A generalized random encounter model for estimating animal density with remote sensor data

Abstract

Acoustic detectors are commonly being used to monitor wildlife. Current estimators of abundance or density require recognition of individuals or the distance of the animal from the sensor, which is often difficult. The random encounter model (REM) has been successfully applied to count data without these requirements. However, count data from acoustic detectors do not fit the assumptions of the REM due to the directionality of animal signals. We developed a generalized REM (gREM), to estimate animal density from count data, derived for different combinations of sensor detection widths and animal signal widths. We tested the accuracy and precision of this model using simulations for different combinations of sensor detection and animal signal widths, number of captures, and animal movement models. The gREM produces accurate estimates of absolute animal density. However, larger sensor detection and animal signal widths, and larger number of captures give more precise estimates. Different animal movement models had no effect on the gREM. We conclude that the gREM provides an effective method to estimate animal densities in both marine and terrestrial environments. As acoustic detectors become more ubiquitous, the gREM will be increasingly useful for monitoring animal populations across broad spatial, temporal, and taxonomic scales.