The Lucky Strike Hydrothermal Field (LSHF) is located ∼400 km to the Southwest of the Azores archipelago on the Mid-Atlantic Ridge at 37^o17′N and 32^o20′W (Figure 1A; Langmuir et al., 1997; Von Damm et al., 1998). This 1 km² hydrothermal vent field consists of 20–30 active vents distributed around a fossil lava lake (apart from Capelinhos vent) surrounded by three ancient volcanic cones (Fouquet et al., 1995; Langmuir et al., 1997; Von Damm et al., 1998; Charlou et al., 2000; Ondréas et al., 2009; Escartin et al., 2015). Figure 1B presents the LSHF bathymetric map [∼1550–1750 m below sea level (mbsl)] with 12 active hydrothermal sites spatially located to the North East (Sintra and Y3), the South East (Aisics, Tour Eiffel, and Isabel), the North West (Jason, and Elisabeth), and the South West (Helene, Nuno, Crystal, South Crystal, Sapins, White Castle, and Cypres) of the fossil lava lake.

During the MoMARsat’19 EMSO-Azores maintenance cruise on-board the R.V. Pourquoi Pas? (June–July 2019; Sarradin and Legrand, 2019), seawater samples were collected at six selected CTD casts. The selected casts cover the LSHF situated to the North (coIN), West (Vortex-W2 and -W4), East (Vortex-E1 and -E3), and at the center (Vortex-centre) of the fossil lava lake (Figure 1B).

Vertical CTD casts used a SeaBird SBE-25 that sampled pressure, temperature, and conductivity at 8 Hz. The mean downcast and upcast CTD velocity was 0.7 m/s. Seawater samples were collected during the upcast between 1722 and 1478 mbsl using a 8 L PVC Ocean Test Equipment (OTE) externally closing bottles and mounted on the CTD carousel. The CTD was stopped for 30 s before firing the bottles to limit the impact of the carousel wake on the samples. Upon recovery, the samples were transferred into a dedicated chemical lab for fluid extraction. The samples were filtered into 250 ml low-density polyethylene bottles (LDPE, Nalgene) washed with diluted hydrochloric acid and rinsed with ultrapure milli-Q water before use. Each subsample was kept in cold storage (∼4°C) until its analysis. In total, 30 samples were analyzed for dissolved cations (DCa, DK, DMg, DNa), anions (DCl, DBr, DSO₄), and dissolved Li (DLi) and Sr (DSr) elemental and isotopic compositions.

CTD data were processed with the Seabird data processing software (filtering, temperature-conductivity alignment, thermal mass correction, data editing). The carousel downcast was stopped 15 m above the seafloor. The carousel was equipped with two 300 kHz RDI LADCP data, one up- and one down-looker. Velocity profiles with a 10-m vertical resolution were produced from the combination of the two LADCPs using the LDEO software (version IX.12).

All the chemical analyses were conducted at the Observatoire Midi-Pyrénées.

The analytical methods used for major dissolved cations (DCa, DK, DMg, DNa), and anions (DCl, DBr, DSO₄) analyses on seawater are detailed in Besson et al. (2014), Chavagnac et al. (2018a), and Leleu (2017) and will be briefly described later. Seawater samples were diluted at 30 fold for cation and anion concentration measurements. Cation concentrations were determined with an inductively coupled plasma atomic emission spectrometer (ICP-AES) Horiba Ultima2 instrument, with an analytical precision better than 2%. Cation concentrations were quantified by the calibration curve method with an IAPSO seawater standard solution (OSIL Ltd., United Kingdom) diluted at 10–100 fold with Milli-Q water. The analytical drift correction was assessed by the standard bracketing method every eight samples. Detection limits were determined by daily repeated measurements of blank solutions (n = 10) at 0.001 mmol/L for DCa, 0.002 mmol/L for both DK and DMg, and 0.1 mmol/L for DNa. Anion concentrations were determined by anionic chromatography (Dionex ICS-2000) equipped with a specific column for a highly charged matrix (DIONEX IC AS19), calibrated with IAPSO seawater standard diluted at 10–50 fold with Milli-Q water, and with an analytical precision better than 2%. The cations and anions concentrations are reported in Supplementary Table S1 with their respective uncertainties (1SD) calculated by error propagation.

