Linkage Analysis Reveals the Independent Origin of Poeciliid Sex Chromosomes and a Case of Atypical Sex Inheritance in the Guppy (Poecilia reticulata)
Namita Tripathi, Margarete Hoffmann, Detlef Weigel, Christine Dreyer


Among different teleost fish species, diverse sex-determining mechanisms exist, including environmental and genetic sex determination, yet chromosomal sex determination with male heterogamety (XY) prevails. Different pairs of autosomes have evolved as sex chromosomes among species in the same genus without evidence for a master sex-determining locus being identical. Models for evolution of Y chromosomes predict that male-advantageous genes become linked to a sex-determining locus and suppressed recombination ensures their co-inheritance. In the guppy, Poecilia reticulata, a set of genes responsible for adult male ornaments are linked to the sex-determining locus on the incipient Y chromosome. We have identified >60 sex-linked molecular markers to generate a detailed map for the sex linkage group of the guppy and compared it with the syntenic autosome 12 of medaka. We mapped the sex-determining locus to the distal end of the sex chromosome. We report a sex-biased distribution of recombination events in female and male meiosis on sex chromosomes. In one mapping cross, we observed sex ratio and male phenotype deviations and propose an atypical mode of genetic sex inheritance as its basis.

LONG-standing models for evolution of sex chromosomes suggest that the process starts when a pair of autosomes acquires sex-determining genes in the vicinity of genes advantageous for only one sex (Muller 1914; Ohno 1967). This chromosome begins to differentiate from its homolog and suppressed recombination—due to accumulation of noncoding repetitive DNA sequences, pseudogenes, and transposable elements or chromosomal rearrangements—consolidates sex chromosome differentiation (reviewed in Charlesworth 2000; Charlesworth et al. 2005; Marshall Graves 2006).

During the gradual conversion of an autosome into a functional sex chromosome, it remains pseudoautosomal across most of its length for a major portion of its evolutionary lifetime. All the sex chromosomes identified in various fish species so far are considered to be at an early stage of differentiation, in contrast with mammals (Marshall Graves and Shetty 2001; Charlesworth 2004; Marshall Graves 2006). Hence, in typical fish sex chromosomes, there is only a small region with reduced crossing over, and homologous recombination is still possible throughout the larger pseudoautosomal portion. On the basis of studies of common ancestors and the viability of homozygous YY males, the age of teleost sex chromosomes has been estimated in a range of 2–30 million years, dependent on the species (Volff et al. 2007), compared to ∼180 million years for the human Y chromosome.

The details of genetic sex determination in the major vertebrate groups vary considerably, especially in the teleost fishes, where a broad spectrum of sex-determining mechanisms can be found within the same order. Teleost species display unisexuality, environmental sex determination, and hermaphrodism in addition to heterogametic sex, which prevails in other classes (Volff 2005). Although male (XY) heterogametic sex is more frequent than female (ZW) heterogamety, both strategies have likely evolved several times independently in teleosts (Mank et al. 2006). In contrast to sex-determining loci (SDL), genes required for differentiation of the gonads are not necessarily located on the sex chromosomes, as deduced from hormone-induced sex reversal and from the presence of autosomal factors that can sometimes overrule the SDL on the male Y chromosome (Winge 1934; Winge and Ditlevsen 1938).

The only sex-determining gene identified so far in a teleost species is the DmrtY gene in the medaka, Oryzias latipes, on its Y chromosome (Matsuda et al. 2002; Nanda et al. 2002). The sex chromosomes of other species in this genus are derived from different ancestral chromosomes (Takehana et al. 2007a; Tanaka et al. 2007). Similarly, the sex-determining loci of different salmonids (Phillips et al. 2005, 2007; Artieri et al. 2006) and of two species of stickleback (Gasterosteus) have apparently evolved from loci originating on distinct ancestral chromosomes (Woram et al. 2003; Charlesworth 2004; Peichel et al. 2004). In the three-spine stickleback Gasterosteus aculeatus (Peichel et al. 2004), the tiger pufferfish Takifugu rubripes (Kikuchi et al. 2007), the rainbow trout (Alfaqih et al. 2009), and the platyfish Xiphophorus maculatus (Volff and Schartl 2002), the SDL could be mapped to a specific linkage group that represents the differentiating Y chromosome. Fine mapping of these loci is impeded by suppression of meiotic recombination and by frequent occurrence of repetitive sequences in the gonosomal region of the Y chromosome (Nanda et al. 1993; Peichel et al. 2004).

