett, B. C. 1992b. Plants and people of the Amazonian rain forests: the role of ethnobotany in sustainable development. Bioscience 42: 599–607. Bennett, B. C. 1992c. The Florida bromeliads: Guzmania monostachia. J. Bromeliad Soc. 42: 266–270. Bennett, B. C. 1995. Ethnobotany and economic botany of epiphytes, lianas, and other host-dependent plants: an overview. In Forest Canopies: A Review of Research on this Biological Frontier, eds. M. Lowman and N. Nadkarni, pp. 547–586. New York: Academic Press. Bennett, B. C. 1997a. An introduction to the Seminole people and their plants, Part I: History and ethnobotany. Palmetto 17: 20–21, 24. Bennett, B. C. 1997b. An introduction to the Seminole people and their plants, Part II: Seminole plant use. Palmetto 17: 16–17, 22. Bennett, B. C., Baker, M. A. and Gomez, P. 1999. The ethnobotany of the Shuar of Eastern Ecuador. Advances in Economic Botany 14, in press. Bennett, R. B. 1954. Spanish moss and some aspects of its commercial possibilities. Engineering Progress 8: 1–11. Benzing, D. H. 1970a. An investigation of two bromeliad myrmecophytes: Tillandsia butzii Mez. and T. caput-medusae E. Morren. Bull. Torrey Bot. Club 97: 109–115. Benzing, D. H. 1970b. Foliar permeability and the absorption of minerals and organic nitrogen by certain tank bromeliads. Bot. Gaz. 131: 23–31. Benzing, D. H. 1970c. Availability of exogenously supplied nitrogen to seedlings of the Bromeliaceae. Bull. Torrey Bot. Club 97: 154–159. Benzing, D. H. 1978a. The life history profile of Tillandsia circinnata (Bromeliaceae) and the rarity of extreme epiphytism among the angiosperms. Selbyana 2: 325–337. Benzing, D. H. 1978b. Germination and early establishment of Tillandsia circinnata Schlecht. (Bromeliaceae) on some of its hosts and other supports in southern Florida. Selbyana 2: 95–106. Benzing, D. H. 1979. Alternative interpretations for the evidence that 624 Literature cited certain orchids and bromeliads act as shoot parasites. Selbyana 5: 135–144. Benzing, D. H. 1980. The Biology of the Bromeliads. Eureka, CA: Mad River Press. Benzing, D. H. 1981a. The population dynamics of Tillandsia circinnata (Bromeliaceae): cypress crown colonies in southern Florida. Selbyana 5: 256–263. Benzing, D. H. 1981b. Bark surfaces and the origin and maintenance of diversity among angiosperm epiphytes: an hypothesis. Selbyana 5: 248–255. Benzing, D. H. 1987a. Major patterns and processes in orchid evolution: a critical synthesis. In Orchid Biology – Reviews and Perspectives, vol. IV, ed. J. Arditti, pp. 33–79. Ithaca, NY: Cornell University Press. Benzing, D. H. 1987b. The origin and rarity of botanical carnivory. Trends Ecol. Evol. 2: 364–369. Benzing, D. H. 1989. The biological status and chemical composition of Spanish moss (Tillandsia usneoides) in the Big Thicket National Preserve: an update for 1984. Special Publication SP-4450-89-10. Research Triangle Park, NC: Northrop Environmental Sciences Inc. Benzing, D. H. 1990. Vascular Epiphytes. New York: Cambridge University Press. Benzing, D. H. 1991. Myrmecotrophy: origins, operations and importance. In Ant–Plant Mutualisms, eds. C. R. Huxley and D. F. Cutler, pp. 353–373. Oxford: Oxford University Press. Benzing, D. H. 1998.Vulnerabilities of tropical forests to climate change: the significance of resident epiphytes. Climatic Change 39: 519–540. Benzing, D. H. and Bermudes, D. 1992. Epiphytic bromeliads as air quality monitors in south Florida. Selbyana 12: 46–53. Benzing, D. H. and Burt, K. M. 1970. Foliar permeability among twenty species of the Bromeliaceae. Bull. Torrey Bot. Club 97: 269–279. Benzing, D. H. and Dahle, C. E. 1971. The vegetative morphology, habitat preference and water balance mechanisms of the bromeliad Tillandsia ionantha Planch. Amer. Midl. Nat. 85: 11–21. Benzing, D. H. and Davidson, E. 1979. Oligotrophic Tillandsia circinnata Schlecht. (Bromeliaceae): an assessment of its patterns of mineral allocation and reproduction. Amer. J. Bot. 66: 386–397. Benzing, D. H. and Friedman, W. E. 1981. Patterns of foliar pigmentation in Bromeliaceae and their adaptive significance. Selbyana 5: 224–240. Benzing, D. H. and Ott, D. 1981. Vegetative reduction in epiphytic Bromeliaceae and Orchidaceae: its origin and significance. Biotropica 13: 131–140. Benzing, D. H. and Pridgeon, A. 1983. Foliar trichomes of Pleurothallidinae (Orchidaceae): functional significance. Amer. J. Bot. 70: 173–180. Benzing, D. H. and Renfrow, A. 1971a. The biology of the atmospheric bromeliad Tillandisa circinnata Schlecht. I. The nutrient status of populations in south Florida. Amer. J. Bot. 58: 867–873. Benzing, D. H. and Renfrow, A. 1971b. The significance of photosynthetic efficiency to habitat preference and phylogeny among tillandsioid bromeliads. Bot. Gaz. 132: 19–30. Benzing, D. H. and Renfrow, A. 1971c. Significance of the patterns of CO2 exchange to the ecology and phylogeny of the Tillandsioideae (Bromeliaceae). Bull. Torrey Bot. Club 98: 322–327. Benzing, D. H. and Renfrow, A. 1974a. The mineral nutrition of Bromeliaceae. Bot. Gaz. 135: 281–288. Literature cited 625 Benzing, D. H. and Renfrow, A. 1974b. The nutritional status of Encyclia tampensis and Tillandsia circinnata on Taxodium ascendens and the availability of nutrients to epiphytes on this host in south Florida. Bull. Torrey Bot. Club 101: 191–197. Benzing, D. H. and Renfrow, A. 1980. The nutritional dynamics of Tillandsia circinnata in southern Florida and the origin of the ‘air plant’ strategy. Bot. Gaz. 141: 165–172. Benzing, D. H. and Seemann, J. 1978. Nutritional piracy and host decline: a new perspective on the epiphyte–host relationship. Selbyana 2: 133–148. Benzing, D. H., Arditti, J., Nyman, L. P., Temple, P. J. and Bennett, J. P. 1992. Effects of ozone and sulfur dioxide on four epiphytic bromeliads. Environ. Exp. Bot. 32: 25–32. Benzing, D. H., Derr, J. A. and Titus, J. E. 1972. The water chemistry of microcosms associated with the bromeliad Aechmea bracteata. Amer. Midl. Nat. 87: 60–70. Benzing, D. H., Givnish, T. J. and Bermudes, D. 1985. Absorptive trichomes in Brocchinia reducta (Bromeliaceae) and their evolutionary and systematic significance. Syst. Bot. 10: 81–91. Benzing, D. H., Henderson, K., Kessel, B. and Sulak, J. 1976. The absorptive capacities of bromeliad trichomes. Amer. J. Bot. 63: 1009–1014. Benzing, D. H., Seemann, J. and Renfrow, A. 1978. The foliar epidermis in Tillandsioideae (Bromeliaceae) and its role in habitat selection. Amer. J. Bot. 65: 359–365. Berlin, B., Breedlove, D. E. and Raven, P. H. 1974. Principles of Tzeltal Plant Classification: An Introduction to the Botanical Ethnography of a MayanSpeaking People of Highland Chiapas. New York: Academic Press. Bermudes, D. and Benzing, D. H. 1989. Fungi in Neotropical epiphyte roots. Biosystems 23: 65–73. Bermudes, D. and Benzing, D. H. 1991. Nitrogen fixation in association with Ecuadorian bromeliads. J. Trop. Ecol. 7: 531–536. Bernardello, L. M., Galetto, L. and Juliani, H. R. 1991. Floral nectar, nectary structure and pollinators in some Argentinian Bromeliaceae. Ann. Bot. 67: 401–411. Beutelspacher, C. R. 1972. Some observations on the Lepidoptera of bromeliads. J. Lep. Soc. 26: 133–137. Biebl, R. 1964. Zum Wasserhaushalt von Tillandsia recurvata L. und Tillandsia usneoides L. auf Puerto Rico. Protoplasma 58: 345–368. Billings, F. H. 1904. A study of Tillandsia usneoides. Bot. Gaz. 38: 99–121. Blakesley, D. and Powell, D. 1992. The UK trade in Tillandsia. TRAFFIC Bull. 13: 38–41. Böhme, S. 1988. Bromelienstudien III. Vergleichende Untersuchungen zu Bau, Lage und Systematischer Verwertbarkeit der Septalnektarien von Bromeliaceen. Trop. subtrop. Pflanzenwelt 62: 86–89. Bokermann, W. C. A. 1978. Observacões sobre hâbitos alimentares do gravião Geranospiza caerulescens (Vieillot, 1817) (Aves, Accipitridae). Rev. Bras. Biol. 38: 715–720. Borchert, R. 1966. Innere Wurzeln als Festigungselement der epiphytischen Bromeliacee Tillandsia incarnata H.B.K. Ber. Deut. Bot. Ges. 79: 253–258. Boresch, K. 1908. Über Gummifluß bei Bromeliaceen nebst Beiträgen zu ihrer Anatomie. Sitzungsber. kaiserl. Akaad. der Wiss. Wein (Math.-nat. Kl.) 117: 1060–1080. 626 Literature cited Borland, A. M. and Griffiths, H. 1989. The regulation of citric acid accumulation and carbon recycling during CAM in Ananas comosus. J. Exp. Bot. 40: 53–60. Borman, M. B. 1976. Vocabulario Cofan. Quito: Instituto Linguistico de Verano. Bradshaw, W. E. 1983. Interaction between the mosquito Wyeomyia smithii, the midge Metriocnemus knabi, and their carnivorous host Sarracenia purpurea. In Phytotelmata: Terrestrial Plants as Hosts for Aquatic Insect Communities, eds. J. H. Frank and L. P. Lounibos, pp. 161–189. Medford, NJ: Plexus. Bradshaw, W. E. and Holzapel, C. M. 1984. Seasonal development of tree-hole mosquitoes (Diptera: Culicidae) and chaoborids in relation to weather and predation. J. Med. Entomol. 21: 366–378. Brewbaker, J. C. and Gorrez, D. D. 1987. Genetics of self-incompatibility in the monocot genera Ananas (pineapple) and Gasteria. Amer. J. Bot. 54: 611–615. Brighigna, L. 1992. Role of nitrogen fixing bacterial microflora in the epiphytism of Tillandsia (Bromeliaceae). Amer. J. Bot. 79: 723–727. Brighigna, L., Cecchi Fiordi, A. and Palandri, M. R. 1990. Structural comparisons between free and anchored roots in Tillandsia (Bromeliaceae) species. Caryologia 43: 27–42. Brighigna, L., Palandri, M. R., Giuffrida, M., Macchi, C. and Trni, G. 1988. Ultrastructural features of the Tillandsia usneoides L. absorbing trichome during conditions of moisture and aridity. Caryologia 41: 111–129. Brokaw, N. V. L. 1983. Groundlayer dominance and apparent inhibition by Aechmea magdalenae (Bromeliaceae) in a tropical forest. Trop. Ecol. 24: 194–200. Brown, G. K. and Gilmartin, A. J. 1984. Stigma structure and variation in Bromeliaceae – neglected taxonomic characters. Brittonia 36: 364–374. Brown, G. K. and Gilmartin, A. J. 1986. Chromosomes of the Bromeliaceae. Selbyana 9: 88–93. Brown, G. K. and Gilmartin, A. J. 1988. Comparative ontogeny of bromeliaceous stigmas. In Aspects of Floral Development, eds. P. Leins, S. Tucker, P. K. Endress and C. Erbar, pp. 191–204. Berlin, Stuttgart: J. Cramer. Brown, G. K. and Gilmartin, A. J. 1989a. Chromosome numbers in Bromeliaceae. Amer. J. Bot. 76: 657–665. Brown, G. K. and Gilmartin, A. J. 1989b. Stigma types in Bromeliaceae – a systematic survey. Syst. Bot. 14: 110–132. Brown, G. K. and Palací, C. A. 1997. Chromosome numbers in Bromeliaceae. Selbyana 18: 85–88. Brown, G. K. and Terry, R. G. 1992. Petal appendages in Bromeliaceae. Amer. J. Bot. 79: 1051–1071. Brücher, H. 1989. Useful Plants of Neotropical Origin and their Wild Relatives. New York: Springer-Verlag. Burlage, H. M. 1968. Index of Plants of Texas with Reputed Medicinal and Poisonous Properties. Austin, TX: Henry M. Burlage. Burt, K. and Benzing, D. H. 1969. The absorption of nutrients by leaves and roots in Billbergia chloristica. Bromeliad Soc. Bull. 19: 5–10. Burt-Utley, K. and Utley, J. F. 1980. Phytogeography, physiological ecology and the Costa Rican genera of Bromeliaceae. J. Bromeliad Soc. 30: 158–170. Bush, S. P. and Beach, J. H. 1995. Breeding systems of epiphytes in tropical montane wet forest. Selbyana 16: 155–158. Calasans, C. F. and Malm, O. 1994. Using Tillandsia usneoides to monitor air pollution by mercury. Bromélia 1: 7–10. Literature cited 627 Caldiz, D. O. and Beltrano, J. 1989. Control of the epiphytic weeds Tillandsia recurvata and Tillandsia aeranthos with simazine. For. Ecol. Manage. 28: 153–159. Calver, F. K., Alaniz, J. R. and Caldiz, D. O. 1983. Tillandsia spp.: epiphytic weeds of trees and bushes. For. Ecol. Manage. 6: 367–372. Calvert, A. M. and Calvert, P. P. 1917. A Year of Costa Rican Natural History. New York: Macmillan. Carcuccio, F. T., Baldwin, J. and Geidel, G. 1975. The use of Spanish moss in identifying the source of atmospherically dispersed nickel particulate matter near Sumter, South Carolina. Geol. Soc. Amer. Abstracts with Program 7: 1021–1022. Carnal, N. W. and Black, C. C. 1989. Soluble sugars as the carbohydrate reserve for CAM in pineapple leaves: implications for the role of pyrophosphate 6phosphofructokinase in glycolysis. Plant Physiol. 