RESUMEN
Productivity benefits from diversity can arise when compatible pathogen hosts are buffered by unrelated neighbors, diluting pathogen impacts. However, the generality of pathogen dilution has been controversial and rarely tested within biodiversity manipulations. Here, we test whether soil pathogen dilution generates diversity- productivity relationships using a field biodiversity-manipulation experiment, greenhouse assays, and feedback modeling. We find that the accumulation of specialist pathogens in monocultures decreases host plant yields and that pathogen dilution predicts plant productivity gains derived from diversity. Pathogen specialization predicts the strength of the negative feedback between plant species in greenhouse assays. These feedbacks significantly predict the overyielding measured in the field the following year. This relationship strengthens when accounting for the expected dilution of pathogens in mixtures. Using a feedback model, we corroborate that pathogen dilution drives overyielding. Combined empirical and theoretical evidence indicate that specialist pathogen dilution generates overyielding and suggests that the risk of losing productivity benefits from diversity may be highest where environmental change decouples plant-microbe interactions.