Bioacoustic monitoring of forest songbirds: interpreter variability and effects of configuration and digital processing methods in the laboratory

Abstract

Omnidirectional bioacoustic recording systems offer the ability to record forest songbirds in the field by technical staff, and then interpret the recordings later in the laboratory by skilled interpreters. Among several advantages to this approach are the ability to estimate variance among interpreters, obtain a permanent archival record of the point count, reduce costs by using regular field crews to collect data vs. those skilled in bird identification, and remove impediments to breeding bird surveys due to lack of available skilled birders. In this study we first evaluated the effects of microphone configuration and digital processing methods on the quality and effectiveness of the recordings, and then evaluated how consistently skilled birders interpreted the same songbird recordings collected under a mix of environmental conditions, and related this to the commonness of the species. At the time of this evaluation, the most cost-effective configuration of the bioacoustic monitoring system included use of a 180°/180° microphone combination, a minidisc digital recording system, analog transfer of the sound data via a digital soundcard, post-processing amplification of the signal, and data storage in an .MP3 format. This combination maintains high sound fidelity while minimizing both expense and data storage requirements. As recording device technology improves, the direct storage and digital transfer of .WAV format files will be the preferred and most effective recording option. Despite noisy conditions due to wind and other ambient sounds for many of the recordings, interpreters showed a high level of similarity in species identification and enumeration for the 34 most abundant species. Standardized coefficient of variance increased sharply when species had fewer than 10 occurrences, suggesting that birders are more variable in their identification of rare or uncommon species. Desktop identification systems that include type specimens of spectrographic signature and sound clips to aid interpreters could improve identification accuracy of rarer species.