HASS cluster analysis: a new method of grouping genotypes or environments in plant breeding

Abstract

A new method of cluster analysis, termed the hierarchical agglomerative sums of squares method (HASS), is proposed to facilitate clustering of genotypes or environments where genotype x environment interaction exists. The method is a modification of that proposed by Lin (1982), who used the equivalent of the genotype x environment interaction mean squares for clustering. Lin fused on the minimum genotype x environment interaction mean square within potential new clusters at each fusion cycle. HASS clustering, however, uses the pooled genotype x environment interaction sums of squares within all clusters (new and old) at each fusion cycle. HASS clustering is shown to more nearly achieve the objective of minimizing the pooled interaction mean square within clusters and maximizing the interaction mean square among clusters when compared with Lin's (1982) method and the widely used average linkage method.