Submarine groundwater discharge derived strontium from the Bengal Basin traced in Bay of Bengal water samples

Evaluating the submarine groundwater discharge (SGD) derived strontium (Sr) flux from the Bengal Basin to the Bay of Bengal (BoB) and determining its isotopic composition is crucial for understanding the marine Sr isotopic evolution over time. Measurements of spatially and temporally distributed water samples collected from the BoB show radiogenic 87Sr/86Sr, high Sr, calcium (Ca) concentrations and high salinity in samples collected dominantly from 100–120 m depth, which can be explained only by the contribution of saline groundwater from the Bengal Basin. These results provide a direct evidence of the SGD-Sr flux to the BoB. This SGD-Sr flux is however, spatially heterogeneous and using conservative hydrological estimates of the SGD flux to the BoB, we suggest a SGD Sr flux of 13.5–40.5 × 105 mol/yr to the BoB. Mass balance calculations using Sr concentrations and 87Sr/86Sr suggest up to 7% contribution of SGD to the 100–120 m BoB water samples. The identification of SGD at 100–120 m depth also provides an explanation for the anomalous variations in barium (Ba) concentrations and the δ18O-salinity relationship in intermediate depths of the BoB.

this value could be biased by groundwater pumping in the Bengal Basin. Using regional hydrological modelling, the recharge rate was estimated to be 0.4 cm/yr, when normalized by a regional land area of 2.5 × 10 11 m 2 , and the SGD flux from the Bengal Basin was estimated to be 3 × 10 8 m 3 /year with post-development pumping taken into account and up to 9 × 10 8 m 3 /year without pumping 16 which was two to three orders of magnitude lower than the original proposal 6 . Additionally, the radiogenic 87 Sr/ 86 Sr ratio of the groundwater (0.7150-0.7200) 6 was not detected in measurements of surface water samples of the BoB 17 collected during the dry months of January and February from the north-eastern BoB (Fig. 1), thereby questioning the role of the Bengal Basin SGD in modulating the marine Sr isotopic composition.

Geochemical and Sr isotopic compositional heterogeneity in Bay of Bengal water samples.
Seawater samples were collected from 21 locations in the northwestern BoB, west of the sampling locations of earlier studies 11,12,17 (Fig. 1). These samples include surface water as well as sub-surface water and were collected over a time span of 29 months from August 2014 to December 2016 from the open ocean ( Fig. 1, Table 1). The samples collected in 2014 (August/September) and 2015 (August/September) were during the summer monsoon season when both river discharge and rainfall are high while samples collected in February 2016 and December 2016 were during the dry winter months (non-monsoon) when river discharge is low 18 . Strontium and Ca concentrations in the water samples were measured using an inductively coupled plasma mass spectrometer (ICPMS, Thermo Scientific X-Series II), while Sr isotope ratio measurements ( 87 Sr/ 86 Sr) were performed using a thermal ionization mass spectrometer (TIMS, Thermo Scientific Triton Plus), both at the Centre for Earth Sciences (CEaS), Indian Institute of Science (IISc). Salinity of the water samples were measured in-situ. See Methods for additional details.
Due to its conservative nature, 87 Sr/ 86 Sr ratio of the open ocean is broadly homogeneous. The homegeneity in seawater 87 Sr/ 86 Sr has been established by direct measurements of deep and shallow water samples from the Hudson Bay and the Pacific, Atlantic, Indian and Arctic oceans [19][20][21] as well as by measurements of biogenic carbonates 22 and marine barite from Holocene sediments from the Pacific, Atlantic and Indian oceans 23 . The maximum variability in the 87 Sr/ 86 Sr ratio of seawater at any given time is expected to be less than 10 ppm 19,20 . In contrast, significant variations in the 87 Sr/ 86 Sr ratio are observed in the water samples of the BoB collected from different depths. Samples collected from depths greater than 250 m show limited variability and the average 87 Sr/ 86 Sr ratio of these samples (0.709171 ± 9, 2 SD, n = 16) ( water 25 while near shore samples also show relatively low values 26 . The Sr concentrations in the BoB water samples overlap with that reported for global oceans [25][26][27] . However, the slightly lower concentrations, especially in the 0-50 m depth samples of the BoB, could be due to large inputs from the Ganges, Brahmaputra and Irrawaddy rivers which have very low Sr concentrations ( Table 2) as well as dilution due to rainwater contribution. It has been suggested that the salinity/Sr value of ocean water is constant 27 although, some studies suggest otherwise 28 . For the BoB water samples, the salinity/Sr values for the surface water samples overlap with those of the deep-water samples ( Fig. 3, Table 1). A couple of surface water samples (locations S5, S9) collected during the monsoon show high salinity/Sr due to the dilution effect of rainwater which has very low Sr 29 . The concentrations of Ca and Sr in the BoB water samples show a positive correlation (R 2 = 0.95, Fig. 2b) which is consistent with the similar behavior of these two elements in aqueous systems. However, the Ca/Sr ratio of the samples collected from 0 m and 50 m show mostly higher values compared to the deeper water samples (Fig. 2c).
