Vive la difference: why, how, and what trait combinations improve functional community ecology

Abstract

Analyses of functional trait variation across ecological communities offer valuable insights into the factors that influence and predict trait composition, deepening our understanding of community structure. These analyses also shed light on patterns of trait dispersion, including underdispersion and overdispersion, which are often linked to the mechanisms underlying community assembly. However, striking a balance between characterizing complex phenotypes through the integration of comprehensive set of traits while ensuring the preservation of the distinct signals of overdispersion and underdispersion can pose significant challenges. This is because some trait combinations can result in high functional dispersion or overdispersion, while others may lead to underdispersion within the same community. In this study, we provide a rationale for the importance of trait selection in community ecology and develop a framework that optimizes patterns of functional trait variation among communities. We explored the responses of six trait sets along large-scale environmental gradients, using [~]700 lake-fish communities in Ontario (Canada). We started by adopting the conventional approaches of combining all traits as well as leveraging prior knowledge in fish biology to group traits that are functionally related (Diet, Morphology, Temperature Preference). We contrasted these approaches with a novel computational method to select combinations of traits that maximise overdispersion and minimise underdispersion. We found that separating traits according to their functions or by their patterns revealed signals in communities that appeared random when all the traits were pooled together. Combining our computational approach with environmental models revealed that functional patterns could be very well explained by multiple environmental predictors (model R2 up to 0.7). Our study also outlines future selection procedures to further uncover more complex patterns of species associations based on their traits.