Distance truncation via sound level for bioacoustic surveys in patchy habitat

Abstract

ABSTRACT Bioacoustic recordings are often used to conduct auditory surveys, in which human listeners identify vocalising animals on recordings. In these surveys, animals are typically counted regardless of their distance from the survey point. When these surveys are carried out in patchy habitat or near edges, detected individuals may frequently occur in a different land-cover type than the survey point itself, which introduces uncertainty regarding species-habitat associations. We propose a method to restrict detections from single microphones to within a pre-specified survey radius. The method uses logistic regression to select a sound level threshold corresponding to the desired distance threshold. We applied this method to acoustic data from the centre of 21 1-ha oil wellsites in northern Alberta. To compare our approach with the results from auditory surveys from the same locations, we used sound localisation to locate birds detected via both methods. Just 22.4% of birds detected on auditory surveys fell within the desired survey area. Using our distance-truncation method, 96% of detections were within the desired survey area. We propose that distance truncation via sound level will be useful for surveys of patchily distributed habitat, or when greater certainty about bird locations or habitat associations is desired.