Biogeosciences Influence of CH 4 and H 2 S availability on symbiont distribution , carbon assimilation and transfer in the dual symbiotic vent mussel Bathymodiolus azoricus

High densities of mussels of the genus Bathymodiolus are present at hydrothermal vents of the Mid-Atlantic Ridge. It was previously proposed that the chemistry at vent sites would affect their sulphideand methane-oxidizing endosymbionts’ abundance. In this study, we confirmed the latter assumption using fluorescence in situ hybridization on Bathymodiolus azoricuspecimens maintained in a controlled laboratory environment at atmospheric pressure with one, both or none of the chemical substrates. A high level of symbiosis plasticity was observed, methane-oxidizers occupying between 4 and 39% of total bacterial area and both symbionts developing according to the presence or absence of their substrates. Using H CO−3 in the presence of sulphide, orCH4, we monitored carbon assimilation by the endosymbionts and its translocation to symbiont-free mussel tissues. Carbon was incorporated from methane and sulphide-oxidized inorganic carbon at rates 3 to 10 times slower in the host muscle tissue than in the symbiontcontaining gill tissue. Both symbionts thus contribute actively to B. azoricusnutrition and adapt to the availability of their substrates. Further experiments with varying substrate concentrations using the same set-up should provide useful tools to study and even model the effects of changes in hydrothermal fluids onB. azoricus’ chemosynthetic nutrition. Correspondence to: V. Riou (virgriou@vub.ac.be)


Introduction
Mytilids of the genus Bathymodiolus are among the dominant fauna inhabiting sulphide-hydrocarbon cold seeps and hydrothermal vents worldwide (Sibuet and Olu, 1998;Van Dover, 2000).Their nutrition seems to be mainly supported by sulphide-and/or methane-oxidizing bacterial (SOB, MOB) endosymbionts, located in specialized gill epithelial cells (Fisher, 1990).In fact, the coexistence of two distinct bacterial symbionts within a single cell of a multicellular eukaryote was demonstrated for the first time ever in the gills of Bathymodiolus spp.(Cavanaugh et al., 1987;1992;Fisher et al., 1993).B. puteoserpentis and B. azoricus, the two mussel species present at Mid-Atlantic Ridge (MAR) vent sites, display the same general characteristics, with two distinct morphotypes of Gram-negative endosymbionts in gill bacteriocytes, and the presence of enzymes specific for sulphide and methane oxidizing metabolisms (type I ribulose 1,5-bisphosphate carboxylase/oxygenase -RubisCO-, ATP sulfurylase, adenylyl sulfate reductase, and methanol dehydrogenase) (Robinson et al., 1998 andFiala-Médioni et al., 2002).Phylogenetic analyses show that the two Mytilid species share the same 16S rRNA phylotype of sulphide-oxidizing γ -proteobacteria, but harbour two distinct strains (Won et al., 2003).Duperron et al. (2006) found that the MOB 16S rRNA phylotype was also shared between the two mussel species, and that vent chemistry could affect the relative abundance of SOB and MOB.The volume occupied by each type of symbiont present within V. Riou et al.: Bathymodiolus azoricus' carbon fixation by chemosynthesis a bacteriocyte, quantified using 3D fluorescence in situ hybridization (FISH) technique, varies from vent site to vent site between B. azoricus specimens (Halary et al., 2008).However, although physiological activity of the symbionts was observed in gills of a hydrothermal vent Bathymodiolid from the Pacific (Fiala-Médioni et al., 1986;Nelson et al. 1995), in the Southern MAR B. puteoserpentis (Robinson et al., 1998) and in live specimens of the seep Bathymodiolid from the Gulf of Mexico (Childress et al., 1986;Fisher and Childress, 1992;Kochevar et al., 1992 andLee andChildress, 1995), it has to date not been demonstrated in live specimens of B. azoricus.
Stable isotope studies on Bathymodiolids have indicated the presence of two isotopically (and hence, nutritionally) distinct groups of mussels at the MAR (Trask and Van Dover, 1999;Fiala-Médioni et al., 2002 andColac ¸o et al., 2002a).A first group (bulk tissue δ 13 C ∼−30‰) would depend mostly on SOB, while a group with heavier isotopic values of −20‰ suggests a larger reliance on MOB.B. azoricus is a complex organism which can derive its carbon and nitrogen from two different symbionts, from particulate organic matter and possibly from dissolved organic matter.In such a case, the analysis of stable isotopes at natural abundance in bulk tissue, which provides information about the "average" diet over a certain period of time, is insufficient to accurately delineate the contributions from the different food sources.Incubations with 13 C-enriched substrates are commonly used to study the activity of pure bacterial cultures as well as the activity of different complex bacterial communities in natural and artificial ecosystems and soils.Autotrophic SOB were already described performing such tracer experiments with 13 C-labelled bicarbonate (Knief et al., 2003a), as were communities of physiologically active MOB using 13 CH 4 (Knief et al., 2003b).In the present study, we investigated symbiont populations and mussel nutrition in specimens of B. azoricus collected from the MAR and kept under various conditions in a controlled laboratory environment (LabHorta facility in the Azores, Colac ¸o and Santos, 2002b).Stable isotope tracer experiments were performed with 13 C-enriched bicarbonate or methane, followed by analysis of 13 C incorporation in mussel tissues.Densities of symbionts were estimated from FISH images using symbiont-specific probes.Results indicate that the symbionts develop in response to the presence or absence of their substrates, and that inorganic carbon and C 1 substrates are assimilated and further translocated into symbiont-free mussel tissues.