90: 91–100. Catling, P. M. 1995. Evidence of partitioning of Belizean ant nest substrates by a characteristic flora. Biotropica 27: 535–537. Catling, P. M. 1997. Influence of aerial Azteca nests on the epiphyte community of some Belizean orange orchards. Biotropica 29: 237–242. Catling, P. M. and Lefkovitch, L. P. 1989. Associations of vascular epiphytes in a Guatemalan cloud forest. Biotropica 21: 35–40. Catling, P. M., Brownell, V. R. and Lefkovitch, L. P. 1986. Epiphytic orchids in a Belizean grapefruit orchid: distribution, colonization, and association. Lindleyana 1: 194–202. Cave, R. D. 1997. Admontia sp., a potential biological control agent of Metamasius callizona in Florida. J. Bromeliad Soc. 47: 244–249. Cecchi Fiordi, A. and Palandri, M. R. 1982. Anatomic and ultrastructural study of the septal nectary in some Tillandsia (Bromeliaceae) species. Caryologia 35: 477–489. Center for Plant Conservation 1991. See www.mobot.org/CPC/ Chapin, F. S. 1993. Preferential use of organic nitrogen for growth by a nonmycorrhizal arctic sedge. Nature 361: 150–153. Chávez Velásquez, N. A. 1977. La Materia Medica in el Incanato. Lima: Editorial Mejia Baca. Cheadle, V. I. 1953. Independent origin of vessels in the monocotyledons and dicotyledons. Phytomorphology 3: 23–44. Cheadle, V. I. 1955. Conducting elements in the xylem of Bromeliaceae. Bromeliad Soc. Bull. 5: 3–7. Chedier, L. M. and Kaplan, M. A. C. 1996. Chemical ecology of three species of Bromeliaceae. Bromélia 3: 25–37. Chodat, R. and Visher, W. 1916. Broméliacées. In La Végétation du Paraguay, ed. R. Chodat, p. 219. Bull. Soc. Bot. Genève (2me Série) 8. Clark, K. L., Nadkarni, N. M. and Gholz, H. L. 1998. Growth, net production, litter decomposition, and net nitrogen accumulation by epiphytic bryophytes in a tropical montane forest. Biotropica 30: 12–23. Clark, L. G., Gaut, B. S., Duvall, M. R. and Clegg, M. T. 1993. Phylogenetic relationships of the Bromeliiflorae–Commeliniflorae–Zingiberiflorae complex of monocots based on rbcL sequence comparisons. Ann. Missouri Bot. Gard. 80: 987–988. Clark, T. F. 1965. Plant fibers in the paper industry. Econ. Bot. 19: 394–405. Clark, W. D. and Clegg, M. T. 1990. Phylogenetic comparisons among rbcL sequences in the Bromeliaceae. Amer. J. Bot. 77: 115 (Abstract). 628 Literature cited Clarkson, D. T., Kuiper, P. J. C. and Lüttge, U. 1986. Mineral nutrition: sources of nutrients for land plants from outside the pedosphere. In Progress in Botany, vol. 48, pp. 81–96. Berlin: Springer-Verlag. Cobley, L. S. 1976. An Introduction to the Botany of Tropical Crops, 2nd edn, revised by W. M. Steele. New York: Longman Inc. Cockburn, W., Goh, C. J. and Avadhani, P. N. 1985. Photosynthetic carbon assimilation in a shootless orchid, Chiloschista usneoides (DON)LDL: a variation on crassulacean acid metabolism. Plant Physiol. 77: 83–86. Cole, L. C. 1954. The population consequences of life history phenomena. Q. Rev. Biol. 29: 103–137. Connor, J. J. and Shacklette, H. T. 1984. Factor analysis of the chemistry of Spanish moss. Open File Report 84–174. Denver, CO: United States Geological Survey. Cook, M. T. 1926. Epiphytic orchids, a serious pest on citrus trees. J. Dept. Agric. Puerto Rico 10: 5–9. Cote, F. X., Andre, M., Folliot, M., Massimino, D. and Daguenet, A. 1989. CO2 and O2 exchanges in the CAM plant Ananas comosus (L.) Merr. Plant Physiol. 89: 61–68. Coutinho, L. M. 1963. Algumas informacões sóbre a ocurrência mata pluvial. [Boletin no 288, Faculdade de Filosofia, Ciênciare Letras da Universidade de São Paulo.] Botânica 20: 81–98. Coxson, D. S. and Nadkarni, N. M. 1995. Ecological roles of epiphytes in nutrient cycles of forest ecosystems. In Forest Canopies, eds. M. D. Lowman and N. M. Nadkarni, pp. 455–495. San Diego, CA: Academic Press. Craighead, F. E. 1963. Orchids and Other Air Plants. Miami, FL: University of Miami Press. Crayn, D. M., Terry, R. G., Smith, J. A. C. and Winter, K. 1999. Molecular systematic investigations in Pitcairnioideae (Bromeliaceae) as a basis for understanding the evolution of crassulacean acid metabolism (CAM). In Systematics and Evolution of Monocots – Proceedings of the Monocots II Conference, eds. K. Wilson and D. Morrison. Collingwood, Australia: CSIRO Publishing, in press. Cronquist, A. 1981. An Integrated System of Classification of Flowering Plants. New York: Columbia University Press. Cummins, K. W., Wilzback, D. M., Gates, D. M., Perry, J. B. and Taliaferro, B. W. 1989. Shredders and riparian vegetation. Bioscience 39: 24–30. Dahlgren, R. and Rasmussen, F. N. 1983. Monocotyledon evolution: characters and phytogenetic estimation. Evol. Biol. 16: 255–395. Dahlgren, R., Clifford, T. H. and Yeo, P. F. 1985. The Families of the Monocotyledons: Structure, Evolution and Taxonomy. Berlin: SpringerVerlag. Damuth, J. E. and Fairbridge, R. W. 1970. Equatorial Atlantic deep-sea arkosic sands and ice-age aridity in tropical South America. Bull. Geol. Soc. Amer. 81: 189–206. Davidson, D. W. 1988. Ecological studies of neotropical ant-gardens. Ecology 69: 1138–1152. Davidson, D. W. and Epstein, W. W. 1989. Epiphytic associations with ants. In Vascular Plants as Epiphytes, ed. U. Lüttge, pp. 200–233. Berlin: SpringerVerlag. Davidson, D. W. and McKey, D. 1993. The evolutionary ecology of symbiotic ant–plant relationships. J. Hym. Res. 2: 13–83. Literature cited 629 Davidson, D. W., Seidel, J. L. and Epstein, W. W. 1990. Neotropical ant gardens II. Bioassays of seed compounds. J. Chem. Ecol. 16: 2993–3013. Davidson, E. H. 1969. Plant quarantine regulation governing the importation of bromeliad plants. Bromeliad Soc. Bull. 19: 105–108. Davis, G. L. 1966. Systematic Embryology of the Angiosperms. New York: Wiley. Davis, J. I. 1995. A phylogenetic structure for the monocotyledons, as inferred from chloroplast DNA restriction site variation, and a comparison of measures of clade support. Syst. Bot. 20: 503–527. De Santo, A. V., Alfani, A. and De Luca, P. 1976. Water vapor uptake from the atmosphere by some Tillandsia species. Ann. Bot. 40: 391–394. Dejean, A. 1990. Influence de l’environnement préimaginal et précoce dans le choix du site de nidification de Pachycondyla (5Neoponera) villosa (Fabr.) (Formicidae-Ponerinae). Behavioural Process 21: 107–125. Dejean, A. and Olmsted, I. 1997. Ecological studies on Aechmea bracteata (Swartz) (Bromeliaceae). J. Nat. Hist. 31: 1313–1334. Dejean, A., Olmsted, I. and Camal, J. F. 1992. Interactions between Atta cephalotes and arboreal ants in the biosphere reserve Sian Ka’an (Quintana Roo, Mexico): efficient protection of the trees (Hymenoptera: Formicidae). Sociobiology 20: 57–76. Dejean, A., Olmsted, I. and Snelling, R. R. 1995. Tree–epiphyte–ant relationships in the low inundated forests of Sian Ka’an Reserve, Quintana Roo, Mexico. Biotropica 27: 57–70. Delaney, K. 1994. Decline of Highlands County’s epiphytes. Newsletter of the Dicerandra Chapter, Florida Native Plant Society Highland County 4: 6–7. DeVries, P. J. 1997. The Butterflies of Costa Rica and their Natural History, vol. II, Riodinidae. Chichester, NY: Princeton University Press. Diesel, R. 1992. Maternal care in the bromeliad crab Metopaulias depressus: protection of larvae from predation by damselfly nymphs. Anim. Behav. 43: 308–312. Diesel, R. and Schuh, M. 1993. Maternal care in the bromeliad crab Metopaulias depressus (Decapoda): maintaining oxygen, pH, and calcium levels optimal for larvae. Behav. Ecol. Sociobiol. 32: 11–15. Dimmitt, M. A. 1984. Tillandsia culture: growing large specimens. J. Bromeliad Soc. 34: 245–251. Dimmitt, M. A. 1985. Intraspecific variation in Tillandsia: selecting superior forms. J. Bromeliad Soc. 19: 101–103. Dimmitt, M. A. 1987. The Bromeliad Society, Inc. conservation goals and code of conduct for growers and collectors. J. Bromeliad Soc. 37: 207–209. Dimmitt, M. A. 1989a. Letter in: Kew Mag. 6: 184–188. Dimmitt, M. A. 1989b. Endangered species: another view. J. Bromeliad Soc. 39: 150–151, 175. Dimmitt, M. A. 1990. Growing atmospheric tillandsias from seed. J. Bromeliad Soc. 40: 17–20, 29–30. Dolzmann, P. 1964. Elektronenmikroskopische Untersuchungen an den Saughaaren von Tillandsia usneoides (Bromeliaceae). I. Feinstruktur der Kuppelzelle. Planta 60: 461–472. Dolzmann, P. 1965. Elektronenmikroskopische Untersuchungen an den Saughaaren von Tillandsia usneoides (Bromeliaceae). II. Einige Beobachtungen zur Feinstruktur der Plasmodesmen. Planta 64: 76–80. Downs, R. J. 1963. Photocontrol of germination of seeds of the Bromeliaceae. Phyton 21: 1–6. 630 Literature cited Downs, R. J. 1974. Anatomy and physiology. In Pitcairnioideae (Bromeliaceae). Flora Neotropica Monograph 14, Part 1, eds. L. B. Smith and R. J. Downs, pp. 2–28. New York: Hafner Press. Duke, J. A. 1986. Ishmian Ethnobotanical Dictionary. Jodhpur, India: Scientific Publishers. Duvall, M. R., Clegg, M. T., Chase, M. W. et al. 1993. Phylogenetic hypotheses for the monocotyledons constructed from rbcL sequences data. Ann. Missouri Bot. Gard. 80: 607–619. Edwards, P. J. and Grubb, P. J. 1977. Studies of mineral cycling in a montane rain forest in New Guinea. I. The distribution of organic matter in the vegetation and soil. J. Ecol. 65: 943–969. Ehler, N. 1977. Bromelienstudien II: Neue Untersuchungen zur Entwicklung, Struktur und Funktion der Bromelien-Trichome. Trop. subtrop. Pflanzenwelt 20: 1–40. Ehler, N. and Schill, R. 1973. Die Pollenmorphologie der Bromeliaceae. Pollen et Spores 15: 13–45. Ehlers, R. 1991. A new Tillandsia species: Tillandsia sueae. J. Bromeliad Soc. 41: 208–213. Ekern, P. C. 1965. Evapotranspiration of pineapple in Hawaii. Plant Physiol. 40: 736–739. Erdtman, G. and Praglowski, J. 1974. A note on pollen morphology. In Pitcairnioideae (Bromeliaceae). Flora Neotropica Monograph 14, Part 1, eds. L. B. Smith and R. J. Downs, pp. 28–33. New York: Hafner Press. Eshbaugh, W. H. 1987. Plant–ant relationships and interactions: Tillandsia and Crematogaster. In Proceedings of the Second Symposium of Botany, Bahamas, ed. R. R. Smith, pp. 7–11. San Salvador: CCFL Bahamian Field Station. Evans, T. M. and Brown, G. K. 1989a. Plicate staminal filaments in Tillandsia subgenus Anoplophytum (Bromeliaceae). Amer. J. Bot. 76: 1478–1485. Evans, T. M. and Brown, G. K. 1989b. Stomata in Tillandsia bryoides. J. Bromeliad Soc. 39: 58–61. Fahn, A. 1979. Secretory Tissues in Plants. London: Academic Press. Farquhar, C. D. and Sharkey, T. D. 1982. Stomatal conductance and photosynthesis. Ann. Rev. Plant Physiol. 33: 317–345. Fetene, M. and Lüttge, U. 1991. Environmental influences on carbon cycling in a terrestrial CAM bromeliad, Bromelia humilis Jacq. J. Exp. Bot. 42: 25–31. Fetene, M., Lee, H. S. J. and Lüttge, U. 1990. Photosynthetic acclimation in a terrestrial CAM bromeliad, Bromelia humilis Jacq. New Phytol. 114: 399–406. Fialho, R. F. 1990. Seed dispersal by a lizard and a tree frog – effect of dispersal site on seed survivorship. Biotropica 22: 423–424. Field, C. and Mooney, H. A. 1986. The photosynthesis–-nitrogen relationship in wild plants. In On the Economy of Plant Form and Function, ed. T. J. Givnish, pp. 25–55. Cambridge: Cambridge University Press. Fischer, E. A. and Araujo, A. C. 1995. Spatial organization of a bromeliad community in the Atlantic rain forest, southeastern Brazil. J. Trop. Ecol. 11: 559–567. Fischer, E. A., Duarte, L. F. L. and Araujo, A. C. 1997. Consumption of bromeliad flowers by the crab Metasesarma rubripes in a Brazilian coastal forest. Crustaceana 70: 118–123. Fish, D. 1976. Structure and composition of the aquatic invertebrate community inhabiting bromeliads in south Florida and the discovery of an insectivorous bromeliad. PhD thesis. Gainesville, FL: University of Florida. Literature cited 631 Fish, D. 1983. Phytotelmata: flora and fauna. In Phytotelmata: Terrestrial Plants as Hosts for Aquatic Insect Communities, eds. J. H. Frank and L. P. Lounibos, pp. 1–27. Medford, NJ: Plexus. Fish, D. and Hall, D. W. 1978. Succession and stratification of aquatic insects inhabiting the leaves of the insectivorous pitcher plant, Sarracenia purpurea. Amer. Midl. Nat. 99: 172–183. Fittkau, E. J. and Klinge, H. 1973. On biomass and trophic structure of the central Amazonian rain forest ecosystem. Biotropica 5: 2–14. Fogg, G. E., Stewart, W. D. P. and Walsby, A. E. 1973. The Blue-Green Algae. New York: Academic Press. Fontoura, T. 1995. Distribution patterns of five Bromeliaceae genera in Atlantic rain forest, Rio de Janeiro State, Brazil. Selbyana 16: 79–93. Fontoura, T., Costa, A. and Went, T. 1991. Preliminary checklist of the Bromeliaceae of Rio de Janeiro State, Brazil. Selbyana 12: 5–45. Fragoso, C. and Rojas-Fernández, P. 1996. Earthworms inhabiting bromeliads in Mexican tropical rain forests: ecological and historical determinants. J. Trop. Ecol. 12: 729–734. Franco, A. C., Ball, E. and Lüttge, U. 1992. Differential effects of drought and light levels on accumulation of citric and malic acids during CAM in Clusia. Plant Cell Environ. 15: 821–829. Frank, J. H. 1983. Bromeliad phytotelmata and their biota, especially mosquitoes. In Phytotelmata: Terrestrial Plants as Hosts for Aquatic Insect Communities, eds. J. H. Frank and L. P. Lounibos, pp. 101–128. Medford, NJ: Plexus. Frank, J. H. 1985. Use of an artificial bromeliad to show the importance of color value in restricting colonization of bromeliads by Aedes aegypti and Culex quinquefasciatus. J. Amer. Mosquito Control Assoc. 1: 28–32. Frank, J. H. 1986. Bromeliads as oviposition sites for Wyeomyia mosquitoes: form and color influence behavior. Fla. Entomol. 