Fluids were recovered using 150ml titanium isobaric gas tight (IGT) samplers (Seewald et al., 2002) operated with the Remotely Operated Vehicle (ROV) MARUM-Quest4000m. Two fluid samples were collected at the orifices of Dog’s Head and Iced Bun at the E2-S segment and one sample was recovered at the newly discovered site called ‘E2-West’ (Bohrmann, 2019, Table 1 ). The fluids were sampled after breaking off the tip of the chimney structure to allow for better flow of fluid. The temperature of the fluids collected with IGTs were monitored in real-time using a thermocouple attached to the IGT or with the ROV’s temperature sensor immediately after fluid sampling. After recovery of the ROV, the IGTs were processed onboard to obtain the fluid as soon as possible (within 12 hours of sampling). The samples were extracted from the IGTs using a HPLC pump and analysed for concentrations of dissolved gases (H₂, CH₄ and H₂S) and pH_(25°C).

Filtered (0.45 µm) fluid aliquots for major and trace elemental analysis were collected in acid-cleansed high density-polyethylene (HDPE) Nalgene™ bottles and acidified using 200 μL of concentrated sub-boiled HNO₃. Filtered and unacidified aliquots were collected for anion analysis and for stable hydrogen (δD_H2O) and oxygen (δ¹⁸O_H2O) water isotope analysis in sealed glass ampoules. To determine the total dissolved inorganic carbon (ΣCO₂, abbreviated as CO₂ hereafter), fluid aliquots were injected from gas-tight syringes into pre-weighed He-filled and subsequently evacuated glass serum vials to prevent atmospheric CO₂ contamination. The serum vials were stored upside down to seal the septa and prevent diffuse gas loss. For determination of δ¹³C and δD in CH₄ and δ¹³C_CO2, 12-15 ml of gas at standard pressure (STP) were injected into 20 ml glass serum vials containing concentrated NaCl solution. Atmospheric contact of the sample gas was prevented.

Concentrations of H₂ and CH₄ were determined onboard using a 7820A Agilent gas chromatograph (GC). A syringe headspace extraction was done first and the H₂ and CH₄ in the headspace gas was measured using different detectors (Reeves et al., 2011). H₂ was quantified using a thermal conductivity detector (TCD), while CH₄ was determined using a flame ionization detector (FID). The GC was equipped with a Molsieve 60/80 column (Sigma-Aldrich, St Louis, MO) and was operated with N₂ as a carrier gas at 90°C. The analytical uncertainties (2s) are considered as ±5% for H₂ and CH₄. pH (at 25°C and 1 atm) was measured potentiometrically using a ‘seven2go’ pH electrode (Mettler Toledo, USA) instantly after the fluid was removed from the gas-tight syringe to minimize the effect of degassing. The electrode was calibrated daily using a 4-point calibration with reference solutions of pH 2, 4, 7 and 11. The uncertainty for pH values can be considered within ±0.1 unit.

Concentrations of major and minor elements were determined using inductively coupled plasma optical emission spectroscopy (ICP-OES, Varian Vista Pro, radial plasma observation) and ion chromatography (IC, Metrohm CompactIC) with a ‘METROSEP A Supp 5-150/4.0’ column at MARUM, University of Bremen. Trace elements including rare earth elements (REEs), were determined using inductively coupled plasma mass spectroscopy (ICP-MS, Perkin Elmer NexION) at Jacobs University, Bremen. The REEs were determined following a matrix separation and a pre-enrichment method (Schmidt et al., 2010). An ion-exchange column (Sep-Pak C18 cartridge™) was used for the matrix separation. The accuracy of major and minor elements was monitored using IAPSO standard seawater (supplied by Ocean Scientific International Ltd., UK). The quality of the trace element concentration measurements was monitored using the certified reference material NASS-7 (seawater) from the National Research Council of Canada. The analytical uncertainties (2s) are ±2% for dissolved Na, Si, Mg, Cl, Br and SO₄, ±5% for dissolved Ca, F, Li, Sr, Ba, Al, Fe, Mn and K, ±8% for dissolved B, Rb, Cs and <±10% for REEs.

Total dissolved sulphide (ΣH₂S, hereafter abbreviated as H₂S) was determined gravimetrically at the Faculty of Geosciences, University of Bremen, following shipboard precipitation as Ag₂S in a 5 wt.% AgNO₃ solution (prepared on a daily basis) in a method adapted from Seewald et al. (2015). The estimated working range for the gravimetric method is >1mM. Low H₂S concentrations (0.2μM to 1mM) were determined photometrically using the methylene blue method by Cline (1969). CO₂ concentrations were determined onshore after the fluid samples were acidified with 25 wt.% phosphoric acid into the aliquots directing the headspace gas directly into the GC (Thermo Scientific Trace GC Ultra), which was equipped with a Haysep 80/100 column (Sigma-Aldrich, St Louis, MO) and was operated with He as a carrier gas at 50°C. This method was adapted after Reeves et al. (2011). The analytical uncertainties (2s) are considered as ±10% for CO₂ and ±15% for H₂S relative to measured values.

Stable oxygen and hydrogen isotope compositions of vent fluid H₂O were analysed using cavity ring-down spectrometry (CRDS, Picarro L-2130i) at MARUM. The measurement consisted of nine injections of 7 μL each and the result is an average of the last three injections out of nine. The isotope ratios were normalized to VSMOW seawater material. For water isotopes the analytical uncertainties are ∼0.09‰ and ∼0.25‰ for δ¹⁸O_H2O and δD_H2O, respectively.

Stable carbon and hydrogen isotope ratios (¹³C/¹²C and ²H/¹H) of CH₄ and CO₂ were determined using GC‐isotope ratio mass spectrometry (GC-IRMS) at MARUM as detailed in Pape et al. (2020a). CH₄ and CO₂ were separated by gas chromatography. Subsequently, CH₄ was either combusted (1,030°C) or pyrolyzed (1,440°C) to generate CO₂ or H₂, respectively. CO₂ or H₂ were then transferred to the IRMS for analysis of ¹³C/¹²C or ²H/¹H. All gas samples were injected at room temperature by manual syringe injection. Reported isotope ratios are arithmetic means of at least triplicate measurements and are given in the δ-notation relative to Vienna Pee Dee Belemnite (VPDB) and Vienna Standard Mean Ocean Water (VSMOW; for carbon and hydrogen, respectively). Reproducibility was proved using commercial CH₄ and CO₂ standards (Isometric Instruments, Canada; Air Liquide GmbH, Germany). Standard deviations of triplicate stable isotope measurements were <0.5‰ (δ¹³C_CH4). Because CH₄ concentrations were below the calibrated range of the GC-IRMS at the MARUM, samples were re-analysed for δ¹³C_CH4 by GC–IRMS at GEO‐data Gesellschaft für Logging‐Service mbH, Germany (see Pape et al., 2020b).

Sulphur isotopes (δ³⁴S_H2S) were measured using an EA-IRMS (elemental analyser isotope ratio mass spectrometry) using a Flash EA isolink interfaced to a ThermoFisher Scientific Data V Advantage mass spectrometer. Analysis was conducted at the Institute of Geology and Palaeontology at the University of Münster. Results are reported in the δ-notation relative to Vienna Canon Diablo Troilite (VCDT). Reproducibility of the measurement was determined by replicate measurements and was better than ±0.3‰. Analytical performance was monitored using international reference materials IAEA S1, S2, S3 and NBS 127 as well as lab internal standards.

