Abstract

Two newly discovered hydrothermal vent fields of the Wallis and Futuna region, Kulo Lasi and Fatu Kapa, were sampled for fluid geochemistry. A great geochemical diversity was observed and assigned to the diversity of lithologies as well as the occurrence of various processes. Kulo Lasi fluids likely formed by interaction with fresh volcanic rocks, phase separation, and mixing with magmatic fluid. Conversely, the geochemistry of the Fatu Kapa fluids would be mostly due to water/felsic lavas reactions. In terms of organic geochemistry, fluids from both fields were found to be enriched in formate, acetate, and semivolatile organic compounds (SVOCs): n-alkanes, n-fatty acids, and polyaromatic hydrocarbons (PAHs). Concentrations of SVOCs reached a few ppb at most. The distribution patterns of SVOCs indicated that several processes and sources, at once of biogenic, thermogenic, and abiogenic types, likely controlled organic geochemistry. Although the contribution of each process remains unknown, the mere presence of organics at the μM level has strong implications for metal dispersion (cycles), deposition (ore-forming), and bioavailability (ecosystems), especially as our fluxes estimations suggest that back-arc hosted vent fields could contribute as much as MOR to the global ocean heat and mass budget.

1. Introduction

Although back-arc settings are favourable environments for the formation of hydrothermal convection cells, hydrothermal exploration has long been conducted to a greater extent on Mid-Oceanic Ridges (MOR). Today, more than 600 active hydrothermal vent fields have been discovered and about half of them are located at MOR against a fifth in back-arc basins (BAB) [1]. Yet back-arc environments are likely to generate more diversity than their MOR homologs in terms of fluid chemistry because of the variety of lithologies the fluids can react with (e.g., basaltic to rhyolitic volcanic rocks with or without arc-like geochemical signature, various alteration mineralogical assemblages) as well as the possible contribution of magmatic-related aqueous fluids [25]. The Wallis and Futuna area was surveyed for hydrothermal activity because of its very peculiar geological settings within a back-arc system and its potential relevance for mineral resources [6, 7]. It is located about 200 km west of the northern tip of the Tonga-Kermadec trench where the fastest subduction rates have been recorded (18 to 24 cm per year) and occur at the junction of 2 BAB: the Lau and the North-Fiji BAB [8]. Here we report on the geochemistry of the fluids of the very first two vent fields discovered in the area and in this type of environment. We chose to bring a special focus on organic geochemistry because it has been hardly studied in modern hydrothermal systems despite the recent growing interest for organic matter (OM) in the ocean. The discussion focuses on processes controlling the geochemistry as well as implications of the presence of organic molecules at the local and regional scales.

Organic geochemistry of hydrothermal fluids has generally been far less studied than the mineral and gas geochemistry. In most cases works focused on small molecules (hydrocarbon gases, volatile fatty acids, and amino acids) and very few data are available on semivolatile organic compounds (SVOCs). Despite the growing interest for OM in the ocean and hydrothermal systems there is still a major lack in identification and quantification of organic compounds [1015]. Notably numbers of studies agree on the major ligand role of organics in metal stabilisation, transportation, bioavailability, and ore-forming but there are hardly any clues on the nature of these ligands in hydrothermal environments [1625]. Organic compounds in hydrothermal fluids may come from marine dissolved organic matter (DOM) recycling [12, 13], subsurface biomass degradation [26], entrainment of organic detritus from local recharge zones, and subsequent degradation, or abiotic formation in the deep subsurface [2730]. The latter is supported by many theoretical [3133] and experimental work summarised in two reviews [34, 35]. Conversely, some other studies reported the absence of organic compounds in hydrothermal fluids except at the Lost City alkaline vent field which is theoretically more favourable for abiotic synthesis [36]. Nevertheless, we report here the presence of semivolatile organic compounds in hydrothermal fluids from the Wallis and Futuna area and provide concentrations of a selection of extractable compounds that have been identified elsewhere as hydrothermally derived [27, 37]: n-alkanes, n-fatty acids (n-FAs), mono-, and polyaromatic hydrocarbons (BTEXs and PAHs). These very first quantitative field data might feed thermodynamic models of abiotic synthesis, guide the design of experiments to better understand hydrothermal organic geochemistry, and help assessing the importance of hydrothermally derived organic compounds in metal complexation and, as a nutrient for microorganisms, complete fluxes calculation and enter in the carbon cycle budget calculations.

2. Geological Settings

Wallis and Futuna Islands are located at the transition between the North Fiji and the Lau back-arc basins. This geodynamical setting accounts for complex volcanic and tectonic activity in the area. Pelletier et al. [38] and Fouquet et al. [6] observed multiple active extensional zones including widespread areas composed of numerous individual volcanoes (e.g., Southeast Futuna volcanic zone (SEVZ)) and well organised spreading centers such as the Futuna and Alofi oceanic ridge. West of Futuna Island, the 20–30° trending Futuna spreading center (FSC), is composed of a series of en echelon spreading segments. The opening rate of this oceanic ridge has been estimated at 4 cm/yr from the interpretation of magnetic anomalies [38]. East and southeast of Futuna Island, bathymetric maps, and reflectivity data clearly reveal that active extension and recent volcanism occur in the SEVZ as well as along the Alofi spreading center [6]. The SEVZ is a broad zone of diffuse volcanism bordered by the ENE-WSW trending volcanic graben (named Tasi Tulo graben) to the north and the NNE-SSW trending Alofi spreading center to the south. The SEVZ includes Kulo Lasi active volcano, the Fatu Kapa, and Tasi Tulo volcanic zones ([7], Figure 1).

Fluids were sampled at the Kulo Lasi and Fatu Kapa sites. Kulo Lasi has been described in detail by Fouquet and collaborators [39]. In summary, it is a shield volcano located about 100 km southeast of Futuna Island (Figure 1). It represents the most recent volcano in the SEVZ and is composed of basaltic to trachy-andesitic lava with no direct geochemical affinity with subduction [39]. The volcanic edifice is ca. 20 km in diameter and appears relatively flat with the top located at a depth of 1200 m and the base only 300 m deeper (ca. 1500 m below sea level). It exhibits a central caldera (5 km in diameter and 200–300 m deep) with a flat bottom covered by recent lavas and a central mound composed of older and tectonised lava flow. By contrast, the Fatu Kapa volcanic area is in a 20 km wide transition zone between the Tasi Tulo graben and the Kulo Lasi volcano. Here only small (<1 km) volcanic edifices are seen to be consisting of young mafic to felsic lavas (Figure 1).

