NUTRIENT SENSITIVITY OF THE COST OF MALE

Transcription

1 American Journal of Botany 84(9): NUTRIENT SENSITIVITY OF THE COST OF MALE FUNCTION IN GYNODIOECIOUS PHACELIA LINEARIS (HYDROPHYLLACEAE) 1 VINCENT M. ECKHART 2,3,4 AND F. STUART CHAPIN, III 2 2 Department of Integrative Biology, 3060 Valley Life Sciences Building, University of California, Berkeley, California ; and 3 Section of Ecology and Systematics, Corson Hall, Cornell University, Ithaca, New York Allocation trade-offs should be measured as opportunity costs, estimating what individuals sacrifice in one function by allocating to others. We investigated opportunity costs of male function in gynodioecious Phacelia linearis, asking whether nutrient limitation contributes to them. This hypothesis predicts that hermaphrodites experience greater nutrient stress than females, and that hermaphrodite disadvantages in production might decrease with nutrient supply. We cultivated hermaphrodites and females at two nutrient levels, scoring individuals for prereproductive leaf number at 5 wk, and biomass, nitrogen concentration, and fruit and seed production at 16 wk. Nutrient treatments caused final growth differences of two orders of magnitude. No gender difference appeared at 5 wk, but at 16 wk hermaphrodites produced less stem, leaf, and inflorescence biomass than females, and made fewer fruits. Hermaphrodites shoot-size disadvantage was marginally significantly more severe at low nutrients than high nutrients. Significant gender nutrient interactions for root fraction and whole-plant nitrogen concentration indicate greater nutrient stress in hermaphrodites than females. Hermaphrodites also acquired less total nitrogen than females. Nutrient limitation contributes to opportunity costs of male function, but there must be other contributors. Possibilities include limitations in other resources, gender effects on morphology, and genetic trade-offs not directly involving allocation or morphology. Key words: cost of reproduction; gynodioecy; nutrient relations; resource allocation; trade-offs. Trade-offs between fitness achieved through male function and fitness achieved through female function can influence the evolution of sex allocation, sexual systems, and other features of life history (Charnov, 1982, 1993). The broad significance of sex-allocation trade-offs has motivated many attempts to measure them. Most studies of sex allocation in seed plants examine trade-offs by estimating correlations between the production of male and female tissues (reviewed in Charnov, 1982; Goldman and Wilson, 1986; Charlesworth and Morgan, 1991; Brunet, 1992), with production measured as the expenditure of various resource currencies (e.g., biomass, energy, carbon, nitrogen) in flowers (see references in reviews above, and, e.g., Ashman and Baker, 1992; Parker, Nakamura, and Schemske, 1995). Ideally, allocation trade-offs should be measured as opportunity costs, which represent lost opportunities for current and future allocation due to current resource expenditure (Gulmon and Mooney, 1986). If plant sex allocation consisted of dividing a fixed pool of resources instantaneously among various independent functions, then expenditures would estimate opportunity costs. But 1 Manuscript received 9 April 1996; revision accepted 21 January The authors thank E. Queathem, H. Reynolds, S. Hobbie, J. Canadel, P. Grogan, and M. Bret-Harte for help in harvesting, K. Kim for help with processing samples, H. Zhong for help with chemical analysis, and E. Weber and two anonymous reviewers for comments that improved the manuscript. This research was supported by USDA CSRS grant to VME. 4 Author for correspondence, current address: Department of Biology, Grinnell College, Grinnell, IA because allocation is not instantaneous, and because resource acquisition usually continues during reproduction, opportunity costs of sex allocation depend on the effect of investment in male and female tissues on the acquisition and loss of resources that could be used for present and future sex allocation (Burd and Head, 1992). If investments in male and female tissues have different effects on resource acquisition and loss, then investments measured at one time may not estimate opportunity costs accurately. In angiosperms, male and female tissues may often have different effects on resource acquisition and loss. Female tissues might be expected to incur high opportunity costs because fruits and seeds involve investments of large amounts of carbon and other resources that are not recovered (e.g., Bierzychudek, 1984; Snow and Whigham, 1989; Ashman, 1992). Male tissues might be expected to incur especially high opportunity costs, however, for at least three reasons. First, stamens are usually not photosynthetic. Developing ovaries often are, and when there is significant photosynthesis by female floral organs, carbon investment overestimates opportunity cost (e.g., Galen, Dawson, and Stanton, 1993). Second, as with seed production, investment in pollen is lost from the plant, so resource resorption is impossible. Finally, the phenology of sex allocation in seed plants, in which male allocation usually precedes much of the allocation to female function, creates the potential for male allocation to incur opportunity costs by reducing resource acquisition and growth between male and female investment periods (Rameau and Gouyon, 1991; Burd and Head, 1992; Seger and Eckhart, 1996).

