Including marker x environment interactions improves genomic prediction in red clover (Trifolium pratense L.)

Area of Science:

• Plant breeding
• Quantitative genetics
• Agricultural science

Background:

• Genomic prediction typically ignores genotype x environment interactions, limiting its application in multi-environment trials.
• Marker x environment (MxE) interaction models have emerged to better capture genotype x environment effects.

Purpose of the Study:

• To evaluate the effectiveness of different genomic prediction models in red clover (Trifolium pratense L.).
• To compare single environment, across environment, and MxE interaction models for predicting yield and quality traits.
• To assess the impact of population structure on prediction accuracy.

Main Methods:

• Field trials were conducted across five European locations within the EUCLEG project.
• Three genomic prediction models were compared: SingleEnv, AcrossEnv, and MxE.
• Predictive ability (PA) was assessed for annual dry matter yield (DMY), crude protein (CP), and date of flowering (DOF).

Main Results:

• The MxE model demonstrated superior predictive ability compared to SingleEnv and AcrossEnv models across traits.
• Dry matter yield (DMY) showed the highest predictive ability, with PAs ranging from 0.40 to 0.87.
• Crude protein (CP) was predicted poorly, while date of flowering (DOF) had moderate to high PAs (0.67-0.87).
• Population structure influenced predictive ability, particularly for DMY and DOF.

Conclusions:

• Incorporating marker x environment interactions significantly enhances genomic prediction accuracy in red clover multi-environment trials.
• The MxE model offers a promising approach for identifying stable and environment-specific genetic markers.
• These findings have implications for optimizing breeding strategies and accelerating genetic gain in forage crops.