Introduction

The Hymenolepididae Perrier, 1897 is the most speciose cestode family, with more than 920 valid species (Mariaux et al. 2017). They are parasitic as adults in birds (mostly aquatic) and mammals. The diversity of hymenolepidids from mammals is estimated at 358 species (Mariaux et al. 2017). Hosts of mammalian hymenolepidids are several groups, with the majority of genera and species occurring in insectivores (Eulipotyphla), rodents (Rodentia) and bats (Chiroptera) (Vaucher 1971; Czaplinski and Vaucher 1994; Georgiev et al. 2006; Binkienė et al. 2011; Mariaux et al. 2017). Earlier views on phylogenetic relationships among hymenolepidids from mammals were expressed by proposals for their classification, especially by the erection of numerous genera based on morphology, host associations and life-cycle patterns, and believed to represent monophyletic groups (Spasskiy 1954; Vaucher 1992; Czaplinski and Vaucher 1994; Gulyaev and Kornienko 1999; Gulyaev et al. 2004; Gulyaev and Mel’nikova 2005).

An entire concept of the phylogenetic relationships among the hymenolepidid cestodes is lacking. Haukisalmi et al. (2010) postulated that hymenolepidids from “rodents and shrews” represented a monophyletic group, though this monophyly had not been tested by inclusion of taxa parasitic in birds. Their assumption is congruent with the results of a preliminary phylogenetic analysis based on 28S rRNA genes (V. V. Tkach, V. V. Kornyushin, B. B. Georgiev and D. T. J. Littlewood, unpublished) revealing hymenolepidids from mammals as forming a monophyletic group together with those from terrestrial (non-aquatic) birds.

Earlier molecular phylogenetic studies were focussed on the examination of the relationships between the major phyletic lineages or orders of cestodes (Mariaux 1998; Mariaux and Olson 2001; Olson and Tkach 2005; Waeschenbach et al. 2007, 2012). No molecular phylogenetic hypotheses have been proposed for the cestode order Cyclophyllidea as a whole or for the family Hymenolepididae. However, special attention was paid to the phylogeny of mammalian hymenolepidids based on sequencing partial (D1-D3) 28S rDNA and a hypothesis for their relationships, though based on a limited number of taxa, was proposed (Haukisalmi et al. 2010). Since the latter pioneer study on molecular phylogeny of mammalian hymenolepidids, it has become a standard practice to include molecular data when describing or redescribing species of this group (e.g. Greiman and Tkach 2012; Greiman et al. 2013; Tkach et al. 2013, 2018; Makarikov et al. 2015, 2018; Binkienė et al. 2015, 2018). This has contributed to a growing set of published gene sequence data available on GenBank, paving the way to more comprehensive elucidation of mammalian hymenolepidid phylogeny.

The aim of the present study is to expand the knowledge on the phylogenetic relationships among mammalian hymenolepidids by analysing a bigger, more comprehensive dataset of species than in previous studies. We used new 28S rDNA sequences of hymenolepidids from shrews obtained by us as well as recently published data by other researchers.

Materials and methods

Cestode sampling and identification

Shrews were collected by trapping from Boyana River, Vitosha Mts. (42.6369°, 23.2598°) and Kalimok Field Station (44.0116° 26.4397°) near Nova Cherna, Bulgaria. Adult cestodes were isolated from intestines. Specimens were preserved in 70% ethanol permitting both morphological and molecular study. Each cestode included in the analysis was divided into two parts. The anterior part (containing the scolex) was stained with iron acetocarmine (Georgiev et al. 1986) and dehydrated in alcohol series, cleared in dimethyl phthalate and mounted in Canada balsam for morphological identification of the species. Some specimens were mounted in Berlese’s medium (Swan 1936) when additional examination of the rostellar armament was deemed necessary. When more specimens were available in a host individual, some of them were mounted as entire specimens in order to be used as comparative material during the identification process. Specimens used for DNA extraction were deposited as voucher slides in the Helminthological Collection of the Institute of Biodiversity and Ecosystem Research, Bulgarian Academy of Sciences (IBER–BAS), Sofia (for acquisition numbers, see Table 1). The posterior parts of the specimens included in the molecular analysis were used as tissue samples for DNA extraction.

