Molecular phylogeny and systematics of native North American lumbricid earthworms (Clitellata: Megadrili)

The family Lumbricidae is arguably the most well-known and well-studied earthworm group due to its dominance in the European earthworm fauna and its invasion in temperate regions worldwide. However, its North American members, especially the genus Bimastos Moore, 1893, are poorly understood. We revised the systematics of the genus Bimastos and tested the hypothesis of the monophyly of North American lumbricids using morphological characters and eight molecular markers. Phylogenetic analyses based on our extensive sampling of Bimastos and inclusion of Dendrodrilus and Allolobophoridella indicated a well-supported clade containing Bimastos and Eisenoides Gates, 1969, and provided the first evidence supporting that North American lumbricids are monophyletic. Assuming the available divergence time estimations and dating of land bridges are correct, it would suggest that the ancestor of this clade arrived North America through Beringia or the De Geer route during Late Cretaceous, and since then the clade has diverged from its Eurasian sister group, Eisenia. The peregrine genera Dendrodrilus and Allolobophoridella are nested within the Bimastos clade; we propose to treat them as junior synonyms of the genus Bimastos, and, contradictory to the commonly held belief of being European, they are indeed part of the indigenous North American earthworm fauna. Morphological characters, such as red-violet pigmentation, proclinate U-shaped nephridial bladders and calciferous diverticula in segment 10 further support this placement. The East Mediterranean–Levantine Spermophorodrilus Bouché, 1975 and Healyella Omodeo & Rota, 1989 are nested within the Dendrobaena sensu lato clade; therefore their close relationship with the North American Bimastos is refuted. Species fit the revised diagnosis of Bimastos are reviewed and keyed, and a new species, Bimastos schwerti sp. nov., is described.


Introduction
Earthworms of the family Lumbricidae are native to the Holarctic. They represent a keystone group of macrofauna in temperate soils, with about 30 common species spread globally by human activity [1]. The family currently has around 750 described species within approximately 40-60 genera [2][3][4]. While the majority of the species and genera are native to the Palearctic region, two genera, Bimastos Moore, 1895 and Eisenoides Gates, 1969, are generally believed to be native to North America [5], with nine and two described species, respectively. In the USA, species belonging to the two genera frequently co-occur with introduced European lumbricids, such as Lumbricus rubellus, Aporrectodea caliginosa, and Octolasion cyaneum [6][7][8], but usually at lower abundance.
The validity, taxonomic boundary and origin of Bimastos, aptly called a "systematic wastebasket" by Gates [9], have been widely debated [5,[10][11][12][13][14][15][16]. Taking into account previously overlooked morphological characters, including the shape and orientation of nephridial bladders and the structure and position of calciferous glands, Gates [5] argued that the name Bimastos should be restricted only for North American species. This concept of Bimastos was later supported by Omodeo and Rota [14], who separated the Balkanic and Anatolian Spermophorodrilus Bouché, 1975 and Healyella Omodeo & Rota 1989 from North American Bimastos. However, Omodeo and Rota's concepts of the three genera suffered severely from ambiguous morphological descriptions and overlapping diagnosis. For this reason, Zicsi [13,15] concluded that the three genera form a homogenous group, and both Spermophorodrilus and Healyella are junior synonyms of Bimastos. In contrast to Gates' restricted definition of Bimastos to the Nearctic, Zicsi's concept of Bimastos encompasses species not only from North America but also from the Balkans and the Anatolia, thus creating a biogeographic puzzle with questions on how this genus achieved its current native range of distribution.
The biogeographic puzzle concerning the two competing hypotheses of Bimastos is further complicated by the close affinity among Bimastos and two monotypic genera, Allolobophoridella and Dendrodrilus [17,18]. This affinity has been suggested in preliminary molecular analyses [19,20], and was recently confirmed in the multigene molecular phylogeny of Lumbricidae [21]. The genus Allolobophoridella was created to host two enigmatic species, Lumbricus eiseni Levinsen, 1884 and Allolobophora parva Eisen, 1874 [17]. The latter was soon transferred to Bimastos [3], making Allolobophoridella monotypic. In the past few decades, Allolobophoridella eiseni has been moved around among Allolobophora [22], Eisenia [15], and Bimastos [18] and has also been suggested to show affinities with several Dendrobaena species [23]. Clearly, a phylogenetic re-evaluation of the species and the status of the genus Allolobophoridella were urgently needed.
