Introduction

Among monogenean flatworms of the genus Gyrodactylus von Nordmann, 1832 (Gyrodactylidae van Beneden & Hesse, 1864), the number of Linnaean species names is still well below half thousand, while the most conservative estimate of the number of species was 20,000 a few years ago (Bakke et al., 2002). From the taxonomist’s point of view, the task of naming the remaining species seems hopeless, but checking of the existing names is equally demanding. Simply, there are not enough professional parasitologists for such a work, in any part of the world.

The use of molecular methods for solving taxonomic questions seems to be unavoidable. Most recently, Hahn et al. (2015) analysed two salmonid parasite species, i.e. Gyrodactylus teuchis Lautraite, Blanc, Thiery, Daniel & Vigneuille, 1999 and G. truttae Gläser, 1974 utilising both ribosomal transcribed spacer ITS2 and mitochondrial cox1 sequences, covering widely the distribution area of both species on salmonid hosts. Wide sampling of G. truttae revealed very limited genetic variation, but G. teuchis was divided into five clades, mostly associated with the divergence of the main host Salmo trutta L. in the major drainage systems in southern and central Europe. The mitochondrial variation between the allopatric clades of G. teuchis was deeper (max. c.10%) than that found in G. lucii Kulakovskaya, 1952 (c.7%) or in G. salaris Malmberg, 1957 / G. thymalli Žitňan, 1960 (max. c.3%). An exception was the “alien” mitochondrial DNA in G. salaris found on rainbow trout farm strains in Macedonia and Poland, differing by c.19% (Ziętara et al., 2010). In all these species, the rDNA variation in ITS1 and ITS2 was very limited, supporting the inclusion of the deviant strains as geographical variants of the species.

The Eurasian minnow Phoxinus phoxinus (L.) (Cyprinidae) is hosting the most diverse fauna of Gyrodactylus described on any fish species in the Palaearctic region, emphasis on “described” (Harris et al., 2004; Bakke et al., 2007). The number of species of Gyrodactylus recorded on P. phoxinus in the latest enumeration was 17 (Harris et al., 2004) and as far as we know, there are no new species added to the list. Therefore, these species might be a suitable “assemblage” to study the ecology and evolution in a host-parasite system, without the complications caused by host switching (Ziętara & Lumme, 2002). The first step in such approach must be the α-taxonomy: the names of the species in the “assemblage”.

The taxon or group of taxa in focus here was originally described on P. phoxinus as Gyrodactylus wageneri aphyae Malmberg, 1957, in the island Nämdö in Stockholm archipelago (59.18°N, 18.69°E, decimal degrees). Later, Malmberg (1964) raised G. w. aphyae to species rank, updating the convention of naming the “difficult” wageneri group members as subspecies. Further, he split the species and accepted only specimens from the freshwater localities Aneboda (Växsjö) and Jämtland into G. aphyae, and left several other samples nameless, including those in Nämdö (Malmberg, 1970). Ergens (1975) revisited the “aphyae-problem” and concluded that the complex contained an acceptable G. aphyae and another species, for which the name G. pannonicus Molnár, 1968 was available.

Ergens (1975) presented a large collection of morphometric data and drawings to study the problem of geographical and seasonal size variation in G. aphyae. His collection contained specimens from the Czech Republic, Slovakia, Hungary, Montenegro and Russia, including the Kola Peninsula. We shall analyse the “aphyae-gasterostei problem” later. From the point of view of the present study, it is a problem that the samples of G. pannonicus measured seasonally (figure 3 in Ergens, 1975), all came from the small brook Rokytka in Prague (roughly 50.09°N, 14.53°E), draining via Vltava (Moldau) and Labe (Elbe) to the North Sea, and thus are not from the watershed of the type-locality of G. pannonicus.

Our molecular survey of minnow parasites was conducted in four sea basins, the White and Baltic Sea basins in Fennoscandia; the Black Sea Basin in the Carpathian mountains in Poland and Slovakia; and the Arctic Ocean Basin at Lake Baikal in Mongolia. In addition to several other recognisable and described or new species of Gyrodactylus, this geographical survey revealed a cluster of four cryptic species closely resembling G. pannonicus, described by Molnár (1968) in River Tisza in Hungary. We recorded and tagged this species by molecular markers in Slovakia, in the watershed near the type-locality. Here we introduce three new sibling species: G. botnicus n. sp., G. albolacustris n. sp. and G. danastriae n. sp., on the basis of DNA characteristics, i.e. by barcoding by nuclear and mitochondrial DNA sequences. We also describe the morphology and standard measurements of the haptors of the species, but we want to warn that morphology cannot be used alone to distinguish these species: they are sibling and cryptic species in the strict meaning of the terms. There are differences in the drawings, but ranges of variation have not been assessed yet.

However, the most interesting observation in this phylogenetic analysis was the mito-nuclear incongruence, as indication and evidence of the reticulate phylogenetic history of the four related taxa (e.g. Funk & Omland, 2003). The first hybridisation leading to an ancient mitochondrial introgression has occurred in the phylogeny of G. albolacustris n. sp. between >1 and 0.6 million years ago. The second analogous case is postglacial: G. botnicus n. sp. and G. albolacustris n. sp. have hybridised and the progeny was found as still segregating, offering a model demonstration of the genetic process leading to reticulate phylogeny.

Materials and methods

Fish and parasite samples

Eurasian minnows Phoxinus phoxinus (L.) were sampled by a hand net or electrofishing in different localities in waters in the Baltic, White Sea and Black Sea basins, and in Mongolia (Table 1; Supplementary Fig. S1). Fish were immediately euthanised and stored in 96% ethanol. Parasites were picked from the skin or fins of the fish later in laboratory, under a stereo microscope.

