Partial degeneration - a solution to LBA and saturation problems
(English)Manuscript (preprint) (Other academic)
Background: There are many factors to consider when studying evolutionary relationships. The effect of substitution saturation and compositional heterogeneity may result in erroneous relationships and long-branch attraction artifacts (LBA) are all too real for a researcher in this field. A well-used method of avoiding LBA is removal or degeneration of the faster evolving third codon positions in protein coding genes and while successful, it possibly reduces valuable information in the process. We want to test the effectiveness of degenerating synonymous sites (i.e. degen1 coding) used on only a selective part of a dataset and how it compares to the common methods otherwise used. Testing these methods, two protein coding gene datasets covering Hymenoptera and Diptera, differing in size, hierarchal level and evolutionary age, was analyzed using maximum likelihood.
Results: Both data sets, with high support across the trees, displayed LBA in the early lineages when all codon positions were included. Both methods of dataset alteration yielded similar and more likely trees, but the partial degeneration method showed higher number of supported nodes and as well as a higher median support.
Conclusions: Partial degeneration is able to solve LBA artifacts and is a more efficient method to use when data is poorly sampled or suffers from substitution saturation and/or compositional heterogeneity.
saturation, Long-branch attraction, phylogeny
Research subject Systematic Zoology
IdentifiersURN: urn:nbn:se:su:diva-127993OAI: oai:DiVA.org:su-127993DiVA: diva2:912056