印度洋长鳍金枪鱼CPUE权重配置对资源状态指标的影响

单位捕捞努力量渔获量（catch per unit effort，CPUE）权重问题对于渔业资源评估而言至关重要。本研究使用印度洋长鳍金枪鱼（Thunnus alalunga）的渔业独立和非独立数据，构建了年龄结构资源评估模型（ASAP）。利用评估模型估计得出的参数，使用年龄结构种群模拟器（PopSim）模拟“真实”的资源种群动态以及相应的捕捞动态。针对不同序列的CPUE数据赋予不同的权重因子，同时考虑种群关键参数（自然死亡系数M和陡度h）的错误设置，进行敏感性分析，阐述CPUE权重的错误设置对评估结果的影响。结果表明，当估算模型中的M和h被正确指定或被低估时，若给具有较高准确性或较长时间序列的CPUE分配更多的权重,模型估算的捕捞死亡系数F和产卵亲体生物量B具有较小的相对误差（RE）和相对均方根误差（RMSE），即估算更为准确。同时，对不确定性较高的CPUE赋予更大的权重会使Flast/Fstart的估计值过高，而Blast/Bstart的估计值准确性较低。因此，当使用多组CPUE数据时，对具有较高准确性或较长时间序列的CPUE分配更高的权重，或可提高资源状态指标估算的准确性。同时，在CPUE权重的分配中应考虑重要生物学参数（例如M和h）的准确性，至少应进行敏感性分析，以涵盖潜在的模型或参数的错误设置对CPUE权重的影响。

Impact of CPUE index weighting on stock status indicators of Indian Ocean albacore Thunnus alalunga

The data weighting for catch per unit effort (CPUE) is essential for integrated fisheries stock assessments. An Age-Structured Assessment Program (ASAP) was developed using fishery-dependent and fishery-independent data of Indian Ocean albacore (Thunnus alalunga). An operating model (OM) was developed to mimic the population dynamics and fishing operations, and estimation models (EMs) based on the Statistical-Catch-At-Age model were used to compare the effect of different CPUE weighting scenarios on the estimation of population attributes. To investigate the confounding impact of misspecification of key model parameters with the CPUE weighting, nine combinations of natural mortality (M) and the steepness (h) of the Beverton-Holt stock-recruit relationship were considered in the EMs. The results showed that when M and h were correctly specified in the EMs, assigning higher weightings on CPUE indices with higher precision or longer time series would lead to better model performance in terms of both median relative errors (REs) and relative root mean square error (RMSE). Meanwhile, putting more weighting on the CPUE indices with higher uncertainty leads to significantly over-estimated fishing mortality and a low accuracy in spawning stock biomass. Therefore, when using multiple sets of CPUE time series, assigning higher weights to CPUE indices with higher precision or longer time series may improve the accuracy of stock status indicators. Furthermore, the reliability and uncertainty of some key parameters (e.g., M and h) should also be taken into account when weighting CPUE; at least a sensitivity analysis should be conducted to cover possible uncertainty of the model or parameter misspecification and its associated CPUE weighting.