Higher level relationships of leeches (Annelida: Clitellata: Euhirudinea) based on morphology and molecular gene sequences

Kathleen Apakupakul*, Mark E. Siddall+ and Eugene M. Burreson[daggerdbl]

*Department of Biology, University of Michigan, Ann Arbor, Michigan 48109; +Laboratory of Phylohirudinology, Museum of Zoology, University of Michigan, Ann Arbor, Michigan 48109; [daggerdbl]Virginia Institute of Marine Science, College of William and Mary, Gloucester Point, Virginia 23062.

Corresponding author:

Mark E. Siddall

Laboratory of Phylohirudinology

Museum of Zoology

University of Michigan

Ann Arbor, Michigan 48109


The evolutionary patterns of divergence of all euhirudinean families were investigated by cladistic analysis of 33 euhirudinidean species. Oligochaetes, Acanthobdella peledina, and branchiobdellidans were included as outgroup taxa. Cladistic analysis employed 1.8 kbp of nuclear 18S ribosomal DNA and 651 bp of mitochondrial cytochrome c oxidase subunit I in addition to morphological data. The use of two molecular data sets, one nuclear gene and one mitochondrial gene, as well as morphological data combined historical information evolving under a variety of different constraints and therefore was less susceptible to the biases that could confound the use of only one type of data. Analyses suggest that the nuclear 18S rDNA gene yields a meaningful historical signal for determining higher-level relationships. The more rapidly evolving mitochondrial gene was more informative in recent or local areas of the evolutionary hypothesis, such as within-family relationships. Analyses combining all data from the three character sets yielded one most-parsimonious tree. Most of the higher taxa in recent leech systematics were well-corroborated in the resulting topology. However, results suggested paraphyly of the order Rhynchobdellida, contradicting the presence of a proboscis as a synapomorphy. The medicinal leech family Hirudinidae was polyphyletic because the Haemadipsidae and the Haemopidae each have a hirudinid ancestor. In addition, all but one of the genera within the family Erpobdellidae must be either abandoned or renamed. Unusual findings included compelling evidence of historical plasticity in bloodfeeding behavior, having been lost at least four times in the course of euhirudinean evolution. Biogeographic patterns supported a North American origin for Arhynchobdellida.


INTRODUCTION

Euhirudinea, or the true leeches, are clitellate annelids exhibiting a marked scope of diversity, including ectocommensalism, parasitic sanguivory and predatory life-history strategies, as well as a variety of reproductive behaviors. They occur in habitats that range from terrestrial to aquatic (both marine and freshwater) environments and are found on all continents. The remarkable diversity in morphology and behavior among species of leeches has been of interest to several fields of biology. Sanguivorous leeches recently have been the focus of pharmaceutical companies seeking to expand their repertoire of anticoagulants, in light of their biomedically important salivary components and clinical use in microsurgeries to prevent blood clot formation (e.g., Walsmann and Markwardt, 1985; Lent, 1986; Munro et al., 1992). As well as being parasites themselves, some leech species serve as vectors of blood parasites for aquatic vertebrates. Blood parasites have been thought to have coevolved with their respective leech hosts as a result of long-term ecological associations (Siddall and Burreson, 1995). Leeches are also used extensively in neurobiological and developmental studies (e.g., Shankland, 1991; Aisemberg et al., 1993; Ramirez et al., 1995). Because non-sanguivorous species play a role in soil mineral balance as well as recycling the benthos of lakes and streams, particularly in eutrophic or polluted situations, they have been used as environmental stress indicators (Della Croze, 1955; McCall and Fisher, 1980). The systematics of leeches is poorly known for most regions of the world, and a phylogenetic system is needed for the development of a database upon which conservation and biomedical initiatives can depend.

Leeches are among the most poorly studied invertebrate taxa with respect to their evolutionary histories. The within-group relationships of leeches have been neglected by most annelid systematists (e.g., Apathy, 1888; Selensky, 1915; Wendrowsky, 1928; Livanow, 1931; Autrum, 1939). Inspection of recent considerations of the systematics of leeches (e.g., Sawyer, 1986; Siddall and Burreson 1995, 1998; Light and Siddall, in review) suggests that some widely accepted groupings may be artificial and precise species-level relationships await resolution of a variety of higher taxonomic questions.

The taxonomic composition of this study was designed to cover the full scope of higher systematic groupings of leeches. Euhirudinea consists of nine principle families that traditionally are divided into two taxonomic groups (Fig. 1). The families Glossiphoniidae, Ozobranchidae, and Piscicolidae constitute the order Rhynchobdellida, so named for their possession of a protrusible muscular proboscis with which they feed. The order Arhynchobdellida, the members of which are marked by the lack of a proboscis, traditionally consists of two suborders: the unarmed nonsanguivorous Erpobdelliformes and the Hirudiniformes, which includes the medicinal leech family Hirudinidae, the terrestrial Haemadipsidae, and the predaceous Haemopidae. Relationships have been assessed at the familial level using either morphology (Siddall and Burreson, 1995), or life history (Siddall and Burreson, 1996) or mitochondrial sequence data (Siddall and Burreson, 1998) on limited taxonomic subsets. Objectives of this project were to assess the monophyly of groups identified in recent taxonomy of leeches in a total evidence approach using multiple sources of data and a broader taxonomic scope.

MATERIALS AND METHODS

Taxa

The taxa sampled and their localities are listed in Table I, along with GenBank accession numbers for their 18S rDNA and CO-I mtDNA sequences. Taxa were chosen to test the monophyly of leech families and of ordinal ranks as well as to represent a broad range of morphological variation among leeches. Our analysis used a global sample consisting of members representing seven euhirudinidean families. Members of the Salifidae and the Cylicobdellidae were not available for analyses. Outgroup taxa consisted of two oligochaetes and the putatively related Branchiobdellida and monotypic Acanthobdellida. The members of Acanthobdellida and Branchiobdellida possess both oligochaete and leech morphological features (Purschke et al., 1993) and are considered to be intermediate between oligochaetes and euhirudinideans (Livanow, 1906, 1931; Michaelsen, 1919; Brinkhurst and Gelder, 1989; but see Holt, 1989).

