Female investment is retarded pending reception of allosperm in a hermaphroditic colonial invertebrate

Abstract

Young colonies of the bryozoan Celleporella hyalinaare capable of acquiring water-borne allosperm and of using it to fertilize ova for a period of 3–6 weeks after reaching female sexual maturity. In these simultaneous hermaphrodites, early allocation to female modules, but not male, is greatly enhanced by the acquisition of allosperm. The degree of enhancement is inversely proportional to coancestry of the recipient and donor colonies, thus promoting outcrossing. This apparently novel mechanism of adjusting operational sex ratio depends on the uptake and storage of sperm by nonreproductive (somatic) modules and subsequent translocation to females.

Theory predicts (1) that gender allocation in simultaneous hermaphrodites is partially constrained by the fixed costs (2) of producing male and female reproductive organs. Fixed costs generally may be difficult to evaluate (1), hindering investigation of their possible influence on gender allocation. A suitable model organism for this purpose, however, is the bryozoan Celleporella hyalina, a modular colonial invertebrate normally growing epiphytically on macroalgae such as kelp. Colonies are composed of feeding male and female zooids (Fig. 1), sex ratio having both environmental and genetic components of variation (3–6). Fixed costs amount to the production and maintenance of the sexual zooids, which rely entirely on nutriment translocated from the feeding zooids (autozooids; ref. 7). Compared with the male function, female function bears the extra cost of producing a brood chamber (ovicell) within which the embryo receives placental nourishment (8). A relatively comprehensive analysis of reproductive resource allocation is possible with this system because mating costs are probably negligible; there is no cost of sexual display or mate acquisition, and there is, at most, a very low risk of disease transmission, whereas the cost of acquiring suspended sperm is largely obviated by use of the feeding current. Moreover, there is no possibility of sexual investment being subsidized by digestion of ejaculates, as there would be in copulating hermaphroditic taxa such as turbellarian flatworms and pulmonate gastropods.

Fig 1.

(A) Zooidal polymorphism in C. hyalina. a, autozooid (feeding zooid); f, female zooid (the triangular part is the female body and the perforated, spherical part is the brood chamber); m, male zooid (note the smaller aperture compared with autozooids). (B) Longitudinal section of an autozooid showing three groups of perforations near the base of the exoskeletal wall (labeled 1–3), through which strands of the funicular system connect to contiguous zooids. SEM photographs by Peter Wright.

C. hyalina is usually an obligate outcrosser (9) in which oocyte development requires the acquisition of allosperm (10) and enforced inbreeding results in severe loss of fitness (11). Female reproductive success, therefore, depends on the acquisition of allosperm. Assuming a limited energy budget with consequent allocation trade-offs (5) and given the greater scope for flexible resource allocation afforded by modularity (12), it may be expected that C. hyalina should minimize the fixed cost of producing relatively expensive female zooids until the reproductive success of such zooids is assured by the acquisition of allosperm. Moreover, because of the severity of inbreeding depression, the ability of allosperm to trigger female allocation should depend on coancestry. We present the results of experiments designed to test these hypotheses.