2 August 1997] ECKHART AND CHAPIN NUTRIENTS AND THE COST OF MALE FUNCTION 1093 In species that exhibit gynodioecy, a sex-allocation dimorphism in which female individuals coexist with hermaphrodites, comparisons of females to hermaphrodites can estimate female output that hermaphrodites sacrifice by investing in male function. Studies of the gynodioecious annual Phacelia linearis suggest that large opportunity costs of male function are mediated by vegetative growth. In this species, direct expenditures by hermaphrodites on male function appear modest. At anthesis, each hermaphrodite flower contains 15% more dry biomass than each female flower (Eckhart, 1992c), and similarly more nitrogen (V. Eckhart, unpublished data). But the main sources of this difference are stamens and corollas (Eckhart, 1992c). These parts are nonphotosynthetic, so they do not acquire carbon, and they are shed, so individuals cannot recover all the resources they invest in them (e.g., Ashman [1994] reports nitrogen resorption efficiencies of 50% for corollas and filaments in Sidalcea oregana [Malvaceae]). Thus investment in these flower parts might be expected to trade off with growth. In P. linearis, females have an approximately twofold seed-production advantage over hermaphrodites (Eckhart, 1992a), arising largely from a vegetative growth-rate advantage (Eckhart, 1992a). Because the principle of compound interest applies to plant growth, a trade-off between male allocation and growth might translate hermaphrodites modest additional expenditures per flower into their large disadvantages in fruit and seed yield (Eckhart, 1992a). In some other gynodioecious species, females have been found to be vegetatively larger than hermaphrodites (Kesseli and Jain, 1984; Kohn, 1989) and/or to grow faster than hermaphrodites during flowering (Delph, 1990a). In some hermaphroditic species, higher male allocation is associated with reduced growth (Rameau and Gouyon, 1991; Delesalle and Mooreside, 1995; but see Silvertown, 1987). These patterns suggest that plant sex allocation may not always conisist of the division of a fixed pool of resources (see also Delph and Meagher, 1995). If individuals with lower male allocation have greater vegetative growth, they may ultimately have larger total resource pools. If male opportunity costs mediated by growth are important, they should be affected by variation in the supply of resources that affect growth. Studies of some other allocation problems, for example, plant defense against herbivores (reviewed by Herms and Mattson, 1992), recognize that trade-offs related to growth can be environmentally sensitive, but sex-allocation studies have paid much less attention to this possibility. This paper asks whether nutrient limitation contributes to male opportunity costs in P. linearis, as nutrients are diverted away from growth to male function. This hypothesis predicts that when hermaphrodite disadvantages in growth and female reproductive output occur at low nutrient supply, hermaphrodites ought to show stronger signs of nutrient stress (e.g., higher root fraction [Davidson, 1969; Hunt and Lloyd, 1987] and lower tissue nitrogen concentration [Chapin, 1980; Mooney and Winner, 1991]) than females. A second prediction is that, if nutrient limitation is an especially important source of male opportunity costs, then the hermaphrodite disadvantages ought to be relatively larger in infertile soil than in fertile soil. Diversion of nutrients (e.g., nitrogen) away from growth would be expected to reduce potential growth most severely when nutrient supply is low (Gulmon and Mooney, 1986). To evaluate these predictions, we cultivated P. linearis hermaphrodites and females at two nutrient levels, assessing the effects of gender and nutrient level on biomass production and allocation, nitrogen concentration, and fruit and seed production. MATERIALS AND METHODS Study species Phacelia linearis Pursh. (Holz.) is an annual of western North America, occurring most commonly in the northern Great Basin and Columbia River drainage (Gillett, 1962). Its pollinators are a diverse assemblage consisting mainly of pollen-collecting solitary bees (Eckhart, 1992b). Most populations in northern Utah are gynodioecious, with female frequencies averaging 3% (Eckhart, 1992a). Hermaphrodites are self-compatible but mainly outcrossing (Eckhart, 1992a). A single nuclear locus has a major effect on sex expression, with male-sterility recessive, while some results suggest the activity of modifying loci (Eckhart, 1992c). Cytoplasmic