Genetic similarities obtained from RAPD data were used to create a cluster diagram. Cluster analysis based on Nei and Li’s (1979) similarity coefficients using UPGMA grouped 16 cultivars into 5 main clusters (Fig. 2) at the similarity coefficient of 0.82. Cluster I consisted of three sub-clusters and the majority of cultivars were placed in it. In this cluster the coefficient of similarity ranged from 0.73 (‘Tabesh’ vs ‘Sepidrood’) to 0.98 (‘Ghaem-1’ vs ‘Ghaem-2’) indicating relatively less divergence among the cultivars of this cluster due to originating from closely related ancestors. For example, both ‘Ghaem-1’ and ‘Ghaem-2’ developed from the cross between ‘Sepidrood’ and ‘Sange-Jo’. So, it is necessary to avoid crossing between cultivars of this cluster.
The remaining cultivars ‘Khazar’, ‘Ghaem-3’, ‘Pouya’ and ‘Hashemi’ individually formed next clusters which were more isolated from the other cultivars. Low similarity indices were observed between ‘Pouya’ and ‘Hashemi’ (0.59) and among ‘Hashemi’ with either of cultivars ‘Neda’, ‘Champa’, ‘Amol-3’ with 0.61 similarity index which indicated more divergence. Crossing between the genotypes with low similarity coefficient will manifest high heterosis. As regards to higher quality status of ‘Hashemi’ among cultivars, it is essential that to pay more attention to it in hybridization programs.