Introduction

Biogeochemical processes in aquatic environments are investigated to understand the pathways by which essential compounds for life are circulated. The sulphur cycle has a key role on these flows given its many interconnected reactions to other nutrients such as carbon or iron, among others1,2,3. Although sulphur cycling in freshwater and marine systems has been studied for many decades and its main pathways have been thoroughly described and reviewed1,2,3,4, the potential for different reactions in saline inland lakes still deserves further exploration.

Sulphate (SO42−) is one of the main electron acceptors in anoxic habitats. Bacterial sulphate reduction (BSR) involves the production of hydrogen sulphide (H2S) which can be either reoxidized back to SO42− or lead to the precipitation of secondary minerals such as pyrite3,4. Organic matter, iron species, oxygen and light availability have a key role in promoting certain processes over others (e.g. oxidation vs reduction or biotic vs abiotic reactions)5,6,7,8. BSR is the main microbial process that remineralizes and recycles organic matter in marine systems, such as euxinic basins and continental margin sediments, because SO42− is highly available with a mean concentration of 28 mM in seawater2. In contrast, SO42− concentrations are generally 2 to 3 orders of magnitude lower in freshwater environments, which restricts BSR in sediments as SO42− becomes rapidly depleted. Consequently, methanogenesis becomes the main anaerobic process of organic matter remineralization in freshwater sediments4. Previous studies in brackish to hypersaline lakes have shown that high salinities do not necessarily inhibit BSR or sulphide oxidation9,10,11,12,13. However, hot-spots of BSR activity and the extent to which the source of SO42− in different athalassic saline systems can come from the lake bed minerals or from groundwater instead of from surface water are poorly documented yet essential to predict methanogenesis potential in these ecosystems.

In ephemeral inland wetlands, variations in chemical and physical parameters are dependent on evaporation, rainfall or groundwater inflows and directly impact biogeochemical cycles14,15. The organic matter sources and the geological characteristics of the setting also play a key role. Coupled geochemical and isotopic characterisation of sulphur compounds in different parts of the lake including vertical profiles (i.e. surface water, porewater, sediment and groundwater), can provide further insight into sulphur cycling in these aquatic environments. This will be especially useful to anticipate biogeochemistry variations upon climate change as increasing temperatures can contribute to the salinization of freshwater lakes.

The saline lake Gallocanta (40°58′00″N, 1°29′50″W), located on a plateau within the Iberian Range at 990 m a.s.l, can be used as a model study site given its following particular characteristics. It is the largest and best-preserved endorheic saline lake in western Europe. It is shallow and has a pH ranging between 8 and 1016. Its maximum water depth for the last 30 years has remained below 1 m, occasionally with periods of complete dryness as the climate of the region is semiarid16,17,18. The water volume of the lake varies mainly due to the rates of evaporation and precipitation, which generate important runoff water flows and influences the groundwater table16,17,18,19. The groundwater flow tends towards the lake and discharges to it through detritic Quaternary material19,20. The multilayer aquifer system surrounding Gallocanta Lake is composed of an unconfined detritic Quaternary aquifer and a partially permeable Mesozoic carbonated aquifer16,19,20. The origin of the depression is karstic and overlies Triassic clays and evaporites (Keuper facies) although quaternary materials are found on the lake edges21,22. Mineralogy of the lake bed is rich in epsomite, hexahydrite, gypsum, quartz and phyllosilicates, halite, bischofite, calcite, dolomite and aragonite, whose proportion presents cross-shore and depth variations23,24,25. Previous studies on gypsum distribution in Gallocanta Lake sediment have shown spatial and depth variations. More specifically, gypsum amount increases from the shores to the centre of the lake. Furthermore, the maximum gypsum concentration is found between 70 and 100 cm depth in the shores of the lake while at the centre of the lake it is found at about 20 cm depth23,25,26,27,28,29. None of these studies report the spatial distribution of other SO42− minerals. Increases in salinity have been related to lower water volumes of the lake and during the dry periods when mainly carbonate and sulphate salts precipitate23. The goal of the present study was to study the occurrence of BSR and its main drivers in Gallocanta Lake using an integrated geochemical and isotopic approach.

