Because most clonal vertebrates have hybrid genomic constitutions, tight linkages are assumed among hybridization, clonality, and polyploidy. However, predictions about how these processes mechanistically relate during the switch from sexual to clonal reproduction have not been validated. Therefore, we performed a crossing experiment to test the hypothesis that interspecific hybridization per se initiated clonal diploid and triploid spined loaches (Cobitis) and their gynogenetic reproduction. We reared two F1 families resulting from the crossing of 14 pairs of two sexual species, and found their diploid hybrid constitution and a 1:1 sex ratio. While males were infertile, females produced unreduced nonrecombinant eggs (100%). Synthetic triploid females and males (96.3%) resulted in each of nine backcrossed families from eggs of synthesized diploid F1s fertilized by haploid sperm from sexual males. Five individuals (3.7%) from one backcross family were genetically identical to the somatic cells of the mother and originated via gynogenesis; the sperm of the sexual male only triggered clonal development of the egg. Our reconstruction of the evolutionary route from sexuality to clonality and polyploidy in these fish shows that clonality and gynogenesis may have been directly triggered by interspecific hybridization and that polyploidy is a consequence, not a cause, of clonality.
The molecular and genetic machinery of meiosis is highly conserved among eukaryotes (Cavalier-Smith 2002; Bengtsson 2009). However, the existence of asexual animals and plants suggests that the recombinant component of meiosis has been disrupted or circumvented multiple times during evolution, an idea that has attracted much theoretical and practical attention (Schön et al. 2009; see Fig. 1 for definitions of sexual reproduction and of the main types of nonsexual reproduction). Molecular genetic studies of extant clonal vertebrates have shown that they are often polyploid and most have genomes combined from at least two ancestral sexual species (Lamatsch and Stöck 2009; Kearney et al. 2009). These findings suggest a causal link between hybridization (and maybe even polyploidization) and clonal reproduction (Schultz 1969; Beukeboom and Vrijenhoek 1998; Lampert 2009a; Neaves and Baumann 2011). However, the correlation between vertebrate clonality and the hybrid state or polyploidy itself in contemporary populations is insufficient for understanding how the latter two phenomena mechanistically relate to the initiation of clonal reproduction. As stated by Kearney et al. (2009), “there are two very different ways to view the link between hybridization and parthenogenesis; either parthenogenesis provides a mean to the end of an advantageous hybrid state, or the hybrid state itself provides a means to become parthenogenetic.”
Uncertainty in this issue has been strengthened by many unsuccessful attempts to synthesize clonal vertebrates (parthenogenetic or gynogenetic, Fig. 1) by crossing the putatively ancestral sexual species, which were identified by phylogenetic analyses. Although laboratory hybridizations have produced hybrid progeny in many cases, these progeny are generally not capable of clonal reproduction because they are either nonviable or infertile or display “normal” Mendelian segregation (Moritz 1991; Murphy et al. 2000; Dries 2003; Lampert et al. 2007; Cole et al. 2010; Stöck et al. 2010), with two exceptions: the fish Poeciliopsis (Schultz 1973; Wetherington et al. 1987; Vrijenhoek 1993, 1994) and the frog Pelophylax (Berger 1968; Hotz et al. 1985). Both vertebrates, however, are hybridogenetic and have hemiclonal inheritance, meaning half of an individual's genome is inherited clonally and the other half sexually (Fig. 1). Rare cases of the laboratory synthesis of parthenogens from bisexual ancestors can be found in invertebrate animals: namely, the grasshopper Warramaba (White et al. 1977; White and Contreras 1978) and planthoppers of the genus Muellerianella (Drosopoulos 1978). It is therefore essential to find appropriate laboratory models among the vertebrate animals to test empirically how interspecific hybridization, clonality and polyploidy are causally related (Simon et al. 2003; Mable 2004; Comai 2005, Hörandl 2009; Kearney et al. 2009) and to understand the circumstances and mechanisms that enable this escape from the paradigm of sexual reproduction.