69: 728–742. Frank, J. H. and Curtis, G. A. 1977a. On the bionomics of bromeliad-inhabiting mosquitoes. III. The probable strategy of larval feeding in Wyeomyia vanduzeei and W. medioalbipes. Mosquito News 37: 200–206. Frank, J. H. and Curtis, G. A. 1977b. On the bionomics of bromeliad-inhabiting mosquitoes. IV. Egg mortality of Wyeomyia vanduzeei caused by rainfall. Mosquito News 37: 239–245. Frank, J. H. and Curtis, G. A. 1981a. Bionomics of the bromeliad-inhabiting mosquito Wyeomyia vanduzeei and its nursery plant Tillandsia utriculata. Fla. Entomol. 64: 491–506. Frank, J. H. and Curtis, G. A. 1981b. On the bionomics of bromeliad-inhabiting mosquitoes. A review of the bromeliad-inhabiting species. J. Fla. AntiMosquito Assoc. 52: 4–23. Frank, J. H. and Lounibos, L. P. 1987. Phytotelmata: swamps or islands? Fla. Entomol. 70: 14–20. Frank, J. H. and O’Meara, G. F. 1984. The bromeliad Catopsis berteroniana traps terrestrial arthropods but harbors Wyeomyia larvae (Diptera: Culicidae). Fla. Entomol. 67: 418–424. Frank, J. H. and O’Meara, G. F. 1985. Influence of micro- and macrohabitat on distribution of some bromeliad inhabiting mosquitoes. Entomol. Exp. Appl. 37: 169–174. Frank, J. H. and Thomas, M. C. 1994. Metamasius callizona (chevrolat) (Coleoptera: Curculionidae), an immigrant pest, destroys bromeliads in Florida. Canadian Entomologist 126: 673–682. Frank, J. H., Curtis, G. A. and Evans, H. T. 1977. On the bionomics of 632 Literature cited bromeliad-inhabiting mosquitoes. II. The relationship of bromeliad size to the number of immature Wyeomyia vanduzeei and W. medioalbipes. Mosquito News 37: 180–192. Frank, J. H., Curtis, G. A. and O’Meara, G. F. 1984. The bionomics of bromeliad inhabiting mosquitoes. X. Toxorhynchites rutilus rutilus as a predator of Wyeomyia vanduzeei (Diptera: Culicidae). J. Med. Entomol. 21: 149–158. Franquemont, C., Plowman, T., Franquemont, E. et al. 1990. The ethnobotany of Chinchero, an Andean community in southern Peru. Fieldiana Botany New Series 24: 1–126. Freeman, C. E., Worthington, R. D. and Corral, R. D. 1985. Some floral nectarsugar compositions from Durango and Sinaloa, Mexico. Biotropica 17: 309–313. Freeze, C. H. and Oppenheimer, J. R. 1981. The capuchin monkeys, genus Cebus. In Ecology and Behavior of Neotropical Primates, vol. 1, eds. A. F. ColmbraFilho and R. A. Mittermeier, pp. 331–390. Rio de Janeiro: Academy of Brazilian Science. Frei, Sister J. K. and Dodson, C. H. 1972. The chemical effect of certain bark substrates on the germination and early growth of epiphytic orchids. Bull. Torrey. Bot. Club 99: 301–307. Freiberg, M. 1996. Spatial distribution of vascular epiphytes on the emergent canopy trees in French Guiana. Biotropica 28: 345–355. Friedman, W. E. 1992. Evidence of a pre-angiosperm origin of endosperm: implications for the evolution of flowering plants. Science 255: 336–339. Friend, D. J. C. and Lydon, J. 1979. Effects of daylength on flowering, growth, and CAM of pineapple (Ananas comosus (L.) Merrill). Bot. Gaz. 140: 280–283. Frölich, D. and Barthlott, W. 1988. Mikromorphologie der epicuticularen Wachse und das System der Monokotylen. Trop. subtrop. Pflanzenwelt 63: 1–135. Galetto, L. and Bernardello, L. M. 1992. Extrafloral nectaries that attract ants in Bromeliaceae: structure and nectar composition. Can. J. Bot. 70: 1101–1106. Garcia-Franco, J. G. and Peters, C. M. 1987. Patron espacial y abundancia de Tillandsia spp. a traves de un gradiente altitudinal en los altos de Chiapas, Mexico. Brenesia 27: 35–45. Garcia-Franco, J. G. and Rico-Gray, V. 1988. Experiments on seed dispersal and deposition patterns of epiphytes – the case of Tillandsia deppeana Steudel (Bromeliaceae). Phytologia 65: 73–78. Garcia-Franco, J. G. and Rico-Gray, V. 1991. Biologia reproductiva de Tillandsia deppeana Steudel (Bromeliaceae) en Veracruz, Mexico. Brenesia 35: 61–79. Garcia-Franco, J. G., Rico-Gray, V. and Zayas, O. 1991. Seed and seedling predation of Bromelia pinguin L. by the red land crab Gecarcinus lateralis in Veracruz, Mexico. Biotropica 23: 96–97. Gardner, C. S. 1982. A systematic study of Tillandsia subgenus Tillandsia. PhD thesis. College Station, TX: Texas A & M University. Gardner, C. S. 1984. Natural hybridization in Tillandsia subgenus Tillandsia. Selbyana 7: 380–393. Gardner, C. S. 1986a. Inferences about pollination in Tillandsia (Bromeliaceae). Selbyana 9: 76–87. Gardner, C. S. 1986b. Preliminary classification of Tillandsia based on floral characters. Selbyana 9: 130–146. Garth, R. E. 1964. The ecology of Spanish moss (Tillandsia usneoides): its growth and distribution. Ecology 45: 470–481. Gentry, A. H. 1974. Flowering phenology and diversity in tropical Bignoniaceae. Biotropica 6: 64–68. Literature cited 633 Gentry, A. H. 1982. Neotropical floristic diversity: phytogeographical connections between Central and South America, Pleistocene climatic fluctuations, or an accident of the Andean orogeny? Ann. Missouri Bot. Gard. 69: 557–593. Gentry, A. H. and Dodson, C. H. 1987. Diversity and biogeography of neotropical vascular epiphytes. Ann. Missouri Bot. Gard. 74: 205–233. Gentry, A. H. and Parker, T. 1992. Rapid Assessment Program Working Paper Two. Washington, D.C.: Conservation International. Germer, B. 1982. Protozoan exclusion in the Bromeliaceae. J. Bromeliad Soc. 32: 154–155. Gilmartin, A. J. 1972. The Bromeliaceae of Ecuador. Phanerog. Monogr. IV. Lehre: J. Cramer. Gilmartin, A. J. 1973. Transandean distributions of Bromeliaceae in Ecuador. Ecology 54: 1389–1393. Gilmartin, A. J. 1983. Evolution of mesic and xeric habits in Tillandsia and Vriesea (Bromeliaceae). Syst. Bot. 8: 233–242. Gilmartin, A. J. and Brown, G. K. 1985. Cleistogamy in Tillandsia capillaris (Bromeliaceae). Biotropica 17: 256–259. Gilmartin, A. J. and Brown, G. K. 1986. Cladistic tests of hypotheses concerning the evolution of xerophytes and mesophytes in Tillandsia subg. Phytarrhiza (Bromeliaceae). Amer. J. Bot. 73: 387–397. Gilmartin, A. J. and Brown, G. K. 1987. Bromeliales, related monocots and resolution of relationships among bromeliad subfamilies. Syst. Bot. 12: 493–500. Gilmartin, A. J., Brown, G. K., Varadarajan, G. S. and Neighbours, M. 1989. Status of Glomeropitcairnia within evolutionary history of Bromeliaceae. Syst. Bot. 14: 339–348. Givnish, T. J., Burkhardt, E. L., Happel, R. E. and Weintraub, J. D. 1984. Carnivory in the bromeliad Brocchinia reducta with a cost/benefit model for the general restriction of carnivorous plants to sunny, moist, nutrient-poor habitats. Amer. Nat. 124: 479–497. Givnish, T. J., Evans, T. M. and Sytsma, K. J. 1998. Polyphyly and convergent morphological evolution in Commelinales and Commelinidae: evidence from rbcL sequence data. Abstracts, Monocots II Conference, Sydney, Australia, pp. 22–23. Sydney: Royal Botanical Gardens. Givnish, T. J., Sytsma, K. J., Smith, J. F., Hahn, W. J., Benzing, D. H. and Burkhardt, E. M. 1997. Molecular evolution and adaptive radiation in Brocchinia (Bromeliaceae: Pitcairnioideae) atop tepuis of the Guayana Shield. In Molecular Evolution and Adaptive Radiation, eds. T. J. Givnish and K. J. Sytsma, pp. 259–311. Cambridge: Cambridge University Press. Goldberg, A. 1989. Classification, evolution, and phylogeny of the families of Monocotyledons. Smithson. Contr. Bot. 71: 50–53. Goldstein, G., Meinzer, F. and Monasterio, M. 1984. The role of capacitance in the water balance of Andean giant rosette species. Plant Cell Environ. 7: 179–186. Golley, F. B., Richardson, T. and Clements, R. G. 1978. Elemental concentrations in tropical forests and soils in northwestern Colombia. Biotropica 10: 144–151. Gómez, L. D. 1972. Karatophyllum bromelioides L. D. Gomez (Bromeliaceae), nov. gen. et. sp., del Terciario Medio de Costa Rica. Rev. Biol. Trop. 20: 221–229. Gómez, M. A. and Winkler, S. 1991. Bromelias in manglares del Pacifico de Guatemala. Rev. Biol. Trop. 39: 207–214. 634 Literature cited Gonzales, J. M., Jaffe, K. and Michelangeli, F. 1991. Competition for prey between the carnivorous Bromeliaceae Brocchinia reducta and Sarraceniaceae Heliamphora nutans. Biotropica 23: 602–604. Goodspeed, T. H. 1961. Plant Hunters in the Andes. Berkeley, CA: University of California Press. Gortan, G. 1991. Narbenformen bei Bromeliaceen: Variationsmöglichkeiten und Überlegungen zu systematisch-taxonomischen Korrelationen. MSc thesis. Wien: Universität Wien. Graham, A. 1997. Neotropical plant dynamics during the Cenozoic – diversification and the ordering of evolutionary and speciation processes. Syst. Bot. 22: 139–150. Grant, J. R. 1993a. New combinations in Mezobromelia and Racinaea (Bromeliaceae: Tillandsioideae). Phytologia 74: 428–430. Grant, J. R. 1993b. True tillandsias misplaced in Vriesea (Bromeliaceae: Tillandsioideae). Phytologia 75: 170–175. Grant, J. R. 1994a. Notes on the coastal Costa Rican endemic Pitcairnia halophila. J. Bromeliad Soc. 44: 170–172. Grant, J. R. 1994b. The reduction of Platyaechmea under Holpophytum and a new name in Tillandsia (Bromeliaceae). Phytologia 77: 99–101. Grant, J. R. 1995a. Addendum to ‘The resurrection of Alcantarea, and Werauhia, a new genus’ (Bromeliaceae: Tillandsioideae). Phytologia 78: 119–123. Grant, J. R. 1995b. The resurrection of Alcantarea, and Werauhia, a new genus. Trop. subtrop. Pflanzenwelt 91: 1–57 and map. Grierson, D. and Covey, S. N. 1988. The plastome and chloroplast biogenesis. In Plant Molecular Biology, 2nd edn, pp. 47–77. New York: Chapman & Hall. Griffiths, H. 1988. Carbon balance during CAM: an assessment of respiratory CO2 recycling in the epiphytic bromeliads Aechmea nudicaulis and Aechmea fendleri. Plant Cell Environ. 11: 603–611. Griffiths, H. and Smith, J. A. C. 1983. Photosynthetic pathways in the Bromeliaceae of Trinidad: relations between life-forms, habitat preference and the occurrence of CAM. Oecologia 60: 176–184. Griffiths, H., Broadmeadow, S. J., Borland, A. M. and Heatherington, C. S. 1990. Short term changes in carbon isotope discrimination identify transitions between C3 and C4 carboxylation during crassulacean acid metabolism. Planta 181: 604–610. Griffiths, H., Lüttge, U., Popp, M. et al. 1989. Ecophysiology of xerophytic and halophytic vegetation of a coastal alluvial plain in northern Venezuela. IV. Tillandsia flexuosa Sw. and Schomburgkia humboldtiana Reichb., epiphytic CAM plants. New Phytol. 111: 273–282. Griffiths, H., Lüttge, U., Simmel, H., Crook, C. E., Griffiths, N. M. and Smith, J. A. C. 1986. Comparative ecophysiology of CAM and C3 bromeliads. III. Environmental influences on CO2 assimilation and transpiration. Plant Cell Environ. 9: 385–393. Griffiths, N. M. 1992. Carbon discrimination and the integration of carbon assimilation pathways in terrestrial CAM plants. Plant Cell Environ. 15: 1051–1062. Gross, E. 1985. Polyembryony in bromeliads: a provisional note. J. Bromeliad Soc. 19: 202–205. Gross, E. 1988a. Bromelienstudien IV. Zur Morphologie der Bromeliaceen-Samen unter Berücksichtigung systematisch-taxonomischer Aspekte. Trop. subtrop. Pflanzenwelt 64: 1–215. Gross, E. 1988b. Über die Keimung der Bromeliaceen. Beitr. Biol. Pflanzen 63: 101–113. Literature cited 635 Gross, E. 1991. A portrait of Ronnbergia petersii with some remarkable features. J. Bromeliad Soc. 41: 172–177. Gross, E. 1993. Die Samen der Bromeliaceae. Die Bromelie 6: 13–18, 53–55. Grubb, P. J. and Edwards, P. J. 1982. Studies of mineral cycling in a montane rain forest in New Guinea. J. Ecol. 70: 623–648. Grubb, P. J., Lloyd, J. R., Pennington, T. D. and Whitmore, T. C. 1963. A comparison of montane and lowland rain forest in Ecuador. J. Ecol. 51: 567–601. Gschneidner, M. 1989. Die Infloreszenzen der Gattung Tillandsia (Bromeliaceae). MSc thesis. München: Universität München. Haberlandt, G. 1914. Physiological Plant Anatomy (translated from the 4th German edn). London: Macmillan and Co., Ltd. Hadley, G. 1982. Orchid mycorrhiza. In Orchid Biology – Reviews and Perspectives, vol. II, ed. J. Arditti, pp. 83–118. Ithaca, NY: Cornell University Press. Halbritter, H. 1988. Bromeliaceae: Pollenmorphologie und Systematik. Die Entwicklung des Pollens von Tillandsia sinuosa L. B. Smith. PhD thesis. Wien: Universität Wien. Halbritter, H. 1992. Morphologie und systematische Bedeutung des Pollens der Bromeliaceae. Grana 31: 197–212. Halbritter, H. 1995. Morphology of Bromeliaceae pollen. Resumos XLVI Congresso Nacional de Botânica 313. Ribeirão Preto: Universidade de São Paulo. Hall, J. 1958. How the native bromeliads took the cold in Florida. Bromeliad Soc. Bull. 8: 6–7. Hallé, F., Oldeman, R. A. A. and Tomlinson, P. B. 1978. Tropical Trees and Forests, pp. 133–250. New York: Springer-Verlag. Hallwachs, W. 1983. Bromelia pinguin and B. karatas. In Costa Rican Natural History, ed. D. H. Janzen, pp. 195–197. Chicago, IL: University of Chicago Press. Hamilton, M. B. 1992. Ex situ conservation of wild plant species: time to reassess the genetic assumptions and implications of seed banks. Conserv. Biol. 8: 39–49. Hancocks, D. 1994. Zoological gardens, arboreta, and botanical gardens: a trilogy of failure? Address to the American Association of Botanical Gardens and Arboreta Annual Conference, June 1994. Hanken, J. and Wake, D. B. 1993. Miniaturization of body size and evolutionary significance. Ann. Rev. Ecol. Syst. 24: 501–519. Harms, H. 1930. Bromeliaceae. In Die Natürlichen Pflanzenfamilien, 2nd edn, 15a, eds. A. Engler and K. Prantl, pp. 65–159. Leipzig: W. Engelmann. Harris, F. S. and Martin, C. E. 1991. Correlation between CAM-cycling and photosynthetic gas exchange in five species of Talinum (Portulacaceae). Plant Physiol. 