Table 1 Sampling sites and GenBank accession numbers of the parasites, all on Phoxinus phoxinus (L.)
Full size table

DNA extraction, amplification and sequencing of the mitochondrial cox1 gene

DNA was isolated by digesting single parasite specimens in 10 μl of lysis buffer (1× PCR buffer, 0.45% (v/v) Tween 20, 0.45% (v/v) NP 40 and 60 µg/ml proteinase K). Tubes were incubated at 65°C for 25 minutes to allow proteinase K digestion, then at 95°C for 10 minutes to denature the proteinase and cooled down to 4°C. Aliquots of 2 µl of this lysate were used as templates for PCR amplification. The remaining impurities in the lysate did not interfere with the PCR process.

The analysis of mitochondrial DNA was conducted as described earlier (Meinilä et al., 2002; Ziętara et al., 2006; Huyse et al., 2007). The 20 μl PCR reaction consisted of 1× PCR buffer, 0.2 mM dNTPs, 2 mM MgCl2, 1 μM of each primer, 0.5 U of Fermentas Taq DNA polymerase and 2 μl of DNA template. The gene cox1 was amplified either as single or as two overlapping fragments. The complete cox1 gene fragment was amplified using primers Trp2F (5′-TTT TAG ACG ATT TGT TTT CA-3′) and Thr2R (5′-ATA GAT TGC TTG GTA TTA CA-3′). The 5′-end fragment of the shorter overlapping fragments was amplified with Trp2F and with either RCox7 (5′-GTA GGT ACA GCG ATT ATC AT-3′; G. pannonicus, G. albolacustris n. sp. or G. danastriae n. sp.) or RCox9 (5′-GTA GGT ACT GCA ATA ATC AT-3′; G. botnicus n. sp.) primers. The 3′-end was amplified with FCox7 (5′-TTT TCA ATA GGT ATG GAC GT-3′) and Thr2R. The PCR program was run as follows: 94°C for 3 min, 38 cycles (94°C for 30 s, 48°C for 1 min and 72°C for 1 min 50 s), 72°C for 7 min, and final cooling to 4°C. The same primers were used for sequencing.


Amplification and sequencing of the ITS of species and hybrids

The entire ITS region of the ribosomal DNA array (spanning ITS1-5.8S-ITS2 and flanking terminal fragments of the 18S and 28S rRNA genes) was amplified with primers ITS1F (5′-GTT TCC GTA GGT GAA CCT-3′) (Rokicka et al., 2007) and ITS2R (5′-GGT AAT CAC GCT TGA ATC-3′) (Ziętara et al., 2000). The PCR reaction contained 2 µl of lysate, 1× PCR buffer, 2 mM MgCl2, 1 µM of each primer, 200 µM of each dNTP and 0.4 unit of Taq polymerase (Fermentas, Vilnius, Lithuania) in a final volume of 20 µl. Amplification mixtures were heated for 3 min at 95°C, then subjected to 37 cycles (94°C, 48°C and 72°C for 1 min each), heated for 7 min at 72°C and cooled down to 4°C. The amplified fragments were purified from the agarose gel and sequenced directly with two additional primers ITS1R (5′-ATT TGC GTT CGA GAG ACC G-3′) and ITS2F (5′-TGG TGG ATC ACT CGG CTC A-3′) as described earlier (Ziętara & Lumme, 2003). The forward allele-specific primers for amplifying the two separate alleles from the heterozygous G. albolacustris n. sp. × G. botnicus n. sp. hybrid worms were ITS1tF (5′-GTG TTG TCA TTG CCT AAA AT-3′) for ITSalb and ITS1aF (5′-GTG TTG TCA TTG CCT AAA AA-3′) for ITSbot . They were designed to start as near of the 5′-end of the sequence as possible.


Data analysis

The phylogenetic trees and genetic distance estimates were based on the complete cox1 (1,545 bp) gene and on specified fragments of the ITS1-5.8S rDNA-ITS2. In the latter, the hypervariable segment of ITS1 (Supplementary Figs. S2 and S3) was removed from the distance calculations which therefore represent minimum values. Sequence alignments were made by ClustalW or Muscle implemented in MEGA6 (Tamura et al., 2013), leading to the same result. The phylogenetic trees were constructed by the Neighbor Joining method based on Kimura’s two parameter distance (K2P for ITS), as implemented in the MEGA6 program package. Distances for the cox1 gene were calculated by the maximum composite likelihood method, correcting the large distances (reaching saturation) upwards. The divergence time estimates were based on the 13% divergence per million years, estimated for G. salaris (see Kuusela et al., 2007). We may give the timing also for the 5.1% suggested by Hahn et al. (2015), but all the calibrations are quite weak (Supplementary Fig. S4).


Morphological descriptions

Worms investigated morphologically were cut in two parts. The opisthaptor was prepared as a microscopy slide in ammonium picrate-glycerin (Malmberg, 1970), and the rest of the body was used for molecular analysis. Isolated haptors were partially digested by Proteinase K in a final concentration of 60 μg/ml prior to their preservation in ammonium picrate-glycerin. Measurements of haptoral hard parts were taken with a microscope and digital camera (Nikon Optiphot - 2) and estimated by the interactive measuring system IMT iSolution Lite (Ver.7.4, IMT iSolution Inc.). The holotype of each species was drawn and several paratypes measured; all measurements are in micrometres.


