Modeling movements improves capture–recapture estimates for mobile species with sparse data: Polar bears (Ursus maritimus) in Viscount Melville sound

Eric V. Regehr,Steven Baryluk,John Boulanger,Marsha Branigan,Faye d'Eon‐Eggertson,Jodie Pongracz,Adam Thom,Evan S. Richardson
DOI: https://doi.org/10.1002/1438-390x.12198
2024-10-02
Population Ecology
Abstract:This paper presents a Bayesian multistate capture–recapture model that is suitable for mobile species with multiple types of sparse data. Demographic parameter estimates indicated that the Viscount Melville (VM) subpopulation of polar bears had recovered from overharvest during the 1970s–1990s and was likely stable in 2014. While VM bears may be experiencing transient benefits of climate warming due to reduced multiyear sea ice, unexplained declines in adult female survival suggest that continued monitoring is warranted. Wildlife management requires estimates of demographic parameters that are difficult to obtain for mobile species at low densities. Biased parameter estimates often result from capture–recapture (CR) studies due to small sample sizes and unequal recapture probabilities, the latter of which can be caused by animal movements with respect to the sampling area. We developed a multistate CR model designed to minimize biases by including multiple data types (capture, harvest, natural mortality, and telemetry) and accounting for temporary emigration. We applied the model to data collected intensively from 2012 to 2014, and intermittently since the 1970s, for the Viscount Melville (VM) subpopulation of polar bears (Ursus maritimus) in the Canadian Arctic. The number of bears within the VM subpopulation boundary likely increased from an average of 145 (Bayesian 95% credible interval [CRI] [109, 221]) in 1989–1992 to 235 (95% CRI [148, 569]) in 2012–2014. Survival probability increased for all sex and age classes except adult females, for which estimates declined due to unknown reasons. Polar bear movements exhibited Markovian dependence with approximately 28% of the subpopulation located outside of the sampling area each spring. This contributed to inaccurate parameter estimates when using a simpler, single‐state CR model that only included capture data. Although the interpretation of demographic status was complicated by statistical uncertainty and changes in study design, our findings suggest that—as of 2014—the VM polar bear subpopulation had likely recovered from an earlier period of overharvest, was stable, and had not exhibited detectable negative effects of climate warming.
ecology
What problem does this paper attempt to address?