| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Biological Science, Florida State University, Tallahassee, Florida 32306-1100
 |
Abstract |
|---|
 TOP Abstract IntroductionMaterials and MethodsResultsDiscussionLiterature Cited |
|---|
|
Introduction |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionLiterature Cited |
|---|
Spatially restricted coexistence ultimately limits the size of colonial invertebrates that might otherwise exhibit indeterminate growth (Harvell and Grosberg, 1988). Because reproductive potential usually increases with colony size, size limits have the potential to impose reproductive costs. Consequently, many authors (Abrahamson, 1975; Hughes and Cancino, 1985; Harvell and Grosberg, 1988) have hypothesized that indeterminately growing clonal organisms should maximize fitness by increasing sexual allocation when extrinsic factors limit growth. This prediction has been supported by empirical data on diverse invertebrate taxa (Hauenschild, 1954; Braverman, 1974; Yamaguchi, 1975; Stebbing, 1980; Harvell and Grosberg, 1988).
In many colonial marine invertebrates (including sponges, cnidarians, bryozoans, and ascidians), conspecific interactions are mediated by allorecognition systems that restrict somatic fusion to self or close kin, thereby limiting spatial competition and spatially restricted coexistence to unrelated individuals (Buss, 1987, 1990; Grosberg, 1988). For taxa in which encounters between relatives are common (often because of limited dispersal of sexually produced progeny), allorecognition may serve as an adaptive mechanism of kin recognition (Grosberg and Quinn, 1986; Hart and Grosberg, 1999). If intraspecific competition is associated with extremely high costs, such as the energetic cost of agonistic behavior and the risk of injury or competitive exclusion (Buss et al., 1984), then intergenotypic fusion may be favorable despite the potentially severe fitness costs associated with it (Buss, 1982, 1987; Rinkevich and Loya, 1983; Barki et al., 2002; Rinkevich, 2002). Conversely, if low costs accompany non-fusion, then fusion may be too costly an alternative. Thus, the fitness consequences of intraspecific competitive encounters provide insight into whether the evolution of allorecognition systems is adaptive.
Colonial hydroids in the genus Hydractinia inhabit shells occupied by hermit crabs, where they often encounter conspecifics (Yund et al., 1987; Buss and Yund, 1988; Yund and Parker, 1989; Hart and Grosberg, 1999; Ferrell, 2004). A genetically coded allorecognition system determines whether contacting colonies will fuse (Grosberg et al., 1996; Mokady and Buss, 1996) or attempt to overgrow one another (Ivker, 1972; Buss et al., 1984). Here I present the results of laboratory experiments demonstrating that the fitness consequences of agonistic interactions between colonies of the undescribed species Hydractinia [GM] depend on the size at which competitive encounters are initiated and the growth form of the colonies involved. The context-dependent nature of competitive outcomes affects the production of immature reproductive zooids—an investment in future reproduction—and permits an assessment of whether allorecognition in Hydractinia is adaptive.
|
Materials and Methods |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionLiterature Cited |
|---|
Hydractinia [GM], as well as many Atlantic Hydractinia species (Yund et al., 1987; Buss and Yund, 1988, 1989; Frank et al., 2001), encrusts the surface of gastropod shells occupied by pagurid hermit crabs. Colonies are gonochoristic and polymorphic, possessing specialized reproductive polyps, or gonozooids, on which the gametes are produced. Sexual reproduction occurs during broadcast-spawning events in which gametes are released in response to light cues (Bunting, 1894; Ballard, 1942; Levitan and Grosberg, 1993). The crawling planula larva of Hydractinia attaches itself to a gastropod shell inhabited by a passing hermit crab (Yund et al., 1987). Metamorphosis subsequently occurs on the shell surface. When two or more colonies recruit to the same shell, the colonies usually grow into contact. The resulting competitive interactions occur commonly in nature (Yund et al., 1987; Buss and Yund, 1988; Yund and Parker, 1989; Hart and Grosberg, 1999; Ferrell, 2004).