Systematic analysis of species resembling G. pannonicus by ribosomal ITS and mtDNA

The minnow parasites were first placed into phylogenetic-systematic context by constructing a phylogenetic hypothesis for wageneri group of the subgenus Gyrodactylus (Limnonephrotus) based on 495 nucleotides (nt) of ITS2 alignment (Supplementary Fig. S5). The clade of the “pannonicus cluster” was monophyletic (93% bootstrap support). Within the pannonicus cluster, the analysed phylogenetic tree was based on 814 nt positions in the ITS1-5.8S-ITS2 alignment using pairwise deletion option (Fig. 1; Supplementary Fig. S6). The species discussed here formed a monophyletic cluster within the wageneri species group, with three deeply separated main clades (99–100% bootstrap support), and the third clade further split into two (91–99% support).

Fig. 1
figure 1

Compressed phylogenetic hypotheses based on ITS and cox1 sequences. The elbow arrow shows the positions of the mitochondrial introgression from pro-danastriae to G. pannonicus Molnar, 1968, leading to the G. albolacustris n. sp. The straight arrow shows the modern hybridisation between G. botnicus n. sp. and G. albolacustris n. sp.

Full size image

In the pannonicus cluster, most of the interspecific variation of ITS1 was in a segment closed with the conservative stem motif 5′-AAAAGUA … UACUUUU-3′ (Supplementary Figs. S2 and S3). The length of this segment was 226 nt in G. albolacustris n. sp. and G. pannonicus, 227 nt in G. botnicus n. sp. and much longer, 268 or 256 nt in G. danastriae n. sp. (Supplementary Fig. S2). While the three first named species had the ITS1 of equal length, the content was changed. The Mfold-predicted (Zuker, 2003) A, B and C types for the secondary structure of the hypervariable segment. A hairpin stem structure depicted in Supplementary Figs. S2 and S3 was the only structurally conserved motif within this hypervariable segment. Transforming this kind of qualitative and taxonomically definitive insertion/deletion variation into distance estimates is difficult and the mechanisms leading to indels remain unexplained.

As a complementation of the ITS analysis, we produced a mitochondrial phylogenetic hypothesis of the subgenus (Supplementary Fig. S7) and the species group (Fig. 1, Supplementary Fig. S6). The trees were based on an alignment of the complete cox1 gene sequence (1,545 nt, 515 amino acids. The “pannonicus cluster” appeared as a perfectly monophyletic clade (92% bootstrap support) and separated for four nearly equidistant clades (Supplementary Fig. S7).


Ancient genome shuffling via hybridisation

The incongruence of the ITS and cox1 trees was serious, with respect of the topology and distances as well (Fig. 1 and Supplementary Fig. S6, Table 2, graphical illustration in Supplementary Fig. S8). This can be explained by an ancient mitochondrial introgression via hybridisation (kinked arrow in Fig. 1) of a member of the ancestral strain from the Dniester Basin (to be G. danastriae n. sp.) with a member of the lineage from the Danube Basin (leading to G. pannonicus). This sidekick hybridisation event was the origin of the lineage leading to G. albolacustris n. sp., which has the ITS quite similar to G. pannonicus (0.8–0.9%), but the cox1 gene more related to G. danastriae n. sp. (16.8%) than to G. pannonicus (19.8%). This introgression has occurred prior to the divergence of the Mongolian and north-east European lineages of G. albolacustris n. sp. which differ by 8.9% in the mtDNA.

Table 2 ITS distances (Kimura 2-parameter model, above diagonal) and cox1 distances (Maximum Composite Likelihood method, MCL, below diagonal) for the species studied
Full size table

Contemporary genome shuffling: G. albolacustris n. sp. × G. botnicus n. sp. hybrid population

The postglacial faunal transfer from the Baltic Basin to the White Sea Basin occurred in the modern Iijoki tributary system in Finland. Lake Kitka was originally connected to the Baltic Basin in the Ancylus lake phase (Heikkinen & Kurimo, 1977; Koutaniemi, 1999). The aquatic fauna, e.g. brown trout (Huusko et al., 1990) or Gyrodactylus salaris on grayling (Meinilä et al., 2004) arrived to the lake from the Baltic. About 8,400 years ago, the isostatic land uplift tilted Lake Kitka to drain east via River Koutajoki system to the White Sea (Heikkinen & Kurimo, 1977; Koutaniemi, 1999). Since then, the faunal components of the primordial Lake Kitka had a free access downstream. Upstream dispersal was prevented by a nine meter waterfall called Jyrävä. How far and how rapidly the eastward dispersal occurred has not been yet studied.

Gyrodactylus botnicus n. sp. had evolved in isolation, until the latest postglacial period, when it succeeded to cross the line less than 8,400 years ago and mate with G. albolacustris n. sp. in the White Sea Basin (straight arrow in Fig. 1). The hybrid population (heterozygous ITS) was found in the creek Merenoja in 2001 (66.3507N, 29.3493E, 209 metres above sea level, masl). Next samplings were conducted half a km upstream, in a creek called Hangaspuro.

In Hangaspuro (66.3406N, 29.3297E, 227 masl) in years 2007 and 2008, homozygous and heterozygous combinations of ITS specific for G. albolacustris n. sp. and G. botnicus n. sp. were found in the sample analysed by rapid PCR-RFLP: seven specimens of ITSalb/alb, nine ITSalb/bot and eight ITSbot/bot. All these specimens carried the mitochondrial DNA of G. albolacustris n. sp., also those with homozygous-looking G. botnicus n. sp. ITSbot/bot. Proper G. botnicus n. sp. (including the proper cox1) was also observed in the River Oulanka, downstream of the creek.

The two allelic forms of ITS representing G. albolacustris n. sp. and G. botnicus n. sp. in the heterozygous specimens were amplified and sequenced by utilising allele-specific primers. The ITS1 sequences in the Supplementary Fig. S2 clearly show that direct sequencing of the hybrids should produce unreadable mess (as it was first observed in 2001 by us). The results suggest segregation after the primary hybridisation, in addition to the fact that the cox1 gene in Hangaspuro parasites was always from G. albolacustris n. sp.