The hybrid origin hypothesis (HOH; Fig. 2A) postulates that clonality is initiated by an interbreeding between two gonochoristic species. The unreduced eggs formed by such hybrid progeny are capable of starting development without subsequent true fertilization. According to the HOH, the hybrid genomic constitution of clonal organisms originates simultaneously with the origin of clonality; thus, all of the necessary steps to switch from sexual to clonal reproduction occur at once, not independently (Schultz 1969). Another explanation for the origin of clonality is offered by the spontaneous or mutation origin hypothesis (MOH; Fig. 2B). This model assumes a nonhybrid origin of clonality, wherein meiosis might be altered by the accumulation of spontaneous mutation(s) in genes required for sex within a single gonochoristic species (Cuellar 1974, 1977; Schurko and Logsdon 2008; Sinclair et al. 2010). Under the MOH, the hybrid state may result from secondary changes in the genomes of established clonal lineages, such as sperm incorporation from a nonparental sexual species (hybridization as a consequence of parthenogenesis). An illustration of the difficulties in reconstructing the evolution of asexual reproduction using only molecular genetic data from contemporary populations comes from the Squalius alburnoides complex. It has been suggested that this hybrid complex has experienced multiple dynamic origins of asexuality (Cunha et al. 2004), but this conclusion has been questioned (Sousa-Santos et al. 2006) by the hypothesis that the apparent polyphyly might have arisen via the postformational introgression of genetic material into an asexual lineage of monophyletic origin. Moreover, a putative ancestor of the entire complex (Carmona et al. 1997) was later identified as a “unisexual” biotype of hybrid origin (Alves et al. 2001).
The two contrasting scenarios, the primary HOH and the spontaneous origin hypothesis, both take into account the fact that most clonal vertebrates are polyploid (Vrijenhoek 1989) and suggest that polyploidization has played a major role in the establishment of clonal reproduction. Within the primary hybrid origins of clonality, the genome addition hypothesis (GAH; [Schultz 1969]; Fig. 2C) denies the direct impact of polyploidization on the formation of clonality, but instead postulates that clonality originates first in a diploid hybrid progeny, and polyploidy arises secondarily by the fertilization of an unreduced heterospecific ovum by a sperm from a sexual male (Avise et al. 1992; Beukeboom and Vrijenhoek 1998). Similarly, the genome duplication hypothesis (GDH; Fig. 2D) suggests that suppression of an equational division in a diploid hybrid could produce unreduced homospecific ova which, if fertilized by a male from a sexual species, would produce triploid progeny (Quattro et al. 1992); alternatively, there could also be direct endogenous production of a triploid ovum by a diploid hybrid female (Vrijenhoek 1994). In contrast, the spontaneous polyploid origin hypothesis (Fig. 2E) suggests that clonality may be initiated by polyploidization (Beukeboom and Vrijenhoek 1998). This polyploidization may be achieved by the fertilization of an unreduced egg, which are occasionally produced by diploid sexual females, thus skipping the intermediate diploid hybrid progeny (Cuellar 1974, 1977). Indeed, the formation of unreduced oocytes is not a rare event in vertebrates (Lampert 2009b).
In this study, we experimentally tested the role of hybridization and polyploidization during a process by which sexual reproduction reverses into clonality in spined loaches (Cobitis), a European bottom-dwelling freshwater fish. Previous molecular phylogenetic approaches have revealed that the two most widespread sexual species, C. taenia and C. elongatoides, produced diploid and polyploid clonal lineages by reciprocal hybridization during the Holocene (Janko et al. 2003, 2007a, b). These unisexual clones are females that produce progeny by gynogenesis; therefore, they must live in sympatry with sexual species that serve as a sperm host (Vasil'ev et al. 1989; Saat 1991; Janko et al. 2007b). Diploid clones (C. elongatoides-taenia) are restricted to the proximity of the hybrid zone in Central Europe, and it has been experimentally shown that the sperm from the sexual species fuse with unreduced eggs from diploid clonal female in approximately 1/3 of cases, resulting in polyploid progeny, whereas 2/3 progeny remain diploid (Janko et al. 2007b). Triploid biotypes of either C. 2elongatoides-1taenia or C. 1elongatoides-2taenia genomic compositions constitute the dominant forms distributed across the entire area of parental species. The triploids apparently arise by the incorporation of sperm of either parental species (Bohlen and Ráb 2001; Janko et al. 2005, 2007a, b). The rarely found hybrid males are infertile, but as in other asexual complexes, the reason for their infertility is unknown (Vasil'ev et al. 2003). Here, we reconstruct the evolutionary route from sexuality to clonality and polyploidy in vertebrate animals under nature-simulating conditions. Our analysis provides support for the theoretical predictions of the HOH and GAH hypotheses. We also discuss the phenomenon of male hybrid infertility that we observed in our synthesized hybrid progeny.