While morphological similarities, such as proclinate U-shaped nephridial bladders, and Nearctic distributions imply that Bimastos and Eisenoides may be closely related, a hypothesized North American clade composed of only Bimastos and Eisenoides were put through a formal test only recently. Molecular phylogeny of Lumbricidae constructed by Dominguez et al. [21] showed that the North American Bimastos is monophyletic, and is nested within a clade consisting Bimastos, Dendrodrilus, and Allolobophoridella, with the former two genera being the sister groups of each other. While the hypothesis that Eisenoides is closely related to the aforementioned clade gained some support in Dominguez et al. [21], the inferred phylogeny did not support a strict Bimastos-Eisenoides monophyly. Moreover, the result was insufficient for making definitive inferences regarding the status of Bimastos, Healyella, and Spermophorodrilus as only three out of the nine nominal species of Bimastos were included, and the genus Spermophorodrilus and the type species of Healyella were missing from the analyses. Nevertheless, the molecular phylogenetic study of Lumbricidae by Dominguez et al. [21] provided a solid basis for further analysis. as endangered or protected. All of the specimens included in this study are archived in the institutions stated above and are publicly accessible.

Histological methods
For histological study of the longitudinal musculature, several postclitellar segments were embedded in paraffin, sliced to 10 μm thin cross-sections using a Microm rotary-microtome, and stained with hematoxylin and eosin [30]. For comparison of the structure of calciferous glands, several longitudinal sections of the preclitellar regions were also sliced and treated as above. The microscopic slides were examined and photographed using a Nikon Eclipse 660 DIC microscope. Taxon sampling for phylogenetic analysis To unravel the phylogenetic relationships among Bimastos, Healyella, Spermophorodrilus, Dendrodrilus and Allolobophoridella, and to test the hypothesis that North American lumbricids are monophyletic, we sampled a total of 14 taxa from the above genera, including two Bimastos species reported in Domínguez et al. [21], and both of the known species belonging to the North American native genus Eisenoides (Table 1). Our overall taxon sampling comprises eight of the 10 valid species of Bimastos, including a new species described in the present study, and fully represents the monotypic genera Allolobophoridella and Dendrodrilus. We also included nine species (three from GenBank; S1 Table) from Dendrobaena and Fitzingeria as the two genera are closely related and species in Healyella and Spermophorodrilus have been classified into Dendrobaena [11]. While our sampling of the latter four genera was far from exhaustive, Healyella, Spermophorodrilus and Fitzingeria have only 10, three, and three valid species, respectively [4]. Furthermore, our samples encompassed the type species of all of the targeted genera (Allolobophoridella eiseni, Bimastos palustris, Dendrobaena octaedra, Dendrodrilus rubidus rubidus, Eisenoides lonnbergi, Fitzingeria platyura platyura, Healyella syriaca, and Spermophorodrilus antiquus, respectively), providing a strong taxonomic basis for drawing unequivocal conclusions. Samples of the lumbricid species Eisenia fetida, Lumbricus rubellus, Octolasion lacteum, Aporrectodea caliginosa and Aporrectodea tuberculata were also included. Overall, sequences from 34 specimens representing 25 species/subspecies were newly acquired, and were combined with selected taxa from the Lumbricidae dataset reported in Domínguez et al. [21] and Pérez-Losada et al. [31] for phylogenetic analyses.

DNA extraction, polymerase chain reactions, and sequencing
Genomic DNA was extracted from earthworm tissues using the Qiagen DNeasy Blood and Tissue Kit (QIAGEN, Valencia, CA, USA). Regions of eight molecular markers, including three nuclear rRNAs (18S, 5.8S and 28S), two mitochondrial rRNAs (12S and 16S), and three mitochondrial protein coding genes (cytochrome c oxidase subunit 1 and 2 (COI, COII) and NADH dehydrogenase subunit 1 (ND1)), were acquired using polymerase chain reaction (PCR) with primers listed in S2

Phylogenetic analysis
Two datasets were analyzed. First, sequences from selected species representing the major clades of Lumbricidae and the outgroup Hormogaster reported in Domínguez et al. [21] and Pérez-Losada et al. [31] were combined with our data (S1 Table). The combined dataset, composed of 12S, 16S, 18S rRNAs, COI, COII, and ND1, contains 64 samples representing 57 species/subspecies and is 3940 bp after alignment (the 'short dataset' hereafter). Second, data acquired in this study were analyzed. Octolasion lacteum was used as the outgroup based on the inferred phylogeny in Domínguez et al. [21]. This dataset, 5715 bp after alignment, is longer (due to both longer sequences and some longer alignments) and allows us to focus on unravelling the phylogeny within Bimastos (the 'long dataset' hereafter).
