There exists extraordinary variation among species in the degree and nature of sex chromosome divergence. However, much of our knowledge about sex chromosomes is based on comparisons between deeply diverged species with different an- cestral sex chromosomes, making it difﬁcult to establish how fast and why sex chromosomes acquire variable levels of divergence. To address this problem, we studied sex chromosome evolution in two species of African clawed frog (Xenopus), both of whom acquired novel systems for sex determination from a recent common ancestor, and both of whom have female (ZW/ZZ) heterogamy. Derived sex chromosomes of one species, X. laevis, have a small region of suppressed recom- bination that surrounds the sex determining locus, and have remained this way for millions of years. In the other species, X. borealis, a younger sex chromosome system exists on a different pair of chromosomes, but the region of suppressed recombination surrounding an unidentiﬁed sex determining gene is vast, spanning almost half of the sex chromosomes. Differences between these sex chromosome systems are also apparent in the extent of nucleotide divergence between the sex chromosomes carried by females. Our analyses also indicate that in autosomes of both of these species, recombination during oogenesis occurs more frequently and in different genomic locations than during spermatogenesis. These results demonstrate that new sex chromosomes can assume radically different evolutionary trajectories, with far-reaching genomic consequences. They also suggest that in some instances the origin of new triggers for sex determination may be coupled with rapid evolution sex chromosomes, including recombination suppression of large genomic regions. Key words: sex chromosomes, linkage map, recombination suppression, differentiation, amphibian, Xenopus. Introduction chromosome that have a sex-biased mode of inheritance may Sex chromosomes originate when an autosome acquires a also have distinct mutation rates (Makova and Li 2002)and mutation that triggers development of one sex or the other. generation times (Amster and Sella 2016). Differences in the Recombination between sex chromosomes (the X and Y or Z variance of reproductive success between each sex can fur- and W) can be suppressed in regions that include and ﬂank ther contribute to the disparity in the extent of genetic drift the sex determining mutation, which causes sex-speciﬁc in- (the effective population size) of these regions (Charlesworth heritance of a sex determining trigger (Charlesworth 1991). 2009). Portions of sex chromosomes that lack recombination (e.g., A lack of recombination causes portions of the two sex the sex speciﬁc portions of the Y or W) and portions that have chromosomes to diverge from one another in nucleotide se- a reduced level of recombination compared with the auto- quence, gene content, and the abundance and distribution of somes (e.g., the nonpseudoautosomal regions of the X or Z) transposable and other repetitive elements (Charlesworth and are subject to distinct population genetic phenomena from Charlesworth 2000; Bachtrog 2013). Additionally, the non- autosomes. These genomic regions generally have a lower recombining region may expand due to accumulation of sex- effective population size than autosomes and thus experience ually antagonistic genes, because sex-biased inheritance can weaker purifying selection (Rice 1994). Portions of each sex mitigate sexual antagonism (Rice 1987; Wright et al. 2017). The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non- commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact firstname.lastname@example.org 742 Genome Biol. Evol. 10(3):742–755. doi:10.1093/gbe/evy045 Advance Access publication February 21, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/3/742/4895093 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Divergent Evolutionary Trajectories of Sex Chromosome Systems GBE Over time, these factors can lead to cytological distinctions refers to the number of chromosomes in an ancestral gamete between the sex chromosomes, a condition known as sex prior to genome duplication. Chromosomes in tetraploids in chromosome heteromorphy. In various taxa (e.g., some mam- subgenus Xenopus are numbered1–18followedby anL or mals, birds, and plants), divergence of sex chromosomes oc- an S, indicating from which of two diploid ancestors each curred incrementally along the length of the sex chromosome was derived (Matsuda et al. 2015). chromosomes due to sequential inversions or natural selection Species in genus Xenopus have homomorphic sex chromo- on recombination modiﬁers, expanding the nonrecombining somes (Tymowska and Fischberg 1973; Tymowska 1991), regions in a stepwise fashion (Coop and Przeworski 2007; and three nonhomologous sex determining systems have Bergero and Charlesworth 2009; Vicoso et al. 2013). been identiﬁed in this group. One is on chromosome 2 L of Interestingly and perhaps counterintuitively, the age of the the allotetraploid species X. laevis (Yoshimoto et al. 2008)and sex chromosomes does not seem to be tightly correlated with also several other allopolyploid Xenopus species (Bewick et al. whether or not sex chromosomes are cytologically distinct 2011). In these species, the W chromosome carries a gene (heteromorphic) or indistinct (homomorphic) (reviewed in called DM-W that triggers female sexual differentiation Wright et al. 2016). In some old sex chromosomes, for exam- (Yoshimoto et al. 2008). DM-W originated after the whole ple, those of neoaves (>100 Myr; Zhou et al. 2014)and the- genome duplication event ancestral to subgenus Xenopus rian mammals (150 Myr; Graves 2006), and also some species (Bewick et al. 2011). A second sex determination sys- young sex chromosomes, such as those of Drosphila miranda tem in Xenopus is located on chromosome 8 L in the allote- (1Myr; Bachtrog and Charlesworth 2002)and Silene latifo- traploid species X. borealis (Furman and Evans 2016). This sex lia (10–20 Myr; Bergero et al. 2007), divergence between the determination system evolved in X. borealis from an ancestor sex chromosomes is pronounced. In contrast, in the old sex that carried DM-W (Furman and Evans 2016). A third sex chromosomes of ratite birds (>100 Myr; Zhou et al. 2014), determination system in Xenopus is located on chromosome recombination is suppressed over large regions of the sex 7 in the diploid species Xenopus Silurana tropicalis (Olmstead chromosomes, but accompanied at the nucleotide level by et al. 2010; Evans et al. 2015). In X. tropicalis,Z,W,and Y relatively modest differentiation between the sex chromo- chromosomes segregate (Roco et al. 2015). Overall then, of somes and minimal cytological differentiation (Vicoso et al. the three sets of sex chromosomes in Xenopus,at least two— 2013; Yazdi and Ellegren 2014). An extreme case of homo- those of X. laevis and X. borealis – are newly evolved, and the morphy exists in the young sex chromosomes of tiger puffer- system of X. borealis is proposed to be derived with respect to ﬁsh, where a single mutation appears to control sexual (i.e., younger than) the system of X. laevis (ﬁg. 1; Furman and differentiation and there is no evidence of suppressed recom- Evans 2016). bination (Kamiya et al. 2012). In the young sex chromosomes This variation in sex chromosomes among Xenopus species of hylid tree frogs (5 Myr old) and Palearctic green toads presents an opportunity to compare the evolutionary trajec- (3.3 Myr old), recombination appears to be low or absent in tories of two newly established sex chromosome systems (i.e., heterogametic males, but there is not substantial nucleotide the sex chromosomes of X. borealis and X. laevis). Some dif- divergence (Sto ¨ ck et al. 2011, 2013). Why sex chromosomes ferences between the W and Z chromosomes of X. laevis have of some species are homomorphic whereas those of others been detected, including differences in gene content, inser- are heteromorphic, and why some heteromorphic sex chro- tion–deletion mutations, and nucleotide divergence, but this mosomes are more cytologically diverged than others remains limited to only a few hundred Kb (<1% of the chromosome enigmatic (Wright et al. 2016). length; Mawaribuchi et al. 2017). However, in general, in X. laevis and most other Xenopus species little is known about fundamental evolutionary genomic characteristics of sex and Sex Chromosomes Evolved Multiple Times in Xenopus recombination, such as sex chromosome-wide levels of diver- Insights into the origin of variation among species in sex chro- gence, the extent of sex-linkage of genes on sex chromo- mosome divergence may be gained by examining whether, to somes, genome-wide variation in rates of recombination, or what extent, why, and for how long recombination is sup- sex differences in rates of recombination. We therefore used pressed in genomic regions ﬂanking the sex determining locus reduced genome sequencing of parents and offspring of each in multiple species. For this reason, we quantiﬁed and com- species to assess sex-linkage of SNPs and to construct sex pared recombination on the sex chromosomes of the African speciﬁc linkage maps for both species. We found that these clawed frog, Xenopus Xenopus laevis, and the Marsabit two systems differ greatly in the extent of sex chromosome clawed frog, Xenopus Xenopus borealis. The most recent recombination suppression during oogenesis, with the youn- common ancestor of these two species experienced allotetra- ger system in X. borealis exhibiting a substantially larger region ploidization 18–34 Ma (Evans et al. 2015; Session et al. than the older system of X. laevis. Whole genome sequence 2016). These and other allotetraploid species in subgenus data indicate that the nonrecombining portions of the X. bo- Xenopus have 2n ¼ 4 s ¼ 36 chromosomes, where