Comparative effects of distorted fishery data on assessment results of two non-equilibrium surplusproduction models

Marine fisheries provide a major source of food and livelihoods for people worldwide. Fishery management plays an important role in achieving sustainable fisheries. Catch per unit effort (CPUE) data from either fishery independent or -dependent surveys are the most informative for variations in population size over time, meanwhile catches from the fishery-dependent survey are also required to assess fishing. If these data are inaccurate, the statistical analyses would be biased, leading to mismanagement of fishery resources. However, systematic distortions appeared in world fisheries catch trends. Moreover, due to lack of fishery scientific investigation, CPUE data were mainly from commercial fishing, and influenced by spatial-temporal factors, environmental factors and also spatial autocorrelation problem. Therefore, it is important to understand the impacts of distorted fishery data on stock assessments. This study used catch and CPUE data of the albacore (Thunnus alalunga) fishery in the South Atlantic. Simulations were conducted to estimate biological reference points (BRPs), i.e., maximum sustainable yield (MSY), B_MSY, F_MSY, B₂₀₁₁/B_MSY, and F₂₀₁₁/F_MSY using non-equilibrium surplus production models based on ASPIC (ASM) and Bayesian state-space modelling (BSM). Simulations were conducted under the following scenarios: ① both catch and CPUE data are accurate; ② only catch data is misreporting; ③ only CPUE data is misreporting, and ④ both catch and CPUE data are misreporting. Five levels (coefficient of variation, CV=1%, 5%, 10%, 20%, and 30%) of stochastic errors were superimposed on catch and CPUE data. The estimated MSYs were 2.866×10⁴ t and 2.836×10⁴ t, B₂₀₁₁/B_MSY were 1.366 and 1.324, F₂₀₁₁/F_MSY were 0.627 and 0.667 by ASM and BSM, respectively, for the first scenario. Larger B_MSY (31.48×10⁴ t) and smaller F_MSY (0.091) were obtained by ASM. These results indicate that this fishery was in a good condition in 2011. Overestimate B_MSY and underestimate F_MSY were obtained using distorted catch and CPUE data by ASM, and distorted CPUE data made more impact than distorted catch data. Absolute percentage bias of estimated BRPs by BSM had a tendency to increase with the stochastic error increasing, and smaller than those by ASM, especially B_MSY and F_MSY. BSM can deal with the stochastic errors better than ASM. Therefore, BSM is suggested to be applied in fishery stock assessment when the fishery data include stochastic error.