Our ability to identify theoretical mechanisms of plant species coexistence has outstripped our ability to test those mechanisms in the field. Here, I provide evidence that supports the Janzen-Connell Hypothesis (JCH) in a temperate grassland. The JCH predicts that the effect of host-specific seed predators should increase with the abundance of their host plant species. This prediction can be tested by experimentally removing seed predators across a gradient of host abundance. Removal is predicted to have a greater effect where a plant species is common than where it is rare. I applied fungicide and insecticide to seedheads of 6 species in a biodiversity experiment at Cedar Creek LTER. Target plant species tended to be relatively common in low-diversity plots and relatively rare in high-diversity plots. Fungicide and insecticide significantly affected the number of seeds produced per seedhead and the fraction of seeds that germinated. However, these results were idiosyncratic across species, and no general patterns emerged. To determine the integrative effects of predator removal and its interaction with abundance, I calculated the expected biomass of seedlings per square meter that would be produced by a given treatment by multiplying seedheads per square meter, seeds per seedhead, germination fraction, and biomass per seedling. As predicted by the JCH, the beneficial effects of fungicide were negligible when target species were rare but increased when they were common. There were no such interactive effects of insecticide. I suggest that the different scales of fungi and insect dispersal relative to the size of the experimental plots led to these disparate results.