Phylogeny structures species' interactions in experimental ecological communities*

The advent of molecular phylogenetics provided a new perspective on the structure and function of ecological communities. In particular, the hypothesis that traits responsible for species' interactions are largely determined by shared evolutionary history has suggested the possibility of connecting the phylogeny of ecological communities to their functioning. However, statistical tests of this link have yielded mixed results. Here we propose a novel framework to test whether phylogeny influences the patterns of coexistence and abundance of species assemblages, and apply it to analyze data from large biodiversity-ecosystem functioning experiments. In our approach, phylogenetic trees are used to parameterize species' interactions, which in turn determine the abundance of species in a specified assemblage. We use a maximum likelihood-based approach to score models parameterized with a given phylogenetic tree. To test whether evolutionary history structures interactions, we fit and score ensembles of randomized trees, allowing us to determine if phylogenetic information helps to predict species' abundances. Moreover, we can determine the contribution of each branch of the tree to the likelihood, revealing particular clades in which interaction strengths are closely tied to phylogeny. We find strong evidence of phylogenetic signal across a range of published experiments and a variety of models. The flexibility of our framework permits incorporation of ecological information beyond phylogeny, such as functional groups or traits, and provides a principled way to test hypotheses about which factors shape the structure and function of ecological communities.