Phylogenies were inferred using maximum likelihood (ML) analyses and Bayesian inferences. ML analyses were conducted using RAxML v8 [35] as implemented in the CIPRES Science Gateway 3.3 web portal [36](www.phylo.org) using the general time reversible model with proportion of invariable sites and gamma distribution (GTR + I + G) estimated for each individual gene partition. Clade support was evaluated using the non-parametric bootstrapping procedure with 1000 bootstrapping replicates. The best ML tree was compared to alternative tree topologies using the Shimodaira-Hasegawa (SH) test as implemented in RAxML v8. Bayesian inferences coupled with Marko chain Monte Carlo (MCMC) were conducted using MrBayes v3.2.6 [37] with default priors and random starting trees. Three independent MCMC searches, each with three heated and one cold chains, were run for 2 x 10 7 generations. The resulting trees were sampled every 1000 generations after discarding the first 20% trees as burn-in. The posterior probabilities and the topologies of the resulting consensus trees from separate analyses were compared for congruency and combined in a 50% majority-rule consensus tree.
We originally considered estimating divergence time and conducting ancestral area reconstruction but decided not to do so for three reasons. First, there are no earthworm fossils available for calibration. Molecular clock estimation in earthworms has been conducted exclusively using geological events [21,24,26]. However, doing so would imply vicariance, an assumption that has been repeatedly questioned [28]. Second, external calibration points are not available in our phylogenetic trees. Third, with the lack of native lumbricid samples from East Asia and a priori knowledge about the origin of Allolobophoridellaa eiseni and Dendrodrilus rubidus, ancestral area reconstruction cannot be properly conducted.

Nomenclatural acts
The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new names contained herein are available under that Code from the electronic edition of this article. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix "http://zoobank.org/". The LSID for this publication is: urn:lsid:zoobank.org:pub:236DD001-3F6B-4D1C-AAB2-D3E08FFEC0B3. The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital repositories: PubMed Central, LOCKSS.

Phylogeny
In both of the short (3940 bp) and the long (5715 bp) datasets, the maximum likelihood (ML) analyses and Bayesian inferences generate similar topologies and have no incongruence regarding supported clades (bootstrap support values ! 50 or posterior probabilities ! 0.90). Therefore, the ML and Bayesian trees are considered together and only the ML trees are shown.
The trees inferred from the short dataset (Fig 1) provided little support for some of the basal internal branches, which is what we expected as the genes included in the short dataset are only a part of those used in Domínguez et al. [21]. As our goal is to understand the phylogenies of Bimastos, Allolobophoridella, Dendrodrilus, Spermophorodrilus and Healyella with an extended sampling, the lack of phylogenetic resolution among other lumbricid genera does not affect our ability to draw meaningful conclusions, and can reasonably be compensated with our current understanding on Lumbricidae phylogeny [21,31].
The phylogenies inferred from both datasets strongly support a monophyletic group (the Bimastos clade hereafter) composed of Bimastos, Allolobophoridella, and Dendrodrilus (bootstrap/posterior probability values = 100/1.0 for both datasets). However, the genus Bimastos sensu Gates [5], is paraphyletic due to the exclusion of Allolobophoridella or Dendrodrilus. Within the Bimastos clade, Bimastos palustris is basal relative to all the other species. Bimastos parvus and Dendrodrilus rubidus are sister species; the two together are the sister group of Allolobophoridella eiseni (Fig 2). We compared the best ML tree (Fig 2) with two alternative topologies using the SH test with the following constraints: (1) monophyly of the Bimastos genus; and (2) monophyly of the Bimastos genus except B. parvus. Both comparisons suggested that the best ML tree and the two alternative topologies are not significantly different from each other (P > 0.05).
The sister group of the Bimastos clade is Eisenoides (bootstrap/posterior probability values = 95/1.0 and 97/1.0 for the short and long datasets, respectively). Together, the two genera form a clade that includes all lumbricid species of North American origin (the North American clade hereafter) (Figs 1 and 2). The phylogenetic trees inferred from the long dataset also suggested that Eisenia is the sister group of the North American clade (bootstrap/posterior probability values = 64/1.0).
