Phylogenomics using low-depth whole genome sequencing: a case study with the olive tribe

Species trees have traditionally been inferred from a few selected markers, and genome-wide investigations remain largely restricted to model organisms or small groups of species for which sampling of fresh material is available, leaving out most of the existing and historic species diversity. The genomes of an increasing number of species, including specimens extracted from natural history collections, are being sequenced at low depth. While these datasets are widely used to analyse organelle genomes, the nuclear fraction is generally ignored. Here we evaluate different reference-based methods to infer phylogenies of large taxonomic groups from such datasets. Using the example of the Oleeae tribe, a worldwide-distributed group, we build phylogenies based on single-nucleotide polymorphisms (SNPs) obtained using two reference genomes (the olive and ash trees). The inferred phylogenies are overall congruent, yet present differences that might reflect the effect of the distance to the reference on the amount of missing data. To limit this issue, the genome complexity was reduced by using pairs of orthologous coding sequences as the reference, thus allowing combining SNPs obtained using two distinct references. Concatenated and coalescence trees based on these combined SNPs suggest events of incomplete lineage sorting and/or hybridization during the diversification of this large phylogenetic group. Our results show that genome-wide phylogenetic trees can be inferred from low-depth sequence datasets for eukaryote groups with complex genomes, and histories of reticulate evolution. This opens new avenues for large-scale phylogenomics and biogeographic analyses covering both the extant and historic diversity stored in museum collections