Character-state space versus rate of evolution in phylogenetic inference

Area of Science:

• Evolutionary biology
• Phylogenetics
• Computational biology

Background:

• Nucleotide characters in phylogenetic inference are limited by few states, leading to frequent parallelisms and reversals.
• Increased character-state space, such as using third codon positions or amino acids, offers potential advantages for phylogenetic accuracy.

Purpose of the Study:

• To quantify the relationship between character-state space and evolutionary rate.
• To assess how tree topology, branch lengths, rate heterogeneity, state change probabilities, and state frequencies affect this relationship.
• To evaluate the impact on inferred tree lengths, consistency and retention indices, and phylogenetic accuracy.

Main Methods:

• Phylogenetic inference simulations were conducted.
• The study varied parameters including tree topology, branch lengths, and site-specific evolutionary rates.
• Probabilities of character state change and frequencies of character states were manipulated.

Main Results:

• Even small increases in character-state space substantially benefit phylogenetic inference accuracy.
• The advantage of greater character-state space is amplified by unequal probabilities of change among states.
• While topology inference accuracy improved significantly, accurate estimation of tree length showed less improvement.
• Initial increases in character-state space yielded the greatest improvements in both accuracy and inferred tree length.

Conclusions:

• Expanding character-state space is a crucial factor for enhancing the accuracy of phylogenetic inference.
• The benefits of increased character-state space are particularly pronounced under conditions of unequal state change probabilities.
• Optimizing character-state space offers significant but diminishing returns; initial increases provide the most substantial gains.