Spatially explicit Bayesian clustering models in population genetics.

Area of Science:

• Population genetics
• Bayesian inference
• Geographic data analysis

Background:

• Bayesian algorithms are increasingly used to infer population structure.
• Current models vary in prior distributions and assumptions, broadly categorized as with or without admixture.
• Spatially explicit models offer advanced population structure analysis.

Purpose of the Study:

• To review recent developments in Bayesian algorithms for population structure inference using geographic information.
• To clarify assumptions of spatially explicit models and test their robustness.
• To guide users in selecting appropriate models for population genetic analysis.

Main Methods:

• Review of Bayesian algorithms incorporating geographical information.
• Testing of models with and without admixture under violated assumptions.
• Comparative analysis of model performance with and without admixed individuals.

Main Results:

• Models without admixture are not robust to admixed individuals, leading to inaccurate population structure assessments.
• Models incorporating admixture are robust even when no admixture is present in the sample.
• Spatially explicit models with admixture provide more reliable population genetic structure inference.

Conclusions:

• Admixture models are superior for population genetic structure inference due to their robustness.
• Users should prefer spatially explicit models that include admixture for data exploration.
• Understanding model assumptions is crucial for accurate population genetic analyses.