Materials and Methods
In 1999, we collected 38 individuals from three geographically distant natural populations of the two sexual species of spined loaches (Fig. 3): seven C. elongatoides males and five C. elongatoides females from Pšovka Creek, Czech Republic (50°22'12″N; 14°33'7″E), two C. elongatoides males from the Krka River, Slovenia (45°47'27″N; 14°58'40″E), and five C. taenia males and nine C. taenia females from Haaren Creek, Germany (53°09'N; 08°06'E). All diploid C. elongatoides-taenia, as well as triploid C. 2elongatoides-1taenia and C. 1elongatoides-2taenia hybrids, were synthesized as laboratory progeny.
DESIGN AND MAINTENANCE OF CROSSING EXPERIMENTS
All fish were raised and maintained under natural light and temperature. The design of the spawning tanks and breeding conditions followed those of Bohlen (1999). Beginning in 2000, we performed reciprocal crosses and mated one male and one female between parental C. elongatoides and C. taenia in 14 separate tanks so that the origin of vital progeny from reciprocal crosses could be tested and F1 hybrids produced (Table 1). The selected progeny were either kept for subsequent crossing experiments (see below) or reared until yolk sac resorption occurred (approximate body length 8 mm) and then used for genetic analyses. For logistical reasons (the experiment spanned 9 years), it was not possible to keep all families until maturity; thus, only randomly selected members of two F1 families (referred to as families F1–1 and F1–2) were raised until maturity for subsequent crossings. Females usually reach maturity in the third year of life (i.e., later than males; Juchno and Boroń 2006). Both families originated from two breeding pairs consisting of C. taenia females from Haaren and C. elongatoides males from Krka. Consequently, we do not have laboratory data on the origin of clonality and polyploidy from the opposite direction of crosses, which are known to occur in natural populations.
Table 1. Summary of crossing experiments with the description of each pair and their progeny.
Sum of laid eggs per clutch/Sum of developed eggs into progeny
Number of analyzed Progeny
ID = internal identification number of the specimens; TT =C. taenia; EE =C. elongatoides; ET = synthesized diploid hybrid specimen C. elongatoides-taenia. An asterisk indicates crosses from which no offspring were analyzed.
215/0, 254/0, 106/0
20/5, 175/53, 47/12, 51/10, 77/22
130/0, 122/0, 38/0, 48/0
146/0, 71/0, 69/0, 81/0, 8/0
250/0, 280/0, 98/0
40/30, 46/46, 49/49, 27/26
220/0, 145/0, 115/0, 307/0, 119/0
148/0, 25/0, 90/0
277/0, 223/0, 205/0, 129/0
162/0, 96/0, 123/0
To permit testing of the fertility of hybrids and to examine the emergence of clonal reproduction and polyploidy, we performed a set of crossing experiments between 2004 and 2008. The randomly selected pairs of synthesized F1 hybrids were mated together, and after up to five spawnings, F1 males were replaced with C. elongatoides males from Pšovka Creek and/or C. taenia males from Haaren Creek (Table 1). Progeny were produced only when F1 females were backcrossed with sexual males (Table 1). The offspring, which were analyzed genetically, are referred to as families B-1 to B-9 (Tables S1 and S2). In 2009, we mated four pairs of C. elongatoides Pšovka females and F1 males, and after two spawnings, we replaced the F1 males with C. taenia Haaren males. The offspring from 2008 and 2009 were not analyzed genetically.
ALLOZYME AND MICROSATELLITE TYPING
Fin clips or muscle tissue samples from parents and offspring from 2000 to 2007 seasons were stored frozen at −80°C in buffer and analyzed by horizontal starch electrophoresis for the diagnostic allozyme loci glucose-6-phosphate isomerase (Gpi-A, EC 18.104.22.168), aspartate aminotransferase (sAat, EC 22.214.171.124), malate dehydrogenase (sMdh-A, EC 126.96.36.199), and phosphoglucomutase (Pgm, EC 188.8.131.52), following the protocols of Šlechtová et al. (2000). Electrophoretic mobility was compared to previously analyzed standards. These allozyme loci constitute fully informative species-specific genetic markers, allowing unambiguous identification of sexual species and their hybrids, which produce heterozygous patterns (Šlechtová et al. 2000; Janko et al. 2007a; Choleva et al. 2008).