Study site and sampling
Acoustically retrievable cages were moored in August 2006 during the MOMARETO cruise (R/V Pourquoi Pas? Ifremer).They were positioned on diffuse venting areas at the Mid-Atlantic Ridge (MAR) hydrothermal vent field Menez Gwen (MG: 37 • 51 N, 32 • 31 W, 817 m) where mussel populations are very dense and cover virtually all available rock surfaces.Cages were filled with approximately 400 mussels using the Ifremer ROV VICTOR 6000.Our experiments were performed on adult mussels (53.6-74.1 mm shell length, mean 65.5±3.7 mm) from two cages recovered by the Portuguese vessel R/V Arquipélago in January and May 2007.After a 20 min surfacing time, the cage was lifted onboard and mussels were transferred to fresh cooled seawater for a 14 h transit to Faial Island (Azores).Some wild specimens were dissected on-board.At the Azorean LabHorta facility, mussels were scrubbed clean of visible adhering material, rinsed in chilled seawater and transferred to the hydrothermal vent laboratory set up.They were kept in maintenance conditions (see Sect. 2.2) until the start of tracer experiments (in Sect.2.3).Maintenance and tracer experiments in LabHorta were monitored daily for CH 4 and H 2 S concentrations, O 2 saturation, pH and temperature.
A first experiment was performed with the mussels collected in January and left to acclimate during 5 days (Table 1), with the aim to describe the effect of sulphide on the incorporation of bicarbonate.Experiments with May mussels were designed to observe (i) B. azoricus' response to prolonged maintenance in the laboratory (Table 2, upper part) and (ii) the evolution of symbiotic populations and carbon incorporation in mussel tissues in response to exposure to only methane, only sulphide or in the absence of stimulation (Table 2, bottom part).Specimens showing the highest 13 C incorporation after 4 or 5 days enrichment in the presence of sulphide or methane, respectively, were selected for FISH analyses in Sect.(2.6).
Dissected tissues were frozen and lyophilized, except for the gill parts analysed for FISH.Gill indexes (GI) were calculated from the specimens' gill tissue and rest of body dry weights (total soft tissue dry weight -gill dry weight) according to the following formula: GI = [gill tissue dry weight (g)/rest of body dry weight (g)] × 100 (1)