Poeciliids are one of the best-studied groups of fishes with respect to sex chromosome evolution, due to the variable nature of sex determination in this family. In the genus Xiphophorus, there are three types of females, WY, WX, and XX, and two types of males, XY and YY. The nature of chromosomal sex determination in other genera is not very clear (Kallman 1973, 1984; Orzack et al. 1980; Volff and Schartl 2001; Schultheis et al. 2006).

Extensive genetic evidence supports a heterogametic XY sex determination in the guppy, Poecilia reticulata (Winge 1922a,b). Guppies are live bearers that display a pronounced sexual dimorphism for body size and for nuptial ornaments that are expressed exclusively in mature males. A substantial fraction of the ornamental genes is faithfully transmitted from father to son in every generation and may therefore represent male-advantageous genes linked to the SDL on Y in wild guppies (Winge 1927; Winge and Ditlevsen 1947; Haskins et al. 1970). Another set of ornamental genes can be maternally inherited, but their expression is largely limited to the sons. Similarly, the derived color patterns in ornamental guppies are linked to the SDL (Khoo 1999a,b,c). Some of the sex-linked traits were located relative to anonymous molecular markers, which were utilized to generate partial linkage maps for this species (Khoo et al. 2003; Watanabe et al. 2005; Shen et al. 2007).

As a first step toward identifying the genes and molecular mechanisms responsible for extreme phenotypic variations in most sex-linked traits of guppies, we have produced a detailed linkage map of the sex chromosomes, which reveals that the guppy sex-linkage group is syntenic with an autosome of medaka, which has an advanced draft of a genome sequence (Kasahara et al. 2007). Although male guppies are generally heterogametic, we describe a case of genetically atypical males and compare the patterns of meiotic recombination between males and females.


Mapping crosses:

Six mapping crosses between Quare and Cumaná guppy are described in a separate article by Tripathi et al. (2009). Briefly, F1 individuals from each intercross were mated in single pairs to obtain F2 progeny. We initiated a total of 26 intercrosses between guppies from Cumaná (Venezuela) and the Quare River (East Trinidad). Of 11 crosses between Quare females and Cumaná males (Q_Cu), 6 had >100 F2 progeny and 2 had no offspring. Of 15 reciprocal crosses (Cu_Q), 2 had >100 F2 progeny and 10 had no offspring. Compromised fertility in reciprocal intercrosses may indicate reproductive isolation (Russell and Magurran 2006) as a consequence of the Cumaná guppy being highly differentiated (Alexander and Breden 2004). Of the intercrosses with >100 F2 progeny, we genotyped 5 (99, 150, 153, 157, 158) Quare ♀ × Cumaná ♂ and 1 (76) Cumaná ♀ × Quare ♂ (see supporting information, Table S2 for the number of individuals in each mapping cross). Anesthetized parents, F1, and F2 specimens were photographed as specified and then preserved in 95% ethanol at −20° (Tripathi et al. 2009).

Screening for sex linkage:

The bulk segregant analysis (BSA) approach (Michelmore et al. 1991) was used for confirming sex linkage of all candidate genes. Briefly, the genomic DNA templates from 15 F2 females and 15 F2 males from a single F1 pair of cross 76 (Cu_Q) were pooled. The primers were selected from the candidate coding genes either flanking an intron or from the 3′-UTR, and PCR was performed on each of the four templates. Single nucleotide polymorphisms (SNPs) were analyzed from these four templates (each F1 parent and the F2 female and male pools) for their segregation patterns to check for sex linkage of the selected candidates (Figure 1). On the basis of the parental genotypes, F2 female and male pools were expected to show typical sex-linked segregation of the alleles (see Figure S1).

Synteny-based candidate gene selection:

Homologs of confirmed sex-linked markers in the guppy were identified on genome scaffolds of other fishes, and adjacent regions were screened for the presence of coding genes ( Sequences of linked genes were then used to identify guppy homologs in an EST database ( using BLASTN. Exon–intron boundaries of candidate genes were predicted on the basis of medaka, fugu, and tetraodon sequences, and intron-flanking primers were designed from each candidate gene. All markers found to be sex linked by BSA were subsequently confirmed by sequencing of PCR products from DNA templates from individual fish (Table S1a).

