Stress promotes maleness in hermaphroditic modular animals

6 downloads 78 Views 222KB Size Report
Sep 2, 2003 - Hughes, R. N. (1989) A Functional Biology of Clonal Animals (Chapman & Hall,. London). 14. ... Lloyd, D. G. & Bawa, K. S. (1984) Evol. Biol.
Stress promotes maleness in hermaphroditic modular animals R. N. Hughes*†, P. H. Manrı´quez*‡, J. D. D. Bishop§¶, and M. T. Burrows储 *School of Biological Sciences, University of Wales, Bangor, Gwynedd LL57 2UW, United Kingdom; §Marine Biological Association of the United Kingdom, The Laboratory, Citadel Hill, Plymouth PL1 2PB, United Kingdom; ¶Department of Biological Sciences, University of Plymouth, Drake Circus PL4 8AA, United Kingdom; and 储Scottish Association for Marine Science, Dunstaffnage Marine Laboratory, Oban, Argyll PA37 1QA, United Kingdom Communicated by John L. Harper, University of Wales, Aberystwyth, United Kingdom, June 30, 2003 (received for review March 4, 2003)

Sex-allocation theory developed for hermaphroditic plants predicts that impaired phenotype or reduced parental survivorship caused by environmental stress should induce relatively greater allocation to the male function. We provide experimental evidence of stress-induced maleness, already well documented in flowering plants, in a modular animal. By using cloned copies of replicate genotypes, we show that the marine bryozoan Celleporella hyalina increases the ratio of male to female modules in response to diverse environmental stressors. Mating trials confirmed that paternity is determined by fair-raffle sperm competition, which should obviate local mate competition at characteristic population density and promote the advantage of increased male allocation. The demonstrated similarity to plants transcends specific physiological pathways and suggests that stress-induced bias toward male function is a general response of hermaphroditic modular organisms to impaired prospects for parental productivity or survival. sex allocation 兩 sperm competition 兩 modularity 兩 hermaphrodites 兩 Bryozoa

M

any flowering plants deploy both male and female functions within the same individual (1), a condition termed hermaphroditism, monoecy, or cosexuality. There is a strong tendency of such plants to increase the ratio of pollen to seed production when growing under environmentally stressful conditions, such as drought (2), nutrient deficiency (3), herbivory (4), or pathogenic infection (5). Stress-induced maleness in plants was first given theoretical consideration, albeit in Lamarckian spirit, by Henslow: ‘‘there seems to be a tolerably uniform consensus of opinion that the female sex in plants is correlated with a relatively stronger vital vigour than the male; and this is just what an a priori assumption would look for, as the duration of existence and the work to be done in making fruit require a greater expenditure of energy than the temporary function of the stamens’’ (ref. 6, p. 230). Despite subsequent experimental demonstrations of stress-induced maleness in plants, quantitative Darwinian treatment remained lacking until Freeman et al. (7) used a newly emergent theory on environmental sex determination (8, 9) to present an evolutionarily stable strategy model predicting stress-induced maleness in plants occupying patchily stressful environments. Freeman et al. (7) considered the example of patch dryness, but stress could mean any condition induced by the physical or biological environment that reduces scope for resource allocation or survivorship and, hence, reduces fitness (10). Colonial invertebrates are a globally dominant life form on sublittoral hard substrata (11). Their sessile habit, modularity, and external dissemination of male gametes are features shared with higher plants; these features generate selection pressures common to the reproductive biology of the two types of organism (12). Parallel trends appear consequently in many aspects of life history, including dispersal of propagules (13), mating strategies (14, 15), resource allocation to sexual and somatic functions (16, 17), and the frequent occurrence of hermaphroditism (18–20). Accordingly, we used the bryozoan Celleporella hyalina to test 10326 –10330 兩 PNAS 兩 September 2, 2003 兩 vol. 100 兩 no. 18

