Reliability of Singing Bird Surveys: Changes in Observer Efficiency with Avian Density

Field trials in which paired observers were used and indoor simulations in which recorded bird songs were used indicated that, as the number of singing birds audible from a listening station increased from 1 to 4, the fraction of them recorded by observers declined by up to 50%. This reduction in efficiency violates one of the basic assumptions of any index-that the proportion of animals detected remains constant-and could cause surveyors who rely primarily on auditory cues to underestimate changes in population density by up to 25% for common species and 33% for abundant species. The change in efficiency, which is best regarded as measurement error, cannot be detected by a statistical examination of the data and thus may pass undetected in many field studies. It seems unlikely that any general procedure for "correcting" the error would be reliable. The results indicate that singing bird surveys of common species should be supplemented by other methods if accurate estimates of changes in density are needed. A general conclusion of the study is that whenever animals "compete" for a place in the survey, for example by filling up traps or suppressing one another's songs, then the index tends to underestimate a change in density. If efficiency increases with density, then the survey tends to overestimate a change in density. If the sign of the bias can be determined, the survey can be used to provide a minimum or maximum estimate of a change in density even if the magnitude of the bias cannot be estimated.