Estimating (Non)Linear Selection on Reaction Norms: A General Framework for Labile Traits.

Area of Science:

• Evolutionary biology
• Quantitative genetics
• Ecology

Background:

• Reaction norms describe how phenotypes change with environment and individual differences.
• Estimating nonlinear selection on reaction norms is crucial for evolutionary theory but empirically challenging.
• Existing methods struggle to account for uncertainty in reaction norm parameters and their fitness consequences.

Purpose of the Study:

• To develop and validate generalized multilevel models for estimating nonlinear selection on reaction norms.
• To provide a flexible Bayesian framework for analyzing labile traits and their evolutionary dynamics.
• To enable robust tests of adaptive theory in heterogeneous and dynamic environments.

Main Methods:

• Proposed generalized multilevel models incorporating stabilizing, disruptive, and correlational selection.
• Utilized a flexible Bayesian framework to simultaneously model reaction norm parameters and fitness effects.
• Validated the models using simulations to assess inference bias and statistical power.

Main Results:

• The proposed models facilitate unbiased Bayesian inference for reaction norm selection.
• Demonstrated desirable statistical power for hypothesis testing with large sample sizes.
• The framework effectively accounts for uncertainty in reaction norm parameters and nonlinear fitness effects.

Conclusions:

• The generalized multilevel models offer a robust approach to empirically estimate nonlinear selection on reaction norms.
• This framework enhances the ability to test adaptive theory for labile traits in natural populations.
• Provided coding tutorials in R using the Stan probabilistic programming language to aid empiricists.