A total of 20 adult reproductive F. vesiculosus (10 × males and 10 × females) per sampling date were collected in 2008 at Praia Norte, Viana do Castelo, Portugal (41°41′59″N; 8°51′19″W). Sampling was carried out beginning at sunrise on Sep 30 and Oct 9, corresponding to spring (S) and neap (N) tide phases, respectively. Sampling was performed in this way to increase transcript coverage during the semilunar reproductive cycles of gamete maturation and release. A receptacle (reproductive tissue; see Additional file 1) was taken from each individual and transverse sections were examined under a field microscope (40 × magnification) to identify and confirm the sexual phenotype. The remaining mature receptacles (separate pools of males and females) were briefly washed in seawater, wiped to remove surface epiphytes, and then flash-frozen in liquid nitrogen for transport to the laboratory. Vegetative tips (pooled from male and female algae) were treated in the same way. In the laboratory, tissues were lyophilized prior to RNA extraction following 43 . RNA was digested with RNase-free DNase (QIAGEN) for 15 minutes at room temperature and then purified with the RNeasy MIDI kit (ca. 1 mg total RNA; Qiagen). RNA concentration was estimated by spectrophotometry (GeneQuant, GE Healthcare); integrity was confirmed by running samples on a 1.2% agarose gel.

Poly-A mRNA was isolated from total RNA (ca. 1 mg) using the Oligotex mRNA Midi kit (Qiagen). Double stranded cDNA was constructed using the SuperScript^® One-Cycle cDNA kit (Invitrogen), following the manufacturer’s instructions. Four independent reactions were carried out for each of the six library samples: two using poly dT priming primers (Oligo d(T)₂₅ VN) and two reactions using random primers (N₁₅). Double-stranded cDNA syntheses were purified using CyScribe GFX Purification Kit (GE Healthcare). A fluorometer was used to estimate DNA concentration (Picofluor, Turner Biosystems). The resulting cDNA (2–3 ug) was adjusted to 50 ng/ul for 454 pyrosequencing at the Max Planck Institute for Molecular Genetics, Berlin (GS FLX Titanium, Life Sciences, Roche).

Sequence quality assessment and trimming were performed using PRINSEQ 44 to remove short (≤ 50 bp) and low quality sequences (average phred score ≤ 20), and tail regions (phred ≤ 20, 5-base sliding window). Sequences from all six libraries that passed this step were combined to produce a single assembly using MIRA v. 3.0 45 . Local BLASTN searches (E ≤ 10^–10) were performed against the Silva rRNA database (LSU and SSU parc, release 108) to identify rRNA. rRNA contigs and reads were filtered with a custom BioPython script. The remaining ESTs (contigs and singletons) were compared against the NCBI non-redundant protein database (nr) using the BLASTX algorithm (E-value ≤ 10^–4). Putative protein sequences were extracted from BLASTX output, and compared against public protein databases (KEGG, Pfam) for further functional annotation. We used the tools and resources available at the CAMERA website (https://portal.camera.calit2.net/gridsphere/gridsphere), utilizing top hits in downstream analyses. A local database (MySQL) containing normalized reads (accounting for library size differences) and annotation information was built for searches and analyses. Additional file 2 provides a schematic outline of the assembly and analysis workflow. The original sequencing data (passing quality control) are available at the NCBI Sequence Read Archive (SRA), accession number SRR575725.

All orthology terms (1,207) identified from BLASTX against the KEGG protein database were mapped onto KEGG pathways using the tools available at http://www.genome.jp/kegg/ko.html. Numbers of normalized reads/library for each pathway were extracted from the local database; non-biologically meaningful pathways (e.g. human disease or other organism-specific pathways) were removed, and the data filtered for minimum KEGG terms/pathway (threshold ≥ 5) and total reads/pathway (≥ 200). This reduced the total from over 150 to a core set of 40. Library expression values (read counts) were compared using Fisher’s exact test using IDEG6 software (Identification of Differentially Expressed Genes; 46 ) with Benjamini-Hochberg correction (α = 0.05; 47 ): FS versus VS, MS versus VS, and MN versus VN. For graphical representation (MeV 4.8.1; 48 ) data were expressed as Log₂ (read counts Library₁ – read counts Library₂).

