BreedToLast : Optimal selection and mating accounting for non-additive genetic effects, genome diversity and Mendelian sampling terms. populus

GWE is a disruptive methodology with potential of attaining high accuracies at short generation intervals. Last decade has seen GWE research focused on factors affecting prediction accuracies. Comparatively, little was done on impacts that this extra accuracy might have on genetic diversity at whole genome scale or on how to use efficiently Mendelian sampling (MS) variation. There are solutions from the pedigree-based era to optimize breeding for maximum gain over the long-term (OCS & MA), but these rely often on average information, like marker-based average coancestry. Thus, information of inbreeding or diversity variation across markers or genomic regions is neglected when constructing such constraints. The general objective here is to devise a breeding tool to improve coherently gain and diversity management with GWE, by accounting explicitly for the available marker information across genomes in the candidate population. For this, we will use quantitative genetics theory, computer simulations and demonstrations via proof-of-concept in poplar, maize and oil palm, for which mating is a key element in their breeding. Three developments will be integrated: (a) imputation expected to act synergistically with the other two; (b) prediction models with explicit non-additive components for predicting crosses; and (c) gain prospections through mate selection while observing diversity and MS terms across genomes.