The tree topologies do not support the hypothesis that Spermophorodrilus and Healyella are associated with Bimastos. Instead, the two genera form a weakly supported clade (bootstrap/ posterior probability values = 59/0.85 and 56/0.72 for the short and long datasets, respectively) nested within a clade (bootstrap/posterior probability values = 86/0.99 and 94/1.0 for the short and long datasets, respectively) that also includes Fitzingeria platyura platyura and nine Dendrobaena taxa. These nine taxa encompass all Dendrobaena species included in our analyses except Dendrobaena byblica byblica. Accordingly, Dendrobaena, as currently defined, is polyphyletic.

Taxonomic treatment
One of our main research goals was to revise the systematics of the genus Bimastos. The inferred phylogenies and the non-significant SH test result suggest two possible relationships among Bimastos, Dendrodrilus and Allolobophoridella: (1) Bimastos is paraphyletic due to the existence of Dendrodrilus rubidus and Allolobophoridella eiseni, or (2) Bimastos is monophyletic. Regardless of the relationship, the three groups formed a highly supported clade (Figs 1 and 2). Given the weak morphological distinction among these three genera [38], we prefer an unambiguously supported, more inclusive Bimastos, and herein propose to treat Dendrodrilus and Allolobophoridella as junior synonyms of Bimastos. This taxonomic treatment accommodated both scenarios of phylogenetic relationships. Furthermore, Mršič [38], in his original description of Allolobophoridella, noted: "Should it be stated, that in the species of the genus Bimastos from North America the glandular part of the nephridial bladder is oriented in the Bootstrap support values (if ! 50) and Bayesian posterior probabilities (if ! 0.90) are shown above and below the branches, respectively. The four genera that were previously hypothesized to be related to Bimastos are colored. The annotated-and-shaded areas on the right correspond to the two major clades characterized in Results. Specimens newly reported in this study were marked with their five-digit HNHM catalog numbers. Diagnosis. Setae strictly paired, pigmentation lacking (Fig 3). Prostomium epilobous, first dorsal pore around 5/6. Male pore on 15 large, just above setal line b, facing ventrad. Female pores small on 14 just above setae b. Clitellum annular, evenly developed, spermathecae and Bootstrap support values (if ! 50) and Bayesian posterior probabilities (if ! 0.90) are shown above and below the branches, respectively. The four genera that were previously hypothesized to be related to Bimastos are colored. The annotated-and-shaded areas on the right correspond to the two major clades characterized in Results. Specimens were marked with their five-digit HNHM catalog numbers. tubercles lacking. Nephridial pores irregularly alternate between b and above d. Two pairs of testes in 10, 11, and two pairs of seminal vesicles in 11, 12. Calciferous glands in segments 10-12, with small diverticula in segment 10. Excretory system holoic, nephridial bladders sausageshaped throughout. Typhlosole bifid, the cross-section of longitudinal muscle layer is of pinnate type.
Type species: Eophila antiqua Černosvitov, 1938 (= Spermophorodrilus albanianus Bouché, 1975) Distribution. From the Balkan Peninsula to North Anatolia. Remarks. Omodeo & Rota [14] noted that the species of Spermophorodrilus differed from the type species of the genus Bimastos, B. palustris, "in many relevant points" (p. 169). However, they did not discuss it in details, and the only difference mentioned is the one segment longer Revision of the earthworm genus Bimastos gizzard (17−19 vs. 17−18, respectively). Perhaps the presence of red pigment in Bimastos was a key point, but in the Omodeo & Rota's diagnosis of Spermophorodrilus, they stated: "epidermis devoid or almost devoid of pigment", a statement that can reasonably be applied also to B. palustris, as the live specimen of B. palustris is almost devoid of pigment throughout its body except for the head dorsad.
Genus Diagnosis. Setae moderately paired or distant, pigmentation brownish or purple at least dorsad on the several first segments. Prostomium epilobous, first dorsal pore around 5/6. Male pore on 15 large, facing ventrad between ab or just above setal line b. Female pores small on 14 between setae ab or just above b. Clitellum annular, evenly developed, spermathecae and tubercles lacking. Nephridial pores irregularly alternate between b and above d. Two pairs of testes in 10, 11, and two pairs of seminal vesicles in 11, 12. Calciferous glands in segments 10 −12, with moderate diverticula in segment 10 (Fig 4). Excretory system holoic, nephridial bladders sausage-shaped throughout. Typhlosole lamellar or bifid, cross-section of longitudinal muscle layer is of pinnate type.
