Genomic selection (GS) can increase the rate of genetic gain in crop breeding programs, but its effectiveness depends on the reliability of phenotypic data, the size and composition of the training population (TP), and the statistical model used to estimate genomic breeding values. These design choices are especially important in resource-limited breeding programs, where additional replication, larger TPs, and more extensive genotyping compete for the same resources. We used an empirical sweet sorghum [Sorghum bicolor (L.) Moench] breeding population developed by CHIBAS to evaluate the effects of phenotyping replication, TP size, training-validation genomic relatedness, and genomic prediction (GP) model on predictive ability (PA) for grain yield, plant height, stem weight, and total soluble solids across three field environments in Haiti. Few studies in sorghum have examined these factors together with comparable empirical rigor. Genotyping-by-sequencing produced 28,785 quality-filtered SNPs for 250 breeding lines. PA was estimated as the Pearson correlation between observed genotype BLUEs and predicted genomic values under repeated cross-validation. Increasing replication improved genomic heritability and PA for all traits and environments, with the largest gains observed for grain yield. For grain yield, PA increased from 0.22 to 0.41 in environment 1, from 0.27 to 0.53 in environment 2, and from 0.16 to 0.38 in environment 3 when one-replicate scenarios were compared with four-replicate scenarios. Larger TPs and increased training-validation genomic relatedness also improved PA, but their effects were most evident when phenotype estimates were based on multiple replicates. GP models showed largely comparable PAs across all evaluated traits. BayesA, BayesB, BayesC, BRR, and rrBLUP consistently produced similar accuracies, with no significant differences among models in most cases. PA ranged from 0.36 to 0.75 depending on the trait. GBLUP generally performed similarly to the other models; however, it showed significantly lower PA for stem weight and total soluble solids in Environment 1. Overall, replication emerged as the key determinant of PA in breeding relevant empirical data, highlighting the importance of accurate phenotypic estimates in improving selection efficiency in resource-limited programs.
CHARLES, J. R., Rice, B., Tovignan, T., Morris, G. P., Pressoir, G.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 0
- Comments 0