Genomic prediction applied to reciprocal recurrent selection scheme in tropical maize for nitrogen use efficiency

Abstract

One of the assumptions of diallel analysis is to accept parents as unrelated, what in practice does not occur and can lead to biased results. Moreover, a critical factor in a hybrid breeding program is the genomic prediction being performed based on the same allelic substitution effects for two complementary heterotic groups. It may lead to a convergent selection of parents and, consequently, increase both the relationship between groups and the endogamy. Combined, these events reduce the probability of obtaining hybrids with high heterosis through the cycles. In this sense, the aim is to identify the impact of the differential modeling of marker effects on heterotic groups in a maize breeding population though reciprocal recurrent selection. For that, mixed model equations will be used to conduct the phenotypic, diallel, and genomic prediction analyses. We will consider two datasets, USP and HELIX, for yield trait. The former was evaluated in two nitrogen regimes (low and ideal) over two growing seasons (2014 and 2015) in two locations. The latter was assessed in five locations over the 2014/15 growing season. Both were genotyped using an array with 660K SNP markers. Based on that, we will apply three prediction models considering: i) just one vector of additive effects for the whole population (standard method); ii) just one matrix of additive-dominant effects for the whole population; iii) a specific vector of additive effects for each heterotic group and for the interaction between them - North Carolina II design. From the results, we intend to evaluate the modeling effect on the prediction ability, the variability inside groups and the genetic distance between heterotic groups throughout selection cycles. Concerning the hybrid breeding, our outcomes may have strong potential to change the way of analyzing phenotypic and genotypic data when heterotic groups are considered toward greater prediction accuracies and avoidance of endogamy increase over the cycles. (AU)