Hybrid Wheat Prediction Using Genomic, Pedigree, and Environmental Covariables Interaction Models

Area of Science:

• Plant breeding
• Quantitative genetics
• Agricultural science

Background:

• Predicting hybrid performance is essential for efficient crop improvement.
• Genotype × environment interactions (G×E) significantly influence trait expression in crops.
• Integrating genomic and pedigree data can enhance prediction accuracy.

Purpose of the Study:

• To evaluate genotype × environment interactions (G×E) models for predicting hybrid performance in wheat.
• To compare the accuracy of different genomic and pedigree models under various cross-validation schemes.
• To assess the potential for predicting unobserved hybrids using G×E models.

Main Methods:

• Utilized five genomic and pedigree models (M1-M5) incorporating G×E.
• Employed four cross-validation schemes (T2FM, T1M, T1F, T0FM) to simulate different training/testing scenarios.
• Assessed similarity using pedigree and molecular markers for lines, and environmental covariables for environments.
• Tested models on 1888 wheat hybrids across three years.

Main Results:

• The most complex model (M5) showed slightly higher prediction accuracy for grain yield under the T2FM scheme.
• Prediction accuracies for grain yield and other traits ranged from 0.50 to 0.55 under the T1F scheme.
• Model M3 achieved high accuracy for flowering traits (0.71) under T1M, while M5 excelled for grain yield (0.5).
• Model M5 achieved a prediction accuracy of 0.61 for grain yield under the T0FM scheme.
• High prediction accuracy was observed even with untested parents, leveraging both genomic and pedigree information.

Conclusions:

• Genotype × environment interactions (G×E) models are effective for hybrid prediction in wheat.
• Integrating genomic and pedigree data improves prediction accuracy, even for untested lines.
• Modeling genomic general combining ability (GCA) and specific combining ability (SCA) with G×E interactions allows for the prediction of unobserved hybrids.