Table 1 Cestode species sequenced and used in the course of present analysis
Full size table

DNA extraction, PCR amplification and sequencing

Total DNA was isolated using Single Worm PCR Protocol (Williams et al. 1992). Samples were incubated in lysis buffer (containing 1× PCR buffer and 0.8 mg/ml Proteinase K) at 56 °C for 12 h followed by inactivation of the enzyme at 95 °C for 15 min. The amplification of a region of 28S rRNA gene that covers the variable regions D2-D3 was accomplished using the following primers D2A 5′-ACAAGTACCGTGAGGGAAAGTTG-3′ and D3B 5′-TCGGAAGGAACCAGCTACTA-3′ as suggested by De Ley et al. (1999). The PCR mixtures contained 25 μL of NZYTaq 2× Colourless Master Mix (Cat. No. MB04002; Nzytech, Lisbon, Portugal), 1 μM of each primer (FOR/REV) and 10 ng of template DNA in a total volume of 50 μL. All PCR reactions were carried out under the following conditions: initial denaturation at 94 °C for 5 min, 30 cycles (denaturation at 94 °C for 30 s; primer annealing at 50 °C for 30 s; extension at 72 °C for 90 s) and final extension at 72 °C for 10 min. PCR products were visualised on 1% agarose gel with GreenSafe staining (NZYTech, Lisbon, Portugal) under UV light. Fragment size was determined using GeneRulerTM 100 bp Ladder Plus (Fermentas, Thermo Scientific, Waltham, USA). All amplicons were purified by GeneJET™ PCR Purification Kit (Fermentas, Thermo Scientific, Waltham, USA) and sequenced in both directions by a PlateSeq kit (Eurofins Genomics, Ebersberg, Germany).

Phylogenetic analysis

The newly obtained 13 nucleotide sequences from D2-D3 28S rDNA were manually edited and aligned with MEGA software version 7.0 (Kumar et al. 2016). An analysis using Basic Local Alignment Search Tool (BLAST analysis, see www.ncbi.nlm.nih.gov/BLAST) was applied for comparison and possible identification with sequences available in GenBank for the family Hymenolepididae (Table 2).

Table 2 Published sequences of 28S rDNA of dilepidid (Dilepis undula, outgroup) and hymenolepidid cestodes deposited in GenBank used in the present phylogenetic analysis
Full size table

For phylogenetic analyses, we used published sequences from several previous studies (Olson et al. 2001; Haukisalmi et al. 2010; Greiman and Tkach 2012; Greiman et al. 2013; Tkach et al. 2013, 2018; Binkienė et al. 2015, 2018; Makarikov et al. 2015, 2018; Nkouawa et al. 2016) and available in GenBank (Table 2). Phylogenetic analysis was performed using Bayesian inference (BI) with MrBayes (Ronquist et al. 2012). Prior to analysis, the best model of nucleotide substitution was selected using MrModeltest2 (Nylander et al. 2004); this was the general time reversible model, with estimates of invariant sites and gamma distributed among-site rate variation (GTR+G+I). The analysis was run for 1.5 × 107 generations, two separate runs, each with four chains, discarding 33% of resulting trees as burn-in. As outgroup, we used a sequence of Dilepis undula (Schrank, 1788) (see Table 2), a species of the family Dilepididae, believed to represent the most closely related family-group taxon, from which matching molecular data were available (Mariaux et al. 2017). For some clades revealed by the present analysis, we used the names proposed by Haukisalmi et al. (2010), though in the majority of cases, additional taxa were added based either on our new data or published sequences by other authors. Average standard deviation of split frequencies below 0.01 was observed and served as a proof of chains reaching convergence. Branches persisting in less than 50% of post burn-in samples were treated as polytomies. Nodal support was expressed as posterior probabilities.

The number of base differences per site from averaging over all sequence pairs between the main clades revealed was calculated in MEGA7 (Kumar et al. 2016). The rate variation among sites was modelled with a gamma distribution (shape parameter = 1). The analysis involved 63 nucleotide sequences. All positions with less than 95% site coverage were eliminated, i.e. fewer than 5% alignment gaps, missing data and ambiguous bases were allowed at any position. There were a total of 930 positions in the final dataset.

Results

Genetic characterisation of newly sequenced hymenolepidid cestodes

Thirteen individuals of 12 cestode species isolated from shrews in Bulgaria were sequenced in the present study (Table 1). The sequences were deposited in GenBank (for accession numbers, see Table 1). New sequences spanned the D2-D3 fragment of 28S rRNA gene from position 376 to position 1522 according to a reference sequence, Acc. No. AY157181 Hymenolepis diminuta (Rudolphi, 1809) (Lockyer et al. 2003).

