I evaluated the effects of selective logging upon a bird community in the Brazilian Atlantic forest. Two areas 500 m apart were selected for quantitative (point censuses) avian surveys: a non-fragmented primary forest (PF) and a selectively logged forest (LF). Six of eight quantitative measures of vegetation structure compared between PF and LF were reduced at LF. Indices of avian species richness and diversity were very similar between PF and LF sites; species composition, however, differed strongly. Although the composition of guilds remained largely the same, most guilds differed in membership between PF and LF in at least one species. Understory and terrestrial insectivores were the most sensitive ecological groups (with most species missing in LF), as also observed in the process of forest fragmentation in other parts of the Neotropics. Based on the results of this study, I recommend the following procedures to minimize adverse effects of selective logging on bird communities in the Atlantic forest: (1) logged areas should be close enough to unfragmented, unlogged forests to allow recolonization of some species, (2) the exploitation of the forest should be carded out using as few roads and as little mechanized equipment as possible, and (3) long-term rotation should be used in areas designed for logging to allow time for forest regeneration.
