AUTHOR=Gill Harsimardeep S. , Halder Jyotirmoy , Zhang Jinfeng , Brar Navreet K. , Rai Teerath S. , Hall Cody , Bernardo Amy , Amand Paul St , Bai Guihua , Olson Eric , Ali Shaukat , Turnipseed Brent , Sehgal Sunish K. 

TITLE=Multi-Trait Multi-Environment Genomic Prediction of Agronomic Traits in Advanced Breeding Lines of Winter Wheat

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 12 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2021.709545

DOI=10.3389/fpls.2021.709545

ISSN=1664-462X

ABSTRACT=Genomic prediction (GP) is a promising approach for accelerating the genetic gain of complex traits in wheat breeding. However, increasing the prediction accuracy (PA) of GP models remains a challenge in the successful implementation of this approach. Multivariate approaches can leverage simultaneous evaluation of several traits under multiple environments by exploring correlations to improve GS performance in breeding programs. Though, these models have been mostly evaluated using diverse panels of unrelated accessions. Here, we used multivariate GP models to predict multiple agronomic traits using 314 advanced and elite breeding lines of winter wheat evaluated at ten site-year environments. We evaluated a multi-trait (MT) model with two cross-validation schemes representing different breeding scenarios (CV1, prediction of completely unphenotyped lines; and CV2, prediction of lines partially phenotyped for correlated traits). Moreover, extensive data from multi-environment trials (METs) was used to cross-validate the Bayesian multi-trait multi-environment (MTME) model that integrates the analysis of multiple-traits including GxE interaction. The MT-CV2 model outperformed all other models for predicting grain yield with significant improvement in PA over the single-trait (ST-CV1) model. The MTME model performed better for all traits, with average improvement over the ST-CV1 reaching up to 19%, 71%, 17%, 48%, and 51% for grain yield, for grain protein content, test weight, plant height, and days to heading, respectively. Overall, our empirical analyses elucidate the potential of both MT-CV2 and MTME models when advanced breeding lines are used as training population to predict related preliminary breeding lines. Further, we also evaluated the practical application of MTME model in our breeding program to reduce phenotyping cost by using sparse testing design. This showed that complementing METs with GP can substantially enhance the resource efficiency. Our results demonstrate that the multivariate GS models hold great potential in implementing GS in breeding programs.