Maintenance in LabHorta
Mussels were placed in 40 L tanks containing 7.5-10 • C seawater (warmer water was found to inactivate methane oxidizers in B. childressi; Kochevar et al., 1992).Seawater was oxygenated using compressed air and replaced every second day (pH 7.5-9).Methane and sulphide were supplied to support the presence of endosymbionts.A sulphide solution (20 mmol L −1 Na 2 S in filtered seawater, pH adjusted with HCl 6 N to 8.6-9.2) was dispensed discontinuously for 15 min every hour, using a peristaltic pump injecting 2 mL min −1 .Resulting sulphide concentrations between 1 and 30 µmol L −1 were measured using the colorimetric diamine test.Such values are well within the range of the 0 to 62 µmol L −1 measured on mussel beds of Northern saturation (±SD) experienced by the sampled mussels over the duration of the experiment (days)."n" represents the number of specimens analysed for their gill index (GI), bulk tissue carbon isotopic signature (δ 13 C) and/or observed by fluorescence in situ hybridisation (FISH).MAR vent sites (Sarradin et al., 1998).Methane was bubbled continuously in seawater and dissolved methane concentration was monitored by sampling 60 mL seawater in a 300 mL bottle which was rolled for 20 min before measuring the headspace CH 4 concentration with a Gasurveyor 524.Since high methane concentrations above 300 µmol L −1 were found inhibitory for symbionts of B. childressi exhibiting high methane consumption rates (Kochevar et al., 1992), we tried to maintain CH 4 concentration below this level (8-92 µmol L −1 ).The effect of the imposed laboratory conditions on the mussels was assessed by statistical Mann-Whitney U test and Kruskal-Wallis non-parametric analyses.

Fatty acids extraction
Lyophilized gill tissues were ground to a fine powder using a mortar and pestle.All glassware was decontaminated for 4 h at 450 • C prior to use.Lipids were extracted using a modified version of the Bligh and Dyer method (White et al., 1979) using chloroform/methanol/phosphate buffer pH 7.4 in 1:1:0.9final volume proportions to form a biphasic system.Total lipids retrieved in the chloroform phase were applied onto activated silicic acid columns and partitioned into apolar, neutral and polar lipids by sequential elution with chloroform, acetone, and methanol, respectively.Lipid fractions were dried under nitrogen flux and immediately resuspended in 2 mL of a fresh solution of methanol/hydrochloric acid 37%/chloroform (10:1:1 volume) to form fatty acid methyl esters (FAMEs) by trans-methylation reaction (60 min at 90 • C).

Stable isotope analyses
FAME extracts were transferred to tin capsules for liquids (previously decontaminated with acetone) and chloroform was evaporated under mild conditions (37 • C, to avoid the loss of the most volatile FAMEs).Aliquots of lyophilized tissue powder were weighed into silver cups (previously decontaminated for 4 h at 450 • C) and acidified with a few drops of dilute HCl (5%) before analysis, to remove any possible trace of carbonates.Tissues were redried overnight at 60 • C.
δ 13 C analyses were performed on a Flash1112 elemental analyzer, coupled to a Delta V via a Conflo III interface (Thermo Finnigan).All carbon stable isotope ratios are expressed relative to the conventional reference (VPDB limestone) as δ values, defined as: Or as Atom percent values, defined as: (3) The reference material used as standard for carbon isotopic ratio measurement was sucrose (IAEA-CH-6).The standard followed the same analytical processes as the unknown samples.
Net carbon incorporation in the mussel tissues (C inc , in µmol C g tissue −1 ) was calculated using the following formula: Where A exp is the 13 C atom percent measured in the mussel after the tracer experiment, A control is the 13 C atom percent measured in the control mussel, C tissue is the carbon content of the tissue analyzed (µmol C g tissue −1 ) and A substrate the 13 C atom percent of the substrate used in the experiment (i.e., 25% for CH 4 , as given by the supplier, or 16.3% for NaHCO 3 and 22.2% for CH 3 OH, as calculated from the preparation).

Fluorescence In Situ Hybridization (FISH)
FISH was performed according to the protocol developed by Halary et al. (2008) with a few modifications.Upon recovery of selected mussel specimens, the most anterior part of their left gill was fixed in 2% formaldehyde (2-4 h, 4 • C), rinsed twice with filtered seawater, and stored at −20 • C in 1:1 ethanol/filtered seawater.In the lab, gill tissue was embedded in polyester wax and 10 µm-thick sections were cut using a microtome (JUNG, Heidelberg, Germany), collected on Superfrost ® Plus slides (Roth, Germany), and hybridized as previously described (Halary et al., 2008) using the ATTO488labelled ImedM probe (5'-ACCAGGTTGTCCCCCACTAA-3', Duperron et al., 2008) specific for the methane-oxidizing symbiont and the Cy5-labelled BangT-642 probe (5'-CCTATACTCTAGCTTGCCAG-3', Duperron et al., 2005) specific for the sulphide-oxidizing symbiont.Images were acquired using a BX61 microscope (Olympus Optical Co., Tokyo, Japan) under a 10× objective (NA 1.30).The green channel corresponding to the ImedM probe (methaneoxidizers) and the blue channel corresponding to the BangT probe (sulphide-oxidizers) were binarized by applying a luminosity threshold to distinguish between background fluorescence and probe signal, and numbers of pixels belonging to each type of symbiont were computed using the ImageJ software (Abramoff, 2001).Total area occupied by bacteria and respective proportions of each phylotype were computed for the whole image (one image per specimen, around 20 gill filaments per image).For an inter-comparison between images, bacterial areas were normalized by dividing areas by the total length of filaments visible and measured on each picture.

