基于不同参数先验假设的CMSY模型敏感性研究

Sensitivity analysis of CMSY based on varying parameter priorassumptions

  • 摘要: 受数据不足的限制,大部分渔业难以采用传统的渔业资源评估方法为渔业管理提供建议。因此,基于数据缺乏方法的评估模型逐渐受到关注,其中渔获量-最大可持续产量 (Catch-maximum sustainable yield, CMSY) 是目前国际上应用广泛的数据缺乏评估方法。然而,现有研究表明,CMSY模型的评估结果对先验假设具有高度依赖性,且不同参数先验对模型输出结果的具体影响尚不明确。为探究CMSY参数先验假设对评估结果的影响,从RAM传统资源评估数据库 (RAM Legacy Stock Assessment Database) 中随机选取200个来自不同海域的鱼类及无脊椎动物种群对CMSY敏感性进行分析。系统分析了内禀增长率 (r) 以及开始、中间、最后年份的资源量水平 (Bstart/KBint/KBend/KB为生物量,K为环境容纳量) 的上下限先验假设变化对模型估算结果生物量轨迹、最大可持续产量 (MSY)、最大可持续产量对应的生物 (BMSY)、最大可持续产量对应的捕捞死亡率 (FMSY)、相对生物量 (B/BMSY)、相对捕捞死亡率 (F/FMSY)的影响。结果表明:1) 生物量轨迹、BMSYFMSYr下限的影响大,B/BMSYF/FMSYBend/K上限的影响大,而MSY受各参数先验的影响较小。2) 目级分类差异对CMSY方法的敏感性无显著性影响。3) 参数先验的改变易导致Kobe图中的种群资源状况发生象限迁移,其中Bend/K上限对其影响最大。研究表明,CMSY模型对参数先验设置具有较高的敏感性,建议使用该模型时谨慎设定参数和进行结果分析。

     

    Abstract: Due to data limitation, most fisheries, most fisheries find it difficult to use traditional fisheries resource assessment methods to provide recommendations for fisheries management. As a result, data-limited assessment models have gainedincreasing attention, among which catch-maximum sustainable yield (CMSY) has become a widely used method internationally. However, studies indicate that CMSY is highly dependent on prior assumptions, and the specific impacts of different parameter priors on model outputs remain uncertain. To address this problem, we conducted a sensitivity analysis of CMSY by randomly selecting 200 fish and invertebrate populations from different sea areas from the RAM Legacy Stock Assessment Database. The model estimation results, including biomass trajectory, MSY, BMSY, FMSY, B/BMSY and F/FMSY were systematically analyzed to examine their variations by incrementally adjusting the upper and lower bounds of the intrinsic growth rate (r), and the biomass levels at the start, middle and end years (Bstart/K, Bint/K, Bend/K). The results indicate that: 1) Biomass trajectory, BMSY, and FMSY were significantly influenced by the lower bound of r, while B/BMSY and F/FMSY were more sensitive to the upper bound of Bend/K. MSY was less affected by the prior settings of the parameters. 2) Taxonomic classification at the order level exhibited no significant influence on the sensitivity of the CMSY method. 3) Variations in parameter priors were observed to induce quadrant transitions in the population stock status depicted in the Kobe plot, among which the upper bound of Bend/K showed the most significant effect. This study reveals that the CMSY model exhibited high sensitivity to parameter prior settings. It is suggested that careful parameter configuration and result interpretation are essential when utilizing this model.

     

/

返回文章
返回