Origin and evolutionary trajectories of brown algal sex chromosomes

Abstract

Research on the biology and evolution of sex chromosomes has primarily focused on diploid XX/XY and ZW/ZZ systems. In contrast, the rise, evolution and demise of U/V systems has remained an enigma. Here we analyse genomes of nine brown algal species with different sexual systems to determine the history of their sex determination. U/V sex chromosomes emerged between 450 and 224 million years ago, when a region containing the pivotal male-determinant MIN ceased recombining. Seven ancestral genes within the sex-determining region show remarkable conservation over this vast evolutionary time, although nested inversions caused expansions of the sex locus, independently in each lineage. We evaluate whether these expansions are associated with increased morphological complexity and sexual differentiation, and show that taxonomically restricted genes evolve unexpectedly often in U and V chromosomes. We also investigate two situations in which U/V-linked regions have changed. First, we demonstrate that convergent evolution of two monoicous species occurred by ancestral males acquiring U-specific genes. Second, the Fucus dioecious system involves new sex-determining gene(s), acting upstream of formerly V-specific genes during development. Both situations have led to the demise of U and V chromosomes and erosion of their specific genomic characteristics.

Main

The mechanisms controlling the development of male or female identities, or co-sexuality, when individuals express both sex functions, vary widely across different organisms^1,2. In species with separate sexes, sex chromosomes may be present, carrying a sex-determining region (SDR)³ that encodes factors directing sex identity and which often does not undergo recombination in the heterogametic sex (XY or ZW)⁴ of diploid species (dioecious), or in the diploid stage of haploid-dominant (dioicous) species. Sex chromosomes have independently evolved from autosomes multiple times and may be subject to specific evolutionary forces, including differential selection between sexes, asymmetrical expression of deleterious mutations and hemizygosity, meiotic silencing and dosage compensation³.

Research on the biology and evolution of sex chromosomes has primarily focused on diploid XX/XY and ZW/ZZ systems in mammals, birds, fish, Drosophila and diploid plants^4,5. U/V haploid sex-determination systems, such as those of bryophytes and algae^6,7, have been less explored. In U/V systems, sex is not determined at fertilization but during meiosis, when haploid spores inherit either a U chromosome, and will develop into a female gametophyte, or a V chromosome, controlling male gametophyte formation⁸. These fundamental inheritance differences between U/V and XX/XY or ZW/ZZ systems have broad evolutionary and genomic implications^9,10. However, so far, only the U/V systems of the brown alga Ectocarpus and the U/Vs of four distantly related bryophyte taxa^11,12,13,14 have been fully sequenced and assembled into chromosomes. While these studies helped understand the genomic structure of bryophyte U- and V-linked regions, the species involved diverged ~500 Ma (million years ago) and do not share homologous U/V chromosomes¹⁵. As a result, we still lack a broad comparative view across multiple homologous U/V systems that would inform a reconstruction of their evolutionary history. Brown algae represent exceptional models for studying sex chromosome evolution because they display diverse reproductive systems, life cycles and sex chromosome systems in a single lineage¹⁶. Their ancestral state probably involved separate sexes¹⁶, suggesting that their sex chromosomes could share a common origin. Here we study the origin, evolution and demise of U/V sex chromosomes in the brown algae.

Results

The origin of brown algal sex chromosomes

We focused on species covering the phylogenetic, morphological and reproductive diversity of the brown algal clade¹⁷ and their closest extant outgroup, Schizocladia ischiensis^17,18. We substantially improved the brown algal genome datasets available^18,19,20 to reach chromosome or near-chromosome-level genome assemblies (Extended Data Table 1 and Supplementary Table 1). This revealed that brown algae have 27–33 chromosomes and largely conserved macrosynteny (Fig. 1a).

Fig. 1: Origins of U/V sex chromosomes in brown algae.

a, Macrosynteny plot comparing genomes of six dioicous (green), two monoicous (red), one dioecious (blue) and one outgroup species (yellow). The chromosomes were originally numbered by their physical size in the Ectocarpus v2 genome¹⁴¹. Note that the dioecious species F. serratus has a fully diploid life cycle (without gametophytes⁷). Syntenic blocks of the V sex chromosome are highlighted in red, with the emergence of U/V chromosomes shown in the phylogeny. Genome sizes are indicated in brackets. b, Microsynteny plot of V chromosomes in five dioicous species, highlighting the male sex-determining regions (blue) and the PARs (green). The PAR genes whose orthologues are found within the SDR of other species are highlighted in purple. c, Microsynteny plot of U chromosomes in two dioicous species, highlighting the female sex-determining regions (peach) and the PARs (green). The PAR genes whose orthologues are found within the SDR of other species are highlighted in purple. Note that the genome assemblies for C. linearis and S. ischiensis are not chromosome level, leading to a high number of contigs.

We identified the female (U) and male (V) sex-determining regions (SDRs) in the dioicous species using a combination of bioinformatic and experimental approaches (see ‘Discovery of the U/V sex determination regions’ in Methods; Supplementary Figs. 1–5). All U/V species share the same, albeit highly rearranged, ancestral sex chromosome, showing remarkable stability despite the large evolutionary time (Fig. 1b,c and Extended Data Table 1). The recombination suppression event leading to the birth of U/V sex chromosomes occurred after the split of S. ischiensis and Dictyota dichotoma, ~450–224 Ma²¹ (Fig. 1a). The male-determining gene MIN²² is the only V-specific gene consistently present in all V-SDRs of the dioicous species. We note that one dioecious (Fucus serratus) and two monoicous (haploid, co-sexual Chordaria linearis and Desmarestia dudresnayi) species lack U/V sex chromosomes but still retain MIN on a chromosome homologous to the ancestral U/V (‘U/V-homologue’ hereafter). The outgroup S. ischiensis has low synteny with the brown algae and exhibits putative fusion-with-mixing events²³ (Fig. 1a and Extended Data Fig. 1).

We next examined the U/V-SDRs by comparing male and female genome assemblies (Methods). The SDRs contain a small number of genes overall (between 18 and 52), and compared with the pseudoautosomal regions (PARs) (between 229 and 904), with considerable variation in gene content and size across species, the smallest being found in the Ectocarpales (Ectocarpus sp. 7, Ectocarpus crouaniorum, Scytosiphon promiscuus; Fig. 1b,c, Extended Data Fig. 2a and Extended Data Table 1). SDR size differences across species are strongly correlated with the number of genes (R² = 0.97; Extended Data Fig. 2b) and the repeat content (R² = 0.99; Extended Data Fig. 2c) inside these regions. Many genes located in the PARs of the Ectocarpales are within the V-SDRs of Undaria pinnatifida, Desmarestia herbacea and D. dichotoma, indicating that the SDR boundaries have changed across species. The boundary differences coincide with extensive structural rearrangements, particularly inversions, even among closely related taxa (Fig. 1b,c and Extended Data Fig. 3). Note that the centromere in the V chromosome of Ectocarpus is found within the SDR¹⁹, so we cannot exclude that a centromere-related suppression of recombination may have preceded the inversion events found on the SDR²⁴.

Together, our results indicate that the brown algal U/V sex chromosomes evolved between 450–224 Ma, via suppressed recombination in a genomic region that contained MIN (henceforth male-determining locus). The presence of MIN in distantly related lineages could push the age of the U/V chromosomes further back in time, but more evidence would be required to establish that dioicy existed in these organisms.

The evolution of the SDRs involved boundary expansions and gene gains

The brown algal U- and V-SDRs carry homologous genes (gametologue pairs), indicating descent from a common ancestral region (Supplementary Table 2). Ectocarpus sp. 7 and D. herbacea show similar ratios of gametologues and U- or V-specific genes (16/14 and 11/7 gametologue/sex-specific genes in Ectocarpus and D. herbacea, respectively; Supplementary Table 2). Only ten genes share SDR orthologues between both species, while the rest were mostly acquired independently in the SDR of each species, with one gene that was retained as a gametologue pair in Ectocarpus sp. 7 but lost both copies in D. herbacea (Supplementary Table 2). Five gametologue pairs conserved both copies in the two species, while another three gametologue pairs lost either the male or the female copy in D. herbacea (Supplementary Table 2). In addition, MIN and a U-specific gene are also conserved between species (Supplementary Table 2). Although the total number of U/V-SDR genes differs between Ectocarpus sp. 7 (18 genes) and D. herbacea (30 genes), each species shows an equal number of gametologues and sex-specific genes in its U- and V-SDRs (Supplementary Table 2). This intraspecies symmetry supports the idea that the U and V chromosomes may have undergone parallel evolutionary changes within each lineage^10,25,26. The V-SDR of D. herbacea contains 20 additional genes that belong to endogenous viral elements, which are common across brown algal genomes¹⁸.

