Evolutionary consequences of gender plasticity in genetically dimorphic breeding systems

Authors


Author for correspondence: Lynda F. Delph Tel: +1 812 855 1831 Fax: +1 812 855 6705 Email: ldelph@indiana.edu

Summary

A functional view of gender helps evolutionary biologists evaluate the mechanisms underlying breeding-system evolution. Evolutionary pathways from hermaphroditism to dioecy include the intermediate breeding systems of gynodioecy and androdioecy. These pathways start with the invasion of unisexual mutants, females or males, respectively, followed by alteration of the hermaphrodites to allocate more to the sexual function that the unisexuals lack. Eventually, hermaphrodites become unisexual and dioecy has evolved. Some species evolving along these pathways stop short of completing this second step, or even revert back from dioecy. We evaluate the hypothesis that gender plasticity is involved in these transitions to and from dioecy. Evidence from studies of subdioecious species that have evolved along the gynodioecy pathway suggests that gender plasticity occurs and stabilizes subdioecy by lowering the cost of producing seed. Factors influencing species evolving toward androdioecy, or reverting to androdioecy from dioecy, appear to be more varied and include reproductive assurance, herbivory and gender plasticity. In general, gender specialization appears to be favored in resource-poor environments regardless of which pathway is taken to dioecy.

Introduction

One goal of evolutionary biology is to understand the maintenance of genetic variation. Because breeding systems are highly tractable, both theoretically and empirically, the study of genetically based gender dimorphism has served an important role in our comprehension of how evolutionary forces influence genetic variation. We use the term ‘breeding system’ to refer to the various ways plants combine sex parts within and between individuals, for example cosexuality, gynodioecy, androdioecy and dioecy. Here we focus primarily on how lability of sex expression, or gender plasticity, might be involved in breeding-system evolution.

Resource-dependent gender plasticity is frequently observed in natural plant populations, but it is conspicuously absent from theoretical treatments of dimorphic breeding systems. Although the available data are limited, plasticity appears to be sufficiently widespread to merit attention as a factor in maintaining gender dimorphism. We describe gender plasticity in populations with genetically determined gender dimorphism and discuss how this plasticity may influence the maintenance of gender dimorphism, especially with regard to gynodioecy, androdioecy and dioecy.

Description of functional gender and breeding systems

An important concept in the study of breeding systems is functional gender (Lloyd, 1974, 1976, 1980; Lloyd & Bawa, 1984), which allows us to quantify the fitness that individuals or groups gain from male (pollen, seed siring) vs female (seed production) function. Flower morphology is the traditional criterion used to classify breeding systems, but the evolutionarily relevant criterion is fitness-based functional gender. Even Darwin (1877), who coined many breeding-system terms, grappled with the issue of function vs morphology in his description of a subdioecious species: ‘the most acute botanist, judging only by structure, would never have suspected that some of the bushes were in function exclusively males.’ Individuals in populations containing only one type of individual (hermaphrodites) are cosexual. The term ‘cosexual’ is not meant to be synonymous with hermaphrodite or bisexual, but instead refers to the fact that, on average, individuals in the population will gain half their fitness via male function and half via female function (Lloyd, 1980). Even if some individuals fail to produce seed or pollen in a given season or over a lifetime, the average functional gender is unimodal in these monomorphic populations.

In genetically dimorphic breeding systems, such as gynodioecy (females and hermaphrodites), androdioecy (males and hermaphrodites), or dioecy (males and females), one genetic morph is functionally more female while the other is more male. In a dioecious population, where each morph produces only pollen or seed, each morph gains 100% of its fitness via either male or female function. Because each morph provides half of the genes to each offspring (each seed has only one mother and one father), the sex ratio is expected to be 1 : 1. Indeed, sex ratios generally are 1 : 1 (Rottenberg, 1998) in the absence of factors biasing sex ratios, such as selfish genetic elements (Taylor, 1994) and differential mortality (Delph, 1999).

In gynodioecious and androdioecious species sex ratios can be highly variable, and the functional gender of hermaphrodites is influenced both by the pollen and seed production of hermaphrodites relative to unisexuals (Lewis, 1941; Lloyd, 1976), and by the frequency of unisexuals (functional gender is frequency dependent; Delph, 2003). If only hermaphrodites are present in a population their functional gender is 50% female, but as unisexuals increase in frequency, the functional gender of hermaphrodites becomes skewed towards the function that unisexuals lack – towards maleness in gynodioecious populations, and towards femaleness in androdioecious species. If the frequency of females or males reaches 0.5, the population will be effectively very close to dioecy, because hermaphrodites will be gaining most of their fitness through the function not provided by unisexuals. In the absence of sex ratio-biasing factors, we expect the frequency of unisexuals to be at least slightly lower than 0.5 if hermaphrodites gain some fitness through the function provided by unisexuals; hence the equilibrium sex ratio provides a clue to the functional gender of hermaphrodites.

