Genomic prediction in maize breeding

Abstract

In recent years the availability of cheap genome-wide markers resulted in a novel approach in maize breeding called genomic prediction (GP). Genomic prediction is a special form of marker assisted breeding, in which genetic markers covering the whole genome are used for estimation genomic breeding value of individuals under selection. Genomic estimated breeding value (GEBV) of an individual represents the sum of effects associated with all the marker alleles present in an individual and included in the GP model applied to the population under selection. The GP is based on training (TP) and breeding (BP) populations. The TP is used for training of the GP model and for estimation of the marker effects required for estimation of GEBV of the individuals in the BP. The BP is the population subjected to GP for identification of the superior lines for use as parents for new hybrid combinations. In real maize breeding programs, successful implementation of the GP is based on data from multi-environment trials. Thus, appropriate modeling and statistical approaches are required to deal with the complexity of the multi-environment grain yield data coupled with genomic and environmental data to speed-up maize breeding.