Reconstructing Phylogenetic Relationships Based on Repeat Sequence Similarities

Abstract

A recent phylogenetic method based on genome-wide abundance of different repeat types proved to be useful in reconstructing the evolutionary history of several plant and animal groups. Here, we demonstrate that an alternative information source from the repeatome can also be employed to infer phylogenetic relationships among taxa. Specifically, this novel approach makes use of the repeat sequence similarity matrices obtained from the comparative clustering analyses of RepeatExplorer 2, which are subsequently transformed to between-taxa distance matrices. These pairwise matrices are used to construct neighbour-joining trees for each of the top most-abundant clusters and they are finally summarized in a consensus network. This methodology was tested on three groups of angiosperms and one group of insects, resulting in congruent evolutionary hypotheses compared to more standard systematic analyses based on commonly used DNA markers. We propose that the combined application of these phylogenetic approaches based on repeat abundances and repeat sequence similarities could be helpful to understand mechanisms governing genome and repeatome evolution.

Surprising amount of stasis in repetitive genome content across the Brassicales

“…While a few relationships were established correctly, none of the resulting dendrograms mirrored the tree obtained through ASTRAL using transcriptome data. At the family level, similar to previous publications, we did observe mirroring relationships within some clades (Figure S3), while the overall dendrogram was still in conflict with the species phylogeny (Dodsworth et al 2015;Vitales et al 2020). Overall, we observed more agreement within genera, with the level of conflict increasing in higher taxonomic ranks, resulting in a poorly resolved tree across the Brassicales.…”

“…To test the congruence of the ASTRAL tree with a repeat clustering based tree, we performed hierarchical clustering of TE abundances for Copia and Gypsy elements, as well as total TE content. We specifically highlight these superfamilies since previous studies which have used TE abundances to reconstruct phylogenetic trees found that Copia and Gypsy elements bear the strongest signal (Dodsworth et al 2015;Harkess et al 2016;Dodsworth et al 2017;Vitales et al 2020). However, we were unable to reproduce the species tree using our TE abundance data (Figure S2).…”

“…One possibility is related to certain technical compromises that were necessary in order to run such a large number of species in Transposome. We did not attempt to pool all 71 species into a single clustering analysis, such as performed in smaller species groups like eight species of Asparagus (Harkess et al 2016), six Nicotiana species (Dodsworth et al 2015;Dodsworth et al 2017;Vitales et al 2020), and nine Fabeae species (Dodsworth et al 2015;Vitales et al 2020). Similarly, we did not adjust for genome size due to the large number of sampled species and drastic genome size differences across the samples.…”

“…The abundances of TEs in genomes have also been shown to be informative when inferring phylogenetic relationships among taxa, especially in groups rife with polyploidy, such as those in the Brassicales, because it adds an analysis complementary to species tree phylogenies (Dodsworth et al 2015;Harkess et al 2016;Dodsworth et al 2017;Vitales et al 2020). Several studies have used maximum parsimony methods to reconstruct phylogenetic trees, treating TE abundances as continuous characters (Dodsworth et al 2015;Dodsworth et al 2017).…”

“…The resulting trees are largely concordant to those produced via traditional phylogenetic methods (i.e., using protein-coding genes). More recently, a study has shown the power of combining TE sequence similarity with TE abundance to understand evolutionary relationships (Vitales et al 2020). Generally, Copia and Gypsy are the most informative elements due to their high abundance in the genomes, whereas low-abundant TEs are insufficient to resolve the phylogenetic relations with these approaches (Dodsworth et al 2015;Harkess et al 2016;Dodsworth et al 2017;Vitales et al 2020).…”