Contrasting reproductive strategies of triploid hybrid males in vertebrate mating systems

doi: 10.1111/jeb.12556 Contrasting reproductive strategies of triploid hybrid malesin vertebrate mating systems *Institute of Evolutionary Biology and Environmental Studies, University of Zurich, Zurich, Switzerland†Department of Zoology, Faculty of Natural Sciences, Comenius University in Bratislava, Bratislava, Slovak Republic ‡Laboratory of Fish Genetics, Institute of Animal Physiology and Genetics, Lib
echov, Czech Republic §Department of Zoology, Faculty of Science, Charles University in Prague, Prague, Czech Republic all-hybrid population; The scarcity of parthenogenetic vertebrates is often attributed to their asexual reproduction; ‘inferior' mode of clonal reproduction, which restricts them to self-repro- gamete production; duce their own genotype lineage and leaves little evolutionary potential with regard to speciation and evolution of sexual reproduction. Here, we show that for some taxa, such uniformity does not hold. Using hybridoge- netic water frogs (Pelophylax esculentus) as a model system, we demonstrate that triploid hybrid males from two geographic regions exhibit very different reproductive modes. With an integrative data set combining field studies, crossing experiments, flow cytometry and microsatellite analyses, we found that triploid hybrids from Central Europe are rare, occur in male sex onlyand form diploid gametes of a single clonal lineage. In contrast, triploidhybrids from north-western Europe are widespread, occur in both sexes andproduce recombined haploid gametes. These differences translate into con-trasting reproductive roles between regions. In Central Europe, triploidhybrid males sexually parasitize diploid hybrids and just perpetuate theirown genotype – which is the usual pattern in parthenogens. In north-wes-tern Europe, on the other hand, the triploid males are gamete donors fordiploid hybrids, thereby stabilizing the mixed 2n-3n hybrid populations. Bydemonstrating these contrasting roles in male reproduction, we draw atten-tion to a new significant evolutionary potential for animals with nonsexualreproduction, namely reproductive plasticity.
(Bengtsson, 2009; rev. in Sch€ on et al., 2009). Partheno- genetic (also called unisexual) vertebrates mostly arose In vertebrates, a little more than 0.1% of extant species by hybridization between two phylogenetically related reproduce by parthenogenesis sensu lato, that is by apo- sexual species (Vrijenhoek et al., 1989; Avise, 2008; mictic and automictic parthenogenesis, gynogenesis or Choleva et al., 2012; but see Sinclair et al., 2010). Com- hybridogenesis (for details see Suomalainen et al., 1987; bining two different, independently evolving genomes Parker & Niklasson, 1999; Vrijenhoek, 1999; Neaves & of sexual progenitors leads to difficulties in pairing of Baumann, 2011). Comparative studies of these repro- divergent homologs during gametogenesis. This has ductive modes are not only important for understand- modified the normal meiotic cycle in hybrids so that ing the evolution of parthenogenesis and explaining chromosome segregation and recombination is absent the paradox of sex (Otto & Lenormand, 2002), they or limited during gametogenesis. Occasionally the mei- also yield a deeper understanding of the origin of otic problems also result in the production of diploid gametes which, after fusion with haploid or diploidones, produce triploid or tetraploid individuals (Sten-berg & Saura, 2009). The preconditions and the evolu- Correspondence: Heinz-Ulrich Reyer, Institute of Evolutionary Biology tionary role of polyploidy plays in animal systems are, and Environmental Studies, University of Zurich, Winterthurerstrasse however, still widely debated (Cunha et al., 2008; 190, Zurich 8057, Switzerland. Tel.: +41 44 635 49 80; fax: +41 44 63557 11; e-mail: [email protected] Christiansen & Reyer, 2009; Choleva et al., 2012).
(e.g. Uzzell et al., 1975; St€ ock et al., 2010b, 2012).
whereas fish and amphibians are sperm-dependent par- Together, these two factors may result in dynamic thenogens. Therefore, the latter can be considered as reproductive relationships (Alves et al., 2001). This can sexual parasites that must live and mate with an ances- lead to the formation of a new bisexual species via tor (the sexual host) to obtain the sperms that are nec- polyploid speciation (Cunha et al., 2008), or play a key essary for the parthenogen's reproduction (Vrijenhoek, role in maintaining bisexual hybrid populations by 1989). In one form, hybridogenesis, the parasitic taxa releasing the hybrid from its reproductive dependence have a hemiclonal heredity mode, because only one of on a sexual progenitor (G€ unther, 1975; G€ their parental genomes is transmitted to the next gener- 1979; Christiansen & Reyer, 2009).
ation, whereas the second parental genome is elimi-nated prior to meiosis. True syngamy between a The Pelophylax esculentus study system haploid clonal gamete (called a hemiclone sensu Vrijen-hoek, 1979) from the hybridogenetic hybrid and a In this study, we address the origin and test ambiguous recombined gamete provided by the parental species reproductive roles of male polyploidy in P. esculentus whose genome has been eliminated in the hybridogens hybrid water frogs by comparing new results from a reconstitutes a hybrid state in the progeny. Therefore, detailed investigation on a local scale with previously maternal and paternal genomes do not recombine, published results on a wide geographic scale (Pruvost except on rare occasions (Vorburger, 2001b; Guex et al., et al., 2013a). The P. esculentus complex includes two 2002; Schmeller et al., 2005; Lamatsch & St€ sexual species, Pelophylax lessonae (Camerano, 1882), The general rareness of unisexual vertebrates is the pool frog (genotype LL), and Pelophylax ridibundus attributed to the necessity to overcome several prob- (Pallas, 1771), the marsh frog (RR). From their primary lems before they can establish themselves within a nar- hybridization originated, and still originates, the bisex- ual hybridogenetic P. esculentus (Linnaeus, 1758), the hypothesis; Moritz et al., 1989). These problems include edible frog (genomic composition LR) (Fig. 1a). In most genetic incompatibilities between nonrelated parental of the species' European range, diploid P. esculentus live genomes in hybrids, segregation of parental genomes in sympatry with P. lessonae. In these so-called L-E sys- during meiosis and finding an ecological niche in com- tems, the hybrid excludes its haploid L genome, trans- petition with their progenitor species. Another conse- mits in its gametes the haploid R genome and restores quence is that most unisexuals maintain a single clonal hybridity in the new generation by obtaining the L reproductive mode within a mating complex, irrespec- genome from mating with P. lessonae (Fig. 1b). In some tive of whether they are of a monophyletic origin (e.g.
populations, the mirror images, so-called R-E systems, North American hybrid fish Poecilia formosa; St€ are found. Here, most diploid P. esculentus hybrids 2010a), or of an ongoing polyphyletic origin (e.g. Euro- exclude the R, transmit their L genomes and mate with pean hybrid fish of the genus Cobitis; Choleva et al., P. ridibundus to perpetuate the hybrid populations 2008; Janko et al., 2012). Hence, unisexual vertebrates (reviewed by Graf & Polls Pelaz, 1989; Pl€ otner, 2005).
are generally considered as taxa with a low evolution- In several areas of the species' range also triploid ary potential in terms of speciation and evolution of hybrids have been found. This is especially true for sex. Their uniform reproductive mode, so the argu- northern European regions belonging to the drainage ment, allows for a single role only: to self-reproduce basins of the North Sea and the Baltic Sea (Berger, their own genotypes or individual lineages (e.g. Vrijen- 1988b; Rybacki & Berger, 2001; Pl€ otner, 2005). In this hoek, 1989; Maynard-Smith, 1992).
area, the most frequent population structure is the one This, however, is not always true. Some of these with no parental species and two or three types of mating systems display reproductive plasticity with hybrids: diploid LR in sympatry with triploid hybrids, signs of an evolutionary potential. This plasticity is mostly with LLR, but also with LRR or both (Christiansen achieved through at least two co-occurring factors.
