Phylogeography by diffusion on a sphere: whole world phylogeography

Area of Science:

• Evolutionary Biology
• Biogeography
• Computational Biology

Background:

• Phylogeographical methods reconstruct species' geographical history using phylogenetic trees.
• Current Bayesian models use diffusion on a plane, ignoring Earth's spherical geometry and causing distortions in global analyses.
• Map projections like Mercator introduce inaccuracies in large-scale phylogeographical reconstructions.

Purpose of the Study:

• To develop a novel Bayesian phylogeographical method accounting for Earth's spherical nature.
• To address limitations of planar diffusion models in global-scale phylogeographical inference.
• To provide a more accurate framework for reconstructing species' geographical origins.

Main Methods:

• Introduced a Bayesian phylogeographical method based on diffusion on a sphere.
• Developed an efficient algorithm for Markov Chain Monte Carlo (MCMC) inference.
• Compared planar and spherical diffusion models through simulation studies.

Main Results:

• The spherical diffusion model yields substantial differences compared to planar models for global datasets.
• The method accurately approximates planar models for small geographical areas.
• Applied the method to infer the origin of Hepatitis B in Eurasia and Africa.

Conclusions:

• Developed a robust framework for phylogeographical inference suitable for both small and large geographical scales.
• The framework accommodates regional sampling and prior information on clade locations.
• Implemented the method in the open-source GEO_SPHERE package for BEAST 2, enabling joint tree and geography inference.