欢迎访问《环境科学学报》编辑部网站！

快速检索			高级检索

研究报告

摘要：营养盐基准是湖泊富营养化控制与管理的基础，目前多采用以生态分区为基础的基准确定方法.由于同一生态分区内湖泊间的差异性，对区域性湖泊营养盐基准的适用性需要做进一步的探究.本研究提出了基于响应关系的聚类方法（RCA），包括：①基于线性混合效应模型和平均绝对百分误差的层次聚类过程；②基于赤池信息准则（AIC）的模型选择过程.基于长时间数据的可得性，选择我国云贵湖区和东部湖区作为对象，采用RCA对叶绿素a与营养盐（总氮和总磷）之间的响应关系进行分析.结果发现，区域性响应关系对应的AIC值均为最高，模型权重均小于0.001，且可能导致生态谬误；表明区域性湖泊营养盐基准不能完全适用于生态分区内全部湖泊.该研究可为我国湖泊营养盐基准的空间尺度确定和浓度限值识别提供参考.

Abstract：Nutrient criteria is the foundation of lake eutrophication control and management. To assess the applicability of ecoregional nutrient criteria, a relationship-based clustering approach (RCA) is proposed in this study. The RCA includes two major steps, including ①the hierarchical clustering based on the linear mixed-effect model and the mean absolute percentage error and ②the model identification based on the Akaike information criterion (AIC). Considering the availability of long-term data, the Yungui Plateau Ecoregion and the Eastern Plain Ecoregion were selected as the study areas. The RCA was then employed to explore the Chlorophyll a-nutrient (total nitrogen and total phosphorus) relationship. The results show that the ecoregional models have the highest AIC values among all the four groups of relationships with Akaike weights of less than 0.001. The ecoregional relationship may result in the ecological fallacy. We thereby concluded that the ecoregional nutrient criteria is not applicable to all the lakes within the ecoregion. It is believed that the modeling results and the RCA approach will enhance the understanding of proper spatial scale and values of nutrient criteria in China.

Key words：ecoregion nutrient criteria relationship-based clustering approach hierarchical clustering linear mixed-effect model