Effect of the maintenance in aquarium on mussel condition
Mussels collected in January were allowed to recover from the decompression stress during 5 days at atmospheric pressure (transport followed by LabHorta) with CH 4 and H 2 S to ensure the survival of endosymbionts ("Acclimated", Table 1).A quick drop of their gill index (GI) could be noticed over this short period (although not statistically significant, Mann-Whitney U test p=0.237,Fig. 1).This drop was also observed for May mussels kept in the presence of H 2 S and CH 4 (see concentrations in Table 2) up to 38 days before starting the tracer experiments (aquarium mean GI=33.7,n=7; wild animals mean GI=47.3,n=10, Kruskal-Wallis p=0.006,Fig. 1).Subsequent to this rapid loss in gill weight, no further significant changes in GI were observed between individuals collected at different moments during the maintenance period (from day 23 to 52, Kruskal-Wallis p=0.115).When compared to mussels kept in maintenance conditions, the GIs of specimens from the tracer experiments did not decrease significantly and rather increased slightly in the 13 CH 4 experiment (Fig. 1, Kruskal-Wallis, n=22: CH 4 p=0.041,H 2 S p=0.598, control p=0.262).

Endosymbiont populations' response to experimental treatments
Gill filaments from May mussels were observed using FISH to assess the effects of the maintenance period in Lab-Horta and of the tracer experiments on symbiont populations (Fig. 2).Estimates of symbiont areas and proportions for a wild specimen, for a specimen collected at the start of the experiments (after the maintenance period) and others after 4 to 5 days in the absence or presence of either CH 4 or H 2 S are displayed in Table 3. Gill filaments of aquarium specimens appeared narrower than the ones from a wild mussel dissected immediately after the cage recovery and carried much less bacteria (Fig. 2A-B).The sections observed showed a decrease in area covered by bacteria by a factor of 4 (from 16.7 to 3.7 µm 2 /µm filament), and indicated a shift in symbiont relative areas, with methane-oxidizing bacteria (MOB) representing 14% of the total area occupied by bacteria in acclimated specimens, versus 24% in fresh specimens (Table 3).At the beginning of the tracer experiments, bacterial densities appear low.A mussel kept four days in filtered seawater devoid of CH 4 and H 2 S also displayed a lower total area occupied by bacteria of 2.1 µm 2 /µm filaments.The exclusive presence of 86.9±98.0µmol L −1 CH 4 seemed to favour MOB which represented 39% of total bacterial area (Fig. 2C).Four days enrichment with exclusively 7.7±1.6µmol L −1 H 2 S resulted in higher overall bacterial areas with a high proportion of SOB representing 96% of total bacterial area (Fig. 2D).

