An ABBA-BABA Test for Introgression Using Retroposon Insertion Data

DNA sequence alignments provide the majority of data for inferring phylogenetic relationships with both concatenation and coalescence methods. However, DNA sequences are susceptible to extensive homoplasy, especially for deep divergences in the Tree of Life. Retroposon insertions have emerged as a powerful alternative to sequences for deciphering evolutionary relationships because these data are nearly homoplasy-free. In addition, retroposon insertions satisfy the no intralocus recombination assumption of summary coalescence methods because they are singular events and better approximate neutrality relative to DNA sequences commonly applied in phylogenomic work. Retroposons have traditionally been analyzed with phylogenetic methods that ignore incomplete lineage sorting (ILS). Here, we analyze three retroposon data sets for mammals (Placentalia, Laurasiatheria, Balaenopteroidea) with two different ILS-aware methods. The first approach constructs a species tree from retroposon bipartitions with ASTRAL, and the second is a modification of SVD-Quartets. We also develop a {chi}2 Quartet-Asymmetry Test to detect hybridization using retroposon data. Both coalescence methods recovered the same topology for each of the three data sets. The ASTRAL species tree for Laurasiatheria has consecutive short branch lengths that are consistent with an anomaly zone situation. For the Balaenopteroidea data set, which includes rorquals (Balaenopteridae) and gray whale (Eschrichtiidae), both coalescence methods recovered a topology that supports the paraphyly of Balaenopteridae. Application of the {chi}2 Quartet-Asymmetry Test to this data set detected 16 different quartets of species for which historical hybridization may be inferred, but significant asymmetry was not detected in the placental root and Laurasiatheria analyses.