A newly developed genetic sex marker and its application to understanding chemically induced feminisation in roach (Rutilus rutilus)

Oestrogenic wastewater treatment works (WwTW) effluents discharged into UK rivers have been shown to affect sexual development, including inducing intersex, in wild roach (Rutilus rutilus). This can result in a reduced breeding capability with potential population level impacts. In the absence of a sex probe for roach it has not been possible to confirm whether intersex fish in the wild arise from genetic males or females, or whether sex reversal occurs in the wild, as this condition can be induced experimentally in controlled exposures to WwTW effluents and a steroidal oestrogen. Using restriction site‐associated DNA sequencing (RAD‐seq), we identified a candidate for a genetic sex marker and validated this marker as a sex probe through PCR analyses of samples from wild roach populations from nonpolluted rivers. We also applied the sex marker to samples from roach exposed experimentally to oestrogen and oestrogenic effluents to confirm suspected phenotypic sex reversal from males to females in some treatments, and also that sex‐reversed males are able to breed as females. We then show, unequivocally, that intersex in wild roach populations results from feminisation of males, but find no strong evidence for complete sex reversal in wild roach at river sites contaminated with oestrogens. The discovered marker has utility for studies in roach on chemical effects, wild stock assessments, and reducing the number of fish used where only one sex is required for experimentation. Furthermore, we show that the marker can be applied nondestructively using a fin clip or skin swab, with animal welfare benefits.


Abstract
Oestrogenic wastewater treatment works (WwTW) effluents discharged into UK rivers have been shown to affect sexual development, including inducing intersex, in wild roach (Rutilus rutilus). This can result in a reduced breeding capability with potential population level impacts. In the absence of a sex probe for roach it has not been possible to confirm whether intersex fish in the wild arise from genetic males or females, or whether sex reversal occurs in the wild, as this condition can be induced experimentally in controlled exposures to WwTW effluents and a steroidal oestrogen. Using restriction site-associated DNA sequencing (RAD-seq), we identified a candidate for a genetic sex marker and validated this marker as a sex probe through PCR analyses of samples from wild roach populations from nonpolluted rivers. We also applied the sex marker to samples from roach exposed experimentally to oestrogen and oestrogenic effluents to confirm suspected phenotypic sex reversal from males to females in some treatments, and also that sex-reversed males are able to breed as females. We then show, unequivocally, that intersex in wild roach populations results from feminisation of males, but find no strong evidence for complete sex reversal in wild roach at river sites contaminated with oestrogens. The discovered marker has utility for studies in roach on chemical effects, wild stock assessments, and reducing the number of fish used where only one sex is required for experimentation. Furthermore, we show that the marker can be applied nondestructively using a fin clip or skin swab, with animal welfare benefits.

