Recent approaches to understanding community composition have built on the possibility that the phylogenetic relatedness of community members might serve as an unbiased and easily obtained surrogate for ecologically relevant trait data. Several observational studies find that communities contain more distantly related species than expected by random assembly, and this has been interpreted as evidence of competitive exclusion. However, experimental tests of the relationship between phylogenetic relatedness and competitive outcomes are rare. We designed a mesocosm experiment to test whether (a) phylogenetic diversity or (b) measured diversity in feeding traits better predicts total production, resource depletion, and the outcome of competition in communities of seagrass invertebrates. From a pool of nine species of herbivorous amphipods, we selected 18 different combinations of three species. These combinations varied independently in their phylogenetic relatedness and feeding trait diversity. After multiple generations, we assessed reduction of plant resources (seagrass, detritus, and macro- and microalgae), as well as total secondary production, population growth by each herbivore species, and invasion of new species.
Results/Conclusions
In mesocosms with a higher diversity of feeding traits, measured as proportional overlap in potential food sources, we found reduced dominance of prolific species, and greater abundance of both other community members and invaders. In contrast, the phylogenetic diversity of the species combination had no significant effect on community composition or population growth of the component species. We also found that diversity in a range of other traits, including body size and fecundity, had no impact on community outcomes. We assessed the conservation of feeding traits, separately examining the number of foods consumed, the identity of foods consumed, and the rate of consumption, and found that some dimensions of the feeding niche are poorly predicted by phylogeny. Rather than capturing the overall ecological similarity underlying competition between species, we conclude that phylogenetic relatedness may be a poor predictor of species interactions unless it happens to correlate with the particular trait or set of traits that drive those interactions.