Nucleotide Usage Biases Distort Inferences of the Species Tree

Area of Science:

• Evolutionary Biology
• Phylogenetics
• Genomics

Background:

• Phylogenetic methods commonly assume neutral evolution, ignoring natural selection's role.
• Genome-wide processes like GC-bias and selection in coding regions may operate in the nearly neutral range.
• The impact of weak, pervasive selection on species tree inference using genome-wide data is largely unexplored.

Purpose of the Study:

• To develop and apply a polymorphism-aware phylogenetic model to assess the impact of nucleotide usage biases (selection) on species tree inference.
• To quantify how selection affects genetic distance estimations and molecular divergence times.
• To test the influence of GC-bias on phylogenetic relationships and divergence dating.

Main Methods:

• Development of a novel polymorphism-aware phylogenetic model.
• Tailoring the model to detect signatures of nucleotide usage biases.
• Application of the model to estimate species trees and divergence times for fruit fly populations.

Main Results:

• Inferred species relationships were largely robust to selection.
• Genetic distances were systematically underestimated, particularly at deeper nodes (node-height-dependent bias).
• GC-bias significantly impacted divergence time estimates (up to 23% under a neutral model).

Conclusions:

• Weak but pervasive selection, such as GC-bias, can bias phylogenetic inferences.
• Underestimation of genetic distances and divergence times necessitates accounting for selection in evolutionary studies.
• Accurate molecular dating and divergence quantification require incorporating selection into phylogenetic models.