For DLi, and DSr measurements, seawater samples were doped with a prepared DLi and DSr mono-elemental solution. The dissolved Li and DSr concentrations were then measured with ICP-AES and quantified by the method of standard addition with a doped IAPSO standard ranging from 24.6 µmol/L to 386.2 µmol/L for DLi, and 88.0–137.4 µmol/L for DSr. The analytical drift correction, as well as the detection limit (0.2 µmol/L for DLi and 0.002 µmol/L for DSr, n = 10), was assessed as for cations. The dissolved Li and DSr measurements are reported in Table 1 with their respective uncertainties (1SD) calculated by error propagation (including instrumental precisions, and mono elemental concentration uncertainties).

For isotopic measurements, all seawater samples were treated in a clean laboratory to isolate the two elements of interest from the matrix using conventional liquid chromatography. For each of them, 1 ml of the seawater sample was evaporated to dryness in a Savillex beaker on a hot plate at 70°C. Additionally, international seawater standards (IAPSO and NASS 6) were processed in the same manner alongside our samples.

Sr was isolated using Sr-Spec resin (Eichrom, United States) using the protocol described in Pin et al. (2014). Sr isotopic composition of individual fluid samples was measured on a Thermo Fisher Triton + Thermal Ionisation Mass Spectrometer at the Observatoire Midi-Pyrénées. The ⁸⁷Sr/⁸⁶Sr ratio was determined as the average values of 200 measurements of ion intensities following a static multi-collection mode. The ⁸⁷Sr/⁸⁶Sr ratios were corrected from mass fractionation based on the normalization value of ⁸⁶Sr/⁸⁸Sr ratio at 0.1194. The repeated measurements of the NBS 987 Sr standard gave a mean ratio of 0.710259 ± 0.000013 (2 SD; n = 24; 2SE = 0.000003) for a recommended value of 0.710250. The accuracy of our technique is verified against the ⁸⁷Sr/⁸⁶Sr ratios of international standards: 1. IAPSO seawater with a measured value of 0.709174 ± 0.000003 (2SD; n = 4), in line with the published values of 0.709179 ± 0.000007 (2SD; n = 7; El Meknassi et al., 2020), and 2. NASS-6 seawater with a measured value of 0.709174 ± 0.000005 (2SD; n = 3), in line with the published values of 0.709179 ± 0.000014 (2SD; n = 8; Neymark et al., 2014).

Li elution and isolation from the NaCl-rich matrix was carried out using two steps ion exchange columns made of AGW-X12 200–400 mesh cation resin bed and eluted with 1N HCl [Protocol adapted to NaCl-solution from James and Palmer (2000)]. Li isotopic composition of individual fluid samples was measured on a Thermo Fisher Triton + Thermal Ionisation Mass Spectrometer at the Observatoire Midi-Pyrénées. The Li samples were diluted with a freshly made 0.15 N H₃PO₄. 1 µL of the resulting solution (∼30–60 ng of Li) was deposited on 1 µL of 0.15 N H₃PO₄ previously loaded on one side of a degassed double Re filament. After loading, the sample dried down gently at filament current at 0.8 A, then heated up to 2.4 A until acid phosphoric fumes were driven off, and then flashed at 2.8 A. During isotopic measurement, the current of the ionization filament was raised to ∼2 A to reach a pyrometer temperature of ∼1210–1220°C whereas the evaporation filament was increased to ∼0.7–0.9 A until a stable ⁷Li beam of ∼5V was reached. The ⁷Li/⁶Li ratio was determined as the average values of 200 measurements of ion intensities following a static multi-collection mode. The ⁷Li/⁶Li ratios are expressed in a δ⁷Li‰ relative to the IRMM-16 Li standard (Li₂CO₃) at a similar Li concentration. The repeated measurements of the IRMM-16 standard gave a ⁷Li/⁶Li ratio mean of 12.08709 ± 0.01525 (2SD; n = 33), an internal precision of 0.22‰ (2SE, uncertainty reported on the error bars in Figure 2), and an external precision of 1.26‰ (2SD; n = 33). The accuracy of our technique is verified against the measured ratios of international IAPSO seawater standard with a measured value of +29.5 ± 0.2‰ (2SE), in line with the published values of +30.8 ± 0.1‰ (2SE, with external precision ≤1‰; Rosner et al. (2007)).