Methods

Colonies of C. hyalina were established from larvae in April 1996 and propagated as cuttings (ramets) to generate a corresponding set of clones (genets; ref. 13). Each clone was isolated from sources of allosperm, and, consequently, no ramets produced larvae before experimentation. Mating pairs E1/A1, D1/Q1, J1/Q1, and M1/A1 were established, where the first clone listed in each pair was the sperm donor and the second was the recipient. One ramet per recipient clone was dosed with a suspension of allosperm, obtained in sufficient quantity by using multiple ramets of the corresponding donor (13). After a period of 3 weeks, when all allosperm recipients bore mature embryos, larval release was induced by the dark/light reaction (14). Three larvae from each maternal ramet, constituting a “family” equivalent to a randomized block in the experimental design, were individually pipetted into plastic multiwell plates filled with 5 ml of 0.2 μm-filtered, UV-irradiated seawater (FSW). A circular piece of acetate, conditioned by prior immersion in seawater for 2 months and placed in the bottom of each well, served as the settlement substratum. Larvae settled within several hours, after which the FSW was replaced with a suspension of the unicellular alga Rhinomonas reticulata at a concentration of ≈100 cells per μl to serve as food. After 2 days, when metamorphosis was complete, the acetate was trimmed to an area of 4 × 4 mm bearing the ancestrula. This piece was glued with cyanoacrylate adhesive (15) to another piece of acetate measuring 75 × 45 mm and placed in a 300-ml glass jar two-thirds filled with FSW at 16 ± 2°C and fed daily with algal suspension. After a further 2 weeks when each ancestrula had budded two autozooids, the three sibling colonies from each mating were assigned to the treatments: reproductive isolation (control), single exposure to allosperm (donor H1), and repeated exposure to allosperm (initial donor H1, subsequent donor N2). At each dosage, sperm concentration was confirmed to be at least 10 per μl, ensuring maximum fertilization potential (13). Comparison between the control and experimental treatments would reveal any effect of allosperm on female allocation, whereas comparison between the single and repeated sperm-dosage treatments would reveal the duration of any sperm storage after precocious uptake. The use of different donor clones, together with microsatellite genotyping of offspring, was intended for studying sperm precedence (our unpublished data). Colonies were drawn at weekly intervals over 10 weeks with a dissection microscope at 40 × 10 magnification, a 1,200 × 1,200 μm graticule placed beneath the acetate substratum, and a camera lucida. Counts were made of autozooids, male zooids, mature female zooids (those bearing ovicells), and embryos within ovicells, from which were derived measures of relative female allocation (sex ratio: females per male), fertilization frequency (embryos per ovicell), and colony size [log₁₀(autozooids)]. The experiment was run from November 1997 to January 1998 and replicated a year later, using ramets from the stock clones to repeat mating between the same pairs of genets.

The weekly series of data were analyzed by repeated measures ANOVA. The treatment–families interaction was used as the error term for the treatment effect, and the time–treatment–families interaction was used as the error term for the treatment–time interaction.

The effect of coancestry on the ability of allosperm to promote female allocation was examined by preparing five receptor ramets, taken as cuttings from clones AE8, AM1, AM2, QD2, QD3, and QJ9 (AE = progeny of parental clones A and E, etc.). Four ramets were initially exposed for 2 h to allosperm suspension that had been obtained from maternal (coefficient of relatedness = 0.5), full-sib (c.r. = 0.5), half-sib (c.r. = 0.25), and unrelated (c.r. ≪ 0.25) donor clones. The fifth ramet was kept in reproductive isolation and served as a control for self-fertilization. Ramets were monitored immediately before experimental treatment and again 2 months later. Data were analyzed by the Kruskall–Wallis test.

Results

Access to allosperm had no significant effect on the production of males [repeated measures (RM) ANOVA, treatment F_2,6 = 1.251, P = 0.352] but significantly influenced the production of mature females (treatment F_2,6 = 321.829, P < 0.0001). Single exposure to allosperm caused a pulse of female production in weeks 3–5, whereas repeated exposure generated a more substantial, sustained increase, both with corresponding inflation of sex ratio (Fig. 2A). Control ramets did not produce embryos (Fig. 2B). Colonies exposed once to allosperm began producing embryos in week 3, peaking in week 4, and ceasing in week 8–9. Colonies repeatedly exposed to allosperm began to produce embryos in week 3 and to reach a sustained level at week 5. Colonies consisting merely of the ancestrula plus two autozooids, therefore, obtained and stored allosperm and, allowing 3–4 weeks for the process of brooding, used the store to fertilize ova for a maximum period of 4–6 weeks. Colonies repeatedly exposed to allosperm grew more slowly than those exposed once or not at all (Fig. 2C, RM ANOVA, time–treatment F_18,54 = 8.40, P < 0.0001). Qualitatively similar results to the above were obtained in a parallel experiment using uniparental families of ramets, cloned as cuttings from mature colonies, rather than biparental families of sibling larvae (15).

Fig 2.

Effect of allosperm acquisition on reproductive performance of colonies of C. hyalina. ▵, control treatment; □, single exposure to allosperm; ○, repeated exposure to allosperm. Data are grand means (n = 8) with SE (too small to be seen in most cases) of two experimental runs. (A) Female allocation expressed as sex ratio (number of mature female zooids per male zooid). (B) Fertilization frequency, represented by the proportion of ovicells containing embryos. (C) Colonial growth in the number of autozooids, which constitutes the basal layer of the colony and is responsible for feeding.