Allorecognition and agonistic interactions in Hydractinia
The adaptive significance of kin fusion to the allorecognition system of Hydractinia is uncertain (Feldgarden and Yund, 1992; Yund and Feldgarden, 1992; Grosberg, 1992). In laboratory assays, the probability of fusion remains high between full siblings or between parent and offspring, but it drops precipitously between colonies of lesser relatedness (Hauenschild, 1954, 1956; Müller, 1964; Ivker, 1972; Grosberg et al., 1996). If kin interactions are rare in the field, however, discrimination between kin and non-kin may simply reflect imperfect discrimination of self from non-self (Feldgarden and Yund, 1992; Yund and Feldgarden, 1992). Although kin interactions may be common in natural populations of H. symbiolongicarpus (Hart and Grosberg, 1999), adaptive kin/non-kin discrimination should persist only if kin fusion results in a net increase in a genotype’s fitness relative to non-fusion.
Upon contact with genetically unrelated conspecific tissue, Hydractinia colonies do not fuse but instead usually mount a well-characterized agonistic response (Ivker, 1972; Buss et al., 1984). The ability to mount an agonistic attack, or competitive ability, has a strong genetic basis and depends directly on colony growth form (Buss and Grosberg, 1990). (See Fig. 1 for an illustration of growth forms.) Some individuals grow solely by expansion of continuous, unbranching tissue ("mat" phenotypes), whereas others grow primarily by proliferating thin branches of tissue, or stolons ("stoloniferous" phenotypes). Colonies exhibit continuous variation between these two morphological extremes ("intermediate" phenotypes). In agonistic interactions, existing stolons swell due to the recruitment of nematocytes (cells bearing the nematocysts) and discharge of nematocysts (highly specialized stinging organelles) (Buss et al., 1984). These "hyperplastic stolons" (Ivker, 1972) discharge nematocysts in an attempt to competitively exclude the opposing colony by tissue destruction and subsequent overgrowth (Buss and Grosberg, 1990). Thus, colonies with greater stolon proliferation (stoloniferous phenotypes) have an increased capacity to mount an agonistic response and competitively exclude others in laboratory competitive interactions between small colonies (Buss and Grosberg, 1990).
|
Laboratory studies
In July 2000, I hand-collected five large, well-established, sexually mature colonies (designated colonies I–V hereafter) of H. [GM] from shallow water (1–1.5 m deep) in St. Joseph’s Bay, Florida, located in the northeastern Gulf of Mexico. Each colony occupied an individual gastropod shell (Fasciolaria lilium hunteria, F. tulipa, Polinices duplicatus, or Phyllonotus pomum) inhabited by the hermit crab Pagurus pollicaris. In the laboratory, small pieces of tissue, or explants, including four or five gastrozooids were excised with a scalpel and transferred with watchmaker’s forceps to plain glass microscope slides (7.64 cm x 3.82 cm, 1.1–1.3 mm thick). These tissue explants were held in the desired position using a single monofilament thread (8-lb. test) and kept in artificial seawater (Instant Ocean Sea Salt, ©1994; temperature = 24 °C, salinity = 32 ppt) in a single aquarium. Tissue explants attached to the glass substrate within a few days. This technique was used to establish all experimental replicates (described in detail below).
Two factors were manipulated in this experiment: (1) inter-competitor spacing and (2) relative competitive ability of competing colonies. Two competitive treatments were established by placing tissue explants from two different field-collected colonies on glass slides at 4-mm or 12-mm apart, resulting in interactions between small and large colonies, respectively, after subsequent colony growth. Such competitive interactions were established between all 10 pairwise combinations of the five experimental colonies. As a control, I also established each experimental colony singly on separate glass slides. All competing pairs and controls were replicated five times, yielding 125 experimental units [(10 pairs x 2 distances x 5 replicates) + (5 controls x 5 replicates) = 125 units]. Glass slides were mounted vertically on one of three acrylic plastic racks lined with nine plastic columns for attachment. Five slides could be mounted in each plastic column, generating five rows of slides on each rack. All slide positions were assigned randomly with respect to rack, column, and row. The three racks, including all glass slides, were maintained in artificial seawater in a single aquarium, as described above. Two underwater pumps ensured adequate circulation. Colonies were fed daily (3–4 h) to repletion with 2-day-old brine shrimp nauplii (Ocean Star International, Inc., Pro 100).