Microthermometry was carried out on anhydrite-hosted fluid inclusions from a chimney sample at the vent orifice of Dog’s Head. A Linkam heating/freezing stage in combination with a LNP2 flow regulator and a Zeiss Axioskop was used to observe phase transitions in two-phase liquid-vapour fluid inclusions. The temperature of the heating/freezing stage was calibrated using FLINC^® synthetic fluid inclusions hosted in quartz. Phase transitions in the H₂O system (freezing point: 0.0°C, critical temperature: 374.1°C) were used to calibrate the temperature sensor of the Linkam stage. The calibration procedure yielded temperatures of -0.1°C and 373.4°C for the freezing point and critical temperature, respectively. These values agree with the range of 15 long term control measurements of 0.04 ±0.07°C and 373.6 ±0.75 °C and show that the temperature sensor provided accurate and precise temperatures. Both accuracy and precision are within ±0.1°C at temperatures around the freezing point of pure H₂O and are better than ±1°C in the high-temperature regime (>>100°C).

Sampled hydrothermal vent fluids inadvertently contain a fraction of seawater that was entrained naturally in the subseafloor immediately prior to sampling or accidently during sampling, due to the dead volume of the inlet snorkel. This typically results in a two-component mixture of seawater and hydrothermal fluid. We have used the Mg concentrations in the vent fluid samples at E2-S and E2-W to calculate the chemical composition of hydrothermal endmember solution for all elements by extrapolating to zero Mg.

Endmember temperatures were calculated using isenthalpic-isobaric mixing, by considering a temperature and salinity-dependent heat capacity of the fluid. This dependence of salinity and temperature was calculated using the scheme by Driesner (2007), which calculates the thermodynamic data of H₂O-NaCl fluids using thermodynamic data of pure water (Haar et al., 1985).

Fluid inclusions in hydrothermal precipitates can be used for tracing salinity and temperature in hydrothermal fluids. When crystals grow from hydrothermal solutions, they tend to incorporate microscopic volumes of the hydrothermal fluid, called fluid inclusions. The study of phase-transitions in these inclusions allows for the reconstruction of physio-chemical conditions in active and fossil hydrothermal systems (Peter and Scott, 1988; Xu, 2000; Baker and Lang, 2001; Bieseler et al., 2018; Diehl et al., 2020a). Measurements of ice-melting temperatures (T_m) and homogenization temperatures (T_hom) in two-phased fluid inclusions enable the calculation of salinities and entrapment temperatures (T_e) (Vityk et al., 1994). Both measurements represent the properties of the hydrothermal fluid during crystal growth if the inclusions have not changed in composition or volume since they were incorporated (Roedder, 1984; Bodnar, 2003) they provide key information on temperature and salinity of hydrothermal fluids during crystal growth. Fluid inclusions in anhydrite are known to re-equilibrate easily and may increase in volume (stretching) during the microthermometric measurement itself. We used the procedure proposed by Vanko and Bach (2005) to avoid inclusions being affected by stretching.

Using thermodynamic relationships in the H₂O-NaCl system, salinities were calculated from ice-melting temperatures (T_m) while entrapment temperatures (T_e) were calculated from the homogenization temperatures (T_hom) and salinities using empirical relationships (Vityk et al., 1994; Atkinson, 2002). The uncertainties in the determination of these two parameters is based on the uncertainty of the temperature determination during the microthermometric work. The accuracy and precision are ±0.1°C and ±1°C for the freezing point and the homogenization temperature (T_hom) determinations, respectively, and result in maximum uncertainties of <±0.2 wt.% NaCl eq. and <±2°C for entrapment temperatures (T_e). The pressure correction for entrapment at seafloor conditions (250 bar) corresponds to a 15-17°C adjustment.

The use of IGTs facilitated the measurement of concentrations of gases dissolved in the fluids, and hence the reconstruction of in-situ pH and redox, which are key parameters controlling speciation of elements and solubility of minerals in the hydrothermal fluids prior to venting. The H₂ and H₂S concentration data were plotted in activity-activity plots, constructed using the R based software package CHNOSZ (Dick, 2019). We assumed unity activity coefficients for both gases.

Sub-surface processes in hydrothermal systems can be examined using thermodynamic reaction-path models. Reaction-path models were constructed using the Geochemists Workbench™ (GWB) software package and a tailor-made 400 bar database assembled by SUPCRT92 (Johnson et al., 1992). For these models, measured bottom seawater (BSW) was heated to hydrothermal temperatures while allowing minerals to precipitate. Using the REACT module of GWB, the heated fluid was then allowed to react with the host rock till the fluid became rock-buffered i.e., attained quartz saturation. Host rock data for samples taken in close vicinity to the vent sites (Sample ID: WX.30, 29, 28, 27, 2 & 1) were utilized from Leat et al. (2000). All major elements were considered for the reaction path models (except P). Ba, Cu and Zn concentrations were formally recalculated and added to the system as barite, chalcopyrite, and sphalerite.

Catabolic energy landscape computations were carried out as described in Amend et al. (2011). The REACT module of GWB was used along with a 250 bar database assembled using SUPCRT92 (Johnson et al., 1992) to simulate mixing of cold BSW with hot hydrothermal fluid represented by the Iced Bun site endmember composition. Minerals were not allowed to precipitate during mixing and redox reactions were also suppressed. Acid-base reactions were allowed to equilibrate. For this model 8 catabolic reactions were considered ( Table 2 ). Five of the reactions represented aerobic respirations using different inorganic electron donors and three denoted anaerobic respiration.

Values of Gibbs energy (ΔG_r) for catabolic reactions were computed using the relation:

where Δ _rG^o denotes the standard Gibbs energy of reaction, R and T represent the gas constant and temperature in Kelvin, respectively, and Q_r stands for the activity product of reaction r, which is evaluated with the relation.

where a_i denotes the activity of species i raised to the stoichiometric reaction coefficient v_i,r . The B-dot equation was used to calculate activity coefficients needed to convert species concentrations to activities. The Q-term accounts for the chemical composition of the mixed hydrothermal fluid, it also accounts for intracellular concentrations for building blocks of cells. Values of Δ _rG^o were calculated at 250 bar and the temperatures of interest with the computer code SUPCRT92 (Johnson et al., 1992), which relies on the relation of

where Δ a G i o represents the apparent standard Gibbs energy of formation of the i^th species in reaction r.

For catabolic reactions the amount of energy available was then calculated as a function of either temperature or mixing ratios. This was done by multiplying the calculated Gibbs energy for the reaction at each temperature with the concentrations of reactants (in the mixed fluid). The stoichiometry of the reaction and the reactant present in limiting supply were taken into account and then multiplied with the amount of mixed fluid at that specific temperature (McCollom, 2007; Amend et al., 2011).