K E Y W O R D S
endocrine disruption, intersex, RAD-seq, Rutilus rutilus, sex marker, sex reversal
Over the past two decades, the roach (Rutilus rutilus; Figure 1) has been widely adopted in central and Northern Europe as a species for studies into endocrine disruption (Burkhardt-Holm, Peter, & Segner, 2002;Gerbron et al., 2014;Jobling et al., 1998;McGee, Brougham, Roche, & Fogarty, 2012;Minier, Caltot, Leboulanger, & Hill, 2000;Noaksson, Tjärnlund, Bosveld, & Balk, 2001;Trubiroha et al., 2010;Wiklund, Lounasheimo, Lom, & Bylund, 1996). These studies have been aided by the fact that the roach is a gonochorist (developing as either a male or female, thus avoiding complications where sex changes occur as part of the natural process of sexual development), a good understanding of the normal reproductive development (Paull, Lange, Henshaw, & Tyler, 2008), and the availability of analytical tools for assessing oestrogenic effects in this species (Hamilton & Tyler, 2008;Harris et al., 2011;Lange et al., 2008Lange et al., , 2009Tyler, van der Eerden, Jobling, Panter, & Sumpter, 1996;Tyler et al., 2009). Surveys of wild roach populations in UK rivers contaminated with wastewater treatment works (WwTW) effluent have shown high incidences of intersex (Enivornment Agency, 2004;Jobling et al., 1998) and moderately to severely intersex fish have reduced milt volume and sperm density, and reduced fertility (Jobling, Beresford, et al., 2002;Jobling, Coey, et al., 2002) which might have population level consequences for roach in some UK rivers (Harris et al., 2011). These feminised phenotypes observed in wild roach populations can be induced through controlled chronic exposures to both EE2 or to a treated WwTW effluent. Furthermore, exposure to both 4 ng EE2/L or to a full strength treated WwTW effluent have been shown to result in phenotypically all-female populations (Lange et al., 2009;Lange, Paull, Hamilton, Iguchi, & Tyler, 2011), although that sex reversal could not be confirmed in the absence of a genetic sex probe. The ability to determine the fish's genetic sex is crucial for assessing whether fish in wild populations undergo sex reversal as a consequence of pollution exposures and understanding whether sex reversed fish reproduce in this ecologically important cyprinid species.
The development of restriction-site-associated DNA sequencing (RAD-seq) (Miller, Dunham, Amores, Cresko, & Johnson, 2007) has made the discovery of genetic markers in wild populations of non-model species more readily attainable. Application of RADseq has resulted in the discovery of informative polymorphic markers and in the construction of linkage maps for a variety of fish species (Houston et al., 2014;Leitwein et al., 2017;Manousaki et al., 2016;Willing, Hoffmann, Klein, Weigel, & Dreyer, 2011).
Here, we used RAD-seq to isolate a male-specific genetic sex marker in roach and verified this using phenotypically assigned males and females derived from clean water study sites. The sex marker was then applied to a series of samples firstly, to confirm male to female sex reversal in roach for controlled chronic exposures to EE2 and treated WwTW effluents. We then applied the sex probe to confirm that induced intersex in wild roach populations arises due to feminisation of genetic males, and to assess evidence for complete sex reversal in wild roach populations of roach inhabiting rivers heavily contaminated with oestrogenic WwTW effluents. We also sought to investigate the use of the genetic marker to sex roach prior to the appearance of a histologically sex differentiated gonad and to sex roach with non-destructive tissue (fin clip and skin swab) sampling methods. Our results illustrate the wide-ranging applications of the sex probe for studies on roach sex and the effects of chemicals (here oestrogens) on roach populations.

| Fish populations and sampling
For RAD-seq and subsequent analysis for sex marker isolation, fish were sampled from two independent sites ( b Fish were only one generation away from the wild and parental fish were all caught from the same location which was also the same as for population E. c Some putative sex-reversed males appeared to have bred as females. the second population were the offspring of parental fish that originated from the River Trent, in the Midlands (site 2) and that had been bred in the laboratory. One hundred males and 100 female roach were sampled from each population.
For validation and application of the sex marker, samples from a total of 429 roach were analysed: (i) Eighty for the initial validation of the isolated sex marker; (ii) Eighty-two fish to verify complete male to female sex reversal can occur for exposure to environmental oestrogens and; (iii) two hundred and sixty-seven fish to test that intersex roach in the wild derive from genetic males. These samples were derived from previous studies (David, Lange, Abdul-Sada, Tyler, & Hill, 2017;Defra, 2009;;Harris et al., 2011;Lange et al., 2009Lange et al., , 2011 and details on the different roach populations used are provided below and in Table 1. The fish used for validation of the sex marker (n = 80, 42 males and 38 females, based on gonadal phenotype) originated from roach populations collected at five independent field sites across England with no known pollution inputs (populations A-E, Table 1).
Application of the sex probe to verify complete male to female sex reversal can occur after exposure to environmental oestrogens, was carried out on samples derived from 82 roach that had been chronically exposed to EE2 or a treated WwTW effluent and resulted in phenotypic all-female populations (Lange et al., 2009(Lange et al., , 2011. In the exposure study to a WwTW effluent by Lange et al. (2011), a subsequent competitive breeding experiment that included parentage analysis on the offspring suggested that presumptive males in the effluent-exposed treatment reproduced as females, but this could not be confirmed at the time due to the lack of a genetic sex marker.
A further application of the sex probe involved testing the hypothesis that intersex roach in the wild derive from genetic males and to investigate for evidence for possible complete reversal in roach. To this end, the sex marker was applied to 267 (177 males and 90 females) samples collected at sites heavily contaminated with oestrogenic WwTW effluents. For this, samples were used from four independent roach populations that varied in the proportions and degrees of intersex (populations F-I, Table 1).
For each individual fish used in the study, a fin clip was collected from the caudal fin and preserved in ethanol until DNA extraction.
All procedures employed were in accordance with UK Home Office regulations (Animals (Scientific Procedures) Act 1986).

