TAXONOMIC STATUS OF STAGNICOLA PALUSTRIS (O. F. MÜLLER, 1774) AND S. TURRICULA (HELD, 1836) (GASTROPODA: PULMONATA: LYMNAEIDAE) IN VIEW OF NEW MOLECULAR AND CHOROLOGICAL DATA

Analyses of nucleotide sequences of 5’and 3’ends of mitochondrial cytochrome oxidase subunit I (5’COI, 3’COI) and fragments of internal transcribed spacer 2 (ITS2) of nuclear rDNA gene confirmed the status of Stagnicola corvus (Gmelin), Lymnaea stagnalis L. and Ladislavella terebra (Westerlund) as separate species. The same results showed that Stagnicola palustris (O. F. Müll.) and S. turricula (Held) could also be treated as separate species, but compared to the aforementioned lymnaeids, the differences in the analysed sequences between them were much smaller, although clearly recognisable. In each case they were also larger than the differences between these molecular features of specimens from different localities of S. palustris or S. turricula. New data on the distribution of S. palustris and S. turricula in Poland showed – in contrast to the earlier reports – that their ranges overlapped. This sympatric distribution together with the small but clearly marked differences in molecular features as well as with differences in the male genitalia between S. palustris and S. turricula strongly support the view that they are separate species and not two subspecies within S. palustris. key woRds: Poland, lymnaeids, COI, ITS2, sympatric/allopatric distribution, species/subspecies status


MATERIAL AND METHODS
Snails were collected in drainage ditches, small pools, ponds, lakes, streams or small rivers, with a sieve (frequently mounted on a two-metre long boom) or a dredge, and hand-collected from plants, stones, or tree branches pulled out of the water.The list of 25 localities in Poland where specimens for molecular studies were collected is presented in Appendix 1.The lymnaeids caught in the field were kept in containers with aerated water of room temperature (21-24ºC).They were sporadically fed with lettuce leaves.
The lymnaeid species were indentified on the basis of morphological and anatomical features (Jackiewicz 1998a, 2000, GlöeR & meieR-bRook 2003, kRuGlov 2005).DNA was extracted from the foot tissue of each specimen.The rest of the body (including reproductive system) of each specimen was preserved in 75% ethyl alcohol.Total genomic DNA was extracted with DNeasy Tissue Kit (Qiagen) according to manufacturer's procedure.Amplifications of 3'-end of the cytochrome oxidase subunit I (3'COI) mitochondrial gene were performed, according to Qiagen procedure, in the reaction mixture composed of 0.2 unit of Taq polymerase, 0.5 μM of each primer, and 0.2 mM of each dNTP in a final volume of 10 μl.Amplifications of 5'-end of the same gene (5'COI) were accomplished in 25 μl volume following modified protocol prepared by Biodiversity Institute of Ontario for Consortium for the Barcode of Life (http:// barcoding.si.edu/PDF/Protocols_for_High_Volume_DNA_Barcode_Analysis.pdf).Amplifications of ITS2 sequence were carried out in a 10 μl volume with 1.25 unit of Taq polymerase, 1 μM of each primer, and 0.2 mM of each dNTP.The PCR reactions were performed with the use of the following primers: CO1PU (5'-TTTTTGGGCATCCTGAGGTTAT-3') and CO1PL (5'-TAAAGAAAGAACATAATGAAAATG-3') (bowles et al. 1992, stothaRd & Rollin son 1997) for the 3'COI, bcsmF1 5'-AAYCATAAAGA YATTGGDACWTTDTA-3' and bcsmR1 5'-TAWACYT CWGGRTGACCAAAAAAYCA-3' (PRoćków et al. 2013, the nucleotides and ambiguity codes were determined according to IUPAC) for 5'COI as well as NEWS2 5'-TGTGTCGATGAAGAACGCAG and ITS2- RIXO 5'-TTCTATGCTTAAATTCAGGGG (almeyda-aRtiGas et al. 2000) for ITS2.The PCR conditions were as follows: 3'COI -initial denaturation step of 5 min at 95ºC, followed by 30 cycles of 30 sec at 94ºC (denaturation), 30 sec at 54ºC (hybridisation), 30 sec at 72ºC (elongation), and final incubation of 5 min at 72ºC; 5'COI -3 min at 96ºC followed by 6 cycles of 1 min at 96ºC, 30 sec at 45ºC, and 1 min 15 sec at 72ºC followed by 36 cycles of 1 min at 96ºC, 30 sec at 51ºC, and 1 min 15 sec at 72ºC, and final elongation of 15 min at 72ºC; ITS2 -2 min at 94ºC followed by 35 cycles of 30 sec at 94ºC, 30 sec at 58ºC, and 30 sec at 72ºC followed by final elongation of 7 min at 72ºC.The PCR products were visualised on 1% agarose gels and sequenced on Applied Biosystems Hitachi 3130xl Genetic Analyser automated sequencer.