3. Sampling and Analytical Procedures

Sampling was achieved at Kulo Lasi and Fatu Kapa by the HOV Nautile during the FUTUNA 1 and FUTUNA 3 cruises conducted by Ifremer in 2010 and 2012. Fluid samples were taken at the nose of smokers to minimise seawater contamination. Samples of volumes up to 750 mL of hydrothermal fluids were collected in titanium syringes that were modified after the model described in Von Damm et al. [40]. The gas-tightness was greatly improved and ensured the majority of the gas to be recovered. Those same syringes have been used in several studies by Charlou et al. and have shown good results notably for gas-Mg correlations (e.g., Charlou et al., 2002). Autonomous temperature sensors (S2T 6000-DH, NKE Instrumentation) were mounted on the sampler nozzle. As soon as the fluids were recovered, pH, H2S, and Cl- concentrations were measured to evaluate the quality of the sample. Total gases were immediately extracted and analysed; then aliquots of gas were conditioned for further stable isotopes measurements. Finally, the gas-free fluid was conditioned for major and minor elements analyses, on the one hand, and for organic compounds analyses, on the other hand.

3.1. Gas Extraction and Analyses

Total gas was extracted as described in Charlou and Donval [41]. Preliminary major gases (CO2, H2, CH4, and N2) concentrations were obtained on board by using a portable chromatograph (Microsensor Technology Instruments Inc.) that was mounted on line with the gas extractor. Extracted gases were conditioned on board in stainless steel pressure-tight flasks and stored until analyses. Gases were separated by Gas-Chromatography (Agilent GC 7890A, Agilent Technologies) and quantitatively analysed by triple detection using mass (MS 5975C, Agilent technologies), flame ionisation, and thermal conductivity detectors. Aliquots of gas were stored both in vacuum tight tubes (Labco, Ltd.) and in copper tubes to be sent for further carbon isotope analyses (Isolab b.v., Netherlands) and He isotopes analyses (CEA, Saclay, France), respectively.

3.2. Inorganic Geochemistry: Sample Preparation and Analyses

pH was measured using a combined glass electrode (Ecotrode Plus, Metrohm). Cl and H2S were measured by potentiometry using AgNO3 (0.05 M) and HgCl2 (0.01 M) as titrating solutions, respectively. NaOH (2 M) was added to the aliquot before H2S measurement. SO4, Br, Na, K, Mg, Ca, Li, and Cl were measured by ionic chromatography (Dionex Ion Chromatograph System 2000) after appropriate dilutions. Fe, Mn, Cu, Zn, Sr, Li, and Rb were measured by flame atomic absorption spectrometry using standard additions (AAnalyst, Perkin Elmer Inc.). Aliquots for silica determination were immediately diluted 100- to 200-fold and analysed by the silicomolybdate automatic colorimetric method [42, 43].

3.3. Organic Geochemistry

Total Organic Carbon (TOC) was measured using a multi N/C 3100 (Analytik Jena AG, Germany). Samples were acidified online with HCl and then purged with to remove inorganic carbon (IC). A TIC control analysis was performed and followed by three TOC measurements on each sample.

Acetate and formate concentrations were determined using a Dionex ICS-2000 Reagent-Free Ion Chromatography System equipped with an AS50 autosampler (Dionex Camberley, UK). Chromatographic separation was conducted using two Ionpac AS15 columns in series at 30°C and the determination of species was carried out using an Anion Self-Regenerating Suppressor (ASRS 300 4 mm) unit in combination with a DS6 heated conductivity cell (35°C). The gradient program was as follows: 6 mmol L−1 KOH (43 min); increase from 27 mmol L−1 KOH min−1 to 60 mmol L−1 (39 min); decrease from 54 mmol L−1 KOH min−1 to 6 mmol L−1 (5 min).

SVOCs were extracted using Stir Bar Sorptive Extraction (SBSE). Basically any compound with a log Ko/w > 2.5 is recovered with a rate > 50% [44]. The method was improved after Konn et al. [37]. The entire content of the titanium syringe was transferred into a precombusted glass bottle and six 90 mL aliquots of the sample were poured into 100 mL precombusted glass vials. 10 mL of MeOH was added to avoid adsorption of the compounds onto the wall of the vials. Internal standards were added to the solutions in 2012 so that quantification could only be achieved in Fatu Kapa fluids. Extraction was performed in sealed vials with ultrainert septum crimps, at 300 rpm and using 48 μL PDMS Twisters® (Gerstel GmbH). We focused on a selection of chemical groups that had previously been described as hydrothermally derived [27]. To that respect, pairs of aliquots were dedicated to analysis of n-alkanes, n-FAs, and both BTEXs and PAHs, respectively. Extraction kinetics experiments showed that chemical equilibrium was reached after 5 h of extraction for n-alkanes, 4 h for PAHs and 14 h for n-FAs (Konn, unpublished results). Twisters were then removed, rinsed with MQ water, dried, and stored at +4°C until analyses by Thermal Desorption-Gas Chromatography-Mass Spectrometry (TD-GC-MS) [37]. Analytical parameters were adjusted for each group of compounds (Table 1).

For each batch of conditioned Twisters, one was spared, stored at +4°C, and analysed in the same run as the other Twisters. This dry blank aimed at showing any contamination that could have occurred during conditioning, storage, and transport. MQ water samples were prepared and extracted on board as regular hydrothermal samples to check if any contaminations could have occurred during the sample preparation step. Deep-sea water was also collected, processed, and analysed, using the same titanium syringes and according to the same protocols as for hydrothermal fluid samples, and thus constitute the reference blank experiment.

Calibration was achieved using a commercial standard solution of BTEXs and 3 custom standard solutions of C9–C20  n-alkanes, C6–C18  n-FAs, and PAHs containing naphthalene (N), Acenaphtene (A), Fluorene (F), Phenanthrene (Ph), Anthracene (An), Fluoranthene (Fl), and pyrene (Py) (LGC Standards, LGC Ltd.). Deuterated n-alkanes (C10D22 and C14D34), methyl esters (C9H18O2 and C15H30O2), and deuterated PAHs (naphthalene-D8, Biphenyl-D8, and Phenanthrene-D10) were used as internal standards (IS). Calibration curves (Concentration (analyte)/Concentration (IS) versus Area (analyte)/Area (IS)) were obtained using at least five concentration levels that were replicated 3 times (Table 2). Although the correlation coefficient of the linear regressions was satisfactory for all compounds, the significance and lack of fit of the model were checked by statistical tests before validation. A series of Student, Barlett, Chi-square, and Fisher tests was run for each individual compound using the Lumière software. The best fitting model was then chosen for each case and confidence intervals were calculated.