The variations in the geochemical and Sr isotopic compositions in the BoB water samples could be due to multiple reasons that include: (i) local hydrothermal input and/or oceanic crust -seawater interaction at mid-oceanic ridges 30 , (ii) differential atmospheric contributions (aerosol and/or rainwater) 31 , (iii) biomineralization and its subsequent dissolution 25 , (iv) dissolution of riverine particulate matter 32 and (v) mixing of different water masses 33 . The absence of any active mid-oceanic ridge in the BoB rules out any local hydrothermal inputs. The atmospheric contribution of Sr to the Bengal basin was estimated to be 21 nmol/l 8 while Sr concentrations in rainwater were estimated to be 32-191 nM 29 . These concentrations are very low to affect the oceanic Sr reservoir with an average Sr concentration of 91.3 μmol/l 5 . In addition, the lack of any variations in Sr, Ca concentrations, salinity and as well as 87 Sr/ 86 Sr ratios between samples collected during the monsoon (2014, 2015) and non-monsoon (2016) (Fig. 2), all indicate that atmospheric contributions could not have affected the 87 Sr/ 86 Sr of the BoB water samples. Biomineralization of Celestite (SrSO 4 ), tests of the protozoan acantharia and its dissolution 25,34 have been invoked to explain the ~5% variation 25 in the Sr concentrations with depth in certain oceanic basins. However, these organisms have not been reported in the BoB. In addition, the lack of seasonal variation in the Sr, Ca concentrations and 87 Sr/ 86 Sr ratios in the monsoon-driven biogeochemical system of BoB, (Fig. 2) rules out the possibility of any biogenic contribution to the Sr isotopic variation of BoB. The dissolution of riverine particulate matter can also change the 87 Sr/ 86 Sr ratio of seawater samples 32,35 . If the dissolution of particulate sediments is the source of the high 87 Sr/ 86 Sr in the BoB water samples, a seasonal variation would be expected in the Sr concentration and 87 Sr/ 86 Sr of the BoB water as the suspended sediment load in the rivers is very high during the monsoon. In addition, the effect of sediment dissolution is expected to be higher in the samples collected from closer to the coast where the riverine influence is higher compared to the samples collected from father away. However, the radiogenic 87 Sr/ 86 Sr ratios observed in the BoB water samples, particularly in the samples collected from 100 m depth, which were collected both during the monsoon and non-monsoon seasons do not show any seasonal variations as well as any relationship with the sampling distance from the coast (Figs 1, 2). Therefore, the contribution from the dissolution of particulate matter can be ruled out. Hence, the observed variations in the Sr concentrations and 87 Sr/ 86 Sr as well as salinity in the BoB samples (Figs 2, 3) can only be explained by the mixing of water from different sources.
Contributions of river water and evidence for submarine groundwater discharge. The samples of this study can be divided into three distinct groups based on their Sr, Ca concentration, salinity, and 87 Sr/ 86 Sr (Figs 2, 3). The first group comprises samples collected from 0-50 m depths which show lower salinity, Sr and Ca concentrations and higher Ca/Sr compared to the other samples. Strontium concentrations, especially in the samples collected from the surface, show a rough positive correlation with salinity and fall on a mixing trend between the average composition of the deep waters of BoB (>250 m, D-BoB) (Fig. 2a), which overlaps with the composition of the average global seawater, and the average compositions of the Ganges, Brahmaputra and Irrawaddy rivers ( Table 2).
