Evolution of sex in crops: recurrent scrap and rebuild

Sexuality is the main strategy for maintaining genetic diversity within a species. In flowering plants (angiosperms), sexuality is derived from ancestral hermaphroditism and multiple sexualities can be expressed in an individual. The mechanisms conferring chromosomal sex determination in plants (or dioecy) have been studied for over a century by both biologists and agricultural scientists, given the importance of this field for crop cultivation and breeding. Despite extensive research, the sex determining gene(s) in plants had not been identified until recently. In this review, we dissect plant sex evolution and determining systems, with a focus on crop species. We introduced classic studies with theoretical, genetic, and cytogenic approaches, as well as more recent research using advanced molecular and genomic techniques. Plants have undergone very frequent transitions into, and out of, dioecy. Although only a few sex determinants have been identified in plants, an integrative viewpoint on their evolutionary trends suggests that recurrent neofunctionalization events are potentially common, in a “scrap and (re)build” cycle. We also discuss the potential association between crop domestication and transitions in sexual systems. We focus on the contribution of duplication events, which are particularly frequent in plant taxa, as a trigger for the creation of new sexual systems.

has low fitness (Blaser et al. 2014). Another possibility is that when sexually antagonistic polymorphism is maintained on an autosome, a new sex determinant that arises in the region becomes advantageous (Van Doorn and Kirkpatrick 2007). Although genome sequencing of sex chromosomes was a difficult task until five years ago, recent developments in high-throughput sequencing technology have allowed (almost) perfect construction of sex chromosomes in some species, as discussed later.

How did plants evolve sex?
One of the most representative theoretical models, the "two-mutation" model Charlesworth 1978a, Westergaard 1958), proposes the evolution of dioecy via gynodioecy. In this model, dioecy is thought to be established via a male-sterility mutation (often called disruption of the "M factor") in an ancestral, functionally hermaphroditic state and, thereafter, a gain-of-function mutation for suppressing feminization (often called "SuF") (Charlesworth and Charlesworth 1978a). Although this "two-mutation" model is predominant in discussions of plant sex evolution, the same authors also proposed another theoretical model for the evolution of dioecy with a single factor, via monoecy-paradioecy (Charlesworth 2013, Charlesworth andCharlesworth 1978b). Indeed, dioecy has been artificially engineered in monoecious maize (Zea mays) (Jones 1934) and cucurbit (Boualem et al. 2015). In any models, resource allocation and the avoidance of inbreeding are presumably both involved whenever a dioecious species evolves from a hermaphroditic species (Charlesworth and Guttman 1999). To verify these models, understanding of the evolution of sex determining gene(s) located on sex chromosomes would be indispensable. However, the sex determinants in plants have been identified in relatively few species, and evolutionary paths are still poorly defined (Renner and Müller 2021). Understanding the mechanisms behind sex determination is important not only for improving pollination success or controlling yield and fruit quality (fruit shape, or parthenocarpy), but also for planning cropping or breeding strategies. The detailed mechanisms for different sex determining gene(s) in crops are discussed in the next section.
Another important question concerns the effect of polyploidy on sexual systems (or transitions into and out of dioecy) (Ashman et al. 2013, Glick et al. 2016. The association between dioecy and polyploidy was documented by Baker (1984) and was highlighted by Miller and Venable (2000). Polyploidy is frequently found in several plant clades in which sex chromosomes have been well studied [e.g., in the genera Salix (Wagner et al. 2020), Silene (Popp and Oxelman 2007), Mercurialis (Russell and Pannell 2015), Diospyros (Akagi et al. 2016a), and Fragaria (Tennessen et al. 2018)]. Polyploidy can affect genome evolution and gene expression, such as chromosomal rearrangements and gene loss (Udall and Wendel 2006), unequal rates of sequence evolution of duplicated genes, and changes in DNA methylation. Such changes can result in novel phenotypes, ecological diversification, and new niche invasion (Otto and Whitton 2000). The acquisition of new functions is thought to facilitate novel reproductive systems, including conversion from dioecy to asexuality or selfing systems (Akagi et al. 2022). Nevertheless, the exact mechanisms that trigger transitions into and out of dioecy, via polyploidy, are not well understood.

Why so many pathways?
