A high-density genetic linkage map for Chinese perch (Siniperca chuatsi) using genotyping-by-sequencing (GBS)

Background: Chinese perch, Siniperca chuatsi (Basilewsky), is one of the most commercially important cultured fishes in China. In the present study, a high-density genetic linkage map of Chinese perch was constructed by genotyping-by-sequencing (GBS) technique with an F1 mapping panel containing 190 progenies. Result: A total of 2328 SNPs were assigned to 24 linkage groups (LGs), agreeing with the chromosome haploid number in this species (n = 24). The sex-averaged map covered 97.9% of the Chinese perch genome, with the length of 1694.3 cM and a marker density of 0.7 cM/locus. The number of markers per LG ranged from 57 to 222, with a mean of 97. The length of LGs varied from 43.2 cM to 108.2 cM, with a mean size of 70.6 cM. The recombination rate of females was 1.50 : 1, which was higher than that of males. In order to better understand the distribution pattern of segregation distortion between the two sexes of Chinese perch, the skewed markers were retained and used to reconstruct the sex-specific maps. The 16 SDRs were identified on 10 LGs of the female map, while 12 SDRs on 8 LGs of the male map. Among these LGs, 6 LGs matched between the sex-specific maps. Conclusion: This high-density linkage map could provide a solid basis for identifying QTLs associated with economically important traits, and for implementing marker-assisted selection breeding of Chinese perch.

oversea market. However, in the past three decades, the artificial breeding projects of Chinese perch only focused on the quantity of fish fry, but not on its quality. The genetic diversity of S. chuatsi stocks has been dramatically decreased during several generations, leading to the recession of excellent properties, such as growth performance and disease resistance. Therefore, genetic improvement programmes should be performed in S. chuatsi to consolidate and improve its excellent properties. Generally, the important agronomical traits are governed by interactions of multiple genetic loci which are difficult to be identified by traditional quantitative genetics methods [3].
The molecular markers has been an effective tool to detect and genetically locate the quantitative trait loci (QTLs) for genetic improvement [4]. Marker-assisted selection (MAS), utilizing molecular markers as indicators for indirect selection of a targeted trait, exhibits considerable advantages over conventional breeding approaches, especially for those traits which are difficult to measure and record on a continuous quantitative scale, and for those species with long generation times [5,6]. Genetic linkage map construction using polymorphic markers is one of the most crucial steps for MAS projects, and also provides a framework for comparative mapping study and chromosome-level assembly of whole-genome sequences [7-9].
The advances in technology of molecular marker exploitation and linkage analysis made it possible to construct genetic linkage map for diverse species [10]. Due to its high abundance, uniform distribution, biallelic nature and genetic stability, single nucleotide polymorphisms (SNPs) has gradually replaced traditional molecular markers such as amplified fragment length polymorphism (AFLP) and microsatellites (SSRs) in molecular genetics studies [11][12][13][14]. Genotyping-by-sequencing (GBS) could discover genome-wide SNPs in a rapid and cost-effective manner, which allows the processes of sequencing, discovery, and scoring of massive SNPs simultaneously in a single procedure [15]. GBS has successfully been applied to construct high-density linkage maps in a variety of different species [16][17][18].
The genomics studies and genetic analysis of the genetic improvement and artificial breeding of S. chuatsi are still in infancy. Although molecular markers have been reported previously [19][20][21], these studies could not meet the increasing industry demands, yet. In the present study, we used a panel of SNPs developed by GBS technique to produce a high-density genetic linkage map of S. chuatsi. The map developed in this study could be an important tool in elucidating genome structure and organization, anchoring the scaffolds in ongoing genome sequencing project, identifying QTLs associated with economically important traits, and implementing MAS for genetic improvement in Chinese perch breeding.

SNP Discovery in S. chuatsi Using GBS approach
To identify SNP markers present in the mapping family for map construction, ApeKI reduced representation libraries were constructed and sequenced using the Illumina HiSeq4000. After filtering the dirty raw reads, a total of 974.48 M reads were obtained with a mean of approximately 10.55 and 5.02 million reads for parents and offspring, respectively. A total of 65,535 putative GBS markers were discovered, and of which, 12,466 were informative. From these 12,466 informative markers, 2519 high-quality SNPs were retained for linkage analyses.