Diploid sex chromosome in animals and plants exhibit evolutionary strata representing different recombination suppression events over time. Strata are identified by analysing synonymous substitutions (K_s) between male/female gametologue pairs²⁷ whose locations in fully X or Z-linked regions are known. However, detecting evolutionary strata in U/V systems is difficult because neither of these fully sex-linked regions recombines and gene movements and chromosome rearrangements disrupt collinearity of both chromosomes between species^25,28,29. Moreover, in the absence of a recombining outgroup (which does not exist in brown algae), the ancestral gene order cannot be reliably inferred. In both Ectocarpus sp. 7 and D. herbacea, the V- and U-SDR rearrangements differ by inversions (Fig. 2a,b and Extended Data Fig. 4), consistent with the idea that inversions may lead to suppressed recombination between sex chromosomes. An analysis of gametologue pair divergence revealed saturated levels of K_s values (Fig. 2b,c and Supplementary Table 3), further limiting the inference of evolutionary strata across brown algal SDRs. Nonetheless, the gametologue K_s values are broadly consistent between orthologues in Ectocarpus sp. 7 and D. herbacea, where shared SDR gametologues between species have higher K_s values and probably spent more evolutionary time diverging than the gametologues that are not shared between species (Supplementary Table 3). Furthermore, the location of gametologues with the lowest K_s values in the U-SDR of D. herbacea, relative to the PAR genes in Ectocarpus sp. 7, suggests that inversions involving the entire U-SDR and adjacent PAR segments probably contributed to the expansion of the U/V-SDR boundaries in D. herbacea, in a process we term ‘engulfment’ (Extended Data Fig. 4). The expansion of the SDR boundaries in D. herbacea led to the engulfment of a region containing four genes in the PAR1 of Ectocarpus sp. 7, and a second region with 13 genes located on the PAR2 (Fig. 2d, Supplementary Table 4 and Extended Data Fig. 4). Twelve of these engulfed genes into the SDR of D. herbacea were retained as gametologues. These observations support a scenario where expansions in the SDR boundaries of brown algae occur through nested inversions. Two chromatin-related transcription factors in the Ectocarpus PARs were independently incorporated into the SDRs of four other dioicous species (Supplementary Tables 4 and 5).

Fig. 2: Lineage-specific U/V-SDR expansion from an ancestral SDR and its association with sexual dimorphism.

a, Microsynteny plot between the U and V chromosomes of Ectocarpus sp. 7 and D. herbacea. b, Synteny between the U and V gametologues within the V-SDRs of both species, coloured by synonymous substitutions per site (K_s). c, Identification of ancestral SDR gametologues (red squares) and independently acquired gametologues (black dots) with respect to their gametologue K_s values and their position in the V-SDR. d, Circos plot linking gametologue pairs in each species with expression levels of all SDR genes (log₂(TPM + 1)) across different life stages in Ectocarpus sp.7 and mature gametophytes (matGA) in D. herbacea. Gametologues are highlighted in dark colours, sex-specific genes are highlighted in light colours, viral insertions are marked in grey and asterisks denote conserved SDR genes (also in f). e, The ancestral state reconstruction of V-SDR gene content across brown algae, showing the expected number of genes in the SDR (white circles), gene retention (blue numbers), gene gain through expansion of the SDR boundaries (orange), gene gain through autosomal translocation (purple), gene birth event inside the SDR (yellow) and gene loss (red) along with changes in gamete dimorphism¹⁶. f, Schematic of the seven ancestral V-SDR gene orthogroups (OGs), with genomic locations marked: retained in the V-SDR (blue), found in the U/V-homologue of non-dioicous species (green), translocated from the V-SDR to an autosome (yellow), present in a non-scaffolded contig (red), and lost (grey). Bold: MIN. See Supplementary Tables 5 and 6.

The observation of greater V-SDR gene content in early diverging lineages (such as D. dichotoma) than in the later-diverging Ectocarpales (Extended Data Table 1) could reflect either gene loss in the V-SDRs of Ectocarpales or independent gene gains in the V-SDRs of each lineage (as predicted in ref. ¹⁰), from an ancestral state with low V-SDR gene content that is retained in Ectocarpales. To distinguish between these possibilities, we reconstructed the ancestral SDR gene content (Supplementary Table 5), focusing on the V chromosome, as the genomic data are of better quality (Supplementary Table 1), and assuming parallel U/V-SDR evolution^10,25,26 as seen in Ectocarpus sp. 7 and D. herbacea (Fig. 2d and Supplementary Table 2). This analysis revealed that brown algal V-SDR evolution occurred via lineage-specific gene gains rather than gene loss in the Ectocarpales (Fig. 2e,f). Gene gains were caused by a combination of three processes: expansions of the SDR boundaries into the PARs, translocation of autosomal genes into the SDR and lineage-specific gene birth events within the SDR (Fig. 2e and Supplementary Table 5). Consistently, ancestral V-SDR genes were associated with higher gametologue K_s values, while independently acquired gametologues in D. herbacea had lower K_s values (Supplementary Table 2 and Fig. 2c).

The seven genes in the ancestral V-SDR (Fig. 2f and Extended Data Fig. 2d) include the male-determinant MIN²² and six V gametologues of genes that are also carried on the U chromosome (Fig. 2f and Supplementary Table 6). As predicted by early models of U/V-SDR evolution¹⁰, all seven genes are probably related to sex determination processes. Gametologue pairs include putative transmembrane proteins that may play a role in gamete recognition³⁰, STE20 serine/threonine kinase gametologues probably involved in pheromone pathways³¹, and a casein kinase, a MEMO-like domain protein and a GTPase-activating protein which may act in signal transduction (Supplementary Table 6). All of these genes are gametologue pairs in Ectocarpus sp. 7, but in D. herbacea the casein kinase was lost from the U-SDR and the putative transmembrane receptor was lost in both sexes. We noticed that these ancestral V-SDR genes remain in the U/V-homologue of the species that have lost their U/V system (Fig. 2f and Supplementary Table 6), emphasizing their importance for pathways in sex organ development even in the absence of sex chromosomes.

The V-SDR size appears to be associated with the level of sexual dimorphism (Fig. 2e and Extended Data Fig. 2), but the small sample size is insufficient for formal statistical analysis. Species with low sexual dimorphism (anisogamous) retained the ancestral V-SDR genes with very few gene gains, further suggesting that they may represent the V-SDR ancestral state. Although the number of SDR changes is small, oogamous species each independently gained diverse V-SDR genes, and one gene (ATP-dependent RNA helicase) was convergently acquired in all (OG0003211 in Extended Data Fig. 2d). All the detected autosomal translocations into the V-SDRs of Ectocarpus sp. 7 and D. herbacea (Fig. 2e) also involve sex-specific genes (Supplementary Tables 2 and 5), consistent with a model where sexual antagonism in autosomal loci may be solved by gaining sex linkage³². In contrast, we found no correlation between autosomal sex-biased gene (SBG) expression and sexual dimorphism level (false discovery rate (FDR)-corrected P > 0.01; Supplementary Table 7), supporting previous studies³³ (Extended Data Fig. 5a). However, we observed an enrichment of male-biased genes on the PARs in all species (chi-square test P < 0.01) except D. dichotoma (Extended Data Fig. 5b).

Most U/V-SDR genes were prominently expressed in fertile haploid gametophytes, consistent with gene preservation via haploid purifying selection (Supplementary Table 8). Gametologues had typically higher expression levels than sex-specific genes (present in only one of the SDRs) (Wilcoxon test, P = 0.00075 in D. herbacea; P = 0.08843 in Ectocarpus sp. 7) (Fig. 2d). A comparative analysis in fertile gametophytes between SDR genes and their autosomal counterparts in other species showed that newly acquired genes on the SDR had similar expression levels to their autosomal counterparts (Extended Data Fig. 6), suggesting either a co-option of autosomal biological activity into male-specific functions in the V-SDR or the general importance of these genes for gametophyte development. Examining expression levels across multiple tissues in Ectocarpus sp. 7 revealed that activity of U/V-SDR genes is not confined to fertile gametophytes (Fig. 2d). Therefore, the SDRs contain not only genes involved in sex determination and gametophyte fertility but also genes playing a broader role in development.

