Molecular Phylogenetics and Evolution 71 (2014) 55–78
Contents lists available at ScienceDirect
Molecular Phylogenetics and Evolution
journal homepage: www.elsevier.com/locate/ympev
Adaptive radiation, correlated and contingent evolution, and net species
diversification in Bromeliaceae
Thomas J. Givnish a,*, Michael H.J. Barfuss b, Benjamin Van Ee c, Ricarda Riina d, Katharina Schulte e,f,
Ralf Horres g, Philip A. Gonsiska a, Rachel S. Jabaily h, Darren M. Crayn f, J. Andrew C. Smith i, Klaus Winter j,
Gregory K. Brown k, Timothy M. Evans l, Bruce K. Holst m, Harry Luther n, Walter Till b, Georg Zizka e,
Paul E. Berry o, Kenneth J. Sytsma a
a
Department of Botany, University of Wisconsin-Madison, Madison, WI 53706, USA
Department of Systematic and Evolutionary Botany, Faculty of Life Sciences, University of Vienna, Vienna A-1030, Austria
c
School of Natural Sciences, Black Hills State University, Spearfish, SD 57799, USA
d
Real Jardín Botánico, CSIC, Plaza de Murillo 2, Madrid 28014, Spain
e
Department of Botany and Molecular Evolution, Research Institute Senckenberg and J.W. Goethe University, Frankfurt am Main D-60325, Germany
f
Australian Tropical Herbarium, James Cook University, Cairns, QLD 4878, Australia
g
GenXPro, Frankfurt am Main 60438, Germany
h
Department of Biology, Rhodes College, Memphis, TN 38112, USA
i
Department of Plant Sciences, University of Oxford, Oxford OX1 3RB, United Kingdom
j
Smithsonian Tropical Research Institute, Balboa, Ancon, Republic of Panama
k
Department of Botany, University of Wyoming, Laramie, WY 82071, USA
l
Department of Biology, Grand Valley State University, Allendale, MI 49401, USA
m
Marie Selby Botanical Gardens, Sarasota, FL 34236, USA
n
Gardens By The Bay, National Parks Board Headquarters, Singapore 259569, Singapore
o
Department of Ecology and Evolutionary Biology, University of Michigan, Ann Arbor, MI 48109, USA
b
a r t i c l e
i n f o
Article history:
Received 22 May 2013
Revised 18 September 2013
Accepted 11 October 2013
Available online 26 October 2013
Keywords:
Biogeography
Evolutionary predictions
Epiphytes
Key innovations
Pollination syndromes
Species richness
a b s t r a c t
We present an integrative model predicting associations among epiphytism, the tank habit, entangling
seeds, C3 vs. CAM photosynthesis, avian pollinators, life in fertile, moist montane habitats, and net rates
of species diversification in the monocot family Bromeliaceae. We test these predictions by relating evolutionary shifts in form, physiology, and ecology to time and ancestral distributions, quantifying patterns
of correlated and contingent evolution among pairs of traits and analyzing the apparent impact of individual traits on rates of net species diversification and geographic expansion beyond the ancestral Guayana Shield. All predicted patterns of correlated evolution were significant, and the temporal and spatial
associations of phenotypic shifts with orogenies generally accorded with predictions. Net rates of species
diversification were most closely coupled to life in fertile, moist, geographically extensive cordilleras,
with additional significant ties to epiphytism, avian pollination, and the tank habit. The highest rates
of net diversification were seen in the bromelioid tank-epiphytic clade (Dcrown = 1.05 My 1), associated
primarily with the Serra do Mar and nearby ranges of coastal Brazil, and in the core tillandsioids
(Dcrown = 0.67 My 1), associated primarily with the Andes and Central America. Six large-scale adaptive
radiations and accompanying pulses of speciation account for 86% of total species richness in the family.
This study is among the first to test a priori hypotheses about the relationships among phylogeny, phenotypic evolution, geographic spread, and net species diversification, and to argue for causality to flow
from functional diversity to spatial expansion to species diversity.
Ó 2013 Elsevier Inc. All rights reserved.
1. Introduction and conceptual framework
Bromeliaceae (58 genera, ca. 3140 species) is the largest of the
37 families of flowering plants found mostly or exclusively in the
Neotropics (Stevens, 2013), and includes more epiphytic taxa than
⁄ Corresponding author. Fax: +1 262 7509.
E-mail address: givnish@wisc.edu (T.J. Givnish).
1055-7903/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.ympev.2013.10.010
any family worldwide except Orchidaceae (Gentry and Dodson,
1987; Benzing, 1987, 2000; Zotz, 2013). Bromeliads often impound
rainwater and detritus in ‘‘tanks’’ formed by the overlapping bases
of rosulate leaves, employ CAM photosynthesis, and bear absorptive trichomes on their leaf surfaces, providing the means to
weather drought and absorb water and nutrients on rocks and epiphytic perches (Pittendrigh, 1948; McWilliams, 1974; Benzing,
1980, 2000; Crayn et al., 2004; Givnish et al., 2007, 2011; Schulte
56
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
et al., 2009). Bromeliads constitute a large fraction of all species of
vascular epiphytes in Neotropical forests, are especially diverse at
mid-elevations and in areas of high rainfall and humidity, and display increasingly narrow endemism at higher elevations (Gentry
and Dodson, 1987; Kessler, 2001; Kreft et al., 2004; Krömer et al.,
2005; Linares-Palomino et al., 2009; Linares-Palomino and Kessler,
2009). They show little variation in chromosome number and have
centers of diversity in four mountainous regions, including Central
America, the Andes, the tepuis of the Guayana Shield, and the Serra
do Mar and nearby coastal ranges of the Brazilian Shield in South
America (Givnish et al., 2011).
Based on a fossil-calibrated phylogeny based on eight plastid
regions, Givnish et al. (2011) concluded that Bromeliaceae arose
in the Guayana Shield ca. 100 million years ago (Mya), with six
of the eight subfamilies containing all but 2% of current species
having diverged from each other over a relatively short period from
15 to 10 Mya, as bromeliads spread into the Andes, Amazonia, Central America, the Caribbean, and the Brazilian Shield. What functional traits did bromeliads acquire between 15 and 10 Mya that
allowed them to invade mountainous or dry regions beyond the
Guayana Shield and evolve the epiphytic habit? Which of these
traits, or the habitats or adaptive zones invaded, helped trigger rapid rates of net species diversification? Did such traits evolve once
or multiple times? Where did they evolve, and under what conditions? Which traits underwent correlated or contingent evolution,
and why? Answers to these fundamental questions are now within
reach, given the well supported, taxonomic ally and ecologically
well-stratified phylogeny for Bromeliaceae provided by Givnish
et al. (2011).
Several key innovations (sensu Simpson, 1944) – including epiphytism, the tank habit, water- and nutrient-absorptive leaf trichomes, CAM photosynthesis, and avian pollination – may have
allowed bromeliads to invade new adaptive zones in rain- and
cloud-forest treetops or arid regions and microsites and speciate
extensively there (Schimper, 1888; Mez, 1904; Pittendrigh, 1948;
McWilliams, 1974; Benzing, 1980, 2000; Givnish et al., 1984,
1997, 2004, 2007, 2010, 2011; Benzing et al., 1985; Gentry and
Dodson, 1987; Smith, 1989; Kessler and Krömer, 2000; Crayn
et al., 2004; Schulte et al., 2005, 2009; Givnish, 2010). Key landscapes (sensu Givnish, 1997) – including moist, fertile, dissected
mountainous regions – may also have triggered adaptive radiation
and pulses of speciation by offering a promising range of ecological
possibilities. We propose that a complex of evolutionary forces
(Fig. 1) can account for spatial and temporal patterns in the origin
of critical bromeliad traits and the invasion of key landscapes; that
this complex implies that these traits and landscapes should have
undergone correlated and contingent evolution (sensu Pagel, 1994),
associated primarily with orogenies of extensive, mineral-rich
cordilleras; and that variation in these traits and landscapes –
across lineages and across time – can help explain patterns in
the distribution and species diversification of the eight bromeliad
subfamilies. Our rationale is as follows:
1. Fertile, humid tropical montane habitats select for the
epiphytic habit, by favoring heavy rains and low evaporation rates at mid elevations (Gentry and Dodson, 1987; Grytnes and Beaman, 2006; Acharya et al., 2011) and a rich
nutrient rain derived from leachate and shed parts from host
trees, and possibly from animals or droppings derived ultimately from fertile soils (Janzen, 1974a,b, 1977; Benzing,
2000; Gentry and Emmons, 1987; Romero et al., 2006,
2010; Benner and Vitousek, 2007).
2. Epiphytism and fertile, humid montane habitats favor the
tank habit – Many bromeliads impound rainwater and
detritus in ‘‘tanks’’ formed by tightly overlapping leaf bases.
Tanks can provide epiphytes with a source of water and
nutrients tapped by absorptive leaf trichomes or adventitious roots (McWilliams, 1974; Benzing, 1980, 2000). However, young tank epiphytes are vulnerable to desiccation
given their high ratio of evaporative surface to water volume, and are unlikely to survive in lowlands with unpredictable rainfall and high temperatures and evaporation rates
(Krömer et al., 2006; Zotz et al., 2011). Tanks should thus
be more common in montane areas with higher rainfall,
greater humidity, and lower seasonality and in larger species
and individuals, and uncommon in small-bodied species and
arid areas. By capturing nutrients from falling debris, tanks
should be favored in areas with richer substrates.
3. Epiphytism should favor the evolution of ‘‘entangling
seeds’’ that permit ready attachment to twigs and
branches, and vice versa – Epiphytes must be able to attach
their seeds to hosts (Schimper, 1884). Bromeliads have
evolved three mechanisms of seed dispersal, involving finely
divided appendages (comas) in subfamily Tillandsioideae,
fleshy berries in subfamily Bromelioideae, and wing-like
appendages in the remaining six subfamilies (Givnish
et al., 2010). Of these, the first two can entangle seeds with
the substrate via wetted comas, or sticky regurgitates or
droppings.
4. Epiphytism or dry terrestrial sites favor the evolution of
CAM photosynthesis – CAM photosynthesis and leaf succulence reduce transpiration and prolong the period over
which carbon uptake can be maintained following the onset
of drought, albeit at the cost of low photosynthetic capacity
(Medina, 1974; Winter and Smith, 1996a,b). CAM should
thus be associated with atmospheric Tillandsia that absorb
water and nutrients solely via absorptive trichomes, given
their exposure, small body size and virtually absent water
Epiphytism
Entangling seeds
Fertile, moist
montane habitats
Tank habit
Dry habitats
Ornithophily
CAM
Absorptive
trichomes
Species richness
Fig. 1. Proposed schema of ecological and evolutionary forces driving the evolution of various traits in the family Bromeliaceae. One-way arrows indicate a single direction of
causality; two-way arrows, bi-directional causality. Although not every trait is expected to affect all others directly, the complex of causal drivers could indirectly tie together
many traits shown that are shown as unlinked.
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
5.
6.
7.
8.
9.
10.
storage (Benzing and Renfrow, 1974), and to a lesser extent
with tank epiphytes. Dry conditions should also favor CAM
in terrestrial bromeliads, at low elevations or above cloud
and elfin forests at high elevations. Low atmospheric CO2
in the past would have also favored CO2-concentrating
mechanisms like CAM (Arakaki et al., 2011).
In bromeliads, epiphytism favors absorptive leaf trichomes and vice versa – Epiphytes lack access to reliable
supplies of water and nutrients in the soil, and their roots
are often reduced to holdfasts (Pittendrigh, 1948; Benzing,
1990). Trichomes on the leaf bases of tank and atmospheric
bromeliads provide an alternative means of absorbing water
and nutrients (Mez, 1904; Benzing, 2000).
Fertile, humid montane habitats favor avian pollination –
Cool, wet conditions select for thermoregulating pollinators,
often hummingbirds in the Neotropics (Cruden, 1972; Mabberley, 1975; Feinsinger, 1983; Bawa, 1990; Givnish et al.,
2009). Floral nectar often lacks critical amino acids, so avian
pollinators must consume substantial amounts of insects as
well (Brice and Grau, 1991; Martínez del Rio, 1994; Fleming
and Nicolson, 2003; Yanega and Rubega, 2004). Extremely
infertile substrates should thus work against avian pollination, because such substrates favor heavy chemical defenses
in plants that, in turn, reduce the density of herbivorous
insects (see Janzen, 1974a,b, 1977).
Fertile, humid montane habitats in extensive cordilleras
favor high rates of net species diversification in epiphytic
lineages – Extensive shifts in ecological conditions over
short horizontal and vertical distances in montane regions
can foster rapid speciation in epiphytes (Gentry and Dodson,
1987). More extensive cordilleras have more geographic barriers for species specialized on particular habitats, and thus,
a greater potential for species diversification rate and total
species richness. Species at higher altitudes should face
more habitat barriers (e.g., valleys), fostering narrower
ranges for individual taxa (Ibisch, 1996; Kessler, 2002a,b)
and partly reversing the trend for species richness to decline
with elevation. As argued above, extremely infertile substrates should work against epiphytic diversity. Net rates
of species diversification should be especially high for
recently uplifted, ecologically unsaturated areas (Gentry,
1982; Gentry and Dodson, 1987; Benzing, 1990; Linder,
2008; Givnish, 2010).
Epiphytism favors high net rates of species diversification
– Epiphytism offers more impetus for speciation by providing a more diverse range of microsites than forest floors,
over a larger, better lit, more fragmented, more dynamic surface (Gentry and Dodson, 1987; Benzing, 1990; Gravendeel
et al., 2004). Selection for short generation times in twig specialists may favor high speciation rates in some lineages
(Benzing, 1990, 2000; Gravendeel et al., 2004; Richter
et al., 2009).
Entangling seeds favor higher net rates of species diversification – Entangling seeds characterize Tillandsioideae
(coma-like flight apparatus) and Bromelioideae (fleshy berries), and should result in greater dispersal ability than the
winged or unappendaged seeds of other subfamilies.
Increased seed movement might increase overall speciation
in epiphyte line ages by facilitating occasional long-distance
dispersal, colonization, isolation, and parallel bouts of speciation along the length of extensive cordilleras (Gentry, 1982;
Gentry and Dodson, 1987; Benzing, 1990; Givnish et al.,
2004, 2007; Gravendeel et al., 2004).
Avian pollination should favor higher species richness
and rates of species diversification – Coevolution with
more than 300 hummingbird species spawned by the recent
57
uplift of the northern Andes may have accelerated bromeliad
speciation (Gentry, 1982; Graham, 1997; Kay et al., 2005).
Hummingbirds may have also accelerated speciation by
favoring gullet-shaped flowers. Once such exclusionary blossoms evolve, their length and shape could easily shift to
attract species with different bill lengths and shapes, providing a rapidly evolved means of plant premating isolation
(Givnish, 2010).
Our model (Fig. 1) predicts that epiphytism should exhibit patterns of correlated and contingent evolution with entangling seeds,
the tank habit, CAM photosynthesis, ornithophily, absorptive trichomes, and fertile, moist montane habitats. It also implies that
these characteristics should, directly or indirectly, result in higher
aggregate levels of species richness and species diversification
associated with radiation into the epiphytic adaptive zone and
dry terrestrial microsites. Here we test these hypotheses by (1)
overlaying key traits on the most recently derived bromeliad phylogeny as a function of time and spatial location; (2) testing
whether these traits exhibit predicted patterns of correlated and
contingent evolution; and (3) assessing whether the such traits
are correlated with species richness and net rates of species diversification within Bromeliaceae. Our conclusions have broad implications for the evolution of epiphytism and the impact of
individual traits and ecological conditions on the genesis of plant
diversity (Givnish, 2010). They also illustrate how phylogeny, ecology, physiology, and biogeography can be integrated to develop
and test a priori hypotheses about the relationships among phenotypic innovation, geographic spread, and species diversification of
a major group.
2. Methods
2.1. Taxon sampling
We used a placeholder approach in this study and our preceding
analysis of phylogenetic relationships and historical biogeography
in Bromeliaceae (Givnish et al., 2011). We employed 90 species
stratified across 46 of 58 bromeliad genera – which collectively include >97.5% of all described bromeliad species – to represent a
family of ca. 3140 species. We believe that this approach is justified, based on the high degree of uniformity within genera (and often within subfamilies) for most of the character-states under
study (see Smith and Down, 1974, 1977, 1979; Crayn et al.,
2004; Schulte et al., 2009), and based on a general presumption
of phenotypic conservatism across close relatives (Webb, 2000;
Donoghue, 2008). Where variation in key traits under study occurs
within genera (e.g., presence/absence of the tank habit in Brocchinia, or of CAM in Puya and Tillandsia), we attempted to include
species representing the taxonomic extent of alternative
character-states based on published phylogenies for those groups
(e.g., Givnish et al., 1997; Barfuss et al., 2005; Schulte et al.,
2009; Sass and Specht, 2010).
2.2. Phylogenetic analyses
Character-state reconstructions, tests of correlated and contingent evolution, estimates of net rates of diversification, and assessments of character determinants of speciation used phylogenetic
trees for Bromeliaceae derived by Givnish et al. (2011) from 9341
aligned plastid nucleotides from the 90 bromeliad species and four
outgroups from Typhaceae, Rapateaceae, and Arecaceae. Very few
natural hybrids have been detected among bromeliads via comparisons of nuclear vs. plastid trees, justifying the use of a plastid phylogeny alone to infer evolutionary relationships (see Schulte et al.,
58
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
2009; Jabaily and Sytsma, 2010; Sass and Specht, 2010). Trees used
were generated based on maximum likelihood (ML) (Givnish et al.,
2011) and Bayesian inference (BI) using BEAST v 1.7.4 (Drummond
and Rambaut, 2007; Drummond et al., 2012a). Vriesea espinosae in
Givnish et al. (2011) is here considered Tillandsia sp. A; this taxon is
embedded in a large clade of Mexican Tillandsia based on matK and
rps16 sequences (M. Barfuss, pers. comm.).
Simultaneous estimations of phylogeny and divergence times
were executed in BEAST under a Yule tree prior (Yule, 1925) and
unlinked clock models, using the GTR + G + I model of evolution
suggested by the Akaike information criterion (Akaike, 1974) obtained in ModelTest v3.7 (Posada and Crandall, 1998). Calibration
dates for the root (Poales: 100 ± 1.0 My (mean ± s.d.)), Typhaceae
(Typha + Sparganium): 69.5 ± 1.5 My, and crown Bromeliaceae:
19 ± 2.0 My) were obtained from the monocot-wide analysis of
Givnish et al. (2011). In all analyses, we constrained Puya to be
monophyletic (Jabaily and Sytsma, 2010). BEAST analyses were
run for 40 million generations, with samples taken every 2000 generations. The first two million generations were discarded as burnin, and we then interpreted the trees in TreeAnnotator v. 1.7.4 prior
to visualization in FigTree v. 1.4. Log files were analyzed in tracer
(Rambaut and Drummond, 2007), and the effective sample size
values (ESS) were over 300 for all parameters. A random subset
of 100 prior probability trees was also saved for additional analyses
(see below).
2.3. Character coding
We obtained data on the taxonomic distribution of the states of
seven key functional and distributional characters from the literature. Binary character states were assigned for habit (terrestrial vs.
epiphytic), growth form (tank vs. non-tank), seeds (winged/unappendaged vs. entangling), photosynthetic pathway (C3 vs. CAM),
pollination syndrome (avian vs. non-avian), elevational distribution (low vs. high [P1000 m]), and arid/semi-arid habitat or
microsite (present vs. absent). To show when particular traits arose, ancestral state reconstructions were superimposed on the
BEAST chronogram. To show where traits arose, the maximumparsimony (MP) reconstruction of ancestral regions – very similar
to those obtained via ML and S-DIVA (Givnish et al., 2011) – were
also overlaid on the chronogram. Given the large differences in
ecology between interior vs. coastal portions of the Brazilian
Shield, and their invasion by different clades, here we separate
the coding of the Atlantic Forest region (including the Serra do
Mar and Serra da Mantiqueira) from the rest of the Brazilian Shield.
Data on habit, growth form, seeds, and habitat were drawn from
Smith and Down (1974, 1977, 1979), Givnish et al. (1997), Schulte
et al. (2009), and Zizka et al. (2009). Data on photosynthetic pathways were drawn from Crayn et al. (2004) (CAM for d13C > 20‰,
and C3 for d13C < 20‰), supplemented by d13C measurements for
Puya mima (R. Jabaily and T. Givnish) and Pitcairnia carinata and P.
feliciana (J. A. C. Smith), and identification of Tillandsia usneoides
(Spanish moss) as CAM by Kluge et al. (1973). For the few cases
in which no d13C value was available for a species, but all congeners studied have the same photosynthetic pathway, we coded
remaining species as having that pathway. Data on pollination syndromes (avian, bat, insect) were drawn from Vogel (1954, 1969),
Smith and Down (1974, 1977, 1979), Gardner (1986), Sazima
et al. (1989, 1995a, 1995b, 1996, 1999), Galetto and Bernardello
(1992), Till (1992), Smith and Till (1998), Benzing (2000), Kessler
and Krömer (2000), Dziedzioch et al. (2003), Canela and Sazima
(2005), Krömer et al. (2005, 2006, 2008), and Tschapka and von
Helversen (2007). For species lacking direct observations of pollinators, pollination syndromes were deduced from floral traits such
as corolla color, size, and shape, position of the anthers, and presence/absence of landing platforms (Vogel, 1954; Baker and Baker,
1990). Waser et al. (1996) has criticized this approach, but pollination syndromes have been shown to predict accurately hummingbird, bat, and hawkmoth pollination in several Neotropical plant
groups (Cruden, 1997; Krömer et al., 2008). Data on elevational
distributions were drawn from Smith and Down (1974, 1977,
1979) and an extensive literature search. Data on absorptive trichomes are too sparse and continuous in nature to be included in
our analyses, but such trichomes are common in tillandsioids
and tank-forming brocchinioids and bromelioids (Benzing et al.,
1976, 1985; Givnish et al. 1984, 1997), while hydrophobic trichomes are common in other groups (Pierce et al., 2001).
