Introduction

Bivalves with chemolithoautotrophic bacterial endosymbionts inhabiting their gills are common in many marine high-sulfide habitats, including hydrocarbon seeps, hydrothermal vents, and coastal reducing sediments (Fisher 1990). Hosts are phylogenetically diverse, with representatives in five families of bivalves (Fisher 1990). All of these symbiotic bivalves are believed to rely to varying degrees on their symbionts for organic carbon, and all of them have dramatically hypertrophied gills (Fisher 1990). For example, gills account for 13–23% of the wet weight of the vesicomyid clam Calyptogena magnifica (Fisher et al. 1988b), 25–38% of the mytilid Bathymodiolus thermophilus (Fisher et al. 1988a), 18–38% of the lucinids Lucinoma borealis and Loripes lacteus (Dando et al. 1986), and 26% of the protobranch Solemya reidi (Fisher and Childress 1986).

Hypertrophied gills are likely to be an adaptation to the symbiotic lifestyle. Large gills can accommodate more symbionts, an advantage given that the symbionts are providing the hosts with organic carbon. Housing the symbionts in a single layer of bacteriocytes adjacent to the seawater source of dissolved gases and other nutrients precludes the necessity for nutrient delivery via a vascular system, but also necessitates large surface areas. Furthermore, the sulfidinous habitats of these bivalves sometimes have low O2 concentrations, due to the rapid reaction of these two dissolved gases with each other (Scott et al. 1994). Low environmental O2 concentrations are a burden, given that the host clams require O2 for cellular respiration (Childress and Fisher 1992), and many of the symbionts utilize O2 as their terminal electron acceptor (Nelson and Fisher 1995). Large gills with large surface areas may be an adaptation not only for symbiont housing, but also to facilitate the acquisition of O2 from low-O2 environments at rates sufficient to supply both symbiont chemolithoautotrophy and host heterotrophy.

Large gill surface areas should also facilitate the exchange of CO2 with the environment. Rapid rates of CO2 exchange are suggested by the extremely 13C-depleted biomass δ13C values found in bivalves with chemolithoautotrophic symbionts, typically falling between −30 and −34‰ (Fisher 1990). This isotopic signature cannot be explained by unusual carbon fixation pathways in the symbionts, as all use Rubisco and the Calvin–Benson–Bassham cycle to fix and reduce CO2 (Fisher 1990; Robinson and Cavanaugh 1995; Cavanaugh and Robinson 1996). Instead, rapid rates of CO2 exchange may keep the pool of CO2 within the symbiont cells in near-isotopic equilibrium with environmental CO2, preventing the isotopic enrichment of intracellular CO2 by the Rubisco reaction, which fixes 12CO2 slightly more rapidly than 13CO2 (Scott et al. 2004). As gill structure and function play a major role in determining the rate of CO2 acquisition from the environment, they are likely to exert a powerful influence on symbiotic bivalve δ13C values.

Solemya velum is a symbiotic protobranch clam inhabiting subtidal sediments near eelgrass beds. Its large gills (Fig. 1) are packed with endosymbiotic bacteria (Zardus 2002). These γ-proteobacterial endosymbionts (Eisen et al. 1992) have been confirmed to be chemolithoautotrophs, based on high activities and immunolocalization of Rubisco within the symbiont cells (Cavanaugh 1983; Cavanaugh et al. 1988), and stimulation of gill carbon fixation by thiosulfate and sulfide (Cavanaugh 1983). Based on the similarity of δ13C values for the symbionts and the host clam (Conway et al. 1989), the reduced feeding apparatus in the clam, and its limited ability to filter feed (Krueger et al. 1992), the symbionts provide the majority of the organic carbon for the clam (Cavanaugh 1994). Accordingly, the host clam provides conditions favorable for chemolithoautotrophy by constructing a Y-shaped burrow in reducing sediments (Stanley 1970), which it flushes with its siphons to obtain O2 from bottom water and H2S from interstitial water (Cavanaugh 1985).

Fig. 1
figure 1

Solemya velum. Ventral view of a clam pried open with pins for dissection (gi gills, f foot, s shell).