Altogether, our analyses illustrate how brown algal U/V-SDRs undergo structural changes, evolving mainly by lineage-specific gene gains associated with increasing levels of sexual dimorphism. We identified a set of conservatively sex-linked genes in dioicous brown algae, suggesting their role in sex determination and/or differentiation, along with genes potentially involved in other developmental pathways.

Structural features and evolutionary dynamics of brown algal U/V sex chromosomes

We next examined the structural features that differentiate the entire U/V sex chromosomes (V-SDR and PARs) from the rest of the genome (Fig. 3a and Supplementary Fig. 12a). As expected for non-recombining regions^25,34, all V sex chromosomes are repeat rich and gene poor (Wilcoxon rank-sum test, FDR-corrected P < 0.01; Fig. 3a, Extended Data Fig. 7 and Supplementary Tables 9–11). V-SDRs have significantly higher repeat density than the PARs or the autosomes (permutation test, FDR-corrected P < 0.001; Extended Data Fig. 8). This low gene density is not influenced by the presence of centromeres within the SDRs, as the coding density in the Ectocarpus sp. 7 V centromere (3.51%) is slightly higher than in the rest of the V-SDR (2.85%), presumably due to the small size of the centromere (153 kbp)¹⁹. The PARs were also significantly enriched in repeats when compared with the autosomes, although less so than the V-SDRs (permutation test, FDR-corrected P < 0.001; Extended Data Fig. 8). Among repetitive elements, ‘unclassified’ transposable elements (TEs) were enriched in the PARs and SDRs of the Ectocarpales (permutation test, FDR-corrected P < 0.01), while the V-SDRs of species that underwent genome expansion (for example, U. pinnatifida, D. herbacea, D. dichotoma) predominantly accumulated long terminal repeat (LTR) elements (Supplementary Fig. 6 and Supplementary Table 9).

Fig. 3: The U/V sex chromosomes are enriched in taxonomically restricted genes.

a, Karyoplots for five dioicous species showing the following features from bottom to top: chromosome compartments (autosomes, PARs and SDR), relative gene ages, interspecies K_s values, and proportion of coding (CDS, red) and repeat (TEs, blue) density. Statistically significant differences for each feature between each autosome and the V chromosome are depicted on top of the track for that autosome (FDR-corrected two-sided Wilcoxon rank-sum test; values indicated with solid colours when P < 0.01 for the tested hypothesis). The precise range of gene age categories and interspecies K_s values for each species can be found in Supplementary Figs. 7–11. b, Violin plots for five dioicous species showing the relative gene age ranks (higher ranks equate to younger ages) of the TRGs across chromosome compartments (autosomes, PARs and SDR). Statistically significant differences in mean values of gene ages (centre line) were assessed using FDR-corrected two-sided permutation tests.

Moreover, sex chromosomes had fewer orthologues conserved between species compared with the autosomes (chi-square test, P < 10⁻⁴, Supplementary Table 12), possibly reflecting increased numbers of taxonomically restricted genes (TRGs; that is, genes that are not detectable outside of a defined taxonomic group). Phylostratigraphy analyses^35,36 confirmed an enrichment of TRGs in the sex chromosomes of all dioicous species (Wilcoxon rank-sum test, FDR-corrected P < 0.01; Fig. 3b and Supplementary Tables 13 and 14). TRG enrichment was localized in the PARs of the Ectocarpales, but this pattern extended to the entire sex chromosome, including the SDRs, in species with larger V-SDRs (permutation test, FDR-corrected P < 0.001; Fig. 3b). Importantly, sex chromosomes have statistically younger TRGs than the last common ancestor of the five dioicous species (same Order or broader taxonomic groups), indicating that TRG enrichment arose independently in each species (Pearson standardized residuals >2.4; Supplementary Figs. 7–11).

We previously proposed a theoretical model where generation-antagonistic selection may favour the retention of young sporophyte-beneficial loci in the PARs of Ectocarpus sp. 7 U/Vs³⁷. Consistent with this model, sporophyte-biased genes are indeed enriched in the sex chromosomes of Ectocarpus sp. 7 and U. pinnatifida (fold change >2, adjusted P < 0.05), but less so in D. dichotoma and S. promiscuus (Supplementary Table 15). Moreover, we explored additional mechanisms underlying TRG emergence by estimating interspecies K_s values between orthologues in closely related species (Supplementary Table 16) and comparing these values between chromosomes and genomic compartments (V-SDR, PAR, autosomes). If synonymous mutations behave neutrally^38,39, then interspecies K_s can be used as a proxy for mutation rates^40,41. Consistently, we found higher interspecies K_s values in the V sex chromosomes compared with autosomes across all dioicous species (Wilcoxon rank-sum test, FDR-corrected P < 0.01; Fig. 3a and Supplementary Tables 16 and 17), suggesting higher mutation rates relative to autosomes. Higher interspecies K_s values are also localized in the PARs, mirroring the pattern observed with the TRGs (Supplementary Figs. 7–10). Therefore, the enrichment of TRGs in the U and V is associated with both enrichment of sporophyte-biased genes and higher synonymous substitution rates.

To test the generality of the pattern of TRG enrichment on U and V chromosomes, we applied the same approach in other organisms with haploid sex determination, the plants Ceratodon purpureum, Sphagnum angustifolium, Marchantia polymorpha^11,12,42 and the fungus Cryptococcus neoformans⁴³. We observed a clear enrichment of TRGs in the V chromosomes of C. purpureum and S. angustifolium (Wilcoxon rank-sum test, FDR-corrected P < 0.01; Supplementary Figs. 12 and 13 and Supplementary Tables 13 and 14), but not in the U/V chromosomes of M. polymorpha or the mating-type chromosome of C. neoformans (Supplementary Figs. 14 and 15 and Supplementary Tables 13 and 14).

Fate of U/V sex chromosomes following loss of dioicy

We studied the evolutionary trajectory of brown algal genomes after the loss of the U/V system, by exploring two independent transitions to monoicy in C. linearis and D. dudresnayi that undergo sexual reproduction and develop male and female gametangia³³. Most genes in the ‘ex’-sex chromosomes (U/V-homologues) of both monoicous species are male derived, indicating that monoicy emerged from a male background (Fig. 4a,b). The U/V-homologue of C. linearis contains several rearrangements spanning the regions that are homologous to the PAR and SDR (SDR-homologue), with 11 V-SDR-derived and 2 U-SDR-derived orthologues located within the SDR-homologue, with an additional V-SDR-derived gene that was translocated elsewhere in the U/V-homologue (Fig. 4a and Supplementary Table 18). Likewise, D. dudresnayi underwent at least two inversion events within the SDR-homologue after splitting from D. herbacea (Fig. 4b), containing 20 V-SDR-derived genes and 4 U-SDR-derived genes (Supplementary Table 19).

Fig. 4: Fate of sex chromosomes during transitions from dioicy to co-sexuality (monoicy).

a, Comparison of the U/V-homologue in C. linearis against the U and V chromosomes of Ectocarpus sp. 7. b, Comparison of the U/V-homologue in D. dudresnayi against the U and V chromosomes of D. herbacea. The colour code represents the identity of the genes alongside the chromosomes, while the shapes represent the evolutionary fate of each SDR gene in the monoicous genome. The matching shades between the SDRs and the U/V-homologue are either colour coded by their ancestral background or they appear as transparent dotted shades if the gametologue of the other sex was retained. c, Karyoplot of D. dudresnayi showing the following features from bottom to top: chromosome compartment (autosomes and U/V-homologue), relative gene ages, interspecies K_s values, proportion of coding (CDS, red) and repeat (TEs, blue) density. Statistically significant differences for each feature between each autosome and the U/V-homologue are depicted on top of the track for that autosome (FDR-corrected two-sided Wilcoxon rank-sum test; values indicated with solid colours when P < 0.01 for the tested hypothesis). d, Violin plot showing the relative gene age ranks (higher ranks equate to younger ages) of the TRGs between the autosomes and the U/V-homologue of D. dudresnayi. Statistically significant difference in mean values of gene ages (centre line) was assessed using an FDR-corrected two-sided permutation test.

