Appropriate statistical methods for analysis of safflower genetic diversity using agglomerative hierarchical cluster analysis through combination of phenotypic traits and molecular markers

Combining phenotypic and genotypic germplasm characterization is a key of efficient and successful safflower breeding program by identifying valuable and confirmed parents. This study aimed to investigate and use appropriate statistical methods for such a characterization, and to identify potential genetic pools in safflower germplasm that may be useful for breeding program implementation. The genetic diversity of 45 accessions from different countries, provided by USDA-ARS, was assessed, during two cropping seasons, using agro-morphological traits and ISSR molecular markers. Agglomerative hierarchical cluster analysis (AHCA) was used with appropriate similarity distances, and Ward and UPGMA linkages. Agreement between distance-linkage combinations was evaluated using cophenetic correlation, Mantel test, Fisher exact test, Cramer's V, overall accuracy, and Cohen's kappa. Both agro-morphological phenotyping and molecular genotyping revealed significant genetic diversity. Ward linkage was better than UPGMA, using simple matching distance for molecular markers and Gower distance for phenotypic traits as well as for combined phenotypic traits and molecular markers. It delineated the studied accessions into four main clusters. Some accessions showed desirable profiles that can be used in future breeding programs. This is the first report of a series of appropriate statistical methods that can be used for assessing genetic diversity in safflower, combining phenotypic traits and molecular markers, and thus identifying relevant genetic pools for breeding program. This article is protected by copyright. All rights reserved