Arid and semi-arid ecosystems often exhibit diverse plant growth forms in water-limited environments, but it is unclear whether resource competition (interference) is actually important in structuring communities. We chose a diverse Chihuahuan desert shrubland to examine the response of the plant community to experimental removals of selected perennial plant species or groups of species. Four treatments involved the removal of all individuals of all species of a single functional group (functional group removals: shrub removal, succulent removal, subshrub removal, perennial grass removal). Three other treatments involved removing species within functional groups. These seven treatments plus a control (no plants removed) were replicated six times each in 25x25 m experimental plots, in summer 1995. Permanent belt transects were surveyed for number and sizes of all vascular plants in spring and fall in 1997, 1999, 2000, and 2001. Those plots from which the dominant shrub, Larrea tridentata, was removed had not recovered in total plant cover or volume by 2001, but cover and volume in all other treatments were similar to those in control plots. Relatively few species demonstrated a positive response to the removal of other species or functional groups. The perennial grass group and forbs were the most responsive; perennial grass cover increased in the shrub removal treatment relative to the control but treatment differences diminished after dry growing seasons in 2000 and 2001. Results over the first five years suggest that either environmental conditions or intrinsic biological characteristics limit the ability of most plant species to respond to the removal of substantial fractions of community biomass and composition in the short term. Such slow response by both dominant and less abundant components of the community has implications for the recovery of semi-arid systems after human disturbance or other events leading to the reduction of biological diversity.