基于约束线性回归的柔鱼栖息地指数渔场预报模型构建

The Construction of Habitat Suitability Index Forecast Model of Ommastrephes bartramii Fishing Ground Based on Constrained Linear Regression

中文关键词: 柔鱼  西北太平洋  约束条件  栖息地指数

英文关键词: Ommastrephes bartramii  Northwest Pacific  Constrained conditions  Habitat suitability index (HSI)

中文摘要:

      柔鱼(Ommastrephes bartramii)是我国在西北太平洋重要的商业捕捞对象，对其渔场进行准确预报是提高渔业生产能力的重要内容。本研究分别选取2005~2013年我国在该海域的柔鱼渔获量和捕捞努力量作为计算适宜度指数(SI)的2种指标，利用包括海表温度、叶绿素a (Chl-a)浓度、表温梯度强度和100 m水深的Argo浮标水温数据在内的海洋环境因子，通过非线性回归，生成了不同环境因子的SI曲线。在考虑约束条件的前提下，建立2种柔鱼渔场的栖息地指数(HSI)模型，并利用逐步回归剔除不显著的解释变量。2种模型拟合优度比较的结果显示，利用渔获量建立的模型具有更高的精度，其中，7~11月模型的调整后相关系数分别为0.853 (P<0.001)、0.773 (P<0.001)、0.789 (P<0.001)、0.745 (P<0.001)和0.724 (P<0.0001)。各环境因子的SI权重系数符合约束条件，并随着季节的变化，权重值有所不同。在主要渔汛期间(7、8和10月)，100 m水深温度的SI对HSI得分起到了最关键作用；而在渔汛末期(11月)，与海表温度相关的SI成为影响HSI的最重要因子。利用该模型对2014年进行预报实验，预报结果与实际渔场在空间分布上具有一致性。全年统计结果显示，高HSI (>0.7)的区域渔获量占总渔获量的49.06%，而低HSI (<0.3)区域渔获量仅占9.06%，表明该模型具有一定的渔场预报能力。

英文摘要:

      Neon flying squid (Ommastrephes bartramii) is an important commercial fishing target for China in Northwest Pacific. Accurate prediction of fishing grounds can improve the squid production capacity. The present study selected historical catch and fishing effort data of Chinese squid-jigging fishery from 2005 to 2013 as suitability index (SI) sources. SI curves were created through nonlinear regression based on 4 environmental factors, including sea surface temperature (SST), chlorophyll-a concentration (Chl-a), SST gradient (Grad) and the temperature of 100 m water layer (T100) from Argo float dataset. On the premise of given constrained conditions, two habitat suitability index (HSI) models were constructed based on catch and fishing effort. Non-significant explanatory variables in the model were eliminated via the stepwise regressions. By comparing the goodness-of-fit of two models, catch-based model provided higher accuracy than fishing effort-based one. The adjusted correlation coefficients were 0.853 (P<0.001), 0.773 (P<0.001), 0.789 (P<0.001), 0.745 (P<0.001) and 0.724 (P<0.0001) from July to November, respectively. The weight coefficients of SI for environmental factors were in accord with the constraint and seasonally varied. In particular, the SI of T100 played the most important role in the HSI score in the main fishing season (July, August and October) while the SI of SST was the major factor to affect HSI in November. The forecast experiment of HSI model was carried out with environmental factors in 2014. Spatial position of forecasted fishing grounds were consistent with actual ones, and catch in high HSI (>0.7) regions accounted for 49.06% of the total catch, while catch in low HSI (<0.3) regions accounted for only 9.06% of the total catch. These findings indicate that the HSI model is able to predict neon flying squid fishing grounds in Northwest Pacific.