96: 1118–1124. Harris, J. A. 1918. On the osmotic concentrations of the tissue fluids of phanerogamic epiphytes. Amer. J. Bot. 5: 490–506. Harris, L. D. 1984. The Fragmented Forest: Island Biogeography Theory and Preservation of Biotic Diversity. Chicago, IL: University of Chicago Press. Haughton, C. S. 1978. Green Immigrants: The Plants that Transformed America. New York: Harcourt Brace, Inc. Hay, J. D., de Lacerda, L. D. and Tan, A. L. 1981. Soil cations increase in a tropical sand dune ecosystem due to a terrestrial bromeliad. Ecology 62: 1392–1395. Hayward, W. 1947. Spanish moss as an economic plant. Plantlife 1: 97–98. 636 Literature cited Hazen, W. E. 1966. Analysis of spatial pattern in epiphytes. Ecology 47: 634–635. Hegnauer, R. 1963. Chemotaxonomie der Pflanzen, vol. 2, pp. 99–109. Basel, Stuttgart: Birkhäuser. Hegnauer, R. 1986. Chemotaxonomie der Pflanzen, vol. 7, pp. 591–594. Basel, Boston, Stuttgart: Birkhäuser. Hensyl, W. R. (ed.) 1989. Stedman’s Medical Dictionary, 25th edn. Baltimore, MD: Williams and Wilkins. Heslop-Harrison, Y. and Shivanna, K. R. 1977. The receptive surface of the Angiosperm stigma. Ann. Bot. 41: 1233–1258. Hess, M. 1991. Ultrastructure of organelles during microsporogenesis in Tillandsia pallidoflavens (Bromeliaceae). Pl. Syst. Evol. 176: 63–74. Heywood, V. H. 1992. Botanic gardens and the aftermath of UNCED. Botanic Gardens Conservation News 1: 3. Hietz, P. 1997. Population dynamics of epiphytes in a Mexican humid montane forest. J. Ecol. 85: 767–775. Hietz, P. and Hietz-Seifert, U. 1995a. Intra- and interspecific relations within an epiphyte community in a Mexican humid montane forest. Selbyana 16: 135–140. Hietz, P. and Hietz-Seifert, U. 1995b. Composition and ecology of vascular epiphyte communities along an altitudinal gradient in central Vera Cruz, Mexico. J. Veg. Sci. 6: 487–498. Hietz, P. and Hietz-Seifert, U. 1995c. Structure and ecology of epiphyte communities of a cloud forest in central Vera Cruz, Mexico. J. Veg. Sci. 6: 719–728. Hietz-Seifert, U., Hietz, P. and Guevara, S. 1996. Epiphyte vegetation and diversity on remnant trees after forest clearance in southern Veracruz, Mexico. Biol. Conserv. 75: 103–111. Hoehne, F. C. 1951. Treetop aquariums. Bromeliad Soc. Bull. 1: 45–49. Hofstede, G. M., Wolf, J. D. H. and Benzing, D. H. 1993. Epiphytic biomass and nutrient status of a Colombian upper montane rain forest. Selbyana 14: 37–45. Holcomb, G. E. 1995. Ball moss: an emerging pest on landscape trees in Baton Rouge. Proc. Louisiana Acad. Sci. 58: 11–17. Holst, B. K. 1994. Checklist of Venezuelan Bromeliaceae with notes on species distribution by state and levels of endemism. Selbyana 15: 132–149. Holst, B. K. 1996. New species of Lindmania and Navia from southern Venezuela. J. Bromeliad Soc. 46: 156–168. Horres, R. 1995. Untersuchungen zur Blattsukkulenz bei Bromeliaceae. MSc thesis. Frankfurt am Main: Universität Frankfurt. Huber, H. 1977. The treatment of monocotyledons in an evolutionary system of classification. Syst. Ecol. Suppl. 1: 285–298. Huber, H. 1991. Angiospermen. Stuttgart, New York: G. Fischer. Huxley, C. R. 1980. Symbiosis between ants and epiphytes. Biol. Rev. 55: 321–340. Ibisch, P. L. 1996. Neotropische Epiphytendiversität-das Beispiel Bolivien. Wiehl: Martina Galunder-Verlag. Ibisch, P. L., Boegner, A., Nieder, J. and Barthlott, W. 1996. How diverse are neotropical epiphytes? An analysis based on the ‘Catalogue of flowering plants and gymnosperms of Peru’. Ecotropica 2: 13–28. Ibisch, P. L., Gross, E., Rauer, G. and Rudolph, D. 1997. On the diversity and biogeography of the genus Fosterella L. B. Smith (Bromeliaceae) with the description of a new species from eastern Bolivia. J. Bromeliad Soc. 47: 211–217. Literature cited 637 Ihlenfeldt, H. D. 1994. Diversification in an arid world: the Mesembryanthemaceae. Ann. Rev. Ecol. Syst. 25: 521–546. Ingram, S. W., Ingram-Ferrel, K. and Nadkarni, N. M. 1996. Floristic composition of vascular epiphytes in a Neotropical cloud forest, Monte Verde, Costa Rica. Selbyana 17: 88–103. Izquierdo, L. Y. 1995. Estructura y variacion genetica en cuatro especies de Aechmea (Bromeliaceae) en México. PhD thesis. Mexico City: Universidad National Autonoma de Mexico. Jaffe, K., Michelangellig, F., Gonzalez, J. M., Miras, B. and Christine, M. 1992. Carnivory in pitcher plants of the genus Heliamphora (Sarraceniaceae). New Phytol. 122: 733–744. Janetzky, W. and Vareschi, E. 1993. Phytotelmata in bromeliads as microhabitats for limnetic organisms. In Animal–Plant Interactions in Tropical Environments, eds. W. Barthlott, C. M. Naumann, K. Schmidt-Loske and K. L. Schuchmann, pp. 199–209. Bonn: Zoologisches Forschungsinstitut und Museum Alexander Koenig. Janos, D. P. 1993. Vesicular-arbuscular mycorrhizae of epiphytes. Mycorrhiza 4: 1–4. Janzen, D. H. 1966. Coevolution of mutualism between ants and acacias in Central America. Evolution 20: 249–275. Jaramillo, M. A. and Cavelier, J. 1998. Fenologia de dos species de Tillandsia (Bromeliaceae) en un bosque montano alto de la cordillera oriental colombiana. Selbyana 19: 44–51. Jebb, M. 1991. Cavity structure and function in the tuberous Rubiaceae. In Ant–Plant Interactions, eds. C. R. Huxley and D. F. Cutler, pp. 374–389. Oxford: Oxford University Press. Jenkins, D. W. 1999. Cold hardiness and cold sensitivity of bromeliads. J. Bromeliad Soc. 49: 32–41. Jensen, A. S. 1982. Spanish moss: symbol of the Southland. Palmetto 2: 1–2. Joel, D. M. 1988. Mimicry and mutualism in carnivorous pitcher plants (Sarraceniaceae, Nepenthaceae, Cephalotaceae, Bromeliaceae). Biol. J. Linn. Soc. 35: 185–197. Johansson, D. R. 1975. Ecology of epiphytic orchids in west African rain forests. Amer. Orchid Soc. Bull. 44: 125–136. Johansson, D. R. 1977. Epiphytic orchids as parasites of their host trees. Amer. Orchid Soc. Bull. 46: 703–707. Johow, F. 1910. Estudios de biologia vejetal. I. Sobre algunos casos de ornithofilía en la flora chilena. Anales University Chile 126: 27–50. Johri, B. M., Ambegaokar, K. B. and Srivastava, P. S. 1992. Comparative Embryology of Angiosperms (2 vols). Berlin: Springer-Verlag. Jones, H. L. and Diamond, J. M. 1976. Short-time-base studies on turnover in breeding bird populations on California’s Channel Islands. Condor 78: 526–549. Jordan, C. F. 1985. Nutrient Cycling in Tropical Forest Ecosystems: Principles and their Application in Management and Conservation, pp. 24–27. New York: Wiley. Joyal, E. 1987. Ethnobotanical field notes from Ecuador: Camp, Prieto, Jorgensen, and Giler. Econ. Bot. 41: 163–189. Junk, W. J. and Furch, K. 1985. The physical and chemical properties of Amazonian waters and their relationships with the biota. In Amazonia, eds. G. B. Prance and T. E. Lovejoy, p. 7. Oxford: Pergamon Press. Kato, M., Inone, T. and Nagamitsu, T. 1995. Pollination biology of Gnetum 638 Literature cited (Gnetaceae) in a lowland mixed dipterocarp forest in Sarawak. Amer. J. Bot. 82: 862–868. Kellman, M., Hudson, J. and Sanmugadas, K. 1982. Temporal variability in atmospheric nutrient influx to a tropical ecosystem. Biotropica 14: 1–9. Kelly, D. L. 1985. Epiphytes and climber of a Jamaican rain forest: vertical distribution, life forms, and life histories. J. Biogeogr. 12: 223–241. Kernan, C. 1994. A method for estimating bark surface in forest canopies. Selbyana 15: 14–17. Kernan, C. and Fowler, N. 1995. Differential substrate use by epiphytes in Corcovado National Park, Costa Rica: a source of guild structure. J. Ecol. 83: 65–73. Kiff, L. F. 1991. A Distributional Checklist of the Genus Tillandsia. Encino, CA: Botanical Diversions, 5404 Encino Ave., CA 91316, USA. Klein, R. M. 1967. Aspectos do problema ‘bromeliad malaria’ no sul do Brazil. Sellowia 19: 125–135. Kleinfeldt, S. E. 1978. Ant-gardens: the interaction of Codonanthe crassifolia (Gesneriaceae) and Crematogaster longispina (Formicidae). Ecology 59: 449–456. Kluge, M., Lange, O. L., v. Eichmann, M. and Schmid, R. 1973. Diurnaler Säurerhythmus bei Tillandsia usneoides: Untersuchungen über den Weg des Kohlenstoffs sowie die Abhängigkeit des CO2-Gas-Wechsels von Lichtintensität, Temperatur und Wassergehalt der Pflanze. Planta 112: 357–372. Knab, F. 1912. New species of Anisopidae (Rhyphidae) from tropical America (Diptera, Nematocera). Proc. Biol. Soc. Wash. 25: 111–114. Knauft, R. L. and Arditti, J. 1969. Partial identification of dark 14CO2 fixation products in leaves of Cattleya (Orchidaceae). New Phytol. 68: 657–661. Knudsen, J. T. and Tollsten, L. 1995. Floral scent in bat-pollinated plants: a case of convergent evolution. Bot. J. Linn. Soc. 119: 45–57. Koniger, M., Harris, G. C., Virgo, A. and Winter, K. 1995. Xanthophyll-cycle pigments and photosynthetic capacity in tropical forest species: a comprehensive field study in canopy gap and understory plants. Oecologia 104: 280–290. Koopowitz, H. 1992. A stochastic model for the extinction of tropical orchids. Selbyana 13: 115–122. Koopowitz, H., Thornhill, A. D. and Anderson, M. 1994. A general stochastic model for the prediction of biodiversity losses based on habitat conversion. Conserv. Biol. 8: 425–438. Koptur, S. 1992. Extrafloral nectary-mediated interactions between insects and plants. In Insect–Plant Interactions, vol. 4, ed. E. Bernays, pp. 81–129. Boca Raton, FL: CRC Press. Krauss, B. H. 1948–49. Anatomy of the vegetative organs of the pineapple Ananas comosus (L.) Merr. Bot. Gaz. 110: 159–217, 333–404, 550–587. Kress, W. J., Luther, H. E. and Roesel, C. S. 1990. Genetic variation in three species of Florida Tillandsia. J. Bromeliad Soc. 40: 59–60, 109–111. Krügel, P. 1993. Biologie und Ökologie der Bromelienfauna von Guzmania weberbaueri im amazonischen Peru, pp. 1–69. Biosystematic and Ecology Series 2. Wien: Österreichische Akademie der Wissenschaften. Krügel, P. and Richter, P. 1995. Syncope antenori – a bromeliad breeding frog with free-swimming nonfeeding tadpoles (Anura, Microhylidae). Copia 61: 955–963. Literature cited 639 Kubisch, F. 1965. The blooming season of some Mexican tillandsias. Bromeliad Soc. Bull. 15: 13. Kugler, H. 1942. ‘Raphidenpollen’ bei Bromeliaceae. Ber. Deut. Bot. Ges. 60: 388–393. L. H. Bailey Hortorium 1976. Hortus Third. A Concise Dictionary of Plants Cultivated in the United States and Canada. New York: Macmillan. Lacerda, L. D. and Hay, J. D. 1982. Habitat of Neoregelia cruenta (Bromeliaceae) in coastal sand dunes of Marica, Brazil. Rev. Biol. Trop. 30: 171–173. Laessle, A. M. 1961. A microlimnolgical study of Jamaican bromeliads. Ecology 42: 499–517. Laferriere, J. E., Weber, C. W. and Kohlhepp, E. A. 1991. Use and nutritional composition of some traditional mountain Pina plant foods. J. Ethnobiol. 11: 93–114. Lange, O. L. and Medina, E. 1979. Stomata of the CAM plant Tillandsia recurvata respond directly to humidity. Oecologia 40: 357–363. Larcher, W. 1980. Physiological Plant Ecology, 2nd edn. Berlin: Springer-Verlag. Lauer, W. 1986. Die Vegetationszonierung der Neotropis und ihr Wandell seit der Eiszeit. Ber. Deut. Bot. Ges. 99: 211–235. Lavelle, P. and Kohlmann, B. 1984. Etude quantitative de la macrofaune du sol dans une forêt tropicale humide du Mexique (Bonampak, Chiapas). Pedobiologia 27: 377–393. Lee, D. W., Lowry, J. B. and Stone, B. C. 1979. Abaxial anthocyanin layer in leaves of tropical rain forest plants: enhancer of light capture in deep shade. Biotropica 11: 70–77. Lee, H. S. J., Lüttge, U., Medina, E. et al. 1989. Ecophysiology of xerophytic and halophytic vegetation of a coastal alluvial plain in northern Venezuela. III. Bromelia humilis Jacq., a terrestrial CAM bromeliad. New Phytol. 111: 253–271. Lehman, P. S. 1987. Tylenchocriconema alleni, a pathogen of the bromeliad Tillandsia flabellata. J. Bromeliad Soc. 37: 108–110. Leme, E. 1997. Bromélias da Mata Atlântica. Rio de Janeiro: Salamandra Consultoria Editorial Ltda. Leme, E. 1998a. Bromeliads of the Atlantic Forest: Canistrum. Rio de Janeiro: Salamandra Consultoria Editorial Ltda. Leme, E. 1998b. Bromeliads of the Atlantic Forest: Canistropsis. Rio de Janeiro: Salamandra Consultoria Editorial Ltda. Leme, E. M. 1984. Bromélias. Ciéncia Hoje 3: 66–72. Leme, E. M. and Marigo, L. C. 1993. Bromélias na Natureza. Rio de Janeiro: Mariago Comunicação Visual Ltda. Lesica, P. and Antibus, R. K. 1991. Canopy soils and epiphyte richness. Natl. Geogr. R. Explor. 