Abstract
Genetic control and heterosis depend directly on genetic divergence among the parents in generating promising hybrids
required by plant breeders. The purpose of this study was to determine the relative importance of heterosis, combining
abilities, regression and correlation estimates in order to develop hybrid cultivars in maize. The 28 F1 hybrids obtained by
partial diallel cross of 8 inbreds in a randomized complete block design were evaluated at the Lower Niger River Basin
Authority, Oke-Oyi, Nigeria in three years. General (GCA) and specific combining abilities (SCA) produced significant (P
< 0.01) effects for all the characters, while non-additive genetic effects were predominant. The levels of heterosis for grain
yield varied widely among crosses, ranging from -16.83 to 9.76%. Positive and significant genotypic and phenotypic
correlations among grain yield and some related characters (days to anthesis and silking; plant and ear heights; number of
ears plant-1 and 1000 seed weight) showed that each character could be used indirectly to selection of grain yield. These
results also indicated that SCA was more effective than heterosis for describing hybrid performance. The regression of
actual hybrid characters on the expected hybrid characters, based on parental GCA values, was highly significant.
Regression analysis also showed that all the agronomic characters jointly contributed 19.4% to grain yield of maize. The
inbreds (Pop 66 SR and 2000 EV DT-Y STRC4) and crosses (Pop 66 SR x, 2000 EV DT-Y STRC4, Pop 66 SR x KU1409
and 9006 x KU1409) featured prominently with respect to better GCA × Year and SCA × Year effects with high heterotic
values for maize grain yield and associated characters. These inbreds could serve as donors to obtain early and short
statured hybrids with higher yield, while the crosses may be exploited and adapted to the Nigerian Savannas. This study
also affirmed that genotypic and phenotypic correlation coefficients as well as combining abilities, heterosis and regression analyses were found to be suitable models for yield improvement in maize breeding.
