Genomic selection for agronomic traits in a winter wheat breeding program.

Area of Science:

• Plant breeding
• Agricultural genomics
• Quantitative genetics

Background:

• Genomic selection (GS) is crucial for optimizing quantitative traits in plant breeding.
• Annual implementation of GS can improve parent selection and reduce phenotyping costs.
• Repeat amplification sequencing (rAMP-seq) offers a cost-effective genotyping approach for large populations.

Purpose of the Study:

• To evaluate rAMP-seq based GS for enhancing winter wheat breeding.
• To determine optimal population structures for GS prediction accuracy.
• To compare the performance of different GS models in wheat breeding.

Main Methods:

• Phenotyped and genotyped 1870 winter wheat genotypes using rAMP-seq.
• Optimized training-to-testing population size, identifying a 70:30 ratio for consistent accuracy.
• Tested three GS models: rrBLUP, RKHS, and feed-forward neural networks.

Main Results:

• The 70:30 population ratio yielded the most consistent prediction accuracy across models.
• rrBLUP, RKHS, and neural networks showed comparable performance for most traits.
• RKHS model demonstrated superior prediction accuracy for yield (r=0.34-0.39).

Conclusions:

• rAMP-seq based GS is a valuable tool for increasing genetic gain in winter wheat breeding.
• Integrating GS into breeding programs enhances efficiency and reduces costs.
• The study validates the utility of GS for selecting optimal parents and accelerating breeding cycles.