the prediction that stress inflates proportional allocation to the male function in modular animals as in plants. C. hyalina is an ephemeral epiphyte whose life span is normally constrained to a time scale of months by deterioration of the substratum (21) or competitive overgrowth (22). The lifespan of C. hyalina, however, can be extended indefinitely in the laboratory by artificial propagation with colonies showing no symptoms of senescence (23). Multiple colonies of each genotype, thus, may be propagated and used to partition genetic and environmental effects experimentally. Sexually mature colonies bear morphologically distinct male and female modules (zooids), which meet their nutritional requirements by the translocation of resources from adjacent feeding modules (autozooids) (24). A relative measure of sex allocation, therefore, can be gained simply by counting the number of male or female zooids per autozooid in a colony (25). Male zooids are of two types: frontal and basal. Frontal males, like females, are produced by frontal (upper surface) budding of the basal layer of autozooids, whereas basal males are produced by lateral budding of autozooids in the peripheral meristem or, occasionally, by conversion of established autozooids remote from the meristem (26). In the laboratory, extensive basal male production has been recorded among colonies of C. hyalina grown under suboptimal combinations of temperature and food supply (27). Hence, the production of basal males probably is a mechanism by which Celleporella adjusts sex ratio in response to stress. Environmental stressors naturally encountered by C. hyalina include physical impediment to growth (11), reduced food supply (28, 29), temperature shock or desiccation during tidal emersion (30), and destruction of zooids by predators (31). We simulated these stressors in the laboratory and recorded their effects on sex allocation in clonally propagated colonies. To interpret any effect of stress on sex allocation successfully, it is necessary also to consider the implications of sperm competition. Fair-raffle competition (32) will cause fitness gained by male function to increase over a greater range of proportional resource allocation and, thereby, will promote maleness (33, 34) in concert with the predicted effect of stress (35). We, therefore, experimentally tested the hypothesis that paternity is determined by fair-raffle sperm competition. Methods Colonies of C. hyalina were established from larvae settled on acetate in April 1996 and propagated by taking cuttings (ramets) to generate a corresponding set of clones (36). Each clone was isolated from sources of allosperm and, consequently, no ramets produced larvae before experimentation (24). Experimental ramets were housed in 300-ml glass jars two-thirds filled with 0.2 ␮m-filtered, UV-irradiated seawater (filtered sea water) at 16°C ⫾ 2°C and fed daily with Rhinomonas reticulata at ⬇100 cells per ␮l. Clones used for outcrossing were first screened for †To

whom correspondence should be addressed. E-mail: [email protected].

‡Present address: Estacio ´ n Costera de Investigaciones Marinas (ECIM), Facultad de Ciencias

Biolo´gicas, Pontificia Universidad Cato´lica de Chile, Casilla 114-D, Santiago, Chile.

www.pnas.org兾cgi兾doi兾10.1073兾pnas.1334011100

reproductive compatibility, shown by the production of larvae in crossed ramets and the lack of larval production by reproductively isolated controls.

Table 1. Effects of stressors on the number of basal male, frontal male, and female zooids produced per autozooid