7: 156–165. Levey, D. J. 1987. Seed size and fruit-handling techniques of avian frugivores. Amer. Nat. 129: 471–485. Lindschau, M. 1933. Beiträge zur Zytologie der Bromeliaceae. Planta 20: 506–530. Lineham, T. U., Jr. 1987. Rusts on bromeliads. J. Bromeliad Soc. 37: 76–77. Linsbauer, K. 1911. Zur physiologischen Anatomie de Epidermis und der Durchlüftungsapparates der Bromeliaceen: Akademie de Wissenschaften in Wien. Sitzungsberichte. Math.-nat. Kl. 120(4): 319–348 and plates. Lobry De Bruyn, L. A. and Conacher, A. J. 1990. The role of termites and ants in soil modification: a review. Aust. Soil Res. 28: 55–93. 640 Literature cited Loeschen, V. S., Martin, C. E., Smith, M. and Eder, S. L. 1993. Leaf anatomy and CO2 recycling during crassulacean acid metabolism in twelve epiphytic species of Tillandsia (Bromeliaceae). Int. J. Plant Sci. 154: 100–106. Long, S. P., Humphries, S. and Falkowski, P. G. 1994. Photoinhibition of photosynthesis in nature. Ann. Rev. Plant Physiol. Mol. Biol. 45: 633–662. Longino, J. T. 1986. Ants provide substrate for epiphytes. Selbyana 9: 100–103. Loope, L., Duever, M., Herndon, A., Snyder, J. and Janzen, D. 1994. Hurricane impact on upland and freshwater swamp forest. Bioscience 44: 238–246. Lopez, L. C. S., Madeira, J. A., Torrey, K. and Rios, R. I. 1993. Composição e dinâmica hidrica de phytotelmata de Aechmea nudicaulis e Neoregelia cruenta (Bromeliaceae – Bromelioideae) na Restinga de Maricá/ R. J. Anais do III. In Simpósio de Ecosistemas da Costa Brasileira, pp. 200–207. Rio de Janeiro. Lounibos, L. P., Frank, J. H., Machado-Allison, C. E., Octanto, C. E. and Navarro, J. C. 1987. Survival, development and predator effects of mosquito larva in Venezuelan phytotelmata. J. Trop. Ecol. 3: 221–242. Lowman, M. and Linnerooth, W. 1995. Population dynamics of some native Florida epiphytes. II. Mortality after a storm. J. Bromeliad Soc. 45: 15–17. Lowman, M. D. 1995. Herbivory as a canopy process in rain forest trees. In Forest Canopies, eds. M. D. Lowman and N. M. Nadkarni, pp. 431–456. San Diego, CA: Academic Press. Lowman, M., Wittman, P. and Murray, D. 1996. Herbivory in a bromeliad of the Peruvian rain forest canopy. J. Bromeliad Soc. 46: 52–55. Lugo, A. E. and Scatena, F. N. 1992. Epiphytes and climate change research in the Caribbean: a proposal. Selbyana 13: 123–130. Luther, H. 1993. An unusual night-flowering Guzmania from southern Ecuador. J. Bromeliad Soc. 43: 195–197. Luther, H. E. 1997. A showy Fosterella from Bolivia. J. Bromeliad Soc. 47: 118–119. Luther, H. E. and Sieff, E. 1996. An Alphabetical list of Bromeliad Binomials. Newberg, OR: The Bromeliad Society, Inc. Lüttge, U. 1987. Carbon dioxide and water demand: crassulacean acid metabolism (CAM), a versatile ecological adaptation exemplifying the need for integration in ecophysiological work. New Phytol. 106: 593–629. Lüttge, U. 1988. Day–night changes of citric acid levels of crassulacean acid metabolism: phenomena and ecophysiological significances. Plant Cell Environ. 11: 445–451. Lüttge, U. and Ball, E. 1987. Dark respiration of CAM plants. Plant Physiol. Biochem. 25: 3–10. Lüttge, U., Klauke, B., Griffiths, H., Smith, J. A. C. and Stimmel, K. 1986a. Comparative ecophysiology of CAM and C3 bromeliads. V. Gas exchange and leaf structure of the terrestrial C3 species Pitcairnia integrifolia. Plant Cell Environ. 9: 411–419. Lüttge, U., Stimmel, K., Smith, J. A. C. and Griffiths, H. 1986b. Comparative ecophysiology of CAM and C3 bromeliads. II. Field measurements of gas exchange of CAM bromeliads in the humid tropics. Plant Cell Environ. 9: 377–383. Lyra, L. T. 1971. Algumas diatomáceas encontradas en bromeliaceas, Brazil. Memórias do Instituto Oswaldo Cruz 69: 129–139. Mabberley, D. J. 1987. The Plant Book. Cambridge: Cambridge University Press. MacArthur, R. H. and Wilson, E. O. 1967. The Theory of Island Biogeography. Princeton, NJ: Princeton University Press. Literature cited 641 Maddison, W. P. and Maddison, D. R. 1992. MacClade: Analysis of Phylogeny and Character Evolution, Version 3.0. Sunderland, MA: Sinauer Associates. Madeira, J. A., Ribeiro, K. T., Lopez, L. C. S. and Rios, R. I. 1995. Colonization processes of associated tank communities in two bromeliads from Maricá ‘restinga’. Bromélia 2: 18–31. Madison, M. 1977. Vascular epiphytes: their systematic occurrence and salient features. Selbyana 2: 1–13. Madison, M. 1979. Additional observation on ant-gardens in Amazonas. Selbyana 5: 107–115. Maguire, B. 1971. Phytotelmata: biota and community structure determination in plant-held waters. Ann. Rev. Ecol. Syst. 2: 439–464. Marchant, C. 1967. Chromosome evolution in the Bromeliaceae. Kew Bull. 21: 161–168. Marlatt, R. and Krauss, J. F. 1974. A new leaf disease of Aechmea fasciata caused by Helminthosporium rostratum. Plant Disease Reporter 58: 445–448. Martin, A. C. 1946. The comparative internal morphology of seeds. Amer. Midl. Nat. 36: 550–551. Martin, C. A. and Siedow, J. N. 1981. Crassulacean acid metabolism in the epiphyte Tillandsia usneoides L. (Spanish moss): responses of CO2 exchange to controlled environmental conditions. Plant Physiol. 68: 335–339. Martin, C. A., Higley, M. and Wang, W. Z. 1988. Ecophysiological significance of CO2-recycling via crassulacean acid metabolism in Talinum calycinum Engelm. (Portulacaceae). Plant Physiol. 86: 562–568. Martin, C. A., McKee, J. M. and Schmitt, A. K. 1989. Responses of photosynthetic O2 evolution to PPFD in the CAM epiphyte Tillandsia usneoides L. (Bromeliaceae). Photosyn. Res. 21: 145–150. Martin, C. A., McLeod, K. W., Eades, C. A. and Pitzer, A. F. 1985. Morphological and physiological responses to irradiance in the CAM epiphyte Tillandsia usneoides L. (Bromeliaceae). Bot. Gaz. 146: 489–494. Martin, C. E. 1994. Physiological ecology of the Bromeliaceae. Bot. Rev. 60: 1–82. Martin, C. E. and Adams, W. W., III. 1987. Crassulacean acid metabolism, CO2recycling, and tissue desiccation in the Mexican epiphyte Tillandsia schiedeana Steud. (Bromeliaceae). Photosyn. Res. 11: 237–244. Martin, C. E. and Peters, E. A. 1984. Functional stomata of the atmospheric epiphyte Tillandsia usneoides L. Bot. Gaz. 145: 502–507. Martin, C. E. and Schmitt, A. K. 1989. Unusual water relations in the CAM atmospheric epiphyte Tillandsia usneoides L. (Bromeliaceae). Bot. Gaz. 150: 1–8. Martin, C. E., Christensen, H. L. and Strain, B. R. 1981. Seasonal patterns of growth, tissue acid fluctuations, and 14CO2 uptake in the crassulacean acid metabolism epiphyte Tillandsia usneoides L. (Spanish moss). Oecologia 49: 322–328. Martin, C. E., Eades, C. A. and Pitner, R. A. 1986. Effects of irradiance on crassulacean acid metabolism in the epiphyte Tillandsia usneoides L. (Bromeliaceae). Plant Physiol. 80: 23–26. Martinelli, G. 1994. Reproductive biology of Bromeliaceae in the Atlantic rain forest of southeastern Brazil. PhD thesis. St Andrews: University of St Andrews. Martinez del Rio, C., Stevens, B. R., Daneke, D. E. and Andreadis, P. T. 1988. Physiological correlates of preference and aversion for sugars in three species of birds. Physiol. Zool. 62: 222–229. 642 Literature cited Martinez, J. D., Nathany, M. and Dharmarajan, V. 1971. Spanish moss, a sensor of lead. Nature 233: 564–565. Maschwitz, J. and Holldobler, B. 1970. Der Nestkartonbau bei Lasius fuliginosus. Z. Vrgl. Physiol. 66: 176–189. Mason, L., Jr. 1976. Microscopy–Macroscopy. J. Bromeliad Soc. 26: 187–191. Mastalerz, J. 1957. Preliminary report on the effects of daylength on the flowering of Billbergia nutans. Bromeliad Soc. Bull. 7: 37–38. Matelson, T. J., Nadkarni, N. M. and Iongino, J. T. 1993. Longevity of fallen epiphytes in a Neotropical montane forest. Ecology 74: 265–269. Maxwell, C., Griffiths, H. and Young, A. J. 1994. Photosynthetic acclimation to light regime and water stress by the C3–CAM epiphyte Guzmania monostachia: gas-exchange characteristics, photochemical efficiency and the xanthophyll cycle. Funct. Ecol. 8: 746–754. Maxwell, C., Griffiths, H., Borland, A. M., Broadmeadow, M. S. J. and McDavid, C. R. 1992. Photoinhibitory responses of the epiphytic bromeliad Guzmania monostachia during the dry season in Trinidad maintain photochemical integrity under adverse conditions. Plant Cell Environ. 15: 37–47. Maxwell, C., Griffiths, H., Borland, A. M., Young, A. J., Broadmeadow, S. J. and Fordham, M. C. 1995. Short-term photosynthetic responses of the C3–CAM epiphyte Guzmania monostachia var. monostachia to tropical seasonal transitions under field conditions. Aust. J. Pl. Physiol. 22: 771–781. Mayo, S. J. and Barroso, G. M. 1979. A new pedate-leaved species of Philodendron from Bahia, Brazil. Aroideana 2: 81–94. McKey, D. 1975. The ecology of coevolved seed dispersal systems. In Coevolution of Animals and Plants, eds. L. E. Gilbert and P. H. Raven, pp. 159–191. Austin, TX: University of Texas Press. McMahon, L. 1987. Plant conservation laws: how effective are they? Nature Conservancy Mag. 37: 21–23. McVaugh, R. 1989. Flora Novo-Galiciana, vol. 15, Bromeliaceae–Dioscoreaceae. Ann Arbor, MI: University of Michigan Press. McWilliams, E. L. 1969. Crabs belonging to the genus Sesarma found living in four species of bromeliads. Bromeliad Soc. Bull. 20: 80–82. McWilliams, E. L. 1970. Comparative rates of dark CO2 uptake and acidification in the Bromeliaceae, Orchidaceae, and Euphorbiaceae. Bot. Gaz. 131: 285–290. McWilliams, E. L. 1974. Evolutionary ecology. In Pitcairnioideae (Bromeliaceae). Flora Neotropica Monograph 14, Part 1, eds. L. B. Smith and R. J. Downs, pp. 33–55. New York: Hafner Press. Medina, E. 1974. Dark CO2 fixation, habitat preference and evolution within Bromeliaceae. Evolution 28: 677–686. Medina, E. 1990. Eco-fisiologia y evolution de las Bromeliaceae. Boletin Academia Nacional Ciencias, Cordoba, Argentina 59: 72–100. Medina, E. and Troughton, S. 1974. Dark CO2 fixation and the carbon isotope ratio in Bromeliaceae. Pl. Sci. Lett. 2: 357–362. Medina, E., Delgado, M., Troughton, J. H. and Medina, J. D. 1977. Physiological ecology of CO2 fixation in Bromeliaceae. Flora 166: 137–152. Medina, E., Popp, M., Lüttge, U. and Ball, E. 1991a. Gas exchange and acid accumulation in high and low irradiance grown pineapple cultivars. Photosynthetica 25: 489–498. Medina, E., Lüttge, U., Leal, F. and Ziegler, H. 1991b. Carbon and hydrogen isotope ratios in bromeliads growing under different light environments in natural conditions. Bot. Acta 104: 47–52. Literature cited 643 Medina, E., Olivares, E. and Diaz, M. 1986. Water stress and light intensity effects on growth and nocturnal acid accumulation in a terrestrial CAM bromeliad (Bromelia humilis Jacq.) under natural conditions. Oecologia 70: 441–446. Medina, E., Popp, M., Olivares, E., Janett, H. P. and Lüttge, U. 1993. Daily fluctuations of titratable acidity content of organic acids (malate and citrate) and soluble sugars of varieties and wild relatives of Ananas comosus L. growing under natural tropical conditions. Plant Cell Environ. 16: 55–63. Melchior, H. 1964. A. Engler’s Syllabus der Pflanzenfamilien, 12th edn. BerlinNikolassee: Bornträger. Méluzin, S. 1997. Ritual use of bromeliads in the maize-planting ceremony of the Lenca of Honduras: Part I. J. Bromeliad Soc. 47: 252–260. Méluzin, S. 1998. Ritual use of bromeliads in the maize-planting ceremony of the Lenca of Honduras: Part II. J. Bromeliad Soc. 48: 17–23. Mercier, H. 1993. Effects of nitrogen source on the growth and development, endogenous hormone production, polypeptide and isozyme profiles of three native Brazilian bromeliads cultivated in vitro. PhD thesis. São Paulo: Universidade de São Paulo. Mercier, H. and Kerbauy, G. B. 1992. In vitro multiplication of Vriesea fosteriana. Plant Cell, Tissue and Organ Culture 30: 247–249. Meyer, F. J. 1929. Bromeliaceae. In Systematische Anatomie der Monokotyledonen, Heft IV (Farinosae), eds. H. Solereder and F. J. Meyer, pp. 80–129. Berlin: Bornträger. Meyer, L. 1940. Zur Anatomie und Entwicklungsgeschichte der Bromeliaceenwurzeln. Planta 31: 492–522. Mez, C. 1894. Bromeliaceae. In Flora Brasiliensis 3, Part 3, ed. C. Martius, pp. 635–816. Lipsiae: F. Fleisher. Mez, C. 1896. Bromeliaceae. In Monographiae Phanerogamarum 9, ed. C. de Candolle, pp. 1–990. Paris: G. Masson. Mez, C. 1904. Physiologische Bromeliaceen-Studien. I. Die Wasser-Ökonomie den extrem atmosphärischen Tillandsien. Jahrbücher für wissenschaftliche Botanik 40: 157–229. Mez, C. 1934–35. Bromeliaceae. In Das Pflanzenreich, vol. IV (32), ed. A. Engler, pp. 1–667. Leipzig: W. Engelmann. Midgiey, J. J. and Stock, W. D. 1998. Natural abundance of d15N confirms insectivorous habit of Roridula gorgonias despite it having no proteolytic enzymes. Ann. Bot. 82: 387–388. Miller, A. C. 1971. The bromeliad ecosystem. J. Bromeliad Soc. 11: 33–35. Miller, G. A. 1986. Pubescence, floral temperature and fecundity in species of Puya (Bromeliaceae) in the Ecuadorian Andes. Oecologia 70: 155–160. Miller, G. A. and Silander, J. A., Jr. 1991. Control of the distribution of giant rosette species of Puya (Bromeliaceae) in the Ecuadorian Páramos. Biotropica 23: 124–133. Mitchell, P. 1974. Proof of insect pollination of Hechtia scariosa. J. Bromeliad Soc. 24: 28–29. Mlot, C. 1989. Blueprint for conserving plant diversity. Bioscience 39: 364–368. Moerman, D. E. 1986. Medicinal plants of native America, vol. 1. University of Michigan Museum of Anthropology Technical Reports No. 19 (Research Reports in Ethnobotany, Contribution 2). Ann Arbor, MI: University of Michigan. Moffat, A. S. 1998. Global nitrogen overload problems grows critical. Science 279: 988–989. 