Molecular Data

DNA Extraction and Purification

Field-collected specimens were identified and then either immediately used for DNA extraction or were preserved in 100% ethanol at ambient temperature for later extraction. Whenever possible, tissue was taken from the caudal sucker so as to prevent possible contamination by host erythrocyte DNA in the gut. Genomic DNA was extracted from specimens using the QIAamp Tissue Kit (QIAGEN Inc., 28159 Avenue Stanford, Valencia, CA 91355).

PCR amplification

Molecular characters for phylogenetic inference were obtained from the small subunit (18S) ribosomal gene and from cytochrome c oxidase I (CO-I) gene sequences. Mitochondrial CO-I sequences (651 bp) were amplified from purified genomic DNA using the universal primers LCO1490: 5'-GGTCAACAAATCATAAAGATATTGG-3' and HCO2198: 5'-TAAACTTCAGG GTGACCAAAAAATCA-3' (Folmer et al., 1994). A nested PCR approach was used to obtain purer 18S rDNA templates. The initial amplification used the primers 5'-AACCTGGTTGATCCT GCCAGT-3' and 5'-TGATCCTTCCGCAGGTTCACCT-3' (primers "A" and "B" respectively, Medlin et al. 1988), yielding a 1.8 kb fragment. Subsequent amplifications used internal primers (primer "L": 5'-CCAACTACGAGCTTTTTAACTG-3', primer "C": 5'-CGGTAATTCCAGCTC CAATAG-3', primer "Y": 5'-CAGACAAATCGCTCCACCAAC-3', primer "O": 5'-AAGGGCA CCACCAGGAGTGGAG-3') to yield three overlapping shorter double-stranded DNA fragments (denoted AL, CY and BO) of approximately 600 bp in length each.

Amplification reaction mixtures for CO-I contained 10xII buffer, 2.5 mM MgCl2, 0.25 mM of each dNTP, 1 ul of each primer, 1.25 units AmpliTaq DNA polymerase (Perkin-Elmer Corporation, 850 Lincoln Centre Drive, Foster City, CA 94404), and template DNA in a 50 ul total volume. The reaction mixtures were heated to 94deg.C for 4 min and then cycled in a PTC-100 Programmable Thermal Controller for 35 cycles at 94deg.C for 135 s, 44deg.C for 20 s, 70deg.C for 90 s with a final extension of 72deg.C for 7 min. 18S rDNA reaction mixtures were heated to 94deg.C for 4 min and then cycled 35 times at 94deg.C (20 s), 47deg.C (20 s), 68deg.C (105 s) with a final extension of 70deg.C (7 min). Amplification reactions for the BO fragment of 18S rDNA also included 10% DMSO to stabilize against secondary structure formation. The amplified DNA was purified in an agarose gel and manually excised over UV light. Further purification was performed either according to the Qiaquick PCR Purification kit protocol (QIAGEN) or by centrifugation through Sephadex G-50 beads (SIGMA Chemical Co., P.O. Box 14508, St. Louis, MO 63178) in Centri-sep columns (Princeton Separations, P.O. Box 300, Adelphia N.J. 67710).

DNA sequencing

Sequencing reactions contained 1 ul primer, 2.5 ul purified amplification product, 2 ul Big DyeTM (Applied Biosystems, Perkin-Elmer Corporation, 850 Lincoln Centre Drive, Foster City, CA 94404) and were cycled 35 times at 96deg.C (70 s), 44deg.C (5 s) and 60deg.C (4 min). Unincorporated dyes were removed from sequencing reaction products with Centri-sep columns (Princeton Separations). Sequencing products were electrophoresed in a 4% polyacrylamide gel in an ABI Prism 377 SequencerTM (Applied Biosystems). 18S rDNA was sequenced in three fragments of approximately 600 bp each in both directions. As well, the light and heavy strands of CO-I mtDNA were sequenced in both directions. Sequences were reconciled using Sequence Navigator (Perkin Elmer) and aligned using Clustal in Gene Jockey (Taylor, 1994), as well as MALIGN (Wheeler and Gladstein, 1994).

Morphology

Characters and character states used are adapted from Siddall and Burreson (1995). The character matrix is shown in Table II.

Character 1: chaetae: present (0); absent (1).

Character 2: testisacs: one pair (0); clusters (1); four pair (4); five pair (5); six pair (6); ten pair (9).

Character 3: coelomic organization: open with complete septa (0); reduced to lacunae without complete septa (1).

Character 4: presence of conducting (vector) tissue: present (0); absent (1).

Character 5: nephridia: complete in genital somites (0); suppressed in genital somites (1).

Character 6: pharynx: not protrusible (0); modified into protrusible proboscis (1).

Character 7: intestine: acaecate (0); caecate (1).

Character 8: cephalic eyespots: dorsolateral (0); dorsal (1).

Character 9: coelomic architecture: internal to the circular muscle (0); external to the circular muscle (1).

Character 10: intestinal blood sinus: absent (0); present (1).

Character 11: body shape: vermiform (0); dorsoventrally flattened (1).

Character 12: surface covering of cocoons: proteinaceous (0); membraneous (1).

Character 13: deposition of cocoons: slipped off head (0); secreted ventrally (1).

Character 14: arrangement of salivary tissue: diffuse (0); discrete glands (1).

Character 15: male bursa: bilobed (0); single bulb (1).

Character 16: eyespots: one pair per annulus (0); at least two pairs per annulus (1).

Character 17: myognaths: armed (0); unarmed (1).

Character 18: testisacs: discretely arranged on vasa deferens (0); hundreds of sacs profusely arranged (1).

Character 19: ovisacs: tubular (0); spheriod (1).

Character 20: nephridia: single funnel (0); multiple funnels located in ciliated organ (1).

Character 21: urinary bladder: absent (0); present (1).

