The family Hirudinidae, long taken for granted to be monophyletic, is not. Hirudinid leeches, characterized as relatively large, vermiform, swimming leeches that feed on blood by making an incision with three armed jaws, fall into two separate clades: one typified by the North American M. decora and the other by the European H. medicinalis. The Hirudinidae is represented by two independent origins of aquatic medicinal leeches, each from a terrestrial ancestor. Both groups create spongy cocoons that are deposited on shore, leaving the hatchlings to search for the water in a manner similar to newly hatched sea turtles. Also, both groups have internal insemination, a behavior common to terrestrial organisms to prevent sperm desiccation, unlike the aquatic leech families Glossiphoniidae and Piscicolidae that exhibit external traumatic insemination. The clade containing M. decora includes additional New World genera, such as the South American Oxyptychus, Semiscolex, and Patagoniobdella, as well as the North American Macrobdella, Philobdella, and Limnobdella. Unexpectedly, within this otherwise New World clade is the Old World bloodfeeding genus Limnatis distributed from Eastern Europe, throughout Africa, and eastward to the Indian subcontinent. The second hirudinid clade contains H. medicinalis and related genera found only in the Old World including Africa (Aliolimnatis and Asiaticobdella spp.), Asia (Hirudinaria spp., some Hirudo spp., and Whitmania spp.), Australia (Goddardobdella spp.), and Europe (Hirudo spp.). This polyphyly of the family Hirudinidae is further complicated by each of the two clades' inclusion of non-bloodfeeding taxa heretofore assigned to the family Haemopidae [15].
The deep divergence between the two hirudinid clades was hinted at by Borda and Siddall [17] in their findings that the Old World Limnatis nilotica (Savigny, 1822) placed closer to the North American M. decora than to other African species of the genus Aliolimnatis. With our addition of members of the genus Limnobdella that group sister to Limnatis species, the nature of this relationship is more precise. Prior work regarding the anticoagulant profiles of various medicinal leeches may have been prescient regarding polyphyly of the so-called "medicinal leeches". A variety of anticoagulants have been characterized from hirudinid leeches, with each compound targeting a different point in the clotting process [20, 21]. It is generally held that the major protease inhibitors employed by Hirudo species and their allies block thrombin, whereas that for M. decora targets platelet aggregation as opposed to the clotting cascade itself [e.g., [22, 23]]. Regarding the close association of Old World Limnatis species and New World Limnobdella species, generalized morphological similarities have previously been noted. Richardson and Oosthuizen lamented in personal letters (in the possession of MES) their inability to find a synapomorphy for the two genera that might allow them to erect a new family.
As noted above, in addition to the polyphyletic origin of the medicinal leeches, both hirudinid clades are paraphyletic in light of members of the family Haemopidae placing within each group. Previously, non-bloodfeeding, relatively large, vermiform, swimming leeches were grouped together on the basis of their macrophagous feeding behavior, regardless of geographic distribution. The family Haemopidae, among other non-bloodfeeding taxa, included the genera Haemopis, Whitmania, Semiscolex, and Patagoniobdella [15]. Our analyses demonstrate that this family is not phylogenetically corroborated because haemopid genera fall variously within the two independent hirudinid clades, thus rendering them paraphyletic. Whitmania laevis is sister to a bloodfeeding species within the genus Hirudo, and not even monophyletic with the other nearby non-bloodfeeding species of Haemopis. The macrophagous genera Semiscolex and Patagoniobdella, while monophyletic, are sister to a clade containing the sanguivorous taxa, Oxyptychus, Macrobdella and Philobdella. Though the ancestral hirudinid was clearly a bloodfeeder [17], what is remarkable is the number of times that sanguivory has been abandoned by this group of annelids otherwise notorious for its ectoparasitic dependence on vertebrate blood. Already the loss of sanguivory has been inferred for other groups of leeches such as Erpobdellidae, with a predilection for chironomid larvae, and the glossiphoniid genera Helobdella, Glossiphonia, and Alboglossiphonia that prefer the hemolymph of gastropods or other annelids. Even the terrestrial haemadipsid, Idiobdella seychellensis Harding, 1913 shifted away from feeding on blood on remote islands where terrestrial gastropods are more plentiful (and often larger) than resident anurans [24].