The depth profiles of DSr (Figure 2A) are distinct from one cast to another as evidenced by 1) constant DSr at seawater value throughout coIN cast, 2) lower DSr concentrations than reference seawater on eastern casts (Vortex-E1 and -E3), and 3) higher DSr concentrations than seawater on western casts (Vortex-W2 and -W4). Regarding the coIN cast, its DSr values are close to the reference seawater value (dashed red line in Figure 2A) and its ⁸⁷Sr/⁸⁶Sr ratios fall within the range of oligotrophic oceanic waters (red box in Figure 2B), evidencing no detection of hydrothermal contribution at this location. Figure 3 illustrates a 3D vision of all datasets at the LSHF scale. Figure 3A dedicated to DSr includes a 91 µmol/L isosurface representative of DSr reference seawater value (Millero et al., 2008), pointing out the opposite trends of DSr variability between the eastern and western casts. Indeed, the LSHF hosts about ∼30 active hydrothermal vents distributed over its 1 km² surface area. At this location, Chavagnac et al. (2018a) and Leleu (2017) showed that high-temperature hydrothermal fluids contain DSr concentrations that are either lower or higher than seawater depending on the vent. In particular, the vents located on the western and north-eastern sides of the LSHF contain 94–179 µmol/L of DSr and all the south-eastern vents display 74–81 µmol/L of DSr (light blue boxes in Figure 2A, and Supplementary Table S2; Von Damm et al., 1998; Charlou et al., 2000; Pester et al., 2012; Chavagnac et al., 2018a). These features are in line with the DSr depletions observed for south-eastern casts at depth and the gradual increase toward the reference seawater value that is reached at 1480 mbsl. They are also in line with the DSr enrichments observed for western casts at depth and the gradual decrease toward the seawater value also reached at 1480 mbsl (Figure 2A). These trends reflect the progressive decreasing contribution of hydrothermal fluid to seawater and the lighter ventilation by hydrothermal plumes of distant water above the seafloor. Clearly, the DSr dataset shows the hydrothermal contribution to the water column, considering the specific chemical characteristics of each vent, which can be tracked up to ∼300 m above the seafloor. This hydrothermal impact cannot be so easily detected from ⁸⁷Sr/⁸⁶Sr ratios of all datasets, as illustrated by the data profiles (Figure 2B) and the 3D representation (Figure 3B). The highest ⁸⁷Sr/⁸⁶Sr ratio difference is obtained at 1630 mbsl between coIN/Vortex-centre and Vortex-E1 with the values of 0.709182 ± 0.000004 and 0.709166 ± 0.000005, respectively (Table 1). Even if this ratio difference is significant compared to the internal precision (±2SE = 0.000003), the mean value of all casts is 0.709174 ± 0.000010 (±2SD; n = 30), which falls within the range of oligotrophic oceanic waters at 0.709172 ± 0.000023 (±2SD, n = 84; global scale; El Meknassi et al., 2020). Thus, the Sr isotopic system does not allow the detection of hydrothermal plume input in the water column at the LSHF km-scale. However, this does not mean that the hydrothermal input in the water column could not be detected with the Sr isotopic system at a smaller scale (e.g., m-scale). Only additional studies at an m-scale could answer this question.

The DLi dataset is represented as depth profiles (Figure 2C) and as a 3D vision (Figure 3C), including the seawater DLi isosurface (24.6 µmol/L). The depth profiles of DLi present the following features: 1) higher DLi concentrations than seawater at all casts and depths apart from Vortex-centre, 2) no clear DLi differences between the eastern and western casts, and 3) DLi concentrations increase at depth at all casts except coIN and Vortex-W2. Among all the CTD casts performed at the LSHF, Vortex-centre is the only one displaying DLi concentrations close to the seawater value (Figure 2C), whereas all the others are enriched by 23–133%. High-temperature hydrothermal fluids are 10–20 times more enriched in DLi (287–398 µmol/L) than seawater without any distinction between the eastern and western sites (light blue box in Figure 2C; Supplementary Table S2; Leleu, 2017; Chavagnac et al., 2018a). The DLi enrichment seen in CTD casts can result from a hydrothermal contribution to the water column by up to 10% considering a DLi conservative behavior. A hydrothermal source of DLi to the water column can be further sustained, especially at Vortex-E3 and Vortex-W4, by the sharp DLi increase at depth up to 10 and 21 µmol/L, respectively (deeper than 1600 mbsl, Figure 2C). Then, as for DSr, hydrothermal contribution to the water column can be detected by measuring DLi concentrations of CTD casts on the LSHF km-scale and up to ∼300 m above the seafloor.

