Optimal Model Complexity in Stock Assessment

Fisheries stock assessments have evolved towards using complex models involving hundreds of parameters. It is not clear that these types of models are optimal for the management goals of sustaining high yields and ensuring stock safety. Their implementation is time-consuming and limited to highly trained technical staff. Additionally, this complexity makes their behavior difficult for even trained analysts to understand, and to decision-makers and stakeholders they are completely opaque.  The performance of stock assessment (as yield and health of a simulated fish stock) using models ranging from very simplistic to very complex is compared using a management strategy evaluation simulation approach. The relationship of optimal model complexity to the types and quality of data available is investigated. The study also examines a variety of precautionary harvest control rules for translating the results of a stock assessment into a harvest quota; these include decision analytic methods.