Genomic Analysis and Prediction within a US Public Collaborative Winter Wheat Regional Testing Nursery

Area of Science:

• Plant breeding
• Genomics
• Agricultural science

Background:

• Allele-based breeding utilizes genomic prediction models to estimate breeding values.
• Genomic selection (GS) can effectively use unbalanced datasets common in breeding programs.
• The Southern Regional Performance Nursery (SRPN) has collected wheat variety performance data since 1931.

Purpose of the Study:

• To evaluate the SRPN as a training population (TP) for genomic selection.
• To assess the effectiveness of GS models for predicting wheat performance across years.
• To compare predictability using program-specific versus whole-set TPs.

Main Methods:

• Generated whole-genome profiles using genotyping-by-sequencing (GBS) for 939 SRPN entries (1992-present).
• Developed and evaluated GS prediction models for traits like grain yield.
• Compared GS model performance against year-to-year phenotypic correlations and analyzed predictability using different TP strategies.

Main Results:

• GS prediction models across years (average = 0.33) outperformed year-to-year phenotypic correlation for yield ( = 0.27) in most evaluated years.
• Genomic selection shows potential to improve selection for low heritability traits in variable environments.
• Predictability of breeding programs was similar when using program-specific or whole-set TPs.

Conclusions:

• The SRPN is a valuable resource for developing GS training populations in wheat.
• Collaborative use of regional testing network data can significantly benefit wheat breeding programs.
• Genomic selection offers a powerful tool to accelerate genetic gain in wheat breeding.