39

Deep divergence among subgenera of Gyrodactylusinferred from rDNA lTS region

M. S. ZIETARA 1.II.S.,T. HUYSEI, J. L UMMEs and F. A. VOLCKAERT1

1Laboratory of Aquatic Ecology, Katholieke Universiteit Leuven, Ch. de Bériotstraat 32, B-3000 Leuven, Belgium2Biological Station, Gdansk University, Ornitologów 26, PL-80-680 Gdansk-Górki Wsch., Poland3Department of Biology, University of Oulu, POB 3000, Fi-90014 Oulun Yliopisto, Finland

(Received 27 April 2001,. revised 6 August 2001,. accepted 6 August 2001)

SUMMARY

This paper adds new insight to a molecular phylogeny of Gyrodactylus, based on a complete sequence of the lTS rDNAregion of 4 sub genera and a more detailed molecular analysis. We propose a hierarchical approach in elucidating thephylogeny of this species-ri eh genus. A total of 37 sequences (915-1239 bp) from 10 representative species from 4 out of6 subgenera, as defined by Malmberg (1970), are included in the analysis. Genetic differences observed at the 5.8S locusprovide objective criteria to separate (sub)genera, while deep genetic differences of the spacers form a sound basis forspecies-specific identification. We demonstrate that each Gyrodactylus subgenus possesses a unique sequence of the 5.8Sgene. Thus, there is concordance between the 5.88 gene and the excretory system used by Malmberg (1970) as a diagnosticcharacter of subgenus status. At the species level, there is a discrepancy between morphological and molecular variation.Whereas the morphological variation, expressed in the shape and size of the attachment apparatus, is very low, themolecular variation, expressed at the lTS 1 and IT82 regions, is very high. This can be attributed either to the fast evolvinglTS region or to the fact that the genus consists of groups of a higher taxonomie level than previously recognized.Perspectives and limitations of using the lTS rDNA region for a molecular phylogeny of th is genus are discussed.

Key words: lTS, 5.88 rRNA gene, Monogenea, phylogeny, Gyrodactylus.

INTRODUCTION

With more than 400 species named (Williams &Jones, 1994), the genus Gyrodactylus is extremelyrich in species. It is only recently th at Gyrodactylusresearch has been given more attention, especiallysince G. salaris caused major losses in the salmonindus try (johnsen & Jensen, 1991). It is likely thatmany more species will be added to this long list.The high taxonomic diversity seems predictabiefrom evolutionary models th at include the high hostand habitat specificity, and the combination of clonaland sexual reproduction (Suomalainen, Saura &Lokki, 1987; Harris, 1993; Keam, 1994; Page &Hafner, 1996). According to Brooks & McLennan(1993) the high species diversity in comparison withtheir sister group the Cestodaria can be attributed toputative key innovations displayed by the Mono-genea. They hypothesized that the evolution of thedirect life-cycle and progenesis influences the rate ofadaptively driven speciation in this group. Thegyrodactylids display the extreme condition of thisdevelopmental trend, viviparity. A high number ofspecies within a genus mayalso indicate that thegenus is too loosely defined. In either case, moretaxonomic attention is required.

· Corrcsponding author: Gdansk U niversity, BiologicalStation, 80-680 Gdansk, Ornitologów 26, Poland. Tel:+48583080714. Fax: +48583080703.E-mail: zietara@biotech.univ.gda.pl

In contrast to high species diversity, Gyrodactylus. shows considerably more anatomical eonservatism in

its attachment apparatus (haptor) and copulatoryapparatus than any other monogenean parasite,which might reflect the highly sueeessful nature ofthe special combination of viviparity and progenesis(Keam, 1994; Cable et al. 1999). Consequently, thenumber of useful taxonomie charaeteristics is ex-tremely reduced. Nevertheless, a morphologicalphylogeny of the genus has been proposed byMalmberg (1970), based on characteristics of theexcretory system studied in living specimens and theattachment apparatus. The genus has been sub-divided into 6 subgenera : Gyrodactylus s.str.,Mesonephrotus, Metanephrotus, Paranephrotus, Neo-nephrotus and Limnonephrotus. The division of thesubgenera into species groups is based on mor-phological differences of the sclerites forming thehaptor. However, identification based solely onmorphometry is diffieult because of intraspecificvariation induced by climate and habitat (Harris,1993; Appleby, 1996; Shinn et al. 1996; Geets,Appleby & Ollevier, 1999). By using charactersindependent of morphology, molecular phylogenycan add a new dimension that promotes 'a tot alevidence' approach towards obtaining a true phy-logeny.

Nowadays, more and more phylogenies are con-structed based upon sequence comparisons (Pagel,1999) but at the same time many questions arise on

Parasitology(2002), 124. 39-52. «;>2002 Cambridge Vniver.ity Pre..OOI: 10.1017/80031182001008939 Printed in the United Kingdom

Reprinted with kind permission from Cambridge University Press

M. S. Ziçtara and others

the interpretation of the molecular data and theirtranslation into functional phylogenies. Page &Charleston (1997) point to the potential danger inobtaining a gene tree different from the species tree.This can be overcome by using more than one geneand by comparing several evolutionary mode Is andtree-building methods.

Cable et al. (1999) presented a first molecularphylogeny of 11 Gyrodactylus species based on ashort 278 bp lTS 1 fragment and a combination ofthe 5.8S gene and ITS2. Species were divided into 2groups having either a short or long lTS 1 sequence,which matched with the sub genera Mesonephrotusand Metanephrotus vs. Limnonephrotus. The analysisbased on the 5.8S gene alone was found to bephylogenetically uninformative and the lTS 1 wastoo variabie for confident alignment.

In this paper, we compare Malmberg's (1970)division of the genus with the molecular phylogenyconstructed with the lTS sequences of 4 differentsubgenera. We show th at the subgenus division ofMalmberg is .natural' and indeed follows thephylogeny, but the subgenus is a very low systematicrank to describe this variation. The genet ic di-vergence presented in this paper seems to be muchdeeper in comparison to other animal groups. Avise& Johns (1999) argued that current classifications failto carry useful comparative information because ofthe lack of standard criteria for taxonomic ranks.The genus Gyrodactylus seems to be one of theextremes in this respect. To minimize problems withalignment of the lTS region, a new hierarchicalapproach for constructing the mol ecu lar phylogenyis proposed. Perspectives and limitations in using thelTS rDNA region for molecular phylogeny of thisspecies-rich genus are discussed.

MATERlALS AND METHODS

Specimens of 7 Gyrodactylus species were collectedfrom the wild in Belgium, Sweden and Finland in1997-1999. Host fish species, locality, habitat (salty,brackish or freshwater) are given in Table 1. ThelTS region encompassing the internal transcribedspacers (ITSI and ITS2) and the small ribosomalsubunit (5.8S rRNA) gene was amplified. Three to6 specimens of each species were analysed. Allparasites were identified morphologically to specieslevel prior to DNA analysis. Each specimen wasindividually placed in 5}t1 of milli-Q water andstored at -20°C until required.

DNA extraction, amplification and sequencingwere as described by Zi~tara et al. (2000). The ITSIprimer of Cunningham (1997) 5'-TTTCCGTAG-GTGAACCT -3' was used as the forward primer.To avoid amplification of the fish rRNA region, anew primer called ITS2R 5' -GGT AATCACGCT-TGAATC-3' was designed based on a comparisonof the 28S rDNA 5' ends between Gyrodactylus

40

arcuatus and 2 fish, Gasterosteus aculeatus andPomatoschistus minutus. Eight sites out of 18 weredifferent. The primer was designed so that therewere 3 divergent nucleotides at the 3' end of theprimer preventing the amplification of fish DNA.

Two or three additional primers, complementaryto either 5.8S or ITSl, were used for sequencing:ITSIR 5'-ATTTGCGTTCGAGAGACCG-3',ITS2F 5'- TGGTGGA TCACTCGGCTCA-3' andlTS 18R 5' -AAGACT ACCAGTTCACTCCAA-3'.