Full-length sequences were edited by eye using the programme BioEdit, version 7.0.5.(hall 1999) and aligned using Prank (löytynoJa & Goldman 2008) for COI, or CLUSTAL W (thomPson et al. 1994) for ITS2.The COI sequences were aligned according to the translated amino acid sequences.The ends of all sequences were trimmed.The length of the sequences after cutting were 354 bp, 558 bp, 421-447 bp for 3'COI, 5'COI and ITS2, respectively.Gaps and ambiguous positions were removed from alignments.The sequences were collapsed to hap-lotypes (3'COI and 5'COI) or consensus sequences (ITS2) prior to phylogenetic analysis.

MOLECULAR ANALYSES
Twenty seven different sequences of mitochondrial 5'COI gene fragments were deposited in GenBank (Table 1).Analysed by the NJ method, the 5'COI se-quences were arranged in a dendrogram in five clusters representing five lymnaeid species (Fig. 1).Five haplotypes (5'COI 1 -5'COI 5) were found among the specimens anatomically identified as S. palustris, three (5'COI 6 -5'COI 8) -as S. turricula, three Table 1.5'COI haplotypes found in specimens of five lymnaeid species (locality numbers according to Appendix 1, number of specimens in parentheses)
The mean values of intraspecific variation between the sequences of 5'COI gene fragments characteristic of particular species, expressed as genetic K2P distances, were 0.3% for S. corvus and L. terebra, 0.7% for S. turricula and for L. stagnalis, and 0.9% for S. palus-tris, whereas the genetic K2P distances between the species were higher (Table 2).The smallest distances between the sequences of 5'COI gene fragments distinguished S. palustris and S. turricula (mean 2.6%).The mean K2P distances between the sequences of S. palustris and the other species were 6.4%, 15.3%, and 21.1% for S. corvus, L. stagnalis, and L. terebra, respectively.Similarly, the mean distances which dis- tinguished S. turricula from S. corvus, L. stagnalis and L. terebra were 5.9%, 14.7% and 21.6%, respectively.The mean values of K2P distances between S. corvus vs. L. stagnalis and S. corvus vs. L. terebra were 14.5% and 19.5%, respectively.The mean K2P distance between the 5'COI sequences for L. stagnalis and L. terebra was 20.2%.Surprisingly, the sequences of 3'COI gene fragments did not provide resolution between species.A total of six 3'COI haplotypes were deposited in GenBank (Table 3): one (3'COI 5) was found in specimens anatomically identified as representing three different species: S. corvus, L. stagnalis and L. terebra, two other were unique for S. corvus and L. terebra (3'COI 4 and 3'COI 6, respectively) (Table 3).In general, these three sequences were very similar.Moreover, we found two unique sequences of the 3'COI gene fragment among specimens anatomically identified as S. palustris (3'COI 1 and 3'COI 2) and one characteristic of S. turricula only (3'COI 3).The dendrogram built with the use of NJ method included three groups of sequences -the first for the sequences of S. corvus, L. stagnalis and L. terebra, the second for the sequences of S. palustris and the third for the sequence of S. turricula (Fig. 2).