The discontinuity of dioecious plant lineages in evolutionary trees suggests that the determinants located in the sex chromosomes evolved independently (Ming et al. 2011, Renner 2014. Consistent with this assumption, physiological reactions to plant hormones, in terms of sex, can vary, with no clear commonality in direction. For instance, cytokinin (CK) treatment can enhance gynoecium growth in male grape accessions, resulting in the production of functional hermaphroditism (Negi andOlmo 1966, Wang et al. 2013). Similar sex conversion (or feminization) has been observed in crop species, including spinach (Spinacia oleracea) (Chailakhyan andTimiriazev 1979, Grant et al. 1994), hexaploid Oriental persimmon (Diospyros kaki) (Yonemori et al. 1993), kiwifruit (Actinidia chinensis or deliciosa) (Akagi et al. 2018), and wild plants, including the genus Mercurialis (Durand and Durand 1991), and Plukenetia volubilis (the family Euphorbiaceae) (Luo et al. 2020). Feminization can also be triggered via ethylene or auxin in hemp (Cannabis sativa), and via gibberellin in maize, as summarized by Grant et al. (1994). This situation is, at least partially, a result of differences in the reaction point in sex determination.
Various organs are potentially responsible for sex determination in plants (Dellaporta and Calderon-Urrea 1993). As represented by the two-mutation model (Charlesworth and Charlesworth 1978a), transitions into dioecy from ancestral functional hermaphroditism would require novel regulatory systems to sterilize either the gynoecium or androecium. Particularly for the androecium, there are various spatiotemporal procedures to form fertile gametes (Goldberg et al. 1993, Koltunow et al. 1990), from primordia formation in the shoot apical meristem to tapetum degradation during anther maturation. Any mutations in these many procedures could be a potential causal factor in the establishment of male or female, which might be a physiological reason that plants have recurrently evolved sexes in lineage-specific manners. For instance, androecium development in the female flowers of hemp is arrested in the primordia initiation stage (Mohan Ram and Nath 1964), while garden asparagus (Asparagus officinalis) or persimmon develop rudimentary anthers in female flowers, which fail during microsporogenesis (Akagi et al. 2014, Lazarte andPalser 1979). Kiwifruit and grapes exhibit "cryptically dioecious" systems (Mayer and Charlesworth 1991), in which female flowers look structurally hermaphroditic but are functionally unisexual, owing to pollen sterility that occurs during the pollen maturation stage , Falasca et al. 2013, Lombardo et al. 1976, 1978. Despite these findings of morphological and physiological diversity in plant sex determination, the molecular pathways have only been characterized in a limited number of dioecious species. In this era of massive sequencing, there would be numerous results for a simple comparison of transcriptomic data between male and female. However, it is worth noting that physiological reactions in sex differentiation often differ among the organs or developmental stages, even under the regulation of a single sex determinant. In persimmon, an example of a well-characterized sex determination process, differentially expressed genes between males and females do not overlap in organ primordia initiation, organ development, and flower maturation stages (Li et al. 2019, Wang et al. 2020, Yang et al. 2019. Thus, with only physiological analyses, it is often hard to define a single molecular pathway for sex determination in a species.

Sex determining genes and molecular pathways
Recent rapid progress in genomics has enabled the identification of sex chromosome-encoded determinants, which was previously challenging because of putative long nonrecombining regions of sex chromosomes (Ming et al. 2011). The first finding of a sex chromosome-encoded determining gene in a plant was, interestingly, made in a wild persimmon, Diospyros lotus (or Caucasian persimmon) with no reference genome information (Akagi et al. 2014; Table 1). Cataloging of subsequences (or k-mer) in Illumina short reads, and a comparison of the male and female k-mer pools, resulted in direct isolation of Y chromosomespecific polymorphisms, including the determinants (Akagi et al. 2014). In persimmon (genus Diospyros), a Y chromosome-encoded small-RNA gene, named OGI, acts to suppress an autosomal counterpart gene, named MeGI, which is a HD-ZIP1 homeodomain gene. MeGI solely orchestrates both repression of androecium growth and promotion of gynoecium development, to integrate "feminization" (Yang et al. 2019). For androecium development, at least during primordia development, an APETALA3 (AP3)/PISTILLATA (PI)-mediated pathway is repressed by SHORT VEGETATIVE PHASE (SVP)/ AGOUMOUS-like 24 (AGL24) and AGAMOUS (AG) (Gregis et al. 2009), putatively under the control of MeGI (Yang et al. 2019). MeGI can hypothetically enhance gynoecium development via direct regulation of the genes reminiscent of CK signaling, such as KNOTTED-LIKE FROM ARABIDOPSIS THALIANA 1 (KNAT1), OVATE FAMILY PROTEIN (OFP). This was supported by the restoration of gynoecium in male flowers with CK treatment in hexaploid persimmon (Yonemori et al. 1993).
