RESULTS OF BAYESIAN METHODS DEPEND ON DETAILS OF IMPLEMENTATION - AN EXAMPLE OF ESTIMATING SALMON ESCAPEMENT GOALS

Bayesian methods have been proposed to estimate optimal escapement goa ls, using both knowledge about physical determinants of salmon product ivity and stock-recruitment data. The Bayesian approach has several ad vantages over many traditional methods for estimating stock productivi ty: it allows integration of information from diverse sources and prov ides a framework for decision-making that takes into account uncertain ty reflected in the data. However, results can be critically dependent on details of implementation of this approach. For instance, unintend ed and unwarranted confidence about stock-recruitment relationships ca n arise if the range of relationships examined is too narrow, if too f ew discrete alternatives are considered, or if data are contradictory. This unfounded confidence can result in a suboptimal choice of a spaw ning escapement goal.