A Feasibility Study of Using Unmanned Aerial Vehicles to Survey Avian Abundance using Audio Recording
Environmental Studies Honors Thesis
Faculty Advisor: Professor Andrew Wilson
Abstract
Roadside counts are commonly used to assess songbird abundance, but they result in oversampling habitat edges and underrepresenting core habitats, areas of step terrain, and wetlands. Accessing off-road habitats can be logistical challenging and time-consuming, resulting in low survey efficiency. Aerial ecological surveys, using unmanned aerial vehicles (UAVs, or drones) have already proven to be valuable in wildlife monitoring. Previous studies used photography or videography to provide permanent documentation of wildlife surveys through low altitude aerial imagery. A significant advantage of UAVs over manned aircraft is their greater safety and lower costs. I propose that UAVs can also be used to conduct audio surveys of vocal species. Here, I report on experiments to test the feasibility of using UAVs to conduct point counts of songbirds. To establish the detection radius of bird songs recorded with a microphone attached to a UAV (DJI Phantom 2.0), recordings of the songs of five regionally abundant bird species (Wood Thrush, Eastern Towhee, Song Sparrow, Chipping Sparrow, and Eastern Meadowlark) were broadcast at distances of up to 140m from the UAV, which was flown at three altitudes (20m, 40m, and 60m). I found that detection rates and radial detection distances of the broadcasts did not differ with UAV altitude for four of the five species (Χ2 tests). Bird recordings were clearly audible at radial distances of 60m. I conclude that it is feasible to use UAVs to conduct aerial point counts that are comparable with traditional terrestrial bird point counts, and describe additional field experimentation needed to refine our survey protocols.