4. Results

Altogether 35 hot fluid samples were collected in the study area from 8 different sites: Kulo Lasi caldera , on the one hand, and Stéphanie , Carla ,   ,   ,   , Fati Ufu , and Tutafi , on the other hand, all located in the Fatu Kapa area (Figure 1). The Kulo Lasi smokers occurred at ~1500 m depth on recent lava flows and consisted in a multitude of short (~25 cm) and narrow (~3–5 cm) diameter anhydrite chimneys containing a small percentage of sphalerite (ZnS), chalcopyrite (CuFeS2), isocubanite (CuFe2S3), pyrrhotite (), and pyrite (FeS2) (Figure 2). The temperature was consistently about 343°C and the pH approached 2.2–2.3 (Table 3). In the Fatu Kapa area we could distinguish two types of hydrothermal environments at 1550–1650 m depth. Translucent 270–290°C fluids associated with anhydrite chimneys (up to 25 m tall and 2.5 m in diameter) characterised Stéphanie, Carla, , , and sites, while >300°C milky to grey fluids associated with sulphide chimneys were characteristic of the southwest region including Fati Ufu and Tutafi sites (Figure 3, Table 3).

4.1. Gas

Concentrations of gases in all fluids as well as stable isotopes data are compiled in Table 4. Samples recovered from Kulo Lasi were extremely poor in CH4 (<0.01 mM) but contained the series of C2–C5 hydrocarbons. Samples from Fatu Kapa had higher concentration of CH4 (0.05–0.235 mM) but only n-pentane (0.5–3.2 μM) could be detected and quantified in terms of longer hydrocarbons. One sample from Kulo Lasi was found to be extremely rich in H2 with nearly 20 mM while the others ranged from 1 to 6 mM and were below 0.05 mM at Fatu Kapa. H2S was highly variable between the 3 sampled chimneys at Kulo Lasi (0.39, 1.66, and 5.05 mM) while it was found rather homogeneous at Fatu Kapa with values around 1 mM. CO2 concentrations were more elevated at Fatu Kapa (4.5–29 mM) compared to Kulo Lasi (1–5 mM).

Helium isotope ratios were in the range 7.0–9.9 Ra over the Fatu Kapa area, in agreement with plume data [7]. They could not be measured at Kulo Lasi unfortunately. Carbon isotopes ratios were around −5 for CO2 at Fatu Kapa whereas at Kulo Lasi the ratio showed very different results ranging from −0.2 to −4.1. As for methane, δ13C were slightly lower at Kulo Lasi (~−28) versus Fatu Kapa (~−23) and δD was about −110 in all samples from Fatu Kapa. δD (CH4) could not be measured in the Kulo Lasi fluids because of the too low concentrations of CH4. Carbon isotope ratios of longer hydrocarbons were in the −27 to −22 at both vent fields. To be noted one sample from Fati Ufu in the Fatu Kapa area showed remarkably lower isotopic ratios with δ13C (CO2) = −2.3, δ13C (CH4) = −6.1 and δD (CH4) = −93. We do not have any explanation for this but do not have any reasons either to consider it as an outlier.

4.2. Major and Minor Elements

Major and minor elements measurements data are compiled in Table 3. Fluids from Fatu Kapa all exhibited a higher salinity than seawater up to 4.6 wt% NaCl whereas at Kulo Lasi fluids with both lower (2.8 wt% NaCl) and higher (4.3 wt% NaCl) salinity were sampled. Mg and SO4 concentrations tend to be zero in the purest samples at Fatu Kapa. But, the purest fluids from Kulo Lasi showed significant levels of Mg and SO4, associated with an extremely acidic pH (<2.5) and a high (343°C). Although we cannot totally discard that some mixing with seawater occurred, endmember concentrations of the Kulo Lasi fluids were then estimated to be close to the purest fluids sampled whereas they were obtained from mixing lines at Fatu Kapa assuming Mg zero (Table 5).

Fluids from Fatu Kapa were enriched compared to seawater in alkali, alkaline Earth, and transition metals as well as in strontium, bromide, and silica. Conversely, the fluids from Kulo Lasi exhibited a much more complex pattern. They were all highly enriched in transition metals and silica compared to seawater and fluids from Fatu Kapa (e.g., Fe up to ~10 mM). The enrichment versus seawater in alkali metals was not as striking as for Fatu Kapa fluids. As for the alkaline Earth metals, the amount of Ca was identical to seawater and fluids were depleted in Sr compared to seawater. Finally, both depletion and enrichment in Br were observed in the fluids from Kulo Lasi.

4.3. Organic Geochemistry

First of all, we would like to mention that because solubility of organic compounds decreases with and because samples were processed at room temperature, the measured concentrations are probably lower than in situ concentrations. Moreover, it is very likely that a portion of the OM was adsorbed on small particles in the fluids which are not taken into account using our extraction and analytical techniques. As a result, the concentrations we report here probably represent lower estimates of in situ concentrations. However, since in situ measurement techniques are not available yet, these values are the best estimates we can obtain. Note that they also are the first to be published for SVOCs.

Formate and acetate reached 16.3 and 15.5 μM, respectively, and covaried with Mg in the Kulo Lasi fluids (Figure 4). Concentrations of formate and acetate were significantly higher in the Fatu Kapa area but no correlation with Mg could be observed. Nevertheless, the purest fluids usually showed the highest concentrations. Formate reached 68 ppb at Stéphanie and 722 ppb at Fati Ufu whereas it could not be detected at and Tutafi and was not measured at Carla. Acetate was detected in all analysed samples and concentrations were an order of magnitude higher than the ones of formate (543–2309 ppb) (Table 6).

