Characteristics of bird species using forest and agricultural land covers in southern Costa Rica

Our understanding of why tropical forest species differ in their ability to inhabit agroecosystems is limited, despite the link between this ability and the likelihood of population decline for species inhabiting regions undergoing widespread conversion of forest to agriculture. We used logistic regression and data from southern Costa Rica to develop a model based on natural history characteristics to distinguish between forest species that did or did not use agricultural land covers. We sampled birds along 15 3.0-km routes, seven in forest and eight in agriculture, five times over three years. Each species was classified as an F-species, detected only in forest, a G-species, detected in both forest and agriculture, or an A-species, detected only in agriculture. Thirty percent of species were F-species, 42% were G-species, and 28% were A-species. Based on the logistic regression model, the likelihood of being a G-species, as opposed to an F-species, was low for species that were dependent on forest interior, had a stenophagous diet, and a small elevational range. Weight, resident versus migrant status, and whether or not a species was insectivorous, were not significant predictors of being a G-species. For all F–G species pairs, the model correctly predicted the G-species 70% of the time. The model provides a first step in identifying those characteristics that predispose forest species to use agricultural land. In addition, our results indicate that the structurally simple agricultural systems of the study region are of limited value for a large proportion of the regional species pool.