Principal component analysis revisited: fast multitrait genetic evaluations with smooth convergence.

Area of Science:

• Population genetics
• Quantitative genetics
• Animal breeding

Background:

• Genetic evaluation is crucial for population management decisions.
• Multivariate mixed models improve accuracy by considering trait correlations.
• Scalability issues arise in multitrait models due to exponential parameter growth with increasing traits.

Purpose of the Study:

• To introduce a novel, computationally efficient method for genetic evaluation.
• To reduce the dimensionality of response variables in multitrait genetic analysis.
• To approximate full multivariate analysis using principal components.

Main Methods:

• Applied principal component analysis (PCA) to reduce response variable dimensions.
• Utilized computed principal components as separate responses in genetic evaluation.
• Compared the PCA-based approach with traditional multivariate and factor analytic methods.
• Evaluated computational time and rank lists of predicted genetic merit on forest tree datasets.

Main Results:

• The PCA-based approach significantly reduced computational time (seconds) compared to traditional methods (hours).
• Rank lists of top individuals showed good agreement across methods.
• The approach demonstrated robustness with missing data and compatibility with standard software.
• Factor analytic approach required 5-10 minutes.

Conclusions:

• PCA-based genetic evaluation offers a scalable and efficient alternative for multitrait analysis.
• This method mitigates computational challenges in breeding and wild populations.
• The approach is practical, requiring no specialized software implementations.