Genomic walking:

A genomic BAC library of eightfold coverage and an average insert size of 160 kb were prepared from pooled Cumaná males (constructed by Bio S&T, Montreal). Filters with spotted BAC clones were prehybridized in 5× SSC, 0.02% (w/v) SDS, 0.1% (w/v) N-lauroylsarcosine, and 1% blocking solution for 4 hr at 60°. Probes representing known sex-chromosome-linked sequences were prepared with a PCR DIG-probe synthesis kit (Roche), and filter hybridization was carried out overnight at 60°. Filters were washed three times each in 0.1% SDS, 2× SSC (30 min at 60°); 0.1% SDS, 2× SSC (20 min at 60°); and 0.1% SDS, 0.5× SSC (15 min at 60°). Membranes were treated with blocking solution, followed by 1:10,000 (v/v) diluted anti-DIG antibodies (Roche). Detection was performed using CSPD chemiluminescent substrate (Roche). Blots were exposed to Lumi-film chemiluminescent detection film (Roche) for 20 min to several hours. Clones identified by significant hybridization signals were streaked on chloramphenicol plates, and single colonies were picked for inoculation of overnight cultures for DNA preparation. BAC DNA was isolated using a Qiagen large-construct kit and tested by PCR for the presence of the genomic fragment used for filter hybridization. Each BAC was sequenced at both ends with Big Dye Terminator chemistry using the standard pIndigoBAC-5 vector-specific sequencing primers forward 5′-GGA TGT GCT GCA AGG CGA TTA AGT TGG and reverse 5′-CTC GTA TGT TGT GTG GAA TTG TGA GC on an Applied Biosystems 3730xl DNA Analyzer. The obtained sequences, which mostly represented noncoding DNA, were used to design primers for PCR on genomic templates. The resulting novel SNP markers were tested by BSA performed with pooled male and female templates as described above to verify sex linkage. PCR on DNA from all BACs isolated during a genomic walk was used to determine the extent of overlap between these BACs.

Mapping of Sex:

For mapping Sex, >2000 offspring from six mapping crosses between the Quare and Cumaná populations were used, as specified in Table S2. Phenotypic sex was scored by presence of adult male ornaments and a differentiated gonopodium. Genotypes for 790 SNP markers of each individual were analyzed with Joinmap4 (Van Ooijen 2006) to determine the location of Sex in the genomewide linkage map. Recombination frequencies of the markers on the sex linkage group were estimated on the basis of X- and Y-linked alleles segregating in each mapping cross, with phenotypic sex as the reference (Table 1). The exact sex chromosome composition for each recombinant individual was predicted from the genotypes of the sex-linked markers and correlated with phenotypic sex.

Synteny determination with medaka:

BLASTN and BLASTX were performed for all sex-linked markers from the guppy against the medaka genome. The exact position for each marker on medaka chromosome 12 was converted on a scale of 1–30 for comparison of the relative orders and gaps between markers with the guppy LG 12 map. The individual maps were plotted using MapChart (Voorrips 2002).


Identification of sex-linked markers:

BSA (see materials and methods) allows identification of sex-linked genes whose male and female alleles may be distinguished by SNPs. As shown in Figure 1, a SNP is informative for sex linkage when a male F1 is heterozygous for this marker. For this strategy, the F1 parents and the pools of F2 male and female genomic DNA were used as a template for PCR with the candidate primers (described in materials and methods). The sequence traces of the PCR products were inspected for segregation of SNPs identified in the grandparents (Figure 1 and Figure S1).

Figure 1.—

BSA of SNP markers for sex linkage. Segregation of SNP markers among F2 female and male pools from a single F1 pair was analyzed for sex-linked pattern (see Figure S1). Among the four possible combinations of a biallellic SNP marker occurring in the F1 parents, half were informative for sex linkage (examples 1 and 2), while the other half were not (examples 3 and 4). When the male F1 parent is heterozygous and the female F1 parent is homozygous (example 1), all F2 individuals are informative. When both F1 parents are heterozygous, 50% of their F2 progeny are informative, resulting in signals of different strength for both alleles in pools of male and female F2 (example 2). SNP markers are inconclusive when the male F1 parent is homozygous (examples 3 and 4). Markers that appeared sex linked in F2 pools were scrutinized in individual fish.

