Abstract
Genomic selection is an alternative for the reduction of selection costs and rapid genetic progress in plant breeding. Previous genomic selection studies on tomato have evaluated the impact of several factors such as linkage disequilibrium and training population size on prediction accuracy. These evaluations were done using simple environment models that do not consider GxE interactions.In our study, we have evaluated the impact of the integration of GxE interaction and environmental cofactors on prediction accuracy. The simple environment, G+E and GxE models were used to evaluate the prediction accuracy in two population datasets (MAGIC and GWAS) that have been genotyped and phenotyped for several traits in different environments.This study revealed that the prediction accuracy of multi-environment models is better than that of the single environment model. The GxE model has a prediction accuracy as high as that of G+E, and even predicted better than G+E depending on the environment or the composition of the training population. Both models have a higher prediction accuracy than the single environment model (G) when accessions are tested in at least one environment (CV2 sampling) compared to models where some accessions are tested in none of the environments (CV1). The prediction accuracy of these different models was strongly correlated to the heritability of the traits. It increased when the environments were correlated with each other. The integration of environmental co-factors in the GxE model improved prediction accuracy. In limited conditions, CV2 sampling is the most appropriate to make genomic predictions with multi-environment models.
Abstract from the paper’s OpenAlex record; open copy licensed other-oa. Read at the source ↗
The text the model read, with the source record’s inline markup (italics, sub- and superscripts) rendered and its character entities decoded; quote offsets refer to the raw text.
Claims 6
Claim · model-written summary
Multi-environment models had better prediction accuracy than the single environment model.
Verbatim
e0“the prediction accuracy of multi-environment models is better than that of the single environment model”
Claim · model-written summary
The GxE model had prediction accuracy as high as G+E and better than G+E depending on the environment or training population composition.
Verbatim
e0“The GxE model has a prediction accuracy as high as that of G+E, and even predicted better than G+E depending on the environment or the composition of the training population”
Claim · model-written summary
Both multi-environment models had higher prediction accuracy than the single environment model when accessions were tested in at least one environment (CV2) rather than in none (CV1).
Verbatim
e0“Both models have a higher prediction accuracy than the single environment model (G) when accessions are tested in at least one environment (CV2 sampling) compared to models where some accessions are tested in none of the environments (CV1)”
Claim · model-written summary
Prediction accuracy was strongly correlated with trait heritability and increased when the environments were correlated with each other.
- Taxon
- tomatet0
- Outcome
- prediction accuracye0
- Qualifier
- It increased when the environments were correlated with each othere0
- Comparator
- heritability of the traitse0
Not reportedDirection
Verbatim
t0“tomate”
e0“The prediction accuracy of these different models was strongly correlated to the heritability of the traits. It increased when the environments were correlated with each other”
Claim · model-written summary
Integrating environmental co-factors in the GxE model improved prediction accuracy.
- Taxon
- tomatet0
- Treatment
- integration of environmental co-factors in the GxE modele0
- Outcome
- prediction accuracye0
- Direction
- improvede0
Not reportedComparator
Verbatim
t0“tomate”
e0“The integration of environmental co-factors in the GxE model improved prediction accuracy”
Claim · model-written summary
In limited conditions CV2 sampling is the most appropriate for genomic predictions with multi-environment models.
Verbatim
t0“tomate”
e0“In limited conditions, CV2 sampling is the most appropriate to make genomic predictions with multi-environment models”