Morphological, Physiological and Metabolic Comparisons Between Runner-Like and Sheet-Like Inbred Lines of a Colonial Hydroid

Heterochrony, an evolutionary change in the timing of development, is a pervasive theme in studies of evolutionary morphology (e.g. Gould, 1977; Alberch et al. 1979; Bonner, 1982; McKinney, 1988; Wake et al. 1991; McNamara, 1997). Nevertheless, the factors underlying most morphological heterochronies remain to be elucidated. Are morphological and life history heterochronies direct consequences of genetic heterochronies? Do metabolic and physiological parameters mediate heterochronic variation in some cases? Can seemingly unrelated genetic changes produce similar heterochronies by affecting the same metabolic or physiological properties? Here, selection and inbreeding experiments (e.g. Green, 1981; Monteiro et al. 1997a,b; Wilkinson et al. 1998) and morphological, physiological and metabolic assays of the resulting inbred lines are used to investigate the mechanistic basis of heterochrony in colonial hydroids.

Patterns of heterochrony in hydractiniid hydroids have been well characterized (Blackstone and Buss, 1991; Blackstone, 1992, 1996). Species of Hydractinia and Podocoryne generally exhibit contrasting suites of morphological and life history traits (‘sheets’ versus ‘runners’) and in this way exemplify patterns characteristic of many colonial animals (Buss and Blackstone, 1991). The morphology of these organisms can be idealized as comprising feeding and reproductive entities (e.g. polyps) that are connected to other entities by a fluid-carrying system (e.g. stolons; see Harper, 1977; Jackson, 1979). Runner-like forms show widely spaced polyps with little stolon branching and anastomosis, while sheet-like forms show closely packed polyps with extensive stolon branching and anastomosis. These different morphological patterns correspond to changes in the timing of the production of polyps and stolon tips relative to rates of stolon growth and colony maturation: high rates of production yield sheets, while low rates yield runners. Furthermore, in hydractiniid hydroids, relative rates of polyp and stolon production show an inverse correlation with rates of gastrovascular fluid flow to peripheral stolon tips. Compared with colonies of Podocoryne carnea, mature colonies of Hydractinia symbiolongicarpus exhibit a low rate of flow to peripheral stolons (Blackstone and Buss, 1992; Blackstone, 1996).

Experimental studies of heterochrony allow the between-species pattern in these hydroids to be mimicked by manipulation of colonies of a single species (Blackstone and Buss, 1992, 1993; Blackstone 1997a, 1998; for analogous studies of other organisms, see for example Stebbins and Basile, 1986; Meyer, 1987; Müller, 1991). Putatively, these hydroids incur substantial energetic costs in circulating the gastrovascular fluid throughout the colony. The application of dilute solutions of 2,4-dinitrophenol to colonies of Podocoryne carnea results in a condition of ‘loose-coupling’ of oxidative phosphorylation, a decrease in the amount of ATP available for generating gastrovascular flow and a consequent diminution of the rate of flow to peripheral stolons. Correlated with this diminished flow are changes that parallel patterns of heterochrony: the rates of production of polyps and stolon tips increase relative to rates of stolon growth and colony maturation. Alternatively, gastrovascular flow can be diminished by increasing the frequency with which a colony is fed (e.g. from three to six times per week), possibly because a higher rate of feeding increases the resistance of the stolon tissues to fluid flow or the absorption of fluid by these tissues, or both (Blackstone, 1997a; Van Winkle and Blackstone, 1997). Similar to treatment with uncouplers, feeding manipulation results in changes which parallel patterns of heterochrony; again, the rates of production of polyps and stolon tips increase relative to rates of stolon growth and colony maturation (see also Braverman, 1974). While increased feeding produces a surfeit of nutrients and seems in many ways the opposite of the energy-poor state produced by uncoupling, its effects on colony physiology (e.g. flow rate) are similar. In combination, the between-species data (Blackstone and Buss, 1992; Blackstone, 1992, 1996) and the experimental manipulations (Blackstone and Buss, 1992, 1993; Blackstone 1997a, 1998) suggest that flow rate is the principal physiological mechanism underlying heterochronic changes in these hydroid colonies.