In addition to sex ratio, the seed and pollen production of hermaphrodites (gender allocation) relative to unisexuals will influence their functional gender. For instance, in androdioecious populations, if hermaphrodites produce much less pollen than males, males will sire most seeds, and very few will be sired by hermaphrodites. Thus the functional gender of hermaphrodites will be highly female-biased. If hermaphrodites and males produce similar amounts of pollen, hermaphrodites are likely to fertilize some seeds, and have a less female-biased functional gender. The seed and pollen production of hermaphrodites relative to unisexuals influences the relative fitness of the two morphs, and therefore influences the sex ratio (Lloyd, 1975). Thus gender allocation and sex ratio interact to determine the functional gender.

Additional complexities arise when gender allocation of hermaphrodites is plastic, meaning that hermaphrodites can alter their production of seeds or pollen depending on environmental conditions. The role of gender plasticity in (genetically) dimorphic breeding systems has not been well studied from a theoretical perspective. However, such plasticity occurs in gender-dimorphic plants.

In this paper we discuss the role of gender plasticity in influencing the evolution of dimorphic breeding systems. Specifically, we discuss the possibility that resource-dependent plasticity may prevent the evolution of complete dioecy from gynodioecy by reducing the cost of producing seeds. Additionally, we discuss the breakdown of dioecy and the role of plasticity in systems resembling androdioecy. In the majority of species whose breeding system resembles androdioecy, plasticity is not present. When it is present, we ask whether plasticity may maintain gender dimorphism, by preventing the complete breakdown of dioecy into hermaphroditism. By discussing the role of plasticity in the evolution of breeding systems, we hope to illuminate the various ways in which gender dimorphism evolves, instead of trying to hammer square pegs (how a species is partitioning male and female function within and among individuals) into round holes (traditional breeding-system nomenclature).

The gynodioecy to dioecy pathway – why not go all the way?

One of the most important evolutionary pathways from hermaphroditism to dioecy is via gynodioecy (Webb, 1999; Weiblen et al., 2000). There are essentially two steps to this process, starting with a population containing hermaphrodites (Fig. 1). Typically, the first step occurs when a cytoplasmic (maternally inherited) mutation causing male sterility arises and spreads, making the population dimorphic for gender, as females now co-occur with hermaphrodites. Because of the maternal inheritance, such a mutation will spread if the individual containing the male-sterile cytoplasm has even only slightly higher seed fitness than one containing the nonsterile cytoplasm (Lewis, 1941). This breeding system is termed gynodioecy. Once the population is dimorphic for gender, the hermaphrodites will acquire more of their fitness via male than via female function. The more females there are, the more the hermaphrodites will achieve their fitness as males.

Figure 1.

The evolutionary pathways from cosexuality (shown here as hermaphroditism) to dioecy via gynodioecy and androdioecy are depicted as they are traditionally modeled. Arrow thickness indicates frequency of occurrence (thicker = more common). The evolution of dioecy is a two-step process in each case: invasion of a unisexual mutant, followed by a decrease and eventual elimination in the original hermaphrodites of the sex function provided by the unisexuals. Different species reduce investment in female or male function in different ways. For example, a decrease in female function along the gynodioecy pathway can occur by producing fewer pistillate flowers; by producing fewer seeds per fruit (depicted here); or by setting fewer flowers to fruit. A decrease in male function along the androdioecy pathway can occur by producing fewer staminate flowers; fewer anthers per flower; or smaller anthers containing less pollen (depicted here).

This skewed way of achieving fitness sets up the second step in the process of gynodioecy evolving into dioecy, as hermaphrodites are selected to allocate more to the sex function from which they gain the most fitness – they are selected to allocate more to making pollen than to making seeds (Charlesworth & Charlesworth, 1978; Maurice et al., 1994; Seger & Eckhart, 1996). If there is a trade-off between pollen and seed production, then this results in a decrease in allocation to female function, and the functional gender of the hermaphrodites is shifted more and more towards maleness until, in at least in some species, they become completely female sterile and are both morphologically and functionally purely male. The population is then composed of females and males, is termed dioecious, and usually has a sex ratio at, or close to, 1 : 1.