To confirm our identification of the species of Stagnicola s.l.we amplified their DNA to obtain sequences of the ITS2 fragment of nuclear rDNA gene.We obtained four sequences, deposited in GenBank, one for each: S. palustris, S. turricula, S. corvus and L. terebra (Table 4).The sequence identified for the specimens of S. palustris (ITS2 1) was identical (MCL distance equal to 0.0%) with the appropriate sequence AJ319620 deposited in GenBank for this species (baRGues et. al. 2001).The sequence found for S. turricula (ITS2 2) differed in one indel and one, two or  turricula), ITS2 3 with AJ319625 (S. corvus), ITS2 4 with AJ457042 (L.terebra), leaving AJ319616 as the fifth independent group (L.stagnalis).

DISCUSSION
Distinctness of L. stagnalis (L.), S. corvus (Gmel.)and L. terebra (Westerlund) was well established on the basis of the conchological, other morphological and reproductive system features (Jackiewicz 1993, 1998a, 2000, FalkneR 1995, FalkneR et al. 2001, GlöeR 2002, GlöeR & meieR-bRook 2003) as well as on the molecular features found in the nucleotide sequences of ITS1 and ITS2 fragments in rDNA gene (baRGues et al. 2001(baRGues et al. , 2003(baRGues et al. , 2006) ) and obtained with the use of RAPD technique (Rybska et al. 2000a(Rybska et al. , b, 2008)).The distinct species status of these three taxa is confirmed by the results of this study.In our analyses we used molecular markers frequently applied for taxonomic or phylogenetic studies, i.e. sequences of mitochondrial 3'COI (loaiza et al. 2013(loaiza et al. , zein-eDDine et al. 2014) and 5'COI (RemiGio & hebeRt 2003, runDell et al. 2004, Perez et al. 2005, kane et al. 2008, Falniowski et al. 2009, Falniowski & szaRowska 2011, szaRowska et al. 2013, 2014) gene fragments as well as nuclear ITS2 fragment of rDNA gene (baRGues & mas-coma 2005, källeRsJö et al. 2005, Puslednik et al. 2009, baRGues et al. 2012, loaiza et al. 2013).The nucleotide sequences of mitochondrial 5'COI gene ("barcode sequence") differentiate between the above three lymnaeids at the level of 14.0-20.8%.These genetic distances are much greater than the threshold established by hebeRt et al. (2003, 2004a, b) in whose opinion a difference exceeding 2% may be treated as the sign of separateness at the species level.Short 3'COI gene fragments are rather useless for the separation of these three species although they were useful in the study on cryptic species of Anopheles (loaiza et al. 2013) and species identification of some Bulinus (zein-eddine et al. 2013).These sequences are very similar (0.0-0.6%) for S. corvus, L. stagnalis and L. terebra.On the other hand the three species differ in their nucleotide sequences of ITS2 fragment, as was previously shown by baRGues et al. (2001,2003).
Also S. palustris (O.F.Müll.) and S. turricula (Held) are the taxa well separated from the three above species, as the genetic distances of their nucleotide sequences of the 5'-end of COI gene are between 5.6 and 22.0%.The distinctness of S. palustris/turricula was established in earlier molecular analyses (baRGues et al. 2001, 2003, 2006, Rybska et al. 2000a, 2008).However, the separateness between S. palustris and S. turricula is under debate (baRGues et al. 2011: table S1).They differ only in 3 or 2-3 nucleotides in the sequences of ITS1 (baRGues et al. 2006) or ITS2 (baRGues et al. 2001, 2003) fragments, respectively, while the difference in comparison with other representatives of Stagnicola (s.l.) are 19-42 nucleotides.Similar small differences were also found in our analyses of the ITS2 sequences of S. palustris and S. turricula (they differ in one indel only).In the RAPD fragments S. palustris and S. turricula differed at the level of 12-18% while each of these two taxa differed from the other Polish Stagnicola species at the level of 43-65% (Rybska et al. 2008).However, it must be stressed that interpopulation genetic variation within each of these species was by one order of magnitude smaller (3.1-4.4% between populations of S. palustris, 3.9-5.9%between populations of S. turricula) (Rybska et al. 2008).Our results on 5'COI sequences shed new light on the taxonomic status of S. palustris and S. turricula.The genetic distances between S. palustris and S. turricula in these sequences vary from 2.2 to 3.3%.Such differences are greater than interpopulation 5'COI differentiation (0.9% and 0.7% for S. palustris and S. turricula, respectively).Moreover, the sequences of 3'COI gene fragments of S. palustris and S. turricula are similar but distinctly different.