At 15, 30, 60, 80, and 90 days, each experimental unit was examined under a dissection microscope. Survival, colony width and length (omitted at the 80-d interval), and the number of immature gonozooids were recorded. The colonized surface area, defined here as the area covered by connecting stolon tips with straight lines or tracing the edge of mat tissue, was estimated by measuring colony width and length and then calculating the ellipsoid area [= 
x (length/2) x (width/2)]. The surface-area growth rate of colonies was then calculated (= ellipsoid area/time). Digital computer images were obtained from 53 additional laboratory-cultured H. [GM] colonies established from clonal explants, and linear regression was performed to evaluate the appropriateness of estimated ellipsoid area as an indicator of colonized surface area, which was calculated using SigmaScan Pro 4 software. Additionally, the presence or absence of contact between competitors was noted to estimate the time in contact for all replicates. For example, after 90 d of growth, the estimated time in contact for conspecifics in contact after only 15 d would be 75 d.
After 65 d of growth, all control replicates of each of the five experimental colonies were analyzed morphometrically to determine colony growth form (i.e., relative allocation to mat and stolonal tissue). In the absence of substrate limitation, mat and stolon growth rates remain constant throughout ontogeny (McFadden et al., 1984; Blackstone and Yund, 1989). Consequently, morphological assessment at this time can be considered to represent conditions throughout the experiment (0–90 d).
I used a Nikon camera (60-mm lens) interfaced to a computer to obtain and store black-and-white images of the colonies. After tracing all tissue (stolonal or mat) of each replicate (see Fig. 1 for examples), I used SigmaScan Pro 4 software to perform the morphometric measurements needed to calculate the shape metric developed by Blackstone and Buss (1991). The unitless shape metric [= perimeter/(area)0.5] is a reliable quantitative indicator of growth form in Hydractinia and other organisms that have similar colony development. Higher values correspond to increasingly stoloniferous phenotypes. The biomass surface area covered by mat tissue and individual stolons was also determined by using SigmaScan Pro 4 software. Note that the biomass surface area determined from these images of control-treatment colonies was calculated differently than the estimated ellipsoid surface area used to evaluate growth rate in all experimental units. Calculations of the shape metric and biomass growth rate allowed a posteriori assessment of the competitive ability of the five experimental colonies used in this study. The competitive-dominance hierarchy constructed using biomass growth rate and shape metric determinants was then compared to the actual competitive hierarchy based on the outcomes (in terms of colony survival) of the pairwise competitive encounters between all colonies.
Statistical analysis
I used Contingency 
2 tests of independence to evaluate whether colony survival was associated with (1) intercolony spacing, (2) colony identity, and (3) competitor identity. In cases in which expected values were sufficiently small (
5), Fisher’s exact test of independence was used instead.
Initially, three-way analyses of variance (ANOVA) were used to explore the effects of slide position on the two continuous response variables (surface-area growth rate and number of immature gonozooids) after 90 days of growth. The main effects of rack (3), column (9), and row (5) were examined, as well as all possible interactions (JMP, ver. 3.2.6). Then, mixed-model, two-way ANOVAs were used to analyze the surface-area growth rate and the number of immature gonozooids after 90 days of growth. Separate ANOVA models were employed for each of the five experimental colonies to meet model assumptions of independence. The main effects of distance between competing colonies (fixed; 4 vs. 12 mm) and competitor identity (random; genotype of competitor) were explored, as well as the interactions between these two main factors. For each distance treatment group, contrasts comparing colony performance against each competitor were performed (12 total contrasts, family level 
= 0.05, Bonferroni adjusted = 0.0042; Sokal and Rohlf, 1995). Differences from control colonies in growth and reproductive variables were examined separately for the 4-mm and 12-mm treatments with one-way ANOVAs; contrasts comparing control versus competitive treatment colonies were performed (4 total contrasts, family level = 0.05, Bonferroni adjusted = 0.0125; Sokal and Rohlf, 1995). In addition, the relationship between immature gonozooid production and quantitative estimates of colony growth form (i.e., shape metric) was examined using simple linear regression analysis for the 4-mm and 12-mm treatments. Natural logarithmic and square root data transformations were used when necessary to meet model assumptions of normality and homoscedasticity.