To reflect the phylogeny, the family Hirudinidae sensu stricto must hereafter exclude those bloodfeeding taxa unrelated to H. medicinalis and minimally includes those more closely related sanguivores [e.g., Hirudo, Goddardobdella, Hirudinaria, Aliolimnatis, Asiaticobdella included here], but must also include the non-sanguivorous genera Haemopis and Whitmania if leech taxonomy is to avoid both polyphyly and paraphyly of this family. The remaining genera previously included in the family Hirudinidae are in want of a unifying taxonomic name. Macrobdellidae [14] could include the genera Macrobdella, Philobdella, and Oxyptychus so as to reflexively retain a family for the non-bloodfeeding Semiscolescidae (Sciban & Autrum, 1934), their sister taxon. Yet, this would leave the genera Limnatis and Limnobdella without a synapomorphy for any family that would be required to include them. Conveniently, the Hirudinidae sensu stricto are easily differentiated from the hirudinid clade typified by M. decora by virtue of their profoundly muscular ejaculatory bulbs in the median male reproductive apparatus that are efferent to the epididymes; a characteristic Hirudinidae shares with the Haemadipsidae. In the absence of a clear morphological synapomorphy for the Limnobdella/Limnatis clade, we acknowedge that the genera Macrobdella, Philobdella, Oxyptychus, Limnobdella, Limnatis, and Semiscolex could presently be considered genera in the family Semiscolescidae (sensu lato), in that this family has taxonomic priority over the alternatives. Ironically, such a revision would leave the characteristically bloodfeeding Hirudinidae encompassing some non-bloodfeeding taxa and the traditionally non-bloodfeeding family Semiscolescidae (sensu lato) including notable bloodfeeders.
The genus Patagoniobdella is, by virtue of its relationships, merely a junior synonym of Semiscolex. Asiaticobdella fenestrata (Moore, 1939) falls within the genus Aliolimnatis. It is likely that these two genera will have to be synonymized, though we are presently reluctant in the absence of either of the type species for the genera. Similarly, though W. laevis falls within the genus Hirudo, formal revision should require the inclusion of the type species, Whitmania pigra (Whitman, 1884).
Both H. nipponia and L. nilotica are known to include multiple morphological variants [25] (Oosthuizen notes in the possession of MES) over a wide distribution (the latter from Eastern Europe through the entire continent of Africa and parts of India, and the former throughout much of East Asia) and so most likely these each represent multiple lineages. Notably, our determinations of the identity of leeches matching the description of L. nilotica represent a paraphyletic assemblage relative to L. paluda. More sampling across the range of these taxa is needed in order to better define lineages and distinguish potentially cryptic species.
While there are no fossil data for correlation in historical interpretations of the Hirudinidae, geologic events can be used as a rough estimate when considering the current distributions of leech taxa. Assuming a vicariance-dominated explanation, both clades would have had to originate on Pangea with significant diversification in all groups prior to the supercontinent's breakup. The Semiscolecidae-related group seems to have originated in South America with diversification into the clades containing Oxyptychus, Semiscolex, and Patagoniobdella on that continent before approximately three Mya when North and South America became proximal. Thereafter, the lineage leading to Macrobdella and Philobdella could have dispersed north, a pattern mirrored in other leech groups, such as Helobdella and Haementeria [26]. Some diversification would have had to occur prior to the breakup of Pangea in order to explain the presence of the genus Limnobdella in the New World and the genus Limnatis in Old World locales. Long distance dispersal of some ancestral Limnatis or Limnobdella species should be considered, though presently this is only known for terrestrial leeches in the family Haemadipsidae feeding on birds.
The clade containing H. medicinalis also seems to have undergone an intense period of diversification around the time of the breakup of Pangea. The node joining the Aliolimnatis/Asiaticobdella, Hirudinaria, and Goddardobdella clades is short and unstable suggesting a rapid diversification associated with the continental breakup of Pangea during the Cretaceous. Closely related taxa from Africa, Australia, and Southeast Asia follow a Gondwanan vicariance distribution, distinctly separate from the Laurasian Haemopis/Hirudo sector of the Hirudinidae sensu stricto. The sister group relationship of H. nipponia and W. laevis reflects the geologic history of Asia with their northerly origin in Laurasia and a later dispersal of the non-bloodfeeder into southern regions following a period of isolation from the remaining Hirudinidae by the presence of the Turgai Sea (93 - 89 Mya) [27]. The unusual recent distribution of H. manillensis in the Caribbean closely related to the others from (for example) Thailand can only be explained by H. manillensis having been introduced to the Caribbean in the 1800s by physicians using leeches on board galleons transporting goods and persons between Spanish holdings in the Pacific and the New World [28, 29]. Clarity regarding this potentially invasive species might be better assessed through haplotype analyses involving individuals from the Philippines and Northern Taiwan, which were under Spanish influence when leech phlebotomy was heavily practiced by European surgeons.
Despite extensive collection efforts, the type species of several genera in the family Hirudinidae have not been included in this analysis. These include Aliolimnatis diversa Richardson, 1972, Asiaticobdella birmanica (Blanchard, 1894), Semiscolex juvenilis Kinberg, 1866, and Whitmania pigra (Whitman, 1884). As such, definitive segregation of genera, and even their proper familial designations remain underdetermined. Approximately 15 genera, an inordinate numberof which are monotypic taxa from Australia described by Richardson [14], are not yet included in phylogenetic analyses. We anticipate that the addition of these and the multitudinous, however poorly distinguished, species described by Sciacchitano from Africa [e.g., [30–32]], might yet provide better support for some nodes, and further our understanding of the interrelationships of these medically important annelids.