The muscle tissue of individuals crossed in 2007 was preserved in 100% ethanol for microsatellite analysis. Genomic DNA was extracted from the tissue using Nucleospin extraction kits (Machery-Nagel GmBH). In total, 11 recently developed microsatellite loci were analyzed for genetic polymorphisms. Nine of these loci were previously described by De Gelas et al. (2008). Two additional loci, cota_010 (GenBank Acc. No. JN034033) and cota_068 (GenBank Acc. No. JN034034), were included and added to polymerase chain reaction (PCR) multiplex 1 (De Gelas et al. 2008). The primer concentrations in the 10-μl PCR reactions were 0.025 and 0.05 μM for cota_010 and cota_068, respectively. The reaction conditions and primer concentrations of the other primers were identical to those described by De Gelas et al. (2008). The PCR products were run on an ABI 3130 Genetic Analyzer (Life Technologies, Carlsbad, CA, USA) and scored relative to a Genescan 500 LIZ size standard using Genemapper v. 3.7 (Life Technologies, Carlsbad, CA, USA).
Genotyping of parents and offspring was performed to study the inheritance patterns and to disentangle hybridization and polyploidization pathways and differentiate between modes of reproduction (i.e., sexual or clonal). We assumed that the paternal genome was incorporated into the progeny (1) if the father and offspring shared alleles not present in the mother, and, for some loci, (2) if we observed in the progeny approximately double gene doses of those allozyme alleles that were shared between the mother and father. An increase in the staining intensity of shared paternal and maternal allozyme alleles has previously been used as a marker of successful genome incorporation into triploids (Vrijenhoek 1975; Šlechtová et al. 2000; Janko et al. 2007a). Moreover, analysis by flow cytometry has demonstrated that such a gene dosage effect in allozyme data are suitable for differentiating between diploid and triploid forms of spined loaches (Janko et al. 2007a). For the microsatellite analysis, the dosage effect or a variant could not be applied due to differences in amplification intensity among the different species at some loci. However, the microsatellites used were far more variable than the allozymes. Hence, the alleles from the father and mother differed completely at one or more loci, rendering the designation of the parental contribution of alleles to the offspring straightforward. We used chi-squared analysis to test the significance of the sex ratio in F1 progeny and the rate of triploidization among progeny of different females.
SYNTHESIS OF F1 STRAINS
Reciprocal combinations between sexual species C. elongatoides and C. taenia produced developing eggs and vital offspring in all 14 experiments from the year 2000 (Tables 1, S1, and S2). Two F1 families selected at random were raised to maturity in 2003, and their resulting sex ratio was not significantly different from a 1:1 ratio (P= 0.5 using the chi-squared test). All of the 74 analyzed F1 individuals were heterozygous at all diagnostic loci and contained one allele from the sexual mother and one allele from the sexual father, clearly confirming their hybrid constitution. The allozyme staining at heterozygous loci was balanced, and we never observed more than two microsatellite alleles per locus, again supporting the diploid state of the F1 hybrids (Tables S1 and S2).
BACKCROSSES INVOLVING F1 FEMALES
All crosses involving 11 F1 hybrid females with sexual males (seven C. elongatoides, with one male used twice; four C. taenia) produced vital progeny (Table 1). The backcross progeny from nine families were subjected to genetic analysis (families B-1 to B-9; Tables S1 and S2) and were found to express a complete set of alleles identical to that of the somatic cells of the mother (100%), suggesting the production of diploid nonrecombinant eggs by all F1 females. Five (3.7%) of the 135 analyzed backcross progeny expressed exclusively maternal alleles; however, 130 (96.3%) specimens also expressed an allele characteristic of the father in addition to the maternal pattern described above, which is suggestive of true fertilization. Consistently, most offspring possessed three alleles of the studied loci, of which one was always attributable to the father, and, simultaneously, we observed a double-dose intensity of the allozyme alleles shared between the mother and father (Tables S1 and S2). This double-dose signal of shared parental allozyme alleles alone (i.e., the absence of three different alleles) was found in 11 triploid individuals. At the loci where the father appeared heterozygous, its progeny usually segregated for both paternal alleles. This finding suggests that the incorporation of haploid sperm resulted in elevated ploidy (Tables S1 and S2). Males from both sexual species were able to induce triploidization in progeny. Two exceptions were observed. First, one backcross individual from the B-9 family possessed an allele that was not present in either of the parents (allele no. 374 at microsatellite locus cota_006; Table S2). The second exception was the B-8 family at locus cota_037. Here, the father appeared homozygous for allele no. 272 at the microsatellite locus; however, 23 of its progeny did not possess this allele, whereas at other loci, they segregated for both paternal alleles, often displaying triallelic combinations.