n refers realis sex chromosomes have a modest, but detectable, level to the number of chromosomes in a haploid gamete and s of nucleotide divergence. Finally, linkage mapping in both Genome Biol. Evol. 10(3):742–755 doi:10.1093/gbe/evy045 Advance Access publication February 21, 2018 743 Downloaded from https://academic.oup.com/gbe/article-abstract/10/3/742/4895093 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Furman and Evans GBE X. borealis Recombination University Institute of Biotechnology Genome Diversity Facility chromosome 8L Suppressed Free on an Illumina HiSeq 2500; other details about these data Daughters Sons ●●● ●●●● ●●●●●●●●●●●●●●●●●●●●●●●●●● Sex linked ● ● available in Furman and Evans (2016).For the X. laevis family, ●●●●●●● ● ● Non−linked we obtained female and male individuals from Boreal Science ●● ●● ● ● ● ●● ● ● ●● ● ●● ● ● ●● ● ●●●●●●●●● (St. Catharines, ON, Canada). We induced breeding with in- ●● ● ●●● ● ● chr8L X. borealis ● jection of human chorionic gonadotropin and determined the X. clivii DM-W ● ● ●● sex of tadpoles using primers for DM-W, which ampliﬁes only ● ● ● 0 ●● ●●● ●●●●●●●●●●●●●●●●●●●●●●●●● ● chr2L X. laevis 050 100 in females, and primers for DMRT1, which is present in both X. laevis sexes, as a positive control (Yoshimoto et al. 2008). The chromosome 2L DM-W RADSeq library was generated by Floragenex (Portland, OR) on both X. laevis parents, 17 daughters, and 20 sons and 150- 1 ● ●● ●●●●●●●●● bp single-end sequencing was performed at the University of ● ● ●●●●●● ● ●●●●● ●● ●● ●●●●●● ●●● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ●●●●●●●●●●● ● ● ● ●●●● ●●●●●●●●●●● ●●●● ●●●●● ● ●●●●● ● ● ● ●●●● ●●● ● ●●● ●●●●● ● ●●●●●●●●●●● ●● ●●●●●●●● ●● ● ● ● ● ●● ●● ● ● Oregon using an Illumina HiSeq 2500 machine. Though ● ● ● ● ● ●●●●● ●●●●● ●●● ● ●●●●●●●●●●●●● ●●●●●●●●●●●●●●●●●● ● ●●●●●●●●●●●●●● ●●●●●●●● ● ● ● ● ●●●●●●● ● ●● ●●●●●●●●●● ● ●●●●●●●●●●●● ● ● ●● ● ●●●●●●●●● ●●●●●● ●● ● ● ● ● ●●●● ●● ●●●● ●●● ●● ●●●● ●●●●●●●● ●●● ● ●●●● ● ●●● ● ● ● ●● slightly different procedures were used to generate reduced ●●●● ● ●●●●● ● ●●●●● ● ● ● ● ● ● representation genome sequences from each species, the na- ●● ●●● ●●●●●●● 0 ● ●●●●● ture of the data is essentially the same—both methods pro- 0 50 100 150 Chromosome Position (Mb) duced sequence data from many homologous regions in most or all individuals from each family. FIG.1.—Sex-linkage of SNPs on sex chromosomes of X. borealis and GBSorRADSeq datafrom each X. borealis or X. laevis X. laevis. In each graph, the x-axis is the position on the sex chromosome individual were demultiplexed, trimmed, and aligned to the using the coordinates of the X. laevis reference genome and the y-axis is X. laevis genome version 9.1 (www.xenbase.org) followed by the major daughter genotype frequency in sons and daughters (see Materials and Methods for details) with colors as deﬁned in the key indi- genotyping and ﬁltering steps that are described in the sup- cating whether or not a SNP is signiﬁcantly associated with sex (FDR plementary S1.1, Supplementary Material online. This yielded corrected P< 0.05). For each species, a diagram of a chromosome is a panel of SNPs for each family that were used to study re- shaded darker in the region of suppressed recombination. The inset phy- combination as described next. We discuss the potential logeny is from Furman and Evans (2016); DM-W is carried by female impacts that the differences in the data sets of X. borealis X. clivii, but its presence on chr2L has not been conﬁrmed. and X. laevis may have on our study in supplementary S1.1, ﬁgure S4, Supplementary Material online. species demonstrates that females have higher rates of re- combination than males of both species, and that the location of crossovers is distinctive between females and males in both Sex-Linked Genomic Regions species, but similar in same sex comparisons across species. In X. laevis and X. borealis, females are the heterogametic sex These ﬁndings demonstrate that newly evolved sex chromo- (Yoshimoto et al. 2008; Furman and Evans 2016). Using the somes in different species may rapidly assume radically differ- ﬁltered data for both families, we thus calculated maternal ent evolutionary trajectories. genotype association with the phenotypic sex (male or fe- male) of each individual SNP following Goudet et al. (1996). Signiﬁcance was assessed using a false discovery rate correc- Materials and Methods tion on the P value of association with sex (a ¼ 0.05, using R; Reduced Representation Genome Sequences from R Core Team 2016) and we discarded from this analysis ma- X. laevis and X. borealis Families ternal SNPs that were also heterozygous in the father. In order To assess genome wide sex-linkage, we used reduced repre- to make inferences discussed below about the region of sup- sentation genome sequencing (genotype by sequencing pressed recombination that ﬂanks the trigger for sex determi- [GBS], Elshire et al. 2011; and restriction site associated nation, for each maternal SNP, we also determined the DNA sequencing [RADSeq], Baird et al. 2008)on parents frequency of the most common genotype in daughters and and offspring of an X. borealis family and an X. laevis family, then the frequency of this same genotype in sons. We refer to respectively. For the X. borealis family, we used GBS data that this frequency as the “major daughter genotype frequency.” we previously reported (Furman and Evans 2016), with a fe- At a completely sex-linked site that was heterozygous in the male and male obtained from XenopusExpress (Brooksville, mother and homozygous in the father, we expected offspring FL). These GBS data included mother, father, 24 daughters, genotypes to be homozygous in one sex and heterozygous in and 23 sons (22 and 17 individuals, respectively, after ﬁltering, the other (which sex is heterozygous depends on whether the see supplementary S1.1, Supplementary Material online), SNP was on the maternal Z or W). Thus, the major daughter with offspring sex determined by dissection after euthanasia. genotype frequency at a completely sex-linked site would be The GBS data were 100 base pairs (bp) single-end sequences; 1.0 for daughters, and 0.0 for sons. Conversely, at an auto- library preparation and sequencing was performed at Cornell somal site the major daughter genotype frequency in 744 Genome Biol. Evol. 10(3):742–755 doi:10.1093/gbe/evy045 Advance Access publication February 21, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/3/742/4895093 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Major Daughter Genotype Frequency Divergent Evolutionary Trajectories of Sex Chromosome Systems GBE daughters should be 50% (but always 50% because we between male and female markers were unable to be esti- excluded from this analysis positions with more than two mated for some chromosomes, and thus the ﬁrst step of cre- variants). In sons, the major daughter genotype frequency ating a joint linkage group could not be performed. For these should also be 50% at autosomal sites, but could be lower chromosomes, we instead selected the largest female-speciﬁc or higher than this value. and largest male-speciﬁc linkage group for each chromosome independently to estimate sex-speciﬁc linkage maps. Thus for these chromosomes, the male and female linkage groups do Linkage Maps not span identical genomic regions. We set out to evaluate rates and locations of recombination events in the mother and the father of our laboratory crosses. Error Correction and Haplotype Estimation To accomplish this, we used the R package OneMap (Margarido et al. 2007) to construct linkage groups based Genotyping errors create genotypes resembling recombined on variable sites from the X. borealis and X. laevis families haplotypes that distort linkage maps and lead to inﬂated map that mapped to each of the 18 X. laevis chromosomes in lengths (Hackett and Broadfoot 2003). Although we ﬁltered the reference genome. For each X. laevis chromosome and incompatible parent–offspring genotypes (supplementary S1. separately for each species, linkage groups were constructed 1, Supplementary Material online), undercalling of heterozy- with a maximum recombination fraction of 0.4 and a LOD gous sites can also produce incorrect homozygous genotypes threshold of ﬁve. With perfect synteny between the X. laevis in offspring that are nonetheless compatible with parental and X. borealis and an even genomic distribution of geno- genotypes. To deal with this problem, we identiﬁed putative typed SNPs, there should be one linkage group per X. laevis genotype errors based on phased offspring haplotypes. Each chromosome. However, we frequently identiﬁed several link- parent has two haplotypes per chromosome, and sites inher- age groups per X. laevis chromosome in each species and we ited by offspring can be assigned to one or the other haplo- suspect that this was a consequence of genotyping and map- type for each parent. Recombination during gametogenesis ping errors (see below) and regions with sparse SNPs due to creates new combinations of the two parental haplotypes poor mapping of X. borealis reads to the X. laevis reference within an offspring, with the “phase” referring to which pa- genome. For the X. borealis family, rearrangements between rental haplotype an offspring site comes from (see supple- X. borealis and X. laevis could also break up a chromosome- mentary ﬁg. S1, Supplementary Material online, for a visual speciﬁc linkage group. For either species, genome assembly explanation). Genotyping errors appear as a change in phase errors could also prevent assembly of one linkage group for a for a single SNPs (or a few SNPs in a row) when compared chromosome. We note that our linkage maps did not include with surrounding SNPs. This pattern at one or few sites can a particularly large number of offspring (39 