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

The material from each locality and date was initially sorted by “morphospecies” based on coloration, size, and habitus of the individuals. Additional characters such as the presence or absence of ocelli, the structure of maxillary palpi, the number of leg spurs, and the size and shape of antennae were used to further refine the sorting process [23]. Within each series, specimens were sorted by sex, and the abdomen of a few males per series were removed and cleared using the lactic acid method proposed by Blahnik et al. [24]. These male genitalia were then compared to the original species descriptions to identify them to the most precise taxo- nomic category possible. If identification to species level was not possible, taxa were consid- ered as morphospecies within each genus [23]. To assess the influence of habitat variables, we constructed an environmental matrix used for the non-metric multidimensional scaling analysis. Relative humidity (RH), Mean (Mean T˚) and minimum average annual air temperature (Min T˚) were derived from HOBO Pro Temp/RH data loggers (U23-001; Onset, USA) in one-hour intervals at 1 m above ground level collected at each location between 2016–2018 [22]. Average annual precipitation (here- after “precipitation”) was obtained from the Climatologies at high resolution for the Earth’s land surface areas (CHELSA) dataset at 30 Arc-sec resolution (~1km) [25]. Elevation, slope, and drainage area of each location were derived from the Shuttle Radar Topography Mission

(SRTM) at 30 m pixel resolution, and land cover of each location was derived from pre-exist- ing vegetation maps for the year 2016 derived from Landsat images at 30 m pixel resolution [26] Data analysis Species diversity. We estimated species richness and species diversity (α-diversity) based on the Shannon-Weiner exponential (N1, Hill number 1) [28]. To assess sampling complete- ness (i.e., the number of species collected per sampling location), we used species and genera accumulation plots with the Chao 1 estimator and sample completeness estimator to assess the completeness of the sampling at each locality (elevation), which allows better comparison among assemblages [27–29]. This calculations were obtained using iNEXT online software [30]. To complement these sampling completeness calculations, we performed species accu- mulations curves with the observed species and Chao 1 estimator at the three samples at each site permuted, using PRIMER 6 software [31]. Community composition analysis. We analyzed patterns of community composition (species and genus abundance patterns) along the altitudinal gradient, for which we con- structed an abundance matrix for each Trichoptera species and locality. We square-root trans- formed taxa abundances to increase the weight of low-abundance taxa in the analysis (68% of the species were represented by 2 or fewer individuals). We then constructed a between-local- ity Bray-Curtis similarity matrix as a similarity measure and performed a non-metric multidi-

mensional scaling analysis (NMDS). Further, we performed a hierarchical agglomerative cluster analysis using a group average linkage procedure [31], and we assessed the significance of these groups using the Similarity Profile routine, SIMPROF, that tests data against the null hypothesis of absence of structure [32]. With the formed groups, according to the SIMPROF routine, we performed a SIMPER procedure to assess the dissimilarity among these groups and the genera and species responsible for the differences among groups [33]. These results were corroborated, correlating the ordination matrix with the genera/species abundance matrix (only Spearman correlations higher than 0.6 were included). Lastly, we correlated the ordination matrix with environmental variables using Spearman correlations to address which variables were responsible for faunal groups (only correlations PLOS ONE Elevational biodiversity gradients of Neotropical caddisflies PLOS ONE | https://doi.org/10.1371/journal.pone.0272229 August 5, 2022 5 / 17