The genotypes detected as homozygous by the PCR-RFLP were sequenced directly using universal ITS primers. Seven specimens of apparent ITSalb/alb homozygotes on the basis of PCR-RFLP were supposed to represent the maternal parent of the hybrids. However, they always showed low intensity ambiguous nucleotide characteristics to the ITSbot allele, when sequenced by using the general Gyrodactylus primers amplifying both alleles. The mean ratio of high/low peak heights in the ABI phenograms was 1: 0.27 (G. albolacustris n. sp.: G. botnicus n. sp. allele). Because the sequences were made from individuals, the aberration from 1:1 ratio indicates recombination within the repeated rDNA cassette. The recombination is modelled in Supplementary Fig. S9.

In contrast, the spontaneous ITSbot/bot homozygotes showed no ‘contaminant’ ambiguities except in a single specimen, which showed a similar variation as described above. However, these ITSbot/bot homozygotes carried the G. albolacustris n. sp. cox1 gene and thus necessarily were recovered from hybrid heterozygotes either via mating of F1 × F1 or, perhaps via backcrossing F1 × G. botnicus n. sp. male. Pure G. botnicus n. sp. was not found just in this creek, but it was observed downstream in the Oulanka River (66.3695N, 29.3352E, 155 masl).

The GenBank accession numbers of the ITS (n = 22) and cox1 (n = 22) sequenced from the various hybrid combinations are given in Table 1.


Family Gyrodactylidae van Beneden & Hesse, 1864

Genus Gyrodactylus von Nordmann, 1832

Subgenus Gyrodactylus ( Limnonephrotus ) Malmberg, 1970

Species group wageneri sensu Ziętara & Lumme (2002)


Gyrodactylus albolacustris n. sp.

Type-host: Phoxinus phoxinus (L.) (Cypriniformes: Cyprinidae), Eurasian minnow.

Other host: Alburnoides bipunctatus (Bloch) (Cypriniformes: Cyprinidae), Schneider.

Type-locality: River Vidlitsa draining to Lake Ladoga, Baltic Basin, Russia (61.1966N, 32.4213E).

Other localities: P. phoxinus: River Bol’shaya Uya draining to Lake Onega, Russia (61.6028N, 34.6603E); River Strel’na, south coast of Kola Peninsula, White Sea Basin, Russia (66.0967N, 38.5248E); Kuusamo, White Sea Basin, Finland, (66.3406N, 29.3297E), only ITS and cox1 DNA found in hybrids of G. albolacustris n. sp. with G. botnicus n. sp.; River Tuul near Ulaanbaatar, Mongolia (47.8863N, 106.9336E), in the Lake Baikal watershed, draining to the Arctic Ocean via River Yenisey. A. bipunctatus: River Vidlitsa draining to Lake Ladoga, Baltic Sea Basin, Russia (61.1966N, 32.4213E).

Site on host: Skin and fins.

Type-material: Type-specimens (the holotype and four paratypes): slides of the haptors are deposited in the Finnish National History Museum in Helsinki University, under accession numbers MZH118057 (holotype) and MZH118058–MZH118061 (paratypes) (all collected on 15.vi.2006). The Mongolian specimens are deposited as MZH118085–MZH118086 (paratypes) (collected on 28.viii.2012).

Representative DNA sequences: GenBank accession numbers for ITS (n = 18) and cox1 (n = 16) sequences are given in Table 1.

Etymology: The name albolacustris refers to the mythic region of Lacvs Albvs in the Carta Marina (1539) of Olavus Magnus, a Swedish cleric and historian. Because the eastern regions of the map were mainly based on stories of fur traders, it is not clear whether the name refers to the Lake Onega or to the White Sea, and therefore, the name is suitable to a parasite living in both basins.

Description (Figs. 2.1, 2.2, 2.6)

[Based on 7 specimens: 5 European (from the type-locality, River Vidlitsa) and 2 Mongolian specimens, see Table 3 for metrical data.] Haptor ovate. Anchor 39–52 long with fold; anchor root 13–19 long, tends to bend ventrally; anchor shaft pronounced, 33–40 long. Ventral bar 4–9 × 18–26, with short processes and medium triangular membrane, 9–10 long. Dorsal bar delicate, 1–4 long, 14–16 wide. Marginal hook 25–30 long, with robust marginal hook sickle, 4–6 long; marginal hook shaft leaning; marginal hook point extending toe and pointing backwards; filament loop long extending almost to middle of handle; transition of shaft to point not distinct.

Fig. 2
figure 2

Morphology of haptoral parts of Gyrodactylus spp. 1, G. albolacustris n. sp. from Europe (River Vidlitsa, holotype); 2, G. albolacustris n. sp. from Mongolia (River Tuul, paratype); 3, G. botnicus n. sp. from the Baltic Basin (brackish water, Varjakka, holotype); 4, G. danastriae n. sp. from south-eastern Poland (River Strwiąż, holotype); 5, G. pannonicus from Slovakia (River Ondava); 6–9, Marginal hook sickles: 6, G. albolacustris n. sp.; 7, G. botnicus n. sp.; 8, G. danastriae n. sp.; 9, G. pannonicus. Scale-bars: a (anchors), 10 µm; b (marginal hooks), 10 µm; c (marginal hook sickles), 10 µm

Full size image
Table 3 Measurements (range in μm) of haptoral hard parts of Gyrodactylus spp. from Phoxinus phoxinus. The relative length of each part (in % in parentheses) was calculated as percentage of the total length of anchor/marginal hook, using the maximum values
Full size table

