Multivariate analyses of variance (MANOVA) were performed
to check for differences in the composition of species and in
the trait frequency/mean values among sampling sites. For
the MANOVAs, we used the same permutation design as in
the previously described ANOVA. Then, principal coordinates
analysis (PCoA) was performed in order to graphically
characterize the changes in species composition across the
areas. The species data were standardized by marginal rows
and column totals, and the similarity measure was based on
Euclidean distances. A principal components analysis (PCA)
was performed to characterize trait variation patterns across
the areas based on a correlation similarity matrix. All analyses
were performed in MULTIV 2.63 statistical software (available
at http://ecoqua.ecologia.ufrgs.br/ecoqua/MULTIV.html).