One of 58 SNP markers tested identified the gene encoding cyclin G2 as sex linked. Additional markers were developed by analyzing the sex linkage of guppy homologs of genes that were located on the same genomic scaffolds as cyclin G2 in medaka, fugu, or tetraodon (see materials and methods and Table S1a and Table S1b). Using this synteny-based candidate selection approach, 20 additional sex-linked markers could be identified in the guppy. As a complementary strategy, we pursued genomic walking with BAC clones.

Additional markers came from the development of a whole-genome linkage map for the guppy, described in Tripathi et al. (2009). Table S1a and Table S1b show the complete list of >60 sex-linked markers generated with the three approaches.

Recombination frequency on the sex linkage group and mapping of Sex:

Most sex chromosome markers were homozygous in the parental strains and were therefore informative in only 50% of the F2 progeny. The markers that could distinguish between X- and Y-linked alleles in a cross were used to estimate recombination frequencies between the two chromosomes (Figure S1). F2 progeny from five mapping crosses between Cumaná and Quare populations were used to estimate the genetic position of the dominantly acting sex-determining locus, Sex.

Utilizing the information from X- and Y-linked alleles, the position of crossover events was noted for every recombinant sex chromosome. Table 1 shows the recombination frequencies detected from all informative sex chromosome markers in different mapping crosses. Cross 16 is a cross between two additional populations (Oropuche ♀ × Tranquille ♂), analyzed for a subset of sex-linked markers. Across all crosses, the highest recombination frequency for any marker with Sex was 2.3% and the closest to Sex was marker 0229, with 1.5% recombination frequency. Recombination between X chromosomes in XX females had identified this marker as the last genetic marker mapped on the sex chromosome, placing Sex distal to marker 0229.

View this table:

Detected recombination frequency between X and Y chromosomes in mapping crosses

Among five mapping crosses with >1300 F2 individuals (Table S2), we found 26 F2 with recombinant sex chromosomes (Figure 2). For 14 of these individuals, the genotype at marker 0229 agreed with the phenotypic sex (XX female, XY male, Figure 2). A single crossover event between the most distal marker (0229) and the Sex locus in their F1 father can explain the phenotypic sex of the other 12 individuals. We therefore infer the occurrence of such crossover events, although additional molecular markers are required to confirm this at the DNA level.

Figure 2.—

Recombinant F2 individuals for X and Y chromosomes in five mapping crosses. Each bar represents the recombined sex chromosomes of an F2 individual. The X chromosome inherited from the F1 female parent is represented by the left half of each bar, while the right half of each bar represents the sex chromosome from the F1 male parent. The light and dark blue colors show the paternal strain (grandfather) derived X and Y chromosomes, respectively. The X chromosomes from the maternal strain (grandmother) are in yellow in each recombinant. The phenotypic sex of each recombinant is shown by symbols at the top. The presence of either an XX or an XY genotype for mapped markers at the distal end of every recombinant is indicated at the bottom as X (yellow) or Y (blue). The putative Sex locus is predicted to be near the distal end of the chromosome, downstream of the last mapped marker (0229). Horizontal black lines mark recombinants, for which the sex chromosome genotype at the distal end disagrees with phenotypic sex. A single crossover event between marker 0229 and the Sex locus could explain their phenotypic sex. These recombinants were used to estimate the total recombination frequency at this end of the linkage group (see Table 1). The positions of the crossover events along the sex chromosome in female and male meiosis are shown by red and blue arrows, respectively. The distribution of the crossover events in female and male meiosis is represented by the vertical red and blue brackets on the right.

Nonrandom distribution of crossovers between X and Y chromosomes:

In cross 157 (Q_Cu), <1% (7/854) of F2 animals had a recombinant sex chromosome, while, in the reciprocal cross 76 (Cu_Q), 6% (12/200) of F2 individuals were found to have a recombinant sex chromosome (Table 1). Recombination between the distal-most marker (0229) and Sex was detected in only one of the cross 157 individuals, but in four of the cross 76 progeny. The resulting estimates of the recombination frequency between marker 0229 and the Sex locus are therefore 0.1% for cross 157 and 2.0% for cross 76 (Table 1).

Crosses 99, 150, and 153 (all Q_Cu) produced 1.6, 3.3, and 0.4% recombinants, respectively. These were phenotypic females in cross 99 and cross 153, and males in cross 150. The recombination events in all of these individuals are predicted to have occurred between marker 0229 and the Sex locus.

