A comparison of methods to estimate cross-environment genetic correlations

Area of Science:

• Quantitative genetics
• Animal breeding
• Statistical genetics

Background:

• Advanced quantitative genetic methods are not always needed for broad genetic inquiries.
• Simpler estimation methods can introduce bias, impacting their utility.

Purpose of the Study:

• Compare family mean correlations with least squares and REML variance component approaches.
• Evaluate methods for estimating cross-environment genetic correlations.

Main Methods:

• Analyzed empirical data from existing studies and internal laboratory experiments.
• Compared estimates derived from family mean correlations and variance component models.

Main Results:

• Agreement between methods improved with full-sib data over half-sib data.
• Increased mean family size enhanced the agreement of estimates.
• Noted potential biases in REML estimation, particularly when testing correlations near 0 or 1.

Conclusions:

• Family mean correlations are adequate for testing the presence of genetic correlations across environments for certain research questions.
• Variance component approaches are recommended for precise parameter estimation or when detailed analysis beyond broad patterns is required.