Application of Multivariate Analysis for Genotype Evaluation in Soybean

Abstract

Multivariate analysis is a frequently used approach in breeding studies. Applied to multiple trait data, Principal Component Analysis (PCA) gives an opportunity to graphically display the relationships between important agronomic traits, evaluate the varieties on the basis of multiple traits, identify valuable breeding sources and recommend possible selection strategies. This study was conducted to evaluate agronomic performance of 31 early varieties (maturity group 0) from the soybean collection of the Maize Research Institute “Zemun Polje” (Belgrade, Serbia). PCA analysis was applied to identify the best performing genotypes considering multiple traits (seed yield, major yield components and parameters of technological quality of grain), and to determine the level of trait interdependence. Genotypes were tested in a randomized complete block design with three replications during two years (2011 and 2012) at two locations in Serbia (Zemun Polje and Pančevo). An analysis of variance (mixed model) exhibited significant effects of genotype (G), environment (E), and genotype × environment (G × E) for most of the traits. According to PCA biplot, seed yield per plant was in positive correlation with number of seeds and pods per plant, indicating that those yield components could be effective selection criteria in breeding for seed yield. Comparison of early-maturing genotypes based on multiple traits revealed two varieties with above-average seed yield per plant and four varieties with high oil and above average protein content, which could be used as potential parents in breeding for important agronomic traits in future breeding programs.