The dideoxy terminator cycle sequencing methodwas used. Both strands of the DNA were sequenced.DNA from G. arcuatus and G. salaris was sequencedusing the ABI PRISM Big Dye Terminator CycleSequencing Kit (PE Biosystems) and ABI 377 DNAsequencer. G. rugiensis DNA was sequenced fol-lowing the protocol of SequiTherm EXCEL IJ(Epicentre Technologies). Samples were denaturedfor 4 min at 97°C, then incubated for 30 s at 95°C,30 s at 58°C and 45 s at 70°C for 30 cycles, followedby a final extension step at 70°C for 7 min. Thereaction products were separated on a LICOR 4200system and visualized on a 6 % Long Ranger gel(FMI BioProducts).

Additional sequences were obtained fromGenBank. Species names and accession number arefor the Digenea: Dolichosaccus symmetrus (LOI631),Echinostoma caproni (U58098), E. hortense (U58101),E.liei (U58099), E. paraensei (U58100), E. revolutum(consensus from AF067850, U58102), E. trivolvis(consensus from AF067851, AF067852, U58097),Schistosoma intercalatum (Z21717), S. haematobium(Z21716), S. mansoni (consensus from L03658,U22168, X85246, AA525615, AA528926, A1559064)and S. mattheei (Z21718); Monogenea: Gyrodactylusarcuatus (AF156668, AJOOI839), G. branchicus(AFI56669), G. derjavini (AJI32259, AJOOI840), G.gasterostei (AJOOI841), G. pungitii (AJOOI845), G.salaris (Z72477, AJOOI841) and G. truttae(AJI32260); Turbellaria: Schmidtea mediterranea(AF047853) and Porifera: Hymeniacidon sanguinea(XOOI32).

Sequences were aligned with the Clustal W(version 1.7) multiple sequence alignment program(Thompson, Higgins & Gibson, 1994). The con-served secondary structure elements that wereinferred for each sequence independently using thealgorithms as implemented in the program MFOLD(Zuker, 1989) were taken into consideration whileadjusting the alignment. A DNA dot matrix com-parison was performed to visualize the alignable sitesof the lTS region, using the Dotlet software (Junier& Pagni, 2000). Based on this information, 3 sets ofaligned sequences were prepared. One consisted ofthe 5.8S rDNA only, the second consisted of 5.8Sand the ITS2 rDNA and the third consisted of tbe 2

separate spacer regions ITSI and ITS2. Allambiguities and gaps longer than 1 nucleotide wereexcluded from further phylogenetic analysis. In a

Mo/ecu/ar phy/ogeny of Gyrodactylus

z

(/)...

'ü...!:>.(/)...'"o:r:

'"...'ü...!:>.

en o

.-ei'"

.

41

first step we used Modeltest 1.05 to select the modelof DNA evolution that best fitted the data based onlog likelihood scores (Posada & CrandaIl, 1998).Secondly, we compared the base composition for aIlsequences using a 5 % X2 test on the ave ragecomposition (PUZZLE, Strimmer & Von Haeseler,1996). The molecular-clock hypothesis was testedassuming the TrN model (Tamura & Nei, 1993)with y-distributed rates across sites, with thelikelihood ratio test for the clock hypothesisimplemented in PUZZLE. Maximum likelihood(ML), using the parameters estimated under thebest-fit model, and maximum parsimony (MP)analysis were carried out on tbe 5.8S rDNAsequences with PAUp. v. 4.01 b (Swofford, 1999).With MP aIl characters were weighted equally andgaps were treated as missing data, for bootstrapping(n = 1000) the branch and bound algorithm wasapplied.

To infer a phylogeny based on 5.8S and ITS2, weused maximum parsimony (MP), maximum like-lihood and distance-based methods (PAUp.). Withmaximum parsimony the foIlowing models wereused: unweighted parsimony with aU charactersunordered and gaps treated as both missing data andfifth character; weighted parsimony with transition:transversion (ti/tv) ratios 10: 5 for 5.8S and 1: 5 forITS2; and transvers ion parsimony (ti/tv weight 0:1). The maximum likelihood analysis was performedusing the parameters estimated under the best-fit

. model. With maximum parsimony and maximumlikelihood we conducted the exhaustive searchmethod and bootstrapped (n = 1000/100) with thebranch and bound algorithm. With the minimum-evolution distance method, the di stance matrix wascalculated using the paralinear/LogDet distances.We conducted the heuristic search method and

bootstrapped (n = 1000) with the tree-bisection-reconnection (TBR) branch-swapping algorithm inforce.

A third dataset was prepared for the 5 Limno-nephrotus species, consisting of the lTS 1 and ITS2rDNA, with aU ambiguities and gaps longer than 1nucleotide excluded (a total of 901 bp). The datawere treated like the second dataset. In order to testthe presence of saturation in the lTS 1 and ITS2sequences, DAMBE 4.0.24 (Xia, 2000) comparedthe saturation index expected when assuming fuUsaturations with the observed saturation index. A t-

test with infinite degrees of freedom was used toassess statistica 1significance.

RESULTS

General characteristics of the rDNA sequenees

Thirty-four parasite sequences of the lTS regionfrom 7 species and 7 localities were obtained(Tabie 1). The amplified fragments encompassingthe 3' end of 18S rRNA gene, ITSl, 5.88 rRNA

·(I) Ziçtara et al. 2000; (2) Cunningham, Aliesky & Coltins, 2000.

Table 3. Intraspecific variation in Gyrodactylusgasterostei, G. pungitii and G. sp. 1

(The numbers correspond to the position counted fromfirst nucleotide aCterthe primer. R - A, G; W - A, T; Y -C, T.)

971

CY

210

GRR

gene, ITS2 and 5' end of the 28S rRNA gene variedin size from 939 bp in the not-yet-describedGyrodactylus sp. 1 from Pomatoschistus lozanoi andP. minutus to 1236 bp in G. pungitii. The sequenceshave been deposited in GenBank under accessionnumbers G. arcuatus (AF328865), G. gasterostei(AF328867), G. micropsi (AF328868), G. pungitii(AF328869), G. rugiensis (AF328870), G. salaris(AF328871) and G. sp. 1 (AF328866).

Coding regions were identified by comparativealignment with previously published sequences(Kane & Rollinson, 1994; Cunningham, Aliesky &Collins, 2000j Zit:tara et al. 2000). All 5.8S rDNAsequences had the same length of 157 bp. The shortfragments of 18S rDNA and 28S rDNA (15 bp and9 bp, respectively) were invariable. The length ofITS2 varied from 392 bp in G. arcuatus to 428 bp inG. gasterostei, G. derjavini, G. pungitii, G. salaris andG. truttae. The pronounced length differences in thetotal amplified fragments we re mainly due to

different lengths of ITS1, varying from 364 bp in G.sp. 1 to 654 bp in G. truttae. Tbe GC content wasslightly biased towards AT in the spacers (41 to48%) and towards GC in 5.8S rDNAs (51 to 52%)(Tabie 2), which is in agreement with related taxa(Luton, Walker & Blair, 1992; Morgan & Blair,1995; Zit:tara et al. 2000).

All sequences obtained were consistent within aspecies although some intro-specific variation wasalso observed. There were 2 varia bIe sites within thespecies of G. gasterostei and G. sp. 1 and 1 in G.pungitii (TabIe 3). No intro-individual length vari.ution of lTS was detected.

