基于CART算法的长江口鱼种丰富度预测  

Prediction of fish species richness in the Yangtze River estuary using CART algorithm

作　　者：戴黎斌 陈锦辉 田思泉[1,2,3,5] 高春霞 王家启[1,2,3] 杜晓雪 王学昉[1,2,3,5] DAI Libin;CHEN Jinhui;TIAN Siquan;GAO Chunxia;WANG Jiaqi;DU Xiaoxue;WANG Xuefang(College of Marine Science,Shanghai Ocean University,Shanghai 201306,China;National Data Centre for Distant-Water Fisheries of China,Shanghai 201306,China;Key Laboratory of Sustainable Exploitation of Oceanic Fisheries Resources,Ministry of Education,Shanghai 201306,China;Superintendence Department of Shanghai Yangtze Estuarine Nature Reserve for Chinese Sturgeon,Shanghai 200092,China;National Distant-water Fisheries Engineering Research Center,Shanghai 201306,China)

机构地区：[1]上海海洋大学海洋科学学院,上海201306 [2]中国远洋渔业数据中心,上海201306 [3]大洋渔业资源可持续开发教育部重点实验室,上海201306 [4]上海市长江口中华鲟自然保护区管理处,上海200092 [5]国家远洋渔业工程技术研究中心,上海201306

出　　处：《中国水产科学》2018年第5期1082-1090,共9页Journal of Fishery Sciences of China

基　　金：长江口中华鲟增殖放流跟踪监测和效果评估项目(170062);上海市科委地方能力建设项目(18050502000)

摘　　要：长江口是西太平洋最大的河口生态系统和典型的生态群落交错区,预测鱼类生物多样性对评价其生态系统有着重要的科学指示意义。结合2012―2013年长江口15个站点的渔业资源和环境调查数据,利用分类与回归树(CART)算法中的回归树算法,构建了长江口鱼种丰富度预测模型。基于1-SE准则,最优决策树的复杂性参数设置为0.067,结果表明,盐度、溶解氧和季节是影响长江口鱼类生物多样性的主要因子。此外,使用2014年的观测数据对回归树模型预测的长江口鱼种丰富度予以验证,均方根误差(RMSE)、平均相对误差(ARE)和平均绝对误差(AAE)值的统计结果显示,回归树模型在春、夏季的预测效能优于秋、冬季,模型总体上呈现出了较好的预测能力,表明利用CART算法对长江口鱼种丰富度进行预测是可行的。The main ecological patterns and functioning of estuarine ecosystems are difficult to evaluate owing to natural and human induced complexity and variability on biodiversity.Therefore,there is an increased demand to analyze and predict the relationships between the environment and the distribution of biota in estuarine ecosystems.Biodiversity is viewed as the variety of life,encompassing variations from the gene to ecosystem levels,and is commonly expressed as species richness.The patterns of biodiversity in the Yangtze River Estuary have remained largely unexplored,despite the increasing understanding of the importance of estuarine ecosystems and the existing knowledge on the variability of fish communities within estuaries and their environmental drivers.As a transitional system,the Yangtze River Estuary,a typical ecotone,is the largest estuarine ecosystem in the western Pacific Ocean.It establishes links between the marine and freshwater ecosystems in the East China Sea;persistent environmental fluctuations in this estuarine ecosystem creates considerable physiological demands on the species that inhabit this ecosystem.Predictive modelling techniques are being increasingly used to determine major habitat requirements that affect species distribution.Important technological advancements have benefited predictive distribution modelling,and new and sophisticated methods have been developed for use in statistical models that are applied to ecology.The prediction of fish biodiversity has important scientific implications for evaluating the Yangtze River Estuary ecosystem.Based on fishery and environmental data collected in 2012-2013,a regression tree model was built to predict fish species richness in the Yangtze River Estuary.The node structure of the optimal decision tree model indicated that salinity,dissolved oxygen,and month(i.e.season)were three factors affecting fish biodiversity in the Yangtze River Estuary.In addition,the data observed in 2014 was used to validate the predictive performance of the tree-base

关 键 词：分类与回归树 长江口 鱼种丰富度 预测 

分 类 号：S931[农业科学—渔业资源]