| DNA extraction and RAD-seq library preparation
For RAD-seq library preparation, total DNA was extracted from individual fish using a small fraction of the preserved fin tissue using Qiagen DNeasy Blood & Tissue columns (Qiagen, Germany), with a few modifications of the manufacturer's protocol: Briefly, tissue lysis was carried out with proteinase K (20 µl of a 20 mg/ml stock) under gentle rotation at 37°C for 1 hr, followed by incubation with RNase A (4 µl of a 100 mg/ml stock) prior to loading onto the extraction column. DNA was eluted from the extraction columns with 100 µl of AE buffer. After DNA extraction, DNA and RNA concentrations were determined by fluorometry using a Qubit fluorometer. DNA quality was assessed spectrometrically with a NanoDrop 1,000 Spectrophotometer to measure the ratio of absorbance at 260 nm and 280 nm. DNA integrity was assessed by gel electrophoresis. Validation and application of the sex marker were performed on DNA extracted from individual fin clips using the HotSHOT method (Truett et al., 2000). Briefly, a small section of fin tissue was incubated in 75 µl alkaline lysis reagent (25 mM NaOH, 0.2 mM Na 2 EDTA) at 95°C for 45 min. The samples were placed on ice for 5 min before the adding 75 µl neutralising reagent (40 mM Tris-HCl, pH 5.0).

| RAD-seq identification of a male-specific marker and locus extension
The software Stacks version 0.99 (Catchen, Amores, Hohenlohe, Cresko, & Postlethwait, 2011) was used to assemble markers from the RAD-seq data. Raw reads were cleaned and demultiplexed using process radtags, with options to clean (-c), quality filter (-q) and rescue barcodes (-r). Demultiplexed clean reads were formed into polymorphic loci using denovo_map.pl with the minimum number of reads set to 2 (-m), number of mismatches set to 2 (-M) and the number of differences between the four DNA pools set to 4 (-n), using parameter optimisation as outlined elsewhere (Paris, Stevens, & Catchen, 2017). Catalogue consensus sequences built from the males and females from both sites were used for downstream coverage analysis.
In order to identify sex-specific markers, we undertook a read mapping coverage analysis of the RAD-seq catalogue consensus sequences. Clean, demultiplexed forward reads from the males and females of both sites were mapped to the Stacks' catalogue consensus sequences (687,048 sequences, see Results) using bwa mem (Li, 2013).
Resulting bam files were used as input for read coverage analysis using a custom script (RAD_coverage_read1.py; available at https://github. com/tceza rd/RADma pper). The script takes each aligned RAD read and calculates per consensus coverage. As well as calculating raw read coverages, samtools (Li et al., 2009)  Given that Stacks version 0.99 was used, the ability to assemble paired-end read contigs within the Stacks software was not possible at the time of analysis. Therefore, we used a different, but widely adopted approach for assembling contigs from the paired-end read data. We did this in the following way: The reverse reads (read 2) were aligned to the catalogue consensus sequences, and were grouped using the read 1 alignment bam files. In order to assemble long and contiguous contigs that could be used to design primers, several different assembler software were used (https://github.com/ tceza rd/RADma pper/blob/maste r/bin/RAD_assem ble_read2.py) and assessed (https://github.com/tceza rd/RADma pper/blob/maste r/bin/RAD_assess_read2_contig.py). The software IDBA-UD version 1.0.9 (Peng, Leung, Yiu, & Chin, 2012), assembled the longest contigs from the paired-end reads, using the following parameters: min_contig = 200, --mink = 40, --min_count = 8 and --min_support = 4).
The paired-end contig from each of the three identified markers were used downstream for PCR probe design. All markers were tested as candidates for sex-specific genetic markers using PCR (see Section 2.4.1., Figures S1 and S2). The paired-end contig of Marker_780797 successfully distinguished between male and female roach (see Results, Figure 2 and Figure S2).
As the read-mapping method relies on the identification of restriction site-associated presence-absence polymorphisms, we explored whether the identified sex marker was more likely to be the result of high sequence divergence between males and females, or a male-specific insertion-deletion polymorphism. If the presence-absence site was the result of the former, we might expect other contigs in the data set to show high sequence divergence. We assessed this by performing an allele frequency divergence analysis. The allele frequencies of all contigs were calculated (RAD_allele_fre-quency_count.py), which reports the number of ATCG nucleotides if coverage Q20 is > 5. The allele frequencies were filtered so that major alleles were reported as different between males and females at a major allele frequency greater than 0.8. Analysis of potential high sequence divergence between male and female haplotypes was also assessed by kmersing the male marker contig and aligning the resulting kmers to the female-derived catalogue consensus tags. The contig was kmerised using jellyfish count version 1.1.4 (Marçais & Kingsford, 2011), by kmerising it to word lengths of 11, 15, 21, 31.
Finally, in order to identify a putative function of the male-specific region, all reads with a match against the putative sex marker sequence were extracted from the RAD-seq data set, including both members of the paired-end read (even if only one member of the pair actually matched the target sequence), and pairs of reads were then assembled using SPAdes version 3.6.2 (Bankevich et al., 2012 Figure S3).