factors that affect gender have not yet been discovered. Low female frequencies in nature would be expected under the combination of nuclear inheritance of sex expression and a seed-production advantage of females slightly greater than twofold (Lewis, 1941; Charlesworth and Charlesworth, 1978). Seed source For this experiment, the first author generated a set of outbred individuals with a diverse genetic background and a gender ratio of 1:1. The original source consisted of 20 naturally pollinated seed families of female maternal parents (collected in 1989 from Red Butte Canyon population a of Eckhart, 1992a). One year after collection, seeds per family were cultivated in an outdoor garden, 2 km from the population s native site. Natural pollinators were abundant and active in the garden, and cultivated plants produced many seeds. On the single-locus inheritance model, the hermaphrodites in the garden were expected to be heterozygous at the gender locus, so equal frequencies of male-fertile and male-sterile alleles would be expected in the pollen pool. Thus the expected segregation ratio in progenies of females in the garden would be 1:1 hermaphrodites to females. Twenty randomly chosen females from the garden population served as parents for the present experiment, with 25 seeds taken per female. We did not keep track of the family membership of individual seeds. Plant care and experimental design Seeds imbibed a dilute aqueous solution of gibberellic acid (100 mg/ L) on filter paper for 2dand were sown in 800-mL pots (three seeds per pot) filled with U.C. Davis soil mix, coarse sand, and perlite (2:1:1). The gibberellic acid treatment hastens germination but does not substantially affect subsequent growth compared to untreated P. linearis seeds (V. Eckhart, unpublished data; see Fox, Evans, and Keefer, 1995). Daily misting with distilled water kept the soil surface moist until seedling emergence ceased (2 wk), after which pots were thinned to leave one randomly chosen seedling per pot. Individuals were randomly assigned to low and high nutrient treatments, with treatments dispersed in a checkerboard fashion on a single greenhouse bench. Systematic dispersion was intended to reduce the possibility that environmental gradients would cause spurious treatment effects, relative to a fully randomized design (Hurlbert, 1984). Greenhouse temperatures averaged 30 C during the day and 20 C at night. High-intensity metal-halide lamps above the bench provided supplemental light for 15 h/d, long enough to stimulate flowering in P. linearis. All individuals received 20 ml of a complete fertilizer (DynaGro, Novato, CA, 7:9:5 NPK plus micronutrients, diluted 1:1500 with water) on day 34. High-nutrient individuals also received three applications of 4 mo release 14:14:14 NPK (Osmocote, Grace-Sierra, Milpitas, CA): (1) 0.25 g on day 20; (2) 0.25 g on day 39; and (3) 0.5 g on day 60. Daily to biweekly distilled-water irrigation kept all individuals well watered throughout the experiment. After flowering began

3 1094 AMERICAN JOURNAL OF BOTANY [Vol. 84 (day 40), flowers were hand-pollinated daily with pollen from a haphazard collection of hermaphrodites from the same experiment, using a small paintbrush. Data collection During cultivation, we scored individuals for the number of true leaves on day 35, and after flowering began (day 40), we scored individuals daily for floral sex expression. Of 246 individuals that began the experiment, pathogens killed 13, and 71 did not flower. Most nonflowering individuals were in the low-nutrient treatment. Among flowering individuals, anthesis date averaged 5 d earlier at high nutrients than low nutrients, but was unaffected by gender. Three intersex individuals with a mixture of fertile and sterile stamens were excluded from the analysis, as were five that showed partial female sterility. Final sample sizes were 58 high-nutrient hermaphrodites, 48 high-nutrient females, and 24 of each gender at low nutrients. Processing began on day 114, when most flowering plants were beginning to senesce. Roots and shoots were harvested, and shoots were clipped into stem, leaf, infructescence, and fresh-flower fractions. After 48 h of forced-air drying in a 65 C oven, tissue fractions were weighed to the nearest 0.1 mg and then ground in an electric mill. Many individuals had no fresh flowers at harvest. Because these individuals had biomass values of 0 for this tissue fraction, the distribution of freshflower biomass could not be made to satisfy the assumptions of