首页|期刊导航|生态学报|基于随机森林算法的鄱阳湖松门山砂坑长江江豚种群环境容纳量评估

基于随机森林算法的鄱阳湖松门山砂坑长江江豚种群环境容纳量评估OACHSSCD

The environmental capacity evaluation of the Yangtze finless porpoise based on the random forest in the Songmenshan sandpit of Poyang Lake

中文摘要英文摘要

评估种群在栖息地的环境容纳量,对栖息地管理、物种及其栖息地保护有着重要意义.基于随机森林算法构建的物种分布模型通过分析物种与各种环境因子之间的关系,广泛运用于探索生态位空间中物种的潜在分布与数量.使用随机森林算法,基于已有的鄱阳湖松门山砂坑水域内长江江豚(Neophocaena asiaeorientalis)分布数据与鱼类资源数据,构建长江江豚物种分布模型,分析长江江豚种群环境容纳量和分布状况,探讨不同鱼类物种对长江江豚分布的影响.模型估算的鄱阳湖松门山砂坑水域长江江豚种群环境容纳量约为 95 头,与当前监测到的种群数量(44 头)相比,研究区域内长江江豚仍然有较大的增长空间;与 2022 年鄱阳湖极枯水位长江江豚应急救护结果(111 头)相比,模型估算的种群环境容纳量偏低,可能是水位的变化导致鱼类资源可获得性发生变化,影响了栖息地种群环境容纳量.不同鱼类对模型贡献度的分析显示南方大口鲇(Silurus meridionalis)、飘鱼(Pseudolaubuca sinensis)、团头鲂(Megalobrama amblycephala)和长须黄颡鱼(Pelteobagrus eupogon)等对长江江豚的分布有较大的影响,意味着保持长江江豚种群的增长既要考虑小型适口鱼类的数量,也要关注大型鱼类与长江江豚在生态位上的竞争关系.本研究显示随机森林算法在长江江豚物种分布模型研究上的潜力,支撑长江江豚保护相关措施的出台,为长江江豚等濒危水生动物的物种分布模型的构建提供理论依据.

Assessing the environmental capacity is critically important for habitat management and the conservation of species and their habitats.In the context of the fishing ban in the Yangtze River basin,the status of the Yangtze finless porpoise(YFP)serves as an important indicator of the ecosystem's health and is a matter of particular concern.However,little is known about the potential distribution of its populations within their habitats.Species distribution models(SDM),constructed using the random forest(RF)algorithm to analyze relationships between species and environmental factors,are widely applied to evaluate species distribution and abundance.This method is suitable for analyzing how the complex structure of fish resources in Poyang Lake affects YFP distribution.In this study,data on the distribution of YFP and fish resources in the Songmenshan sandpit of Poyang Lake were used to construct SDMs for YFP based on RF.A total of 22 fish species were used as environmental variables in constructing the SDM.The SDM was applied to analyze environmental capacity and YFP distribution,and to investigate the influence of various fish species on this distribution.The SDM explained 56.53%of the variance in YFP distribution and abundance related to fish resources in the study area.Consequently,when employing SDMs to evaluate YFP population distributions,it is imperative to consider environmental variables,such as water depth,to enhance model efficacy.We propose to develop multiple SDMs to perform assessments of primary habitats and estimate the environmental carrying capacity for the YFP population.The environmental capacity for YFP in the study area,as estimated by the SDM,was 95 individuals,compared to the current surveyed population size of 44 individuals,indicating significant potential for population growth.This estimated environmental capacity was lower than the previously reported value of 111 individuals during the 2022 Yangtze finless porpoise emergency rescue operation.The discrepancy may be attributed to changes in water level,which affect fish availability and consequently influence environmental capacity.The analysis of the contribution of fish species to the SDM indicated that Silurus meridionalis,Pseudolaubuca sinensis,Megalobrama amblycephala,and Pelteobagrus eupogon significantly affected the distribution of YFP.This indicates that while increasing the abundance of small fish is beneficial,attention must also be paid to the competitive relationship between large fish and YFP to support YFP population growth.This study demonstrates the effectiveness of RF in constructing SDMs for YFP,supporting the implementation of targeted conservation measures.It also provided a theoretical basis for future SDM studies on endangered aquatic animals,including YFP.

饶榕城;李彩刚;朱国才;王生;翁佳桥;阙江龙;余进祥;闵佳玲;刘芳宁;黄怡;魏玲霞;尹恒;张韵;杨英

江西省水生生物保护救助中心,南昌 330029江西省水生生物保护救助中心,南昌 330029江西省水生生物保护救助中心,南昌 330029江西省水生生物保护救助中心,南昌 330029江西省水生生物保护救助中心,南昌 330029江西省水生生物保护救助中心,南昌 330029江西省水生生物保护救助中心,南昌 330029江西省水生生物保护救助中心,南昌 330029江西省水生生物保护救助中心,南昌 330029江西省水生生物保护救助中心,南昌 330029江西省水生生物保护救助中心,南昌 330029江西省水生生物保护救助中心,南昌 330029江西省水生生物保护救助中心,南昌 330029江西省水生生物保护救助中心,南昌 330029

鄱阳湖水生哺乳动物物种分布模型机器学习算法鱼类资源

Poyang Lakeaquatic mammalsspecies distribution modelmachine learning algorithmsfish resources

《生态学报》 2026 (9)

4856-4864,9

江西省科技重大专项项目(20252ABF010001)江西省农牧渔业科研指导性课题(2025-43)

10.20103/j.stxb.202506271633

评论