| Roach sex marker
In the first instance, all three markers were tested as candidates for sex-specific genetic markers using polymerase chain reaction (PCR) (see Supporting Information Methods and Figure S2). Using Primer-BLAST (Ye et al., 2012), primers were designed to the sequences identified as present in males and absent in females (see

| Internal transcribed spacer
Roach can breed (thus hybridise) with other cyprinid fish species including bream (Abramis brama) and rudd (Scardinius erytrophthalmus) (Pitts, 1994). Therefore, all individuals used in this study (those used for the RAD-library pools and those for the validation and application approach) were verified as genetically pure-bred roach using species-specific forward primers and an universal cyprinid reverse primer designed to the ITS1 nuclear ribosomal DNA region according to Wyatt, Pitts and Butlin (2006). In addition, this amplification also served as positive control for successful DNA extraction.

| Locus extension
Primers were designed at the outer ends of the newly assembled scaffold to experimentally confirm the sequences (Table 2; Figure S3).
PCR reactions were carried out using Q5 High-Fidelity 2X Master Mix (New England BioLabs), 0.5 μM of each forward (Rr_sml_f1,

| Genetic sex identification prior to histological gonadal differentiation
Roach fry at 54 days post hatch (dph) were obtained from the of phenotypic sex was carried out as described previously .
DNA was extracted from fin clips and mucus swabs using the HotSHOT method or Chelex extraction protocol (Estoup, Largiadèr, Perrot, & Chourrout, 1996), respectively. All samples were genetically identified as pure roach origin using the ITS1 nuclear ribosomal DNA region according to Wyatt et al. (2006). For the early life stage fish, genetic sex of each fish was determined by PCR using the Multiplex PCR Kit (Qiagen) with individual reactions for each primer combination described above for marker_780797.
The PCR reaction mix included the kit's Q-solution, 0.2 µM of each primer and 2 µl DNA in a total reaction volume of 20 µl. An initial activation step of 95°C for 15 min was followed by 45 cycles of denaturation (30 s at 94°C), annealing (90 s at 60°C) and extension (1 min at 72°C), followed by a final extension of 10 min at 72°C. Amplicons were resolved on 1.5% agarose gels. For the later roach stage fin clips and mucus swabs, PCRs were performed as described above (Section 2.4.1).

| Restriction-site associated DNA sequencing (RAD-seq) and bioinformatical identification of putative sex markers
Sequencing of the PstI RAD libraries produced between 30,435,402 and 37,086,239 raw reads, of which between 24,950,577 and 31,846,048 were retained after cleaning (Table S1). De novo assembly of the first read was achieved using Stacks (Catchen et al., 2011), resulting in 737,393 consensus RAD-tags, for which coverage varied from 14x to 17x (Table S1).
To detect presence-absence sex markers, we undertook a read mapping analysis and selected sites where five or more reads mapped in the male sex for both sites, and no reads mapped in either of the female pools. This identified three male-specific contigs:  (Table S2). The associated paired-end reads were assembled into contigs using IDBA_UD, which successfully assembled 687,048 paired-end contigs. The paired-end contigs resulting from all three markers were tested as candidates for sex-specific genetic markers using PCR ( Figure S2) and marker_780797 successfully distinguished between male and female roach (Figure 2 and Figure S2).
In an attempt to disentangle the origin of the presence-absence polymorphism, two additional analyses were performed. Analysis of allele-frequency divergence between males and females on other contigs (using a major allele frequency cutoff of 0.8), did not result in the recovery of any other sex-linked contigs. We also kmerised the male marker and aligned it to the female consensus catalogue tags.
Kmers of length 11,15 and 21 did not align to any female consensus tags, and although word length 31 did show alignment to two female consensus tags, further inspection of this alignment using mafft, suggested that the alignment does not represent true sequence homology (Table S3). We therefore hypothesise that the identified sex marker is probably the result of a male-specific insertion-deletion polymorphism, rather than the result of high haplotype divergence between the sexes (although further research would be required to confirm this; see Discussion).