analysis of variance (ANOVA), so fresh-flower biomass was combined with infructescence biomass (becoming inflorescence biomass ). All individuals were scored for fruit number before grinding, but 47 individuals infructescences were ground before their seeds could be counted, reducing the power to detect treatment effects on seed number. For tissue samples from which at least 2 mg of ground material could be recovered, total nitrogen concentrations were determined with a Carlo-Erba NA 1500 CN analyzer (CE Instruments, Milan). Roots and stems were ground together for nitrogen analysis. We made two indirect estimates of total nitrogen within plants. For lower-bound estimates, we assumed that nitrogen resorption from corollas and stamens was complete (literally impossible, since nitrogen in pollen cannot be recovered) and that tissue losses (e.g., root mortality) during growth were negligible. Under these assumptions, nitrogen content at final harvest equals total nitrogen uptake. Multiplying wholeplant nitrogen concentration by total biomass gives this estimate of total nitrogen. For something closer to upper-bound estimates, we assumed that plants recovered no nitrogen from abscised flower parts and that other tissue losses were negligible. Under these assumptions, the product of the average biomass of corollas and androecia in hermaphrodite and female flowers (Eckhart, 1992c), nitrogen concentrations in these tissues (V. Eckhart, unpublished data), and individual fruit production from this study can be added to the lower-bound estimates as a second estimate of total nitrogen, albeit one for which statistical comparisons would not be valid. Statistical analysis We analyzed most variables by two-way ANO- VA, with gender and nutrient level treated as fixed effects, and with Type III sums of squares used to construct F statistics. For the data collected at the final harvest, we also examined the effect of adding prereproductive leaf number as a covariate, to control for early differences in shoot size. This had no qualitative effect on the statistical significance of main effects and interactions. Thus the significance categories reported below apply to both the ANOVA and the analysis of covariance. We used two-way multivariate analysis of variance (MAN- OVA) to analyze the effects of gender and nutrients on selected sets of related characters. SAS for personal computers performed these analyses (procedure GLM) (SAS, 1990). Several data transformations preceded analysis. To test whether relative production differences between females and hermaphrodites were sensitive to nutrients, log-transformed data are appropriate, so log 10 transformation of dry masses, fruit number, and seed number preceded ANOVA. This transformation also made variances homogeneous. The log 10 of estimated total nitrogen did not satisfy ANOVA assumptions, so for this variable the genders were compared within nutrient levels by nonparametric Mann-Whitney U tests of the untransformed values (SAS procedure NPAR1WAY). For root fraction and nitrogen concentrations, hypothesis testing did not require log transformations, but these variables were logit-transformed (ln[y/1-y]) to improve normality and homogenize variances for ANOVA. RESULTS Prereproductive size Fertilization significantly increased leaf number by day 35 (for hermaphrodites, mean 1SE leaves at high nutrients vs leaves at low nutrients; for females leaves at high nutrients vs leaves at low nutrients; P 0.001). There was no significant gender effect on day 35 leaf number (P 0.40), nor any evidence of interaction (P 0.80). Nutrient effects on final production Nutrient effects on the final production of biomass, fruits, and seeds were large and highly statistically significant. Average differences between nutrient levels approached two orders of magnitude for all measures of biomass, showing that growth in the low-nutrient treatment was strongly nutrient limited (Figs. 1 3). MANOVA of the four tissue fractions together (roots, stems, leaves, and inflorescences) revealed a highly significant effect of nutrients (P ). Compared to high-nutrient individuals, low-nutrient plants had significantly larger root fractions and lower nitrogen concentrations in whole plants (Fig 4) and in all tissue fractions (Fig. 5). Gender effects on final production Significant hermaphrodite disadvantages in biomass production were found at both nutrient levels and for all tissues except roots (Figs. 1 2). MANOVA of the four tissue fractions together revealed a significant effect of gender (P 0.003). Though MANOVA detected no overall gender nutrient interaction, the interaction term approached significance for total shoot biomass analyzed alone (P 0.08 without prereproductive size as a covariate; P 0.06 with the covariate; Fig. 1). On average, hermaphrodite shoots were 50% as large female shoots at low nutrients, while at high nutrients they were 80% as large. For stems, leaves, and inflorescences there were significant gender effects but no significant interactions (Fig. 2). Hermaphrodites produced fewer fruits and seeds than females at both nutrient levels, significantly so for fruits (Fig. 3). Gender and signs of nutrient stress In low-nutrient conditions, hermaphrodite growth disadvantages appeared to be related to nutrient status in the predicted way. At low nutrients, hermaphrodites had larger root fractions than females; at high nutrients there was a smaller gender difference, leading to a significant interaction as well as a significant main effect of gender (Fig. 4). Because hermaphrodites and females had similar root biomass, this relative allocation difference means that a given amount of root biomass supported less shoot growth in hermaphrodites than in females. Whole-plant nitrogen concentration was lower in hermaphrodites than females at low nutrients, as would be expected if her-

4 August 1997] ECKHART AND CHAPIN NUTRIENTS AND THE COST OF MALE FUNCTION 1095 Fig. 1. Mean (A) total biomass, (B) shoot biomass, and (C) root biomass of hermaphrodites and females cultivated at two nutrient levels, on a log 10 scale. Error bars are 1 SE. Statistical significance levels are denoted by, *, **, and *** for P 0.10, P 0.05, P 0.01, and P 0.001, respectively; ns means nonsignificant. maphrodites were more nutrient stressed; at high nutrients, nitrogen concentration was actually lower in females than hermaphrodites, leading to a significant nutrient gender interaction (Fig. 4). For nitrogen concentration in different tissues, the dominant pattern was interaction between gender and nutrient supply. MANOVA of nitrogen concentrations of all four fractions together (roots plus stems, leaves, infructescences, and flowers) revealed a significant overall effect of nutrients (higher concentrations at high nutrients, P 0.001) and a significant interaction (P 0.05), but no main effect of gender. For leaves, infructescences, and flowers, gender rankings changed between nutrient levels, with hermaphrodites lower at low nutrients and the reverse at high nutrients (Fig. 5). The most dramatic difference was in leaf nitrogen concentration, which was 20% higher in high-nutrient hermaphrodites than highnutrient females. The exceptional tissue fraction was the combination of roots and stems, for which nitrogen concentration was marginally higher in hermaphrodites than females at both levels. Together, these patterns suggest that the hermaphrodite deficit in whole-plant nitrogen Fig. 2. Mean (A) stem biomass, (B) leaf biomass, and (C) inflorescence biomass of hermaphrodites and females cultivated at two nutrient levels, on a log 10 scale. Lines and abbreviations as above. concentration at low nutrients was due to lower nitrogen concentrations in most tissues and also to smaller relative allocation to tissues with high nitrogen concentrations. Gender and total nitrogen Gender differences in estimated total nitrogen are similar to those in whole-plant and shoot biomass. By the lower-bound estimates (assuming complete nitrogen resorption), total nitrogen was lower in hermaphrodites than females at low nutrients (mean 1SE mg for hermaphrodites vs mg for females; P 0.05). At high nutrients the gender difference was significant and in the same direction, but the female mean was only 16% greater than that of hermaphrodites ( mg for hermaphrodites vs mg for females; P 0.05), compared to a 113% difference at low nutrients. By the alternative estimate (assuming no resorption), hermaphrodites also had much less total nitrogen than females at low nutrients (an average of 0.32 mg for hermaphrodites vs mg for females), and at high nutrients there was a relatively smaller difference in the same direction (29.3 mg for hermaphrodites vs mg for females). DISCUSSION Nutrient limitation and opportunity costs of male function The experiment detected large hermaphrodite