Sulphide induced CO 2 incorporation
A low incorporation of 13 C from HCO − 3 was observed in the absence of H 2 S, which increased considerably when B. azoricus was supplied with H 2 S (Fig. 3).Highest incorporation was found in the gill tissue, followed by other tissues such as the digestive tract and vital organs which displayed www.biogeosciences.net/5/1681/2008/Biogeosciences, 5, 1681-1691, 2008  much lower incorporations.The analyses of gill tissue phospholipid fatty acid methyl esters (PLFA) also showed a coupling between enriched inorganic carbon incorporation and the presence of H 2 S (Fig. 4).δ 13 C analysis of fatty acid methyl esters (FAMEs) from gill lipid fractions of mussel A (the mussel specimen displaying the highest δ 13 C values after January experiments) showed the highest enrichment in the polar PLFA, followed by the FAMEs from the neutral fraction.A significantly lower δ 13 C was found in fatty acids associated with the apolar lipid fraction among which can be found di-and tri-acylglycerols and cholesterol derivatives.
3.4 CO 2 and CH 4 incorporation rates After 4 and 5 days of exposure to 13 C-labelled HCO − 3 (in the presence of H 2 S) or CH 4 , B. azoricus gill tissue already showed 13 C abundances significantly higher than background levels (Kruskal-Wallis p=0.049 for both treatments).At each sampling time, δ 13 C values tended to be higher in gill than in muscle tissue.Labelling of the gill tissue appeared uniform between the different mussel specimens after 4 days in H 2 S.However, strong differences in δ 13 C between individuals were observed after 15 days with enriched CH 4 and 20 days with H 2 S and H 13 CO − 3 (Fig. 5, left panel).In the gill tissue, carbon incorporation rates varied from 151 to 323 nmol C g dry tissue −1 h −1 after 20 days in the H 2 S+H 13 CO − 3 experiment (sulphide concentration: 0-32 µmol L −1 ), and from 56 to 228 nmol C g dry tissue −1 h −1 with 13 CH 4 (14-200 µmol L −1 ).Muscle tissues net incorporation rates ranged from 52 to 104 nmol C g dry tissue −1 h −1 in the H 2 S+H 13 CO − 3 experiment, and from 11 to 21 nmol C g dry tissue −1 h −1 in the 13 CH 4 experiment (Fig. 5, right panel).