In this section we focus on species that are not strictly dioecious, but that are near the dioecy end of the gynodioecy–dioecy continuum, and ask how gender plasticity might be involved in maintaining fruit set on the pollen-producing morph in the population. A variety of terms have been used to describe the breeding system of such species, including leaky dioecy, polygamodioecy, near dioecy, or subdioecy. We will use the term subdioecy, but note that not all species to which we ascribe this term combine their sex parts in the same ways or make allocation decisions at the same developmental stages. The key point is that we will use this term to describe those species (or populations of species) far along the gynodioecy pathway to dioecy (Lloyd, 1976; Webb, 1979) in which at least some of the pollen-producing individuals produce seeds, but the females are constant in that they never make pollen. This latter feature is indicative of the gynodioecy pathway via the spread of male-sterile mutants. Theoretically, it occurs because pure females (those that do not produce any pollen) produce only outcrossed progeny, which do not suffer from inbreeding depression, whereas progeny from partially male-sterile individuals may have lower fitness if they suffer from inbreeding depression (Lloyd, 1975).

A plethora of terms have been used to describe the pollen-producing individuals in subdioecious species. In some studies, clear distinctions are made between a pollen-producing individual that produces fruit and one that does not (e.g. fruiting male or inconstant male vs male), and in others the distinction among individuals is less clear cut and only one term is used to describe the morph that is not a female, such as hermaphrodite, polleniferous morph, or male. We follow the fairly widespread convention of calling them males, in line with the focus on functional gender that Lloyd initiated and championed, in which the gender of a plant is quantitative rather than qualitative (Lloyd, 1974, 1976, 1980; Lloyd & Bawa, 1984). Moreover, we emphasize that we are discussing those species closer to the dioecy end of the gynodioecy–dioecy continuum.

We review studies on several subdioecious species that have evolved along the gynodioecy pathway (invasion of a male-sterile morph into a population containing hermaphrodites), and focus on the question of whether or not allocation to female function is a labile (plastic) trait in males (see Delph, 2003 for a review of this question in gynodioecious species). We then discuss how this resource-mediated gender plasticity might help to stabilize subdioecy. Two different lines of evidence are considered. First, if males growing in better environments (in space or time) are more likely to set fruit than when in environments of lower quality, then we take this as evidence for plasticity. Second, if manipulative experiments to alter resource levels result in an alteration of fruit set or production, then this is also taken as evidence for plasticity.

Species studied

The species are discussed in the order in which they were studied; the cases are summarized in Table 1.

Table 1.  Summary of characteristics of subdioecious species reviewed here
SpeciesSex ratioNo fruit production on some males in some years?Evidence for trade-off between pollen and seed production?Evidence for plasticity in fruit production by males?Sources
  1. Following the species name, the next two columns relate to whether or not the species should be characterized as subdioecious rather than gynodioecious or dioecious. The question whether there is a trade-off between pollen and seed production relates to theory suggesting that such a trade-off will promote a decrease in allocation to female function in the males; the question whether fruit production in males is plastic relates to the theory that plasticity will contribute to the stability of the breeding system at subdioecy rather than evolving into strict dioecy.

Fuchsia lycioides1 : 1 or slightly male-biasedYesNone availableLimited evidenceAtsatt & Rundel (1982)
Hebe subalpina1 : 1 or slightly male-biasedYesYesYesDelph (1990b, 1990c, 1993); Delph & Lloyd (1991)
Schiedea globosa1 : 1YesNoYesSakai & Weller (1991)
Ochradenus baccatus1 : 1 or male-biasedYesYesYesWolfe & Shmida (1995, 1997)
Wurmbea dioica1 : 1 or male-biasedYesYesYesBarrett (1992), Barrett et al. (1999); Ramsey & Vaughton (2001)
Dombeya ciliata1 : 1YesNone availableLimited evidenceHumeau et al. (1999, 2000)
Astilbe biternatavariable, but usually male-biasedYesYesYesOlson & Antonovics (2000), Olson (2001)

Fuchsia lycioides (Onagraceae)  The genus Fuchsia has breeding systems that range from hermaphroditism to gynodioecy to dioecy, and potentially even androdioecy (Godley, 1955, 1963; Arroyo & Raven, 1975; Arzimendi et al., 1996). Some species are not strictly dioecious, but exhibit fruit production on some but not all males, including the Chilean shrub F. lycioides (Atsatt & Rundel, 1982). The sex ratios of this species hover near 1 : 1, with a very slight excess of males in three of four sites. Males produce perfect flowers, in addition to the occasional style-less flower, but these often fail to set fruit. Evidence for plasticity in fruit production by the males is scanty, but the authors note that fruit production varies seasonally and males in more mesic southern sites set more fruit than those in drier northern sites (Atsatt & Rundel, 1982).

