Demography and population status of polar bears in western Hudson Bay

Abstract

We evaluated the demography and population status of the Western Hudson Bay (WH) polar bear subpopulation for the period 1984-2011, using live-recapture data from research studies and management actions, and dead-recovery data from polar bears harvested for subsistence purposes or removed during human-bear conflicts. We used a Bayesian implementation of multistate capture-recapture models, coupled with a matrix-based demographic projection model, to integrate several types of data and to incorporate sampling uncertainty, and demographic and environmental stochasticity across the polar bear life cycle. This approach allowed for estimation of a suite of vital rates, including survival and reproduction. These vital rates were used to parameterize a Bayesian population model to evaluate population trends and project potential population outcomes under different environmental scenarios. Survival of female polar bears of all age classes was significantly correlated with sea ice conditions; particularly with the timing of sea ice break-up in the spring and formation in the fall and the interaction of the two. This is consistent with previous findings linking body condition and survival of WH polar bears to environmental changes associated with climatic warming and supports the ecological dependence of polar bears on the availability of sea ice. Survival of male polar bears was not correlated with sea ice conditions. This was likely because a higher proportion of mortality for males was caused by humans rather than by natural factors. For example, approximately 73% of mortality for young male bears (i.e., 5-9 years old) was due to direct human-caused removals, largely because of sex selectivity in the subsistence harvest. The declining trend in size of the WH subpopulation over the period 1987-2004 was similar to a previous analysis (Regehr et al. 2007), suggesting consistency between the two demographic evaluations. Point estimates of abundance were somewhat lower using the updated statistical approach. It is important to recognize that the analyzed data were not collected in a manner that is optimal for estimating abundance and that the goal of the current analysis was to estimate vital rates and demographic trends. Estimates of population growth rate were also derived using a Bayesian population model based on estimated survival and reproductive rates from the multistate capture-recapture model. For the recent decade 2001-2011, the growth rate of the female segment of the population was 1.02 (95% CI = 0.98-1.06). Apparently stable to positive population growth for females may be due in large part to nonlinearity (i.e., short-term stability) in the long-term observed and forecasted trend toward earlier sea ice break-up in western Hudson Bay. The 2011 abundance estimate from this analysis was 806 bears with a 95% Bayesian credible interval of 653-984. This is lower than, but broadly consistent with, the abundance estimate of 1,030 (95% confidence interval = 745-1406) from a 2011 aerial survey (Stapleton et al. 2014). The capture-recapture and aerial survey approaches have different spatial and temporal coverage of the WH subpopulation and, consequently, the effective study population considered by each approach is different.