A comparison of conventional and state-space production models in fisheries stock assessment and management

Production model with only observational error (conventional model) and a state-space production model incorporating both observational and process errors were evaluated by means of Monte Carlo simulation. The state-space model generally provided more accurate and precise estimates of model parameters even for the scenarios in which process error was absent. When the uncertainty increased in simulated “true” fishery dynamics, the estimates of model parameters would be subject to large bias. Estimation of errors from different sources were difficult in the state-space model, implying that the real uncertainty of “true” dynamics was difficult to define. In contrast, the error in conventional production model reflected the overall uncertainties better. Two biological reference points (BRP) based management strategies, Fmsy-based management strategy and MSY-based management strategy, derived from the two models were compared. The state-space model outperformed the conventional model in most scenarios in achieving management goal of having no overfished stock. The MSY-based strategy led to lower probability of having stock overfished for both conventional and state-space production models for almost all the scenarios. This might be due to that estimate of MSY derived from state-space model is more reliable. One implication of the poor performance of management strategies was that a precautionary reduction factor should be applied to biological reference points in practice.