in X. borealis and also arise biologically from a double recombination (a cross- 37 in X. laevis), and this contributed to a lack of statistical over on either side of a variable position). However, double power to form whole-chromosome linkage groups. recombination events in small genomic windows are consid- However, this was not a concern for (or an objective of) our ered to be rare because of recombination interference analyses, which focus on genomic regions for which assembly (reviewed in Zickler and Kleckner 2016). of linkage groups was possible. To identify putative genotype errors, we used the parental In order to evaluate rates of recombination in the mother phase estimated during linkage map construction (using and father of each species, we selected the largest linkage OneMap; see Wu et al. 2002 for details on phase estimation group from each chromosome and divided the markers in of outcross maps) to estimate the parental haplotypes inher- each linkage group into those that were heterozygous in ited by each offspring individual, for each chromosome- the mother, in the father, or in both parents. Then, using speciﬁc linkage map (supplementary ﬁg. S1a and b, each of the maternal and paternal sets of markers from Supplementary Material online). Under the assumption that each of the largest linkage groups per chromosome, we double recombination events are rare in small genomic win- recomputed recombination fractions between the sets of dows, we set to missing data any single genotype supporting sex-speciﬁc markers and constrained marker order to match a phase change in an individual at just that site (i.e., sites the mapping position in the v.9.1 X. laevis genome. For the whose ﬂanking genotypes were consistent double recombi- X. borealis family, some chromosomes had very few or no nation event around a single genotyped site). As well, any double heterozygous sites (sites that were heterozygous in genotypes in an individual that indicated a double recombi- both parents), which is a consequence of the lower overall nation event that only encompassed a small genomic window amount of data for this cross compared with the X. laevis of <5 Mb were set to missing data (i.e., a series of sites within cross (due to mapping of X. borealis but not X. laevis data 5 Mb who were in an alternate phase compared with adja- to a diverged reference genome, and the lower overall cov- cent sites). For the X. laevis cross, which involved substantially erage we obtained from the GBS data compared with the more markers than the X. borealis cross, thereweremoreof RADSeq data). This meant that the recombination fractions these potential genotyping errors (4% of all genotyped sites in Genome Biol. Evol. 10(3):742–755 doi:10.1093/gbe/evy045 Advance Access publication February 21, 2018 745 Downloaded from https://academic.oup.com/gbe/article-abstract/10/3/742/4895093 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Furman and Evans GBE individuals indicated a double recombination at either a single that have evolved quickly. With sex chromosomes, faster-Z site or phase changes encompassing <5MBinthe X. laevis (i.e., rapid evolution of Z-linked genes) or degeneration of the maps, compared with <0.5% for either in the X. borealis W sequences could lead to an underrepresentation of rapidly map; supplementary table S1, Supplementary Material on- evolved sequences, leading to an underestimation of diver- line). Over 90% of the putative genotyping errors that were gence. Contrary to this expectation, however, the number of identiﬁed based on double recombination like phase changes reads mapped to the sex linked region of chromosome 8 L in in the X. laevis maps were homozygous, which is consistent the female (10.2 million) was similar to other identically sized with the bulk of these putative errors having been generated regions of other chromosomes (range 7.8–11 million). by undercalled heterozygous positions (supplementary table One concern in the quantiﬁcation of divergence in the S1, Supplementary Material online). After setting these gen- nonrecombining portion of the sex chromosomes is that inter- otypes to missing data in the affected individuals, we reesti- genic regions may have many mapping errors due to repeti- mated linkage maps for each chromosome, for each parent tive sequences. For this reason, we focused our calculation of for each species. Map distances were then calculated using nucleotide diversity on genomic regions that are within and the Kosambi function (Kosambi 1943). ﬂank genes, because these areas contain less repetitive DNA To quantify recombination events across all maps, we (at least in X. tropicalis; Shen et al. 2013). We used the counted all phase changes in each linkage map for each in- X. laevis genome annotation (version 9.1 primary gene mod- dividual based on haplotypes that were constructed from els gff ﬁle; www.xenbase.org) to separately calculate nucleo- phased SNPs in each offspring. The location of recombination tide diversity (p) in each parent for coding sequence of genes 0 0 events was approximated as half the distance between the (hereafter CDS), introns, 5 and 3 untranslated regions (here- two markers bordering a recombination event in the X. laevis after UTR), 5,000-bp upstream of the 5 -UTR, and 5,000- bp reference genome. We assessed the relationship between downstream of the 3 -UTR for genes on all chromosomes. We linkage map length and the amount of bp covered (on the considered only estimates that were generated from at least X. laevis genome) by each map using a linear model, ﬁtting an 200 bp of contiguous data from both X. borealis individuals. interaction between sex and species, along with a three-way Overall, we measured p in 30,876 CDS regions, 3,092 0 0 interaction between sex, species, and the Mb covered by a 5 -UTRs, 14,954 3 -UTRs, 119,420 introns, 30,326 upstream linkage map (after scaling and centering Mb) using R. This regions, and 30,270 downstream regions (for a total of strategy allowed us to assess for each sex and species slopes 230,016 genomic regions) in the female and the male for the relationship between cM and Mb. We then used the X. borealis individuals. conﬁnt function to compute conﬁdence intervals on the To test whether the W and Z chromosomes were more estimates. diverged in the mother than the homologous Z region in the father, we used a linear mixed model implemented by the lme4 package in R (Bates et al. 2015). We set as ﬁxed effects Divergence between the W and Z Chromosomes of sex (female or male) and sex-linkage (deﬁned as sex-linked if X. borealis between bp 4,605,306 and 51,708,524 [corresponding to As discussed below, our analysis identiﬁed a large region of 100% sex linked tags; ﬁg. 1] of chromosome 8 L as deﬁned the X. borealis sex chromosomes that had sex-linked inheri- by the analysis of sex-linked GBS tags discussed below). The 0 0 tance. If recombination has been suppressed in this region for six categories of gene regions (CDS, 5 -and 3 -UTRs, introns, a protracted period of evolutionary time, we expected molec- up/down-stream) were set as a random effects. The model ular polymorphism in the mother to be higher than the ho- also included an interaction between the two ﬁxed effects mologous region of the father due to the accumulation of (sex and sex-linkage). We then used likelihood proﬁles (using diverged sites between the W and Z. For this reason, we also the proﬁle command in lme4) to calculate conﬁdence inter- predicted that polymorphism in this region of the maternal vals on the estimated coefﬁcients. sex chromosomes would be higher than other recombining To visualize and test for differences in divergence within portions of the maternal genome. the sex-linked region, we calculated median p for the mother To explore the effects of this lack of recombination at the and father in 1-Mb windows of chromosome 8 L, using the p nucleotide level, we performed whole genome sequencing on estimates from each of the genomic regions (intragenic, 5 - the parents of our X. borealis family using the Illumina HiSeqX UTR, 3 -UTR, introns, 5,000 bp upstream of genes, and platform at The Center for Applied Genomics (Toronto, 5,000 bp downstream of genes). Because the mother and Canada), with both individuals multiplexed across two lanes. father had different levels of polymorphism, we needed to We trimmed the data, mapped it to the X. laevis reference control for this difference in our comparisons between geno- genome, and genotyped and ﬁltered the data as described in mic regions of each individual. We therefore ﬁrst calculated the supplementary S1.4, Supplementary Material online. the median p value of all 1-Mb windows across chromosome Mapping to a diverged reference genome could lead to a 8 L for each individual. We then standardized the maternal bias of more conserved sequences mapping, than sequences and paternal estimates of p by dividing by their corresponding 746 Genome Biol. Evol. 10(3):742–755 doi:10.1093/gbe/evy045 Advance Access publication February 21, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/3/742/4895093 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Divergent Evolutionary Trajectories of Sex Chromosome Systems GBE chromosome-wide median. In order to compare these stan- inherited molecular variation in this genomic region was dardized values of diversity, we then divided the standardized also almost entirely sex-linked, with exceptions discussed estimates of p measured in each 1-Mb window of the mother below. Across the entire genome after ﬁltering, the SNP by the standardized estimates of p measured in the homolo- data set consisted of 1,813 variable positions and there gous window of the father. With no difference in level of were more heterozygous SNPs in the mother than the fa- divergence between alleles, we expected this ratio to