These samples also show large variations in 87 Sr/ 86 Sr, with mostly radiogenic values compared to the average D-BoB which overlaps with the composition of the global average seawater 21 (Table 2) (Figs 2b, 4). In a plot of 87 Sr/ 86 Sr versus 1/Sr, the composition of these samples can be explained by mixing between the average D-BoB and river water end-members comprising Ganges, Brahmaputra, and Irrawaddy (Fig. 4). The compositional uniqueness of the 0-50 m samples is also evident in a plot of salinity/Sr versus 1/Sr (Fig. 3b) where, unlike the deeper water samples, these samples show large variations in Sr concentrations and no correlation between salinity/Sr and 1/Sr. The low salinity of these samples suggests the dominance of the riverine input although, the influence of groundwater cannot be ruled out. The exchange of shallow groundwater (<40 m) with the Ganges-Brahmaputra rivers has been observed in the Bengal basin, where the Sr and Li composition of both end-members overlap with each other 14,36 . The influence of groundwater is consistent with the composition of some of these samples that plot within the mixing domain of high saline groundwater and average seawater (Figs 2a and 4). The second group comprises samples collected mostly from the deeper waters (>250 m depth, D-BoB) that are characterized by limited variations in Sr concentrations, average global seawater-like 87 Sr/ 86 Sr ratios, high values of salinity tightly clustered around 34.967 ± 0.129 psu (2 SD) and a strong correlation between salinity/Sr and 1/Sr (Fig. 3b). The Sr concentrations of these samples overlap with those of the Pacific and Atlantic oceans as reported by Bernat et al. 27 and De Villiers 25 but are lower than that reported by Wakaki et al. 26 . The third group consists of samples that are collected from 100-120 m depth and are characterized by significantly radiogenic 87 Sr/ 86 Sr ratios compared to D-BoB (average 87 Sr/ 86 Sr = 0.709171), as high as 0.709339, but unlike the 0-50 m samples, these have markedly higher salinity and Sr, Ca concentrations (Table 1, Fig. 2). The salinity of these samples is however, slightly lower than the >250 m samples (Table 1, Figs 2d and 3a). Eight out of 14 samples from 100-120 m depth show more radiogenic 87 Sr/ 86 Sr ratios than that of the D-BoB. This excess in 87 Sr/ 86 Sr is expressed as Δ 87 Sr/ 86 Sr (defined as [ 87 Sr/ 86 Sr (sample) /0.709171-1] × 10 6 ) and ranges from 24 ppm to as high as 237ppm (Fig. 4). We consider samples whose Δ 87 Sr/ 86 Sr values are higher than the analytical uncertainty of the measurements (9 ppm, 2 SD) based on the external reproducibility of the average D-BoB (n = 16) ( Table 1). The BoB is highly stratified with a strong shallow pycnocline at 10-30 m depth that inhibits mixing between surface and sub-surface water 37 . Rare exceptions can occur during tropical cyclones. For example, water characteristic of the Andaman Sea was observed within the thermocline of the western Bay of Bengal during the passage of the Tropical Cyclone Lehar 38 . However, the relatively high salinity of the water samples from 100-120 m depth (33.250-34.853) compared to the water samples from 0-50 m depth (28.501-33.262) ( Table 1) and the lack of seasonal variability in their 87 Sr/ 86 Sr ratios, which is expected due to the varying fluxes of the monsoon-driven rivers draining into the BoB, both rule out the possibility of a riverine water input at 100-120 m depth. Therefore, we propose that the high Δ 87 Sr/ 86 Sr of the water samples collected from the 100-120 m depths are a result of radiogenic Sr input of SGD from the Bengal Basin. The Sr input from a SGD is further corroborated in a plot of 87 Sr/ 86 Sr versus 1/Sr (Fig. 4), where compositions of the high Δ 87 Sr/ 86 Sr water samples can only be explained by the mixing between average seawater and saline groundwater 6,14 from the Bengal Basin (Table 2). This observation is also in-line with the results of regional groundwater modeling for the Bengal Basin which indicates the presence of large-scale groundwater flow at depths of 100 m or greater 16 . Submarine groundwater discharge and Ba, δ 18 O profiles. High concentrations of the non-conservative tracer Ba in suraface waters of BoB during non-monsoon was interpreted as a signature of SGD from the Bengal Basin 11 . In depth profiles of Ba concentrations in the BoB 39 , dissolved Ba in