Garden asparagus (Asparagus officinalis) is a dioecious  (Harkess et al. 2017, Murase et al. 2017, Tsugama et al. 2017. Whole genome sequencing of a doubled haploid 2A + YY individual, with the Illumina-PacBio-BioNano hybrid method, and resequencing of YY and XX siblings, found limited numbers of the sex determinant candidates in the hemizygous Y-specific region (Harkess et al. 2017). Further gamma irradiation-induced and EMS-induced mutants confirmed the two described genes as the sex-determining genes (Harkess et al. 2017(Harkess et al. , 2020. The gene aspTDF1 is an ortholog of Arabidopsis R2R3-type Myb gene, DEFECTIVE IN TAPETAL DEVEL-OPMENT AND FUNCTION1 (TDF1), which plays an important role in the degradation of the tapetal layer in anther maturation (Zhu et al. 2008). Consistent with its function in Arabidopsis, aspTDF1 is substantially expressed in garden asparagus' tapetal layer (Harkess et al. 2020), while its high expression has also been validated in early flower development (Tsugama et al. 2017). This situation is feasible because the female phenotype exhibits rudimentary anthers, which are derived from a lack of M factor functioning at the anther primordia stage. Nevertheless, molecular pathways regulated not only by aspTDF but also by SOFF are not well known. Sexuality of kiwifruit (or the genus Actinidia) is also regulated by two genes in the Y-chromosome, named Shy Girl (SyGI) and Friendly Boy (FrBy), acting as SuF and M factors, respectively (Akagi et al. 2018. SyGI is a type-C Arabidopsis Response Regulator 22/24 (ARR22/24)-like gene. SyGI negatively regulates the CK signaling pathway (Kiba et al. 2004) and affects the expression patterns of genes involving gynoecium identity or development, such as AGAMOUS (AG), SHATTERPROOF (SHP), or SUPER-MAN (SUP) (Varkonyi-Gasic et al. 2021). FrBy is a gene with the fasciclin domain (FAS1 domain), which plays an important role in tapetum cell degradation for proper pollen maturation. The orthologs of FrBy are putatively monophyletic and functionally conserved in a range of angiosperm species , Tan et al. 2012. SyGI arose recently via a lineage-specific duplication event that resulted in a new expression pattern in the gynoecium (Akagi et al. 2018). This evolutionary pattern is thought to be similar to that in garden asparagus, which is consistent with the scenario proposed in the two-mutation model (Charlesworth and Charlesworth 1978a). The timeline for establishment of the two factors, which could be important for defining their evolutionary significance, remains unknown. The closest outgroup lineage (genus Saurauia, in the family Actinidiaceae) is also dioecious (Haber and Bawa 1984), and the sex determinant would be independent of the genus Actinidia (Akagi et al. 2018).