Type species: Allolobophora syriaca Rosa, 1893 Distribution. From West Anatolia to the Levant. Remarks. Similar to Spermophorodrilus, Omodeo and Rota's concepts of Healyella and Bimastos suffer overlapping diagnoses. The only distinguishing characters of Healyella were the position of the genital pores between setal line a and b and the calciferous glands devoid of lateral diverticles [14]. However, the description of Healyella naja [14] (p. 176) has male pores above b and the description of Healyella schweigeri [13] (described as Bimastos schweigeri) has calciferous diverticula in 10, make the defining characters of Bimastos and Healyella sensu Omodeo & Rota [14] overlapping. Description of the nephridial bladders, which is the definitive difference [5] between Bimastos and the two Eurasian genera, was entirely missing in Omodeo & Rota [14,45].
Genus Diagnosis. Setae strictly or moderately paired, pigmentation red-violet at least dorsad on the several first segments. Prostomium epi-or tanylobous, first dorsal pore around 5/6. Male pore on 15 large, just above setal line b. Female pores small on 14 just above setae b. Clitellum annular or saddle-shaped. Spermathecae usually lacking, if present, frequently empty in 9/10-10/11 and open in setal line c. Tubercles usually lacking, if present, indistinct bands on the ventral edge of the clitellum. Nephridial pores irregularly alternate between b and above d. Two pairs of testes in 10, 11, and two pairs of seminal vesicles in 11, 12 (sometimes lacking or three pairs in 9, 10, 11). Calciferous glands in segments 10-12, 13 with variable sized diverticula in segment 10 (Fig 4). Excretory system holoic, nephridial bladders in the anterior part of the body U-shaped with proclinate ental limb, that might partly merge with the ectal limb toward the hind end of the body. Typhlosole lamellar or bifid, the cross-section of longitudinal muscle layer is of pinnate or fasciculated type.
Type species: Bimastos palustris Moore, 1895. Distribution. Western Canada (Vancouver Island) and eastern USA. Several species are peregrine, and introduced all over the temperate regions.
Remarks. This extended definition of Bimastos includes Lumbricus eiseni Levinsen, 1884 (placed previously in Allolobophoridella by Mršić [17]) and also Dendrodrilus rubidus (Savigny, 1826). It keeps apart, however, the species of Healyella and Spermophorodrilus differing in the structure of the nephridial bladders, which are simple, sausage-shaped throughout the body in Healyella and Spermophorodrilus and U-shaped in at least the first several segments in Bimastos (Table 2). Although it is clear that Spermophorodrilus, Healyellla and Bimastos are distinct clades based on Domínguez et al. [21] as well as our study, and that there are numerous taxonomic descriptions in the literature [13][14][15], we have summarized the differences among these clades in Table 2 in order to resolve the long standing taxonomic confusions and help both taxonomists and ecologists better distinguishing them. In the genus Bimastos the phoral insemination prevails. It means that the spermathecae disappeared and the sperm of the copulating partner is stored in spermatophores [63,64]. This type of insemination is regarded as a plesiomorphy in annelids; however in Lumbricidae it is surely a secondary reversion from thecal (by means of spermathecae) insemination [63]. Spermatophores can be of variable shape from the falciform in B. palustris to the flattened spermatophores found in B. schwerti sp. n. and usually they are attached to the body in the region of either the male pores or the clitellum. As they are present only for a short period after copulation, the species without spermathecae were frequently regarded as parthenogenetic [65,66].
Remarks. With no knowledge on the nepheridial system of B. palustris and the majority of North American Bimastos, Mršić [17] kept his Allolobophoridella separated from Bimastos, but noted that if the nephridial bladders of the North American genus Bimastos is "oriented in the same way as in eiseni. . .. . .the genus Allolobophoridella will be just a synonym of Bimastos" [17]. It turns out Mršić was correct on this point. Pop [11] erroneously synonymised B. eiseni with B. parvus, causing a long lasting confusion. Perel [62] demonstrated that the two species names should not be in synonymy because eiseni has fasciculated type of musculature, whereas parvus has pinnate type of musculature. Furthermore, the prostomium of B. eiseni is tanylobous and not epilobous as in case of B. parvus. Our molecular results completely support this view.
Remarks. B. gieseleri is similar to B. tumidus, but differs from it in the presence of thickened septa, the bifid type of the typhlosole and by its larger size. Therefore, in line with other authors (e.g. [43,78,79]) we regard it as a valid species.