It should be stressed that the genetic difference between S. palustris and S. turricula in the "barcode sequence" is also greater than the above-mentioned 2% threshold of hebeRt et al. (2003, 2004a, b).Probably the low 5'COI divergence between S. palustris and S. turricula, which is much smaller than the divergences between each of these two taxa and each of the other three lymnaeid species studied in this paper, may suggest their recent origin.The conclusion that S. palustris and S. turricula are closely related is supported by the fact that the differences in the mitochondrial gene (5'COI,3'COI) are larger than those in the nuclear (ITS2) gene fragment.Mitochondrial genome evolves much faster than the nuclear one (RemiGio & hebeRt 2003).Therefore analyses of ITS2 sequences can be useless when distinguishing between closely related species which was demonstrated in the earlier studies on Trochulus hispidus species complex (kRuckenhauseR et al. 2014) and Monacha claustralis vs. M. cartusiana identification (Pieńkowska et al. 2015).On the other hand, ITS2 sequences are very useful for analyses of phylogenetic relationships between more distant taxa (baRGues & mas-coma 2005, källeRsJö et al. 2005, baRGues et al. 2012, loaiza et al. 2013).
We are aware that taxonomic decisions based on molecular features only are precarious as they are based on arbitrarily adopted rules.Therefore we tried to use also characters used in classical taxonomy which is often practiced by other researchers (e.g. szaRowska 2006, haase et al. 2007, Falniowski et al. 2009, Puslednik et al. 2009, Falniowski & szaRowska 2011, szaRowska et al. 2013, 2014).Each specimen taken for molecular studies was identified on the basis of its genital anatomy (according to Jackiewicz 1998a, 2000and GlöeR & meieR-bRook 2003).
Jackiewicz (1993, 1998a, 2000) showed distinct differences in the structure of male genitalia.The praeputium : penis sheath length ratio was 1 : 1 for S. palustris and 1 : 3 to 1 : 5 in S. turricula (see e.g.Jackiewicz 1998a: fig.68 for S. palustris and fig.70 for S. turricula).Consequently, Jackiewicz (1993, 1998a, 2000) treated S. palustris and S. turricula as two distinct species.However, GaRbaR (2001) and kRuGlov (2005) did not confirm this difference in the male genitalia between S. palustris and S. turricula.They showed several other differences in their anatomy, and placed them in two different sections of Stagnicola -S.palustris in Stagnicola s. str., S. turricula in Fenziana Servain, 1881.According to vinaRski et al. ( 2011) "S.palustris and S. turricula were obviously distinct by their morphological traits, but were shown to be almost indistinguishable with the genetic markers".The differences in male genitalia prompted baRGues et al. (2001,2003,2006) to treat them as two subspecies, S. palustris palustris and S. p. turricula.GaRbaR et al. (2004), studying karyotypes of Ukrainian representatives of Lymnaeidae, found that S. palustris and S. turricula had the same number of chromosomes (n=36) however they differed in the chromosome types (14 matecentric, 10 submetacentric, 12 subtelocentric for S. palustris vs. 12 metacentric, 8 submetacentric, 16 subtelocentric for S. turricula).Although we cannot confirm by molecular features that S. turricula from Ukraine is conspecific with S. turricula from Poland, their localities (especially those in the Bieszczady Mts, Poland) are not very far apart.
