指标聚类构建复合渔业资源丰度指数的研究

Building CPUE index of multi-gear by R type clustering

由于复合渔业中作业方式较多,需要考察CPUE指数也较多,使得渔业资源监测工作复杂化,需要将之删繁为简。本研究以渔港抽样调查积累的南海北部二长棘犁齿鲷不同作业方式与功率段的CPUE数据为例,探讨复合渔业的CPUE指数的优化构建方法。尝试了对CPUE指数采取指标聚类,再根据典型CPUE指数的选择标准和相应的权重系数,计算出综合CPUE指数,然后纳入剩余产量模型开展敏感性分析,比较不同综合CPUE指数与对照CPUE指数以及模型测试输出结果的差异。结果显示,20个CPUE指数可聚成4个类別,并由4个类別中选出了4个典型CPUE指数。权重系数与CPUE指标的不同并未使综合CPUE指数表现出与对照CPUE指数的显著差异,但这些综合CPUE模型分析的结果却存在较大的差异。比较发现,用传统的指标聚类的典型指标挑选方法和复相关系数的倒数计算的权重系数计算的综合CPUE指数适用性最好。指标聚类构建的复合渔业CPUE指数既能减少观测指标,也能满足资源评估要求。

As there are many CPUE indexes in compound fisheries, the monitoring of fishery resources is complicated and needs to be simplified. We discussed the estimation method of CPUE index of mixed fisheries based on CPUE data of the Evynnis cardinalis collected from stratified sampling survey at fishing ports along the northern South China Sea. 20 CPUE indexes of different fishing method and different power were classified using index clustering analysis. The typical CPUE indexes of each category were selected to multiply by the corresponding weight coefficient to get the synthetic CPUE indexes, and sensitivity analysis was test by fitting the surplus production model, then the differences between different synthetic CPUE index and control CPUE index as well as the model test results were compared. As a result, 20 CPUE indexes were classified into four category, and four typical CPUE indexes were selected. The differences of weight coefficients and typical CPUE indexes did not make the synthetic CPUE index significantly different from the control index. However, there are great differences in MSY analyzed by the model that fitted to synthetic CPUE indexes. The synthetic CPUE index is the best which is calculated by the weight coefficient inferred by the inverse of complex correlation coefficient and the typical index selection method of traditional index clustering. Synthetic CPUE index which is estimated by index clustering, not only reduce the observational indexes, but also meet the requirement of stock assessment.