We scored elevational distribution in three ways (low vs. high
[>1000 m]; low or infertile substrates [i.e., Guayana Shield sandstones and quartzites] vs. high and fertile [i.e., found in the Andes,
Serra do Mar or similar mountains in the Brazilian Shield]; and low
or infertile or dry [i.e., Puya and Deuterocohnia taxa above the
cloud-trapping inversion layer on tropical mountains, ca. 3000 m
on large massifs and much lower on smaller mountains] vs. high,
fertile, and moist). Multiple scorings were needed because we
hypothesized (see above) that fertile, moist conditions at higher
elevations favor epiphytism and the tank habit, while fertile conditions at higher elevations (including those in cold, dry conditions
above the inversion layer) favor ornithophily. Highly infertile conditions at any elevation should be inimical to epiphytism, the tank
habit, and avian pollination. We also scored a composite eighth
character, epiphyte or arid/semi-arid habit (present/absent), to
test hypothesis 4 (see above).
2.4. Reconstruction of character-state evolution
We characterized evolutionary transitions of each of our focal
characters in terms of the number, directionality, and timing of inferred shifts. We implemented maximum parsimony (MP) and
Bayesian inference (BI) optimization of character evolution. MP
reconstruction utilized the ‘‘trace character’’ option in MacClade
(Maddison and Maddison, 2005) with the resolving option of ‘‘all
most parsimonious states at each node’’. The resulting ancestralstate reconstructions were visually displayed by color-coding the
branches of the ML chronogram.
BI (MCMC – Pagel, 1999) reconstructions were implemented in
BayesTraits v.1.0 (Pagel and Meade, 2007) using MultiState and a
random set of 100 Bayesian prior probability trees. To reduce some
of the uncertainty and arbitrariness of choosing priors under
MCMC, we used the hyperprior approach (the rjhp command) as
recommended (Pagel et al., 2004; Pagel and Meade, 2007). Combinations of hyperprior values (exponential or gamma, mean and
variance) and rate parameter values were explored to find acceptance rates when running the Markov chains of between 20% and
40% (as recommended by Pagel and Meade, 2007). All subsequent
analyses used the reversible-jump hyperprior command (rjhp
gamma 0 30 0 10) that seeded the mean and variance of the gamma prior from uniform hyperpriors on the interval 0 to 10, and a
rate parameter of 150 (rate parameters of 100 and 350 were used
only for pollination and elevation, respectively). All Bayesian analyses used 25 million generations, with sampling every 1000 generations and a burn-in period of 20,000 generations. Ancestral
reconstruction of character evolution under BI with the 100 randomly chosen PP trees was represented using pie charts to indicate
state probabilities at each node in the bromeliad chronogram.
2.5. Tests of correlated evolution and directionality
We tested for correlated evolution between each pair of characters using BayesTraits (Pagel and Meade, 2008) under BI using the
same methods and prior probability trees just described. We
implemented the BayesDiscrete module, which investigates
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
correlated evolution between a pair of discrete binary traits by
comparing the log likelihood of two models for independent vs.
dependent evolution of binary traits. The first model assumes that
the two states of two traits, such as habit (terrestrial vs. epiphytic)
and photosynthetic pathway (C3 vs. CAM), evolve independently
on the tree. This creates two rate coefficients per trait, or four rate
coefficients in all, that must be estimated. The second model allows
the traits to evolve in a correlated fashion, such that the rate of
change in one trait depends on the background state of the other.
The dependent model has four states, one for each combination of
the two binary traits, creating eight rate coefficients that must be
estimated in all (see Electronic appendix, Fig. A1). To determine
whether a character (e.g., habit) shows correlated evolution with
another trait (e.g., photosynthetic pathway), we compared the likelihood estimate of the independent model (L(I)) to that for the
dependent model (L(D)). The pattern of correlated evolution is considered significant when 2[L(D) L(I)] exceeds the critical value
(P < 0.05) for the v2 distribution with 4 d.f., based on a comparison
of eight vs. four estimated rate coefficients in the dependent vs.
independent model, respectively (Pagel, 1999). We also determined whether each of the eight transition parameters (qij in
Fig. A1) in the dependent model is significantly greater than zero.
Individual transition parameters were restricted to zero (e.g.,
q12 = 0) and the likelihood score of this seven-parameter dependent model was compared to the likelihood score of the full
eight-parameter, dependent model using a v2 distribution with 1
d.f. Comparisons involving the equivalent of autocorrelation (e.g.,
habitats P1000 m elevation vs. fertile, moist habitats P1000 m
elevation, or epiphytes vs. epiphyte or arid/semi-arid habitat) or
associations between different environmental variables were excluded from analysis.
2.6. Tests of contingent evolution
We tested hypotheses of contingent evolution between each
pair of characters (e.g., does CAM evolve equally frequently in terrestrial and epiphytic clades?) using BayesDiscrete. We tested specific hypotheses by restricting two rates to be equal (e.g., for
q12 = q34, where the former is the rate of evolution of CAM from
C3 in a terrestrial clade, and the latter is the same rate in an epiphytic clade; Fig. A1). This seven-parameter model was then compared to the full eight-parameter model described above using a v2
test with 1 d.f. A significant outcome indicates that the state of the
second character (terrestrial vs. epiphytic) influences the evolution
of the first (Pagel, 1994). Critical a values for v2 tests were adjusted
using Bonferroni corrections for the four comparisons involved in
each independent test (Friedman and Barrett, 2008).
2.7. Determinants of net rates of species diversification
We used phylogenetically unstructured and phylogenetically
structured analyses to estimate rates of species diversification.
We then related differences in diversification rate to variation
among clades in age, habit, growth form, pollination syndrome,
photosynthetic pathway, mode of seed dispersal, and geographic
distribution.
2.7.1. Phylogenetically unstructured analyses
We used phylogenetically unstructured analyses (t-tests and
regressions) focused on non-overlapping lineages (subfamilies or
subfamily subsets) (Magallón and Castillo, 2009), recognizing that
the rapid divergence of most subfamilies from each other between
15 and 10 million years ago implies an almost star-shaped phylogeny relating those subfamilies to each other (Givnish et al., 2011).
Species numbers per genus and subfamily (Table A1), or major
clades of similar rank (i.e., the bromelioid tank-epiphyte clade sis-
59
ter to Acanthostachys, the Xeric clade (Deuterocohnia–DyckiaEncholirium), the core tillandsioids (excluding Catopsis-Glomeropitcairnia); see Schulte et al. 2009 and Givnish et al. 2011) were
obtained from Luther (2008), with Pepinia sunk into Pitcairnia
(Robinson and Taylor, 1999). We estimated the rate of net species
diversification D for each subfamily and key subclade as D = (ln N)/
t, where N is the current number of species in a clade and t is its
estimated stem age, based on a simple, birth-only model:
N(t) = N(0) eDt (Magallón and Sanderson, 2001; Jansson and Davies,
2008). Similar calculations were performed using crown ages.
More complex, birth–death models to estimate D were used in
phylogenetically structured analyses (see below).
We tested the hypotheses that species richness and net diversification rates should be greater in younger vs. older lineages
(Magallón and Sanderson, 2001; Givnish et al., 2005, 2009; Rabosky and Lovette, 2008; Givnish, 2010); in epiphytic vs. terrestrial
lineages (Gentry and Dodson, 1987; Benzing, 1990, 2000); in hummingbird- vs. insect-pollinated lineages (Gentry, 1982; Kay et al.,
2005; Schmidt-Lebuhn et al., 2007); in CAM vs. C3 lineages (Silvera
et al., 2009); in line ages with entangling seeds vs. winged or
unappendaged seeds (Givnish et al., 2007); and in lineages occupying the extensive, relatively fertile, topographically complex
cordilleras of the Andes/Central America and Serra do Mar vs.
other areas (Gentry, 1982; Luteyn, 2002; Hughes and Eastwood,
2006) (Fig. 1). Given the one-sided nature of our predictions, we
used one-tailed t-tests to evaluate the significance of differences
in the average values of richness or net diversification of subfamilies (or groups of similar rank) characterized more or less entirely
by one state of a given character vs. the other state. For comparisons of richness or net diversification as a function of lineage
age, we used one-tailed t-tests to evaluate the significance of
regressions of those variables against lineage age. Tests were conducted on both raw and log-transformed data.
2.7.2. Phylogenetically structured analyses
We employed two recently developed ML methods that incorporate nonrandom/incomplete sampling and extinction to infer
significant shifts in diversification rates across a time-calibrated
phylogeny (MEDUSA 3.0.0, Alfaro et al., 2009) and test for correlations between character states and diversification rates (BiSSE:
Maddison et al., 2007; FitzJohn et al., 2009). MEDUSA requires a
time-calibrated phylogeny in which extant taxonomic diversity
can be assigned to monophyletic terminal clades in order to avoid
the problem of non-randomly sampled clades or unresolved clades.
After removing outgroups while leaving Bromeliaceae rooted, we
reduced the BEAST chronogram to 26 terminals (Table A1) to produce the most resolved tree representing bromeliad clades at the
generic level or above, to which we could confidently assign essentially all members of the family. Due to the present lack of phylogenetic knowledge within the large core Bromelioideae and
Tillandsioideae, several terminals within these groups are large
(up to 868 spp.). Based on Schulte et al. (2009), we included Orthophytum with Cryptanthus as the sister to Acanthostachys + the
tank-epiphyte clade of Bromelioideae. Based on Barfuss et al.
(2005), we maintained the monophyly of both Tillandsieae (Guzmania, Mezobromelia, Racinaea, Tillandsia) and Vrieseae (Alcantarea,
Vriesea, Werauhia) in Tillandsioideae. Steyerbromelia was included
in the Navia/Brewcaria clade. Only Disteganthus (2 spp.) and Fernseea (2 spp.) were not placed in the 26-terminal tree. We used
the March 2013 version of MEDUSA embedded in GEIGER 1.99-3
in R version 3.0.0 (R Core Team, 2012) to estimate the best fit for
rates of net species diversification (r = birth–death) and relative
extinction (e = death/birth) among applications of both pure-birth
and birth–death models to crown and stem ages. MEDUSA employs
a stepwise process to evaluate the support for increasingly complex models of diversification based on the difference in sample-
60
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
size corrected AIC scores. We also calculated diversification rates in
GEIGER for clades in which MEDUSA identified significantly acceleration of such rates (Magallón and Sanderson, 2001).
The BiSSE (Binary State Speciation and Extinction) model of
Maddison et al. (2007), as implemented in Diversitree (FitzJohn
et al., 2009) in R, was used to test the hypotheses that diversification rates should be greater in epiphytic vs. terrestrial lineages, in
hummingbird- vs. insect-pollinated lineages, in CAM vs. C3, in
line ages with entangling seeds vs. winged or unappendaged
seeds, and in lineages occupying the extensive, relatively fertile,
topographically complex cordilleras of the Andes/Central America and Serra do Mar vs. other areas. Proportions of taxa in each
clade bearing each character-state were estimated from the literature (Table A2). BiSSE employs ML optimization to estimate absolute rates (vs. relative rates in MEDUSA) of asymmetric character
change (q), speciation (k), and/or extinction (l) by maximizing
the likelihood of these parameters on our 26-terminal chronogram.
For each character, we compared three constrained models in BiSSE against the unconstrained model in which q, k, and l were allowed to vary. The three constrained models individually forced
q01 = q10, k01 = k10, and l01 = l10, respectively. For example, if
diversification rates are correlated with epiphytism, then the
unconstrained model should be significantly favored over the constrained model with k01 = k10. Because BiSSE often cannot complete
computations involving unresolved tips with >100 species (R. FitzJohn, pers. comm.), we had to scale down the size of the terminals
to a maximum of 90 species. This scaling roughly preserved relative clade size, although the smallest clades (always rounded up
to 1.0 species) are larger relative to species-rich clades than in
the unscaled data. Our rationale was that if BiSSE provided
evidence for significant correlation of diversification rates with a
character-state, despite the relative upscaling of the less diverse
clades, then this would be a conservative and hence strong finding
regarding the impact of that character-state on diversification.
2.8. Cumulative effects of multiple radiations
In addition to assessing whether particular traits accelerate net
diversification within the clades bearing them, we calculated the
fractions of the total number of present-day bromeliads added by
the acquisition of particular traits or suites of traits (e.g., epiphytism, CAM in terrestrial species) strongly associated with laterdivergent clades and the regions they invaded outside the ancestral
Guayana Shield. Our aim was to estimate the augmentation of total
bromeliad diversity due to the additive effects of broad-scale adaptive radiations, of the clades added and geographic areas invaded
as a result of the morphological and physiological innovations,
irrespective of any acceleration of diversification within particular
clades or in association with particular traits.
3. Results
3.1. Phylogeny and time-line
The BEAST chronogram is nearly identical in topology to the ML
tree of Givnish et al. (2011), but resolves Puya as monophyletic and
places Acanthostachys as the non-tank, epiphytic sister to the bromelioid tank-epiphyte clade (Fig. 2). Nodal ages under BEAST are
very similar to those derived using r8s (y = 1.107x
0.683,
r2 = 0.98), with BEAST tending to produce slightly older dates for
all but the shallowest and deepest events. The bromeliad stem is
dated to 97.5 Mya; the bromeliad crown, to 22.7 Mya. The six subfamilies sister to Lindmanioideae, containing 98% of present-day
taxa, arose within a relatively narrow window between 16.9 and
10.1 Mya.
3.2. Reconstruction of character-state evolution
Ancestral bromeliads appear to have been terrestrial, non-tankforming, and insect-pollinated, had winged seeds and C3 photosynthesis, and grew on infertile, moist substrates >1000 m elevation
(Figs. 2 and 3, A2–A4). Each focal character appears to have undergone at least two state transitions, and reconstructions based on
MP and BI broadly agree. Almost all cases of epiphytism trace to
two origins among the taxa surveyed (Fig. 2A). The first occurred
at the base of Tillandsioideae ca. 16.9–15.2 Mya, with dispersal
from the Guayana Shield into the Andes, Central America, and/or
the northern littoral of South America or Caribbean. The second occurred ca. 5.9 Mya in the Atlantic Forest, in the clade subtended by
and including Acanthostachys. Other scattered gains and losses are
detailed in the Electronic appendix.
Tanks appear to have arisen three times: at the base of Tillandsioideae, coincident with the rise of epiphytism; ca. 10 Mya in
Brocchinia within the Guayana Shield; and in Bromelioideae sister
to Acanthostachys ca. 5.8 Mya in the Atlantic Forest region, soon
after invasion from the Andes/central Chile via the Brazilian Shield
(Fig. 2B). The bromelioid tank-epiphyte clade is sister to epiphytic
but tankless Acanthostachys. Scattered losses occurred in five isolated taxa. Entangling seeds evolved without reversal from winged
seeds twice, at the base of Tillandsioideae (see above), and at the
base of Bromelioideae among terrestrial taxa ca. 10.1–9.4 Mya, in
the Andes/central Chile (Fig. A2A). CAM photosynthesis arose from
C3 photosynthesis at least five times, at the base of Bromelioideae–
Puyoideae ca. 10.7 Mya, in the Andes/central Chile; at the base of
the Xeric Clade (Deuterocohnia–Dyckia-Encholirium) ca. 11.3–8.1
Mya, in the Andes and/or the Brazilian Shield; at the base of Hechtia ca. 16.2–9.9 Mya, in Central America; in Tillandsia utriculata-T.
sp. A ca. 5.7 Mya, in Central America; and in widespread Tillandsia
usneoides ca. 3.5 Mya (Fig. 3A). Other possible gains and losses of
CAM are discussed in the Electronic appendix. CAM appears to be
evolutionarily labile in Puya and Tillandsia – or reflect developmental plasticity – given that close relatives have d13C values typical of
C3 and CAM photosynthesis.
Avian pollination – predominantly by hummingbirds – appears
to have arisen 2–3 times, in each case from insect pollination
(Fig. 3B). Ornithophily arose once in ancestral tillandsioids, or separately in Glomeropitcairnia 14.6 Mya (in the northern littoral of
South America and the Caribbean) and in the core tillandsioids
ca. 15.2 Mya (in the Andes/central Chile). Ornithophily also evolved
in the ancestor of Pitcairnioideae–Puyoideae–Bromelioideae ca.
15.9–14.1 Mya, in the Andes/central Chile. Reversions to insect
pollination from avian pollination appear to have occurred at least
seven times in Bromelioideae during the last 6 Mya; once in
Fosterella ca. 11.3 Mya, in the Andes; and four to five times in
Tillandsioideae. Bat pollination appears to have arisen four times
among the taxa surveyed – in Alcantarea, Encholirium, Werauhia
viridiflora, and Tillandsia viridiflora – in each case from birdpollinated ancestors.
Elevations P1000 m above sea level appear to have been invaded up to 13 times based on MP character-state reconstruction
(Fig. A2B). Reconstructions based on Bayesian inference reveal a
reasonable probability of six independent origins of high-elevation
habitats, including the common ancestor of all bromeliads, and
persistence in Navioideae and the bromeliad spine through one
of the two basal lineages in Bromelioideae. Reversion to habitats
<1000 m under BI appears to have occurred two to three times in
tillandsioids, once in Hechtia, up to four times in Navioideae, three
times in Pitcairnia, once in Dyckia-Encholirium, and once in Chilean
Puya (see Appendix). The evolution of moist habitats on fertile
substrates P1000 m elevation under MP largely parallel those
for moist habitats P1000 m, and appear to involve 9–12 separate
origins (Fig. A3A and B). Arid or semi-arid habitats or microsites
61
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
A. Habit
Wittrockia superba
Neoregelia pineliana
Eduandrea selloana
Aechmea nudicaulis
Quesnelia quesneliana
Canistropsis billbergioides
Edmundoa perplexa
Aechmea organensis
Billbergia decora
Lymania alvimii
Aechmea racinae
Araeococcus pectinatus
Araeococcus goeldianus
Canistrum aurantiacum
Aechmea haltonii
Aechmea bromeliifolia
Aechmea sphaerocephala
Hohenbergia stellata
Aechmea drakeana
Ronnbergia petersii
Aechmea lingulata
Acanthostachys strobilacea
Cryptanthus beuckeri
Ananas ananassoides
Ananas nanus
Pseudananas sagenarius
Fascicularia bicolor
Ochagavia carnea
Ochagavia elegans
Deinacanthon urbanianum
Terrestrial
Epiphyte
unresolved
Guayana Shield
Brazilian Shield
Atlantic Forest
Amazonia
Andes, S Chile
N South America,
Caribbean
C America
W Africa
B. Growth Form
Wittrockia superba
Neoregelia pineliana
Eduandrea selloana
Aechmea nudicaulis
Quesnelia quesneliana
Canistropsis billbergioides
Edmundoa perplexa
Aechmea organensis
Billbergia decora
Lymania alvimii
Aechmea racinae
Araeococcus pectinatus
Araeococcus goeldianus
Canistrum aurantiacum
Aechmea haltonii
Aechmea bromeliifolia
Aechmea sphaerocephala
Hohenbergia stellata
Aechmea drakeana
Ronnbergia petersii
Aechmea lingulata
Acanthostachys strobilacea
Cryptanthus beuckeri
Ananas ananassoides
Ananas nanus
Pseudananas sagenarius
Fascicularia bicolor
Ochagavia carnea
Ochagavia elegans
Deinacanthon urbanianum
Non-tank
Tank
Greigia sphacelata
Puya raimondii
Puya aequatorialis
Puya mima
Puya castellanosii
Puya laxa
Puya alpestris
Puya chilensis
Puya venusta
Dyckia dawsonii
Dyckia ferox
Encholirium scrutor
Encholirium irwinii
Deuterocohnia lotteae
Deuterocohnia glandulosa
Deuterocohnia longipetala
Fosterella petiolata
Greigia sphacelata
Puya raimondii
Puya aequatorialis
Puya mima
Puya castellanosii
Puya laxa
Puya alpestris
Puya chilensis
Puya venusta
Dyckia dawsonii
Dyckia ferox
Encholirium scrutor
Encholirium irwinii
Deuterocohnia lotteae
Deuterocohnia glandulosa
Deuterocohnia longipetala
Fosterella petiolata
Pitcairnia corallina
Pitcairnia hirtzii
Pitcairnia poortmanii
Pitcairnia wendlandii
Pitcairnia orchidifolia
Pitcairnia feliciana
Pitcairnia carinata
Pitcairnia heterophylla
Navia splendens
Navia phelpsiae
Pitcairnia corallina
Pitcairnia hirtzii
Pitcairnia poortmanii
Pitcairnia wendlandii
Pitcairnia orchidifolia
Pitcairnia feliciana
Pitcairnia carinata
Pitcairnia heterophylla
Navia splendens
Navia phelpsiae
Navia saxicola
Sequencia serrata
Navia saxicola
Sequencia serrata
Hechtia glomerata
Hechtia lindmanioides
Hechtia guatemalensis
Tillandsia complanata
Tillandsia usneoides
Tillandsia dodsonii
Racinaea ropalacarpa
Tillandsia utriculata
Tillandsia sp. A
Vriesea malzinei
Hechtia glomerata
Hechtia lindmanioides
Hechtia guatemalensis
Tillandsia complanata
Tillandsia usneoides
Tillandsia dodsonii
Racinaea ropalacarpa
Tillandsia utriculata
Tillandsia sp. A
Vriesea malzinei
Guzmania roezlii
Guzmania roezlii
Guzmania monostachia
Mezobromelia hutchisonii
Vriesea glutinosa
Guzmania monostachia
Mezobromelia hutchisonii
Vriesea glutinosa
Mezobromelia pleiosticha
Alcantarea duarteana
Mezobromelia pleiosticha
Alcantarea duarteana
Lindmania longipes
Connellia cf. nutans
Lindmania guianensis
Brocchinia uaipanensis
Brocchinia acuminata
Brocchinia prismatica
Lindmania longipes
Connellia cf. nutans
Lindmania guianensis
Brocchinia uaipanensis
Brocchinia acuminata
Brocchinia prismatica
Bromelioideae
Puyoideae
Pitcairnioideae
Navioideae
Hechtioideae
Tillandsioideae
Miocene
20
15
Pliocene Quaternary
10
5
0 Mya
Miocene
20
15
Lindmanioideae
Brocchinioideae
Pliocene Quaternary
10
5
0 Mya
Fig. 2. Evolution of (A) epiphytic vs. terrestrial habit, and (B) tank vs. non-impounding growth form. Colors of branches reflect reconstruction of ancestral character-states
using MP; pie-diagrams represent the probabilities of different states at each node under Bayesian inference. Colors of shaded boxes indicate the ancestral distribution of taxa
under MP after Givnish et al. (2011); absence of color shows uncertainty of inferred ancestral region. The inset map shows the geographic arrangement of different ancestral
regions. Branch lengths represent inferred times, keyed by the scales at the bottoms of the figures. Brackets at right delimit bromeliad subfamilies.
were invaded six times, by Hechtia, the Xeric Clade, and Puya plus
Bromelioideae; by Alcantarea ca. 6.8 Mya in the Brazilian Shield; by
Ananas ananassoides ca. 6.6–1.0 Mya in the Brazilian Shield,
Guayana Shield, or Amazon Basin; and by Tillandsia sp. A. ca. 2.4
Mya in Central America (Fig. A4A). There were no reversions to mesic habitats or microsites. Finally, epiphytism or arid/semi-arid
62
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
Fig. 3. Evolution of (A) C3 vs. CAM photosynthesis, and (B) bee, bat, and bird pollination. See Fig. 2 legend for additional details.
habitats or microsites evolved at least four times, at the bases of
Tillandsioideae, Hechtia, the Xeric Clade, and Puya plus Bromelioideae (Fig. A4B). Reversions to terrestrial habit and mesic or wet
sites appear to have occurred at least seven times.
