LAUSR

Interactions between plants and the soil are an important ecological process in terrestrial ecosystems as they affect plant community structure-when and where we find different plant species. Those interactions are typically thought of as one-directional: local soil conditions filter through dispersing species to produce a community of locally adapted plants. However, plants can modify local physicochemical soil conditions via their roots and associations with soil microbes. These may in turn affect the local fitness of other plants, making plant-soil interactions bidirectional. In order to understand how they differ from other ecological processes that structure plant communities, we need a theory connecting these individual-level plant-soil feedbacks to community-level patterns. Here, we build this theory with a mathematical model of plant community dynamics where soil conditioning is explicitly modelled over time and depends on the density of the plants. We analyze this model to describe the long-term composition and spatial distribution of the plant community. Our main result is that positive plant-soil feedbacks will create clustering of species with similar soil preferences. The composition of these clusters is further influenced by niche width and conditioning strength. In contrast with competitive dynamics driven by niche overlap, only species belonging to the same cluster can maintain high relative abundance in the community. Spatial heterogeneity in the form of an environmental gradient generates patches, each representing a single cluster. However, such patchiness is disfavored when species differ in dispersal ability. We show that stronger dispersers cannot take over the habitat as long as an exogenous driver favors soil conditions that benefit the other species. If exogenous drivers supersede soil conditioning by plants, we retrieve classic habitat filtering, where species are selected based on their suitability to the local environment. Overall, we provide a novel mathematical model for positive plant-soil feedback, which we use to describe the spatial patterns of plant abundance and traits related to soil preference and conditioning ability.