Measuring sound detection spaces for acoustic animal sampling and monitoring

Abstract

Sound recordings obtained from passive acoustic monitoring systems are increasingly used to sample animal biodiversity. However, sound recorders sample variable detection spaces, so that data may not be comparable between sampling sites and recording setups.Focusing on terrestrial systems, we measured understory vegetation, tree structure, sound transmission, ambient sound pressure level, and derived sound detection spaces of 38 plots in lowland rainforest, jungle rubber, and oil palm and rubber plantations, using different combinations of sound frequency (0.05 to 40kHz) and source height (0 to 5m).We show that simple vegetation structure measures poorly predict sound transmission, so that direct sound transmission measurements are indispensable. We depict highly variable sound detection spaces in different land-use types. Finally we estimated species richness of exemplary animal groups and found considerable differences between land-use types on the basis of variable detection space areas alone.Sound detection spaces respond non-linearly to sound frequency and source height, and they need to be quantified in acoustic surveys to avoid substantial bias in biodiversity estimates between sampling sites. Detection spaces also determine species detection probabilities and allow comparing data between recording setups. We provide guidelines and computer scripts for measuring sound transmission and ambient sound level using consumer audio equipment, and for computing detection spaces. Appreciating the effective sampling area of acoustic recorders closes a gap between acoustic and traditional animal survey methods. Species richness estimates can now be reported for measured sampling areas, and animal population variables such as abundance, density, and activity can be compared at equal areas.