Heavier extractable organic compounds were not detected in the dry control experiment and only a few were detectable but below limit of quantification (LOQ) in the MQ water blank experiment (Table 6). This showed that sample preparation and storage could be considered as contamination-free steps. The levels of heavier extractable organic compounds appeared rather high in the reference water at Fatu Kapa certainly because of the overall spread hydrothermal discharges and diffuse venting in the region [7] (Table 6, Figure 5). This sample was indeed taken mid-way between and fields at about 20 m above the seafloor. As a consequence, it is difficult to assess possible contamination originating from sampling device or seawater contribution in the present case. However, earlier studies have shown that they generally did not represent major sources of contamination as for the studied compounds [27, 37]. Nevertheless, in comparison to deep-sea water both the qualitative (Kulo Lasi) and quantitative (Fatu Kapa) data obtained suggested enrichment of the fluids in hydrothermally derived compounds, namely, n-alkanes (C9–C12), n-FAs (C9, C12, C14–C18), and PAHs (fluorene, phenanthrene, pyrene) ([39]; Table 6, Figures 5 and 6). Such enrichment was unclear for >C12  n-alkanes; C10, C11, C13  n-FAs; BTEXs; naphthalene, acenaphthene, and fluoranthene because of their very low concentration and/or the measurement uncertainty.

Differences in concentrations seemed to exist among the vents over the Fatu Kapa area. Fluids from the Stéphanie vent field had concentrations in hydrocarbons equal or below the reference water sample whereas they were clearly enriched in C9, C12, C14–C18 n-FAs. The Carla fluids were slightly enriched in C9–C12  n-alkanes and showed the highest concentrations in PAHs. Fluids from , Fati Ufu, and Tutafi shared some similarities: a strong enrichment in decane and undecane, alike concentrations in PAHs, and the presence of significant amounts of xylene. However, fluids expelled at the Tutafi vent appeared the most enriched in C9–C11  n-alkanes and xylenes. In terms of fatty acids and considering the analytical error, the 5 vents showed consistent concentrations with C9, C16, and C18 being major. Note that fluids from Fati Ufu seemed depleted in C17 and C18.

Generally we did not observe strong linear correlation between the concentration of individual compounds and Mg. Nonetheless, these relations showed that both enrichment and depletion of organic compounds seemed to occur in hydrothermal fluids versus deep-sea water.

5. Discussion

The elemental and gas composition of hydrothermal fluids is mainly affected by water/rock interactions and thus the nature of the host rocks, phase separation, magmatic fluid contribution, conductive cooling, and seawater mixing in local recharge zones [45]. In the following discussion we attempt to unravel the occurrence of these various processes both at Kulo Lasi and at Fatu Kapa. Much less is known on processes that control organic geochemistry and are therefore discussed here as well as some implications of the presence of organic compounds in hydrothermal fluids. Implications related to the composition of the fluids are dependent on fluxes; therefore, we give here an attempt to provide order of magnitude estimates of heat and mass fluxes.

5.1. Plume-Fluids Relations

The geochemistry and dynamics of the plumes over the Wallis and Futuna region have been studied elsewhere [7]. The Kulo Lasi plume has been proposed to be the result of both high- and diffuse venting from multiple vents located both on the floor and on the wall of the caldera. Consistently, both types of venting have been observed [6]. Helium, nephelometry, and Mn profiles recorded above the northern sampling area showed constant elevated concentrations in the 300 masf and were assumed to be the results of diffuse venting. Our results show that they are obviously the result of the numerous small black smokers observed on the seafloor (Figure 2). The methane concentration in the sampled fluids was extremely low which cannot account for the elevated concentration of CH4 in the water column reported by Konn et al. [7]. The strong difference in the CH4/Mn ratios between the plume (0.7–4.5) and the sampled fluids (0.001–0.01) is another line of evidence that the methane plume has another origin compared to hydrothermal fluids and likely come from degassing of the lava flows as suggested by the authors. Although other fluid discharges likely remain undiscovered, this is consistent with a past eruption and accumulation of the water mass in the caldera [39].

A great diversity of the fluid compositions was expected from the geological settings and the water column survey and was indeed confirmed by the mixing lines that point to as many endmembers as sampled areas (Figure S1). CH4/TDM ratios also differed among the vents but it was not due to sole CH4 concentration variations as suggested earlier (Table 5) [7]. Finally, the very weak nephelometry of the Fatu Kapa plume is likely best explained by the low metal contents of the fluids.

5.2. Reaction Zone Depth

The solubility of Quartz in hydrothermal fluids has been studied by different authors (e.g., [46]). According to these works silica concentration in the fluid may be used to estimate the depth of the reaction zone. The silica concentration measured in the Kulo Lasi and Fatu Kapa fluids indicates a hydrothermal reaction zone at seafloor or in the water column (Figure S2). Both observations suggest that, in this area, fluids are not in equilibria with Quartz at the pressure and temperature of the fluid emission. And this prevents using Si as a geothermometer to determine the depth of the reaction zone.

All fluids at Fatu Kapa were indeed highly depleted in Si with respect to the Quartz saturation curve at 170 bar, 300°C (Si ~12 mM in Figure S2). A higher temperature in the reaction zone (>350°C at 200 bar) may explain a lower Si concentration in the fluid at equilibrium as Quartz solubility decreases (Figure S2). The dispersion of a great number of vent fields over a large area of recent lava flows may be due to complex fluid pathways that favour conductive cooling of the fluid and subsurface loss of silica before venting on the seafloor. Consistently, amorphous silica was common in the seafloor deposits at Fatu Kapa where opal was abundant as a late mineral in sulphides and as silica crusts (slabs) at the surface of the deposits [6]. In conclusion, this would indicate a fairly shallow reaction zone at Fatu Kapa (a few 100 mbsf) in agreement with the geological settings and the possible occurrence of dikes.

5.3. Chlorinity

Phase separation is often accounted for salinity deviation in hydrothermal fluids versus seawater [47, 48]. Phase separation is of great importance in metal transportation and ore-forming processes, for example, [24, 4951]. It also implies that seawater experiences dramatic changes in its physical and chemical properties as it reaches the super- or subcritical state. In particular, strong modification of the density and ionic strength of seawater enables unconventional chemical reactions, hence a likely importance in hydrothermal organic geochemistry, for example, [52]. The measured and of the Kulo Lasi fluids are almost on the critical curve of seawater meaning that liquid and vapor phase may coexist at Kulo Lasi. An adiabatic decompression of supercritical seawater (initial fluid and equivalent to 3.2 wt% NaCl) as it rises towards the seafloor would cause it to separate, at about 320–350 bar and 415–420°C, into two phases having the NaCl percentages observed at Kulo Lasi (Figure S3) [53, 54].