Linkage mapping
Linkage analyses were conducted by Joinmap 4.0 software with LOD threshold of 5.0 for female and male maps, and of 6.0 for sex-averaged map. After eliminating the markers with significant segregation distortion, there were 24 linkage groups in the final linkage maps with 1030, 880 and 2328 SNPs markers in the male, female and sex-averaged map, respectively (Table 1; Table 2; Figure 1). The male, female and sex-averaged map covered a total genetic distance of 1885.9 cM, 1262.8 cM, and 1694.3 cM respectively, with an average interval of 1.2 cM, 2.1 cM, and 0.7 cM (Table 2).

Marker distribution and intermarker space
The SNP markers located in the linkage groups were unevenly distributed.
LG 1 was the largest linkage group in all the three maps, containing a total of 113 (male map), 169 (female map), and 222 markers (sex-averaged map), respectively. LG 24 was the smallest one, with 7, 13 and 37 markers, respectively (Table 1). To further investigate the marker distribution, intermarker space of each linkage group was calculated in sex-averaged map.
Most intervals in this sex-averaged map were less than 5 cM with its percentage of 97.9% (2256/2304) ( Figure 2).

Estimation of genome size and coverage
The genome sizes of the male, female, and sex-average maps were estimated to be 1328.3 cM, 2010.2 cM, and 1731.0 cM, respectively. Based on these estimated sizes, genome coverages were calculated to be 97.9% (male map), 93.8% (female map) and 95.1% (sex-average map), respectively. The haploid genome of S. chuatsi was reported to be approximately 0.8 Gb in a previous study [22]. In our study, linkage map length was estimated to be 1731.0 cM. Thus the relationship of physical to genetic distance could be estimated to be 462.2 Kb/cM. Therefore, the estimated physical distance between the adjacent markers ranged from 0 Kb to 9.2 Mb with an average of 323.5 Kb in the sexaveraged map.

Segregation distortion
In the present study, 150 and 239 segregating markers (P < 0.05) were assembled into the SD male and female map, respectively (Table 3). These markers accounted for 12.7% in the SD male map and 21.4% in the SD female map. The segregation distortion markers among LGs also showed an uneven distribution. In the male map, the percentage of segregation distortion markers ranged from 0.0% (MSD-LG12, MSD-LG15, MSD-LG23, MSD- LG24) to 41.7% (MSD-LG22), while in the female map, it ranged from 0.0% (FSD-LG5, FSD- LG6) to 90.8% (FSD-LG24). The 12 and 16 SDRs were detected on 8 LGs in the male map and 10 LGs in the female map, respectively. Among these linkage groups, 6 LGs containing SDRs were shared by male map and female map.

Sex differences in recombination rate
The differences in both the number of markers and recombination rate were observed between the sex-specific genetic linkage maps. Overall, the marker number in the male map was larger than that in the female map, whereas, the total length was shorter. The relative recombination ratio between female and male was 1.5/1 for the overall LGs, but the recombination ratios were different for individual LG, ranging from 0.5 to 10.6 ( Table   4). On 18 out of 24 LGs, higher recombination rates were observed in male map than in female map. While, on the other 6 LGs, recombination rates were higher in female map than in male map.

