Alternative Bayesian Synthesis Approaches to Bering-Chukchi-Beaufort Seas Bowhead Whale Stock Assessment: Uncertainty in Historic Catch and Hitting with Fixed MSYR

We report the results of two investigations requested by the Scienti c Committee regarding alternative methods for Bayesian synthesis assessment of the Bering-Beaufort-Chukchi Seas stock of bowhead whales (Balaena mysticetus). The rst series of studies examines the e ect of assuming that the historic pelagic catch record is known without error. We propose a model for variability in the historic catch record and we investigate the potential in uence of catch record uncertainty on assessment results. We also examine the in uence of systematic bias in the record. Our analyses indicate that variability in the kill record is not a dominant source of uncertainty in the Bayesian synthesis results. Consistent, systematic bias in the record does e ect conclusions about initial stock size, depletion, and maximum sustainable yield. There is evidence against the hypothesis that the true historic catch was substantially greater than previously assumed, but the combined analysis can not rule out the possibility that the true catch was less than previously assumed. The second series of studies examines an alternative method for tting population trajectories from the BALEEN II (de la Mare and Cooke, 1993) population dynamics model. This method, known as `hitting with xed MSYR', avoids the need to make assumptions about the unexploited stock size. Our analyses indicate that the use of this method has a substantial impact on the results. This impact may be due to a constraint which prevents population extinction even when biological parameters would otherwise favor extinction. Our analyses indicate that the di erences between methods are not due primarily to the assumptions about initial stock size used by the Scienti c Committee in their original analysis. Addition of the Scienti c Committee evidence about initial stock size to the hitting analysis does not change the results. Our analyses also show that assumptions about population model inputs pre-determine the results of a Bayesian synthesis analysis for the hitting method to a larger extent than for the original method.