Testing Empirical Support for Evolutionary Models that Root the Tree of Life

Area of Science:

• Evolutionary biology
• Bioinformatics
• Phylogenetics

Background:

• Constructing rooted Trees of Life (ToLs) requires direct methods optimizing character state information.
• Protein structural domains in proteomes are used to build ToLs with different evolutionary models.

Purpose of the Study:

• To empirically test the support for reversible and non-reversible evolutionary models in phylogenetic reconstruction.
• To evaluate model adequacy using character state reconstruction methods and proteomic datasets.

Main Methods:

• Utilized character state reconstruction methods on two published proteomic datasets.
• Compared fully-reversible models (additive characters, Wagner parsimony) with non-reversible models (unordered characters, asymmetric stepmatrices).
• Rooted unrooted trees using Weston's generality criterion for reversible models.

Main Results:

• Reversible models align with reconstructed character change frequencies and the distribution of serial homologies.
• Non-reversible models produce inconsistencies, placing organisms with large proteomes at the tree base and violating distance inequalities.
• Demonstrated model inadequacy of non-reversible approaches.

Conclusions:

• Fully-reversible models provide a more faithful representation of evolutionary history based on proteomic data.
• Non-reversible models exhibit significant limitations and can lead to erroneous phylogenetic reconstructions.
• Emphasizes the critical need to incorporate empirical evidence and avoid aprioristic assumptions in natural history reconstruction.