3.3. Tests of correlated evolution
All nine associations between pairs of the focal plant traits, life
in fertile, moist habitats P1000 m, and life in arid and semi-arid
63
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
habitats predicted from our schema of direct evolutionary and ecological forces are statistically significant (Table 1). That is, epiphytism shows significant patterns of correlated evolution with the
tank habit, entangling seeds, and moist, fertile habitats P1000 m
elevation. The association between epiphytism and the tank habit
is especially strong (P < 0.00004). In addition, the tank habit and
ornithophily exhibit strong patterns of correlated evolution with
moist, fertile habitats P1000 m elevation. Both the non-tank habit
and CAM show significant patterns of correlated evolution with life
in arid or semi-arid habitats, as well as with epiphytism or life in
arid and semi-arid habitats. C3 photosynthesis is correlated with
moist habitats P1000 m elevation.
Seventeen of the 25 pairwise comparisons in which direct evolutionary forces were not expected to drive correlated evolution
among pairs of focal characters or environmental characteristics
do, in fact, fail to show significance. However, seven cases (28%) exhibit unexpected patterns of correlated evolution (Table 1). With
the exception of photosynthetic pathway, plant characters generally exhibit increasingly strong patterns of correlated evolution
with elevation P1000 m, moist habitats P1000 m elevation, and
fertile, moist habitats >1000 m elevation (Table 1).
3.4. Tests of contingent evolution
Epiphytism always evolves in a tank background (P < 0.0017,
Fig. 4A), and always in clades with entangling seeds (P < 0.0002,
Fig. 4B). When C3 photosynthesis evolves de novo, it only does so
in terrestrial clades, not epiphytic clades (P < 0.05, Fig. 4C), but this
pattern is non-significant with the Bonferroni correction. Other
strong patterns that are not significant with the Bonferroni correction include: (1) C3 photosynthesis evolves only in terrestrial
clades, and CAM photosynthesis evolves four times more frequently in epiphytic vs. terrestrial clades (Fig. 4C); (2) C3 photosynthesis evolves only in non-tank clades (P < 0.0384, Fig. 4E); (3)
avian pollination arises more frequently in tank than non-tank
clades (P < 0.0161, Fig. 4F), while non-avian pollination arises more
frequently in non-tank clades (P < 0.0179, Fig. 4F); (4) avian pollination is nearly 20 times more likely to evolve in an entanglingseed vs. non-entangling seed background (P < 0.0153, Fig. 4G);
(5) C3 photosynthesis arises only in a non-entangling seed background (P < 0.0195, Fig. 4H); and (6) shifts to moist, fertile habitats
P1000 m elevation are more likely to occur in an entangling-seed
back ground (P < 0.0148, Fig. 4I),
Epiphytism evolves significantly more frequently in moist, fertile habitats >1000 m elevation than elsewhere (P < 0.0002,
Fig. 4D). Epiphytic or arid/semi-arid conditions are significantly
more likely to evolve in an entangling-seed background
(P < 0.001, Fig. 4J). CAM only evolves in bird-pollinated clades
(P < 0.0005, Fig. 4K), and non-avian pollination is more likely to
evolve in CAM clades than C3 clades (P < 0.01, Fig. 4K). CAM evolves
almost only in low-elevation, arid/semi-arid conditions
(P < 0.0119, Fig. 4L), and evolves significantly more frequently in
arid/semi-arid conditions (P < 0.0017, Fig. 8M). CAM evolves only
from epiphytism or arid/semi-arid conditions (P < 0.0014,
Fig. 4N), and the latter evolve significantly more frequently from
CAM (P < 0.0061, Fig. 4N). Life in moist, fertile habitats P 1000 m
evolves only in bird-pollinated clades (P < 0.0013, Fig. 4O), and
avian pollination is 25 times more likely to evolve in moist, fertile
habitats P 1000 m elevation, although the latter pattern is nonsignificant (Fig. 4O). All other rates of contingent evolution in all
other comparisons studied are non-significant.
As expected, tank epiphytes appear to evolve mainly from tankbearing terrestrials (Fig. 4A); epiphytes with entangling seeds
evolve from terrestrial species with the same (Fig. 4B); CAM epiphytes evolve mostly from C3 epiphytes (Fig. 4C); epiphytes evolve
mainly in moist, fertile, montane habitats >1000 m elevation
(Fig. 4D); non-tank C3 species evolve from both C3 tank species
and CAM non-tank species (Fig. 4E); bird-pollinated tank species
evolve mainly from tank species pollinated by insects (Fig. 4F);
bird-pollinated species with entangling seeds evolve from those
with the same kinds of seeds but pollinated by insects, reflecting
the high conservatism of entangling seeds (Fig. 4G). Similarly,
CAM species with entangling seeds evolve only from C3 species
with entangling seeds, and C3 species with non-entangling
species evolve only from CAM species with the same (Fig. 4H);
species with entangling seeds in moist, fertile, habitats P1000 m
elevation evolve from species with the same kinds of seeds
elsewhere (Fig. 4I); and species with entangling seeds and living as
epiphytes or as terrestrials in arid conditions evolve from species
with the same kinds of seeds but different life forms or habitats
(Fig. 4J). C3 species in high, moist habitats evolve about equally from
CAM species in such habitats and from C3 species in arid, low-elevation habitats (Fig. 4L). CAM species in arid conditions evolve from C3
species in similar conditions (Fig. 4M); CAM species living as epiphytes or as terrestrials in arid habitats evolve from C3 species growing under similar conditions (Fig. 4N); and bird-pollinated species in
moist, fertile habitats P1000 m elevation evolve mainly from species in the same conditions but pollinated by other animals
(Fig. 4O). CAM species pollinated by birds evolve from C3 species pollinated by birds (Fig. 4K).
3.4.1. Determinants of net rates of species diversification –
phylogenetically unstructured analyses
Species number per subfamily ranges from 20 in Brocchinioideae and 45 in Lindmanioideae to 856 in Bromelioideae and
1256 in Tillandsioideae (Table 2). Stem rates of net diversification
Table 1
Probability of observed pattern of correlated evolution between pairs of characters. Underlines indicate a significant pattern (P < 0.05) where expected as a result of the direct
evolutionary forces hypothesized to affect bromeliad evolution (Fig. 1). Italics with underlines indicate the statistical significance of an unexpected pattern. Patterns of correlated
evolution among environmental measures were not calculated (see text). Note that, in each case, significant patterns of correlated evolution involve the expected association of
character-states, not their converse (e.g., epiphytism and tanks frequently co-occur, as do the terrestrial habit and non-impounding leaf rosettes).
Habit (E)
Growth form (T)
Entangling seeds (S)
Photosynthetic pathway (CAM)
Pollination syndrome (A)
Elevation > 1000 m (H1)
Moist > 1000 m (H2)
Moist fertile >1000 m (H3)
Arid/semi-arid regions and microsites
Epiphytism or arid/semi-arid conditions
T
S
CAM
A
H1
H2
H3
Arid
0.00001
0.0007
0.1116
0.1085
0.0798
0.0776
0.0183
0.0884
A&E
0.0096
0.2172
0.036
0.2669
0.0091
0.0006
0.0139
0.0635
0.1024
0.7818
0.076
0.0028
0.5075
0.0098
0.0004
0.3844
0.0442
0.1598
0.00003
0.00001
0.7422
0.7785
0.0089
0.1288
0.0029
0.025
64
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
B
C
8
Te, Tank
Te, NES
15
0
0
260
84
0
100
680
Epi, NTa
142
54
Epi, NES
E
∩
∩
∩
Epi, Lo
Ar NF
142
0
0
54
Epi,
HiMF
∩
G
C3, NTa
54
NTa, NAv
12
16
0
233
167
12
CAM, NTa
C3, NES
0
82
C3, ES
0
306
135
21
Ar Epi,
ES
0
53
0
65
0
C3, Ar
79
0
2
212
59
145
HiMF, NES
C3, LoAr
91
71
HiMF, ES
97
0
C3, HiM
5
90
108
CAM, Av
CAM, LoAr
176
CAM, HiM
70
O
C 3,
NAr∩Te
0
0
CAM, Ar
CAM,
NAr∩Te
0
CAM, NAr
111
94
CAM, NAv
7
77
0
235
47
C 3,
Ar Epi
75
0
CAM,
Ar Epi
∩
C3, NAr
0
112
C3, Av
N
16
Lo Ar
NF, ES
0
∩
M
10
L
C3, NAv
0
∩
NAr∩Te,
ES
4
10
5
∩
0
63
Ar Epi,
NES
Tank, Av
0
CAM, ES
K
9
Lo Ar
NF, NES
0
J
14
299
47
CAM, NES
0
NAr∩Te,
NES
Tank, NAv
0
Av, ES
9
I
11
Av, NES
NTa, Av
502
CAM, Tank
39
39
Epi, CAM
0
113
∩
51
170
58
11
NAv, ES
20
15
C3, Tank
H
0
12
49
86
7
NAv, NES
Epi, C3
12
0
13 8
43
F
8
Te, HiMF
Te, CAM
73
27
Epi, ES
0
D
0
6
0
Epi, Tank
202
Te, Lo Ar
NF
Te, C3
∩
0
39
Te, ES
∩
Te, NTa
8
∩
A
NAv,
LoArNF
17
0
0
89
Av,
LoArNF
NAv,
HiMF
416
269
150
Av, HiMF
185
Fig. 4. Contingent patterns of evolution among pairs of binary character-states. Rates corresponding to thick vs. hollow arrows differ significantly (P < 0.05) in magnitude.
Black arrows (red in the online color figure) highlight the net direction of transitions between adjacent pairs of character-states when the difference in rates exceeds 20. Boxes
highlight pairs of character-states likely to account for a preponderance of taxa, given the ancestral condition in Bromeliaceae and pattern of transitions. Contingent patterns
shown are for the evolution of Terrestrial vs. Epiphytic habit and (A) Tank vs. Non-Tank growth form, (B) Entangling Seeds vs. Non-Entangling Seeds, (C) C3 or CAM
photosynthetic pathway, and (D) High, Moist, Fertile conditions vs. Low, Arid, or Non-Fertile conditions; for the evolution of Tank vs. Non-Tank growth form and (E) C3 and
CAM photosynthesis, and (F) Avian and Non-Avian pollination; for the latter and (G) Entangling Seeds vs. Non-Entangling Seeds, (K) C3 and CAM photosynthesis, and (O) High
(P1000 m elevation), Moist, Fertile habitats vs. others; for Entangling Seeds vs. Non-Entangling Seeds and (I) High (P1000 m elevation), Moist, Fertile habitats vs. others, and
(J) Epiphytism or Arid or semi-arid conditions vs. Non-Arid or semi-arid conditions; and for C3 and CAM photosynthesis and (H) Entangling Seeds vs. Non-Entangling Seeds,
(M) Arid or semi-arid conditions vs. Mesic conditions, (K) C3 or CAM photosynthetic pathway vs. Avian or Non-Avian pollination, (L) Low-elevation and Arid/semiarid
habitats vs. High-elevation, Mesic conditions, and (N) Epiphytism or Arid or semi-arid conditions vs. Non-Arid or semi-arid conditions.
65
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
vary from 0.13 sp sp 1 My 1 (=0.13 My 1) in Brocchinioideae to
0.63 My 1 in Bromelioideae. Crown rates range from 0.16 My 1
in Brocchinioideae to 0.64 My 1 in Bromelioideae (Table 2). Rates
of net diversification are especially high in the bromelioid tankepiphyte clade native to the Atlantic Forest region (1.11 and 1.05
My 1 for stem and crown rates, respectively), Brazilian ShieldAtlantic Forest bromelioids (0.70 and 0.78 My 1), and the core tillandsioids (i.e., Tillandsioideae minus Catopsis-Glomeropitcairnia)
(0.47 and 0.67 My 1).
Rates of net species diversification drop sharply with subfamily
stem age (y = 2.17 e 0.122x, r2 = 0.85, P < 0.0006 for 1-tailed t-test, 6
d.f.). This pattern is even stronger if the two most diverse subclades
– the tank-epiphytic bromelioids and the core tillandsioids – are
substituted for their respective subfamilies (y = 2.18 e 0.124x,
r2 = 0.94, P < 0.0001 for 1-tailed t-test, 6 d.f.). The increased explanatory power of the latter regression is driven largely by an outlier;
when the tank-epiphytic bromelioids are excluded, r2 = 0.88. There
is no significant correlation of species number with subfamily age
(r = 0.35, P > 0.38 for 6 d.f.).
Based on reconstruction of ancestral character-states, each subfamily or subfamilial subclade can be categorized by the sole/dominant character-states of habit, growth form, seeds, photosynthetic
path way, pollination syndrome, and occurrence in extensive,
moist, fertile cordilleras (Figs. 2 and 3, A2–A4). Based on this, net
diversification rates based on stem ages are significantly higher
in the two major epiphytic clades, Tillandsioideae and the bromelioid tank-epiphytes (mean Dstem = 0.77 ± 0.49 My 1), than in the
remaining, almost entirely terrestrial subfamilies (mean
Dstem = 0.31 ± 0.14 My 1) (P < 0.028 for 1-tailed t-test with 6 d.f.).
Weakly supported differences apply to tank- vs. non-tank lineages
(mean Dstem = 0.77 vs. 0.34, P < 0.047), after Brocchinioideae is excluded due to its possessing large fractions of both tank and nontank species. Similar conclusions apply to calculations based on
ln D rather than D, but with the difference between tank and
non-tank lineages being marginally non-significant (Table 3).
Diversification rates are higher in the two subfamilies with
entangling seeds than in the six without (Dstem = 0.57 ± 0.15
My 1 vs. 0.33 ± 0.15 My 1), but the difference is not significant
for untransformed or log-transformed data (Table 3). Comparisons
of diversification rates in CAM and C3 lineages are more difficult,
given the haphazard variation in reconstructed ancestral states
within Puyoideae. Setting Puyoideae aside and splitting Pitcairnioideae into its largest, purely CAM and C3 lineages, we found no significant difference in diversification rates between CAM lineages
(Bromelioideae, Dyckia-Encholirium-Deuterocohnia, Hechtia) and
C3 lineages (Pitcairnia, Navioideae, Tillandsioideae, Lindmanioideae, Brocchinioideae) (Dstem = 0.47 ± 0.23 My 1 vs. 0.31 ± 0.14
My 1) for arithmetic or log-transformed data (Table 3).
Net diversification rates based on stem ages are significantly
higher for the two major clades pollinated primarily by hummingbirds, Tillandsioideae and Pitcairnioideae–Puyoideae–Bromelioideae
(mean Dstem = 0.44 ± 0.03 My 1), than for the non-bird-pollinated
subfamilies (mean Dstem = 0.22 ± 0.07 My 1) (P < 0.007 for 1-tailed
t-test with 4 d.f.). Mean diversification rates are thus more than 2.5
times higher in epiphytic clades than terrestrial clades, and 2.0
times higher in hummingbird-pollinated clades than in those not
pollinated by birds and mainly by insects (Table 3).
Four lineages – the higher tillandsioids, pitcairnioids, puyoids,
and tank-epiphytic bromelioids – are composed primarily of species native to the Andes and/or the Serra do Mar region of south
eastern Brazil. Substrates in these two regions are also more fertile
than the marine-derived Roraima sandstone of the tepuis in the
Guayana Shield, and presumably more congenial to the evolution
of tank bromeliads. The mean diversification rate of lineages in
the Andes and Serra do Mar regions is 0.43 ± 0.23 My 1, nearly
twice that for lineages outside these regions (Dstem = 0.22 ± 0.07
My 1); when comparisons are made based on log-transformed
data, this is the most highly significant difference observed in comparisons of subfamilies based on traits (P < 0.0017 for 1-tailed
t-test with 6 d.f.) (Table 3).
3.4.2. Determinants of net rates of species diversification –
phylogenetically structured analyses
MEDUSA identified significant acceleration of net speciation
rates at three points in the simplified 26-taxon tree, including (1)
the crown group of the core tillandsioids, (2) the stem group of
the Navioideae–Pitcairnioideae–Puyoideae–Bromelioideae clade,
and (3) a further acceleration in the crown group of the tank-epiphytic bromelioids (Fig. 5). These accelerations were detected
under a mixed birth–death model for (1) and (3), and a Yule
birth-only model for (2). Net diversification rates were highest
for the tank-epiphytic bromelioids (r = 0.815) despite a high rate
of extinction (e = 0.864). The core tillandsioids had a somewhat
slower diversification rate (r = 0.482) but also had a high rate of
extinction (e = 0.864). A GEIGER analysis of crown diversification
rates identified significant accelerated diversification at these same
three nodes. GEIGER also identified a significant acceleration of net
diversification (to r = 0.435) in bromeliads other than Brocchinioideae and Lindmanioideae, and decelerations (to r = 0) in
Table 2
Species richness, stem and crown ages, and net diversification rates (Dstem and Dcrown) for bromeliad subfamilies and key subclades.
Clade
# spp.
Stem age (My)
Crown age (My)
Dstem (My
Brocchinioideae
Lindmanioideae
Tillandsioideae
Core tillandsioids
Hechtioideae
Navioideae
Navia/Brewcaria
Pitcairnioideae
Deuterocohnia–Dyckia-Encholirium
Pitcairnia
Puyoideae
Bromelioideae
Brazilian Shield
Tank epiphytes
Pu + Brom
Pit + Pu + Brom
Nav + Pit + Pu + Brom
Entire family
20
45
1256
1236
52
107
99
587
170
387
217
856
753
629
1073
1660
1764
3140
22.7
18.1
16.9
15.2
16.3
15.9
7.9
14.1
11.3
12.5
10.7
10.7
9.4
5.8
14.1
15.9
16.3
97.5
14.1
5.8
15.2
9.6
9.9
10.5
6.7
11.9
0.13
0.21
0.42
0.47
0.24
0.29
0.58
0.45
0.45
0.48
0.50
0.63
0.70
1.11
0.49
0.47
0.46
0.08
9.4
9.4
7.5
5.5
22.7
1
)
Dcrown (My
0.16
0.54
0.42
0.67
0.33
0.38
0.58
0.48
0.50
0.64
0.79
1.05
0.32
1
)
66
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
Table 3
Significance of differences between net diversification rates (Dstem) associated with states of individual characters across bromeliad subfamilies and key subclades (see text).
Comparison
Epiphytes vs. terrestrials
Tank vs. non-tank taxa
Seeds entangling vs. not
CAM vs. C3
Avian vs. non-avian pollination
Moist, fertile cordilleras vs. not
a
Significancea
Result
Higher
Higher
Higher
Higher
Higher
Higher
in
in
in
in
in
in
1
1
Epi (0.77 ± 0.49 My ) than Te (0.31 ± 0.14 My )
Ta (0.77 ± 0.49 My 1) than NTa (0.34 ± 0.13 My 1)
ES (0.53 ± 0.15 My 1) than NES (0.31 ± 0.14 My 1)
CAM (0.47 ± 0.23 My 1) than C3 (0.31 ± 0.14 My 1)
Av (0.44 ± 0.03 My 1) than NAv (0.224 ± 0.07 My 1)
HiF (0.43 ± 0.31 My 1) than not (0.22 ± 0.07)
<0.028
<0.047
>0.055
>0.13
<0.007
<0.029
<0.042
>0.052
>0.077
>0.15
<0.023
<0.010
First column gives P for untransformed data; the second, for log-transformed data.
Table 4
Significant differences in rates of speciation and extinction associated with different character-states under BiSSE.
