Environmental seasonality regulates community evenness in neotropical bat communities

Supplementary Material 1. List of sites and their descriptors. We selected sites below 1000 m.a.s.l., that had a sampling effort greater than 10000 m2.h (sampling effort calculated following Straube and Bianconi 2002) and a sampling coverage > 0.8 according to Chao and Jost (2012). Site IDs and descriptors from the ATLANTIC BATS data base (Muylaert et al. 2017).Supplementary Material 2. Correlations between original values from the 19 WorldClim layers (Fick and Hijmans 2017) and the four axes selected after from the PCA. PCA 1 explained 47.1% of the variation across climatic variables and shows a gradient of temperature seasonality across our sampling sites in the Atlantic Forest. Supplementary Material 3. List of all path models for evenness indices J′, 1E and 2E . Most significant model in bold (lowest AIC and non-significant P value from a Ficher’s C test). Env = Environmental seasonality derived from 19 WorldClim layers; S = species richness; M = biomass. Numbers above the arrows are standardized path coefficients. The “null” path model is saturated with zero degrees of freedom, and no Fisher’s C score and P-value could be calculated. Therefore, the “null” model at the bottom of the table for comparison.Supplementary Material 4. Matrix of species occurrences across sites. Sites are ordered by environmental seasonality; sites with less seasonable climate have more rare species (left side of the matrix) compared to sites with more seasonable climate (right side of the matrix). Env.var = Environmental seasonality. ID = Site ID according to Muylaert et al. (2017). Environmental seasonality is the first axis from a PCA that accounted for 47.1% of the variation across climatic variables. Higher values indicate sites that experience greater temperature seasonality, lower mean annual temperature, and warmest and coldest quarters.