One-way ANOVAs were used to compare the mean biomass growth rate and mean shape metric of control replicates of the five experimental colonies. All pairwise comparisons were performed (10 total contrasts, family level = 0.05, Bonferroni adjusted = 0.005; Sokal and Rohlf, 1995).
|
Results |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionLiterature Cited |
|---|
Morphological assessment
Growth forms of experimental colonies included those that were entirely mat and those that were highly stoloniferous as well as intermediate types (Fig. 1). For ease of interpretation, I assigned Roman numerals I–V to colonies according to the extent of stolon proliferation. After 65 d of growth, none of the control replicates (established in the absence of competition) of colony I possessed any peripheral stolons branching out from the central ectodermal mat. In contrast, control replicates of colony V were highly stoloniferous, with most growth occurring via stolon rather than as mat tissue. Control replicates of colonies II, III, and IV displayed intermediate growth forms with various proportions of mat and stolon tissue.
Figure 1 compares quantitative estimates of growth form and growth rate of the experimental colonies. The growth rate, in terms of biomass area covered after 65 d of growth, was similar among experimental colonies, except colony V. Although ANOVA results indicated a difference in the mean biomass area of the colonies (F4,20 = 6.76, P = 0.001), Bonferroni pairwise comparisons (adjusted = 0.005) suggested that this is attributable solely to colony V, which produced biomass coverage at a rate greater than the others. The clear differences in growth form of colonies I, II, III, and IV (Fig. 1) did not result in significant differences in biomass production. Thus, morphological variation in H. [GM] does not necessarily reflect differences in the rate of biomass production. In contrast to biomass growth rate, the shape metric reveals a relationship between morphology and the growth differences between experimental colonies. As with biomass growth rate, ANOVA showed overall highly significant differences in the mean shape metric (F4,20 = 24.78, P < 0.001). However, Bonferroni pairwise comparisons indicated that the mean shape metric of colonies IV and V both differ significantly from all other colonies. The colonies can be ranked in order of decreasing allocation to stolon growth: V > IV > III = II = I. The "equal" (=) symbol denotes no statistical difference.
None of the five control replicates of colony I exhibited any stolon growth. This attribute reflects a qualitative morphological distinction between colony I and the other four colonies. Thus, while morphometric analysis did not indicate significant quantitative differences in stolon production between colony I and colonies II and III, an important qualitative difference was observed. Noting the absence of stolons in colony I, the experimental colonies can be ranked in order of decreasing stolon production as follows: V > IV > III = II > I.
Survival
Colony survival differed markedly as a function of intercolony spacing. After only 30 d of growth, competing colonies that had been established 4-mm apart experienced reduced survival compared to those established 12-mm apart or in the absence of competition (calc2 = 27.3, crit, 0.052 = 3.8, df = 1, P < 0.001), and these differences in survival remained significant for the duration of the experiment. For example, after 60 d of growth, survival of 4-mm colonies dropped to 55% and was significantly lower than the 85% survival seen in 12-mm colonies after 90 d (calc2 = 42.7, crit, 0.052 = 3.8, df = 1, P < 0.001). After 90 d, colonies in the 4-mm treatment showed less than half the survival rate of those in the 12-mm treatment (42% compared to 85%; calc2 = 39.9, crit, 0.052 = 3.8, df = 1, P < 0.001; Fig. 2). Thus, competitive encounters between smaller colonies incur greater (or at least earlier) mortality than encounters between larger colonies. By contrast, all control colonies had 100% survival, suggesting that intraspecific competition caused the observed mortality.
|
calc2 = 1.7, crit, 0.052 = 9.5, df = 4, P > 0.05). Although 4-mm clones experienced some mortality in all competitive pairings (Fig. 2), some 4-mm competitors imposed greater overall mortality than others (i.e., data pooled within each competitor column in Fig. 2; calc2 = 12.2, crit, 0.052 = 9.5, df = 4, P < 0.05). Colonies were least likely to survive competition with colony V and most likely to survive competition with colony I.
