Estimation of genetic variance for macro- and micro-environmental sensitivity using double hierarchical generalized...

Area of Science:

• Animal genetics
• Quantitative genetics
• Statistical modeling

Background:

• Animals exhibit genetic variation in their response to environmental factors, termed environmental sensitivity.
• Environmental factors can be identifiable (macro-environmental) or unknown (micro-environmental).

Purpose of the Study:

• To develop a statistical method for simultaneously estimating genetic parameters for macro- and micro-environmental sensitivities.
• To assess the bias and precision of these genetic parameter estimates.
• To evaluate Akaike's information criterion (AIC) with h-likelihood for model selection.

Main Methods:

• A reaction norm model for macro-environmental sensitivity was integrated with a structural model for residual variance (micro-environmental sensitivity) using a double hierarchical generalized linear model.
• Akaike's information criterion was constructed using approximated h-likelihood for model selection.
• Simulations of sire populations with large half-sib offspring groups were used to investigate parameter estimation bias and precision.

Main Results:

• Designs with at least 100 sires, each with 100 offspring, are necessary for variance estimates with standard deviations below 50% of the true value.
• Increasing offspring numbers substantially reduced estimate standard deviations, particularly for genetic variances of macro- and micro-environmental sensitivities.
• No significant bias was observed in parameter estimates; AIC correctly selected the true genetic model in over 90% of replicates with 100 offspring per sire.
• Application to dairy cattle lactation milk yield confirmed the existence of genetic variance for both micro- and macro-environmental sensitivities.

Conclusions:

• The developed algorithm and model selection criterion enhance understanding of the genetic control of environmental sensitivities.
• Optimal study designs require a minimum of 100 sires, each with at least 100 offspring, for reliable genetic parameter estimation.