Both monoicous species retained mostly male and a few female copies for most of the U/V-SDR-derived gametologues (91% in C. linearis and 100% in D. dudresnayi), whereas several U- and V-specific orthologues were lost in these species (10 and 17 sex-specific genes in C. linearis and D. dudresnayi, respectively). Of these lost orthologues, 60% and 43% present closely related autosomal paralogues in C. linearis and D. dudresnayi, respectively (Supplementary Tables 18 and 19), although it is unclear whether the expression of these autosomal paralogues is compensating the activity of the lost genes. The only three U-SDR-derived orthologues in C. linearis are flanked by PAR orthologues translocated at the end of the V-SDR-derived region, suggesting that the U/V-homologue (contig 12) of C. linearis acquired its U-SDR-derived genes through two translocations (Extended Data Fig. 9). Three U-SDR-derived genes in D. dudresnayi are dispersed across the V-SDR-derived region of the U/V-homologue, suggesting independent U-SDR translocations into the V-SDR, while the fourth U-SDR-derived gene was translocated to an autosome (Extended Data Fig. 9).

The seven ancestral V-SDR genes are transcriptionally active (log₂(TPM + 1) > 2) during reproductive stages of both monoicous species (Supplementary Table 20), emphasizing their role in reproduction despite the absence of a U/V system, particularly MIN²² which is retained in both species. While most U-SDR-derived genes are absent in monoicous species, a single intracellular cholesterol transporter gene was convergently preserved in both monoicous genomes (Supplementary Table 20) and actively expressed during fertility in both Ectocarpus sp. 7 and D. herbacea (Supplementary Table 8).

The U/V-homologue of D. dudresnayi retains some vestiges of its past as a U/V chromosome, such as low coding density, high repeat density and enrichment of TRGs (Wilcoxon rank-sum test, FDR-corrected P < 0.01), although we found non-significant differences in interspecies K_s values across the genome (Fig. 4c, Extended Data Fig. 7, Supplementary Fig. 16 and Supplementary Tables 9–14, 16 and 17).

Finally, we examined the transition from haploid to diploid sex determination, which has remained unstudied in eukaryotes. Although the ancestral state for the brown algae is a U/V sexual system, the Fucales recently transitioned to a diploid life cycle⁴⁴, with many species, such as F. serratus⁴⁵, exhibiting diploid separate sexes (dioecy)⁴⁶. Dioecy probably evolved from monoecy (both sexes in the same diploid individual) in the last common ancestor of the Fucus genus⁴⁷ (25–5 Ma¹⁸), consistent with a young sex chromosome in F. serratus. Our extensive bioinformatic analysis and PCR sex-linkage testing for candidate genes such as MIN (Methods) failed to identify sex-linked sequences in F. serratus (Supplementary Fig. 17), suggesting that the SDR is small and undifferentiated. However, male F. serratus conserves the MIN gene and all the ancestral V-SDR genes in its U/V-homologue (Figs. 2f and 5a). Importantly, although none of the U/V-SDR-derived genes are sex-linked in F. serratus, MIN and four other ancestral V-SDR genes are exclusively expressed in males (fully silenced in females) (fold change (FC) > 2, p_adj < 0.05; Fig. 5a and Supplementary Table 21). This pattern is consistent in three other Fucales species (Supplementary Table 22). Therefore, the ancestral V-SDR genes probably still play roles in male sex determination or differentiation pathways.

Fig. 5: Transition from haploid to diploid sex determination.

a, Expression of ancestral U/V-SDR genes in the diplontic species F. serratus. Gene expression of mature algae (using 3 males and 3 females, see Methods) is given as log₂(TPM + 1) and bars represent standard deviation of the mean. P values shown on the plot are derived from differential expression analysis performed using DESeq2, which applies a two-sided Wald test with Benjamini–Hochberg correction for multiple testing. Bold text represents whether the gene in F. serratus corresponds to an ancestral male or the female gametologue. b, Karyoplot of F. serratus showing the following features from bottom to top: chromosome compartment (autosomes and U/V-homologue), relative gene ages, interspecies K_s values (between 0.00079 and 6.838, with an average value of 0.148), proportion of coding (CDS, red) and repeat (TEs, blue) density. Statistically significant differences for each feature between each autosome and the U/V-homologue are depicted on top of the track for that autosome (FDR-corrected two-sided Wilcoxon rank-sum test; values indicated with solid colours when P < 0.01 for the tested hypothesis). c, Violin plot showing the relative gene age ranks (higher ranks equate to younger ages) of the TRGs between the autosomes and the U/V-homologue of F. serratus. The mean values of gene ages (centre line) are not significantly different (FDR-corrected two-sided permutation test).

Contrary to the observations in D. dudresnayi, the U/V-homologue of F. serratus lacks the TRG enrichment pattern and all the other distinctive features of the U/V chromosomes (Wilcoxon rank-sum test, FDR-corrected P > 0.01). Thus, this ‘ex’-sex chromosome has lost all the evolutionary vestiges of its past as a U/V chromosome (Fig. 5b, Extended Data Fig. 7, Supplementary Fig. 18 and Supplementary Tables 9–14, 16 and 17).

Discussion

Here we characterized the evolutionary trajectory of brown algal sex chromosomes (Fig. 6). Brown algal sex chromosomes date back 450–244 Ma²¹, at the origin of brown algae. We propose that the male-determining gene MIN underlies the birth of this U/V system²². The ancestral cassette with seven V-SDR genes suggests a very early evolution of these genes into a non-recombining locus during the evolution of the U/V-SDRs. The ancestral V-SDR genes probably contribute to reproduction, but may also be involved in broader developmental functions. Despite their old age, brown algal U/V chromosomes retain large PARs bordering the SDR, unlike haploid systems in non-vascular plants that mostly lack detectable PARs^11,12,13,14.

Fig. 6: Hypothetical model for U/V sex chromosome evolution.

U/V sex chromosomes arose from an ancestral autosome, via suppression of recombination that probably occurred via an inversion. The SDR boundaries expanded into neighbouring PAR via inversions, but also by recruitment of genes from autosomes; expansion occurred in a lineage-specific fashion, concomitant with increased sexual dimorphism of the different species. SDR genes are maintained within the SDR if they have roles in sex, whereas genes with no role in sex are lost. Faster substitution rates, probably a consequence of the heterochromatic context of the sex chromosome, may promote the rise of taxonomically restricted genes, which are selectively maintained on the sex chromosome if they have advantages to the sporophyte generation. In species that switch to a diploid life cycle, the U/V system disappears, but the genes that are in the V-specific region retain roles in sex, although they are no longer masters. Transition from U/V separate sexes to co-sexuality (monoicy) occurred when a male haploid individual acquired female-specific genes via translocations. During the demise of the U/V sex chromosomes, their structural and evolutionary footprints slowly disappear.

Brown algal genomes have a high degree of synteny conservation. The U/V-SDR is however prone to accumulate structural rearrangements, including inversions that may have caused the initial recombination suppression event in proto-sex chromosomes and the later expansion of the U/V-SDRs into the PARs. Similar to other haploid systems^48,49, TEs conspicuously accumulated in the SDRs following recombination suppression⁵⁰, possibly causing further rearrangements through TE-mediated inversions⁵¹.

Models of XX/XY and ZW/ZZ sex chromosome evolution suggest that sexually antagonistic selection may lead to SDR expansions, making former PAR genes fully sex linked and gaining new genes^52,53. We show that the three brown algal species with greater sexual dimorphism in gametes (oogamy) indeed have enlarged U/V-SDR gene contents, although statistical testing is not possible given the small sample size. Oogamy is proposed to be ancestral in the brown algae^16,21, but this trait seems to be highly labile and our analyses suggest multiple independent transitions to oogamy from a less dimorphic ancestor with a small ancestral SDR, accompanied by SDR gene gain. Anisogamous species only experienced few gene gains in their SDRs, as predicted in ref. ¹⁰. Similar to M. polymorpha, the relatively simple gene content of the U/V-SDRs in the brown algae could regulate an autosomal effector gene network controlling sexual dimorphism, including differences in somatic development between male and female gametophytes⁵⁴. Accordingly, we observed substantial sex-biased expression of autosomal genes between mature male and female gametophytes carrying sex organs. This contrasts with C. purpureus, where sex chromosomes carry thousands of genes and few sex-biased autosomal genes¹¹. Consistent with other studies^33,55, we found no correlation between levels of sex-biased expression and sexual dimorphism in brown algae. Sexual dimorphism in this lineage may be controlled by a relatively small subset of genes, and most sex-biased genes may influence gametophyte physiology or vegetative development.

