Abstract We compared observed bird species richness in the Brazilian Atlantic Forest with predicted richness derived from BirdLife International distribution data to evaluate the effectiveness of point-count surveys in representing local and regional species pools. Observed richness was obtained from seasonal point-count surveys conducted over two years in three forest types (dense ombrophilous, mixed ombrophilous, and seasonal semideciduous). Predicted species lists were derived by intersecting global species distribution polygons with the geographic coordinates of each sampling site and filtering species according to habitat affinity to reduce commission errors. For each site, we estimated species richness, sample completeness, and the proportion of observed to predicted species. A total of 270 species were recorded during point counts, whereas 366 species were predicted based on distribution data. Point counts detected 73.7% of the predicted richness. Sample completeness ranged from 0.75 to 1.0, indicating adequate local sampling effort. Nevertheless, the proportion of observed to predicted species was consistently lower (0.35–0.52) and showed limited variation among forest types. Our results suggest that point-count surveys capture a substantial portion of local bird communities. However, discrepancies between observed and predicted richness likely reflect limitations of coarse-scale distribution maps as well as the low detectability of rare and intermittently occurring species.
Rodrigues et al. (Tue,) studied this question.