欢迎访问《应用生态学报》官方网站,今天是 分享到:

应用生态学报 ›› 2020, Vol. 31 ›› Issue (6): 2076-2086.doi: 10.13287/j.1001-9332.202006.034

• 研究报告 • 上一篇    下一篇

基于集成模型的小黄鱼越冬群体适宜生境及其环境影响因素

刘尊雷1,2, 杨林林1,2, 袁兴伟1,2, 金艳1,2, 严利平1,2, 程家骅1,2*   

  1. 1中国水产科学研究院东海水产研究所, 上海 200090;
    2农业农村部东海渔业资源开发利用重点实验室, 上海 200090
  • 收稿日期:2019-11-08 出版日期:2020-06-15 发布日期:2020-06-15
  • 通讯作者: * E-mail: ziyuan@sh163.net
  • 作者简介:刘尊雷, 男, 1982年生, 副研究员。主要从事渔业资源评估与管理研究。E-mail: liuzl@ecsf.ac.cn
  • 基金资助:
    农业农村部近海渔业资源调查项目和农业农村部中日暂定水域渔业资源调查项目(2015—2017)资助

Overwintering distribution and its environmental determinants of small yellow croaker based on ensemble habitat suitability modeling

LIU Zun-lei1,2, YANG Lin-lin1,2, YUAN Xing-wei1,2, JIN Yan1,2, YAN Li-ping1,2, CHENG Jia-hua1,2*   

  1. 1East China Sea Fisheries Research Institute, Chinese Academy of Fishery Sciences, Shanghai 200090, China;
    2Key Laboratory of East China Sea Fishery Resources Exploitation and Utilization, Ministry of Agriculture and Rural Affairs, Shanghai 200090, China
  • Received:2019-11-08 Online:2020-06-15 Published:2020-06-15
  • Contact: * E-mail: ziyuan@sh163.net
  • Supported by:
    This work was supported by the Ministry of Agriculture and Rural Affairs Fisheries Resource Survey in China Sea and Fisheries Resource Survey In Provisional Measures Zone between China and Japan (2015-2017).

摘要: 小黄鱼是中韩渔业共同利用鱼种,其跨界洄游习性限制了对越冬场范围的调查和评估,导致对越冬群体适宜栖息地分布缺乏了解。本研究基于越冬期我国自然海域的物种分布点位数据和5个环境数据,运用8个物种分布模型(SDM)分析了小黄鱼越冬场分布范围,采用5折交叉验证,利用受试者工作特征曲线下面积(AUC)评价模型预测性能,并通过加权集成方法构建综合生境模型预测越冬场核心分布位置。结果表明: 出现/未出现数据模型预测准确度普遍高于仅出现模型;在出现/未出现数据模型中,机器学习方法预测准确度高于经典回归模型,支持向量模型(SVM)准确度最高(AUC=0.85),广义线性模型(GLM)准确度最低(AUC=0.73)。集成模型AUC较单一独立模型的准确度有所提升,表明集成模型能有效降低单一独立模型所带来的不确定性,提高模型预测准确度。变量重要性分析结果显示,盐度和温度是决定小黄鱼越冬场地理分布的重要因素,适宜分布区集中在黄海南部外海、东海北部外海和浙江省沿岸海域,而黄海南部沿岸海域和东海中南部外海为不适宜越冬区。研究结果为预测小黄鱼潜在越冬场提供了理论基础,可支撑越冬场渔业资源的空间规划和可持续利用。

Abstract: Small yellow croaker is a trans-boundary fish resource shared by China and South Korea. Information on the distribution and preferred habitats of overwintering populations is lacking, parti-cularly in southern waters of Yellow Sea where the species is regulated together by China and South Korea. We simulated the geographic distribution under current condition with eight species distribution models (SDM) based on the presence-absence data and five environmental variables. The performance of model’s prediction was evaluated using the area under the receiver operating characteris-tic curve (AUC) based on 5-fold cross-validation. Ensemble SDMs were constructed using a weighted average of eight habitat suitability model types to identify core areas with high probability of small yellow croaker occurrence. The results suggested that predictions based on presence-absence data generally perform better than those based on presence-only data and classical regression models under-performed compared to machine learning approaches. Among all the approaches that supported presence-absence data, support vector machine was the best performing technique and GLM was the worst. The ensemble model outperformed individual SDM models, demonstrating higher effectiveness of ensemble modelling approaches than individual models in reducing the predictive uncertainty. Salinity and temperature were important factors in predicting the overwintering distribution of small yellow croaker. The core areas with high probability of occurrence were concentrated in three areas, the open waters of southern Yellow Sea, the open waters of northern East China Sea, and the coastal sea of Zhejiang Province. Coastal waters in southern Yellow Sea and open waters in central and southern East China Sea were not suitable for overwintering of small yellow croaker. Our results provided a basis for predicting the potential overwintering distribution to guide spatial planning in support of sustainable utilization of small yellow croaker.