Hebe subalpina (Scrophulariaceae)  This genus of over 100 species also exhibits variation in breeding system, from hermaphroditism to dioecy. The gynodioecious pathway has been documented, with male sterility having arisen a minimum of four times within the genus (Delph, 1990a). Hebe subalpina is relatively close to dioecy even though the males make perfect flowers: the sex ratio is 1 : 1 or very slightly male-biased and males achieve, on average, at least 89% of their fitness via male function, although variation among males is high and continuous for fruit production (Delph, 1990a, 1990b; Delph & Lloyd, 1991).

Several approaches show that fruit production on the males is highly plastic. Vegetative cuttings taken from males growing in a natural population and planted in a common garden exhibited a several-fold increase in fruit production (Delph & Lloyd, 1991). Variation in flowering and fruiting among shoots within males was found to be dependent on shoot size and hierarchy (Delph, 1993). Hence manipulation of resources before fruit set was performed (1) by changing the hierarchy of shoots within plants to see if removal of a dominant shoot increased fruit set in lateral shoots; and (2) by defoliating shoots to reduce resource availability. Both types of resource alteration resulted in a change in fruit production of the shoots (Delph, 1990b, 1993). Furthermore, a trade-off between pollen and seed production within the males was shown by their response to defoliation: whereas defoliated shoots on females fruited but reduced growth, defoliated shoots on males grew but reduced fruit production (Delph, 1990b).

Schiedea globosa (Caryophyllaceae)  Breeding systems of species in the genus Schiedea range from hermaphroditism to dioecy (Weller et al., 1990). Sex ratios of S. globosa, a subdioecious species, vary from year to year but are typically near 1 : 1 (Sakai & Weller, 1991). Fruit production on individual males varied both within and among years, suggesting that this trait is plastic. Growing plants in the glasshouse also revealed plasticity: vegetative cuttings of individuals that did not produce fruit in the field did so in half the cases when grown in the resource-rich glasshouse. Additionally, Sakai & Weller (1991) found that males producing fruit tended to be larger than those that did not produce fruit, and tended to occur on the sunnier and wetter side of the slopes where resources were more abundant. A trade-off within males for pollen and seed production was not found, and may have been obscured by these habitat and size differences (van Noordwijk & de Jong, 1986). These results suggest that a least a subset of males are capable of setting fruit, but it is not known whether there is a discrete set of individuals that cannot.

Ochradenus baccatus (Resedaceae)  This species was once thought to be strictly dioecious, until detailed sex-expression studies by Wolfe & Shmida (1995, 1997) revealed that males in some populations produced fruit. Flowers on males vary from those without pistils to those with mature pistils. Males that did not produce any fruit (those without pistils) produced more stamen biomass than those with flowers containing pistils, suggesting a pollen and seed production trade-off. The sex ratio of 19 out of 24 populations did not differ significantly from 1 : 1, whereas it was male-biased in the remaining five, with one population having only 23% females. Hence some populations are subdioecious, while others are best described as gynodioecious. Plasticity in whether or not a male produced fruit was shown in two ways. First, size mattered, in that larger plants were more likely to produce fruit than smaller plants (Wolfe & Shmida, 1995); second, variation in fruit production along a geographical aridity gradient showed that it was more likely to occur where water was more available (Wolfe & Shmida, 1997).

Wurmbea dioica ssp. dioica (Colchicaceae)  The breeding systems of Wurmbea species range from hermaphroditism to strict dioecy, which evolved via the gynodioecy pathway (Barrett, 1992). Two studies on the sex ratio of W. dioica ssp. dioica populations revealed 1 : 1 or slightly male-biased sex ratios, in which the frequency of females stayed constant from year to year, but the frequencies of males that produced staminate vs perfect flowers varied (Barrett et al., 1999; Ramsey & Vaughton, 2001). One of these studies found support for a pollen vs seed trade-off within males, as pollen production was reduced in individuals producing functional pistils (Ramsey & Vaughton, 2001). Investment in female function by males is determined by the type of flower produced (staminate vs perfect) and therefore occurs before fruit set. The production of perfect flowers was found to be size dependent in the field, leading Barrett et al. (1999) to conclude that gender in males is conditionally dependent on resource status and is therefore plastic. Evidence for plasticity was strengthened by an experiment in which corms were excavated and grown in the glasshouse for 3 yr: 71% of males that had produced only staminate flowers in the field produced perfect flowers in the glasshouse (Ramsey & Vaughton, 2001). Furthermore, a recent study on W. dioica ssp. alba has shown that the percentage of pistillate flowers produced by males increases along a geographical gradient of increasing rainfall (Case & Barrett, 2004).