DNA characteristics

The length of the nuclear ribosomal ITS1-5.8S-ITS2 is 1,218 nt (EU554412, KU365754, HM192922–HM192924). The Mongolian sequence is one nt shorter (KU365755), due to a variation G5 or G4 in sites 236–240 within the hypervariable segment of ITS1 (Supplementary Fig. S2). The hypervariable segment is 226 nt long. Its predicted secondary RNA structure is with three clear hairpins and few unpaired segments (Supplementary Fig. S3). The sequences deviate from those of G. pannonicus by ten single nt changes (0.81%). There are no long indels observed. Mitochondrial cox1 sequences (EU554409–EU554411, HM488981–HM488985) are deeply divergent from the other species in the pannonicus cluster (Fig. 1, Supplementary Figs. S6, S7, S8). Three amino acid substitutions between local isolates are found (Val > Ile at sites 419 and 460, Leu > Phe 459; Supplementary Fig. S10). The Mongolian cox1 sequence (KU365756) differs from the European sequences by 8.9% (Supplementary Fig. S8).

Remarks

The general haptor morphology of G. albolacustris n. sp. is in accordance with wageneri group sensu Malmberg (1970) as follows: folded anchors; ventral bar with short processes; marginal hook shape typical for wageneri group, with rounded heel and pointed toe; sickle point extends sickle toe which makes distal part of sickle wider than proximal part. Further, sickle point is not pointing more upwards like in G. gracilihamatus Malmberg, 1964, the outgroup (see figure 1A in Ziętara & Lumme, 2002 and figure 2 in Ziętara & Lumme, 2004) of wageneri group sensu Ziętara & Lumme (2002). Gyrodactylus albolacustris n. sp. possesses 5.8S rDNA identical with 5.8S rDNA of G. aphyae Malmberg, 1970 (GenBank AF484527). Sequence of 5.8S rDNA is the main DNA character of wageneri group sensu Ziętara & Lumme (2002).

The size range of G. albolacustris n. sp. is wider than in the other species. Specimens from Mongolia were smaller than those collected in Europe. Still, the size of G. albolacustris n. sp. hardly overlaps with the smallest values for the range of G. aphyae (Table 3). The relative point length in G. albolacustris n. sp. is 67% which is clearly more than in G. danastriae n. sp. (54%) and G. pannonicus (51%), as well as in G. aphyae samples measured by us from Topl’a, Slovakia (52%) and Tuul, Mongolia (51%), but the values given by Ergens (1975) indicate relatively longer anchor point (74%) in of G. aphyae. The marginal hook sickle of G. albolacustris n. sp. is the most prominent among the three new species, but less robust than that in G. aphyae, and differs from the latter by leaning shaft and then less prominent transition from the shaft to the point. The only constant difference from G. aphyae is the tendency of bending the roots of anchors ventrally.

The two specimens from Mongolia did not differ in shape from those collected in Europe, their dimensions fall within the lower range specified in Table 3. All the measurements overlap with the other three species in the pannonicus cluster, so they are of limited use to distinguish the species.

Many reports of G. pannonicus in north-eastern Europe may actually concern G. albolacustris n. sp., at least within the postglacially colonised triangle Ladoga - Kola Peninsula - Mongolia.

Gyrodactylus albolacustris n. sp. was found on P. phoxinus as sympatric (often also on the same host specimens) with G. botnicus n. sp., G. macronychus Malmberg, 1957, G. aphyae, G. jussii Ziętara & Lumme, 2003, G. phoxini Malmberg, 1957, G. magnificus Malmberg, 1957 and G. laevis Malmberg, 1957 (unpublished, molecularly confirmed data of the authors, Supplementary Fig. S11).


Gyrodactylus botnicus n. sp.

Type-host: Phoxinus phoxinus (L.) (Cypriniformes: Cyprinidae), Eurasian minnow.

Type-locality: Varjakka, Finland, brackish water (64.9213N, 25.1014E).

Other localities: P. phoxinus: Oulu, Finland, brackish water Baltic Basin (65.0333N, 25.4117E); Marjaniemi in Hailuoto, Finland, brackish water Baltic Basin (65.0333N, 25.4117E); Kiiminkijoki, Oulu, Finland, freshwater Baltic Basin (65.1488N 25.6629E); River Oulanka, Kuusamo, Finland, freshwater White Sea Basin (66.3695N, 29.3352E).

Site on host: Skin and fins.

Type-material: Type-specimens (the holotype and four paratypes): slides of the haptors are deposited in the Finnish National History Museum in Helsinki University, under accession numbers MZH118063 (holotype) and MZH118064–MZH118067 (paratypes) (all collected on 11.ix.2005).

Representative DNA sequences: GenBank accession numbers for ITS (n = 17) and cox1 (n = 14) sequences are given in Table 1.

Etymology: The name botnicus is derived from the Latin name Mare Botnicvm, referring to the Bothnian Bay of the Baltic Sea between Sweden and Finland in the old map of Olavus Magnus.

Description (Figs. 2.3, 2.7)

[Based on 5 specimens from the type-locality, Varjakka, Finland, brackish water, see Table 3 for metrical data.] Haptor ovate. Anchor 49–57 long with fold; anchor root 12–17 long, tends to bend ventrally; anchor shaft shorter, 24–28 long. Ventral bar 4–8 × 18–23, with short processes and medium triangular membrane less pointed, 8–9 long. Dorsal bar delicate, 1–2 long, 13–14 wide. Marginal hook 21–30 long, with prominent marginal hook sickle, 4–5 long; marginal hook shaft more strait; marginal hook point extending toe and pointing less backwards; filament loop long extending almost to middle of handle; transition of shaft to point not distinct.