& Reyer, 2009; Arioli et al., 2010; Jakob et al., 2010; Pru- First, although hybrid males are usually rare and sterile vost et al., 2013a). In those all-hybrid populations, trip- (e.g. Choleva et al., 2012), functional hybrid males loids of both genomic compositions (LLR and LRR) are occur regularly in some taxa. These include the usually formed by fusion of diploid clonal LR eggs pro- hybridogenetic water frog Pelophylax esculentus (Graf duced by LR females with haploid recombined L or R & Polls Pelaz, 1989; Polls Pelaz, 1994; Christiansen sperm of LLR or LRR males, respectively (see Fig. 1d for & Reyer, 2009), the fishes Squalius alburnoides (Alves the LR/LLR populations). Diploid hybrids (LR) can arise et al., 2001) and Hypseleotris (Schmidt et al., 2011), and from the fusion of haploid-recombined L and R gametes the Palearctic green toad of the Bufotes viridis complex of male and female LLR and LRR, respectively, and from ock et al., 2010b). Second, a single hybrid genotype the fusion of recombined L eggs of LLR females and hap- of the above mentioned taxa can often produce more loid clonal R sperm of LR males (for details see Christian- than one type of gametes with some level of recombi- sen, 2009 and Christiansen & Reyer, 2009). Therefore, nation between the conspecific genomes in polyploids by providing recombined haploid gametes in E-E

Reproductive strategies in triploid frogs We, therefore, sampled the area with Central Euro- pean triploid P. esculentus populations to address the ori-gin understand how polyploid vertebrates can evolve fromtheir sexual ancestors and to investigate whether theyuse different reproductive modes in different geographicareas. We particularly studied the following four topics:(I) The structure of populations in terms of genotypes,ploidy levels and sex ratios; (II) Gamete types of dip-loids and triploids, and formation of triploids; (III) Therole of triploids within the breeding system; and (IV) Single or multiple origin and nature of hemiclonallytransmitted genomes. Here, we integrate multiple typesof data from European water frogs to demonstrate con-trasting reproductive pathways (self-reproducing modeor contributing to perpetuate the hybrid population)found within a single parthenogenetic mating system(P. esculentus complex), genotype and sex.
Materials and methods To address the origin and role of polyploidy in waterfrog systems, we combined multiple types of data. Wedid a comparative field study (for topic I), performedartificial crossing, conducted microsatellite analysesexperiments and flow cytometry on sperms (for topicsII and III) and compared gamete production patterns,triploid formation and hemiclonal lineages among theeight populations from our study area in Central Eur-ope (for topic IV).
Fig. 1 Origin of (a) diploid Pelophylax esculentus (LR) from primaryhybridization between Pelophylax lessonae males (LL) and Pelophylax ridibundus females (RR) and (b) perpetuation of diploid and (c) During springs 2008–2010, we collected both published triploid hybrid lineages in Central Europe and (d) north-western and unpublished data on the assumed presence of trip- Europe. Gamete types are shown in circles and underlined in caseof diploid gametes. X and Y indicate female and male sex- loid P. esculentus in Central Europe and sampled a total determining factors, respectively.
of 524 specimens from eight populations in Slovakiaand one in the Czech Republic (see Fig. 2 for locations systems, triploid males substitute the role of parental spe- and Table 1 for names, coordinates, frog sample size cies in L-E and R-E systems (G€ unther et al., 1979), turn- and type of each population). We also sampled twice ing sperm-dependent hybridogens into independent (May 2009 and June 2014) in an area studied by Tun- ‘sexually' reproducing units with an evolutionary poten- ner & Heppich-Tunner (1992) (see ellipse in Fig. 2), tial (Christiansen & Reyer, 2009).
but in contrast to these authors, there we did not find In contrast to this pattern, hybrids from Central Eur- a single polyploid frog in a total of more than 200 indi- ope are mostly of diploid genomic constitution (Berger viduals. Frogs were hand-collected at night and kept et al., 1988; Vorburger, 2001a; Pl€ otner, 2005; Mikul
ıcek separated by sexes in spacious plastic containers. They et al., 2014a). So far, triploids have been reported only were assigned to taxa (P. lessonae, P. ridibundus and from two Central European regions and only in the P. esculentus) according to species-specific morphological form of LLR males (Tunner & Heppich-Tunner, 1992; characters (Pl€ otner, 2005). All specimens were mea- Tunner, 2000; Mikul
ıcek & Kotl
ık, 2001; Mikul
ıcek sured, photographed and toe clipped. Ploidy levels of et al., 2014a). In contrast to the well-studied species' the P. esculentus hybrids were determined by erythro- north-western range, where triploid hybrids flourish, cytes' size in field conditions (erythrocytes of triploids the gamete production pattern and the reproductive are significantly larger than diploid ones; Berger, role of LLR males in Central Europe were poorly 1988a; Vinogradov et al., 1990) and later confirmed by known. It was also not known whether triploids in the DNA microsatellite analyses in the laboratory. Frogs two geographical areas originated from the same or dif- selected for crossing were individually transpondered ferent hyridization events.
(RFID PIT tag AEG ID-162, Ulm, Germany), separated Table 1 Population types and number of frogs sampled in each of them (N LL = number of Pelophylax lessonae, N LR = number of diploidPelophylax esculentus, N LLR = number of triploid LLR P. esculentus, N RR = number of Pelophylax ridibundus).
Sajd
ıkove Humence Sajd Numbers in brackets give the number of males, females and individuals of unknown sex, respectively; ?? indicates that in Bahno and Koz
ıChrb
at, no LR males were caught, although their presence is likely.
by sex and population of origin and transported to the were achieved following the Berger et al. (1994) proto- University of Z€ urich. During transport, they were stored col with slight modifications: To induce ovulation, in cloth bags containing small pieces of rubber sponge females were injected with 100 lL per 10 g body mass and showered daily with fresh water. All frogs survived of a 20 mg L
1 LHRH hormone in Holtfreter solution the journey. Once in Z€ urich, they were kept separated (59 mM NaCl, 0.7 mM KCl, 0.9 mM CaCl2, 2.4 mM NaH- by sexes, released in outdoor cages and fed ad libitum CO3 and 1.6 mM MgSO4, pH 7.4). Males were anesthe- with live crickets.
tized in a buffered solution of MS-222 (0.15 g L
1)before having one of their testes removed and laceratedinto a Petri dish to obtain the sperm solution. This pro- Artificial crossing experiments tocol permits the use of the same sperm solution to fer- We studied the gamete production pattern of hybrid tilize eggs from different females and to cross the same frogs coming from the populations where triploids were female with different males. After about 15 days, the found (Table 1), with the exception of Bahno, because obtained embryos reached free swimming stage (stage this population was discovered later in the course of 25, Gosner, 1960) and were euthanized using an over- this study. Instead, we included one population without dosed MS-222 buffered solution (2 g L
1). The off- Sast
ın-Str
aze) where we caught a large num- spring of a few crosses were used for other experiments ber of diploid hybrids of both sexes. The original experi- (Pruvost et al., 2013b), but their genotypic data could mental design was to cross each hybrid both with other also be used for our purpose.
hybrids and with at least one specimen of each parentalspecies to determine whether they produce clonal or recombined gametes. Because some females had a lim-ited number of eggs, the full design could not be Forty-three hybrids were analysed by flow cytometry to applied in the populations of  Sajd
ıkove Humence and confirm their ploidy level and, if males, to determine Borovec (see results in Table 2). Based on results of ploidy level of their sperms. Blood and sperm samples previous studies using four to eight allozyme markers were stabilized in buffer (40 mM citric acid trisodium and crossing experiments with frogs from four Central salt, 0.25 M saccharose and 5% DMSO) and immedi- European populations (Tunner, 1980; Tunner & Hepp- ately frozen at
80 °C (Cunha et al., 2008). Samples of ich-Tunner, 1992; Mikul
ıcek & Kotl
ık, 2001), we tested both parental species were used as a diploid standard.
with a set of microsatellites a prediction whether dip- Relative nuclear DNA content was measured using loid and LLR triploid hybrids produce haploid R and DAPI fluorochrome applying a commercial kit Cystain diploid LL gametes, respectively. Artificial fertilizations two Step High Resolution DNA Staining (Partec GmbH,

Reproductive strategies in triploid frogs unster, Germany). Fluorescence intensity of 5000 stained nuclei was measured in Partec PA II flow cy-tometer with a speed 0.5 lL s
1. Flow cytometric histo-grams were evaluated using FloMax 2.52 software(Partec GmbH, M€ unster, Germany).
The combination of 18 microsatellite loci was used todetermine and/or confirm the genomic composition ofthe crossed specimens and their offspring, in terms oftaxon and ploidy level to understand the heredity modeof polyploids. To address the evolution of water frogpolyploidy, we used a population genetic approach.