Dombeya ciliata (Sterculiaceae)  The genus Dombeya contains more than 300 tree species which range in breeding system from hermaphroditism to dioecy (Humeau et al., 1999). Populations of eight endemic Dombeya species on the island of La Réunion have 1 : 1 sex ratios. Five of these appear to be strictly dioecious, but three exhibit some fruit production by males and hence are subdioecious (Humeau et al., 1999). These subdioecious species occur in the lowlands, and populations of one species, D. ciliata, were dioecious at high altitude and subdioecious at low altitude. This, together with a decrease in the size of plants and leaf area with altitude for this species, suggests a plasticity in fruiting ability that is related to how stressful the habitat is (Humeau et al., 1999, 2000).

Astilbe biternata (Saxifragaceae) Olson & Antonovics (2000) describe this species as having ‘a breeding system that is close to dioecy’, refer to it as subdioecious, and document a genetic trade-off in male and female reproduction within males; but they note that they do not know if it is evolving towards gynodioecy or dioecy, as other species in the genus are all hermaphroditic. Most males growing in their natural populations typically fail to produce fruit; however, once again, almost all clonally propagated males grown in the glasshouse can be induced to produce fruit (Olson & Antonovics, 2000). The sex ratio of 22 populations was found to be typically male-biased, with an average female frequency of 38% (Olson, 2001). The proportion of males that produced fruit varied within populations from year to year, and fruit production was greater for males growing along roadsides as compared with those in the understory, which Olson attributes to plasticity. He also states ‘one future challenge will be to determine why some males … retain the ability to produce seeds.’

Plasticity of sex expression in males – a common feature of subdioecious species that evolved from gynodioecy

Three possible options exist for males of subdioecious species that evolved via the gynodioecy pathway: (1) be canalized to never make any flowers into fruit (i.e. become pure males); (2) be canalized to always set a given, albeit relatively low, proportion of flowers to fruit; or (3) be plastic and set a variable proportion of flowers to fruit depending on available resources (Delph, 1990c). Interestingly, dioecy exists either for the species (some populations) or the genus in six of the seven subdioecious species reviewed above, suggesting that strong barriers, genetic or otherwise, to the evolution of strict dioecy do not exist for these groups. Males of all of the subdioecious species reviewed appear to be plastic in their allocation to female function, although the timing of the allocation decisions does vary among species. Some species are plastic by altering which flowers contain functional pistils, whereas others alter the number of flowers they turn into fruit. The consistent plasticity of sex expression by the males of these species, all from different angiosperm families, is concordant with the hypothesis that plasticity reduces the strength of the trade-off between male and female function and promotes the retention of fruiting ability in these individuals. Thus, in species where males can allocate excess resources to female function under high resource conditions, there may be reduced selection for males to eliminate all fruit production, stabilizing subdioecy.

The hypothesis that lability of sex expression in the males would influence the stability of subdioecy was proposed by Delph (1990b, 1990c; Delph & Lloyd, 1991), who suggested that this plasticity should be incorporated into future models dealing with the evolution of dioecy from gynodioecy. Theoretical work that does not incorporate such plasticity shows that fitness differences between fruiting vs nonfruiting males of subdioecious species are often small, which allows for dioecy to evolve easily via drift or selection on other characters (Seger & Eckhart, 1996). Given that plastic sex expression appears to occur in several subdioecious species evolving from gynodioecy, as well as in gynodioecious species (Delph, 2003), a call for models on the evolution of dioecy that incorporate such plasticity still seems warranted.