644 Literature cited Montaña, C., Dirzo, R. and Flores, A. 1997. Structural parasitism of an epiphytic bromeliad upon Cercidium praecox in an intertropical semiarid ecosystem. Biotropica 29: 517–521. Mors, W. B. and Rizzini, C. T. 1966. Useful Plants of Brazil. San Francisco, CA: Holden-Day Inc. Morton, J. F. 1981. Atlas of Medicinal Plants of Middle America. Springfield, IL: Charles C. Thomas. Murawski, D. A. and Hamrick, J. L. 1990. Local genetic structure in the terrestrial bromeliad, Aechmea magdalenae. Amer. J. Bot. 77: 1201–1208. Murillo, R., Palacios, J., Labougle, J. et al. 1983. Variación estacional de la entomofauna asociada a Tillandsia spp. en a zona de transitión biótica. Southern Entomologist 8: 292–300. Nadkarni, N. M. 1981. Canopy roots: convergent evolution in rain forest nutrient cycles. Science 214: 1023–1024. Nadkarni, N. M. 1984. Epiphyte biomass and nutrient capital of a Neotropical elfin forest. Biotropica 16: 249–256. Nadkarni, N. M. 1986. The nutritional effects of epiphytes on host trees with special reference to alteration of precipitation chemistry. Selbyana 9: 44–51. Nadkarni, N. M. and Matelson, T. J. 1989. Bird use of epiphyte resources in neotropical montane forest and pasture tree crowns. Condor 91: 891–897. Nadkarni, N. M. and Matelson, T. J. 1991. Litter dynamics within the canopy of a neotropical cloud forest, Monteverde, Costa Rica. Ecology 72: 849–860. Nadkarni, N. M. and Primack, R. B. 1989. The use of gamma spectroscopy to measure within plant nutrient allocations of the tank bromeliad Guzmania monostachia. Selbyana 11: 22–25. Naeem, S. 1990. Resource heterogeneity and community structure: a case study in Heliconia imbricata phytotelmata. Oecologia 84: 29–38. Napp-Zinn, K., Schmidt, R. and Genscher, H. 1978. Vergleichend-anatomische Untersuchungen an petaloiden Hochblättern. I. Bromeliaceen. Trop. subtrop. Pflanzenwelt 24: 1–87. Neales, T. F., Sale, P. J. M. and Meyer, C. P. 1980. Carbon dioxide assimilation by pineapple plants, Ananas comosus (L.) Merr. II. Effects of variation of the day/night temperature regime. Aust. J. Pl. Physiol. 7: 375–385. Nelson, E. C. and Zizka, G. 1997. Fascicularia (Bromeliaceae): which species are cultivated and naturalized in northwestern Europe? New Plantsman 1: 232–239. Netolitzky, F. 1926. Anatomie der Angiospermen-Samenin. In Handbuch der Pflanzenanatomie, vol. 10, ed. K. Linsbauer, pp. 62–63, 71–73. Berlin: Verlag von Gebruder Borntraeger. Newman, E. I. 1995. Phosphorus inputs to terrestrial ecosystems. J. Ecol. 83: 713–726. Newsham, K. K., Fitter, A. H. and Watkinson, A. R. 1995. Multi-functionality and biodiversity in arbuscular mycorrhizas. Trends Ecol. Evol. 10: 407–411. Nievola, C. C. and Mercier, H. 1996. The importance of leaf and root systems in nitrate assimilation in Vriesea fosteriana. Bromélia 3: 14–17. Nobel, G. K. 1929. The adaptive modifications of arboreal tadpoles of Holophryne and torrent tadpoles of Staurois. Bull. Am. Mus. Nat. Hist. 58: 291–334. Nobel, P. S. 1988. Environmental Biology of Agaves and Cacti. New York: Cambridge University Press. Nobel, P. S. 1991. Achievable productivities of certain CAM plants: basis for high values compared to C3 and C4 plants. New Phytol. 119: 183–205. Literature cited 645 Nobel, P. S. and Sanderson, J. 1984. Rectifier-like activities of roots of two desert succulents. J. Exp. Bot. 35: 727–737. Nose, A., Heima, K., Miyazato, K. and Murayama, S. 1986. Effects of day-length on CAM-type CO2 and water vapour exchange in pineapple plants. Photosynthetica 20: 20–28. Nose, A., Matake, S., Miyazato, K. and Murayama, S. 1985. Studies on matter production in pineapple plants. III. Effects of nitrogen nutrition on the gas exchange of shoots. Japan J. Crop Sci. 54: 195–204. Nose, A., Miyazato, K. and Murayama, S. 1981. Studies on matter production in pineapple plants. II. Effects of soil moisture on gas exchange of pineapple plants. Japan J. Crop Sci. 50: 525–535. Nose, A., Shiroma, M., Miyazato, K. and Murayama, S. 1977. Studies on matter production in pineapple plants. I. Effects of light intensity in light period on the CO2 exchange and CO2 balance of pineapple plants. Japan J. Crop Sci. 46: 580–587. Núñez Meléndez, E. 1982. Plantas Medicinales de Puerto Rico. Puerto Rico: Editorial de la Universidad de Puerto Rico. Nyman, L. P., Davis, J. P., O’Dell, S. J., Arditti, J., Stephens, G. C. and Benzing, D. H. 1987. Active uptake of amino acids by an epiphytic vascular plant, Tillandsia paucifolia (Bromeliaceae). Plant Physiol. 83: 681–684. Oberbauer, F. S., Van Kleist, K., III, Whelan, R. T. and Koptur, S. 1996. Effects of hurricane Andrew on epiphyte communities within cypress domes of the Everglades National Park. Ecology 77: 964–967. Okahara, K. 1932. On the role of microorganisms in the digestion of insect bodies in insectivorous plants. Bot. Mag. (Tokyo) 46: 353–357. Oliveira, M. G. N., Rocha, C. F. D. and Bagnall, T. 1994. The animal community associated with the tank bromeliad Neoregelia cruenta. Bromélia 1: 22–29. Oliver, W. R. B. 1930. New Zealand epiphytes. J. Ecol. 18: 1–50. Olmsted, I. and Dejean, A. 1987. Tree–epiphyte–ant relationships of the low inundated forest in Sian Ka’an Biosphere reserve, Quintana Roo, Mexico. 38th Annual AIB Meeting, Ohio State University, Columbus, 9–13 August 1987, p. 88 (Abstract). Orivel, J., Dejean, A. and Errard, C. 1998. Active role of two ponerine ants in the elaboration of ant gardens. Biotropica 30: 487–491. Orr, C. and Wrisely, B. 1981. Vocabulario Quichua. Quito: Instituto Linguistico de Verano. Ortiz-Crespo, F. I. 1973. Field studies of pollination of plants of the genus Puya. J. Bromeliad Soc. 23: 3–7, 54–58. Ortlieb, U. and Winkler, S. 1977. Ökologische Differenzierungsmuster in der Evolution der Bromeliaceen. Bot. Jahrb. Syst. 97: 586–602. Owen, T. P., Jr. and Thomson, W. W. 1988. Sites of leucine, arginine and glycine accumulation in the absorptive trichomes of a carnivorous bromeliad. J. Ultrastructure and Molecular Structure Res. 101: 215–223. Owen, T. P., Jr., Benzing, D. H. and Thomson, W. W. 1988. Apoplastic and ultrastructural characterizations of the trichomes from the carnivorous bromeliad Brocchinia reducta. Can. J. Bot. 66: 941–948. Owen, T. P., Jr., Platt-Aloia, A. and Thomson, W. W. 1991. Ultrastructural localization of lucifer yellow and endocytosis in plant cells. Protoplasma 160: 115–120. Palací, C. A. 1991. Enzyme electrophoresis and systematic relationships among four species of Tillandsia subgen. Anoplophytum (Bromeliaceae) from northwest 646 Literature cited Argentina and adjacent Bolivia. Thesis. Laramie, WY: University of Wyoming. Palací, C. A. 1997. A systematic revision of the genus Catopsis (Bromeliaceae). PhD thesis. Laramie, WY: University of Wyoming. Palacio-Vargas, J. G. 1982. Microarthropodos asociados a Bromeliaceas. Zool. Neotrop. 1: 535–545. Paoletti, M. G., Taylor, R. A. J., Stinner, B. R., Stinner, D. D. H. and Benzing, D. H. 1991. Diversity of soil fauna in the canopy and forest floor of a Venezuelan cloud forest. J. Trop. Ecol. 7: 373–383. Paroz, P. R. 1981. Notes from down under. J. Bromeliad Soc. 31: 119–120. Peixoto, O. L. 1977. Anfíbios anuros associados ás bromeliáceas nos estados do Rio de Janeiro e Espírito Santo. Dissertacão de Mestrado. Rio de Janeiro: University Fed. Rio de Janeiro, p. 55. Penfound, W. T. and Deiler, F. G. 1947. On the ecology of Spanish moss. Ecology 28: 455–458. Pennisi, E. 1992. Pharming frogs: chemist finds precious alkaloids in poisonous amphibians. Science News 142: 40–42. Percival, M. S. 1961. Types of nectar in angiosperms. New Phytol. 60: 235–281. Pérez-Arbeláez, E. 1956. Plantas Utiles de Colombia. Bogota: Liberia Colombiana. Pfitsch, W. A. and Smith, A. P. 1988. Growth and photosynthesis of Aechmea magdalenae, a terrestrial CAM plant in a tropical moist forest, Panama. J. Trop. Ecol. 4: 199–207. Physician’s Desk Reference 1984. Oradell, NJ: Medical Economics Company. Picado, C. 1911. Les broméliacées épiphytes comme milieu biologique. C. R. Acad. Sci. Belgique 153: 960–963. Picado, C. 1913. Les broméliacées épiphytes considérées comme milieu biologique. Bull. Sci. France et Belgique 5: 215–360. Pittendrigh, C. S. 1946. Bromeliad malaria in Trinidad, B. W. I. Amer. J. Trop. Med. Hyg. 26: 47–66. Pittendrigh, C. S. 1948. The bromeliad–Anopheles–malaria complex in Trinidad. I. The bromeliad flora. Evolution 2: 58–89. Pizo, M. A. 1994. Bromeliad use by Atlantic forest birds at Fazenda Intervales in southeastern Brazil. Bromélia 1: 3–7. Plummer, G. L. and Kethley, J. B. 1964. Foliar absorption of amino acids, peptides and other nutrients by the pitcher plant Sarracenia flava. Bot. Gaz. 125: 245–259. Poisson, M. J. 1877. Du siége des matiéres colorées dans la graine. Bull. Soc. Bot. France 24: 280–290. Prance, G. T. 1973. Phytogeographic support for the theory of Pleistocene forest refuges in the Amazon Basin, based on evidence from distribution patterns in Caryocaraceae, Chrysobalanaceae, Dichapetalaceae and Lecythidaceae. Acta Amazon. 3: 5–28. Prance, G. T. 1982. A review of the phytogeographic evidences for Pleistocene climate changes in the Neotropics. Ann. Missouri Bot. Gard. 69: 594–624. Privat, F. 1979. Les bromeliacées, lieu de developpement de quelques insects pollinisateurs des fleurs de cacao. Brenesia 16: 197–211. Puente, M. and Basham, Y. 1994. The desert epiphyte Tillandsia recurvata harbours the nitrogen fixing bacterium Pseudomonas stutzeri. Can. J. Bot. 72: 406–408. Purseglove, J. W. 1972. Tropical Crops: Monocotyledons. Essex: Longman Group, Ltd. Literature cited 647 Raack, J. 1985. Observations on the blooming periodicity of bromeliads. J. Bromeliad Soc. 35: 106–111. Rabatin, S. C., Stinner, B. R. and Paoletti, M. G. 1993. Vesicular-arbuscular mycorrhizal fungi, particularly Glomus tenue, in Venezuelan bromeliad epiphytes. Mycorrhiza 4: 17–20. Ramírez, I. 1991. A systematic revision of Neoregelia subgenus Hylaeaicum (Bromeliaceae). MSc thesis. St Louis, MO: University of Missouri. Ramírez, I. 1994. Notes on Neoregelia subgenus Hylaeaicum (Bromeliaceae; Bromelioideae). Selbyana 15: 82–83. Ramírez, I. 1996. Systematics, phylogeny, chromosome number, and evolution of Cryptanthus (Bromeliaceae). PhD thesis. St Louis, MO: University of Missouri. Ranker, T. A., Soltis, D. E., Soltis, P. S. and Gilmartin, A. J. 1990. Subfamilial phylogenetic relationships of the Bromeliaceae: evidence from chloroplast DNA restriction site variations. Syst. Bot. 15: 425–434. Rauh, W. 1983a. Catopsis pisiformis, a new species from central Panama. J. Bromeliad Soc. 33: 200–204. Rauh, W. 1983b. Bromelienstudien I. Neue und wenig bekannte Arten aus Peru und anderen Ländern (14. Mitteilung). Trop. subtrop. Pflanzenwelt 43: 33–38. Rauh, W. 1990. Bromelien, 3rd edn. Stuttgart: E. Ulmer. Rauh, W. 1992. Are tillandsias endangered plants? Selbyana 13: 138–139. Raven, J. A. 1985. Regulation of pH and generation of osmolarity in vascular plants: a cost–benefit analysis in relation to efficiency of use of energy, nitrogen and water. New Phytol. 101: 25–77. Raven, J. A. 1988. Acquisition of nitrogen by the shoots of land plants: its occurrence and implications for acid–base regulation. New Phytol. 109: 1–20. Read, M. 1989. Bromeliads threatened by trade. Kew Mag. 6: 22–29. Read, R. W. 1969. Crabs in bromeliads on the high mountains of Jamaica. Bromeliad Soc. Bull. 19: 78–79. Redford, K. H. 1992. The empty forest. Bioscience 42: 412–422. Rees, W. E. and Roe, N. A. 1980. Puya raimondii (Pitcairnioideae: Bromeliaceae) and birds: a hypothesis on nutrient relationship. Can. J. Bot. 85: 1262–1268. Reinert, F. and Meirelles, S. T. 1993. Water acquisition strategy shifts in the heterophyllous saxicolous bromeliad, Vriesea geniculata (Wawra). Selbyana 14: 80–88. Reinert, F., Griffiths, H., Ribas, L. et al. 1995. Neoregelia cruenta (Bromeliaceae) from the restinga vegetation of Brazil. XLVI Congresso National de Botânica, IV Simpósio de Bromeliáceas, Universidade de São Paulo, p. 307 (Abstract). Sociedad Botanica do Brazil. São Paulo: Gráfica Canavaci Ltda. Reitz, P. R. 1956. Elpidium bromelarium, a crustacean living in bromeliads. Bromeliad Soc. Bull. 6: 61–62. Reitz, P. R. 1959. A diagram of bromeliad habitats. Bromeliad Soc. Bull. 5: 70–71. Reitz, R. 1983. Bromeliáceas e a malária-Bromélia endémica. Flora Ilustrada Catarinense 1: 1–559. Remsen, J. V., Jr. 1985. Community organization and ecology of birds of high elevational humid forest of the Bolivian humid forest of the Bolivian Andes. In Neotropical Ornithology, eds. P. S. Buckley, M. S. Foster, E. S. Mortan, R. S. Ridgely and F. G. Buckley, pp. 732–756. Ornithological Monograph No. 36. Washington, D.C.: American Ornithologists Union. Rico-Gray, V. and Thien, L. B. 1989. Ant–mealybug interaction decreases reproductive fitness of Schomburgkia tibicinis (Orchidaceae) in Mexico. J. Trop. Ecol. 5: 109–112. 