For seven species, i.e. Coronacanthus integrus (Hamann, 1891), C. magnihamatus Vasileva, Tkach and Genov 2005, C. omissus (Baer and Joyeux, 1943), C. vassilevi Genov, 1980, Vaucherilepis trichophorus Tkach, Vasileva and Genov, 2003, Staphylocystis tiara (Dujardin, 1845) and S. uncinata (Stieda, 1862), the obtained sequences represent the first available genetic data. Sequences were also obtained for an unidentified species of the genus Neoskrjabinolepis Spassky, 1947, probably representing an undescribed species; it differs by a single nucleotide substitution from N. schaldybini Spassky, 1947 (0.1% genetic difference) sequenced by Haukisalmi et al. (2010). For four species that were previously sequenced on the basis of specimens from elsewhere, i.e. Staphylocystis furcata (Stieda, 1862), Lineolepis scutigera (Dujardin, 1845), Ditestolepis diaphana (Cholodkowsky, 1906) and Spasskylepis ovaluteri Schaldybin, 1964, we obtained additional 28S rDNA sequences based on specimens from Bulgaria. The sequence of S. furcata was identical with one published by Haukisalmi et al. (2010). For the remaining three species, they were very similar to previously published examples for the same species: L. scutigera differed by three substitutions (0.26% difference), D. diaphana differed by 22 substitutions (1.96% difference) and S. ovaluteri differed by three deletions of totally 18 nucleotides and 14 substitutions (2.79% difference) from those sequenced by Haukisalmi et al. (2010).

Phylogenetic analysis

The phylogenetic tree of hymenolepidids from mammals (Fig. 1) supported early, deep divergence between the group of taxa with serial maturation of strobila (“Ditestolepis clade” of Haukisalmi et al. 2010) consisting of cestodes from soricids (the genera Ditestolepis Soltys, 1952, Spasskylepis Schaldybin, 1964 and Gulyaevilepis Kornienko and Binkiene, 2014) and the remaining mammalian hymenolepidids characterised (with a few exceptions) by gradual strobilar maturation and comprising parasites of shrews, rodents, chiropterans, humans and erinaceomorph mammals. Among the latter group, the deeper (“backbone”) branching was relatively weakly supported. However, several strongly supported clades were outlined, each of them characterised by a constant composition of taxa included. Since they confirmed the main clades revealed by Haukisalmi et al. (2010), we used for names proposed by these authors for consistency. These clades were as follows:

  • “Rodentolepis clade”. This strongly supported clade consisted of parasites of rodents, chiropterans and soricids. Several well-defined subgroups were outlined. One of them was the subclade consisting of the genera Nomadolepis Makarikov, Gulyaev and Krivopalov, 2010, Pararodentolepis Makarikov and Gulyaev, 2009 and species of the genera Rodentolepis Spassky, 1954 and Staphylocystis Villot, 1877 representing parasites of either shrews or rodents. Other well-supported monophyletic groups in this clade were as follows: (i) hymenolepidids of glirid rodents (Armadolepis spp. and unidentified species from the same host group); (ii) species of Vampirolepis Spassky, 1954, a genus parasitic in bats; and (iii) a group of taxa parasitizing soricids plus the rodent cestode Rodentolepis straminea (Goeze, 1782); in the latter subclade, there was a clear separation of cestodes from Sorex spp. from those isolated from other hosts. The first sequence for Staphylocystis tiara, a parasite of crocidurine shrews, which had been obtained in the course of the present study, placed this species separately and distantly from other species considered its congeners.

  • “Hymenolepis clade”. It was strongly supported by our analysis and consisted of species of the genus Hymenolepis Weinland, 1858 (parasitic in murid and geomyid rodents, erinaceomorphs and humans) plus Rodentolepis asymmetrica (Janicki, 1904), a parasite of cricetid rodents.

  • “Arostrilepis clade”. It consisted of cestodes of soricine shrews (of the tribes Soricini and Nectogalini), with a notable exception representing a species of the genus Arostrilepis Mas-Coma and Tenora, 1997 parasitic in cricetid rodents. The backbone of this clade is weakly supported by our analysis. However, several monophyletic groups had strong support. These were the following: (i) Neoskrjabinolepis plus Lineolepis Spassky, 1959; (ii) the species of the genus Coronacanthus Spassky, 1954; (iii) and (iv) two strongly supported groups consisting of three and two Nearctic species of the genus Staphylocystoides Yamaguti, 1952, however being distant from one another; (v) Soricinia quarta (Karpenko, 1983) plus S. genovi Binkiene, Kornienko and Tkach, 2015. Though being weakly supported, a clade consisting of hymenolepidids of the genera Arostrilepis, Soricinia Spassky and Spasskaya, 1954 and Pseudobothrialepis Schaldybin, 1957, all characterised by the lack of rostellar apparatus or its rudiment, was outlined (Fig. 1, see also Fig. 3, bar E).