The rDNA of the following species G. arcuatusand G. sp. 1. (Mesonephrotus); G. miCTopsi, and G.rugiensis (Paranephrotus) and G. derjavini(AJI32259); G. gasterostei, G. pungitii, G. truttae(AJl32260); and G. salaris (Limnonephrotus) werealigned separately. The alignment of 5 species ofsubgenus Limnonephrotus was 1259 bp long with 291(23'1 %) variabIe sites, 244 (36'3 %) in ITSI and 47(10'9 %) in ITS2. There was a 176 bp long fragmentwithin ITSI th at hampered alignment considerablyas visualized in Fig. 1. The alignments of only 2species from subgenus Mesonephrotus and Para-nephrotus were more reliable than the alignment ofthe 5 species belonging to the sub genus Limno-nephrotus. There were no regions with very longgaps; the longest gap of 11 bp was observed in lTS 1of Mesonephrotus and the other gaps were not longerthan 4 nucleotides. The final alignment of Meso-nephrotus resulted in a length of 954 bp. Altogetherthere were 120 (12'6%) variabIe sites, 83 (20'9%) inlTS 1, 1 (0'6%) in 5.8S rDNA and 36 (9'0%) inITS2. The lTS region alignment of Paranephrotuswas 1179 bp long with 196 (16'6 %) variabIe sites,145 (23'3 %) in ITSI and 51 (12'7%) in ITS2.

Evaluation of GenBank sequences

We checked the quality of sequences deposited inGenBank. This is a delicate task because even if a

M. S. Ziftara and others 42

Table 2. Length of components of the lTS region and base composition for Gyrodactylus speciesanalysed in this study

ITSI 5.8S ITS2 TotalSpecies. bp GC% bp GC% bp GC% bp GC%

G. arcuatus 392 48 157 52 392 47 941 48G. branchicusl 372 47 157 52 402 43 931 45G. derjavini2 654 44 157 52 428 42 1195 44G. gasterostei 612 44 157 52 428 42 1197 44G. micropsi 607 45 157 52 401 43 1165 45G. pungitii 627 44 157 52 428 42 1212 44G. rugiensis 599 39 157 52 401 41 1157 41G. salaris 623 43 157 52 428 42 1208 44G. sp. 1 364 48 157 51 394 46 915 48G. truttae2 654 44 157 52 428 41 1239 44

Number ofSpecies individunis Substitution-G. gasterostei 581

-1 W4 W

G. pungitii 340

1 W4 T

G. sp. 1 209

4 C1 C1 Y

Molecular phylogeny of Gyrodactylus

A G. mi erop si

G. truttae

.. ..IT51 5.85 ITS2

43

B G. salaris

ITSl

...5.85 ITS2

Fig. 1. DNA dot matrix of the lTS region (ITS1, 5.8S and ITS2): Gyrodactylus truttae vs. G. micropsi (A); G.truttae vs. G. salaris (B). The analysis was performed using the Dotlet software (Junier & Pagni, 2000). Every dotrepresents an l1-nucleotide stretch with a similarity greater than 60 %.

Table 4. Comparison of sequences obtained from differentlaboratories

Number of substitutions

ITS2

28 (6)

3

2 (1)

1 (1)o

. (1)This study; (2) Zi~tarael al. (2000); (3) Cableel al. (1999); (4) Cunningham(1997); numbers in parentheses refer to insertions/deletions within the variatioll.

region is recognized as highly conserved, substi-tutions might occur. There are 4 sources of Gyro-dactylus lTS sequences, which have been processedeither by Cunningham (1997), Cable et al. (1999),Zi~tara et al. (2000), or in this paper. We comparedaU the sequences reported for each species. Thesequences obtained by these authors differed intra-specificaUy by 1 to 65 nuc1eotides. Because this isfar beyond the expected intraspecific variation,additional measures were taken to evaluate thecorrectness of the sequences.

First, 5.88 rDNA sequences were compared(Tabie 4). The subgenera differ from each other by 1

or several nuc1eotides, but within the subgeneraLimnonephrotus (5 species) and Paranephrotus (2species), no variation was detected. Among Meso-nephrotus, a GIT substitution in position 112separates the 2 species (Appendix Fig. Al). AU 5.88rDNA sequences submitted by Cunningham (1997)were invariable and identical with ours unlike thosesubmitted by Cable et al. (1999).

Secondly, the spacer sequences were compared(Tabie 4). Some of the observed variation could beexplained as intraspecific variation such as forexample the sequence of G. arcuatus (AF 156668vs.ours) or G. salaris (Z72477 vs. ours). However, it is

Species and accession numbers. 'ITSl 5.88-G. arCllatus

(AF328865)1/(AFI56668)2 0 0

(AF328865)1/(A]001839)" 3 (2) 4 (3)G. derjavini

(A] 132259) /(A]001840)3 7 (4) 30 (6)G. gasterostei

(AF328867)1 /(A]001841)3 1 (1) 5 (3)G. pungitii

(AF328869)1/(A]001845)3 12 9 (1)G. salaris

(AF328871)1/(A]001847)" 0 1

(AF328871)1/(Z72477)' 1 0

M. S. Ziçtara and others

obvious that some of the variation must be due tomisinterpretation, e.g. the deletions or insertions.They were never observed within the same species inthe subset including our samples and tbose ofCunningham (1997) (A]132259. Z72477, AJl32260).Similarly, unique substations in regions with severalunidentified nucleotides were considered unreliable,for example in G. derjavini (A]001840). Our finalconclusion was that aU the Gen Bank sequences withan aberrant 5.8S genotype were judged less reliableand therefore excIuded from further phylogeneticanalysis.

DNA dot matrix of the lTS sequences

Since tbe spacer sequences are known to be highlyvariabie among different Gyrodactylus species (Cabieet al. 1999; Zi~tara 2000), problems arise during dataanalysis, especiaUy with regard to sequence align-ment. A dotplot is a graphical representation of theregions of similarity between 2 sequences and hencevisualizes the useful sections for further data analy-sis. Fig. 1 shows a dotplot of 2 sequences of speciesbelonging to different subgenera (a) and the same (b)subgenus. The lTS region consists of an alternationof highly variabie and more conservative regions(Fig. lA). The most conserved region is under-standably the 5.8S gene and the 3' end of ITS2 thatfolds in a long unbranched loop in the secondarystructure of aU subgenera. The second most con~served regions are the 3' end of ITS1 and 5' end ofITS2. There is a shift in the middle of ITS2 whichpoints to an insertion or deletion event. This regioncorresponds to the second loop th at is the mostvariabie one in the secondary structure of bothspecies. Other fragments of ITS1 near the 5' end arealmost impossible to align unambiguously due tolarge insertions/deletions and repeats (Fig. lA).However, when we deal with species belonging tothe same subgenus, the proportion of difficultalignable regions decreases significantly (Fig. 1B).We therefore pro pose a novel approach by preparinga hierarchical phylogeny. First. the sequences areanalysed by the 5.8S part, which permits a sub-division into subgenera. Subsequently they could bedivided in species-groups using the combined 5.8Sand ITS2 sequences. Finally, for increasing therCllolution up to species level, IT81 and IT82 can beused. These fragments might even incIude intra-specific variation between geographicaUy separatedpopulations (Zi~tara et al. 2000).

A deep division based on the 5.8S rDNA

The deepest division can be recognized using the5.88 sequences. Each subgenus is characterized bydiagnostic sequence features from this highly con-served gene. The 5 species of subgenus Limno-

44

nephrotus and tbe 2 species of subgenus Para-nephrotus respectively, each shared a unique genesequence. The 2 species of Mesonephrotus differed in1 nucleotide (Appendix Fig. Al). The overaUvariation of the 5.88 gene among the differentGyrodactylus subgenera varied from 0,6 to 6'4%.