DNA characteristics

The length of nuclear ribosomal ITS1-5.8S-ITS2 is 1,214 nt (AF484542, EU554413, EU554416, EU727177). It is most basal when the pannonicus cluster is embedded in wageneri group phylogeny (Supplementary Fig. S5). The hypervariable segment is 227 nt long. Its predicted secondary RNA structure is with five tight hairpins and many unpaired segments (Supplementary Figs. S2, S3). The sequence deviates from G. pannonicus by numerous single nucleotide changes (4.2%) in well-aligned segments. There are longer indels in the hypervariable segment. Mitochondrial cox1 sequence (EU554421–EU554422, HM488975–HM488990) is deeply divergent from the other species in the pannonicus cluster (Supplementary Figs. S6, S7, S8).

Remarks

The general haptor morphology of G. botnicus n. sp. is in accordance with wageneri group sensu Malmberg (1970) (see the remarks in the first paragraph for G. albolacustris n. sp.). Gyrodactylus botnicus n. sp. possesses 5.8S rDNA identical with G. aphyae Malmberg, 1970 (GenBank AF484527). Sequence of 5.8S rDNA is the main DNA character of wageneri group sensu Ziętara & Lumme (2002).

Gyrodactylus botnicus n. sp. is the largest of the new species described here, but the samples are from northern part of the Baltic Sea Basin, thus perhaps influenced by the colder waters. The species G. danastriae n. sp. and G. pannonicus hardly overlap with G. botnicus n. sp. in the total length of the anchor. Also the anchor point is relatively longer (75%). The sickle point of G. botnicus n. sp. marginal hook less extended its sickle toe. It has a very thin dorsal bar and a ventral bar with small processes. Ventral bar membrane is rounded.

Specimens from Oulu and Oulanka were originally reported as G. pannonicus (Ziętara & Lumme, 2002) but it was confirmed in this study as misidentification, and the GenBank entries were corrected. Blazek et al. (2008) most probably were experimenting with G. botnicus n. sp. in Finnish inland lakes, instead of G. pannonicus, but the species name has no relevance with respect of their results: in a multiple-choice experiment, pannonicus-like parasite remained on P. phoxinus.

The species has been found on P. phoxinus together with G. aphyae, G. macronychus, G. phoxini, and G. laevis.


Gyrodactylus danastriae n. sp.

Type-host: Phoxinus phoxinus (L.) (Cypriniformes: Cyprinidae), Eurasian minnow.

Type-locality: River Strwiąż near Brzegi Dolne draining to River Dniester, Black Sea Basin, southeaster corner of Poland in Carpathian Mountains (49.4544N, 22.6393E).

Site on host: Skin and fins.

Type-material: Type-specimens (holotype and four paratypes): slides of haptors are deposited in the Finnish National History Museum in Helsinki University, specimen numbers MZH118051 (holotype) and MZH118052–MZH118055 (paratypes) (all collected on 31.vii.2007).

Representative DNA sequences: GenBank accession numbers for ITS (n = 5) and cox1 (n = 5) sequences are given in Table 1.

Etymology: The name danastriae is from an ancient Latin literature name Danastris of the river Dniester.

Description (Figs. 2.4, 2.8)

[Based on 5 specimens from the type-locality, River Strwiąż near Brzegi Dolne.]. Haptor ovate. Anchor 45–48 long with fold; anchor root 12–15 long, tends to bend ventrally; anchor shaft 33–35 long. Ventral bar 5–7 × 17–19, with short processes and longer triangular membrane prominently pointed, 10–12 long. Dorsal bar delicate, 1–2 long, 12–16 wide. Marginal hook 21–25 long, with prominent marginal hook sickle, 4–5 long; marginal hook shaft more strait; marginal hook point less extending toe and pointing downwards; filament loop long extends almost to middle of handle; transition of shaft to point more distinct.


DNA characteristics

The length of nuclear ribosomal ITS1-5.8S-ITS2 is either 1,258 (HM192925–HM19292) or 1,246 nt (HM192928). The hypervariable segment is either 268 or 256 nt long. Intraspecific variation is due to 12 nucleotides long heterozygous indel GAGAGTGTAAGA, situated in tandem with more GAGA repeats. Predicted secondary RNA structure is with five tight hairpins and many unpaired segments. The sequence deviates from G. pannonicus by many single nucleotide changes (3.9%) in well-aligned segments. There are several long indels observed in the hypervariable segment (Supplementary Figs. S2, S3). Mitochondrial cox1 sequence (HM488971–HM488974) is deeply divergent from the other species in the pannonicus cluster, least from European G. albolacustris n. sp. (16.8%; Supplementary Fig. S8). A single unique amino acid substitution Leu > Met is present in site 442 (Supplementary Fig. S10.).

Remarks

The general haptor morphology of G. danastrae n. sp. is in accordance with wageneri group sensu Malmberg (1970) (see the remarks in the first paragraph for G. albolacustris n. sp.). Gyrodactylus danastriae n. sp. possesses 5.8S rDNA identical with 5.8S rDNA of G. aphyae Malmberg, 1970 (GenBank AF484527). Sequence of 5.8S rDNA is the main DNA character of wageneri group sensu Ziętara & Lumme (2002). The anchor in Gyrodactylys danastriae n. sp. is slightly smaller than in the other two new species and the relative point length is only 54%, almost as small as in G. pannonicus in our measured material. In spite of all effort, measuring the haptors is not exact science, and the dimensions vary seasonally (Ergens, 1975). The marginal hook sickle is clearly more curved than in other species and resembles more G. aphyae sickle but is less massive. Sickle point is almost not extending sickle toe. The dorsal bar is very thin and the ventral bar is with small processes. Ventral bar membrane is longer and more oval in comparison with the three other species.