Poplar (Populus species) has multiple dioecious mecha-nisms, which grouped into both XY (heterogametic male) and ZW (heterogametic female) systems in lineage-specific manners during evolution of the genus (Müller et al. 2020). Translocations, duplications, or simple mutations of a single ancestral regulator, ARR17, have formed various structures to cause unisexuality at least three times in the genus Populus (Müller et al. 2020). However, other possibilities cannot be excluded, such as a potential two-gene model in P. deltoides (Xue et al. 2020). ARR17 is a type-A Arabidopsis Response Regulator (ARR), with functions that may be overlapped with SyGI (type-C ARR) in terms of negative regulation of CK signaling. However, their structures are quite distant (Kiba et al. 2004, To and Kieber 2008 Recent studies identified sex determinant candidates in some other crops, including grapevine (Vitis species) and date palm (Phoenix species). Cultivated grapevine species (V. vinifera, V. labrusca, and their hybrid V. labruscana) are hermaphroditic (as discussed in detail later), while wild grapes are mostly dioecious. Efforts to trace the history of recombination from which hermaphroditism was derived in cultivated grapevines, suggest limited numbers (fewer than 10) of two-factor sex determinant candidates in a small sexlinked region (>150 kb) (Badouin et al. 2020, Massonnet et al. 2020. Importantly, no candidates in the grapevine overlap with the sex determinants identified in other species. However, Badouin et al. (2020) suggested that regulation of gynoecium development (or function of SuF) might involve CK signaling. A comparative genome analysis of dioecious date palm species and monoecious oil palm revealed the sex determination region in the Y chromosome, which included the three most likely candidates (Torres et al. 2018). Of these, a duplicated LONELYGUY (LOG)-like gene, which putatively involves CK activation (Kuroha et al. 2009), is thought to be an SuF candidate. These situations may indicate a potential link among dioecious species for the function of gynoecium regulators, i.e., they commonly affect CK metabolism or signaling.

Theoretical models and sex chromosome evolution
No uniform evolution has been proposed for plant sex determination, either in terms of molecular mechanisms or theoretical frameworks. The "two-mutation" theory Charlesworth 1978a, Westergaard 1958) is a representative model that explains many dioecious systems, and has been validated in garden asparagus (Harkess et al. 2017) and kiwifruit . However, sex determination in persimmon, in which the Y-encoded smRNA OGI acts alone, is apparently distinct from the canonical two-gene model. This single-gene framework could be a result of the bidirectional function of the target of OGI, MeGI, which simultaneously represses and promotes androecium and gynoecium development, respectively (Yang et al. 2019). From an evolutionary aspect, this OGI-MeGI interactive system may share a consistent concept with the two-mutation model (Charlesworth 2019). A single-gene system is also applicable to P. tremula, in which gene-editing of the Y-encoded ARR17 resulted in conversion from male to female (Müller et al. 2020). In contrast with the two-mutation model, concrete evolutionary paths for the single-gene system have not yet been defined. An assumption for this evolution is the conversion from monoecy to dioecy, by establishment of a genetic factor to complement a biased female/male ratio in an individual (Charlesworth 2013), which would be consistent with the artificial evolution of dioecy in Cucurbit (Boualem et al. 2015).
Recent progress in DNA sequencing techniques has enabled chromosome-scale draft genome construction, even in dioecious species with sex chromosomes (e.g., Harkess et al. 2017Harkess et al. , 2020 Ma et al. 2022 for spinach). With these genomic results (along with functional analyses of sex determining genes), evolutionary mechanisms to form sex chromosomes in plants may have to be reconsidered. Conventional theory suggests that gradual expansion of a non-recombining region or structural degeneration of the Y chromosome would be a result of maintenance of linkages between the sex determining gene and the surrounding sexually antagonistic genes, or maintenance of sexual dimorphism. Thus, the age of sex chromosomes is thought to be important for their degree of heteromorphism. This hypothesis is, at least partially, supported by some dioecious species (Ming et al. 2011). However, the recent accumulation of sequencing data of sex chromosomes suggests that this tendency might not be common in plant species (Renner and Müller 2021). Furthermore, functional analyses of sex-determining genes from persimmon and kiwifruit suggest that representative sexual dimorphisms in these species can be explained by the pleiotropic functions of sex determining genes, and be independent of genes surrounding the sex determinants Charlesworth 2019, Akagi et al. 2023). Importantly, a recent theoretical model also proposed the potential independence of Y chromosome degeneration and selection of sexual dimorphisms (Lenormand and Roze 2022). There are clearly still many mysteries in plant sex chromosome evolution.

