Studies on the multivariate statistics oriented genetic parameters and their uses

Abstract

Multivariate statistical methods have rarely been used in the plant breeding field. Their intricate theories have made themselves unappealing to most experimentalists. However, recent advancements of electronic computers virtually eliminate the problems associated with complicated calculations required to execute a multivariate analysis. Therefore, in this study, attention is paid primarily to the evaluation of one of the multivariate statistical methods, Factor Analysis, as a new breeding tool. A new concept of hereditary parameters and a new index were proposed. As the proposed method is intended for the supplement of a Smith-Hazel selection index, the efficiency and the consistency of the selection index were also examined. The Smith-Hazel selection index constructed for maize demonstrated its efficiency and pronounced robustness under the violation of one of the hypotheses--correct parameters must be known. For the evaluation of the multivariate analysis oriented method, two sets of paired sorghum subpopulations as well as their original mother population were analyzed by a factor analysis method and heritability like parameter vectors were derived from their factor scores for each population. ...