Conditioning on the causal network prevents indirect response to selection.

Area of Science:

• Animal genetics
• Quantitative genetics
• Statistical modeling

Background:

• Multiple trait animal models (MTM) estimate breeding values (BV) considering genetic and environmental correlations.
• Causal relationships among traits may necessitate a different approach, such as structural equation models (SEM).
• While parametrically equivalent, MTM and SEM yield EBVs that require different interpretations.

Purpose of the Study:

• To investigate the impact of MTM vs. SEM on selection response in a five-trait causal network.
• To compare selection outcomes for traits at different positions (upstream, midstream, downstream) within the causal network.
• To evaluate scenarios with purely causal relationships and combined causal and genetic correlations.

Main Methods:

• Stochastic simulation experiment to model a five-trait causal network.
• Application of both MTM and SEM for estimating breeding values.
• Analysis of selection response under different selection targets and network structures.

Main Results:

• MTM absorbed causal relationships as genetic correlations, altering selection response compared to SEM.
• No differences in selection response were observed when targeting the top trait.
• Significant differences emerged for upstream traits when targeting midstream or downstream traits.
• SEM reduced indirect response in upstream traits when both causal effects and genetic correlations were present.

Conclusions:

• SEM provides a more accurate interpretation of EBVs in causal networks, leading to improved selection response.
• The choice between MTM and SEM significantly impacts selection outcomes, particularly for traits influenced by upstream factors.
• SEM is crucial for accurately dissecting direct and indirect selection responses in complex genetic architectures.