Impact of domestication
Domestication is the most important process of selection by humans which transformed wild plant forms into crops. The gradual transition from hunting and gathering to plant cultivation and animal husbandry began between the end of the Pleistocene and the beginning of the Holocene, some 12,000-10,000 years ago (Arranz-Otaegui et al. 2018, Fuller et al. 2014. As a result of this process, crop plants and domesticated animals share suites of modified traits, referred to as 'domestication syndrome', which differentiates them from their wild ancestors (Meyer et al. 2012, Stetter et al. 2017. In plants, domestication syndrome often involves changes in plant architecture, shattering, the sugar content of fruit, fruit size, and the shift from outcrossing sexual systems. Of these changes, the shift from an outcrossing system is thought to be a requirement for the stable production (cultivation) of a crop with extremely low genetic diversity, as first implied by Darwin (1877b). Although a number of agronomically important plants still exhibit dioecy (see Table 1) (and other outcrossing systems, such as self-incompatibility or dichogamy), some major horticultural crop species successfully switched from dioecious progenitors to self-fertile hermaphroditism during domestication, e.g., grape, papaya, carob, and strawberry. This switch was likely driven by selection for reproductive assurance (Crossman and Charlesworth 2014).

Artificial selection of hermaphroditism in domestication
Wild populations of papaya are strictly dioecious, while cultivated populations are almost always gynodioecious, including hermaphroditic and female individuals (Brown et al. 2012). These findings imply a breakdown of dioecy during domestication. Hermaphroditic papaya plants reliably produce fruit because they can self-pollinate and do not require male plants. In addition, the shape of hermaphroditic fruit, which is more elongated than female fruit, is preferred for commercial production (Liao et al. 2021). Sex determination in papaya is controlled by a recently evolved heterogametic male system (or XY system). A putatively mutated Y chromosome, named Y h , corresponds to hermaphroditic individuals , Wang et al. 2012. All genotypes without an X chromosome (YY, YY h , and Y h Y h ) die in early development, resulting in 25% aborted seeds in self-pollination of hermaphrodites and in crosses between hermaphrodites and males (Ming et al. 2007). This finding indicates that both normal and mutated Y alleles have lost at least one gene essential for development. The Y h chromosome has a non-recombining male-specific region (MSY), as does the normal Y chromosome, which limits the hermaphroditicspecific region of Y h (HSY) (VanBuren et al. 2015, Wang et al. 2012. The Y and Y h chromosomes share much of their genomic context (99.6% sequence similarity), except in the hermaphrodite-specific HSY region. Y h was putatively derived from mutation of the female suppressing factor (SuF), as proposed in the two-mutation model. Importantly, the HSY region including the mutated SuF, exhibits low genetic diversity and some indexes for selective  . Similar to papaya, hermaphroditism in cultivated grapevine is also thought to be a result of a mutated allele of the Y chromosome, named the H allele. The H allele appears to lack the SuF gene and is derived from a part of the Y chromosome in wild male grapevines (Picq et al. 2014). Population genetics approaches have identified indexes of selective sweeps on loci that appear to be associated with hermaphroditism, as well as berry color, implying artificial selection during domestication (Zhou et al. 2019). Genome sequencing analysis across 556 genotypes found two distinct hermaphrodite haplotypes (H1 and H2) among the cultivated grapevines, both derived from chimeras of male (M) and female (f) haplotypes (Zou et al. 2021). H1 haplotypes are distributed globally and are common in table, raisin, and wine grapes. H2 haplotype is only found in a group of wine grape cultivars in a limited area (north and west Europe). The time of divergence between the H1 and H2 haplotypes is estimated to be ~6 million years ago (Zou et al. 2021), which apparently predates the domestication of grapevines (~8,000 years ago). This inconsistency might suggest two independent selection histories of these hermaphrodite haplotypes.

Escape from a dioecious system associated with polyploidization
The genus Diospyros consists of about 400 species that are widely distributed in the (sub)tropical areas of Asia, Africa, and the Americas, and are generally dioecious. An exception to this dioecy is a cultivated hexaploid species, D. kaki (2n = 6x = 90), known as Oriental persimmon, which is considered to be polygamous, with three types of sex expression: gynoecious, monoecious, and polygamomonoecious (hermaphroditic, female, and male flowers in an individual). In D. kaki, genetically female individuals (or individuals with hexaplex X: 6A + 6X) produce no male flowers, as well as in diploid wild relatives. Genetically male individuals (or individuals carrying at least one Y chromosome) can exhibit monoecious expression (Akagi et al. 2016b). In the Y chromosome of D. kaki, the sex determinant OGI was fundamentally silenced by the insertion of a short interspersed nuclear element (SINE)-like TE, named Kali, into its promoter region (Akagi et al. 2016a). Instead, the autosomal counterpart of OGI, MeGI, established a novel expression regulatory system based on DNA methylation in its promoter region that acts as an epigenetic switch to produce male or female flowers (Akagi et al. 2016a). Importantly, the Kali SINE in the OGI promoter is conserved in all accessions with a Y chromosome, including landraces that hypothetically have not experienced domestication events and exhibit no clear signals for recent selection. This situation would suggest that the conversion from male to monoecy was independent of domestication and might be coincident with (or possibly predate) the hexaploidization event.
