Exploiting multi-trait genomic selection to improve the prediction accuracy of processing and end-use quality traits in hard winter wheat breeding.

Authors: GILL, HARSIMARDEEP - South Dakota State University; BRAR, NAVREET - South Dakota State University; HALDER, JYOTIRMOY - South Dakota State University; HALL, CODY - South Dakota State University; Seabourn, Bradford; Chen, Yuanhong - Richard; St Amand, Paul; Bernardo, Amy; Bai, Guihua; GLOVER, KARL - South Dakota State University; TURNIPSEED, BRENT - South Dakota State University; SUNISH, SEHGAL - South Dakota State University

Submitted to: The Plant Genome
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/1/2023
Publication Date: N/A
Citation: N/A

Interpretive Summary: Evaluation of end-use quality is expensive and labor extensive and also can only be done in later stages of wheat development. Indirect selection using genome-wide DNA markers proven to be an alternative for selecting end-use quality traits in earlier generations. To improve genomic prediction accuracy for complex traits, we evaluated different models for various end-use quality traits and optimized the multiple trait models. We found that incorporating simple traits like flour protein and flour sedimentation weight substantially improved the prediction accuracy of multi-trait models. By measuring rapid and inexpensive traits like flour protein, flour sedimentation weight, mixograph and baking, traits can be selected in earlier breeding generations to increase selection accuracy and genetic gains for end-use quality traits that are otherwise impossible to identify in earlier generations.

Technical Abstract: Improvement of end-use quality remains one of the most important goals in hard winter wheat (HWW) breeding. Nevertheless, evaluation of end-use quality is confined to later development generations owing to cumbersome phenotyping of these traits. Genomic selection (GS) has been suggested to select for end-use quality in earlier generations. However, lower prediction accuracy (PA) for complex traits remains a challenge in the implementation of GS. Multi-trait genomic prediction (MTGP) models can improve PA for complex traits by incorporating information on correlated secondary traits. However, optimization of the multi-trait (MT) models with the most effective combinations of secondary traits to predict traits of interest has not been well studied in HWW. We evaluated MTGP for various end-use quality traits that are otherwise impossible to phenotype in earlier generations. A set of 300 advanced breeding lines from 2015 to 2021 was genotyped using genotyping-by-sequencing (GBS) and evaluated for various end-use quality traits and used to evaluate MTGP models. The MTGP models outperformed the single-trait models showing up to a two-fold increase in PA for important traits. For instance, PA for bake absorption was improved from 0.38 to 0.75 using MTGP model and 0.32 to 0.52 for loaf volume. Further, we compared MTGP models by including different combinations of easy-to-score traits as covariates to predict complex traits. Our results suggest that incorporating simple traits like flour protein (FLRPRO) and flour sedimentation weight (FLRSDS) can substantially improve the PA of MT models. Thus, measurement of rapid and inexpensive traits like FLRPRO and FLRSDS can help to implement GP to predict Mixograph and baking traits in earlier generations. MTGP can provide breeders an opportunity for selection on end-use quality traits in earlier generations and cull inferior lines, thus increasing selection accuracy and genetic gains for end-use quality traits.