| Validation of the sex-specific marker
Seventy-nine of the 80 roach (corresponding to 99%) used to validate the sex probe were identified correctly according to their gonadal sex, suggesting that the sex marker distinguishes genetic sex in these roach populations with high fidelity (Figure 3).

| Verification of complete male to female sex reversal for oestrogen exposed fish
The marker proved complete male-to-female sex reversal occurred in samples analysed from experiments for controlled long-term exposures to EE2 (4 ng/L) and to a full strength WwTW effluent.
These studies used roach from independent populations and both oestrogenic exposures resulted in phenotypic all-female populations (based on gonadal histopathology), whereas the sex marker identified genetic male:female ratios of 52%:48% and 53%:47% for the EE2 and effluent experiment, respectively (Figure 4).

| Identifying genetic sex in populations with intersex fish
The sex marker was subsequently applied to identify the genetic sex

| Determining genetic sex prior to gonadal sex differentiation
Of the 44 roach that were sexed via histopathology at seven months old, 19 were phenotypic males and 25 phenotypic females and the genetic sex determined agreed with these phenotypic sexes for all females and for 14 of the 19 of the males. Thus, five phenotypic males were identified as genetic females. All fish were kept individually to allow us to trace each fish. For the same individuals at 70 dph, prior to gonadal differentiation, all 25 phenotypic females were correctly assigned with the genetic sex probe, but six of the 19 males were identified as genetic females ( Figure 6).

| Extending the probe sequence
Extracting all reads with a match against the sex marker sequence from the full data set yielded 2,308 pairs of reads. The subsequent assembly resulted in 53 scaffolds, one of which contained the original target sequence. The length of this scaffold was 1,514 bp, around three times the length of the original sequence. A BLASTN

F I G U R E 3
Validation of the sex-specific marker. The genetic sex of 80 roach collected from five independent, unpolluted river sites across England (populations A-E; see Table 1) identified by PCR. The phenotypic sex of these fish had been determined previously via gonadal histopathology, or based on the release of gametes in mature fish. Each group of stacked bars shows phenotypic sex (left hand bars) and the identified genetic sex (right hand bars) for each population; the lower segment of each bar corresponds to female fish and the upper segments male fish F I G U R E 4 Identification of genetic males in phenotypic allfemale roach populations. Genetic sex of fish derived from two controlled long-term laboratory exposures to EE2 or effluent (Lange et al., 2009(Lange et al., , 2011. The majority of fish were phenotypic females with a small proportion of fish where sex could not be assigned by gonadal histopathology search against the NCBI Nucleotide database mapped this scaffold to GenBank accession EU621898.1, representing a 300 kb region of the Atlantic salmon (Salmo salar) genome that contains the gene for growth hormone 1 (GenBank: ACD37713.1) among other proteincoding genes (von Schalburg et al., 2008). Primers designed to the outer ends of this newly assembled scaffold amplified a PCR product of the expected size in males, but did not yield a PCR product in females. The PCR products were subsequently sequenced. After trimming low quality ends, 1,332 bp of the experimental sequence aligned to the assembled scaffold with 100% identity ( Figure S3).