Nevertheless, when the rate of gastrovascular flow in P. carnea is constant, the rate of polyp and stolon tip initiation can be increased by further shifting the cellular redox state in the direction of oxidation (Blackstone, 1997a, 1998). In fact, uncoupling of oxidative phosphorylation generally triggers metabolic activation and a shift of the redox state in the direction of oxidation (Heytler 1981; Hajnóczky et al. 1995), and these effects are observed in hydroids (Tardent, 1962; Blackstone and Buss, 1992, 1993; Blackstone, 1997a, 1998). Uncouplers function as proton ionophores, diminishing proton-motive force, raising the level of ADP, and thus strongly activating oxidative phosphorylation and shifting the mitochondrial matrix redox state in the direction of oxidation. In parallel, feeding triggers strong contractions of the polyp musculature (Wagner et al. 1998; Dudgeon et al. 1998), and the increased metabolic demands of these epitheliomuscular cells are probably responsible for the observed increase in oxygen uptake and the shift of the redox state in the direction of oxidation subsequent to feeding (Blackstone, 1997a). Thus, the experimental data are consistent with a direct effect of cellular redox state on the rate of polyp and stolon tip initiation, such that relative oxidation favors a high rate of initiation, while relative reduction leads to a low rate of initiation. Such a pattern is, in fact, concordant with the extensive work of an earlier generation of hydroid biologists (for reviews, see Child, 1941; Tardent, 1963; Rose, 1970).

To investigate the mechanistic basis of heterochrony further in these hydroids, a sheet-like and a runner-like inbred line of Podocoryne carnea were produced by selecting colony morphology at the time of the initiation of medusa production. The resulting inbred lines were evaluated at the F₅ and F₆ generations for differences in colony and medusa morphology, gastrovascular flow physiology and polyp epitheliomuscular cell redox state.

Mature colonies of Podocoryne carnea (Sars) release free-swimming medusae that must feed prior to producing gametes, and the swimming planula larvae subsequently colonize hermit crab shells or other surfaces (Edwards, 1972). Colony development in P. carnea begins with the metamorphosis of the planula larva into a primary polyp. Stolons extend from the primary polyp. The stolons encase fluid-filled canals that are continuous with the gastrovascular cavity of the polyp. In cross section, stolons consist of a fluid-filled lumen encased by endoderm, ectoderm and a rigid, acellular perisarc. Gastrovascular fluid circulates in the lumen of the stolons and carries food and possibly other metabolites from the feeding polyp to other parts of the colony; contractions of the muscular polyp largely propel the gastrovascular fluid (Schierwater et al. 1992; Buss and Vaisnys, 1993; Dudgeon and Buss, 1996; Van Winkle and Blackstone, 1997). Colony development from a primary polyp can be mimicked by surgically explanting 1–2 polyps from a colony onto a new surface. In both cases, P. carnea develops by lineal extension of the stolons, initiation of new stolonal tips by branching, formation of connecting stolons by anastomosis and iteration of feeding polyps (gastrozooids) on the stolons, forming a loose network of polyps and stolons typical of many runner-like forms. Once the available surface is covered, P. carnea colonies increase polyp and stolon tip formation, producing a more closely knit network of stolons and ultimately initiating the sexual (medusoid) phase of the life cycle as reproductive polyps (gonozooids) develop.

In colonies of various sizes, polyp epitheliomuscular cell contractions commence upon feeding and continue actively for less than 24 h; waste material is subsequently regurgitated, and the polyp becomes relatively quiescent until the next feeding (Wagner et al. 1998; Dudgeon et al. 1998). In response to feeding, colony oxygen uptake increases and the redox state of the epitheliomuscular cells shifts in the direction of oxidation; when polyps become quiescent, the redox state of these cells shifts in the direction of reduction (Blackstone, 1997a). Polyp and stolon tip initiation and stolon growth may occur principally when polyps are quiescent and not actively contracting (Blackstone, 1996; N. W. Blackstone, unpublished data).