equal ther (1,103 and 644 SNPs in the mother and father, re- one; if the W and Z chromosome were more diverged from spectively, and 66 positions were heterozygous in both each other in the mother than the two Z chromosomes were parents, with 15–133 SNPs per chromosome, and a from each other in the father, this ratio should be greater than mean of 61.8 maternal SNPs per chromosome). For ma- one. We tested for a difference between the sex-linked and ternal heterozygous positions used for assessing sex link- nonsex-linked portions using a Wilcoxon rank sum test and age in X. borealis, daughters had a median depth of 68 the measured disparity between parents of each 1 Mb esti- and genotype quality of 99 (maximum possible value), mates of standardized p. We also explored whether there was sons had a depth of 31 and a genotype quality of 99 (sup- a higher rate of synonymouns and nonsynonymous substitu- plementary ﬁg. S4, Supplementary Material online). tions in genes on the nonrecombining portion of the sex Aligning to the diverged X. laevis genome substantially chromosomes to the rest of the genome using the WGS se- reduced the number of SNPs recovered to 10% of the quence data as described in detail in the supplementary S1.5, de novo SNP discovery method that did not involve map- Supplementary Material online. Finally, we explored the pos- ping to the X. laevis genome (Furman and Evans 2016). sibility of an accumulation of deletions and/or insertions on In sharp contrast, on the X. laevis sex chromosomes (2 L) the sex chromosomes. Further details of these analyses are signiﬁcant sex-linkage was only detected at only six maternal presentedinthe supplementary S1.6, Supplementary Material SNPs spanning 2 Mb (1%; positions 178,144,865 to online. 180,779,644, and possibly to the end of the chromosome at 181,296,000; P< 0.05 after FDR correction; ﬁg. 1 and supplementary ﬁg. S5, Supplementary Material online). In Validation of X. borealis Sex Chromosomes and X. laevis, SNPs immediately adjacent to the statically associ- Recombination Suppression ated SNPs also had a strongly sex-biased pattern of inheri- To explore whether the expansive region of suppressed re- tance, which is consistent with recombination suppression combination in X. borealis was limited to our lab raised family, of this region (ﬁg. 1). A lack of a statistically signiﬁcant sex- we raised a second family of X. borealis using different linkage of some SNPS in this small genomic region may be a parents. We then sequenced two genes (SOX3 and NR5A-1 consequence of undercalled heterozygous positions (supple- [alternatively, SF-1]) located 25 Mb apart within the sex linked mentary table S1, Supplementary Material online and see region (according to placement in the X. laevis genome v9.1) Materials and Methods). Across the entire genome, there to look at coinheritance of alleles from parents to offspring. were 7,779 SNPs, and in this family. The father was more We also surveyed a panel of adults that were not used in polymorphic (1,618 and 4,547 in mother and father, respec- either cross from both sexes to assess linkage of alleles at tively, and 1,614 positions were heterozygous in both these two genes. Further details of these assessments are in parents). For maternal heterozygous positions used in the the supplementary S1.3, Supplementary Material online. sex linkage analysis of X. laevis, daughters had a median depth of 67, and a genotype quality of 99, sons had a depth of 61 and a genotype quality of 99 (supplementary ﬁg. S4, Results Supplementary Material online). Diverse Evolutionary Fates of Newly Evolved Sex Within the sex-linked region of X. borealis,there was a Chromosomes section with no recombination, and an adjacent section Our analysis of the sex chromosomes of X. borealis and with reduced recombination between positions 51,708, X. laevis identiﬁed a far larger region of sex-linked SNPs 524–56,690,925 of chromosome 8 L (ﬁg. 1). Seven con- in X. borealis (ﬁg. 1). In X. borealis, 40 maternal SNPs span- secutive SNPs on the end of this region indicated recom- ning 52 Mb (43%) of the sex chromosome (8 L) had a bination between the W and Z in one daughter, who had signiﬁcant association with the phenotypic sex of offspring the same genotype as the sons at these positions (ﬁg. 1). (positions 4,605,306–56,690,925 of a total chromosome Additionally, by inspecting changes in parental phase in length of 120 Mb in the X. laevis genome assembly; the offspring (see below), another maternal recombination P < 0.05 after FDR correction; ﬁg. 1 and supplementary event was observed immediately adjacent to the region of ﬁg. S2, Supplementary Material online). Within this region, completely suppressed recombination in one of the sons daughters had identical genotypes at 34 of the 40 SNPs, (supplementary ﬁg. S3, Supplementary Material online). with only one daughter differing for the last seven in the We note that additional information from more offspring region (see below). Similarly in most sons, maternally or other families could potentially identify more Genome Biol. Evol. 10(3):742–755 doi:10.1093/gbe/evy045 Advance Access publication February 21, 2018 747 Downloaded from https://academic.oup.com/gbe/article-abstract/10/3/742/4895093 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Furman and Evans GBE recombination events within the genomic region where we did not observe recombination. The genomic locations of several SNPs in the X. borealis family suggested genotyping or mapping error (supplemen- tary ﬁg. S2, Supplementary Material online). For a few sites within the otherwise completely sex-linked region of chromo- some 8 L, different individual sons had the same genotype as their sisters (ﬁg. 1). If this were due to a real recombination event, we would expect these sons to have the same geno- type as their sisters at adjacent SNPs as well. Although this pattern could arise from independent double recombination events around these single sites in different sons, a more plausible explanation is that these are genotyping errors. We observed three SNPs that mapped to the middle of the sex-linked region of chromosome 8 L that were not associated with sex (P> 0.05, following FDR, two sites are overlapping on the plot; ﬁg. 1), and we also found ﬁve SNPs that were FIG.2.—Linkage map length (in cM) is positively correlated with the number of bp spanned by the map (based on the X. laevis genome) for completely sex-linked that mapped chromosome 8 S. These maternal but not paternal linkage maps. Black “sex chr” dots indicate the genotypes are best explained by mapping error between linkage map of the sex chromosome of each species (chromosome 8 L in X. borealis sequence reads and the X. laevis genome, or per- X. borealis, chromosome 2 L in X. laevis). Lines reﬂect linear model relation- haps assembly error in the X. laevis genome wherein homeol- ships; gray shading indicates the 95% conﬁdence interval of this relation- ogous portions of the 8 L and 8 S chromosomes are ship. Additionally, chromosome 8S is highlighted for X. borealis, because it is intermingled in the assembly. It is also possible that sections the homeolog of the sex chromosome 8 L (see Results for details). of homeologous sequences of X. laevis and X. borealis were lost in an asymmetric fashion after whole genome duplication, such that chromosome 8 L in X. laevis is missing portions that female ¼ 0.96 Gb, male ¼ 1.72 Gb; ﬁg. 2). Consistent with were not lost in X. borealis. This could cause reads from this, the number of crossovers is higher in oogenesis than X. borealis to map to homeologous sequence in the X. laevis spermatogenesis in both species (X. laevis: oogenesis ¼ 558 genome, instead of to the missing orthologous sequence in X. total; 15.1/offspring, spermatogenesis ¼ 467 total; 12.6/off- laevis. spring; X. borealis: oogenesis ¼ 270 total, 7.3/offspring; We also identiﬁed a sex-linked site in X. borealis that spermatogenesis ¼ 62 total; 1.6/offspring). mapped to X. laevis chromosome 5S (supplementary ﬁg. S2, Also of note is that the locations of crossovers were dis- Supplementary Material online). We blasted sequence from tinctive in females and males of both species. Female cross- the GBS tag that contained this SNP to a de novo assembly of overs we more concentrated in the middle of the the maternal X. borealis HiSeqX data that were assembled chromosomes, whereas male crossovers occurred more of- using SOAPdenovo v.2.04, with a kmer ¼ 23, and default ten at the ends of chromosomes (ﬁg. 3). Possibly related to parameters. We then blasted the top hit scaffold back to this (see Discussion), the length in cM of female linkage the X. laevis genome and found that its best matches were maps of both species was positively correlated with the chromosomes 8S and 8 L with similar afﬁnities. This suggests number of bp covered by a map, but this relationship was that that this site could be a translocation between X. borealis not found in the male linkage maps from either species and X. laevis, anassembly error inthe X. laevis genome, or a (linear model slope estimates, 95% conﬁdence intervals: mapping error due to the short sequence length (<100 bp) of X. borealis female ¼ 36.96, 24.78–49.13, male¼0.50 each GBS tag. 14.96–13.95, X. laevis female ¼ 40.80, 30.66–50.94, male ¼ 5.40, 8.04–18.83; ﬁg. 2). Similar results were recovered when total length of chromosome was used in- Recombination Is Higher in Females of Both Species stead of the number of bp covered by the linkage map, or Sex differences in the linkage maps revealed higher recombi- when the number of crossover events was used instead of nation rates in females of both species. The female linkage total cM (results not shown). maps of both species were longer (X. laevis ¼ 1,572 cM; For the X. borealis family, the largest female linkage group X. borealis ¼ 719 cM) than the same-species male linkage on chromosome 8 L (the sex chromosome, which includes maps (X. laevis ¼ 1,275 cM; X. borealis ¼ 165 cM; ﬁg. 2). both the Z and the W chromosomes) was formed from Longer female maps were recovered despite female markers markers that mapped to the sex-linked portion (ﬁg. 1), and spanning fewer base pairs of the X. laevis genome in both spe- did not include markers from the nonsex-linked portion (see cies (X. laevis female ¼ 1.76 Gb, male ¼ 2.28 Gb; X. borealis Materials and Methods for possible explanations). This region 748 Genome Biol. Evol. 10(3):742–755 doi:10.1093/gbe/evy045 Advance Access publication February 21, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/3/742/4895093 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Divergent Evolutionary Trajectories of Sex Chromosome Systems GBE 2.0 SNPs (ﬁg. 1). Therefore, we did not detect any restricted re- combination in this map, and the size (in cM) of the linkage 1.5 map of this chromosome was similar to the size of the linkage maps for other chromosomes spanning similar amounts of 1.0 X. borealis Mbp (ﬁg. 2). 