Microsatellites for linkage mapping in tilapia species
Although some microsatellites developed in one species can be used in other closely related species, the success rate of cross-species amplification is usually low . In this study, microsatellites derived from the genome and EST sequences of Nile tilapia were used to construct a linkage map for saline tilapia. Among 576 microsatellites from Nile tilapia, 407 amplified specific and polymorphic products in Mozambique tilapia and red tilapia, indicating that the majority (70.7%) of microsatellites could be universally used in Nile tilapia, Mozambique and red tilapia for genetic and genomics studies.
The linkage maps in tilapias and recombination rates
Several genetic linkage maps have been constructed in tilapias previously [18–20]. The first two maps, one in Nile tilapia, and another based on a three-way cross family, were constructed mainly using dominant AFLP markers [18, 20]. Though a female map of Mozambique tilapia was constructed in the three-way cross family, it only consists of 14 linkage groups with 62 loci . The latest linkage map in tilapia was constructed using an F2 interspecific hybrid family between Nile tilapia and blue tilapia, consisting of 24 linkage groups . In the present study, we constructed the first integrated linkage map in Mozambique and red tilapia. Well-known for their high salt tolerance, Mozambique tilapia and its hybrid including red tilapia have been widely used in the aquaculture and breeding of saline tilapia [4–6]. The consensus and linkage maps of Mozambique tilapia and red tilapia, all consisted of 22 linkage groups. 282 mapped markers were derived from ESTs, and 125 of them were annotated by known genes, including genes related to immunity and growth. The potential inversion between Mozambique tilapia and red tilapia found in LG14 in the present study, along with the differences of karyotypes among tilapias reported before , indicates that some significant differences may exist in chromosomes between different tilapia species. With an average inter-marker distance of 3.3 cM, the present map provides a useful resource for QTL mapping of important commercial traits, comparative mapping and positional cloning of interesting genes in saline tilapia. However, we have noticed that one marker space on LG3 was still larger than 20 cM. Therefore, it is essential to map more DNA markers in this space to facilitate QTL mapping for important traits.
One hundred and nineteen microsatellite markers from 24 linkage groups of a previous map, along with 282 markers from ESTs were assigned to 22 linkage groups. The marker order on each linkage group among the present and the previous maps were almost identical (see Additional file 5: Table S3). The LG21 and LG24 of the previous map merged into LG16 and LG8, respectively. These merges reduced the number of linkage groups from 24 to 22, equal to the chromosome pair number in Mozambique tilapia and Nile tilapia , and resolved the discrepancy between linkage group number and chromosome number in the latest linkage map of tilapia .
Comparing the linkage groups between Mozambique tilapia and red tilapia revealed significant differences in the recombination rate on LG2 and LG15. Similar results have been reported in other species, such as fox and dog . These results suggest that recombination rate is unequal in different genome regions and species. Different ratios of recombination rates between females and males have been reported in a number of species. Females usually have a higher recombination frequency than males. For example, the female: male recombination ratios were 8.26:1 in Atlantic salmon , 1.6:1 in channel catfish , 2.74:1 in zebrafish  and 2:1 in grass carp . However, the linkage map of females was shorter than that of males in striped bass . Nearly identical recombination rates between females and males have been referred in hybrid tilapia previously . In this study, the ratio of lengths of common intervals in females and males was 1.08, indicating that males had a similar recombination frequency of the whole genome as females in tilapia.
Syntenies between different fish species
The sequences of 226, 188, 159, and 88 of the 401 mapped markers had significant hits in the whole genome sequences of stickleback, medaka, puffer fish, and zebrafish, respectively, suggesting that stickleback is more closely related to tilapia than the other three model fishes. All linkage groups in tilapia mainly corresponded to one or two linkage groups, or chromosomes in the four model fishes (Table 3), implying the high evolutionary conservation of chromosomes in these five fish species. Some conserved syntenies of fish chromosomes also were reported in medaka , pufferfish , seabream , catfish , grass carp  and striped bass .
Among all 22 linkage groups in tilapia, LG7 corresponded to two chromosomes in all four model fishes, and LG22 corresponded to two chromosomes in stickleback and medaka, implying that these two groups may be formed by two independent fusion events. It is believed that the ancestral karyotype of cichlids consisted of 24 chromosome pairs . However, how the haploid chromosome number of most tilapiines reduced to 22 remains unclear. One hypothesis is that the largest chromosome in tilapiines came from the fusion of three chromosomes . However, our results suggested that the modern karyotype of tilapiines may be formed by two separate fusions of two sets of two independent chromosomes. These two fusions may have lead to the reduction of the 24 chromosome pairs to 22 pairs in most tilapiines.
