A comparison of clustering methods for river benthic community analysis

Four commonly used clustering methods (UPGMA, Ward Linkage, Complete Linkage and TWINSPAN) were compared in their ability to recognise the structure of three river macroinvertebrates datasets which were pre-determined based on habitat and biological characteristics or chemical water quality of sampling sites. DCA, NMDS and ANOSIM were applied to the same datasets to provide further information about data structure, and nonparametric tests were also undertaken on major chemical variables to justify the predeterminations. The modified Rand Index was used to measure the agreement between a particular solution and the pre-determined classification. The results showed that Ward Linkage performed best when its use was broadened and used with the CY Dissimilarity Measure, followed by TWINSPAN and Complete Linkage with UPGMA being least successful. There was evidence to suggest that the effectiveness of some clustering methods (e.g. UPGMA) may vary at different clustering levels, and simulation techniques which have been used to assess clustering methods could leave some properties of clustering methods unexamined.