The species was found in River Strwiąż with G. aphyae and an undescribed species related to G. aphyae.


Gyrodactylus pannonicus Molnár, 1968

Type-host: Phoxinus phoxinus (L.) (Cypriniformes: Cyprinidae), Eurasian minnow.

Other hosts: Barbus barbus (L.) (Cypriniformes: Cyprinidae), barbel (Molnár, 1968 ).

Localities: Present study only. River Ondava draining to Danube, Black See Basin, Slovakia (49.3513N, 21.4635E); River Topl’a draining to Danube, Black Sea Basin, Slovakia (49.3133N, 21.1932E).

Site on host: Skin and fins.

Voucher material: Specimens collected from the same watershed as the holotype: slides of the haptors are deposited in the Finnish National History Museum in Helsinki University, under accession numbers MZH118040–MZH118047 (River Ondava, near Nižný Mirošov) and MZH118048–MZH118049 (River Topl’a, near Rokytov) in Slovakia (all collected on 30.vii.2007).

Representative DNA sequences: GenBank accession numbers of ITS (n = 15) and mitochondrial cox1 (n = 15) sequences are given in Table 1.

Description (Figs. 2.5, 2.9)

[Based on 10 specimens: 8 from River Ondava and 2 from River Topl’a.] Haptor ovate. Anchor 45–48 long with fold; anchor root 12–17 long, tends to bend ventrally; anchor shaft 33–36 long. Ventral bar 4–5 × 15–20, with short processes and triangular membrane prominently pointed, 9–11 long. Dorsal bar delicate 1–2 long, 12–17 wide. Marginal hook 24–27 long, with delicate marginal hook sickle, 5–6 long; marginal hook shaft leaning prominently; marginal hook point visibly extending toe and pointing backwards; filament loop long extends almost to middle of handle; transition of shaft to point not distinct.


DNA characteristics

The length of nuclear ribosomal ITS1-5.8S-ITS2 is 1,217 nt (HM192915–HM192921). The hypervariable segment is 226 nt long. It is similar to G. albolacustris n. sp.. The predicted secondary RNA structure of the hypervariable segment is with three tight hairpins and few unpaired nucleotides. Both isolates from River Ondava and River Topl’a share A/R/G heterozygosity (Supplementary Figs. S2, S3). Mitochondrial cox1 sequence (HM488961–HM488970) are deeply divergent from the other species in the pannonicus cluster, least divergent from Mongolian G. albolacustris n. sp. (18.1%; Supplementary Fig. S8). A unique amino acid substitution Ala > Val in site 134 in isolate from River Ondava is observed (Supplementary Fig. S10.).

Remarks

The general haptor morphology of G. pannonicus is in accordance with wageneri group sensu Malmberg (1970) (see the remarks in the first paragraph for G. albolacustris n. sp.). Gyrodactylus pannonicus possesses 5.8S rDNA identical with G. aphyae Malmberg, 1970 (GenBank AF484527). Sequence of 5.8S rDNA is the main DNA character of wageneri group sensu Ziętara & Lumme (2002). Parasites from authentic description of Molnár (1968) did not differ in size and shape from those found and investigated in the Rivers Ondava and Topl’a. All sizes vary within the limits specified in Table 3. The maximum values of all diagnostic characters of G. pannonicus measured in this study are either shorter or equal to minimum metrical values given by Gusev et al. (1985) for G. aphyae.

The Slovakian Rivers Ondava and Topl’a are connected upwaters of the tributary system Tisza to Danube to the Black Sea, i.e. from the tributary system of the type-locality of G. pannonicus, and therefore we feel safe to suggest that the DNA sequences reported here represent the genuine G. pannonicus Molnár, 1968. We suggest that it is the best way of tagging with DNA markers already described Gyrodactylus species. The topotype specimens should be collected from the same watershed as the type-material, although not necessarily at the same time (Winston, 1999).

Because the specimens described and measured as G. pannonicus and compared with G. aphyae by Ergens (1975) were from the brook Rokytka in Prague, North Sea Basin, it is possible in the present context that they do represent some other, more western lineage than the genuine G. pannonicus Molnár, 1968.

Gyrodactylus pannonicus was reported by Ziętara & Lumme (2002) from the Bothnian Bay area, but because the ITS sequence (AF484542) was provided, we here confirm it as a misidentification, and the GenBank entry was corrected to represent G. botnicus n. sp. The experiments done by Blazek et al. (2008) in Finland most probably were made using G. botnicus n. sp. and not G. pannonicus. The species name G. pannonicus has been reported in many regions from the Baltic Basin to Mongolia, e.g. by Rumyantsev & Ieshko (1997), Gusev et al. (1985), Pugachev (2002), and Pugachev et al. (2009). Perhaps Ergens & Dulmaa (1967) described parasites of this cluster also as Gyrodactylus sp. from skin of minnow from River Kerlen (Mongolia). Specimens from River Kolyma (Pugachev, 1983) clearly represent larger parasites, and the molecular comparison might be very interesting.

This list of the wide use of the name G. pannonicus in wide geographical areas is in contrast to our results, and demonstrates that even the best specialists have not observed anything suspicious among the samples by measuring and comparing the parasites visually, in distant localities.

In River Ondava, G. pannonicus was found on minnows with G. aphyae. In River Dunajec on the Baltic side of the watershed, G. pannonicus was not found among twenty one G. aphyae specimens.

