Environmental covariates
Bioclimatic covariates are successfully used in SMDs since their beginning (Booth et al., 2014), and perhaps reflect physiological constraints. Soil type and geology have been used in SDMs for Neotropical flying vertebrates (Ramoni-Perazzi et al., 2012, 2017, 2020) likely echoing deeper ecological and historical constraints. Soil characteristics modulate aspects of vegetation as functional traits, biodiversity, and speciation, among others (Hulshof & Spasojevic, 2020; Le Stradic et al., 2015; Mori et al., 2021; Nunes et al., 2015; Rajakaruna, 2018). Besides, plant species distributions can shape primary (Freeman & Mason, 2015) and secondary (Sanín & Anderson, 2018) consumers distributions. Furthermore, geology underpins several soil characteristics (Bockheim et al., 2014), as well as topography and geological events that can influence biological diversification processes over short timespans and over regional or local scales (Antonelli et al., 2018; Gillespie & Roderick, 2014; S. L. Pereira & Baker, 2004).
We used an ad hoc database of bioclimatic covariates for Brazilian mainland (Ramoni-Perazzi et al., in press), information on elevation (GMTED2010; Danielson and Gesch 2011), geological substrate (hereafter geology; Gómez Tapias et al. 2019), and soil type (Hengl et al. 2017). All variables were used at (or resampled to) 30 arc seconds resolution.
We performed all the analyses using R 3.6.3 (R Core Team, 2020). To remove collinearity, we reduced the number of continuous (bioclimatic + elevation) covariables through a principal component analysis using ‘RStoolbox’ (Leutner et al., 2018), keeping the first four components, whose eigenvalues were higher than one and explained 90.4% of the variance (Supplementary material A, Appendix 1, Fig. A). The first component (PC1) can be interpreted as a contrast between the temperature during the most extreme conditions and its seasonality/variability (Supplementary material A, Appendix 1, Fig. B1). Similarly, the second component (PC2) contrasts the precipitation during extreme conditions and its seasonality. The third component (PC3) involves the effect of water availability, since involves temperatures under extreme conditions/seasonality and precipitation during the warmest quarter. Finally, the fourth component (PC4) has a weak negative association between Mean Diurnal Temperature Range (Bio 02) and elevation.