聚类分析法在高寒草甸生态系统健康评价中的应用——以青海玉树县为例

Application of Cluster Analysis in the Health Assessment of Alpine Meadow Ecosystem: A Case Study in Yushu County of Qinghai, China

利用离差平方和法,对青藏高原玉树县72个样方及样方指标进行了聚类分析。结果表明,当欧式距离系数为5时,可将72个样方聚为8类;当欧式距离为10时,可将72个样方聚为4类,样方指标聚为7类。样方指标聚类结果与主成分分析法、双变量相关性分析结果比较一致,植被初级生产力是高寒草甸生态系统健康评价的主要指标依据,而且土壤理化性质与植被群落初级生产力呈显著相关(P≤0.05)。以上结果表明:系统聚类分析法能比较合理的将不同健康水平的样方进行分类。

Seventy-two sites and sites indexes were clustered by means of ward method. The results showed that 72 sites of grassland were clustered into 8 classes when Euclidean distance coefficient was 5. When Euclidean distance coefficient was 10, 72 sites and sites indexes of grassland were clustered to 4 classes and 7 classes, respectively. Hierarchical cluster results of sites indexes were in accordance with principal component analysis and bivariate correlation results. Vegetation primary productivity which was significantly correlated to physical and chemical properties of soil was important to evaluate alpine meadow ecosystem. The above findings suggested that hierarchical cluster could classify different health level grassland sites reasonably.