Effects of Stress on Sex Allocation. Intercolonial contact. Single ramets of clones A1, D1, E1, and M1 (37) were used as replicates to assess the effect of paired contact with ramets of clone Q1, applied as a standard treatment with regard to genotype and associated growth characteristics. To stage intercolonial contact, ramets growing on separate pieces of acetate were glued 4–5 mm apart to a third piece of acetate measuring 45 ⫻ 75 mm and placed in a culture jar. When ramets had grown into contact over a length of 5–8 mm, the sexual zooids were counted in 10 quadrats taken randomly within the first row of functioning autozooids along the edge facing the other ramet. The quadrats measured 1,600 ⫻ 1,600 ␮m, equivalent to the area occupied by 26–28 autozooids, and were marked on acetate placed on the microscope stage. Data were recorded as means per quadrat. Because uptake of allosperm triggers investment in female function (24, 38) that in turn may influence male function, control ramets were exposed to Q1 allosperm suspension on alternate days. Physical obstruction. The same protocols were used as for intercolonial contact, except that each replicate was glued 1–2 mm from a piece of glass measuring 15 ⫻ 15 mm in area and 1 mm in thickness. Measurements began once the ramet had grown in contact with the glass over a length of 5–8 mm. Control ramets were not exposed to allosperm. Food deprivation. Two ramets were taken from each of clones A1, E1, J1, and M1. One ramet was placed in a culture jar containing filtered sea water for 3 d and then fed daily over a period of 1 mo. The other ramet was treated similarly but not deprived of food. At the end of the experiment, sexual zooids were counted within seven quadrats, measuring 1,600 ⫻ 1,600 ␮m, and taken randomly among the first autozooids budded after food deprivation. The demarcation of postdeprivation growth was identified from camera lucida drawings of the colonial perimeter, which were made at the beginning of the experiment. Temperature shock. Protocols were the same as for food deprivation except that experimental ramets were exposed for 2 h to a temperature of 4–5°C then returned to standard culture conditions. Desiccation. Protocols were the same as for food deprivation except that experimental ramets were subjected to 10 alternating periods of 10 min in air and 2 min in filtered sea water then returned to standard culture conditions. Physical damage. Three ramets were taken from each of clones A1, E1, J1, and M1. In one ramet, the distal part of five or six autozooids was destroyed with a needle at 20 points around the colonial perimeter. In the second ramet, total destruction of autozooids was effected within areas 1 mm2 at 20 points equally spaced around the circumference but separated by several zooids from the edge of the colony. The third (control) ramet was left undamaged. After a period of 1 mo, new autozooids and sexual zooids were counted at the destruction points in experimental ramets and at equivalent areas in control ramets by using a 1,600 ⫻ 1,600-␮m quadrat. Comparison of the effects of outer and inner damage was used to assess whether damage to the meristem itself had a critical influence on sex allocation.

Intercolonial contact Treatment Control Physical obstruction Treatment Control Food deprivation Treatment Control Temperature shock Treatment Control Desiccation Treatment Control Peripheral damage Treatment Control Inner damage Treatment Control

ning the following experiments with ramets that had been kept in reproductive isolation for several months and were, therefore, devoid of stored allosperm before experimentation (24). Allosperm suspension was obtained by placing 10 ramets of the donor clone in a 2-liter glass beaker filled with aged filtered sea water. After 12 h in darkness, the colonies were transferred to a similar beaker and illuminated to induce sperm release (36). Sperm were counted in a 30-ml aliquot taken after 15 min of illumination. Allosperm suspensions used in the experiments

Hughes et al.

Basal males

Frontal males

Females

0.045 ⫾ 0.009 0

0.043 ⫾ 0.011 0.028 ⫾ 0.006

0.038 ⫾ 0.008 0.033 ⫾ 0.005

0.058 ⫾ 0.003 0

0.040 ⫾ 0.004 0.008 ⫾ 0.003

0.042 ⫾ 0.003 0

0.027 ⫾ 0.008 0

0.008 ⫾ 0.002 0.015 ⫾ 0.003

0 0.017 ⫾ 0.008

0.088 ⫾ 0.009 0

0.010 ⫾ 0.007 ⬍0.001

0 0

0.023 ⫾ 0.006 0

0.008 ⫾ 0.003 0.020 ⫾ 0.004

0 0.025 ⫾ 0.003

0.048 ⫾ 0.008 0

0.013 ⫾ 0.003 0.013 ⫾ 0.003

0 0.013 ⫾ 0.003

0.013 ⫾ 0.003 0

0.060 ⫾ 0.009 0.080 ⫾ 0.009

0.027 ⫾ 0.008 0.065 ⫾ 0.006

Data are means (n ⫽ 4) ⫾ standard errors.

