Improvement of key agronomical traits in soybean through genomic prediction of superior crosses

Abstract

AbstractMaximizing yield is very important when developing new cultivars. However, yield must usually be improved jointly with other key traits, which can prove challenging when they are unfavorably correlated. Genomic predictions can facilitate the selection of promising lines among the progeny of crosses, but it may also help in selecting crosses that are more likely to produce improved lines by predicting progeny performance for the various key traits considered jointly. To assess whether genomic predictions of cross performance could help breeders simultaneously improve multiple traits, yield and maturity were predicted for 60,000 soybean [Glycine max (L.) Merr.] crosses. These predictions were then compared with the persistence of 101 biparental crosses throughout the selection process measured as the success in advancing progeny lines through to registration and commercialization. All but 2 of the 22 superior crosses retained by breeders had been predicted to display above‐average mean yield within different maturity windows. At the opposite end of the spectrum, 96.2% of all crosses predicted to produce progeny with a below‐average mean yield within a specific maturity window were eliminated during selection. Our results therefore suggest that by making crosses predicted to produce progeny meeting target requirements for multiple key traits, breeders could either achieve the same genetic gains with fewer resources or invest the same resources on a more promising set of crosses and thereby achieve greater gains.