The isotopically depleted δ13C values in S. velum, which lie between −31 and −34‰ (Conway et al. 1989), are likely the result of several factors acting in concert. The degree of isotopic discrimination by S. velum symbiont Rubisco is similar to other Rubiscos (ε=24.4‰, where \( \varepsilon {\text{ = \{ [}}R_{{{\text{CO}}_{{\text{2}}} }} {\text{/}}R_{{{\text{fixed}}}} {\text{] - 1\} $ \times $ 1,000}} \); Scott et al. 2004), so these depleted δ13C values do not arise simply from enzymatic causes. The δ13C of the dissolved inorganic carbon from interstitial water near its burrows is somewhat isotopically depleted, ranging from +1 to −6‰ (Scott et al. 2004), which may contribute. S. velum symbionts use extracellular CO2, and not HCO3 (Scott and Cavanaugh, unpublished data), which eliminates isotopic enrichment due to HCO3 uptake (Goericke et al. 1994). Based on the Rubisco ε value and biomass δ13C values, the intracellular CO2 pool is in near-isotopic equilibrium with environmental CO2 (Scott et al. 2004). Isotopic equilibrium is only possible if the rate of supply substantially outpaces the rate of fixation, and this will only be the case if gill surface areas are large enough to sustain rapid CO2 supply and exchange.

In an effort to better understand nutrient uptake, δ13C values, and other aspects of gill function in symbiotic bivalves, the allometry of S. velum gill size and surface area, as well as δ13C values, were determined. Allometric relations were developed for gill weights as a portion of total weight, as well as gill surface areas, as none of these parameters had been previously explored for any symbiotic bivalve. Gill weights were also measured for the nonsymbiotic protobranch clams Yoldia limatula and Nucula proxima for comparison. The δ13C values were measured for clams of a range of sizes to determine whether CO2 supply outpaced fixation for clams of all sizes.

Materials and methods

Bivalve collection and dissection

Solemya velum Say, 1822, Nucula proxima Say, 1822, and Yoldia limatula Say, 1831 were collected near Woods Hole, Massachusetts in December 2002. Clams were kept in chilled seawater (5–10°C) until their dissection, which occurred within 24 h of collection.

Gill weights

Clam adductor muscles were severed to open the shells, and gills and other soft tissues were gently blotted to remove seawater before transferring them to squares of prebaked, preweighed aluminum foil. Wet weights were measured immediately after dissection. Dry weights were measured after keeping the samples at 65°C for 36–40 h. Samples were weighed again after they were at 65°C for 6–10 h more to confirm that they had reached a constant weight.

Gill surface areas

S. velum gills, weighing 6 to approximately 80 mg, were dissected with great care for surface area measurements by grasping them with microforceps by their suspensory membrane to free them from the rest of the soft tissues. Gills were placed on glass microscope slides and the leaflets were gently spread with the microforceps. A ProgRes 3012 camera mounted on a Leica dissecting microscope was used to record digital images, and the camera was calibrated in accordance with the manufacturer’s instructions along both x and y axes to prevent image distortion. Two pictures were taken for each gill: from above (Fig. 2a) and obliquely (Fig. 2b). Digital images were scanned into Adobe Photoshop 5.5 and saved as PICT files. Gill micromeasurements were taken in NIH Image, which was calibrated in millimeters by measuring 1-mm rulers cemented to the glass slides next to the gills.

Fig. 2
figure 2

Solemya velum. Dissecting micrographs of gills illustrating the measurements taken to estimate their areas. a Top view. b Oblique view. Symbols described in Materials and methods

Gill surface areas were calculated from these micromeasurements. Measurement efforts were focused on the dorsal leaflets of the filaments (Zardus 2002), as the ventral leaflets had a tendency to curl on the slide, which made accurate measurements of their dimensions difficult. An attempt to flatten the ventral leaflets by mechanical means and by exposing the clams to the sedative propylene phenoxetol to relax the gill muscles prior to dissection resulted in gill degradation; accordingly, these methods were abandoned and dissections were conducted without pre-incubation or excessive gill manipulation.