Among all four model fishes compared, only medaka from superorder Acanthopterygii has a haploid chromosome number of 24, which conforms to the presumed ancestral karyotype of cichlids. In addition, each linkage group in tilapia correlates to one homologous chromosome in medaka except LG7 and LG22, each of which coincides with two chromosomes. The syntenies between proto-chromosomes in vertebrates and chromosomes in medaka have been previously speculated . In the present study, the simple and clear correspondences between 22 linkage groups of tilapia and 24 chromosomes of medaka indicate that medaka may possess the most possible ancestral karyotype of cichlids in four model fishes.
A potential inversion on LG 14 was found between Mozambique tilapia and red tilapia. Differences of karyotype between Nile tilapia and Mozambique tilapia have been reported previously. Mozambique tilapia had four metacentric or submetacentric chromosomes and 40 acrocentric or subtelocentric ones, while Nile tilapia only had two metacentric or submetacentric chromosomes . Our red tilapia originated from the hybrid of Mozambique tilapia and Nile tilapia. The difference of marker order on LG14 between the two tilapias may be caused by the hybrid origination of red tilapia.
Mapping of sex determination loci
Two sex determination systems, XY and WZ have been identified in tilapias [26, 28, 29, 42, 43]. Three sex determining loci have been identified on LG1, LG3 and LG22 of tilapias [26, 27], respectively, indicating the complexity of the sex determination in tilapia,. A sex-determining locus on LG3 was reported in female heterogametic (WZ-ZZ) tilapias including T. mariae, O. karongae, O. tanganicae and Israeli stain of O. aureus. A sex determining locus on LG1 was found in male heterogametic (XX-XY) Nile tilapia and T. zillii, and the sex determining locus on LG22 was only found in Nile tilapia . In the present study, only one sex determination locus on LG1 was found in our reference families produced by Mozambique tilapia males, which were identified as male heterogametic. However, Cnaani et al. found that markers on both LG1 and LG3 were associated with sex in three families of Mozambique tilapia and two families of the Egyptian strain of blue tilapia, and the sex determination of these reference families could not be defined as male or female heterogametic . Mozambique tilapia and blue tilapia were known as male heterogametic and female heterogametic, respectively [28, 29]. Our results are identical to the traditional view and differ from the results of Cnaani et al. . This divergence may be caused by the different genetic backgrounds of reference families and strains. Interspecies crosses were prevalent in the tilapias, and most of the hybrids were fertile and could reproduce offspring as purebred fish [44, 45]. These hybrids may spread in the farmed strains as well as in wild populations, and lead to the complex pattern of sex determination in some tilapia strains.
Since the sex-determining locus on LG1 was identified mainly in tilapias with the XY sex determination system, and the sex-determining locus on LG3 was identified mainly in tilapias with the WZ system , we may conclude that the sex determining locus on LG1 determined the male heterogamete and the sex determining locus on LG3 determined the female heterogamete in tilapiine species. As these sex determining loci existed in closed species from the genus of both Oreochromis and Tilapia, it seems that the two sex-determining loci may both emerge prior to the differentiation of Oreochromis and Tilapia, and underwent independent evolution in different species. Alternatively, the tilapiine species may only have one ancestral sex determination locus, which more likely to be the sex-determining locus on LG3 as predicted by some researchers . After the differentiation of Oreochromis and Tilapia, another sex-determining locus appeared and spread to specific species by interspecific hybridization.
We have also identified a XY sex-determining locus on LG22 in red tilapia. This sex-determining locus was reported only in Nile tilapia . Our local red tilapia strain in Malaysia and Singapore originating from the hybrid between Nile tilapia and Mozambique tilapia . The sex-determining locus on LG22 in red tilapia may originate from Nile tilapia instead of Mozambique tilapia. As it only was found in Nile tilapia and red tilapia, the sex-determining locus on LG22 may have a later origination than the sex-determining loci on LG1 and LG3. In this study, the sex-determining loci on LG1 and LG22 were identified in different types of families, which were produced by Mozambique tilapia males or red tilapia males, respectively. There was no family found to have these two sex-determining loci at the same time. Further research is needed to understand the interactions between the sex-determining loci on LG1 and LG22.
About 30% of individuals from two families produced by the red tilapia males showed no association between sex and genotypes of LG1 or LG22, indicating that there are more genetic or environment factors which may be involved in the sex determination in red tilapia. Due to its hybrid origination, red tilapia may have more complex mechanisms of sex determination than Mozambique tilapia.