Fig 1.
Summary of the importance of environmental variables in the development of suitability models for Agave dryland-farming (a, b) and Agave species (c, d, e) relative to regularized training gain. Green bars indicate model gain when only including individual environmental variables. Light-grey bars show the gain when the individual environmental variables are excluded from the full model. Dark-grey bars indicate the gain achieved in the full model, including all environmental variables in the model. Abbreviated environmental variables are defined as follows: Bio 2 = mean diurnal range (mean of monthly (maximum temperature—minimum temperature)), Bio 7 = temperature annual range (maximum temperature–minimum temperature), Bio 8 = mean temperature of wettest quarter, Bio 15 = precipitation seasonality, Bio 16 = precipitation of wettest quarter, and Bio 19 = precipitation of coldest quarter.
Table 1.
Sites and numbers of samples evaluated in our study of rock piles (top portion of table) and grasslands cultivated with Agave angustifolia (bottom portion of table) using MaxEnt.
Table 2.
Environmental variables derived from monthly temperature and rainfall values.
The variables were selected from the WorldClim 2 database to generate dryland-farming MaxEnt models. All the environmental variables analyzed for the study can be found in (S4 Fig).
Table 3.
Suitability and overlapping statistical assessment using permutation tests.
Fig 2.
Suitability models for rock piles in the Sonoran Desert region (a), Arizona, USA (b), and for Agave grasslands in Sonora, Mexico (c). Coloration in red indicates high suitability potential for rock piles in (a) and (b). Coloration in red in (c) indicates high suitability for grasslands.
Fig 3.
Suitability models of Agave palmeri (a), Agave parryi (b), and Agave angustifolia (c) in the borderlands of Arizona, USA and Sonora, Mexico. Red color indicates high suitability for Agave species.
Table 4.
Summary of permutation importance, via MaxEnt analysis, for each individual variable in the dryland-farming and Agave-species models.
Fig 4.
Habitat suitability for rock piles, grasslands, and Agave species (A. angustifolia, A. palmeri, and A. parryi) during the wettest season in a given year in terms of three environmental variables: (a) elevation, (b) precipitation seasonality, and (c) annual precipitation.
Table 5.
Summary of niche models of individual species for dryland-farming and Agave species with test gains of all the variables used to build the distribution models with only one variable.