Elevational biodiversity gradients in the Neotropics: Perspectives from freshwater caddisflies (Insecta: Trichoptera)

higher than 0.6 were included, and collinear environmental variables were discarded before correlations). All these analyses were performed using PRIMER 6 [31]. Geographic distance and β-diversity. We also analyzed the importance of geographic distance in explaining community composition similarity between localities. For this, species and genera abundance matrices for each locality was built based on the sampling carried out at each locations (see sampling procedure). Then, we calculated a Bray-Curtis similarity matrix for species and genera between localities, using the same square-root transformed values described above, which were correlated with a geographic distance matrix using Spearman’s correlation coefficients in the RELATE procedure in PRIMER 6. We evaluated the significance of these correlations using a permutation procedure (9999 permutations). We chose this analy- sis over a classical Mantel test; in spatial analysis, the assumptions of linearity and homoscedas- ticity of the Mantel test do not hold in most cases [34, 35]. Further, the partial Mantel test interpoint distance values are treated independently as a single vector. Instead, the RELATE procedure works on the distance/dissimilarity matrix directly, and therefore it retains the spa- tial structure from the response data [34]. In addition, Trichoptera community turnover (i.e., β-diversity) among the location along the elevation gradient was quantified for both species and genera, using the Whittaker index [36] as follows: bw ¼ S= / 1

where, S is the total number of species recorded at both localities (S = a + b + c); component a comprises the number of shared species between both localities, and components b and c com- prise the number of species present exclusively in each of the two localities; α is the mean num- ber of species found within localities, expressed as (2a+b+c)/2. The index was calculated in R using the vegan package [37]. Null models & Beta diversity. To separate the contribution of local community assembly mechanisms from sampling effects owing to variation in species pools, we compared observed Whitaker β-diversity to values expected under a null model across the elevation gradient, and across the 3 elevation groups found in the NMDS and Cluster analysis [38–40]. Our null model accounts for regional sampling effects due to the size and structure of species pools. We used the algorithm c0_both implemented in the commsim function of the R vegan package, version 2.5–7 [37]. The c0_both model preserves the local richness and the species present at each locality, but individuals among species at each location are shuffled. Further, as the null model maintains the observed distributional patterns of species and their abundances, the observed dispersal limitation of species is conserved in the null model. We implemented 1,000 iterations of the null model, producing a frequency distribution of null β-diversity values for each locality across the elevation gradient. Based on this frequency distribution, we calculated

a β-deviation or the standardized effect size (SES), using the mean and standard deviation of the null distribution as follows: SES ¼ Xobs Xnull s: d: Xnull ð Þ Xobs where Xobs is the observed dissimilarity value (i.e., Whitaker) between two communities, Xnull the mean of the null distribution, and s.d. (Xnull) the standard deviation of the null distribu- tion. SES is a standardized measure of the difference between observed and null β-diversity. It can be interpreted as the relative effect of local assembly mechanisms on β-diversity (e.g., envi- ronmental filtering) after removing the sampling effects from the observed species pools. Val- ues greater than 1.96 (P < 0.05) indicate a higher-than-expected beta dissimilarity between the communities (i.e., non-random) and values below -1.96 indicate a lower than expected PLOS ONE Elevational biodiversity gradients of Neotropical caddisflies PLOS ONE | https://doi.org/10.1371/journal.pone.0272229 August 5, 2022 6 / 17