A method to predict the response to directional selection using a Kalman filter.

Area of Science:

• Evolutionary Biology
• Quantitative Genetics
• Animal Breeding

Background:

• Predicting evolutionary trajectories is a significant challenge in quantitative genetics.
• The breeder's equation, a standard tool, often yields biased predictions due to omitted variables, inaccurate genetic variance estimates, and nonlinearities.
• These prediction errors are frequently non-zero, indicating systematic bias in current models.

Purpose of the Study:

• To develop an improved method for predicting evolutionary change by utilizing systematic prediction biases.
• To enhance the accuracy of evolutionary predictions in quantitative genetics.

Main Methods:

• A new prediction method was developed, incorporating a bias term alongside the breeder's equation.
• A Kalman filter was employed to integrate predictions from the breeder's equation with historical trait mean data.
• Filter parameters were optimized generationally using a learning algorithm on past evolutionary changes.

Main Results:

• The proposed method demonstrated superior performance compared to the traditional breeder's equation in artificial selection experiments (Drosophila melanogaster wing) and complex simulations.
• Outperformance was particularly notable when selection on certain traits was omitted, data were noisy, or additive genetic variance estimates were inaccurate.
• The method requires only trait means from past generations for application.

Conclusions:

• Systematic biases in evolutionary prediction, often considered a nuisance, can be leveraged to improve predictive accuracy.
• The novel Kalman filter-based approach offers a more robust and accurate method for predicting evolution.
• This method provides a practical advancement for quantitative genetics and evolutionary studies, especially in scenarios with data limitations.