Biased detection of bird vocalizations affects comparisons of bird abundance among forested habitats

Community studies of birds often rely on abundance estimates that are obtained from counts of bird vocalizations, yet vocalizations are not equally detectable in all habitats. I broadcast vocalizations for nine bird species to evaluate biases in detection of bird vocalizations among four forested habitats (young, mature, and old aspen dominated forests, and white spruce dominated forests), and in relation to height of the broadcast, whether the broadcast occurred before or after leaf formation, and the frequency of the broadcast vocalization. Virtually all of the broadcast vocalizations were detected at 50 m from the speaker. However, at 100 m from the speaker, 27% of the broadcast vocalizations were not detected and detection was highest in white spruce forest, lowest in young aspen forest, and intermediate in mature and old aspen forests. Detection of broadcasts was negatively related to the minimum frequency of the vocalization, higher for broadcasts from the canopy than for broadcasts from the shrub layer, and higher for broadcasts before than after leaf formation. I reanalyzed abundance data that were obtained from a study involving point counts of wild birds in young and old aspen-dominated forest. Biases among habitats in the detection of vocalizations had moderate influence on the resulting measures of habitat preferences for birds. I suggest that if a detection distance of more than 50 m is used for bird censuses within forested habitats, then comparisons among forest types should be interpreted cautiously unless the researchers demonstrate that biased detection of vocalizations does not affect their conclusions.