Similarly, the excess salinity of the Fatu Kapa fluids (9 to 41%) could be explained by phase separation and is supported by the Br/Cl ratios which significantly differed from seawater [45, 55]. Since we have not sampled any Cl-depleted fluids we may infer that phase separation may have occurred in the past and that only the brine phase was venting at the time of the cruise. Alternatively water-rock reactions could represent a significant Cl source to the fluids [56]. Indeed, the felsic lavas collected in the Fatu Kapa area contained up to 10 times more Cl than MORB (Aurélien Jeanvoine, personal communication).

5.4. Water-Rock Reactions

Generally, fluids from Kulo Lasi and Fatu Kapa were not typical of back-arc settings but shared similarities with ridge, arc, and back-arc settings fluid signatures [3]. The Kulo Lasi fluids have unusually high concentrations of Mg (24.6 to 34.9 mM) and SO4 (6.2 to 12.0 mM) at low pH (2.24 to 3.32) and high (338–343°C) which indicate that significant seawater mixing at subsurface or during sampling is rather unlikely. In back-arc context, the occurrence of Mg and SO4 in endmember fluids can be explained by a magmatic fluid input as observed at the Desmos [5, 57], Rota 1 and Brother sites [58, 59]. Magmatic-derived SO2 would disproportionate according to reaction (1) at temperatures measured at Kulo Lasi (e.g., [5, 60]). This is consistent with widespread occurrences of native sulfur on fresh lava near the active vents [39] as well as the low pH of the fluids.

Yet CO2 concentrations are low and the Na : K : Mg ratios are strongly different to seawater. The latter suggests a contribution of Mg by dissolution of magnesium silicates [39]. Besides, the high Li and Rb concentrations and the presence of recent lava injected in the caldera point to water/fresh hot volcanic rocks interactions. Notably, such interactions are capable of producing the extremely high concentration of H2 measured in the Cl-depleted sample and thus the very unusual H2/CH4 observed [61] (Figure S4). High concentrations of metals are consistent with the highly acidic nature of the fluids coupled with high H2/H2S ratios [62, 63].

The relatively mild pH, 3He/CO2, and R/Ra ratios of the Fatu Kapa fluids are diagnostic of the occurrence of seawater/MORB interactions [6466] (Figure S5). Consistently, the geochemistry of the Fatu Kapa fluids was very similar to the Vienna Woods ones whose composition is mainly the result of interactions with basalts [3, 4]. Yet metal concentrations were lower at Fatu Kapa while Ca, K, and Rb were higher and Li is similar. Plausible explanations for the extremely low metal concentrations observed in the Fatu Kapa fluids are conductive cooling; water/metal-poor rocks interactions; subsurface metal trapping under silica and barite slabs [6]. Given the wide variety of lithologies sampled in the area, fluid compositions are likely the results of interactions with a wide range of rock source chemistries. To that respect, the composition of the local lavas that are characteristic of andesite, trachy-andesite, dacite, and trachy-dacite probably best explains the enrichment in Ca and in the mobile alkali metals K and Rb.

5.5. What Controls Organic Geochemistry?

The origin of hydrocarbon gases and SVOCs in natural systems including hydrothermal systems has been the focus of many studies since the abiotic origin of some hydrocarbons was postulated ([67, 68] for a review). Both field and experimental studies have tried to unravel the origin of hydrocarbons making use of stable isotopes (e.g., reviews of [34, 35]). Although there are strong discrepancies among studies, the variation of δ13C with the carbon number may be a reasonable indicator of the origin. The trend observed in the Cl-depleted sample of Kulo Lasi was very similar to the ones attributed to an abiogenic origin in the Precambrian shields or in the Lost City hydrothermal field [69, 70]. The Kulo Lasi Cl-rich sample exhibited a pattern that has been observed in several Fischer-Tropsch type (FTT) experiments [34]. The strong positive or negative fractionation between C1 and C2 observed in the hot fluids of Kulo Lasi is likely due to chain initiation [71]. Conversely, the low- (135°C) sample that was collected in a beehive-type smoker covered with bacterial mats showed a regular positive trend which has been proposed to be diagnostic of a thermogenic origin. Although we concede that the abiogenic origin of C2+ hydrocarbon gases in the Kulo Lasi field will need more investigation, methane is clearly at the border of abiogenic and thermogenic domains both at Kulo Lasi and at Fatu Kapa with δ13C values ranging from −29 to −6.1 ([72] and Figure 7). Carbon isotopes of CH4 and CO2 suggest that methane underwent oxidation, possibly by bacteria, at both sites and may explain the extremely low concentrations observed (Figure in [73]). Consistently and according to thermodynamic calculations, methanogenesis should be limited under the , , and redox conditions present at the Futuna sites and CH4 consumption might be prevalent [31].

By contrast, carbon isotopes have not appeared to be useful up to date in determining the origin of heavier organic compounds [74]. Several processes are likely to occur simultaneously and to use several C sources, resulting in a nondiagnostic bulk δ13C signature. Several experimental and theoretical studies indicate that a range of organic compounds including linear alkanes and FAs could form and persist in natural hydrothermal systems (e.g., [3135]). However, according to the calculated H2 at and of the study sites, the redox conditions are likely buffered by Hematite-Magnetite (HM) or an even more oxidizing mineral assemblage which appear less favourable for abiotic synthesis than Pyrite-Pyrrhotite-Magnetite, Fayalite-Magnetite-Quartz, or ultramafic rocks assemblages [27, 32, 33] (Table 4). The occurrence of organic compounds in our fluids must thus be attributed to a great part to other processes. Microbial production and thermal degradation of microorganisms, OM detritus, and/or refractory dissolved OM represent good candidates to produce soluble organic compounds. PAHs are indeed common products of pyrolysis of OM [26, 75, 76]. Long chained fatty acids are major constituent of organisms and their presence in the Futuna fluids could be easily associated with thermal degradation of biomass or OM [26, 77]. Yet the distribution of the compounds found in the fluids does not match a simple process of OM degradation. Only >C13  n-FAs occurred in sediments with C16 being the most abundant (Figure S6). However, similar to our samples, both odd and even carbon number n-FAs were observed in the C14–C20 range with odd FAs being less abundant. Petroleum exhibits nearly equal levels of C14–C20  n-FAs. Only the even series has been reported in both massive sulphide deposits (MSD) and hydrothermal mussels with C16 being the most abundant. Short chain FAs (<C13) have only been reported in Lost City fluids but here again only the even series occurred. In any case C9 was reported whereas it was nearly the most abundant in our fluids. Abiotic processes may still be considered as nonanoic acid could be synthesized from CO2 and H2 [31], nonane [78], or undecane [79]. As a difference the presence of C16 and C18 n-FAs in significant amount in the fluids from Fatu Kapa may represent a direct microbial contribution. The distribution observed in the Fatu Kapa fluids likely reflects the occurrence of several concomitant processes possibly including production reactions (abiotic and thermogenic) and consumption mechanisms (adsorption and complexation).

