Comparison of similarity coefficients and clustering methods with amplified fragment length polymorphism markers in Colletotrichum gloeosporioides isolates from yam

Abstract

The choice of the similarity coefficient used in clustering could have great impact on the resulting classification, there is need to study and understand these coefficients better to be able to make the right choice for specific situations. In this study, variations caused by three similarity coefficients: Dice, Jaccard and Simple matching with five clustering methods: (Unweighted Pair-Group Mean Arithmetic (UPGMA), Weighted Pair-Group Mean Arithmetic(WPGMA), complete linkage, single linkage and Neighbour-Joining with AFLP markers in Colletotrichum gloeosporioides isolates from yam were assessed. Comparison among the similarity coefficients and clustering methods were made using correlation analysis, multidimensional scaling and principal component analysis. Dendrogram topology was compared using consensus fork index (CFI) and node counts. The grouping of the pathogens by the markers is not related to their agro-ecological zones. The CFI results showed varying level of similarity for the cluster analysis CA methods. It was observed that high correlation does not necessarily imply similarity in the topology of a tree, therefore care should be taken in its interpretation. The cophenetic correlation with original distances suggests that the UPGMA method gives consistent results with respect to grouping irrespective of the similarity coefficient. The use of UPGMA method is therefore recommended for its consistency