Comparison of autonomous and manual recording methods for discrimination of individually distinctive Ovenbird songs

Many animals produce individually distinctive vocalizations with increased outputs during the breeding season. Many animals, including birds, can recognize other individuals based on the distinctive features of their songs and researchers can use bioacoustics tools to discriminate among individuals. Typically, bioacoustics analyses use recordings made with highly directional microphones that are free of background noise and spectral overlap. However, recent technological advances in automated recording have made it possible to record remotely and cover larger areas simultaneously. We tested whether spectrogram cross-correlation can be used to discriminate among songs of 19 individual Ovenbirds (Seiurus aurocapillus). We used two microphone types: directional (Sennheiser MKH-70) and omnidirectional (SMX-II) microphones. Because birds may vary in their distance from the SMX-II microphones, songs were selected as either high-quality (close to the recorder) or low-quality (further away from the recorder). We found that all recording types could be used to discriminate the songs of individual male Ovenbirds from other males in the population. Discrimination among directional recordings was significantly better than among omnidirectional recordings, and high-quality recordings could be used to discriminate among individuals significantly better than low-quality recordings. Taken together, our results suggest that automated omnidirectional recording could be valuable for future behavioural research allowing individuals to be followed over an entire breeding season. In addition, acoustic surveys of communities could provide information about abundance as well as presence and/or absence of species.

多数动物在繁殖季会发出具有个体辨识度的鸣唱，且鸣唱产出量会有所提升。包括鸟类在内的诸多动物可通过鸣唱的独特特征识别同类个体，研究者亦可借助生物声学（bioacoustics）工具实现个体间的区分。常规生物声学分析多采用无背景噪音且无频谱重叠的高指向性麦克风录制的音频。然而，自动化录音技术的新近进展使得远程录制、同时覆盖更大范围区域成为可能。本研究验证了声谱图互相关分析（spectrogram cross-correlation）是否可用于区分19只个体灶巢鸟（Ovenbird，学名*Seiurus aurocapillus*）的鸣唱。本研究使用了两种类型的麦克风：指向性麦克风（森海塞尔MKH-70）与全向性麦克风（SMX-II）。由于灶巢鸟与SMX-II全向麦克风的距离存在差异，我们将录制的鸣唱分为高质量（靠近录音设备）与低质量（远离录音设备）两类。结果显示，所有录音类型均可用于区分种群内雄性灶巢鸟个体间的鸣唱差异。其中，指向性录音的个体区分效果显著优于全向性录音，且高质量录音的个体区分性能亦显著优于低质量录音。综上，本研究结果表明，自动化全向录音技术可用于未来的行为学研究，实现整个繁殖季内对个体的追踪监测。此外，群落声学调查不仅可获取物种的存在与缺失信息，还能估算其种群丰度。