Survival after 90 d in 12-mm treatments was not significantly different among colonies when pooled over all competitors (i.e., data pooled within each colony row in Fig. 2; calc2 = 8.63, crit, 0.052 = 9.5, df = 4, P > 0.05). Most of the observed mortality occurred among replicates of colonies I and IV, however. Mortality in all other colonies (II, III, and V) was nearly negligible, as survival was 90% or higher. As in 4-mm treatments, some 12-mm competitors imposed greater overall mortality than others (i.e., data pooled within each competitor column in Fig. 2; calc2 = 15.7, crit, 0.052 = 9.5, df = 4, P < 0.01). In 12-mm competitive encounters in which mortality occurred (Fig. 2), colonies were killed by competitors IV and V almost exclusively.
Overall, colonies V and I were the superior and inferior competitors, respectively. Competitive relationships between intermediate competitors were less distinct, with the genotypes that exhibited the more stoloniferous growth forms generally dominating (Fig. 2).
Growth rate
Estimates of ellipsoid area were highly correlated with measurements of colonized surface area (R2 = 94.9%, n = 53), and were therefore used as an indicator of surface-area growth rate for experimental colonies. Colonies established 4 mm apart often suffered marked reductions in growth rate relative to controls; Bonferroni comparisons indicated that all competitor treatments (except one) were significantly different from controls for colonies IV and V, but not significantly different for colonies I, II, and III (Table 1, Fig. 3). In contrast, the growth rate of those in the 12-mm treatment was most often similar to controls; Bonferroni comparisons indicated that no competitor treatments were significantly different from controls, with two exceptions in colony V (Table 1, Fig. 3). Differences in growth rate after 90 d between 4-mm and 12-mm treatments were statistically significant for colonies II, IV, and V (Table 2).
|
|
|
Production of immature gonozooids
The number of immature gonozooids differed remarkably as a function of intercolony spacing for four out of five experimental colonies. Whereas 4-mm colonies invested very little or no energy into immature gonozooids (Fig. 4), 12-mm colonies exhibited immature gonozooid investment that was similar to control colonies initially (up to 30 d), and then in some cases greater than controls for the remainder of the experiment (see analysis below).
|
|
|
|
Discussion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionLiterature Cited |
|---|
Competitive outcomes
The fitness consequences of intraspecific competition in H. [GM] were heavily size-dependent. Not only may the duration of encounters between larger colonies be prolonged, but also the outcome may be entirely different. Whereas encounters at small size usually result in competitive exclusion (Ivker, 1972; Buss et al., 1984; Yund et al., 1987; Buss and Grosberg, 1990; this study), this may not always be true for the encounters of larger colonies. Indeed, when given a distinct size advantage over its superior interspecific competitor, the colonial hydroid Podocoryne carnea, Hydractinia colonies often survived interspecific contests and formed intercolony boundaries stable for 2–3 months (McFadden, 1986). Work with other spatial competitors of Hydractinia on artificial substrata similarly suggests that colonies may persist by inhibiting the growth of adjacent competitors rather than by attempting overgrowth (Karlson, 1978). In the present study, mat and mat-like Hydractinia phenotypes accrued zero mortality in intraspecific competitive interactions initiated between larger colonies.
Yund et al. (1987) also report coexistence, rather than overgrowth, of two competing Hydractinia genotypes established at distant recruitment sites after 9.5 weeks of growth. Even after 18 weeks, only four of eight replicates had resulted in exclusion of one of the two genotypes, despite significant differences in growth rate (and presumably competitive ability) between them. Had such interactions been investigated between genotypes with relatively low and similar growth rates (as in the present study), Yund et al. (1987) might have observed even less or no overgrowth in large-colony competitive interactions. Yund et al. (1987) proposed that the costs of non-fusion are actually much greater in interactions between large colonies. This conclusion implies that complete overgrowth, albeit prolonged, will proceed in all competitive encounters. If this were the case, fitness costs indeed would be more severe, because even the "winner" faces the immense energetic costs of such a prolonged struggle. My results are consistent with this interpretation in that highly stoloniferous colonies showed evidence of overgrowth regardless of intercolony spacing; however, in the absence of highly stoloniferous phenotypes, this does not appear to be the case. Overgrowth, while likely predominating in many Hydractinia populations as a result of recruitment patterns and available gastropod substrata (Buss and Yund, 1988; Yund and Parker, 1989), may not be the inevitable product of intraspecific competition in H. [GM].