In eight of the nine backcross families (89%), we observed signs of sperm incorporation in 100% of the progeny, suggesting complete triploidization of the progeny. The B-3 family was different because only five of the 10 analyzed progeny (50%) were triploid. In addition, they expressed a unique paternal allele at the locus Gpi-A and exhibited double-dose staining for alleles that were identical to those in the mother and father, namely in loci sAat and sMdh-A. The other half of B-3 progeny did not exhibit any sign of paternal genome incorporation and had a diploid constitution with an identical genetic pattern to that of the mother, which is an indication of gynogenetic reproduction with exclusion of the paternal genome.
BACKCROSSES INVOLVING F1 MALES
Crosses involving 11 F1 hybrid males and 11 F1 hybrid females resulted in 4519 laid eggs, of which only 15 exhibited some level of development and hatching, although none survived longer than 2 days (Table 1). The F1 males also never produced any offspring from 908 laid eggs when mated with four sexual females, although the same females produced progeny when they were mated with sexual C. taenia (Table 1).
HYBRIDIZATION TRIGGERS CLONAL REPRODUCTION
Our experimental approach studied the origin of and transition rates to clonality and polyploidy from sexual reproduction by crossing the parental species C. elongatoides and C. taenia, which are also thought to have formed naturally occurring diploid and triploid unisexual gynogenetic hybrids. The synthesized F1 hybrid females produced diploid nonrecombinant eggs based on allozyme and microsatellite patterns. Because all alleles from the F1 mothers were inherited (100%) and no loss of heterozygosity was observed in the backcross progeny, the most likely explanation for the formation of the diploid eggs with unchanged chromosome contents is the premeiotic doubling of chromosomes or apomictic reproduction. To achieve unreduced ploidy and a nonrecombinant DNA content in the eggs of clonal vertebrates, chromosomes can be doubled either by two rounds of premeiotic endoreplication without an intervening mitosis or by the fusion of two diploid oogonial cells (Neaves and Baumann 2011). The important maintenance of fixed heterozygosity over many generations along a clonal lineage is achieved by the homologous pairing of sister chromosomes instead of homologs during meiosis (Neaves and Baumann 2011). In apomixis, unreduced eggs are produced ameiotically via mitotic divisions (Hughes 1989; Lampert et al. 2007). However, the cytological mechanism responsible for the production of nonrecombinant eggs in hybrid spined loaches remains unknown and merits future study.
The formation of diploid nonrecombinant eggs (either via premeiotic doubling or apomixis), which is spontaneously initiated in synthesized spined loaches by interspecific hybridization, suggests that the same mechanism is also likely to occur in natural populations of spined loaches and in other clonal vertebrates. Similar to hybrid spined loaches captured in the field (Janko et al. 2007b), most clonal vertebrates are of hybrid origin (Vrijenhoek 1989; Kearney et al. 2009; Lamatsch and Stöck 2009) and exhibit a fixed F1 hybrid state without an indication of loss of heterozygosity in multiple nuclear markers, for which they are sometimes referred to as “frozen F1s” (Vrijenhoek 1979, 1998). Alternatively, animals with automictic reproduction have usually maintained meiosis, and the diploid chromosome number in the egg is typically restored through a duplication or by the random fusion of meiotic products. This process can lead to variable offspring because crossing over takes place between homologs instead of between sister chromosomes (Suomalainen et al. 1987). Therefore, automixis, which often causes a loss of fixed heterozygosity and occurs, for example, during oocyte formation in synthesized interspecific hybrids of poeciliid fish (Lampert et al. 2007), appears to be an unlikely explanation for our results.
