US-IALE 2018

AUTHORS: Jonathan Eiseman*, Western Michigan University; Maarten Vonhof, Western Michigan University; Sharon Gill, Western Michigan University

ABSTRACT: Soundscapes vary over time and space reflecting dynamic inputs from biological, geophysical and anthropogenic activity. For example, the early morning peak of singing by birds or the rush hour traffic of humans change the composition of soundscapes from those experienced at other times of day. To characterize temporal complexity of soundscapes, researchers may record them over the course of a 24-hour cycle, and when they do so over a season will end up with large datasets that demand considerable storage capacity as well as processing time for analysis. Given these constraints, we investigated how to sample soundscapes, asking two questions: with what frequency should researchers record soundscapes to capture their overall variation? And does an optimal sampling schedule exist for different habitats across different locations? We conducted two tests in forest and grassland habitats at nature preserves in southwest Michigan by recording soundscapes continuously over two 24-hour periods in the spring and summer of 2015. From these recordings, we calculated three acoustic indices for every minute of the 24-hour period to document the actual pattern of soundscape variation over space and time. We then simulated different recording schedules by subsampling the complete dataset (e.g. 1 min every 5, 1 min every 10, up to 1 min every 60). We compared hourly means from continuous sampling versus simulated schemes to assess which recording schedules best reflected overall soundscape variation. Preliminary analyses showed that hourly means from infrequent sampling (1 min every 30, 1 min every 60) were weakly correlated when compared to hourly means from more frequent sampling schemes (1 min every 5) in forests and grasslands. However, subsampled forest recordings generated substantially lower correlations with continuous data than did the subsampled grassland recordings. Expanding analysis to other preserves will inform strategies for capturing temporal complexity of grassland and forest soundscapes.