| D ISCUSS I ON
Globally, feminised responses, including intersex, have been reported for freshwater fish living in rivers containing EDCs with reports of sex ratio skewed towards females for populations in some of the more polluted rivers. Extensive research conducted on roach in UK rivers has established this species as a sentinel for endocrine disruption in wild freshwater fish populations and it has since been adopted widely for this purpose across central and Northern Europe.
To date, however, the lack of a genetic sex marker has prevented answering the question on whether wild intersex roach arise as a consequence of the demasculisation (oestrogenisation) of males or the masculinisation (androgenisation) of females and whether female-biases observed in some wild roach populations are a consequence of sex-reversal in males. This lack of a genetic probe has thus been a limiting factor for studies into the population level consequences of sexual disruption in roach. Using RAD-seq, we isolated and subsequently validated a genetic sex-specific marker for roach, and applied it using PCR to samples derived from previous studies on chemical and effluent exposures and wild populations to confirm genetic sex in sexually disrupted fish and to sex fish prior to gonadal sex differentiation.
RAD-seq has been applied recently in the discovery of sex markers in various species of fish, reptiles and crustaceans (Brown et al., 2016;Carmichael et al., 2013;Fowler & Buonaccorsi, 2016;Gamble et al., 2015Gamble et al., , 2017Gamble & Zarkower, 2014;Mathers et al., 2015;Palaiokostas, Bekaert, Davie, et al., 2013;Palaiokostas, Bekaert, Khan, et al., 2013;Palaiokostas et al., 2015) and in some cases, these studies have been able to construct linkage maps based on single nucleotide polymorphisms (SNPs) and map a major sex determining locus to certain linkage groups. In our study, we have been able to identify presence-absence markers pertaining to sex only, and so more extensive sequencing of the roach genome would be required to construct linkage maps and further characterise this marker.
We performed two additional analyses in order to provide insight into the potential origin of the RAD-seq derived male marker, i.e. whether the presence-absence marker is the result of high sequence divergence between the sexes, or is missing in females due to it being a male-specific insertion-polymorphism. As the coverage analysis was performed on read 1 (where the PstI cut-site is), it is possible that a female-specific mutation in the cut-site, due to high sequence divergence between male and female haplotypes at this region, resulted in male-read recovery, and no female read mappings. As the PCR region sits in the paired-end contig, and the size-selection step targeted regions of 250-700 bp, the pairedend contig could reside some distance from the read 1. This means a divergent sex-specific haplotype of between 250-700 bp (or even larger) could indeed produce such a signal. An alternative hypothesis is the presence of an insertion-deletion polymorphism in which the female haplotype lacks both the restriction enzyme cut site and a proportion of the PCR region in the paired-end contig.
The assessment of allele frequency divergence did not identify any sex-linked divergence in other contigs, and the kmer analysis provided further support that the identified sex-linked marker is truly male-specific (and is missing in females). Due to the lack of a reference genome for roach, a more comprehensive analysis of these two hypotheses is not possible, yet this represents an interesting area of future research.
Sex markers alone, however, can be indicative of the underlying sex determining system with male sex markers indicative of a XY system, whereas female sex markers are indicative of a ZW system (Fowler & Buonaccorsi, 2016). It is not known which sex chromosome system roach have, if indeed they have one, but the isolation of a sex marker present in males and absent in females suggests male F I G U R E 5 Application of the genetic sex marker to wild roach populations with varying levels of intersex condition. The genetic sex of 267 fish from four independent roach populations (populations F-I; see Table 1) with varying proportions of intersex males was identified by PCR. For these fish, gonadal phenotypic sex had been determined previously via histopathology, or via release of their gametes in male fish (Defra, 2009;;Harris et al., 2011).
Each group of stacked bars shows phenotypic sex (left hand bars) and the identified genetic sex (right hand bars) for each population; the lower segment of each bar (in case of phenotypic sex also the middle segment) represents male fish and the upper segments female fish. Numbers indicate the number of fish represented in each segment heterogamety and as such an XX/XY as opposed to the ZW/ZZ sex determination system. To date, only a few sex-determining genes have been identified in fish but male heterogamety seems to predominate, including in medaka (Matsuda et al., 2002), three-spined stickleback (Peichel et al., 2004), the cichlids Pseudocrenilabrus philander and Astatotilapia burtoni (Böhne et al., 2019;Böhne, Wilson, Postlethwait, & Salzburger, 2016) and salmonids (Woram et al., 2003). There is little evidence of a dominant female-determining sex-determination gene, i.e., female heterogamety (Pan et al., 2016). The exception to this is an indication that in zebrafish (wild populations), the closest species studied evolutionary to the roach, where there is a female heterogamety sex determination system (Tong, Hsu, & Chung, 2010;Wilson et al., 2014).