The hexaploid D. kaki has further invented a new pathway to occasionally convert male flowers to hermaphroditic ones, independent of the OGI-MeGI system (Masuda et al. 2022). This sex conversion can occur naturally or be induced by cytokinin treatment (Yonemori et al. 1993). The mechanism depends on hexaploid-specific activation of the cytokinin-or abscisic acid-responsive signaling pathways and their putative integrator, RADIALIS-like DkRAD (Masuda et al. 2022). Notably, this DkRAD-mediated hermaphrodite expression occurs randomly in a tree and has not been utilized for fruit production, suggesting no involvement in the domestication/improvement process. The RADIALIS gene is well known as a regulator in flower morphology, particularly petal architecture (Lucibelli et al. 2020). Hence, the DkRAD function in persimmon as a sex converter could be a result of a neofunctionalization (Bergero 2022). Although the expression of hermaphroditism in D. kaki may look like a reversion to the ancestral state via loss of the existing pathway, this evolution could be a reinvention of hermaphroditism via establishment of a novel pathway. As shown in Fig. 1, persimmon provides a good example of frequent "scrap & rebuild" evolution of sex expression, which may be triggered by ancient and recent duplication events (Masuda et al. 2022).

Perspective: the power of duplication
We focus on gene or genome-wide duplication as a potentially common mechanism to trigger transitions into and out of dioecy in plants. The involvement of genomeduplication (or polyploidization) in such transitions has been theoretically discussed, based on available literature on sexual systems in various plant species (Glick et al. 2016, Goldberg et al. 2017). Recent genome duplications or polyploidization events have also enabled plasticity in plant sexual reproduction systems, including sexuality (Ashman et al. 2013, Comai 2005. The focus on duplications is due to the nature of plant genome behavior. Plants have undergone frequent genome-wide duplications, in a lineage-specific manner, as exemplified by the frequent paleo-ploidizations in the Cretaceous-Paleogene (K-Pg) boundary (Van de Peer et al. 2017). Genome duplications rapidly produce various phenotypic consequences associated with gene expression variation-resulting from novel allele dosages or regulatory interactions-or with genetic and epigenetic rearrangements/modifications (Fox et al. 2020, Osborn et al. 2003, Van de Peer et al. 2021, Wendel 2000. In the long term, duplicated gene pairs often actively exhibit sub-functionalization, pseudogenization, or neofunctionalization, mainly to release them from functional redundancy or adaptive conflicts (Des Marais and Rausher 2008, Flagel and Wendel 2009, Ohno 1970. In other words, a lineage-specific duplication can provide an opportunity to acquire novel functions or traits representing each species. For instance, oil production in olive (Olea europaea) (Unver et al. 2017), fruit ripening processes in tomato (Solanum lycopersicum) (Tomato Genome Consortium 2012), or adaptation to the sea in seagrass (Zostera marina) (Olsen et al. 2016), were reportedly established via genome duplications specific to each lineage, followed by dynamic functional diversifications between the duplicated genes.
