Soil pathogens and spatial patterns of seedling mortality in a temperate tree

The Janzen–Connell hypothesis1,2 proposes that host-specific, distance- and/or density-dependent predators and herbivores maintain high tree diversity in tropical forests. Negative feedback between plant and soil communities could be a more effective mechanism promoting species coexistence because soil pathogens can increase rapidly in the presence of their host3, causing conditions unfavourable for local conspecific recruitment4,5,6. Here we show that a soil pathogen leads to patterns of seedling mortality in a temperate tree (Prunus serotina) as predicted by the Janzen–Connell hypothesis. In the field, the mean distance to parent of seedling cohorts shifted away from maternal trees over a period of 3 years. Seedlings were grown in soil collected 0–5 m or 25–30 m from Prunus trees. Sterilization of soil collected beneath trees improved seedling survival relative to unsterilized soil, whereas sterilization of distant soil did not affect survival. Pythium spp., isolated from roots of dying seedlings and used to inoculate healthy seedlings, decreased survival by 65% relative to controls. Our results provide the most complete evidence that native pathogens influence tree distributions, as predicted by the Janzen–Connell hypothesis, and suggest that similar ecological mechanisms operate in tropical and temperate forests.