Discussion
Because of its capacity to recover and survive in aquaria for several months after the post-retrieval decompression stress, increasing data is being obtained from live Bathymodiolus azoricus maintained in aquaria at atmospheric pressure in the presence or absence of CH 4 and H 2 S for the endosymbionts.Immune responses (Bettencourt et al., 2007), biochemical responses to recovery stress (Dixon et al., 2004) and to metal exposure (Company et al., 2008) have been investigated using such settings.Dixon et al. (2004) saw higher levels of DNA damage in mussels kept in the presence of methane and sulphide than in some kept without any "food supplement".In a recent study, Bettencourt et al. (2008) investigated the physiological state of mussels during acclimatization to atmospheric pressure in the absence of dissolved gases over the course of several months post-capture and concluded that our laboratory set up was a suitable system to study physiological reactions.Response of symbionts to the experimental conditions was barely examined.A preliminary study on sulphide-oxidizing bacterial symbionts (SOB) showed, on transmission electron micrographs, the loss and potential reacquisition of SOB (Kadar et al., 2005).
In the present study we tried to gain insight on the symbiont distributions, physiology and impact on the nutrition of B. azoricus.Gill indexes and fluorescence in situ hybridisation observations of gill filaments indicate that despite a strong post-collection initial stress, with a marked decrease in total symbiont abundances and gill dry weight, both bacterial populations can be maintained in the gill tissue of mussels kept in aquaria.Furthermore, incubation experiments suggest that the presence of either sulphide or methane  alone favours the increase in absolute and relative abundance of SOB and methane-oxidizing bacterial symbionts (MOB), respectively (Table 3).This supports the hypothesis that symbiont populations react to changes in environmental parameters, in particular to the balance between sulphide and methane, and thus optimize the use of available compounds as suggested by previous authors (Trask and Van Dover, 1999;Colac ¸o et al., 2002a;Salerno et al., 2005;Duperron et al., 2007).The range of variation observed in symbiont relative abundances, with MOB occupying between 4 and 39% of total bacterial area in specimens of this study (Table 3), confirms the high level of plasticity of the dual symbiosis.Although our data must be interpreted with caution as only one individual per experimental condition was analyzed, the fact that observations qualitatively agree with expected results is a strong indication that our set-up is appropriate for the enrichment of each endosymbiont.Recently, a 3D-FISH technique was employed to quantify the effect of a sulphide pulse in a similar experimental set-up, on the volume and relative abundance of symbionts in mussel gills (Halary et al., 2008).We herein provide a first indication that MOB also respond to a methane pulse.
Stable isotope tracer experiments demonstrate that H 2 S stimulates the assimilation of inorganic carbon, most probably by SOB endosymbionts.Indeed, SOB carry out chemolithoautotrophic organic production via the Calvin-Benson cycle, whereby energy for CO 2 fixation by the enzyme RuBisCO derives from sulphide oxidation (Fisher, 1990).Our experiments show that most of the incorporated tracer is found in the gill tissue that hosts high densities of bacteria, as opposed with mantle, muscle or other tissues (Fig. 3).The analysis of 13 C enrichment in fatty acids from the different fractions of gill extracted lipids (Fig. 4) reveals that for mussels exposed to H 2 S, the polar fraction (consisting in phosphoglycerolipids) is the most 13 C-enriched fraction in terms of fatty acids, while the apolar fraction shows almost no enrichment.Apolar lipids consist mainly in acylglycerols, free fatty acids, hydrocarbons, pigments, sterols and waxes.Bacteria do not generally synthesise sterols de novo (except for some methylotrophs, Pearson et al., 2003) and do not store acylglycerols as in plant or animal cells.The fact that inorganic carbon is first incorporated in the main components of Gram-negative bacterial membranes (phospholipids, Kates, 1964) and not in apolar lipids originating from the host tissues is an additional indication that CO 2 was fixed by SOB.However, we also see a low level of 13 C assimilation in mussel tissues in the absence of H 2 S (Fig. 3), which was also observed in an earlier study using radio-active bicarbonate with B. thermophilus gill homogenates (Belkin et al., 1986).Various mechanisms can be invoked to explain this inorganic carbon fixation in the absence of H 2 S: i) Mussels encountered severe hypoxia (more than 6 h), during prolonged closing of their valves probably in reaction to physical stress (light or vibrations in the LabHorta container, for example).In such conditions, stored glycogen (which can reach 10% of the dry mass in B. thermophilus, Smith, 1985) becomes the major energy source for the mussel, and the Embden-Merehof-Parnas glycogenolysis pathway deviates from its normal aerobic route, to carboxylate phosphoenolpyruvate (PEP), producing oxaloacetate (Grieshaber et al., 1994).PEP carboxykinase could thus incorporate 13 C-enriched inorganic carbon entrapped in the valve cavity into oxaloacteate and subsequent end-products such as succinate (as seen in Mytilus edulis, de Zwaan et al., 1983) or other organic acids.However, the effect of any enhanced glycogenolysis would probably result in all tissues displaying the same level of label incorporation, whereas we observe 13 C incorporation mainly in gill tissue, and only to a lesser extent in other tissues.Other mechanisms must therefore be involved; ii) CO 2 is fixed by one or both symbiont types via anaplerotic reactions using PEP Carboxylase identified in numerous chemolithoautotrophic and even in methylotrophic bacteria (Glover, 1983;Schobert and Bowien, 1984;Müller-Kraft et al., 1991); iii) Sulphur or thiosulfate reserves are present in the gill tissue and anaerobic sulphide production occurs after prolonged valve closing (see Arndt et al., 2001 andPruski et al., 2002, who reports that a large fraction of the sulphur in gills of B. azoricus specimens is present in the zero oxidation state); iv) MOB harbour a RubisCO activity as observed in type X-methane oxidizers (Baxter et al., 2002) and in Bathymodiolus sp.(Elsaied et al., 2006) and would also incorporate inorganic carbon by chemoorganoautotrophy (using formate as a growth substrate, Baxter et al., 2002).More experiments and biochemical analyses would be needed to understand the processes involved in inorganic carbon fixation in the absence of stimulation with sulphide, and at this stage we cannot eliminate any of the abovementioned hypotheses.
Distribution of labelled carbon in the different mussel tissues suggests that in the case of carbon uptake by SOB, inorganic carbon is first incorporated by bacteria, then transferred to the rest of the tissues including the digestive system and vital organs, muscle and mantle tissues being the last ones to get this carbon.Carbon transfer from the gills to symbiontfree tissue was observed earlier in the MOB-housing seep Bathymodiolus sp. by Fisher and Childress (1992) who detected an increase of radiolabel coming from 14 CH 4 in symbiont-free tissues during a chase period.Evidence for intracellular digestion of symbionts in the same mussel species was provided later by Streams et al. (1997).Partially degraded bacteria in bacteriocyte bodies with lysosomal activity were also observed in B. azoricus (Fiala-Médioni et al., 2002), indicating that this could be the mode of carbon transfer from the symbiont to the host.The slow transfer of 13 C from either CH 4 or HCO − 3 (in the presence of H 2 S) to the muscle tissue relative to the gill (Fig. 5) may reflect differences in tissue turnover: gill tissue hosts the bacteria that continuously incorporate the tracer, and hence show a 13 C abundance more like that of the symbionts; muscle tissue, in contrast, contains exclusively C accumulated through the entire life of the mussel, and its 13 C abundance, therefore, is less responsive to recently ingested food.Our calculations based on the results of May tracer experiments thus indicate 3 and 5-10 times slower carbon incorporation rates in muscle tissue than in symbiont-containing gill tissue for SOB and MOB, respectively (Fig. 5).
We finally measured a median incorporation rate of 0.28±0.06µmol C g −1 dry gill h −1 in adult B. azoricus (n=13, mussels collected after 4 and 20 days, median shell length=65.8±4.7 mm) with enriched bicarbonate in the presence of sulphide at concentrations close to the ones measured over mussel beds at Menez Gwen and Lucky Strike vents (Table 2).In contrast, mussels kept for up to 15 days with 28 µmol L −1 enriched CH 4 (median concentration) displayed slow carbon incorporation rates (0.07±0.07 µmol C g −1 dry gill h −1 , n=10, shell length=62.2±1.8 mm, medians±SD).The high standard deviations in carbon incorporation rates measured for both ex-periments could be the result of differential valve opening behaviour or of differences in initial symbiont abundances between specimens following the maintenance period.Mussels maintained in LabHorta for 38 days before labelling showed a quick drop in symbiont abundance compared to wild mussels, which could explain part of the very low carbon incorporation from enriched methane.Kochevar et al. (1992) described in more details the characteristics inherent to MOB from a seep Bathymodiolid in a study using 14 CH 4 .They showed that about 70% of the consumed methane was incorporated as organic compounds.The highest net carbon incorporation rate (5 µmoles C g −1 wet mussel h −1 ) was obtained at 250 µmol L −1 dissolved methane.Our rate estimates cannot really be compared to the rates obtained by Kochevar et al. (1992), as they are reported per dry gill weight as opposed to wet whole mussel soft tissue weight, plus our experiments were performed with a 9 times lower median methane concentration with mussels maintained in aquarium.However, we provide the first experimental data on carbon fixation in "acclimated" Bathymodiolus azoricus.
Although working at atmospheric pressure, the LabHorta aquarium setting can thus be employed to study the dynamics of symbiont populations.In the future, pulses of varying intensities and durations or parallel measurements of rates of symbiont digestion by the host should be foreseen.Together with the identification of biomarkers for the endosymbionts following tracer experiments, such experiments should provide useful tools to study the effects of changes in hydrothermal fluids on the chemosynthetic nutrition of the Mid-Atlantic Ridge mussel and gain insight in the extent to which the mussel beds interact with and impact their surroundings.

