Evolutionary relationships have remained unresolved in many well-studied groups, even though advances in next-generation sequencing and analysis, using approaches such as transcriptomics, anchored hybrid enrichment, or ultraconserved elements, have brought systematics to the brink of whole genome phylogenomics. Recently, it has become possible to sequence the entire genomes of numerous non-biological models in parallel at reasonable cost, particularly with shotgun sequencing. Here we identify orthologous coding sequences from whole-genome shotgun sequences, which we then use to investigate the relevance and power of phylogenomic relationship inference and time-calibrated tree estimation. We study an iconic group of butterflies - swallowtails of the family Papilionidae - that has remained phylogenetically unresolved, with continued debate about the timing of their diversification. Low-coverage whole genomes were obtained using Illumina shotgun sequencing for all genera. Genome assembly coupled to BLAST-based orthology searches allowed extraction of 6,621 orthologous protein-coding genes for 45 Papilionidae species and 16 outgroup species (with 32% missing data after cleaning phases). Supermatrix phylogenomic analyses were performed with both maximum-likelihood (IQ-TREE) and Bayesian mixture models (PhyloBayes) for amino acid sequences, which produced a fully resolved phylogeny providing new insights into controversial relationships. Species tree reconstruction from gene trees was performed with ASTRAL and SuperTriplets and recovered the same phylogeny. We estimated gene site concordant factors to complement traditional node-support measures, which strengthens the robustness of inferred phylogenies. Bayesian estimates of divergence times based on a reduced dataset (760 orthologs and 12% missing data) indicate a mid-Cretaceous origin of Papilionoidea around 99.2 million years ago (Ma) (95% credibility interval: 68.6-142.7 Ma) and Papilionidae around 71.4 Ma (49.8-103.6 Ma), with subsequent diversification of modern lineages well after the Cretaceous-Paleogene event. These results show that shotgun sequencing of whole genomes, even when highly fragmented, represents a powerful approach to phylogenomics and molecular dating in a group that has previously been refractory to resolution.

Appendix S1 - Pipeline from raw reads to phylogenomics and dating analysis

Scripts used to perform the analyses presented in this study.

Appendix S2 - Dataset 1 with 760 genes (amino acids format)

Phylogenomic dataset of Papilionoidea including 760 orthologous genes in amino acid format (Dataset 1).

Appendix S3 - Dataset 2 with 6621 genes (amino acids format)

Phylogenomic dataset of Papilionoidea including 6,621 orthologous genes in amino acid format (Dataset 2).

Appendix S4 - Dataset 3 with 760 genes (nucleotides format with gene partitions)

Phylogenomic dataset of Papilionoidea including 760 orthologous genes in nucleotide format (Dataset 3).

Appendix S5 - Dataset 4 with 6407 genes (nucleotides format with gene partitions)

Phylogenomic dataset of Papilionoidea including 6,621 orthologous genes in nucleotide format (Dataset 4).

Appendix S6 - Phylogenomic trees

Phylogenomic trees of Papilionoidea inferred with both maximum-likelihood and Bayesian inference using 760 orthologous genes (Dataset 1) and 6,621 orthologous genes (Dataset 2).

Appendix S7 - tree files

Tree files of the molecular phylogenomic analyses of Papilionoidea as inferred with IQ-TREE and PhyloBayes.

Appendix S8 - Gene- and site-concordance and discordance factors

Appendix S8. Gene- and site-concordance and discordance factors estimated with the Datasets 1 and 3 (760-gene) and Dataset 2 (6621-gene).

Appendix S9 - Dated trees with GTS

Appendix S9. Bayesian dated trees of Papilionoidea inferred with the 760-gene dataset and the mixture model CAT-GTR (PhyloBayes).

Appendix S10 - Chronogram files

Appendix S10. Tree files of molecular divergence-time estimation of Papilionoidea as inferred with the PhyloBayes CAT-GTR model following four different analyses of the 760-gene dataset.

Appendix S11 - Prior and posterior distributions

Appendix S11. Comparison of prior and posterior distributions for nodes with set fossil calibrations. Bayesian posterior distributions are not driven by the uniform prior distributions used to calibrate the five nodes with fossil calibrations.

Appendix S12 - Phylogenomic tree with cross-contam

Appendix S12. Phylogenomic tree of Papilionoidea inferred with the Bayesian mixture model using an amino-acid dataset comprised of 2,993 orthologous genes selected without the cross-contamination check.

Appendix S13 - Correlation cross-conta vs. no cross-conta

Appendix S13. Correlations of branch lengths as inferred with the 2,993-gene (without CroCo) versus the 6,621-gene (with CroCo) datasets (a), and as inferred with the 760-gene versus the 6,621-gene datasets (both with CroCo) (b). Units are the number of substitutions per site per branch. Note the higher correlation (R2) obtained when comparing branch lengths between the 760-gene and 6,621-gene datasets with cross-contamination excluded.

Data from: Whole genome shotgun phylogenomics resolves the pattern and timing of swallowtail butterfly evolution

Data files

Abstract

Usage notes