In diploid sexual systems, recombination suppression between sex chromosomes explains Y and W degeneration⁵⁶ (but see refs. ^57,58,59). Differences in the evolution of U/V systems are predicted, since sex determination occurs in the haploid stage, in which deleterious mutations are more efficiently removed by selection than in diploid systems^29,60 and some of the predicted differences have been documented in bryophytes²⁹. In brown algae, as in diploid systems, the SDRs lost recombination, leading to TE accumulation and consequent reduction in gene density. However, unlike the degeneration documented in some XX/XY and ZW/ZZ systems, our ancestral state reconstruction revealed more gene gains than losses or gene movements out of the V-SDR during evolution of brown algal lineages. Unclassified repeats accumulate in the Ectocarpales sex chromosomes, including the PARs, but not in species with larger genomes, where LTR retroelements become dominant. DNA transposons are overrepresented among unclassified repeats⁶¹, and they often insert near the progenitor locus in a process called local hopping⁶². We propose that the U/V-SDR may thus act as a source of DNA transposons that hop to the PARs, thus increasing their repeat density, whereas increased colonization of LTR elements obscures this pattern in larger genomes.

Brown algal U/V chromosomes display an excess of TRGs. What mechanisms underlie this pattern? The SDR and the PARs of U/V chromosomes are enriched in heterochromatin⁶³, involved in repressing TEs⁶⁴, and heterochromatic regions tend to have higher mutation rates due to reduced access of the DNA repair machinery during replication⁶⁵. Accordingly, we consistently observe higher interspecies K_s values in U/Vs, particularly in the PARs, since these regions recombine between sexes and thus have a higher rate of neutral fixation than the SDR, which experiences the Hill–Robertson effect that reduces fixation probabilities of neutral mutations due to its linkage with other sites under selection⁶⁶. We speculate that this feature could facilitate the evolution of TRGs. Alternatively, the high density of DNA transposons within the U/V could also promote the co-option of their regulatory motifs and enable de novo transcript birth, as seen in Drosophila⁶⁷. Note that the pattern could be reinforced through generation-antagonistic selection³⁷, but DNA transposons and higher mutation rates may be sufficient to initiate this pattern in species lacking sporophyte-biased gene expression. Importantly, the TRG enrichment is unique to U/V systems and gradually disappears when these systems are lost. This pattern extends beyond brown algae to other eukaryotes with U/V systems, such as C. purpureus^11,68 and S. angustifolium¹². However, we could not detect enrichment of TRGs on the small U/V sex chromosomes of M. polymorpha^42,69, or on the mating-type chromosome of C. neoformans⁴³. Mosses such as C. purpureus and S. angustifolium display relatively more complex sporophyte body plans than liverworts such as M. polymorpha⁷⁰, which could underlie stronger generation-antagonistic selection³⁷. Unlike brown algae, most of the sex chromosomes in bryophytes are sex linked, with minimal space for PARs²⁹. In this context, the sex chromosome of C. purpureus expanded its SDR very recently in evolutionary time through fusions of autosomes with earlier established U and V chromosomes⁷¹, retaining its evolutionary footprints as PARs, while the sex-linked region in M. polymorpha is much older⁶⁹, which could also limit the formation of TRGs that are predominantly observed in the PARs of the brown algae. The pattern thus appears to be specific to U/V systems where chromosomal degeneration is mild and linked to haploid–diploid life cycles where the sporophyte stage is sufficiently complex, highlighting a key role for generation selection³⁷. Therefore, our study hints at a unique interplay between complex life cycles, heterochromatic landscape, DNA transposons and higher mutation rates that may lead to TRG enrichment in U/V chromosomes, and this process is pervasive across distant, independently evolved eukaryotic kingdoms.

Monoicous brown algae have transcriptomic profiles resembling ancestral females³³. However, our results show that monoicy arose at least twice from a male ancestor that acquired female genes. The male pathway requires MIN^22,25,72, which could have facilitated the evolution of monoicy from males, as also seen in the green lineage^73,74. We note the presence of U-SDR-derived gene(s) in all monoicous species, particularly a cholesterol transporter gene that is found in all U-SDRs, suggesting that the U-SDR contains femaleness-promoting factor(s), consistent with reports in kelps⁷⁵. We thus propose that monoicy evolved via translocation events adding essential femaleness-promoting genes to a male genetic background. Since the resulting monoicous individuals were capable of producing both male and female reproductive structures, individuals with U chromosomes were no longer essential for sexual reproduction, ultimately leading to the loss of the U-SDR in monoicous species. Although the combination of key female and male genes is essential for this evolutionary transition, the retention of a sex chromosome is not. For example, in Volvox africanus, monoicy required the retention of female SDR-like regions, while most male SDR genes were lost except for a multicopy array of the male-determining gene MID⁷⁶.

The evolution of a dioecious system in F. serratus is associated with an irreversible transition to diploidy in Fucales^16,47. The U/V to dioecy transition has remained elusive^2,47, but our data in brown algae imply that it involved an intermediate monoicous stage, supporting previous predictions from ancestral state reconstruction analyses¹⁶. A small, undifferentiated Y-specific region consistent with a young XY system may explain why the sex chromosome in F. serratus was undetectable. Nonetheless, all ancestral V-SDR genes are found in the U/V-homologue of F. serratus, several showing a male-biased gene expression across Fucales species, particularly MIN²². Our findings imply that MIN and possibly other ancestral V-SDR genes are still involved in male differentiation, but shifted downwards in the sex determination cascade. These results thus support and extend the ‘bottom-up’ hypothesis of sex determination, where downstream components of sex differentiation are conserved across taxa, and new master sex regulators can replace older ones⁷⁷.

Methods

Biological material

Scytosiphon promiscuus, Dictyota dichotoma, Undaria pinnatifida and Desmarestia dudresnayi haploid gametophytes were cultivated under laboratory conditions as in ref. ⁷⁸. We cultivated the gametophytes at 14 °C with a photoperiod of 12:12 h light:dark at an irradiance of 25 µmol photons m⁻² s⁻¹. The media consisted of filtered natural seawater (NSW), which was autoclaved and enriched with half-strength Provasoli nutrient solution (Provasoli-enriched seawater; PES)⁷⁸. We grew the first biomass in 140 mm Petri dishes and the gametophytes were later transferred to a 1 l flask with gentle aeration. The gametophytes were fragmented once a month and the media were changed every 2 weeks to promote biomass production. Before freezing, gametophytes were treated with antibiotics for 3 days with gentle agitation and under the same culture conditions. The first day, gametophytes were treated with a mix of streptomycin (2 g l⁻¹ of PES), penicillin G (0.5 g l⁻¹ of PES) and chloramphenicol (0.1 g l⁻¹ of PES); the next day with ampicilin (1 g l⁻¹ of PES), and on the last day with kanamycin (1 g l⁻¹ of PES). Between each day of treatment and before freezing, gametophytes were rinsed with 500 ml of NSW to remove traces of antibiotic.

Samples of fucoid algae sexual and vegetative tissue were collected in the intertidal zone during low tides in June 2012 from Viana do Castelo (F. vesiculosus, A. nodosum) and Caminha (Rio Minho; F. ceranoides), northern Portugal. Sexual phenotypes were verified in the field by sectioning and observing receptacles under a field microscope. Tissue samples were flash frozen in liquid nitrogen on the shore and transported to the laboratory in a cryoshipper, after which they were lyophilised and stored dry at room temperature on silica crystals (see Supplementary Table 1 for a list of strains used in this study).

DNA and RNA extraction and sequencing

Genomic DNA was isolated from algal tissue (~100 mg) by grinding into fine powder under liquid nitrogen and subsequent cell lysis in 500 μl of Genomic Lysis Buffer (OMNIPREP for plant kit) for 1 h at 60 °C. The lysate was cleaned up with 200 μl of chloroform and DNA was precipitated in ethanol. The DNA pellet was digested in CF buffer (Macherey–Nagel) for 45 min at 65 °C and purified using NucleoBond AXG20 Mini columns according to the user manual (Macherey–Nagel). Final high molecular weight genomic DNA was quantified (Qubit), analysed for purity (Nanodrop) and checked for size distribution (Femto Pulse System) before preparing the sequencing libraries. We sequenced the libraries using an Oxford Nanopore Technologies (ONT) MinION Mk1B. We prepared the ONT libraries using an SQK-LSK110 library preparation kit for R9.4.1 flow cells and an SQK-LSK114 library preparation kit for R10.4.1 flow cells. Two libraries were sequenced for D. dudresnayi on R9.4.1 flow cells and a third library was sequenced on an R10.4.1 flow cell.