Fig. 1
figure 1

Bayesian inference tree of phylogenetic relationships among hymenolepidid cestodes from mammals based on analysis of 28S rRNA gene of 56 ingroup taxa. Dilepis undula (family Dilepididae) is used as outgroup. The GenBank numbers are added after the binomial name of each species. Newly sequenced taxa are in bold. The major clades in this group recognised by Haukisalmi et al. (2010) and also outlined and complemented by the present study are marked by circles. Nodal support is given by posterior probabilities. Strongly supported subclades (posterior probability more than 0.95) are marked by rectangles

Full size image

Each of the two species, Urocystis prolifer Villot, 1880 and Vigisolepis spinulosa (Cholodkowsky, 1906), cannot be attributed to any of the aforementioned clades.

The analysis of the number of base differences per site from averaging over all sequence pairs between the main clades of hymenolepidids from mammals resulted in values between 0.060 and 0.086 (Table 3), indicating similar levels of divergence between them.

Table 3 The number of base differences per site from averaging over all sequence pairs between the main clades of hymenolepidids from mammals as revealed by the present study
Full size table

Discussion

Comparison with the previous phylogenetic hypotheses for mammalian hymenolepidids

The general configuration of our phylogenetic tree (Fig. 1) is similar to the previous phylogenetic hypothesis for relationships among hymenolepidids from mammals (Haukisalmi et al. 2010). However, our study includes 56 taxa of the family Hymenolepididae, including 12 taxa sequenced by us (Table 1) and 25 species sequenced by other authors after 2010 (Table 2) (versus 31 taxa in the analysis of Haukisalmi et al. 2010). Our analysis revealed the same main phyletic lineages (clades): “Hymenolepis clade”, “Rodentolepis clade”, “Ditestolepis clade” and “Arostrilepis clade” (Haukisalmi et al. 2010), exhibiting a similar level of divergence between them (Table 3). The former two clades are well supported in our study. However, in contrast to the hypothesis of Haukisalmi et al. (2010), the Arostrilepis clade is weakly supported, a feature suggesting that the inclusion of additional species may lead to its breaking down into smaller monophyletic groups. In addition, the backbone of the phylogenetic tree is characterised by weak resolution, in contrast to that presented in the previous hypothesis (see Fig. 1 of Haukisalmi et al. 2010). This weak support is not surprising, since mostly cestodes from shrews from temperate latitudes have been used in the present analysis. The vast majority of species of the subfamily Crocidurinae (where, inter alia, the genera Crocidura Wagler, 1832 and Suncus Ehrenberg, 1832 belong, both occurring in Europe) and the soricine tribe Nectogalini (harbouring 7 genera, including Neomys Kaup, 1829 from temperate latitudes of Eurasia) are distributed in tropical areas (Wilson and Reeder 2005). It could be expected that the unexplored diversity of their hymenolepidids in tropics is much higher than in temperate latitudes. Denser sampling of these host groups, for both morphological and molecular studies, may contribute for developing more reliable phylogenetic hypotheses for this parasite group. The current diversity of soricine shrews in Europe consists of two genera only while, for the same territory, the number of fossil genera of this subfamily is 24 (Rzebik-Kowalska 2003). The present fauna of European Soricidae (and, apparently, the fauna of their cestodes) has been formed after the last glaciation when the species of this family (mostly those preserved in the South-European refugia) recolonised the continent (Hewitt 1999). Therefore, episodes of mass extinction of pre-glaciation shrews (and their cestodes) as well as the lack of parasite specimens originating from tropical shrews included in the phylogenetic analysis may partly explain the low resolution of the backbone of our tree. That more recently derived nodes were more robustly supported than deeper nodes in the phylogeny suggesting additional gene loci, and particularly those evolving more slowly than this section of 28S rDNA, may help in providing stability to the phylogeny. However, for the purposes of this study, where we are building on an extensive taxon set for which only one gene locus is available, our results provide some useful and robust insights.

