Most explanations for the positive effect of plant species diversity on productivity have focused on the efficiency of resource use, implicitly assuming that resource supply is constant. To test this assumption, we grew seedlings of Echinacea purpurea in soil collected beneath 10-year-old, experimental plant communities containing one, two, four, eight, or 16 native grassland species. The results of this greenhouse bioassay challenge the assumption of constant resource supply; we found that bioassay seedlings grown in soil collected from experimental communities containing 16 plant species produced 70% more biomass than seedlings grown in soil collected beneath monocultures. This increase was likely attributable to greater soil N availability, which had increased in higher diversity communities over the 10-year-duration of the experiment. In a distinction akin to the selection/complementarity partition commonly made in studies of diversity and productivity, we further determined whether the additive effects of functional groups or the interactive effects of functional groups explained the increase in fertility with diversity. The increase in bioassay seedling biomass with diversity was largely explained by a concomitant increase in N-fixer, C4 grass, forb, and C3 grass biomass with diversity, suggesting that the additive effects of these four functional groups at higher diversity contributed to enhance N availability and retention. Nevertheless, diversity still explained a significant amount of the residual variation in bioassay seedling biomass after functional group biomass was included in a multiple regression, suggesting that interactions also increased fertility in diverse communities. Our results suggest a mechanism, the fertility effect, by which increased plant species diversity may increase community productivity over time by increasing the supply of nutrients via both greater inputs and greater retention.