Our conclusion about the dominant role of premeiotic doubling or apomixis during the production of nonrecombinant eggs in laboratory-born spined loaches is not violated by the two exceptions mentioned in the Results section. First, the occurrence of a new allele in one individual in backcross B-9 may have resulted from a mutation in the tested microsatellite marker, a situation that has also been reported in the lizard Darevskia unisexualis (Badaeva et al. 2008). Second, in the case of the allele at locus cota_037 that was missing in approximately half of progeny in the B-8 family, this finding most likely results from the father possessing a null allele at locus cota_037.
The easy synthesis of F1 hybrid females with nonrecombinant eggs points to relatively high transition rates to clonality from sexuality in spined loaches. We achieved this genesis of clonality in the laboratory using natural test-crosses; that is, we simply put males and females together from previously identified sexual parental species without any enforced crossing conditions or any of the genetic or physiological manipulations that are usually used for the artificial induction of fish clonality in aquaculture (Lampert 2009a). Moreover, our laboratory-born F1 females were obtained by crossing sexual C. elongatoides males and C. taenia females that were field-captured from randomly selected, geographically distant populations (Slovenia and Northern Germany, respectively) far away from the contact hybrid zone (Fig. 3). It is therefore likely that clonality in hybrid spined loaches is not restricted to any geographically isolated populations but arises whenever parental species come into reproductive contact. These experimental data are consistent with a previous interpretation based on molecular phylogenic analysis (Janko et al. 2005); that is, that the formation of clonality in spined loaches is an ongoing polyphyletic process and that distinct clones arise on multiple occasions. In contrast, molecular genetic studies of other clonal vertebrates such as the lizards Aspidoscelis (Cole et al. 2010) and Lepidophyma (Sinclair et al. 2010) and the fish Fundulus (Hernández et al. 2007), Phoxinus (Angers and Schlosser 2007), Poecilia (Lampert et al. 2007; Stöck et al. 2010), and triploid Poeciliopsis (Quattro et al. 1992; Mateos and Vrijenhoek 2005) found that clonal strains harbor a limited subset of the genetic variability of their sexual ancestors, probably reflecting strong geographic or genetic limitations on the induction of clonal reproduction (Mateos and Vrijenhoek 2005).
Taken together, our experimental evidence for the origin of clonal reproduction in vertebrates indicates that clonality is a direct consequence of hybridization in spined loaches and clearly supports the concept of the HOH (Figs. 2A and 4A–C). Here, the combination of two different genomes in a hybrid can modify gametogenesis and lead to the production of unreduced eggs capable of starting development without subsequent true fertilization. In this manner, hybridization per se may induce clonal reproduction and initiate the development of a clonal lineage.
The potential for the dynamic recruitment of “fresh” clones may also have important evolutionary implications for the maintenance of sexual versus asexual reproduction. This influx of diversity may aid asexual individuals in the arms race with coevolving pathogens (King et al. 2011) or even buffer the long-term disadvantages of asexuality (e.g., deleterious mutation accumulation [Paland and Lynch 2006], thus potentially maintaining asexuality over much longer periods than the life spans of individual clonal lineages. However, the genetic variability of Cobitis clones does not notably deviate from neutral expectations [Janko et al. 2011]. Hence, the maintenance of clonal diversity may also be governed by the stochastic turnover of clones (Janko et al. 2008), a process analogous to the equilibria between mutation and drift or between speciation and extinction inherent in neutral theories of genetics and macroecology (Kimura and Crow 1964; Hubbell 2001). The possibility that we have discovered of the recurrent formation of new clonal strains implies that such an equilibrium could be established. In any case, the rate of formation of new clones plays a crucial role in our understanding of the evolution of asexual complexes (Butlin et al. 1999; Janko et al. 2008; Vrijenhoek and Parker 2009), and the present study will be helpful in parameterizing explicit models of the maintenance of asexuality.
