Application of Bayesian decision analysis to management of a sockeye salmon (Oncorhynchus nerka) fishery

Abstract

We developed a decision-making framework for management of a sockeye salmon (Oncorhynchus nerka) fishery on the Nass River, British Columbia, that explicitly accounts for uncertainties in (i) the stock-recruitment relationship, (ii) annual recruitment, (iii) run timing, and (iv) catchability. The method used Monte Carlo simulation within a decision analysis framework and used Bayesian statistics to calculate probabilities for parameter values in the Shepherd stock-recruitment model. The decision dealt with when to open a fishery, upstream of all normal fishing areas, that is intended to harvest fish that are considered surplus to spawning requirements. The optimal decision rule for opening this fishery depended on (i) the relative importance of different management objectives and (ii) the range of shapes of the stock-recruitment relationship that were admitted as possible within the decision analysis. The management decision that was optimal if we assumed a dome-shaped stock-recruitment relationship was not optimal when we admitted the possibility of other shapes of the relationship. Therefore, given the variability in salmon stock-recruitment data, uncertainty in the shape of the stock-recruitment relationship should be routinely considered in analyses of management decisions.