To compare this variation with published data, aphylogenetic reconstruction with other Platy-helminthes was prepared. Sequences of Platy-helminthes 5.8S rDNA were extracted fromGenBank and aligned for phylogenetic analysis.Identical sequences were reduced to a single rep-resentative one. As such, 5 species of Echinostomawere reduced to 2 sequences, one for the 37-coUar-spined group (E. caproni, E. liei, E. paraensei, E.revolutum and E. trivolvis) and the other for the 28-coUar-spined group (E. hortense). Four species ofSchistosoma were also reduced to 2 sequencesalthough a deletion in position 141 may be anartefact because aU Digenea have a double C inposition 140-141 except S. mansoni. The final 5.8SrRNA alignment incIuding 12 sequences was 157 bplong i the sponge Hymeniacidon sanguinea was usedas outgroup. Only 1 gap was needed to align thePlatyhelminthes sequences with the sponge, whichproves that the 5.8S rRNA gene is highly conservedin length. Appendix Fig. Al shows the overaU vari-ation within the 5.8S rDNA of the Platyhelminthesstudied here. The base composition for aUsequenceswas compared with PUZZLE (Strimmer & VonHaeseler, 1996) and revealed a significant differencefor the turbeUarian Schmidtea mediterranea. There-

fore this species was excluded from further analysis,which left the more distant sponge H. sanguinea asthe only possible outgroup. Modeltest assigned theSYM + r. model (submodel of the general-time-reversible model) with gamma shape parameter =0,99 and equal base frequencies as the most suitedfor the 5.8S data. Tree topologies constructed withML and MP are generaUy in agreement, althoughML generated lower bootst rap support (Fig. 2).Digenean and monogenean sequences are clearlyseparated from each other. The position of D.symmetrus remains unclear and is represented by apolytomy, The division in the genus G,'rodactylur ismore pronounced in comparison with the othergenera, where branch lengths are considerablyshorter. 'l'he .most advanced' subgenera Para-nephrotus and Limnonephrotus arc strongly clustcrcdand the sub genus Metanephrotus (represented by G.branchicus) is found intermediately, grouping ratherweakly with both Mesonephrotus genotypes. Theoutgroup H. sanguinea appeared too distant and thegene considered too short for inferring the relativeposition among the gyrodactylid subgenera. Thesubgenera Limnonephrotus and Paranephrotusbranched off earlier than the sub genus Mesonephrotus(Fig. 2). When H. sanguinea was replaced with 1 ofthe digenean sequences as outgroup, the opposite

Molecular phylogeny of Gyrodactylus 45

0-05

Fig. 2. Maximum likelihood phylogram constructed with 5.88 rDNA sequences of 11 taxa belonging to tbePlatyhelminthes, using tbe SYM+r. model with gamma shape parameter = 0'99, Ln L = -598'53, bootstrap support(percentage of 500 replicates, branch and bound algorithm) shown below branches. Above branches bootstrap support(percentage of 1000 replicates) of parsimony analysis based on 42 parsimony informative characters, length = 65; Cl= 0'892; RI = 0,950 (PAUp., Swofford et al. 1999). Echinostoma 1 - 37-collar-spined E. caproni, E. liei, E.

paraensei, E. revolutum, E. trivolvis; Echinostoma 2 - 28-collar-spined E. hortense;Schistosoma 1 - S. haematobium,S.intercalatum. S. mattheei: Schistosoma 2 - S. mansoni; Mesonephrotus 1 (G. arcuatus); Mesonephrotus 2 (G. sp. 1):Metanephrotus (G. branchicus); Limnonephrotus (G. gasterostei, G. pungitii, G. salaris, G. truttae); Paranephrotus (G.micropsi, G. rugiensis).

Fig. 3. Parasimony consensus tree of the 5.8S and ITS2 dataset, based on 190 parsimony informative characters withbootstrap support (percentage of 1000 replicates, branch and bound algorithm) shown above branches: length = 1017:Cl = 0'789; RI =0,832. Bootstrap support (percentage of 100 replicates, branch and bound algorithm) for themaximum likelihood method is given below branches (TVM+r., gamma = 0'48, Ln L = -2511'6). G. arcuatus(Mesonephrotus): G. sp. 1 (Mesonephrotus 2); G. branchicus (Metanephrotus); G. gasterostei, G. pungitii, G. salaris, G.truttae (Limnonephrotus); G. micropsi, G. rugiensis (Paranephrotus).

was found. In that case the resulting phylogeny is inagreement with the polygeny proposed by Malmberg(1998) based on the protonephridial system: theprimitive subgenus Mesonephrotus may have givenri se to the G. metanephrotus system, in turn givingri se to the most advanced subgenera Limnonephrotusand Paranephrotus (Fig. 3).

A shallow division based on lTSJ and lTS2 rDNA

To resolve further the division of the Gyrodactylussubgenera, we used the 5.88 gene in combinationwith the more variabie IT82 rDNA. Although theIT82 is quite variabie among subgenera, the align-ment can be checked by eye considering the presence

G. Mesonephfotus1

G. Mesonephfotus2

100 I G. Metanephrotus-68.

!!2 G.UmnonephrotuS86 L- G.Paranephrotus

.!!2.I Schistosoma195 I Schistosoma2

86 D.SymmetJus

72r Echilostoma1

62L-- Echinostoma2

H.SangLinea

.- I80

10085

100100

100 100 I100 100

100

GtnJtae

Gpitii

GderjaW1i I G.Urmonephrotus

G8a1aril

Ggasterostei

Gmicropli I G. ParanephrotusGrugiensis

GbranchicusI

G. Metanephrotus

Garcuatus

Gap1I G. Mesonephrotus

M. S. Ziftara and others

Table 5. Calculated values of the 6 possible base changes by PAUp. (Swofford et al. 1999)

46

AG TC

lTS 1lTS2

3-23Hl

5-463-47

(:) I52

~~

Ggasterostei

Fig. 4. Maximum likelihood tree for lTS2 data of the 5 Limnonephrotus species. Bootstrap support above branches forthe ML analysis (Ln L = -1189'6), under the branches in parentheses for the Paralinear/Logdet distance basedmethod, without parentheses for the MP analysis (Iength = 76; Cl = 0'74; Rl =0'44, PAUp., Swofford et al. 1999).

of known monophyla (here subgenera) and theposition of structural elements in the secondarystructure (Schulenberg, Englisch & Wägele, 1999).The folding pattern consists of 4 main domains forall species studied and is in agreement with thestructure proposed by Cunningham, Aliesky &Collins (2000). However, the domains Band C werethe least sta bIe as they contained the most dele tionsand insertions. Domain A and D were very consistentand remained almost invariant in shape, length andprimary sequence.

Comparison of the different models of evolutionrevealed that the TVM + r4 model (submodel of thegeneral-time-reversible model) with gamma shapeparameter = 0'48 as the most suited for the 5.8S andITS2 data. The molecular clock was not enforcedsince PUZZLE showed rejection of the molecularclock hypothesis. PUZZLE showed a stronglydeviating base composition for the sub genera Meso-nephrotus and Paranephrotus. Therefore, LogDet/paralinear di stances were calculated and the treeconstructed with neighbor-joining generated thesame topology as the ML tree. The resulting tree ispresented in Fig. 3. The tree is rooted with thesubgenus Mesonephrotus as suggested by the 5.88data (see above) and proposed by Malmberg (1998).The different subgenera are clearly separated, andthe overall tree topology is consistent regardless ofthe tree-building method used. The division in theLimnonephrotus subgenus, however, remains un-resolved. The unweighted and transvers ion par-simony, the Logdet and ML analysis all supportedthe clustering of G. truttae and G. pungitii (> 81 %BP, Fig. 3), in contrast to the weighted parsimonywhere only G. gasterostei clustered apart (66% HP,not shown). When treating gaps as a fifth character,the parsimony informative sites increased from 188to 229 without changing the topology. Differences inpairwise LogDet distances between the subgeneraare very high (up to 36 %). However, plotting

transitions and transversipns against divergence ofthe 5.8S and ITS2 sequences (DAMBE, 4.0.24)showed no sign of saturation.

Aiming to increase the resolution, the 5 species ofLimnonephrotus were aligned separately and used forphylogenetic analysis. ITS1, 5.8S rDNA and ITS2alignments were 683 bp, 157 bp and 430 bp long,respectively. Most gaps were within the morevariabIe ITSI region. All gaps longer than 1nucleotide, all unreliable fragments and ambiguitieswere rejected from phylogenetic analysis. Finallengths of the fragments used were 473 bp and428 bp long for ITSI and ITS2, respectively. AH thevariation of the regions is given in the Appendix,Figs A2 and A3. Neighbour-joining, maximum par-simony and maximum likelihood trees weregenerated for separate and combined spacers. Itturned out that both spacers and the various tree-building methods generated different topologies.When looking at the distribution of the various typesof base changes (TabIe 5), it turns out that lTS 1 andITS2 have a clearly different frequency. Theresulting ti/tv ratio is 0'67 and 1'44 for ITSI andITS2, .respectively. However, at-test between theobserved saturation index and the expected valueassuming full saturation (DAM BE 4.0.24) showsthat the lTS 1 sequences are not saturated. TheGC % is for both spacers from 41 to 48 %, theestimated alpha amounts to 0'21 and 0.01 for ITSIand lTS2, respectively.

