Modeling and determination of relationship between kernel yield and its related traits in maize inbred lines and their hybrids using multiple regression and path coefficient analysis

ABSTRACT: For to evaluation and modeling of the relationships between yield and its related traits of maize, forty one inbred lines with fourteen selected hybrids resulted from crossing between the inbred lines were grown under two different geographical locations. Randomized Complete Block Design (RCBD) with four replications was used in both experiments. Multiple regression and stepwise regression and also path coefficient analysis were used to study the relationships. Since variance inflation and tolerance indices in regression model with all measured traits showed existence of some multicollinearity in regression model, stepwise selection technique was used to introduction best modelers of kernel yield. This model selected tassel length (TL), number of tassel branches (TB), plant height (PH), leaf area (LA) and 1000- kernel weight (1000-KW) as the most important traits contributing kernel yield so that plant breeders can use these traits as indirect selection indices to increase kernel yield. Besides, predicting kernel yield with these traits showed little differences between real value of yield and its predicted value by model kernel yield of maize, predicted by this model. The assumptions on residuals such as their normal distribution were reliable. Regression coefficient and direct effect of TL was negative, indicating negative effect of this trait on yield therefore breeders may consider this trait as a limiting factor for yield. LA and 1000- KW had highest direct effect on yield and they may have the higher effects on kernel yield.