Predictability of maize grain yield using various biometrical methods helps in formulating a breeding program for its improvement, due to its potentials in solving the food security challenges across the globe. Twenty-one genotypes of maize were evaluated at the experimental field of Federal University of Agriculture, Abeokuta, Nigeria, in 2013 and 2014, with the aims of examine the employability of discriminant analysis in studying the grain-yield of selected maize genotypes and predicting model for the grain-yield. The experiment was laid out in randomized complete block design with three replications. Eight agronomic characters were measured from flowering to harvesting. Data collected were subjected to analysis of variance and stepwise multiple linear regression analysis to predict a model for grain yield. Principal component analysis was also performed to identify traits with overall significant contribution to variation. Correlation coefficient and discriminant analyses were performed to determine the level of association and relatedness of the observed traits to grain-yield. Significant (P ≤ 0.01) differences were observed for most of the characters. Grain-weight, plant height, ear height and number of days to silking contributed more to variation among the genotypes while regression analysis predicted grain yield from grain-weight, plant height and number of days to silking. Correlation coefficient and discriminant analyses revealed that grain-yield, grain-weight, ear height, plant height and days to anthesis were highly related. The study showed that grain-weight, days to silking, plant height could serve as part of good indices for selecting high grain-yield in maize breeding programme.

Keywords: Association. Zea. Univariate. Discriminate.