the coastal and open waters of the BoB show depleted concentrations in the surface and enrichments in the deeper layers, a pattern similar to other oceanic systems 40 . Compared to the north-eastern BoB 11 (Fig. 1), the Ba concentrations in the surface water samples of western BoB were found to be lower 39 . However, an increase in the Ba concentration below the shallow mixed layer (~60-120 m depth) was observed in most of the profiles 39 . One possible explanation for the increase in Ba concentrations at 60-120 m depths in northern BoB 39 is the desorption of Ba from particulate matter supplied by the SGD. However, without detailed characterization of the nature of the SGD to the BoB, the above explanation is speculative. At the same water depths, the correlation between δ 18 O and salinity in the BoB was found to be poor 24 , which was explained by the mixing of two distinct water masses having very different δ 18 O values. One of these two water masses were suggested as Indonesian Throughflow (ITF) while the second water mass was unidentified 24 . This unidentified water mass at intermediate depths in the BoB could be the Bengal Basin SGD, as identified in the present study.
Spatial heterogeneity in the submarine groundwater discharge. The high concentration of Sr in the Bengal Basin groundwater and its radiogenic 87 Sr/ 86 Sr composition led to the suggestion that SGD is an important source of continental Sr to the oceans and that it carries an equal in magnitude of dissolved Sr as the Ganges-Brahmaputra rivers 6 . The importance of constraining the submarine groundwater Sr flux from the Bengal Basin for evaluating the global marine Sr budget was recognized but signatures of the SGD with radiogenic 87 Sr/ 86 Sr were not found in measurements of the surface waters of BoB collected during non-monsoon (January-February) near Bangladesh 17 (Fig. 1). It may be noted that Beck et al. 17 did not analyze any deeper water samples. In the present study, the radiogenic Sr isotopic signature of the SGD is observed in the intermediate water layers (100-120 m depth) in the north-western BoB, off the coast of India. However, the radiogenic 87 Sr/ 86 Sr signature is not observed in all the samples collected from these depths. Additionally, while the SGD signature is observed in samples away from the coast, it was not detected in four depth profiles along a traverse from the Swatch of No  Ground (locations S18-S21) (Fig. 1), where the Sr-isotopic composition is very similar to the D-BoB (Table 1). The above observations suggests a spatially heterogeneous nature of the SGD flow in BoB. In the BoB, spatial heterogeneity is observed in the flow of river water in the surface BoB which is stirred by mesoscale eddies, often creating filament-like structures of low salinity waters and sharp fronts between freshwater and relatively saltier water 41 . A heterogeneous flow pattern of SGD from the Bengal Basin is consistent with spatially heterogeneous flow pattern of coastal SGD observed in Ubatuba, Brazil 42 and could explain the contrasting findings of the present study and that of Beck et al. 17 .
Sr flux to the BoB from submarine groundwater discharge. The identification of the SGD signature with radiogenic 87 Sr/ 86 Sr, high Sr and high salinity at intermediate water depths of the BoB is consistent with the suggestion of Basu et al. 6 regarding Sr inputs from the SGD to the BoB. However, given that the Sr-flux of SGD to the BoB from the Bengal Basin is heterogeneous, the SGD flux estimates to the BoB by Basu et al. 6 were possibly overestimates, as also suggested by regional hydrological modelling studies 16 . Using the Bengal Basin groundwater fluxes of Michael and Voss 16 which range from 3 × 10 8 m 3 /yr (post-groundwater pumping) to 9 × 10 8 m 3 /yr (pre-groundwater pumping) and the average Sr concentration of Bengal Basin groundwater of 4.5 μmol/l 6 , the SGD driven Sr flux to the BoB is estimated between 13.5 × 10 5 mol/yr and 40.5 × 10 5 mol/yr, which is three orders of magnitude lower than the estimates of Basu et al. 6 and Dowling et al. 14   the Bengal Basin groundwater (~7%), although, spatially restricted due to the heterogeneous nature of the SGD, is higher than estimates for Li contribution from the Bengal basin groundwater (~2%) to the BoB 36 and could have implications for estimates of continental Sr input to the oceans back in time.