Coancestry strongly influenced the ability of allosperm to trigger female allocation. Thus, although initial sex ratio was not significantly different among treatments, final sex ratio was significantly higher in ramets exposed to unrelated and half-sib sperm than in any others (Fig. 3, Kruskall–Wallis test: initial sex ratio, χ² = 0.17, df = 4, P ≈ 1; final sex ratio, χ² = 21.05, df = 4, P < 0.001). Embryos were produced in all ramets exposed to unrelated and half-sib sperm but in only two ramets exposed to full-sib sperm, one ramet exposed to sperm from the maternal clone, and none exposed to autosperm suspension.

Fig 3.

Effect of coancestry of mating partners on female allocation by the colony receiving allosperm, expressed as sex ratio as in Fig. 2. White bars, initial sex ratio; black bars, sex ratio after exposure to allosperm. Data are means with SE, n = 6.

Discussion

Modulation of female allocation by C. hyalina in response to the availability of allosperm requires the capture and storage of sperm by nonreproductive organs and subsequent translocation to female zooids, often including females not yet budded at the time of sperm capture. Water-borne sperm are entrained by the lophophore current of autozooids and pass through the dorsal coelomopore (16). The mechanisms of allosperm storage and translocation remain to be discovered but, as shown above, are already present at the three-zooid stage of astogeny and perhaps even earlier. The funicular system, comprised of strands of tissue passing through perforations of the lateral exoskeletal walls of contiguous zooids (refs. 17 and 18; Fig. 1), might be the route of sperm translocation. Microsatellite analysis has confirmed that stored allosperm sire progeny (our unpublished data).

Simultaneous hermaphroditism is potentially advantageous if it promotes selfing as a fail-safe in the absence of opportunity for outcrossing (19, 20). Although this principle of reproductive assurance (21) could have influenced the origin of hermaphroditism in the Bryozoa, it cannot explain the persistence of hermaphroditism in obligate outcrossers such as C. hyalina. Whatever its historical or adaptive significance, simultaneous hermaphroditism in C. hyalina incurs fixed costs owing to the extreme zooidal polymorphism whereby sexual zooids are nutritionally dependent on somatic feeding zooids. Retarding the development of female zooids pending reception of allosperm reduces the fixed cost, leaving greater scope for allocation to colonial growth and male production. Spatially isolated colonies might, thereby, maximize reproductive success through the male function, although the effectiveness of disseminated sperm may be compromised by competing allosperm from other colonies nearer to potential recipients (22). The present data show that without access to allosperm, production of females by C. hyalina is severely retarded, although ultimately it is not suppressed (15) when, after several months, colonies approach the limits of their normal size and life expectancy (23). Male production is not similarly influenced, with the result that operational sex ratio is passively biased toward males in the absence of opportunity for outcrossing. This result is unlikely to be peculiar to our experimental population, because Chilean colonies of C. cf. hyalina also failed to produce female zooids when reared in reproductive isolation (24).

If the elevation of female allocation in response to allosperm acquisition is a mechanism for promoting outcrossing, strength of response should decline with increasing levels of coancestry, as indeed was the case. The step function relating sex ratio to degree of coancestry (Fig. 3) suggests the involvement of a Mendelian allorecognition system, qualitatively similar to those regulating fusion compatibility in the hydroid Hydractinia symbiolongicarpus (25) and the ascidian Botryllus schlosseri (26). Interestingly, the frequency of colonial fusion in C. hyalina follows precisely the opposite trend to sex ratio, declining abruptly among half-sibs (15), although it remains to be established whether or not gametic and fusion compatibility have the same genetic basis (cf. 27).

When mating strategies involve the recapture of water-borne sperm, an increase in male or hermaphrodite population density would be expected to translate into increased sperm competition (22). Intense sperm competition would yield a linear, nonsaturating, male gain curve, because male fertilization success continues to rise with increased allocation to sperm production (6, 28). However, at low densities, a saturating male gain curve is expected (28). Assuming a flexible two-way trade-off between male and female reproductive allocation and a nonsaturating female gain curve in this modular organism, increased female allocation might be predicted at low-population density (1). Inasmuch as experimental reproductive isolation simulates the extreme low end of population density, maximal female investment might be predicted in isolation. Exactly the opposite trend is observed here, and, particularly in nonselfing modular animals such as in many populations of C. hyalina, it may be vital to consider a three-way trade-off between male, female, and somatic investment (29).