Colonies of P. carnea were collected from the Yale Peabody Museum Field Station in Connecticut. A male and a female colony were identified, and clonal replicates of these colonies were propagated by explanting 1–2 polyps and connecting stolons onto glass microscope slides. Generally, colonies (i.e. polyp stages) were grown on glass slides or coverslips suspended in floating racks and in 120 l aquaria containing Reef Crystals artificial sea water (salinity 35 ‰) with temperature control to 20.5±0.5 °C, undergravel filtration and 50 % water changes weekly. Ammonia, nitrites and nitrates were maintained below detectable levels (Aquarium Systems test kits). Colonies were fed to repletion with brine shrimp nauplii 3 days per week. Analysis has shown that, under these culture conditions, ‘random’ statistical effects (e.g. time effects, tank effects, rack effects; see Sokal and Rohlf, 1995) are negligible (Blackstone and Buss, 1991).

All mating experiments were carried out using water from a tank that was completely free of any hydroids. Medusae were isolated from mature colonies contained overnight in finger bowls in an incubator at 20.5±0.2 °C. Medusae were cultured in finger bowls under similar conditions with daily feedings of brine shrimp followed by water changes. Under these conditions, medusae matured and produced gametes in 3–5 days. Subsequent to mating and larval maturation (3 days), competent larvae were induced to metamorphose by means of ionic imbalance (53 mmol l⁻¹ CsCl solution, see Blackstone and Buss, 1991), and metamorphosing larvae were deposited on 12 mm diameter glass coverslips. Small coverslips were used in the mating experiments to hasten maturation. In later assays of morphology and physiology using clonal replicates, 15 mm coverslips were used in part to allow more precise assays and in part to assess whether colony morphology remained consistent on a larger surface. On both sizes of coverslip, colonies were effectively confined to one side of the coverslip by cutting back encrusting stolons from the reverse side on a daily basis.

Medusae were isolated from the pair of field-collected colonies. Male and female medusae were cultured together, and embryos were isolated after 5 days. Larvae were matured and metamorphosed onto 12 mm coverslips. F₁ colonies were grown on one side of these coverslips until medusa buds appeared on gonozooids. Colonies were then imaged, and the areas of the empty (i.e. unencrusted) coverslip and of the individual polyps were measured (see below). These measurements were then used as a guide in selecting two sets of parents, one set sheet-like, the other set runner-like. These parents and their offspring defined the two inbred lines. Subsequent generations of each line were then propagated in a similar manner. At each generation for the sheet-like line, extreme sheet-like parents were selected, while for the runner-like line, extreme runner-like parents were selected. Often, however, these choices of parents were tempered by deficiencies in medusa production. Many extreme sheet-like colonies grew asexually with little or no medusa production, whereas some extreme runner-like colonies had very few gonozooids and thus produced too few medusae to be useful for breeding. The most extreme phenotypes could not therefore be used for breeding (Fig. 1). In later generations, inbreeding depression became apparent largely in terms of the viability of the gametes. Thus, only one F₆ colony of the sheet-like line was successfully produced (Fig. 1). Despite these difficulties, the inbred lines differed dramatically in colony morphology (Fig. 2). For the assays conducted, four runner-like F₆ colonies, one sheet-like F₆ colony and three sheet-like F₅ colonies were used.

Three clonal replicates of each of the four sheet-like and four runner-like colonies were explanted onto 15 mm coverslips. Generally, colonies covered the surface at the same rate and initiation of medusa production occurred at the same chronological time in both sheet- and runner-like lines (16±1 days after explanting; mean ± S.E.M.). At the time that medusa buds became visible on the gonozooids of a replicate, that replicate was measured using image analysis technology (e.g. Marcus et al. 1996). Briefly, a high-resolution MTI CCD-72 camera attached to a macro lens was used to project each colony onto a color monitor interfaced with a PC-compatible microcomputer (Pentium, 90 MHz CPU, 32 Mbytes RAM) equipped with an overlay frame grabber board (640×480 pixels with 12-bit depth per pixel). Using OPTIMAS software, background-subtracted images of the colonies were acquired using illumination appropriate to produce three distinct luminance thresholds: the polyps (lightest), the stolons (intermediate) and the empty coverslip (darkest). Using these thresholds, the software identified and measured the areas of the empty (i.e. unencrusted) coverslip and of the individual polyps (Blackstone and Buss, 1992). Classification macros were used to identify and exclude coverslip areas outside the edge of the colony. The total colony area and perimeter were also measured. Data files were analyzed using PC-SAS software.

