Phenotyping replication strongly influences genomic predictive ability in sweet sorghum (Sorghum bicolor [L.] Moench)
Abstract
Article summaryGenomic selection can accelerate breeding when the phenotypic and genomic information used to train prediction models is sufficiently reliable and relevant to the prediction target. Using a sweet sorghum breeding population evaluated in three Haitian field environments, we quantified the effects of phenotyping replication, training population (TP) size, TP–validation population (VP) genomic relatedness, and prediction model on genomic predictive ability for grain yield, plant height, stem weight, and total soluble solids. Increasing phenotyping replication consistently improved predictive ability across traits and environments and increased the precision of phenotypic estimates, with particularly large effects observed for grain yield. Larger TPs generally improved predictive ability; however, when TP sizes of 150 and 200 individuals were compared, the effect of increasing phenotyping replication was consistently greater than that of increasing TP size. Greater TP–VP genomic relatedness was also generally associated with higher predictive ability, whereas differences among prediction models were comparatively small and inconsistent across traits and environments. These results provide empirical evidence that genomic prediction performance in sweet sorghum depends not only on prediction model choice but also on the design and quality of the TP. In particular, for larger TPs, improving phenotyping precision through additional replication may provide greater gains in predictive ability than further increasing TP size.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 particularly important in resource-limited breeding programs, where phenotyping replication, TP size, and genotyping compete for the same resources. Using empirical data from a sweet sorghum [Sorghum bicolor (L.) Moench] breeding population developed by CHIBAS, we evaluated the effects of phenotyping replication, TP size, training–validation population (TP–VP) genomic relatedness, and genomic prediction (GP) model on predictive ability (PA). Grain yield, plant height, stem weight, and total soluble solids were evaluated across three field environments. Few studies have evaluated these factors jointly using an empirical dataset. Increasing the number of phenotypic replications consistently increased PA across traits and environments and improved the precision of genotype performance estimates, with the largest effects observed for grain yield. Larger TPs also increased PA, but the magnitude of the benefit diminished as TP size increased. In particular, when larger TPs above 150 individuals were compared, the effect of increasing phenotyping replication was consistently greater than that of further increasing TP size. Greater TP–VP genomic relatedness was generally associated with higher PA, whereas differences among GP models were comparatively small and inconsistent across traits and environments. These results provide empirical evidence that GP performance in sweet sorghum depends strongly on the design and quality of the training data. Within the range evaluated here, when TP size increased from 150 to 200 individuals, improving phenotypic precision through additional replication generally produced larger effects on PA than further increasing TP size.
Related
- Designed nutrient-gradient phenotyping enables predictive ionomics of tropical leafy greens
- Intrinsically disordered regions navigate DNA replication licensing factors to specific genomic loci.
- Genomic and functional characterization of compost-derived bacteria identifies a candidate bioinoculant that influences tomato physiology and leaf metabolism
- Persistence of viral full-length genomic RNA and replication-associated RNA transcripts and proteins in sheep tissues following acute Crimean-Congo hemorrhagic fever virus infection
- Proprietary Processing of Placental Tissue Allografts Influences Wound Healing Trajectories in a Rodent Excisional Skin Wound Model
BibTeX
@article{charles2026phenotyping,
title = {{Phenotyping replication strongly influences genomic predictive ability in sweet sorghum (Sorghum bicolor [L.] Moench)}},
author = {Jean Rigaud Charles and Brian Rice and Geoffrey P. Morris and Klanvi Thierry Tovignan and Gael Pressoir},
journal = {Frontiers in Plant Science},
year = {2026},
month = nov,
doi = {10.3389/fpls.2026.1919713},
url = {https://www.frontiersin.org/articles/10.3389/fpls.2026.1919713},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Phenotyping replication strongly influences genomic predictive ability in sweet sorghum (Sorghum bicolor [L.] Moench)
AU - Jean Rigaud Charles
AU - Brian Rice
AU - Geoffrey P. Morris
AU - Klanvi Thierry Tovignan
AU - Gael Pressoir
PY - 2026
DA - 2026/11/03
JO - Frontiers in Plant Science
DO - 10.3389/fpls.2026.1919713
UR - https://www.frontiersin.org/articles/10.3389/fpls.2026.1919713
AB - Article summaryGenomic selection can accelerate breeding when the phenotypic and genomic information used to train prediction models is sufficiently reliable and relevant to the prediction target. Using a sweet sorghum breeding population evaluated in three Haitian field environments, we quantified the effects of phenotyping replication, training population (TP) size, TP–validation population (VP) genomic relatedness, and prediction model on genomic predictive ability for grain yield, plant height, stem weight, and total soluble solids. Increasing phenotyping replication consistently improved predictive ability across traits and environments and increased the precision of phenotypic estimates, with particularly large effects observed for grain yield. Larger TPs generally improved predictive ability; however, when TP sizes of 150 and 200 individuals were compared, the effect of increasing phenotyping replication was consistently greater than that of increasing TP size. Greater TP–VP genomic relatedness was also generally associated with higher predictive ability, whereas differences among prediction models were comparatively small and inconsistent across traits and environments. These results provide empirical evidence that genomic prediction performance in sweet sorghum depends not only on prediction model choice but also on the design and quality of the TP. In particular, for larger TPs, improving phenotyping precision through additional replication may provide greater gains in predictive ability than further increasing TP size.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 particularly important in resource-limited breeding programs, where phenotyping replication, TP size, and genotyping compete for the same resources. Using empirical data from a sweet sorghum [Sorghum bicolor (L.) Moench] breeding population developed by CHIBAS, we evaluated the effects of phenotyping replication, TP size, training–validation population (TP–VP) genomic relatedness, and genomic prediction (GP) model on predictive ability (PA). Grain yield, plant height, stem weight, and total soluble solids were evaluated across three field environments. Few studies have evaluated these factors jointly using an empirical dataset. Increasing the number of phenotypic replications consistently increased PA across traits and environments and improved the precision of genotype performance estimates, with the largest effects observed for grain yield. Larger TPs also increased PA, but the magnitude of the benefit diminished as TP size increased. In particular, when larger TPs above 150 individuals were compared, the effect of increasing phenotyping replication was consistently greater than that of further increasing TP size. Greater TP–VP genomic relatedness was generally associated with higher PA, whereas differences among GP models were comparatively small and inconsistent across traits and environments. These results provide empirical evidence that GP performance in sweet sorghum depends strongly on the design and quality of the training data. Within the range evaluated here, when TP size increased from 150 to 200 individuals, improving phenotypic precision through additional replication generally produced larger effects on PA than further increasing TP size.
SN - 1664-462X
ER -