had concentrations of 10–102 per ml, sufficient for maximum fertilization success (37). Metamorphosed larvae (ancestrulae) appeared on acetate sheeting, lining the inner wall of each jar, 18–43 d after sperm had been administered. Ancestrulae chosen to be genotyped were grown for a further 3 wk to form small colonies. Samples of three or four zooids were excised from each 3-wk-old colony, whose date of settlement was recorded, and stored in 1.5-ml microtubes containing 10 ␮l of TE buffer (10 nM Tris兾1 nM EDTA, pH 7). Within 20–30 min of storage, DNA was extracted (14) and paternity was identified by using the microsatellite CHY1 (39). Fair-raffle competition. We administered an equal mixture of suspended sperm derived from ramets of three unrelated donor clones (A1, E1, M1) to a virgin ramet of each of five unrelated receptor clones (D1, F1, H1, J1, QJ9). Equal contribution of the donor clones to the pool of offspring would be consistent with fair-raffle sperm competition (32). Autosperm–allosperm interaction. To test for any influence of autosperm suspension on the uptake of allosperm, we first administered the sperm mixtures A1, E1, and M1 to a virgin ramet from each of clones A1, E1, and M1, respectively. Offspring were counted, and 20 of these offspring were genotyped. We then repeated the experiment by using an equal mixture of sperm from the two unrelated donor clones (i.e., by omitting autosperm) in each case. Influence of multiple paternity on egg production. Because exposure to allosperm triggers egg growth (38), we tested the possibility that two sources of allosperm trigger more eggs to develop than one source of allosperm and, thereby, ameliorate sperm competition; we administered single sperm suspensions from A1, E1, and M1 to three virgin ramets from each of the same clones, respectively. Offspring were counted and samples of 20 offspring were genotyped. Reproductive output was compared with that in the previous experiment, in which ramets from the same clones were given two sources of allosperm. Results Effects of Stress on Sex Allocation (Table 1). Intercolonial contact. Basal males were produced only by treatment ramets, but treatment PNAS 兩 September 2, 2003 兩 vol. 100 兩 no. 18 兩 10327

EVOLUTION

Sperm Competition. We investigated sperm competition by run-

Stressor

Table 2. Number of offspring sired by each of three donor clones whose sperm had been presented in equal proportions to each of five receptor clones Intermediate progeny†

First progeny*

Last progeny‡

Receptor clone

A1

E1

M1

A1

E1

M1

A1

E1

M1

D1 F1 H1 J1 QJ9

17 16 16 13 11

10 10 12 10 13

13 14 11 17 15

9 12 11 9 16

11 11 6 10 9

10 7 13 11 5

14 13 16 17 14

13 16 17 14 17

13 11 7 9 8

*First 40 offspring produced. †Offspring selected randomly during midrelease period. ‡Last 40 offspring produced.

had no significant effect on the production of frontal males (t6 ⫽ 1.197, P ⫽ 0.297) or females (t6 ⫽ 0.530, P ⫽ 0.615). Physical obstruction. Basal males were produced only by treatment ramets, which also produced more frontal males than controls (t6 ⫽ 6.400, P ⬍ 0.001). Food deprivation. Basal males were produced only by treatment ramets, frontal male production was not significantly different between treatment and control ramets (t6 ⫽ 1.960, P ⫽ 0.097), and females were produced only by control ramets. Food deprivation, therefore, stimulated the production of basal males but depressed the production of frontal sexual zooids, having a relatively greater influence on frontal females than on males. Temperature shock. Basal males were produced only by treatment ramets, forming a complete ring around the colonial perimeter. Frontal males were produced occasionally by treatment ramets but not by controls. Neither treatment nor control ramets produced females. Desiccation. Basal males were produced only by treatment ramets. Desiccation significantly depressed the production of frontal males (t6 ⫽ 2.611, P ⫽ 0.04) and completely inhibited the production of females. Physical damage. Peripheral physical damage induced the production of basal males, which remained absent in controls. Frontal males occurred with equal mean frequency in treatment and control ramets. Destruction of peripheral autozooids totally inhibited the production of females. Destruction of areas of zooids deeper within the colony stimulated the production of basal males, had no significant effect on the production of frontal Table 3. Number of offspring sired by donor clones whose mixed sperm suspensions had been presented in equal proportions to separate receptor ramets of the same clones