Saline endorheic Gallocanta Lake

Three sampling locations in the lake (A, B, C) were selected as representative areas of different cross-shore subenvironments to study sulphur cycling (Fig. 1). Surface water, porewater and sediment samples were collected in November 2020 (sites A and B) and June 2021 (sites A, B and C). These two seasons covered a daily temperature range from 6 to 14 °C and from 15 to 26 °C, respectively. Also, groundwater samples were obtained from available sources nearby the lake (Fig. 1). Conductivity measurements of surface water at sites A and B in November was 18.4 ± 1.6 mS/cm (Fig. S1). Calcium, magnesium, sodium and potassium concentrations were 15.5 ± 0.5, 75.5 ± 4.4, 240 ± 15 and 6.7 ± 0.7 mM, respectively (Fig. S2). The measured SO42− was 64.2 ± 2.2 mM (Fig. S3). The concentration of these major ions in surface water of Gallocanta Lake did not vary throughout the day but increased with depth in the water column (e.g. SO42− varied from 21 mM at the sub-surface to 37 mM in the bottom at 9:00 in site B on June). Furthermore, site C, which is closer to the centre of the lake, had a higher conductivity compared to A and B (45 ± 4 mS/cm). Chlorine and bicarbonate concentrations were not measured in this study but concentrations reported previously, considering cross-shore variations, reached up to 1890 and 14 mM, respectively23,30.

Figure 1
figure 1

Study site. Sampling points in Gallocanta Lake (blue (A), orange (B) and grey (C)) and nearby sources (violet, S1 to S7). Image modified from Google Earth, © 2022.

Full size image

Detailed data on the hourly measurements of temperature, conductivity, dissolved O2, pH and organic carbon in surface water; major ions concentrations in surface and porewater; SO42− concentrations and isotopic composition in surface water, porewater, and groundwater; sulphide concentration in porewater; and bulk carbon content and isotopic composition in sediment is reported in the Supporting Information.

Organic matter inputs

Gallocanta Lake is surrounded by vegetation and the lake bed is colonized by Ruppia drepanensis, which is a rooted submerged macrophyte commonly found in ephemeral saline inland wetlands of the Mediterranean region31,32,33. Also, it harbours a large community of migratory crane (Grus grus) from November to March each year34,35,36. Therefore, birds excrements and vegetation decay (mainly R. drepanensis) represent a significant source of organic matter and nutrients to the lake. Furthermore, daily measured dissolved O2 concentrations and pH values in surface water were highest at midday as a consequence of high rates of photosynthetic activity both in November and June, despite its salinity and alkalinity (e.g. pH increased from 9 at 11:00 to 9.8 at 15:00, in site B in November). Measured non-purgeable dissolved organic carbon (NPDOC) in surface water in November was 3.3 ± 0.03 mM for sites A and B and showed no daily variations. Nevertheless, the NPDOC concentration was higher in the bottom of the water column compared to the surface (e.g. 2.9 vs. 1.7 mM at 9:00 in site B in June, Fig. S4).

Bulk carbon detected in dry sediment ranged between 5.4 and 6.3% for site A, and between 6.8 and 9.0% for site B. In both cases the content decreased with sediment depth (up to 10 cm, Fig. S5a), which is consistent with mineralogy changes. Organic carbon content after decarbonation was 1% (site B, 4 cm), demonstrating the inorganic nature of the substrate with the exception of a layer rich in organic matter. This black layer is found immediately below the water–sediment interface (< 4 cm) and induces anoxic conditions. Previous studies in Gallocanta Lake also reported the highest organic carbon content on the first 10 cm of the sediment with values ranging from 1 to 6%25,29,30. The measured δ13C for bulk C in sediment ranged from − 7.5 to − 11.1 ‰ and showed an enrichment in the heavier isotopes with depth that coincides with the bulk C content decrease (Fig. S5b). δ13C values ranging from − 1.5 to − 11 ‰ have been previously reported for calcite, magnesite and dolomite in Gallocanta Lake sediments24,37. An increase on the heavy isotopes with depth was also observed in these studies. The δ13C measured for organic C was − 22.3 ‰ (site B, 4 cm). According to this result, the decreasing bulk C content accompanied by an increase in the δ13C values, points to a lower contribution of organic matter with depth38. According to this mass balance assumption, organic matter content is highest in the top layers of the sediment and the upper sediment has the higher mineralization potential.

Sulphur cycling processes in a sulphate rich system

The SO42− concentrations in the water column of Gallocanta Lake (21 to 65 mM) were similar or lower than those found for the upper layers of porewater (59 to 80 mM, < 2 cm depth). Below 2 cm, SO42− concentration increased with sediment depth to a maximum of 235 mM SO42− (sites A, B and C, Fig. 2). Sulphate concentrations were higher closer to the centre (site C) compared to the shore of the lake (sites A and B).

Figure 2
figure 2

Sulphate concentration in surface and pore water. The water column depths ranged from 30 to 60 cm depending on the season and sampling point. In the plot, “S” = surface and “M” = middle. Deeper cores were obtained in June 2021 compared to November 2020.