Comparison
Result
Significance
Epiphytes vs. terrestrials
Tank vs. non-tank taxa
Avian vs. non-avian pollination
Moist, fertile cordilleras vs. not
Speciation higher in Epi (k1 = 1.176) than Te (k0 = 0.221)
Speciation higher in Ta (k1 = 0.870) than NTa (k0 = 0.293)
Speciation higher in Av (k1 = 0.870) than NAv (k0 = 0.293)
Speciation higher in HiMF (k1 = 1.146) than not (k0 = 0.207)
Extinction higher in HiMF (k1 = 0.962) than not (k0 = 0.091)
NS
NS
<0.015
<0.011
<0.02
<0.0001
<0.006
>0.55
>0.25
CAM vs. C3
Seeds entangling vs. not
Acanthostachys (sister to the tank-epiphyte bromelioids), in Deinacanthon, and in Cottendorfia.
BiSSE analyses on the 26-terminal tree showed that tank formation, avian pollination, epiphytism, and especially life on moist, fertile cordilleras significantly accelerated speciation to progressively
larger degrees, and that life on moist, fertile cordilleras also significantly accelerated the apparent rate of extinction (Table 4). CAM
photosynthesis and entangling seeds had no significant effect on
speciation or extinction rates (Table 4). Estimates of the incidence
of various traits in all species in each clade (Table 2) were close to
those implied by the traits of the 90 place-holder taxa, except that
our 90-taxon sampling greatly underrepresented atmospheric nontank Tillandsia, which appear to comprise 40% of Tillandsieae.
and (3) Puya (ca. 200 spp.) at high elevations in the Andes (Table 5).
One additional radiation, not considered here in any detail, includes mostly broad-leaved Pitcairnia (ca. 390 spp.), a genus of
exclusively C3 species in rain- and cloud-forest understories. The
three major clades – Brocchinioideae, Lindmanioideae, Navioideae
– that did not (except for Cottendorfia florida) invade areas outside
the ancestral region and habitats of the Guayana Shield contributed fewer than 200 species to the total of ca. 3140 species of present-day bromeliads.
3.5. Effects of broad-scale adaptive radiation on diversity
Multiple, often closely associated origins of epiphytism, the
tank habit, entangling seeds, CAM photosynthesis, and avian
(mainly hummingbird) pollination in bromeliads occurred mostly
during the mid to late Miocene (Figs. 2 and 3, A2–A4), and largely
agree with our predictions (Fig. 1), with the timing of the orogenies
of the Andes and of the Serra do Mar and nearby ranges, and with
past global declines in moisture availability and atmospheric CO2
concentrations (see below). Transitions to habitats >1000 m arose
most frequently, at least 13 times, often in Atlantic Forest tank-epiphytic bromelioids (Figs. A3 and A4); transitions to the tank habit
and epiphytic growth form arose least frequently, 2–4 times
(Fig. 2A and B).
All nine predicted patterns of correlated evolution among these
traits, life in moist, fertile habitats >1000 m elevation, and life in
epiphytic or dry terrestrial microsites are phylogenetically significant (Table 1). Seventeen of the remaining 24 pairwise combinations of traits with each other or ecological conditions showed no
significant pattern; seven did, however. Unexpected patterns include the rise of tanks with entangling seeds, avian pollination,
and moist habitats P1000 m elevation; of entangling seeds with
fertile, moist habitats P1000 m elevation and with epiphytism or
arid/semi-arid conditions; of CAM with avian pollination; and of
avian pollination with epiphytism or arid/semi-arid conditions
(Table 1). With the exception of the tie between tanks and avian
pollination, these patterns are probably best viewed as indirect
results of the direct forces in our model, which tie together most
of the key functional traits and environmental conditions studied
in this paper. Below we discuss the factors favoring each predicted
Evolution of epiphytism and associated traits is associated with
the generation of some 1200 species of Tillandsioideae, mainly in
the Andes and Central America, and some 600 species of the
tank-epiphyte clade of Bromelioideae, mainly in the Atlantic Forest
region (Table 5; graphic abstract). Subsequently, a limited number
of species of both groups invaded almost all areas occupied by the
family. Three major radiations of terrestrial bromeliads in arid and
semi-arid regions appear to have been facilitated by the evolution
of CAM photosynthesis, including (1) Hechtia (ca. 50 spp.) at low to
mid elevations in Central America/Mexico; (2) Deuterocohnia–Dyckia-Encholirium (ca. 170 spp.) at mid to high elevations in the Andes, and then across the Brazilian Shield to the Horn of Brazil;
Table 5
Summary of the six broad-scale adaptive radiations adding substantial diversity to
Bromeliacae.
Clades
Number of % Species in
family
species
1. C3 + CAM epiphytes (Tillandsioideae)
1256
2. CAM epiphytes (tank-epiphytic Bromelioideae) 629
3. CAM terrestrials (Hechtia, Central America)
52
4. CAM terrestrials (Xeric clade, South America)
170
5. CAM terrestrials (Puya, South America)
217
6. C3 understory herbs (Pitcairnia, South America) 387
Total radiations
2711
Total Bromeliaceae
3140
40.0
20.0
1.7
5.4
6.9
12.3
86.3
100.0
4. Discussion
4.1. Origin and correlated evolution of key functional traits
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
pattern of correlated evolution, at the times and places where the
associated character-states originated.
4.1.1. Fertile, humid montane habitats favor epiphytes
Mid-elevation tropical habitats experience high rainfall and low
evaporation, as well as a rich nutrient rain of detritus derived from
leachate and shed parts from host trees, and possibly from animals
or drop pings derived from nutrient-rich soils. The great abundance and diversity of epiphytes in mid-elevation cloud forests –
and in areas of greater precipitation/fog at lower elevations (Kreft
et al., 2004; Haeger and Dohrenbusch, 2011; Obregon et al., 2011) –
are consistent with higher moisture availability resulting in greater
epiphyte density and diversity. The mid-elevation maximum in
epiphyte density often occurs at somewhat higher elevations than
that for epiphyte species richness (Gentry and Dodson, 1987), but
both (and especially that for species richness) appear to be better
explained by gradients in moisture supply than by mid-domain
sampling artifacts (Currie and Kerr, 2008; cf. Cardelus et al.,
2006; Colwell and Rangel, 2010). High elevations above the inversion layer support very few epiphytic species or individuals (Gentry and Dodson, 1987). Experiments on 11 bromeliad species
showed that water availability has a much stronger and consistent
positive effect on epiphyte growth than nutrient or CO2 supplies
(Zotz et al., 2010).
Soil fertility appears to enhance epiphyte density and diversity.
Our findings support this oft-debated proposition for bromeliads,
given that the epiphytic habit only shows correlated evolution
with fertile, moist habitats P1000 m elevation, and not with habitats P1000 m elevation, alone or in combination with substantial
moisture supply (Table 1). On the rainy tepuis of the Guayana
Shield, composed of highly leached sandstones and quartzites of
marine origin, at most two epiphytic species have evolved in Brocchinioideae, none in Lindmanioideae, and perhaps one in Navioideae (Givnish et al., 1997, 2007; Holst, 1997). Outside South
America, lowland kerangas vegetation on extremely infertile white
sands on Borneo lack epiphytes except those fed by ants (Janzen,
1974a), and dipterocarp forests on infertile soils in southeast Asia
often have very low epiphyte abundance except near roads where
mineral input via dust is likely to be greater (Janzen, 1974b, 1977).
Ground-level P fertilization of wet forests on 4.1 My-old soils on
Kauai increased the abundance and diversity of epiphytic cyanolichens, chlorolichens, mosses, and ferns, where as fertilization with
N or with S, cations, and micronutrients had no significant effect
(Benner and Vitousek, 2007). Natural occurrences of such epiphytes on several Hawaiian islands are significantly greater on
trees with higher P content (Benner, 2011), suggesting that P limits
epiphyte growth there. Depletion of 15N and higher N:P ratios in
response to N + P fertilization of the tank epiphyte Werauhia (formerly Vriesea) sanguinolenta in Panama indicate that its growth is
often P-limited, with a breakpoint at a foliar N:P ratio of 10.4 (Wanek and Zotz, 2011). A strong correlation of foliar P content – but
not N content – between epiphytes and their hosts along an extensive elevational gradient in Costa Rica also suggests a close link of
epiphytes to hosts in terms of P (but not N) nutrition (Cardelus and
Mack, 2010). Widespread natural depletion of 15N of foliar tissue in
epiphytes vs. hosts, as well as epiphytic foliar N:P ratios often
greater than 12, strongly suggest that P limitation or co-limitation
is widespread in bromeliads and other epiphytes (Cardelus and
Mack, 2010; Wanek and Zotz, 2011). Tree litter in tropical rain forests exhibit the highest N:P ratios on Earth (McGroddy et al., 2004),
and canopy soils – derived from litterfall, with both feeding rooted
as well as tank epiphytes – have much higher N:P ratios than ordinary soils, in rain forests and especially cloud forests (Nadkarni
and Solano, 2002; Cardelus et al., 2009; Wanek and Zotz, 2011).
Levels of available soil P should be highest over young rocks and
in areas of rapid uplift and erosion (Chadwick et al., 1999; Vitousek
67
et al., 2003; Porder et al., 2007). Such conditions prevail in the Andes, the mountains of Central America, and the Serra do Mar and
nearby ranges in the Brazilian Shield. Factorial fertilization of three
bromeliad species with three different levels of N and P pointed to
P as the nutrient limiting growth (Zotz and Asshoff, 2010; see also
Zotz and Richter, 2006).
The two main origins of epiphytism in bromeliads occurred in
tillandsioids with the uplift of the northern Andes ca. 15 Mya
(Hoorn et al., 1995, 2010; Potter, 1997; Givnish et al., 2011) and
in the tank-epiphytic bromelioids with the accelerated uplift of
the Serra do Mar and Serra da Mantiqueira in the Atlantic Forest region during the Pliocene–Pleistocene (Almeida, 1976; Amorim and
Pires, 1996) (Fig. 2A). Origins of these radiations correspond reasonably well with the independently derived dates of origin of diving-beetle lineages endemic to bromeliad tanks ca. 12 Mya in
northern South America and ca. 1.6–4.3 Mya in the Atlantic Forest
region (Balke et al., 2008).
Uplift of the Serra do Mar coincided with the rise of the central
Andean Altiplano toward the end of the Miocene (Garzione et al.,
2008), which should have triggered a cooler, rainier climate in
the Atlantic forest region, with enhanced advection of moisture
off the tropical south Atlantic as winds off the Pacific were blocked
by the Altiplano (Ehlers and Poulsen, 2009). This effect, combined
with the actual uplift of the Serra do Mar would have favored the
onset of cooler, rainier, more humid conditions starting ca. 5.6
Mya (Vasconcelos et al., 1992; Grazziotin et al., 2006), the time
of origin of the bromelioid tank-epiphyte clade (Fig. 2A). Today,
the Atlantic Forest region – including remnant, highly diverse rain
and cloud forests, sandy coastal restingas, mangroves, campos de
altitude, and granitic outcrops of the Serra do Mar and Serra da
Mantiqueira and adjacent coastal plains – is the wettest part of
eastern South America, and the highlands, the coolest (Safford,
1999). Local bromeliad floras can be extremely rich, with up to
92 species in one 1° 1° cell in Minas Gerais (Versieux and Wendt,
2007), and vary greatly geographically (Fontoura et al., 2012). The
Serra do Mar and Serra da Mantiqueira form the elevated south
east rim of the Brazilian Shield. Both experienced strong climatic
fluctuations during the Pleistocene (Auler and Smart, 2001; Behling and Negrelle, 2001), as did the northern Andes (Gentry,
1982; van der Hammen, 1995).
4.1.2. Epiphytism and fertile, moist montane habitats favor the tank
habit
Two origins of the tank habit appear to coincide with the rise of
epiphytism in tillandsioids in the northern Andes as they began
their uplift ca. 15 Mya, and the rise of epiphytism in bromelioids
as the Brazilian Highlands began their uplift ca. 5.6 Mya. The third
origin of tanks, in Brocchinia, occurred ca. 9 Mya in mid- to highelevation in the long-uplifted Guayana Shield (Fig. 2B), apparently
as an adaptation to capture nutrients from falling debris, commensalants, or insect prey under exceedingly nutrient-poor and
humid conditions atop tepuis (Givnish et al., 1997, 2007). Though
not captured in our taxon sampling, at least one origin of epiphytism was associated with the acquisition of the tank habit in Brocchinia, in B. tatei and the unstudied B. hitchcockii (Givnish et al.,
1997). The tank habit was lost at least five times. In three cases,
this occurred in species with CAM photosynthesis, which have an
additional means of conserving water. In two cases – Spanish moss,
Tillandsia usneoides, and T. sp. A – there was a shift to the atmospheric habit.
Atmospheric species of Tillandsia are non-tank epiphytes that
depend almost exclusively on absorptive leaf trichomes for water
and nutrient uptake (Benzing, 1980, 2000). Abundant trichomes
also make their dry leaves highly reflective, possibly reducing
water loss and photoinhibition (cf. Pierce et al., 2001). Dense arrays
of tillandsioid trichomes are highly efficient at capturing moisture
68
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
from fog when combined with narrow leaves that efficiently intercept fine droplets (Martorell and Ezcurra, 2007). Atmospherics tolerate desiccation better than tank species but have lower rates of
photosynthesis per unit leaf mass (Benzing and Burt, 1970; Benzing and Renfrow, 1974; Reyes-Garcia et al., 2008), and almost
all have CAM photosynthesis (Crayn et al., 2004). Many tank bromeliads exhibit developmental heterophylly, with juveniles starting as atmospherics (Adams and Martin, 1986; Benzing, 2000;
Zotz et al., 2011) and then later forming tanks as body size increases. Krömer et al. (2006) found that large, tank-epiphyte bromeliads dominated the rainy, moist eastern slope of the Andes in
Bolivia, while atmospheric Tillandsia taxa dominated the drier,
more seasonal leeward slope. Atmospheric species of Tillandsia account for ca. 40% of the Tillandsieae (Tillandsia–Racinaea–Guzmania–Mezobromelia), and that that clade is nearly 2.4 times as
species-rich as its sister, the Vrieseae, which is also nearly 100%
epiphytic but lacks atmospherics.
4.1.3. Epiphytism favors the evolution of entangling seeds, and vice
versa
Such seeds characterize Tillandsioideae and Bromelioideae,
which together contain almost all epiphytic bromeliads. Entangling seeds, epiphytism, tanks, and invasion of fertile, moist habitats P1000 m elevation evolved simultaneously in the northern
Andes, Central America, or the northern littoral of South America
in tillandsioids. In bromelioids, fleshy fruits evolved 1 My before
invasion of the Brazilian Shield, and 4.4 My before epiphytism
and the tank habit arose. Fleshy fruits frequently arise in forest
understories, where wind dispersal is less effective, and the incidence of such fruits increases toward rainier habitats (Croat,
1978; Gentry, 1982; Givnish, 1998; Smith, 2001; Givnish et al.,
2005; Sytsma et al., 2002). Life in forest and scrub understories is
widespread in the basal bromelioid grade, including Greigia, Ochagavia, Bromelia, Pseudananas, Ananas, Cryptanthus, and Acanthostachys. Evolution of fleshy fruits under such conditions may be
seen as a preadaptation for the later evolution of epiphytism in
the Serra do Mar. An interesting case of apparent convergence involves the coma-like appendage in the epiphyte Brocchinia tatei, in
which one end of the seed’s bicaudate wing splits into several
small processes; this is the only such dispersal apparatus in Brocchinioideae (Givnish et al., 1997).
4.1.4. Dry conditions favor CAM
CAM dominates three terrestrial lineages from arid sites (Dyckia–Deuterocohnia-Encholirium, Puya, Hechtia) as well as some epiphytic lineages (especially bromelioids from humid Atlantic
forests), with a high incidence of CAM in atmospheric Tillandsia
species in dry, seasonal areas (Krömer et al., 2006). As predicted,
CAM photosynthesis shows a significant pattern of correlated evolution with epiphytism or terrestrial growth in arid or semi-arid
microsites (Fig. 3A). Correlations with dry sites alone are also significant, but not those with epiphytism (Table 1). CAM is more
common in epiphytes from several families in warmer, drier, lower-elevation tropical forests than elsewhere (Earnshaw et al., 1987;
Zotz and Ziegler, 1997; Silvera et al., 2009), and in bromeliads of
the upper canopy vs. the lower canopy or understory (Griffiths
and Smith, 1983; Zotz and Ziegler, 1997; Zotz, 2004).
The three origins of CAM on terrestrial sites all occurred 16.2–
8.1 Mya, at a time of increasing aridification, warm but dropping
temperatures, and declining atmospheric concentrations of CO2
(Zachos et al., 2001; Crayn et al., 2004; Sage, 2004; Kürschner
et al., 2008; Tripati et al., 2009). Selection should have favored
the water-conserving and CO2-concentrating mechanism of CAM
photosynthesis at that time, in much the same way as it apparently
favored multiple origins of C4 photosynthesis in terrestrial plants
during the same period (Ehleringer et al., 1991; Sage, 2004; Ed-
wards et al., 2010; Christin et al., 2011; Sage et al., 2011). Arakaki
et al. (2011) have made the same argument for other groups, and
observed that several succulent lineages with CAM photosynthesis
(higher cacti, Agavaceae, Aizoaceae) began to diversify extensively
about 15–10 Mya. Most likely, the drop in global CO2 levels from
500 to 300 ppm in mid-Miocene reflected increased carbon burial
in the titanic sediment train from the newly uplifting Himalayas
(France-Lanord and Derry, 1997), increased productivity of the Pacific Ocean (Vincent and Berger, 1985), or the global formation of
large brown-coal deposits (Holdgate and Clarke, 2000).
Near closure of the Isthmus of Panama by the Middle Miocene
(Farris et al., 2011; Montes et al., 2012) would have made dispersal
by CAM-bearing Hechtia to arid and semi-arid sites in Central
America easier than envisioned by Givnish et al. (2011), who assumed a much later closure (see Kirby et al., 2008). Uplift of the
northern Andes beginning in the mid-Miocene (Hoorn, 1994;
Hoorn et al., 1995, 2010) helped to create dry habitats in rain shadows in northwestern and southeastern South America, and at high
elevations above the inversion layer along the length of the Andes.
The latter were invaded independently by Puya and early-divergent members of the Xeric clade (Deuterocohnia, several Dyckia)
(Fig 3A; see also Givnish et al., 2007). The basal bromelioid grade –
consisting of seven small, terrestrial genera – colonized dry, rocky
microhabitats or open, relatively dry forests in many cases. Of the
taxa involved, all but the three small Chilean genera exhibit CAM.
The occurrence of C3 photosynthesis in these is likely related to
their occurrence in deciduous forest understories (Greigia, Ochagavia carnea) and/or cool microsites immediately next to the coast
(Fascicularia on the mainland, O. elegans on Robinson Crusoe Island
in the Juan Fernández archipelago) (see habitat descriptions by
Zizka et al. (2009)). Dyckia-Encholirium secondarily invaded dry,
rocky sites through the interior of the Brazilian Shield to the Horn
of Brazil (Givnish et al., 2007, 2011). Hechtia invaded arid and
semi-arid areas near 30°N in Central America. These and similar
areas in the Caribbean, as well as dry areas in the lee of the Andes,
were colonized by the fourth bromeliad CAM clade, composed of
epiphytes in subfamily Tillandsioideae (e.g., Tillandsia utriculata,
T. sp. A).
Our data support five origins of CAM in Bromeliaceae, compared
with the three recognized by Crayn et al. (2004). Our more extensive data allowed us to recognize additional origins of CAM in the
Xeric clade of Pitcairnioideae and within Tillandsia, as captured in
T. usneoides, a placeholder for the large number of CAM atmospherics in that genus. C3 photosynthesis characterizes the basal grade of
pitcairnioids, consisting of broad-leaved Pitcairnia (native to rainand cloud-forest understories, especially along the Andes) and
soft-leaved Fosterella (native to open, mesic habitats at mid-elevations in the Andes). The mesic terrestrial sites inhabited by all navioids and basal pitcairnioids clearly implies that CAM arose in the
Xeric clade after a shift from mesic conditions (Figs. 2A and 3A).
Crayn et al. (2004) surveyed d13C values for 1837 species stratified across all but three bromeliad genera; of these, 44% possessed
CAM, almost all of which can be assigned to the five lineages
shown in Fig. 3A. Only three genera – Acanthostachys, Puya, Tillandsia – have photosynthetic pathways that differ among species; the
four species of Vriesea known to possess CAM will likely be re-assigned to Tillandsia (Grant, 1993). More research on Puya and Tillandsia is needed to exclude the possibility that variation in d13C
within these genera reflects developmental plasticity related to
moisture supply. An MP overlay (not shown) of CAM vs. C3 photosynthesis on the Bayesian phylogeny of Tillandsia (Barfuss et al.,
2005) suggests that CAM may have evolved three times in that
genus (in Barfuss’ clades N and O, and the common ancestor of
all Tillandsia excluding clades H, I, and J). The latter origins are recent and may reflect adaptation to drying cycles during the Pliocene–Pleistocene.
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
Paradoxically, CAM occurs in several epiphytic bromelioids and
tillandsioids in wet, often shaded microsites, where C3 photosynthesis should be favored. Under frequently rainy conditions, CAM
can permit the recycling of respiratory CO2 through CAM idling
when leaf gas exchange is blocked by dense arrays of wet leaf trichomes (Pierce et al., 2002; Freschi et al., 2010). The high diversity
of CAM epiphytic bromelioids in the Atlantic Forest region may primarily reflect this advantage of CAM. Other factors that may help
account for the presence of CAM epiphytes in shaded, rainy microsites – despite the relatively small additional energetic cost of CAM
vs. C3 photosynthesis (Winter and Smith, 1996a,b) – include (1)
efficient use of sunflecks for carbon gain during Phase III when
intercellular CO2 is high (Skillman and Winter, 1997), and (2)
greater allocation to light-absorbing foliage and less to roots, at
least in the CAM terrestrial Aechmea magdalenae vs. sympatric C3
species (Skillman et al., 1999).