Inbred lines were compared using nested univariate (ANOVA) and multivariate analysis of variance (MANOVA) (clonal replicates nested within colonies, colonies nested within inbred lines) for size (total area), for ‘size-free’ shape (perimeter divided by the square root of area, see Blackstone and Buss, 1991) and for the relationship between the total area of polyps and the total area of empty, unencrusted coverslip enclosed within the colony. Both polyp area and empty, unencrusted inner area were expressed as a fraction of the total area (note that the total area of stolons can be calculated as 1 minus this combined fraction, although this third variable was not used in the analyses). Polyp area is clearly a measure of polyp development; empty, unencrusted inner area is largely a measure of stolon branching and anastomosis (i.e. as these aspects of stolon development increase, inner area decreases). While polyps can shield empty inner area from observation and measurement, in practice this is a minor source of error because stolon development is generally most extensive at the base of the polyps. This is particularly true at the time of the initiation of the sexual (medusoid) phase of the life cycle and at subsequent times. Thus, at the time in development when morphology was measured, polyp area and unencrusted inner area behave as largely independent measures of two different aspects of colony development (Fig. 2; see further discussion in Blackstone, 1996). While some heterogeneity of variances was apparent in some of the measures used, generally all of these data approximately meet the assumptions of parametric statistics. Both natural logarithm and arcsine transformations provided a poorer fit to these assumptions.

Subsequent to imaging, all colonies were grown until medusae began to be released. All three replicates of one runner-like colony failed to release a single medusa despite repeated isolations; with the exception of this colony, the initiation of medusa release occurred at the same chronological time in both lines (21.9±0.6 days after explanting for sheet-like lines and 22.1±0.6 days after explanting for runner-like lines; means ± S.E.M.). Each colony was isolated nightly until a total of five medusae had been obtained. The morning after its release, each medusa was imaged while swimming horizontally. Images were taken when the medusa was relaxed, not contracted. The length of the bell (excluding the base of the tentacles) and the maximum width were recorded for each medusa. As a consequence of the failure of one runner-like colony to release any medusae, the nested analysis of variance (medusae nested within replicates, replicates nested within colonies, colonies nested within inbred lines) was highly unbalanced and could not be used effectively (see Sokal and Rohlf, 1995). Differences among the medusae from the sheet-like and runner-like inbred lines were therefore compared graphically and using a simple one-way multivariate analysis of variance of logarithm_e-transformed data.

At the time that colonies covered one side of the coverslips (2–2.5 weeks after explanting and immediately prior to the time that medusa buds became visible on the gonozooids), gastrovascular flow to three peripheral stolons was measured in each replicate for all four colonies of both inbred lines. Gastrovascular flow reaches a maximum 2–8 h after feeding (Schierwater et al. 1992; Wagner et al. 1998; Dudgeon et al. 1998); all these studies were carried out 3–5 h after feeding. The colony was placed in a flow-through chamber with a no.1 coverslip base (Warner Instruments). The chamber reservoir was maintained at 20.5±0.1 °C (Neslab RTE-100D). The temperature of the in-flowing sea water was further adjusted with a thermoelectric device to maintain a constant chamber temperature (20.5±0.3 °C; chamber temperature was monitored using a YSI cuvette thermometer with a flexible probe). Colonies were viewed on an inverted light microscope (Zeiss Axiovert 135) with a 40× Plan-Neofluar objective in differential interference contrast (DIC). Using the MTI CCD camera, three primary stolon tips from each colony were video-taped (at 30 frames s⁻¹) for 10 min each.