Previous δ⁷Li values measured on different water masses collected in Pacific and Atlantic oceans range between +29.3 and +32.5‰ (Figure 2D; James and Palmer, 2000; Nishio and Nakai, 2002, among others). Moreover, the signatures of the different water masses in the Atlantic Ocean are not precisely defined contrarily to other isotopic systems such as neodymium isotopes (Lacan and Jeandel, 2005; Hartman, 2015). Our δ⁷Li dataset will then be compared to that of the IAPSO seawater standard supplied and prepared from natural mid-Atlantic surface waters, and therefore, outside any hydrothermal influence (Bacon et al., 2007). Our measured δ⁷Li value of IAPSO is +29.5 ± 0.2‰ (±2SE). Taking this value as a reference for seawater, δ⁷Li values of all CTD casts are lower, equal, and higher, i.e., from +25.9 to +36.1‰ (Table 1; Figures 2D, 3D). Regarding LSHF hydrothermal fluids, δ⁷Li values range from +4.4 to +6.4‰ with a mean value of 5.4 ± 0.9‰ (2SD, n = 26; Leleu, 2017, and references therein). As a result, any hydrothermal input to the water column at LSHF should be evidenced by δ⁷Li values lower than that of seawater, if no additional fractionation processes are at play.

Figure 4A presents the δ⁷Li values as a function of 1/DLi*100 to scrutinize and discriminate between hydrothermal contribution to seawater and isotopic fractionation processes. The mixing between hydrothermal fluid and seawater involves physicochemical gradients that lead to mineral precipitation (sulfate-bearing minerals, polymetallic sulfides, iron oxyhydroxides, and manganese oxides) which can be further dissolved and reprecipitated in the buoyant to the non-buoyant hydrothermal plume (Chavagnac et al., 2018b; Lee et al., 2021). Such mineral—seawater interactions (mineral precipitation and dissolution, element adsorption) can induce Li isotope fractionation, i.e., the preferential adsorption of ⁶Li over ⁷Li on mineral (Decitre et al., 2004; Chan and Hein, 2007; Vigier et al., 2008; Araoka et al., 2016; Wilckens et al., 2019). The δ⁷Li and DLi datasets of all casts do not follow the conservative mixing curve between hydrothermal fluid and seawater end-members (Figures 4A,B). However, the DLi versus DSr concentrations define a linear trend (Supplementary Figure S1), suggesting that conservative mixing occurs between these two end-members. Moreover, high DLi concentration (discussed above) necessarily involves a DLi source that at the LSHF location is most likely of hydrothermal origin. Therefore, we can calculate the hydrothermal fluid contribution to the water column for all samples considering a DLi conservative behavior (fHF column in Table 1). Figures 4C,D presents the δ⁷Li values as a function of the fraction of hydrothermal fluid (fHF expressed in %). The difference between the δ⁷Li signatures of our samples and the mixing curve becomes larger as fHF increases (Figures 4C,D). In other words, the δ⁷Li signatures of our samples become heavier as the hydrothermal fluid contribution increases. Thus, the main effects impacting the LSHF water column are DLi input of hydrothermal fluid leading to heavy δ⁷Li values. We hypothesize that these positive deviations from conservative mixing are due to mineral—seawater interactions. This has already been evidenced in buoyant hydrothermal fluids along the mixing gradient where anhydrite/baryte precipitation and dissolution disrupt dissolved rare-earth element concentrations and neodymium isotopic compositions (Chavagnac et al., 2018b). Regarding hydrothermal lithium, Chan and Hein (2007) showed that the hydrated form of Li (LiOH) can be adsorbed preferentially on negatively charged surfaces of Mn oxides contrarily to positively charged surfaces of Fe oxy-hydroxides. The preferential adsorption of ⁶Li over ⁷Li on hydrothermal minerals is evidenced in our dataset by the heaviest δ⁷Li measured on water samples with the highest hydrothermal contribution. However, Chan and Hein (2007) showed that Li adsorbed on Mn oxides surface continuously exchanges its Li isotopes with those of seawater, losing its original hydrothermal δ⁷Li signature. This progressive re-equilibration toward δ⁷Li seawater signature is observed in our samples with decreasing hydrothermal contribution (Figures 4C,D).