Contigs of F. vesiculosus were compared against the Ectocarpus proteome using BLASTX with an E-value cutoff of ≤ 10^-10, and contigs matching the same Ectocarpus accession were combined to define an orthologous set of F. vesiculosus proteins (and hereafter referred to as genes). Differential expression (read counts/orthologue for each library, with a minimum cut-off of 20 reads/orthologue) between male, female and vegetative tissue was tested with Fisher’s exact tests and Benjamini-Hochberg correction, and displayed graphically as Log₂ expression values. Distributions of expression values for vegetative-, male- and female-biased genes were compared using non-parametric Kolmogorov-Smirnov tests.

Material for quantitative real-time PCR (qPCR) analysis was collected at Praia Norte, Viana do Castelo, Portugal. Sampling was carried out on rising tides in July 2011 during spring and neap tidal phases (as for the EST libraries), and an additional sample was taken two days after the gamete release peak, as determined in the field (see Figure 7c in 30 ). Female and male individuals were sexed by examining sectioned receptacles under the microscope, and tagged in the field prior to sampling. Individual receptacles (female and male) were collected, while apical tips (vegetative tissue) were pooled; tissue was fast frozen in liquid nitrogen and stored at -80° before further analysis. qPCR analyses were performed according to previous work on F. vesiculosus 49 . Briefly, total RNA was extracted from lyophilized tissue from four individual female and four individual male receptacles; for non-reproductive tissue (Veg) pools of apical tips of at least four individuals (including both female and male material) were used. RNA concentration was estimated by spectrophotometry and integrity was confirmed by running samples on a 1.2% denaturing agarose gel. First strand synthesis was performed using SuperScript III (Invitrogen) from 1 ug of total RNA in two parallel 20-μl reactions, which were then pooled. cDNAs quality were tested under standard PCR conditions (amplification conditions: 95°C for 2 min and then 35 cycles at 95°C for 30 s, 65°C for 30 s and 68°C for 20 s) and the product electrophoresed in a 2% agarose gel.

Primers for candidate genes were designed using the Primer3 web application (http://frodo.wi.mit.edu/), with a Tm of 68–70°C and an amplicon size between 100 and 250 bp. From an initial screening panel, eight candidate genes were selected for further qPCR analysis. Selection was carried out based on amplification quality and transcript expression pattern in the 454 dataset. qPCR was performed in a 96 well format, with triplicate amplifications and SYBR-green-based detection (Maxima SYBR Green/ROX, Fermentas in the Applied Biosystem step-one system). Reactions contained a 1:100 dilution of cDNA template and 0.5 μM of primers in a total volume of 20 μl. Cycle parameters were 95°C for 2 min and then 45 cycles at 95°C for 10 s and 68°C for 30 s. The amplification efficiency of each primer pair was calculated from dilution curves, using pooled cDNA from all treatment conditions as template. Relative expression was analysed with Applied Biosystems StepOne, using the ΔΔCT method corrected for amplification efficiency 50 and following 51 for applying multiple reference (housekeeping) genes: EF1d/b (elongation factor-1 B beta) and SUMO3 (small ubiquitin-like modifier). A 1:100 dilution of the pooled cDNA from all treatment conditions was used as the internal reference.

After adaptor trimming, quality assessment and removal of rRNA, 309,281 high quality ESTs were assembled into 41,918 contigs and 3,719 singletons (Table 1). The total number of contigs with a BLASTX hit E-value ≤ 10^-4 was 17,170 (44.7%) (Table 1). Annotation success by BLASTX was aided by the availability of the genome sequence for the brown alga E. siliculosus, which was the top hit species (7,591 contigs = 16.6%, corresponding to 3,853 unique protein accessions). Second, with ca. two orders of magnitude fewer accessions was the Leguminosae Medicago truncatula (1,136 contigs with 39 unique accessions), followed by the diatom Thalassiosira oceanica (850 contigs with 43 unique accessions) and F. vesiculosus (385 contigs with 91 unique accessions). A preliminary analysis of expression values amongst the six libraries indicated potential contamination of the female neap (FN) library by male tissue. After confirmation by qPCR tests (data not shown) the FN library was not considered further, reducing the assembly to 241,870 reads in 38,415 contigs and 2,761 singletons. Differences in expression between spring and neap tidal samples for male and vegetative tissue were low (35 and 25 genes, respectively: Additional file 3). Kolmogorov-Smirnov tests indicated that the distributions of significant DE terms did not differ between these tissue types (data not shown). Further analysis was therefore confined to the spring tide libraries.