Observation of a low genetic diversity and little geneticdifferentiation within clonally transmitted genomeswould suggest a single rather than a multiple origin ofhemiclones. DNA was extracted from toe or tail tips ofthe adult frogs or tadpoles, respectively, stored in 96%ethanol. The Qiagen BiosprintTM 96 DNA Blood Kit(Qiagen, Venlo, the Netherlands) was used for extrac-tion following supplier's protocol. We used a set of 18microsatellite primer pairs which were run in four pri-mer mixes: 1 Primer Mix 1A – CA1b6, Ga1a19 redesigned (Arioli et al., 2010), RlCA1b5, RlCA5 (Garner et al., 2000),Rrid064A (Christiansen & Reyer, 2009) 2 Primer Mix 1B – Re2CAGA3 (Arioli et al., 2010), Res16, Res20 (Zeisset et al., 2000) RlCA2a34 (Chris-tiansen & Reyer, 2009) 3 Primer Mix 2A – ReGA1a23, Rrid169A, Rrid059A redesigned (Christiansen & Reyer, 2009), Res22(Zeisset et al., 2000), Rrid013A (Hotz et al., 2001) 4 Primer Mix 2B – Re1CAGA10 (Arioli et al., 2010), RlCA18 (Garner et al., 2000), RlCA1a27, Rrid135A(Christiansen & Reyer, 2009).
Details on PCR protocols are given by Christiansen (2009) and Christiansen & Reyer (2009). Fragmentlength analysis of the PCR products was run on an ABI3730 Avant capillary sequencer with internal size stan-dard (GeneScan-500 LIZ; Life Technologies EuropeB.V., Zug, Switzerland), and the alleles were scoredwith the GeneMapper software 3.7 (Applied Biosys-tems, Foster City, CA, USA).
We knew from previous studies that three microsatel- lite loci are species specific for the P. lessonae genome Fig. 2 (a) Areas of water frog populations with triploid hybrids (Res20, RlCA1a27 and RlCA18), four are specific for relevant to this study. The ellipse shows the approximate P. ridibundus genome (Re2CAGA3, Res22, Rrid169A major distribution around the Baltic Sea, with four localities and Rrid135A), and 11 loci amplify in both the L and R (black dots) for which the north-western European pattern of genomes (Christiansen, 2005, 2009; Arioli et al., 2010).
gamete production has been documented. The rectangleindicates the Central European area investigated in this study.
(b) Enlarged map of the study area shows locations of the nine Estimation of null alleles and selection of investigated populations (black dots) and a previously studied microsatellite loci area (ellipse; Tunner & Heppich-Tunner, 1992) with the sameCentral European gamete production pattern as found in our Because L and R genomes do not recombine in hybrids, the two genomes were considered separately in the Table 2 Origin, genotype, sex and individual numbers of the absence of an allele as evidence for a null allele. We frogs used in artificial crossing experiments.
then excluded all loci showing an estimated null allelefrequency > 0.2 in any of the populations. This led us to exclude locus Re1CAGA10 for the L genome, Genotype Sex Ind. Num.
crosses offspring Gametes RlCA2a34 for the R genome, and RlCA5 and Res16 for both genomes from subsequent analyses. We also Rrid059A redesigned for the L genome and locus ReGA1a23 for the R genome, because in all samples they showed only one allele per locus and, thus, pro- vided no variation for the genetic analysis. This left us with 8 loci for the L genome and 11 for the R genome: Res20, RlCA2a34, ReGA1a23, RlCA1a27 and RlCA18 (L genome); Ga1a19 redesigned, Rrid064, Re2CAGA3, Res22, Rrid169A, Rrid059A redesigned, Re1CAGA10 and Rrid135A (R genome); and CA1b6, RlCA1b5 and Rrid013A (for both L and R genomes).
Analysis of genetic diversity and differentiation at individual and population levels We calculated the gene diversity, corrected for sample size, expressed by the expected heterozygosity (H 1978), using the program SPAGeDi 1.3 (Hardy & Veke- mans, 2002) and the allelic richness (AR) using the program FStat 2.9.3.2 (Goudet, 2001). Genetic differen- tiation between populations and genotypes was mea- sured using FST statistics following the method of Weir & Cockerham(1984), which is implemented in the pro- gram SPAGeDi 1.3. The program allows the combina- tion of multiple ploidy levels in the same analysis.
Concerning genetic diversity, we used two tailed pair- wise t-tests on the values of He for each locus, to test N crosses = number of different partners the individual was the significance of differences between different frog crossed with; N offspring = number of resulting tadpoles that were types, independent of their origin, and we used ANOVAs analysed; and Gametes = gamete type produced by each individual to look for differences in He between population types.
as deducted from the parents' and the offspring's genotypes. All Statistical tests were run using the program R 2.15.1 individuals exclusively produced the indicated gamete type.
(http://www.r-project.org/). Differences in AR amonggenomes present in different genotypes were carriedout using two-sided permutation tests implemented in subsequent genetic analyses. Prior to these steps, we tested raw data for the presence of null alleles. Nonam- To test whether R and L genomes present in hybrido- plifying loci were rerun for PCR two to three times.
genetic hybrids are related to those present in the local When even then no allele was amplified, we attributed parental species, Bayesian assignment programs STRUC- the result to the presence of a null allele, rather than to TURE 2.3.3 (Pritchard et al., 2000; Falush et al., 2007) low DNA quality, because this individual DNA ampli- and BAPS 5.3 (Corander et al., 2003) were applied.
fied for other loci. Potential genotyping errors like stut- These programs use an iterative approach to assign tering, allelic dropout or presence of null alleles were genotypes into K populations without a priori knowl- tested separately for parental RR and LL taxa using the edge of the population membership of individuals, min- program Micro-Checker 2.2.3 (Van Oosterhout et al., 2004). We estimated frequencies of null alleles with the disequilibria within populations. Both parental genomes Brookfield 2 null allele estimator, which treats nonam- were analysed separately. Models implemented in both plifications as data and regards them as null homozyg- programs assume that loci are unlinked and in H-W equilibrium. These assumptions are unlikely to be met (Brookfield, 1996). Because this method cannot be in clonal and hemiclonal hybrid populations because of applied to the diploid hybrids, we inspected the L and fixed heterozygosity and linkage of multilocus haplo- R genomes in hybrids visually and considered the types. Therefore, we did not infer the most likely Reproductive strategies in triploid frogs number of K, that is clusters with H-W and linkage equilibria. Instead, only fixed K = 2 and K = 3 were microsatellite alleles, producing a chain of allele sizes used, assuming hybrids and a parental species (K = 2), which represent our MLGs. We then compared these and diploid hybrids, triploid hybrids and a parental spe- MLGs to find whether they were present in other pop- cies (K = 3) as the clusters, respectively. Using STUC- ulations under study. The detected MGLs were named TURE, admixture and uncorrelated allele models were after the hemiclone type (L, R, LL), followed by a capi- applied. The analyses were based on runs of 106 itera- tal letter attributed in accordance to the descending tions, following a burn-in period of 100 000 iterations.
overall frequency (e.g. L-A = P. lessonae hemiclone- A series of ten independent runs for each K was made A = most frequent L hemiclone). For more details, see with the same parameters to test the accuracy of results. In BAPS, a clustering of groups of individualswas run first, followed by an admixture clustering (Cor- ander & Marttinen, 2006; Corander et al., 2008). Thenumber of iterations that were used to estimate the Structure of water frog populations (topic I) admixture coefficients for the individuals, and thenumber of reference individuals from each population, The genomic composition of the 524 sampled speci- was 200. The number of iterations that were used to mens analysed with 18 microsatellite loci showed that estimate the admixture for the reference individuals all but one population contained two or three water was set to 20.
frog genotypes (DNA microsatellite data are given inData S1). The exception was the  Sprinclov majer local- ity, where we found only P. ridibundus. The exact num- Analysis of hemiclonal diversity bers, including sex ratio for each genotype, are listed in As coined by Vrijenhoek (1979), the term ‘hemiclone' Table 1. Based on their genotype composition, the eight refers to the clonally transmitted haploid genome, populations were classified into four types (1–4), each which in our case can be of the L or R type. We deter- represented by two localities. The four populations of mined them by a multilocus genotype (MLG), defined types 1 and 2 contained only diploid hybrids, whereas by the identical combination of alleles found in our mi- the populations of types 3 and 4 were inhabited also by crosatellite analysis. The same MLG can be, however, triploids. In all but two populations, diploid genotypes found also in two or more unrelated sexual individuals were always found in both sexes with a male bias in when discrimination power of used molecular markers the parental species LL and RR and a female bias in LR is low. Therefore, we first calculated two statistics, hybrids (see totals in Table 1). The two exceptions were probability of identity (PI) and probability of identity Bahno and Koz
ı Chrb
at where no LR males were siblings (PIsibs), that estimate the probability that two caught. In contrast, triploid LLR hybrids in the four individuals randomly chosen from a population have populations of types 3 and 4 occurred as males only; the same MLG on a set of markers (Waits et al., 2001).