Full size image

Sulphate concentration decreases in porewater are usually related to BSR, especially in anoxic environments with high organic matter inputs. Indeed, up to 600 µM for sulphide species (ΣS2− = H2S + HS− + S2−) were measured in the first mm of Gallocanta Lake sediment for site B and in a lesser extent for site A using a microprobe in November (Gonzalez-Álvarez et al., in preparation). The generation of sulphide in site B was confirmed during an additional sampling campaign performed in October 2022 (Fig. S6). Dissolved sulphide and iron sulphides such as framboidal pyrite, which can form in sediments containing ferrous iron and sulphide26, have been previously detected in Gallocanta Lake25,26,30. Also, in site B purple bacteria were observed in the water overlying the sediment cores immediately after collection, suggesting further sulphide oxidation on the water–sediment interface (Fig. S7). This is consistent with the low O2 concentrations measured at the bottom of the water column in site B in June (< 0.1 mM). Sulphate reducing and sulphur oxidizing microorganisms have been previously reported in bulk sediment samples from Gallocanta Lake39. However, although evidence of BSR and sulphide oxidation were robust, it is unlikely that these biological processes could produce almost a 200 mM increase over a 25 cm depth.

We hypothesized that SO42− concentration in porewater increased with depth due to SO42− dissolution from minerals or salts originating from the evaporites of the geological substrate. Also, that the most active location for BSR was the organic matter rich layer close to the water–sediment interface. Porewater concentration of Mg2+ correlate strongly with SO42− (r2 > 0.99, Fig. S8), and epsomite (MgSO4·7H2O) is highly undersaturated throughout the sediment (Saturation Index (SI) ranges from − 1.8 at depth to − 1 at the surface, based on Visual MINTEQ calculation) suggesting that dissolution of epsomite is the main source of SO42−. Porewater Ca2+ concentrations are uniform with depth and the very low saturation of gypsum (SI ranges from 0.23 at depth to − 0.11 at the surface) indicate a possible equilibrium. Moreover, the much lower SO42− concentrations measured in groundwaters nearby the lake (4.7 mM, Figure S9), suggests that these sources are not contributing SO42− to the lake.

δ34S data as a proxy to depict sulphate cycling pathways

The sulphur isotopic composition of SO42− (δ34S-SO42−) can be used to trace its sources and transformation processes40,41,42. It has been established for some decades that BSR generate an isotopic fractionation leading to an increase of the δ34S values of the residual substrate in contrast to dilution that do not modify the isotopic signature43. Under closed system conditions, with no substrate renewal and in the absence of isotopic exchange, the isotopic fractionation (ε) is calculated by means of the Rayleigh distillation equation which involves the analyte concentration (C) and the determined isotopic composition:

$${\delta }_{residual} - {\delta }_{initial}= \varepsilon x Ln\left(\frac{{C}_{residual}}{{C}_{initial}}\right)$$

The use of the Rayleigh equation also implies a unidirectional and irreversible reaction. Instead, in open systems such as sediments influenced by mass exchange across the water–sediment interface and diffusive flows, the isotopic fractionation can be derived from the following equation proposed by Canfield (2001)44:

$${\delta }_{residual}= \frac{\left[\frac{{\delta }_{initial}}{\left(1-\frac{{C}_{residual}}{{C}_{initial}}\right)}\right]+\left[\frac{\varepsilon }{\left(\frac{\varepsilon }{1000}+1\right)}\right]}{\left[\frac{1}{\left(\frac{\varepsilon }{1000}+1\right)}\right]+\left[\frac{\left(\frac{{C}_{residual}}{{C}_{initial}}\right)}{\left(1-\frac{{C}_{residual}}{{C}_{initial}}\right)}\right]}$$

The ε values previously reported in the literature for BSR range between − 4 and − 66 ‰43,45,46,47,48,49,50. The main causes for these variations are related to the microbial SO42− metabolism and therefore to the SO42− reduction rates, the type and availability of electron donors, the temperature of the media and the active bacterial community43,45,46,47,48.