By analyzing the distribution of crossover events along the sex chromosomes, we found a stark difference between female and male meiosis. Recombination events in the F1 female meiosis were evenly distributed along the length of the X chromosome. In contrast, crossovers in F1 male meiosis clustered toward the distal end of the chromosome (blue arrows in Figure 2). This observation was consistent with the difference between the reciprocal crosses 76 and 157, which had produced 12 and 7 recombinant individuals, respectively.

Synteny between the guppy sex chromosome and medaka chromosome 12:

We identified homologs of the guppy sex chromosome genes in the medaka O. latipes using BLASTN. All but three of these are located on medaka chromosome 12, with the order along the chromosome largely syntenic with the order of the guppy sex chromosome loci (Figure 3). Medaka homologs of several additional sequences identified by genomic walking in a guppy BAC library are also located on chromosome 12. Many sequences identified by genomic walking toward the distal end of the guppy sex chromosome were rich in repeat sequences, making it impossible to place these on the genetic map.

Figure 3.—

Synteny relationships between medaka chromosome 12 and guppy LG 12. Markers common between the two maps, with known positions in both, are underlined in gray. Markers in pink on guppy LG 12 map to different chromosomes of medaka (0380 to chromosome 15, 0517 to chromosome 03, and 0073 to chromosome 14 of medaka). Markers in brown on guppy LG 12 do not have a significant BLAST hit on the medaka genome. Markers in black on medaka chromosome 12 are sex linked in the guppy, and their approximate positions are predicted from the synteny of linked markers, but not shown on the guppy map. BAC linked1,2 (thin blue lines on the right) include markers derived from BAC end sequences during genomic walking. Additional BAC-linked markers from the same region, with no homolog identified on medaka chromosome 12, are listed in Table S1. Several BAC clones resulting from genomic walk toward the distal end of guppy LG 12 were rich in repeat sequence that interfered with reliable detection of sex linkage. The putative location of the master SDL is at the distal end of the guppy LG 12.

Comparative mapping has demonstrated that chromosome 12 of O. luzonensis, in the same genus as O. latipes, contains its SDL (Tanaka et al. 2007) at a position corresponding to between markers 0090 and 0210 on the guppy sex chromosome. Considering that Sex is located at the extreme distal end of LG 12 in the guppy, we think that it is unlikely that the sex determination locus is conserved between the guppy and O. luzonensis.

Atypical sex inheritance:

We found an atypical segregation ratio of Y-linked male traits in cross 158 (Q_Cu) with more males than females (Figure 4, A and B), in contrast to the expected 1:1 sex ratio that was normally observed in all other mapping crosses.

Figure 4.—

Summary of cross 158 atypical sex determination. (A) The three YLTs segregating in cross 158 are marked on the left image of the Cumaná male parent of this cross: (1) dorsal fin orange and black (DFOB), (2) central blue white spot (CBWS), and (3) posterior ventral black stripe (PVBS). The X- or Y-linked traits (X-YLT) of the Cumaná population are(4) the hind-fin black spot (HFBS) and (5) the hind-fin lower orange (HFLO). The right fish shows the Quare X- or Y-linked trait (6), a black spot on the caudal peduncle, which shows a mutually exclusive expression with respect to the HFBS (4) from Cumaná. (B) Representative males showing segregation of the phenotypes of three YLTs and the sex ratios in F1 and F2 generations. Three kinds of F1 males with genotypes as explained in C were found. F1 males without YLTs gave rise to all F2 males with no YLT (pairs 11, 14, 16). F1 males of genotype XQ1YC-sired F2 males with YLTs (pairs 6, 7, 15). F1 males of genotype XQ2YC had YLTs and gave rise to predominantly male F2 of which 50% had YLTs (pairs 5, 9, 10; shown with red lines). See Table S2 for the number of individuals in each category and sex ratios. (C) The segregation of the sex chromosomes in cross 158, which correspond to each of the phenotypes displayed in B. The four sex chromosomes segregating in this cross were identified by specific SNP markers in each (Table S3) and are shown in four different colors. All males with YC had three YLTs whereas males whose sex was conveyed by XQ2 had no YLT. The recombinant sex chromosomes for any of the genotypes are not shown in the figure.