RNA was isolated from mature gametophytes of U. pinnatifida and S. promiscuus following modified procedure of the Qiagen RNAeasy kit, and the TruSeq RNA Library Prep Kit v.2 was used to sequence the transcriptomes in an Illumina NextSeq 2000 platform (150 bp, PE reads). Extraction of total RNA from fucoid algae (F. vesiculosus, A. nodosum and F. ceranoides) was performed following ref. ⁷⁹ and RNA libraries were sequenced on an Illumina HiSeq 2000 machine (100 bp, PE reads).

Genome assembly and annotation

High-quality, chromosome-level assemblies of brown algal genomes have been notoriously difficult to obtain due to technical challenges in extracting nucleic acids. Whole-genome assemblies and annotations of S. promiscuus male, D. dichotoma male, D. herbacea male and female, E. crouanorium male, C. linearis, S. ischiensis and F. serratus male were obtained from ref. ¹⁸. We also downloaded the genome of Ectocarpus sp. 7 (ref. ¹⁹) and the male genome of U. pinnatifida²⁰, which were already assembled at a chromosome level. For D. dudresnayi, we performed genome sequencing, de novo genome assembly and ab initio gene annotation. Base calling was done using ONT Guppy⁸⁰ with the configuration files ‘dna_r9.4.1_450bps_sup.cfg’ and ‘dna_r10.4.1_e8.2_400bps_sup.cfg’ and the options ‘–trim_adapters –trim_primers’, yielding 17.4 Gbp of data in 2,871,152 reads. We merged all the reads and analysed them using Kraken (v.2.1.2)⁸¹ and the bacteria database (August 2022) to remove potential contaminant sequences. All data classified as bacterial reads by Kraken were screened using blastN (v.2.13.0+)⁸² (-evalue 0.001 -num_alignments 20) against the NCBI genbank bacterial database (downloaded November 2023). The blastN output was visualized in MEGAN (v.6.23.4)⁸³, and all reads that were declared as bacterial were extracted and removed from further analyses. We obtained 1,908,772 decontaminated reads with an average length of 5.1 Kbp (9.8 Gbp of data, 20× coverage), which were deposited on the NCBI Sequence Read Archive (Supplementary Table 1).

The decontaminated reads were assembled de novo using flye (v.2.9.1-b1780)⁸⁴ with the options ‘–nano-raw -g 450 m -t 28 -i 3 –scaffold’. The draft assembly consisted of 1,032 contigs with a total size of 425 Mbp, an N50 of 4.6 Mbp and an L50 of 29 contigs. We used TransposonPSI (http://transposonpsi.sourceforge.net/) to predict the TEs and RepeatScout (v.1.0.6)⁸⁵ to predict the simple repeats in the genome assembly. Both predictions were combined to soft mask the repetitive content in the genome assembly using bedtools maskfasta (v.2.27.1)⁸⁶. We mapped the RNA-seq data of D. dudresnayi from the PhaeoExplorer database¹⁸ to the soft-masked genome assembly using STAR (v.2.7.1a)⁸⁷. We used BRAKER v.2.1.6 alongside the RNA-seq data⁸⁸ to predict the protein-coding genes in the soft-masked genome assembly.

Hi-C library preparation and sequencing for chromosome-level assemblies

We generated Hi-C libraries for three male genomes (S. promiscuus, D. herbacea and D. dichotoma) and two female genomes (Ectocarpus sp. 7 and D. herbacea). Fresh algal tissue was cross-linked for 20 min at room temperature in a solution of 2% formaldehyde with filtered NSW and then transferred into a 400 mM glycine solution with filtered NSW for 5 min to quench the formaldehyde. The samples were then stored at −80 °C until use. The Hi-C libraries were prepared as follows. The samples were de-frosted in 1 ml of 1× DpnII buffer with protease inhibitors (Roche cOmplete), transferred to Precellys VK05 lysis tubes (Bertin) and disrupted using the Precellys apparatus with five grinding cycles of 30 s at 7,800 r.p.m., followed by 20 s pauses. SDS was added to the lysate at 0.5% final concentration and samples were incubated for 10 min at 62 °C, followed by the addition of Triton X-100 to a final concentration of 1% and 10 min of incubation at 37 °C under gentle shaking. We added 500 U of DpnII to 4.6 ml of the digestion mixture and incubated the samples for 2 h at 37 °C under gentle shaking (180 r.p.m. in an inclined rack to prevent sedimentation), followed by the addition of another 500 U of DpnII and an overnight incubation under the same conditions. The digested samples were centrifuged at 4 °C for 20 min at 16,000 × g. The supernatant was discarded and the pellet was incubated for biotinylation at 37 °C for 1 h under a constant shaking (300 r.p.m.) in a 500 ml biotinylation mix with a concentration of 1× ligation buffer, 0.09 mM dATP-dGTP-dTTP, 0.03 mM biotin-14-dCTP and 0.64 U ml⁻¹ Klenow fragments. After biotinylation, the samples were incubated for 3 h at room temperature in a 1.2 ml ligation reaction with a concentration of 1× ligation buffer, 100 mg ml⁻¹ BSA, 1 mM ATP and 0.4 U ml⁻¹ T4 DNA Ligase. The samples were then incubated overnight at 65 °C after adding 20 μl 0.5 M EDTA, 80 μl 10% SDS and 1.6 mg Proteinase K. DNA was extracted with 1 volume of phenol/cholorform/isoamyl alcohol (24:24:1), followed by 30 s of vortexing at top speed and a 5-min centrifugation at top speed. We precipitated the DNA by adding 1/10 volume of 3 M NaAC pH 5 and two volumes of 100% cold ethanol, followed by a 30-min incubation at −80 °C and a 20-min centrifugation at 14,000 × g and 4 °C. The DNA pellet was washed with 1 ml 70% ethanol, then dried at 37 °C for 10 min and resuspended in 100 μl 1× TE buffer with 1 mg ml⁻¹ RNase. DNA was sheared to 250–500 bp fragments using a Covaris S220 ultrasonicator, purified with AMPure beads (0.6×) (Beckman) and eluted in 20 μl 10 mM Tris pH 8.0. Biotinylated but not ligated DNA fragments were first removed by T4 DNA polymerase treatment (final concentration, 300 U per pellet; NEB), and the biotin-labelled fragments were selectively captured by Dynabeads MyOne Streptavidin C1 beads (Invitrogen). The libraries were prepared using the NEB Ultra II library preparation system and sequenced on the NextSeq 2000 Illumina platform (2 × 150 bp) (Supplementary Table 1).

We scaffolded the genomes from ref. ¹⁸ into chromosome-level assemblies using the Hi-C data. We filtered the low-quality Hi-C reads using Trimmomatic (v.0.39)⁸⁹ (ILLUMINACLIP:2:30:10 LEADING:25 TRAILING:25 SLIDINGWINDOW:4:15 MINLEN:75 AVGQUAL:28). We mapped the Hi-C reads against each genome assembly using BWA-mem v.0.7.17-r1188d in the Juicer v.1.6 pipeline⁹⁰ to generate a contact map, which was then fed to 3D-DNA v190716 (ref. ⁹¹) to scaffold the genomes into chromosomes. The obtained scaffolds were manually inspected against the contact maps to solve the limits of each chromosome using Juicebox (v.1.11.08)⁹². The PhaeoExplorer gene annotations¹⁸ were lifted into the new assemblies using Liftoff (v.1.6.1)⁹³, while the annotation of TEs was performed using RepeatModeler2 (ref. ⁹⁴). We scaffolded the genomes of E. crouaniorum and D. dudresnayi into chromosomes using a reference-guided assembly with RagTag (v.2.0.1)⁹⁵ against the chromosome-level assemblies of Ectocarpus sp. 7 and D. herbacea, respectively. All genes within the SDRs in the brown algal species studied (see below) were manually curated to exclude any TE-related genes from the annotation.

Discovery of the U/V sex determination regions

Male sex-determining regions (V-SDR) in S. promiscuus, U. pinnatifida, D. herbacea and D. dichotoma, as well as female sex-determining region (U-SDR) in D. herbacea were analysed following two complementary methods: (1) a k-mer-based YGS approach, originally designed to detected Y-linked sequences in heterogametic systems, developed in ref. ⁹⁶ and (2) genomic coverage analysis, designed to identify sex-linked regions through differences in read depth between male and female individuals⁹⁷. These methods are well suited for organisms with divergent sex chromosomes, such as brown algae, where U and V haplotypes have diverged over extended evolutionary time.

