Tropical montane landscapes harbor diverse flora and fauna, and many species there are ecological specialists with narrow elevational distributions, limited geographic ranges, and small global populations. Along elevational gradients, environmental conditions and community composition change dramatically over small spatial scales. As forests are disturbed and edges formed with modified habitat, natural communities could be affected differently across elevations by the many physical and biotic changes at edges. We asked whether forest edges produced altered patterns of avian species composition along a cloud forest - dry forest gradient on the Pacific slope of the Tilarán mountains in Monteverde, Costa Rica. A strong moisture gradient produces cloud forests near the ridgetops, with a concentration of species endemic to the Costa Rica – Panama highlands that are habitat specialists. We conducted 552 point counts across 110 locations from 1100 to 1800 m elevation, yielding 6586 detections of 115 species in 10 km² of montane forest. We analyzed differences in species composition and single-species abundances between interior and near-edge forest habitats for species grouped by geographic range size. Species composition changed markedly from forest edge to interior in cloud forest habitats, but not in drier forests downslope. Endemic species, especially in cloud forest, were detected less frequently in mature forest near edges than in mature forest interior, and this difference was more pronounced than for cosmopolitan species. On tropical mountainsides, we can expect habitat-specialist endemic species to be more sensitive to further habitat modification. This sensitivity could limit the resilience of tropical bird communities.

These data come from audio-visual surveys (point counts) conducted in May-July (the avian breeding season) of 2006 from 1100-1800 m elevation, at 110 points, including 77 points located in forest interior and 33 points near forest edges. Points were separated by >200 m to avoid double counting individuals across points, and were distributed across a moisture gradient defined by the distance of points to the continental divide. Edge points were placed inside the forest 50 m from a boundary with a clearing to keep the sampled area within forest. Interior points were placed >200 m from a boundary. All birds seen or heard in 10 min within a 50 m radius by two observers were recorded with the estimated distance of detection. Counts were repeated at each point weekly, a total of five times over the season (see published article for further details). We used the maximum number of simultaneously detected conspecifics per point across all visits for each species’ abundance value at each point, and created a site by species matrix using these values.

The dataset provided shows the site by species matrix, with sites (point labels="SITE") as rows, and associated point data (distance of point from the continental divide in meters = "DTD"; elevation of point in m.a.s.l. = "ELEV"; categorical zone of distance from the continental divide = "DTD_ZONE"; categorical variable labeling each site as a forest dge or interior point = "INT-EDG"; zone-habitat as a merged column of DTD_ZONE and INT-EDG = "ZONE_HAB") and the remaining columns as species (shown by each species' letter code). Values in the data matrix show the maximum number of simultaneously detected individuals of a species at a point over the five visits to the site. A table of letter codes of species corresponding to the columns of this data matrix can be found in Table A2 of published Supplementary Material.