Drone Flights Reduce Bat Detection Rates
Between 2018 and 2020, researchers at the Kenauk Institute in western Quebec conducted a series of surveys to investigate how small commercial drone flights affect the detection of bat activity. The work began in 2018 with traditional transect-based surveys using an Anabat SD2 detector, mapping species presence across open-canopy and closed-canopy habitats. Eight species known in Quebec were observed, with markedly higher detection rates in open-canopy areas—seven times greater than in closed-canopy sites. Species with similar acoustic signatures were pooled into complexes to reduce identification errors: the EPNO complex (big brown bat and silver-haired bat) and the MYSP complex (three Myotis species). Hoary bats, eastern red bats, and tri-coloured bats were identified individually.
In 2019 and 2020, the team shifted focus to open habitats, where bat activity was greater. The Anabat was too heavy for aerial deployment, so they used lightweight Echometer Touch detectors attached to iPod 7 devices, mounted on a DJI Phantom 4 quadcopter. To mitigate propeller noise interference, a 2-inch Sonoflat acoustic foam divider was installed between the detector and the drone, following recommendations from earlier studies.
Surveys followed a three-phase design: Phase 1 recorded from the ground without UAV; Phase 2 recorded during UAV flight in a 10–15 m diameter circle at canopy height; Phase 3 repeated ground recording post-flight. In 2020, simultaneous ground-based detection during Phase 2 and control sites over 1 km away were added to assess UAV noise impact. Sites were selected based on high bat activity, often near water or buildings, both preferred hunting grounds.
Data collection proceeded only if Phase 1 detected at least three bat passes in five minutes. In some cases, phase durations were extended to ten minutes to capture more data. Each flight required a certified UAV pilot and an assistant, with manual hand-launch and hand-landing procedures due to the obstructed landing gear from the mounted equipment. All protocols adhered to Canadian Council for Animal Care guidelines and were approved under McGill University’s animal care protocol.
Analysis using generalized linear models in R examined the effects of phase and detector location on total bat passes. Tukey tests compared differences between phases. Species-specific detectability changes were calculated as the difference between average detection rates in Phases 1 and 3 versus Phase 2. Four species categories were analyzed: MYSP, EPNO, LABO (eastern red bat), and LACI (hoary bat). Tri-coloured bats were absent from these later surveys.
To isolate technological limitations, a secondary experiment measured how drone propeller noise affected detection range. A 40 kHz sine wave at 40 dB SPL was broadcast from a loudspeaker, and the Echometer Touch detector was moved away until the signal was lost. This was repeated with the detector mounted on a flying UAV. The team recorded spectrograms via the Echometer Touch app to visually assess the distance at which the ultrasonic signal became indistinguishable from drone noise.
Spectral analysis was performed using Avisoft SASLab Pro, comparing recordings of the drone in flight, motors running without propellers, and ambient noise. Recordings were normalized and processed with Fast Fourier Transform to achieve fine frequency and temporal resolution. Logarithmic Power Spectra revealed frequencies where drone noise exceeded −80 dB, highlighting spectral overlap with bat echolocation pulses.
The findings demonstrated a consistent reduction in bat passes detected during UAV flight compared to pre- and post-flight ground recordings. Ground-based detectors during UAV operation often recorded bat activity that aerial detectors missed, indicating that propeller noise and possibly flight presence reduced detection capability. The combination of field surveys and controlled acoustic tests provided clear evidence that even small commercial drones can interfere with bioacoustic monitoring, with implications for both ecological research and UAV operational practices in sensitive wildlife habitats.