Bimastos lawrenceae
Remarks. Unfortunately, the original description is incomplete. Most importantly, the descriptions of the male pore and biometry are missing. Fender [56] compared B. lawranceae with B. zeteki and wrote that the new species is slightly smaller than B. zeteki, which is the largest species in the genus with 100-135 mm by 5-6 mm measures. Therefore B. lawranceae might be around 100 mm long. Diagnosis. Body length 60-70 mm, diameter 2-3 mm. Color reddish dorsally and paler ventrally. Prostomium epilobous, first dorsal pore in 5/6. Setae closely paired (Fig 3), glandular tumescences lacking. Clitellum saddle-shaped, ventrally extends to setae a on segments ½22, 23-32, ½33. Male pore on 15, equatorial just above setae b, surrounded by prominent glandular crescents bulging slightly into the neighbouring segments. Female pore on 14 small, dorsad of b. Nephropores irregularly alternate between b and above d. Flat spermatophores are scattered ventrally on the clitellar region (Fig 8). Septa on 6/7 slightly thickened, on 7/8-11/12 strongly thickened. Calciferous glands in segments 10-12 with moderate diverticula in Revision of the earthworm genus Bimastos segment 10. Excretory system holoic. Nephridial bladders U-shaped throughout, with forward-bent ental limbs. Typhlosole bifid with manicate limbs. There is no data on the structure of the longitudinal musculature.

Bimastos longicinctus (Smith and Gittins 1915
Remarks. Based upon the clitellar position, this species is very close to B. heimburgeri, but differs from it by its paler color, and the presence of thickened septa. The structure of the clitellum is also somewhat different, being almost circular in B. heimburgeri and ventrally incomplete in B. longicinctus. Some authors consider B. longicinctus to be a synonym of B. parvus (cf. [9,43]). However, the presence of thickened septa, larger size and longer clitellum clearly separate the two species and, as demonstrated in the present study, they are also genetically distinguishable.  Diagnosis. Body length 18-30 mm, diameter 1.5-2.5 mm. Color pale with reddish hints on dorsum. Prostomium epilobous, first dorsal pore in 5/6. Setae moderately paired (Fig 3), glandular tumescences lacking. Clitellum annular extends on segments 23-28. Male pore on 15, postsetal and ventral, just above setae b, surrounded by well-developed glandular crescents (Fig 9). Female pore on 14 small, slightly dorsad of b. Nephropores irregularly alternate between b and above d. Septa 7/8-8/9 slightly thickened. Crop in 15-16, gizzards small in 17-18. Calciferous glands in segments 10-12 with moderate diverticula in segment 10. Sperm duct open through a muscular copulatory chamber. Excretory system holoic. Nephridial bladders proclinate U-shaped throughout. Typhlosole small, lamellar, the cross-section of longitudinal muscle layer is of pinnate type (Fig 10).

Bimastos palustris
Remarks. The species, contrary to the widely accepted view [66], does not seem to be parthenogenetic but biparental with phoral insemination. Almost all specimens collected in April bore several falciform spermatophores attached around the male pore (Fig 9).

Discussion
Our phylogenetic analyses showed that the genus Bimastos as revised in this study is monophyletic and also includes the former Allolobophoridella eiseni and Dendrodrilus rubidus. Largely in agreement with Gates [5], the revised Bimastos is characterized by shared character states that include proclinate U-shaped nephridial bladders, the presence of calciferous diverticula in 10, and having porphyrin-based red pigments. Contrary to Zicsi [13] and Zicsi and Michalis [15], the genera Spermophorodrilus and Healyella are not related to Bimastos. They form a clade nested among several Dendrobaena species, and can be distinguished morphologically from Bimastos by their sausage-shaped nephridial bladders.
Close relationships among Bimastos, Allolobophoridella eiseni and Dendrodrilus rubidus have been consistently revealed in phylogenetic analyses based on DNA sequences [19][20][21]. With DNA data comparable in length (5715 bp) to that in Dominguez et al. [21] (5866 bp) while having extensive sampling of Bimastos for the first time, our results corroborated previous molecular results and led to our taxonomic treatment that both Allolobophoridella eiseni and Dendrodrilus rubidus indeed belong to the Nearctic genus Bimastos. Although B. eiseni and B. rubidus have been exclusively referred to as "European" and considered non-native in North America [9,55,91], there has been no genetic evidence suggesting that the two species are truly of European origin. Based on our phylogenetic analyses and the fact that the two species are also reported in North America, of which one is widespread [3,43,78,91], we herein propose that B. eiseni and B. rubidus, like B. parvus, are of North American origin and thus non-native in Europe. The discovery of an earthworm cocoon attributed to B. rubidus from lake sediment dated over 7,000 years old in Ontario, Canada [92] provides additional proof that the species was present thousands of years before Europeans had reached the continent.