EVOLUTION OF GYNOGENESIS
Our experimental hybridizations not only initiated the production of nonrecombinant eggs but also resulted in the de novo origin of clonal progeny (3.7%) in the family B-3, which indicates how the transition from sexual reproduction to clonal reproduction could have occurred in natural strains of gynogenetic C. taenia-elongatoides hybrids. Half of the synthesized progeny from the B-3 family possessed a true clonal copy of the genotype of the F1 mother and did not inherit any paternal genetic contribution. Two different modes of unisexual reproduction could explain this genotype pattern: parthenogenesis or gynogenesis. Parthenogenesis, or “virgin birth,” is an unlikely modus operandi of reproduction in diploid backcross spined loach clones because obligate parthenogens have hitherto only been identified among vertebrates in reptiles (Kearney et al. 2009), and we have never observed egg development in spined loaches without male attendance. Gynogenesis is further implied by the fact that all F1 females produced polyploid progeny in which the paternal contribution was detectable, hence demonstrating reproductive contact between each mated pair. Moreover, all females had visible spawning marks on their bodies, which result from the specific male behavior of making a ring around the female during spawning in spined loaches (Bohlen 2008). We conclude that the sperm of the sexual male triggered the diploid nonrecombinant egg to undergo clonal development in half of the B-3 progeny; subsequently, the sperm DNA was degraded without true fertilization, and thus the male DNA made no genetic contribution to the developing embryos.
How gynogenesis evolves is an important subject in evolutionary biology (Schlupp 2005; Rogers and Vamosi 2010; Stöck et al. 2010; Neaves and Baumann 2011). It is generally assumed that genomic changes must occur to establish reproduction via gynogenesis (Vrijenhoek 1994; Schlupp 2005). The previous failure to establish clonal reproduction in the laboratory and the extreme rarity of clonality in natural vertebrates have been attributed to the complex genetic preconditions and evolutionary changes required (Rogers and Vamosi 2010; Stöck et al. 2010) or to the requirement for mutations in the genes that regulate meiosis (Sinclair et al. 2010). However, gynogenesis is not that improbable, as it has evolved multiple times independently across various animal genera (Vrijenhoek 1989; Schlupp 2005). Our study furthermore suggests that whatever the number of evolutionary changes required for the genesis of clonal reproduction, they all may happen simultaneously upon the hybridization of suitable genotypes/species. The cytological mechanisms for the occurrence of clonal reproduction require further investigation (Scali 2009).
LINKING CLONALITY AND POLYPLOIDY: STEPWISE GENOME ADDITION
Ploidy elevation with saltatory increase in heterozygosity is considered to be an important mechanism of evolutionary adaptation, allowing asexual organisms to exploit various ecological niches and compensating for the disadvantages of unisexual reproduction (Neaves and Baumann 2011). The results of our microsatellite and allozyme analyses indicate that the elevated ploidy observed in backcross progeny (96.3%) from F1 females resulted from the incorporation of the haploid paternal genome of the sexual male into an unreduced egg. Given that the synthesized diploid F1 females most likely produced diploid nonrecombinant eggs, our observation provides direct support for the GAH (Figs. 2C and 4B–D) and confirms that polyploidy is not a trigger of clonality in spined loaches, but rather a consequence. Nonetheless, in 11 triploid individuals we only observed a double-dose signal of shared parental allozyme alleles (absence of three different alleles at loci in seven out of 11 individuals in the B-1 family and in four out of 11 individuals in the B-2 family). Although genome addition is also likely for the origin of these 11 individuals, the allozyme pattern alone does not distinguish the GAH from the GDH sensu, Vrijenhoek (1994). Therefore, we cannot exclude the possibility that these individuals arose either from triploid eggs (Fig. 2) by genome duplication (Vrijenhoek 1994) or from the combination of one maternal set and two paternal sets resulting from dispermy or diandry (Niebuhr 1974; Lampert et al. 2007). However, given the genetic patterns observed in the majority of triploid progeny, we consider these alternative mechanisms unlikely.
Our study provides another interesting contribution to the discussion of the evolution of polyploidy: we found that individual spined loach hybrid females substantially differed in their rates of paternal genome incorporation. Studies of wild-caught diploid hybrid females found that each fish produced mixed clutches of approximately 2/3 diploid and 1/3 triploid progeny (Janko et al. 2007b). In the present study, eight females produced completely triploid progeny, whereas only one female (family B-3) produced a mixed clutch, in which half of the progeny did not exhibit any sign of paternal genome incorporation. This difference in polyploidization rate between wild-caught females and those analyzed in this study was highly significant (P < 0.01 using the chi-squared test). The reason underlying this difference is currently unknown. In theory, the success of sperm incorporation may depend on the choice of male, as occurs in the gibel carp Carassius gibelio (Gui and Zhou 2010). However, we consider this explanation unlikely because the same breeding stocks of males were used in our previous crossing experiments (Janko et al. 2007b) and in this study. Furthermore, all crosses were performed in the same air-conditioned facility using standardized protocols (Bohlen 1999). Therefore, it is unlikely that contrasting polyploidization rates result from different laboratory conditions, although has been shown that physical or chemical stimuli can affect polyploidization in fish (Felip et al. 2001).