0.5 Divergence between the Sex-Linked Portions of the W and 0.0 maternal Z Chromosomes of X. borealis paternal We analyzed genotypes inferred from whole genome se- 1.5 quencing data from the mother and the father to test whether we could detect evidence of sex chromosome diver- 1.0 X. laevis gence between sex-linked portions of the W and Z sex chro- mosomes. Compared with the pseudoautosomal portion of 0.5 chromosome 8 L and also to the autosomes, the sex-linked portion of chromosome 8 L had the highest median nucleo- tide diversity in the female (pairwise nucleotide diversity 0.00 0.25 0.50 0.75 1.00 ½p¼ 0:012; ﬁg. 4a). In this female genome, diversity within Relative Chromosome Position the nonsex-linked (pseudoautosomal) portion of chromosome FIG.3.—Density plots of recombination events with respect to the 8 L was similar to that of other chromosomes (p ¼ 0:009; relative position along chromosomes (chromosome length scaled to be ﬁg. 4a). In the male genome, diversity of each portion of between 0 and 1) in the maternal and paternal linkage maps of X. borealis chromosome 8 L fell within the range of estimates from other and X. laevis. chromosomes from this genome (sex linked: p ¼ 0:0072; nonsex linked: p ¼ 0:009; ﬁg. 4a). The nucleotide diversity spanned52 Mb(43% of the total X. laevis chromosome 8 L) measured for these chromosomes is far less than the 7% and was only 5 cM in length. That this recombination prob- divergence of homeologous sequences (Evans and Kwon ability is not 0 cM is attributable to two recombination events 2015); the considerably lower p estimates reported here sug- at the end of the region, each of which is illustrated in plots of gest that cross mapping of reads across subgenomes was offspring haplotype assignment (supplementary ﬁg. S3, relatively rare. Supplementary Material online). The female linkage map of Analyses of nucleotide diversity in and around genes (di- chromosome 8 L was much shorter in recombination proba- vided into six categories; see Materials and Methods), which bility (cM) than other female and male linkage maps that used a linear mixed model, recovered a signiﬁcant interaction spanned similar numbers of bp on other chromosomes between sex and sex-linkage, indicating that the mother had (ﬁg. 2). Themalemap of chromosome 8L in the X. borealis a higher p than the father in the sex-linked portion of chro- family, which corresponds to a pair of Z chromosomes, mosome 8 L compared with the rest of the genome, and after spanned almost the entire chromosome, and had a length controlling for differences in polymorphism between these of 13 cM, which is similar to other chromosomes (ﬁg. 2). In individuals (estimate of the increase in female diversity in the father, we detected ﬁve recombination events within the the sex linked region ¼ 0.0018, 0.0009–0.0027 95% CI, t- portion of chromosome 8 L (i.e., between two Z chromo- stat ¼ 4.09; ﬁg. 4b). For this analysis, we discarded the ﬁrst somes) that had suppressed recombination in the mother four million base pairs of chromosome 8 L because we lacked (i.e., the region where there was almost no recombination information on whether this region is also sex-linked (ﬁg. 1). between the W and Z chromosomes; supplementary ﬁg. We note that nucleotide diversity in the sex-linked portion S3, Supplementary Material online). of the female sex chromosomes includes ﬁxed differences Interestingly, even though it is not a sex chromosome, the between the W and Z chromosomes and also positions that maternal linkage map of the X. borealis chromosome that is are segregating on the Z chromosome. Thus, this measure- homeologous to the sex chromosome—chromosome 8 S— ment is inﬂuenced by demographic differences between the was also substantially shorter in cM than other linkage maps female and male (the female genome is more polymorphic; spanning a similar amount of megabases (it was below the ﬁg. 4). However, we found that standardizing the estimates best ﬁt line; ﬁg. 2). This suggests that recombination is less of nucleotide diversity by the genome-wide average for each frequent on this homeologous chromosome than other auto- individual (by dividing diversity estimates from the male or somes, even though it is not sex-linked. female genome by the corresponding genome-wide mean The X. laevis female linkage map of chromosome 2 L did for each genome) did not affect the results of the linear mixed not include the last 20 Mb, which is where DM-W resides model (see Results and supplementary S1.4, Supplementary (Session et al. 2016), and where we detected sex-linked Material online). In the analysis of nucleotide diversity, the sex Genome Biol. Evol. 10(3):742–755 doi:10.1093/gbe/evy045 Advance Access publication February 21, 2018 749 Downloaded from https://academic.oup.com/gbe/article-abstract/10/3/742/4895093 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Crossover Density Furman and Evans GBE FIG.4.—Nucleotide diversity (p)in X. borealis based on WGS data mapped to the X. laevis reference genome. (a)Median p by chromosome as measured in the six genomic categories; error bars indicate 95% CI bootstrap estimates (for further information on differences see supplementary S1.4, Supplementary Material online). The 8 L_NL category refers to the diversity measured on chromosome 8 L in the nonsex-linked region (57–120 Mb). (b) Box and whisker plot of p across six genomic categories (described in Materials and Methods); the y-axis is truncated at 0.05 for clarity. (c) Standardized nucleotide diversity of the female divided by the standardized nucleotide diversity of male in 1-Mb windows across chr8L; the completely sex-linked region is highlighted in dark purple, and the signiﬁcantly sex linked region with suppressed recombination in light purple (see ﬁg. 1). linked portion of chromosome 8 L stood out as the most poly- There was also a peak of divergence near end of the chromo- morphic region in the female genome, supporting the exis- some in the nonsex-linked region (ﬁg. 4c), that overlapped tence of ﬁxed divergent sites between the W and Z with a region where X. borealis daughters were mostly inher- chromosomes. iting the same allele, suggesting partial sex-linkage (ﬁg. 1). The disparity between the female and male in nucleotide This could be due to an inversion, although we did not explore diversity along chromosome 8 L was greater in the sex-linked this possibility in our data. portion than the pseudoautosomal portion of chromosome Within coding regions, dN and dS were very slightly, but 8 L (Wilcoxon rank sum test: P< 0.001; ﬁg. 4c). This result is signiﬁcantly (statistically) elevated in the sex-linked region of consistent with the results of the linear mixed model (above). X. borealis compared with the rest of the genome for both the 750 Genome Biol. Evol. 10(3):742–755 doi:10.1093/gbe/evy045 Advance Access publication February 21, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/3/742/4895093 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Divergent Evolutionary Trajectories of Sex Chromosome Systems GBE female and male, but dN/dS was not (based on a permutation reads and relatively low coverage). However, there were two test; see Supplement 1.5). But, unlike the analysis of all SNPs crossover events detected in the sex linked region (ﬁg. 1 and (above), which included more data, the sex linked region was supplementary ﬁg. S3, Supplementary Material online). As not the highest for any value (dN,dS,or dN/dS)compared well, the level of divergence between the W and Z was lower individually to the other chromosomes. This emphasizes the in the last 1/3 of the sex linked region, consistent with a more subtlety of the divergence in the sex linked region and indi- recent cessation of recombination (and possibly indicating the cates that the time since recombination suppression is recent. presence of genomic regions—strata—with different levels of We did not recover evidence of substantial differences in cov- divergence). These results suggest that a single large scale erage between the female and male on the sex chromosomes inversion encompassing the entire sex-linked region is not a (see supplementary S1.6, Supplementary Material online). likely reason for suppressed recombination. We cannot rule out the possibility that there are smaller inversions within the sex linked region that causes recombination suppression in ﬂanking regions. In some sex chromosome systems, inversions Discussion are not thought to be the driver of recombination suppres- More Expansive Recombination Suppression on Younger sion. For example, in the plant S. latifolia, inversions in the Sex Chromosomes nonrecombining portion of the sex chromosomes may have The homomorphic sex