RAD sequencing and linkage mapping
Here, we present the first linkage map of Gnathopogon, which is also the first for the Gobioninae, a diverse group of fishes within the family Cyprinidae. Taking advantage of massively parallel sequencers, we obtained a high-density linkage map with 25 linkage groups and an average marker distance of approximately 0.87 cM that covers 86.9% of the genome. The number of identified linkage groups is congruent with the karyotypes of G. caerulescens (2n = 50) and G. elongatus (2n = 50)
. Such a dense linkage map contains detailed information on the genomic structure of an organism and is therefore useful for studies involving comparative genomics and QTL mapping.
To date, AFLP and microsatellite markers have been popular options for linkage analyses in organisms without genomic information. Although AFLP markers require no prior information about the genome of a target species, they are anonymous dominant markers bearing no sequence information for genomic comparison; microsatellite markers are sequence-based, but they are costly and time-consuming if hundreds or thousands of markers are involved. Our linkage map is solely based on RAD-seq. In contrast to AFLP and microsatellite markers, RAD-tag markers have advantages for the genomic analysis of non-model organisms. These markers are sequence-based, allowing the practice of comparative genomics
[10, 11], which aids in exploring candidate genes for traits of interest
 and even assembling de novo genomic sequences
. Moreover, allelic information on a large number of markers is readily available without prior curation and labourious experiments. The present study further demonstrates the utility of RAD-seq in the genomic study of a non-model organism, yielding a wealth of genomic information without prior knowledge of the genome of a subject species.
We found a female-specific TRD in the alleles of marker loci homologous to LG3M. TRD refers to a phenomenon in which the alleles of a locus of a heterozygous parent are not transmitted equally, resulting in deviation from the Mendelian 1:1 segregation
[32, 36]. This phenomenon is an extension of segregation distortion, referring to the unequal segregation during meiosis; TRD also includes cases in which postmeiotic effects or unknown causes yield distorted transmission of alleles. The extent of TRD is also correlated with genomic divergence, which is empirically shown as the abundance of distorted markers in interspecific crosses relative to intraspecific crosses
[37–39]. The divergence time of the two Gnathopogon species used to construct the mapping family is estimated at 4 million years ago (mya)
. This might have caused substantial differences between the genomes of the two species, due to the accumulated genomic changes following the divergence. Taken together, the genomic data suggest that TRD occurred in the interspecific cross of the Gnathopogon lineages. To test this hypothesis, intraspecific crosses of Gnathopogon species would be needed.
Sex-specific TRD occurs in several animals
[40–42]. A study of female-specific TRD in the mouse Mus musculus suggested that the TRD was caused by the post-fertilisation reduction of female viability that involved a specific region of a chromosome
. The female-specific TRD in Gnathopogon also seems to be due to postzygotic causes, such as the reduced viability of the female embryo or fry involving a deleterious gene on LG3 or a deleterious gene regulated by a gene on LG3. This gene might be a recessive lethal allele derived from the female founder, G. elongatus. Male viability might not be reduced because the lethality of hybrids is rescued by a gene in the male-determining region. This explanation seems likely because male offspring exhibited no such TRD on LG3M and because the allele frequency in female progeny exhibited the trend along LG3M. Further studies are needed to elucidate the mechanism of TRD by investigating the survivability of gametes and zygotes and the allele transmission using interspecific and intraspecific crosses.
TRD can affect the transmission of alleles in the hybrid zone. In mapping populations of the iris Iris fulva and Iris brevicaulis, for example, TRD causes an asymmetric introgression of alleles of I. fulva, which is attributable to the more frequent introgression of I. fulva alleles into I. brevicaulis in the natural hybrid zones between the iris species
. Our subject species, G. caerulescens and G. elongatus, show parapatric distribution in the Lake Biwa basin. G. elongatus inhabits the tributaries, lagoon, and shallow littoral zone of the lake. Conversely, G. caerulescens inhabits the offshore limnetic zone. However, G. caerulescens spawns in the lagoon and littoral zone, and the reproductive seasons of these species overlap, resulting in reproductive season sympatry the coexistence during the reproductive season
. These species occasionally hybridise in natural habitats, that is, the premating barrier is incomplete (
, Kokita, unpublished data). The TRD might contribute to reproductive barriers between sympatric Gnathopogon species by lowering the fitness of hybrids because hybrid individuals produce a smaller number of viable offspring
There was high synteny between Gnathopogon and zebrafish. Majority of the RAD loci located on a Gnathopogon linkage group are colocalised to a single zebrafish chromosome. Considering the old divergence of the lineages leading to each species, which date back to 117 mya (95% CI, 100–135 mya)
, this is a substantial conservation. It is therefore likely that extensive interchromosomal rearrangements have not occurred in either of the lineages leading to Gnathopogon and zebrafish since they diverged. This conclusion supports the findings from the comparative analysis of genomic structure among fish and mammalian species indicate that interchromosomal rearrangements are less frequent in teleost fishes than in mammals
[47–50]. Collinearity was also general between Gnathopogon and zebrafish, yet interruptions of collinearity were not rare. These data suggest that intrachromosomal rearrangements, such as inversions, occurred in either or both of the two lineages after the divergence of their ancestors.