4.1.5. Epiphytism favors absorptive leaf trichomes and vice versa
We lack enough data on absorptive trichomes across Bromeliaceae to overlay them on our phylogeny and rigorously test predictions regarding their evolution. However, such trichomes are
known throughout tillandsioids, in many species of the bromelioid
tank-epiphyte clade, and in tank-forming Brocchinia (Benzing et al.,
1976, 1985; Givnish et al., 1984, 1997; Benzing, 2000; Pierce et al.,
2001). Absorptive trichomes appear to exhibit the same pattern of
evolution as tanks, and would – based only on current data – seem
likely to exhibit significant patterns of correlated evolution with
epiphytism and fertile, moist habitats P1000 m elevation. Within
Brocchinia, absorptive trichomes evolved to acquire nutrients in
the extremely rainy and infertile habitats on tepuis and adjacent
sandplains (Givnish et al., 1997). Brocchinioid trichomes retain live
cap cells (Givnish et al., 1984; Benzing et al., 1985; Owen et al.,
1988), and thus do not provide the one-way valves for water and
nutrient uptake – and thus for life in somewhat drier circumstances – that tillandsioid trichomes provide. This may have
greatly limited the brocchinioid radiation.
4.1.6. Fertile, humid montane habitats favor avian pollination
Avian (mainly hummingbird) pollination showed significant
correlated evolution with fertile, moist habitats P1000 m elevation. Ornithophilous flowers, adapted primarily to hummingbirds,
include an estimated 60% of all bromeliad species in the Bolivian
Andes (Kessler and Krömer, 2000) and 85% of those in Atlantic Forest fragments (Piacentini and Varassin, 2007). Such flowers have
arisen at least twice, in the tillandsioid ancestor ca. 15.4 Mya, associated with the Andes/central Chile, Central America, or the Caribbean and northern littoral of South America, and in the common
ancestor of pitcairnioids, puyoids, and bromelioids ca. 14.4 Mya,
associated with the Andes/central Chile (Fig. 3B). These origins occurred at about the same time as the uplift of the northern Andes
and the initial origin of two main Andean lineages of hummingbirds (Bleiweiss, 1998), although McGuire et al. (2007) recently argued suggested that these clades represent one invasion of
montane sites. No comparably large lineages of hummingbirds
evolved in the Serra do Mar or the Guayana Shield (Bleiweiss,
1998). Hummingbirds in these areas appear to represent eclectic
and not especially diverse mixes of taxa with various geographic
affinities (R. Bleiweiss, pers. comm.).
Kessler and Krömer (2000) and Krömer et al. (2006) found that
birds pollinate an increasing fraction of bromeliads at higher elevations and heavier rainfalls in the Bolivian Andes, as expected. Bat
pollination is most common in the humid lowlands, and insect pollination, in drier low-elevation sites. The dearth of insect pollination in cooler/wetter sites accords with the expectation that
thermoregulating pollinators should dominate such sites (Cruden,
1972; Mabberley, 1975; Feinsinger, 1983; Bawa, 1990; Givnish
69
et al., 2009). The shift from bat to hummingbird pollination toward
higher elevations is harder to explain. The high absolute caloric
requirement of nectarivorous bats – based on their larger body
mass (10.2 ± 2.5 g [Patterson et al., 1996]) than Andean hummingbirds (ca. 6 g excluding Patagonia gigas [Altschuler et al., 2004]) –
may make bat pollination, with its large amounts of associated
nectar (nearly ten times that of hummingbird-pollinated flowers
in two Atlantic forest remnants [Sazima et al., 1996, 1999]) too
costly for plants in nutrient-poor, relatively unproductive cloud
forests. High elevations may also not be favorable for bats given
their large heat and water losses from naked flight membranes at
low air temperatures relative to well-feathered hummingbirds.
Shifts from bird to bat pollination in bromeliads (Fig. 3B) accord
with studies on other families (Kay, 2003; Perrett et al., 2007; Knox
et al., 2008; Martén-Rodríguez, 2008), and may also reflect the
greater efficiency of pollen transfer by bats (Muchhala and Thomson, 2010) as well as the context-specific advantages just
discussed.
4.1.7. The tank habit favors ornithophily
This relationship was not predicted by our original model, but
receives support from observed patterns of correlated evolution
(Table 1), and marginal support from observed patterns of contingent evolution (Fig. 4E). In retrospect, it also appears to be explicable in terms of bromeliad biology. Water-filled tanks are likely to
lure and/or support large numbers of small insects, which may
be attractive prey to hummingbirds. In addition, once an inflorescence adapted to avian pollination opens, a water-filled tank can
help provide the fluid required to secrete the large amounts of nectar required to attract hummingbirds.
4.2. Contingent evolution of key functional traits
Nine of the 16 significant patterns of contingent evolution
involving pairs of traits, or of traits and environmental conditions,
mirror expectations from our model or are consistent with them.
One additional pattern – life in moist, fertile habitats P1000 m elevation evolving more frequently in entangling-seed backgrounds –
might have arisen indirectly from the close association of epiphytism with such habitats and such seeds. Another – non-avian (i.e.,
insect) pollination evolving more from a CAM background – might
have arisen in similar fashion from the close association of insect
pollination and CAM with arid/semi-arid conditions (see above).
The same may be true for avian pollination evolving 20 times more
frequently in an entangling-seed background, given the correlation
of both with tanks (Table 1). Yet another significant pattern – avian
pollination evolving more frequently in tank backgrounds – may
reflect the potential availability of insect prey for hummingbirds
at or near tanks, and the ability to secrete large nectar volumes
with access to tank fluid (see above). The remaining four significant
patterns remain puzzling, and join the 120 cases that did not show
significant patterns of contingent evolution and were not expected
to do so. The four anomalous cases thus represent, in some sense,
an error rate of 4/124 = 3.2%, below the 5% level of error expected
from chance alone.
4.3. Determinants of net rates of species diversification
The sharp drop in the rate of net species diversification with
subfamily stem age could reflect a slowing of speciation as adaptive radiations fill ecological space (Givnish et al., 2005, 2009; Rabosky and Lovette, 2008), or a statistical artifact of regressing D = (ln
N)/t (or its more complicated analogue for birth–death processes)
against t (Givnish, 2010). The decline in net diversification with
stem age across angiosperm families (Magallón and Castillo,
2009) and adaptive radiations generally might reflect either
70
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
process, or both. Across bromeliad subfamilies, D declines at t < 1,
suggesting that at least part of the pattern reflects a slowing of speciation in older bromeliad lineages, possibly via ecological
saturation.
Based on phylogenetically unstructured comparisons among
subfamilies, significantly higher rates of net diversification occur
in lineages that occupy moist, fertile, extensive cordilleras, are pollinated by hummingbirds, are epiphytic, possess tanks, or bear
entangling seeds (Table 3). Claims by Quezada and Gianoli (2011)
that CAM is associated with higher diversification rates in bromeliads are based on their incorrect identification of four of five sistergroup comparisons they employed.
Although our comparisons are based on non-overlapping clades
defined by particular character-states, the comparisons themselves
are not independent of each other. Six character-states are broadly
associated with epiphytic lineages – tanks, entangling seeds, avian
(usually hummingbird) pollination, CAM photosynthesis, absorptive trichomes, and occupation of extensive, moist, fertile montane
regions – and a contrasting suite of states is associated with terrestrial lineages (Figs. 2 and 3, A2–A4). We lack enough degrees of
freedom to dissect the relative contributions of each trait to net
diversification. But, based on the magnitude of effects of individual
character-states, epiphytism appears to accelerate diversification
to the greatest extent, by a factor of 2.4, while life in fertile, extensive cordilleras and hummingbird pollination have the greatest effects in terms of statistical significance.
Phylogenetically structured analyses similarly identify tank formation, avian pollination, epiphytism, and especially life on moist,
fertile cordilleras as accelerating speciation with progressively
greater support, and that the latter also significantly accelerated
the apparent rate of extinction (Table 4). MEDUSA identified significant accelerations of net diversification in the core tillandsioids
and the common ancestor of Navioideae–Pitcairnioideae–Puyoideae–Bromelioideae, and a further, nested acceleration in the bromelioid tank-epiphytic clade (Fig. 5). The inclusion of Navioideae in
an accelerated clade is not predicted by our model, but may reflect
high rates of local speciation in Navia-Brewcaria associated with a
complete loss of long-distance seed dispersal (naked, unappendaged seeds), combined with it being sister to a large clade (Pitcairnioideae–Puyoideae–Bromelioideae) strongly associated with
moist, fertile, extensive cordilleras and hummingbird pollination.
The general concordance of results based on phylogenetically
unstructured and structured analyses of the relationship of net
diversification rate to lineages and individual plant characterstates is itself notable, and most likely reflects the nearly star-like
divergence of most subfamilies from each other during the socalled ‘‘bromeliad revolution’’.
Among vascular epiphytes, 80% of all species are concentrated
in just four families: Orchidaceae, Bromeliaceae, Polypodiaceae,
and Araceae (Gentry and Dodson, 1987; Zotz, 2013). Gentry and
Dodson (1987) and Benzing (2000) proposed that, for these families, the evanescent nature of epiphytic substrates, of unoccupied
twigs and branches, select for short-generation species that are
likely to speciate rapidly. However, Ibisch (1996) showed that epiphytic Tillandsia adpressa takes 6–8 years to flower, about 10 years
after the twigs first formed, making even such epiphytic pioneers
among bromeliads more similar to slow-growing forest herbs than
to rapidly speciating desert annuals (Givnish, 2005, 2010).
We propose that epiphytism is associated with high rates of
diversification in bromeliads mainly as a result of (1) the extensive
area and dynamism of habitat available for colonization and speciation; (2) the recent invasion of a new adaptive zone; (3) coevolution with a rapidly diversifying clade of pollinators, the
hummingbirds; (4) partitioning habitats by elevation and exposure
in montane regions that favor epiphytes; and – most importantly –
(5) acceleration of speciation via invasion of the fertile, geographic
ally extensive, topographically dissected, edaphically and climatically complex Andes and Serra do Mar.
4.3.1. Extensive area and dynamism of the epiphytic biotope
Gentry and Dodson (1987) and Benzing (2000) noted the potential importance for diversification of frequent shifts in the location
of optimal microsites for epiphytes in mountainous terrain, due to
geological and climatic dynamism and their impacts on the concentration of mist in a few favorable microsites (e.g., narrow wet
ridges). Temporal shifts in the location of such microsites could
plausibly lead to frequent founder events and rapid speciation
even in plants with long generation times, as could divergence in
elevation and seasonal time of reproduction. The surface area
available for epiphyte to colonize is also generally much bigger
than the ground below, which might also act to increase epiphyte
diversification.
4.3.2. Recent invasion of a new adaptive zone, key innovations, and
key landscapes
Key innovations that open access to new adaptive zones (Simpson, 1944), such as epiphytism and associated traits, are likely to
be associated with high net rates of diversification only if they have
arisen recently (Linder, 2008) and are likely to arise only in the
‘‘key landscapes’’ that favor their evolution (e.g., canopies of tropical rain and cloud forests: Givnish, 1997). Epiphytism arose in
both tillandsioids and the tank-epiphyte bromelioids at just about
the times that the northern Andes and the Serra do Mar – two key
areas occupied by these clades – began their major uplifts in the
Middle and Late Miocene, respectively. Access to volcanically active Central America after the Isthmus of Panama narrowed by
the Middle Miocene (Farris et al., 2011; Montes et al., 2012) as well
as continued uplift of the Andes 3–5 Mya (Steinmann et al., 1999;
Coltorti and Ollier, 2000; Gregory-Wodzicki, 2000) continued to
provide a diversifying influence for tillandsioids. Uplift of the Serra
do Mar during the Pliocene apparently triggered a radiation of epiphytic bromelioids there. While the newly uplifted northern Andes
and Serra do Mar would have provided unoccupied montane regions likely to trigger adaptive radiations in epiphytic groups, we
must recognize that any such radiation in bromeliads would have
had to compete at least with the epiphytic epidendroid orchids,
which began diversifying 50 Mya (Ramirez et al., 2007).
Two unique features of bromeliads may have expanded their
adaptive zone beyond that of other epiphytes: (1) the tillandsioid
trichome and the atmospheric habit that it makes possible (Benzing, 2000), and (2) augmentation of nutrient inputs to some
tank bromeliads by resident ants (Benzing, 1970, 1990; Huxley,
1980; Givnish et al., 1997) and by predators (e.g., spiders, damselfly larvae, frogs and tadpoles) that consume aerial prey (Romero
et al., 2006), submersed detritivores (Ngai and Srivastava, 2006)
or fruits and vegetable debris (da Silva and Britto-Pereira,
2006), and thereby add or retain nutrients in the tank ecosystem,
with microbes living in tanks capable of rapidly mineralizing many
forms of added nitrogenous compounds (Inselbacher et al., 2007).
Modern data – often using d15N to trace the contribution of animals
to a plant’s nitrogen supply – are beginning to support Picado’s
(1913) proposal that many tank-epiphytic bromeliads depend on
animals, at least indirectly, for much of their nutrient supply.
4.3.3. Coevolution with rapidly diversifying hummingbirds
Acquisition of hummingbird pollination in cool montane habitats would open another adaptive zone for bromeliads: a range
of new pollinators with different bill lengths and shapes to be
partitioned. In bromeliads, bird-pollinated lineages do have
significantly higher rates of net diversification, roughly twice that
of lineages pollinated by other animals, primarily insects. In
five of six cases reviewed by Schmidt-Lebuhn et al. (2007),
71
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
Geiger crown rates
Best estimate of r
(r for ε = 0; ε = 0.9)
r = 0.8150
ε = 0.863768
Tank-epiphytic bromelioids
Core bromelioids
r = 1.0039
(1.0039 – 0.7244)
r = 1.0044
ε = 0.0000006
Acanthostachys
Cryptanthus-Orthophytum
Ananas
Pseudananas
Greigia
Bromelia
Deinacanthon
Ochagavia-Fascicularia
Puya - Andes
Puya - central Chile
Deuterocohnia
r = 0.4209
ε = 0.440097
Dyckia-Encholirium
Navioids through
bromelioids
r = 0.4265
(0.6693 – 0.3971)
Fosterella
r = 0.4634
ε = 0.000004
r = 0.2253
ε = 0.0000007
Pitcairnia
Sequencia
Navia-Brewcaria-Steyerbromelia
Cottendorfia
Hechtia
Tillandsia-RacinaeaGuzmania-Mezobromelia
Core tillandsioids
r = 0.1607
(0.6693 – 0.3971)
r = 0.1089
ε = 0.997015
Alcantarea-Vriesea-Werauhia
Catopsis
Glomeropitcairnia
Lindmania
Brocchinia prismatica
Brocchinia all others
Fig. 5. MEDUSA 26-taxon tree, with grey arrows (green in the online color figure) highlighting nodes at which there was a significant acceleration of speciation in the Yule
model, and black arrows (red in the online color figure) highlighting nodes at which there was a significant acceleration of speciation in the birth–death model.
hummingbird-pollinated lineages had a higher net diversification
rate than sister groups pollinated by other animals, including hummingbird-pollinated Tillandsia subg. Tillandsia (227 spp.) having
three times as many species as its sister clade, insect-pollinated Tillandsia subg. Phytarrhiza and Racinaea (77 spp.). To these, we add
our finding that mostly hummingbird-pollinated Pitcairnioideae–
Puyoideae–Bromelioideae (1306 spp.) has 13 times as many species as its sister, insect-pollinated Navioideae (96 spp). The reason
for higher speciation rates in hummingbird-pollinated lineages is
not clear (Castellanos et al., 2003; Schmidt-Lebuhn et al., 2007),
but association with continued uplift of the northern Andes and
climatic oscillations during the Pliocene–Pleistocene and coevolution there with >300 recently derived hummingbird species may
have played a role (Gentry, 1982; Graham, 1997; Berry et al.,
2004; Kay et al., 2005). We propose that hummingbird pollination
may also stimulate speciation by favoring the rise of gullet-shaped
flowers from ancestral cup-shaped blossoms. Once such exclusionary
flowers are present, their length and shape can quickly evolve to
attract different hummingbird species, leading to premating barriers for the plants. Coevolution with rapidly radiating hummingbirds cannot account, however, for rapid diversification of tankepiphyte bromelioids centered in the Serra do Mar. Indeed, the reliance of large numbers of Atlantic bromelioids on just a few hummingbird species, or strongly overlapping sets of species, begs
the question as to how such bromelioids are reproductively isolated (Piacentini and Varassin, 2007; Wendt et al., 2008).
4.3.4. Epiphyte partitioning of habitats by elevation and exposure in
montane regions
Gentry and Dodson (1987) argued that one factor promoting
high diversity in epiphytes should be partitioning of habitats by
elevation and exposure/rainfall. The data of Krömer et al. (2006)
show substantial turnover in the composition of epiphytic bromeliads along one wet elevational transect in the Bolivian Andes, as do
72
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
the data of Kessler and Krömer (2000) for Bolivia generally. Kessler
(2001) and Kreft et al. (2004) document large shifts in the total
number of epiphyte species with rainfall in Bolivia and Ecuador.
It remains to be seen whether epiphytes show greater compositional turnover with a given change in elevation, rainfall, or exposure than other kinds of plants, but their strong association with
montane regions creates the possibility of substantial shifts in
composition with total changes of elevation and precipitation.
4.3.5. Invasion of extensive, fertile, topographically, climatically, and
edaphically complex montane regions
In general, limited dispersal should elevate speciation rate, by
decreasing the scale at which genetic differentiation occurs within
species and increasing the likelihood that such differences would
become large enough to lead to partial mating barriers and, ultimately, to speciation and endemism at small spatial scales (Mayr,
1970; Diamond et al., 1976; Givnish et al., 1995, 2009). This prediction is in accord with the relative high rate of net species diversification seen in Navia–Brewcaria (D = 0.55 My 1), associated with no
apparent means of long-distance seed dispersal (Table 1), and with
a general tendency for parallel, species-rich adaptive radiations to
occur in poorly dispersing groups, from fossorial rodents, brooding
sea anemones, and heavy seeded monocots, to fleshy-fruited plants
dispersed by sedentary forest-interior birds, to philopatric, mouthbreeding cichlids of the African Rift Lakes (Givnish, 1998, 1999a,b,
2010).
The capacity for occasional medium- to long-range dispersal
can, however, also lead to rapid speciation if it permits invasion
of an extensive, topographically dissected region, but does not allow the frequent crossing of habitat barriers within such a region.
The rise of epiphytism in tillandsioids and bromelioids is likely to
have accelerated speciation in this way by increasing seed dispersal ability and consequent ability to colonize different sites
and differentiate populations in parallel along the lengths of the
Andes and the Serra do Mar. Across 172 bromeliad species in Bolivia, Kessler (2002a) found that range size is greater in (a) species
with fleshy fruits (bromelioids) or plumose seeds (tillandsioids)
vs. those with winged seeds; (b) epiphytic vs. terrestrial taxa;
and (c) species at lower vs. higher elevations. The first pattern supports our assumption that entangling, plumose seeds and fleshy
fruits adapted for epiphytism also have greater dispersal ability
than winged seeds. The second follows immediately from the first,
given that almost all epiphytic taxa are tillandsioids and bromelioids. In bromelioids, dispersal of fleshy fruits over longer distances
may be more likely in epiphytic than terrestrial taxa, given that
canopy birds are more mobile than those in the understory (Diamond et al., 1976; Burney and Brumfield, 2009). Many bromelioid
tank epiphytes, however, occur in the understory of the Atlantic
forests (Leme and Marigo, 1993; Siqueira Filho and Leme, 2007),
raising the possibility for differentiation and, ultimately, speciation
and endemism at small spatial scales within such groups. The third
pattern almost surely results from epiphytism being favored in
cloud forests and the increasing spatial discontinuity of bands of
such vegetation at higher elevations.
Given the greater dissection of higher-elevation habitats by
drier valleys, it is not surprising that range size decreases with elevation (Kessler, 2002a, 2002b), or that closely related species at
mid elevations often have peripatric ranges that abut deep Andean
valleys (see Berry, 1982, 1989; Norman, 2000; Andersson, 2006;
Smith and Baum, 2006; Antonelli et al., 2009). The ability of a
lineage to invade an elongate geographic area with a wide range
of elevations to partition and many natural barriers to dispersal –
especially in the latitudinally extensive, topographically and
edaphically complex Andes and Serra do Mar (Luteyn, 2002; Young
et al., 2002) – coupled with a tendency to speciate at small spatial
scales at mid elevations, should lead to high levels of species
diversity at continental scales. As previously argued, richer substrates are likely to favor epiphytes even more within such regions;
the much greater diversity of (mainly epiphytic) Ericaceae in the
Andes (586 spp.) vs. the Guayana Highlands (71 spp.) (Luteyn,
2002) is consistent with this argument. Furthermore, if such montane regions were newly formed, ecologically empty slates, we
would expect several lineages to diversify rapidly within them,
irrespective of life form (Linder, 2008).
The above ideas are confirmed by the association of the Andean
and Serra do Mar orogenies – marked by newly formed montane
habitats, numerous natural barriers to gene flow, and highly dynamic climates and landscapes during the Plio-Pleistocene (Gentry,
1982; Berry, 1989; Vasconcelos et al., 1992; van der Hammen,
1995; Gregory-Wodzicki, 2000; Grazziotin et al., 2006; Garzione
et al., 2008; Antonelli and Sanmartin, 2011) – with several highly
diverse plant groups that are widely distributed within each region
(e.g., Calceolaria, Centropogon-Burmeistera, Epidendrum, Espeletia,
Fuchsia, Lepanthes, and Pleurothallis in the Andes; and Asteraceae
[Lychnophora], Bromeliaceae-Bromelioideae, Eriocaulaceae [Leiothrix, Paepalanthus], Gentianaceae, Lentibulariaceae, Myrtaceae,
and Velloziaceae in the Serra do Mar) (Gentry, 1982; Peixoto and
Gentry, 1990; Luteyn, 2002; Young et al., 2002; Knox et al., 2008;
Alves and Kolbek, 2009). The northern Andes are a hotspot for
diversity of angiosperm families, especially herbaceous ones (Hawkins et al., 2011). Several clades with the highest rates of net diversification known in plants occur wholly or partly in the Andes,
including Andean Lupinus (2.49–3.79 My 1 [Hughes and Eastwood,
2006], Andean Valeriana (1.71–3.2 My 1 [Bell and Donoghue,
2005], Neotropical Costus (0.6–2.6 My 1 [Kay et al., 2005], and Andean Astragalus (2.01–2.07 My 1 [Scherson et al., 2008]). Many of
these and other plant and animal clades from the Andes have origins around or after the uplift of the northern cordilleras ca. 15
Mya (Hoorn et al., 2010). To this roster we now add the largely Andean, largely epiphytic core tillandsioids (0.67 My 1) and, especially, bromelioid tank epiphytes from the Atlantic Forest region
(1.05 My 1).