LLR females were neither caught during this study Both statistics were calculated for both parental species (Table 1) nor found during previous samplings per- using GenAlEx 6.4 (Peakall & Smouse, 2006). PI and formed by Mikul
ıcek et al. (2014a).
2.4 9 10
4, respectively. PI and PIsibs for P. lessonae Gamete production (topics II and III) were 1.6 9 10
7 and 2.3 9 10
3, respectively. Thesevalues are reasonably low (cf. Waits et al., 2001), indi- To identify the heredity mode among hybrids, we per- cating there is low probability that two P. ridibundus or formed flow cytometry on sperms of 28 males and P. lessonae individuals share the same MLG on a set of genotyped 2216 offspring from 96 crosses through mi- used microsatellites. Following this calculation, we crosatellite analyses. Flow cytometric analysis allowed applied a conservative approach and recognized a hemi- us to distinguish different ploidy levels among frogs clone when the same MLG was present in our sample (2n, 3n and one 4n individual) and between parental more than three times.
genotypes (RR and LL) (Data S2a). It also allowed dis- As different hemiclonal gametes may fuse (syngamy) tinguishing between haploid sperms (produced by and develop into diploid zygotes on the basis of parental males and 2n and 4n hybrids) and 2n sperms hybrid 9 hybrid matings (Hotz et al., 1992), we also (produced by 3n males) (Data S2b). In all these cases, searched for possible hemiclonal MLG combinations in the flow cytometric histograms were clearly nonover- the individual genomes of the parental species (LL and lapping. In contrast, overlapping histograms of blood RR) and in diploid and triploid hybrids (LR and LLR).
samples did not allow distinguishing between genotypes Because some triploid LLR hybrids may produce also of diploid hybrids (LR) and parental species, nor was it diploid (LL) hemiclonal gametes (Tunner & Heppich- possible to tell whether LLR males produced LL or LR Tunner, 1992; Mikul
ıcek & Kotl
ık, 2001), we tested the sperms. However, in combination with results from the data for the presence of LL hemiclones as well. To do artificial crossing experiment, we unambiguously iden- tified the gamete production pattern, including for (AR = 1.625) and LL (AR = 8.125, two-sided female eggs which cannot be analysed through flow permutation test, P = 0.0001), and between LLR and cytometry. All specimens of the two sexual parental LR (AR = 7.272, P = 0.003), but not between diploid species used for the crosses acted as normal haploid hybrids and P. lessonae (P = 0.308). For the P. ridibundus gamete donors (L in P. lessonae, R in P. ridibundus) with genome, highly significant differences have been found chromosome segregation in accordance with the second Mendel's law. Both sexes of LR hybrids produced hap- P = 0.009), and between RR and LLR (AR = 3.000, loid R gametes only. The triploid LLR hybrid males P = 0.002), but not between diploid and triploid exclusively produced diploid clonal LL gametes, a pat- hybrids (P = 0.825).
tern supported by two independent analyses: flow Global FST values showed significant and substantial cytometry on sperm (Data S2) and microsatellite analy- differentiation among populations for both genomes.
ses on parents and offspring from the crossing experi- The mean FST values were 0.271 for the L genome and 0.114 for the R genome, respectively. For the L (WFB015-54 from Koz
ı Chrb
at) produced haploid R sperms and a few diploid cells of unknown genotypic between LL and LR individuals (FST = 0.021), but very large differentiation between LLR and both LL and LRindividuals Table 4). For R genomes, the genetic differentiation Population genetics (topic IV) was small between LR and LLR hybrids (FST = 0.019), Genetic diversity and differentiation whereas it was large between RR and both LR and LLR The genetic diversity estimates for the L genomes (HeL) hybrids (FST = 0.133 and FST = 0.129, respectively).
and for the R genomes (HeR), are presented in Table 3 Pairwise FST values clearly separated the L genomes of and Data S3. Pooled over all eight populations, gene triploid LLR hybrids from those of LR and LL individu- diversity in the P. lessonae genome was significantly als (Table 5). Hence, the triploids were in their L ge- lower in LLR triploids (HeL = 0.256) compared to P. les- nomes genetically not only strongly differentiated from the parental LL individuals, but also from the diploid LR hybrids in syntopic populations. In contrast, there t(7) = 2.364, P = 0.011). No significant differences in was little to only moderate genetic differentiation in L HeL were found between the P. lessonae genome of dip- genomes of parental LL individuals and diploid LR loids and the parental species (t 2.364, P = 0.111).
hybrids in all population types. The only exception was For the P. ridibundus genome, significant differences in represented by diploid LR hybrids from the Czech pop- ulation of Borovec, whose L genome was distinct from 0.631) and both LR (HeR all other populations (Table 5).
P = 0.001) and LLR hybrids (HeR = 0.413, t(10) = 0.228, Concerning the R genomes, parental RR individuals P = 0.006), but not between diploid and triploid from different localities revealed mostly little to large 0.228, P = 0.996). Significant differences genetic differentiation between themselves and mostly in gene diversity between different population types moderate to large differentiation between them and were not observed.
both diploid and triploid hybrids from all population In terms of AR, highly significant differences in types (Table 5). In contrast, there was only little to the P. lessonae genome have been found between moderate differentiation among R genomes of both Table 3 Gene diversity (He) corrected for sample size (Nei, 1978) for Pelophylax lessonae genomes (HL) and Pelophylax ridibundus genomes(HR) in the different frog genotypes (LL, LLR, LR, RR). Sample size is given in brackets.
PT2 (LL + LR + RR) PT2 (LL + LR + RR) PT4 (LLR + LR + RR) Sajd
ıkove Humence PT4 (LLR + LR + RR) Reproductive strategies in triploid frogs hybrid types (LR and LLR) from all populations. Again, a separate cluster with high probability. Structuring in the population in Borovec stood out, because both the the P. ridibundus genome between the genotypes RR, diploid and the triploid hybrids genetically differed in LR and LLR was not so straightforward (Fig. 3b). More their R genome from parental RR individuals and than 90% of P. ridibundus individuals were assigned to hybrids found elsewhere.
the cluster 1 regardless of the number of expected K, The results of the two Bayesian programs were whereas 64% of both diploid and triploid hybrids were concordant and revealed substantial structuring in the assigned to cluster 2 (including almost all diploid LR P. lessonae genome (Fig. 3a). Triploid hybrids on the from Borovec); remaining hybrids were assigned to the one hand and diploid hybrids and P. lessonae on cluster 1 (assuming K = 2) and clusters 1 or 3 (assum- the other were unequivocally assigned to two separate ing K = 3). Only few individuals were assigned to more clusters assuming K = 2. Assuming K = 3, Bayesian than one cluster revealing admixture across analyses.
clustering was very similar, with the exception of LRhybrids from Borovec – most of them were assigned to Analysis of hemiclonal diversityWith respect to the R genomes present in hybrid indi-viduals, we detected a total of 14 hemiclones with dif- Table 4 Pairwise FST values for L (below the diagonal) and R ferent relative frequencies among populations (Table 6 (above the diagonal) genomes between the hybrid types listed in and Data S4). In the Czech population of Borovec, we the left column (LLR, LR) and the hybrids and parental species found only a single hemiclone (R-B), whereas the shown in the top horizontal row (LL, LLR, LR, RR) and parental Slovakian populations contained multiple R hemi- species (LL, RR), pooled over all populations.
clones, ranging from four in Brodsk
e to eight in  Str
aze (Table 6). Hemiclone R-A occurred in all fourpopulation types (PT1-4); five (R-F, R-H, R-K, R-L and R-N) occurred only in populations with parental LL frogs (PT1 and/or PT2); and three hemiclones (R-B, R- Table 5 Pairwise FST value comparisons between all genotype-population combinations for L (below the diagonal) and R (above thediagonal) genomes. Darker colours correspond to lower FST values. For abbreviations of population names, see Table 1.