Our analysis confirms the following: the monophyly of the genus Hymenolepis (sensu stricto) parasitic in hosts from various mammalian orders as revealed by Binkienė et al. (2018); the monophyly of hymenolepidids from glirid rodents (including armed and unarmed subgenera) and the paraphyly of the nominotypical subgenus of the genus Armadolepis Spassky, 1954 (Makarikov et al. 2018); the position of the genera Pararodentolepis and Nomadolepis as sister groups (Makarikov et al. 2015); the position of Rodentolepis asymmetrica as the sister group of Hymenolepis (sensu stricto) and distantly from other Rodentolepis spp. (Haukisalmi et al. 2010; Greiman and Tkach 2012; Binkienė et al. 2018); the position of Neoskrjabinolepis and Lineolepis as sister taxa (Haukisalmi et al. 2010; Greiman and Tkach 2012); the close phylogenetic relationships among the unarmed genera of the Arostrilepis clade, i.e. Arostrilepis, Soricinia and Pseudobothrialepis (Haukisalmi et al. 2010; Binkienė et al. 2015).

Novelties revealed by our analysis are mostly associated with the position of taxa, which have not been sequenced or have not been involved in similar analyses previously. For example, we provide the first evidence based on molecular data about the monophyly of the genus Coronacanthus erected or recognised on the basis of morphological data (Spasskiy 1954; Genov 1980; Czaplinski and Vaucher 1994; Vasileva et al. 2005). We have also revealed the non-monophyletic status of the genus Staphylocystis (harbouring parasites of shrews). Its polyphyly is due to the distant position of S. tiara and its relationship with the subclade harbouring the genera Nomadolepis and Pararodentolepis. In addition, the remaining congeners form a paraphyletic group, with Rodentolepis straminea, a parasite of rodents, belonging to the same clade. Our results also suggest the polyphyly of Staphylocystoides in its current taxonomic concept (Czaplinski and Vaucher 1994; Greiman et al. 2013; Tkach et al. 2018), since its species, though all belonging to the Arostrilepis clade, are distributed among three subclades (Fig. 1).

The position of the newly sequenced genus Vaucherilepis Tkach, Vasileva and Genov, 2003 remains unclear as its type and only species forms a weakly supported clade with a species of Staphylocystoides.

Morphological support of the revealed relationships

Hoberg et al. (1999) considered the rostellar apparatus of hymenolepidids as consisting of a sac-like rostellum, typically in combination with a rostellar sheath; they believed that the loss of rostellum is a secondary evolutionary event that happened independently in various phyletic lineages of cyclophyllideans. Therefore, we accept that the presence of a sac-like rostellar apparatus is the plesiomorphic state for this family. This type of rostellar apparatus consists of a rostellum, a protrusible apical part of the scolex, termed rhynchus (often bearing a crown of rostellar hooks) and a rostellar sheath (Fig. 2a, b; for further details of the structure of the sac-like rostellar apparatus, see Stoitsova et al. (1997)). Almost half of the species included in our analysis (23 out of 56) are characterised by this general (plesiomorphic) pattern of rostellar apparatus.

Fig. 2
figure 2

Major types of rostellar apparatus of hymenolepidid cestodes from mammals (schematic presentation). a, b Scoleces with fully developed (not rudimentary) rostellar apparatus with invaginable rostellum (a) and retractile rostellum (b). c–e Scoleces with rudimentary rostellar apparatus. c Rostellar apparatus characterised by the entire reduction of the rostellar sheath and partial reduction of the rostellum, the latter being transformed into a glandular apical organ. d Rostellum entirely reduced and the rhynchus (“anterior canal”) invaginated into the rostellar sheath. e The entire reduction of the rostellar apparatus (no vestigial structures of rostellum or rostellar sheath)

Full size image

Notably, the major clades and subclades outlined by our analysis reveal various degrees of reduction or modification of the general rostellar pattern. The species of the Ditestolepis clade are characterised by the entire reduction of the rostellar sheath and partial reduction of the rostellum; the latter is transformed into a small “glandular apical organ” (termed by Gulyaev and Karpenko 1998), with almost entirely reduced musculature (Fig. 2c). In the Hymenolepis clade, where there are both cestodes with armed rostellum (Rodentolepis asymmetrica) and unarmed species (Hymenolepis s. str.), the rostellum of the latter is entirely reduced and the rhynchus (termed “anterior canal”, see, e.g. Specian and Lumsden 1980) is invaginated into the rostellar sheath (Fig. 2d). Similarly, an entire reduction of the rostellum can also be observed in the Rodentolepis clade, among the subclade consisting of cestodes from glirid rodents, where the subgenus Bremserilepis Makarikov, 2017 is characterised by the presence of unarmed rhynchus invaginated into the rostellar sheath (Makarikov 2017; Makarikov et al. 2018). In the Arostrilepis clade, the subclade consisting of the genera Arostrilepis, Soricinia and Pseudobothrialepis exhibits entire reduction of the rostellar apparatus, i.e. no vestigial structures of rostellum or rostellar sheath (Fig. 2e). Therefore, we confirm that the partial or entire reduction of the rostellar apparatus is an apomorphic condition (Hoberg et al. 1999) that appeared independently in all major phyletic lineages (clades) of hymenolepidids.

