Rapid extraction of ecologically meaningful information from large-scale acoustic recordings

Abstract

Acoustic recordings have the potential to address a suite of important conservation questions, from assessing phenology shifts due to climate change, to examining the impact of anthropogenic noise on wildlife, to monitoring biodiversity at enormous spatio-temporal scales. However, consistent methods are required to extract meaningful information from these large datasets. Here we apply a method of calibrating recordings to standardize acoustic data collected at over 50 unique sites in a diversity of habitats across the continental U.S. using a variety of recording units and parameters. The calibration method results in a coarser data resolution, decreasing storage space and computation time of further analysis. We then apply recently developed acoustic indices to evaluate biodiversity in our recordings. A review of existing acoustic indices and degree of correlation with bioacoustic activity, species richness, functional diversity, landscape attributes, and anthropogenic influence guided our decisions about what indices to implement. Resulting indices were compared with the diversity of birds from observer point counts and from animal vocalizations observed in the recording spectrograms and to anthropogenic sounds observed in the recordings. The results provide important insight on the utility of each index, or group of indices, to investigate dynamics of ecological communities across large scales.