F female, M male and V vegetative tissue; N neap tide, S spring tide samples.

*BLASTX; E ≤ 10^–4.

**Possible contamination with male tissue – not analysed further.

A relatively small subset of contigs were expressed in all tissues (1,809, 16%), but accounted for 64.5% of reads (Figure 1a). Many small contigs were assigned uniquely to individual tissues. Pairwise comparisons indicated that male and female tissues were the most differentiated, sharing the fewest contigs and reads (Figure 1a). At the functional level, a core set of 558 Kegg terms (>50%) were common to all tissues, accounting for 94% of annotated reads. Male tissue showed the greatest number of unique Kegg orthology terms (13%) and corresponding numbers of reads (Figure 1b).

A total of 1,084 KEGG orthology terms were identified in the dataset (FS, MS, VS; see Additional file 4 for the list of KEGG terms and Additional file 5 for the list of Pfams identified), and were mapped onto 123 pathways (KEGG ko; list available as Additional file 6). ‘Translation’ was the most common functional class, represented mostly by the biochemical pathway ‘Ribosome’ (ko03010). The number of transcripts assigned to this pathway is about 7.5x higher than the second most expressed pathway, ‘Photosynthesis’ (ko00195), class ‘Metabolism’.

Analysis of KEGG pathway expression showed that 32 of the 40 core pathways had significant differences in transcript representation (Figure 2). Relative to vegetative tissue, both female and male sexual tissues were under-represented for transcripts belonging to energy metabolism pathways, including photosynthetic carbon fixation (ko00710) and antennae proteins (ko00196). Under-expression of nitrogen metabolism (ko00910) was mainly due to fewer transcripts for carbonic anhydrases (Additional file 6), which are likely to mediate inorganic carbon acquisition in brown algae 52 53 54 . These differences were most obvious in the male tissue (Figure 2), also supported by the observation that transcripts for photosynthesis (ko00195; i.e. generation of reducing power in the light reactions) were under-expressed in males, but not in females. Several pathways for carbohydrate metabolism were also under-expressed in both sexes; glycolysis and gluconeogenesis (ko00010), pentose phosphate pathway (ko00030), fructose and mannose metabolism (ko00051) (Figure 2). Under-expression of transcripts for peroxisome (ko04146) and PPAR signaling (ko03320) pathways in sexual tissues was also seen. However, the main peroxisomal transcripts affected are involved in free radical detoxification and/or redox signalling (peroxiredoxin and Mpv17), rather than lipid homeostasis. Furthermore, peroxisomal, rather than chloroplast or mitochondrial targeting of these transcripts could not be confirmed. Similarly, down-regulation of the peroxisome proliferator-activated receptor (PPAR) pathway was due almost entirely to under-expression of acyl-CoA-binding protein (K08762), which binds medium- and long-chain acyl-CoA esters and has multiple cellular roles.

Male-biased expression of the ribosome pathway, as well as RNA transport (ko03013), RNA degradation (ko03018) and mRNA surveillance (ko03015), RNA polymerase (ko03020), spliceosome (ko03040), nucleotide excision repair (ko03420), protein processing in the endoplasmic reticulum (ko04141), ubiquitin mediated proteolysis (ko04120), oocyte meiosis (ko04114) and cell-cycle pathways (ko04110, ko04111) is indicative of actively differentiating and replicating cells (Figure 2, Additional file 6).

After mapping of F. vesiculosus genes to Ectocarpus orthologues, a single female-specific gene was identified: a member of the mannuronan C-5-epimerase gene family. This gene catalyses the final step of alginate biosynthesis 55 (Table 2). Another female-biased gene was also identified, and related to carbohydrate modification with a likely role in cell wall biogenesis (chondroitin-D-glucuronate 5-epimerase). A gene encoding aquaporin (APQ1; K09864) that was over-expressed in female tissue in spring tide library was also over-expressed in males in the neap tide library, and was barely detectable in either vegetative tissue library (Table 2; Additional file 4).

The table shows the number of F. vesiculosus contigs mapping (BLASTX, E-value ≤ 10^-10) to an E. siliculosus protein; descriptions are based on the accession and/or annotations in the KEGG, Pfam and GO databases. Expression is shown for Spring Female (F), Male (M) and Vegetative (Veg) libraries (as normalized reads; see Results for more details).