Fig. 4 shows the ML tree based on ITS2, withgamma distributed rates (alpha = 0'01) using theGTR model. In the weighted parsimony (ti/tv = 1:

5), only G. salaris and G. derjavini cluster togetherwith 67 % bootstrap value, while the ML anddistance based topologies are more resolved bygrouping subsequendy G. pungitii and G. truttae.The topology generated by the lTS 1 conflicts withFig. 4 in the sense that G. salaris and G. gasterosteicluster together.

AT AC TG CG Ts Tv Ts/Tv

7'51 3'38 1 1.26 8'69 13'15 0,67Hl 0'55 1 0'28 4,58 3-19 H4

GIJU!ae

Gpungiti

Gdeijavii

Gsalaris

Molecular phylogeny of Gyrodactylus

DISCUSSION

This paper adds new insights to a molecularphylogeny of Gyrodactylus as presented by Cable etal. (1999). Representatives from 4 out of 6 subgenera,as defined by Malmberg (1970), are included in thegenet ic analysis based on the lTS rDNA region. The5.88 gene was found to be the most conserved partof the IT8 region, but still phylogenetically in-formative among subgenera. We demonstrated thatamong 10 species, each Gyrodactylus subgenuspossesses a unique sequence of the 5.88 gene.Therefore this region can be used as a tooI for a firstdivision of Gyrodactylus into subgenera. To godeeper into the phylogeny of Gyrodactylus, morevariabie regions like ITS2 can be used. Thecombination of 5.88 and IT82 provided a valuabletooI to separate to species level, although theLimnonephrotus subgenus remains partly unresolved.lTS 1 is the most variabie region consisting of manyrepeats and insertions/deletions. Therefore, phylo-genetic analysis based on lTS 1 should be restrictedto the subgenus level.

Although it is expected that a short and conservedsequence like the 5.88 rDNA gene yields lowphylogenetic information (Joffe et al. 1995; Page &Holmes, 1998), the coding regions of the rRNAtranscription unit have been extensively used toinvestigate phylogenetic relationships from phylumto genus level. According to Hershkovitz & Lewis(1996), 5.88 contains considerable phylogeneticinformation, particularly with respect to deep basalbranches. The 5.88 sequence variation found in thegenus Gyrodactylus (up to 6'4%) is higher thanintrageneric differences reported for tropicalcnidarians (up to 2'6 %) and even higher than thevariation found between nematode families and

superfamilies (up to 5'2 %) (Chen, Willis & Miller,1996; Chilton, Gasser & Beveridge, 1997; Zhu,Gasser & Chilton, 1998). This is also visualized inthe maximum likelihood phylogram based on 5.88sequences of 11 Platyhelminthes where branchesleading to the Gyrodactylus subgenera are relativelylonger than those separating digenean genera. MLanalysis clearly separated the Monogenea and theDigenea. However, the correct polarization of thegyrodactylid clade could only be inferred by usingless distant digeneans as outgroup.

The division generated by the 5.88 data becomesmore pronounced when IT82 is added to theanalysis. The various tree-building methods gen-erate the same topology, however, the division oftheLimnonephrotus subgenus remains partly unresolved.The overall tree based on 5.88 and IT82 is rootedwith the presumably oldest subgenus according toMalmberg (1970,1998) and as suggested by the 5.88analysis. IC Mesonephrotus is indeed the oldestsubgenus, then Metanephrotus adopts an intermedi.ate position while Paranephrotus and Limnonephrotus

47

appear to be the most recent subgenera. This is inagreement with the phylogeny based on the proto-nephridial system (Malmberg, 1998). At the mo-ment, more closely related outgroups are neededbefore a final conclusion can be made. It would beinteresting to use Macrogyrodactylus polypteri asoutgroup to test Malmberg's (1998) phylogeny.

The overall variation of IT8 within the genus isvery high, especially the IT81 region of the Limno-nephrotus group that consists of any array of repeatsnear the 5' end. This is also reported for ITSIsequences of Digenea (Schulenberg et al. 1999; vanHerwerden, Blair & Agatsuma, 1998, 1999). Theti/tv ratio lies below I, which could indicate that theITSI sequences are saturated, but this is rejected byat-test between the observed saturation index and

the expected value assuming full saturation(DAMBE 4.0.24). The 3' end is less influenced byinsertion/deletion repeats, and can be used to inferrelationships within a subgenus. The IT82sequences, however, have a ti/tv ratio of about 1.31and the overall variation is thus lower. This region ismore easily aligned, and can even be used acrosssubgenera. The big difference in among-site rateheterogeneity, base composition and frequency ofthe various types of base change could indicate thatIT81 and IT82 are subjected to different sub-stitution pressures (Carranza, Baguna & Riutort,1999). Another important characteristic of the IT8region is the evolutionary rate across sites. A likeli-

. hood ratio test (Modeltest) showed that the likelihoodof the tree with gamma rates is significantly betterthan the likelihood of the tree without gamma rates.Consequently, for performing a reliable analysisITSI and 2 should be analysed separately and thegamma rates have to be taken into account.

The overall variation within Gyrodactylus isalmost impossible to evaluate due to the highlyunreliable alignment when all species are pooled.Nevertheless, variation as estimated for the sub-genera Limnonephrotus (23 %), Mesonephrotus (13 %)and Paranephrotus (17 %), already reaches the upperlimits repDrted for the most related genera, whichwas already suggested by the differentiation at the5.88 level (see abo"Ve). lTS variation in relatedgroups is presented in Table 6. These findingssupport the idea that Gyrodactylus species are mucholder and genetically more differentiated than mightbe deduced from their morphological similarity. Animportant trait of gyrodactylids is progenesis, whichmight have had a tendency towards uniformizationof the body plan and associated organs. An alterna-tive explanation is that the substitution rate isunusually high due to the special combination ofclonal and sexual selection with cross-insemination.However, it is difficult to substantiate such anexplanation. From the observed level of variation, itcan be expected that the present genus Gyrodactyluscontains species from a much higher taxonomie level

M. S. Zi~tara and otlu!rs

Table 6. Variation in lTS region

(1) Chen, Willis & Miller (1996); (2) Newtonetal. (1998);(3) Bowles, Blair & McManus (1995); (4) Morgan & Blair(1995); (5) Grabda-Kozubska et al. (1998); (6) Adlard etal. (1993); (7) Jousson et al. (1998); (8) Audebert, Durette-Dessett & Chilton (2000); (9) Bowles et al. (1993).

than species grouped in thc Schistosoma or Echino-stoma genera. Conventional taxonomy groups taxa,which are not equivalent in agc, diversity, disparity,or any other consistent property of biology orevolutionary histories. In other words, any taxo-nomie rank such as for example 'genus' may not beequivalent across taxa and therefore hamper com-parative evolutionary studies (Avise & Johns, 1999).

Although very low, intraspecific variation isobserved in the lTS region of G. sp. 1, G. gasterosteiand G. pungitii. This has also been reported for G.arcuatus and G. branchicus (Ziçtara et al. 2000) but,in contrast, Cabie et al. (1999) reported nadifferences in ITS2 sequences of G. kobayashii fromthc UK and Australia.

However, no intra-individual variation in length isobserved in the lTS region of the Gyrodactylusspecies studied here. It is reported for lTSloftropical cnidarians (Chen et al. 1996), Paragonimuswestermani (Digenea) (van Herwerden et al. 1999),2species of Trichostrongylus and Nematodirus battus(Nematoda) (Hoste et al. 1995). The intraindividualvariation of the latter was sometimes greater than theinterspecific variation.

