Practical application of genomic selection in a doubled-haploid winter wheat breeding program

Area of Science:

• Agricultural Science
• Genetics
• Plant Breeding

Background:

• Crop improvement is lengthy and costly.
• Genomic selection (GS) enhances genetic gain in animals using SNP data.
• GS applicability in crops is largely limited to algorithm performance evaluation.

Purpose of the Study:

• To assess genomic selection applicability in wheat line development for grain yield.
• To identify factors influencing GS predictive ability in hard red winter wheat.
• To support the implementation of GS in practical wheat breeding programs.

Main Methods:

• Utilized a doubled-haploid population of hard red winter wheat.
• Evaluated GS performance for grain yield prediction over two years.
• Compared semi-parametric (RKHS) and parametric (GBLUP) prediction algorithms.
• Investigated the impact of cross-year validation and phenotype adjustment for GxE effects.
• Assessed the effect of SNP subset selection on predictive ability and computational load.

Main Results:

• Semi-parametric RKHS algorithm generally outperformed parametric GBLUP.
• Within-year cross-validation showed upward bias; cross-year validation is crucial for reliable prediction.
• Adjusting phenotypes for genotype by environment (GxE) effects improved GS predictive ability.
• A subset of SNPs (correlation > 0.4) yielded comparable results to the full set, reducing computation.
• GS results provide confidence for line selection in wheat variety development.

Conclusions:

• Genomic selection is a viable tool for accelerating wheat variety development.
• Cross-year validation and GxE adjustment are essential for robust GS models.
• SNP selection can optimize GS efficiency without sacrificing predictive power.