Acoustic metrics predict habitat type and vegetation structure in the Amazon. (October 2020)

• Record Type: Journal Article
• Authors: Do Nascimento, Leandro A.; Campos-Cerqueira, Marconi; Beard, Karen H.
• Is Part Of: Ecological indicators. Volume 117(2020)
• Publication Date: 2020-10
• Subjects: Acoustic index -- Ecoacoustics -- Habitat heterogeneity -- Habitat-specific soundscapes -- Rapid biodiversity assessment; Environmental monitoring -- Periodicals; Environmental management -- Periodicals; Environmental impact analysis -- Periodicals; Environmental risk assessment -- Periodicals; Sustainable development -- Periodicals; 333.71405
• DOI: 10.1016/j.ecolind.2020.106679
• Languages: English
• ISSNs: 1470-160X

Abstract

Highlights: Each of the eight habitats studied had unique and predictable soundscapes. Statistical indices better distinguished habitat types than complexity indices. The acoustic evenness index was strongly connected to changes in canopy cover. Soundscapes can track changes in biodiversity across habitats in the Amazon. Abstract: The rapidly developing field of ecoacoustics offers methods that can advance multi-taxa animal surveys at policy-relevant extents. While the field is promising, there remain foundational assumptions that need to be tested across different biomes before the methods can be applied widely. Here we test two of these assumptions in the Amazon: 1) that acoustic indices can be used to predict soundscapes of different habitat types, and 2) that acoustic indices are related to vegetation structure. We recorded soundscapes and collected vegetation data in 143 sites spanning six natural and two human-modified habitats in Viruá National Park, Roraima, Brazil. We grouped the eight habitats into three categories based on vegetative characteristics and flooding regime: open habitats, flooded-forests, and non-flooded forests. Thirteen acoustic indices were calculated from 92, 283 one-minute recordings to describe the soundscapes of the habitats. We found that each habitat type had unique and predictable soundscapes. Random forest models were 74% accurate at predicting the eight habitats types and 87% accurate at predicting the three broader habitats categories. The most important acoustic indices to distinguish habitats were the third quartile and centroid. Canopy cover significantly affected 11 of 13 acoustic indices, and while other vegetation variables (e.g., shrub cover and number of trees) appeared in top models for some indices, their effects were not significant. The best indices linking soundscapes to vegetation structure were the acoustic evenness index and skewness, with canopy cover explaining 81% and 52% of the variance in these indices, respectively. These results expand our knowledge regarding which acoustic indices best connect changes in habitats to changes in soundscapes. These findings are particularly important for diverse ecosystems, like the Amazon, which are known to have complex soundscapes with sound-producing animals that are difficult to detect with traditional survey methods (e.g., visual transects). Ultimately, our results suggest that soundscapes are able to track changes in biodiversity levels across major habitat types of the Amazon.