Diversification rates for the fastest diversifying bromeliad
clades are comparable to those for the fastest adaptive radiations
of plant lineages on islands, including the Hawaiian silversword
alliance (0.56 My 1 [Baldwin and Sanderson, 1998]), Macaronesian
Sideritis and Echium (0.79 and 0.60 My 1 [Kim et al., 2008]), and
Hawaiian Cyanea (maximum of 1.36–2.09 My 1 [Givnish et al.,
2009]). They are, however, substantially less than those for South
African ice plants (1.75 My 1 [Klak et al., 2004]), European Dianthus (2.2 My 1 [Valente et al., 2010]) and certain groups of African
rift-lake cichlids (2.93 My 1 for Bermin Tilapia, 6.1 My 1 for Malawi Astatotilapia, and 178.8 My 1 for Natron Oreochromis [based on
data compiled by Seehausen, 2006; see also McCune, 1997]).
The isolation of many bromeliad taxa on isolated peaks or inselbergs (especially in the Serra do Mar) may select for reduced dispersal and thereby trigger additional pulses of local speciation, as
Moyle et al. (2009) have argued for the ‘‘great speciator’’ lineage
of Asian and Pacific white-eyes (Zosterops). Our explanations for rapid speciation in epiphytic tillandsioids and bromelioids differ
from those of Qian and Ricklefs (2000), Schluter (2000) and Linder
(2008) in incorporating the effect of specific biological features of
epiphytes on the ability to infiltrate, speciate, and diversify across
extensive montane regions.
4.4. Bromeliad diversity in relation to regional patterns and broadscale adaptive radiation
Our findings confirm the widespread pattern for many Neotropical plant groups to have diversified mostly since the Middle Miocene, associated with uplift of the northern Andes and the Serro do
Mar and nearby ranges in southeastern Brazil (Givnish et al., 2004,
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
2007, 2011; Hughes and Eastwood, 2006; Antonelli et al., 2009;
Antonelli and Sanmartin, 2011; Arakaki et al., 2011; Hoorn et al.,
2010; Nagalingum et al., 2011; Drew and Sytsma, 2013; Perret
et al., 2013; Roncal et al., 2013; cf. Rull, 2008; Lohmann et al.,
2013). Bromeliads differ from most of these groups, however, in
containing the only major lineages of epiphytes studied to date,
and in having an ancient stem age (97.5 Mya) but a recent crown
age (22.7 Mya). Although bromeliads arose about when continental
drift split South America from Africa, they show no trace of ancient
colonization of the Paleotropics (Givnish et al., 2011). While diversification of several Neotropical plant groups appears related to
edaphic diversity or rain shadows created by northern Andean uplift, our results indicate that bromeliad diversity is tied more to
epiphytism and associated traits arising on moist, fertile, montane
conditions in the northern Andes and southeastern Brazil, to the
occupancy of extensive and heterogeneous cordilleras, and to
coevolution with hummingbird pollinators whose own diversity
is centered on the Andes.
The two regions worldwide with the highest average rates of
net species diversification in previously studied plant groups are
the Andes and the Atlantic forests (Jansson and Davies, 2008) –
the two areas where we have documented the highest rates of
diversification in bromeliads. The tropical Andes are the world’s
leading hotspot for plant biodiversity, with ca. 15% of all angiosperms worldwide (Gentry, 1982; Myers et al., 2000). The Andes
and the Brazilian coastal forests and mountains rank first and third,
respectively, in terms of the number of species endemic to individual regions (Jansson and Davies, 2008). Lineages within Bromeliaceae parallel these global patterns of diversification, species
richness, and richness of endemic species at the family level, and
detailed studies of factors in each that promote genetic differentiation, ecological specialization, and regional colonization could
thus provide a model for future studies of plant diversity and
diversification. Elevational heterogeneity is a major driver of plant
species richness worldwide (Kreft and Jetz, 2007). Our thesis is
that, for different reasons, elevational heterogeneity and life on
extensive, fertile cordilleras are also major drivers of epiphytic
plant diversity.
On a global scale, Jansson and Davies (2008) found that differences in species richness between sister families of angiosperms
are most strongly correlated with differences in the areas of their
geographic ranges. This pattern also appears to hold, at least qualitatively, across bromeliad subfamilies, with the four having the
smallest ranges – Brocchinioideae, Lindmanioideae, Navioideae,
Hechtioideae – having the fewest species (20–107), while the
two with the largest ranges (Tillandsioideae, Bromelioideae) have
the greatest number of species (1256 and 856 species,
respectively).
Previous attempts to explain differences among lineages in
diversity based on differences in the spatial extent of regions invaded, or the diversity of adaptive zones occupied (e.g., Ricklefs
and Renner, 1994; Davies et al., 2004), have foundered on the question of whether broad-scale adaptive radiation or geographic
spread caused extensive speciation, or extensive speciation permitted adaptive differences and geographic differences in distribution to accumulate (Dodd et al., 1999; Givnish, 2010). We have
attempted to cut this Gordian knot by coupling specific traits (or
suites of traits) to individual bromeliad clades and to the particular
regions and environmental conditions they were able to invade by
evolving those traits, tracing the causal arrow as flying from functional diversification to geographic diversification to species diversification. This does not remove the problem entirely. But invasion
of geographic regions and ecological zones by particular lineages
facilitated by particular traits, with speciation ensuing as those lineages spread, makes it hard to argue that speciation drives the
accumulation, through some random process, of adaptive and
73
geographic/functional diversity within a lineage, rather than functional divergence, then geographic spread driving speciation in a
straightforward fashion. Further, we argue that causes of bromeliad diversity should be sought not only in the rise of traits that
accelerate the net rate of diversification, but also in the cumulative
effect of broad-scale radiations on the total range of adaptive zones
and geographic areas added, and their associated complements of
clades and species.
Specifically, the subfamilies Brocchinioideae, Lindmanioideae,
and Navioideae – which continue to be limited to the nutrientpoor, extremely rainy habitats of the Guayana Shield where the
family arose – comprise 172 species or 5.4% of present-day bromeliad species. Evolution of epiphytism and a suite of related traits
added 1256 species of tillandsioids, mainly in the Andes, but
spreading throughout Central and South America, the Caribbean,
and southeastern North America, as well as 629 species of bromelioid tank epiphytes, mainly in the Serra do Mar and nearby areas
(Table 5). Epiphytism could thus be said to have increased the
number of bromeliad species 11-fold, and the geographic area
occupied by at least 30-fold and to account for 60% of current species richness. Evolution of CAM in terrestrial species in arid habitats and microsites permitted the radiation of (1) Hechtia at low
to mid elevations in Central America, adding 52 species; (2) Deuterocohnia–Dyckia-Encholirium at low to high elevations in the Andes
and drier parts of the Brazilian Shield, adding 170 species; and (3)
Puya at high elevations in the Andes, adding 217 species (Table 5).
CAM in terrestrial plants thus appears responsible for adding ca.
430 species, or 13.6% of current bromeliad diversity. The final major radiation in bromeliads outside the Guayana Shield was that of
Pitcairnia (387 spp., 12.3% of all bromeliads), many bearing large,
broad leaves and adapted to rain- and cloud-forest understories.
Two radiations based on epiphytism, three based on terrestrial
CAM plants, and one based on tropical forest understories thus
underlie, in some sense, 86% of all current bromeliad species. This
is not to say that divergence in pollinators, elevation, rainfall, exposure, or frugivores, or the acquisition of mutualists like ants, were
unimportant, or that they did not, in fact, contribute other axes and
specific shares of biodiversity to the six radiations just listed. But
we can now see that these six large-scale radiations account, directly or indirectly, for more than six-sevenths of total presentday bromeliad diversity, partly through the acceleration of net species diversification in epiphytic lineages inhabiting the Andes and
the Atlantic Forest region, and partly through additions to the total
range of physiological capabilities in bromeliads and thus, to the
geographic regions occupied by different bromeliad clades in
aggregate (see graphical abstract).
This study is the first to relate determinants of net diversification within a major group to a priori predictions that particular
key innovations would lead their bearers to invade specific ecological zones and geographic areas (including ‘‘key landscapes’’) and
diversify at different rates. Such predictions, when combined with
differences among clades in dispersalability and thus the tendency
to speciate, can help explain the overall diversity of innovations,
species, ecological zones, and geographic areas invaded, summed
across radiations. Moore and Donoghue (2007) and Drummond
et al. (2012b) addressed similar questions involving two families
of Dipsacales and for Lupinus. Both studies identified invasion of
extensive cordilleras as a major determinant of diversification,
but failed to recognize the cumulative contribution of different
radiations to overall diversity even when diversification showed
no acceleration in any particular radiation, failed to identify key
innovations that might underlie the invasion of different regions
and ecological zones, and failed to recognize the importance of
shared key landscapes (e.g., recently uplifted, extensive cordilleras)
that could themselves trigger multiple radiations (Givnish, 1997).
The assertion by Drummond et al. (2012b) that perenniality is
74
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
the key innovation driving high rates of diversification in Lupinus is
not plausible: most montane lineages are composed of perennials
but none have as high a diversification rate as Lupinus.
We believe that an approach similar to ours might help revolutionize the study of diversification, biogeography, and adaptive
radiation in many other groups. It provides a direct means for testing a priori hypotheses about the relationships among geographic
spread, phenotypic evolution, and net species diversification, and
thus for making historical biogeography a hypothesis-driven enterprise (Crisp et al., 2011). Just as importantly, the results of our
study show that – while phylogenetic niche conservatism, or the
tendency for closely related species to have similar ecologies and
traits, may be a broad rule in many plant groups (Donoghue,
2008; Wiens et al., 2010) – in bromeliads, at least, it is precisely
when adaptive radiation and the invasion of new geographic areas
jump the tracks of phylogenetic niche conservatism that lineages
can spawn large amounts of new species diversity.
Acknowledgments
The authors gratefully acknowledge financial support for this
investigation by Grants from the National Science Foundation to
PEB, KJS, and TJG (DEB-9981587), TJG (DEB-0830036 and DEB0830036), KJS (DEB-0431258), and GKB and TME (DEB-0129446
and DEB-0129414), and from the Hertel Gift Fund to TJG; from
the Commission for Interdisciplinary Ecological Studies (KIÖS) at
the Austrian Academy of Sciences (ÖAW; 2007-02 to WT and
MHJB; and from the Deutsche Forschungsgemeinshaft (ZI 557/71, SCHU 2426/1-1) and the Hessian initiative for the development
of scientific and economic excellence (LOEWE) at the Biodiversity
and Climate Research Centre, Frankfurt am Mein, to GZ and KS.
Plant material was kindly supplied by the Selby Botanical Garden,
the Palmengarten Frankfurt am Main, and the Botanical Garden of
Heidelberg. Cecile Ané provided invaluable advice on implementing MEDUSA and BiSSE.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.ympev.2013.
10.010.
References
Acharya, K.P., Vetaas, O.R., Birks, H.J.B., 2011. Orchid species richness along
Himalayan elevational gradients. J. Biogeogr. 38, 1821–1833.
Adams III, W.W., Martin, C.E., 1986. Physiological consequences of changes in life
form of the Mexican epiphyte Tillandsia deppeana (Bromeliaceae). Oecologia 70,
1432–1439.
Alfaro, M.E., Santini, F., Brock, C., Alamillo, H., Dornburg, A., Rabosky, D.L., Carnevale,
G., Harmon, L.J., 2009. Nine exceptional radiations plus high turnover explain
species diversity in jawed vertebrates. Proc. Natl. Acad. Sci. USA 106, 13410–
13414.
Almeida, F.F.M., 1976. The system of continental rifts bordering the Santos Basin.
Brazil. An. Acad. Bras. Ciênc. 48, 15–26.
Altschuler, D.L., Dudley, R., McGuire, J.A., Wake, D.B., 2004. Resolution of a paradox:
hummingbird flight at high elevation does not come without a cost. Proc. Natl.
Acad. Sci. USA 101, 17731–17736.
Alves, R.J.V., Kolbek, J., 2009. Can campo rupestre vegetation be floristically delimited
based on vascular plant genera? Plant Ecol. 207, 67–79.
Akaike, H., 1974. A new look at the statistical model identification. IEEE Trans.
Automatic Control 19, 716–723.
Amorim, D.deD., Pires, M.R.S., 1996. Neotropical biogeography and a method for
maximum biodiversity estimation. In: Bicudo, C.E.M., Menezes, N.A. (Eds.),
Biodiversity in Brazil: A First Approach. CNPq, São Paulo, pp. 183–291.
Andersson, S., 2006. On the phylogeny of the genus Calceolaria (Calceolariaceae) as
inferred from ITS and plastid matK sequences. Taxon 55, 125–137.
Antonelli, A., Nylander, J.A.A., Persson, C., Sanmartin, I., 2009. Tracing the impact of
the Andean uplift on Neotropical plant evolution. Proc. Natl. Acad. Sci. USA 106,
9749–9754.
Antonelli, A., Sanmartin, I., 2011. Why are there so many plant species in the
Neotropics? Taxon 60, 403–414.
Arakaki, M., Christin, P.A., Nyffeler, R., Lendel, A., Eggli, U., Ogburn, R.M., Spriggs, E.,
Moore, M.J., Edwards, E.J., 2011. Contemporaneous and recent radiations of the
world’s major succulent plant lineages. Proc. Natl. Acad. Sci. USA 108, 8379–8384.
Auler, A.S., Smart, P.L., 2001. Late Quaternary paleoclimate in semiarid northeastern
Brazil from U-series dating of travertine and water-table speleothems. Quat.
Res. 55, 159–167.
Baker, H.G., Baker, I., 1990. The predictive value of nectar chemistry to the
recognition of pollination types. Israel J. Bot. 39, 157–166.
Baldwin, B.G., Sanderson, M.J., 1998. Age and rate of diversification of the Hawaiian
silversword alliance (Compositae). Proc. Natl. Acad. Sci. USA 95, 9402–9406.
Balke, M., Gómez-Zurita, J., Ribera, I., Viloria, A., Zillikens, A., Steiner, J., Garcia, M.,
Hendrich, L., Vogler, A.P., 2008. Ancient associations of aquatic beetles and tank
bromeliads in the Neotropical forest canopy. Proc. Natl. Acad. Sci. USA 105,
6356–6361.
Barfuss, M.H.J., Samuel, R., Till, W., Stuessy, T.F., 2005. Phylogenetic relationships in
subfamily Tillandsioideae (Bromeliaceae) based on DNA sequence data from
seven plastid regions. Am. J. Bot. 92, 337–351.
Bawa, K.S., 1990. Plant-pollinator interactions in tropical rain forests. Annu. Rev.
Ecol. Syst. 21, 399–422.
Behling, H., Negrelle, R.R.B., 2001. Tropical rain forest and climate dynamics of the
Atlantic lowland, southern Brazil, during the late Quaternary. Quat. Res. 56,
383–389.
Bell, C.D., Donoghue, M.J., 2005. Phylogeny and biogeography of Valerianaceae
(Dipsacales) with special reference to the South American valerians. Org. Divers.
Evol. 5, 147–159.
Benner, J.W., 2011. Epiphytes preferentially colonize high-phosphorus host trees in
unfertilized Hawaiian montane forests. Bryologist 114, 335–345.
Benner, J.W., Vitousek, P.M., 2007. Development of a diverse epiphyte community in
response to phosphorus fertilization. Ecol. Lett. 10, 628–636.
Benzing, D.H., 1970. An investigation of 2 bromeliad mymecophytes – Tillandsia butzii
Mez, T. caput-medusae Morren, and their ants. B. Torrey Bot. Club 97, 109–115.
Benzing, D.H., 1980. Biology of the Bromeliads. Mad River Press, Eureka, CA.
Benzing, D.H., 1987. Vascular epiphytism – taxonomic participation and adaptive
diversity. Ann. Missouri Bot. Gard. 74, 183–204.
Benzing, D.H., 1990. Vascular Epiphytes: General Biology and Associated Biota.
Cambridge University Press, New York.
Benzing, D.H., 2000. Bromeliaceae: Profile of an Adaptive Radiation. Cambridge
University Press, Cambridge.
Benzing, D.H., Burt, K.M., 1970. Foliar permeability among 20 species of
Bromeliaceae. B. Torrey Bot. Club 7, 269–279.
Benzing, D.H., Givnish, T.J., Bermudes, D., 1985. Absorptive trichomes in Brocchinia
reducta (Bromeliaceae) and their evolutionary and systematic significance. Syst.
Bot. 10, 81–91.
Benzing, D.H., Renfrow, W.B., 1974. Mineral nutrition of Bromeliaceae. Bot. Gaz.
135, 281–288.
Benzing, D.H., Renfrow, W.B., Henderson, K., Kessel, B., Sulak, J., 1976. The
absorptive capacities of bromeliad trichomes. Am. J. Bot. 63, 1009–1014.
Berry, P.E., 1982. The systematics and evolution of Fuchsia sect. Fuchsia
(Onagraceac). Ann. Missouri Bot. Gard. 69, 1–198.
Berry, P.E., 1989. A systematic revision of Fuchsia sect. Quelusia (Onagraceae). Ann.
Missouri Bot. Gard. 76, 532–584.
Berry, P.E., Hahn, W.J., Sytsma, K.J., Hall, J.C., Mast, A., 2004. Phylogenetic
relationships and bigeography of Fuchsia (Onagraceae) based on noncoding
nuclear and chloroplast DNA data. Am. J. Bot. 91, 601–614.
Bleiweiss, R., 1998. Tempo and mode of hummingbird evolution. Biol. J. Linn. Soc.
65, 63–76.
Brice, A.T., Grau, C.R., 1991. Protein requirements of Costa’s hummingbirds Calypte
costae. Physiol. Zool. 64, 611–626.
Burney, C.W., Brumfield, R.T., 2009. Ecoogy predicts levels of genetic divergence in
Neotropical birds. Am. Nat. 174, 358–368.
Canela, M.B.F., Sazima, M., 2005. The pollination of Bromelia antiacantha (Bromeli
aceae) in southeastern Brazil: ornithophilous versus melittophilous features.
Plant Biol. 7, 411–416.
Cardelus, C.L., Colwell, R.K., Watkins, J.E., 2006. Vascular epiphyte distribution
patterns: explaining the mid-elevation richness peak. J. Ecol. 94, 144–156.
Cardelus, C.L., Mack, M.C., 2010. Nutrient status of epiphytes and their host trees
along an elevational gradient in Costa Rica. Plant Ecol. 207, 25–37.
Cardelus, C.L., Mack, M.C., Woods, C.L., DeMarco, J., Treseder, K.K., 2009. Nutrient
cycling in canopy and terrestrial soils at a lowland rainforest site, Costa Rica.
Plant Soil 318, 47–61.
Castellanos, M.C., Wilson, P., Thomson, J.D., 2003. Pollen transfer by hummingbirds
and bumblebees, and the divergence of pollination modes in Penstemon.
Evolution 57, 2742–2752.
Chadwick, O.A., Derry, L.A., Vitousek, P.M., Huebert, B.J., Hedin, L.O., 1999. Changing
sources of nutrients during four million years of ecosystem development.
Nature 397, 491–497.
Christin, P.-A., Osborne, C.P., Sage, R.F., Arakaki, M., Edwards, E.J., 2011. C4 eudicots
are not younger than C4 grasses. J. Exp. Bot. 62, 3171–3181.
Coltorti, M., Ollier, C.D., 2000. Geomorphic and tectonic evolution of the Ecuadorian
Andes. Geomorphology 32, 1–19.
Colwell, R.K., Rangel, T.F., 2010. A stochastic, evolutionary model for range shifts
and richness on tropical elevational gradients under Quaternary glacial cycles.
Phil. Trans. R. Soc. B 365, 3695–3707.
Crayn, D.M., Winter, K., Smith, J.A.C., 2004. Multiple origins of crassulacean acid
metabolism and the epiphytic habit in the Neotropical family Bromeliaceae.
Proc. Natl. Acad. Sci. USA 101, 3703–3708.
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
Crisp, M.D., Trewick, S.A., Cook, L.G., 2011. Hypothesis testing in biogeography.
Trends Ecol. Evol. 26, 66–72.
Croat, T.B., 1978. Flora of Barro Colorado Island, Panama. Stanford University Press,
Stanford, CA.
Cruden, R.W., 1972. Pollinators in high-elevation ecosystems: relative effectiveness
of birds and bees. Science 176, 1439–1440.
Cruden, R.W., 1997. Implications of evolutionary theory to applied pollination
ecology. Acta Hort. 437, 27–51.
Currie, D.J., Kerr, J.T., 2008. Tests of the mid-domain hypothesis: a review of the
evidence. Ecol. Monogr. 78, 1–18.
da Silva, H.R., Britto-Pereira, M.C., 2006. How much fruit do fruit-eating frogs eat?
An investigation on the diet of Xenohyla truncata (Lissamphibia: Anura:
Hylidae). J. Zool. 270, 692–698.
Davies, T.J., Barraclough, T.G., Savolainen, V., Chase, M.W., 2004. Environmental
causes for plant biodiversity gradients. Phil. Trans. R. Soc. B 359, 1645–1656.
