Although we know that competition sometimes controls the composition of plant communities, we still do not understand its significance in communities having high species richness. We removed an abundant and apparently dominant grass (Andropogon virginicus) in species-rich pine grassland in southeastern Louisiana and evaluated the effects on species richness and composition. At 2 sites, moist and dry, we located twenty 1- x 1-m plots with 10 randomly assigned control plots and 10 treatment plots, in which herbicide was applied to individual shoots of A. virginicus over 2 y. Plant cover, species richness, and species composition were recorded 4 times over this period. Repeated measures analyses of variance and Mantel tests were used to evaluate differences between control and removal plots. Although there were more than 90 species of vascular plants that might have responded to the removal of A. virginicus, no significant effect on cover, species richness, species composition, or functional group composition was found. Competition apparently played a minor role in determining the composition of this subtropical grassland. The general model of competition in temperate grasslands, which assumes a few species of grasses dominate the community through competition and other species survive in the interstices, does not seem to apply. Infertile soils may reduce rates of competitive exclusion and establishment, minimize interactions between grasses and forbs, or produce a fundamentally different kind of competition that is inherently slower and more symmetrical than in most experimental situations. © Ecoscience.