Statistical analysis
To detect significant differences in single metrics between the two sampling campaigns we applied the Wilcoxon Signed Test for paired samples. To highlight possible correlations between biological metrics (such as taxa richness, biodiversity etc.) and indices derived from RHS (i.e., HQA, HMS, and LRD) we applied the Spearman test. For community analyses, we ran an NMDS on taxa composition and performed the PERMANOVA test to detect possible significant differences between years. To test biotic homogenization in both taxonomic diatom and macroinvertebrate communities, we performed the test of homogeneity for multivariate dispersion (Anderson, 2006). These analyses were based on the Bray-Curtis dissimilarity metric for diatoms, while the Jaccard metric for macroinvertebrates.
We followed the approach of Legendre & De Cáceres (2013) to estimate the proportion by which each sampling site and each taxon contributed to the overall area diversity. This approach calculates: 1) the local contribution to ß-diversity (LCBD). LCBD is a measure of the ecological uniqueness of the single sampling site: the higher the site uniqueness, the higher its contribution to the total richness; 2) the species contribution to ß-diversity (SCBD), which is a score calculated for each species (diatoms) or taxon (macroinvertebrates). The higher is SCBD for a single species, the higher its contribution to the variance among sites. We calculated LCBD and SCBD Hellinger-transformed abundance data with the “beta.div” function of the values using adespatial R package (Legendre & De Cáceres, 2013). All the statistical analyses were performed in R (R Core Team, 2020).
and has been used as a tool to identify sites to prioritize in conservation efforts (Ruhı´, Datry, & Sabo, 2017).