Diamond, J.M., Gilpin, M.E., Mayr, E., 1976. Species-distance relation for birds of the
Solomon Archipelago, and the paradox of the great speciators. Proc. Natl. Acad.
Sci. USA 73, 2160–2164.
Dodd, M.E., Silvertown, J., Chase, M.W., 1999. Phylogenetic analysis of trait
evolution and species diversity variation among angiosperm families.
Evolution 53, 732–744.
Donoghue, M.J., 2008. A phylogenetic perspective on the distribution of plant
diversity. Proc. Natl. Acad. Sci. USA 105, 11549–11555.
Drew, B.T., Sytsma, K.J., 2013. The South American radiation of Lepechinia
(Lamiaceae): phylogenetics, divergence times, and evolution of dioecy. Bot. J.
Linnean Soc. 170, 171–190.
Drummond, A.J., Rambaut, A., 2007. BEAST: Bayesian evolutionary analysis by
sampling trees. BMC Evol. Biol. 7, 214.
Drummond, A.J., Suchard, M.A., Xie, D., Rambaut, A., 2012a. Bayesian phylogenetics
with BEAUTi and the BEAST 1.7. Mol. Biol. Evol. 29, 1969–1973.
Drummond, C.S., Eastwood, R.J., Miotto, S.T.S., Hughes, C.E., 2012b. Multiple
continental radiations and correlates of diversification in Lupinus
(Leguminosae): testing for key innovation with incomplete taxon sampling.
Syst. Biol. 61, 443–460.
Dziedzioch, C., Stevens, A.-D., Gottsberger, G., 2003. The hummingbird plant
community of a tropical montane rain forest in southern Ecuador. Plant Biol. 5,
331–337.
Earnshaw, M.J., Winer, K., Ziegler, H., Stichler, W., Cruttwell, N.E.G., Kerenga, K.,
Cribb, P.J., Wood, J., Croft, J.R., Carver, K.A., Gunn, T.C., 1987. Altitudinal changes
in the incidence of crassulacean acid metabolism in vascular epiphytes and
related life forms in Papua New Guinea. Oecologia 73, 566–572.
Edwards, E.J., Osbourne, C.P., Strömberg, C.A.E., Smith, S.A., 2010. The origins of C4
grasslands: integrating evolutionary and ecosystem science. Science 328, 587–
591.
Ehleringer, J.R., Sage, R.F., Flanagan, L.B., Pearcy, R.W., 1991. Climate change and the
evolution of C4 photosynthesis. Trends Ecol. Evol. 6, 95–99.
Ehlers, T.A., Poulsen, C.J., 2009. Influence of Andean uplift on climate and
paleoaltimetry estimates. Earth Planet. Sci. Lett. 281, 238–248.
Farris, D.W., Jaramillo, C., Bayona, G., Restrepo-Moreno, S.A., Montes, C., Cardona, A.,
Mora, A., Speakman, R.J., Glascock, M.D., Valencia, V., 2011. Fracturing of the
Panamanian Isthmus during initial collision with South America. Geology 39,
1007–1010.
Feinsinger, P., 1983. Coevolution and pollination. In: Futuyma, D.J., Slatkin, M.
(Eds.), Coevolution. Sinauer Associates, Sunderland, MA, pp. 282–310.
FitzJohn, R.G., Maddison, W.P., Otto, S.P., 2009. Estimating trait-dependent
speciation and extinction rates from incompletely resolved phylogenies. Syst.
Biol. 58, 595–611.
Fleming, P.A., Nicolson, S.W., 2003. Osmoregulation in an avian nectarivore, the
whitebellied sunbird Nectarinia talatala: response to extremes of diet
concentration. J Exper. Biol. 206, 1845–1854.
Fontoura, T., Scudeller, V.V., da Costa, A.F., 2012. Floristics and environmental
factors determining the geographic distribution of epiphytic bromeliads in the
Brazilian Atlantic rain forest. Flora 207, 662–672.
France-Lanord, C., Derry, L.A., 1997. Organic carbon burial forcing of the carbon
cycle from Himalayan erosion. Nature 390, 65–67.
Freschi, L., Takahashi, C.A., Cambui, C.A., Semprebom, T.R., Cruz, A.B., Mioto, P.T.,
Versieux, L.D., Calvente, A., Latansio-Aldar, S.R., Aldar, M.P.M., Mercier, H., 2010.
Specific leaf areas of the tank bromeliad Guzmania monostachia perform distinct
functions in response to water shortage. J. Plant Physiol. 167, 526–533.
Friedman, J., Barrett, S.C.H., 2008. A phylogenetic analysis of the evolution of wind
pollination in the angiosperms. Int. J. Plant Sci. 169, 49–58.
Galetto, L., Bernardello, J., 1992. Nectar secretion pattern and removal effects in 6
Argentinean Pitcairioideae (Bromeliaceae). Bot. Acta 105, 292–299.
Gardner, C.S., 1986. Inferences about pollination in Tillandsia (Bromeliaceae).
Selbyana 9, 76–87.
Garzione, C.N., Hoke, G.D., Liarkin, J.C., Withers, S., MacFadden, B., Eiler, J., Ghosh, P.,
Mulch, A., 2008. Rise of the Andes. Science 320, 1304–1307.
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., Dodson, C.H., 1987. Diversity and biogeography of Neotropical
vascular epiphytes. Ann. Missouri Bot. Gard. 74, 205–233.
Gentry, A.H., Emmons, L.H., 1987. Geographical variation in fertility, phenology, and
composition of the understory of tropical forests. Biotropica 19, 216–227.
75
Givnish, T.J., 1997. Adaptive radiation and molecular systematics: aims and
concepts. In: Givnish, T.J., Sytsma, K.J. (Eds.), Molecular Evolution and
Adaptive Radiation. Cambridge University Press, New York, pp. 1–54.
Givnish, T.J., 1998. Adaptive radiation of plants on oceanic islands: classical
patterns, molecular data, new insights. In: Grant, P. (Ed.), Evolution on Islands.
Oxford University Press, New York, pp. 281–304.
Givnish, T.J., 1999a. On the causes of gradients in tropical tree diversity. J. Ecol. 87,
193–210.
Givnish, T.J., 1999b. Adaptive radiation, dispersal, and diversification of the
Hawaiian lobeliads. In: Kato, M. (Ed.), The Biology of Biodiversity. Springer
Verlag, Tokyo, pp. 67–90.
Givnish, T.J., 2005. Hallmarks of the New World. Syst. Biol. 54, 340–344.
Givnish, T.J., 2010. Ecology of plant speciation. Taxon 59, 1326–1366.
Givnish, T.J., Ames, M., McNeal, J.R., McKain, M.R., Steele, P.R., dePamphilis, C.W.,
Graham, S.W., Pires, J.C., Stevenson, D.W., Zomlefer, W.B., Briggs, B.G., Duvall,
M.R., Moore, M.J., Heaney, J.M., Soltis, D.E., Soltis, P.S., Thiele, K., Leebens-Mack,
J.H., 2010. Assembling the tree of the monocotyledons: plastome sequence
phylogeny and evolution of Poales. Ann. Missouri Bot. Gard. 97, 584–616.
Givnish, T.J., Barfuss, M.H.J., Van Ee, B., Riina, R., Schulte, K., Horres, R., Gonsiska, P.A.,
Jabaily, R.S., Crayn, D.M., Smith, J.A.C., Winter, K., Brown, G.K., Evans, T.M., Holst,
B.K., Luther, H., Till, W., Zizka, G., Berry, P.E., Sytsma, K.J., 2011. Adaptive
radiation and diversification in Bromeliaceae: insights from a 7-locus plastid
phylogeny. Am. J. Bot. 98, 872–895.
Givnish, T.J., Burkhardt, E.L., Happel, R.E., Weintraub, J.W., 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. Am.
Nat. 124, 479–497.
Givnish, T.J., Millam, K.C., Evans, T.M., Hall, J.C., Pires, J.C., Berry, P.E., Sytsma, K.J.,
2004. Ancient vicariance or recent long-distance dispersal? Inferences about
phylogeny and South American-African disjunction in Rapateaceae and
Bromeliaceae based on ndhF sequence data. Int. J. Plant Sci. 165, S35–S54.
Givnish, T.J., Millam, K.C., Berry, P.E., Sytsma, K.J., 2007. Phylogeny, adaptive
radiation, and historical biogeography of Bromeliaceae inferred from ndhF
sequence data. In: Columbus, J.T., Friar, E.A., Porter, J.M., Prince, L.M., Simpson,
M.G. (Eds.), Monocots: Comparative Biology and Evolution – Poales. Rancho
Santa Ana Botanic Garden, Claremont, California, pp. 3–26.
Givnish, T.J., Millam, K.C., Theim, T.T., Mast, A.R., Patterson, T.B., Hipp, A.L., Henss,
J.M., Smith, J.F., Wood, K.R., Sytsma, K.J., 2009. Origin, adaptive radiation, and
diversification of the Hawaiian lobeliads (Asterales: Campanulaceae). Proc. R.
Soc. B 276, 407–416.
Givnish, T.J., Pires, J.C., Graham, S.W., McPherson, M.A., Prince, L.M., Patterson, T.B.,
Rai, H.S., Roalson, E.R., Evans, T.M., Hahn, W.J., Millam, K.C., Meerow, A.W.,
Molvray, M., Kores, P., O’Brien, H.E., Kress, W.J., Hall, J.C., Sytsma, K.J., 2005.
Repeated evolution of net venation and fleshy fruits among monocots in shaded
habitats confirms a priori predictions: evidence from an ndhF phylogeny. Proc.
R. Soc. B 272, 1481–1490.
Givnish, T.J., Sytsma, K.J., Smith, J.F., Hahn, W.J., 1995. Molecular evolution, adaptive
radiation, and geographic speciation in Cyanea (Campanulaceae, Lobelioideae).
In: Wagner, W.L., Funk, V.A. (Eds.), Hawaiian Biogeography: Evolution on a Hotspot Archipelago. Smithsonian Institution Press, Washington, D.C., pp. 288–337.
Givnish, T.J., Sytsma, K.J., Smith, J.F., Hahn, W.J., Benzing, D.H., Burkhardt, E.M., 1997.
Molecular evolution and adaptive radiation in Brocchinia (Bromeliaceae:
Pitcairnioideae) atop tepuis of the Guayana Shield. In: Givnish, T.J., Sytsma,
K.J. (Eds.), Molecular Evolution and Adaptive Radiation. Cambridge University
Press, New York, pp. 259–311.
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., 1993. True tillandsias misplaced in Vriesea. Phytologia 75, 170–175.
Gravendeel, B., Smithson, A., Silk, F.J.W., Schuiteman, A., 2004. Epiphytism and
pollinator specialization: drivers for orchid diversity? Phil. Trans. R. Soc. B 359,
1523–1535.
Grazziotin, F.G., Monzel, M., Echeverrigaray, S., Bonatto, S.L., 2006. Phylogeography
of the Bothrops jararaca complex (Serpentes: Viperidae): past frag mentation
and island colonization in the Brazilian Atlantic Forest. Mol. Ecol. 15, 3969–
3982.
Gregory-Wodzicki, K.M., 2000. Uplift history of the central and northern Andes: a
review. Geol. Soc. Am. Bull. 112, 1091–1105.
Griffiths, H., 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.
Grytnes, J.A., Beaman, J.H., 2006. Elevational species richness patterns for vascular
plants on Mount Kinabalu. Borneo. J. Biogeogr. 33, 1838–1849.
Haeger, A., Dohrenbusch, A., 2011. Hydrometeorology and structure of tropical
montane forests under contrasting biophysical conditions in north-western
Costa Rica. Hydrol. Process. 25, S392–S401.
Hawkins, B.A., Rodríguez, M.Á., Weller, S.G., 2011. Global angiosperm family
richness revisited: linking ecology and evolution to climate. J. Biogeogr. 38,
1253–1266.
Holdgate, G.R., Clarke, J.D.A., 2000. A review of Tertiary brown coal deposits in
Australia: their depositional factors and eustatic correlations. AAPG Bull. 84,
1129–1151.
Holst, B.K., 1997. Bromeliaceae. In: Berry, P.E., Holst, B.K., Yatskievych, K. (Eds.),
Flora of the Venezuelan Guayana, Missouri Botanical Garden, vol. 2. St. Louis,
Missouri, pp. 548–676.
76
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
Hoorn, C., 1994. An environmental reconstruction of the palaeo-Amazon river
system (middle-late Miocene, NW Amazonia. Palaeogeogr. Palaeocl. 112, 187–
238.
Hoorn, C., Guerrero, J., Sarmiento, G.A., Lorente, M.A., 1995. Andean tectonics as a
cause for changing drainage patterns in Miocene northern South America.
Geology 23, 237–240.
Hoorn, C., Wesselingh, F.P., ter Steege, H., Bermudez, M.A., Mora, A., Sevink, J.,
Sanmartin, I., Sanchez-Messeguer, A., Anderson, C.L., Figueirado, J.P., Jaramillo,
C., Riff, D., Negri, F.R., Hooghiemstra, H., Lundberg, J., Stadler, T., Särkinnen, T.,
Antonelli, A., 2010. Amazonia through time: Andean uplift, climate change,
landscape evolution, and biodiversity. Science 330, 927–931.
Hughes, C., Eastwood, R., 2006. Island radiation on a continental scale: exceptional
rates of plant diversification after uplift of the Andes. Proc. Natl. Acad. Sci. USA
103, 10334–10339.
Huxley, C.R., 1980. Symbiosis between ants and epiphytes. Biol. Rev. 55, 321–340.
Ibisch, P.L., 1996. Neotropisiche Epiphytendiversität – das Beispiel Bolivien. Martina
Galunder Verlag, Wiehl.
Inselbacher, E., Cambui, C.A., Richter, A., Stange, C.F., Mercier, H., Wanek, W., 2007.
Microbial activities and foliar uptake of nitrogen in the epiphytic bromeliad
Vriesea gigantea. New Phytol. 175, 311–320.
Jabaily, R.S., Sytsma, K.J., 2010. Phylogenetics of Puya (Bromeliaceae): placement,
major lineages, and evolution of Chilean species. Am. J. Bot. 97, 337–356.
Jansson, R., Davies, T.J., 2008. Global variation in diversification rates of flowering
plants: energy vs. climate change. Ecol. Lett. 11, 173–183.
Janzen, D.H., 1974a. Epiphytic myrmecophytes in Sarawak: mutualism through the
feeding of plants by ants. Biotropica 6, 237–259.
Janzen, D.H., 1974b. Tropical blackwater rivers, animals, and mast fruiting by the
Dipterocarpaceae. Biotropica 6, 69–103.
Janzen, D.H., 1977. Promising directions of study in tropical animal-plant
interactions. Ann. Missouri Bot. Gard. 64, 706–736.
Kay, E., 2003. Floral evolutionary ecology of Passiflora (Passifloraceae), subgenera
Murucuia, Pseudomurucuja, and Astephia. Ph.D. dissertation, Saint Louis
University.
Kay, K.M., Reeves, P.A., Olmstead, R.G., Schemske, D.W., 2005. Rapid speciation and
the evolution of hummingbird pollination in Neotropical Costus subgenus
Costus (Costaceae): evidence from nrDNA ITS and ETS sequences. Am. J. Bot. 92,
1899–1910.
Kessler, M., 2001. Patterns of diversity and range size of selected plant groups along
an elevational transect in the Bolivian Andes. Biod. Conserv. 10, 1897–1921.
Kessler, M., 2002a. Environmental patterns and ecological correlates of range size
among bromeliad communities of Andean forests in Bolivia. Bot. Rev. 68, 100–
127.
Kessler, M., 2002b. Species richness and ecophysiological types among Bolivian
bromeliad communities. Biodiv. Conserv. 11, 987–1010.
Kessler, M., Krömer, T., 2000. Patterns and ecological correlates of pollination modes
among bromeliad communities of Andean forests in Bolivia. Plant Biol. 2, 659–
669.
Kim, S.C., McGowen, M.R., Lubinsky, P., Barber, J.C., Mort, M.E., Santos-Guerra, A.,
2008. Timing and tempo of early and successive adaptive radiations in
Macaronesia. PLoS ONE 3, e2139.
Kirby, M.X., Jones, D.S., MacFadden, B.J., 2008. Lower Miocene stratigraphy along the
Panama Canal and its bearing on the Cental American peninsula. PLoS ONE 3,
e2791.
Klak, C., Reeves, G., Hedderson, T., 2004. Unmatched tempo of evolution in Southern
African semi-desert ice plants. Nature 427, 63–65.
Kluge, M., Lange, O.L., Eichmann, M.V., Schmid, R., 1973. CAM in Tillandsia
usneoides: studies on pathway of carbon and dependency of CO2 exchange
on light intensity, temperature and water content of plant. Planta 112, 357–
372.
Knox, E.B., Muasya, A.M., Muchhala, N., 2008. The predominantly South American
clade of Lobeliaceae. Syst. Bot. 33, 462–468.
Kreft, H., Jetz, W., 2007. Global patterns and determinants of vascular plant
diversity. Proc. Natl. Acad. Sci. USA 104, 5925–5930.
Kreft, H., Köster, N., Küper, W., Nieder, J., Barthlott, W., 2004. Diversity and
biogeography of vascular epiphytes in western Amazonia, Yasuní, Ecuador. J.
Biogeogr. 31, 1463–1476.
Krömer, T., Kessler, M., Gradstein, S.R., Aceby, A., 2005. Diversity patterns of
vascular epiphytes along an elevational gradient in the Andes. J. Biogeogr. 32,
1799–1809.
Krömer, T., Kessler, M., Herzog, S.K., 2006. Distribution and flowering ecology of
bromeliads along two climatically contrasting elevational transects in the
Bolivian Andes. Biotropica 38, 183–195.
Krömer, T., Kessler, M., Lohaus, G., Schmidt-Lebuhn, A.N., 2008. Nectar sugar
composition and concentration in relation to pollination syndromes in
Bromeliaceae. Plant Biol. 10, 502–511.
Kürschner, W.M., Kvacek, Z., Dilcher, D.L., 2008. The impact of Miocene atmospheric
carbon dioxide fluctuations on climate and the evolution of terrestrial
ecosystems. Proc. Natl. Acad. Sci. USA 105, 449–453.
Leme, E.M.C., Marigo, L.C., 1993. Bromeliads in the Brazilian Wilderness. Marigo
Comunicacao Visual Ltda.
Linares-Palomino, R., Kessler, M., 2009. The role of dispersal ability, climate and
spatial separation in shaping biogeographical patterns of phylogenetically
distant plant groups in seasonally dry Andean forests of Bolivia. J. Biogeogr. 36,
280–290.
Linares-Palomino, R., Cardona, V., Hennig, E.I., Hensen, I., Hoffmann, D., Lendzion, J.,
Soto, D., Herzog, S.K., Kessler, M., 2009. Non-woody life-form contribution to
vascular plant species richness in a tropical American forest. Plant Ecol. 201,
87–99.
Linder, H.P., 2008. Plant species radiations: where, when, and why? Phil. Trans. R.
Soc. B 363, 3097–3105.
Lohmann, L.G., Bell, C.D., Calió, M.F., Winkworth, R.C., 2013. Pattern and timing of
biogeographical history in the Neotropical tribe Bignonieae. Bot. J. Linnean Soc.
170, 154–170.
Luteyn, J.L., 2002. Diversity, adaptation, and endemism in Neotropical Ericaceae:
biogeographical pattersn in the Vaccinieae. Bot. Rev. 68, 55–87.
Luther, H., 2008. An Alphabetical List of Bromeliad Binomials, 11th ed. Bromeliad
Society International, <http://www.selby.org/clientuploads/research/Bromeliaceae/
Binomial_BSI_2002.pdf>.
Mabberley, D.J., 1975. Giant lobeliads: pachycauly, biogeography, ornithophily and
continental drift. New Phytol. 74, 365–374.
Maddison, D.R., Maddison, W.P., 2005. MacClade 4: Analysis of Phylogeny and
Character Evolution (version 4.08). Sinauer Assoc., Sunderland, Massachusetts.
Maddison, W.P., Midford, P.E., Otto, S.P., 2007. Estimating a binary character’s effect
on speciation and extinction. Syst. Biol. 56, 701–710.
Magallón, S., Castillo, A., 2009. Angiosperm diversification through time. Am. J. Bot.
96, 349–365.
Magallón, S., Sanderson, M.J., 2001. Absolute diversification rates in angiosperm
clades. Evolution 55, 1762–1780.
Martén-Rodríguez, S., 2008. Evolution of pollination and breeding systems of
Antillean Gesneriaceae. Ph.D. Dissertation, University of Maryland, College Park.
Martínez del Rio, C., 1994. Nutritional ecology of fruit-eating and flower-visiting
birds and bats. In: Chivers, D., Langer, P. (Eds.), The Digestive System in
Mammals: Food Form and Function. Cambridge University Press, Cambridge,
pp. 103–127.
Martorell, C., Ezcurra, E., 2007. The narrow-leaf syndrome: a functional and
evolutionary approach to the form of fog-harvesting rosette plants. Oecologia
151, 561–573.
Mayr, E., 1970. Populations, Species, and Evolution. Belknap Press, Cambridge, MA.
McCune, A.R., 1997. How fast is speciation: molecular, geological and phylogenetic
evidence from adaptive radiations of fishes. In: Givnish, T.J., Sytsma, K.J. (Eds.),
Molecular Evolution and Adaptive Radiation. Cambridge University Press, New
York, pp. 585–610.
McGuire, J., Witt, C.C., Altshuler, D.L., Remsen Jr., J.V., 2007. Phylogenetic
systematics and biogeography of hummingbirds: Bayesian and maximum
likelihood analyses of partitioned data and selection of an appropriate
partitioning strategy. Syst. Biol. 56, 637–656.