A1 E1 M1

A1 E1 M1

A1

E1

Sperm Competition. Fair-raffle competition. Genotyping was success-

ful for all but 3 of the 550 offspring sampled. Offspring were sired in approximately equal proportion by each donor clone (Table 2; goodness-of-fit ␹2 summed across receptors ⫽ 6.460, df ⫽ 10, P ⫽ 0.775). The relative contributions of the three donor clones were not significantly different in the first and last batches of offspring to be produced (Table 2; contingency ␹2 summed across receptors ⫽ 11.162, df ⫽ 10, P ⫽ 0.345), indicating that sperm had been captured and stored randomly. Autosperm–allosperm interaction. No offspring were sired by autosperm, and, both in the presence and absence of autosperm, the numbers of offspring sired by unrelated sperm donors were not significantly different from random expectation (Table 3). Influence of multiple paternity on egg production. The donor clone(s) sired all offspring, and consequently, there was no selffertilization. The number of offspring produced by ramets receiving allosperm from either one or two donor clones was not significantly different (Table 4), indicating that the mixed sperm must have shared a limited supply of ova. Discussion Normally, basal males are much less frequent within wild colonies than frontal males (26, 37, 40). The wide variation in sex ratio and sexual allocation recorded in apparently healthy wild colonies is, therefore, attributable to variation in the extent and composition of frontal budding (25, 37, 40, 41). Stressed wild Table 4. Number of offspring produced by ramets receiving sperm from two or one allosperm donor clones Receptor

Donor clone Receptor ramet

males (t6 ⫽ 1.549, P ⫽ 0.172), and depressed the production of females significantly (t6 ⫽ 3.790, P ⫽ 0.009).

M1

Sperm mixture containing autosperm 0 13 11 8 0 9 12 7 0 Sperm mixture lacking autosperm — 13 7 14 — 6 11 9 —

Probability*

0.503 0.824 0.263

0.263 0.115 0.824

The experiment was run first by using triple mixtures that contained autosperm of the receptor clones and again by using double mixtures lacking autosperm. A sample of 20 offspring was collected from each receptor ramet. *Binomial probability of departure from equal representation of outcrossed matings. 10328 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.1334011100

A1 E1 M1 A1 E1 M1 A1 A1 E1 E1 M1 M1

Donor(s) Two donors* E1, M1, (A1) A1, M1, (E1) A1, E1, (M1) E1, M1 A1, M1 A1, E1 One donor E1 M1 A1 M1 A1 E1

No. of offspring 37 44 56 44 39 61 47 39 52 34 62 58

Two-way ANOVA. Receptor identity: F2,6 ⫽ 9.471, P ⫽ 0.014, number of donors: F1,6 ⫽ 0.241, P ⫽ 0.641; interaction: F2,6 ⫽ 0.008, P ⫽ 0.992. *Discounted autosperm donor is shown in parentheses.

Hughes et al.