Lineages including Gnathopogon, zebrafish, or the common carp Cyprinus carpio are major lineages within Cyprinidae that diverged in the early stage of the diversification of cyprinid fishes
[46, 51]. Cyprinid fishes show great cytogenetic variation. Their chromosome numbers range from 2n = 42 (Acheilognathus gracilis, or 2n = 30 if the taxonomically controversial Paedocypris carbunculus is placed within Cyprinidae
) to 2n = 417–470 (Ptychobarbus dipogon[54, 55]), with a mode at 2n = 50, followed by 2n = 48
. Thus, it has been suggested that the ancestral karyotype of cyprinid fishes was 2n = 48–50, and that polyploidisation occurred in several groups within the Cyprinidae
[57–63]. Genome-scale syntenic analyses between zebrafish and other cyprinid fishes have been conducted for common carp (2n = 100)
 and grass carp Ctenopharyngodon idella (2n = 48)
, both of which revealed some cases of interchromosomal rearrangements. The majority of the linkage groups in common carp have two-to-one relationships with zebrafish chromosomes, suggesting tetraploidisation in the common carp lineage. Those analyses also revealed a common carp linkage group sharing loci with two zebrafish chromosomes, which is speculated to have resulted from a chromosome recombination or transposition followed by fusion of homologous chromosomes during the process of diploidisation following tetraploidisation
. Most grass carp linkage groups are syntenic with zebrafish chromosomes, many of which have one-to-one relationships. One grass carp linkage group exhibits a one-to-two relationship with zebrafish chromosomes, suggesting chromosomal fusion. In grass carp, substantial macrosynteny and several cases of interchromosomal rearrangements are suggested. Considering the macrosynteny between Gnathopogon and zebrafish and the substantially straightforward trace of autopolyploidisation in the genome of common carp, the ancestral karyotype of Cyprinidae seems to be 2n = 50, concordant with the inference from the comparative karyological studies
[62, 63]. On the other hand, syntenic analysis between zebrafish and Mexican cave tetra Astyanax mexicanus (Characidae; 2n = 50) revealed cases of putative interchromosomal rearrangements
, such that syntenic loci of an Astyanax linkage group resided on several zebrafish chromosomes. These were suggested to be caused by gene duplications after the divergence of the lineages 248 mya (95% CI, 227–268 mya)
. Nevertheless, the analysis also revealed that synteny was conserved between Astyanax and zebrafish in numerous genomic regions. Combining the syntenic relationships and the genomic information of zebrafish, candidate genes for ecologically and evolutionarily important traits were identified in Astyanax.
Cyprinidae is the largest family of freshwater fishes. They have highly diverse morphology, ecology, and physiology, which are adapted to the vast range of habitats and resources they exploit
[55, 67]. Evolutionary ecological studies have been conducted in various cyprinid species concerning, e.g., adaptive radiation
[69, 70], and resource polymorphism
. However, the genomic basis and consequences of their diversification have not been extensively explored. In this study, Gnathopogon are suggested to be able to take advantage of the genomic information of a model cyprinid species, zebrafish, and its conserved synteny and collinearity with Gnathopogon. They may provide a prediction of candidate genes responsible for the traits related to phenotypic divergence that have ecological and evolutionary significance. Conservation of synteny and collinearity might be expected among cyprinid fishes, which could be advantageous for transferring genomic information between species
[3–5, 72]. This raises the prospect that evolutionary genomic studies of cyprinid fishes are accelerated by the interspecific exchange of information and by complementary studies between species.