Clitellum annular evenly developed also on the ventral side 2
Clitellum saddle shaped, ventrally always less developed 3 Bimastos as herein re-defined is the sister group of Eisenoides Gates, 1969, the other North American genus in the family Lumbricidae, confirming the 'Bimastos-Eisenoides clade that also contains Allolobophoridella and Dendrodrilus revealed in Dominguez et al. [21]. The two genera, therefore, form a clade that comprises exclusively North American Lumbricidae, suggesting for the first time that North American lumbricids are indeed monophyletic. This relationship is consistent with that presented in Dominguez et al. [21], and is supported by the proclinate U-shaped nephridial bladders shared by both genera.
Eisenia is the sister group of the North American clade both in our extensive dataset and in Dominguez et al. [21]. With the exception of a few Central European species (E. lucens, E. spelaea and E. balatonica), Eisenia is primarily a temperate Asian genus originally from Central and Northeast Asia (the Turanian-Far Eastern earthworm domain [93]), whereas the majority of the endemic North American lumbricids are distributed in eastern US [94]. This geographic distribution, an East Palearctic-East Nearctic disjunction, presents a biogeographic challenge that requires proper explanation.
Molecular clock estimation in Dominguez et al. [21] suggested that the North American Bimastos-Eisenoides clade diverged from the Eurasian Eisenia in the Late Cretaceous, about 72.6 (69.2-76.1) mya, providing some insights into several possible biogeographic scenarios. Three land bridges between Eurasia and North America during the Late Cretaceous (100-66 mya) and Paleocene (66-56 mya) have been proposed to explain distributions of fauna and flora across the two continents: Beringia (connecting Siberia and Alaska), the De Geer route (connecting Northern Europe and Northeastern North America), and the Thulean land bridge (connecting Northwestern Europe and Northeasten North America). When the divergence between closely related taxa is dated back to the Late Cretaceous-Paleocene, the majority of present-day East Palearctic-East Nearctic disjunctions in animals have been attributed to the North Atlantic land bridges (i.e. the De Geer and/or Thulean routes) [95], while those involved Beringia took place more recently (e.g. [96][97][98]). However, recent evidence has shown that the Thulean land bridge existed only around 57 and 56 mya [99], which is later than the estimated time of cladogenesis between Eisenia and the North America clade. Meanwhile, Beringia and the De Geer route existed earlier than the Thulean land bridge (Beringia: 100-75 mya, 65.5 mya, 58 mya; De Geer route: 71-63 mya; [95,99]), providing possible paths for faunal exchange. A viable hypothesis would thus require ancestral range expansion from Eurasia to North America through either Beringia or the De Geer route followed by extinction in western North America or Europe, respectively. The current native distribution of most Eisenia (Central Asia and Northeast Asia,) and the existence of Beringia during Late Cretaceous appear to favor the Beringia hypothesis, but dispersal through the De Geer route is logically equally likely and cannot be ruled out. To test these hypotheses requires more extensive sampling of Eisenia covering the native range of the genus and rigorous ancestral range reconstruction using likelihood-based evolution models.
The inferred phylogeny also suggested that Spermophorodrilus and Healyella form a weakly supported clade. As only three out of the 13 described species from the two genera were sampled, we were unable to conclude whether any of the two is monophyletic. In both of our short and long datasets, the Spermophorodrilus-Healyella clade, together with Fitzingeria platyura platyura, was nested among as many as eight Dendrobaena species, including the type species D. octaedra. Therefore, without synonymizing any of the other three genera, Dendrobaena as currently defined is at best paraphyletic. This heterogeneous genus has more than 100 described species [4] and, with some species falling outside of the "Dendrobaena clade" as revealed in this study, is in urgent need of revision. Our phylogenetic analyses encompassed the type species of all four genera, and from the nomenclature point of view, it is justifiable to synonymize the four names. However, we strongly oppose to do this, because (1) it would lead to losing evolutionarily meaningful morphological information that is currently included in the diagnosis of the respective genera, and (2) it would impede further research towards a comprehensive revision of the genus Dendrobaena. We also believe that any nomenclature act regarding the involved genera should only be done along with a full revision of the genus Dendrobaena.
