Phylogenetic Accuracy Under Non-Stationary and Non-Homogeneous Conditions: A Simulation Study.

Area of Science:

• Evolutionary Biology
• Phylogenetics
• Computational Biology

Background:

• Phylogenetic inference commonly assumes data evolves under Stationary, Reversible, and Homogeneous (SRH) conditions.
• Violations of SRH assumptions can lead to significant errors in phylogenetic reconstruction.
• Previous simulations often focused on extreme non-SRH conditions, not representative of typical empirical data.

Purpose of the Study:

• To simulate datasets under varying degrees of non-SRH conditions using empirically derived parameters.
• To examine the impact of assuming SRH conditions on phylogenetic inference when these assumptions are violated.
• To assess the robustness of phylogenetic inference methods to realistic evolutionary model violations.

Main Methods:

• Simulation of datasets mimicking real biological data under diverse non-SRH evolutionary conditions.
• Application of maximum likelihood inference to phylogenetically analyze simulated datasets.
• Evaluation of phylogenetic accuracy and systematic bias under different levels of SRH model violations.

Main Results:

• Maximum likelihood phylogenetic inference demonstrates considerable robustness across a broad spectrum of SRH model violations.
• Inference accuracy significantly degrades under conditions of extreme convergent evolution.
• Simulations using empirically derived parameters provide a more realistic assessment of model violation impacts.

Conclusions:

• The maximum likelihood method is generally reliable for phylogenetic inference even with moderate deviations from SRH conditions.
• Extreme convergent evolution represents a critical scenario where SRH assumption violations severely compromise phylogenetic accuracy.
• This study highlights the need for careful consideration of evolutionary models in phylogenetic analyses, especially when non-SRH processes are suspected.