Gastrovascular flow must be reversed in each distal ‘dead-end’ tip. Stolon tips fill as fluid enters; the velocity of the fluid then decreases to zero. Tips then empty, and the fluid velocity again decreases to zero. In the region of the stolon immediately behind the tip, the difference between the width of the stolon lumen when it is at a maximum (and fluid velocity is zero) and when it is at a minimum (and velocity is again zero) provides a measure of the rate of gastrovascular flow if this difference is measured over time. These width measurements are taken at the base of the lumen. With the image analysis system connected to the video recorder, the width of the stolon lumen was measured at a point 250 μm behind the tip itself. In this region of the stolon, gastrovascular fluid velocity falls to zero as the lumen width approaches its maximum and minimum (thus, velocity itself need not be measured). Lumen width was measured when the stolon was full and when it was empty for three consecutive, but non-overlapping, cycles. For each cycle, the net change in the lumen width, i.e. the difference between the maximum and minimum lumen widths, was calculated. Perisarc-to-perisarc total stolon width (which is invariant throughout the contraction cycle) and the period (in seconds) of each cycle were also measured. The interpretation of these measures in terms of the volumetric rate of gastrovascular flow is discussed in detail in Blackstone (1996).

Statistical analysis thus focused on the three measured outcomes: the net change in lumen width per cycle, contraction cycle period and stolon width. These measures can be combined into a biologically meaningful measure of gastrovascular flow rate for each stolon contraction cycle: net change in lumen width divided by cycle period and stolon width (meters of lumen width expansion and contraction per total meters of stolon width per second). Biologically, this rate measure illuminates the ‘rate of supply’ of food to the tissues of the stolon tip. Both this rate measure and the individual flow parameters generally meet the assumptions of parametric statistics (see Sokal and Rohlf, 1995). To compare inbred lines, a nested analysis of variance was used with cycles nested within stolons, stolons nested within replicates, replicates nested within colonies and colonies nested within inbred lines.

The characteristic fluorescence of NADH and NADPH compared with the oxidized forms of these molecules has been used extensively to measure cellular redox state (for a review, see Chance, 1991). Currently, this technique is widely used (e.g. Pralong et al. 1992, 1994; Heineman and Balaban, 1993; Hajnóczky et al. 1995; Rohács et al. 1997). NAD(P)H fluorescence includes both mitochondrial and cytosolic compartments. Under physiological conditions, these compartments are in a slowly equilibrated steady state, and the redox states show corresponding behavior when subject to perturbation (Scholz et al. 1969; Hajnóczky et al. 1995).

Localized measures of NAD(P)H fluorescence were obtained using the Zeiss Axiovert 135 and ultraviolet light (excitation at 365 nm, barrier filter at 420 nm). Brief exposures were used, since hydroids are sensitive to ultraviolet light. A colony was contained in the flow-through chamber at 20.5±0.3 °C as described above. Images were recorded on film (10 s exposure, ASA 160 balanced for tungsten filaments), digitized and quantified with densitometry in OPTIMAS (brighter values relative to the dark background signal greater reduction). In such images, stolons appear dark, except for a weak signal from the chitinous perisarc (stolons lack the muscular fibers characteristic of polyp epitheliomuscular cells; Schierwater et al. 1992), while polyps show a much stronger signal. Because polyps are highly contractile in vivo, only the base can be used in precise between-polyp comparisons. In cross-sectional images of the base of a living polyp, the fluorescence of the base of the polyp epitheliomuscular cell fibers or myonemes can be clearly identified (Fig. 3). These fibers form a longitudinal network in a polyp, and their contractions drive the gastrovascular flow. Both the number of fibers fluorescing visibly above the dark background and the relative luminance of the fibers can be calculated (Fig. 3). Since all the fibers of a polyp are part of the same epitheliomuscular cell network, it is not clear that these individual fibers can be considered as being statistically independent. Therefore, the relative luminances of all visible fibers of a polyp were averaged, and this mean value was used in statistical comparisons.

