Inclusion of Dominance Effects in the Multivariate GBLUP Model

Area of Science:

• Agricultural Science
• Genetics
• Plant Breeding

Background:

• Molecular markers and prediction models can reduce costs and improve genotype selection in maize breeding.
• Genomic Best Linear Unbiased Prediction (GBLUP) models including dominance effects have not been extensively studied in univariate and multivariate contexts for maize.

Purpose of the Study:

• To evaluate univariate and multivariate GBLUP models with additive and additive plus dominance effects for maize.
• To compare the predictive ability and variance component estimation of these models using phenotypic and molecular marker data.

Main Methods:

• Constructed 400 simple maize hybrids from 4,091 lines using phenotypic and molecular marker data.
• Analyzed hybrids using univariate and multivariate GBLUP models, considering additive effects and additive plus dominance effects.
• Compared model performance under various heritability and genetic architecture scenarios.

Main Results:

• Inclusion of dominance effects significantly improved predictive ability and variance component estimation.
• Marginal differences were observed between univariate and multivariate models, attributed to low-to-moderate trait correlations and moderate heritabilities.
• Dominance effects proved more impactful than the choice between univariate and multivariate approaches.

Conclusions:

• Incorporating dominance effects into GBLUP models is an effective strategy for enhancing prediction accuracy and estimating genetic parameters in maize.
• Univariate and multivariate GBLUP models showed comparable performance in this study due to specific genetic correlations and heritability levels.
• The findings support the use of GBLUP models with dominance for more precise maize genotype selection.