quently, it is reasonable to expect that sperm competition should linearize the male gain curve in large breeding populations of C. hyalina. Ultimately, there must be a maximum number of active males that can be supported per autozooid, perhaps limiting the scope of a colony to respond to sperm competition. Indeed, total costs of sperm production are known to be significant in other taxa (51, 52). From the present data, however, we may conclude reasonably that the male gain curve of C. hyalina is increasing monotonically and is probably approximately linear in typical populations (obviating local mate competition). Moreover, local mate competition among kin is prevented by random larval settlement in C. hyalina (53). Shape of the female gain curve is less certain, but whereas the relatively short duration of dispersal (53) may cause larvae to settle at densities likely to generate competition for space or food (21, 28, 29), random assortment with respect to kin should discourage local resource competition that otherwise might decelerate the female gain curve. On the other hand, progeny through male function of a focal colony will be dispersed by sperm movement and subsequent larval dispersal from the female mates, whereas progeny through female function of the same focal colony will be spread only by larval dispersal. Local competition, therefore, should promote male allocation because paternal offspring of a competing colony are more likely than maternal offspring to reach less densely populated substratum. Stress-induced maleness is predicted to be less sensitive to the shapes of gain curves if stress influences the female function more than the male function (35). This assumption applies well to C. hyalina, in which brooding demands far more time and energy per gamete than spermatogenesis. Energetic consequences of the longer time required for producing seeds or embryos than pollen or sperm have preoccupied much theoretical work (6, 35, 54, 55), but the implication of survivorship is also critical. When mortality risk is high, the parent is likely to die before progeny are released but will probably survive long enough to produce copious male gametes. Day and Aarssen (56) use the time-commitment hypothesis to develop an evolutionarily stable strategy model predicting that smaller plants should allocate proportionally more to the male function because they are more vulnerable to mortality factors. This prediction is applicable to any aspect of individual status (35) that influences survivorship. If stress is confined to an area whose radius is less than the mean distance of pollen or sperm dispersal, afflicted individuals will still have access to nonstressed potential partners. Thus, in cases like C. hyalina where maternal investment prolongs risk to mortality, the flurry of male activity induced by factors such as mechanical damage or desiccation may be regarded as an act of reproductive bailout. Another example of reproductive bailout may be found in the tape worm Schistocephalus solidus, if it is assumed that the observed inverse relationship between relative male allocation and body size is a response to the decreased life expectancy of individuals stunted by environmental conditions (57). We conclude that increased proportional allocation to the male function is a general response of hermaphroditic modular organisms to constrained productivity and increased risk of mortality associated with localized environmental stress.

1. Darwin, C. (1877) The Different Forms of Flowers on Plants of the Same Species (John Murray, London). 2. Galen, C. (2000) Am. Nat. 156, 72–83. 3. Heslop-Harrison, J. (1957) Biol. Rev. 32, 38–90. 4. Cobb, N. S., Trotter, R. T. & Whitham, T. G. (2002) Oecologia 130, 78–87.

5. Lokesha, R. & Vasudeva, R. (1993) Curr. Sci. 65, 238–242. 6. Henslow, G. (1888) The Origin of Floral Structures through Insect and Other Agencies (Kegan Paul, Trench & Co., London). 7. Freeman, D. C., Harper, K. T. & Charnov, E. L. (1980) Oecologia 47, 222–232. 8. Charnov, E. L. & Bull, J. (1977) Nature 266, 828–830.

Hughes et al.

Nick Hardman conducted microsatellite genotyping. We thank Lisa Angeloni, Ric Charnov, John Harper, Howard Lasker, Geoff Parker, and Phil Yund for incisive comments; any errors remain our own. We acknowledge John Harper’s conceptual objection to use of the term ‘‘stress’’ in an ecophysiological context. This work was supported by Natural Environment Research Council (London) Grant GR310226 (to R.N.H. and J.D.D.B.).

