Acoustic space occupancy: Combining ecoacoustics and lidar to model biodiversity variation and detection bias across heterogeneous landscapes

Ecological Indicators
By: , and 

Links

Abstract

There is global interest in quantifying changing biodiversity in human-modified landscapes. Ecoacoustics may offer a promising pathway for supporting multi-taxa monitoring, but its scalability has been hampered by the sonic complexity of biodiverse ecosystems and the imperfect detectability of animal-generated sounds. The acoustic signature of a habitat, or soundscape, contains information about multiple taxa and may circumvent species identification, but robust statistical technology for characterizing community-level attributes is lacking. Here, we present the Acoustic Space Occupancy Model, a flexible hierarchical framework designed to account for detection artifacts from acoustic surveys in order to model biologically relevant variation in acoustic space use among community assemblages. We illustrate its utility in a biologically and structurally diverse Amazon frontier forest landscape, a valuable test case for modeling biodiversity variation and acoustic attenuation from vegetation density. We use complementary airborne lidar data to capture aspects of 3D forest structure hypothesized to influence community composition and acoustic signal detection. Our novel analytic framework permitted us to model both the assembly and detectability of soundscapes using lidar-derived estimates of forest structure. Our empirical predictions were consistent with physical models of frequency-dependent attenuation, and we estimated that the probability of observing animal activity in the frequency channel most vulnerable to acoustic attenuation varied by over 60%, depending on vegetation density. There were also large differences in the biotic use of acoustic space predicted for intact and degraded forest habitats, with notable differences in the soundscape channels predominantly occupied by insects. This study advances the utility of ecoacoustics by providing a robust modeling framework for addressing detection bias from remote audio surveys while preserving the rich dimensionality of soundscape data, which may be critical for inferring biological patterns pertinent to multiple taxonomic groups in the tropics. Our methodology paves the way for greater integration of remotely sensed observations with high-throughput biodiversity data to help bring routine, multi-taxa monitoring to scale in dynamic and diverse landscapes.

Study Area

Additional publication details

Publication type Article
Publication Subtype Journal Article
Title Acoustic space occupancy: Combining ecoacoustics and lidar to model biodiversity variation and detection bias across heterogeneous landscapes
Series title Ecological Indicators
DOI 10.1016/j.ecolind.2020.106172
Volume 113
Year Published 2020
Language English
Publisher Wiley
Contributing office(s) Patuxent Wildlife Research Center
Description 106172, 9 p.
Country Brazil
State Mato Grosso
City Feliz Natal, Nova Ubirita
Google Analytic Metrics Metrics page
Additional metadata about this publication, not found in other parts of the page is in this table