Because these measures of NAD(P)H fluorescence were somewhat time-consuming, only one sheet-like and one runner-like colony were used. Nine clonal replicates of each colony were explanted onto 15 mm round coverslips and grown for 1 week. These replicates were divided into three groups of three replicates each. Three replicates were assayed 3–5 h after feeding, three replicates were assayed after being starved for 27–29 h, and three replicates were assayed after being starved for 27–29 h and treated with 30 μmol l⁻¹ 2,4-dinitrophenol in sea water for this period (for detailed protocols and discussions of uncoupler treatments, see Blackstone and Buss, 1992, 1993; Blackstone, 1997a, 1998). For each replicate of each treatment, three polyps were imaged and measured. A mixed-model analysis of variance (ANOVA) was used to detect between-treatment effects for each colony (polyps nested within replicates, replicates nested within treatments). Since the runner-like and sheet-like colony were tested at different times, direct statistical comparisons between runners and sheets were not made, but the effects of the treatments on the two colonies were compared graphically.

Colonies inbred for sheet-like morphologies initiate medusa production at larger total areas than runner-like colonies (Fig. 4A; using the colonies-within-lines effect as the error term, F=14.9, d.f.=1,6, P<0.01). At the initiation of medusa production, sheet-like colonies also exhibit more regular colony shapes, that is, smaller perimeter/area0.5 values (Fig. 4B; using the colonies-within-lines effect as the error term, F=15.0, d.f.=1,6, P<0.01). Sheet-like colonies also show a significant difference from runner-like colonies in the relationship between polyp area and unencrusted inner area (Fig. 4C; using the colonies-within-lines effect as the error term in a MANOVA, F=15.5, d.f.=2,5, P<0.01). For the most part, sheet-like colonies show greater stolon branching and anastomosis and thus considerably less unencrusted inner area enclosed within the colony. Note that three runner-like replicates cluster fairly close to the sheet-like colonies (Fig. 4C). These three replicates are all from colony 4 in Fig. 4A,B. Examination of the images suggests that this colony grows in a manner similar to other runner-like colonies; however, it initiates medusa production at such small sizes that anastomoses between stolons have not yet formed and thus little or no unencrusted area is enclosed within the stolons. This produces not only small total areas at the initiation of medusa production (Fig. 4A) and small inner area/total area ratios (Fig. 4C) but also relatively large polyp area/total area ratios (Fig. 4C). In general, some differences are apparent between the inner area and polyp area measures of the experimental colonies (Fig. 4C) and the same colonies as they were originally grown (Fig. 1). Differences in substratum size (12 mm versus 15 mm coverslips) and in initial polyp size (small primary polyp versus fully grown polyp explant) probably produce this discrepancy.

In general, medusae from sheet-like colonies are distinctly smaller than those from runner-like colonies (Fig. 5; MANOVA of log-transformed data, F=381, d.f.=2,102, P⪡0.001).

Prior to the initiation of medusa production, sheet-like colonies exhibit a significantly smaller gastrovascular flow rate than runner-like colonies (Fig. 6; using the colonies-within-lines effect as the error term, F=41.1, d.f.=1,6, P<0.001). Since flow rate is a composite of three measured flow parameters (net change in lumen width per cycle, contraction cycle period and stolon width), it is useful to examine the between-inbred line difference in these variables individually. Stolon width and contraction cycle period show weak and non-significant differences (Fig. 7B,C; F=4.6, d.f.=1,6, P>0.05, and F=3.9, d.f.=1,6, P>0.05, respectively, for each variable using the colonies-within-lines effect as the error term). The net change in lumen width during each cycle, however, shows a large and statistically significant difference (Fig. 7A; F=29, d.f.=1,6, P<0.002, again using the colonies-within-lines effect as the error term). Thus, the between-line difference in flow rate derives primarily from large differences in the amount that the lumen opens and closes with each contraction cycle in response to the polyp-driven gastrovascular flow, rather than from differences in stolon width or contraction cycle period.

