Can quartet analyses combining maximum likelihood estimation and Hennigian logic overcome long branch attraction in...

Area of Science:

• Phylogenetics
• Computational Biology
• Evolutionary Biology

Background:

• Model-based phylogenetic methods like maximum likelihood and Bayesian inference can be misled by systematic biases.
• Long branch attraction (LBA) is a common artifact arising from model misspecification, particularly with empirical data.

Purpose of the Study:

• To introduce PhyQuart, a novel method for evaluating binary phylogenetic trees for quartets of taxa.
• To address limitations of existing methods in handling LBA due to model misspecification.

Main Methods:

• PhyQuart identifies site-patterns supporting specific quartet trees, incorporating Hennig's distinction between apomorphic and plesiomorphic similarity.
• It employs corrections to observed site-pattern frequencies based on maximum likelihood estimation expectations.

Main Results:

• Extensive simulations show PhyQuart outperforms maximum likelihood estimation in specific scenarios.
• PhyQuart is more robust to LBA artifacts caused by extreme branch length asymmetries and minor model misspecification.

Conclusions:

• PhyQuart offers a more accurate alternative to maximum likelihood for phylogenetic tree evaluation.
• The method is particularly valuable when dealing with datasets prone to long branch attraction artifacts.