5 1096 AMERICAN JOURNAL OF BOTANY [Vol. 84 Fig. 3. Mean (A) fruit production and (B) seed production of hermaphrodites and females cultivated at two nutrient levels, on a log 10 scale. Lines and abbreviations as above. disadvantages in growth and fruit production, which we interpret as opportunity costs of male function. These costs were expressed as reduced shoot growth after the onset of flowering. Subsequent studies show that hermaphrodites have a shorter reproductive life-span than females (V. Eckhart, unpublished data), so these costs Fig. 5. Mean nitrogen concentration of (A) roots stems, (B) leaves, (C) infructescences, and (D) flowers of hermaphrodites and females cultivated at two nutrient levels. Lines and abbreviations as above. Fig. 4. Mean (A) root fraction and (B) nitrogen concentration of hermaphrodites and females cultivated at two nutrient levels. Lines and abbreviations as above. may have been even larger had the experiment not been terminated while some plants were still flowering. Nutrient limitation does appear to contribute to opportunity costs of male function. In conditions where growth was strongly nutrient limited, male opportunity costs were associated with greater signs of nutrient stress (larger root fractions and lower nitrogen concentrations) in hermaphrodites. There is also some indication (the suggestion of a gender nutrient interaction for shoot biomass) that these costs were relatively larger at low nutrients. Perhaps most significantly, especially at low nutrients, where nitrogen is an appropriate allocation currency for plants (Chapin, 1989), male function appeared to reduce the size of the total pool of nitrogen available for allocation. An alternative possibility would have been similar nitrogen pool sizes in hermaphrodites and females, with larger per-flower expenditures by hermaphrodites leading to females having more nitrogen available for other functions. Instead, male function appeared to reduce the amount of a limiting resource available for allocation. The possibility that trade-offs between male function and growth are especially severe in infertile soil suggests

6 August 1997] ECKHART AND CHAPIN NUTRIENTS AND THE COST OF MALE FUNCTION 1097 that the seed-production advantage of P. linearis females in nature might decline with soil fertility. The equilibrium frequency of females in gynodioecious populations is expected to increase with the size of the female-fertility advantage (Lewis, 1941; Charlesworth and Charlesworth, 1978). Thus it is interesting to note that in some gynodioecious species, female frequency is greater in lowresource sites (e.g., Delph, 1990b). These findings suggest that dimorphic sexual systems are more likely to be evolutionarily stable in low resource environments. A recent ESS (evolutionarily stable strategy) analysis makes the same prediction (De Laguérie, Olivieri, and Gouyon, 1993). That study reports that for many plausible relationships between resource availability and fitness through male and female function, tradeoffs are most severe at low resources, and the ESS changes from dimorphic sexual systems to hermaphroditism as resource availability increases. Though the severity of sex allocation trade-offs does not always decline monotonically with resource levels, e.g., in gynodioecious Plantago lanceolata (Plantaginaceae) the nutrient response of female-fertility advantages varies among studies (Poot, 1996), some form of environmental sensitivity of sex allocation trade-offs may be common. Other sources of the trade-off Another important finding is that male opportunity costs remain substantial in high-nutrient conditions. In the high-nutrient treatment, hermaphrodites had higher nitrogen concentrations than females and only slightly smaller total pools of nitrogen, yet still grew less and produced fewer fruits. Mechanisms other than nutrient limitation must be responsible for the trade-off between male function and growth at high nutrients. What else might contribute? One possibility is that allocation trade-offs in other resources play a role. For example, diverting carbon away from growth might be expected to incur opportunity costs whenever carbon acquisition limits growth. Carbon opportunity costs would be expected to be especially large when growth is rapid, as in this experiment s high-nutrient treatment (Gulmon and Mooney, 1986). Another possibility is that some indirect costs of male function involve not the allocation of limited chemical resources or energy but the activation of meristems. Subsequent experiments with P. linearis show that hermaphrodites produce axillary branches (which in this species consist of inflorescences subtended by pairs of leaves) at lower rates than females, and this is not merely due to hermaphrodites