Subspecies niche analysis
We used ‘phyloclim’ (Heibl & Calenge, 2018) to quantify the niche overlap measurements for each pair of conspecific subspecies based on predictions (Warren et al., 2008): D (Schoener, 1968) andI (van der Vaart, 1998). Both report whether the average agreement of environmental requirements calculated between subspecies pairs is greater or smaller than expected if such subspecies were independent of each other, ranging from zero (no overlap) to one (identical niches). Then, we assessed the null hypothesis of niche identity, evaluating the equivalency between pairs of predictions, comparing the respective observed values of D and Iagainst those generated through 100 pseudoreplicates, assigning the occurrence points for both subspecies to one lineage or the other at random to simulate the potential overlap of a group of points occurring across a given geographic space (Warren et al. 2008).
Moreover, we used background randomization tests to contrast the observed niche overlap values against a null distribution of 100 overlap values resulting from contrasting the predictions of one subspecies against those created from points taken at random from the geographic space of the another subspecies (Warren et al., 2008). We considered that: (1) there was niche conservation when the backgrounds overlapped and niche distance was not significantly different from zero; (2) there was niche divergence when the distance between both subspecies was significantly different from zero, independently of backgrounds’ overlap.