Differences in the outcome of competitive encounters based on colony size are important ecologically and evolutionarily only if the frequency of encounters between small and large colonies varies predictably in natural populations. Field surveys indicate that competitive interactions between large colonies occur regularly in some H. [GM] populations with a frequency that varies predictably with the availability of different sizes and types of gastropod shells (Ferrell, 2004). Shell size and morphology affect the likelihood that multiple larvae will recruit to distant positions on a shell (Buss and Yund, 1988; Yund and Parker, 1989). Yund and Parker (1989) acknowledge that a sizeable portion of interactions may occur between large colonies in certain Hydractinia species. However, the mean shell length (
42 mm) of encrusted gastropod shells in some H. [GM] populations (Ferrell, 2004) greatly exceeds even the maximum shell length (25 mm) utilized in many northwestern Atlantic Hydractinia populations (as reported in Buss and Yund, 1988). On bigger shells, the spacing between colonies can be greater; thus competitive interactions between extremely large colonies, in which overgrowth seems increasingly improbable, occur in H. [GM] (pers. obs.). As in other marine cnidarians (e.g., Francis, 1973; Karlson, 1980; Purcell, 1977), agonistic behavior may result in territorial defense of spatial resources rather than competitive exclusion.
Production of immature gonozooids—an investment in future reproduction
Not only were instances of overgrowth attributable to colony morphology, but immature gonozooid production (and growth rate to a lesser degree) also varied as a function of the morphology of the competitors. When growth rate differed significantly among opponents, the less stoloniferous competitor experienced the greater in growth rate and produced few immature gonozooids, or none at all. These results are consistent with expectations of the intensity of aggression based on colony growth form.
Although production of immature gonozooids was higher in colonies encountering less formidable opponents (i.e., mat or mat-like intermediate growth forms), production in some cases was even higher than in the controls. This effect was not apparent in small-colony interactions, in which overgrowth proceeded much more rapidly. Even Hydractinia colonies with as few as two feeding polyps are capable of producing mature gonozooids (Hauenschild, 1954; Müller, 1964); nevertheless, their extreme vulnerability to overgrowth and the scarcity of tissue not directly involved in colony defense most likely kept the small colonies from producing many gonozooids. Only colonies of mat and mat-like intermediate morphologies increased their production of immature gonozooids. Hauenschild (1954) made several observations of Hydractinia that shed light on this trend: (1) the formation of mat tissue is a precondition for the production of gonozooids, (2) gonozooids form much earlier in mat colonies, (3) entirely mat phenotypes produce gonozooids naturally, even when occupying an unlimited substrate, (4) intermediate phenotypes do not develop gonozooids unless confronted with growth limitations, and (5) stoloniferous colonies require full colonization of surface before gonozooids arise. Thus, the greater production of immature gonozooids observed in mat and mat-like intermediates is simply a by-product of their larger supply of mat tissue upon which to form gonozooids, coupled with an earlier onset of sexual reproduction. If Hauenschild’s (1954) third observation is true, it may explain why the entirely mat phenotype, unlike the mat-like intermediate phenotypes, did not increase production of immature gonozooids substantially above control levels. (Table 1, Fig. 4). That is, with no growth limitations (as in the control treatment), the entirely mat phenotype would still exhibit significant immature gonozooid production, whereas the intermediate phenotypes, which react in this manner only in growth-limiting conditions, would not.