Replicates of both the runner-like and the sheet-like colony exhibit similar responses 3–5 h after feeding, after more than 24 h of starvation and after more than 24 h of starvation combined with treatment with 30 μmol l⁻¹ 2,4-dinitrophenol (Fig. 8). Both feeding and treatment with the uncoupler dinitrophenol shift the cellular redox state in the direction of oxidation (fewer fibers are visible against the dark background and those that are visible exhibit lower values of relative luminance), while starvation for more than 24 h shifts the cellular redox state in the direction of reduction (more fibers are visible against the dark background and these fibers exhibit higher values of relative luminance). For replicates of the runner-like colony, both the number of fibers fluorescing (Fig. 8A) and the relative luminance of these fibers (Fig. 8C) show statistically significant between-treatment differences (F=5.9, d.f.=2,6, P<0.05, and F=6.3, d.f.=2,6, P<0.05, respectively, for each variable using the replicates-within-treatments effect as the error term). For replicates of the sheet-like colony, only the relative luminance of these fibers (Fig. 8D) shows statistically significant between-treatment differences (F=9.6, d.f.=2,6, P<0.05, using the replicates-within-treatments effect as the error term), while the number of fibers fluorescing (Fig. 8B) shows a non-significant effect (F=2.7, d.f.=2,6, P>0.05, using the replicates-within-treatments effect as the error term).

While the synthesis between development and evolution can be framed in strictly genetic terms (e.g. Pennisi and Roush, 1997), such an approach is probably an oversimplification. Genes are but one aspect of developmental mechanisms (Nijhout, 1990), and an understanding of the metabolic and physiological aspects of these mechanisms will be crucial to any definitive synthesis (e.g. Sinervo and Basolo, 1996; Zera et al. 1998). In this context, selection and inbreeding experiments (e.g. Green, 1981; Monteiro et al. 1997a,b; Wilkinson et al. 1998) can provide useful insights, particularly when combined with an experimental approach. In colonial animals, selection and inbreeding experiments have generally not been used; such organisms can nevertheless be examined using these methods (Mokady and Buss, 1996).

In the case of P. carnea, experimental manipulations have suggested a relationship between heterochronic variation and gastrovascular flow physiology. Selection and inbreeding of heterochronic variants of P. carnea support this hypothesis. Colonies selected on the basis of morphology and subsequently inbred produced distinct lines of runner- and sheet-like morphologies. Representatives of these inbred lines showed large differences in the rate of gastrovascular flow to peripheral stolon tips. Examination of the flow parameters suggests that this difference in flow rate derives primarily from the amount that the stolon lumen opens and closes with each contraction cycle. A strikingly similar pattern of flow parameter differences is obtained by treating colonies with uncouplers of oxidative phosphorylation such as dinitrophenol, in contrast to feeding manipulations, for instance (Blackstone, 1997a). The oscillations of the stolon lumen are in response to the gastrovascular flow, which is largely driven by contractions of the polyps. Polyps of runner-like colonies may thus have a greater capacity for large and sustained contractions than polyps of sheet-like colonies. Alternatively, polyps of sheet-like colonies may be supplying greater numbers of stolons with flow, thus diminishing the amount of flow to each stolon.

If polyps of runner-like colonies have a greater capacity for large and sustained contractions than polyps of sheet-like colonies, differences in metabolic capacity should be apparent between these inbred lines. Polyps of P. carnea initiate contractions in response to feeding and continue these contractions for less than 24 h (Wagner et al. 1998; Dudgeon et al. 1998). At wavelengths suitable for detecting NAD(P)H, fluorescent microscopic measurements of both the visible number of polyp epitheliomuscular cell fibers and their relative luminance indicate that the cellular redox state oscillates dramatically in a runner-like colony in response to these feeding-related contractions. At 3–5 h after feeding, polyps are contracting maximally, and the few muscle fibers visible against the dark background exhibit a low relative luminance. The redox state of these cells is thus probably shifted in the direction of oxidation because of the heavy metabolic demand and the consequent high levels of ADP (Chance and Baltscheffsky, 1958; Chance and Thorell, 1959; Scholz et al. 1969; Hajnóczky et al. 1995). Perturbations of the system using dilute solutions of dinitrophenol support this interpretation; polyps treated with dinitrophenol exhibit a similar number of muscle fibers and with similar luminance to polyps 3–5 h after feeding. In contrast, polyps starved for more than 24 h are relatively quiescent, and the large number of muscle fibers visible against the dark background exhibit a high relative luminance. Cellular redox state is thus probably shifted in the direction of reduction owing to the low metabolic demand and the abundance of substrate and ATP.

