Species richness and species turnover

We estimated the species richness per degree latitude based on two methods, one which controls for differences in sample size and hence also controls the potential bias in our richness estimates resulting from unequal sampling effort across the latitudinal extent of the WG, while the other was independent of our sampling effort and an estimate of the total regional species pool. First, we estimated interpolated species richness (S_R) based on individual based rarefaction curves, which controls for the differences in the number of individuals sampled across the gradient. We carried out this analysis for every one degree latitudinal zone within the Western Ghats by aggregating all samples within a latitudinal zone and estimated the average number of species per 450 individuals, the lowest sample size in our dataset amongst the 11 latitudinal zones. We also extrapolated the rarefaction curves to estimate species richness [17] up to4000 individuals which was approximately the number of individuals in the most well sampled latitudinal zone. Additionally, we also used two non-parametric richness estimators, Chao1 and ACE (Abundance based coverage estimator) to estimate the lower bound of undetected species richness. Second, we also estimated the total regional species pool (S_T) for each one degree latitudinal belt based on the count of all the species ranges that pass through it. Assuming range cohesion, species latitudinal ranges were estimated based on species occurrences (Fig 2) from our primary data as well as based on species occurrences and records from secondary literature such as floras and other published inventories (S2 Appendix). This method is independent of the sampling effort since (i) species were assumed to be present as long as their latitudinal ranges fell within a latitudinal belt irrespective of whether those species were recorded in our sampling plots and (ii) the latitudinal ranges were estimated considering species records from multiple sources.

In addition to this, we also estimated species richness at smaller spatial scales i.e. at the level of a plot and at the level of a cluster. For every cluster, we calculated mean alpha and gamma diversity, where alpha diversity is the mean number of species observed in a single plot, and gamma diversity is the cumulative number of species recorded for all plots within a cluster. Based on the values of alpha and gamma we calculated beta diversity using Whittaker’s index (Beta_W = γ / α), where ‘γ’ is gamma diversity estimated at the level of a cluster while ‘α’ is mean species richness at the level of a plot within a cluster. We also estimated beta diversity using the Simpson’s index (Beta_SIM) which is purely a measure of species turnover or replacement and disregards contribution to beta diversity (species dissimilarity) resulting from differences in species richness between samples.