Fig. 2 .
Fig. 2. FISH observations of transverse sections of gill filaments, displaying methane-(green) and sulphide-oxidisers (red) in May mussels.A-B: gills from a fresh specimen (A) and a specimen maintained 38 days (B) in LabHorta with H 2 S+CH 4 .C-D: specimens observed after 5 days with 13 CH 4 (C) or 4 days with H 2 S+NaH 13 CO 3 (D).

Fig. 4 .
Fig. 4. δ 13 C values of FAMEs (uncorrected for the additional carbon from the methylation) from polar lipids extracted from January tracer mussels' gills (white bars, n=3, mean ±SD).For mussel A from the 13 C+H 2 S incubation, polar, neutral and apolar fatty acids were measured (bars from left to right, respectively).

Table 1 .
January experimental design, with median CH 4 and H 2 S concentrations (µM) in the presence or absence of 10% H 13 CO − 3 , and O 2 saturation (±SD) experienced by the sampled mussels over the duration of the acclimation and of the labelling experiment (days)."n" represents the number of specimens analysed for their gill index (GI), bulk tissue and/or phospholipid fatty acids (PLFA) carbon isotopic signature (δ 13 C).

Table 2 .
May experimental design, with median (13) CH 4 and H 2 S concentrations (µM) in the presence or absence of 16.3% H 13 CO −

Table 3 .
Desbruyères et al., 2000)rocessing.The CH 4 /H 2 S ratio of 1.2 reported for fresh mussels, is a mean from end-member fluids at Menez Gwen (0.67-1.75,Desbruyères et al., 2000), which might not be representative of the situation in their immediate environment.