Among the hymenolepidid genera with the general (plesiomorphic) pattern of rostellar apparatus, two types of rostella were recognised (Czaplinski and Vaucher 1994). “Invaginable rostellum” is that with an anterior part able to invaginate and, as a result, hooks are with anteriorly directed blades when rostellum is withdrawn (Fig. 2a); in contrast, “rectractile rostellum” is with hook blades directed posteriorly when the rostellum is withdrawn (Fig. 2b). These two types of armed rostellar apparatus correspond to the “Type A” and “Type C” recognised by Vaucher (1971) who has explained their morphological and functional differences on the basis of examination of histological sections of intestinal walls with scoleces attached in situ. In our phylogenetic tree (Fig. 1), excluding forms with rudimentary rostella, the retractile type of rostellum is characteristic for the taxa belonging to the Rodentolepis clade and the Hymenolepis clade while the invaginable type of rostellum occurs only in the Arostrilepis clade (Fig. 3).

Fig. 3
figure 3

Distribution of the various types of rostellar apparatus and strobilar development of hymenolepidid cestodes from mammals across the phylogenetic tree produced by the present analysis. The labels A–E of rostellar apparatus correspond to those in Fig. 2. For the strobilar development, dotted rectangles correspond to taxa (or clades) characterised by serial maturation; striped rectangles correspond to taxa (or clades) with gradual maturation

Full size image

In addition to the abovementioned peculiarities of the rostellar apparatus, other morphological characters may support the clades revealed by the present analysis. In the Ditestolepis clade, all the members are characterised by serial maturation of strobila; however, this character is homoplasious, since serial maturation of strobila occurs also in the genus Neoskrjabinolepis of the Arostrilepis clade (Kornienko et al. 2006, 2008). The majority of mammalian hymenolepidids are characterised by gradual maturation of the strobila (Fig. 3).

The subclade consisting of the genera Nomadolepis Makarikov, Gulyaev and Krivopalov, 2010 and Pararodentolepis Makarikov and Gulyaev, 2009 (within the Rodentolepis clade) is characterised by fraternoid shape of rostellar hooks, median position of vitellarium, bi-lobed uterus not extending beyond the osmoregulatory canals and embryophores with polar filaments (Makarikov and Gulyaev 2009; Makarikov et al. 2010, 2015; Greiman and Tkach 2012). Some of these characters fit well to the preliminary data on the morphology of Staphylocystis tiara (Dujardin, 1845), a species placed by our analysis in this clade and distantly from its congeners. This demonstrates the need for redescription of the latter species and its possible affiliation to a genus belonging to this clade, probably to Pararodentolepis.

The remaining species of Staphylocystis are grouped into two clades, one consisting of species parasitic in soricine and the other parasitic in crocidurine shrews (Fig. 1). These two clades are also supported by specific morphological characters of their rostellar hooks, being fraternoid in the former and cricetoid in the latter clade. Further morphological studies are needed in order to justify the possible erection of a new genus for one of these groups.

The subclade consisting of the genera Neoskrjabinolepis and Lineolepis (within the Arostrilepis clade) is morphologically supported by the number of rostellar hooks (10) and the transformation of gravid proglottides into oophores responsible for the group dispersion of eggs (Kornienko et al. 2008; Korneva and Kornienko 2013).

The subclade consisting of the genera Arostrilepis, Soricinia and Pseudobothrialepis (belonging also to the Arostrilepis clade), in addition to the entire reduction of the rostellar apparatus, is also supported by the elongate fusiform embryophores with polar processes of its members as described by Karpenko and Gulyaev (1990), Makarikov et al. (2011) and Binkienė et al. (2015).

