Full Bayesian Comparative Phylogeography from Genomic Data.

Area of Science:

• Evolutionary Biology
• Population Genetics
• Genomics

Background:

• Understanding biological diversification requires accounting for community-scale processes like co-speciation.
• Existing approximate-likelihood Bayesian computation (ABC) methods for inferring co-speciation are sensitive to prior assumptions and can be biased.

Purpose of the Study:

• To introduce ecoevolity, a full-likelihood Bayesian approach for inferring population history and divergence events from genomic data.
• To improve accuracy, precision, and computational efficiency in estimating divergence times and numbers of events compared to existing methods.

Main Methods:

• Developed ecoevolity, a full-likelihood Bayesian method that analytically integrates over gene trees to calculate the likelihood of population history from genomic data.
• Employed Markov chain Monte Carlo algorithms for efficient sampling of the model-averaged posterior.
• Tested method robustness using simulations, including scenarios with linked loci and character-acquisition biases.

Main Results:

• Simulations show ecoevolity is significantly more accurate and precise than existing approximate-likelihood methods for estimating divergence events.
• The new method is orders of magnitude faster than existing ABC approaches.
• Application to Gekko lizard genomic data strongly supports independent divergences for all four island population pairs, even with recent divergences.

Conclusions:

• Ecoevolity provides a robust and efficient full Bayesian approach for inferring population divergence histories from genomic data.
• The method accurately estimates the number and timing of divergence events, outperforming previous techniques.
• Independent divergences were confirmed in Gekko lizard populations, demonstrating the method's effectiveness in real-world scenarios.