The evolutionary processes involved in the establishment of sex determinants could also be explained by genome or gene duplications to invent novel functions (summarized in Fig. 2). As partially suggested in the previous sections, in persimmon, the Y-encoded OGI was derived from gene duplications to form an inverted repeat (or small-RNA encoding locus), which is specific to the genus Diospyros (Akagi et al. 2014). The target of OGI, MeGI, was neofunctionalized to act as a feminization factor via a lineage-specific genome-wide duplication (or paleo-polyploidization) at the K-Pg boundary (Akagi et al. 2020). The genus Asparagus underwent multiple genome-wide or partial genome duplication. A recent duplication that is specific to dioecious lin-eages, such as garden asparagus, derived the Y-encoded SOFF gene (Harkess et al. 2017). Shy Girl in the genus Actinidia also arose from an Actinidia-specific genomewide duplication (Huang et al. 2013), and resulted in novel expression patterns in the gynoecium (Akagi et al. 2018. Furthermore, in the family Salicaceae, multiple recent lineage-specific duplications of ARR17 formed inverted repeats (or small-RNA coding loci), resulting in at least three independent Y-encoded sex-determining loci (Müller et al. 2020). Extending this concept, frequent duplications (or translocations) of a transposable cassette putatively including the sex determinants in the genus Fragaria, triggered various lineage-specific sex determination systems, with the formation of neo-sex chromosomes (Tennessen et al. 2018). There are two potential mechanisms through which these duplications could trigger the establishment of sex determinants: (i) a novel SuF (in the "two-mutation" model) or a dominant suppressor, which is often essential for the establishment of dioecy, is thought to be hard to generate from a single (non-redundant) gene; and/or (ii) transition to dioecy may need dynamic changes in the regulatory path adjusted/specialized to each sex, represented by sexual dimorphisms, to which genome-wide duplication(s) may substantially contribute.
Regarding the first hypothesis, transitions other than those in the two-mutation model (e.g., from monoecy to dioecy) might not necessarily involve dominant suppressors but rather work only with a simple loss of function, as indicated by artificial evolution in Cucurbit (Boualem et al. 2015). Most of the identified paths to XY system dioecy include at least one dominant suppressor, as described, but the mechanisms remain unsolved. With the exception of genes that are highly redundant for their function (although these would often be generated by recent duplications), mutations that alter a single gene to act in a new distinct function (such as a dominant suppressor) would be evolutionarily disadvantageous, owing to the loss of the original function(s) (Flagel andWendel 2009, Wendel 2000). However, duplication also provides a chance to establish functional redundancy and facilitate neofunctionalization into a dominant suppressor. Consistent with this hypothesis, a genome-wide duplication in the ancient lineage of the genus Diospyros (or the family Ebenaceae), named Dd-α, derived a paralogous pair of HD-ZIP1 type homeobox genes. One of these duplicates underwent strong positive selection on specific mutated residues to be neofunctionalized into the current MeGI, which is a novel dominant suppressor of male function (Akagi et al. 2020). In terms of the second hypothesis, there are no clear genetic or physiological clues yet. In whole-genome duplication (or even simple gene duplication) events, the evolution of cisregulatory elements has made more rapid and substantial contributions to lineage-specific acquisition of representative traits than that of trans-acting elements in plants (Charoensawan et al. 2010, Roulin et al. 2013, as well as in animals (Carroll 2008, Lynch and Conery 2000, Wray et al. 2003. In actuality, numerous paralogous gene pairs ex-hibit fundamental differentiation in the expression patterns in flower buds, represented by the Dd-α in persimmon (Akagi et al. 2020). This finding is in contrast to limited gene pairs with strong positive selection potentially deriving neofunctionalized trans-functions. However, the functions of these differentially expressed genes, particularly involving sex expression, are not well defined. In the future, if a new determinant involving the establishment of dioecy (or even escape from dioecy) were to be derived via genome duplication, we may have to focus not only on the sex determining genes but also on a birds-eye view of "the whole" genome evolution, such as genome/epigenome rearrangements adjusted to the transitions into new sexual systems. Genome-wide surveys for changes in evolutionary pressures (or evolutionary rates: dN/dS) or expression patterns between ancestral and new sexual systems could also provide insights into the adaptations to new sexual systems. These would accelerate our understanding of the evolution of sex determination systems in plants, as well as providing future directions to harness various plant sexualities for breeding, via recent genome techniques, such as geneediting.

Evolution of crop sexes
Breeding Science Vol. 73 No. 2

Acknowledgments
We thank Marianne Fish, PhD, from Edanz (https:// jp.edanz.com/ac) for editing a draft of this manuscript. This work was supported by the Japan Science and Technology Agency through grants from PRESTO (JPMJPR20D1 T.A.), JSPS Grant-in-Aid for Transformative Research Areas (22H05172 and 22H05173 T.A.). Funding sources had no involvement in the study.