海州湾双斑蟳栖息分布特征与环境因子的关系

摘要

There are internal relationships between the spatial and temporal distribution of species and environ-mental factors; however, the randomness and uncertainty of marine ecosystems prevent demonstrating the relation-ships in an easy way. Based on the seasonal bottom trawl survey data collected from 2011 to 2016 in Haizhou Bay and its adjacent waters, we used three species distribution models, including GLM, GAM and Random Forest to study the spatial distribution of Portunid crab (Charybdis bimaculata) and analyze the relation with environmental factors. We selected influential environmental variables and built models according to criteria such as AIC, devi-ance explanation and cross validation. The effects of environmental variables on the distribution of C. bimaculata were evaluated on the basis of the SDMs. The results showed similar results among the three models in the inter-pretation of the relationship between environmental variables and crab distribution. The GAM provided better fit to data, whereas random forest have superior predictive performances than the other models. The models illustrated significant variations of crab abundance among years and seasons, which contributed to over 18 percent and 3.8 percent of total deviance, respectively. Depth and sea surface salinity were influential environmental factors for re-lative abundance, both positive correlated with relative catches. The spatial distribution of Portunid crab was fea-tured by high catches in northeast area and low in southwest, in accordance with the water depth in Haizhou Bay. We predicted the spatial distribution of C. bimaculata across seasons using FVCOM simulation data and the ran-dom forest model to facilitate the exploitation and conservation of the fishery resources.%为了解双斑蟳栖息分布规律,实验根据2011―2016年多个季度航次在海州湾进行的渔业资源和环境调查数据,采用广义线性模型(GLM)、广义可加模型(GAM)以及随机森林3种物种分布模型(SDMs)方法,结合AIC(akaike information criterion)准则、累积偏差解释率和交叉检验等评判指标筛选和构建了双斑蟳栖息分布模型,并分析了环境因子对双斑蟳分布的影响.结果显示,3种模型在解释因子与响应变量间的关系上基本一致;其中GAM在模型拟合上具有优势,而随机森林的预测性能明显高于传统的GLM和GAM.双斑蟳相对渔获量在年份和月份间的变异性最为显著,两个因子的解释率分别在18%和3.8%以上.水深和表层盐度对双斑蟳资源分布的影响较大,均与双斑蟳相对丰度呈正相关关系;双斑蟳分布总体呈现冬季相对较高,夏季东北部海域高、西南部低的特点,与海州湾水深分布特点基本一致.本研究还根据FVCOM(finite-volume coasta ocean model)模拟环境数据,利用随机森林分布模型估计了双斑蟳在海州湾海域2011年各个季节的空间分布,为渔业资源的开发和保护提供依据.