Discussion

The convention of using the internal transcribed spacers of the tandem repeats of nuclear ribosomal gene (ITS rDNA) for species identification of Gyrodactylus was initiated in the mid-nineties (Cunningham et al., 1995a, b; Cunnigham, 1997).

The ITS of most species of the genus can be amplified by a set of optimised primers, not only among salmonid parasites (Cable et al., 1999; Cunningham et al., 2001) but widely in parasites found from many fish species in European freshwaters (Matějusová et al., 2001, 2003; Ziętara & Lumme, 2002, 2003, 2004), as well as among subgenera, species groups and species in European marine environments (Huyse & Volckaert, 2002, 2005; Huyse et al., 2003, 2015; Huyse & Malmberg, 2004), in North American coasts (e.g. Gilmore et al., 2012), in the Caribbean (Xavier et al., 2015), in the Far East Asia (Hayward et al., 2001), in Amazonas (Boeger et al., 2003), Antarctica (Rokicka et al., 2009), Chile (Ziętara et al., 2012) and Africa (Vaughan et al., 2010; Přikrylová et al., 2012).

The ITS marker serves perfectly the purposes of a taxonomist. The most important rationale confirmed later is that ITS is almost invariable in conspecific samples, and homozygous and homogeneous in all the specimens studied (Ziętara et al., 2000). It has been suggested that a genetic distance of 1% between homozygous ITS sequences might justify species separation within the genus (Ziętara & Lumme, 2003; Huyse et al., 2004). This suggestion was based on observations that this small genetic differentiation was detected between sympatric parasites on different hosts, which were originally described as separate species on the basis of morphology by one of the best specialists (Malmberg, 1970). The lack of heterozygous ITS types a posteriori justified the separation of species (Ziętara & Lumme, 2003). Considering the material in this study, the genetic distance 0.81% of the ITS of G. albolacustris n. sp. and G. pannonicus just remains below the 1% level we have ourselves suggested to tell apart species (Ziętara & Lumme, 2003). However, the 19.8% difference in the mtDNA supports the decision made.

The rarity of heterogeneity in the rDNA is the evidence that the mechanisms of concerted evolution (e.g. Polanco et al., 1998, 2000) homogenise the radical changes in ITS observed between species of Gyrodactylus. Even among the closely related wageneri species group, the length of ITS1 is variable, indicating that insertions and deletions are frequent in the evolutionary time scale (Ziętara & Lumme, 2002 ). Indels were also separating the species studied in this paper, and 11 bp indel was found even segregating in the population of G. danastriae n. sp. (Supplementary Fig. S2).

A clear disadvantage of the ITS is that it is not well suitable for molecular clock approach. Especially the ITS1 seems not to be evolving via point nucleotide substitutions, but rather as large blocks of insertions and/or deletions. In the Supplementary Fig. S4, some tentative transformations of the genetic distances to divergence timing are given. For the ITS, Huyse & Volckaert (2005) suggested divergence rate of 0.7%/My when studying the rugiensis species group. For mitochondrial DNA divergence rate within the subgenus Limnonephrotus, 13%/My was suggested by Kuusela et al. (2007) and 5.1%/My by Hahn et al. (2015). According to the phylogenetic hypothesis provided here, the comparisons between ITS and cox1 of G. danastriae n. sp., G. botnicus n. sp. and G. pannonicus distances are without shuffling effects, and the 5.1%/My seems to be more in concert with the ITS suggestion. However, both calibrations are quite unreliable.

A very good advantage of the discontinuous variation of the ITS is that it is easy to construct single step DNA-based identification tools for faunistic comparisons (e.g. Rokicka et al., 2007; Supplementary Fig. S11). From experience we claim that species identification by PCR-RFLP method is easier, faster, applicable to a technician, and more accurate than morphological inspection, also avoiding seasonal variation. Therefore, species description based mainly on DNA characteristics, or at least supported by DNA is clearly preferable (Ziętara & Lumme, 2003).

As demonstrated in the results of this study, the gene trees may depart, and the species trees are another phenomenon (Vanhove et al., 2013). In some animal groups, nuclear and mitochondrial evolutionary trajectories depart by continuous or repeated hybridisations (Linnen & Farrell, 2007, 2008; Toews & Brelsford, 2012). In the pannonicus cluster, two hybridisations were demonstrated, one ancient and the other contemporary. Without analysing all four lineages by nuclear and mitochondrial marker, the ancient hybridisation had remained undetected.

Some contemporary cases of hybridisation between divergent lineages of Gyrodactylus have been reported in fish farms, on introduced host species Oncorhynchus mykiss Walbaum, 1972 (see Rokicka et al., 2007; Kuusela et al., 2008; Ziętara et al., 2010; Ieshko et al., 2015). Hybridisation might instantly lead to higher degree of heterozygosity than a transient change, and it should therefore be very obvious and easily detectable by molecular analysis. In the above fish farm cases, the hybrid clones were permanently heterozygous and showed no segregation. As a hermaphroditic parthenogenic flatworm, a single Gyrodactylus hybrid is capable to found a viable deme. Evidently, the ancient and modern hybrids observed here among the minnow parasites have been sexual and the heterozygosity and variation in the ribosomal gene cassette has been or is going to be eliminated via Mendelian segregation and/or by the process of concerted evolution.

This specific taxonomic and systematic case will contribute significantly to the understanding of the evolution and speciation process among Gyrodactylus. Adaptive radiation via host-switching has been postulated as an explanation of the high species number in this genus, but the genetic processes associated with host-switching remain elusive (Ziętara & Lumme, 2002). In the present case, the evolving parasites remained on minnow and experienced allopatric mode of divergence, but the hybridisation might be the mechanism to obtain genetic variation needed for adaptation to new hosts as well.