Marker-based estimation of heritability in immortal populations

Area of Science:

• Quantitative genetics
• Plant breeding
• Genomics

Background:

• Heritability is crucial for evolutionary and breeding insights.
• Traditional methods estimate broad-sense heritability, not accounting for genetic relatedness.
• Marker-based narrow-sense heritability estimation is gaining traction but faces methodological hurdles in plants.

Purpose of the Study:

• To investigate the feasibility of using mixed models at the individual plant or plot level for heritability estimation in plants.
• To compare this novel approach with traditional genotypic mean-based methods.
• To assess the impact on genomic prediction and genome-wide association studies.

Main Methods:

• Statistical analysis, simulations, and real plant data were employed.
• Mixed models were applied at both the genotypic mean and individual plant/plot levels.
• Evaluated effects on best linear unbiased predictor (BLUP) for genomic prediction and genome-wide association studies (GWAS).

Main Results:

• Heritability estimates from genotypic means showed large standard errors and unrealistic values.
• Individual plant/plot level mixed models produced more realistic estimates with up to 13x smaller standard errors for simulated traits.
• Genomic prediction accuracy increased by up to 49% using individual plant data.
• Genome-wide association studies showed minimal power increase with individual plant data for simulated traits.

Conclusions:

• Mixed models at the individual plant or plot level offer a more robust method for heritability estimation in plants.
• This approach significantly improves genomic prediction accuracy.
• The methodology is applicable to complex traits in various organisms, including crops, bacteria, and fungi.