Including autapomorphies is important for paleontological tip-dating with clocklike data, but not with non-clock data

Area of Science:

• Paleontology
• Evolutionary Biology
• Phylogenetics

Background:

• Tip-dating, incorporating dated fossils into Bayesian phylogenetic inference, is a common method.
• Morphological datasets often exclude autapomorphies (unique character states) due to ascertainment bias in parsimony analyses.
• The impact of this ascertainment bias on tip-dating has not been previously studied.

Purpose of the Study:

• To investigate the effect of excluding autapomorphies on divergence time estimates in tip-dating.
• To determine if excluding autapomorphies introduces bias in phylogenetic dating.
• To evaluate potential solutions for ascertainment bias in tip-dating.

Main Methods:

• Tested the effect of autapomorphy exclusion using a morphological matrix of Carboniferous-Permian eureptiles.
• Simulated strict clock and non-clock datasets to assess the influence of evolutionary models.
• Evaluated the computational practicality of the ascertainment bias correction (Mkparsinf).

Main Results:

• Excluding autapomorphies had a negligible effect on divergence time estimates for the eureptile dataset.
• Significant changes were observed in morphological rate estimates and minor effects on topological and dating confidence.
• Autapomorphy exclusion biased dating estimates only in simulated strict clock datasets, not non-clock datasets.
• The ascertainment bias correction (Mkparsinf) was found to be computationally impractical for large datasets.

Conclusions:

• The exclusion of autapomorphies does not appear to bias divergence time estimates in tip-dating when evolutionary rates are variable (non-clocklike).
• Paleontologists should prioritize coding autapomorphies in morphological matrices to mitigate potential biases, especially when dealing with datasets that may approximate clocklike evolution.
• While theoretical corrections for ascertainment bias exist, their practical application is limited by computational constraints.