Integrating genomic predictions into an applied Central European wheat breeding program.

Area of Science:

• Plant breeding
• Genetics
• Agricultural science

Background:

• Genomic prediction is crucial for accelerating genetic gain in plant breeding.
• Accurate genomic predictions depend on factors like data quality, sample size, and diversity.
• Understanding how breeding stages influence prediction accuracy is vital for optimizing wheat improvement.

Purpose of the Study:

• To evaluate the impact of phenotypic data quality, sample size, and diversity across breeding stages on genomic prediction accuracy in winter bread wheat.
• To compare prediction abilities within and across different breeding stages.
• To identify optimal strategies for genomic selection in wheat breeding programs.

Main Methods:

• Utilized extensive phenotypic data (57,000 plots) and genotypic data (6,228 genotypes, 7,000 SNPs) from a winter bread wheat breeding program.
• Implemented genomic best linear unbiased prediction (GBLUP) models.
• Tested prediction abilities across three scenarios: within breeding stages, across breeding stages, and for advanced genotypes.

Main Results:

• Models trained on mid- or late-stage phenotypic data yielded higher prediction accuracy for most traits compared to early-stage data.
• Combining mid- and late-stage data significantly improved predictions for complex traits like grain yield and yellow rust resistance.
• Genomic selection effectively reduces generation intervals and costs, increasing genetic gain in wheat breeding.

Conclusions:

• Balancing heritability and population size is key for effective genomic prediction.
• Advanced breeding stages and comprehensive training sets are essential for accurate predictions, particularly for complex traits.
• Genomic selection is a foundational tool for enhancing efficiency and genetic progress in modern wheat breeding programs.