Reconciling stock assessment paradigms to better inform fisheries management

Abstract

Abstract For several decades, the fisheries stock assessment paradigms of virtual population analysis (VPA) and statistical catch-at-age (SCA) models have been routinely applied to major fish stocks, and their prevalence often dictated by historical continuity, local experience, and geographical differences in standard practices. Similarly, there is a growing split among models using short and long time-series. In one approach, only the recent time-series, where the data are relatively complete, and the assumptions about stationarity in population and sampling processes are relatively simple, are included. In the other, long time-series include far more historical data, but necessitate the relaxation of many common assumptions regarding stationarity. Unlike scientific paradigms in fields outside of fisheries science where empirical validation can provide a growing body of irrefutable evidence (such as physics), there is no expectation that some “truth” will emerge or that a single best stock assessment modelling approach will ultimately displace the others. The 2013 Pacific halibut SCA stock assessment, with the addition of a VPA-based analysis, is used to illustrate how an ensemble approach can represent a more complete description of the uncertainty in management quantities, relative to selecting just one of these competing model paradigms. We suggest that risk assessment for fisheries management, based on stock assessment models, should seek to avoid binary decisions about which models to include, and instead seek better approaches to incorporate alternative models. The ensemble approach to stock assessment also provides a conceptual link between traditional “best model” analyses and fully developed management strategy evaluation of harvest policy and management procedures.