PAPER On the measurement of dimensionality of biodiversity

Aim To characterize relationships among indices of taxonomic, phylogenetic, functional and phenetic diversity (TPFP diversity) for Neotropical bats and examine if dimensionality is different from null model expectations after control-ling for sampling effects owing to underlying variation in species richness. Location Neotropics. Methods Bat distributions were estimated using geographic range maps. By over-laying range maps on a gridded map of the New World, the species composition of each cell was determined. Then, species richness and three indices each of phylogenetic, functional and phenetic diversity were calculated. A principal com-ponents analysis (PCA) determined the dimensionality of the entire multivariate data set. This was followed by two additional PCAs that examined the dimension-ality of spatial and non-spatial fractions of the original data. Spearman rank cor-relations determined pair-wise association among indices. Correlations and dimensionality were compared with two different null models that account for species richness gradients. Results Ten measures characterizing TPFP diversity exhibited much spatial struc-ture across the New World. Pair-wise correlations between indices were typically different from null model expectations. While patterns of multivariate covariation were similar across spatial and non-spatial data sets, the dimensionality of biodi-versity was low and either lower than or no different from null model expectations. Main conclusions Most measures of biodiversity exhibit at least some level of quantitative redundancy, and this redundancy is often higher than expected given sampling effects owing to species richness gradients. Considerations of uniqueness should be made when examining conceptually different dimensions of biodiversity.