Discussion
In this study, the linkage map was described for S. chuatsi, a commercially important freshwater fish in China, based on 2328 SNP markers. These markers were located on 24 LGs, consistent with the haploid chromosome number of S. chuatsi (n = 24). The sexaveraged total map length was 1694.3 cM, covering 97.9% of the expected length (1731.0 cM). According to the S. chuatsi genome size of 800 Mb reported in previous study and the mean intermarker space of 0.7 cM estimated in our study, the present map contained a marker approximately 0.3235 Mb, could provide sufficient marker density for QTL mapping and other genetic or genomic analyses. The majority of the marker intervals were found to be less than 5 cM (97.9%), and only 0.7% were more than 10 cM in the present genetic linkage map of S. chuatsi, indicating that SNP markers were well-distributed along the whole chromosomes of S. chuatsi (Figure 2). Large intervals observed in genetic linkage maps were a common phenomenon, which has been reported in other animals [23][24][25][26].
These large intervals indicated that some regions in the genome exhibited the very little recombination or they were highly conserved regions. In these regions, the informative SNP markers were reported to be rare or difficult to identify [27,28].
Sex difference in recombination rates was a common phenomenon across species with the least recombination usually observed in the heterogametic sex [29]. Though no heteromorphic chromosomes were reported in fishes [30], similar sex difference in recombination rates was reported in fish, such as a ratios of 1.18:1 in the white grouper [31], 2.00:1 in grass carp [10], 1.45:1 in the kelp grouper [32], 1.03:1 in orange-spotted groupers [33], and 1.69:1 in Arctic char [34]. This study found that females also had a higher recombination rate (1.50:1) than male for S. chuatsi.
The knowledge of relative recombination rates of females and males is of great practical significance. High recombination rate plays a positive role in distinguishing closely-linked markers, which could improve the linkage map resolution for QTL mapping of economic traits. In contrast, low recombination rate is helpful for identifying the approximate location of mutation in a chromosome arm [30]. Moreover, double mutants in cis with a low recombination rate were reported to be linked frequently, which was extremely useful for epistasis analysis [30]. Therefore, future genetic research such as QTL mapping could take the recombination differences between sexes into consideration in S. chuatsi and other Siniperca species.
Segregation distortion is a phenomenon that genotype frequencies at a locus deviate from their Mendelian expectations [35], and it is considered to be a feature of genetic maps in many species, such as red drum [6], Atlantic Killifish [36], bighead carp [37], and Nile tilapia [38]. Many studies have already reported that inclusion of distorted markers in map construction has slight impact on the accuracy for determining maker orders and map length [39,40].For better understanding the distribution pattern of segregation distortion between the two sexes in S. chuatsi, all the distorted markers were retained and used to reconstruct the sex-specific maps. Finally, 150 and 239 distorted markers were integrated into the male and female map, respectively.
In the present study, part of segregation-distorted loci was randomly scattered on the sexspecific maps of S. chuatsi, which could be caused by abiological factors, such as the bias generated from the GBS by which sequencing is difficult in highly repetitive regions and data points missing in SNP calling. However, majority of distorted SNPs were clustered on some LGs and/or in some regions within an individual LG. Such a clustering is impossible to result from abiological factors, Instead, it might be due to SDR, a region containing at least three adjoining markers exhibiting distorted segregation. It was reported that SDRs might be related to deleterious genes that potentially affected survival and fitness of organism [41,42]. In the present study, 16 SDRs were identified on 10 LGs in the female map, while 12 SDRs were found on 8 LGs in the male map. Among these LGs, 6 LGs showed corresponding relations between the male and female maps. Therefore, SDRs on the 6 shared LGs might be used to identify potential lethal or deleterious genes on certain chromosomes of S. chuatsi, thus providing useful instruction for breeding projects.

Conclusions
In the present study, a high-density genetic linkage map was constructed for Chinese No reference genome of S. chuatsi was available to align raw reads for SNP detection. Therefore, the program Stacks was utilized to discover SNP markers. This program is available for free download from http://creskolab.uoregon.edu/stacks/ [44]. After the removal of adaptor sequences, short DNA sequences (reads) were sorted to individuals from the sequencing pools by barcode sequences by using the program process_radtags.
Reads of each individual were merged into populations and the consensus sequences were generated using the "ustacks" and "cstacks" program. SNP calling and genotyping were conducted by sstacks and genotypes programs, respectively. Any SNP with > 20% missing data was excluded from further segregation analysis.

Map construction
Mendelian segregation at each locus was subjected to Chi-square test. Only SNPs with P ≥ 0.05 were selected for the map construction. Linkage groups were created with JoinMap 4.0 program under CP algorithm. Four marker segregation types were used in the genetic linkage mapping: Type I (lm × ll) was heterozygous in the male (lm) and homozygous in female (ll) which was recorded for paternal linkage mapping; Type II (nn × np) were homozygous in the male (nn) and heterozygous in female (np), which was recorded for maternal linkage mapping; Type III (eg × ef) was heterozygous in both parents with three alleles (two sex-specific alleles and one shared allele); Type IV (hk × hk) was heterozygous in both parents with two common alleles, which was recorded as anchor

Consent for publication
Not applicable.

Availability of data and materials
We declared that materials described in the manuscript, including all relevant raw data, will be freely available to any scientist wishing to use them for non-commercial purposes, without breaching participant confidentiality.

Competing Interests
The authors declare no conflict of interest. The founding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results.     Table 4 Comparative recombination rates between male and female S. chuatsi.
LG Sex-specific linkage groups Female: male ratio