Using the image taken from above the gill (e.g., Fig. 2a), the length of the gill axis was measured, and the number of filaments per millimeter along the axis (n) was calculated by choosing three areas of the gill that were in especially sharp focus and counting the filaments along a measured portion of the axis. The gill axis was divided into 16 sectors of equal length (l) and the length of the middle leaflet on the dorsal side of the gill (m i ) was measured. The number of leaflets in each sector (p) was calculated by multiplying n×l.

Measurements from the digital image of an oblique view of the gill (Fig. 2b) were calibrated by using the width of the top view (w). The oblique view was divided into the dorsal and ventral leaflets. The dorsal leaflet was divided into 16 sectors of equal width (x) and the height of the midpoint of each sector was measured (h j ). As the gill filaments did not lie perpendicular to the gill axis, parallel to the oblique view (Fig. 2a), the true width of each sector (y) was calculated as

$$ y = x \times m_{{\max }} /m_{{{\text{obl}}}} $$

where mmax is the true length of the longest leaflet (from the top view), and mobl is the apparent length of the longest leaflet in the oblique view. The area on both sides of the largest leaflet (Amax) was calculated as

$$ A_{{{\text{max}}}} = 2 \times {\left[ {{\left( {{\sum\nolimits_{j = 1}^{j = 15} {y \times h_{j} } }} \right)} + 1/2{\left( {y \times h_{{16}} } \right)}} \right]} $$

The surface area in a leaflet of a particular length (A i ) was calculated from

$$A_{i} = A_{{\max }} \; \times {\left( {m_{i} /m_{{\max }} } \right)}^{2} $$

and the total surface area of leaflets of this size (T i ) was estimated from

$$T_{i} = A_{i} \times p$$

The surface area of filaments on the dorsal leaflet of the gill (Ad) was calculated using

$$ A_{d} = {\left( {{\sum\nolimits_{i = 1}^{i = 16} {T_{i} } }} \right)} $$

and Ad was multiplied by four to account for the surface areas of the ventral leaflet and the other gill. Gill surface areas were divided by the total wet weight of the individual to obtain weight-specific gill surface areas in square centimeters per gram. Unfortunately, gills from Y. limatula and N. proxima were too small and delicate for these measurements to be possible.

Biomass δ13C values

Feet were dissected from S. velum clams of a range of sizes and dried at 65°C overnight. Samples were prepared for isotopic analysis at the Boston University Stable Isotope Laboratory by adding a drop of 1% platinum chloride in HCl into tin capsules containing 1 mg of material, allowing them to stand overnight, and placing them in a 60°C drying oven to dry fully. Samples were flash combusted with a Eurovector Euro EA elemental analyzer, which injected the sample CO2 into a GV Instruments IsoPrime mass spectrometer in continuous flow mode. The fidelity of the δ13C values [δ13C={[(13C/12C)sample/(13C/12C)standard]-1}X103 (‰)] was monitored by running internal glycine standards in parallel with the samples.

Results

Gill weight

Of the three species of protobranch clams examined, Solemya velum had the largest gills (Fig. 3). Power functions describing the allometric relation between gill and total weight, G=aMb, where G is the gill wet weight (in grams), and M is the total wet weight of the clam (in grams), indicate a decrease in gill weight per total weight for larger individuals (Table 1; b<1 in all cases; P<0.05 for S. velum and Nucula proxima). Dry gill weights are a similar proportion of dry total weights as wet weights (Table 2), indicating that the large gill wet weights are not due to high water content but to a substantial allocation in biomass. For the power equations describing gill weights as a fraction of total weights, the values of a are largest for S. velum, consistent with the generally larger gill weights measured for this organism, followed by N. proxima and Yoldia limatula (Tables 1 and 2).

