Quantifying uncertainty and incorporating environmental stochasticity in stock assessments of marine mammals

Incorporating scientific uncertainty and accounting for potential effects of environmental variability in stock assessments are vital aspects of providing sound management advice. This dissertation focuses on these issues in the context of estimating sustainable quotas for aboriginal subsistence hunting of marine mammals.;Bayesian model averaging was used to take into account both parameter and model uncertainty in the assessment of the Bering-Chukchi-Beaufort Seas (BCB) stock of bowhead whales. The lower 5th percentile of the Bayesian model-averaged posterior for an aboriginal whaling catch quantity &parl0;Q1+1&parr0; was estimated to be 155 whales in 2002. This estimate provides confirmatory evidence that current catch quotas for this stock are sustainable.;In order to assess the robustness of Bayesian assessments to alternative methods for constructing a joint prior distribution which respects biological realism, sensitivity analyses were performed in the context of a risk assessment. The probability of meeting a management objective for aboriginal subsistence hunting was found to be robust for the data-rich BCB bowhead example. However, the data-poor East Greenland walrus example was shown to be sensitive to this issue.;A framework was developed and applied to the assessment of the eastern North Pacific stock of gray whales which incorporated environmental stochasticity by estimating the relationship between residuals in birth and survival rates to an environmental time series. The scenarios which accounted for the effect of the mortality event in 1999 and 2000 led to less optimistic estimates of population status during recent decades with concomitant recovery generally attributable to higher survival rates as opposed to higher birth rates. This framework was then used as an operating model conditioned on forecasts of sea-ice, which has been hypothesized to regulate recent calf production in the population. Future population trajectories were then used to test the performance of the current catch control rule for this stock. Performance was measured by the ability to satisfy different levels of future aboriginal subsistence need, and was shown to be satisfactory under the range of trials examined here.;The methods developed in this dissertation are generally applicable for providing management advice for exploited populations of marine mammals and other renewable natural resources.