Focused venting of black smoker-type fluids at Dog’s Head and Iced Bun in the E2-S area took place at maximum temperatures of 344°C and 320°C, respectively ( Table 1 ). The fluids have a measured pH _{(25°C, 1bar)} of 3.3 at Iced Bun and 3.2 at Dog’s Head. Calculated pH _(in-situ) values for these fluids were 4.8 for Iced Bun and 5.3 for Dog’s Head ( Table 1 ). Isenthalpic-isobaric mixing models indicate endmember temperatures of 386°C and 342°C for Dog’s Head and Iced Bun, respectively ( Supplementary Figure 1 ). These temperatures are accurate if all the Mg measured was contributed from entrainment of ambient seawater upon sampling.

The fluid at E2-W was characterized by a maximum measured temperature of 53°C ( Table 1 ) and a relatively high Mg concentration (42.6 mM). An isenthalpic-isobaric mixing model indicates an endmember temperature of 273°C ( Supplementary Figure 1 ). The fluid had a pH _(25°C) of 6.7 and a pH _(in-situ) of 6.6 ( Table 1 ).

The dissolved gases investigated in the fluids at E2-S (H₂, H₂S, CO₂ and CH₄) show variable concentrations and isotopic compositions ( Table 1 ). The fluids at Dog’s Head and Iced Bun have endmember H₂ concentration of 45.2 and 65.1 μM, respectively ( Figure 4A ). Endmember concentrations of CO₂ are 8.65 and 12.5 mM, respectively. CH₄ concentrations are 32.0 μM at Dog’s Head and 53.5 μM at Iced Bun ( Figures 4B, C ). The endmember fluids at E2-S have δ¹³C_CO2 values of -4.1 and -5.6‰ ( Table 3A ) and δ¹³C_CH4 values of -7.0 and -7.6‰. H₂S concentrations in the endmember fluids are 3.3 and 4.0 mM ( Figure 4D ) and δ³⁴S_H2S values are between +4.0 and +6.7‰.

The endmember concentrations for H₂ and CH₄ at E2-W were 3.75 and 9.03 μM, respectively ( Figures 4A, B ). The H₂S concentration in this fluid was below the detection limit (<0.2 μM). The CO₂ endmember concentration is 8.55 mM ( Figure 4C ). These concentrations of dissolved gases at E2-W are lower than that at E2-S. The δ¹³C_CO2 value of -10.8‰ is lower than for the fluid at E2-S. Endmember values were not calculated for δ¹³C_CO2 due to processes affecting the CO₂ concentrations of the fluid in the mixing zone (cf. Formation of fluid at E2-West).

The high-temperature fluid feeding E2-S is depleted in Cl by ∼9% relative to seawater with the lowest Cl endmember measuring 491 mM ( Figure 4E ). Sulphate is depleted in the Iced Bun and Dog’s Head fluids with endmember concentration of ≤1 mM ( Figure 4F ). The Na_CB (Na calculated based on charge balance of the endmember) endmember concentration is depleted by ∼10% relative to seawater ( Figure 4G ). Endmember Ca concentrations in the fluids are 31.8 and 26.8 mM ( Figure 4H ). Endmember Br and F is depleted relative to seawater with 809 – 754 μM and 47.5 - 59.4 μM, respectively ( Table 3A ). Endmember Sr concentrations show both, depletion and enrichment relative to seawater at Dog’s Head and Iced Bun, respectively ( Table 3A ). The endmember concentrations of alkali metals K, Li, Rb and Cs (33.7 – 35.1 mM, 383 – 545 μM, 52.6 – 55.7μM, 468 – 538 nM) are higher than those of seawater ( Table 3A ).

Fluids sampled at E2-W have seawater-like concentrations of Na, Cl and Br. Endmember concentrations of K, Ca and Li are 29.3 mM, 30.2 mM and 328 μM, respectively, and are like endmember E2-S concentrations. However, SO₄, F and Si have endmember concentrations of 13 mM, 8.36 μM and 11.5 mM, respectively. These concentrations are markedly different from the endmember concentrations of the E2-S fluids.

The K/Cl, Sr/Cl, Li/Cl, Cs/Cl, Rb/Cl and Br/Cl are elevated in the fluids venting at E2-S and E2-W, except for Br/Cl at E2-W, which is seawater-like ( Table 4 ). F/Cl ratios in all fluids are lower than in seawater. The E2-S element-to-Cl ratios measured in this study show deviations to those reported by James et al. (2014) ( Table 4 ). Dog’s Head fluids have lower Ca/Cl, K/Cl, Sr/Cl, Li/Cl, Cs/Cl and Br/Cl ratios than those sampled in 2010, whereas Na_CB/Cl ratios are slightly higher compared to the 2010 ratios.

E2-S endmember fluids are characterized by concentrations of Fe and Mn of 624 – 1074 μM and 1424 – 1821 μM, respectively. The metal concentrations in the endmember fluids from E2-S are lower than in the endmembers calculated from the 2010 samples ( Figures 4K, L ). E2-W has an endmember Fe and Mn concentrations of 69.0 and 351 μM ( Table 3A ; Figures 4K, L ).

The total endmember REE (ΣREE) concentrations in the fluids from E2-S are 27 nmol/kg and 29 nmol/kg at Iced Bun and Dog’s Head, respectively ( Table 3B ). Chondrite-normalized REE_N distribution pattern shows the characteristic light REE (La-Nd) enrichment and a positive Eu anomaly (Douville et al., 1999; Douville et al., 2002; Craddock et al., 2010, Figure 5 ).

The δ¹⁸O_H2O endmember value of +0.4 and +0.6‰ for fluids from E2-S are ¹⁸O-enriched compared to the endmember ratio for the fluid sample at E2-W (-1.7‰, Table 3A ). The δD_H2O values for all fluids are lower than seawater, with the lowest values in the fluid at E2-W (-2.3‰). The endmember δD_H2O and δ¹⁸O_H2O values recorded in this study are lower than those of the fluids sampled in 2010 ( Table 3A ).

Microthermometry of 26 fluid inclusions was conducted in anhydrite precipitates from a chimney sample of Dog’s Head ( Supplementary Table 1 ). Two fluid inclusion types were identified during the investigation: Inclusion Type I is single phased and was not analysed. Inclusion Type II contains two-phases, with vapour-liquid inclusions and can be subdivided into Type IIa (<50 vol.% vapour) and Type IIb (≥50 vol.% vapour). Type IIa is the dominant inclusion type with 22 out of 26 measured inclusions. The remaining of 4 inclusions is classified as Type IIb. All Type II inclusions exclusively homogenize into the liquid phase suggesting a similar origin. T_m (ice-melting temperature) in individual inclusions ranged from -1.6 to -2.1°C. The corresponding calculated salinities range between 2.7 wt.% NaCl eq. (minimum) and 3.6 wt.% NaCl eq. (maximum). The average T_m of -1.8°C corresponds to a mean salinity of 3.1 wt.% NaCl eq., just slightly below seawater, for all inclusions. Calculated salinities (mean) for different inclusion types (Type IIa and Type IIb) are 3.1 and 3.2 wt.% NaCl eq. According to measured salinities, the inclusions can be grouped into two populations ( Supplementary Figure 3A ): 9 inclusions were determined to contain fluids with salinities significantly below seawater (<3.0 wt.% NaCl eq.) and 22 inclusions contained fluids with near-seawater salinities (3.0-3.6).