I explored two alternate explanations for the observed patterns in immature gonozooid production. First, perhaps the interacting pairs benefiting from positive effects on immature gonozooid production were relatives. None of the interacting colonies exhibited the fusion one might expect if they were genetically related. However, about 50% of Hydractinia full siblings do not fuse upon contact (Grosberg et al., 1996), and non-fusing kin could conceivably exhibit reduced aggression accompanied by reduced fitness costs. On the other hand, the probability of finding close relatives of H. [GM] on separate shells is extremely low (as indicated by extremely low rates of fusion between such colonies in laboratory assays; D.L. Ferrell, unpubl. data), and this explanation requires that three of the five experimental colonies (I, II, and III) be closely related. Consequently, this first possibility is unlikely. A second explanation could be found in the gender of the competitors. It is conceivable that only male-female (M-F) interactions result in increased production of immature gonozooids. In this experiment, only colony I was female. Unfortunately, only two of the four clear instances of augmented immature gonozooid production involved M-F pairs, suggesting these data may be inconclusive with regard to this explanation. There are six other M-F interactions, however, and only one (I vs. II) of these shows any evidence of increased production. Because the majority of M-F interactions contradict this hypothesis, the second explanation also appears unlikely. In contrast, the growth form of the interacting colonies explains much of the variation in immature gonozooid production (Fig. 5).
Because the rate at which H. [GM] extends across a substratum increases with stolon proliferation, mat and mat-like phenotypes grow more slowly. Although dissimilar morphologies may not differ in the rate of tissue production (y-axis of Fig. 1), they differ considerably in the rate at which a colony expands from its initial site of establishment. The surface-area growth rate measurement used in this study reflected this rate of expansion. ANOVA comparing the mean surface-area growth rate of control colonies revealed significant differences between the five experimental colonies (F4,20 = 17.9, P < 0.001; Fig. 3). Bonferroni comparisons (adjusted = 0.008) further indicated that the surface-area growth rates of colonies IV and V are both greater than those of colonies I, II, and III, but not significantly different from one another. Thus, colonies IV and V extended outward (via stolons) more quickly than colonies I, II, and III. As a result, competitive encounters involving colonies IV and V most likely initiated contact with competing conspecifics sooner than others did.
The greater production of immature gonozooids observed in some encounters in the 12-mm treatment might be attributable to the timing of contact between competitors such that more recently contacting colonies showed less reduction in production. Indeed, greater immature gonozooid production was observed in more slowly extending mat and mat-like phenotypes (i.e., colonies I, II, III). However, these competitive encounters yielded greater immature gonozooid production not only relative to other competing colony pairs, but also relative to control replicates, suggesting that the increases were not attributable solely to the timing of contact between conspecifics. Moreover, when the multiple regression model for immature gonozooid production (12-mm treatment) included the estimated time in contact as an independent variable, the result indicated negligible explanatory power for this added variable. The growth phenotypes of competitors (i.e., sum of shape metrics) remained significantly correlated with immature gonozooid production even after the time in contact between competitors was accounted for (Table 4). Interestingly, time in contact was not statistically significant in this model (Table 4). After a second independent variable (time in contact) was added to the model, the R2 increased only slightly, from 60.1% to 62.9% (Table 4), indicating that time in contact explained very little of the variation in immature gonozooid production.
|
Adaptive allorecognition and costs of competition
An adaptive allorecognition system is molded evolutionarily by the fitness consequences of the behavioral alternatives it mediates. Extremely high costs associated with non-fusion increase the likelihood that intercolony fusion will optimize colony fitness in intraspecific competitive interactions. Despite potentially severe costs of fusion between genetically distinct but closely related colonies (Buss, 1982, 1987; Rinkevich and Loya, 1983; Barki et al., 2002; Rinkevich, 2002), fusion may still increase fitness. On the other hand, if the costs of non-fusion and fusion are comparable, allorecognition systems may be nonadaptive or linked to other adaptive processes (Chadwick-Furman and Weissman, 2003). The current paper demonstrates that non-fusion entails significantly different fitness costs depending on the size and competitive ability of competitors. Thus, fusion behavior is likely to be adaptive in some ecological scenarios, but not others. As a consequence, complexities should be expected in the allorecognition system of Hydractinia if it functions adaptively in governing intergenotypic fusion.
