Identification of Genomic Regions from an Exotic Germplasm Source Under Breeding Selection and Insight into Genomic Prediction for Yield in Soybean

Expanding genetic diversity in U.S. soybean breeding populations has become a necessity to improve yield. Over 100 high yielding lines developed at eight public institutions derived from exotic PI416937 have been entered into USDA Tests over the past 15 years. Using SoySNP50k Infinium Chips, we attempted to understand the contribution of PI416937 to these high yielding lines. Using a prior yield region originating from PI416937 to define a significance threshold, we discovered eight regions under positive selection. Interestingly, PI416937 regions under strong positive selection are targets for introgression into low diversity regions related to the historical bottleneck associated with North American introduction. Using SoySNP6k Infinium Chips, we genotyped four RIL populations (total 336 elite breeding lines) containing PI416937 in their pedigrees to investigate regions under selection during early breeding stages before advanced yield testing. We found four regions with significant evidence for positive selection, though none of these regions were discovered in the aforementioned analysis. This may indicate that regions significantly associated with agronomic performance are often population as well as environment specific. Utilizing the aforementioned RILs and over 150 diverse advanced lines, we also explore optimization of a genomic prediction model for yield via Ridge Regression BLUP. We were able to achieve cross validation prediction accuracies of 0.57 when combining all material. Our ability to predict separate test sets was significantly lower. We finally are looking to combine the two discussed approaches for improving yield by including regions under selection from PI416937 as fixed effects for improving prediction accuracy.