chromosomes of X. borealis and occurred after recombination suppression evolved (Bergero X. laevis experienced distinctive evolutionary histories since et al. 2008). We did not recover any evidence of major cov- they originated. In X. laevis, the sex-linked region is restricted erage differences between the sequenced female and male to a small portion on the end of a chromosome (2 L). In X. borealis (supplementary S1.4, Supplementary Material on- X. borealis, however, the sex-linked region encompasses al- line), suggesting a lack of deletions or insertion differences most half of a chromosome (8 L; ﬁg. 1), even though this sex between the Z and W. However, our inference is limited by a chromosome system is thought to be derived with respect to lack of a con-speciﬁc reference genome, because unique the sex determination system of X. laevis (Furman and Evans or rapidly evolving sequences on the sex chromosomes of 2016). Within the region of suppressed recombination of both X. borealis may not map to the homologous portion of or of these species, there is evidence of sex chromosome diver- be present in the X. laevis reference genome. gence at the molecular level (X. borealis: ﬁg. 4a–c and supple- Alternatively, modiﬁers of recombination can be favored mentary S1.5, Supplementary Material online; X. laevis: by natural selection to suppress recombination (Charlesworth Mawaribuchi et al. 2017). Although the magnitude of sex et al. 2005; Coop and Przeworski 2007). These genetic factors chromosome divergence in the large sex-linked region of control chiasmata formation during meiosis, possibly by mod- X. borealis is modest, it appears that recombination has ifying chromosome structure, or via the action of genes or been suppressed over sufﬁcient evolutionary time for these repetitive elements (Ji et al. 1999; Otto and Lenormand 2002). differences to be detectable, presumably for many thousands Curiously, chromosome 8 S in X. borealis also had a lower of generations or more. Supporting this, our second family of recombination rate that other chromosome linkage maps of lab-reared X. borealis and the surveyed panel of adults also had similar size (ﬁg. 2). This chromosome is homeologous (i.e., completely suppressed recombination in this large region related by genome duplication) to the sex chromosomes 8 L (there were some sex linked female heterozygous sites that (Session et al. 2016). This result offers the intriguing possibility appeared in both families and others that were unique to one that whatever is acting to suppress recombination on the sex family or the other, see supplementary S1.3, Supplementary chromosome may also inﬂuence recombination of homeolo- Material online). Together, these ﬁndings are consistent with gous sequence on chromosome 8 S (genome-wide, the L and observations made in other, more diverged species that the S nucleotide divergence is 6%; Session et al. 2016). This is extent of recombination suppression need not be more expan- unlikely to be an artifact of mapping errors because linkage sive in older than younger sex chromosomes (reviewed in groups would not form from markers that were a mix of Wright et al. 2016). They further demonstrate that newly chromosome 8 L and 8 S, because SNPs on different chromo- established sex chromosomes may assume radically different somes should have a recombination fraction of 0.5 (above evolutionary trajectories. our threshold; Materials and Methods). We infer here that the younger sex chromosomes of Sex-linkage with minimal divergence (similar to our obser- X. borealis have a larger region of suppressed recombination vations in X. borealis) has also been found in other species. For than the older sex chromosomes of X. laevis. One possibility is instance, the Japan sea population of stickleback ﬁsh have a that this is due to a large scale genomic change, such as an recently evolved set of sex chromosomes, which were gener- inversion or deletion leading to widespread recombination ated by a fusion of the ancestral sex chromosome and an suppression (Charlesworth et al. 2005). We were unable to autosome (Kitano et al. 2009). In this system, recombination characterize rearrangements in the sex chromosomes of suppression spread from the point of sex chromosome fusion X. borealis here due to the nature of our WGS data (short to an ancestral autosome along a large fraction of the neosex Genome Biol. Evol. 10(3):742–755 doi:10.1093/gbe/evy045 Advance Access publication February 21, 2018 751 Downloaded from https://academic.oup.com/gbe/article-abstract/10/3/742/4895093 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Furman and Evans GBE chromosome (Natri et al. 2013). Sex-linked genomic regions chromosomes (ﬁg. 4). Likewise, homeologous coding sequen- with variable levels of divergence suggest that the boundaries ces (including nonsynonymous and synonymous sites) also of recombination suppression evolve over time, and may en- have higher divergence (7%; Evans and Kwon 2015)than compass areas that are not yet diverged. As such, recombi- the nonrecombining region of the X. borealis sex chromo- nation may occasionally happen in these regions until a hard somes. These genomic patterns are consistent with the pro- recombination boundary is established (Bergero and posal that suppressed recombination in the sex chromosomes Charlesworth 2009). In some other amphibians, periodic re- of X. borealis occurred after allotetraploidization. Thus, even if combination may prevent divergence of the sex chromo- previous phylogenetic inferences (Furman and Evans 2016) somes (Perrin 2009; Sto ¨ ck et al. 2011; Dufresnes et al. are incorrect, the level of divergence between these sex chro- 2014). Though recombination was not detected in this region mosomes still argues that the expansion of the nonrecombin- for either family of X. borealis, it is possible that over long ing region occurred after the origin of DM-W (i.e., post-whole timescales the sex chromosomes of X. borealis may occasion- genome duplication in subgenus Xenopus) after or at least ally recombine. However, the divergence detected here be- within a similar time frame. tween the Z and W, though modest, indicates that recombination is not happening frequently enough to More Recombination in Females than Males, and in completely prevent divergence (ﬁg. 4). Different Genomic Regions Heterochiasmy refers to differences in sex-speciﬁc rates of The Relative Ages of the Sex Chromosomes of X. laevis and recombination. Here, in two independently derived sex chro- X. borealis mosome systems with female heterogamy, we observed het- Our inference that recombination suppression expanded erochiasmy with females having a higher rate of more quickly in X. borealis than X. laevis is based on (i) the recombination than males. In some species of bird and crab inferred origin of DM-W in subgenus after the whole genome with female heterogamy, recombination rates appear to be duplication event shared by all extant subgenus Xenopus spe- similar between the sexes (Groenen et al. 2008; Backstro¨m cies (Bewick et al. 2011) and (ii) inferred phylogenetic relation- et al. 2010; Cui et al. 2015; Nietlisbach et al. 2015). But in ships within subgenus Xenopus (Furman and Evans 2016), some ﬁsh and other bird species the rate of recombination is which indicates that the DM-W based sex determination sys- higher in heterogametic females (Hansson et al. 2010; Ruan tem is ancestral to the system of X. borealis (ﬁg. 1). If this et al. 2010), or higher in homogametic males (Kawakami phylogenetic inference were erroneous and instead the sex et al. 2014). In vertebrates with male heterogamy, the rate determining system of X. borealis were ancestral to the DM-W of recombination is often higher in females, particularly in XY based system of X. laevis, the rate that recombination sup- mammals (Wong et al. 2010; Ottolini et al. 2015), though pression expanded over the sex chromosomes of X. borealis exceptions are known where rates are similar between the could be slower than it seems here. sexes, or higher in males (Mank 2009a; Johnston et al. 2016, However, there are several lines of evidence that argue respectively). against X. borealis having the older sex chromosomes than In several other frog species with male heterogamy, heter- X. laevis. First, the strongest phylogenetic signal found using ochiasmy has been observed with a higher recombination rate 1,585 genes supports a paraphyletic clade of DM-W possess- in females (Berset-Brandli et al. 2008; Brelsford et al. 2016). ing species (ﬁg. 1; Furman and Evans 2016). More speciﬁcally, This was interpreted to be consistent with the Haldane– the alternate hypothesis of monophyly of DM-W-possessing Huxely Rule (Haldane 1922; Huxley 1928) which postulates species is supported by substantially fewer genes than the that when one sex does not recombine (i.e., when one sex is hypothesis of paraphyly of DM-W-possessing species with a achiasmatic), that sex is the heterogametic sex (Berset-Brandli sister relationship between DM-W-possessing Xenopus clivii et al. 2008; Brelsford et al. 2016). Our results suggest instead and X. borealis (as presented in ﬁg. 1; Furman and Evans that in species with heterochiasmy, the sex with lower recom- 2016). In fact, the hypothesis of monophyly of DM-W-pos- bination is not strongly linked to which sex is heterogametic sessing species has an equal support to another paraphyletic (Lenormand and Dutheil 2005). Heterochiasmy may be more relationship among DM-W-possessing species where X. bore- prominently inﬂuenced by haploid selection (Lenormand and alis is more closely related to X. laevis than X. clivii is to X. laevis Dutheil 2005), sexual antagonism (Mank 2009a), or other (Furman and Evans 2016). explanations. Additional evidence against the possibility of older sex The locations of recombination events