In the sheet-like colony, similar feeding-related oscillations in cellular redox state occur, but they appear to be less dramatic. There is a tendency for polyps 3–5 h after feeding and polyps treated with dinitrophenol to have fewer visible muscle fibers than polyps starved for more than 24 h but, in contrast to the runner-like colony, this trend is not statistically significant. In terms of relative luminance, the sheet-like colony exhibits a significant effect in the same direction as in the runner-like colony. Nevertheless, the relative luminance of muscle fibers from polyps starved for more than 24 h is not dramatically greater than that of polyps 3–5 h after feeding and that of polyps treated with dinitrophenol. These data suggest that the sheet-like colony contains less functionally oxidizable NAD(P)H than the runner-like colony, although further assays should be undertaken to support this interpretation. A lower level of functionally oxidizable NAD(P)H suggests diminished ‘reducing power’ and thus a diminished capacity to perform metabolic work (e.g. Harold, 1986). Polyp epitheliomuscular cells in sheet-like colonies may therefore lack the metabolic capacity to match the rates of gastrovascular flow found in runner-like colonies. Nevertheless, at this time, alternative explanations must also be considered. For instance, the extensive development of stolons in the sheet-like colony may require substantial amounts of substrate to be maintained. Because of this high allocation of substrate into the stolons, sheet-like polyps may be somewhat depleted of substrate more than 24 h after feeding, thus shifting the cellular redox state more in the direction of oxidation.

An unanticipated consequence of the selection on colony morphology was a strong effect on medusa size. Sheet-like colonies exhibited considerably smaller medusae than runner-like colonies. While this may represent a stochastic consequence of inbreeding, it also may be a genuine ‘trade-off’ between colony and medusa morphology. Additional inbreeding experiments should be carried out to test this hypothesis. Possibly, medusa size may be a consequence of selection on colony morphology. For instance, large polyps with extensive muscular development may be capable of the large and sustained contractions that produce high rates of gastrovascular flow to peripheral stolon tips and result in the development of runner-like colonies. Thus, selection for runner-like colonies may produce colonies with large polyps, and large medusae may typically develop from such large polyps. Apparent genetic ‘trade-offs’ between colony morphology and medusa-related life history traits (e.g. dispersal distance) might in this way derive from such simple structural consequences (Buss and Blackstone, 1991). To test this hypothesis definitively, polyp size must be experimentally manipulated and the effects on colony and medusa morphology must then be assessed (see Dudgeon and Buss, 1996; Sinervo and Basolo, 1996).

The metabolic and physiological differences discussed here might underlie the heterochronic differences in any or all of several ways. Changes in the timing of the production of polyps and stolon tips relative to rates of stolon growth and colony maturation require different patterns of signaling in runner- and sheet-like colonies. Different rates of gastrovascular flow in these colonies may provide hydromechanical signals which influence colony development (Van Winkle and Blackstone, 1997; Dudgeon et al. 1998). Gradients of morphogens emanating from polyps (Plickert et al. 1987) may be affected by cellular redox state (Blackstone, 1997b; Jantzen et al. 1998), thus also differentially influencing colony development. Finally, specific characteristics of the feeding-related oscillations of both gastrovascular flow and polyp redox state may signal different patterns of colony development. Biological and biochemical oscillations are ubiquitous (Chance et al. 1973), and amplitude and frequency modulation of such oscillations may be a common signaling mechanism (Berridge, 1997; Dolmetsch et al. 1997). Since some observations suggest that colony growth and morphogenesis are periodic, the last hypothesis may be the most likely.

Helpful comments were provided by L. Buss, B. Chance, S. Dudgeon and an anonymous reviewer. The National Science Foundation (IBN-94-07049) provided support.