Path analysis under multicollinearity in soybean

Abstract

This study aimed to establish the phenotypic correlations among several soybean traits with grain yield in direct and indirect effects using path analysis, and to compare alternative methods for minimizing the adverse effects of multicollinearity in estimating path coefficients. The experiment was conducted in greenhouse in a randomized complete block design with four replications. Nine soybean genotypes belonging to three seed size categories were used. The correlation studies and the path analysis showed that the seed size was not important for increased yield. The number of nodes and plant height at maturity showed significant correlation with grain yield. Using the least square methodology, the results obtained by path analysis under multicollinearity were not satisfactory. The ridge path analysis and the trait culling were efficient in reducing the adverse effects of multicollinearity. Both methods showed that only the number of nodes at maturity trait had a high direct effect on grain yield per plant.