Inconsistency of parsimony under the multispecies coalescent

Area of Science:

• Phylogenetics
• Evolutionary Biology
• Computational Biology

Background:

• Parsimony is a common phylogenetic inference method, but its statistical consistency under the multispecies coalescent (MSC) model is not fully understood.
• Previous research shows conflicting results regarding parsimony's consistency under MSC, particularly concerning the number of taxa and tree rooting.

Purpose of the Study:

• To investigate the statistical consistency of concatenated parsimony under the multispecies coalescent (MSC) model for various tree sizes and rootings.
• To develop a novel method for calculating expected internal branch lengths of gene trees under MSC to assess parsimony's performance.

Main Methods:

• Developed a technique to compute expected internal branch lengths of gene trees under the MSC.
• Analyzed the parameter space of species trees to identify conditions leading to concatenated parsimony inconsistency.

Main Results:

• Concatenated parsimony is statistically consistent for unrooted 5-taxa trees under MSC.
• Identified regions of statistical inconsistency for concatenated parsimony in rooted trees with 5 or more taxa and unrooted trees with 6 or more taxa under MSC.

Conclusions:

• Parsimony methods are not universally dependable for phylogenetic inference under the multispecies coalescent model.
• The findings highlight the limitations of parsimony, especially for complex evolutionary scenarios involving more than 5 taxa or unrooted trees.