Many of the authors of the work reviewed here noted the plasticity in males and its possible effect on breeding-system evolution. For example, Wolfe & Shmida (1997) conclude that conditional sex expression in males affords a selective advantage and will remain ‘a stable feature of populations of O. baccatus.’Barrett et al. (1999) discuss the trade-off between pollen and seed production with regard to plasticity, by pointing out that while greater allocation to pollen should reduce ovule production, if seed is produced only when resource status is high and therefore does not compromise male fitness, then plasticity may ‘hinder the evolution of dioecy’. Ramsey & Vaughton (2001) explicitly undertook their study to gain insight into the stability of subdioecy, and concur that if the plasticity lessens the trade-off between pollen and seed production and ‘seed production augments plant fitness’, then it could be considered adaptive. Lastly, Olson & Antonovics (2000) note that the coexistence of males that do and do not produce fruit may be stable. We concur that plasticity appears to be an important aspect in stabilizing subdioecy because it reduces the cost to males of producing fruit.

Androdioecy – a pathway to dioecy

Historically, androdioecy has been considered an alternative pathway from hermaphroditism to dioecy (Bawa, 1980). This requires invasion of a female-sterile (male) mutant, creating an androdioecious population, followed by a gradual or abrupt reduction of pollen production by hermaphrodites such that they become purely female (Fig. 1). However, the first step does not occur frequently: a female-sterility mutation is unlikely to invade a hermaphroditic population (Lloyd, 1975; Charlesworth, 1984). Why are female-sterile mutations unlikely to invade, while male-sterile mutations are commonly seen? One reason concerns the genetics of sex determination, and the fact that cytoplasmic genes are generally inherited maternally (Corriveau & Coleman, 1988). Male sterility commonly arises via cytoplasmic mutations – the genes that eliminate pollen are in a genome that is usually not inherited through pollen (Charlesworth & Laporte, 1998; Dudle et al., 2001). In most plants there is no genome that is exclusively inherited paternally, so a mutation cannot cause female sterility without reducing its own fitness. A biparentally inherited female-sterility mutation can invade only if its pollen fitness increased to make up for the loss of seed fitness. Diminishing fitness gains make this difficult to achieve. Because androdioecy was considered unlikely to serve as a pathway to dioecy, requirements for the second part of this pathway, when hermaphrodites become pure females, have not been studied. In fact, because there are so few androdioecious (or subandrodioecious) species, most of the phenomena that are well known in gynodioecious species have not been addressed in androdioecious species.

Despite theoretical predictions that androdioecy should not evolve from hermaphroditism, it may have done so on several occasions (discussed by Wolf & Takebayashi, 2004: Phillyrea sp. (Oleaceae), Vassiliadis et al., 2000; Neobuxbaumia mezcalensis (Cactaceae), Valiente-Banuet et al. (1997); Sagittaria lancifolia ssp. lancifolia (Alismataceae), Muenchow (1998); Fraxinus (Oleaceae), Wallander, 2001). Additionally, phylogenetic evidence suggests that dioecy subsequently evolved from androdioecy in the genus Fraxinus (Wallander, 2001).

The genus Fraxinus is a group of wind- and insect-pollinated plants in which 10 species are morphologically androdioecious (Wallander, 2001). Hermaphroditism and dioecy are also found in the genus, and Wallander's phylogenetic work suggests that in one clade dioecy was derived from androdioecy, which was derived from hermaphroditism. This analysis, however, assumes that the morphologically androdioecious species are actually functionally androdioecious, rather than functionally dioecious, or that they at least passed through an androdioecious stage. Most populations of the three species that have been studied have 1 : 1 sex ratios (Fraxinus ornus, Dommée et al., 1999; Wallander, 2001; Verdúet al., 2004; Verdú, 2004; Fraxinus lanuginosa, Ishida & Hiura, 2002; Fraxinus longicuspis, Wallander, 2001). Thus, although the seed-producing morph also produces functional pollen (Ishida & Hiura, 1998; Dommée et al., 1999), it seems unlikely that they are acquiring much fitness via male function (Verdúet al., 2004), and it might be more appropriate to consider these species as nearly dioecious (here we refer to these individuals as females, following the protocol of using terminology that highlights functional gender rather than morphology). The exception is F. lanuginosa, in which low-density populations have male frequencies as low as 0.11, and females have been shown to sire some seeds (Ishida & Hiura, 2002), suggesting that some populations may be androdioecious, and that dioecy could be derived from androdioecy. The role of genetics and plasticity in sex expression is not clear in these species, but there is some evidence for genetic sex determination in F. ornus (Dommée et al., 1999; Verdú, 2004). However, a small number of individuals changed from one gender to the other in F. lanuginosa (Ishida & Hiura, 2002). Additional research is necessary to uncover the role of genetics and gender plasticity in the breeding-system evolution of this clade.

