基于不同模型研究环境因子对中西太平洋鲣资源丰度的影响

根据1998—2013年中西太平洋鲣(Katsuwonus pelamis)生产数据,选取时空因子(年、月、经纬度)和环境因子海表面温度(SST)、海表面高度(SSH)、尼诺指数(ONI)和叶绿素a浓度]Chl-a)],通过两种不同的模型(广义加性模型GAM和提升回归树模型BRT)研究各因子对鲣资源丰度(以CPUE表示)的影响。研究结果认为,GAM模型中,经度对CPUE的影响最大,累计解释偏差超过50%,其次为纬度、年和月;在环境因子中,SSH最为重要,其次为ONI,而SST和Chl-a的影响相对较低。BRT模型分析结果与GAM分析结果类似,时空因子相对占据了重要的地位,其中经度的影响最大,其次为年、纬度和月;而在环境因子中,ONI的重要性相对更高,其次为SSH,SST和Chl-a同样影响较低。研究认为,两种模型均能较好地反映出因子对CPUE的影响。由于厄尔尼诺/拉尼娜现象引起的海洋环境变化会使鲣资源分布产生差异,因此在后续的渔情预报研究中,应该更多地考虑将ONI因子纳入渔情预报模型中,以提高预测精度。

Influence of environmental factors on the abundance of skipjack tuna (Katsuwonus pelamis) in west-central Pacific Ocean determined using different models

Correlations of the catch per unit effort (CPUE) (based on the catch data of skipjack tuna, caught using the purse seine technique in the west-central Pacific Ocean) with spatial-temporal factors (year, month, latitude, and longitude) and environmental factors (sea surface temperature, SST; sea surface height, SSH; oceanic nino index, ONI; and chlorophyll-a, Chl-a) were analyzed, and the relative importance of CPUE was estimated using two different types of models (Generalized additive model:GAM, and Boosted regression tree:BRT). The results showed that longitude is the most important factor in determining the importance of CPUE using GAM, accounting for more than 50% of the total CPUE, while latitude, year, and month had decreasing importance in the order mentioned. SSH is the most important environmental factor in GAM, and ONI, SST, and Chl-a are less important in determining the importance of CPUE. The result of BRT was similar to that of GAM; longitude is the most important spatial-temporal factor, accounting for 60% of the total importance of CPUE, while year, latitude, and month were of less importance, with their importance decreasing in the order mentioned. ONI is the most important environmental factor in BRT, followed by SSH, SST, and Chl-a, in that order. In conclusion, the two types of models can effectively reflect the influence of CPUE. ENSO induced oceanographic variation will change the abundance distribution of skipjack tuna; so, ONI should be included in fishery forecasting models to improve the accuracy of prediction in future.