The YGS method principle is to identify male or female sex-linked scaffolds by comparing k-mer frequencies between reference genome assembly and k-mers generated from DNA-seq reads of the opposite sex. Regions in the male reference genome that contain k-mers that are absent in female reads will indicate candidate male SDR sequences; similarly, female genomic scaffolds with low coverage in male k-mers will denote female SDR region. For each species, 15-base-pair k-mer sequences were generated separately from male and female Illumina reads (see Supplementary Table 1 for data accession numbers) using Jellyfish v.2.3.0 count (-m 15 -s 10 G -C –quality-start=33 –min-quality=20) and converted to fasta format with Jellyfish dump (–lower-count=5)⁹⁸. Next, non-overlapping 500-kb sliding windows (Desmarestia, Dictyota and Undaria) or 200-kb sliding windows (Scytosiphon) of the reference chromosome genomes (from the sex whose SDR was to be identified) were created using seqkit (v.2.3.1)⁹⁹ and used as input for the YGS.pl script⁹⁶ together with the fasta k-mer files produced in the previous step. Each window was then analysed to calculate the proportion of k-mers in the reference window that are not present in the opposite-sex k-mer database. Genomic windows with a minimum of ≥50% of unmatched single-copy k-mers were then retained as candidate male or female SDR sequences. Because the borders of the SDRs cannot be precisely defined at the single-nucleotide level with the available data, we focused on genes within these regions and defined the SDR boundaries on the basis of the flanking genes located at the transition to pseudoautosomal regions (PARs).

Candidate SDR regions identified by YGS were further validated by analysing sex-specific differences in read coverage. In detail, the short Illumina reads coming from males and females of each investigated species were trimmed with Trimmomatic⁸⁹ (see above) and mapped to the reference genome for which the SDR was to be studied, using HISAT2 (ref. ¹⁰⁰) (default settings). Bam files produced by HISAT2 were used as input for Mosdepth¹⁰¹ to calculate coverage in 10-kb windows along the genome sequence (-m -n -b 10000 –fast-mode -Q 30). Read mapping depth in genomic windows was normalized by the genome-wide mean for each sex, and the coverage in genomic intervals was then compared between males and females. Because V-SDR-linked sequences are present only in males, we expect them to have similar read coverage as autosomal regions in males, but little or no coverage in females (and conversely for the U-SDR sequences in D. herbacea). The comparison focused on regions within male reference genomes where the coverage in males fell within the range of 75–125% of the genome average, while the coverage in females remained below 50% of the genome average.

Both coverage and k-mer analysis identified identical genomic regions, providing high-confidence candidate SDRs (Extended Data Table 1 and Supplementary Figs. 1–5). In D. herbacea, where both male and female chromosome-level genome assemblies were available, we directly compared U and V chromosomes to further confirm the SDR borders by analysing the collinearity of pseudoautosomal regions flanking the SDRs. The SDR scaffolds for all studied species were further validated by PCR amplification (Supplementary Table 1) using 4 males and 4 females.

Genetic mapping and search for the sex chromosome in F. serratus

Three different sets of materials were used in this study: (1) 12 male and 12 female field samples, hereafter denoted the 24-individual natural population; (2) 157 sporophyte progeny population derived from a cross between one male sample and one female sample collected from the field and (3) 3 male and 3 female samples collected from the field for whole-genome sequencing. The 157-progeny population and 24-individual natural population were genotyped using the double digest RAD sequencing approach (ddRAD-seq). Briefly, individual genomic DNA was digested with the restricted enzymes PstI and HhaI to obtain fragments that were size selected between 400 and 800 bp before sequencing on in Illumina HiSeq 2500 platform (paired-end 2 × 125 bp). See ref. ¹⁰² for detailed protocol of the ddRAD-seq.

We performed whole-genome sequencing on an Illumina HiSeq 2500 system (2 × 150 bp paired-end) for the 3 male and 3 female samples. For ddRAD-seq data, raw reads were cleaned and trimmed with Trimmomatic as above and mapped to the draft genome of F. serratus male. For the progeny population, genotypes were called from the obtained bam files using the Stacks pipeline (v.2.5)¹⁰³. The obtained vcf files were filtered with VCFtools (v.0.1.16)¹⁰⁴ and bcftools¹⁰⁵ (max missing per locus: 30%, max missing per sample: 40%, max mean coverage:30, minQG:20).

The filtered vcf file of the progeny population was used to construct a genetic map with Lep-MAP3 (ref. ¹⁰⁶). Briefly, the ParentCall2 module was used to call parental genotypes, the SeparateChromosomes2 module was used to split the markers into linkage groups and the OrderMarkers2 module was used to order the markers within each linkage group using 30 iterations per group and finally computing genetic distances. Phased data were converted to informative genotypes with the script map2genotypes.awk.

We used different approaches to identify the SDR in F. serratus

Coverage analysis

We combined whole-genome sequence data from the 3 males and 3 females alongside the ddRAD-seq data of the 24-individual natural population, mapping both datasets to the F. serratus male genome assembly using bwa-mem¹⁰⁷. Coverage analyses was done in several ways:

• Using SATC (sex assignment through coverage)¹⁰⁸, a method that uses sequencing depth distribution across scaffolds to jointly identify: (1) male and female individuals and (2) sex-linked scaffolds. This identification was achieved by projecting the scaffold depths into a low-dimensional space using principal component analysis and subsequent Gaussian mixture clustering. Male and female whole-genome sequences were used for this analysis.

• Using the method SexChrCov described in ref. ¹⁰⁹ with the 24-individual natural population.

• Using the method DifCover¹¹⁰ which identifies regions in a reference genome for which the read coverage of one sample is significantly different from the read coverage of another sample when aligned to a common reference genome. The 24-individual natural population was used for this analysis.

• Using soap.coverage (v.2.7.9)¹¹¹ to calculate the coverage (number of times each site was sequenced divided by the total number of sequenced sites) of each scaffold in each sample. For each scaffold, the male to female (M:F) fold change coverage was calculated as log₂(average male coverage) – log₂(average female coverage). The 24-individual natural population was used for this analysis.

Fixation index (F_ST) and sex-biased heterozygosity

This approach has been previously used to find sex-linked genomic regions in several studies^112,113. Using the 24-individual natural population, F_ST was calculated using vcftools¹⁰⁴. Sex-biased heterozygosity was defined as the log₁₀ of the male heterozygosity:female heterozygosity ratio, where heterozygosity was measured as the fraction of sites that are heterozygous. This ratio is expected to be zero for autosomal scaffolds and elevated on young sex scaffolds due to excess heterozygosity in males.

Identification of eventual female scaffolds that failed to map to the male reference genome

Vcftools and bedtools were used to extract female regions that did not map to the reference genome consistently in the 3 resequenced female samples.

All candidate contigs were tested by PCR in 4 males and 4 females.

Synteny analyses, K _s analysis and transitions to co-sexuality

Whole-genome synteny comparisons were performed for each pair of chromosome-level assemblies using MCscan (v.1.2.14)¹¹⁴, both between different species, between sex chromosomes in the same species and between monoicous species and their closest relatives with U/V chromosomes. The putative gametologues between sex chromosomes that were predicted with MCscan were reassessed using OrthoFinder (v.2.5.4)¹¹⁵ and best reciprocal DIAMOND (v.2.1.8.162)¹¹⁶ hits.

We calculated the number of synonymous substitutions per synonymous site (K_s) for each pair of male and female gametologues as a proxy to assess the relative time at which both genes diverged from each other. The amino acid sequences of each pair of gametologues were aligned with MAFFT (v.7.520)¹¹⁷ and subsequently aligned into codons using pal2nal (v.14)¹¹⁸. The gametologue K_s values were calculated using the model in ref. ¹¹⁹ as implemented in KaKs_calculator (v.2.0)¹²⁰.

We evaluated the male or female identity of the genes in the monoicous species whose orthologues were found within the SDR in their closest dioicous relatives. For this, we compared the results obtained with MCscan¹¹⁴ against the orthogroup prediction performed with OrthoFinder¹¹⁵, with best reciprocal DIAMOND¹¹⁶ hits and by calculating gene trees for each orthogroup using an amino acid alignment with MAFFT¹¹⁷ and gene tree reconstructions using FastTree (v.2.1.11)¹²¹.