smaller shoot size (V. Eckhart and T. Chen, unpublished data). This pattern is not unique to this species. In gynodioecious Plantago lanceolata, female genotypes also have higher rates of branching than hermaphrodites (Olff et al., 1989; Poot, 1996). If male function tends to increase apical dominance, and if reproductive growth in high-resource conditions is limited more strongly by meristem availability than by resources (Geber, 1990), then male function could reduce growth and female yield through its effect on morphology. This experiment s findings that hermaphrodites had relatively high leaf nitrogen concentration at high nutrients and higher root and stem nitrogen concentration at both nutrient levels are consistent with the possibility of sink limitation of hermaphrodite growth; it is as if females were able to translocate nitrogen into reproductive tissues more effectively than hermaphrodites. If morphological correlates of sex expression cause costs of male function, this would constitute a genetic trade-off that is not directly related to allocation. Mole (1994) points out that genetic trade-offs unrelated to allocation should be considered as alternatives to resourcebased explanations whenever the latter appear not to add up from strictly budgetary considerations. This is an important suggestion, though we should highlight two additional points. First, budgetary calculations for trade-offs involving growth and time must account for the possibility that modest differences in early allocation can generate much larger differences later on. Second, studies of allocation trade-offs need to consider the possibility of genotypic differences in resource acquisition (Van Noordwijk and de Jong, 1986). For example, Delph and Meagher (1995) suggest that in dioecious Silene latifolia (Caryophyllaceae), higher sink source ratios in females than males (because of females developing fruits) lead to faster carbon assimilation and larger total resource pools in females. In summary, nutrient limitation appears to contribute to the cost of male function in gynodioecious P. linearis. Nutrient sensitivity of the cost of male function may be an important factor in the evolution of plant sex allocation, sexual systems, and life histories. Because for most agronomic species, it is vegetative or female tissues that are valuable, it may also be useful to consider costs of male function in breeding programs (see Garnier, Maurice, and Olivieri, 1993), perhaps especially in programs for yield improvement in poor soils. Nutrient limitation is clearly not the only source of opportunity costs of male function, however, for this study detected significant costs in conditions where hermaphrodites actually had higher nutrient status than females. Limitations in other kinds of resources, morphological effects of sex expression, and other kinds of genetic trade-offs might lead to male opportunity costs where nutrients do not limit growth. Understanding the fitness payoffs for alternative resource allocations requires looking into and beyond the physiology of allocation. LITERATURE CITED ASHMAN, T.-L Indirect costs of seed production within and between seasons in a gynodioecious species. Oecologia 92: A dynamic perspective on the physiological cost of reproduction in plants. American Naturalist 144: , AND I. BAKER Variation in floral sex allocation with time of season and currency. Ecology 73: BIERZYCHUDEK, P Determinants of gender in jack-in-the-pulpit: the influence of plant size and reproductive history. Oecologia 65: BRUNET, J Sex allocation in hermaphroditic plants. Trends in Ecology and Evolution 7: BURD, M., AND G. HEAD Phenological aspects of male and female function in hermaphroditic plants. American Naturalist 140: CHAPIN, F. S., III The mineral nutrition of wild plants. Annual Review of Ecology and Systematics 11: The cost of tundra plant structures: evaluation of concepts and currencies. American Naturalist 133: CHARLESWORTH, D., AND B. CHARLESWORTH A model for the evolution of dioecy and gynodioecy. American Naturalist 112:

