Sound localization for estimating population densities of birds: Progress, challenges, and opportunities

Abstract

Autonomous recording units (ARUs) are seeing rapid adoption as a research tool to estimate population densities of birds. One challenge for estimating population density is that the distance to the vocalizing bird is unknown. An emphasis has been placed on assessing the detection radius for different species on ARUs, and more recent work has had success relating sound amplitude to distance. We argue, however, that the true variable of interest is not density, per se, but the relationship between density and habitat type or other variables. In an ideal world, ARUs would collect information on the absolute location of each sound source, rather than the number of each species within earshot of the microphone, since the detection radius of an ARU commonly encompasses multiple habitat types. Sound localization is a promising solution to this problem, with some localization methods allowing sound sources to be pinpointed within 1m of their true location. We will present localization results comprising several thousand songs from several species in the Boreal forest, to highlight how localization data can be translated into estimates of population density. We will also highlight several challenges and opportunities for future contributions in the field of sound localization.