The Cumaná male in cross 158 had three Y-linked traits (YLT) typical for this population, including a central blue white spot (CBWS) ventral to the dorsal fin, a dorsal fin orange and black (DFOB), and a posterior ventral black stripe (PVBS) on the caudal peduncle. In addition, some color traits are X and Y linked (X-YLT), the most important being the hind-fin black spot (HFBS) and the hind-fin lower orange (HFLO), in Figure 4A, left fish.

The F1 generation from cross 158 included 21 females and 41 males (♀:♂ = 1:2). Among the F1 males, two strikingly different phenotypes could be distinguished by either the presence or the absence of the three Y-linked traits, with 22 males lacking and 19 males having them (ratio 1:1). The F2 sons of F1 males lacking the Y-linked traits (pairs 11, 14, and 16) also did not express these traits. Among the normal-appearing F1 males, some faithfully passed on the YLT to the F2 generation (pairs 6, 7, and 15) while other normal-looking F1 males from three pairs (5, 9, and 10) had half their F2 male offspring without the YLT. In addition, F2 sex ratios were very skewed (Figure 4B).

All the F1 and F2 progeny of this cross were genotyped with a set of SNP markers for genomewide linkage analysis (Tripathi et al. 2009). In a few cases, it was possible to differentiate between all four sex chromosomes involved in the cross from the Quare female (XQ1XQ2) and from the Cumaná male (XCYC) using the marker genotypes (see Table S3) and to follow their segregation in the subsequent generation. Correlating the presence of the different chromosomes and the abnormal segregation of sex and Y-linked traits revealed that the XQ2 chromosome was responsible for both. This was evident from the observation that all F1 females of this cross had the XQ1XC genotype, while F1 males not only were as expected—XQ1YC and XQ2YC—but also were XQ2XC. These three genotypes furthermore were clearly correlated with the three classes of the F1 males described above. All XQ1YC F1 males and their F2 progeny showed normal Y-linked traits and normal sex ratios. In contrast, F2 progeny of XQ2XC males displayed a 1:1 sex ratio but the males lacked YLT. Whereas XQ2YC males expressed normal YLT, half of their F2 male progeny lacked them. Moreover, there was an up to sevenfold excess of sons among the F2 progeny.

All these results can be explained by postulating that the XQ2 chromosome from the Quare female parent carries a factor that supports the development of fully fertile males in the absence of a Y chromosome (Figure 4C). An important clue was provided by the presence of two sex-chromosome-linked phenotypic traits, the HFBS on Cumaná and the posterior black spot on the caudal peduncle of Quare (Figure 4A), which is X and Y linked in Quare individuals. The HFBS from Cumaná is dominant over the Quare black spot and is always expressed if either an XC or YC allele for this trait is present.

Analysis of the sex-chromosome-recombinant F2 individuals from cross 158 indicated that the markers at the distal end of the sex chromosome were correlated with the phenotypic sex of the XX individuals. The unusual XQ2 locus responsible for male phenotypic sex is therefore predicted to be toward the distal end of the sex chromosome.

We compared the distribution of crossover events in the F1 females (XCXQ1) with atypical F1 males (XCXQ2) from cross 158. For this purpose, we analyzed a total of 87 F2 offspring from three F1 pairs (11, 14, and 16) of cross 158, which had XCXQ2 males as their F1 parent. Interestingly, recombination frequency on the sex chromosome was lower in the F1 males compared to F1 females of this cross, even though the genotype of males is XX and not XY (supporting information, Figure S2; supporting information, Table S4b. In addition, the distribution of the crossover events along the length of the chromosome displayed a pattern similar to that observed from meiosis in normal XY males of other crosses (Figure 2). The majority of crossovers in male meiosis occurred toward the distal end of chromosome, while no such bias was observed in female meiosis.

The different recombination frequencies observed in female and male meiosis could be caused by sequence differences between the X and Y chromosomes. However, the clustering of crossovers does not correlate with the expected gradient of recombination suppression in the vicinity of the differentiated region of the sex chromosomes. In addition, the XQ2XC males in cross 158, like normal XY males, show a similar pattern of reduced recombination as well as a biased distribution of crossover events along the sex chromosome (Figure S2; Table S4a and Table S4b). These results indicate that recombination is controlled by phenotypic sex.