Bors Kala Brod Sast Bahn Kozi
Sajd Boro Bors Kala Brod Sast Bahn Kozi
Sajd Boro Brod Sast Sajd Boro Spri
Bahn LLR .280
very great

Fig. 3 Structuring in the L genome (a)and the R genome (b) according to aBayesian analysis assuming two (K = 2)and three (K = 3) clusters, respectively.
G, and R-M) were found only in populations where genomes combined from two hemiclones, namely as a triploid LLR hybrids are present (PT3 and PT4). The combination of LL-A hemiclone and one of six R hemi- remaining four hemiclones (R-C, R-E, R-I and R-J) clones (Comb-B to G, Table 6). Three LLR males from were not specific to any population type.
Borovec were composed of LL-B and R-B hemiclones Concerning the L genome, the number of hemiclones (Comb-H, Table 6).
was much smaller than within the R genome. Wedetected only a single L hemiclone (L-A) and two LL (diploid) hemiclones (LL-A and LL-B). L-A occurredonly in diploid hybrids from Borovec but there in a Population composition and gamete production very high proportion (38 of 50 sampled LR frogs). Both (topics I and II) LL hemiclones were present in all male LLR triploidhybrids (N = 83). One hemiclone (LL-B) was restricted In the nine sampling sites that we studied in Central to Borovec, and the other one (LL-A) was present in Europe, we have identified four population types, the three Slovak populations (Bahno, Koz
ı Chrb
at and three where hybrids live in sympatry with one or Sajd
ıkove Humence). The two LL hemiclones differed both parental species and one with hybrids only. Our by only one allele in their MLG comparison, showing combined data from flow cytometry, crossing experi- one dinucleotide repetition difference at the locus ments, analysis of genetic diversity and gene flow RlCA18. In Borovec, the locus RlCA18 amplified for between genotypes show that even in the two appar- alleles 177 and 181, whereas LLR frogs from Slovakia ent all-hybrid populations of type 3, the parental spe- carried alleles 179 and 181. We further found that 35 cies P. lessonae rather than triploid hybrids provide L of 50 LR hybrids from Borovec likely originated from a combination of two hemiclones, namely L-A and R-B hybrids. This is further supported by the FST statistics (called as Comb-A, Table 6). Triploid LLR individuals in and clustering of L genomes from LR hybrids with Koz
ı Chrb
at,  Sajd
ıkove Humence and Bahno had also P. lessonae and not with LLR hybrids (Tables 4 and 5), Reproductive strategies in triploid frogs Table 6 Multilocus genotypes (MLGs) for L, R and LL hemiclones and their combinations found in the study and their distribution overthe populations.
Hemiclonal MLG name Distribution of hemiclonal MLGs in populations 17 Sast (17 LR), 13 Kozi (9 LLR, 4 LR), LR and LLR hybrids 10 Sajd (4 LLR, 6 LR), 6 Bahn (2 LLR, 4 LR), 5 Brod (LR), 3 Bors (LR), 2 Kala (LR) 50 Boro (3 LLR, 47 LR), 2 Bahn (LR), 21 Kozi (15 LLR, 6 LR), 14 Kala (LR), 8 Bahn (1 LLR, 7 LR), 4 Sajd (3LLR, 1 LR) 10 Bors (LR), 8 Sast (LR), 7 Kozi (4 LLR, 3 LR), 6 Bahn (LR), 6 Brod (LR), 5 Sajd (2 LLR, 3 LR), 14 Kozi (10 LLR, 4 LR), 7 Bahn (2 LLR, 5 LR), 6 Sajd (5 LLR, 1 LR), 2 Kala (LR) 18 Sats (LR), 1 Kala (LR) 11 Kozi (8 LLR, 3 LR), 1 Sajd (LLR) 6 Bors (LR), 5 Sast (LR), 1 Kala (LR) 8 Kala (LR), 1 Bahn (LR), 1 Bors (LR) 5 Sast (LR), 3 Brod (LR), 1 Bahn (LR) 6 Sast (LR), 1 Bors (LR) 5 Kozi (LLR), 1 Sajd (LLR) 2 Sast (LR), 1 Bors (LR) 1 Brod (LR) L hemiclone in LR hybrids, LL hemiclone in LLR hybrids 52 Kozi (LLR), 20 Sajd (LLR), 5 Bahn (LLR) Comb. LL+R hemiclones 15 Kozi, 3 Sajd, 1 Bahn (composed of LL-A + R-C) 10 Kozi, 5 Sajd, 2 Bahn (composed of LL-A + R-E) 9 Kozi, 4 Sajd, 2 Bahn (composed of LL-A + R-A) 8 Kozi. 1 Sajd (composed of LL-A + R-G) 4 Kozi, 2 Sajd (composed of LL-A + R-D) 5 Kozi, 1 Sajd (composed of LL-A + R-M) 3 Boro (composed of LL-B + R-B) Comb L+R hemiclones 35 Boro (composed of L-A + R-B) Letters (A-N) behind the genomes indicate different hemiclones: ‘single MLG' refers to allele combinations that were found in only one ortwo copies and, hence, were not considered to form a hemiclone. For abbreviations of population names, see Table 1.
as well as by contrasting levels of genetic diversity (He between RR individuals sampled in different sites and AR), which is comparable between LR hybrids and (Table 5; cf. Mikul
ıcek et al., 2014b). Moreover, the R P. lessonae but substantially lower in the LL genome of genome of hybrids reveals lower genetic diversity in LLR triploids. L genome provisioning through P. lesso- comparison to P. ridibundus, thus showing that only nae is characteristic for L-E system (here represented part of the P. ridibundus individuals contributed to the by the population type 1), where diploid hybrids clon- formation of hybridogenetic lineages. Our combined ally transmit R genomes (see Table 2) and receive L data show that LLR hybrids exclusively produce diploid gametes from P. lessonae (Tunner, 1974; Uzzell & Ber- LL rather than haploid L sperms. Thus, matings ger, 1975; Graf & Polls Pelaz, 1989). Even where between LLR males and LR females will result in LLR parental P. ridibundus individuals exist, as in population offspring. This raises the question how then diploid LR types 2 and 4, they do not seem to be the major con- hybrids are produced and maintained in the four pop- tributors of R gametes to hybrid progeny. This is indi- ulations of types 3 and 4. At present, the answer cated by the fact that genetic differentiation among R remains open, but we develop three not mutually genomes (FST, Bayesian analysis) is larger between LR exclusive hypotheses in Data S5.
hybrids and RR sexuals in the same population than

The population data and gamete production modes (a) Central Europe
are in agreement with an XX–XY sex determinationsystem in which the hemiclonal genome may be cou-pled with either an X or Y haploid set of chromosomes(Graf & Polls Pelaz, 1989). In hybrids, the R genomelikely carries a female determining factor (X), whereasL genomes carry female (X) or male (Y) determiningfactors with equal probability (Berger et al., 1988;Christiansen, 2009). Therefore, in principle, when thediploid LL sperms of LLR males fertilize haploid R eggsof LR females, only LLR males (LYL?RX) will be pro-duced (Fig. 1c). In contrast, diploid LR hybrids come inboth sexes, but with an excess of females because LR (b) North-western Europe
males sire daughters only (cf. Fig. 1b). This female biasseems to be particularly extreme in the PannonianBasin to which all but one (Borovec) study populationbelongs. In this basin, male proportions as low as 3% Fig. 4 Heredity pathways of L genomes (black arrows) and R have been found (Tunner & Dobrowsky, 1976; Berger genomes (white arrows) between different genotypes in et al., 1988; Gubanyi & Creemers, 1994; Mikul
ıcek & populations with triploid hybrids from (a) Central Europe (this Kotl
ık, 2001). Therefore, the virtual absence of LR study) and (b) north-western Europe (simplified from Som & males in Bahno and Koz
ı Chrb
at may be the result of a Reyer, 2006; Christiansen, 2009).
sampling bias that is due to low abundance. In contrast,the lack of LLR females in Central Europe is to beexpected from the gamete production pattern (Fig. 1c).
produce LLR offspring of both sexes (LYL?RX, LXLXRX)(Fig. 1d).