In contrast to females, many more male-biased or male-specific genes were identified (Table 2). Amongst male-specific terms, signalling-related genes were conspicuously over-represented. Most striking was a mitogen activated protein kinase (MAP2K; K04368), a cAMP-dependent protein kinase regulator (PRKAR; K04739), a PAS-PAC histidine kinase and putative blue-light photoreceptor, calmodulin genes and a Ca²⁺/calmodulin-dependent protein kinase (CaMK; K00908). This provides a general picture of male tissue as very active in signalling for a potentially diverse range of cellular processes. The over-expression of dynein-related molecular motor proteins (K10410, K10411), tenascin-like Sig (sexually induced) genes and creatine kinases involved in energy storage and release are likely related to specific flagella localization and functions. We also identified male-biased genes for histones, centrin, as well as α- and β-tubulins (Table 2).

Quantitative RT-PCR analysis of independent samples of sexual and vegetative tissues confirmed the results obtained by pyrosequencing. Seven male-biased and one female-biased transcripts were selected after tests for amplification efficiency. The specific contig and primer sequences are given in Additional file 7. In agreement with pyrosequencing results, alpha-tubulin (TUBA) and centrin were confirmed as male-biased, with low but detectable expression in non-male tissue (Figure 4a,c), while a beta-tubulin isoform, a Mec-17 domain protein (pfam05301; touch receptor protein), and 2 tenascin-like proteins homologous to sexually induced (Sig) genes 1 and 3 from the diatom Thalassiosira weissflogii were male-specific (Figure 4d,e,f). A cAMP-dependent protein kinase regulatory subunit (PKA), which was only detected in male libraries, was highly male-biased in qPCR tests, with > 20 fold lower expression in females and below detection limits in vegetative tissue (Figure 4g). Variation in expression within the tidal cycle was less clear, although increased expression in males at neap tide for some transcripts (centrin, PKA) agrees with pyrosequencing results. Finally, the female-biased expression of a mannuronan C-5-epimerase was also confirmed (Figure 4h). Interestingly, expression of this transcript was not detected in the post-neap sample (i.e., following gamete release) (Figure 4h), suggesting that it may be localized to mature oogonia.

Primary metabolism for energy and carbohydrate pathways differed between the sexes and between sexual and vegetative tissue, presumably reflecting functional constraints and trade-offs associated with different developmental programs and sexual phenotype. Receptacle photosynthetic rates are reported to be ca. 50% lower than in non-reproductive tissue in F. serratus 56 , although no information on sex-specific variation is available. At the transcriptional level, male under-expression of photosynthesis-related pathways exceeded that in females, particularly for light energy capture (antennae proteins) and photosynthetic electron transport (neither of which differed between female and vegetative tissue at spring tide). This could reflect greater female investment in gamete provisioning.

While respiratory pathways were under-expressed in sexual tissue generally, neither oxidative phosphorylation nor the citrate (TCA) cycle in males were significantly different from vegetative tissue during the spring tide phase of reproduction, the period when energy expenditure in spermatogenesis, and the maturation of anteridia for release on the following neap tide cycle should be greatest. The fucoid sperm is the only motile phase of the life cycle, showing a heterokont biflagellated cell pattern 57 . Sperm flagella have two main functions requiring high energy consumption: swimming and mediation of gamete-gamete interactions during fertilization (via the anterior mastigoneme-bearing flagellum; 58 ). Transcripts for creatine and arginine kinases (arginine and proline metabolism pathway; Additional file 4), were also highly expressed in males, and virtually absent from other tissues. Phosphocreatine and phosphoarginine are storage molecules forming an energy reservoir from which ATP can be rapidly generated to replenish cellular energy balance in highly active cells.

The estimated sperm-egg ratio is 400:1 in F. vesiculosus 23 , so elevated levels of cell cycle activity and associated processes during gametogenesis in male tissue may arise simply because a greater number of cells (and proportion of mRNA) in the tissue sample are affected. Cell cycle and oocyte meiosis pathways were over-expressed uniquely in males, and were accompanied by male over-expression of pathways involved in transcription (RNA polymerase, spliceosome), translation (ribosome, mRNA surveillance), protein folding, sorting and degradation (ubiquitin-mediated proteolysis, protein processing in the endoplasmic reticulum, RNA degradation), and replication and repair (nucleotide excision and repair). Overall, pathway analysis strongly reflects higher replicative activity in male relative to female reproductive tissue.