In cases discussed in the previous section in which subdioecy evolved from gynodioecy, the males were the morphologically plastic sex (sometimes producing fruit, sometimes not). We might expect that females would be the plastic morph when breeding systems that are near dioecy have evolved via the androdioecy pathway. This does not appear to be the case. In Fraxinus, females always produce functional pollen. There is likely to be high variation in the actual fitness gained through pollen, perhaps only gaining substantial male fitness in newly founded populations that lack males (Dommée et al., 1999), but no morphological plasticity. Perhaps this is because pollen is a reward for pollinators. Like the gynodioecious examples, Fraxinus moved towards dioecy and gender specialization in resource-poor environments, mainly because of an increase in the frequency of the specialized gender (males). A balance between gender specialization and reproductive assurance may be the key to maintaining what is perhaps a mix of breeding systems in these Fraxinus species.

There are several species that are not strictly androdioecious, but possess characteristics of androdioecious systems. Studies of these nearly androdioecious species contribute to our understanding of the androdioecious breeding system. These species range from the dioecy end of the spectrum, such as Fraxinus, to the hermaphroditic end, at which males produce a low number of seeds. These examples also suggest that gender specialization may be favored in harsh or resource-poor environments.

Sagittaria lancifolia ssp. lancifolia (Alismataceae) is closer to the hermaphroditic end of the spectrum. The breeding system varies from population to population, such that some populations appear to be monoecious (the likely ancestral breeding system), while others resemble androdioecy (Muenchow, 1998). There are two genetically determined morphs: one morph produces 35% pistillate flowers, while the other produces 0–2% pistillate flowers and is clearly functioning almost entirely through male function. The predominantly male morph made up only 16% of the population, and the other morph clearly could and did function through both male and female function (Muenchow, 1998).

A closely related species, S. latifolia, has a mix of monoecious and dioecious populations. Several populations studied by Sarkissian et al. (2001) consist of individuals from both types of population, giving the appearance of androdioecy. However, this is probably the result of migration between monoecious and dioecious populations, rather than the result of natural selection. Sarkissian et al. (2001) suggest that the same may be true in S. lancifolia. There are no reports of dioecious S. lancifolia populations, so subandrodioecious populations are not likely to be the result of migration between dioecious and monoecious populations. Additionally, there is an apparent selective advantage to males in S. lancifolia (Muenchow, 1998). Although there is no purely male morph in this species, and study of additional populations is needed, current evidence suggests that some populations of S. lancifolia have a breeding system that resembles androdioecy.

There is clearly a trade-off between male and female function in S. lancifolia, because staminate flowers in males replace pistillate flowers in hermaphrodites, but there is no evidence of resource-based gender plasticity in either sex. Rather, herbivores that primarily destroy the upper, staminate flowers in hermaphrodites appear to allow the coexistence of males, which produce staminate flowers in lower whorls that are not destroyed by herbivores. Populations without strong herbivore pressure lack males. Thus there may be selection pressure in populations with herbivores to reduce fruit production in the predominantly male morph, and strict androdioecy or dioecy may eventually evolve (Muenchow, 1998). This is similar to the gynodioecious cases discussed in the previous section, in which males make both pollen and seeds in the benign environment, but specialize more towards one gamete type (pollen) in the harsh environment. Here the trade-off between the two sex functions is minimized when herbivory is low. As in Fraxinus, however, the functional gender of hermaphrodites varies in this species: it is 50 : 50 in the monoecious populations, and female-skewed in the subandrodioecious (nearly androdioecious) populations.

Androdioecy – the breakdown of dioecy

Other examples of androdioecy and subandrodioecy are likely to represent the breakdown of dioecy rather than a pathway to dioecy (discussed by Pannell, 2002a and Wolf & Takebayashi, 2004). The first step is the invasion into a dioecious population of self-compatible individuals that produce both seed and pollen; the second is for these individuals to replace females, but allow males to persist. Males will persist only if the other morph has a female-skewed allocation pattern, producing just a small amount of pollen. When female reproduction is limited by pollen availability, individuals that are able to produce a small amount of pollen may replace females such that androdioecy evolves (Wolf & Takebayashi, 2004). Additionally, metapopulation dynamics may favor the replacement of females by these seed-and-pollen-producing individuals, because the latter can found new populations that males may subsequently invade (Pannell, 2000).