T_hom were determined between 211°C and 316°C ( Supplementary Figures 3B, C ). The calculated T_e range between 226-333°C with 9 of 22 T_e determinations above 300°C. The highest T_e was determined as 333°C, just 11°C lower than the measured temperature of 344°C during fluid sampling. Both these temperatures are significantly below the boiling temperature of 386°C at seafloor pressure.

Chloride is the dominant anion in hydrothermal fluids, it ranges in concentration due to phase separation and plays a key role in complexing dissolved metals. Hydrothermal vent fluids commonly show a tight correlation between the concentrations of cations and chloride. Metal concentrations in vent fluids can therefore be normalized to Cl for a comparison of fluids affected by different degrees of phase separation. It is known that partitioning of cations in the low-salinity fluid and the brine phase is not uniform during supercritical phase separation. Thus Cl normalization allows a better understanding of the dissolved species ( Table 4 , Berndt and Seyfried, 1990; Foustoukos and Seyfried, 2007a). Na/Cl and Ca/Cl ratios at E2-S are in the range for unsedimented mafic-hosted hydrothermal systems (Gallant and Von Damm, 2006; McDermott et al., 2018; Diehl and Bach, 2020b; Diehl and Bach, 2021). E2-W has a seawater like Na/Cl and Ca/Cl ratios, which are a result of excessive mixing of hydrothermal fluid with seawater. Theoretical studies and laboratory experiments have shown that thermodynamic equilibrium between plagioclase and epidote solid solutions in basalt hosted hydrothermal systems control the Na and Ca concentrations of the fluids (Berndt et al., 1989; Seyfried et al., 1991; Berndt and Seyfried, 1993; Seyfried and Ding, 1995). Higher temperature conditions favour the stability of anorthite, clinozoisite of the plagioclase and epidote solid solutions respectively, due to the following reactions:

These reactions result in Ca fixation and decrease in Ca/Cl ratios in hydrothermal systems. The Na/Cl ratios are lower and the Ca/Cl ratios are higher than seawater, indicating that albitization and Ca-release occur in the root zone of E2-S. However, in 2019 the fluids at Dog’s Head had higher Na/Cl and lower Ca/Cl ratios than in 2010 ( Table 4 ). This may indicate a subtle switch to less albitization and more Ca fixation. Likewise, the Li/Cl and Cs/Cl ratios were noticeably lower in 2019 than in 2010. These ratios are proxies for water/rock mass ratios (w/r). Field and experimental data have shown that alkalis are highly fluid mobile and get leached from basalt during fluid-rock interactions at high temperatures (Mottl and Holland, 1978; Seyfried et al., 1984; Von Damm et al., 1985a; Von Damm, 1990). The high mobility results in higher concentrations of dissolved alkalis in the fluids that have interacted with large masses of rock ( Table 3A ). Assuming that Li, Rb and Cs are fully leached from the rocks in the root zone w/r ratios can be calculated (e.g., Von Damm et al., 1985a). The glassy lava from the E2 segment contains ca. 0.15 ppm of Cs, ca. 8 ppm of Rb (Leat et al., 2000). Basaltic andesites commonly contain ca. 8 ppm of Li (Ryan and Langmuir, 1987). Using endmember compositions for Li, Rb and Cs at E2-S and assuming 100% extraction of the elements the w/r ratios were ca. 2 for Cs and ca. 2 for Rb at both Iced Bun and Dog’s head. The w/r ratios of ca. 2 for Iced Bun and ca. 3 for Dog’s Head were calculated using Li. Moreover, the ratios in this study are higher compared to that calculated in 2010 (ca. 1.9) because of lower alkali abundances in these fluids.

Dog’s Head and Iced Bun have very similar vent fluid compositions in terms of Na/Cl, K/Cl, Br/Cl, Si and predicted w/r ratios, indicating that the high-temperature fluids venting at E2-S are fed by a common root zone. Likewise, back in 2010, Dog’s Head and Sepia had identical Na/Cl, K/Cl, Ca/Cl, Li/Cl and Cs/Cl which also points to a single sourced fluid at E2-S. We next discuss how the composition of the Dog’s Head fluids has evolved between 2010 and 2019 and how this variation may indicate changes in the common root zone of the E2-S vents.

Continuous alteration of fresh rock over time can result in the loss of elements from the rock, thereby reducing the amount of fresh rock available for w/r interactions (high Na/Cl ratios). This results in higher w/r ratios and decrease in concentration of mobile elements. Higher Na/Cl ratios coupled with higher w/r ratios in 2019 as compared to 2010 provide evidence that the root zone at E2-S has undergone effective alteration over the duration of the past 9 years. Furthermore, the fluids sampled in 2019 have low δD_H2O and δ¹⁸O_H2O compared to the fluids sampled in 2010 ( Table 3A ). While the low δD_H2O at E2 may be related to magmatic water input, the low δ¹⁸O_H2O cannot be explained by this process (cf. Magmatic water input). The fluid sampled at Dog’s Head in 2010 had an endmember δ¹⁸O_H2O value of 1.1‰ while the fluid from 2019 has a lower value of 0.4‰. Hydrothermal alteration of the igneous crust by circulating seawater-derived fluids at high temperature will have the δ¹⁸O_H2O values of the altered rocks decrease with time (Shanks, 2001 and references therein). The lower δ¹⁸O_H2O values of the 2019 vent fluid may indicate that the rocks in the root zone are more altered, which is consistent with the higher w/r ratios derived from the decreased concentration of fluid mobile elements (Rb, Li).

Experimental work has shown that REE concentrations are also affected by the intensity of hydrothermal alteration: the REE concentrations decrease and the positive Eu anomaly becomes more pronounced as the extent of hydrothermal alteration of mafic rocks increase (Beermann et al., 2017). REE_CN (chondrite normalised) pattern observed in the fluids from E2-S are typical for mature mafic hosted hydrothermal systems ( Figure 5 ). ΣREE measured in this study is lower than 2010 (Cole et al., 2014). This decrease in total REE content would be consistent with an increased extent of rock alteration in the root zone. However, REE solubility in hydrothermal fluids is also controlled by temperature, pH and redox conditions during high-temperature fluid-rock interactions (Craddock et al., 2010). A lower temperature in the root zone could therefore also explain the observed drop in ΣREE.

Our fluid inclusion study provides insights into the temporal evolution of salinity and phase separation processes taking place at E2. The fluid inclusion data at E2-S show that the fluids salinity was either slightly lower than or close to that of seawater ( Supplementary Figure 3C ). Although the time of entrapment of these inclusions was not determined, the low salinity in most of the inclusions in the young top of the chimney sample provide evidence for low Cl venting fluid at E2 during the past years. In addition to the low Cl concentrations measured in the fluids in 2019 and 2010, the fluid inclusion salinities indicate that phase separation has been affecting fluid compositions at E2-S for a prolonged period, despite the apparent changes in the extent of alteration in the root zone.

In summary, the root zone of E2-S may have changed slightly towards a more altered basement and slightly lower temperatures, but it appears that the process of phase separation (supercritical) taking place at E2 is constant and has not changed significantly.

