Managing Global Resources for a Secure Future

2017 Annual Meeting | Oct. 22-25 | Tampa, FL

64-2 Integration of Genomics-Assisted Selection into Soybean Breeding to Accelerate Genetic Gain.

See more from this Division: C07 Genomics, Molecular Genetics and Biotechnology
See more from this Session: Symposium--Next Generation Trait Mapping & Molecular Breeding for Accelerating Genetic Gains

Monday, October 23, 2017: 10:35 AM
Tampa Convention Center, Room 21

Benjamin Stewart-brown, Institute of Plant Breeding, Genetics and Genomics, University of Georgia, Athens, GA, Justin Vaughn, USDA-ARS, Athens, GA, Qijian Song, USDA-ARS, Beltsville, MD and Zenglu Li, Institute for Plant Breeding, Genetics and Genomics & Department of Crop and Soil Sciences, University of Georgia-Athens, Athens, GA
Abstract:
Increasing soybean yield potential is the ultimate goal for soybean breeding programs. Improved genetic yield potential is critical to the success of any new soybean cultivar and the main component that drives profitability of soybean. Utilizing genomic technology will help improve the rate of genetic gain in soybean yield. One of the current challenges to plant breeders remains identifying functional markers for traits of interest and effectively integrating genomics-assisted selection approaches into plant breeding programs in a cost-effective and efficient way to unlock genetic potential and improve breeding selection efficiency that leads to significant soybean improvement. In this study, we attempted to employSoySNP6k Infinium Chips to genotype RIL populations and diverse advanced lines which have been yield tested to build a genomic prediction reference population. A genomic prediction model for yield and seed composition traits were built using Ridge Regression BLUP method. Using this reference population, we were able to achieve a cross validation prediction accuracy of 0.60 for yield across populations when predicting using all material in the training population, but we illustrate this is deceiving and inflated due to population structure. To negate this inflated prediction accuracy, we predicted each RIL population separately. Prediction accuracy for predicting “within population” was often greater than “across populations” when the training population size was held constant, but varied depending upon the population. Prediction accuracies for more heritable seed composition traits tended to be higher. We also investigated how increasing training population size can improve prediction accuracy and the impact of G x E on the potential to properly implement genomic selection within a soybean breeding program. We will also exemplify the use of trait markers for effective selection in a soybean breeding workflow to accelerate genetic gain.

See more from this Division: C07 Genomics, Molecular Genetics and Biotechnology
See more from this Session: Symposium--Next Generation Trait Mapping & Molecular Breeding for Accelerating Genetic Gains