Character 22: cocoons: cemented to a substrate (0); not cemented to a substrate (1).

Character 23: female median reproductive apparatus: simple pocket (0); modified into vaginal tube (1).

Character 24: respiratory auricles: absent (0); present (1).

Character 25: epididymes: loosely arranged (0); tightly coiled mass (1).

Character 26: cocoons: without spongy covering (0); with spongy covering (1).

Character 27: pairs eyes: zero (0); one (1); two (2); three (3); four (4); five (5).

Character 28: annuli per somite: one (1); two (2); three (3); five (5); six (6); >10 (9).

Character 29: location of male gonopore: on ring (0); in furrow (1).

Character 30: salivary papillae: absent (0); present (1).

Phylogenetic Analysis

Parsimony analyses with unweighted, unordered characters were conducted with PAUP* (Swofford, 1998) in Macintosh and DOS environments with 20 random sequence additions of taxa and tree bisection reconnection (TBR) branch breakup. AutoDecay (Eriksson and Winstöm, 1996) was utilized in calculating Bremer support values (Bremer, 1988).

RESULTS

Parsimony analysis of 18S rDNA alone (2023 characters) resulted in 48 equally parsimonious trees, each of which had 4426 steps and a retention index (RI) of 0.802. The strict consensus of these trees (Fig. 2a) resolved 27 clades out of a total 37 possible for the 39 taxa. Within the strict consensus, relationships are unresolved among Marsupiobdella africana, Theromyzon pallens, Glossiphonia complanata and Alboglossiphonia heteroclita, as well as between Mooreobdella bucera and a Mooreobdella melanostoma + Erpobdella punctata clade (Fig. 2a). Analysis of CO-I alone (651 characters) resulted in two equally parsimonious trees with 2909 steps and an RI of 0.411. These trees differed only in the placement of Lumbricus terrestris and Acanthobdella peledina (Fig. 2b). One hypothesis suggested that A. peledina was sister to the Euhirudinea, whereas the other had A. peledina as as sister to L. terrestris. The CO-I strict consensus tree is largely compatible with the 18S rDNA strict consensus tree, save that the root of the tree was applied to the ingroup at a different internode. Use of CO-I alone resulted in an unexpected placement of the Branchiobdellida well within the Euhirudinea. When molecular data sets were combined, heuristic searches yielded one most parsimonious tree (length=4426; RI=0.5902).

Use of all of the available data (the two molecular data sets in addition to 30 morphological characters) in parsimony analysis also resolved one most parsimonious tree (tree length=4497; RI=0.6031) identical to that found with the combined molecular data (Fig. 3). In this hypothesis, the following groups were recognized as monophyletic: the order Arhynchobdellida (combining Erpobdeliiformes and Hirudiniformes), the suborders Erpobdelliformes and Hirudiniformes, the families Glossiphoniidae, Piscicolidae, Haemopidae and Haemadipsidae, as well as the subfamilies Piscicolinae and Hirudininae. Polyphyly was indicated for the family Hirudinidae (Hirudo, Limnatis, Macrobdella), the subfamilies Haementeriinae and Glossiphoniinae, and the genera Dina and Erpobdella. The order Rhynchobdellida and the genus Mooreobdella each were paraphyletic. The sister group relationship of piscicolids to Arhynchobdellida was supported by a Bremer support index of only two, and only six additional steps were needed to make Rhynchobdellida (Glossiphoniidae + Ozobranchidae + Piscicolidae) a monophyletic group.

Within the Glossiphoniidae, neither of the subfamilies Haementeriinae nor Glossiphoniinae were monophyletic. Thirty-two additional steps were required for a monophyletic Haementeriinae and 34 for a monophyletic Glossiphoniinae as these subfamilies are presently constituted (Sawyer, 1986). Within the Piscicolidae, the subfamily Piscicolinae (Branchellion, Calliobdella, Piscicola) was monophyletic (Bremer support index=4). Stibarobdella macrothela, a pontobdellid, was sister to Piscicolinae. Ozobranchus margoi, the sole included representative of the family Ozobranchidae, was corroborated as sister to the Piscicolidae.

Groupings within the suborder Erpobdelliformes appeared to be geographically arranged but not taxonomically consistent. For example, the European Erpobdella octoculata (type species for the genus and for the family Erpobdellidae) and Erpobdella testacea formed a group sister to Dina japonica, obtained from Korea (Bremer support index=11). Twenty-five additional steps were required to group Dina japonica with the North American Dina dubia. The North American representative of Erpobdella, however, grouped with Mooreobdella species (Bremer support index=7). Ten additional steps were required for a monophyletic Mooreobdella clade, while grouping Erpobdella punctata with the other two species of Erpobdella would require 63 extra steps.

Although the families Haemopidae (Bremer support index=51) and Haemadipsidae (Bremer support index=22) each had only two representatives in this analysis, both were monophyletic groups within the suborder Hirudiniformes. Hirudinidae, however, was thus rendered polyphyletic because Macrobdella decora did not group with Hirudo medicinalis and Limnatis michaelseni. The haemopids were sister-group to the hirudinids (Bremer support index=31). Grouping M. decora with the hirudinids would require 48 additional steps.

A total of 160 additional steps were needed when constraints based on current systematic taxonomic groupings (Sawyer, 1986) were placed on the analysis. A tree with a monophyletic bloodfeeding clade or with a monophyletic nonbloodfeeding clade each required 200 extra steps. Acanthobdella peledina was not the sister group to Euhirudinea in this analysis but rather was sister to a leech + branchiobdellidan clade.