Fig. 3
figure 3

Solemya velum, Yoldia limatula, and Nucula proxima. a Soft tissue wet weights of gills and whole clams (all soft tissues: gills, foot, visceral mass, etc.). b Soft tissue dry weights of gills and whole clams

Table 1 Solemya velum, Yoldia limatula, and Nucula proxima. Gill wet weight as a function of total wet weight. Values of a and b are for the scaling equation G=aMb, where G is gill wet weight (in grams) and M is total tissue wet weight of the clam (in grams)
Table 2 Solemya velum, Yoldia limatula, and Nucula proxima. Dry gill weight as a function of total dry weight. Values of a and b are for the scaling equation G=aMb, where G is dry gill weight (in grams) and M is total dry tissue weight of the clam (in grams)

Gill surface areas

S. velum gills had large surface areas (Fig. 4). When gill areas were normalized to the total weight of the organism (square centimeters gill surface area per gram wet weight), values of square centimeters per gram were extremely high (Table 3), and smaller individuals had higher values (b<1, P<0.05; Table 3). Interestingly, gill surface area scaled with gill weight; when gill surface areas were normalized to gill weight, b~1 (Table 3).

Fig. 4
figure 4

Solemya velum. Gill surface areas as a function of soft tissue weights (open circles), and the power function y=axb (solid curves). Values of a and b were estimated from linear regressions of log(y) versus log(x). a Gill surface areas are plotted against the weights of whole clams (all soft tissues: gills, feet, visceral mass, etc.). b Gill surface areas are plotted against the weights of the gills.

Table 3 Solemya velum. Gill surface areas as a function of total tissue and gill wet weights. Values of a and b are based on the scaling equation SA=aMb, where SA is gill surface area and M is total or gill weight

Biomass δ13C values

There does not appear to be a correlation between the δ13C value of foot biomass organic carbon and the wet weight of the whole clam; all values fell within 1‰ of the average value of −31.7±0.5‰ (SD; Fig. 5; r=0.55; Spearman’s rank correlation coefficient = 0.014).

Fig. 5
figure 5

Solemya velum. Stable carbon isotope composition of feet from clams with a range of soft tissue wet weights

Discussion and conclusions

Solemya velum gills have been dramatically modified to accommodate their dual roles of gas exchange and symbiont housing. They are substantially larger than those of other protobranch bivalves. This is especially so in smaller individuals as the scaling exponent (0.85) indicates that gill weights are a smaller portion of total weight in larger clams. This relation may be due to higher mass-specific requirements in small S. velum for oxygen from the environment and organic carbon from the symbionts to fuel growth. For S. velum of all sizes, mass-specific gill surface areas are the largest measured for a bivalve, exceeding 1 to 13.9 cm2 g−1 measured in other species (reviewed by Pouvreau et al. 1999; Andersen et al. 2002).

Large gill surface areas have been described for other organisms from high-sulfide habitats. The hydrothermal vent vestimentiferan Riftia pachyptila has a mass-specific gill surface area of 22 cm2 g−1 (Andersen et al. 2002), whereas a hydrocarbon seep orbiniid polychaete has values of about 10 cm2 g−1 (Hourdez et al. 2001). Previous to this study, the record holder was the hydrothermal vent polychaete Paralvinella grasslei, whose mass-specific gill surface area is 47 cm2 g−1 (Jouin and Gaill 1990). It has been suggested that these large gill surface areas are an adaptation to living in low-oxygen habitats (Jouin and Gaill 1990; Hourdez et al. 2001; Andersen et al. 2002). Given the extraordinarily powerful aroma of sulfide in sediments that S. velum inhabits (K. Scott, personal observation), and the rapidity of sulfide oxidation in seawater (Scott et al. 1994), oxygen concentrations may be chronically or periodically low in S. velum burrows. Alternatively, the clam may ventilate its burrow rapidly enough to keep oxygen tensions moderately high. In any case, S. velum must provide a sufficient supply of oxygen for its own metabolic needs, while simultaneously meeting the (substantial) oxygen demand of its chemolithoautotrophic symbionts. Large gills with high surface areas must be a benefit under these circumstances. Furthermore, large gills provide the possibility of housing more symbionts. The extraordinarily high gill surface areas reported here for S. velum are likely to be an adaptation both to ensure high rates of oxygen supply, and to meet the need for symbiont housing. Given that other bivalves with chemoautotrophic symbionts face similar pressures, large mass-specific gill surface areas are expected for them as well.