The concentration of dissolved Si in hydrothermal vent fluids can be used to estimate the depth (pressure) and temperature in the hydrothermal root zone (Bischoff and Rosenbauer, 1985; Von Damm et al., 1991; Foustoukos and Seyfried, 2007c). This thermobarometer is calibrated for quartz which is commonly present in hydrothermally altered mafic rocks. Since quartz solubility is a function of temperature and pressure, either one can be estimated if the other one is known. Using this relation, sub-surface conditions of fluid-quartz equilibration at E2-S was calculated ( Figure 6 ). With the calculated endmember Si concentrations and the measured temperature at E2-S, the equilibrium pressure was calculated at ∼300 bar and ∼500 bar at Dog’s Head and Iced Bun, respectively. Silica concentrations at Dog’s Head in this study is slightly lower than in 2010. Silica concentration in the Sepia fluids was high (22.3 mM, James et al., 2014), corresponding to unreasonably high 800-900 bar pressure. James et al. (2014) argued that the root zone temperatures at ‘Sepia’ were probably higher than the measured 353°C and that cooling (but no quartz precipitation) in the upflow zone gave rise to the high apparent pressures. The low pressures (300- 350 bar in 2019) obtained for Dog’s Head may indicate that quartz did precipitate. At Iced Bun, the estimated depth of quartz equilibria is 500 bar. It is highly unlikely that the depth of the hydrothermal root zone for vent orifices that are in such proximity [ca. 60 m of each other, Tyler, (2012)] have such a huge variability. More plausibly, the estimated pressures are an artefact of conductive cooling (James et al., 2014). When conductive cooling takes place, silica precipitation may or may not take place. The shallow pressures obtained for Dog’s Head indicate some quartz precipitation in the upflow zone. In contrast, the high silica concentrations in the Sepia vent fluids in 2010 (James et al., 2014) indicate that quartz did not precipitate upon cooling. The effect of conductive cooling and quartz precipitation kinetics on the Si concentrations of the fluids makes it difficult to obtain a realistic depth of the reaction zone from quartz equilibrium pressures at E2-S.

The high-temperature fluids venting at E2-S have a Cl concentration that is significantly lower than that of seawater. The lowest endmember concentration was 491 mM ( Figure 4E ). No major Cl sink is known to exist in mafic-hosted hydrothermal systems and therefore Cl depletions observed in such systems are typically attributed to phase separation (Von Damm, 1990; Von Damm, 1995; Foustoukos and Seyfried, 2007b). Seafloor pressure of ∼260 bar places the two-phase boundary at 386°C ( Supplementary Figure 2 ; Driesner and Heinrich 2007), while the critical point of seawater is 407°C at 298 bars (Bischoff, 1991). Our investigation shows that the observed variability in salinities and temperatures in hydrothermal fluid samples and fluid inclusions cannot be produced by subcritical phase separation ( Supplementary Figure 3 ). The Cl concentrations lower than seawater in the hydrothermal fluids at E2-S are therefore likely a result of phase separation at supercritical conditions. Subcritically phase separated fluids would produce mixing lines with considerably lower Cl concentrations when mixing with cold seawater. The temperature-salinity plane observed for fluid samples and inclusion data is characteristic for fluids produced by phase separation at pressures of at least over 300 bar. Elevated Br/Cl ratios at E2-S relative to seawater corroborate the idea of phase separation under supercritical conditions (cf. McDermott et al., 2018). Laboratory experiments have shown that Br partitions preferentially into the low salinity phase under supercritical conditions, resulting in increased Br/Cl ratios (Berndt and Seyfried, 1990; Foustoukos and Seyfried, 2007a). Using the Cl normalized partition coefficients and the following formula:

where ( M b r M C l ) v and ( M b r M C l ) b represents the molal concentration of Br/Cl in the low salinity and high salinity phase, respectively (Berndt and Seyfried, 1990; Foustoukos and Seyfried, 2007a). It is known that vapour-brine partition coefficients for Br increases with decreasing fluid density and can be related with the following expression:

where p_w = density of pure seawater (Foustoukos and Seyfried, 2007a).

Endmember Br/Cl ratios at E2-S are 1.54 and 1.58 ( Table 4 ), which are higher relative to seawater value of 1.52. These Br/Cl ratios give values of D_Br/Cl of 1.01 and 1.04. Values of D >1 is known to be a result from the formation of a low salinity phase (Foustoukos and Seyfried, 2007a). Moreover, the calculated D values correspond to densities of ∼0.25 g/ml for the fluid at E2-S. This fluid density allows us to confine the conditions of phase separation to a pressure and temperature range of 320 to 390 bar and 420 to 450°C. The Cl-depleted nature of the fluids from E2-S represents a supercritical phase separated fluid owing to their depleted Cl concentrations relative to seawater and density constraints (Bischoff and Pitzer, 1989; Bischoff, 1991). The cooler temperatures of 320 to 344°C measured at seafloor conditions suggest that the fluids have been undergoing conductive cooling prior to venting.

Phase separation is known to affect the concentrations of dissolved gases in hydrothermal fluids. Cl-depleted fluids have a higher concentration of dissolved gases relative to conjugate brine phases. This is also true for the fluids at E2-S. Fluids with low concentration of Cl show the highest concentration of CO₂ and H₂S ( Figures 4C–E ), while concentrations of dissolved H₂ and CH₄ do not show similar trends. CH₄ abundances likely represent a CH₄-Cl rich fluid (brine phase) mixing with the vapour rich fluid at E2-S.

The pH of the fluid and metal-chloro complexes (under the effect of pressure and temperature) play a key role for the concentrations of dissolved metals in hydrothermal fluids. The temperatures required to have high concentrations of metals stable in seawater-like solutions decrease in the order Cu>Fe>Zn>Mn (Seewald and Seyfried, 1990). It is therefore expected that a cooled hydrothermal fluid will lose Cu and Fe by mineral precipitation faster than Zn and Mn. The fluids at E2-S have a relatively high endmember Mn concentration but low Fe concentrations ( Table 3A ). The low Fe/Mn ratios at E2-S (0.20 to 0.59) may represent the loss of Fe relative to Mn because of extensive conductive cooling of the fluid. To test this hypothesis, we applied the Fe/Mn geothermometer by Pester et al. (2012), derived from an empirical relation between the Fe/Mn ratios of hydrothermal fluids and temperature. The calculated temperatures (e.g., 305°C for Dog’s Head) are lower than the measured temperatures (here: 344°C). The same applies for the Dog’s Head fluids analysed in 2010 ( Table 3A ), which give Fe/Mn temperatures of 308°C (at measured 351°C). Sepia and Sepia Flange show even lower Fe/Mn ratios indicative of even more cooling-related Fe-precipitation there. These overall low Fe/Mn ratios are a clear indication that the fluids venting at E2-S undergo extensive cooling and pyrite precipitation prior to venting at the seafloor ( Figure 8 ). The 40°C cooling of Sepia Flange relative to Sepia vent (James et al., 2014) come with a drop in Fe concentrations of >200 µM. The associated changes in H₂S concentrations (drop by 500 µM) and pH (drop from 3.1 to 2.9) are consistent with the idea of pyrite precipitation.