McGroddy, M.E., Daufresne, T., Hedin, L.O., 2004. Scaling of C:N:P stoichiometry in
forest ecosystems worldwide. Ecology 85, 2390–2401.
McWilliams, E.L., 1974. Evolutionary ecology. In: Smith, L.B., Downs, R.J. (Eds.),
Pitcairnioideae (Bromeliaceae). Flora Neotropica Monograph. New York
Botanical Garden Press, Bronx, New York, pp. 40–55, Vol. 14.
Medina, E., 1974. Dark CO2 fixation, habitat preference and evolution within the
Bromeliaceae. Evolution 28, 677–686.
Mez, C., 1904. Physiologische Bromeliaceen-Studien I. Die Wasser-Ökonomie der
extrem atmosphärischen Tillandsien. Jahrb. Wissensc. Bot. 40, 157–229.
Montes, C., Cardona, A., McFadden, R., Morón, S.E., Silva, C.A., Restrepo-Moreno, S.,
Ramírez, D.A., Hoyos, N., Wilson, J., Farris, D., Bayona, G.A., Jaramillo, C.A.,
Valencia, V., Bryan, J., Flores, J.A., 2012. Evidence for Middle Eocene and younger
land emergence in central Panama: implications for Isthmus closure. Geol. Soc.
Amer. Bull. 124, 780–799.
Moore, B.R., Donoghue, M.J., 2007. Correlates of diversification in the plant clade
Dipsacales: geographic movement and evolutionary innovations. Am. Nat. 170,
S28–S55.
Moyle, R.G., Filardi, C.E., Smith, C.E., Diamond, J., 2009. Explosive Pleistocene
diversification and hemispheric expansion of a ‘‘great speciator’’. Proc. Natl.
Acad. Sci. USA 106, 1863–1868.
Muchhala, N., Thomson, J.D., 2010. Fur versus feather: pollen delivery by bats and
hummingbirds and consequences for pollen production. Am. Nat. 175, 717–726.
Myers, N., Mittemeier, R.A., Mittemeier, C.G., da Fonseca, G.A.B., Kent, J., 2000.
Biodiversity hotspots for conservation priorities. Nature 403, 853–858.
Nadkarni, N., Solano, R., 2002. Potential effects of climate change on canopy
communities in a tropical cloud forest: an experimental approach. Oecologia
131, 580–586.
Nagalingum, N.S., Marshall, C.R., Quental, T.B., Rai, H.S., Little, D.R., Mathews, S.J.,
2011. Recent synchronous radiation of a living fossil. Science 334, 796–799.
Ngai, J.T., Srivastava, D.S., 2006. Predators accelerate nutrient cycling in a bromeliad
ecosystem. Science 314, 963.
Norman, E., 2000. Buddlejaceae. Flora Neotropica Monograph, vol. 81. New York
Botanical Garden Press, Bronx, New York.
Obregon, A., Gehrig-Downie, C., Gradstein, S.R., Rollenbeck, R., Bendix, J., 2011.
Canopy level fog occurrence in a tropical lowland forest of French Guiana as a
prerequisite for high epiphyte diversity. Agr. Forest Meteorol. 151, 290–300.
Owen, T.P., Benzing, D.H., Thomson, W.W., 1988. Apoplastic and ultrastructural
characterizations of the trichomes of the carnivorous bromeliad Brocchinia
reducta. Can. J. Bot. 66, 941–948.
Pagel, M., 1994. Detecting correlated evolution on phylogenies: a general method
for the comparative analysis of discrete characters. Proc. R. Soc. B 255, 37–45.
Pagel, M., 1999. The maximum likelihood approach to reconstructing ancestral
character states of discrete characters on phylogenies. Syst. Biol. 48, 612–622.
Pagel, M., Meade, A., 2007. BayesTraits, version 1.0. Computer Program and
Documentation Distributed by the Author. <http://www.evolution.rdg.ac.uk>
(accessed 10.10.11).
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
Pagel, M., Meade, A., 2008. Modelling heterotachy in phylogenetic inference by
reversible-jump Markov chain Monte Carlo. Phil. Trans. R. Soc. B 363, 3955–
3964.
Pagel, M., Meade, A., Barker, D., 2004. Bayesian estimation of ancestral character
states on phylogenies. Syst. Biol. 53, 673–684.
Patterson, B.D., Pacheco, V., Solari, S., 1996. Distributions of bats along an
elevational gradient in the Andes of southeastern Peru. J. Zool. 240, 637–658.
Peixoto, A.L., Gentry, A.H., 1990. Diversidade e composição florística da mata de
tabuleiro na Reserva Florestal de Linhares (Espírito Santo, Brasil). Rev. Bras. Bot.
13, 19–25.
Perret, M., Chautems, de Araujo, A.O., Salamin, N., 2013. Temporal and spatial origin
of Gesneriaceae in the New World inferred from plastid DNA sequences. Bot. J.
Linnean Soc. 170, 61–79.
Perrett, M., Chautems, A., Spichiger, R., Barraclough, T.G., Savolainen, V., 2007. The
geographical pattern of speciation and floral diversification in the Neotropics:
the tribe Sinningieae (Gesneriaceae) as a case study. Evolution 61, 1641–1660.
Picado, C., 1913. Les Bromeliacées epiphytes considérée comme milieu biologique.
Bull. Sci. France Belgique. 47 (215), 360.
Piacentini, V.D., Varassin, I.G., 2007. Interaction network and the relationships
between bromeliads and hummingbirds in an area of secondary Atlantic forest
in southern Brazil. J. Trop. Ecol. 23, 663–671.
Pierce, S., Maxwell, K., Griffiths, H., Winter, K., 2001. Hydrophobic trichome layers
and epicuticular wax powders in Bromeliaceae. Am. J. Bot. 88, 1371–1389.
Pierce, S., Winter, K., Griffiths, H., 2002. The role of CAM in high rainfall cloud
forests: an in situ comparison of photosynthetic pathways in Bromeliaceae.
Plant, Cell Environ. 25, 1181–1189.
Pittendrigh, C.S., 1948. The bromeliad-Anopheles-malaria complex in Trinidad. I. The
bromeliad flora. Evolution 2, 58–89.
Porder, S., Vitousek, P.M., Chadwick, O.A., Chamberlain, C.P., Hilley, G.E., 2007.
Uplift, erosion, and phosphorus limitation in terrestrial ecosystems. Ecosystems
10, 158–170.
Posada, D., Crandall, K.A., 1998. Modeltest: testing the model of DNA substitution.
Bioinformatics 14, 817–818.
Potter, P.E., 1997. The Mesozoic and Cenozoic paleodrainage of South America: A
natural history. J. S. Am. Earth Sci. 10, 331–344.
Quezada, I.M., Gianoli, E., 2011. Crassulacean acid metabolism photosynthesis in
Bromeliaceae: an evolutionary key innovation. Biol. J. Linn. Soc. 104, 480–486.
Qian, H., Ricklefs, R.E., 2000. Large-scale processes and the Asian bias in species
diversity of temperate plants. Nature 407, 180–182.
R Core Team, 2012. R: A Language and Environment for Statistical Computing. R
Foundation for Statistical Computing, Vienna, Austria. <http://www.Rproject.org/> (accessed 15.12.12).
Rabosky, D.L., Lovette, I.J., 2008. Density-dependent diversification in North
American wood warblers. Proc. R. Soc. B 275, 2363–2371.
Rambaut, A., Drummond, A. J., 2007. Tracer 1.4. <http://beast.bio.ed.ac.uk/Tracer>.
Ramirez, S.R., Gravendeel, B., Singer, R.B., Marshall, C.R., Pierce, N.E., 2007. Dating
the origin of the Orchidaceae from a fossil orchid with its pollinator. Nature 448,
1042–1045.
Reyes-Garcia, C., Griffiths, H., Rincon, E., Huante, P., 2008. Niche differentiation in
tank and atmospheric epiphytic bromeliads of a seasonally dry forest.
Biotropica 40, 168–175.
Richter, M., Diertl, K.-H., Emck, P., Peters, T., Beck, E., 2009. Reasons for an
outstanding plant diversity in the tropical Andes of southern Ecuador.
Landscape Online 12, 1–35.
Ricklefs, R.E., Renner, S.S., 1994. Species richness within families of flowering plants.
Evolution 48, 1619–1636.
Robinson, H., Taylor, D.C., 1999. A rejection of Pepinia (Bromeliaceae:
Pitcairnioideae) and taxonomic revisions. Harvard Pap. Bot. 4, 203–217.
Romero, G.Q., Mazzafera, P., Vasconcellos-Neto, J., Trivelin, P.C.O., 2006.
Bromeliad-living spiders improve host plant nutrition and growth. Ecology
87, 803–808.
Romero, G.Q., Nomura, F., Gonçalves, A.Z., Dias, N.Y.N., Mercier, H., Conforto, E.deC.,
Rossa-Ferres, D.deC., 2010. Nitrogen fluxes from treefrogs to tank epiphytic
bromeliads: an isotopic and physiological approach. Oecologia 162, 941–949.
Roncal, J., Kahn, Millan, B., Couvreur, T.L.P., Pintaud, J.-C., 2013. Cenozoic
colonization and diversification patterns of tropical American palms: evidence
from Astrocaryum (Arecaceae). Bot., J. Linnean Soc. 170, 120–139.
Rull, V., 2008. Speciation timing and Neotropical biodiversity: the TertiaryQuaternary debate in the light of molecular phylogenetic evidence. Mol. Ecol.
17, 2722–2729.
Safford, H.D., 1999. Brazilian paramos. I. An introduction to the physical
environment and vegetation of the campos de altitude. J. Biogeogr. 26, 693–712.
Sage, R.F., 2004. The evolution of C4 photosynthesis. New Phytol. 161, 341–370.
Sage, R.F., Christin, P.-A., Edwards, E.J., 2011. The C4 plant lineages of planet Earth. J.
Exp. Bot. 62, 3155–3169.
Sass, C., Specht, C.D., 2010. Phylogenetic estimation of the core bromelioids with an
emphasis on the genus Aechmea (Bromeliaceae). Mol. Phylogenet. Evol. 55, 559–
571.
Sazima, I., Buzato, S., Sazima, M., 1995a. The saw-billed hermit Ramphodon naevius
and its flowers in southeastern Brazil. J. Ornithol. 136, 195–206.
Sazima, I., Buzato, S., Sazima, M., 1996. An assemblage of hummingbird-pollinated
flowers in a montane forest in southeastern Brazil. Bot. Acta 109, 149–160.
Sazima, M., Buzato, S., Sazima, I., 1999. Bat-pollinated flower assemblages and at
visitors at two Atlantic forest sites in Brazil. Ann. Bot. 83, 705–712.
Sazima, I., Vogel, S., Sazima, M., 1989. Bat pollination of Encholirium glaziovii, a
terrestrial bromeliad. Plant Syst. Evol. 168, 167–179.
77
Sazima, M.S., Vogel, M., Sazima, I., 1995b. Bat pollination of Vriesea in southeastern
Brazil. Bromelia 2, 29–37.
Scherson, R.A., Vidal, R., Sanderson, M.J., 2008. Phylogeny, biogeography, and rates
of diversification of New World Astragalus (Leguminosae) with an emphasis on
South American radiations. Am. J. Bot. 95, 1030–1039.
Schimper, A.F.W., 1884. Ueber Bau und Lebensweise der Epiphyten Westindiens.
Bot. Centralbl. 17, 192–195, et seq.
Schimper, A.F.W., 1888. Die epiphytische Vegetation Amerikas. Botanische
Mitteilungen aus den Tropen, Heft 2. Jena, Gustav Fischer Verlag.
Schluter, D., 2000. The Ecology of Adaptive Radiation. Oxford University Press,
Oxford.
Schmidt-Lebuhn, A.N., Kessler, M., Hensen, I., 2007. Hummingbirds as drivers of
plant speciation? Trends Plant Sci. 12, 329–331.
Schulte, K., Barfuss, M.H.J., Zizka, G., 2009. Phylogeny of Bromelioideae
(Bromeliaceae) inferred from nuclear and plastid DNA loci reveals the
evolution of the tank habit within the subfamily. Mol. Phylogenet. Evol. 51,
327–339.
Schulte, K., Horres, R., Zizka, G., 2005. Molecular phylogeny of Bromelioideae and its
implications on biogeography and the evolution of CAM in the family.
Senckenbergiana Biologica 85, 1–13.
Seehausen, O., 2006. African cichlid fish: a model system in adaptive radiation
research. Proc. R. Soc. B 273, 1987–1998.
Silvera, K., Santiago, L.S., Cushman, J.C., Winter, K., 2009. Crassulacean acid
metabolism and epiphytism linked to adaptive radiations in the Orchidaceae.
Plant Physiol. 149, 1838–1847.
Simpson, G.G., 1944. Tempo and Mode in Evolution. Columbia University Press,
New York.
Siqueira Filho, J.A.S., Leme, E.M.C., 2007. Fragments of the Atlantic Forest of
Northeast Brazil: Biodiversity, Conservation and the Bromeliads. Andrea
Jakobsson Estudio, Rio de Janeiro, Brazil.
Skillman, J.B., Garcia, M., Winter, K., 1999. Whole-plant consequences of
crassulacean acid metabolism for a tropical forest understory plant. Ecology
80, 1584–1593.
Skillman, J.B., Winter, K., 1997. High photosynthetic capacity in a shade-tolerant
crassulacean acid metabolism plant. Implications for sunfleck use,
nonphotochemical energy dissipation, and susceptibility to photoinhibition.
Plant Physiol. 113, 441–450.
Smith, J.A.C., 1989. Epiphytic bromeliads. In: Lüttge, U. (Ed.), Vascular Plants as
Epiphytes: Evolution and Ecophysiology. Springer-Verlag, Heidelberg, pp. 109–
138.
Smith, J.F., 2001. High species diversity in fleshy-fruited tropical understory plants.
Am. Nat. 157, 646–653.
Smith, L.B., Down, R.J., 1974. Pitcairnioideae (Bromeliaceae). Flora Neotrop. 14, 1–
662.
Smith, L.B., Down, R.J., 1977. Tillandsioideae (Bromeliaceae). Flora Neotrop. 14,
663–1492.
Smith, L.B., Down, R.J., 1979. Bromelioideae (Bromeliaceae). Flora Neotrop. 14,
1493–2142.
Smith, L.B., Till, W., 1998. Bromeliaceae. In: Kubitzki, K. (Ed.), The Families and
Genera of Vascular Plants, Alismatanae and Commelinanae (except Gramineae),
vol. IV. Springer-Verlag, Berlin, pp. 74–99.
Smith, S.D., Baum, D.A., 2006. Phylogenetics of the florally diverse Andean clade
Iochrominae (Solanaceae). Am. J. Bot. 93, 1140–1153.
Steinmann, M., Hungerbuhler, D., Seward, D., Winkler, W., 1999. Neogene
tectonic evolution and exhumation of the southern Ecuadorian Andes: a
combined stratigraphy and fission-track approach. Tectonophysics 307, 255–
276.
Stevens, P.F., 2013. Angiosperm Phylogeny. <http://www.mobot.org/mobot/
research/apweb/> (accessed 07.09.13).
Sytsma, K.J., Morawetz, J., Pires, J.C., Nepokroeff, M., Conti, E., Zjhra, M., Hall, J.C.,
Chase, M.W., 2002. Urticalean rosids: circumscription, rosid ancestry, and
phylogenetics based on rbcL, trnLF, and ndhF sequences. Am. J. Bot. 89, 1531–
1546.
Till, W., 1992. Systematics and evolution of the tropical-subtropical Tillandsia
subgenus Diaphoranthema (Bromeliaceae). Selbyana 13, 88–94.
Tripati, A.K., Roberts, C.D., Eagle, R.A., 2009. Coupling of CO2 and ice sheet stability
over major climatic transitions of the last 20 million years. Science 326, 1394–
1397.
Tschapka, M., von Helversen, O., 2007. Phenology, nectar production and visitation
behaviour of bats on the flowers of the bromeliad Werauhia gladioliflora in a
Costa Rican lowland rain forest. J. Trop. Ecol. 23, 385–395.
Valente, L.M., Savolainen, V., Vargas, P., 2010. Unparalleled rates of species
diversification in Europe. Proc. R. Soc. B 277, 1489–1496.
van der Hammen, T., 1995. Global change, biodiversity, and conservation of
Neotropical montane forests. In: Churchill, S.P., Balslev, H., Forero, E., Luteyn, J.L.
(Eds.), Biodiversity and Conservation of Neotropical Montane Forests. New York
Botanical Garden, Bronx, New York, pp. 603–607.
Vasconcelos, P.M., Becker, T.A., Renne, P.R., Brimhall, G.H., 1992. Age and duration of
weathering by 40K–40Ar and 40Ar/39Ar analysis of potassium-manganese oxides.
Science 258, 451–455.
Versieux, L.M., Wendt, T., 2007. Bromeliaceae diversity and conservation in Minas
Gerais state. Brazil. Biodiv. Conserv. 16, 2989–3009.
Vincent, E., Berger, W.H., 1985. Carbon dioxide and polar cooling in the Miocene:
the Monterey hypothesis. Geophys. Monogr. 32, 455–468.
Vitousek, P.M., Chadwick, O.A., Matson, P.A., Allison, S., Derry, L.A., Kettley, L., Luers,
A., Mecking, E., Monastra, V., Porder, S., 2003. Erosion and the rejuvenation of
78
T.J. Givnish et al. / Molecular Phylogenetics and Evolution 71 (2014) 55–78
weathering-derived nutrient supply in an old tropical landscape. Ecosystems 6,
762–772.
Vogel, S., 1954. Blütenbiologische Typen als Elemente der Sippengliederung.
Botanische Studien 1. G. Fischer, Jena.
Vogel, S., 1969. Chiropterophilie in der neotropischen Flora. Neue Mitteil. III. Flora
158, 289–323.
Wanek, W., Zotz, G., 2011. Are vascular epiphytes nitrogen or phosphorus limited?
A study of plant 15N fractionation and foliar N:P stoichiometry with the tank
bromeliad Vriesea sanguinolenta. New Phytol. 192, 462–470.
Waser, N.M., Chittka, L., Pirce, M.V., Williams, N.M., Ollerton, J., 1996. Generaliza
tion in pollination systems, and why it matters. Ecology 77, 1043–1060.
Webb, C.O., 2000. Exploring the phylogenetic structure of ecological communities:
an example for rain forest trees. Am. Nat. 156, 145–155.
Wendt, T., Coser, T.S., Mattalana, G., Guilherme, F.A.G., 2008. An apparent lack of
prezygotic reproductive isolation among 42 sympatric species of Bromeliaceae
in southern Brazil. Plant Syst. Evol. 275, 31–41.
Wiens, J.J., Ackerly, D.D., Allen, A.P., Anacker, B.L., Buckley, L.B., Cornell, H.V.,
Damschen, E.I., Davies, T.J., Grytnes, J.A., Harrison, S.P., Hawkins, B.A., Holt, R.D.,
McCain, C.M., Stephens, P.R., 2010. Phylogenetic niche conservatism as an
emerging principle in ecology and conservation biology. Ecol. Lett. 13, 1310–
1324.
Winter, K., Smith, J.A.C., 1996a. An introduction to crassulacean acid metabolism:
biochemical principles and ecological diversity. In: Winter, K., Smith, J.A.C.
(Eds.), Crassulacean Acid Metabolism – Biochemistry, Ecophysiology and
Evolution. Springer Verlag, Berlin, pp. 1–13.
Winter, K., Smith, J.A.C., 1996b. Crassulacean acid metabolism: current status and
perspectives. In: Winter, K., Smith, J.A.C. (Eds.), Crassulacean Acid Metabolism –
Biochemistry, Ecophysiology and Evolution. Springer Verlag, Berlin, pp. 389–
426.
View publication stats
Yanega, G.M., Rubega, M.A., 2004. Feeding mechanisms: hummingbird jaw bends to
aid insect capture. Nature 428, 615.
Young, K.R., Ulloa, C.U., Luteyn, J.L., Knapp, S., 2002. Plant evolution and endemism
in Andean South America: an introduction. Bot. Rev. 68, 4–21.
Yule, G.U., 1925. A mathematical theory of evolution, based on the conclusions of
Dr. JC Willis, FRS. Phil. Trans. R. Soc. B 213, 21–87.
Zachos, J., Pagani, M., Sloan, L., Thomas, E., Billups, K., 2001. Trends, rhythms, and
aberrations in global climate 65 Ma to present. Science 292, 686–693.
Zizka, G., Schmidt, M., Schulte, K., Navoa, P., Pinto, R., König, K., 2009. Chilean
Bromeliaceae: diversity, distribution and evaluation of conservation status.
Biodiv. Conserv. 18, 2449–2471.
Zotz, G., 2004. How prevalent is crassulacean acid metabolism among vascular
epiphytes? Oecologia 138, 184–192.
Zotz, G., 2013. The systematic distribution of vascular epiphytes – a critical update.
Bot. J. Linn. Soc. 171, 453–481.
Zotz, G., Wilhelm, K., Becker, A., 2011. Heteroblasty – a review. Bot. Rev. 77, 109–
151.
Zotz, G., Asshoff, R., 2010. Growth in epiphytic bromeliads: response to the relative
supply of phosphorus and nitrogen. Plant Biol. 12, 108–113.
Zotz, G., Richter, A., 2006. Changes in carbohydrate and nutrient contents
throughout a reproductive cycle indicate that phosphorus is a limiting
nutrient in the epiphytic bromeliad, Werauhia sanguinolenta. Ann. Bot. 97,
745–754.
Zotz, G., Ziegler, H., 1997. The occurrence of crassulacean acid metabolism among
vascular epiphytes from Central Panama. New Phytol. 137, 223–229.
Zotz, G., Bogush, W., Hietz, P., Ketteler, N., 2010. Epiphytic bromeliads in a changing
world: the effects of CO2, water and nutrient supply. Acta Oecol. 36, 659–665.
Ricarda Riina
Darren Crayn