Our approach to obtain a population-independent sex marker for the roach was to sequence four genomic DNA pools created from equal numbers of individuals from two independent roach populations. Adopting this approach, one of the three markers identified proved to be specific for sex. Previous approaches to obtain sex markers have sequenced individuals rather than pool individuals (e.g. Fowler & Buonaccorsi, 2016;Gamble et al., 2015;Gamble & Zarkower, 2014) which then requires extensive bioinformatics to confirm the identification of many putative of sex-loci/specific markers, most of which turn out to be false-positives in the subsequent sex validation (Gamble & Zarkower, 2014). We acknowledge that potential genotyping and sex-identification error can also be introduced by pooling males and females, and the biases in pooled approaches, especially in allele frequency determination, is well recognised (Gautier et al., 2013;Rellstab, Zoller, Tedder, Gugerli, & Fischer, 2013). However, pooled-based approaches have shown great promise for delineating diversity in species lacking reference genomes (Kurland et al., 2019;Neethiraj, Hornett, Hill, & Wheat, 2017), and, moreover, pool-RAD-seq approaches have been used to successfully identify species-specific SNPs across a range of species (Delord et al., 2018).
The PCR approach developed to determine genetic sex of individuals is rapid and easily applied, but the reliance on a presence/ absence marker means false assignment is possible due to a failure in the DNA extraction or PCR. In the roach this could lead to assigning genetic males as females. We circumvented this potential problem through the inclusion of a positive control (ITS1 nuclear ribosomal DNA region) to verify successful DNA extraction and through routinely running PCR assays with three different primer combinations for each sample. Yields on two PCRs in bands of the correct size were taken as a confident call on a genetic male.
Sex specificity of the RAD marker was validated based on the analysis of 80 individual roach (42 males and 38 females) from five different river sites for which there was 99% match between phenotypic (gonadal) and genetic sex. One phenotypic female was identified as being genetic male. As all of these roach were derived from the wild, sex-reversal could be a possible explanation for the observed mismatches between phenotypic and genetic sex, although this seems unlikely given that rivers sites from where they were derived are reported to have low levels of pollution. In the case for the phenotypic female, that was a genetic male, such a sex-reversal is possible as this has been shown to occur in fish chronically exposed to low oestrogen levels including during the period of sexual differentiation (Kidd et al., 2007;Lange et al., 2009). Female-to-male sex-reversal on the other hand would imply complete masculinisation. This could result from exposure to androgens, but unlike that for specific cases in the USA for exposure to cattle feedlots containing high concentrations of the androgen trenbolone (Ankley et al., 2003), androgens are not present at concentrations in English rivers to cause such a phenotypic change in roach.
The isolated sex marker shows high fidelity for identifying sex-reversed individuals also based on our analysis of roach derived from two controlled, life-long exposures to oestrogens (a full strength WwTW effluent and 4 ng EE2/L) both of which resulted in all-female populations as determined by gonad histopathology (Lange et al., 2009(Lange et al., , 2011. In both of these experiments, the genetic sex ratios of exposed fish were near 50:50 (52:48 and 53:47) male:female, respectively. The phenotypic sex ratios of the controls in these experiments were 69:31 (EE2) and 56:44 (effluent) (Lange et al., 2009(Lange et al., , 2011. Previous studies on roach maintained in the laboratory have reported sex ratios of 50:50, but also male and female biased populations (Lange et al., 2009(Lange et al., , 2011Paull, Filby, & Tyler, 2009;Rodgers-Gray et al., 2001) which may reflect the fact that growth rates and/or social factors potentially affect sex assignment in this species . In wild roach populations evenly balanced as well as both male-or female-biased populations have been reported (Geraudie, Gerbron, & Minier, 2017;Jobling, Beresford, et al., 2002) and outside the spawning season, even the existence of monosex populations has been suggested (Tyler, Lange, Paull, Katsu, & Iguchi, 2007).