were sex-biased in chromosomes in X. borealis is provided by divergence of both species of Xenopus investigated, with recombination orthologous autosomal genes of X. borealis and X. laevis most frequent in the center of chromosomes in females, ver- (e.g., divergence of synonymous site of 14%; Chain et al. sus the ends of chromosomes in males (ﬁg. 3). Sex speciﬁc 2008) that is substantially greater than that observed between differences in crossover location have been observed in other the nonrecombining regions of the X. borealis sex taxa, including, for example, frogs, dogs, and primates (Wong 752 Genome Biol. Evol. 10(3):742–755 doi:10.1093/gbe/evy045 Advance Access publication February 21, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/3/742/4895093 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Divergent Evolutionary Trajectories of Sex Chromosome Systems GBE et al. 2010; Venn et al. 2014; Ottolini et al. 2015; Brelsford ratio in X. laevis of 1.2:1 is within the range of a wide variety et al. 2016). Female linkage map length (in cM) and the num- of other species (1.4:1 for a ﬁsh, Ruan et al. 2010; 1.2:1 for a ber of crossover events was positively correlated with the mammal, Wong et al. 2010; 1.1:1 for a bird, Kawakami et al. amount of bp covered by the map and the total length of a 2014). Thus, the sex speciﬁc differences detected in X. laevis chromosome, whereas in males this relationship was not ob- are likely genuine. We note that the magnitude of the sex served (ﬁg. 2). A similar disparity between the sexes in the difference in recombination rate for X. borealis (females: relationship of cM and Mb spanned by linkage maps has been 719 cM and males: 165 cM) may be exaggerated due to observed in the frog Hyla arborea (Brelsford et al. 2016)and in lower genomic coverage in the X. borealis family (though humans (Ottolini et al. 2015). This sex speciﬁc difference could large differences in recombination between closely related be due to the differences in recombination location. In speciesisknown Kawakami et al. 2014). Furthermore, our females, because recombination is spread out across the mid- linkage maps are not capturing all recombination events in dle of chromosomes, longer chromosomes may permit more either species because the per gamete rates of recombination recombination events to occur without crossover interfer- are much less than the expectation of one event per chromo- ence. In males, where recombination occurs mostly on the some of 18 (Results). As such, caution should be used when tips of chromosomes, crossover interference is less likely to interpreting linkage maps from reduced genome sequencing vary among chromosomes with different lengths. Similar ﬁnd- technologies (e.g., RADseq, GBS), especially when a closely ings have been recovered in soay sheep, where male recom- related reference genome is lacking to assess marker distribu- bination is mostly biased to the last 18 Mb of each of the tion across chromosomes. chromosome tips, with chromosomes ranging in size from 50–200 Mb (Johnston et al. 2016), encompassing the chro- Drivers of Sex Chromosome Evolution and Stasis mosome length variation of Xenopus (Session et al. 2016). Why females and males have differences in recombination Information from a diversity of organisms suggest that the locations is potentially due to differences in meiosis. During age of sex chromosomes is not a strong predictor of the speramtogenesis there appears to be more control over for- amount divergence between sex chromosomes within a spe- mation and number of crossover events compared with oo- cies (Wright et al. 2016). Our ﬁndings from the sex chromo- genesis, with crossovers stopping in the presence of errors somes of X. borealis and X. laevis support this inference. One and more often restricted to one per arm (Hunt and possible explanation for these observations is that the geno- Hassold 2002; Hassold et al. 2004; Coop and Przeworski mic context in which a new sex chromosome system is estab- 2007). As well, maintenance of favorable allelic combination lished plays a large role in determining the extent of by haploid selection, which is generally stronger in males, may divergence a newly established will experience. For example, limit the breadth of possible crossover locations to genomic the ability to cope with dosage imbalances or the potential for regions, such as chromosome tips, that have low gene density dosage compensation mechanisms to evolve could strongly (Lenormand and Dutheil 2005). inﬂuence whether sex chromosomes become heteromorphic One possible caveat to our conclusions on sex speciﬁc dif- or not (Batada and Hurst 2007, but see Mank 2009b). If, for ferences in recombination rate is that in some cases maternal instance, the sex chromosomes of X. laevis (chromosome 2 L), and paternal linkage groups spanned nonoverlapping geno- contains more dosage sensitive genes than the sex chromo- mic regions, which themselves may vary in the local rate of somes of X. borealis (chromosome 8 L), this could hinder the recombination (Groenen et al. 2008; Kawakami et al. 2014; expansion of recombination suppression in X. laevis but not X. Ottolini et al. 2015). Since male recombination rate is biased borealis. In ratites, for example, an inability to accommodate toward tips of chromosomes (ﬁg. 3), it is possible that cross- dosage imbalances may prevent sex chromosome divergence over events were not accounted for in these linkage maps if beyond the limited regions thought to no longer recombine tags do not span to the ends of chromosomes. Kawakami (Adolfsson and Ellegren 2013; Vicoso et al. 2013; Yazdi and et al. (2014) also noted that RAD based studies in birds may Ellegren 2014). As well, the life history or ecological context of also underestimate linkage map lengths, because they under- a population can inﬂuence the fate of sex chromosomes. represent underrepresent microchromosomes and ends of Guppies, which similar to X. borealis have a large sex linked chromosomes. In this study, the disparity between female region without extensive degeneration, show variability in the and male linkage map lengths in X. laevis (1.2:1 ratio of extent of sex linkage on the chromosomes depending on an map length) is much less than X. borealis (4.4:1). The total interplay between the strength of sexual antagonism and map lengths in X. laevis (females: 1,572 cM and males: predation pressures in the population (Wright et al. 2017). 1,275 cM) was not far from a total map length of A compelling direction for further inquiry is to explore 1,800 cM, which is the expected length if there were an ob- factors that govern sex chromosome divergence and stasis ligate rate of one crossover per chromosome arm. This sug- in African clawed frogs, including the role of natural se- gests our estimate of recombination in X. laevis is not lection (e.g., favoring balanced gene dosage between the unreasonably low. As well, the female to male map length sexes, sexually antagonistic selection, haploid selection; Genome Biol. Evol. 10(3):742–755 doi:10.1093/gbe/evy045 Advance Access publication February 21, 2018 753 Downloaded from https://academic.oup.com/gbe/article-abstract/10/3/742/4895093 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Furman and Evans GBE Charlesworth B. 2009. Fundamental concepts in genetics: effective pop- Rice 1994; Lenormand 2003; Adolfsson and Ellegren ulation size and patterns of molecular evolution and variation. Nat Rev 2013), and nonselective events (e.g., recombination in Genet. 10(3):195. sex reversed individuals; Perrin 2009, or large scale Charlesworth B. 1991. The evolution of sex chromosomes. Science inversions). 251(4997):1030–1033. Charlesworth B, Charlesworth D. 2000. The degeneration of y chromo- somes. Philos Trans R Soc Lond B Biol Sci. 355(1403):1563–1572. Supplementary Material Charlesworth D, Charlesworth B, Marais G. 2005. Steps in the evolution of heteromorphic sex chromosomes. Heredity 95(2):118–128. Supplementary data areavailableat Genome Biology and Coop G, Przeworski M. 2007. An evolutionary view of human recombi- Evolution online. nation. Nat Rev Genet. 8(1):23–34. Cui Z, et al. 2015. High-density linkage mapping aided by transcriptomics documents ZW sex determination system in the chinese mitten crab Eriocheir sinensis. Heredity 115(3):206. Acknowledgments Dufresnes C, et al. 2014. Sex-chromosome differentiation parallels post- We thank Brian Golding for providing computational resour- glacial range expansion in european tree frogs (hyla arborea). Evolution 68(12):3445–3456. ces. We also thank Natural Sciences and Engineering Research Elshire RJ, et al. 2011. A robust, simple genotyping-by-sequencing (GBS) Council (NSERC) for funding support (CGSD3-475567-2015 approach for high diversity species. PLoS One 6(5):e19379. to B.L.S.F.; RGPIN/283102-2012 and RGPIN-2017-05770 to Evans BJ, et al. 2015. Genetics, morphology, advertisement calls, and his- B.J.E.). torical records distinguish six new polyploid species of african clawed frog (Xenopus, pipidae) from west and central africa. PLoS One 10(12):e0142823. Literature Cited Evans BJ, Kwon T. 2015. Molecular polymorphism and divergence of du- Adolfsson S, Ellegren H. 2013. Lack of dosage compensation accompanies plicated genes in tetraploid african clawed frogs (Xenopus). Cytogenet the arrested stage of sex chromosome evolution in ostriches. Mol Biol Genome Res. 145(3–4):243–252. Evol. 30(4):806–810. Furman BLS, Evans BJ. 2016. Sequential turnovers of sex chromosomes in Amster G, Sella G. 2016. Life history effects on the molecular clock of africanclawedfrogs (Xenopus) suggest some genomic regions are autosomes and sex chromosomes. Proc Natl Acad Sci U S A. good at sex determination. G3 (Bethesda) 6(11):3625–3633. 