Heredity 78 (1997) 50—56 Received 19 January 1996 Mixed genetic and environmental sex determination in an androdioecious population of Mercurialls annua JOHN PANNELL* Department of P/ant Sciences, University of Oxford, South Parks Road, Oxford OX1 3R8, U.K. Mercurialisannua is a ruderal of pan-European distribution. In southern Spain, southern Portugal and northern Morocco, males coexist with monoecious (cosex) individuals at frequen- cies which vary between zero and about 30 per cent. Here, I report the results of two glasshouse experiments which aimed to characterize the mode of sex determination in one such androdioecious population. In a breeding experiment, cosexes isolated from males produced only cosexual progeny, whereas those allowed to mate with males produced both cosexual and male offspring. The results are consistent with a single-locus model of sex inheritance, with maleness determined by a dominant allele. In a density experiment, the frequency of males correlated positively with density, confirming field observations. Differential mortality or germination between the sexes was excluded by experimental design, so that the result supports the hypothesis of density-dependent gender choice. A negative correlation between density and the proportion of 'late cosexes' (males which turned cosexual late in their development), and the lack of any differences in the proportion of pure (unchanged) cosexes between density, suggest that only males were capable of sex change. This interpretation is consistent with the results of the breeding experiment, in which late cosexes occurred only in the offspring of cosexes allowed to mate with males. I argue that these findings help to explain the maintenance of androdioecy in M annua. keyworcis: androdloecy,dioecy, reproductiveassurance, sex change, sex choice, sexual dimorphism. Introduction In gynodioecious species (i.e. where females coexist with cosexes), male sterility is frequently the Insexually dimorphic (e.g. dioecious, androdioe- outcome of interactions between nuclear and mater- cious or gynodioecious) plant species, gender may nally inherited cytoplasmic genes (e.g. Gouyon & be determined genetically, environmentally, or by a Couvet, 1987; Durand & Durand, 1991a; Van genotype—environment interaction (Charnov, 1982; Damme, 1991), whereas female sterility in andro- Lloyd & Bawa, 1984; Meagher, 1988). Some genetic dioecious species (i.e. where males coexist with systems of sex determination in dioecious species cosexes) is most likely to be controlled solely by appear to be chromosomal in nature (reviewed in nuclear genes (paternal inheritance of cytoplasmic Westergaard, 1958; Meagher, 1988). Others involve genes is uncommon in Angiosperms; Reboud & the expression of genes at one or more loci on Zeyl, 1994). In the androdioecious notostracan clam chromosomes which cannot be identified cytologi- shrimp Eulimnadia texana, sex is inherited at a single cally. In diploid, dioecious populations of Mercurialis locus, with a recessive allele determining maleness annua, for example, sex is determined by the inter- (Sassaman & Weeks, 1993). However, this study of active effects of dominant or recessive sex alleles at Mercurialis annua presents the first attempt to eluci- three independently segregating loci (Irish & date the genetic system of sex determination for an Nelson, 1989; Durand & Durand, 1991b; see below). androdioecious plant species. Environmental factors are also known to influence *Present address: Department of Botany, University of Toronto, sex ratios in several plant species (Lloyd & Bawa, 25WillcocksStreet, Toronto, Ont., Canada M55 3B2. E-mail: 1984; Meagher, 1988), with males most typically [email protected],ca tending to be over-represented in resource-limited 50 1997 The Genetical Society of Great Britain. SEX DETERMINATION IN MERCURIAL/S ANNUA 51 or less favourable environments (reviewed in loci, A, Bi and B2, segregating independently in the Freeman & Vitale, 1985; Meagher, 1988). This has nucleus. For an individual to be phenotypically male, been attributed both to lower mortality rates of it must possess a dominant A allele with at least one males than females under less favourable conditions, dominant B allele (i.e. either Bi or B2). An indivi- as well as to environmental influences on sex expres- dual with a dominant A allele but homozygous sion itself (Sakai & Weller, 1991). In androdioecious recessive at both B-loci is phenotypically female, as populations of M. annua (see below), sex appears to are all genotypes homozygous recessive at the be determined to a large extent genetically; seed A-locus. Thus, for the A, Bi and B2 loci, respect- sampled from populations lacking males produced ively, genotypes (Al-, B1/-, -I-) and (Al-, -I-, B2/-) are