The low concentrations of dissolved H₂ in the fluids at E2-S (<1 mM) are typical for unsedimented mafic hosted hydrothermal systems (e.g., Welhan and Craig, 1983; Merlivat et al., 1987; Evans et al., 1988; Proskurowski et al., 2008; Diehl and Bach, 2020b; Diehl and Bach, 2021). Dissolved H₂ in hydrothermal fluids are known to be controlled as a result of fluid-mineral equilibria and phase separation processes, where H₂ abundances are buffered as a result of equilibration of Fe-bearing sulphide, oxides and aluminosilicate minerals (Seyfried and Ding, 1995; Seyfried et al., 2003). Phase separation can also affect the concentration of dissolved gases such as H₂. Cl-depleted vapour phases are usually associated with high concentration of dissolved gases. However, phase separation alone at E2-S is likely unable to explain the concentration of H₂ in these fluids.

High-temperature fluid-rock interaction in the sub-surface root zone results in buffering the H₂ concentration of these fluids. Experimental basalt alteration studies have shown that pyrite-pyrrhotite-magnetite (PPM) and hematite-magnetite (HM) assemblages are predicted to exist in the range of redox conditions for natural hydrothermal systems (Seewald and Seyfried, 1990; Seyfried et al., 1991; Seyfried and Ding, 1995). The stability of these buffers can be predicted using thermodynamic data at various sets of temperature and pressure conditions. The known endmember H₂ concentrations for the fluids at Iced Bun and Dog’s Head, predict H₂ concentrations that are consistent with the HM buffer at temperatures of ∼358 and ∼340°C at 300 bar ( Figure 7 ). The higher degree of conductive cooling at Iced Bun relative to Dog’s Head results in the offset of the data from the predicted HM buffer. The measured concentrations of H₂ at Iced Bun correspond to an equilibrium fluid temperature of ∼358°C ( Figure 7 ), while the fluid at Dog’s Head is predicted to have a temperature of ∼340°C. The fluid temperature predicted at Dog’s Head are close to the measured temperatures of 344°C. The higher temperature predicted for fluids emitting at Iced Bun could be a result of addition of H₂ from rapid pyrite precipitation after equilibration with the HM buffer, during fluid upflow. However, we cannot rule out the fact that the fluid venting at Iced Bun underwent a higher degree of conductive cooling compared to Dog’s Head, and therefore the H₂ concentrations could also be the result of higher equilibration temperatures (>320°C).

Endmember concentrations of dissolved H₂S of 3.61 and 5.02 mM in the fluids emitted at E2-S are within the known range of basalt-hosted hydrothermal systems (Diehl and Bach, 2020b; Diehl and Bach, 2021). δ ³⁴ S_H2S values of fluids from E2-S range from +4.0 to +6.7‰. Such δ ³⁴ S_H2S values have been previously reported from the Lau Basin (McDermott et al., 2015a). The H₂S in unsedimented basalt hosted hydrothermal systems can be either a result of mantle derived sulphide present in rocks in the form of pyrite and pyrrhotite with δ ³⁴ S_H2S values of +0.1 ±0.5‰ (Sakai et al., 1984) and/or reduced seawater SO₄, with a δ ³⁴ S_H2S value of +21.0 ±0.2 ‰ (Rees et al., 1978). At E2-S the δ ³⁴ S_H2S values are a result of mantle derived sulphur that varies from 81 and 69% relative to seawater derived sulphur of about 19 and 31% in these fluids.

Conductive cooling of fluids during ascend to the seafloor can result in pyrite precipitation, which can in turn affect the concentration of H₂, H₂S and Fe in the fluid (McDermott et al., 2018). Pyrite precipitation produces H₂ and acidity, while consuming dissolved Fe and H₂S:

Indeed, the fluid venting at Iced Bun appears to be affected by pyrite precipitation as its fluid is characterized by low concentrations of Fe and H₂S, a higher concentration of H_2, and lower pH _(in-situ) relative to Dog’s Head ( Table 1 , 3A ). Fluid venting at Dog’s Head could also be affected by pyrite precipitation in the subsurface (low Fe/Mn ratios), but it appears that the pyrite precipitation at Iced Bun is more pronounced than at Dog’s Head. Activity-activity diagrams for H₂S and H₂ indicate that Dog’s Head and Iced Bun plot near the hematite-magnetite-pyrite invariant point in the Fe-S-O-H system ( Figure 8 ). Deviations from the invariant point can be accounted for by the facts that (1) the natural system is compositionally more complex than the Fe-O-H-S system for which the diagram is representative of and (2) there is considerable uncertainties in concentration-activity relations (Scheuermann et al., 2019). H₂S concentrations at E2-S may also have been influenced by phase separation, because Dog’s Head fluid (low Cl) have higher concentration of H₂S than the high-Cl fluids venting at Iced Bun ( Table 3A ). As shown, the dissolved H₂ and H₂S concentrations at E2-S are mainly a result of fluid-mineral equilibria, but phase separation and cooling-induced pyrite precipitation are additional influences.

Elevated concentrations of dissolved CO₂ relative to seawater in BAB-hosted hydrothermal systems result from magmatic degassing (Takai et al., 2008; Reeves et al., 2011; Seewald et al., 2015; Seewald et al., 2019). However, a small fraction of CO₂ in the fluids could also result from leaching of CO₂ trapped in rocks as inclusions. Basaltic glasses from the ESR have a CO₂ concentration of <200 ppm (Mattey et al., 1984). Assuming a w/r of 2 (calculated for E2-S in this study), complete leaching of CO₂ from the rocks would contribute a CO₂ concentration of no more than 4 mmol/kg CO₂. Therefore, the higher abundances of dissolved CO₂ ( Figure 4C ) at E2-S are primarily a result of magmatic degassing with a minor possible contribution of CO₂ leached from basalts.

The carbon isotopic signatures of fluids from E2-S reveal endmember δ¹³C_CO2 values of -4.1 and -5.6‰, which are within the known range of BAB hosted hydrothermal system (Reeves et al., 2011; Seewald et al., 2015; Diehl and Bach, 2020b; Diehl and Bach, 2021). In contrast, basaltic glass at ESR has more negative δ¹³C_CO2 values of -15.0 to -19.9‰ (Mattey et al., 1984). Degassed CO₂ from basaltic melts at MOR have more positive δ¹³C_CO2 values, such as those at Lucky Strike hydrothermal system (-3.9 to -5.6‰) that are interpreted to reflect replenishing of the magmatic reservoir (Javoy et al., 1978; Mattey, 1991; Pester et al., 2012; Holloway and Blank, 1994). The high abundance and ¹³C-enriched nature of CO₂ in the fluids at E2-S relative to the basaltic glass at ESR are likely a result of degassing from a magma reservoir that has undergone replenishment.