The revealed polyphyly of the genus Staphylocystoides with molecular data, with species belonging to three subclades, is morphologically supported by several characters. The type species S. sphenomorphus (Locker and Rausch, 1952) together with S. oligospinosus Tkach, Kinsella and Greiman, 2018 and S. parasphenomorphus Tkach, Kinsella and Greiman, 2018 form a well-supported monophyletic group; these species are characterised by slender, more elongate strobila (Tkach et al. 2018). The type species has rostellar hooks forming sigmoid curvature in dorso-ventral plane (Gulyaev et al. 2007); however, this character is not mentioned in the description of the two recently described species (Tkach et al. 2018). In contrast, the remaining two Nearctic species included in our analysis, S. parvissima (Voge, 1953) and S. gulyaevi Greiman, Tkach and Cook, 2013 are characterised by very short strobila of a few proglottides and the shape of hooks as presented by Greiman et al. (2013) is rather cricetoid than fraternoid, thus differing substantially from the shape of the hooks in the other Nearctic clade. The Palaearctic S. stefanskii (Zarnowski, 1954) is also characterised by rostellar hooks with sigmoid configuration in dorso-ventral plane (Gulyaev et al. 2007); in addition, it has medial vitellarium in contrast to the Nearctic species having vitellarium situated antiporally, at the level between the antiporal testis and medial testis. Therefore, our results suggest the heterogeneity of the genus Staphylocystoides in its current taxonomic concepts (Czaplinski and Vaucher 1994; Gulyaev et al. 2007; Greiman et al. 2013; Tkach et al. 2018); a re-evaluation of the genus with additional morphological and molecular data is warranted.

Our analysis reveals that two characters that have been considered important from taxonomic point of view are homoplasious. The embryophore is provided with polar filaments in two subclades of the Rodentolepis clade, i.e. in Pararodentolepis–Nomadolepis subclade (Makarikov and Gulyaev 2009; Greiman and Tkach 2012; Makarikov et al. 2015) as well as in Rodentolepis microstoma (see Cunningham and Olson 2010). The gravid uterus is restricted entirely in the median field of proglottis in all the members of the Ditestolepis clade (Gulyaev 1991; Gulyaev et al. 2004; Kornienko and Binkiene 2014), Pararodentolepis–Nomadolepis subclade (Makarikov and Gulyaev 2009; Makarikov et al. 2015) and Armadolepis subclade (Makarikov 2017; Makarikov et al. 2018) of the Rodentolepis clade as well as Lineolepis–Neoskrjabinolepis subclade (Kornienko et al. 2008; Korneva and Kornienko 2014) of the Arostrilepis clade. Since these characters have been widely ignored in previous taxonomic descriptions, we need more detailed morphological studies in order to understand the possible independent origin of these structures in various groups.

The distribution of the species currently recognised as members of Rodentolepis in four distinct monophyletic groups (clades and subclades) suggests the artificial character of this genus and requires deeper morphological and taxonomic examination in order to justify the erection of new genera.

Definitive-host and life-cycle associations

Among the four major clades outlined in hymenolepidids from mammals, only the Ditestolepis clade is strongly associated with soricid shrews (Fig. 4). Each of the remaining three major clades contains parasites of two or more orders of mammals: Arostrilepis clade—parasites of cricetid rodents and soricids; Hymenolepis clade—parasites of rodents, primates and erinaceomorphs; and Rodentolepis clade—parasites of rodents, chiropterans and soricid shrews. Such distribution of parasite–host associations across the phylogenetic tree resolved by our study suggests the presence of multiple events of host switching in the course of the formation of the current diversity of mammalian hymenolepidids, including host switching between members of different mammalian orders.

Fig. 4
figure 4

Distribution of definitive host associations of hymenolepidid cestodes from mammals across the phylogenetic tree produced by the present analysis

Full size image

However, we revealed several lineages as closely associated with certain mammalian groups, with their present diversity probably reflecting the host–parasite coevolution or multiple colonisations of hosts of the same group. These are the following: (i) the Ditestolepis clade, probably corresponding to the tribe Ditestolepidini (in the concept of Gulyaev 1991), strongly associated with hosts of the family Soricidae; (ii) the subclade containing the genera with entirely reduced rostellar apparatus (Soricinia and Pseudobothrialepis), also parasitizing soricids; (iii) the genus Coronacanthus hosted by Neomys shrews and (iv) the genus Armadolepis diversified within glirid rodents.