At the km-scale LSHF, hydrothermal input can be tracked in the water column using Sr and Li elemental concentrations up to ∼300 m above the seafloor. At depth, the variations of DSr and DLi depict the influence of hydrothermal fluids with their specific chemical characteristics. Further processes such as hydrothermal contribution and mineral-seawater interactions can be detected by δ⁷Li signature once combined with DLi concentrations. We anticipate that the closer the CTD cast is to a hydrothermal source, the stronger the influence of hydrothermal inputs on the water samples. This explanation is overall in agreement with our dataset. Nevertheless and surprisingly, some geographically distant CTD casts with different element profiles display identical DSr and DLi concentrations or similar δ⁷Li signatures at the same depth, e.g., in particular at 1630 mbsl (Figures 2A,C,D). Moreover, the coIN cast located at the northern edge of the fossil lava lake and outside the immediate influence of LSHF show high DLi concentrations (between 39 and 49 µmol/L; Figure 2C) and lower δ⁷Li signatures compared with seawater (between 26.6 and 28.6‰, Figure 2D). Then, although coIN could be considered the most representative of seawater composition based on its DSr concentrations and ⁸⁷Sr/⁸⁶Sr ratios (Table 1; Figures 2A,B), its Li signatures suggest hydrothermal contribution present up to 1480 mbsl. The Sr and Li results contradict each other, therefore questioning the eventual occurrence of an active hydrothermal vent in the vicinity of the coIN cast. The only available Sr and Li data for the LSHF northwestern area are for the Jason (in 1996, Von Damm et al., 1998) and Elisabeth (in 2008, Pester et al., 2012) sites prior to their sampling exclusion as part of the Lucky Strike Marine Protected area (Mullineaux et al., 1998; Leleu, 2017). Without further investigation in this area, the occurrence of an active hydrothermal site remains an open question. At LSHF, hydrothermal vents are distributed around a fossil lava lake, which is surrounded by three ancient volcanic cones (Ondréas et al., 2009). This peculiar topography can constrain the current dynamics in the lava lake and produce specific dispersion patterns. Moreover, the injection of high-temperature hydrothermal fluids (up to 360°C) in a cold seawater mass (∼4°C) may contribute to the formation of currents at depth (Dutay et al., 2004).

Figure 5 presents the spatial variations in the direction and intensity of oceanic currents at the six CTD casts as measured by the LADCPs (current intensity values available in Supplementary Table S3). Currents vary from one cast to the other, are stronger at the LSHF western side (up to 0.25 m/s at Vortex-W4), and show variable direction from the bottom up to 1450 mbsl. Western casts (Vortex-W2 and -W4) display their maximum current speed intensities up to 0.17 m/s and 0.25 m/s, respectively, between 1560 and 1650 mbsl with a general North East direction (Figure 5; Supplementary Figure S2). On the contrary, eastern casts as well as Vortex-Centre display their maximum current intensities (from 0.07 to 0.17 m/s) at 1450 mbsl toward the SW direction. Given the small size of the lava lake (1 km²) and the intensity of the measured currents, the hydrothermal signature of the western part of the lake can easily be advected toward the center and eastern parts in less than an hour (for a current intensity of 0.25 m/s). This may explain why CTD casts distanced from one another by less than a km can exhibit similar and/or identical Sr and Li elemental and isotopic signatures.

Resing et al. (2015) showed that the dispersion of hydrothermal input can be tracked in the water column up to 4,000 km away from its source using dFe concentrations, i.e., Fe has a short residence time in the ocean (200–500 yr; Millero, 1996; Johnson et al., 1997). While Sr and Li have a much longer residence time than Fe, i.e., 2–3 Myr (Hodell et al., 1990; Millero, 1996), we show that hydrothermal input can be detected with these tracers at a km-scale investigation. This suggests that chemical tracers of hydrothermal input have to be chosen depending on the spatial scale of the studied area.