The endmember CH₄ concentrations (25.6 and 48.9 μM) are low relative to other BAB-hosted hydrothermal systems (Reeves et al., 2011; Seewald et al., 2015; Seewald et al., 2019). A variety of processes can contribute towards the CH₄ concentrations in hydrothermal systems that can range from abiotic sources, microbial activity and thermogenesis of sediments and/or organic matter (Von Damm et al., 1985b; Welhan, 1988; Seewald and Seyfried, 1990; Seewald et al., 1994; Von Damm et al., 2005; Cruse and Seewald, 2006; Proskurowski et al., 2008; McDermott et al., 2018; Fiebig et al., 2019). Low concentrations of CH₄ at E2-S coupled to a relatively ¹³C-enriched δ¹³C_CH4 isotopic composition of -7.0 and -7.6‰ rule out the influences of sediment thermogenesis and microbial derived CH₄ which would be expected to impart more ¹³C-depleted signatures. Such ¹³C-enriched δ¹³C_CH4 in combination with ²H-enriched δD_CH4 values (here: -98.7 and -99.9‰) have previously been associated to a ‘potentially abiotic source’ of origin (Kelley, 1996; Kelley and Früh-Green, 1999; Wang et al., 2018; McDermott et al., 2018; Klein et al., 2019). However, recent studies indicate that similar values can also be produced as a result of thermogenesis of seawater derived DOM – ‘volcanic thermogenesis’ (Fiebig et al., 2019; Reeves and Fiebig, 2020). As a result, the CH₄ abundances at E2-S could be a combination of volcanic thermogenesis with a possible contribution from an abiotic source.

An inverse relationship in CO₂ and CH₄ concentrations is observed in the fluids at E2-S such that high concentrations of CO₂ are matched with low CH₄ abundances. High CO₂ abundances are found in the most Cl-depleted fluid, while CH₄ concentrations are highest in the fluid containing with Cl concentrations. This relationship could result from mixing of CH₄- and Cl-bearing hydrothermal fluids and a CO₂-enriched and Cl-poor magmatic vapour in the sub-surface of the E2-S vent site.

Stable oxygen and hydrogen isotopes in water in hydrothermal systems are mainly affected by three factors: i) interaction between the hydrothermal fluid and the host rock and/or sediments in the sub-surface, ii) phase separation and/or iii) contribution of mantle derived magmatic water. Hydration reactions between the hydrothermal fluids and the oceanic igneous crust yields increasing δD_H2O and δ¹⁸O_H2O values with decreasing w/r ratios ( Supplementary Figure 4 , Bowers and Taylor, 1985; Bowers, 1989; Shanks et al., 1995; Shanks, 2001). The δD_H2O values in fluids from E2-S are depleted in ²H relative to seawater (-2.0 to -1.2‰; Supplementary Figure 4 ). These low δD_H2O values are an indication that factors other than w/r interactions affect the H isotopic composition of these fluids.

Experimental work has shown that phase separation can affect the H and O isotope ratios of hydrothermal fluids (Horita et al., 1995; Berndt et al., 1996; Shmulovich et al., 1999; Foustoukos and Seyfried, 2007b). Vapours are slightly elevated in δD_H2O values relative to the brine phase (Von Damm et al., 2003). Oxygen isotopes show the opposite fractionation direction: brines have higher δ¹⁸O_H2O values than vapours (Berndt et al., 1996; Foustoukos and Seyfried, 2007b). The low chlorinity fluids analysed in this study show low δD_H2O and high δ¹⁸O_H2O values (0.4 to 0.6‰), which is opposite of what would be expected for a vapour phase produced by phase separation. An influence of sediments on the fluid chemistry can also explain low δD_H2O combined with high δ¹⁸O_H2O values (Baumberger et al., 2016; Toki et al., 2016). No sediment was observed at E2-S, and therefore sediment-fluid interactions are likely not responsible for the observed δD_H2O and δ¹⁸O_H2O data.

The negative δD_H2O (-2.0 to -1.2‰) and positive δ¹⁸O_H2O values (+0.4 to +0.6‰) in the E2-S hydrothermal vent fluids may indicate that a small fraction of the venting H₂O is derived from magmatic degassing. A constricted range of δD_H2O (-65.0 ± 20‰) and δ¹⁸O_H2O (+6 ± 1‰, Taylor, 1979; Ohmoto, 1986) has been estimated to be of mantle derived magmatic water. Whereas water resulting from subduction related volcanic vapours (SRVV) have δD_H2O values ranging from -10 to -30‰ and δ¹⁸O_H2O of +6 to +10‰ (Giggenbach, 1992; Hedenquist and Lowenstern, 1994). Fractionation related to degassing and slab-derived input of water can result in an increase in δD_H2O of magmatic waters at convergent margins (Taylor, 1986; Giggenbach, 1992; Hedenquist and Lowenstern, 1994; Taylor, 1997). Linear extrapolation of the fluids at E2-S indicate a possible small input of SRVV component into the fluids ( Figure 9 ). Such an SRVV component input has been previously suggested to affect the PACMANUS hydrothermal vent fluids, Manus spreading centre, Papua New Guinea (Reeves et al., 2011), which show CO₂, δ¹³C_CO2, δD_H2O and δ¹⁸O_H2O characteristics similar to those found at E2-S. The most plausible explanation for the δD_H2O and δ¹⁸O_H2O data of the fluids venting at E2-S is that seawater-derived hydrothermal fluids underwent extensive water/rock interactions and mixing with small fractions of magmatic water vapour with a SRVV isotopic signature. Moreover, it appears that the fluid in 2010 did not have any involvement of an SRVV component as no highly negative δD_H2O values were reported ( Table 3A ). These variations seen at E2 in 2019 further provide a temporal evolution into the ESR systems.

The fluids diffusively venting at E2-W are low in temperature (53°C) relative to those emitted at E2-S (320°C, 344°C). Such diffuse low- to moderate-temperature fluids can be a result of either subseafloor mixing of high-temperature fluid with entrained seawater or they can represent conductively heated seawater (Cooper et al., 2000; McDermott et al., 2015b). The Mg depletion (42.6 mM) and K enrichment (13.4 mM) relative to seawater are a clear indication that the fluid at E2-W is a result of a high-temperature fluid that has undergone mixing with seawater prior to venting.

The E2-W hydrothermal system is located about ca. 1.25 km north of the E2-S system. and the two vents investigated at E2-S, Dog’s Head and Iced Bun, are fed by a single source fluid. E2-W is proximal enough to E2-S to be fed by fluids from the same root zone. To investigate the plausibility of the idea that a single deeply sourced fluid supplies all the E2 systems sampled in this study, an average composition of the endmember fluid (EF) for E2-S was computed and mixed with seawater (SW) until the Mg measured for the fluid at E2-W, was met ( Supplementary Table 2 ). The Mg concentrations of the E2-W fluid reveal an EF: SW mixing ratio of 1:5. Enrichments and depletions of dissolved species in this hypothetical E2-W fluid (E2-W_calc) were then compared with the measured E2-W fluid ( Supplementary Table 2 ). The excellent correspondence of calculated and measured data indicate that the E2-W fluid could indeed result from mixing of E2-S hydrothermal fluid with seawater. Na, Cl, Br, K, Ca and CO₂ values estimated are either within the analytical errors of the measurements (K, Ca and CO₂) or have seawater-like concentrations (Na, Cl and Br) indicating the fluid may derive from mixing between seawater and the common source fluid that feeds the E2-S vent sites.

