Validation of cross progeny variance genomic prediction using simulations and experimental data in winter elite bread wheat

Abstract

AbstractThe utilization of genomic prediction is increasing in crop breeding to parental selection and mating. The Usefulness Criterion (UC) that considers Parental Mean (PM), progeny Standard Deviation (SD) and selection intensity has been shown to increase the likelihood to get outstanding progenies compared to mating using PM alone while maintaining more diversity in the germplasm for next generations.This study estimates our ability to predict UC and its two components (PM and SD) using simulations and experimental data (73-101 winter bread wheat crosses depending on the trait, with 54.8 progenies on average) including heading date, plant height, grain protein content and yield evaluation. The training population comprises 2,146 French varieties registered during the last 20 years and INRAE-AO breeding lines.According to simulations, prediction ability increases with heritability and progeny size and decreases with QTL number, most notably for SD. We used as a reference a TRUE scenario,i.e. an optimal situation where TP is infinite and where marker effects are perfectly estimated. SD was strongly impacted by the quality of marker effect estimates. In simulations, considering the error in marker effect estimates improved SD predictions for quantitative traits with low heritability. In experimental data, the interest of this method was limited.PM and UC were reasonably predicted for all traits, while SD was more challenging. This pioneering study experimentally validates genomic prediction of progeny variance. The ability of prediction depends on trait architecture while the realization of cross potential in the field necessitates a sufficient number of progenies.Key messageFrom simulations and experimental data, the quality of cross progeny variance genomic predictions may be high, but depends on trait architecture and necessitates sufficient number of progenies.