Species-Specific Responses of Bird Song Output in the Presence of Drones

Abstract

Drones are now widely used to study wildlife, but applications for studying bioacoustics have been limited. Drones can be used to collect data on bird vocalizations, but an ongoing concern is that noise from the drones could change bird vocalization behavior. To test this we conducted an experiment using 30 sound localization arrays to track the song output of seven songbird species before, during, and after a 3-minute flight of a small quadcopter drone hovering at 50 m above ground level. We analyzed 8,303 song bouts, of which 2,285 song bouts of 184 individual birds were within 50 meters of the array centers. We used GLMMs to assess patterns in song output showed patterns that could be attributed to the drone's presence. We found no evidence of any effect of the drone for five species: American Robin Turdus migratorius, Common Yellowthroat Geothlypis trichas, Field Sparrow Spizella pusilla, Song Sparrow Melospiza melodia, and Indigo Bunting Passerina cyanea. However, we found a substantial decrease in Yellow Warbler Setophaga petechia song detections during the 3-minute drone hover, such that there was an 81% drop in detections after 2 minutes (Wald-test, p<0.001), compared with before the drone's introduction. In contrast, the number of singing Northern Cardinal Cardinalis cardinalis increased after the drone was introduced, and remained almost five-fold higher for 4-minutes after the drone departed (p<0.001). Further, we found an increase in cardinal contact/alarm calls when the drone was overhead, with the elevated calling-rate sustaining for 2 minutes after the drone had departed (p<0.001). Our study suggests that responses of songbirds to drones may be species-specific, an important consideration when proposing the use of drones in avian studies. We note that recent advances in drone technology have resulted in much quieter drones, which makes us hopeful that the impacts that we detected could be greatly reduced.