The immediate costs of rejection were clearly greater in interactions between small colonies than between larger colonies. Small colonies experienced significantly greater fitness costs in terms of colony survival, growth rate, and immature gonozooid production. Competitively inferior colonies suffer higher rates of mortality (by definition) and are typically eliminated, whereas dominant colonies generally survive but suffer significant setbacks in growth and reproduction.
Because the size of the competitors affects the fitness consequences in conspecific interactions, it probably also affects the selective pressures that influence allorecognition (assuming its adaptive kin-discriminatory role) in Hydractinia. Interestingly, the allorecognition system of Hydractinia exhibits ontogenetic variation in full-sibling fusibility, or the tendency to fuse (Shenk and Buss, 1991; R.K. Grosberg, Univ. of California, Davis, pers. comm.). Shenk and Buss (1991) claim that fusibility declines with ontogeny and that some colonies disengage from fusion behavior with the onset of sexual maturity, suggesting that the costs of fusion are greater at this time. On the other hand, while noting some sort of ontogenetic effect, others have been unable to corroborate the decreased fusibility with colony ontogeny and the coupling of sexual maturity with the onset of rejection (Gild et al., 2003; R.K. Grosberg, pers. comm.). Moreover, the costs of intergenotypic fusion may be greatest early in ontogeny (R.K. Grosberg, pers. comm.) and are probably not averted by delayed rejection (Gild et al., 2003). That is, subsequent colony separation may provide little protection from somatic cell parasitism, the primary cost thought to accompany fusion, in which one genotype in a chimera contributes very little to somatic growth while contributing greatly to gamete production (Buss, 1982). Unfortunately, these conflicting reports make it difficult to incorporate the size-dependent variability in the costs of non-fusion reported in this study into a discussion of the possible adaptive significance of changes in fusibility with colony ontogeny.
My findings have a straightforward interpretation if the fitness costs of fusion truly increase with ontogeny and, hence, colony size. In this scenario, low costs of non-fusion would be accompanied by increased costs of fusion later in ontogeny. Thus, fusion would not be favorable in competitive interactions between large colonies, and decreased fusibility with ontogeny (Shenk and Buss, 1991) would be expected. Initially, it would seem that this expectation would be valid even if the costs of fusion actually were heavily skewed in favor of intense costs early in ontogeny. However, given intensified costs of early ontogenetic fusion, the costs of fusion late in ontogeny could be even lower than the low, but measurable, costs of non-fusion I have reported here. An opposite trend in fusibility then would be expected. Of course, ridding ourselves of such speculative deliberations depends on identifying and quantifying the fitness consequences of fusion in Hydractinia and understanding whether these consequences vary with colony ontogeny. Nevertheless, recognizing that the costs of non-fusion vary ontogenetically (Yund et al., 1987, this study), the presence of ontogenetic variation in fusibility is consistent with an adaptive allorecognition system intended to distinguish kin from non-kin.
Several avenues for future research remain unexplored. Long-term monitoring of prolonged encounters is needed to verify the nature of their dynamics and eventual outcome (overgrowth or coexistence). In particular, estimates of lifetime reproductive output, although difficult to obtain, would be most helpful. This study reveals that, in some circumstances, reproductive allocation may be modified in the presence of conspecifics, even in species that exhibit agonistic behavior. Similar effects may have been previously overlooked in other cnidarians with analogous inducible defenses. Lastly, close examination and quantification of the costs of fusion is needed to better evaluate the adaptiveness of kin fusion in Hydractinia.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
* [GM] is the designation given by Cunningham et al. (1991) to a species of Hydractinia that is found in the Gulf of Mexico but has not been formally described. Reference: C. W. Cunningham, L. W. Buss, and C. Anderson. 1991. Molecular and geological evidence of shared history between hermit crabs and the symbiotic genus Hydractinia. Evolution 45: 1301–1316. 
E-mail: ferrell{at}bio.fsu.edu
|
Literature Cited |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionLiterature Cited |
|---|
| ||||||||||||||||||||||||||||||||||||||
