Predators and avian community organization: an experiment in a semi-desert grassland.

We provide experimental evidence that predators are a major factor organizing a community of granivorous grassland birds (mostly emberizid finches). Our focus is not on the lethal effects of predators, but on the simple idea that (i) birds will not settle where they perceive a high risk of predation, and (ii) species differ in their perception of the safety of woody vegetative cover due to differences in antipredator escape behavior. Consistent with this idea is the fact that the composition of this grassland community responds markedly to minor manipulations in the distribution of woody cover. In particular, with the addition of cover to open grasslands, species with cover-dependent escape tactics increase in abundance, while the abundance of "cover-independent" species decreases greatly; this decrease may reflect aggression from cover-dependent species, but evidence suggests that some cover-independent species may actively avoid cover-rich areas per se. Non-predatory effects of cover, most notably those concerning food resources and microclimate, appear unable to explain these results. Predators may influence many communities of terrestrial vertebrates via species-specific responses to cover.