The problem of reconciling earthworm taxonomy and phylogeny has shown advances with the application of molecular techniques, yet they have proven insufficient. Integrative systematics could solve this by combining multiple sources of evolutionary information. Relatively low diversity, restricted range and low nomenclatural conflict make Hormogastridae Michaelsen, 1900 a desirable target for an integrative systematics approach. The main systematic conflicts within this family are the polyphyly of the species Hormogaster pretiosa Michaelsen, 1899, the widespread presence of cryptic lineages, the lack of resolution of supraspecific relationships and the paraphyly of the genus Hormogaster Rosa, 1877 (found to be composed of four well-supported genus-level clades by molecular phylogenetic inference). This work integrates all the existing information by performing phylogenetic inference based on morphological, molecular and total evidence datasets, comparing their performance with the topology obtained by phylogenomic analyses. It also includes a comparative study of representatives of the main clades based on microcomputed tomography (µCT) reconstructions. The addition of morphological characters improved the resolution of the Hormogastridae tree; ancestral state reconstruction displayed the evolution of character states and provided morphological diagnoses for the genera within a new system, which incorporates information about ecological niches and biogeography.

The earthworm family Hormogastridae is a relatively diverse group in the Western Mediterranean basin. Since 1887, around thirty species have been described and assigned to four genera. However, from 2010 on, molecular, ecological and morphological studies have questioned the validity of those genera. Meanwhile, new species were discovered and assigned to them, pending a formal systematic revision; such a revision has been performed recently by integrating all the existing sources of information. The resulting classification consists of nine genera, including four newly erected ones. This revised systematic background is used in the current work as a base for the description of six new hormogastrid species: Diazcosinia sacrarium Marchán, Fernández, Díaz Cosín & Novo, sp. nov., Boucheona martae Marchán, Fernández & Díaz Cosín, sp. nov., Boucheona rosae Marchán, Díaz Cosín & Novo, sp. nov., Norana emiliae Marchán, Fernández, Díaz Cosín & Novo, sp. nov., Norana xylocerasi Marchán, Fernández, Díaz Cosín & Novo, sp. nov., and Norana beatrizae Marchán, Fernández, Díaz Cosín & Novo, sp. nov. Norana is a new replacement name for the preoccupied Nora Marchán, Fernández, Díaz Cosín & Novo, 2018. Likewise, Xanina is proposed to replace the preoccupied Xana Díaz Cosín, Briones & Trigo, 1989. We provide an overview of the currently known diversity of the different genera, and we further propose common names in several languages for some of the species of Hormogastridae.

Corsica and Sardinia are amongst the largest islands of the Western Mediterranean. Their complex geological history included belonging to the European–Iberian continental margin (close to current-day Catalonia and Provence) and varying degrees of isolation for the last 30 million years, leading to peculiar, highly endemic faunas and floras. This is especially true for their earthworm faunas, which include endemic species of several Lumbricoidea genera and the endemic family (or subfamily) Diporodrilidae. Only three species have been described for the morphologically unique Diporodrilus, but there exists evidence for wide morphological variability within them and the existence of several species-level genetic lineages within Corsica. This work aimed to investigate the genetic diversity and phylogenetic relationships between the genetic lineages of Diporodrilus from Corsica and Sardinia (based on the sequences of 5 mito-nuclear markers), to perform an integrative systematics revision combining species delimitation techniques and morphological data, and to obtain a time-calibrated phylogeny of Diporodrilus and other Corso-Sardinian Lumbricoidea. Within 15 populations of the morphospecies Diporodrilus omodeoi and Diporodrilus pilosus across Corsica and Sardinia, 10 species-level genetic lineages were detected. Phylogenetic independence, high genetic divergence and morphological differences provided the support for the description of five new pseudocryptic species: Diporodrilus rotundus sp. nov., Di. jorgei sp. nov., Di. minor sp. nov., Di. meridionalis sp. nov. and Di. telti sp. nov. Time-calibrated phylogenetic inference estimated the age for genus Diporodrilus at 65.9 Mya; even if other Corso-Sardinian genera were significantly younger, all of them presented deep divergences predating the break-off of the microplate from the continent. The almost threefold increase in the known diversity of Diporodrilus stresses the need for comprehensive earthworm diversity inventories in both Sardinia and Corsica, and for studies on their ecological role and conservation status. The close correspondence between some geological and cladogenetic events suggest that the distribution of Corso-Sardinian earthworms could be used to inform standing geological controversies.

Although DNA barcodes-based operational taxonomic units (OTUs) are increasingly used in earthworm research, the relative efficiency of the different methods available to delimit them has not yet been tested on a comprehensive dataset. For this study, we used three datasets containing 651, 2304 and 4773 COI barcodes of earthworms from French Guiana, respectively, to compare five of these methods: two phylogenetic methods—namely Poisson Tree Processes (PTP) and General Mixed Yule Coalescence (GMYC)—and three distance matrix methods—namely Refined Single Linkage (RESL, used for assigning Barcode Index Numbers in the Barcode of Life Data systems), Automatic Barcode Gap Discovery (ABGD), and Assemble Species by Automatic Partitioning (ASAP). We found that phylogenetic approaches are less suitable for delineating OTUs from DNA barcodes in earthworms, especially for large sets of sequences. The computation times are unreasonable, they often fail to converge, and they also show a strong tendency to oversplit species. Among distance-based methods, RESL also has a clear tendency to oversplitting, while ABGD and ASAP are less prone to mismatches and have short computation times. ASAP requires less a priori knowledge for model parameterisation than AGBD, provides efficient graphical outputs, and has a much lower tendency to generate mismatches.