A five-year removal experiment in which plant functional group diversity was manipulated found strong limitation of ecosystem functioning caused by the differing abilities of remaining functional groups to recruit into space left unoccupied by the plants removed. We manipulated functional group diversity and composition by removing all possible combinations of zero, one, or two plant functional groups (forbs, C3 graminoids, and C4 graminoids), as well as randomly chosen biomass at levels corresponding to the functional group removals, from a prairie grassland community. Although random biomass removal treatments showed no significant effect of removing biomass in general on ecosystem functions measured (P > 0.05), the loss of particular functional groups led to significant differences in above- (P < 0.001) and belowground (P < 0.001) biomass, rooting-zone (P = 0.001) and leached (P = 0.01) nitrogen, nitrogen mineralization (P < 0.001), and community drought resistance (P = 0.002). Many of these differences stemmed from the marked difference in the ways remaining functional groups responded to the experimental removals. Strong recruitment limitation of C4 graminoids resulted in large areas of open ground, high nutrient leaching, and high community drought resistance in plots containing just this functional group. In contrast, rhizomatous C3 graminoids quickly colonized space and used soil resources made available by the removal of other groups, leading to lower soil nitrate in plots containing C3 graminoids. These effects of recruitment limitation on ecosystem functioning illustrate possible effects of diversity loss not captured by synthetic experiments in which diversity gradients are created by adding high densities of seeds to bare soil.