Approximate Bayesian computation for Markovian binary trees in phylogenetics.

Area of Science:

• Evolutionary biology
• Computational phylogenetics
• Statistical modeling

Background:

• Phylogenetic trees reveal evolutionary relationships and diversification rates.
• Understanding macroevolutionary mechanisms requires realistic models of diversification.
• Multitype branching processes offer flexible frameworks for phylogenetic modeling.

Purpose of the Study:

• To introduce Markovian binary trees (MBTs) as a flexible model for trait-based diversification rates.
• To develop and evaluate an approximate Bayesian computation (ABC) scheme for inferring MBT parameters from phylogenetic tree shapes.
• To apply the developed method to real-world phylogenetic data to test evolutionary hypotheses.

Main Methods:

• Utilized Markovian binary trees (MBTs), a class of multitype branching processes.
• Developed an approximate Bayesian computation (ABC) scheme to infer MBT parameters.
• Employed phylogenetic tree shapes as the primary source of information for inference.
• Validated the ABC method using simulation studies and compared it with likelihood-based methods.

Main Results:

• The ABC scheme accurately infers diversification rates from phylogenetic tree shapes.
• The method demonstrates capability in detecting variation in diversification rates.
• Accuracy of the ABC method is comparable to, and often exceeds, that of likelihood-based methods.
• Application to Squamata phylogeny supports bidirectional transitions between oviparity and viviparity.

Conclusions:

• The developed ABC method provides a powerful tool for phylogenetic inference using complex evolutionary models.
• Likelihood-free inference schemes enable the use of more realistic phylogenetic models.
• Findings reinforce evolutionary hypotheses regarding reproductive mode transitions in Squamata.