Mapping of the Y-specific region of the sex chromosome:

The Y-specific “gonosomal” region containing the male SDL and a number of male-specific color genes has previously been postulated to be at one end of the guppy sex chromosomes by Winge (1927). Later, the Y-specific segment has been clearly detected at a terminal location in mitotic and meiotic spreads by chromosome in situ hybridization, comparative genomic hybridization, and synaptonemal complex analysis (Nanda et al. 1992; Traut and Winking 2001). Consistent with this, predictions from our results suggest a distal location of the Y-specific segment and the Sex locus. An extensive region of sequence homology, as well as several polymorphisms detected in the pseudoautosomal regions of X and Y chromosomes, indicate that the molecular rearrangements differentiating the sex chromosomes is still ongoing. In support of this prediction, no difference was found in allele distributions between X and Y when ∼250 specimens from 42 wild guppy populations were genotyped, using all mapped sex-linked markers (data not shown).

Population-specific divergence of sex chromosomes:

It is striking that cross 76, which is the only genotyped cross between a Cumaná mother and a Quare father, shows a much higher number of recombination events between X and Y compared to the five reciprocal crosses. Although this difference could be a random variation, it could also indicate that the sex chromosomes are diverged to different degrees between these two populations. If the Y is more differentiated from the X in Cumaná compared to the Quare population, a higher rate of recombination is expected between the X and Y chromosomes of Quare (as in cross 76) relative to the recombination rate between the X and Y chromosomes of Cumaná (reciprocal crosses). Since the order of mapped sex-linked markers showed no differences between cross 76 and the reciprocal crosses, any potentially significant differences between the sex chromosomes of the Cumaná and Quare populations should be limited to the distal region of the chromosome not yet covered by our markers.

Synteny of the guppy sex chromosome with medaka chromosome 12:

The order of markers between the sex linkage group of the guppy and homologous chromosome 12 of O. latipes (Figure 3) are mostly conserved. One significant translocation is revealed by the position of marker 1053 at the proximal end, which is present toward the distal end of medaka chromosome 12. The gap between marker 1053 and marker 0229 is only 2.93 Mb on medaka chromosome 12, whereas they map to two ends of guppy chromosome 12. A comparison with other fish genomes revealed these markers to be linked within a 1.9-Mb segment on stickleback group XVI and to occur within 600 kb on homologous scaffold 106 in Fugu. These observations suggest that the chromosomal segments mapping to either end of guppy chromosome 12 were linked closely in the ancestral populations of these fish species.

Linkage analysis of the sex chromosomes of guppy and O. luzonensis indicated that despite a shared synteny with O. latipes chromosome 12, these two species have different sex-determining loci. This is not surprising, considering the general lack of conservation of the master sex-determining gene among teleost species (Mank et al. 2006; Volff et al. 2007). At least four different chromosomes gave rise to the SDL in different species of Oryzias, suggesting that several master sex-determining genes may have evolved even in closely related species (Takehana et al. 2007a,b; Tanaka et al. 2007). The O. latipes Dmrt1bY containing the Y-specific segment is only 250 kb in length, while the rest of the Y is completely homologous with the X chromosome (Kondo et al. 2006), suggesting that a small segment of DNA, having at least one important functional gene for male sex determination, is sufficient to serve as a focal point for sex chromosome divergence.

Sex determination in the guppy can deviate from the XY paradigm:

Winge (1930, 1934) has shown that it was possible to obtain XX males in the guppy, and he also proposed that multiple antagonistic autosomal factors may contribute to the ultimate outcome of genetic sex determination. The male sex differentiation linked to one specific X chromosome was not universally observed in the Quare population but appears to be specific to the female Quare parent of cross 158. From the available markers, we cannot deduce whether the putative sex-determining gene near the distal end of the XQ2 chromosome is derived from the normal Y-linked male sex determinator itself or by recombination from another cryptic sex chromosome occurring in the Quare population. Alternatively, this XQ2-linked gene might encode a downstream component of the male sex-determining pathway, and a mutation or recombination might have altered its regulation. In either of these scenarios, a gene now found on the XQ2 may function as a male sex-determining gene. A distinction between these possibilities will require denser marker coverage and sequence information from the distal end of this chromosome. The XQ2 was found to behave as a stable male-determining chromosome in the Quare genomic background throughout more than five generations of backcrosses derived from cross 158 F1 males (data not shown). Therefore, different combinations and ratios of additional autosomal factors might direct the regulation of the sex differentiation toward the male or female pathway.