In contrast to hydrothermal vent vestimentiferans, which demonstrate a 4‰ increase in δ13C values over a similar size range (Fisher et al. 1990), the smallest (0.013 g) and largest (0.196 g) clams only differed by 0.6‰. Either the CO2 supply-to-demand ratio is similar for clams of all sizes, or the rate of supply always greatly outpaces demand. The first possibility seems unlikely, given that clams at different stages of growth probably have quite different demands for organic carbon.

To determine whether the rate of CO2 supply from the environment is large relative to demand, a simple model was constructed. Fick’s first law, which is

$${\text{Flux}} = D{\text{ }} \times {\text{ }}{\left[ {C_{1} - C_{2} } \right]}/x$$

where D is the diffusion constant, C1 and C2 are the concentrations of CO2 outside and inside the gill, respectively, and x is the diffusive distance, was used as a starting point. A value of 1.8×10−5 cm2 s−1 was used for D (Zeebe and Wolf-Gladrow 2003), and x was set equal to 0.001 cm, which is the maximum thickness of the microvilli and bacteriocyte cytosol separating the S. velum symbionts from the mantle cavity (Cavanaugh 1983). C1 is the concentration of CO2 in the mantle cavity, which should be similar to CO2 concentrations measured in the habitat, 20–200 μM (Scott et al. 2004). C2 was set to equal zero, as the purpose of this calculation was to estimate the gross rate of CO2 diffusion into the clam gill, not the net rate of CO2 diffusion into and out of the gill. The allometric relation between the gill surface area and clam soft tissue wet weight was used to estimate gill surface areas for clams spanning the weights of those dissected for this study (0.013 to 0.289 g).

The estimated flux, or delivery rate, of CO2 to the gills of a clam weighing 150 mg is approximately 120 to 1,200 μmol g−1 wet weight h−1 (Fig. 6), or 32 to 320% of clam organic carbon per day, which greatly exceeds carbon fixation rates measured in Solemya spp. The maximum rate of net CO2 fixation measured in S. velum gills is 9.1 μmol g−1 wet weight h−1 (Cavanaugh 1983). On average, gill weight is 38% of total weight (present study); therefore, this gill-specific carbon fixation rate amounts to 3.5 μmol g−1 wet weight h−1. Given that this is a net fixation rate measured on intact gills, the possibility exists that host cell respiration is masking a much more rapid gross carbon fixation rate by the symbionts. However, maximum carbon fixation rates for symbionts purified from S. velum gills are similar to this value from intact gills (K. Scott, unpublished data), and maximum gross carbon fixation rates measured on intact S. reidi clams (net carbon fixation rates plus host respiratory rates) are about 3 μmol g−1 wet weight(Anderson et al. 1987).

Fig. 6
figure 6

Solemya velum. Rates of carbon delivery to S. velum gills (per gram total clam wet weight) at a range of CO2 concentrations. These rates were predicted from Fick’s law and the power function G=69.8M0.85, where G is gill surface area (square centimeters), and M is total wet weight of the soft tissues of the clam (grams)

Furthermore, the estimated rates of supply are probable underestimates, as they do not include the influence of the high activities of carbonic anhydrase, from 270 to 400 Wilbur-Anderson units g−1 protein, present in S. velum gills (K. Scott, unpublished data). Net flux, equal to the rate of delivery minus the rate of loss, will of course be much lower, as it is highly unlikely that the symbionts are able to fix a substantial portion of the CO2 that diffuses to them from the mantle cavity. Instead, most of the intracellular CO2 will diffuse back out to the mantle cavity. Consequently, the isotope-enriching effect of Rubisco-catalyzed carbon fixation will be obliterated. Thus, δ13C values may be depleted in this and other bivalve–chemoautotroph symbioses due to the high gill surface areas necessary for symbiont housing and meeting a high demand for O2.