基于GAM模型的天津海域鱼类资源和环境因子关系的初步研究  

Preliminary Study on Fish Resources and Its Relationship with Environmental Factors in Tianjin Sea Area Based on Generalized Additive Model(GAM)

作　　者：谷德贤[1] 刘国山[1] 王晓宇[1] 王婷[2] 尤宏争[1] 钱红[1] 徐海龙[3] 

机构地区：[1]天津市水产研究所,天津300221 [2]天津渤海水产研究所,天津300457 [3]天津农学院水产学院,天津300384

出　　处：《天津农学院学报》2017年第1期38-43,共6页Journal of Tianjin Agricultural University

基　　金：天津市水产局青年科技创新项目"天津海域口虾蛄资源开发现状及补充量研究"(J2016-04青)及"天津附近水域钵水母类生物量变动研究"(J2014-29青)

摘　　要：根据2013、2014年5、8、10月6个航次的渔业资源及环境因子调查结果,分析了天津海域鱼类资源的种类组成及生物量,同时利用GAM模型,对天津海域鱼类资源生物量与环境因子之间的关系进行了分析研究。天津海域共采集到鱼类7目17科26种,其中鲈形目种类最多;鱼类资源平均生物量为3.37 kg/h,与1983年天津市海岸带和海涂资源综合调查结果相比,资源量降低了90.39%。8月航次的鱼类资源生物量最大,其次为10月航次,5月航次最小。本研究选取了月份、水温、盐度等11个环境因子,GAM分析结果表明,对鱼类生物量影响较大的4个因子为温度、仔稚鱼密度、盐度和经度,水温、仔稚鱼密度和盐度对鱼类资源生物量影响均显著,经度在p<0.1的水平上显著,模型的累计偏差解释率为44.2%。根据AIC准则,包含上述4个变量的广义加性模型为最佳模型。各预测变量的相对重要性依次为盐度>温度>仔稚鱼密度>经度。According to the survey of fish resources and environmental factors on May, August, October in 2013 and 2014, analyses were performed on the species composition and biomass of fish resources in Tianjin sea area. Furthermore, the relationship between the fish biomass and environmental factors was studied based on the generalized additive model （GAM） . The results showed that： there were 7 orders, 17 families and 26 species of fishes in Tianjin sea area, and the preponderance of the species belonged to perciformes. The average biomass was 3.37 kg/h, and the biomass had dropped by 90.39%, compared with the results of Tianjin multipurpose investigation of the coastal zone and tidal wetland resources （1983） . The maximum number of biomass appeared in August, followed by that in October, and the minimum number appeared in May. This study selected 11 environmental factors as predictor variables, and the biomass of fishes were mainly affected by four factors based on GAM, namely, salinity, temperature, fish larvae density and Longitude. The effects of the foregoing three factor were significant, while the effect of longitude was significant on the level p 〈 0.1. The accumulation of deviance explained by the optimized GAM to the fish biomass was 44.2%. According to Akaike information criterion （AIC） , the model including the four factors was optimal. Salinity had the greatest influence on the biomass of fish resources, followed by temperature and fish larval density, longitude had the smallest effect on fish resources.

关 键 词：天津海域 鱼类资源 广义加性模型(GAM) 环境因子 

分 类 号：P641.131[天文地球—地质矿产勘探] S152.72[天文地球—地质学]