High throughput sequencing technologies have increased our focus on sex-biased gene expression with the realization that a large proportion of the transcriptome in many organisms shows distinct patterns of male and female expression 4 16 59 60 61 . Sexual dimorphism occurs against an almost identical genomic background, implying that the majority of sexually dimorphic traits may experience differential, or even conflicting, selection pressure in males and females 4 . Several studies have identified demasculinization (under-representation of male-biased genes) and/or femininization (greater numbers of female-biased genes) on X chromosomes in organisms with heterogametic male (XY) chromosomes ( 4 19 62 63 64 65 66 ; see also 15 for ZW sex determination in birds). However, the evolutionary dynamics of demasculinization may be complex, with evidence in Drosophila that recent male-biased genes may initially accumulate on the X chromosome and be lost over time 67 .

This is the first dataset to our knowledge that can begin to address the question of sex-biased gene expression in brown algae, an independent multicellular lineage 20 . Despite a relatively low sequencing depth, we found more genes over-expressed in males than in females (ca. 13% versus 4% of the total), and that both the mean and variance of expression were significantly greater in males. We currently know little about sex determination in brown algae other than that it is genetically based 68 , although the recent discovery of a non-recombining chromosomal locus in Ectocarpus promises progress in this area (SM Coelho and JM Cock, pers commun; 69 ). Since most evidence from the literature suggests greater sex-biased expression in the heterogametic sex (males in Drosophila, Silene, females in birds), it will be interesting to further study sex-determination and the occurrence or not of heterogametic sex in Fucus.

Relatively few over-expressed female annotations were identified in our dataset. Overall, our data are consistent with widespread pleiotropic effects in female-effect genes 1 17 18 59 .

The three clearest examples of female-biased genes are all carbohydrate-modifying enzymes. Two were epimerases - mannuronan C-5-epimerase and chondroitin-D-glucuronate 5-epimerase (COG0451) - with roles in cell wall biogenesis and/or modification. Mannuronan C-5-epimerase is a multigene family in brown algae 55 , performing the final step in alginic acid biosynthesis that is critical for the formation of the alginate-rich brown algal cell wall. Of the 13 contigs in our dataset with top Blast hits to 12 predicted mannuronan C-5-epimerase proteins in Ectocarpus, three were female-unique, and one was expressed at a relatively higher level in female reproductive tissue. A third transcript encodes a homologue to an Ectocarpus glucosylceramidase. This transcript(s) contained both an O-glycosyl hydrolase (hydrolysis of glycosidic bonds) and a lectin-like (carbohydrate-binding) domain. It seems unlikely that this gene product participates directly in the fertilization process, as the presence of lectin domains on female gamete recognition proteins runs contrary to what is known about gamete recognition in Fucus, or invertebrates, where sperm-localized glycoproteins bind carbohydrate ligands on the egg 31 41 42 70 . The functional significance of this gene therefore remains to be seen.

Gametangial expulsion from fucoid receptacles is a rapid (min) process 34 , mediated by osmotic adjustments associated with periods of K⁺ and Cl^- efflux from oogonia into the conceptacle extra-cellular matrix (ECM) 36 . Subsequent swelling of the ECM and/or expansion of anchoring gametangial stalk cells may be key events driving gametangial release. We identified a sex-biased (FS and MN) aquaporin-1 that could potentially play a key role in controlling rapid movements of water between cells and the ECM. Although previous work with inhibitors provided evidence for the involvement of K⁺ uptake, slow-type anion channels, and tyrosine phosphorylation in gamete expulsion 34 , our transcript profiles failed to reveal candidate genes with potential roles in these processes in sexual tissue.

