论文部分内容阅读
摘要：分析南极磷虾分布与环境因子的非线性和空间非静态性关系，对南极磷虾的高效捕捞和管理具有重要意义。本研究基于“龙腾”船2015、2016年在南设得兰群岛捕捞作业的渔捞日志数据，应用广义加模型（Generalized additive model，GAM）和地理权重回归模型（Geographical weighted regression，GWR）探究南极磷虾（Euphausia superba）渔场分布与环境因子的非线性和空间非静态性关系，并比较这2种模型的模拟性能，为南极磷虾的渔场渔情预报、资源评估和渔业管理提供基础数据。GAM模型结果显示，2015、2016年单位捕捞努力量渔获量（Catch per unit effort, CPUE）与作业水深均呈显著负相关关系（P<0.01），表明在作业水深范围内，南极磷虾在较浅水域集群密度较高；2015年CPUE与表层水温呈显著正相关关系（P<0.01），但在2016年呈显著负相关关系（P<0.01），推测是由于2年调查作业位置不同所致；CPUE与离岸距离关系不显著（P>=0.05）。GWR模型结果显示，作业水深对CPUE的影响无显著的空间变化（P>0.05）；海水表温和离岸距离对CPUE的影响具显著的空间变化（P<0.01），表明这2个因子对南极磷虾渔场分布的影响在空间上不连续，存在显著空间非静态性。GAM模型可用于研究资源分布与驱动因子的一般规律；GWR模型作为全局回归模型的有效补充，可用于探究一般规律不适合的特殊区域，便于发现资源分布的“热点”区域，未来在海洋生物资源分布研究中将有广阔的应用前景。Abstract: Antarctic krill is one of zooplanktonic crustaceans, and environmental factors are crucial for its distribution. The relationship between the fishing ground distribution of Antarctic krill and environmental factors is often indicated to be nonlinear and variable among spatial locations. Carrying out the researches in this area is meaningful for forecasting shoals of fish and fishing grounds, as well as assessing and managing krill resource. In this study, we used a generalized additive model (GAM) and a geographically weighted regression (GWR) model to analyze nonlinear and spatial nonstationary relationships between Antarctic krill distribution and environmental factors with the results from the two models based on the fishing-log data of “Longteng” boat in 2015 and 2016 compared. Results from the GAM indicated a significant negative relationship (P<0.01) between catch per unit effort (CPUE) and fishing depth in both years, implying that the aggregation density was high in the shallow water; a significant positive relationship of CPUE with sea surface temperature (SST) in 2015 which was conformed with the previous studies; a significant negative relationship in 2016 probably caused by the different sampling sites in two years; and an insignificant relationship (P>0.05) of CPUE with offshore distance in both years. Results from the GWR indicated that fishing depth affected CPUE, which changed among sample sites insignificantly and was in accordance with the results from the GAM; and the SST and offshore distance affected CPUE, which changed among sample sites significantly (P<0.01) and was not correlated with these two factors. Global regression models, such as GAM, are used to study the general rules between resource distribution and driving factors, and GWR as a complementary tool is used to explore the special zones where general rules are not applicable and have broad prospects in the research of distribution of marine living resources.