Data analysis

Tag data were adjusted for the frequency response of the microphone and high-pass filtered by a 4-pole 10 kHz high pass Butterworth filter to extract only the echolocation calls. Accelerometer data were low-pass filtered with a delay-free linear phase finite impulse response (FIR) filter with a cut-off frequency of 30 Hz. All analyses were conducted using custom-written scripts (Matlab, 2019a, The Mathworks, Natick, MA, USA). All calls in the recordings were automatically extracted and visually inspected for correct detections. Calls were not extracted in time periods where loud sounds from e.g. wind or conspecific calls appeared in the recordings to avoid false detections. As the calls of the bats were emitted in a directional beam in front of the bat, the tag-recorded call levels were lower than the actual on-axis call levels. The difference between the off and on-axis call levels for this species was estimated at 14 dB (Stidsholt et al., in review). Source levels were therefore estimated by adding 14 dB to the call levels measured in energy flux density (dB re 20μPa²s) over a -6 dB energy window from the tag-recordings. This conversion does not take head movements into account, which may shade some calls, but not in an extend affecting our conclusions of the study.

The call source level was also quantified in RMS approximated by adding 25 dB (corresponding to a fixed 3 ms call duration) to the call levels in EFD to facilitate comparison to the literature.