Nonvolatile n-alkanes are usually associated with lower processes such as in oil fields or at the Middle Valley hydrothermal vent field [80]. In the Guaymas basin, where n-alkane-rich sediment samples have been reported, it is less clear what temperature they were exposed to. However and as far as we understood high temperatures were rather associated with absence of n-alkanes and presence of PAHs consistently with high-temperature OM pyrolysis [26, 81, 82]. Pyrolytic processes resulted in the presence of light hydrocarbon gases with an exception of some high (>200°C) fluids containing C9 and C10  n-alkanes. n-Alkanes also occur in solids from unsedimented hydrothermal vent fields ([76] and references therein). Notably, the n-alkanes distribution in our fluids does not resemble any aspects neither the ones resulting of low- processes nor the ones created by high- FTT reactions [83, 84] (Figure S7). C10–C20  n-alkanes usually occur in equivalent amounts in petroleum or show a consistent decrease with molecular weight. Experimental FTT reactions produced consistent increasing concentrations from C9 to C12 and then consistent decreasing concentrations to C20. Similar patterns are also associated with the kerosene fraction of petroleum [85]. Distribution patterns in hydrothermal solids are difficult to picture as usually only chromatograms are provided in the studies, for example, [86], but they largely differ by the simple fact that <C14 alkanes were not detected in most cases as in sediments from various locations [87]. The smaller alkanes may well be preferentially entrained in fluid circulation but they are more likely the result of other processes, especially high-temperature ones, and including abiotic reactions. Note that the latter should not be reduced to sole FTT reactions because supercritical water is a fabulous medium for unconventional reactions [8890].

Formate and acetate have been given more attention in both laboratory [9193] and field hydrothermal studies [28, 29] as these small molecules are likely to prevail according to thermodynamic studies (e.g., [3133]). Where usually formate dominates, acetate was found to be more abundant in fluids from Fatu Kapa. According to Shock studies, at 280–300°C, formic acid concentrations should not be much higher than acetic acid, but this is not enough to explain our “reverse” concentrations. And especially it is not consistent with higher amounts in the 300°C fluids. A ratio close to 1 was observed at Kulo Lasi which may indicate that different production/consumption processes occur. Also the concentrations of the formate and acetate plotted on a line versus Mg which suggest that the fate of these volatile fatty acids at Kulo Lasi is controlled by simple mixing; that is there would be no consumption/production when fluid mixes with seawater (Figure 4). The deep-seawater concentrations were high compared to what is usually reported in the literature which was most likely due to plume contribution [27, 28]. This supports the simple mixing model hypothesis and is consistent with the near absence of organisms around those chimneys.

5.6. Organic Compounds: Implications for Biology, Mineral Resources, and C Cycling

The idea that life could have originated in hydrothermal systems from abiotic reactions was postulated in the late 70s [94]. However, the question of the origin of organic compounds in hydrothermal systems has remained ever since they were evidenced in natural environments [95]. On the one hand, a biogenic or thermogenic origin seems most likely for most compounds investigated so far; on the other hand, one cannot exclude that some of the formate and aliphatic hydrocarbons form abiotically [13, 2628, 30, 96]. As detailed in the previous section, our results are consistent with a mix of origin although abiotic synthesis likely occurs to a far smaller extent than other processes that would overprint an abiotic signature.

Upon the hot topic of the origin of life, the mere presence of organic compounds is highly important for the fauna at the local and regional scales. It is well established that VFAs constitute a significant food source for some microorganisms and thus help sustaining hydrothermal ecosystems [97100]. Besides, some bacteria have proven to be capable of using naphthalene [101] and tubeworms, hydrocarbons [102].

Organics can form complexes with metals [20, 21]. This greatly improves the dispersion of metals in the ocean and prevents them from precipitation as sulphides or oxyhydroxydes [23, 103]. Notably fatty acids are efficient ligands that play a major role in making metals bioavailable as well as in transporting them both through the upper crust ([17] and references therein) and through the water column in the plume [11, 104106]. In addition, they have been shown to be involved in growth/dissolution processes of some minerals [19, 107]. For these reasons, they are of particular importance in ore-forming processes. Hydrocarbons which are weaker ligands would react with sulfates to generate bisulfide (HS), which in turn would easily react with metal chlorides to form metal sulphides according to the following mass balance equations:where R is a carbonated chain, either aliphatic or aromatic, and represents OM [108]. To that respect hydrocarbons are likely to be involved in depositional processes of metals. Notably associations of aliphatic and aromatic hydrocarbons with mineral deposits have also been observed on the EPR [109] and in sulphide sedimentary deposits on land [104].

5.7. Fluxes: Importance of Back-Arc Hydrothermal Systems to the Ocean Geochemistry

Hydrothermal input to the ocean via plumes has long been neglected but recent results of the GEOTRACES program clearly show its importance in terms of metals and trace elements transportation and implications for ocean biogeochemistry [13, 110113]. While it is now well established that MOR hydrothermal discharge has a large impact on the global ocean chemistry and element cycles, the relative impact of hydrothermal activity from other hydrothermal settings has not been established. The extensive hydrothermal activity reported in the Wallis and Futuna region suggests that back-arc system hydrothermalism may be of much greater importance than previously anticipated [7, 114]. Estimation of hydrothermal fluxes is generally challenging and very few data are available in the literature [115118]. Therefore we believe that any kind of estimation, even orders of magnitudes are of importance to make advances in this field. We propose to combine two different approaches based on geophysical data and video recordings (see here), respectively, to propose such estimates with some confidence.