The use of the sex probe in these instances has proven that full male-to-female sex reversal can occur for exposure to a full strength WwTw effluent and to EE2 (4 ng/L) when they are exposed during the periods of sex differentiation and gonadal development supporting the possibility for this to occur in wild populations (but see later in the discussion). Dilution of WwTW effluents in UK rivers, however, averages 10%-30% and in only a few exceptional circumstances and during periods of low flow do they exceed 50% , albeit in exceptional circumstances the full flow of the river can be made up of treated wastewater effluent (Jobling et al., 1998). In the case of EE2, concentrations in the river reaches where roach live are most likely to be in the range of 0.15-1.14 ng/L (Williams, Churchley, Kanda, & Johnson, 2012), although concentrations up to 3.4 ng/L have been reported previously (Williams, Johnson, Smith, & Kanda, 2003).
Application of the genetic sex marker to phenotypic female roach derived from the controlled WwTW effluent exposure and then subsequently subjected to a competitive breeding experiment furthermore identified that sex reversed genetic males spawned as females, producing viable offspring. These findings add a new and fascinating dimension for studies investigating potential impacts of exposure to WwTW effluents on roach (and potentially other fish) populations.
Based on modelling approaches applied to investigate for the effect of compromised reproductive fitness, as a result of endocrine disruption, on other fish populations (An, Hu, Giesy, & Yang, 2009;Cotton & Wedekind, 2009) the levels of effect seen in roach would indicate this is unlikely for wild populations living in UK rivers. This should, however, not be discounted, as predictive modelling for EDCs has shown that it may be many generations of chemical exposure before fish populations are reduced substantially and such changes are difficult to monitor in the wild, particularly for long-lived species such as the roach (Hamilton et al., 2016).
In the studies on wild roach populations exposed to WwTW effluents with varying proportions and degrees of intersex males, we confirmed that the intersex condition occurs due to feminisation of males.
For these study populations, the sex marker assigned ≥ 97% of phenotypic males (including intersex males) as genetic males and ≥ 94% of phenotypic females as genetic females. Of the intersex fish, 98% were assigned as genetic males. As mentioned above, some studies have reported female-biased populations (Geraudie et al., 2017;Jobling, Beresford, et al., 2002), leading to the question of possible complete male to female sex reversal. In our analyses of 128 phenotypic females, 90 of which were derived from river sites with oestrogenic contamination, only six registered as genetic males, strongly indicating that complete sex reversal in roach does not occur commonly, if at all, in roach in UK rivers as a consequence of pollution exposure.
By extending the initial sequence of the sex marker to around three times the original sequence length, the subsequent BLASTN search mapped the sequence to a 300 kb region of the Atlantic salmon (Salmo salar) genome containing a gene encoding growth hormone (GH) (von Schalburg et al., 2008). In fish, GH is involved in a variety of physiological processes, including the regulation of ionic and osmotic balance, lipid, protein and carbohydrate metabolism, skeletal and soft tissue growth, reproduction and immune function (Reinecke et al., 2005). Interestingly, salmonids are considered to have a male heterogamety sex determination system (Yano et al., 2013) and amongst several Y chromosome-specific gene markers that are located near the sex-determining gene, one is a GH pseudogene (GHp). This has been successfully applied to determine males of various Pacific salmon species including chum salmon (Oncorhynchus keta), pink salmon (Oncorhynchus gorbuscha), coho salmon (Oncorhynchus kisutch), Chinook salmon (Oncorhynchus tshawytscha) and masu salmon (Oncorhynchus masou) (Devlin, Biagi, & Smailus, 2001;Du, Devlin, & Hew, 1993;Micheletti & Narum, 2018;Zhang et al., 2001), whereas it did not produce sex-specific patterns in other Oncorhynchus salmonid species including sockeye salmon (Oncorhynchus nerka), rainbow trout (Oncorhynchus mykiss), or in Atlantic salmon . Given that the GHp appears sex-specific for some Pacific salmon species only, the BLAST hit does not align with the coding region for Atlantic salmon GH, and salmonids are distant in evolutionary terms to cyprinids it is unlikely that there is a relationship between our isolated genetic sex marker and GH.
The finding that the sex marker can also be applied to fish at a life stage prior to gonadal sex differentiation and the analysis can be undertaken nondestructively from fin clips and even skin swabs gives the marker further considerable utility. For example, where monosex populations are required for chemical exposures or for other laboratory biological investigations the sex probe can be applied to preselect sexed individuals, as immature or mature fish, prior to experimentation and thus halving the number of experimental animals used and thus addressing the principle of 3Rs. The same principle applies for surveys of wild roach populations (even as juveniles) where only one sex is required for subsequent termination and analysis. The ability to identify the genetic sex of roach non-destructively also has utility for studies on the wider ecology of the roach, including understanding of sex related behaviours that include fish movement and feeding patterns, and for appropriate sex balancing in the re-stocking of rivers and other water bodies for recreational fishing.