What is functional androdioecy?

Androdioecy denotes the co-occurrence of males and hermaphrodites in a sexually reproducing population. As Darwin (1877) first noted, the breeding system is exceedingly rare, and theoretical studies have shown why this is so (Charlesworth 1984; Charlesworth & Charlesworth 1978; Lloyd 1975). In particular, males can invade a hermaphroditic population only under the restrictive condition that they sire more than twice as many successful offspring as do the hermaphrodites. Moreover, unlike gynodioecy, androdioecy cannot evolve from hermaphroditism as a result of selection for inbreeding avoidance, because selfing immediately makes ovules unavailable for males to fertilize. Seemingly in agreement with theory, Charlesworth (1984) found that there was no convincing example of the breeding system in nature. Instead, she concluded that populations with morphological male and hermaphrodite individuals, which have often been called androdioecious, were cryptically dioecious, with a 1 : 1 sex ratio and hermaphrodites that function only as females (see also Mayer & Charlesworth 1991). Although functional androdioecy has been confirmed for several species of plants and animals in the past couple of decades (reviewed in Pannell 2002), the breeding system must still be regarded as very rare. Given this rarity, and because cryptically dioecious species have so often been misconstrued as androdioecious, functional interpretations of putatively androdioecious populations have been understandably cautious. One issue that has been raised concerns the extent to which sex or gender (see below) in androdioecious populations ought to have a genetic basis. Functional males and hermaphrodites co-occur, for example, in populations of the herb Mercurialis annua (Durand & Durand 1992; Pannell 1997b). However, because sex in M. annua is determined by genetic and environmental factors (Pannell 1997a), several authors have suggested that this species should strictly not be regarded as functionally androdioecious (e.g. Dommée et al. 1999; Ishida & Hiura 1998; Webb 1999; Wolf, Rieseberg & Spencer 1997). The issue is important as it brings into focus the relative weight given to functional (ultimate) vs mechanistic (proximal) causes in naming and understanding repro-ductive strategies. A strong case can be made for favouring the former of these two alternatives. Indeed, sex allocation theory is founded on interpretations of gender that are based on prospective and realized reproductive success through male and female functions, irrespective of how allocation patterns are determined mechanistically (Charnov 1982). A definition of functional androdioecy based on mechanistic rather than purely functional considerations is therefore retrogressive. This note aims to clarify a functional basis for the definition of androdioecy, with particular reference to the questionable relevance of the sex determination mechanism. A key point is that populations should be regarded as functionally androdioecious only if they comprise two classes of individual: one in which all fitness is gained through male function, the other in which fitness is gained substantially through both male and female functions. How these two strategies are determined mechanistically is of secondary importance. Consistent with the notion of a stable polymorphism, individuals in the male class of an androdioecious population should not differ in average fitness from those in the hermaphroditic class when the population is at equilibrium. I first illustrate the significance of these points by highlighting the apparent similarity, but functional difference, between androdioecy and a particular type of size-dependent sex allocation, where individuals are male when small and hermaphrodite as they grow. I then argue against a definition of androdioecy that gives primacy to the mode of sex determination. While I restrict this note to a discussion of androdioecy, it illustrates the general importance of functional, as opposed to mechanistic, interpretations of gender. Small individuals are often found to emphasize their male function, with larger plants being more female (Lloyd & Bawa 1984). Where this size-dependent sex allocation is extreme, gender may fall into two distinct categories, with small plants being fully male and larger plants expressing male and female function. Such cases of gender diphasy, exemplified by Arisaema dracontium (Clay 1993), may easily be confused with androdioecy. However, they do not represent a dimorphic strategy, and each individual is expected to make an equal genetic contribution over its lifetime through each of the two sexual functions. A related issue concerns the importance of distinguishing between gender switching, or ‘gender diphasy’ (Lloyd & Bawa 1984; Schlessman & Lovett Doust 1988), and functional androdioecy. Presumably because of the heavy reproductive burden of fruit production, hermaphrodites that fruit one year may be incapable of producing fruits the next, and may thus function solely as males for a season. Such populations, surveyed only once, may appear to be androdioecious. The herbaceous alpine perennial Lloydia serotina, for instance, was first suspected as being androdioecious (Jones & Gliddon 1999), but subsequent study has shown that individuals possess only a single strategy, with switches in gender from year to year (Manicacci & Despres 2001). A concern over gender switching may have fuelled the idea that sex or gender in functionally androdioecious populations should be determined genetically. Certainly, populations in which males and functional hermaphrodites differ at one or more sex-determining loci are likely to be good examples of functional androdioecy, particularly as sexual diphasy and gender switching are at once ruled out as an explanation of observed phenotypic variation in any one reproductive season. However, the requirement for androdioecious populations specifically to be genotypically dimorphic is unhelpful because it does not emerge clearly from sex-allocation theory. The theory of sex allocation attempts to explain the phenotypic mix of allocation strategies found in populations in terms of evolutionarily stable strategies (ESS) (Charnov 1982). This includes the ESS sex ratio in dioecious species (Fisher 1930; Hamilton 1967), the ESS allocation of limited resources to male and female functions in hermaphrodites (Charnov, Maynard Smith & Bull 1976), and