BioTrack: Adaptive Population Dynamics Model for Long-Term Monitoring of Threatened Wildlife Species

Abstract

Long-term monitoring of endangered animal populations faces increasing difficulties due to disjointed data streams, non-stationary environmental factors, and poor adaptability of classical demographic models to changing monitoring conditions. This paper proposes BioTrack, an adaptive population dynamics model capable of accommodating heterogeneous biologging, telemetry, and survey data into a unified framework for estimation of endangered species' populations. Contrary to static demographic models, BioTrack continuously updates its parameter set as more observations become available and thus allows estimating population trajectories, survival probabilities, and dynamic demographic signals triggered by individual movements. The proposed framework was tested on historic monitoring data of endangered terrestrial and avian species' populations, with the evaluation performed via the prediction accuracy of population trajectories obtained with respect to the periods of observation withheld for test purposes. The results show that adaptability of the BioTrack framework allows for improving tracking of short-term fluctuations in population trajectories and individual behavior over static approaches while maintaining interpretability for conservationists. The implications of the proposed approach for long-term monitoring of conservation, its comparison with other population modeling approaches, and its applicability for wildlife management agencies are considered.