DISCUSSION

The resulting cladogram from the combined data indicates that current leech systematics for the most part has been reasonably accurate. Current higher taxonomic categories of leeches are found to be consistent with monophyletic groups identified in the analysis. The order Arhynchobdellida, the suborders Erpobdelliformes and Hirudiniformes, the families Glossiphoniidae, Piscicolidae, Erpobdellidae, Haemopidae and Haemadipsidae, and the subfamily Piscicolinae each are groups that are well supported. The most parsimonious tree, however, does not support all of the traditional groups. The orders Rhynchobdellida and Arhynchobdellida are distinguished by the possession of a protrusible muscular proboscis in the former and its absence in the latter. Usually, it is the lack of a character that denotes the plesiomorphic state and is indicative of paraphyly (Eldredge and Cracraft, 1980) such that a priori one would expect to find a paraphyletic Arhynchobdellida. Relationships depicted in figure 3 however show Arhynchobdellida as a derived, monophyletic group having lost the proboscis. Rhynchobdellida is paraphyletic in that Ozobranchidae and Piscicolidae form the sister group to Arhynchobdellida to the exclusion of Glossiphoniidae. This further corroborates Siddall and Burreson's (1998) hypothesis based only on CO-I that the presence of a proboscis is not a synapomorphy for Rhynchobdellida. Because only an extra six steps are needed to have a monophyletic Rhynchobdellida, and because there is a Bremer support index of only two for the Ozobranchidae + Piscicolidae + Arhynchobdellida clade, formal revision of ordinal taxonomy still would be premature. Moreover, although Trontelj (1997) also found a paraphyletic Rhynchobdellida, his analyses placed glossiphoniids as sister to Arhynchobdellida which differs from our results.

Nuclear genes such as 18S rDNA have a slow rate of change appropriate for resolving deep branching patterns and therefore higher-level relationships (Hillis and Dixon, 1991). If evolutionary rates are too slow however, there is not sufficient change within lineages to provide resolution in local areas of the emergent phylogenetic hypothesis. The use of 18S rDNA data alone poorly resolved within-family relationships of leeches as is evident from the Bremer support indices in figure 3. 18S rDNA is a non-protein coding structural gene that folds on itself and thus may involve compensatory changes as a possible source of error in phylogenetic analysis (Wheeler and Honeycutt, 1988). These data also are subject to insertions and deletions. Notably, Clustal alignment and two equally optimal alignments from MALIGN returned trees of different length for 18S rDNA. However, all of these trees were topologically identical, suggesting that the total evidence hypothesis is robust to indel placements. Mitochondrial genes such as the protein coding CO-I gene usually evolve more rapidly and can provide for resolution of more recent relationships. It has been argued that if rates of change are too fast, phylogenetic signal can be swamped by extraneous noise and could yield spurious deeper groupings (Miyamoto and Boyle, 1989; Hillis and Moritz, 1990; Swofford and Olsen, 1990; Cracraft and Helm-Bychowski, 1991). Because analysis of CO-I alone roots the ingroup within the glossiphoniids, it yields unexpectedly paraphyletic groupings for all major leech groups except Erpobdelliformes. CO-I has no insertions or deletions and is also very AT rich, which may introduce unique biases. The use of two independent molecular data sets in addition to morphological data combines historical information evolving under a variety of different constraints (nuclear and mitochondrial; coding and non-coding; fast rate of change and slow) and should be less susceptible to the biases that can confound the use of only one type of data (Wheeler et al., 1994). Where these data offer mutually corroborating support should be due to some extrinsic commonality; that is, history (Eernisse and Kluge, 1993).

In the total evidence analysis, the subfamilies Glossiphoniinae and Haementeriinae were found to be polyphyletic within a monophyletic Glossiphoniidae. Traditionally, the defining character in the two subfamilies has been the mode of cocoon deposition: members of the Glossiphoniinae attach cocoons directly onto a substrate whereas those of Haementeriinae attach cocoons onto the venter of the parent (Sawyer, 1971). Our findings corroborate those originally found by Light and Siddall (in review) based on CO-I and ND-I and continue to show that these are unnatural groupings. This suggests that these characters either may have arisen independently or are poorly characterized for the group.

Ozobranchus margoi, the sole included representative of the family Ozobranchidae, was supported as sister to the Piscicolidae. The ozobranchids are ectoparasitic on marine turtles and are distinguished by the presence of lateral digitiform branchiae (MacCallum and MacCallum, 1918). Traditionally, the family Piscicolidae is divided into three subfamilies: Piscicolinae, Pontobdellinae, and Platybdellinae. The shark leech Stibarobdella macrothela, a pontobdellid, was found to be sister to the subfamily Piscicolinae, which is consistent with the possession of external circulatory vessels, or pulsatile vesicles. These appendages may have been a recent adaptation to marine environments, and it has been speculated that they serve an osmoregulatory or circulatory function (Herter, 1936). Myzobdella lugubris, the only representative of the subfamily Platybdellinae, falls as sister to a Piscicolinae + Pontibdellinae clade. Although Myzobdella lugubris also occurs in marine environments, platydellines do not have pulsatile vesicles. Evaluating historical patterns of freshwater and marine immigration awaits further data from a more extensive taxonomic sample of these subfamilies.

Members in the suborder Erpobdelliformes appear to be descended from a common ancestor. However, most generic groupings within this clade were found to be unnatural. The genus Erpobdella is usually defined by having a preatrial loop on the paired seminal ducts and by body somites being five-annulate, with each annulus of approximately equal size (Sawyer, 1986). Erpobdella octoculata, the type species of this genus and of the family Erpobdellidae appears to be more closely related both to Nephelopsis and Dina species than to its North American congener Erpobdella punctata. The genus Dina also is characterized by being five-annulate but differs from Erpobdella species in that every fifth annulus is distinctly widened and subdivided (Sawyer, 1986). Although the type-species of Dina (the European Dina lineata) was not included, D. japonica often has been mistaken for it (Sawyer, 1986), and the failure of the nearctic D. dubia to group with palearctic D. japonica suggests that this genus also is in need of revision. The genus Mooreobdella, distinguished by the lack of preatrial loops, was not found to be monophyletic unless it included Erpobdella punctata. Without specifying his rationale, Sawyer (1986) placed Mooreobdella species, all of which are North American, in the genus Erpobdella. The most parsimonious tree however suggests either the placement of Erpobdella punctata in a genus (i.e., Mooreobdella) separate from the European erpobdellids or the expansion of the genus Erpobdella to include all of the Erpobdellidae. The lack of specimens from Salifidae (see Nesemann, 1995) and the South American family Cylicobdellidae, members of which have both erpobdellid and hirudinid-like characteristics, presently precludes a comprehensive revision of that group.

