Phylogenomic analyses of Crassiclitellata support major Northern and Southern Hemisphere clades and a Pangaean origin for earthworms

Abstract

Background: Earthworms (Crassiclitellata) are a diverse group of annelids of substantial ecological and economic importance. Earthworms are primarily terrestrial infaunal animals, and as such are probably rather poor natural dispersers. Therefore, the near global distribution of earthworms reflects an old and likely complex evolutionary history. Despite a long-standing interest in Crassiclitellata, relationships among and within major clades remain unresolved.

Methods: In this study, we evaluate crassiclitellate phylogenetic relationships using 38 new transcriptomes in combination with publicly available transcriptome data. Our data include representatives of nearly all extant earthworm families and a representative of Moniligastridae, another terrestrial annelid group thought to be closely related to Crassiclitellata. We use a series of differentially filtered data matrices and analyses to examine the effects of data partitioning, missing data, compositional and branch-length heterogeneity, and outgroup inclusion.

Results and discussion: We recover a consistent, strongly supported ingroup topology irrespective of differences in methodology. The topology supports two major earthworm clades, each of which consists of a Northern Hemisphere subclade and a Southern Hemisphere subclade. Divergence time analysis results are concordant with the hypothesis that these north-south splits are the result of the breakup of the supercontinent Pangaea.

Conclusions: These results support several recently proposed revisions to the classical understanding of earthworm phylogeny, reveal two major clades that seem to reflect Pangaean distributions, and raise new questions about earthworm evolutionary relationships.

Keywords: Clitellata; Crassiclitellata; Earthworm; Phylogenomics.

Fig. 1

PhyloBayes 50%-majority-rule consensus phylogram for the 75% data set (59 loci, 16,458 amino acid characters, CAT-GTR model, 500-generation burn-in; see text for details). Posterior probabilities are shown at nodes; nodes without values have posterior probabilities of 1.0. Members of Metagynophora (Moniligastridae + Crassiclitellata) are highlighted in bold font; with Drawida sp. representing Moniligastridae. Transcriptomes downloaded from the Sequence Read Archive are labeled “(SRA)”. Crassiclitellate taxa are color coded by biogeographic region; Dichogaster saliens and Pontodrilus litoralis are cosmopolitan species. Dates for nodes labeled 1 and 2 were estimated with PhyloBayes; see text for details

Fig. 2

a Strict consensus tree of ML trees for unfiltered, filtered (with TreSpEx and BaCoCa) and deleted outgroup 75, 50 and 25% data matrices and two 75% data matrices from which ?Haplotaxidae sp. was deleted (one with all sites included, the other with sites with >50% gaps deleted) (eleven analyses/trees total). Numbers at nodes highlight well-supported discrepancies across analyses. Taxa in bold black or gray were not included in the “deleted outgroup” analyses. b Resolutions of conflicting clades across eleven analysis/matrix combinations, with ML bootstrap values or ranges shown

Fig. 3

Quartet-based supernetwork based on all single-gene ML trees for the 75% data set (49 genes). Geographically delineated subclades of the main crassiclitellate clade (the clade subtending node 2 in Fig. 1) are circled

Fig. 4

Summary tree of relationships among the taxa included in this study, highlighting the names of relevant higher taxa. The topology shown is a strict consensus of all trees recovered for the 25, 50 and 75% data matrices generated in this study (filtered with TreSpEx and BaCoCa and unfilitered, with and without long-branch outgroups, etc.), with families and major taxa highlighted. The two numbered nodes were the focus of divergence time estimation. See Fig. 2 and text for details