摘要
总初级生产力(GPP)是陆地生态系统碳汇的重要组分之一,明晰GPP的变化规律及其影响机制,对深入认知生态系统碳循环过程有着重要意义。然而,目前鲜有研究基于多站点通量监测阐明不同类型生态系统GPP的季节变化特征及其主要影响因素。基于ChinaFLUX,沿着东西水热梯度分布的草地样带选择内蒙古温带草原(内蒙古站)、高寒灌丛-草甸(海北站)和高寒草地(当雄站)作为研究对象,沿着南北森林样带选择长白山温带针阔混交林(长白山站)、千烟洲亚热带常绿针叶人工林(千烟洲站)和鼎湖山亚热带常绿针阔混交林(鼎湖山站)作为研究对象,结合机器学习等统计方法,探讨GPP的季节变化特征、主要影响因素以及甄别不同生态系统间的差异性。研究发现,(1)3种典型生态系统GPP的季节动态整体呈单峰分布,且森林生态系统GPP高于草地和灌丛。(2)全子集回归等分析发现,草地生态系统GPP与土壤含水量(C_(SW))、叶面积指数(LAI)和气温(t_(a))呈正相关,与饱和水汽压差(D_(vp))呈负相关,而灌丛和森林生态系统GPP的影响因子主要为t_(a)和光合有效净辐射(R_(PA)),与水分因子的关系较弱。在季节干旱期间,千烟洲站GPP与C_(SW)和D_(vp)存在显著的相关关系。(3)方差分解和随机森林结果表明,水分是草地生态系统GPP季节变异的主导因子,其解释比重高于30%;温度是高寒灌丛(海北站)和温带落叶针叶林(长白山)生态系统GPP变异的主导因子,其解释比重接近或高于50%;辐射是亚热带森林生态系统(鼎湖山和千烟洲站)GPP变异的主导因子,其解释比重高于50%。在季节性干旱期间,水分因子对千烟洲站GPP变异的解释比重高于对其他灌丛和森林站点的。因此,在中国不同类型生态系统中,生长季GPP变异的驱动要素存在明显的异质性,其中在干旱和半干旱草地生态系统中,水分(C_(SW)和D_(vp))是GPP变异的主要影响要素,而在森林和灌丛生态系统中,温度或辐射为GPP的主要驱动要素。该研究有助于深入认识GPP的变化趋势及其影响要素,也可为中国碳循环的准确模拟提供数据基础和参数依据。
Gross primary productivity(GPP)plays an important role in driving the global carbon sinks.Clarifying the variability of GPP and its impact mechanisms is crucial for predicting the future terrestrial carbon balance.However,those mechanisms are still largely unknown in the literature.In this research,we used the observation data from six eddy covariance sites(including DX,NM,HB,CBS,QYZ and DHS)based on ChinaFLUX to represent three types of ecosystems(forest,grassland,and shrublands).We applied multiple statistical methods(e.g.,machine learning)to explore the seasonal variability in GPP and its driving factors.The results showed that the seasonal dynamics of GPP displayed an obviously unimodal seasonal cycle in all of the three ecosystems.The GPP in forest was larger than that in grassland and shurbland.The complete subset regression results showed that the soil water conten(C_(SW)),leaf area index(LAI),and air temperature(t_(a))were positively correlated with GPP in grassland ecosystems,while the vapor pressure deficit(D_(vp))had negative effects on GPP.t_(a) and photosynthetically active radiation(R_(PA))played a comparatively more important role in driving the seasonal variability of GPP in forest and shurbland ecosystems.Our random forest and variance partitioning analysis revealed that more than 30%of the seasonal variability in GPP can be explained by C_(SW)in grassland sites.In the forest and shurbland sites,the relative contribution of t_(a) or R_(PA)to GPP was greater than 50%.In addition,water conditions(moisture related factors)significantly influenced GPP in QYZ during seasonal droughts and the contributions of water factors were greater than those in the other forest and shurbland sites.Therefore,the driving factors of the seasonal variability in GPP showed strong ecosystem heterogeneities.Among arid and semi-arid grassland ecosystems,the dominant driver of GPP was water availability,but were temperature and radiation in forest and shurland ecosystems.This study improves the existing understanding of the dynamics of GPP and its driving factors in the three typical biomes in China,and also provides a basis for the accurate simulation of China’s carbon cycle.
作者
姜鹏
秦美欧
李荣平
孟莹
杨霏云
温日红
孙沛
方缘
JIANG Peng;QIN Mei’ou;LI Rongping;MENG Ying;YANG Feiyun;WEN Rihong;SUN Pei;FANG Yuan(The Institute of Atmospheric Environment,China Meteorological Administration,Shenyang 110166,P.R.China;China meteorological Administration training center of Liaoning,Shenyang 110166,P.R.China;Regional Climate Center of Shenyang,Shenyang 110016,P.R.China;China Meteorological Administration Training Center,Beijing 100081,P.R.China)
出处
《生态环境学报》
CSCD
北大核心
2022年第4期643-651,共9页
Ecology and Environmental Sciences
基金
中国气象局沈阳大气环境研究所联合开放基金课题(2021SYIAEKFMS35)
中国气象局气象干部培训学院2021年科研项目(2021-13)
2020年度辽宁省气象局博士科研专项(D202003)
辽宁省气象局科研项目(BA202003)干旱气象科学研究基金项目(IAM202010)。
关键词
GPP
季节变异
影响要素
典型生态系统
机器学习
异质性响应
GPP
seasonal variations
influencing factors
typical ecosystems
machine learning
heterogeneous responses