Retroposon Insertions within a Multispecies Coalescent Framework Suggest that Ratite Phylogeny is not in the ‘Anomaly Zone’

Abstract

coalescence methods were developed to address the negative impacts of incomplete lineage sorting on species tree estimation with concatenation. Coalescence methods are statistically consistent if certain requirements are met including no intralocus recombination, neutral evolution, and no gene tree reconstruction error. However, the assumption of no intralocus recombination may not hold for many DNA sequence data sets, and neutral evolution is not the rule for genetic markers that are commonly employed in phylogenomic coalescence analyses. Most importantly, the assumption of no gene tree reconstruction error is routinely violated, especially for rapid radiations that are deep in the Tree of Life. With the sequencing of complete genomes and novel pipelines, phylogenetic analysis of retroposon insertions has emerged as a valuable alternative to sequence-based phylogenetic analysis.Retroposon insertions avoid or reduce several problems that beset analysis of sequence data with summary coalescence methods: 1) intralocus recombination is avoided because retroposon insertions are singular evolutionary events, 2) neutral evolution is approximated in many cases, and 3) gene tree reconstruction errors are rare because retroposons have low rates of homoplasy. However, the analysis of retroposons within a multispecies coalescent framework has not been realized. Here, we propose a simple workaround in which a retroposon insertion matrix is first transformed into a series of incompletely resolved gene trees. Next, the program ASTRAL is used to estimate a species tree in the statistically consistent framework of the multispecies coalescent. The inferred species tree includes support scores at all nodes and internal branch lengths in coalescent units. As a test case, we analyzed a retroposon dataset for palaeognath birds (ratites and tinamous) with ASTRAL and compared the resulting species tree to an MP-EST species tree for the same clade derived from thousands of sequence-based gene trees. The MP-EST species tree suggests an empirical case of the 'anomaly zone' with three very short internal branches at the base of Palaeognathae, and as predicted for anomaly zone conditions, the MP-EST species tree differs from the most common gene tree. Although identical in topology to the MP-EST tree, the ASTRAL species tree based on retroposons shows branch lengths that are much longer and incompatible with anomaly zone conditions. Simulation of gene trees from the retroposon-based species tree reveals that the most common gene tree matches the species tree. We contend that the wide discrepancies in branch lengths between sequence-based and retroposon-based species trees are explained by the greater accuracy of retroposon gene trees (bipartitions) relative to sequence-based gene trees. Coalescence analysis of retroposon data provides a promising alternative to the status quo by reducing gene tree reconstruction error that can have large impacts on both branch length estimates and evolutionary interpretations.

Deep-Time Demographic Inference Suggests Ecological Release as Driver of Neoavian Adaptive Radiation

“…Herein, we combine estimates of CBL obtained using genome-scale indel data with information from timetrees and reconstructions of ancestral generation times to yield what we believe to be the first empirical estimates of the relative effective population size (Ne of ancestral lineages in very deep time. Our approach bears some similarity to recently described approaches [81,82] and those of Tiley et al [83], who estimated CBL using binary data (TE insertions and "perfect" transversions, respectively). However, we implemented these analyses in a modified version of the commonly used multispecies coalescent program Accurate Species TRee Algorithm (ASTRAL) [84,85].…”

“…Perhaps as importantly as questions of alignment and genome assembly, indels lag far behind nucleotides in terms of 'substitution' models that faithfully describe differences in rates of insertion versus deletion, as well as differences in rate, frequency, and homoplasy associated with differences in indel length and mechanism of gain or loss. We are only beginning to learn their behavior in multispecies coalescent analysis [82,89].…”

Whole genome phylogeny of Gallus: introgression and data-type effects

“…However, the congruence between estimates of the species tree generated using ASTRAL on gene trees and those generated using perfect transversions indicates that intralocus recombination cannot be distorting our topology because the transversion data treats each perfect transversion site as a partially resolved gene tree. Scoring biallelic markers as partially resolved is expected to yield accurate estimates of coalescent branch lengths as long as the markers are truly homoplasy-free (Springer and Gatesy 2019). However, our observation that perfect transversion branch lengths are shorter than those estimated using gene trees (Table 2) suggests that the transversion data are unlikely to be homoplasy-free.…”

An ABBA-BABA Test for Introgression Using Retroposon Insertion Data

“…2). First, we employed the protocol outlined previously by Springer and Gatesy (2019). Each retroposon matrix of presence/absence characters was converted into a set of incompletely resolved 'gene trees' with each gene tree representing a single retroposon character and supporting just a single bipartition.…”

“…Retroposon insertions have commonly been analyzed using one or more variants of parsimony (e.g., Camin-Sokal parsimony [Nikaido et al 1999]; polymorphism parsimony [Suh et al 2015]; Dollo parsimony [Lammers et al 2019]), distance methods (e.g., neighbor joining [Lammers et al 2019]), or networks (e.g., Doronina et al, 2017a;Lammers et al 2019). More recently, Springer and Gatesy (2019) suggested that retroposons have several properties that are well matched to the assumptions of summary coalescent methods such as ASTRAL (Mirarab and Warnow 2015). First, summary coalescent methods assume that ILS is the only source of gene tree heterogeneity.…”