Hence, our datapoint to the possibility that individual diploid hybrids differ in their affinity for sperm incorporation. It is possible that one type of “stable diploid” hybrids (wild-caught females [Janko et al. 2007b] and the B-3 family from this study) produces a rather small fraction of polyploid progeny, whereas another type of “transient diploid” hybrids (families B1, B2, B4–B9) has a high triploidization rate. If this is the case, it is easy to understand why all wild-caught diploid hybrids belonged to the “stable diploid” type. Any “transient diploid” clones would persist only for a single generation, and it would be difficult to collect them in nature. Our observation that the eggs of some hybrid females fully accept sperm whereas other hybrid females tend to produce clonal progeny points to the existence of a new phenomenon that requires further investigation. At this stage, however, it remains unknown whether the basis for such differences is genetic or physiological.
FERTILITY OF HYBRID FEMALES AND INFERTILITY OF HYBRID MALES
However, spined loaches challenge this notion, as we observed male infertility a single generation after crossings, which is too short for mutation accumulation. This observation may be explained in principle by Haldane's rule, which suggests that the heterogametic sex suffers more from hybridization than the homogametic sex. Although we lack any direct observations of sex chromosomes in the studied spined loaches, multiple sex chromosome systems with heterogametic male have been proposed to exist in two other loach species (Saitoh 1989; Vasil'ev 1995). In any case, the present results indicate that different mechanisms of gene control and expression affect oogenesis and spermatogenesis in asexuals (Moritz et al. 1989) and suggest that mechanisms other than mutation accumulation may account for the dysfunction of males in asexual complexes.
Although the phylogenetic relationships between clonal and polyploid hybrids and their sexual or diploid ancestors have been reconstructed in many cases, no study has explained why so many in vitro attempts to recreate a clonal animals, or at least clonal vertebrates, have failed (the only recreated asexual vertebrates known from nature were hemiclonal; see section Introduction). Furthermore, it has remained unclear why it is generally easier to cross clonal hybrids once they have been produced in nature than to obtain a clonal F1 hybrid itself (Vrijenhoek 1989; Mallet 2005). This uncertainty underlines our poor understanding of the interplay between hybridization, clonality, and polyploidy. The present study of spined loaches has allowed us to reconstruct the entire evolutionary route from sexual reproduction to clonality and polyploidy without any cell manipulation (Fig. 4A–D). Our experimental data illustrate a manner in which clonality in vertebrates could have originated, providing direct support for the hypothesis that hybridization per se triggers clonality and that polyploidy is a consequence of clonal reproduction. Because the establishment of clonality is easy in spined loaches, this group appears to be a suitable model organism to study the proximate mechanisms, and their evolutionary implications, for the disruption or circumvention of meiosis after hybridization events.
Associate Editor: K. Petren
We thank J. Kopecká for her help in the laboratory. We thank R. C. Vrijenhoek and one anonymous reviewer for their useful comments and corrections that improve the quality of this article. We also thank M. R. Kearney for his advices concerning this article. We gratefully acknowledge the grant support from Grant Agency of Academy of Sciences of the Czech Republic by grant no. KJB600450902 to LC and KJ. KDG acknowledges the support received for an exchange grant from the European Science Foundation (ESF) in the framework of the project ConGen: “Integrating population genetics and conservation biology: Merging theoretical, experimental and applied approaches.” We thank the Grant Agency of the Czech Republic by grant nos. 206/06/1763 and 206/09/1298 with both grants to LC, MR, PR, and KJ, by grant no. 206/09/1154 to LC, JB, and PR, and by grant no. 523/08/0824 to MR and PR. Further support was provided by the Academy of Sciences of the Czech Republic Grant IRP IAPG AV0Z 50450515 and by the Biodiversity Research Centre LC06073.