7 1098 AMERICAN JOURNAL OF BOTANY [Vol. 84, AND M. T. MORGAN Allocation of resources to sex functions in flowering plants. Philosophical Transactions of the Royal Society of London B 332: CHARNOV, E. L The theory of sex allocation. Princeton University Press, Princeton, NJ Life history invariants: some explorations of symmetry in evolutionary ecology. Oxford University Press, NY. DAVISON, R. L Effect of root:leaf temperature differentials on root:shoot ratios in some pasture grasses and clover. Annals of Botany 33: DE LAGUÉRIE, P., I. OLIVIERI, AND P.-H. GOUYON Environmental effects on fitness-sets shape and evolutionary stable strategies. Journal of Theoretical Biology 163: DELESALLE, V. A., AND P. D. MOORESIDE Estimating the costs of allocation to male and female functions in a monoecious cucurbit, Lagenaria siceraria. Oecologia 102: DELPH, L. F. 1990a. Sex-differential resource allocation patterns in the subdioecious shrub Hebe subalpina. Ecology 71: b. Sex-ratio variation in the gynodioecious shrub Hebe strictissima Scrophulariaceae. Evolution 44: , AND T. R. MEAGHER Sexual dimorphism masks life history trade-offs in the dioecious plant Silene latifolia. Ecology 76: ECKHART, V. M. 1992a. Resource allocation and the evolution of gynodioecy in Phacelia linearis (Hydrophyllaceae). Evolution 46, b. Spatio-temporal variation in abundance and variation in foraging behavior of the pollinators of gynodioecious Phacelia linearis (Hydrophyllaceae). Oikos 64, c. The genetics of gender and the effects of gender on floral characters in gynodioecious Phacelia linearis (Hydrophyllaceae). American Journal of Botany: 79: FOX, G. A., A. S. EVANS, AND C. J. KEEFER Phenotypic consequences of forcing germination: a general problem of intervention in experimental design. American Journal of Botany 82: GALEN, C., T. E. DAWSON, AND M. L. STANTON Carpels as leaves: Meeting the carbon cost of reproduction in an alpine buttercup. Oecologia 95: GARNIER, P., S. MAURICE, AND I. OLIVIERI Costly pollen in maize. Evolution 44: GEBER, M. A The cost of meristem limitation in Polygonum arenastrum: negative genetic correlations between fecundity and growth. Evolution 44: GILLETT, G. W Evolutionary relationships of Phacelia linearis. Brittonia 14: GOLDMAN, D. A., AND M. F. WILLSON Sex allocation in functionally hermaphroditic plants. Botanical Review 52: GULMON, S. L., AND H. A. MOONEY Costs of defense and their effects on plant productivity. In T. J. Givnish [ed.], On the economy of plant form and function, Cambridge University Press, London. HERMS, D. A., AND W. J. MATTSON The dilemma of plants: to grow or defend. Quarterly Review of Biology 67: HUNT, R., AND P. S. LLOYD Growth and partitioning. New Phytologist 106 (Supplement): HURLBERT, S. H Pseudoreplication and the design of ecological field experiments. Ecological Monographs 54: KESSELI, R., AND S. K. JAIN An ecological genetic study of gynodioecy in Limnanthes douglasii (Limnanthaceae). American Journal of Botany 71: KOHN, J. R Sex ratio, seed production, biomass allocation, and the cost of male function in Cucurbita foetidissima HBK (Cucurbitaceae). Evolution 43: LEWIS, D Male sterility in natural populations of flowering plants. New Phytologist 40: MOLE, S Trade-offs and constraints in plant-herbivore defense theory: a life-history perspective. Oikos 71: MOONEY, H. A., AND W. E. WINNER Partitioning response of plants to stress. In H. A. Mooney, W. E. Winner, and E. J. Pell [eds.], Responses of plants to multiple stresses, Academic Press, San Diego, CA. OLFF, H., D. KUIPER, J.M.M.VAN DAMME, AND P. J. C. KUIPER Gynodioecy in Plantago lanceolata VI: functions of cytokinins in growth, development, and reproduction of two sex types. Canadian Journal of Botany 67: PARKER, I. M., R. K. NAKAMURA, AND D. W. SCHEMSKE Reproductive allocation and the fitness consequences of selfing in two sympatric species of Epilobium (Onagraceae) with contrasting mating systems. American Journal of Botany 82: POOT, P Ecophysiological aspects of the maintenance of male sterility in Plantago lanceolata. Ph.D. dissertation, University of Utrecht. PRIMACK, R. B Evolutionary aspects of wind pollination in the genus Plantago (Plantaginaceae). New Phytologist 81: RAMEAU, C., AND P.-H. GOUYON Resource allocation to growth, reproduction, and survival in Gladiolus: the cost of male function. Journal of Evolutionary Biology 4: SAS SAS/STAT user s guide, version 6, fourth edition. SAS Institute, Cary, NC. SEGER, J., AND V. M. ECKHART The evolution of sexual systems and sex allocation in hermaphroditic plants when growth and reproduction overlap. Proceedings of the Royal Society of London B 263: SILVERTOWN, J The evolution of hermaphroditism: an experimental test of the resource model. Oecologia 72: SNOW, A. A., AND D. F. WHIGHAM Costs of flower and fruit production in Tipularia discolor (Orchidaceae). Ecology 70: VAN NOORDWIJK, A. J., AND G. DE JONG Acquisition and allocation of resources: their influence on variation in life-history tactics. American Naturalist 45: