Dominant power spectrums as a tool to establish an ecoacoustic baseline in a premontane moist forest

Abstract

The establishment of a soundscape baseline might allow the rapid detection of significant environmental changes and could potentially be a holistic instrument for biome conservation. Due to the diversity and variability of the sounds produced in an ecosystem, the establishment of a baseline is complex. Thus we evaluated if, in an ever-changing soundscape, it was possible to find dominant acoustic features on a monthly and hourly scale. Dominant power spectrums (PSs) in an area of premontane moist forest were identified by correlating PSs from February to December 2015, calculating the linear correlation between them, and defining a set of PSs that were strongly correlated (Pearson > = 0.95) to most of the obtained PSs. For any two PSs, the Pearson coefficient > = 0.95 accounted for an equivalence relation between the PSs; these relations allowed us to group spectrums into a few sets. In the daytime, 7.3% (eight out of 109 PSs) of the PSs (i.e., the dominant spectrums) were strongly correlated to 80.7% of the obtained PSs; the remaining 19.3% were singular PSs, weakly correlated to other diurnal PSs. During the night, 6.06% (four out of 66 PSs) of the PSs (i.e., the dominant spectrums) were strongly correlated to 53.0% of the obtained PSs; the remaining 47.0% were singular PSs, weakly correlated to other nocturnal PSs. Strong correlations between the PSs on an hourly scale and monthly scale could be used to denote features that prevail over time. The occurrence of strong associations between PSs (i.e., > 0.98) in spectrums from different months suggests that the generation of sound in the studied forest has a well-defined frequency distribution.