PERFORMANCE OF FUNCTIONAL DIVERSITY METRICS APPLIED AS MEASURES OF DISPARITY

Ecologists have made much progress examining the functional diversity of communities, an approach that places primary emphasis on the traits of organisms. These approaches parallel those by paleobiologists examining morphological and ecological disparity. Because of their limited interaction, the metrics used in these disciplines have been developed largely independently. Here I evaluate the performance of metrics used in both disciplines and offer recommendations for future analyses.

Metrics adopted from functional ecology include functional richness (FRic), evenness (FEve), divergence (FDiv), dispersion (FDis), Rao's quadratic entropy (Q), and functional group richness (FGR). Most of these distance-based metrics are calculated from the distribution of taxa in a multivariate ordination defined by taxon functional traits. PCoA based on Gower's distance typically is used because of its robust flexibility in handling different data types, missing trait information, and its ability to incorporate abundance weighting in analyses. FRic measures convex hull volume, FEve measures regularity of spacing along the minimum spanning tree, FDiv measures deviation of species from community centroid, FDis is similar but incorporates information on species abundance, Q is also similar but uses abundance data to measure the average pairwise functional difference between taxa, and FGR measures functional richness using a cluster analysis of traits. To these metrics, I add four common disparity metrics: unique trait richness, mean Euclidean distance, maximum range, and total variance.

Metrics were evaluated by their sensitivity to sample size, correlation with competing metrics, and performance against four null models of trait-space assembly. The four models compared include a random model, a model of trait redundancy (partial trait overlap), partitioning (progressive trait differentiation), and expansion (progressive evolution of novel traits). Overall, functional ecology metrics share many benefits with traditional metrics of ecological and morphological disparity. However, they are recommended as important complements to traditional metrics because of their robustness, flexibility, cross-linkage with ecological theory, and ease of computation in the R statistical environment.