Prediction of Maize Hybrid Performance Using the Information On the Alikeness in State and Coancestry Among Relatives.

The identification of better hybrid combinations demands lots of efforts from breeders and financial resources for the realization of the crosses and evaluation of the hybrid combinations that may have satisfactory performance on maize breeding programs. In this context, some tools have been arising and can help the breeders work. Among these, information on the state of similarity and relationship between relatives obtained by use of molecular markers can be used in the best linear unbiased predictor (BLUP) as a method of non evaluated hybrid prediction. The aim of this study was to evaluate the BLUP efficiency and the influence of the use of information on the state of similarity and relationship between relatives on non evaluated hybrid prediction. It was used nine inbred lines which were genotyped with 48 microsatellite markers (SSR) that were associated to QTLs for  grain productivity. These inbred lines were crossed in a complete diallel scheme, and single hybrids were obtained for the correlation studies. Information on the alikeness in state and coancestry among relatives were used to construct the additive and dominance genetic matrices that were used for the prediction of the genotypic values on BLUP and the non evaluated hybrids specific combining ability (SCA). Five unbalanced degrees were simulated (3, 6, 9, 12 and 15 withdraw hybrids). The information on the alikeness in state compared to coancestry among relatives had a superior performance on the correlation of the better hybrids and selection efficiency, demonstrating that it was more efficient on hybrid ranking, besides it would have a more practical application for breeders. The relationship between relatives showed lower values for PRES, demonstrating that the obtained predictions are closer to the observed value, showing that they were more accurate than the one obtained from the information on the state of similarity. The hybrid ranking difference might be due to the lowest values of similarity detected among the inbred lines; which leads to the inference that they are from different genetic backgrounds. In this context, although it has a bias due to overestimation of the relationship, the similarity in the state can be used when it is detected lower values of relationship between the parents which hybrid performance one wants to predict.