PNAS 兩 September 2, 2003 兩 vol. 100 兩 no. 18 兩 10329

EVOLUTION

colonies have not been examined specifically, but in the congener Celleporella bathamae males often occur ‘‘where normal budding pattern has been affected by compaction or overcrowding’’ (ref. 42, p. 46). Because basal males are usually budded from peripheral autozooids, adjustment primarily affects the sex ratio of zooids budded during the stressful period. A lesser influence may impinge deeper within the colony when occasional autozooids are converted directly into basal males. Such conversion not only will increase the relative number of local males but also will deprive neighboring females of the nutrition necessary to sustain embryogenesis (37). We examined only the new populations of zooids budded under experimental conditions because scanning the preexisting inner colony for scattered basal males would have been too time consuming for the small amount of information to be gained. The increased proportional allocation to males in response to growth obstruction, reduced food supply, temperature shock, desiccation, and physical damage (Table 1) appears to represent a general response of C. hyalina to environmental stress, as predicted for plants by Freeman et al. (7). Specifically, the complete suppression of female production in most treatments matches prediction for the limiting case when environmental stress acts locally on individuals within an otherwise nonstressed population (figure 1 in ref. 7). More generally, sex-allocation theory (33–35, 43) emphasizes the potential importance of local mate competition and local resource competition on gender modification. Local mate competition occurs when male gametes of the focal individual and its kin in socially structured populations saturate the supply of eggs and, therefore, compete with each other for fertilization; local resource competition occurs when offspring compete with kin for resources such as nutrients, water, food, or space. Both types of competition tend to decelerate the respective gain curves describing the fitness accrued by each gender as proportional resource allocation increases (33, 35, 44). Preferential male allocation is predicted if the male gain curve is linear and the female gain curve is decelerating, which might occur if pollen is widely dispersed but restricted seed dispersal generates sibling competition (45). Breeding populations of C. hyalina are typically large; hundreds of colonies often occur within a radius of a few meters (R.N.H., unpublished data), and the efficient uptake and storage of dilute sperm suspension extends the effective fertilization distance by water-borne sperm (46). In such a situation, fairraffle sperm competition should prevent deceleration of the male gain curve because greater proportional representation in the sperm pool will improve the chance of paternity when not overridden by differential sperm quality. This prediction is supported by data (41, 47) showing that among colonies of C. hyalina grown in the field, increased numbers of male zooids per colony (and, therefore, total colonial sperm production) were correlated with increased frequency of paternity (cf. ref. 48). Furthermore, a linearizing effect of sperm competition on the male gain curve has been demonstrated in the compound ascidian Botryllus schlosseri (49), which, like C. hyalina, fertilizes retained eggs with water-borne sperm. Paternity analysis in flowering plants shows that male fitness is not necessarily correlated with the amount of pollen produced; this finding implicates differential pollen quality (50). The present data, however, suggest that qualitative differences among sperm suspensions released by donor colonies unrelated to the receptor are inconsequential in C. hyalina because receptor colonies accept sperm randomly (Tables 2 and 3). Conse-

9. Charnov, E. L., Gotshall, D. W. & Robinson, J. G. (1978) Science 200, 204–206. 10. Sibly, R. M. & Calow, P. (1986) Physiological Ecology of Animals: An Evolutionary Approach (Blackwell Scientific, Oxford). 11. Jackson, J. B. C. (1977) Am. Nat. 111, 743–768. 12. Harper, J. L. (1977) Population Biology of Plants (Academic, London). 13. Hughes, R. N. (1989) A Functional Biology of Clonal Animals (Chapman & Hall, London). 14. Hoare, K., Hughes, R. N. & Goldson, A. J. (1999) Mar. Ecol. Prog. Ser. 188, 3–79. 15. Hoare, K. & Hughes, R. N. (2001) Mar. Biol. (Berlin) 139, 147–162. 16. Bloom, A. J., Chapin, F. S. & Mooney, H. A. (1985) Annu. Rev. Ecol. Syst. 16, 363–392. 17. Hughes, R. N. & Cancino, J. M. (1985) in Population Biology and Evolution of Clonal Organisms, eds. Jackson, J. B. C., Buss, L. W. & Cook, R. E. (Yale Univ. Press, New Haven, CT), pp. 153–186. 18. Ghiselin, M. T. (1969) Q. Rev. Biol. 44, 189–208. 19. Knowlton, N. & Jackson, J. B. C. (1993) in The Natural History of Inbreeding and Outbreeding: Theoretical and Empirical Perspectives, ed. Thornhill, N. W. (Univ. of Chicago Press, Chicago), pp. 200–249. 20. Jarne, P. & Charlesworth, D. (1993) Annu. Rev. Ecol. Syst. 24, 441–466. 21. Cancino, J. M. (1986) Monogr. Biol. 4, 279–308. 22. Seed, R. & O’Connor, R. J. (1981) Annu. Rev. Ecol. Syst. 12, 49–74. 23. Martı´nez, D. E. (2002) in Reproductive Biology of Invertebrates, ed. Hughes, R. N. (Wiley, London), Vol. 11, pp. 115–140. 24. Hughes, R. N., Manrı´quez, P. H. & Bishop, J. D. D. (2002) Proc. Natl. Acad. Sci. USA 99, 14884–14886. 25. Cancino, J. M. & Hughes, R. N. (1987) J. Exp. Mar. Biol. Ecol. 112, 109–130. 26. Cancino, J. M. & Hughes, R. N. (1988) J. Zool. Soc. London 215, 167–181. 27. Hunter, E. & Hughes, R. N. (1995) Mar. Ecol. Prog. Ser. 120, 193–201. 28. Buss, L. W. (1979) Nature 281, 475–477. 29. Okamura, B. (1992) Ecology 73, 1502–1513. 30. Wood, V. & Seed, R. (1980) Cah. Biol. Mar. 21, 135–154. 31. Harvell, C. D. (1984) Science 224, 1357–1359. 32. Parker, G. A. (1990) Proc. R. Soc. London Ser. B 242, 120–126.

