Ecology, 82(9), 2001, pp. 2601±2616 q 2001 by the Ecological Society of America ECOLOGICAL DIFFERENTIATION OF COMBINED AND SEPARATE SEXES OF WURMBEA DIOICA (COLCHICACEAE) IN SYMPATRY ANDREA L. CASE1 AND SPENCER C. H. BARRETT Department of Botany, University of Toronto, 25 Willcocks Street, Toronto, Ontario, Canada M5S 3B2 Abstract. The evolution and maintenance of combined vs. separate sexes in ¯owering plants is in¯uenced by both ecological and genetic factors; variation in resources, partic- ularly moisture availability, is thought to play a role in selection for gender dimorphism in some groups. We investigated the density, distribution, biomass allocation, and physi- ology of sympatric monomorphic (cosexual) and dimorphic (female and male) populations of Wurmbea dioica in relation to soil moisture on the Darling Escarpment in southwestern Australia. Populations with monomorphic vs. dimorphic sexual systems segregated into wet vs. dry microsites, respectively, and biomass allocation patterns and physiological traits re¯ected differences in water availability, despite similarities in total ramet biomass between the sexual systems. Unisexuals ¯owered earlier at lower density, and they allocated sig- ni®cantly more biomass below ground to roots and corms than did cosexuals, which al- located more biomass above ground to leaves, stems, and ¯owers. Females, males, and cosexuals produced similar numbers of ¯owers per ramet, but unisexuals produced more ramets than cosexuals, increasing the total number of ¯owers per genet. Contrary to ex- pectation, cosexuals had signi®cantly higher (more positive) leaf carbon isotope ratios and lower leaf nitrogen content than unisexuals, suggesting that cosexuals are more water-use ef®cient and have lower rates of photosynthesis per unit leaf mass despite their occurrence in wetter microsites. Cosexuals appear to adjust their stomatal behavior to minimize water loss through transpiration while maintaining high investment in leaves and reproductive structures. Unisexuals apparently maximize the acquisition and storage of both water and nitrogen through increased allocation to roots and corms and enhance the uptake of CO2 by keeping stomata more open. These ®ndings indicate that the two sexual systems have different morphological and physiological features associated with local-scale variation in water availability. Key words: carbon isotope ratios; cosexuality; geophyte; habitat specialization; leaf nitrogen content; plant gender; resource allocation; sexual systems; sympatry; unisexuality; water limitation; Wurmbea dioica. INTRODUCTION Numerous empirical studies have implicated ecolog- The evolution of separate sexes from combined sexes ical conditions, particularly harsh environments, in the has occurred many times among angiosperm families. evolution and maintenance of gender dimorphism The ecological and genetic conditions favoring these (Webb 1979, Hart 1985, Arroyo and Squeo 1990, Delph 1990a, b, Barrett 1992, Costich 1995, Weller et al. contrasting sexual systems therefore represent a central 1995, Wolfe and Shmida 1997, Ashman 1999, Sakai question in evolutionary biology (reviewed in Geber and Weller 1999). These studies have reported in- et al. 1999). Theoretical models have invoked three creased gender specialization and a greater incidence key factors affecting selection for gender dimorphism, of gender dimorphism in resource-limited habitats. For including the genetic control of sex expression, the example, all of the dimorphic species of Schiedea (Car- ®tness consequences of sel®ng and outcrossing, and yophyllaceae) in Hawaii are found in dry habitats, the optimal allocation of resources to female and male whereas all but four monomorphic species occur in wet function (Charnov 1982, Lloyd 1982, Charlesworth habitats (Weller and Sakai 1990). In the Iberian pen- 1999). Because ecological context can affect the rel- insula, the sexual systems of Ecballium elaterium (Cu- ative importance of these factors, recent work on gen- curbitaceae) are geographically segregated, with the der dimorphism has focused on the environmental con- monoecious subspecies restricted to the wetter northern ditions that contribute to the relative success of com- region and the dioecious subspecies occurring in the bined vs. separate sexes. drier south (Costich 1995). In several dimorphic (gyn- odioecious and subdioecious) species, increased gender Manuscript received 22 February 2000; revised 8 September specialization (e.g., higher frequencies of female plants 2000; accepted 26 September 2000; ®nal version received 25 Oc- and reduced fruit production by hermaphrodites) is as- tober 2000. 