The use of multivariate statistics in breeding: discussions of common mistakes

Abstract

Predictive methods of genetic diversity have been widely used in recent years mainly through the use of multivariate statistical methods, since with these methods the multiple information of each cultivar is expressed in measures of dissimilarity, which represent the existing diversity in the set of analyzed accesses. However, there are several errors and mistakes that can be made in the wrong choice of one of the methods that cover multivariate analysis. Two of them are: wrong choice of the hierarchical grouping method and the error in deciding on the ideal number of groups in the final stage of cluster analysis.In this context, the main objective of this article is to discuss the main questions that occur in the analysis of the groupings obtained, suggesting some solutions.