Robust termite phylogenies built using transposable element composition and insertion events

Phylogenetic trees are typically reconstructed using conserved sequence alignments.^1,2 Other genomic elements, such as transposable elements (TEs), make up a large fraction of eukaryotic genomes³ but are ignored for phylogenetic reconstruction, despite potentially containing phylogenetic information,^4,5 which could be used to resolve nodes that remain contentious. Here, we reconstructed accurate phylogenetic trees of 45 termites and two cockroaches using two types of characters derived from the TE landscape: (1) genome-wide presence and absence of 37,966 TE families and (2) presence/absence data of 37,966 TE families in the flanking regions of orthologous ultraconserved elements (UCEs), which was a proxy for TE insertions. The topologies of our TE-based phylogenetic trees were largely congruent with phylogenetic trees inferred from alignments of UCEs and single-copy orthologous genes, only differing for a few nodes variably reconstructed in other phylogenetic analyses. Notably, trees based on genome-wide TE family composition were more accurate than trees inferred from mitochondrial genome alignments, and trees based on TE family composition in regions flanking UCEs achieved comparable accuracy with trees inferred from single-copy orthologous gene alignments. Our results demonstrate that the TE landscape is phylogenetically informative, representing an additional set of markers for robust phylogenetic reconstructions, with potential use to resolve ambiguous nodes in the tree of life.