all cosex progeny, whereas seed from androdioe- phenotypically male, whereas genotypes (Al-,bilbi, cious populations produced both male and cosex b2/b2) and (a/a, -I-, -I-) are phenotypically female progeny when grown under uniform conditions (Durand & Durand, 1991b). (Pannell, 1995). However, male frequency correlated In hexaploid androdioecious populations of M. significantly with stand density in two populations in annua, males have the same inflorescence morph- the field, suggesting that there may also be an ology as those of the diploid populations (they have environmental cOmponent governing sex expression sessile staminate flowers arranged in tight spiral in the species. clusters along erect axillary peduncles). Hexaploid In this paper, I attempt to resolve the environ- cosexes are similar in morphology to diploid mental and genetic components of sex determina- females, except that their subsessile axillary pistillate tion in M. annua. In particular, I ask whether gender flowers are surrounded basally by a tight spiral in androdioecious M. annua is inherited by a simple cluster of staminate flowers. Occasionally, cosexes 'active-Y' mechanism (Meagher, 1988), such that are found with male pedunculate inflorescences in maleness is determined by the expression of a single their lower leaf axils; the significance of these 'late or several linked dominant alleles (as has been cosexes' is interpreted in the discussion below. In demonstrated for polyploid species of Rumex (Love both the diploids and the polyploids, plants produce & Kapoor, 1967)), or whether a more complex flowers and fruits indeterminately in their leaf axils system of inheritance is involved. I also aim to test usually from the age of two weeks after germination the hypothesis experimentally that stand density until they die at the end of the growing season affects male frequency. (Pannell, 1995, 1997b). Although it is likely that floral development and differentiation are governed Thespecies by similar physiological gradients in both diploids and polyploids (Durand & Durand, 1991b), nothing Mercurialisannua L. s.l. (Euphorbiaceae) is a is yet known about the genetics of sex determination ruderal colonizer of pan-European distribution. in polyploid populations of M annua —neither Diploid (2n =16)populations are exclusively dioe- whether the same (duplicated) sex-determining loci cious and are widespread across central and western are responsible, nor how these may be expressed. Europe. Polyploids occur around the western Medi- terranean in southern Europe and north Africa and Materialsand methods are largely monoecious (Durand, 1963; Durand & Seedwas collected for both the density and the Durand, 1992). In the genus Mercurialis, all seven breeding experiment in the spring of 1994 from a species are dioecious (apart from M. annua polyp- large androdioecious population of M. annua in the bids), and cosexuality would thus appear to be a precincts of the Pabellón de Cuba in Sevilla, south- derived trait (Tutin et al., 1968; Valdes et a!., 1987). ern Spain. The frequency of males in the quadrats In southern Spain, southern Portugal and northern from which seed was taken ranged between about 10 Morocco, hexaploid androdioecious populations of per cent and 20 per cent (Pannell, 1995, 1997a). M. annua are widespread, with males co-occurring Plants were harvested and allowed to dry at room with monoecious plants (cosexes) at frequencies of temperature to disperse their seeds. The seed was less than about 30 per cent (Pannell, 1995, 1997a,b). thoroughly mixed before subsampling for the The physiology and genetic system of sex determi- nation in diploid dioecious M annua (i.e. where only experiments. males and females occur) have been characterized by Durand and coworkers (Louis et a!., 1990; Densityexperiment Durand & Durand, 1991b). They have concluded Inlate summer 1995, 12 seeds were sown in potting that dioecy is controlled by three diallelic nuclear compost with a 1 cm covering of soil in each of 375 The Genetical Society of Great Britain, Heredity, 78, 50—56. 52 J. PANNELL pots 8 cm in diameter. Two weeks later, when seed- were harvested, dried at room temperature, and lings had emerged and were at the cotyledon stage, their seeds were collected and stored. In the their numbers were thinned to leave densities as summer of 1995, the seeds produced by plants in follows: 200 pots with one plant; 100 pots with 2 each growth box were sown out into trays in potting plants; 50 pots with 4 plants; and 25 pots with 8 compost, arranged randomly on the glasshouse plants. (Increased numbers of pots were allocated to bench, and were left to germinate and grow. At 8 the lower density treatments so that sex ratios could weeks after sowing, plants were