Polar bears (Ursus maritimus) depend on sea ice for most aspects of their life history, including access to seals, their main prey. Climate warming induced losses in sea ice are therefore expected to result in a loss of feeding opportunities, with consequent reductions in polar bear body condition, survival, reproduction, and eventually population abundance. To date, no quantitative predictions for changes in body condition, reproduction, or survival under climatic warming exist despite long-term empirical research, largely because historic and predicted sea ice conditions differ substantially, making extrapolation from current observations difficult.
Here, we develop a mechanistic dynamic energy budget model for adult female polar bears to predict how their body condition and reproductive success would be affected by expected losses in sea ice and feeding opportunities. To this end, we first develop a body composition model that allows estimating the amount of energy stored in the fat and protein reserves of individual bears, given their body mass and body length. Based on the body composition model, we formulate the dynamic energy budget model to track changes in energy stores due to somatic maintenance, movement, and feeding. Applying the models to the population of western Hudson Bay, we give quantitative predictions for changes in adult female energy stores as a function of predicted changes in sea ice dynamics and associated feeding opportunities. Using the predictions for changes in maternal energy stores, we obtain predictions for changes in reproduction under predicted climate change scenarios, specifically regarding the probability of successful pregnancy and expected mean litter size.
We present quantitative predictions for the probabilities of a pregnant female producing zero, one, two, or three cubs as a function of spring sea ice break-up date and on-ice feeding rates. Severe declines in both the probability of reproduction and expected mean litter size can be expected, but the precise rates of change can only be approximated due to a lack of data on current feeding rates.