Multi-trait multi-environment models in the genetic selection of segregating soybean progeny

Area of Science:

• Plant breeding and genetics
• Quantitative genetics
• Agricultural science

Background:

• Single-trait best linear unbiased prediction (BLUP) is standard for soybean genetic selection.
• Analyzing correlated traits individually can introduce selection bias.
• Accounting for trait correlations can improve genetic estimates under environmental influences.

Purpose of the Study:

• To evaluate the efficiency and applicability of multi-trait multi-environment (MTME) models for soybean progeny selection.
• To compare residual maximum likelihood (REML/BLUP) and Bayesian approaches within MTME frameworks.
• To assess genetic parameter estimation and prediction accuracy for soybean traits.

Main Methods:

• Utilized data from 203 soybean F2:4 progeny across two environments.
• Assessed traits: days to maturity (DM), 100-seed weight (SW), and seed yield per plot (SY).
• Estimated variance components, breeding values, and genetic gains using REML/BLUP and Bayesian MTME models.

Main Results:

• REML/BLUP and Bayesian MTME methods yielded similar variance components, breeding values, and genetic gains.
• MTME models (both frequentist and Bayesian) showed higher broad-sense heritability and progeny accuracy than single-trait models.
• Bayesian analysis provided credibility intervals for heritability estimates.

Conclusions:

• Multi-trait multi-environment models enhance genetic gain predictions in soybean breeding.
• Both REML/BLUP and Bayesian MTME approaches are effective for selecting segregating soybean progeny.
• MTME models provide more accurate genetic parameter estimates and selection accuracy.