A comparison of phylogenomic inference pipelines for low-coverage whole-genome sequencing in Formica ants

Abstract

A rapid proliferation in the availability of whole genome sequences (WGS), often with relatively low read depth, offers an unprecedented opportunity for phylogenomic advances using publicly available data, but there are several key challenges in applying these data. Using low-coverage WGS data for the ant species of Formica, we conducted detailed comparisons on two different analytical pipelines (reference-based vs. de novo genome assembly), four types of datasets (5-kbp-window, ultra-conserved element [UCE], single-copy ortholog [BUSCO] and mitogenome), and a series of analytical procedures (e.g. concatenation vs. coalescent analyses) to identify which are robust to typical WGS data. The results show that at a shallow scale of phylogenetic relationships of closely related species 5-kbp-windows from the reference-based pipeline and UCEs from the de novo assemblies are more successful than the BUSCOs in recovering informative markers for phylogenetic inference. Compared with concatenation analyses, coalescent analyses often resulted in disparate deeper relationships in the phylogeny. This study also uncovers evident mito-nuclear discordance and demonstrates genome-wide gene conflicts in phylogenetic signals, both pointing to possible incomplete lineage sorting and/or hybridization during the early, rapid radiation of Formica ants. Divergence dating analyses show that different types of data and analytical methods could result in inconsistent time estimates, highlighting the potential need for multiple approaches to better understand species divergence. The strengths and weaknesses of different analytical pipelines and strategies are discussed. Findings from this study provide valuable insights for large-scale phylogenomic projects using WGS data.