Since primary sequence similarity appeared to beassociated with secondary structure conservation, itis suggested th at similarity is due. to identity bydescent and not chance (Schulenberg et al. 1999). Allspecies studied share a folding pattern in which 4main domains can be identified. The structure of theITS2 sequence of G. salaris is identical with thatpresented by Cunningham et al. (2000).

When focusing on tbe Limnonephrotus group,different results we re generated according to theregion and tree-building method used. It seems thatboth spacers are subjected to different selectionpressures. The unresolved tree constructed on theITS2 data may be ascribed to a radiation event. Thisradiation can be the outcome of speciation by hostswitching, which refiects an ecological rather thanphylogenetic host specificity since the hosts occur in

48

the same habitat rather than being dosely related.According to Harris (1985) this group consistslargely of sibling species, still undergoing speciation.This study confirms the observation of Cable et al.(1999) that molecular studies based on tbe lTS regioncannot separate G. salaris from the other 4 Limno-nephrotus species (representatives of G. wagenerispecies-group) as suggested by Malmberg (1993).

If a molecular dock is applied to the ITS2 data,evolving at 0'3-0'7 %/Myr (Despres et al. 1992), thespeciation event of the Limnonephrotus species tookplace about 10 Myr BP. This is after the hostsspeciated. However, if the same rate is applied to allspecies, G., rugiensis and G. arcuatus (41 %, un-corrected p-distance) for example would havespeciated around 80 Myr BP (assuming a rate of0'5 % Myr BP). Here we arrive at a point where themolecular dock should be treated with caution.When dealing with all species together, the moleculardock hypothesis is rejected, but it is accepted whendealing within each subgenus. More analysis isneeded before hypothesizing about differences inevolutionary rate between subgenera.

Cable et al. (1999) presented a first phylogeny ofGyrodactylus based on lTS 1 and combined 5.8SrDNA with ITS2 sequences. They conduded thatthe 5.8S rRNA gene also was phylogeneticallyuninformative; 11 species were separated into 2groups based on combined 5.8S and ITS2sequences. Here, their division is supported whereLimnonephrotus and Paranephrotus have long ITSIsequences (599 to 654 bp) while Mesonephrotus andMetanephrotus have dearly shorter sequences (364 to392 bp and 372 bp, respectively). The differencesbetween our conclusions and those of Cable et al.

(1999) seem related to the quality of the 5.8S rDNAsequence. Cable et al. (1999) mentioned that their5.8S rDNA and ITS2 sequences were subject tosequencing errors, which resulted in the inclusion ofambiguous or unidentified bases. This might be themost straightforward explanation why aU published5.8S rDNA sequences of G. derjavini differed by30 % although they should be identical. It is obviousthat when using short and highly conserved regionswith a low number of variabie sites, the sequenceshave to be of tbe highest quality.

In conclusion, we suggest a hierarchical approachto elucidating tbe phylogeny of the genus Gyro-dactylus based on the lTS region. The conservative5.8S gene is proven to be phylogenetically in-formative and it may even be used as an aid to detectthe (sub)generic position of difficult species orperhaps even as a tooI to define a (sub)genus. Assuch, there is a concordance between the 5.88 geneand the excretory system used by Malmberg (1970)as a diagnostic character in designating the subgenusstatus. The molecular phylogeny for 10 species is atfirst sight in agreement with the morphologicalphylogeny presented by Malmberg (1998). Accord-

ITS1 ITS2 lTS

Cnidaria1 5-15 %Cooperiaz - 2-5%Echinococcus3 1-18 %Echinostoma - - 5-19 %Echinostomatidae3 11%Fasciola' - 2-13 %Mesometridae 7 7-19% 3-15%Nematodirus8 2-17%Schistosoma' - 5-11 %

Molecular phylogeny of Gyrodactylus

ing to Milinkovitch et al. (2001), covariation betweena priori morphological/physiological designationsand a minimum of 1 molecular character is asufficient condition for biological species recog-nition.

However, when moving to species level, there is adiscrepancy between the morphological and mol-ecular variation. Whereas morphological variation,expressed in shape and size of the attachmentapparatus, is very low, molecular variation,expressed by variation in the lTS 1 and ITS2regions, is very high. This can be attributed to thefast evolving nature of the lTS region, or to the factthat this genus is eonstituted of groups of a highertaxonomie level than previously reeognized. Byincluding different tests and comparing different treebuilding methods, we tried to overcome the pitfallsof phylogenetie analysis. However, exp"loringanother genetic marker and including more speciesshould shed more light on this intriguing issue.

M. S. Zi~tara did this research as a Post-doctoral Fellow atthe K. U. Leuven funded by the Belgian Ministry ofScientific, Technical and Cultural Affairs, and as a Post-doctoral Fellow of the Finnish Academy at the Universityof Oulu, Finland. T. H. was funded by the lnstitute ofScientific and Technological research (IWT-Vlaanderen).F. V. is a research associate of the Fund for ScientificResearch (FWO- Vlaanderen). This project was funded bythe Belgian Ministry of Scientific, Technical and CulturalAffairs, program MM/DD/42 'Sustainable managementof the North Sea' and the Finnish Aeademy, grant no.63797. We thank A. Arndt, G. Geets, L. Kvist, L. vanHerwerden and espeeially G. Malmberg for helpfulinsights.

REFERENCES

ADLARD,R. D., BARKER,S. C., BLAIN,D. & CRIBB,T. H.(1993). Comparison of the seeond internal transcribedspacer (ribosomal DNA) from populations and speciesof Faseiolidae (Digenea). International Journal forParasitology 23, 423-425.

APPLBBY,c. (1996). Variability of the opisthaptoral hardparts of Gyrodactylus caLlariatis Malmberg, 1957(Monogenea: Gyrodaetylidae) from Atlantic eodGadus morhua L. in the Oslo Fjord, Norway.Systematic Parasitology 33, 199-207.

AUDEBBRT,F., DUIlliTTE-DESSI!1T,M. &CHII.TON,N. (2000).lnternal transcribed spaeer rDNA can be used to inferthe phylogenetie relationships of species within thegenus Nematodirus (Nematoda: Molineoidea).International Journal for Parasitology 30, 187-191.

AVISE,J. C. &JOHNS,c. c. (1999). Proposal for astandardized temporal scheme of biologicalclassification for extant species. Proceedings of theNational Academy of Sciences, USA 96, 7358-7363.

BOWLES,J., HOPE, M., TIU, M. W., LIU, X. & MCMANUS,D. P. (1993). Nuclear and mitochondrial markershighly conserved between Chinese and PhilippineSchistosoma japonicum. Acta Tropica 55, 217-229.

BOWLIiS,J., BLAJR,D. & MCMANUS,P. P. (1995). Amolecular phylogeny of the genus Echinococcus.Parasitology 110, 317-328.

49

DROOKS,D. 11.&MCLI!NNAN,D. A. (1993). Parascript.Parasites and the Language of Evolution. Smithsonianlnstitution Press, Washington and London.

CABLE,J., HARRIS,P. D., TlNSI.EY,R. C. & LAZARUS,C. M.(1999). Phylogenetic analysis of Gyrodactylus spp.(Platyhelminthes: Monogenea) using ribosomal DNAsequences. Canadian Journalof Zoology 77,1439-1449.

CARRANZA,S., BAGUNA,J. &RIUTORT,M. (1999). Originand evolution of paralogous rRNA gene clusterswithin the flatworm family Dugesiidae(Platyhelminthes, Tricladida). Journalof MolecularEfJolution 49, 250-259.

CHEN,C. A., WI1.I.s, B. L. & MILLER,D. J. (1996).Systematic relationships between tropicalCorallimorpharians (Cnidaria: Anthozoa:Corallimorpharia): Utility of the 5.8S and internaltrans cri bed spaeer (lTS) regions of the rRNAtranscription unit. Bulletin of Marine Science 59,196-208.