Second, LLR males in Central Europe sexually para- The reproductive role of triploid males in Central sitize LR females for self-reproduction, because the and north-western Europe (topic III) resulting progeny are 100% LLR males which will In none of the nine Central European populations exclude the R genomes at gametogenesis (Fig. 4a).
that we studied in this paper did we find evidence These diploid LR females, in turn, are also sexual para- for the type of all-hybrid populations that are typical sites because for successful reproduction of hybrid off- for north-western Europe. Even the two populations spring they require a donor of L gametes, which likely containing only diploid LR and triploid LLR hybrids (Fig. 4a). In contrast, in all-hybrid E-E populations Although both regions share a presence of LLR males, from the north-western Europe, L and R alleles are they differ in several aspects which are summarized passed on between diploid and triploid males and females (Berger, 1988b; G€ First, sex ratios differ markedly among LLR triploids.
& Reyer, 2006) (Fig. 4b). In these latter populations, Because of the XX/XY sex determining mechanism LLR males are sexual hosts for the diploid LR females described above, fusion of LL sperms from LLR males and, hence, fulfil the key role that P. lessonae has in L-E and R eggs from LR females in Central Europe will systems. Thus, in north-western Europe, triploids help result in LLR males only (LYL?RX) (Fig. 1c). In contrast, in stabilizing (all-hybrid) populations by substituting fusion of haploid L sperms from LLR males with diploid the role of sexual species, whereas in Central Europe, LR eggs from LR females in north-western Europe will they do not.
Table 7 Differences between water frog populations with triploid LLR individuals in north-western and Central Europe; occasionaldeviations from this pattern do occur, based on this study and data from Christiansen (2009) Christiansen & Reyer (2009) and Jakob et al.
(2010).
Features of LLR frogs North Western Europe Gamete production Clonal LL gametes Recombined L gametes Origin of triploids LL sperms from LLR males 9 R eggs from LR females L sperms from LLR males 9 LR eggs from LR females Reproductive role Self-reproduction sexual parasites L gamete donor substitute sexual hosts Reproductive strategies in triploid frogs present, we do not know whether the two geographic The origin of male polyploidy in Central Europe regions represent single or multiple hybrid origin.
However, our results strongly suggest that partheno- Our results from microsatellite genotyping, crossing genetic animals (sensu lato) originating from the same experiments and population genetic statistics consis- parental species and carrying even the same genotype tently indicate that LLR from all populations were very (here LLR) can independently develop various repro- similar with respect to the multilocus genotype (MLG) ductive roles. These findings place hybrid water frogs of their two lessonae genomes: in the three Slovakian in contrast to most other vertebrate parthenogenetic populations, the MLG was identical, and in the Czech systems. For example, all taxa of parthenogenetic rep- population of Borovec (130 km apart), it differed by tiles are virtually constrained into a single reproduc- only a single allele mutation at the locus RlCA18. We therefore believe that LL hemiclones represent a single represents a genetic copy of the mother (see a list of clonal lineage which diversified by mutation after taxa in Kearney et al., 2009). Similarly, most parthe- nogenetic fish (either diploid or polyploid) show a The geographic origin of this LL hemiclonal lineage, uniform reproductive system, for example in the however, remains puzzling. Given the high genetic dif- genus Cobitis, Poecilia, Poeciliopsis and others (Lamatsch ferentiation in L genomes between the LLR triploids and the group of Slovakian sexual LL and hybrid LR S. alburnoides complex produce eggs of various ploidies frogs, the origin is unlikely to have been in situ, at least within a single genotype and individual, their role in not in a recent time. On the other hand, the LL hemi- a mating system is rather complex than contrasting clone of triploid males is genetically more similar to the (Alves et al., 2001). Fertile diploid and triploid hybrid L genome of diploid hybrids in Borovec than to the L males in Squalius maintain only clonal spermatogene- genome of diploid hybrids in Slovakia. Therefore, we sis, whereas tetraploids produce one type of meiotic suggest that the LL hemiclonal lineage might have orig- sperms (Collares-Pereira et al., 2013). Therefore, a inated somewhere in the area of Borovec, a sample site demonstration of contrasting roles in reproduction of situated in the proximity of the European watershed of a single genotype in vertebrate parthenogens in gen- the Baltic Sea, the North Sea and the Black Sea. Subse- eral, and in male sex in particular (i.e. to be a donor quently, it may have spread southerly through the of gametes vs. to be a sexual parasite as we evi- Danubian Basin. The origin of the haploid L hemiclone denced in LLR triploids), gives an example of a new found in LR hybrids in Borovec remains unclear, significant evolutionary potential (reproductive plastic- because we were not able to recognize its donor in the ity) in animals with nonsexual reproduction. The present data also open research questions for future The presence of several R hemiclones in the Slovak studies, namely how these triploid male lineages with populations suggests their multiple origins. This pattern different inheritance modes evolutionarily affect the has also been documented for other populations of dynamics of hybrid populations and what happens in water frogs (e.g. Tunner, 1974; Uzzell & Berger, 1975; a contact zone between the two geographic regions' Hotz et al., 2008). Instead of a scenario of ongoing pri- populations where the two lineages may meet in the mary hybridizations between P. lessonae and P. ridibun- same population.
dus, we suppose the existence of several R hemiclonesto be explained rather by past than current primary hybridization events. If primary hybridization wasongoing and common, then we would expect low We thank A. Hoffman, D. Hollinger, M. Kautman and genetic differentiation in R genome between sexual M. Konvicka for help in collecting frogs, S. R€ (RR) and hybrid (LR) genomes, i.e. primary hybridiza- ger for assistance in the laboratory, S.  tion should tend to decrease genetic differentiation with flow cytometry analysis and S. Rauch and Y. Willi between sexual and hemiclonal genomes. Contrary to for helpful input on microsatellite analyses. Specimens this expectation, we have found substantial genetic dif- involved in the crosses are lodged in the European ferentiation (i.e. low gene flow rate) between both Water Frogs Collection of the Zoological Museum of genomes, corroborating results based on AFLP markers the University of Z€urich. Sampling was performed (Mikul
ıcek et al., 2014b).
under the permissions of the Ministry of Environmentof the Slovak Republic No. 6846/06-3.1 and 9303/2009-2.1 and of the Nature Conservation Agency of the General evolutionary implications Hybrid water frog triploids in north-western Europe gratefully acknowledge financial support through Grant and in the Central European area represent indepen- No. 383911 provided by the Charles University Grant Agency (GA UK) and support No. RVO 67985904 by characterized by contrasting inheritance modes. At the Academy of Sciences of the Czech Republic to L.
Choleva, and Grant No. 3100A0-120225/1 by the Swiss Corander, J., Waldmann, P. & Sillanp€a€a, M.J. 2003. Bayesian National Science Foundation to H.-U. Reyer.
analysis of genetic differentiation between populations.
Genetics 163: 367–374.
Corander, J., Marttinen, P., Sir
en, J. & Tang, J. 2008.
Enhanced Bayesian modelling in BAPS software for learn-ing genetic structures of populations. BMC Bioinformatics 9: Alves, M.J., Coelho, M.M. & Collares-Pereira, M.J. 2001. Evo- lution in action through hybridisation and polyploidy in an Cunha, C., Doadrio, I. & Coelho, M.M. 2008. Speciation Iberian freshwater fish: a genetic review. Genetica 111: towards tetraploidization after intermediate processes of non-sexual reproduction. Philos. Trans. R. Soc. Lond. B Biol.
Arioli, M., Jakob, C. & Reyer, H.-U. 2010. Genetic diversity in Sci. 363: 2921–2929.
water frog hybrids (Pelophylax esculentus) varies with popula- Falush, D., Stephens, M. & Pritchard, J.K. 2007. Inference of tion structure and geographic location. Mol. Ecol. 19: population structure using multilocus genotype data: domi- nant markers and null alleles. Mol. Ecol. Notes 7: 574–578.
Avise, J.C. 2008. Clonality – The Genetics, Ecology, and Evolution Garner, T.W.J., Gautschi, B., R€ othlisberger, S. & Reyer, H.-U.
of Sexual Abstinence in Vertebrate Animals. Oxford University 2000. A set of CA repeat microsatellite markers derived from Press, New York, NY.
the pool frog, Rana lessonae. Mol. Ecol. 9: 2173–2175.
Bengtsson, B.O. 2009. Asex and evolution: a very large-scale Gosner, K.L. 1960. A simplified table for staging anuran overview. In: Lost Sex. The Evolutionary Biology of Parthenogen- embryos and larvae with note on identification. Herpetologica on, K. Martens & van Dijk P., eds), pp. 1–19.
16: 183–190.
Springer, Dordrecht, the Netherlands.