1 Present address: Department of Biological Sciences, Uni- sociated with more stressful environments (Webb 1979, versity of Pittsburgh, Pittsburgh, Pennsylvania 15260, USA. Arroyo and Squeo 1990, Delph 1990a, Barrett 1992, E-mail: [email protected] Wolfe and Shmida 1997, Ashman 1999, Case 2000). 2601 2602 ANDREA L. CASE AND SPENCER C. H. BARRETT Ecology, Vol. 82, No. 9 Why should separate sexes be favored under harsh investigating the ecology of gender variation, because ecological conditions? Darwin (1877) was ®rst to note the transition between the two sexual systems has likely that under resource-limited conditions, unisexuals occurred relatively recently, and plants with combined would be favored over those investing in both pollen vs. separate sexes should be similar in most other as- and seed if unisexuals could outperform hermaphro- pects of their morphology and ecology. However, rel- dites in the comparable sex function (i.e., reproductive atively few such taxa are known (e.g., Ecballium ela- compensation). This prediction has been supported by terium, Costich 1995; Elatostema spp., Lahav-Ginott a recent theoretical model, which further proposes that and Cronk 1993; Leptinella spp., formerly Cotula, [As- the sex with the lower total reproductive expenditure teraceae], Lloyd 1972; Mercurialis annua [Euphorbi- should be favored in poor environments (De LagueÂrie aceae], Pannell 1996; Sagittaria latifolia [Alismata- et al. 1993). Delph (1990b) showed that the seed ®tness ceae], Wooten 1971, Sarkissian et al. 2001; Wurmbea of female plants relative to hermaphrodites is greatest dioica [Colchicaceae], Barrett 1992), and rarely do the in harsh environments in Hebe strictissima (Scrophu- contrasting sexual systems occur sympatrically. Evi- lariaceae), and proposed that the mechanism for this dence for ecological differentiation and habitat spe- effect involves greater plasticity of fruit set on her- cialization of sympatric monomorphic and dimorphic maphrodites than females. The negative relation be- sexual systems would suggest an important role for tween reproductive investment and stress tolerance is resource availability in the evolution and maintenance often observed in dioecious species, where males, of gender variation. which do not bear the high cost of fruiting, are better Here we investigate the ecology of monomorphic and able to tolerate low resource conditions than females dimorphic populations of Wurmbea dioica [(R. Br.) F. (reviewed in Delph 1999). Within populations, male Muell] ssp. alba (Macfarlane) at a site where the sexual plants often occur in greater abundance in dry micro- systems occur in sympatry. Wurmbea dioica ssp. alba sites (Freeman et al. 1976, Bierzychudek and Eckhart is a diminutive, self-compatible, insect-pollinated geo- 1988, Dawson and Bliss 1989, Dawson and Ehleringer phyte that is widespread in southwestern Australia. 1993), and many have sex-speci®c morphological and Population surveys indicate continuous variation in physiological mechanisms for dealing with drought sexual systems, ranging from monomorphic popula- (reviewed in Dawson and Geber 1999). Thus, gender- tions through gynodioecious and subdioecious to fully based differences in morphology and physiology, when dioecious populations (Barrett 1992), with contrasting associated with spatial segregation by habitat quality, sexual systems occurring sympatrically at several sites may re¯ect the direct in¯uence of disparate environ- near Perth, Western Australia (WA). Recent molecular mental conditions, or selection for strategies that max- studies of a monomorphic and a dimorphic population imize resource acquisition, improve resource use ef®- of W. dioica ssp. alba in WA indicate that the sexual ciency, or enhance survival under harsh environmental systems differ by several mutations in a relatively con- conditions. served region of the chloroplast (Case 2000). This rais- These arguments based on studies of dioecious spe- es the possibility that they may represent different spe- cies may be extended to ecological studies of combined cies. However, because our sampling to date is limited, vs. separate sexes. Stressful habitats may contribute to we will follow recent taxonomic treatments (Macfar- the maintenance of gender dimorphism if: (a) limited lane 1980, 1987) and refer to plants of the different resources reduce reproductive allocation, providing an sexual systems as W. dioica ssp. alba. advantage to gender specialists with lower total repro- In the present study, we asked two speci®c questions. ductive expenditure; or (b) poor microsite quality re- First, what is the pattern of density, distribution, and duces the reproductive ®tness of hermaphrodites rel- ¯owering phenology of monomorphic and dimorphic ative to unisexuals (Delph 1990b). Thus, in comparing sexual systems in sympatry, and are these patterns re- the ecology