Sex-specific meiotic recombination on chromosome 12:

The crossover events found in male meiosis were predominantly restricted to a limited boundary region between the pseudoautosomal and the putative sex-differentiated segment of the sex chromosome (Figure 2). Differences in meiotic recombination frequencies depending on phenotypic sex have been described in medaka fish (Yamamoto 1961; Matsuda et al. 1999; Kondo et al. 2001), where the recombination frequencies on the sex chromosomes of normal and sex-reversed individuals showed a clear correlation with their phenotypic sex. Irrespective of their sex chromosomal genotypes, phenotypic males displayed a lower recombination frequency on the sex chromosomes than phenotypic females did. Sex-specific variation in recombination frequency and distribution is observed in many, including mammals, but no clear molecular basis has yet been found to explain these differences (Lynn et al. 2005).

Model for the organization and evolution of guppy sex chromosomes:

On the basis of our observations on the nonrandom distribution of meiotic recombination on guppy sex chromosomes and our mapping results, we propose a model for the organization of the guppy sex chromosomes. According to this model (Figure 5), the guppy sex chromosome is constituted by three distinct regions, namely male-specific nonrecombining 1 (MSNR1), freely recombining (FR), and male-specific nonrecombining 2 (MSNR2). The set of mapped markers in this study covers the MSNR1 region and a part of the FR region. From the distribution of all the mapped markers on the sex chromosome and the preliminary results of chromosome in situ hybridization (I. Nanda, N. Tripathi, C. Dreyer and M. Schantl, unpublished results), we predict these regions to represent ∼60% (MSNR1), 10–20% (FR), and 20–30% (MSNR2) of the total length of the sex chromosome (Figure 5).

Figure 5.—

Model for the organization and evolution of guppy sex chromosomes. The three regions of the sex chromosome differentiated on the basis of recombination in male meiosis: male-specific nonrecombining 1 (MSNR1), freely recombining (FR), and male-specific nonrecombining 2 (MSNR2). The phenotypic sex of the individuals depends on the genotype of the distal (MSNR2) segment only, XX female and XY male, irrespective of the composition of the rest of the chromosome.

These three regions can be differentiated only on the basis of the recombination suppression in male meiosis, as all three regions recombine freely between the X chromosomes during female meiosis. MSNR1 shows suppressed recombination during meiosis in phenotypic males, in contrast to FR, which either has escaped sex-specific recombination suppression or overrules it. MSNR2 is the region that represents the differentiated part of the sex chromosome, including the gonosomal region of Y with the sex-determining and Y-linked color loci (Figure 5). The sequence composition and length of this segment might be variable between guppy populations.

While sex-dependent recombination suppression of the distal gonosomal region of the Y chromosome can be explained by sequence divergence, e.g., by inversions or insertions, the mechanism by which meiotic recombination also becomes restricted in the pseudoautosomal region (MSNR1) remains unclear. Positioning of a recombining region at the boundary (FR) of the diverged segment of the sex chromosome (MSNR2) could be under favorable. Meiotic recombination of the chromosome at the interphase between the pseudoautosomal and gonosomal regions offers a mechanism that protects the major portion of the sex chromosome against accumulation of deleterious mutations while maintaining the diverged region of the chromosome, which links male-advantageous traits to a SDL. Recombination in this region may also provide a mechanism for maintenance of enhanced variation among X- and Y-linked male color traits. The characteristic male color variation is regarded as important for both natural and sexual selection in guppies. Consequently, enhanced recombination of the FR could be advantageous for at least two reasons: purging of deleterious mutations and enhanced natural variation. From an evolutionary perspective, this model allows for an easy translocation of a small diverged segment of a sex chromosome to an autosome, potentially initiating the evolution of novel sex chromosomes.


We thank Christa Lanz for help with sequencing, Eva-Maria Willing for statistical analysis, Aleks Basara for help with illustrations, Indrajit Nanda for discussion, Manfred Schartl for comments on the manuscript, and the reviewers for constructive criticism. This work was supported by a Gottfried Wilhelm Leibniz Award of the Deutsche Forschungsgemeinschaft and by the Max Planck Society.


  • Received November 10, 2008.
  • Accepted March 2, 2009.


View Abstract