Goudet, J. 2001. FSTAT, a program to estimate and test gene Berger, L. 1988a. Principles of studies of European water frogs.
diversities and fixation indices (2.9.3.2.). http://www2.unil.
Acta Zool. Cracov. 31: 563–580.
Graf, J.D. & Polls Pelaz, M. 1989. Evolutionary genetics of the persisting in agrocenoses of central Poland (Amphibia, Sali- Rana esculenta complex. In: Evolution and Ecology of Unisexual entia, Ranidae). Proc. Acad. Nat. Sci. Philadelphia 140: Vertebrates, Vol. 466 (R.M. Dawley & J.P. Bogart, eds), pp.
289–301. New York State University Museum Bulletin, Berger, L., Uzzell, T. & Hotz, H. 1988. Sex determination and sex-ratios in Western Palearctic water frogs – XX and XY Gubanyi, A. & Creemers, R. 1994. Population structure of female hybrids in the Pannonian Basin. Proc. Acad. Nat. Sci.
water frogs in a flood plain of the river Danube (Szigetk€ Philadelphia 140: 220–239.
in Hungary. Zool. Pol. 39: 441–445.
Berger, L., Rybacki, M. & Hotz, H. 1994. Artificial fertilization Guex, G.-D., Hotz, H. & Demlitsch, R.D. 2002. Deleterious of water frogs. Amphib.-Reptil. 15: 408–413.
alleles and differential viability in progeny of natural hemicl- Brookfield, J.F.Y. 1996. A simple new method for estimating onal frogs. Evolution 56: 1036–1044.
null allele frequency from heterozygote deficiency. Mol. Ecol.
unther, R. 1975. Zum nat€ urlichen Vorkommen und zur Mor- 5: 453–455.
phologie triploider Teichfr€ osche "Rana esculenta", L., in der Choleva, L., Apostolou, A., Rab, P. & Janko, K. 2008. Making DDR (Anura, Ranidae). Mitt. Zool. Mus. Berlin 51: 145–158.
it on their own: sperm-dependent hybrid fishes (Cobitis) unther, R. & Pl€ otner, J. 1990. Mating pattern in pure hybrid switch the sexual hosts and expand beyond the ranges of populations of water frogs, Rana kl. esculenta (Anura, Rani- their original sperm donors. Philos. Trans. R. Soc. Lond. B Biol.
dae). Alytes 8: 90–98.
Sci. 363: 2911–2919.
unther, R., Uzzell, T. & Berger, L. 1979. Inheritance patterns Choleva, L., Janko, K., De Gelas, K., Bohlen, J., Slechtova, V., in triploid Rana "esculenta" (Amphibia, Salientia). Mitt. Zool.
Rabova, M. et al. 2012. Synthesis of clonality and polyploidy Mus. Berlin 55: 35–57.
in vertebrate animals by hybridization between two sexual Hardy, O.J. & Vekemans, X. 2002. SPAGeDi: a versatile com- species. Evolution 66: 2191–2203.
puter program to analyse spatial genetic structure at the Christiansen, D.G. 2005. A microsatellite-based method for individual or population levels. Mol. Ecol. Notes 2: 618–620.
genotyping diploid and triploid water frogs of the Rana escul- Hotz, H., Beerli, P. & Spolsky, C. 1992. Mitochondrial-DNA enta hybrid complex. Mol. Ecol. Notes 5: 190–193.
reveals formation of nonhybrid frogs by natural matings Christiansen, D.G. 2009. Gamete types, sex determination and between hemiclonal hybrids. Mol. Biol. Evol. 9: 610–620.
stable equilibria of all-hybrid populations of diploid and trip- Hotz, H., Uzzell, T., Guex, G.-D., Alpers, D., Semlitsch, R.D. & loid edible frogs (Pelophylax esculentus). BMC Evol. Biol. 9: 135.
Beerli, P. 2001. Microsatellites: a tool for evolutionary Christiansen, D.G. & Reyer, H.-U. 2009. From clonal to sexual genetic studies of western Palearctic water frogs. Zoosyst. Evol.
hybrids: genetic recombination via triploids in all-hybrid 77: 43–50.
populations of water frogs. Evolution 63: 1754–1768.
Hotz, H., Guex, G.-D., Beerli, P., Semlitsch, R.D. & Pruvost, Collares-Pereira, M.J., Matos, I., Morgado-Santos, M. & Coel- N.B.M. 2008. Hemiclone diversity in the hybridogenetic frog ho, M.M. 2013. Natural pathways towards polyploidy in ani- Rana esculenta outside the area of clone formation: the view mals: the Squalius alburnoides fish complex as a model from protein electrophoresis. J. Zool. Syst. Evol. Res. 46: system to study genome size and genome reorganization in polyploids. Cytogent. Genome Res. 140: 97–116.
Jakob, C., Arioli, M. & Reyer, H.-U. 2010. Ploidy composition Corander, J. & Marttinen, P. 2006. Bayesian identification of in all-hybrid frog populations in relation to ecological condi- admixture events using multi-locus molecular markers. Mol.
tions. Evol. Ecol. Res. 12: 633–652.
Ecol. 15: 2833–2843.
Reproductive strategies in triploid frogs Janko, K., Kotusz, J., De Gelas, K., Slechtova, V., Opoldusova, Pruvost, N.B.M., Hollinger, D. & Reyer, H.-U. 2013b. Geno- Z., Drozd, P. et al. 2012. Dynamic formation of asexual dip- type-temperature interactions on larval performance shape loid and polyploid lineages: multilocus analysis of Cobitis population structure in hybridogenetic water frogs (Pelophy- reveals the mechanisms maintaining the diversity of clones.
lax esculentus complex). Funct. Ecol. 27: 459–471.
PLoS One 7: e45384.
Rybacki, M. & Berger, L. 2001. Types of water frog populations Kearney, M., Fujita, M.K. & Ridenour, J. 2009. Lost sex in the (Rana esculenta complex) in Poland. Mitt. Mus. Nat. Berlin, reptiles: constraints and correlations. In: Lost Sex. The Evolu- Zool. Reihe 77: 51–77.
tionary Biology of Parthenogenesis (I. Sch€ on, K. Martens & van Schmeller, D.S., Seitz, A., Crivelli, A. & Veith, M. 2005. Cross- Dijk P., eds), pp. 447–474. Springer, Heidelberg.
ing species' range borders: interspecies gene exchange medi- Lamatsch, D.K. & St€ ock, M. 2009. Sperm-dependent partheno- ated by hybridogenesis. Proc. Biol. Sci. 272: 1625–1631.
genesis and hybridogenesis in teleost fishes. In: Lost Sex. The Schmidt, D.J., Bond, N.R., Adams, M. & Hughes, J.M. 2011.
Evolutionary Biology of Parthenogenesis (I. Sch€ Cytonuclear evidence for hybridogenetic reproduction in van Dijk P., eds), pp. 399–432. Springer, Heidelberg.
natural populations of the Australian carp gudgeon (Hypsel- Maynard-Smith, J. 1992. Age and the unisexual lineage. Nat- eotris: Eleotridae). Mol. Ecol. 20: 3367–3380.
ure 356: 661–662.
on, I., Martens, K. & van Dijk, P., eds. 2009. Lost Sex. The Mikul
ıcek, P. & Kotl
ık, P. 2001. Two water frog populations Evolutionary Biology of Parthenogenesis. Springer, Heidelberg.
from western Slovakia consisting of diploid females and dip- Sinclair, E.A., Pramuk, J.B., Bezy, R.L., Crandall, K.A. & Sites, loid and triploid males of the hybridogenetic Rana esculenta J.W. Jr 2010. DNA evidence for nonhybrid origins of parthe- (Anura, Ranidae). Mitt. Zool. Mus. Berlin 77: 59–64.
nogenesis in natural populations of vertebrates. Evolution 64: Mikul
ıcek, P., Kautman, M., Kautman, J. & Pruvost, N.B.M.
2014b. Mode of hybridogenesis and habitat preferences Som, C. & Reyer, H.-U. 2006. Demography and evolution of influence population composition of water frogs (Pelophylax pure hybridogenetic frog (Rana esculenta) populations. Evol.
esculentus complex, Anura: Ranidae) in a region of sympatric Ecol. Res. 8: 1235–1248.
occurrence (western Slovakia). J. Zool. Syst. Evol. Res. doi: 10.
Stenberg, P. & Saura, A. 2009. Cytology of asexual animals.
