When Do Traits Tell More Than Species about a Metacommunity? A Synthesis across Ecosystems and Scales

AbstractLinking species traits with the variation in species assemblages across habitats has often proved useful for developing a more mechanistic understanding of species distributions in metacommunities. However, summarizing the rich tapestry of a species in all of its nuance with a few key ecological traits can also lead to an abstraction that provides less predictability than when using taxonomy alone. As a further complication, taxonomic and functional diversities can be inequitably compared, either by integrating taxonomic-level information into the calculation of how functional aspects of communities vary or by detecting spurious trait-environment relationships. To remedy this, we here synthesize analyses of 80 datasets on different taxa, ecosystems, and spatial scales that include information on abundance or presence/absence of species across sites with variable environmental conditions and the species' traits. By developing analyses that treat functional and taxonomic diversity equitably, we ask when functional diversity helps to explain metacommunity structure. We found that patterns of functional diversity explained metacommunity structure and response to environmental variation in only 25% of the datasets using a multitrait approach but up to 59% using a single-trait approach. Nevertheless, an average of only 19% (interquartile range = 0%-29%) of the traits showed a significant signal across environmental gradients. Species-level traits, as typically collected and analyzed through functional diversity patterns, often do not bring predictive advantages over what the taxonomic information already holds. While our assessment of a limited advantage of using traits to explain variation in species assemblages was largely true across ecosystems, traits played a more useful role in explaining variation when many traits were used and when trait constructs were more related to species' status, life history, and mobility. We propose future research directions to make trait-based approaches and data more helpful for inference in metacommunity ecology.