Field-Testing of Unmanned Aerial Vehicles to Monitor Avian Abundance Using Audio Recordings
Environmental Studies Honors Thesis
Point counts are the most common method of surveying birds, but they are subject to temporal, spatial, and observer biases. To eliminate several of these biases, we propose attaching a recorder to a consumer-grade quadcopter (Unmanned Aerial Vehicle, or UAV) to estimate bird populations based on audio data. We obtained information on behavior (song output) in response to the drone as well as recordings of bird song. Birds had a slightly negative response to the quadcopter. Species richness was similar between the terrestrial and aerial point counts, but species composition differed. For most species, the number of birds detected on aerial recordings were similar to ground-based counts, but they severely under-sampled mourning doves (Zenaida macroura), gray catbirds (Dumetella carolinensis), and willow flycatchers (Empidonax traillii). Birds with quiet or low frequency songs are likely to be under-detected by UAV-based methods, due to masking by the drone noise of the quadcopter. We demonstrate that low-cost consumer-end UAVs provide a useful new method of surveying songbirds that is accessible to organizations and researchers with restricted budgets.