Ambient noise decreases detectability of songbird vocalizations in passive acoustic recordings in a consistent pattern across species, frequency, and analysis method

摘要

The presence of ambient noise is a potential constraint that can impact studies that rely on passive acoustic recording units to compare biotic signals across a range of ambient noise levels. To investigate the hypothesis that ambient noise will impact the detectability of recorded acoustic signals, we projected noise with different amplitudes and frequency distributions into one microphone on a passive acoustic recording unit, while recording simultaneously from a second unit free of noise stimuli. We compared detection rates for five frequently recorded bird species derived from three different analysis methods. As predicted, amplitude had a strongly negative impact on detectability. There were also absolute differences in detection rates between species and the analysis methods, but the pattern of change as amplitude increased did not differ across treatments. The frequency distribution of the noise stimuli had no impact on detectability. Based on the consistent impact of amplitude on signal detection, we formulated a single correction factor to estimate the proportion of detections being lost to background ambient noise. While this formula will benefit from additional testing across an array of species and conditions, it is clear that ambient noise alters detection rates in passive acoustic recordings.