648 Literature cited Rivero, J. A. 1984. Bromeliad frogs of Puerto Rico. J. Bromeliad Soc. 32: 64–67. Rivero, J. A. 1989. Two beautiful bromeliad frogs from the Andes of Venezuela. J. Bromeliad Soc. 39: 26–27. Rivero, J. A. and Barard, M. A. 1983. An interesting association between an ant and a bromeliad. J. Bromeliad Soc. 29: 7–8. Robinson, H. 1969. A monograph on foliar anatomy of the genera Connellia, Cottendorfia and Navia (Bromeliaceae). Smithson. Contr. Bot. 2: 1–41. Robinson, J. W., Christian, C. M., Martinez, J. D. and Madhusudan, N. 1973. Spanish moss as an indicator of lead in the atmosphere before the use of leaded gasoline. Environ. Lett. 4: 87–93. Rocha, C. F. D., Van Sluys, M., Ornellas, A. B. et al. 1996. The effect of fire on natural populations of Vriesea neoglutinosa in a relict restinga of Espírito Santo State. Bromélia 3: 16–26. Röhweder, O. 1956. Die Farinosae in der Vegetation von El Salvador. Abhandl. Gebiet Auslandskunde 61: 1–197. Rossi, M. R., Méndez, V. H. and Monge-Nájera, J. 1997. Distribution of Costa Rican epiphytic bromeliads and the Holdridge life zone system. Rev. Biol. Trop. 45: 1021–1031. Rowe, J. H. 1963. Urban settlements in ancient Peru. Ñawpa Pacha 1: 1–28. Rudolph, D., Rauer, G., Neider, J. and Barthlott, W. 1998. Distributional patterns of epiphytes in the canopy and phorophyte characteristics in a western Andean rain forest in Ecuador. Selbyana 19: 27–33. Ruess, B. R. and Ellers, B. M. 1985. Correlation between crassulacean acid metabolism and water uptake in Senecio medley-woodii. Planta 166: 57–66. Ruinen, J. 1953. Epiphytosis. A second view on epiphytism. Ann. Bogor. 1: 101–157. Ruschi, A. 1949. A polenizacão realizada pelos Trochilideos, a sua área de alimentação e o repovoamento. Boletim do Museu de Biologia Professor Mello Leitão, Biologia 2: 1–51. Sakai, W. S. and Sanford, W. G. 1979. Ultrastructure of water-absorbing trichomes of pineapple (Ananas comosus, Bromeliaceae). Ann. Bot. 46: 7–12. Sale, P. J. M. and Neales, T. F. 1980. Carbon dioxide assimilation by pineapple plants, Ananas comosus (L.) Merr. I. Effects of daily irradiance. Aust. J. Pl. Physiol. 7: 363–373. Sazima, I., Vogel, S. and Sazima, M. 1989. Bat pollination of Encholirium glaziovii, a terrestrial bromeliad. Pl. Syst. Evol. 168: 167–179. Sazima, M., Buzato, S. and Sazima, I. 1995. Bat pollination of Vriesea in southeastern Brazil. Bromélia 2: 29–37. Scarano, F. R., Ribeiro, K. T., Luiz, F. D., De Moraes, F. D. and De Lima, H. C. 1997. Plant establishment on flooded and unflooded patches of a freshwater swamp forest in southeastern Brazil. J. Trop. Ecol. 14: 793–803. Scarano, F. R., Mattos, E. A. De, Franco, A. C. et al. 1999. Habitat segregation of C3 and CAM Nidularium (Bromeliaceae) in response to different light regimens in the understory of a swamp forest in southeastern Brazil. Flora 194: 281–289. Schaffer, W. M. and Gadgil, M. D. 1975. Selection for optimal life histories in plants. In Ecology and Evolution of Communities, eds. M. L. Cody and J. M. Diamond, pp. 142–157. Cambridge, MA: Harvard University Press. Schill, R., Dannenbaum, C. and Jentzsch, E. M. 1988. Untersuchungen an Bromeliennarben. Beitr. Biol. Pflanzen 63: 221–252. Schimper, A. F. W. 1884. Über Bau und Lebensweise der Epiphyten Westindiens. Bot. Centralbl. 17: 192–195. Literature cited 649 Schimper, A. F. W. 1888. Die epiphytische Vegetation Amerikas. Botanische Mitteilungen aus den Tropen, Heft 2. Jena: Gustav Fischer Verlag. Schimper, A. F. W. 1898. Pflanzengeographie auf physiologischer Grundlage. Jena: Gustav Fischer Verlag. Schindler, R. 1957. Vergleichende typologische und histogenetische Untersuchungen an den Sproßachsen einiger Bromeliaceen. Inaugural dissertation. Heidelburg: Universität Heidelburg. Schlesinger, W. J. and Marks, P. L. 1977. Mineral cycling and the niche of Spanish moss, Tillandsia usneoides L. Amer. J. Bot. 64: 1254–1262. Schmid, R. and Schmid, M. J. 1977. Fossil history of the Orchidaceae. In Orchid Biology – Reviews and Perspectives, vol. I, ed. J. Arditti, pp. 27–45. Ithaca, NY: Cornell University Press. Schmidt, C. 1992. Tillandsias: the Nicaraguan experience. J. Bromeliad Soc. 42: 9–11. Schmidt, J. and Kaiser, W. M. 1987. Response of the succulent leaves of Peperomia magnoliaefolia to dehydration. Plant Physiol. 83: 190–194. Schmidt, R. 1985. Functional interpretations of the morphology and anatomy of septal nectaries. Acta Bot. Neerl. 34: 125–128. Schrimpff, E. 1981. Air pollution patterns in two cities in Colombia S.A. according to trace substance content of an epiphyte (Tillandsia recurvata L.). Water, Air and Soil Pollut. 21: 279–315. Schroeder, H. A. 1970. A sensible look at air pollution by metals. Arch. Environ. Health 21: 798–806. Schubart, C. D., Diesel, R. and Blair Hedges, S. 1998. Rapid evolution to terrestrial life in Jamaican crabs. Nature 393: 363–365. Schulte, P. L. and Nobel, P. S. 1989. Responses of a CAM plant to drought and rainfall: capacitance and osmotic pressure influences on water movement. J. Exp. Bot. 40: 61–70. Schultes, R. E. and Raffauf, R. F. 1990. The Healing Forest. Portland, OR: Dioscorides Press. Schulz, E. 1930. Beiträge zur physiologischen und phylogenetischen Anatomie der vegetaten Organe der Bromeliaceen. Bot. Arch. 29: 122–209. Schulze, E. D., Gebauer, G., Schulze, W. and Pate, J. S. 1991. The utilization of nitrogen from insect capture by different growth forms of Drosera from southeast Australia. Oecologia 87: 240–246. Schürhoff, P. N. 1926. Die Zytologie der Blütenpflanzen. Stuttgart: F. Enke. Scogin, R. 1985. Floral anthocyanins in the genus Puya. Biochem. Syst. Ecol. 13: 387–389. Scogin, R. and Freeman, C. E. 1984. Floral pigments and nectar constituents in the genus Puya (Bromeliaceae). Aliso 10: 617–619. Seidel, J. L., Epstein, W. W. and Davidson, D. W. 1990. Neotropical ant gardens. I. Chemical constituents. J. Chem. Ecol. 16: 1791–1816. Sengupta, B., Nandi, A. S., Samanta, R. K., Pal, D., Sengupta, D. N. and Sen, S. P. 1981. Nitrogen fixation in the phyllosphere of tropical plants: occurrence of phyllosphere nitrogen-fixing microorganisms in eastern India and their utility for the growth and nitrogen nutrition of host plants. Ann. Bot. 48: 705–716. Shacklette, H. T. and Connor, J. J. 1973. Airborne chemical elements in Spanish moss. Geological Survey Professional Paper 574-E. Washington, D.C.: US Government Printing Office. Sharma, A. K. and Ghosh, I. 1971. Cytotaxonomy of the family Bromeliaceae. Cytologia 36: 237–247. 650 Literature cited Sheline, J. R. and Winchester, J. W. 1976. Trace element similarity groups in north Florida Spanish moss: evidence for direct uptake of aerosol particles. J. Geophys. Res. 81: 1047–1050. Shubert, T. S. 1990. Epiphytic bromeliads on Florida trees. Plant Pathology Circular No. 333. Gainesville, FL: Florida Department of Agriculture and Consumer Service. Sideris, C. P. and Krauss, B. H. 1928. Water relations of pineapple plants. Soil Sci. 26: 305–315. Sideris, C. P. and Krauss, B. H. 1955. Transpiration and translocation phenomena in pineapples. Amer. J. Bot. 42: 707–709. Sieber, J. 1955. Untersuchungen über die Wasser- und Nährstoffaufnahme bei epiphytischen trichterbuildenden Bromeliaceen. Gartenbauwissenschaft 2: 141–164. Sillett, T. S. 1994. Foraging ecology of epiphyte-searching insectivorous birds in Costa Rica. Condor 96: 863–877. Simpson, B. B. and Connor-Ogorzaly, M. 1995. Economic Botany: Plants in our World, 2nd edn. New York: McGraw-Hill, Inc. Simpson, M. G. 1988. A critique of ‘Bromeliales, related monocots, and resolution of relationships among Bromeliaceae subfamilies’. Syst. Bot. 13: 610–614. Skillman, J. B. and Winter, K. 1997. High photosynthetic capacity in a shadetolerant crassulacean acid metabolism plant. Plant Physiol. 113: 441–450. Skillman, J. B., Garcia, M. and Winter, K. 1999. Whole cell consequences of crassulacean acid metabolism for a tropical forest understory plant. Ecology 80: 1584–1593. Skotak, C. 1989. Endangered species, my view. J. Bromeliad Soc. 39: 147–149. Smith, J. A. C. 1989. Epiphytic bromeliads. In Vascular Plants as Epiphytes: Evolution and Ecophysiology, ed. U. Lüttge, pp. 109–138. Berlin: SpringerVerlag. Smith, J. A. C., Griffiths, H., Bassett, M. and Griffiths, N. M. 1985. Day–night changes in the water relations of epiphytic bromeliads in the rain forest of Trinidad. Oecologia 67: 475–485. Smith, J. A. C., Griffiths, H., Lüttge, U., Crook, C. E., Griffiths, N. M. and Stimmel, K. 1986. Comparative ecophysiology of CAM and C3 bromeliads. IV. Plant water relations. Plant Cell Environ. 9: 395–410. Smith, L. B. 1934a. Geographical evidence on the lines of evolution in the Bromeliaceae. Bot. Jahrb. 66: 446–468. Smith, L. B. 1934b. Studies in Bromeliaceae – V. Contr. Gray Herb. 104: 71–83. Smith, L. B. 1962. Origins of the flora of Southern Brazil. Contr. US Nat. Herb. 35: 215–249. Smith, L. B. 1967. Notes in Bromeliaceae. Phytologia 15: 180–200. Smith, L. B. 1986. Revision of the Guayana Highland Bromeliaceae. Ann. Missouri Bot. Gard. 73: 689–721. Smith, L. B. and Downs, R. J. 1974. Pitcairnioideae (Bromeliaceae). Flora Neotropica Monograph 14, Part 1. New York: Hafner Press. Smith, L. B. and Downs, R. J. 1977. Tillandsioideae (Bromeliaceae). Flora Neotropica Monograph 14, Part 2. New York: Hafner Press. Smith, L. B. and Downs, R. J. 1979. Bromelioideae (Bromeliaceae). Flora Neotropica Monograph 14, Part 3. Bronx, NY: New York Botanical Garden. Smith, L. B. and Kress, W. J. 1989. New or restored genera of Bromeliaceae. Phytologia 66: 70–79. Literature cited 651 Smith, L. B. and Spencer, M. A. 1992. Reduction of Streptocalyx (Bromelioideae: Bromeliaceae). Phytologia 72: 96–98. Snow, B. K. and Snow, D. W. 1971. The feeding ecology of tanagers and honeycreepers in Trinidad. Zoologia 46: 27–48. Snow, D. W. and Snow, B. K. 1986. Feeding ecology of hummingbirds in Serra do Mar, southeastern Brazil. El Hornero 12: 286–296. Soltis, D. E., Gilmartin, A. J., Rieseberg, L. and Gardner, S. 1987. Genetic variation in the epiphyte Tillandsia ionantha and T. recurvata. Amer. J. Bot. 74: 531–537. Soukup, J. 1970. Vocabulario de los Nombres Vulgares de la Flor Peruana. Lima: Colegio Salesiano. Spencer, M. A. and Smith, L. B. 1993. Racinaea, a new genus of Bromeliaceae (Tillandsioideae). Phytologia 74: 151–160. Stearns, S. C. 1976. Life history tactics: a review of the ideas. Q. Rev. Biol. 51: 3–47. Steele, A. R. 1964. Flowers for the King. Durham, NC: Duke University Press. Stephenson, S. C. and Landolt, J. C. 1998. Dictyostelid cellular slime molds in canopy soils of tropical forests. Biotropica 30: 657–661. Sternberg, L., De Niro, M. J. and Johnson, H. B. 1984. Isotope ratios of cellulose from plants having different photosynthetic pathways. Plant Physiol. 74: 557–561. Stewart, G. R., Schmidt, S. C., Handley, L. L., Turnbull, M. H., Erskine, P. D. and Joly, C. A. 1995. 15N natural abundance in vascular rainforest epiphytes: implications for nitrogen source and acquisition. Plant Cell Environ. 18: 85–90. Steyermark, J. A., Berry, P. E. and Holst, B. K. 1995. Flora of the Venezuelan Guayana, vol. 1. Portland, OR: Timber Press. Stiles, E. W. 1993. The influence of pulp lipids on fruit preference by birds. Vegetatio 107/108: 227–235. Stiles, F. G. 1978. Temporal organization of flowering among the hummingbird foodplants of a tropical wet forest. Biotropica 10: 194–210. Stiles, F. G. and Freeman, C. E. 1993. Patterns of floral nectar characteristics of some bird-visited plant species from Costa Rica. Biotropica 25: 191–205. Stiles, K. C. and Martin, C. E. 1996. Effects of drought stress on CO2 exchange and water relations in the CAM epiphyte Tillandsia utriculata (Bromeliaceae). J. Plant Physiol. 149: 721–728. Strehl, T. and Winkler, S. 1983. Vergleichende Studien zum Bau der Bromeliaceenwurzeln. Beitr. Biol. Pflanzen 58: 219–235. Strehl, V. T. 1983. Forma, distribuicão e flexibilidade dos trichomas foliares usados na filogenia de bromeliáceas. Iheringia. Sér. Bot., Porto Alegre 31: 105–119. Strehl, V. T. and Winkler, S. 1981. Vergleichende Untersuchungen über die Trichome der Bromeliaceen. Beitr. Biol. Pflanzen 56: 415–438. Stuart, G. E. 1992. Maya heartland under siege. Nat. Geog. 182: 95–107. Subils, R. 1973. Poliembrionía en especies argentinas de Tillandsia (Bromeliaceae). Kurtziana 7: 266–267. Suessenguth, K. 1921. Beiträge zur Frange des systematischen Anschlusses der Monokotylen. Beih. Bot. Centralbl. 