Ancestral reconstruction of the male SDR

The brown algal phylogeny was obtained from ref. ¹⁸. The species tree is based on 32 single-copy nuclear genes whose protein sequences were aligned manually using AliView¹²², and whose best-fit substitution models were assessed independently using the Akaike information criterion. The tree was generated using a maximum likelihood approach implemented in RAxML bootstraps and the gamma model. Every node in the phylogeny has 99–100% bootstrap support values. Divergence times were subsequently calculated using MCMCtree¹²³ and three calibration points. The MCMC chains were run for 1.5 million generations and the first 200,000 MCMC chains were discarded as burn-in.

We searched for orthologue genes within the V-SDR of five species (Ectocarpus sp. 7, S. promiscuus, U. pinnatifida, D. herbacea and D. dichotoma) in our OrthoFinder results. For each V-SDR gene, we coded its orthologue in the other species as ‘present’ (1) if it is also sex linked in the V-SDR, whereas it was coded as ‘absent’ (0) if the orthologue resides in the PARs, in an autosome or if there is no detectable orthologue in that species. Once we generated this presence/absence matrix with the evolutionary relationship of the genes within the V-SDR (Supplementary Table 5), we used it as the input file for the software Count (v.10.04)¹²⁴ to estimate the ancestral content of the V-SDR throughout a phylogeny and determine the most likely scenario of V-SDR evolution in the brown algae. We employed posterior probabilities under a phylogenetic birth-and-death model with independent gain and loss rates across each branch in the phylogeny. We modelled the independent gain and loss rates through 10 gamma categories and performed 1,000 optimization rounds with a convergence threshold on the likelihood >0.1 to find the best fitting model for the data. The branch lengths in the tree that were used for the ancestral state reconstruction were retrieved from the molecular clock analysis performed in ref. ¹⁶. We distinguished between conserved V-SDR genes that are ancestral and parallel acquisitions of the same gene in the V-SDR by analysing gene trees between male and female genomes, in addition to female transcriptome assemblies of D. dichotoma and U. pinnatifida. Sequence alignments were done using MAFFT¹¹⁷ with default settings and uploaded to the https://www.phylogeny.fr/ platform. Alignments were further curated using Gblocks (v.0.91b)¹²⁵ (min. seq. for flank pos.: 85%, max. contig. nonconserved pos.: 8, min. block length: 10). Trees were produced using PhyML (v.3.1l)¹²⁶ with the default model and visualized in TreeDyn (v.198.3)¹²⁷. The approximate likelihood-ratio test was chosen as the statistical test for branch support. We inferred the function of the ancestral V-SDR genes through the annotation of genes in Ectocarpus sp. 7 belonging to that orthogroup. The most likely acquisition mechanism of each SDR gene in each species was assessed on the basis of the position of each orthologue in the other species (pseudoautosomic, autosomic or missing; Supplementary Table 5).

Genomic content across chromosomes

We used closely related genome assemblies available in the PhaeoExplorer database¹⁸ to assess the depletion of orthologues in the sex chromosome. We predicted one-to-one orthologues using OrthoFinder¹¹⁵ between the following species pairs: Ectocarpus sp. 7 with Ectocarpus siliculosus, S. promiscuus with C. linearis, U. pinnatifida with Saccharina japonica, F. serratus with Fucus distichus, D. herbacea with D. dudresnayi, and D. dichotoma with Halopteris paniculata (Supplementary Table 16). We calculated the expected number of detectable orthologues for each chromosome and compared it against the observed number of detected orthologues using chi-square tests. We performed Benjamini–Hochberg corrections to the P values of the chi-square tests to control the FDR in the analysis¹²⁸.

GenEra³⁶ was used by running DIAMOND in ultra-sensitive mode¹¹⁶ against the NCBI NR database and all the PhaeoExplorer proteins¹⁸ to perform a phylostratigraphic analysis (e-value threshold of 10⁻⁵) and calculate the relative ages of each gene in each genome (Supplementary Table 13). Phylostratigraphy is a genetic statistical method developed to date the putative origin of all the genes contained in the genome of a target species by detecting homologues across species at different evolutionary distances (all the way from species within the same genus to species from different domains of life). Finding the most distant homologues of each gene can link them to their founder events (that is, the first instance where a gene homologue is found in the history of that lineage), allowing us to then determine their relative ages, coded as the taxonomic group where that gene is detected^35,36,129. The gene age categories outside of the brown algae and S. ischiensis were based on the taxonomic classification of each species within the NCBI Taxonomy database¹³⁰, while the gene ages within the brown algae were manually assessed to reflect the evolutionary relationships obtained in the PhaeoExplorer maximum likelihood tree¹⁸. We performed Wilcoxon rank-sum tests in R (v.4.3.1)¹³¹ to assess non-random differences in gene age distributions between pairs of chromosomes (Supplementary Table 14). We performed Benjamini–Hochberg corrections to the P values of the Wilcoxon rank-sum tests to control the FDR in the analysis¹²⁸. The gene ages responsible for these differences were found by evaluating the standardized residuals using mosaic plots.

We used the interspecies K_s values between pairs of species as a proxy for neutral mutation rates across six of the seven chromosome-level assemblies by using the most closely related genome assemblies available in the PhaeoExplorer database¹⁸. We used the same set of one-to-one orthologues detected between species pairs as for the orthologue-depletion test (Supplementary Table 16). However, the evolutionary distance between D. dichotoma and H. paniculata prevented us from calculating reliable interspecies K_s values for this species since synonymous substitutions reached the point of saturation. The amino acid sequences of each pair of orthologues were aligned with MAFFT¹¹⁷ and subsequently aligned into codons using pal2nal¹¹⁸. The interspecies K_s values were calculated using the model in ref. ¹¹⁹ as implemented in KaKs_calculator (v.2.0)¹²⁰. We also evaluated the difference in interspecies K_s values between the autosomes and the sex chromosomes through FDR-corrected Wilcoxon rank-sum tests (Supplementary Table 17). We calculated the protein-coding density, the density of TEs and the taxonomic identity of these TEs within 100-kb non-overlaping windows across each chromosome using bedtools⁸⁶ (Supplementary Table 9). The differences in protein-coding space and repeat content between the autosomes and the sex chromosomes were also tested using FDR-corrected Wilcoxon rank-sum tests (Supplementary Tables 10 and 11). The differences in repeat density, percentage of unclassified repeats and gene ages across genomic compartments (SDR, PARs and autosomes) were tested using FDR-corrected permutation tests with 10,000 permutations. All genomic features were plotted using karyoploteR (v.1.20.3)¹³².

Gene expression analysis

We used kallisto (v.0.44.0)¹³³ to calculate gene expression levels using 31-base-pair-long k-mers and 1,000 bootstraps. Transcript abundances were then summed within genes using the tximport v.3.19 package¹³⁴ to obtain the expression level for each gene in transcripts per million (TPM). Differential expression analysis was done in the DESeq2 v.3.19 package¹³⁵ in R v.4.3.1, applying FC ≥ 2 and p_adj < 0.05 cut-offs. Sex-biased gene expression analysis in Ectocarpus sp. 7, S. promiscuus, U. pinnatifida, D. herbacea and D. dichotoma was performed between mature male and female gametophytes (gametophytes bearing reproductive structures). To discover genes with sporophyte-biased expression in Ectocarpus sp.7, S. promiscuus, U.pinnatifida and D. dichotoma, we first calculated the differential expression between male gametophytes and sporophytes, as well as female gametophytes and sporophytes. Genes that showed significant sporophyte-biased expression (FC ≥ 2, p_adj < 0.05) in both comparisons were considered sporophyte biased.

A total of 314.2 M RNA-seq reads from F. vesiculosus male, female and vegetative tissue were assembled de novo with rnaSPAdes¹³⁶ using k-mer values of 33 and 49. Assembly quality was assessed by (pseudo)mapping reads back onto the resulting assembly and retaining ‘good’ contigs as defined using TransRate (v.1.0.3)¹³⁷ with default settings. The resulting 159,108 contigs were aligned with BLASTx⁸² against a database of Stramenopile proteins, and those with top hits against brown algae (Phaeophyceae) were retained as the final curated reference transcriptome (36,394 contigs, N50 = 1,770 bp). Transcript expression levels were determined by mapping the reads from all samples against the reference transcriptome using Bowtie2 (ref. ¹³⁸) and the RSEM-EBSeq (v.1.3.3)¹³⁹ pipeline, and relative expression values were recorded as TPM. All samples used in the gene expression analysis can be found in Supplementary Table 1.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.