【摘要】 相似性系数的选择可能会影响生物集群间的β多样性分析。Bray-Curtis相似性系数是目前使用最广的系数之一,还有一些研究使用了欧氏距离等系数。通过非度量多维标度的结果直观对比了两个相似性系数在处理生物数据时的结果,以便了解两个系数的差异。分析结果表明,Bray-Curtis相似性系数比欧氏距离更适合处理包含很多零值的生物数据。生物数据中的零值与环境数据中的不同,不应该影响集群间相似性的计算。如果遇到完全没有生物的集群,可以选用零调整的Bray-Curtis系数,或者剔除此类集群后再计算Bray-Curtis相似性系数。一些研究忽略了相似性系数的差异,在比较和分析多元分析结果时应该慎重,以免在生态环境评价时被误导。更多还原

【Abstract】 Bray-Curtis similarity coefficient has become one of the most widely used indexes for the analysis of β-diversity among biological assemblages or communities. Euclidean distance is also employed by some scholars. This study aims to determine which coefficient is better for biological data. Twelve simulated assemblages were built to compare the difference between Bray-Curtis coefficient and Euclidean distance, and the MDS results were used to show their visualized difference. In order to reduce the influence of the dominance species on the quantitative coefficients, the abundance data were transformed with a square root prior to calculating MDS. The results showed that there was obvious difference between the two coefficients. Compared with the Euclidean distance, the Bray-Curtis coefficient was more suitable for dealing with the biological data. This difference was resulted from most possibly the zero value in the biota dataset. The Bray-Curtis coefficient showed the obvious advantage in dealing with this kind of zero value while Euclidean distance was not suitable. When some samples had no species, which were not rare events in field surveys nowadays, the zero-adjusted’ Bray-Curtis measure could be chosen, or these samples should be deleted prior to calculation. In addition, the default coefficient settings of some popular software are usually Euclidean distance, rather than Bray-Curtis coefficient. Therefore, attentions should be paid to the software selection and their settings when different results were analyzed or compared.更多还原