Of the three families within the Hirudiniformes that were analyzed, the families Haemadipsidae and Haemopidae were found to be monophyletic. While Siddall and Burreson (1995, 1998) did not find a monophyletic Haemopidae, here it was supported because of the basal placement of Macrobdella decora. This in turn renders the medicinal leech family Hirudinidae polyphyletic (Figure 3). The arrangement corroborates the notion of two separate medicinal leech families, the old world Hirudinidae and the new world Macrobdellidae, as previously suggested by (Richardson, 1969).

Evolution of bloodfeeding

Because leeches are best known for their bloodfeeding habits, it is perhaps not widely acknowledged that several common species of leeches do not bloodfeed and instead are predaceous on invertebrates. Members of the freshwater family Erpobdellidae, such as the popular bait leech Nephelopsis obscura, are carnivorous on other oligochaetes (Klemm, 1972). Haemopis species, a group that is closely related to the sanguivorous medicinal leeches, also are predaceous and many have large teeth for shredding prey as they are ingested. Sawyer (1986) reasoned that the evolution of feeding behavior in the hirudiniformes originated with macrophageous feeding by haemopids and culminated with sanguivory in the hirudinids and haemadipsids. Neither of these hypotheses is corroborated in the most parsimonious tree. Rather, it appears that the ancestral hirudiniform (and the ancestral leech more generally) was sanguivorous, and that sanguivory has been lost at least four times in the course of leech evolution (Figure 3). Bloodfeeding was lost twice within the Glossiphoniidae by the ancestor of the Glossiphonia complanata + Alboglossiphonia heteroclita clade and that of the Helobdella stagnalis + Desmobdella paranensis clade (Figure 3). As well, a carnivorous mode of nutrition has been adopted independently over sanguivory by the erpobdellids and the haemopids. This corroborates the notion previously raised by Siddall and Burreson (1996) that bloodfeeding is a plastic character easily lost by leeches. Because of the omission of certain taxa, our findings are likely to be an underestimate of the number of times sanguivory has been lost. For example, Mysidobdella borealis, a piscicolid which is known to parasitize mysid shrimp, was not included in the analysis and yet we predict it to group with other piscicolids. If so, every major group of leeches would then have taxa indicative of independent losses of this behavior for which leeches are so well known.

Cocoonology

Within the Euhirudinea there is a diversity in cocoon types and the manner in which cocoons are produced and deposited. Like the rest of the Clitellata, most leeches slip the secreted cocoon off the head, which then hardens and darkens to form a proteinaceous covering. Hirudiniform cocoons, which have a spongy covering unlike the cocoons of other leeches, are deposited out of water and abandoned. The cocoons of piscicolids and of the Erpobdelliformes normally are attached to a submerged substrate, whether it be an inanimate object or the body of a host organism. Glossiphoniids have the distinction of being the only annelids that exhibit parental care characterized by a protective brooding behavior. As opposed to a hardened protective cocoon, these leeches secrete a thin membranous cocoon in which fertilized eggs are deposited either on a substrate or on the venter side of the parent (Sawyer 1971, 1986). Even after hatching, glossiphoniids remain with their young, a behavior that corresponds historically to the loss of the protective proteinaceous covering of those species that abandon their cocoons (Siddall and Burreson, 1996). Mann (1962) suggested that membranous cocoons are plesiomorphic to hardened cocoons and therefore that glossiphoniids are primitive to other leeches. This notion is unwarranted because Acanthobdella peledina and branchiobdellidans deposit proteinaceous cocoons (Siddall and Burreson, 1995). Moreover, the most parsimonious hypothesis indicates that brooding behavior and membranous cocoons are not primitive states but rather are unreversed synapomorphies for the monophyletic Glossiphoniidae. As well, uncemented spongy cocoons that are abandoned appear to be synapomorphies for the Hirudiniformes.

Biogeography

The sampling of taxa in our study includes representatives from all continents and from a diversity of environments. As discussed previously, the revision of some generic-level groupings is recommended to better characterize some groups. Some of these revisions stem from observations that where traditional leech systematics fail, many species group geographically (e.g., erpobdellid species, hirudinid species). Indeed, we found that in many cases, North American species are more closely related to other North American species than they are to their European counterparts.

The terrestrial Haemadipsidae, represented here by Chtonobdella bilineata and Haemadipsa sylvestris, is the only leech group that is not found on all continents. These leeches are only known to occur in Australia, in the Wallacean archipelago, Southeast Asia, India, and Madagascar. Because they do not also occur in Africa, this distribution appears to post-date the breakup of Gondwanaland. Alternatively, this particular distribution may be attributable to recent dispersal via Indonesia and not to vicariance biogeography. Additional haemadipsid taxa could provide more information that may determine which of these hypotheses is accurate.

This study, the first to combine molecular data in addition to morphology, depicts the most complete phylogenetic higher-level analysis of the Euhirudinea to date. The results establish a foundation for more in-depth phylogenetic determination of species relationships and form a basis for investigating the nature of historical ecological associations. With the support of a well corroborated analysis of evolutionary relationships, studies can then be undertaken to discover patterns in historical ecology, character and life history evolution, including those of parental care among leeches as well as bloodfeeding, habitat preference and the historical invasion of terrestrial, freshwater and marine environments. Although this analysis includes 33 leech taxa and six outgroup taxa representing seven leech families from all continents except Antarctica, the inclusion of additional taxa is desirable in order to further stabilize the hypothesis. Because this has a North American bias in the taxa sampled, representatives from other continents would be particularly useful. Of particular interest would be to see where members of the unrepresented families Salifidae and Cylicobdellidae would fall. The Salifidae traditionally has been grouped as an erpobdelliform family, and morphological data suggest a basal paraphyletic relationship to the family Erpobdellidae. Members of the South American family Cylicobdellidae traditionally are considered a group of the Hirudiniformes. Cylicobdellid species have a typical hirudinidiform eye arrangement, but their median male reproductive apparatus is more erpobdellid-like. The inclusion of molecular data in addition to these morphological traits may confirm the placement of the Cylicobdellidae within Hirudiniformes or may suggest for the its grouping with the erpobdellids. As well, expansion of analyses to include more oligochaetes and some polychaetous outgroups should eventually lead to an understanding of which oligochaete family is most closely related to leeches, branchiobdellidans, and Acanthobdella peledina.