The number of male-biased and -unique annotations confirms male reproductive tissue as being highly differentiated compared with either female or vegetative tissue. The majority of these gene products are likely components of the flagella (e.g., tubulin, dynein), or flagella basal bodies (e.g. centrin; 71 ). In addition, a number of genes with known functions in signal perception and transduction and/or cytoskeletal regulation were male-specific. A phototropin-like blue light receptor was uniquely expressed in males, with homology to an Ectocarpus PAC/PAS sensor histidine kinase and green algal phototropins. Fucoid sperm are negatively phototactic, and this gene is a prime candidate for the sperm photoreceptor. The absence of any transcript for a putative blue-light receptor in female tissue (including putative opsin-like proteins; 72 ) is curious, since photopolarization by blue-light is one of the first axis-forming responses of the zygote after fertilization 73 74 . It remains possible however, that expression of some female genes at the neap tide (gamete-release) phase of the reproductive cycle are under-represented in the spring tide library, when numbers of mature oogonia may be lower 30 34 .

Eukaryotic cilia and flagella use both cyclic adenosine monophosphate (cAMP) and calcium as second messengers to regulate the beating of dynein-driven microtubules in the axomene 75 . We detected a male-specific cAMP-dependent protein kinase (PKA) regulatory subunit. PKA is known to play a key role in mammalian sperm capacitation 76 via the cAMP/PKA signaling pathway. cAMP-dependent phosphorylation of flagella proteins, particularly axonemal dynein, is essential for the initiation and maintenance of mammalian sperm motility, with PKA the likely downstream kinase responsible for phosphorylation 77 78 . Therefore, although we were unable to confirm the presence of adenylate cyclase or potential PKA-anchoring proteins in our dataset, the male-specific expression of PKA regulatory subunit hints at shared components in sperm signalling pathways in brown algal and mammalian systems.

Male transcriptome data revealed the expression of a specific CaMK, a male-specific calmodulin isoform, and a dynein-light chain (both the latter with CaM EF-hand Ca²⁺-binding domains). Calcium and Ca²⁺ signalling play a central role in flagella function and sperm motility from mammals to green algae 79 80 , with evidence for the involvement of calmodulin (CaM) and Ca²⁺/CaM-dependent kinases (CaMK) that modulate dynein activity in the axoneme 81 82 83 . The identification of a male-specific 1-phosphatidylinositol-4-phosphate 5-kinase (PIP5K) annotated as a axonemal radial spoke protein in Ectocarpus may further indicate a link between the phosphatidylinositol signalling pathway, releasing Ca²⁺ from intracellular stores, with Ca²⁺ induced modulation of flagella activity in sperm.

Finally, a highly over-expressed mitogen activated protein kinase kinase (MAPKK) was identified in male reproductive tissue. The MAP kinase pathway has many important roles mainly in the transduction of extracellular signals, particularly in plant stress signalling pathways 84 , but also more generally in the control of cell growth, death, proliferation and differentiation. In the context of male-specific expression, MAP kinases are implicated in the control of flagella development in Chlamydomonas 85 .

Homologues to the sexually induced protein (Sig) family, first identified in Thalassiosira weissflogii by Armbrust 86 , showed male-specific expression in Fucus vesiculosus. Based on homology with Ectocarpus predicted protein sequences, four independent transcripts were identified, with sequence similarities to T. weissflogii Sig1-3. The most highly expressed member was Sig1-like and contained cysteine-rich epidermal growth factor (EGF)-like domains and similarity to extracellular tenascin-like glycoproteins involved in vertebrate cell adhesion 87 88 . Interestingly, some Ectocarpus homologues are annotated as tubular mastigoneme proteins of the anterior flagellum. Mastigonemes are tubular hairs present on the anterior flagellum of heterokonts (but lacking in diatoms) that play an important role in establishing sexual contact in Ectocarpus 58 . A second group of five male-specific contigs homologous to Ectocarpus CBN79192, and with limited homology to Herpes virus major outer envelope glycoprotein (Herpes_BLLF1; Pfam05109) could also be involved in cell-cell interactions.

After earlier work established some of the characteristics of sperm-egg recognition in Fucus 41 42 89 , further progress has slowed due mainly to a lack of genomic information. Candidate genes for potential gamete recognition proteins will stimulate further molecular analysis (e.g. the Sig gene family) in the context of speciation, one of the primary motivations for this study. Diversification within the genus Fucus began fairly recently, and is ongoing 21 90 . While barriers to interspecific fertilization between more distantly-related members operate under normal conditions 91 , historical and/or ongoing hybridization has been inferred in several studies 22 23 24 25 92 . Species complexes where barriers to hybridization range from ecological factors 24 30 to (potential) divergence in gamete recognition proteins promise insight into the evolutionary processes underlying species divergence.