4种多元聚合新策略在德国干草地类型划分中的应用

Application of Four PolytheticAgglomerative ClusteringStrategies to Dry Grasslands in Hopsten FRG

基于扩展了的Ｌａｎｃｅ和Ｗｉｌｉａｍｓ的多元聚合模型之上，现已提出了不少新的聚合策略。笔者用其中４种新策略，即新组内离差平方和法、新组内方差法、新组内平均距离法和加权平均距离法，对德国西北部Ｈｏｐｓｔｅｎ干草地进行了分类。结果表明，４种方法结果趋于一致，吻合度很高；所划分的类型与实际相符，生态意义明确，影响该草地分异的主要因素是土壤砂质含量和土壤保水能力，人类活动也是一重要影响因素。

Based on extended Lance & Williams model, many new clustering strategies have been introduced to ecologists. Four of them, sum of squares of new cluster, variance of new cluster, average distance of new cluster and weighted average distance were applied to the classification of dry grasslands in Hopsten, Germany. The results of these four methods are consistent. The community groups can be clearly interpreted by ecological knowledge; the contents of sand in soils and the capacity of keeping water in soils are the main factors which affect grasslands distribution. Additionally, human activities are also important to grassland communities.