Methods
Seawater samples from 21 locations in the northwestern BoB were collected during four different sampling missions aboard the ORV Sagar Kanya and ORV Sagar Nidhi over a time span of 29 months between August 2014 and December 2016 (Fig. 1, Table 1). The samples were collected from the open ocean. Sub-surface seawater samples were collected using 5 L Niskin bottles, mounted on a 12-bottle rosette with a Seabird CTD. The surface water samples were collected using a plastic bucket while the ship was stationed for CTD measurements. Water sampling at greater depths (depth-profiles) was not possible at all sampling locations. All samples were acidified with double-distilled nitric acid (pH ~2) immediately after sampling. The salinity data was measured in-situ from CTD (SeaBird SBE 19 Plus CTD system). The concentrations of Sr and Ca were measured using an Inductively Coupled Plasma Mass Spectrometer (ICPMS, Thermo Scientific X-Series II) at the Centre for Earth Sciences (CEaS), Indian Institute of Science (IISc). For these measurements, ultra-pure ICPMS solutions from Alfa Aesar (traceable to NIST standards) were used as calibrations standards (four-point calibration). A 10 ppb Be, In solution was used as an internal standard to correct instrumental drift. External reproducibility (2 SD) of the Sr (±2.33) and Ca (±0.47) concentrations and 1/ Sr (±0.0004) were estimated based on multiple measurements of a seawater standard NASS 6 (n = 46) measured in two different analytical sessions along with repeat measurements of the samples. The concentrations of Sr and Ca in the deep-water samples of the BoB are consistent with the deep-water measurements of global oceans [25][26][27] .
For isotopic measurements, Sr was separated from the samples using cation-exchange chromatography. The purified Sr, was loaded on single Ta filaments and measured using a Thermal Ionization Mass Spectrometer (TIMS, Thermo Scientific Triton Plus) at CEaS, IISc. The NIST SRM 987 standard analyzed during the course of this study yielded an average value of 0.710268 ± 0.000022 (n = 22). The measured 87 Sr/ 86 Sr ratios were corrected for instrumental mass fractionation using 86 Sr/ 88 Sr = 0.1194. The normalized 87 Sr/ 86 Sr of the samples were re-normalized to the recommended NIST SRM 987 value of 0.710244 43 . After re-normalization, the 87 Sr/ 86 Sr ratio of the NASS 6 seawater standard, analyzed during the course of this study, yielded a value of 0.709180 ± 0.000020 (n = 4), which is consistent with the average global seawater 87 Sr/ 86 Sr value 21 indicating that our measurements are accurate. The 87 Sr/ 86 Sr ratio of sixteen deep BoB water samples (D-BoB) collected from depths greater than 250 m yield an average value of 0.709170 ± 9 (2 SD) which is consistent with 87 Sr/ 86 Sr of global seawater value of 0.709179 ± 8 21 . The D-BoB measurements indicate that our analyses are accurate and the external reproducibility of our 87 Sr/ 86 Sr measurements is better than 9 ppm. Additional details of the elemental and isotopic measurements are described in Banerjee et al. 44 .  (Tables 1, 2). The samples from 0-50 m and 100-120 m show significant variations in 87 Sr/ 86 Sr ratios, which are greater than our analytical uncertainty (±9 ppm) (gray shade and cross-hairs). The compositional variations in the BoB samples from 0-50 m can be explained by mixing between the average D-BoB and the Ganges 6 , Brahmaputra 6 and Irrawaddy 10 rivers. Compositions of several samples from 100-120 m depth (squares) with very radiogenic 87 Sr/ 86 Sr and low 1/Sr (2 SD, ±0.0004, shown as cross-hairs) require contributions from saline groundwater from the Bengal Basin 14 .