113(6):1588–1593. Goudet J, Raymond M, de Meeu ¨ s T, Rousset F. 1996. Testing differenti- Bachtrog D. 2013. Y chromosome evolution: emerging insights into ation in diploid populations. Genetics 144(4):1933–1940. processes of y chromosome degeneration. Nat Rev Genet. Graves JAM. 2006. Sex chromosome specialization and degeneration in 14(2):113. mammals. Cell 124(5):901–914. Bachtrog D, Charlesworth B. 2002. Reduced adaptation of a non- Groenen MA, et al. 2008. A high-density snp-based linkage map of the recombining neo-y chromosome. Nature 416(6878):323–326. chicken genome reveals sequence features correlated with recombi- Backstro ¨ m N, et al. 2010. The recombination landscape of the zebra ﬁnch nation rate. Genome Res. 19(3):510–519. Taeniopygia guttata genome. Genome Res. 20(4):485–495. Hackett C, Broadfoot L. 2003. Effects of genotyping errors, missing values Baird NA, et al. 2008. Rapid SNP discovery and genetic mapping using and segregation distortion in molecular marker data on the construc- sequenced rad markers. PLoS One 3(10):e3376. tion of linkage maps. Heredity 90(1):33. Batada NN, Hurst LD. 2007. Evolution of chromosome organization driven Haldane JB. 1922. Sex ratio and unisexual sterility in hybrid animals. J by selection for reduced gene expression noise. Nat Genet. Genet. 12(2):101–109. 39(8):945–949. Hansson B, et al. 2010. Avian genome evolution: insights from a linkage Bates D, M€ achler M, Bolker B, Walker S. 2015. Fitting linear mixed-effects map of the blue tit (Cyanistes caeruleus). Heredity 104(1):67. models using lme4. J Stat Softw. 67(1):1–48. Hassold T, et al. 2004. Cytological studies of meiotic recombination in Bergero R, Charlesworth D. 2009. The evolution of restricted recombina- human males. Cytogenet Genome Res. 107(3–4):249–255. tion in sex chromosomes. Trends Ecol Evol. 24(2):94–102. Hunt PA, Hassold TJ. 2002. Sex matters in meiosis. Science Bergero R, Charlesworth D, Filatov DA, Moore RC. 2008. Deﬁning regions 296(5576):2181–2183. and rearrangements of the Silene latifolia y chromosome. Genetics Huxley J. 1928. Sexual difference of linkage in Gammarus chevreuxi.J 178(4):2045–2053. Genet. 20(2):145–156. Bergero R, Forrest A, Kamau E, Charlesworth D. 2007. Evolutionary strata Ji Y, Stelly DM, De Donato M, Goodman MM, Williams CG. 1999. A on the X chromosomes of the dioecious plant Silene latifolia: evidence candidate recombination modiﬁer gene for Zea mays l. Genetics from new sex-linked genes. Genetics 175(4):1945–1954. 151(2):821–830. Berset-Br€ andli L, Jaqui ery J, Broquet T, Ulrich Y, Perrin N. 2008. Extreme Johnston SE, B er enos C, Slate J, Pemberton JM. 2016. Conserved genetic heterochiasmy and nascent sex chromosomes in european tree frogs. architecture underlying individual recombination rate variation in a Proc R Soc Lond B Biol Sci. 275(1642):1577–1585. wild population of soay sheep (Ovis aries). Genetics 203(1):583–598. Bewick AJ, Anderson DW, Evans BJ. 2011. Evolution of hte closely related, Kamiya T, et al. 2012. A trans-species missense snp in Amhr2 is associated sex-related genes DM-W and dmrt1 in africanclawedfrogs (Xenopus). with sex determination in the tiger pufferﬁsh, Takifugu rubripes (fugu). Evolution 65(3):698–712. PLoS Genet. 8(7):e1002798. Brelsford A, Dufresnes C, Perrin N. 2016. High-density sex-speciﬁc linkage Kawakami T, et al. 2014. A high-density linkage map enables a second- maps of a european tree frog (Hyla arborea) identify the sex chromo- generation collared ﬂycatcher genome assembly and reveals the pat- some without information on offspring sex. Heredity 116(2):177–181. terns of avian recombination rate variation and chromosomal evolu- Chain FJ, Ilieva D, Evans BJ. 2008. Duplicate gene evolution and ex- tion. Mol Ecol. 23(16):4035–4058. pression in the wake of vertebrate allopolyploidization. BMC Evol Kitano J, et al. 2009. A role for a neo-sex chromosome in Stickleback Biol. 8(1):43. speciation. Nature 461(7267):1079–1083. 754 Genome Biol. Evol. 10(3):742–755 doi:10.1093/gbe/evy045 Advance Access publication February 21, 2018 Downloaded from https://academic.oup.com/gbe/article-abstract/10/3/742/4895093 by Ed 'DeepDyve' Gillespie user on 16 March 2018 Divergent Evolutionary Trajectories of Sex Chromosome Systems GBE Kosambi D. 1943. The estimation of map distances from recombination markers and its application in qtl analysis. Aquaculture 308(3): values. Ann Hum Genet. 12(1):172–175. 89–100. Lenormand T. 2003. The evolution of sex dimorphism in recombination. Session AM, et al. 2016. Genome evolution in the allotetraploid frog Genetics 163(2):811–822. Xenopus laevis. Nature 538(7625):336–343. Lenormand T, Dutheil J. 2005. Recombination difference between sexes: a Shen JJ, Dushoff J, Bewick AJ, Chain FJ, Evans BJ. 2013. Genomic dynam- role for haploid selection. PLoS Biol. 3(3):e63. ics of transposable elements in the western clawed frog (Silurana Makova KD, Li WH. 2002. Strong male-driven evolution of DNA sequences tropicalis). Genome Biol Evol. 5(5):998–1009. in humans and apes. Nature 416(6881):624–626. Sto ¨ ck M, et al. 2011. Ever-young sex chromosomes in european tree frogs. Mank JE. 2009a. The evolution of heterochiasmy: the role of sexual selec- PLoS Biol. 9(5):e1001062. tion and sperm competition in determining sex-speciﬁc recombination Sto ¨ ck M, et al. 2013. Low rates of X-Y recombination, not turnovers, rates in eutherian mammals. Genet Res. 91(5):355–363. account for homomorphic sex chromosomes in several diploid species Mank JE. 2009b. The W, X, Y and Z of sex-chromosome dosage compen- of palearctic green toads (Bufo viridis subgroup). J Evol Biol. sation. Trends Genet. 25(5):226–233. 26(3):674–682. Margarido GRA, de Souza AP, Garcia AAF. 2007. Onemap: software for Tymowska J. 1991. Polyploidy and cytogenetic variation in frogs of the genetic mapping in outcrossing species. Hereditas 144(3):78–79. genus Xenopus. In: Green DM, Sessions SK, editors. Amphibian cyto- Matsuda Y, Uno Y, Kondo M, Gilchrist MJ, Zorn AM, Rokhsar DS, Schmid genetics and evolution. San Diego (CA): Academic Press. p. 259–297. M, Taira M. 2015. A new nomenclature of Xenopus laevis chromo- Tymowska J, Fischberg M. 1973. Chromosome complements of the genus somes based on the phylogenetic relationship to Silurana/Xenopus Xenopus. Chromosoma 44(3):335–342. tropicalis. Cytogenet Genome Res. 145(3–4):187–191. Venn O, et al. 2014. Strong male bias drives germline mutation in chim- Mawaribuchi S, et al. 2017. Sex chromosome differentiation and the W- panzees. Science 344(6189):1272–1275. and Z-speciﬁc loci in Xenopus laevis. Dev Biol. 426(2):393–400. Vicoso B, Kaiser VB, Bachtrog D. 2013. Sex-biased gene expression at Natri HM, Shikano T, Meril€ a J. 2013. Progressive recombination suppres- homomorphic sex chromosomes in emus and its implication for sex sion and differentiation in recently evolved neo-sex chromosomes. Mol chromosome evolution. Proc Natl Acad Sci U S A. Biol Evol. 30(5):1131–1144. 110(16):6453–6458. Nietlisbach P, et al. 2015. A microsatellite-based linkage map for song Wong AK, et al. 2010. A comprehensive linkage map of the dog genome. sparrows (Melospiza melodia). Mol Ecol Resourc. 15(6):1486–1496. Genetics 184(2):595–605. Olmstead AW, Lindberg-Livingston A, Degitz SJ. 2010. Genotyping sex in Wright AE, Dean R, Zimmer F, Mank JE. 2016. How to make a sex chro- the amphibian, Xenopus (Silurana) tropicalis, for endocrine disruptor mosome. Nat Commun. 7:12087. bioassays. Aquat Toxicol. 98(1):60–66. Wright AE, et al. 2017. Convergent recombination suppression suggests Otto SP, Lenormand T. 2002. Resolving the paradox of sex and recombi- role of sexual selection in guppy sex chromosome formation. Nat nation. Nat Rev Genet. 3(4):252. Commun. 8:14251. Ottolini CS, et al. 2015. Genome-wide maps of recombination and chro- Wu R, Ma CX, Painter I, Zeng ZB. 2002. Simultaneous maximum likelihood mosome segregation in human oocytes and embryos show selection estimation of linkage and linkage phases in outcrossing species. Theor for maternal recombination rates. Nat Genet. 47(7):727–735. Popul Biol. 61(3):349–363. Perrin N. 2009. Sex reversal: a fountain of youth for sex chromosomes? Yazdi HP, Ellegren H. 2014. Old but not (so) degenerated–slow evolution Evolution 63(12):3043–3049. of largely homomorphic sex chromosomes in ratites. Mol Biol Evol. R Core Team. 2016. R: a language and environment for statistical com- 31(6):1444–1453. puting. Vienna (Austria): R Foundation for Statistical Computing. Yoshimoto S, et al. 2008. A w-linked dm-domain gene, dm-w, participates Rice WR. 1987. The accumulation of sexually antagonistic genes as a se- in primary ovary development in Xenopus laevis. Proc Natl Acad Sci U S lective agent promoting the evolution of reduced recombination be- A. 105(7):2469–2474. tween primitive sex chromosomes. Evolution 41(4):911–914. Zhou Q, et al. 2014. Complex evolutionary trajectories of sex chromo- Rice WR. 1994. Degeneration of a nonrecombining chromosome. Science somes across bird taxa. Science 346(6215):1246338. 263(5144):230–231. Zickler D, Kleckner N. 2016. A few of our favorite things: pairing, the bou- Roco AS, Olmstead AW, Degitz SJ, Amano T, Zimmerman LB, Bullejos M. quet, crossover interference and evolution of meiosis. In: Seminars in cell 2015. Coexistence of Y, W, and Z sex chromosomes in Xenopus and developmental biology. Vol. 54. Elsevier. p. 135–148. tropicalis. Proc Natl Acad Sci U S A. 112(34):E4752–E4761. Ruan X, Wang W, Kong J, Yu F, Huang X. 2010. Genetic linkage mapping of turbot (Scophthalmus maximus L.) using microsatellite Associate editor: Judith Mank Genome Biol. Evol. 10(3):742–755 doi:10.1093/gbe/evy045 Advance Access publication February 21, 2018 755 Downloaded from https://academic.oup.com/gbe/article-abstract/10/3/742/4895093 by Ed 'DeepDyve' Gillespie user on 16 March 2018
Genome Biology and Evolution – Oxford University Press
Published: Mar 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera