When can we reliably estimate the productivity of fish stocks?

In modern fishery stock assessments, the productivity of exploited stocks is frequently summarized by a scale-invariant "steepness" parameter. This parameter, which describes the slope of the spawner-recruit curve, determines resilience of a stock to exploitation and is highly influential when estimating maximum sustainable yield. In this study, we examined conditions under which steepness can be estimated reliably. We applied a statistical catch-age model to data that were simulated over a broad range of stock characteristics and exploitation patterns and found that steepness is often estimated at its upper bound regardless of underlying productivity. The ability to estimate steepness reliably was most dependent on the true value of steepness, the exploitation history of the stock, natural mortality, duration of the time series, and quality of an index of abundance: this ability was relatively unaffected by levels of stochasticity in recruitment and sampling intensity of age compositions. We further explored the method of inverse prediction to improve estimates of steepness and conclude that this approach holds promise. We illustrate the utility of simulation and inverse prediction methods with two fish stocks located off the southeastern United States, greater amberjack (Seriola dumerili) and gag grouper (Mycteroperca microlepsis).