Variation Across Mitochondrial Gene Trees Provides Evidence for Systematic Error: How Much Gene Tree Variation Is Biological?

Abstract

The use of large genomic data sets in phylogenetics has highlighted extensive topological variation across genes. Much of this discordance is assumed to result from biological processes. However, variation among gene trees can also be a consequence of systematic error driven by poor model fit, and the relative importance of biological vs. methodological factors in explaining gene tree variation is a major unresolved question. Using mitochondrial genomes to control for biological causes of gene tree variation, we estimate the extent of gene tree discordance driven by systematic error and employ posterior prediction to highlight the role of model fit in producing this discordance. We find that the amount of discordance among mitochondrial gene trees is similar to the amount of discordance found in other studies that assume only biological causes of variation. This similarity suggests that the role of systematic error in generating gene tree variation is underappreciated and critical evaluation of fit between assumed models and the data used for inference is important for the resolution of unresolved phylogenetic questions.

Phylogenomics — principles, opportunities and pitfalls of big‐data phylogenetics

“…In node-dating approaches, estimating divergence times requires a phylogenetic tree and associated branch length information, and a list of fossil taxa of reliable age and phylogenetic placement to be used as calibration points. Depending on the focal organism, there may be little information in the Betancur-R (2013); Copetti (2017); Richards (2018) fossil record, such that obtaining fossils with reliable age and phylogenetic placement may be challenging. Therefore, palaeontological literature should be carefully reviewed so that fossil placement is critically evaluated.…”

FrogCap: A modular sequence capture probe set for phylogenomics and population genetics for all frogs, assessed across multiple phylogenetic scales

“…The widespread adoption of high-throughput DNA sequencing technologies to address questions in evolutionary biology has revealed new challenges for analyzing these expansive genomic datasets (Shendure & Ji 2008;Kircher & Kelso 2010;Jones & Good 2016). Despite the increased ability to collect large phylogenomic datasets, many phylogenetic relationships remain ambiguous because of both biological and non-biological processes that can lead to gene trees that are discordant with the species tree (Knowles 2009;Liu et al 2009;Hobolth et al 2011;Blom et al 2017;Hahn & Nakhleh 2016;Reddy et al 2017;Knowles et al 2018;Richards et al 2018). Further exacerbating this challenge, data from different genomic regions can demonstrate conflicting phylogenomic results Lemmon & Lemmon 2013;Chen et al 2017;Reddy et al 2017).…”

“…Additionally, natural selection of protein coding and other selected regions can be problematic for phylogenetic inference if phylogenetic signal is reduced as a result of selection (Castoe et al 2009;Edwards 2009;Liu et al 2009;Hobolth et al 2011). Discordant relationships can also be caused by non-biological mechanisms, such as gene tree estimation error resulting from model inadequacy, short alignment lengths, low levels of phylogenetic informativeness, anomaly zones, or errors in sequence assembly or alignment (Xi et al 2015;Hahn & Nakhleh 2016;Blom et al 2017;Reddy et al 2017;Richards et al 2018). As a result, selecting the best method and types of molecular markers to address these issues remains challenging.…”

Phylogenetic signal is associated with the degree of variation in root-to-tip distances

“…One approach to carrying out a phylogenomic study is to employ a criterion to select genes for analysis, a practice known as “data filtering” or “gene shopping” (Molloy and Warnow 2018). Some of the criteria that have previously been used for data filtering include phylogenetic branch supports (Blom et al 2016), the amount of missing data (Molloy and Warnow 2018), measures of substitution model adequacy (Duchêne et al 2018c; Richards et al 2018), and base composition (Dávalos and Perkins 2008; Martijn et al 2018). It not clear which of these criteria is the most effective (Molloy and Warnow 2018), but it is likely that no single criterion is universally applicable (Reddy et al 2017).…”