ACKNOWLEDGMENTS

We thank Soraya Bartol, BioPharm (UK), Gustavo Calvo, Kathryn Coates, Pierre Delaporte, Louis du Preez, Stuart Gelder, Richard Kraus, Brad Moon, Bob Murphy, Cynthia Sims Parr, Peter Rothlisberg, and Jessica.E. Light for their assistance in obtaining specimens. This research was supported by funds from the University of Michigan Museum of Zoology and Rackham School of Graduate Studies and grants from the University of Michigan Department of Biology and the National Science Foundation (BIO/DEB-9615211).

REFERENCES

Aisemberg, G.O. and Wysocka-Diller, J., Wong, V.Y., and Macagno, E.R. (1993). Antennapedia-Class Homeobox Genes Define Diverse Neuronal Sets in the Embryonic CNS of the Leech. Journal of Neurobiology. 24(10): 1423-1432.

Apathy, St. V. (1888) Analyse der äusseren Körperform der Hirudineen. Mittheil. Zool. Stat. Neapel. 8: 153-232.

Autrum, H. (1939). Hirudineen. Geographische Verbreitung. In "Klassen und Ordnungen des Tierreichs," (H. S. Bronns, Ed.), Vol. 4, Section III, Book 4, No. 1, pp.1-96, Akademische Verlagsgesellschaft, Leipzig.

Bremer, K. (1988). The limits of amino-acid sequence data in angiosperm phylogenetic reconstruction. Evolution, 42: 795-803.

Brinkhurst, R.O. (1994). Evolutionary relationships within the Clitellata: an update. Megadrilogica 5(10): 109-112.

Brinkhurst, R.O., and Gelder, S.R. (1989) Did the lumbriculids provide the ancestors of the branchiobdellidans, acanthobdellidans and leeches? Hydrobiologia 180: 7-15.

Cracraft, J., and Helm-Bychowski, K. (1991). Parsimony and phylogenetic inference using DNA sequences: Some methodological strategies. In "Phylogenetic analysis of DNA sequences," (Miyamoto, M.M. and J. Cracraft, eds). Oxford Univ. Press, New York.

Eernisse, D.J., and Kluge, A.G. (1993). Taxonomic congruence versus total evidence, and amniote phylogeny inferred from fossils, molecules, and morphology. Molecular Biology and Evolution 10: 1170-1195.

Della Croze, N. (1955). The conditions of the sedimentation and their relationship with Oligochaeta populations of Lake Maggiore. Mem. Inst. Idrobiol. Suppl. 8: 39-62.

Eldredge, N., and J. Cracraft. (1980). "Phylogenetic Patterns and the Evolutionary Process," Columbia University Press, New York.

Eriksson, T., and Winström,N. (1996). "AutoDecay," Version 3.0. Botaniska Institutionen, Stockholms Universitet 106 91 Stockholm, Sweden.

Folmer, O., Black, M., Hoeh, W., Lutz, R., and Vrijenhoek, R. (1994). DNA primers for amplification of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates. Molecular and Marine Biology Biotechnology 3: 294-299.

Herter, K. (1936). Die Physiologie der Hirudineen. In "Bronn's Klassen und Ordnungen des Tierreichs," 4. Bd., 3. Abt., 4. Buch, Teil 2,2. Lief, pp. 123-319. Leipzig.

Hillis, D.M., and Dixon, M.T. (1991). Ribosomal DNA: Molecular Evolution and Phylogenetic Inference. The Quarterly Review of Biology 66(4): 411-452.

Hillis, D.M., and Moritz, C. (1990). An overview of applications in molecular systematics. In "Molecular Systematics," (Hillis, D.M. and C. Moritz eds.). Sinauer, Sunderland, MA. pp. 502-515.

Holt, P.C. (1989). Comments on the classification of the Clitellata. Hydrobiologia 180: 1-5.

Klemm, D.J. (1972). "The leeches (Annelida: Hirudinea) of Michigan," The Michigan Academician 4: 405-444.

Lent, C. (1986). New medicinal and scientific uses of the leech. Nature 323: 494.

Light, J.E., and Siddall, M.E. in review. Phylogeny of the leech family Glossiphoniidae based on mitochondrial gene sequences and morphological data. Journal of Parasitology.

Livanow, N. (1906) Acanthobdella peledina Grube, 1851. Zool. Jrb. Anat. 22: 637-866.

Livanow, N. (1931). die Organisation fer Hirudineen und die Beziehungen dieser Gruppe zu de Oligochaeten. Erg. Fortschr. Zool. 7: 378-484

Mann, K. H. (1962). "Leeches (Hirudinea) their structure, physiology, ecology and embryology," Pergammon Press, New York.

MacCallum, W.G., and MacCallum, G.A. (1918). On the Anatomy of Ozobranchus branchiatus (Menzies). Bulletin American Museum of Natural History 38: 395-408.

McCall, P.J., and Fisher, J. B. (1980). Affects of tubificid oligochaetes on physical and chemical properties of Lake Erie sediments. In "Aquatic Oligochaete Biology," (R. O. Brinkhurst and D. G. Cook, eds.), Plenum Press. pp. 253-318.

Medlin, L., Elwood, H.J., Stickel, S., and Sogin, M.L. (1988). The characterization of enzymatically amplifed eukaryotic 16S-like rRNA-coding regions. Gene, 71: 491-499.