The chaos in the systematics of Bimastos, Allolobophoridella, Dendrodrilus, Spermophorodrilus, and Healyella highlighted the difficulty of understanding character evolution and the respective synapomorphy and autapomorphy in earthworm systematics. The striking morphological similarity among Bimastos, Spermophorodrilus, and Healyella is an example of homoplasy. The annular clitellum was believed to be the defining synapomorphy uniting Bimastos and Healyella, [13,15], but it turned out to be acquired independently in the Bimastos clade and in the Spermophorodrilus-Healyella clade. All three genera shifted to phoral dissemination which resulted in disappearance of the spermathecae and tubercles. Disappearance of tubercles made it possible to develop an annular clitellum, which is more appropriate for cocoon formation. However, to facilitate the successful copulation, glandular genital ridges (pseudotubercles) developed in the male pore region [14,100].
The revised Bimastos genus shares proclinate U-shaped nephridial bladders, the presence of calciferous diverticula in 10, and porphyrin based red pigments among its species, while at the same time allows both tanylobous and epilobous prostomium and both pinnate and fasciculated longitudinal musculature in the genus. The presence of tanylobous prostomium, an autapomorphy in the case of Bimastos, was part of the reason why B. eiseni was originally classified as a Lumbricus [67], a genus whose species shares tanylobous prostomium as synapomorphy. Similarly, Pop [11] believed that inclusion of more than one type of musculature in a genus renders it polyphyletic. However, two types of musculature can be seen in Bimastos, as is also the case in Eisenia, in which the musculature of Eisenia lucens (Waga, 1857) is fasciculated while the other three species in Dominguez et al.'s [21] monophyletic Eisenia genus possess pinnate musculature [3].
The native range of most Bimastos species is currently restricted to Eastern North America, with the distribution of some species, such as Bimastos welchi, well extending into the Great Plains, states of Kansas and Colorado in the USA [5,91,101]. Gates [9] believed that the genus once had a much wider range that had significantly reduced due to glaciations. This hypothesis is supported by the discovery of Bimastos lawrenceae Fender, 1994 from Vancouver Island in Northwestern North America, a location far north of the last glacial boundary [56]. The fossil cocoon attributed to B. rubidus from lake sediment in Ontario, Canada [92] lends further support that Bimastos might have covered also the terrain that is now Canada, and its representatives were able to survive, at least in a few favorable refuges, during the Ice Age.
Our understanding of the peregrine Bimastos eiseni and B. rubidus, has been misled by the poor knowledge of the systematics of this genus, especially in Europe, where they have been mistakenly categorized as indigenous for more than a century [3,9]. The interception of the two species in imported soil at US ports was further viewed, again mistakenly, as direct evidence for introduction to North America [102]. It is now clear that like B. parvus, both B. eiseni and B. rubidus are native to North America, and, in contrast to the commonly held belief, should be categorized as introduced in Europe.
Bimastos parvus is considered a peregrine species that also has been intercepted in shipments at North American ports [9]. Similar to other peregrine earthworms, the three Bimastos species were likely to be transported back and forth between continents and even today can easily enter into countries without strict regulations on soil, plants and timber import. Both B. eiseni and B. rubidus have been found in different types of wetlands, along streams, and under bark of trees, rotting logs, pile of leaf litter, compost, moss, and dung [55,103]. It appears that cool, moist, sometimes acidic habitats with high organic matter content are optimal for these species [55]. This might explain the occurrence of B. rubidus at high elevations in the tropics [9,43] and at high latitudes, although temperature may still limit their distribution if it does not reach the degree days necessary for individual development [104]. It also highlights why they might travel well, as they were often intercepted in potted plants covered with Sphagnum moss.
We conclude that the combined morphological, molecular and biogeographical anaysis was necessary to reveal interesting, novel information about the phylogeny and distribution of the Bimastos genus. On the one hand, the classical, purely morphology-based taxonomy proved to be wrong in the past; on the other hand, relying on DNA analysis alone is insufficient to define genera. It is our hope that such integrated approach will lead to a more consistent system of Lumbricidae.
The study also highlights that not all peregrine lumbricid species in North America came from Europe; reverse introduction also occurred. Additionally, our results generated exciting biogeographical hypotheses to be tested in the future.
Supporting information S1