According to Jackiewicz (1988bJackiewicz ( , 1998aJackiewicz ( , 2000) ) S. palustris and S. turricula were differently distributed in Poland: the former was a northern species common in the lowland (west-northern) part of Poland while the latter lived in the south-eastern part of the country (Bieszczady Mts).Similarly in Germany -S. palustris lives rather in the northern part, S. turricula -in the Danube region of southern Germany   (1988b, 1993, 1998a, 2000) papers).Jackiewicz (1998aJackiewicz ( , 2000) ) reported the occurrence of S. turricula in Hungary and Bulgaria.The above allopatric distribution supported the subspecies status of S. palustris and S. turricula established by baRGues et al. (2001,2003,2006).However we found some localities of S. palustris in south-eastern Poland and some localities of S. turricula much farther west-and north.It should be stressed that Piechocki (1972,2000) found both species at one locality in the Niebieskie Źródła Nature Reserve near Tomaszów Mazowiecki (Central Poland), identifying individuals based on their anatomical features.We also found S. palustris near Odrzykoń where S. turricula was reported earlier (identified anatomically, see lewin & cebula 2003: fig.3).It proves that S. palustris and S. turricula occur sympatrically.They co-occur in Hungary (kilias 1992, Jackiewicz 1996), Ukraine (GaRbaR 2001, GaRbaR et al. 2004, kRuGlov 2005), Eastern Europe (nekhaev 2011) and Bulgaria (GeoRGiev & hubenov 2013).
Sympatric occurrence of two taxa excludes their subspecific status within one species.Subspecies of the same species are by definition allopatric.Therefore, taking into account the well established differences in the male genitalia and the very small but clearly recognisable differences in the molecular features, combined with the sympatric distribution, we conclude that S. palustris and S. turricula should be treated as two distinct species.
Finally, it is noteworthy that our results of the comparison of the 5'COI gene fragment sequences support FalkneR's (1995) opinion that S. corvus should be included with S. palustris and S. turricula in the genus Stagnicola Jeffreys, 1830, leaving L. stagnalis within the genus Lymnaea Lamarck, 1799.The genetic distances between L. terebra and all other species (19.0-22.0%)also suggest that it was reasonable to create a new genus Catascopia Meier-Brook et Bargues, 2002(= Ladislavella B. Dybowski, 1913) to accommodate G. occulta Jackiewicz, 1959(meieR-bRook & baRGues 2002, vinaRski 2012).

Fig. 1 .
Fig. 1.Neighbour-Joining tree based on 558-nt-long fragment of 5'COI sequences of five lymnaeid species: Stagnicola palustris, S. turricula, S. corvus, Lymnaea stagnalis and Ladislavella terebra.5'COI sequence of Planorbarius corneus (FR797858) was used as outgroup.Figures at the nodes indicate bootstrap support from 1,000 replicates.Bootstrap values below 50 not shown.Evolutionary distances computed using Kimura two-parameter method, expressed as number of base substitutions per site.Codon positions included were 1st+2nd+3rd+Noncoding.All positions containing gaps and missing data were eliminated from dataset (complete deletion option).
hoRsák et al. (2013) listed S. palustris among Czech and Slovak molluscs only, but they synonymised it with S. turricula, listing features treated by Jackiewicz (1993, 1998a) as characteristic of S. turricula both in the text and in the figures (see: hoRsák et al. 2013: p. 49, fig.9; the figure taken from Jackiewicz's

Table 3 .
3'COI haplotypes found in specimens of five lymnaeid species (locality numbers according to Appendix 1, number of specimens in parentheses) Fig. 2. Neighbour-Joining tree based on 354-nt-long fragment of 3'COI sequences of five lymnaeid species: Stagnicola palustris, S. turricula, S. corvus, Lymnaea stagnalis and Ladislavella terebra.3'COI sequence of Planorbarius corneus (AY577512) was used as outgroup.Figures at the nodes indicate bootstrap support from 1,000 replicates.Bootstrap values below 50 not shown.Calculation parameters same as for Fig. 1.

Table 4 .
ITS2 sequences found in specimens of four lymnaeid species (locality numbers according to Appendix 1, number of specimens in parentheses) Fig.3.Neighbour-Joining tree for five lymnaeid species using ITS2 fragment.Percentage of replicate trees in which the associated taxa clustered together in bootstrap test (1,000 replicates) shown next to the branches.Bootstrap values below 50 not shown.Evolutionary distances computed using Maximum Composite Likelihood method, expressed as number of base substitutions per site.All positions containing gaps and missing data were eliminated from the dataset (complete deletion option).