Genomic selection for lentil breeding: Empirical evidence

Area of Science:

• Plant breeding
• Genetics
• Agricultural science

Background:

• Genomic selection (GS) is a powerful marker-based breeding tool, increasingly adopted in crop improvement.
• Its application and optimal models for lentil (Lens culinaris Medik.) breeding remain unexplored.

Purpose of the Study:

• To evaluate and compare the predictive accuracy of various genomic selection models in lentil breeding.
• To provide recommendations for developing effective genomic selection strategies for lentil.

Main Methods:

• Assessed nine single-trait (ST), two multiple-trait (MT), and genotype × environment interaction (GEI) models.
• Utilized lentil diversity panels and RILs, phenotyped for five traits and genotyped via exome capture.
• Conducted within-population, across-population, and across-environment genomic predictions.

Main Results:

• Prediction accuracy varied significantly across models, populations, scenarios, and traits.
• BayesB model excelled when large-effect QTL were present; GEI and MT-GS models enhanced accuracy for low heritability traits.
• High accuracies were achieved for within-population and across-environment predictions; across-population predictions were generally poor.

Conclusions:

• Genomic selection is viable for lentil breeding, particularly for within-population and across-environment predictions.
• Across-population predictions require careful consideration, especially with small population sizes, due to low accuracy.