Whereas most elements in the fluid at E2-W fit the conservative mixing model, enrichments (SO₄ and Sr) and depletions (H₂, CH₄, H₂S, Si and Fe) relative to the E2-W_calc are observed. Si in hydrothermal systems is believed to behave conservatively and has been extensively used to constrain sub-surface conditions in these systems (Von Damm et al., 1991; Reeves et al., 2011; McDermott et al., 2018). However, depletions in Si relative to calculated E2-W_calc observed is likely a result of sub-surface precipitation of silica. Janecky and Seyfried (1984) have shown that removal of silica through precipitation is only possible with the fluids that have undergone extensive cooling, which is also likely the case for the fluid at E2-W. Reaction path models further indicate this loss of Si in the modelled fluid because of quartz precipitation ( Supplementary Figure 5 ).

The fluids at E2-W are enriched in SO₄ (by 2.27 mM) and Sr (by 1.98 μM) relative to the conservative mixing line. Sr in hydrothermal systems is known to substitute Ca in anhydrite precipitation (Kuhn et al., 2003). Therefore, the observed enrichments in Sr and SO₄ can be directly attributed to sub-seafloor dissolution of anhydrite. Endmember extrapolation of SO₄ at E2-W reveals a concentration of 13 mM. These strong SO₄ excesses have commonly been interpreted as evidence of anhydrite dissolution (e.g., Reeves et al., 2011).

Aerobic sulphide oxidation (equation 1) is another possible source of excess sulphate. Indeed, thermodynamic computations indicate that the major source of catabolic energy in the E2-W subseafloor is aerobic sulphide oxidation ( Figure 10 ). Furthermore, non-sedimented hydrothermal systems are known to host a variety of sulphide oxidizing microorganisms (Dahle et al., 2015; Dahle et al., 2018). Microbial sulphide oxidation could also further help explain the depletion of H₂S in the fluid (<0.2 μM, below limit of quantification) compared to the calculated 0.81 mM abundance that would be expected if conservative mixing were occurring. However, even if 0.8 mM of H₂S were oxidized to sulphate, the sulphate enrichments of 2.27 mM cannot be fully explained by this process. Moreover, abiotic precipitation of pyrite is another possible sink for H₂S. Indeed, depletion of H₂S is matched by a loss of >90% of the dissolved Fe, which could point to pyrite precipitation. While Fe is likely only affected by abiotic reactions, SO₄ and H₂S concentrations at E2-W could be a result of both biotic as well as abiotic reactions taking place in the sub-surface. We suggest that most of the sulphate excess is due to anhydrite dissolution, whereas H₂S is lost as consequence of pyrite precipitation as well as aerobic sulphide oxidation.

A strong depletion of H₂ relative to the conservative mixing line is also observed in the fluids at E2-W ( Supplementary Table 2 ). H₂ in mixing zones in hydrothermal systems can be consumed by the reduction of CO₂ to formic species (McDermott et al., 2015b). If CH₄ was a metabolic product of CO₂ reduction, CH₄ concentrations in the fluids should be higher than predicted from conservative mixing. This is not the case, as the fluid at E2-W is depleted in CH₄ relative to the mixing line by 5.27 μM ( Supplementary Table 2 ). Such a depletion of CH₄ at E2-W is likely the result of aerobic oxidation of CH₄ ( Figure 10 ). CO₂ and H₂ concentrations of the fluids are inconclusive in this regard, as a H₂ consumption of 8 µM would deplete CO₂ only by 2 µM, which is small relative to the 300 µM mismatch between measured and calculated composition ( Supplementary Table 2 ). Moreover, CO₂ could be produced by oxidation of CH₄ by microorganisms. δ¹³C_CO2 at E2-W is -10.8‰ and is more ¹³C-depleted than the measured δ¹³CO₂ values at E2-S of -3.7‰ and -4.4‰. If the CO₂ at E2-S is a result of magmatic degassing and the entire E2 system is fed by a common root zone, then a range of δ₁₃ C_CO2 between -4.4‰ to 0.3‰ would be predicted at E2-W if it simply reflects mixing between the E2-S endmember and seawater. The far more ¹³C-depleted values of δ¹³C_CO2 at E2-W suggest an additional, non-magmatic source of CO₂ in the vent fluids. Such more negative values of δ¹³C_CO2 relative to the high-temperature fluid have been previously observed at EPR and have been attributed to methanotrophy (Proskurowski et al., 2008). The idea that methane oxidation may contribute to the dissolved CO₂ pool is further supported by catabolic landscape models, which indicate that this aerobic methane oxidation is indeed exergonic at E2-W ( Figure 10 ). It is likely that the δ¹³C_CO2 at E2-W represents a combination of the endmember fluid mixed with seawater with contributions from microbial metabolism.

A more general assessment of the catabolic energy landscape at E2 is warranted as endmember fluids at E2-S are enriched in H₂S, CH₄, CO₂, H₂, Fe and Mn relative to seawater. The enrichments of these components are a result of high-temperature fluid-rock interaction and magmatic degassing taking place in the sub-surface. A combination of these processes creates an energy gradient representing an appropriate environment for life in the deep-sea at E2 to thrive on.

Just like we established for the E2-W site, this reduced fluid can mix with the oxic seawater entrained in the subseafloor and create an energy rich environment for catabolic reactions to take place ( Figure 10A , McCollom and Shock, 1997; Amend et al., 2011). These reactions ( Table 2 ) help support a chemosynthetic ecosystem at E2. The catabolic energy computations we conducted indicate that aerobic sulphide oxidation (ASO) is clearly the dominant energy source at both low and high temperatures ( Figure 10B ). Between 120 and 60°C (i.e., at low SW: HF mixing ratios) ASO has an energy yield between -0.2 and -0.4 kJ/kg of vent fluid. At temperatures <60°C (i.e., high SW: HF mixing ratios), the models predict a higher energy yield as high as -2.6 kJ/kg of vent fluid.

Aerobic methane oxidation (AMO) is the next most abundant energy source predicted at E2 ( Figure 10B ). At high temperatures (120 to 60°C), AMeO has an energy yield of ∼ -0.03 kJ/kg of vent fluid, and the energy yield may approach -0.8 kJ/kg for vent fluid at lower temperature. Aerobic iron oxidation (AFeO) and aerobic manganese oxidation (AMnO) account for the lowest energy availability at E2.

The modelling results presented are valid for conservative mixing of hydrothermal fluids with seawater. Conductive cooling, mixing with entrained seawater and related abiotic processes at E2 can greatly affect the composition of the hydrothermal fluids. Studies have shown that abiotic reactions can indeed reduce the energy available for catabolic reactions (McDermott et al., 2020).

The effect of mixing on the energy availability is evident at E2-W, where the fluids have been extensively mixed with BSW prior to venting and have a SW: HF mixing ratio of 5:1. Although the H₂S in the fluid was below <0.8 mM, large microbial mats were observed that covered the flange. In contrast, such features were not evident in the E2-S system. Thermodynamic models predict that these H₂S-depleted and O₂-enriched fluid-seawater mixtures coincide with peak energy availabity to chemosynthetic microbial communities ( Figures 10A, B ).

ASO and AMO are the dominant energy source for all the E2 vents. Large areas of the E2-S have been known to be inhabited by biota that are dependent on mainly ASO for their metabolic pathways (Rogers et al., 2012). The temperature range, extensive mixing with BSW, fluid chemistry and predicted thermodynamic modelling indicate that the conditions at E2-W are suitable to host a robust sub-surface biosphere.