Most of the species derived from dioecy do not exhibit any gender plasticity (Sassaman & Weeks, 1993; Haag & Kimble, 2000; Wolf et al., 2001). Mercurialis annua (Euphorbiaceae) is the exception, in which plasticity may be important in the maintenance of gender dimorphism. Mercurialis annua is a self-compatible, annual plant that exhibits a great deal of breeding-system variation and ploidy variation across its range (Pannell, 1997a). Diploid populations are dioecious (the ancestral breeding system), while polyploid populations are mostly monoecious (Durand & Durand, 1992). Hexaploid populations in the western Mediterranean region contain plants with two genetically determined strategies, one of which is canalized to always produce both types of flowers (staminate and pistillate), and one that is plastic in that it undergoes a distinct developmental switch between two forms, producing only staminate flowers (male phase) or both staminate and pistillate flowers (hermaphrodite phase) (Pannell, 1997b). In dense populations, where resources for seed production may be limiting, the frequency of plants in the male phase increases and the canalized morph produces a higher proportion of pistillate flowers (Pannell, 1997a, 1997b). Thus dense populations tend towards the dioecious end of the spectrum, while less-populated populations, in which mates may be limiting, tend towards the hermaphroditic end of the spectrum.

Plasticity is a contentious issue in the androdioecy literature (Pannell, 2002b). Can a species be considered androdioecious if the male can switch from making both pollen and seed to making only pollen? Plasticity of this sort is not part of the traditional definition of androdioecy, which is defined as having individuals that are genetically determined to be pure males (just as in gynodioecy there are individuals that are genetically determined to be pure females; Fig. 1). Some M. annua individuals gain fitness only through male function throughout their lifetime, so their lifetime functional gender is male. However, from a population-genetic perspective, it is the fitness and functional gender of the morphs across both gender phases that will determine the persistence of each gender morph, and is the relevant factor to define the breeding system. If the plastic morph takes the male phase most of the time, it will clearly gain its fitness primarily through male function and populations will be essentially androdioecious. However, if the plastic morph is usually in the hermaphrodite phase, it may be more suitable to consider the population cosexual. A better understanding of the breeding system of M. annua would result from a comparison of the proportion of fitness that the two morphs (plastic and canalized) gain through male function in the field. This is analogous to looking at the relative seed fitness of the two morphs in gynodioecious or subdioecious species. Because the plastic morph can produce seeds when resources are abundant and mates are limited, conditions for the maintenance of this complex dimorphism should be less stringent than for the maintenance of pure males, and plasticity may be important in stabilizing this dimorphism.

Conclusions

Are there any common patterns between the species with breeding systems resembling androdioecy, and those that have evolved subdioecy from gynodioecy? Several well studied species that have evolved from gynodioecy exhibit resource-based plasticity in the fruit production of males (derived from the original hermaphrodites), such that males produce fruit only when resources are abundant. Likewise, in species whose breeding system resembles androdioecy, gender specialization is favored when resources are limited or the environment is harsh. Resource-poor populations moved towards dioecy in both gynodioecious and androdioecious systems.

As males are the plastic morph in subdioecious populations that evolved via gynodioecy, we might expect females to be the plastic morph in systems involving androdioecy. No clear pattern exists. In Fraxinus, females always produce pollen, even when they are unlikely to be gaining much fitness through male function. This could be because pollen production is less costly than fruit production (Lloyd, 1984), lessening the benefits of plasticity for pollen production, or it may be because pollen production aids in pollinator attraction (Dommée et al., 1999). However, the pattern of gender specialization in resource-poor environments is consistent. The trade-off between the two sex functions may indeed be minimized when resource levels are high, but maximized when resource levels are low, favoring a strategy with plastic gender expression.

We conclude that there is a need for rigorous theoretical treatment of how gender plasticity influences both the evolution of dioecy and the maintenance of systems with genetically based gender dimorphism, and whether selection acts differently when the plastic morph is the predominantly male morph rather than the predominantly female morph. While gender plasticity makes it difficult to assign a particular term for characterizing the breeding systems of some species, documenting the plasticity illuminates the evolutionary pathway the species has taken, and helps to clarify how trade-offs between the two sex functions drive breeding-system evolution.

Acknowledgements

We are grateful for helpful comments from M. Bailey, S. Barrett, J. Busch, M. Olson, J. Pannell, J. Steven, N. Takebayashi, M. Verdú, S. Weller and L. Wolfe, and appreciate the invitation to contribute to this issue. This work was supported a National Science Foundation grant to L.D. (DEB-0075318) and a National Science Foundation EPSCoR grant to D.W. (EPS-0092040).

Ancillary