10330 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.1334011100

33. Charnov, E. L. (1982) The Theory of Sex Allocation (Princeton Univ. Press, Princeton). 34. Lloyd, D. G. (1983) J. Theor. Biol. 105, 525–539. 35. Lloyd, D. G. & Bawa, K. S. (1984) Evol. Biol. 17, 255–338. 36. Manrı´quez, P. H., Hughes, R. N. & Bishop, J. D. D. (2001) Mar. Ecol. Prog. Ser. 224, 87–92. 37. Manrı´quez, P. H. (1999) Ph.D. thesis (University of Wales, Bangor). 38. Bishop, J. D. D., Manrı´quez, P. H. & Hughes, R. N. (2000) Proc. R. Soc. London Ser. B 267, 1165–1169. 39. Hoare, K., Hughes, R. N. & Gliddon, C. J. (1998) Mol. Ecol. 7, 355–356. 40. Hughes, D. J. (1992) J. Anim. Ecol. 61, 291–306. 41. McCartney, M. A. (1997) Evolution (Lawrence, Kans.) 51, 127–140. 42. Ryland, J. S. & Gordon, D. P. (1977) J. R. Soc. N. Z. 7, 17–49. 43. Charnov, E. L. & Bull, J. J. (1985) J. Theor. Biol. 115, 619–624. 44. Brunet, J. (1992) Trends Ecol. Evol. 7, 79–84. 45. McKone, M. J., Lund, C. P. & O’Brien, J. M. (1998) Am. J. Bot. 85, 776–783. 46. Bishop, J. D. D. (1998) Proc. R. Soc. London Ser. B 265, 725–731. 47. Yund, P. O. & McCartney, M. A. (1994) Ecology 75, 2151–2167. 48. Petersen, C. W. & Fischer, E. A. (1996) Evolution (Lawrence, Kans.) 50, 636–645. 49. Yund, P. O. (1998) Ecology 79, 328–339. 50. Snow, A. A. & Lewis, P. O. (1993) Annu. Rev. Ecol. Syst. 24, 331–351. 51. Nakatsuru, K. & Kramer, D. L. (1982) Science 216, 753–755. 52. Olsson, M., Madsen, T. & Shine, R. (1997) Proc. R. Soc. London Ser. B 264, 455–459. 53. Orellana, M. C. & Cancino, J. M. (1991) Bull. Soc. Sci. Nat. Ouest France HS 1, 309–316. 54. Schlessman, M. A. (1988) in Plant Reproductive Ecology: Patterns and Strategies, eds. Lovett Doust, J. & Lovett Doust, L. (Oxford Univ. Press, Oxford), pp. 139–153. 55. De Jong, T. J. & Klinkhamer, P. G. L. (1994) J. Ecol. 82, 399–402. 56. Day, T. & Aarssen, L. W. (1997) Evolution (Lawrence, Kans.) 51, 988–993. 57. Scha¨rer, L., Karlsson, L. M., Christen, M. & Wedekind, C. (2001) J. Evol. Biol. 14, 55–67.

Hughes et al.