Michaelsen, W. (1919). Über die Beziehungen der Hirudineen zu den Oligochaeten. Mitt. hamb. Zool. Mus. Inst. 36: 131-153.

Miyamoto, M.M., and Boyle, S.M. (1989). The potential importance of mitochondrial DNA sequence data to eutherian mammal phylogeny. In "The hierarchy of life. Molecules and morphology in phylogenetic analysis," (Bremer and H. Jornvall, eds.), Elseviers Sci. Publ., B.V., Amsterdam, The Netherlands.

Munro, R., Siddall, M., Desser, S.S. and, Sawyer, R.T. (1992). The leech as a tool for studying comparative haematology. Comparative Haematology International, 2: 75-78.

Nesemann, H. (1995). On the morphology and taxonomy of the asian leeches (Hirudinea: Erpobdellidae, Salifidae). Acta Zool. Acad. Scient. Hungaricae 41: 165-182.

Purschke, G., Westheide, W., Rhode, D., and Brinkhurst, R. O. (1993). Morphological reinvestigation and phylogenetic relationship of Acanthobdella peledina (Annelida: Clitellata). Zoomorphology 113: 91-101.

Ramirez, F., Wedeen, C.J., Stuart, D.K., Lans, D., and Weisblat., D.A. (1995). Identification of a neurogenic sublineage required for CNS segmentation in an Annelid. Development 121: 2091-2097.

Richardson, L.R. (1969). A contribution to the systematics of the hirudinid leeches, with description of new families, genera and species. Acta Zool. Acad. Scient. Hung. 15: 97-149.

Sawyer, R.T. (1971). The phylogenetic development of brooding behaviour in the Hirudinea. Hydrobiologia 37(2): 197-204.

Sawyer, R.T. (1986). "Leech Biology and Behavior," Oxford University Press, Oxford.

Selensky, W. (1915.) "Morphological Studies and Systematics of Hirudinids. I. Organization of the Ichthyobdellids." Petrograd. [In Russian]

Shankland, M. (1991). Leech Segmentation: Cell Lineage and the Formation of Complex Body Patterns. Developmental Biology 144: 221-231.

Siddall, M.E., and Burreson, E.M. (1995). Phylogeny of the Euhirudinea: Independent evolution of blood feeding by leeches? Canadian Journal of Zoology 73: 1048-1064.

Siddall, M.E., and Burreson, E.M. (1996). Leeches (Oligochaeta?: Euhirudinea), their phylogeny and the evolution of life-history strategies. Hydrobiologia 334: 277-285.

Siddall, M.E., and Burreson, E.M. (1998). Phylogeny of Leeches (Hirudinea) Based on Mitochondrial Cytochrome c Oxidase Subunit I. Molecular Phylogenetics and Evolution, 9: 156-162.

Swofford, D.L. (1998). PAUP - Phylogenetic analysis using parsimony, version 4.0d64.1. Sinauer Associates, Sunderland Massachusetts.

Swofford, D.L., and Olsen, G.J. (1990). Phylogeny reconstruction. In "Molecular Systematics," (Hillis, D.M. and C. Moritz eds.), Sinauer, Sunderland, MA. pp. 411-501.

Taylor, P.L. (1994). GeneJockey-II Sequence processor. Software distributed by BIOSOFT, Cambridge.

Trontelj, P. (1997). "Molekulare Systematik der Egel (Hirudinea): Phylogenetische Analyse nuklearer und mitochondrialer ribosomaler DNA-Sequenzen." Dissertaion der Fakultät für Biologie der Eberhard-Karls-Universität Tübingen. 161 pages.

Walsmann, P., and Markwardt, F. (1985). On the isolation of the thrombin inhibitor Hirudin. Thrombosis res. 40: 563-569.

Wendrowsky, V. (1928). Über die Chromosomenkomplexe der Hirudineen. Z. Zellforsch. 8: 153-175.

Wheeler, W.C., and Gladstein, D.S. (1994). MALIGN: a multiple sequence alignment program. The Journal of Heredity, 85: 417-418.

Wheeler, W.C., and Honeycutt, R.C. (1988). Paired sequence difference in ribosomal RNAs: evolutionary and phylogenetic implications. Mol. Biol. Evol. 5: 90-96.

Wheeler, W.C., Cartwright, P., and Hayashi, C.Y. (1993). Arthropod phylogeny: A combined approach. Cladistics 9: 1-39.


Table I.   Taxa used in phylogenetic analyses of leeches.  
	Taxon	Locality	GenBank Accession #
			18S rDNA	CO-I__
OLIGOCHAETA
Tubifex tubifex	Great Britain	XXXXXXXX	U24570
Lumbricus terrestris	Great Britain	XXXXXXXX	U74076
		
ACANTHOBDELLIDA		
Acanthobdella peledina	Sweden	XXXXXXXX	AF003264
		
BRANCHIOBDELLIDA		
Xironogiton victoriensis	Oregon	XXXXXXXX	XXXXXXXX
Cambarincola holti	Tennessee	XXXXXXXX	XXXXXXXX
Cronodrilus ogygius	Georgia	XXXXXXXX	XXXXXXXX
		
RHYNCHOBDELLIDA		
GLOSSIPHONIIDAE		
Glossiphoniinae		
	Glossiphonia complanata	England	XXXXXXXX	AF003277
	Hemiclepsis marginata	France	XXXXXXXX	AF003259
	Placobdella parasitica	Ontario	XXXXXXXX	AF003261
	Desserobdella picta	Ontario	XXXXXXXX	XXXXXXXX
Haementeriinae		
	Alboglossiphonia heteroclita	Michigan	XXXXXXXX	XXXXXXXX
	Desmobdella paranensis	Uruguay	XXXXXXXX	XXXXXXXX
	Haementeria gracilis	Uruguay	XXXXXXXX	AF003276
	Haementeria ghilianii	Brazil	XXXXXXXX	XXXXXXXX
	Helobdella stagnalis	France	XXXXXXXX	XXXXXXXX
	Marsupiobdella africana	South Africa	XXXXXXXX	XXXXXXXX
	Oligobdella biannulata	North Carolina	XXXXXXXX	XXXXXXXX
Theromizinae		
	Theromyzon pallens	Ontario	XXXXXXXX	XXXXXXXX
		
OZOBRANCHIDAE		
	Ozobranchus margoi	Virginia	XXXXXXXX	AF003268
		
PISCICOLIDAE		
Piscicolinae		
	Branchellion torpedinis	South Carolina	XXXXXXXX	AF003265
	Calliobdella vivida	Virginia	XXXXXXXX	AF003260
	Piscicola geometra	France	XXXXXXXX	AF003280
Platybdellinae		
	Myzobdella lugubris	Virginia	XXXXXXXX	AF003269
Pontobdellinae		
	Stibarobdella macrothela	Virginia	XXXXXXXX	XXXXXXXX
		
		
ARHYNCHOBDELLIDA		
ERPOBDELLIFORMES		
ERPOBDELLIDAE		
	Dina dubia	Michigan	XXXXXXXX	XXXXXXXX
	Dina japonica	Korea	XXXXXXXX	XXXXXXXX
	Erpobdella punctata	Ontario	XXXXXXXX	AF003275
	Erpobdella octoculata	France	XXXXXXXX	AF003274
	Erpobdella testacea	France	XXXXXXXX	XXXXXXXX
	Mooreobdella melanostoma	Michigan	XXXXXXXX	XXXXXXXX
	Mooreobdella bucera	Michigan	XXXXXXXX	XXXXXXXX
	Nephelopsis obscura	Ontario	XXXXXXXX	AF003273
			
HIRUDINIFORMES		
HAEMADIPSIDAE		
	Chtonobdella bilineata	Australia	XXXXXXXX	AF003267
	Haemadipsa sylvestris	Vietnam	XXXXXXXX	AF003266
HAEMOPIDAE		
Haemopinae		
	Haemopis lateromaculata	Michigan	XXXXXXXX	XXXXXXXX
	Haemopis marmorata	Michigan	XXXXXXXX	AF003270
HIRUDINIDAE		
Hirudininae		
	Hirudo medicinalis	France	XXXXXXXX	AF003272
	Limnatis michaelseni	Congo	XXXXXXXX	XXXXXXXX
Macrobdellinae		
	Macrobdella decora	Michigan	XXXXXXXX	AF003271

Table II. Character and state matrix of morphological data used in phylogenetic analyses of leeches.
Taxon
Characters
         111111111122222222223
123456789012345678901234567890
Tubifex tubifex
00010000000000?0---000?00011-0
Lumbricus terrestris
00010000000000?0---000?00011-0
Cambarincola holti
10010000000000?0---000?00011-0
Cronodrilus ogygius
10010000000000?0---000?00011-0
Xironogiton victoriensis
10010000000000?0---000?00011-0
Acanthobdella peledina
00010000000000?0---000000011-0
Marsupiobdella africana
1611011101111000-0000100002300
Theromyzon pallens
1611011101111000-0000100004300
Hemiclepsis marginata
1?11011101111000-0000100004300
Glossiphonia complanata
1911011101111000-0000100003300
Alboglossiphonia heteroclita
1911011101111000-0000100003300
Haementeria gracilis
1511011101111100-0000100002300
Haementeria ghilianii
1511011101111100-0000100002300
Helobdella stagnalis
1611011101111000-0000100002300
Desmobdella paranensis
1611011101111000-0000100002300
Desserobdella picta
1611011101111000-0000100002300
Oligobdella biannulata
1611011101111000-0000100002200
Placobdella parasitica
1611011101111100-0000100002300
Ozobranchus margoi
1410010000000000-0000000002300
Calliobdella vivida
1610010110000000-0000000002600
Branchellion torpedinis
1511010110000000-0000000002300
Myzobdella lugubris
1511010100000000-0000000002900
Piscicola geometra
1610010110000000-0000000002900
Stibarobdella macrothela
1610010110000000-0000000002300
Dina dubia
111110000000000111000000004500
Mooreobdella bucera
111110000000000111000000003500
Mooreobdella melanostoma
111110000000000111000000003510
Dina japonica
111110000000000111000000003500
Erpobdella octoculata
111110000000000111000000003500
Erpobdella punctata
111110000000000111000000003510
Erpobdella testacea
111110000000000111000000003510
Nephelopsis obscura
111110000000000111000000004610
Haemadipsa sylvestris
191110000000001000111111115511
Chtonobdella bilineata
191110000000001000111111115511
Macrobdella decora
191110000000001000111110115500
Haemopis marmorata
191110000000001000111110115510
Haemopis lateromaculata
191110000000001000111110115510
Limnatis michaelseni
191110000000001000111110115511
Hirudo medicinalis
191110000000001000111110115510

Figure Legend

Figure 1: Current systematics of leeches, sensu Sawyer, 1986.

Figure 2: Strict consensus (a) of 48 equally parsimonious trees obtained from 18S rDNA and (b) of 2 equally parsimonious trees obtained from CO-I.

Figure 3: Most parsimonious hypothesis resulting from the combination of morphological, 18S rDNA, and CO-I data. Numbers above internodes indicate the Bremer support for that branch based on the combined data set. Upper and lower values below internodes indicate Bremer support based on CO-I and 18S rDNA alone respectively.

Figure 4: Most parsimonious reconstruction of the evolution of sanguivory in leeches indicates a common origin for the trait in the ancestral leech and four independent reversions for a carnivorous mode of nutrition.


Class                         Hirudinea
Subclass                     Euhirudinea
Order         Rhynchobdellida           Arhynchobdellida
Suborder                       Erpobdelliformes Hirudiniformes
Family        Glossiphoniidae  Erpobdellidae    Haemadipsidae
              Piscicolidae     Salifidae        Haemopidae
              Ozobranchidae                     Cylicobdellidae
                                                Hirudinidae