Sampling effort for community composition higher by a magnitude compared to species richness in passive acoustic monitoring

1. Comprehensive estimations of relative abundances and changes therein are one of the most important measures of conservation monitoring. The use of passive acoustic monitoring (PAM) has the potential to greatly enhance information availability, yet has mainly focused on species diversity or coverage, yielding results comparable to those from expert observers. Comparisons with conventional monitoring data have been made mainly for point counts, even though area-wide territory mapping is a widely used standard in many monitoring programs. 2. We compare data derived from the combination of PAM and automated identification via machine learning with an area-wide conventional breeding bird survey conducted by expert observers across the Marburg Open Forest in Hesse, Germany. By varying the number of survey intervals, recording duration and locations, we then determined the sampling effort needed to adequately reflect both species coverage and community composition of the conventional survey. 3. Only small subsamples of PAM data were required to reach maximum species richness of the conventional survey; minimum requirements were as low as 1) one survey interval or 2) 30 min duration or 3) three recording locations if other factors were allowed to vary accordingly. Revealing the community composition of the conventional survey, however, required sampling over three to six survey intervals with recording durations between 10 and 100 h. While communities were similar between methods in terms of species activity and relative abundance, PAM also partially reflected the composition of territories in the conventional survey. 4. Our study demonstrates the relative importance of greater sampling effort especially for monitoring community composition, requiring more recording locations, duration and intervals in comparison to species richness. We provide insight into the current applicability of PAM in monitoring practice and present best-use scenarios on how to make the most of high spatio-temporal resolution acoustic monitoring data. • Composition requires up-scaling sampling effort by a magnitude. • Species richness requires low sampling effort in both methods. • Recording duration, intervals and locations can vary for composition. • Territory mapping results partially mirrored in PAM.