Only few of the species included in the present analysis are with known intermediate hosts (Fig. 5). Coleopterans are intermediate hosts of Ditestolepis diaphana, Hymenolepis diminuta, H. weldensis Gardner and Schmidt, 1988, H. hibernia Montgomery, Montgomery and Dunn, 1987, H. erinacei (Gmelin, 1789), Neoskrjabinolepis schaldybini, Staphylocystis furcata, S. uncinata, Rodentolepis microstoma (Dujardin, 1845) and R. fraterna (Stiles, 1906) (Prokopic 1971, Genov 1984, Sulgostowska et al. 2015). Siphonapterans (fleas) are hosts of Hymenolepis diminuta and Lineolepis scutigera (see Genov 1984) as well as of a species of Armadolepis (see Vaucher and Quentin 1975). Metacestodes of Urocystis prolifer have been recorded in myriapods (Baer and Della Santa 1960; Kisielewska 1960). Mites are known as intermediate hosts of Rodentolepis asymmetrica (see Prokopic and Mauer, 1969), collembolans of Arostrilepis spp. (Ishigenova et al. 2018) and Vigisolepis spinulosa (see Prokopic 1968), while dipterans have been recorded as hosts of Staphylocystis brusatae Vaucher, 1971 (see Quentin et al. 1971). The taxonomic affiliations of intermediate hosts are distributed across the phylogenetic tree generated by the present study, and no clear pattern of association can be inferred. The formation of the suprageneric monophyletic groups revealed by the present study does not seem to be associated with adaptation to type of invertebrate group as intermediate hosts.

Fig. 5
figure 5

Distribution of intermediate host associations of hymenolepidid cestodes from mammals across the phylogenetic tree produced by the present analysis

Full size image

The majority of life cycles of mammalian hymenolepidids are associated with terrestrial environments due to the habitat preferences of their definitive hosts. However, the genera Vaucherilepis and Coronacanthus, which use aquatic amphipods as intermediate hosts (Prokopic et al. 1970; Tkach 1994; Tkach et al. 2003), are nested among species from other mammalian hosts with obviously terrestrial life cycles (Fig. 5). This position suggests the secondary acquisition of aquatic life cycles of cestodes parasitic in Neomys shrews, which is congruent with the hypothesis for the secondary adaptation of the shrews of the tribe Nectogalini to an aquatic environment (He et al. 2010). In view of understanding the origin of the cestode parasites from Neomys spp., it could be worth comparing its cestode parasites with those from the other members of the tribe Nectogalini that includes mostly shrews living in terrestrial habitats and having great taxonomic diversity in East Asia. Neomys is the only genus of this tribe studied in detail for cestode parasites, with 24 hymenolepidid species recorded in its members (Binkienė et al. 2011, 2015). Other Nectogalini examined for cestode parasites are three species of the genus Soriculus Blyth, 1854 from Nepal, with eight cestode species recorded from them (Sawada et al. 1995); however, their descriptions are inadequate and do not allow detailed morphological comparisons.

Interestingly, we found Spasskylepis ovaluteri from Neomys fodiens in Bulgaria. This parasite is believed to be a specific cestode of Sorex spp. (Gulyaev et al. 2004). The only published sequence for this species was based on a specimen isolated from Sorex caecutiens Laxmann, 1788 from Finland (Haukisalmi et al. 2010). The specimens from Bulgaria used in the present study correspond very well to the description by Gulyaev et al. (2004) and a redescription based on our samples is to be published elsewhere. The record of S. ovaluteri in Neomys is congruent with the observation that Eurasian water shrews feed mainly on terrestrial prey; aquatic prey comprises only 11–27% of their diet (He et al. 2010). In support of the wide overlap of diets of Sorex and Neomys shrews, Binkienė et al. (2011) listed 11 cestode species believed to be specific for Sorex spp. and having also records in Neomys.

Concluding remarks

Except for the Ditestolepis clade diversified within the Soricidae, the composition of the remaining phyletic lineages of mammalian hymenolepidids suggests multiple events of host switching, including between hosts belonging to different mammalian orders. However, for the entire reconstruction of the evolutionary history of mammalian hymenolepidids, we need denser taxon sampling, mostly from tropical representatives of the major host groups that better reflect the huge diversity of this cestode group as revealed by morphological studies. Mariaux et al. (2017) summarised that mammalian hymenolepidids include 358 species belonging to 52 genera, suggesting adequate representative sampling remains a considerable task. Additionally, we recognise that further molecular data capable of resolving deeper nodes will be important to estimate more robust phylogenies and test further the hypotheses raised in this study.