JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.. Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecology.Abstract. Herbivory by Atta cephalotes was measured in four plant communities of different complexity in Costa Rica. The four communities were a monoculture of cassava {Manihot esculenta) and three diverse assemblages, each 1.5 yr old: (1) successional vegetation, unmodified by the investigators; (2) imitation of succession, a community of investigator-introduced species designed to mimic the unmodified succession; and (3) enriched succession, a successional vegetation that the investigators had augmented by propagule inputs. Each ant colony had access to all four community types simultaneously. The average herbivory rate (daily cutting of leaf per square metre of ground, all treatments combined) was ~ 150 mg (or 38 cm2).In communities with greater leaf area index, structural complexity, and species richness, A. ceph? alotes cut lower portions of total leaf area. Before harvest of the cassava monoculture, the herbivory rate of A. cephalotes was 87.9 cm2m~2d_1 in the monoculture, 21.4 in the imitation, 14.7 in the succession, and 6.8 in the enriched succession community. These amounts represented 0.3% of total leaf area in the monoculture and a mean of 0.03% of total leaf area in the three complex ecosystems. Cassava, which occurred in three of the four communities, was attacked most heavily (per unit leaf area) in the imitation successional community, least heavily in the enriched succession, and at intermediate intensity in the monoculture. In response to loss of their preferred forage (cassava) through harvesting by humans, the ants cut more leaf tissue in the three species-rich communities, especially the imitation. As cassava resprouted in the monoculture, A. cephalotes' rate of attack on the three diverse treatments reurned to preharvest levels.Atta cephalotes cut only 17 of 332 available plant species. They cut proportionally more woody than herbaceous species, more introduced species than natural colonizers, and species with belowaverage water contents. Plant relative abundance alone did not determine host plant selection, but most of the attacked species were cut in proportion to their total leaf area.