38: 31. Sugden, A. M. 1981. Aspects of the ecology of vascular epiphytes in two Colombian cloud forests. II. Habitat preferences of Bromeliaceae in the Serrania de Macuria. Selbyana 5: 264–273. Sugden, A. M. and Robins, R. J. 1979. Aspects of the ecology of vascular 652 Literature cited epiphytes in Colombian cloud forest. I. The distribution of the epiphyte flora. Biotropica 11: 173–188. Sun, G., Dilcher, D. L., Zheng, S. and Zhou, Z. 1998. In search of the first flower: a Jurassic angiosperm, Archaefructus, from northeast China. Science 282: 1692–1695. Szidat, L. 1922. Die samen der Bromeliaceen in ihrer Anpassung an den Epiphytismus. Bot. Arch. 1: 29–46. Tanner, E. V. J. 1977. Four montane rain forests of Jamaica: a quantitative characterization of floristics, the soils and the foliar mineral levels, and a discussion of the interrelations. J. Ecol. 65: 883–918. Tanner, E. V. J. 1980. Studies on the biomass and productivity in a series of montane rain forests in Jamaica. J. Ecol. 68: 573–588. Taylor, D. W. and Hickey, L. J. 1992. Phylogenetic evidence for the herbaceous origin of angiosperms. Pl. Syst. Evol. 180: 137–156. Ter Steege, H. and Cornelissen, J. H. C. 1989. Distribution ecology of vascular epiphytes in lowland rain forest of Guayana. Biotropica 21: 331–339. Terry, R. G. and Brown, G. K. 1991. Evaluation of generic limits in Bromeliaceae subfamily Tillandsioideae using chloroplast DNA (cp DNA) restriction site analysis. Suppl. Amer. J. Bot. 78: 223 (Abstract). Terry, R. G., Brown, G. K. and Olmstead, R. G. 1997a. Examination of subfamily phylogeny in Bromeliaceae using comparative sequencing of the plastid locus ndhF. Amer. J. Bot. 84: 664–670. Terry, R. G., Brown, G. K. and Olmstead, R. G. 1997b. Phylogenetic relationships in subfamilly Tillandsioideae (Bromeliaceae) using ndhF sequences. Syst. Bot. 22: 333–345. Thompson, J. N. 1981. Reversed animal–plant interactions: the evolution of insectivorous and ant-fed plants. Biol. J. Linn. Soc. 6: 147–155. Thorne, B. L., Haverty, M. I. and Benzing, D. H. 1996. Associations between termites and bromeliads in two dry tropical habitats. Biotropica 28: 781–785. Tietze, M. 1906. Physiologische Bromeliaceen-Studien II. Die Entwicklung der wasseraufnehmenden Bromeliaceen-Trichome. Z. Naturwissen. 78: 1–50. Till, W. 1984. Sippendifferenzierung innerhalb Tillandsia subgenus Diaphoranthema in Südamerika mit besonderer Berücksichtigung des Andenostrandes und der angrenzenden Gebiete. PhD thesis. Wien: Universität Wien. Till, W. 1989a. Die Untergattung Diaphoranthema (Beer) C. Koch von Tillandsia Linnaeus. 1. Das Tillandsia capillaris Aggregat. Die Bromelie 2/89: 31–34. Till, W. 1989b. Die Untergattung Diaphoranthema von Tillandsia Linnaeus. 2. Teil: Das Tillandsia loliacea Aggregat. Die Bromelie 3/89: 55–59. Till, W. 1991. Die Untergattung Diaphoranthema von Tillandsia Linnaeus. 3. Teil: Das Tillandsia rectangula Aggregat. Die Bromelie 1/91: 15–19. Till, W. 1992a. Systematics and evolution of the tropical-subtropical Tillandsia subgenus Diaphoranthema (Bromeliaceae). Selbyana 13: 88–94. Till, W. 1992b. Die Untergattung Diaphoranthema von Tillandsia Linnaeus. 4. Teil: Das Tillandsia recurvata Aggregat. Die Bromelie 1/92: 15–20. Till, W. 1995. Tillandsia – evolution of an artificial genus? Resumos XLVI Congresso Nacional de Botânica 306, 349. Ribeirão Preto: Universidade de São Paulo. Till, W., Halbritter, H. and Gortan, G. 1997. Some notes on the remarkable bromeliad genus Glomeropitcairnia. J. Bromeliad Soc. 47: 65–72. Ting, I. P., Bates, I., Sternberg, L. O. and DeNiro, M. J. 1985. Physiological and isotopic aspects of photosynthesis in Peperomia. Plant Physiol. 78: 246–249. Literature cited 653 Todzia, C. 1986. Growth habits, host tree species, and density of hemiepiphytes on Barro Colorado Island, Panama. Biotropica 18: 22–27. Tomlinson, P. B. 1969. Commelinales–Zingiberales. In Anatomy of the Monocotyledons, ed. C. R. Metcalfe, pp. 193–294. Oxford: Clarendon Press. Tomlinson, P. B. 1970. Monocotyledons: toward an understanding of their morphology and anatomy. Adv. Bot. Res. 3: 207–292. Towle, M. A. 1961. The Ethnobotany of Pre-Colombian Peru. Chicago, IL: Aldine Publishing Co. TRAFFIC (Trade Analysis of Flora and Fauna In Commerce), sponsored by the World Wide Fund for Nature (WWF) and the World Conservation Union (IUCN). For regional office addresses contact: TRAFFIC International, 219c Huntingdon Road, Cambridge CB3 0DL, UK. TRAFFIC Germany 1988. Trade in grey-leafed tillandsias in the Federal Republic of Germany. Unpublished interim report. TRAFFIC Europe 1992. See www.traffic.org TRAFFIC USA 1992. Unpublished report; data compiled from United States Department of Agriculture, Animal and Plant Health Inspection Service (APHIS). Tukey, H. B. 1970. The leaching of substances from plants. Ann. Rev. Plant Physiol. 21: 305–324. Turner, R. M., Alcorn, S. M. and Olin, G. 1969. Mortality of transplanted saguaro seedlings. Ecology 50: 835–844. Ueno, C. 1989. Morphologische Studien am Gynözeum der Bromeliaceae. PhD thesis. Mainz: Johannes Gutenberg-Universität. Ule, E. 1906. Ameisenpflanzen. Bot. Jahrb. Syst. 37: 335–352. Ulubelen, A. and Mabry, T. J. 1982. Flavonoids of Tillandsia utriculata L. (Bromeliaceae). Rev. Latinoamer. Quím. 13: 35. Usher, G. 1974. A Dictionary of Plants used by Man. New York: Hafner Press. Utley, J. F. 1983. A revision of the middle American thecophylloid Vriesea (Bromeliaceae). Tulane Stud. Zool. Bot. 24: 1–81. Valdivia, P. E. 1977. Estudio botanico y ecológico de la región del Rio Uxpanapa, Vera Cruz. IV. Las epifitas. Biotica 2: 55–81. Van Sluys, M. and Stotz, D. F. 1995. Patterns of hummingbird visitation to Vriesea neoglutinosa in Spirito Santo, southeastern Brazil. Bromélia 2: 27–35. Vance, E. D. and Nadkarni, N. M. 1990. Microbial biomass and activity in canopy organic matter and the forest floor of a tropical cloud forest. Soil Biol. Biochem. 22: 677–684. Varadarajan, G. S. 1990. Patterns of geographic distribution and their implications on the phylogeny of Puya (Bromeliaceae). J. Arnold Arbor. 71: 527–552. Varadarajan, G. S. and Brown, G. K. 1988. Morphological variation of some floral features of the subfamily Pitcairnioideae (Bromeliaceae) and their significance in pollination biology. Bot. Gaz. 149: 82–91. Varadarajan, G. S. and Gilmartin, A. J. 1988a. Phylogenetic relationships of groups of genera within the subfamily Pitcairnioideae (Bromeliaceae). Syst. Bot. 13: 283–293. Varadarajan, G. S. and Gilmartin, A. J. 1988b. Seed morphology of the subfamily Pitcairnioideae (Bromeliaceae) and its systematic implications. Amer. J. Bot 75: 808–818. Vasak, V. 1969. The viability of bromeliad seeds. Bromeliad Soc. Bull. 19: 102–104. Veloso, H. P. 1952. O problema ecológico: vegetacao-Bromelias-Anofelinos. I. A 654 Literature cited presenca relativa das formas aquaticas do A. (Kerteszia) spp. como indice de positividade das espécies de Bromeliaceas. Sellowia 4: 187–240. Veloso, H. P. and Klein, R. M. 1957. As communidades e associçoes vegetais de mata pluvial do sul do Brazil I. Sellowia 8: 81–235. Vijayaraghavan, M. R. and Bhatia, K. 1985. Cellular changes during microsporogenesis, vegetative and generative cell formation: a review based on ultrastructure and histochemistry. Int. Rev. Cytol. 96: 263–297. Vogel, S. 1969. Chiropterophilie in der neotropischen Flora. Neue Mitteil. III. Flora 158: 289–323. Vogelmann, T. C. and Martin, G. 1993. The functional significance of palisade tissue: penetration of directional versus diffuse light. Plant Cell Environ. 16: 65–72. von Reis Altchul, S. 1973. Drugs and Foods from Little Known Plants: Notes in the Harvard University Herbaria. Cambridge, MA: Harvard University Press. von Reis, S. and Lipp, F. J., Jr. 1982. New Plant Sources for Drugs and Foods from the New York Botanical Garden Herbarium. Cambridge, MA: Harvard University Press. Wake, D. B. 1987. Adaptive radiation of salamanders in middle American cloud forests. Ann. Missouri Bot. Gard. 74: 242–264. Walker, L. R. 1991. Tree damage and recovery from hurricane Hugo in Luquillo Experimental Forest, Puerto Rico. Biotropica 23: 379–385. Wanderly, M. das Graças Lapa 1984. Contribuição à paliotaxonomia da família Bromeliaceae. MSc thesis. São Paulo: Universidade de São Paulo. Weir, J. S. and Kiew, R. 1986. A reassessment of the relations in Malaysia between ants (Crematogaster) on trees (Leptospermum and Dacrydium) and epiphytes of the genus Dischidia (Asclepiadaceae) including ‘ant-plants’. Biol. J. Linn. Soc. 27: 113–132. Weiss, H. E. 1965. Étude caryologique et cyto-taxonomique de quelques Broméliacées. Mém. Mus. Nat. Hist. Naturelle, Sér. B, Bot. 16: 9–35. Weyl, R. 1964. Die paläogeographische Entwicklung des mittelamerikanischwestindischen Raumes. Geol. Rundschau 54: 1213–1240. Wheeler, W. M. 1921. A new case of parabiosis and the ‘ant gardens’ of British Guiana. Ecology 2: 89–103. Wheeler, W. M. 1942. Studies of Neotropical ant plants and their ants. Bull. Mus. Comp. Zool. Harvard Coll. 90: 131–154. Wherry, E. T. and Capen, R. G. 1928. Mineral constituents of Spanish moss and ball moss. Ecology 9: 501–504. Wiley, E. O. 1988. Vicariance biogeography. Ann. Rev. Ecol. Syst. 19: 513–542. Williams, R. O. 1951. The Useful and Ornamental Plants in Trinidad and Tobago, 4th edn. Port-of-Spain: Guardian Commercial Printery. Williams, C. A. 1978. The systematic implications of the complexity of leaf flavonoids in the Bromeliaceae. Phytochemistry 17: 729–734. Wilson, C. R. 1989. Plant uses. In Encyclopedia of Southern Culture, eds. C. R. Wilson and W. Ferris, pp. 352–353. Chapel Hill, NC: University of North Carolina Press. Wilson, E. O. 1987. The arboreal ant fauna of Peruvian Amazonian forests: a first assessment. Biotropica 19: 245–282. Winkler, S. 1980. Ursachen der Verbreitungsmuster einiger Bromeliaceae in Rio Grande do Sul (Südbrasilien). Flora 170: 371–393. Winkler, S. 1986. Differenzierungen und deren Ursachen innerhalb der Bromeliaceen. Beitr. Biol. Pflanzen 61: 283–314. Winkler, S. 1990. Zur Evolution der Gattung Tillandsia L. Bot. Jahrb. Syst. 112: 43–77. Literature cited 655 Wittmack, L. 1888. Bromeliaceae. In Die Natürlichen Pflanzenfamilien, vol. II (4), eds. A. Engler and K. Prantl, pp. 32–59. Leipzig: W. Engelmann. Wolf, K. H. 1991. Protein coding genes in chloroplast DNA: a compilation of nucleotide sequences, data base entries and rates of molecular evolution. In The Photosynthetic Apparatus: Molecular Biology and Operation, eds. L. Bogorad and I. K. Vasil, pp. 467–482. Vol. 7 in Cell Culture and Somatic Cell Genetics in Plants, editor-in-chief I. K. Vasil. New York: Academic Press. Wollenweber, E. 1990. On the distribution of exudate flavonoids among Angiosperms. Rev. Latinoamer. Quím. 21: 115–121. Wollenweber, E. and Mann, K. 1992. A myricetin tetramethyl ether from the leaf and stem surfaces of Tillandsia usneoides. Z. Naturforsch. 47c: 638–639. World Resources Institute and The International Institute for Environment and Development 1986. New York: Basic Books Inc. Wright, J. S. and Calderon, O. 1995. Phylogenetic patterns among tropical flowering phenologies. J. Ecol. 83: 939–948. Wülfinghoff, R. 1967. Collecting tillandsias in Mexico. Bromeliad Soc. Bull. 17: 28–32. Wunderlin, R. P. 1998. Guide to the Vascular Plants of Florida. Gainesville, FL: University Press of Florida. Wurthman, E. 1984. Brazilian vrieseas prove to be freeze stalwarts. J. Bromeliad Soc. 34: 252–254. Yeaton, R. I. and Gladstone, D. E. 1982. The pattern of colonization of epiphytes on Calabash trees (Crescentia alata HBK) in Guanacaste Province, Costa Rica. Biotropica 14: 137–140. Young, A. M. 1979. Arboreal movements and tadpole-carrying behavior of Dendrobates pulmilio (Dendrobatidae) in northeastern Costa Rica. Biotropica 11: 218–219. Yu, W. 1994. The structural role of epiphytes in ant gardens. Biotropica 26: 217–221. Zahl, P. A. 1975. Hidden worlds in the heart of a plant. Natl. Geogr. Mag. 147: 389–397. Zavortink, T. J. 1973. Mosquito studies (Diptera: Culicidae). XXIX. A review of the subgenus Kerteszia of Anopheles. Contr. Am. Ent. Inst. 9: 1–62. Ziereis, H. and Arnold, F. 1986. Gaseous ammonia and ammonium ion in the free troposphere. Nature 321: 503. Zimmerman, J. K. and Olmsted, I. C. 1992. Host tree utilization by vascular epiphytes in a seasonally inundated forest (Tintal) in Mexico. Biotropica 24: 402–407. Zoller, W. H., Gordon, G. E., Gladney, E. S. and Jones, A. G. 1973. The sources and distribution of vanadium in the atmosphere. In Trace Elements in the Environment, pp. 31–47. Advances in Chemistry Series No. 123. Washington, D.C.: American Chemical Society. Zotz, G. 1997a. Substrate use by three bromeliads. Ecography 20: 264–270. Zotz, G. 1997b. Photosynthetic capacity increases with plant size. Bot. Acta 110: 306–308. Zotz, G. and Andrade, J. L. 1997. Water relations of two co-occurring epiphytic bromeliads. J. Plant Physiol. 152: 545–554. Zotz, G. and Thomas, V. 1999. How much water is in the tank? Model calculations for two epiphytic bromeliads. Ann. Bot. 83: 183–192.