CHII.TON,N. 11.,GASSEIt,R. B. & BEVERIDGB,I. (1997).Phylogenetic relationships of Australian strongyloidnematodes inferred from ribosomal DNA sequencedata. International Journal for Parasitology 27,1481-1494.

CUNNINGHAM,c. o. (1997). Species variation within theinternal transcribed spacer (lTS) region ofGyrodactylus (Monogenea: Gyrodactylidae) ribosomalRNA genes. Journalof Parasitology 83,215-219.

CUNNINGHAM,c. 0., ALiESKY,H. &COLLlNS,C. M. (2000).Sequence and secondary structure variation in theGyrodactylus (Platyhelminthes, Monogenea) ribosomalRNA gene array. Journalof Parasitology 86, 567-576.

DESPRES, 1.., IMBERT-ESTABLIIT, D., COMBES, C. &

IIONHOMME,F. (1992). Molecular evidence linkinghominid evolution to recent radiation of Schistosomes(Platyhelminthes: Trematoda). MolecularPhylogenetics and Evolution 4, 295-304.

GEE'fS,A., APPLUBY,C. &OLLEVIER,F. (1999). Host-dependent and seasonal variation in opisthaptoral hardparts of Gyrodactylus cf. arcuatus from threePotnatoschistus spp. and G. arcuatus from Gasterosteusaculeatus: a multivariate approach. Parasitology 119,27-40.

GNABDA-KOZUBSKA, B., BORSUK, P., LASKOWSKJ, Z. 80;MONÉ,

H. (1998). A phylogenetic analysis of trematodes ofthe genus Echinoparyphium and related genera basedon sequencing of internal transcribed spacer region ofrDNA. Acta Parasitologica 43, 116-121.

HARRIS,P. D. (1985). Species of Gyrodactylus vonNordmann, 1832 (Monogenea: Gyrodactylidae) fromfreshwater fishes in southern England, with adescription of Gyrodactylus rogatensis sp. nov. from.the bullhead Cuttus gobio L. Journalof NaturalHistury 19, 791-809.

HANIUS,P. D. (1993). Interactions between reproductionand population biology in gyrodactylid monogeneans- a review. Bulletin Français de Pêche et de laPisciculture 328, 47-65.

HBRSHKOVITZ,M. A. & LEWIS,L. A. (1996). Deep-Ieveldiagnostic value of the rDNA-ITS region. MolecularBiology and Evolution 13, 1276-1295.

HOSTB,H., CHJLTON,N. B., GASSBR,R. B. & BBVERIDGB,I.(1995). Differences in the 2nd intemal transcribed

M. S. Ziçtara and others

spacer (ribosomal DNA) between 5 species ofTrichostrongylus (Nematoda, Trichostrongylidae).International Jrmrnalfor Parasitology 25,75-80.

JOFFE,B. I., VALIEJOROMAN,K. M., BIRSTEIN,V. Y. A. '"TROITSKY,A. v. (1995). 5S rRNA sequences of 12species of flatworms: implications for the phylogenyof the Platyhelminthes. Hydrobiologia 305, 37-43.

JOHNSEN,B. O. '" JENSEN,A.J. (1991). The Gyrodactylusstory in Norway. Aquaculture 98, 289-302.

JOUSSON,0., BARTOLl,P., ZANINETTI,L. & PAWLOWSKI,J.(1998). Use of the lTS rDNA for elucidation of somelife-cycles of Mesometridae (Trematoda, Digenea).International Journalfor Parasitology 28, 1403-1411.

JUNIBR,T. '" PAGNI,M. (2000). Dodet: diagonal plots in aweb browser. Bioinformatics 16, 178-179.

KANE,R. A. &.ROLLINSON,D. (1994). Repetitive sequencein the ribosomal DNA internaJ transcribed spacer ofSchistosoma haematobium, Schistosoma inttrcalatumand Schistosoma mattheei. Molecular and Biochemical

Parasitowgy 63, 153-156.KEARN,G. c. (1994). Evolutionary expansion of the

Monogenea. International Journal for Parasitology 24,1227-1271.

LUTON,K., WALKER,D. &BLAIR,D. (1992). Comparison ofribosomal internaJ transcribed spacers from twocongeneric species of ftukes (Platyhelminthes:Trematoda: Digenea). Molecular and BiochemicalParasitology 56, 323-328.

MALMBERG,G. (1970). The excretory systems and themarginal hooks as a basis for the systematics ofGyrodactylus (Trematoda, Monogenea). Arkiv fiirZoologi 23, 1-235.

MALMBERG,G. (1993). Gyrodactylidae and gyrodactylosi~of Salmonidae. Bulletin Français de la PêchePisciculture 328, 5-46.

MALMBERG,G. (1998). On the evolution within the familyGyrodactylidae (Monogenea). International Journal forParasitology 28, 1625-1635.

MILINKOVITCH,M. C., LEDUC,R., TlEDEMANN,R. & DIZON,A. (2001). Applications of molecular data in cetaceantaxonomy and population genetics with specialemphasis on defining species boundaries. In MarineMammals : Biology and ConseTtlation (ed. Evans,P. G. H. & Raga, J. A.), pp. 00-00. Kluwer AcademiePress, Amsterdam.

MORCAN,J. A. T. &BLAIR,D. (1995). Nuclear rDNA lTSsequence variation in the trematode genusEchinostoma: an aid to establishing relationshipswithin the 37-coUar-spine group. Parasitology 111,609-615.

NEWTON,L. A., CHILTON,N. B., BEVERIDCI!,I. &GASSER,R. B. (1998). Genetic evidence indicating th at Cooperiasurnabada and Cooperia oncophora are one species.International Jrmrnal for Parasitology 28, 331-336.

PAGB,R. D. M. &.CHARI.BSTON,M. A. (1997). From gene toorganismal phylogeny: reconciled trees and the genetree/species tree problem. Molecular Phylogenetics andEvolution 7, 231-240.

PAGE,R. D. M. &.HAFNER,M. s. (1996). Molecularphylogenies and host-parasite cospeciation: gophersand lice a model system. In New Uses for NewPhylogenies (ed. Harvey, P. H., Brown, A. J. L.,Smith, J. M. & Nee, S.), pp. 255-270. OxfordUniversity Press, New York.

50

PAGE,R. D. M. &.HOLMES,B. c. (1998). Molecular

Evolution. A Phylogenetic Approach. Blackwell ScienceLtd, Oxford.

PAGEL,M. (1999). Inferring the historical pattems ofbiologica I evolution. Nature, London 401, 877--884.

POSADA,D. '" CRANDALL,K. A. (1998). MODEL TEST:testing the model of DNA substitution. Bioinformatics14, 817-818.

SCHULENBURG,J. H. G. v. d., BNGI.ISCH,U. '" WÄGELE,J. H.(1999). Evolution of ITSI rDNA in the Digenea(Platyhelminthes: Trematoda): 3' end sequenceconservation and its phylogenetic utility. JournalofMolecular Evolution 48, 2-12.

SHINN,A. P., DESCLI!RS,S., GIBSON,D. I. '" SOMMBRVILLE,c. (1996). Multivariate analysis of morphometricalfeatures from Gyrodactylus spp. (Monogenea)parasitizing British saJmonids: Light microscopebased studies. Systematic Parasitowgy 33, 115-126.

SUOMALAINEN,B., SAURA,A. &.LOKKt,J. (1987). Cytologyand Evolution in Parthenogenesis. CRC Press, BocaRaton, USA.

STRIMMER,K. '" VONHAESBLER,A. (1996). Quartetpuzzling: a quartet maximum likelihood method forreconstructing tree topologies. Molecular Biowgy andEvolution 13, 964-969.

SWOFFORD,D. L. (1999). PAUp.: Phylogenetic AnalysisUsing Parsimony (and other methods), Version 4.0.Sinauer Associated, Massachussetts.

TAMURA,K. '" NEl, M. (1993). Estimation of the numberof nucleotide substitutions in the control region ofmitochondrial DNA in humans and chimpanzees.Molecular Biology and Evolution 10, 512-526.

