东南太平洋竹?鱼资源评估与捕捞控制规则模拟研究

以东南太平洋智利竹鱼为对象、以资源量动态模型为基础,使用模拟方法构建了"真实"的智利竹鱼种群及其渔业,评估了观测误差和过程误差对智利竹鱼资源评估和管理的影响。模拟的"真实"的智利竹鱼种群及其渔业结果显示,1997—2014年太平洋智利竹鱼资源量总体上呈逐年下降趋势,且远低于B_(MSY)的50%;捕捞死亡系数波动剧烈,仅在2012—2014年低于F_(MSY)且相对稳定。渔业资源评估模拟结果显示,观测误差和过程误差使资源量和B_(MSY)被低估,捕捞死亡系数和F_(MSY)被高估,且随机误差越大,资源量、B_(MSY)被低估,而捕捞死亡系数、F_(MSY)被高估的程度越大。渔业管理模拟的结果表明,捕捞控制规则采用恒定捕捞死亡系数时,未来10年基于50%2014年捕捞死亡系数的管理措施为最佳管理措施。由于捕捞死亡系数被高估,最佳管理措施实施后使得年总可捕捞量高于预期,而年资源量低于预期,资源量增长或恢复的速度变慢,资源可能同时处于过度捕捞状态和正遭受过度捕捞。过度捕捞的风险与随机观测误差和过程误差的大小成正比。

Stock assessment and harvest control rules simulation for jack mackerel (Trachurus murphyi) in the southeast Pacific

Observation errors and process errors are the main source for the uncertainties of the biomass dynamics model. These errors are harmful to fisheries management because they can lead to biased estimation for some key management quantities such as biomass, fishing mortality and their related biological reference points. Simulation approach has been widely used to examine the effects of observed and process errors and evaluate the ability of the stock assessment model for providing robust catch advice. In this study, we use simulation approach to examine and quantify the impacts of observation and process errors on the population dynamics, assessment and management of Chilean jack mackerel in the southeast Pacific. Based on the biomass dynamics model and real catch and catch per unit fishing effort (CPUE) data, the operating model was built to describe the "true" population dynamics and fisheries for jack mackerel stock, and generate "true" time-series CPUE data. Assuming errors in CPUE has lognormal distribution, 100 sets of simulated time-series CPUE data were generated respectively by adding low level and high level random errors to the "true" CPUE. The assessing model was also based on the biomass dynamics model but low and high level process errors were added by setting low and high coefficient of variation, CV. We refer to the assessing model as "high" error when both observed and process errors were subject to high level and "low" error when both errors were subject to low level. Relative errors (RE) of biomass, fishing mortality, B_MSY and F_MSY were used to measure the disparity between the operating model and assessment model. Constant fishing mortality rate was considered as harvest control rules. Four constant fishing mortality scenarios with current level (F₂₀₁₄) and at 1.25, 0.75 and 0.5 were input to operating model and assessing models to do projections respectively. The results of the projections of the operating and assessing models were compared to analyze the impacts of random observed and process errors on the "true" and simulated population dynamics. The management advice derived from the assessing models, i.e., values of a certain level of F₂₀₁₄ estimated by the assessing models, was employed to the "true" population and the projections were done again to predicted biomass and total allowable catch (TAC). We refer to these predicted biomass and TAC as the theoretical biomass and TAC. The "true" population dynamics and fisheries showed that jack mackerel biomass was less than 50% B_MSY during the period of 1997-2014 while fishing mortality rate was greater than F_MSY except the last three years. The estimated parameters instantaneous growth rate r and hatchability coefficient q by assessing model were greater than those by operating model, but carry capacity K and biomass of the first year (B₁₉₉₇) were smaller than that by the operating model. Median REs of biomass and B_MSY were negative while median REs of fishing mortality rate and F_MSY were positive, and absolute value of these median REs were proportional to the error level. These results indicated that B_MSY and time-series biomass were overestimated, and F_MSY as well as time-series fishing mortality rate were underestimated because of the observed and process errors, and the degree of over- or under- estimation related to the error level. The projections of operating and assessing models showed that simulated biomass trend in the future ten years (from 2015 to 2024) will increase for all the four fishing mortality rate scenarios, but the increase rate or the recovery rate of jack mackerel biomass will decrease with the fishing mortality rate increasing, i.e., the stock needs more time to reach the level of B_MSY. The predicted biomass and TAC in future by the assessing model were overestimated when compared to the "true" future biomass predicted by the operating model. Furthermore, the predicted annual TAC, biomass and the 80% confidence intervals of biomass by the high error assessing model were greater than those predicted by the low error assessing model. When the values of the four level F₂₀₁₄ estimated by the lower and high error assessing models were input to the operating model, the predicted theoretical biomass, ratio between biomass and B_MSY and TAC in future ten years were less than their simulated and "true" values. The projection results of low and high error assessing model indicated that the fishing mortality rate was set to 0.5F₂₀₁₄ (equal to 0.073 and 0.074 for the low and high error assessing model respectively) in future was best for the jack mackerel stock. Under this fishing mortality rate (0.073 and 0.074), jack mackerel biomass is lower than the simulated and "true" biomass. Although the theoretical value will still increase, the rate of increase will be slow and the stock can recover until the year of 2023. On the other side, the theoretical TAC which means the actual catch for the fisheries, are higher than the expected TAC (i.e., the simulated TAC). Moreover, the higher the error level is, the bigger differences between the theoretical and simulated biomass and TAC are, and this can result in the increase of risk that the jack mackerel stock will be overfished with overfishing.