Blood from a stone: performance of catch-only methods in estimating stock biomass status

Abstract

Demand for data-limited stock assessment models is increasing and new methods are being developed rapidly. One class of these models requires only catch time series and, in some cases, information about life history or fishery characteristics, to estimate stock status. These catch-only models (COMs) range from statistical models trained on data-rich stocks to mechanistic population models that make assumptions about changes in fishing effort. We review 11 COMs, comparing performance through application to data-rich stocks and simulated fisheries. The catch-only models evaluated here produce imprecise and biased estimates of B/BMSY, especially for stocks that are lightly exploited. They were also generally poor classifiers of stock status. While no method performed best across all stocks, ensembles of multiple COMs generally performed better than individual COMs. We advocate for testing new COMs using this common platform. We also caution that performance in estimating stock status is not sufficient for gauging the usefulness of COMs in managing fisheries. Greater use of management strategy evaluation is needed before COMs can be considered a reliable tool for management.