THOMPSON,J. D., HIGGINS,D. G. &.GIBSON,T. J. (1994).CLUSTAL W: improving the sensitivity ofprogressive multiple sequence aJignment throughsequence weighting positions-specific gap penaJtiesand weight matrix choice. Nucleic Acid Research 22,4673-4680.

VANHERWERDEN,L., BLAIR,D. '" AGATSUMA,T. (1998).Intra- and inter-speci/ic variation in nuclear ribosomaJinternal transcribed spacer I of the Schistosomajaponicum species complex. Parasitology 116, 311-317.

VANHERWERDEN,L., BLAIR,D. '" AGATSUMA,T. (1999).Genetic diversity in parthenogenetic triploidParagonimus westermani. International Journal forParasitowgy 29, 1477-1482.

WILLIAMS,H. &.JONES,A. (1994). Parasitic Worms ofFish. Taylor and Francis, London.

XIA,x. (2000). DAMBE: Data Analysis in Molecular

Biology and Evolution. Department of Ecology andBiodiversity, University of Hong Kong.

ZIP'ARA, M. s., ARNDT,A., CI!ETS,A., HELLEIlfANS,B. '"VOLCKAERT,F. A. M. (2000). The nuclear rDNA regionof Gyrodactylus arcuatus and G. branchicus

(Monogenea: GyrodactyJidae). Journalof Parasitology86, 1368-1373.

ZHU,x. Q., GASSI!R,R. B. '" CHILTON,N. B. (1998).Differences in the 5.8S rDNA sequences of ascaridnematodes. International Journal for Parasitology 28,617-622.

ZUKER,M. (1989). Computer prediction of RNAstructure. Methoels in Enzymology 180, 262-268.

APPENDIX1 1 1 1 1 1 1 1 1 1 1 1 1 111 1 1 111 1 1

1 1 1 2 3 3 3 334 4 4 5 5 566 6 677 7 7 8 8 8 8 8 9 9 9 9 9 900 1 1 1 2 2 2 2 2 2 333 3 3 3 4 4 4 4 4 5 5

367 890 1 3 9 0 1 2 3 4 4 5 7 1 4 603 4 9 0 3 4 8 0 1 5 6 9 3 5 678 9 1 928 904 5 6 7 9 0 1 3 4 690 1 2 4 560 7

Smed _ T T A A G C A G C A CGT ACT A A A G C T T G C T C A C C T T T A C ACT TAG T A T T C A ACT A Ace A T C T T C T G A

Dsym AC. G

Echl A C G

G . T G T G . C C CG. T TC. C T C T . G . C . G G _ GC C . TT C TT. A C

TC. T T C T . G G G G . G C C . TG . C C CG. T A T C TT. A CG . T G T

Ech2 AC. G . A . G . T G T G . C C CG. T . C A T C TT. A C TC. T T C G G G . G C C . TT . G

Schl A G . T G T G Tee CG. T T C TT. A C TC. T T CAT . T C . TG G G . G CG

Sch2 A G . T G T G T C C CG. T T C.T T . A C TC. T T CAT . T G G G . G C C . TG

Gmel A G T . T GA. G TA. G TC. CGT . C A C G C T TC. CGT C T C C T . C T TG. CT. G . G . T . A T C T T

Gme2 A G T . T GA. G TA. G TC. CGT . C A C G C T TC. CGT C T CC. T . T . C T TG. CT. G . G . T . A T C T T

Gmet A CC. T . T GA. G TA.' G TC. CGT . C A C G C T TC. CGT C Tee T . C T TG. CT. G . G . T . A T C T T

Glim A CC. T . T GA. G T . A G TC. CGT . C A C ACT TC. CGT C T C C ACT T G T CT. G . G . T CAT C T T

Gpar A CC. T . T GA. G TA. G TC. CGT . C . . A C ACT CC. CGT C T CC. . . . ACT T G T CT. G . G . T CAT C T T

Fig. At. Variabie sites in 5.8S rDNA of 12 Platyhelminthes taxa. Numbering starts from the first nucleotide of the 5.8S gene. (.) indicates identity to Smedsequence. Abbreviations are as follows: Smed - Schmidtea mediterrane a ; Dsym - Dotichosaccus symetrus; Ech 1 - 37-collar-spined Echinostomacaproni, E. tiei , E.paraensei,E. revolutum, E. tri volvis; Ech2 - 28-collar-spined E. hortense; Sch1 - Schistosomahaematobium,S. intercalatum, S. mattheei; Sch2 - S. mansoni;Mes1 - Mesonephrotus(G. arcuatus); Mes2 - Mesonephrotus(G. sp. 1); Met - Metanephrotus (G. branchicus); Lim - Limnonephrotus (G. gasterostei, G. pungitii, G.salaris, G. truttae); Par - Paranephrotus (G. micropsi, G. rugiensis).

~2i:>...'I:>~~..~~Q'<.,oc:>.~()-<i:"'

VI-

2 2 2 2 2 2 2 2 2 2 222 2 333 3 3 3 3 3 3 3 3 3 444 4 4 4 4 4 4 4 4

00011 134 4 4 4 4 6 9 0 2 4 4 6 6 6 6 667 9 022 2 2 2 4 4 5 6 6

1 8 9 1 4 7 9 0 1 567 024 2 3 912 4 6 7 8 0 8 612 3 4 657 146

Gder A T T C A CAT ACT A A A T T T T A A A T C T T A T A C C A A T A C A C

Ggas G G A A T TT. T . A A A G C CAT. C . A GAT C . G C.A T -

Gpun G

Gsal G

TA T G G G AG . T G TTT.GCATG

T G G. . G A . A A TGCATG

Gtru G T A. GA AG G .GCATG

Fig. A2. Varia bIe sites of the lTS! fragment within the subgenus Limnonephrotus. Numbers correspond to the firstnucleotide of tbe aligned lTS! sequence. " indicates identity to Gder sequence. Abbreviations are as follows:Gder - G. derja vini; Ggas - G. gasterosteijG. pun - G. pungitii; Gsal- G. salaris; Gtru - G. trottae.

~ 1 1 1 1 1 1 1 1 1 1 1 1 1.1 1 2 2 2 2 2 2 2 2 2 3 3 3 3 3 3 3 3 3 3 4 4

7 7 7 7 8 8 900 3 3 4 4 5 5 667 7 7 8 8 8 0 1 155 6 7 7 7 0 1 2 8 8 8 999 922

1 2 3 7 6 7 904 375 9 7 924 3 4 7 2 5 6 7 192 4 804 8 4 2 5 4 5 824 7 926

Gder C T T G C G G C T A T C T G TAG A CAT G G C TGG eTA C T A A A A A A T A T TAG

Gqas T C GAG . TAG T . TG. A T G A A G GTAG. .T T . T

GpunT . G . A TGG. T AA TA . G . G G T . G CGT

Gsal T . G . G G . A TGGT.A .A.AAG . A

Gtru T . G A AC. T . CT. A AG .AA.AATA . G . G T TT. G

Fig. A3. Varia bIe sites of ITS2 in subgenus Limnonephrotus. Numbers correspond to the first nucleotide of thealigned ITS2. (.) indicates identity to Gder sequence. Abbreviations are as follows: Gder - G. derjavinij Ggas - G.gasterostei;G. pun - G. pungitii. Gsal- G. salaris; Gtru - G. truttae.

M. S. Zit'taTa and others 52

122 3 3 344 4 555 666 6 7 7 7 7 8 8 8 9 9 9 0 0 0 0 1 1 112 3 4 4 5 5 7 8 9 9

9071792 5 948 907 8 912 3 4 4 5 9 0 180 158 044 6 0 8 3 8 453 6 1 2

Gder TC GAAT TT C - TGCAAAGATACT GGA- GTGG T CCAATAATACT C T

Gqas - TAG CA.ATT T ATACCAC A A - A. . T . C

Gpun A - T T G CAATTG.ATTC

Gsal . TG. .AAA. GAT TA C . A . G T AT C

Gtru A-A.TG.AA. G . .CAA. ATTC.A-