Genome-based prediction of multiple wheat quality traits in multiple years

Area of Science:

• Agricultural Science
• Genetics
• Plant Breeding

Background:

• Wheat quality is crucial but costly and time-consuming to assess late in breeding.
• Genomic prediction offers a potential solution to improve efficiency in wheat quality selection.

Purpose of the Study:

• To evaluate genomic prediction models for 13 wheat quality traits.
• To compare the performance of Bayesian multi-trait multi-environment (BMTME) and multi-trait ridge regression (MTR) models.

Main Methods:

• Utilized five datasets of wheat lines evaluated over two years.
• Trained models using first-year data to predict second-year quality traits.
• Assessed prediction accuracy using Mean Arctangent Absolute Percentage Error (MAAPE) and Pearson's correlation.

Main Results:

• Moderate to high prediction accuracies were observed for most wheat quality traits.
• The BMTME model consistently outperformed the MTR model across all traits.
• Test weight showed the best predictions with BMTME, while ALVPL had the worst.

Conclusions:

• Genomic selection is feasible for efficient wheat quality improvement.
• The BMTME model is recommended for multi-trait prediction due to its accuracy and ability to model trait and environment correlations.