the optimum sequence and timing of sex reversals in sequential hermaphrodites in terms of the age and size of individuals (Charnov 1982). Whether sex is environmentally or genetically determined is subsumed under the more general question of whether it is evolutionarily optimal for sex to be determined at the time of conception and zygote formation, or at some later time in development (Charnov & Bull 1977). The timing of allocation decisions in androdioecious populations, and thus whether sex is determined genetically or environmentally, raises interesting questions and may be an important component in explaining the evolution or stability of the breeding system in certain cases. However, understanding the mode and timing of sex determination is logically secondary to the primary task of describing a population's breeding system. Models of androdioecy have typically been coined in terms of contrasting sexual strategies, with little attention paid to how these strategies are determined (Charlesworth 1984; Charlesworth & Charlesworth 1978; Charnov et al. 1976; Lloyd 1975; Maurice & Fleming 1995; but see Vassiliadis et al. 2000). Nevertheless, they have at times been discussed in the language of genetics. For example, Lloyd (1975) noted that if ‘androdioecy [is] controlled by nuclear genes, the effects [of population parameters] on the equilibrium sex ratio do not depend on the precise mode of inheritance.’ It is important to note that this does not presume sex must be determined by nuclear genes. Rather, the reference to genetic control simply emphasizes that if the reproductive strategy in a population were transmitted uniparentally rather than biparentally, the model's predictions of the sex ratio would no longer hold. Lloyd (1975) did not pursue this issue further, noting only that ‘[a]ndrodioecy, unlike gynodioecy, cannot be maintained by cytoplasmic inheritance.’ His statement here assumes that cytoplasmic genes are maternally inherited. The transmission of cytoplasmic genes through pollen is not unknown (Reboud & Zeyl 1994), and in such cases androdioecy might indeed be maintained by cytoplasmic inheritance, at least in principle. Similarly, segregation distorters linked to a nuclear male-determining gene could conceivably maintain androdioecy (Vassiliadis et al. 2000), although no such system has yet been described. In both these cases, Lloyd's sex ratio predictions for androdioecious populations would not apply. However, were such an example to be found in nature, androdioecy would be the appropriate term (just as populations with cytoplasmic inheritance of male sterility are gynodioecious). It is worth commenting briefly on a distinction that has sometimes been made between ‘sex’ and ‘gender’. In arguing for the primacy of a genetic mode of sex determination in defining androdioecy, Wolf et al. (1997) have interpreted Lloyd as suggesting that whereas the sex of a plant is genetically determined, gender ‘is determined by the other plants in the population or by the environment’. In fact, Lloyd (e.g. page 258 of Lloyd & Bawa 1984) distinguished between these two terms in a rather different sense that makes no reference to the mechanism of sex determination: ‘we […] consider not only the “sex” of a plant in the strict sense of the production and disposition of its gametes, but more broadly its “gender”, its maleness or femaleness as a parent of adults of the next generation.’ This distinction emphasizes that reproductive success gained through male and female functions depends as much on the relative production of seeds vs pollen (a plant's sex or ‘phenotypic gender’sensuLloyd 1980), as on the fate of those seeds or pollen grains (its ‘functional gender’sensuLloyd 1980). Lloyd (1980, page 107) referred to androdioecy in particular as a gender dimorphism in which ‘there are two distinguishable morphological classes of plants …’ (my italics). In keeping with the conventions of breeding-system terminology, and in accordance with usage adopted in theoretical discussions, a description of functional androdioecy should be based on the following criteria. Androdioecious populations should display a clear sexual or gender dimorphism in which there are two distinguishable classes of plants: males, which lack female function; and hermaphrodites which, as a class, make a substantial genetic contribution through both sexual functions. Populations displaying a uniform gradation in sex allocation from pure maleness through to female-biased hermaphroditism should not be described as androdioecious. Populations in which morphological hermaphrodites fail to sire seeds through their male function should be viewed as cryptically dioecious rather than androdioecious (Mayer & Charlesworth 1991). Ideally, paternity analysis could confirm the male function of hermaphrodites (Vassiliadis et al. 2002). Populations in which males co-occur with females or morphological hermaphrodites that disperse some pollen but sire only a limited number of progeny might better be viewed as examples of ‘leaky dioecy’ or subdioecy (Baker & Cox 1984; Lepart & Dommée 1992). Bimodality in sex expression is a necessary, but insufficient, condition for the definition of androdioecy. This is because functional gender, as defined by Lloyd (1980), characterizes individuals in terms of the proportion of their genes that they contribute through their female (or male) function, without reference to their total fitness. This is useful if we wish to determine the extent to which individuals emphasize one sex function over the other in different contexts, for example as a function of age or size. However, for the maintenance of polymorphic strategies, the absolute genetic contributions made by members of each class are important (Sarkissian, Barrett & Harder 2001). At equilibrium in a polymorphic population, we expect individuals within each phenotypic class, on average, to possess equal fitness. Thus an androdioecious population at equilibrium ought to comprise males and hermaphrodites that make, on average, equal genetic contributions to the next generation. A positive correlation between quantitative gender in a population, expressed as prospective femaleness, and total reproductive success, for example, would therefore be inconsistent with functional androdioecy and may rather reflect size-dependent sex allocation. I thank Deborah Charlesworth and an anonymous referee for comments on an earlier version of the manuscript.