||In a 4-yr field experiment, we tested the hypotheses that insectivorous birds (1) controlled densities of herbivorous grasshoppers in an ungrazed semiarid grassland in southeastern Arizona, and (2) functioned as keystone predators, by limiting abundances of grasshoppers that otherwise might change vegetation cover and species composition, and/or by mediating the effects of otherwise competitively superior members of the grasshopper
assemblage. We measured grasshopper densities and vegetation on 32 464-m2 grassland plots for 1 yr, then enclosed 16 of these plots with bird exclosures and continued data collection for 3 yr. Eight of the 16 experimental plots were further modified in the last 2 yr of the study by installing fine-mesh 1 m high barriers designed to retard grasshopper dispersal. Microclimates of caged plots differed only slightly from open plots. Lizards and rodents increased inside the exclosures, but they were removed and released elsewhere such that their average abundances did not differ among treatments.
By the final year of the study, mean annual adult grasshopper density was > 2.2 times higher on plots from which birds were excluded, and where grasshoppers were enclosed by dispersal barriers, than on unmanipulated control plots. Mean nymph density was > 3.0 times higher in the same comparison. Grasshoppers were significantly more abundant in bird enclosures with insect dispersal barriers, indicating that experimental plots were dispersal sources rather than sinks. Seven of 12 common grasshopper species were more abundant inside the bird exclosures, while none was less abundant. Among the more abundant taxa, those responding most positively were grass feeders: Eritettix simplex, Opeia obsciira, Paropomala Tyomingensis, and Phoetaliotes nebrascensis. We found no evidence that grasshoppers competed with one another under increased densities inside the bird exclosures. Although the amount of insect herbivory was somewhat
higher inside the bird exclosures, and was positively correlated with grasshopper density across all 32 plots (r = 0.87), overall vegetation cover and species composition did not differ among treatments by the end of the study. Dactylotum variegatut, an aposematic species apparently immune to avian predation, showed no significant responses to the experiment. Birds clearly limited grasshoppers in this grassland ecosystem, but they failed to qualify
as keystone predators, at least in the short term, for two reasons: (1) in their absence, increased grasshopper densities had no appreciable impact on vegetation cover or species composition; and (2) there was no evidence that birds mediated competition among grasshoppers.