Predicting the structure of soil communities from plant community taxonomy, phylogeny, and traits

There are numerous ways in which plants can influence the composition of soil communities. However, it remains unclear whether information on plant community attributes, including taxonomic, phylogenetic, or trait-based composition, can be used to predict the structure of soil communities. We tested, in both monocultures and field-grown mixed temperate grassland communities, whether plant attributes predict soil communities including taxonomic groups from across the tree of life (fungi, bacteria, protists, and metazoa). The composition of all soil community groups was affected by plant species identity, both in monocultures and in mixed communities. Moreover, plant community composition predicted additional variation in soil community composition beyond what could be predicted from soil abiotic characteristics. In addition, analysis of the field aboveground plant community composition and the composition of plant roots suggests that plant community attributes are better predictors of soil communities than root distributions. However, neither plant phylogeny nor plant traits were strong predictors of soil communities in either experiment. Our results demonstrate that grassland plant species form specific associations with soil community members and that information on plant species distributions can improve predictions of soil community composition. These results indicate that specific associations between plant species and complex soil communities are key determinants of biodiversity patterns in grassland soils.