In: Lost Sex. The Evolutionary Biology of Parthenogenesis (I.
Mikul
ıcek, P., Kautman, M., Demovic, B. & Janko, K. 2014a.
on, K. Martens, P. van Dijk, eds), pp. 63–74. Springer, When a clonal genome finds its way back to a sexual species: evidence from ongoing but rare introgression in the hybrido- ock, M., Lampert, K.P., Moeller, D., Schlupp, I. & Schartl, genetic water frog complex. J. Evol. Biol. 27: 628–642.
M. 2010a. Monophyletic origin of multiple clonal lineages in Moritz, C., Brown, W.M., Densmore, L.D., Wright, J.W., an asexual fish (Poecilia formosa). Mol. Ecol. 19: 5204–5215.
Vyas, D., Donnellan, S. et al. 1989. Genetic diversity and ock, M., Ustinova, J., Lamatsch, D.K., Schartl, M., Perrin, N.
the dynamics of hybrid parthenogenesis in Cnemidophorus & Moritz, C. 2010b. A vertebrate reproductive system (Teiidae) and Heteronotia (Gekkonidae). In: Ecology and Evo- involving three ploidy levels: hybrid origin of triploids in a lution of Unisexual Vertebrates, Vol. 466 (R.M. Dawley & J.P.
contact zone of diploid and tetraploid Palearctic green toads Bogart, eds), pp. 87–112. New York State Museum, Albany, (Bufo viridis subgroup). Evolution 64: 944–959.
ock, M., Ustinova, J., Betto-Colliard, C., Schartl, M., Moritz, Neaves, W.B. & Baumann, P. 2011. Unisexual reproduction C. & Perrin, N. 2012. Simultaneous Mendelian and clonal among vertebrates. Trends Genet. 27: 81–88.
genome transmission in a sexually reproducing, all-triploid Nei, M. 1978. Estimation of average heterozygosity and genetic vertebrate. Proc. Biol. Sci. 279: 1293–1299.
distance for small number of individuals. Genetics 89: Suomalainen, E., Saura, A. & Lokki, J. 1987. Cytology and Evo- lution in Parthenogenesis. CRC Press, Boca Raton, FL.
Otto, S.P. & Lenormand, T. 2002. Resolving the paradox of sex Tunner, H.G. 1974. Die klonale Struktur einer Wasserfrosch- and recombination. Nat. Rev. Genet. 3: 252–261.
population. Z. Zool. Syst. Evol. 12: 309–314.
Parker, E.D. Jr & Niklasson, M. 1999. Genetic structure and Tunner, H.G. 1980. Kreuzungsexperimente mit Wasserfr€ evolution in parthenogenetic animals. In: Evolutionary Genet- osterreichischen und polnischen Mischpopulationen ics: from Molecules to Morphology (R.S. Singh, C.B. Krimbas, (Rana lessonae + Rana esculenta). Z. Zool. Syst. Evol. 18: 257–297.
eds), pp. 456–474. Cambridge University Press, Cambridge.
Tunner, H.G. 2000. Evidence for genomic imprinting in uni- Peakall, R. & Smouse, P.E. 2006. GENALEX 6: genetic analysis sexual triploid hybrid frogs. Amphib.-Reptil 21: 135–141.
in Excel. Population genetic software for teaching and Tunner, H.G. & Dobrowsky, M.T. 1976. Morphological, sero- research. Mol. Ecol. Notes 6: 288–295.
logical and enzymological differentiation of Rana lessonae and otner, J. 2005. Die westpal€aarktischen Wasserfr€osche. Laurenti- hybridogenetic Rana esculenta from Seewinkel and Neusie- Verlag, Bielefeld.
dlersee (Austria, Burgenland). Zool. Anz. 197: 6–22.
Polls Pelaz, M. 1994. Modes of gametogenesis among kleptons Tunner, H.G. & Heppich-Tunner, S. 1992. A new population of the hybridogenetic water frog complex: an evolutionary system of water frogs discovered in Hungary. In: Proc. 6th synthesis. Zool. Pol. 39: 123–128.
Ord. Gen. Meet. Soc. Europ. Her. pp. 453–460. 19–23 August Pritchard, J.K., Stephens, M. & Donnelly, P. 2000. Inference of 1991. Budapest, Hungary.
population structure using multilocus genotype data. Genetics Uzzell, T. & Berger, L. 1975. Electrophoretic phenotypes of Rana 155: 945–959.
ridibunda, Rana lessonae, and their hybridogenetic associate, Pruvost, N.B.M., Hoffmann, A. & Reyer, H.-U. 2013a. Gamete Rana esculenta. Proc. Acad. Nat. Sci. Philadelphia 127: 13–24.
production patterns, ploidy, and population genetics reveal Uzzell, T., Berger, L. & G€ unther, R. 1975. Diploid and triploid evolutionary significant units in hybrid water frogs (Pelophy- progeny from a diploid female of Rana esculenta (Amphibia lax esculentus). Ecol. Evol. 3: 2933–2946.
Salientia). Proc. Acad. Nat. Sci. Philadelphia 127: 81–91.
Van Oosterhout, C., Hutchinson, W.F., Wills, D.P.M. & Ship- Zeisset, I., Rowe, G. & Beebee, T.J.C. 2000. Polymerase chain ley, P. 2004. Microchecker: software for identifying and cor- reaction primers for microsatellite loci in the north European recting genotyping errors in microsatellite data. Mol. Ecol.
water frogs Rana ridibunda and R. lessonae. Mol. Ecol. 9: Notes 4: 535–538.
Vinogradov, A.E., Borkin, L.J., G€unther, R. & Rosanov, J.M.
1990. Genome elimination in diploid and triploid Rana escul-enta males – Cytological evidence from DNA flow cytometry.
Supporting information Genome 33: 619–627.
Additional Supporting Information may be found in the Vorburger, C. 2001a. Fixation of deleterious mutations in clo- online version of this article: nal lineages: evidence from hybridogenetic frogs. Evolution55 Data S1 DNA microsatelitte data file for the 16 loci : 2319–2332.
Vorburger, C. 2001b. Non-hybrid offspring from matings used in the study (RlCA5 and Res16 were not used for between hemiclonal hybrid water frogs suggest occasional any of the analyses).
recombination between clonal genomes. Ecol. Lett. 4: 628–636.
Data S2 Genome sizes measured by flow cytometry Vrijenhoek, R. 1979. Factors affecting clonal diversity and based on blood (b) and sperm (s) samples.
coexistence. Am. Zool. 19: 787–797.
Data S3 Gene diversity (He) corrected for sample size Vrijenhoek, R.C. 1989. Genetic and ecological constraints on (Nei, 1978) for P. lessonae and P. ridibundus genomes in the origins and establishment of unisexual vertebrates. In: sexual and hybrid genotypes sampled in populations.
Evolution and Ecology of Unisexual Vertebrates, Vol. 466 (R.M.
Data S4 Hemiclonal and single multilocus genotypes Dawley & J.P. Bogart, eds), pp. 24–31. New York State Uni- (MLGs) in four genotypes (LL, LR, LLR and RR) from versity Museum Bulletin, Albany, NY.
nine study populations (Pop.).
Vrijenhoek, R.C. 1999. Parthenogenesis and natural clones. In: Encyclopedia of Reproduction, Vol. 3 (E. Knobil & J.D. Neill, Three not mutually exclusive hypotheses to eds), pp. 695–702. Academic Press, San Diego, CA.
explain how diploid LR hybrids are maintained in pop- Vrijenhoek, R.C., Dawley, R.M., Cole, C.J. & Bogart, J.P. 1989.
ulations types PT3 (LR females and LLR males) and PT4 A list of the known unisexual vertebrates. In: Evolution and (LR males and females, LLR males and RR males and Ecology of Unisexual Vertebrates, Vol. 466 (R.M. Dawley & J.P.
Bogart, eds), pp. 19–23. New York State University MuseumBulletin, Albany, NY.
Data deposited at Dryad: doi:10.5061/dryad.hn4tn Waits, L.P., Luikart, G. & Taberlet, P. 2001. Estimating the Received 4 December 2013; accepted 17 November 2014 populations: cautions and guidelines. Mol. Ecol. 10: 249–256.
Weir, B.S. & Cockerham, C.C. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38: 1358–1370.

Source: http://www.uli-reyer.ch/index_htm_files/Pruvost%20et%20al%20(2014).pdf