Genomic prediction for rust resistance in pea.

Area of Science:

• Plant genetics and breeding
• Genomic selection applications
• Crop disease resistance

Background:

• Genomic selection (GS) is crucial for complex traits in plant breeding.
• Pea rust (Uromyces pisi) resistance is a complex trait requiring advanced breeding strategies.
• Understanding genotype-environment interactions (GEI) is vital for accurate resistance prediction.

Purpose of the Study:

• To evaluate genomic selection models for predicting pea rust resistance.
• To assess the impact of marker × environment (M×E) interactions on prediction accuracy.
• To compare single-trait versus multi-trait index approaches for complex trait prediction.

Main Methods:

• Utilized 320 pea accessions and 26,045 Silico-DArT markers.
• Employed genomic best linear unbiased prediction (GBLUP) models, including those with M×E interactions.
• Compared various cross-validation strategies and assessed trait-index approaches like FAI-BLUP.

Main Results:

• GBLUP models, especially with M×E interactions, showed superior predictive ability for rust resistance.
• The FAI-BLUP approach within the Bayesian Lasso (BL) model achieved the highest predictive ability (0.635).
• M×E interactions significantly improved prediction accuracy in diverse environments but not for un-phenotyped lines.

Conclusions:

• Genomic selection, particularly multi-trait FAI-BLUP and M×E interaction modeling, is effective for pea rust resistance breeding.
• GEI significantly influences predictive ability, highlighting the need for robust models.
• Advanced GS strategies enhance the development of disease-resistant pea varieties.