5.7.1. Estimation Using Geophysics

We can make an order of magnitude estimate of the heat flux from the different hydrothermally active areas based on the physical characteristics of the plumes. Marshall and coworkers [119121] have proposed a scaling relationship between the heat flux at an interface, Hf, the ambient buoyancy frequency () in the surroundings of the plume, the characteristic size () of the heat transfer region, and the equilibrium height (or depth; ) reached by the plume, aswhere is seawater density (1030 kg·m−3), is heat capacity of seawater (~4000 J·kg−1·K−1), and is gravitational acceleration (9.81 m·s−2), and is the thermal expansion coefficient of seawater (10−4 K−1). The ambient buoyancy frequency () can be estimated to be between 0.001 and 0.002 s−1 from CTD profiles in the area using a routine in the UNESCO Sea Water Library described by Jackett and Mcdougall [122]. The radius () of the Kulo Lasi caldera is 2500 m and the plume rose in average about 200 m above seafloor (top layer boundary) [7]. For order of magnitude estimates, the ~130 km2 Fatu Kapa area can be approximated by a disk of radius 6400 m, with a similar plume height. Introducing these numbers in (4) leads to heat flux estimates ranging between 100 and 800 W·m−2 for the Kulo Lasi caldera and 50 and 400 W·m−2 for the Fatu Kapa area, which is greatly dependent on the value for buoyancy frequency. While these estimates scale proportionally with the area considered to be hydrothermally active, they integrate the sources within the area which do not demand that the whole area be active. We estimate that total heat inputs are in the 2–16 GW for the Kulo Lasi caldera and 5–40 GW for the Fatu Kapa areas.

5.7.2. Estimation Based on Video Postprocessing

Video recordings could be used to estimate fluid velocities of the Carla and chimneys and of a few smokers at Kulo Lasi using for instance the Typhoon algorithm [123] (Figures S8S11). This optical flow method recovers the (2D) fluid flow from the apparent displacements, in an image sequence, of tracers advected by the flow. Here, the plume acts as the tracer. Compensation for the camera and vehicle motion and parallax correction were not possible, so the investigated video sequences where chosen according to (i) the overall stability of the camera and vehicle and (ii) the plume being as perpendicular to the camera as possible. The relevant lengths scales (image spatial resolution and diameter of the chimneys) had to be estimated from known object sizes in the same ground, typically shrimps. External diameters of the chimney were used for calculation as any estimation of the internal diameter on video recordings would be too speculative. Note that (i) chimney samples taken at Kulo Lasi exhibited similar internal and external diameters; (ii) the large anhydrite chimneys (e.g., , Carla) at Fatu Kapa did not seem to have any central conduit but rather exhibited a sponge-like structure leaking fluid at a high velocity from the entire volume. As a result, we believe the overestimation resulting from this assumption to be limited. Finally, the observed flow velocity is assumed to be constant across the jet section. Given all these limitations and assumptions, the resulting fluxes values should be taken as an indication of their order of magnitude.

The fluid velocity was estimated to be on average 0.05, 0.15, and 1 m s−1 for Kulo Lasi, Carla, and , respectively. In terms of heat fluxes, Carla (diameter ca. 70 cm) would generate ~6 MW while (diameter ca. 250 cm) would produce ~5.7 GW, respectively. Associated mass fluxes would be 54 L s−1 and 5 m3 s−1 which means, for instance, that the single chimney could generate an input of 2.6 × 107 mol y−1 CH4 to the ocean. Comparatively, estimation of the total efflux of methane from serpentinisation ranges from 15 to 84 × 109 mol y−1 including 9 × 109 mol y−1 for the sole slow spreading ridges. Similarly, the Carla chimney would release about 5.7 × 103 mol y−1 of dodecane that may help forming 1.4 mol y−1 of metal sulphides (see Section 5.6). Cumulative observations during the dives brought to a total of 220 smokers of various sizes (~5 cm to ~2.5 m in diameter) and apparent flows (strong, medium, slow). We assigned the strong, medium, and slow flows observed to the velocities of , Carla, and Kulo Lasi, respectively. At Kulo Lasi about 100 smokers were counted during the Nautile dives and all appeared very similar in diameter (~3 cm) and fluid flow (0.05). Keeping in mind these uncertainties, an order of magnitude of the heat and mass fluxes generated by hot smokers at Fatu Kapa are estimated to be 6.8 GW and 6 m3 s−1 for the Fatu Kapa area versus 9 MW and 6 L s−1 for the Kulo Lasi caldera. This means, for example, that the total Fe flux from hot fluids emanating from the caldera would be up to 1.9 × 106 mol y−1 versus recent estimations of the global hydrothermal iron input that are about 109 mol y−1 [112, 124]. The average nonanoic acid concentration in Fatu Kapa purest fluids is 7.25 ppb, which would result in 1.4 × 106 mol y−1 released in the ocean by the Fatu Kapa hot smokers. The carboxylic acid functional group of fatty acids makes them good potential ligands to form coordination complexes with iron, which stabilises iron in the plume in its reduced form [103, 125]. Hence, the example of nonanoic acid suggests that fatty acids could largely contribute to iron stabilisation.

However, the high-temperature fluxes calculated above failed to include heat fluxes from diffusive venting, which was largely present in both areas and is thought to be an important part of the global hydrothermal heat flux (up to 98%) [126]. The surface of the diffusive areas was also assessed on the videos. However, because the velocity of diffusive fluids could not be estimated using Typhoon, we assumed hydrothermal waters are exiting the seafloor at the minimum velocity reported for low temperature flow (0.04 m s−1) [127]. The cumulative surface of diffusive areas with a typical temperature of 10°C reached 100 m2 at Kulo Lasi and 2885 m2 at Fatu Kapa. In addition, a particular area of about 300 m2 at Kulo Lasi consisted in hundreds of silica chimney diffusing a 40°C fluid [6]. The resulting contribution of diffuse venting to the heat flux would be 2.14 GW and 5.3 GW at Kulo Lasi and Fatu Kapa, respectively. This brings the total heat flux estimates at 2.15 GW and 12 GW, respectively, which is consistent with the estimates obtained using the lower value as well as the fact that only a small portion of the total surface of the sites was explored with the submersible.

5.7.3. Summary

According to these different estimates heat efflux at Kulo Lasi and Fatu Kapa are conservatively estimated to be at least for 1-2 GW and 5–10 GW, respectively; this estimate is based on the low value, whereas using the higher suggests a flux almost 10x higher. It seems highly likely that the Wallis and Futuna active areas combined with the 3 calderas to the East [114] have a heat flux of >10 GW. Vent fields on MOR have been reported to generate between 10 MW and 25 GW ([116, 117, 127, 128] and references therein) and the total hydrothermal heat flux at MORs is estimated to be about 1000 GW [129, 130]. This suggests that the presently discovered area might be of significant importance in the global budget and that back-arc hydrothermal activity contributes as much as MOR systems, and, possibly more, to the global ocean chemistry and cycles. Few estimates of hydrothermal heat flux have been published and the relative importance of heat, fluid, and geochemical hydrothermal fluxes from different environments will require studies designed to more accurately gauge these fluxes.

6. Concluding Remarks

The study of the geochemical characteristics of hydrothermal fluids from the Wallis and Futuna area confirmed the great potential of the region to generate a variety of fluid chemistries, as it was expected considering its particular geological context. This supports the idea that the hydrothermal contribution of back-arc environments is of great interest for the global ocean chemistry. Our order of magnitude estimates of fluxes suggest that back-arc hydrothermal activity contributes as much as MOR systems and possibly more. Notably the sole chimney could generate ~1 of the total hydrothermally derived CH4. The diversity observed in the Wallis and Futuna area also emphasizes that each new field presents its own characteristics and that exploration should continue. A huge number of sites remain to be discovered according to the newly published estimation of vent fields occurring on Earth [131].

A special focus was brought on organic geochemistry because of the few data available in modern hydrothermal systems despite the recent growing interest for oceanic OM. Concentrations of SVOCs are the first to be reported which will have implications in a wide range of questions and fields. Our results are relevant to the understanding of C cycling and complete the works by Hawkes and Rossel who demonstrated that DOM is recycled if not removed partially through hydrothermal systems but who could not identify compounds in DOM. Identification of organic molecules is especially needed to better understand organometallic chemistry at hydrothermal vents and thus utilisation by microbes, metal export, and ore-forming processes. The distribution patterns obtained revealed the occurrence of several processes controlling organic geochemistry and notably that one cannot exclude abiotic synthesis to occur in the study area but very likely to a so small extent that the signature would be overprinted.

This brings the idea that using natural concentrations to feed thermodynamic models of abiotic synthesis and/or guide the design of experimental work should enable making progress in unravelling the origin of organic compounds in hydrothermal systems. In addition, growing techniques as clumped isotopes [132] and position specific isotopes measurements [133] are available and should also help answering this question.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Acknowledgments

The 2010 and 2012 cruises to the French EEZ of Wallis and Futuna were financed through a public/private consortium of the French state, Ifremer, AAMP, and BRGM and industrial groups including Eramet, Technip, and Areva. The authors are very grateful to the ship crew and the ship captains JR. Glehen, P. Moimeaux, and R. Picard for running these three cruises with skills and professionalism. They acknowledge all the scientific parties of these cruises for their collaboration. They are also grateful to D. Pierre, C. Guerin, and A. Normand for processing bathymetric data on board and thank A.-S. Alix for providing the final maps. They are indebted to the physical oceanographers L. Marié and B. Le Cann who helped a lot with fluxes estimations and water mass physics. Finally, many thanks are due to P. Dérian from the Fluminance team (Inria, Rennes, France) who postprocessed video recordings using Thyphoon for fluxes estimation.

Supplementary Materials

Supplementary 1. S1: mixing lines used for calculation of the endmember composition of the Fatu Kapa fluids. S2: modified after Von Damm et al. [46]. Plot of the molality of dissolved SiO2 in equilibrium with quartz in seawater versus temperature for isobars from 1500 to 1000 bar according to Von Damm et al. model. The Si most enriched fluid collected at Kulo Lasi is represented by the blue star. The red circle covers the range of Si concentrations and encountered in fluids from the Fatu Kapa vent field. S3: modified after Bischoff and Pitzer [53]. Stars stand for Kulo Lasi fluid phases characteristics. They nearly plot on the 150-bar isobar. The close-up of the 400°C, 300-bar region shows that seawater could produce the observed salinities at Kulo Lasi by phase separation at about 320–350 bar and 415–420°C. S4: modified after Kawagucci et al. [134]. Plots of H2 concentration versus CH4 concentration in various hydrothermal fluids. The grey area represents values observed in a hydrothermal experiment using natural seafloor sediments. Values obtained for the Wallis and Futuna vent fields are reported: Kulo Lasi brine and condensed vapour phases are marked by the red square and the blue diamond, respectively, and the blue shaded area covers the range of values obtained in the Fatu Kapa field. S5: modified after Lupton et al. [66]. (a) Plot summarizing 3He/4He ratio versus C/3He for various mantle provinces, including mid-ocean ridges (black-filled symbols), submarine arc volcanoes (blue), and sub aerial arc volcanoes (green). Values for the Fatu Kapa vent field are reported as orange diamonds. 3He/4He is expressed as R/Ra. Crosses indicate average values for MORBs and for subaerial arcs from. (b) Similar plot including values for hotspot volcanoes such as Loihi, Kilauea fumarole, Yellowstone Park gases, Reunion, and Fatu Kapa (orange diamonds). S6: distribution of linear fatty acids in various environments. Data are from [135] for Massive Sulphide Deposits (MSD); [36] for Lost City (LC) fluids; [136] for petroleum and recent and ancient sediments; [137] for 13°N mussels. S7: distribution of linear alkanes obtained by thermogenic maturation in various crude oil basins and abiotic Fischer-Tropsch type experiment [84]. S11: time series of the estimated displacements corresponding to the video sequences shown in Figures S8, S9, and S10.

Supplementary 2. S8: example of a postprocessed video sequence using the Typhoon algorithm to estimate displacements (instantaneous, left panel; averaged on 25 frames, right panel) on one of the small black smokers in the Kulo Lasi caldera.

Supplementary 3. S9: example of a postprocessed video sequence using the Typhoon algorithm to estimate displacements (instantaneous, left panel; averaged on 25 frames, right panel) at the base of the Carla chimney.

Supplementary 4. S10: example of a postprocessed video sequence using the Typhoon algorithm to estimate displacements (instantaneous, left panel; averaged on 25 frames, right panel) at the top of the massive chimney. S11: time series of the estimated displacements corresponding to the video sequences shown in Figures S8, S9, and S10.