The δ34S-SO42− values of surface water of Gallocanta were uniform across all sites, time, and season (+ 21.7 ± 0.3‰, Fig. S10). Also, no significant variation of δ34S-SO42− were observed in porewater of Gallocanta Lake from the middle to the bottom of the sediment cores collected in the sites A, B and C with an average of + 21.0 ± 0.5‰ (Fig. 3a). However, δ34S-SO42− increases up to + 24.5‰ in porewater in the site B from the middle to the top of the sediment core . This enrichment coincided with a decrease in SO42− concentrations, suggesting the process of BSR (Fig. 3b). The calculated ε value using the Rayleigh equation was − 6.6‰ (Fig. S11), which is in the range of those reported in the literature for BSR43,45,46,47,48. The use of this equation can be valid for Gallocanta sediment since the percentage of sulphur from SO42− reduction in the study site might be extremely low compared to the infinite amount of available substrate and because BSR rates can outcompete diffusive fluxes. To validate it, we compared the sample δ34S-SO42− values to the modelled trends with the open system equation. For this representation, we employed the determined ɛ values with Rayleigh and those estimated with the open system equation (Fig. S12). The obtained good fitting points that BSR in Gallocanta operates as in closed conditions, showing low or negligible effect from diffusion or advection processes on the SO42− isotopic fractionation.

Figure 3
figure 3

Porewater sulphate isotopic composition. Isotopic signature evolution with depth (3a) and with respect to concentration (3b). Data correspond to June 2021 samples.

Full size image

The δ34S determined for bulk S in sediment (Site B, 9 and 21 cm depth) was + 21.2 ± 1.6‰, which is extremely close to the measured average δ34S-SO42− in surface water (+ 21.7 ± 0.3‰) and porewater (+ 21.0 ± 0.5‰). δ34S values reported for nearby Triassic evaporites (anhydrite) ranged from + 12.5 to + 14.5‰51. Precipitation of dissolved SO42− during gypsum formation can lead to an enrichment of heavy isotopes of up to 2‰52. Therefore, what was measured as bulk S seems to correspond to SO42− salts precipitated after several cycles of dissolution–precipitation of secondary SO42− minerals such as epsomite. On the other hand, significantly lower δ34S-SO42− values (from + 8.0 to + 14.5‰) were measured in groundwater (Fig. 4). Therefore, the lack of variations in the measured δ34S-SO42− for porewater samples in which concentration varied to almost 180 mM, elucidated that source of dissolved SO42− was the dissolution of local epsomite. In a similar study in the Salton Sea, SO42− dissolution from evaporite deposits and subsequent diffusion was observed but the possibility of BSR was not assessed53. Also, in the saline Devils lake (North Dakota), a bidirectional SO42− diffusion (from the lake bed and the water column) was observed at the BSR layer11. The SO42− concentration decreased from ~ 16 mM at the water–sediment interface to ~ 10 mM at 2 cm depth and then increased to ~ 24 mM at 8 cm depth11. Given the much high SO42− levels found in porewater compared to surface water of Gallocanta Lake, we only considered an upwards diffusion. Also, these high levels SO42− ensure BSR outcompetes methanogenesis.

Figure 4
figure 4

Sulphate isotopic composition versus concentration in water samples from Gallocanta Lake. The average δ34S of bulk S in sediment, including standard deviation, is presented as an ochre horizontal bar (bulk S content in sediment was 0.3–0.9%, which is not reflected in this figure).

Full size image

A progressive decrease in SO42− concentrations from top to the bottom is commonly caused by BSR in oxygen depleted sediment in freshwater and marine environments3,54. Instead, in Gallocanta Lake SO42− concentrations increase with depth upon dissolution of SO42− rich minerals and salts. Sulphur isotopic data support this hypothesis and demonstrate that BSR is active in the organic rich layer below the water–sediment interface (Fig. 5). This was also supported by the high variations of SO42− concentrations, the detection of H2S and the observation of purple phototrophic bacteria. The sulphur cycling pathways observed in Gallocanta Lake could also be found in other systems containing soluble SO42− mineral beds and should be considered in studies aiming to determine the fate of nutrients. That is because in freshwater and marine environments, the main electron acceptors for the organic matter mineralization are available after diffusion from the water column (e.g. oxygen, nitrate or sulphate) or sedimentation (e.g. iron or manganese oxides). Contrarily, in saline inland lakes such as Gallocanta one of the major oxidants, SO42−, can be available from the lake bed. This will likely constrain other nutrient cycling processes such as the methanogenesis.

Figure 5
figure 5

Sulphur cycling processes revealed by isotopic data. Theoretical trends for SO42− minerals/salts dissolution and BSR have been drawn by using ε = 0‰ and from − 4 to − 66‰, respectively. These lines are plotted together with the obtained results for our samples. For both axes, “final” corresponds to the values obtained at each depth while “initial” correspond to the deepest points (highest SO42− concentration).

Full size image

Concluding remarks

BSR in Gallocanta Lake occurs near the water–sediment interface. The source of SO42− in the system is found on the lake bed mineralogy and leads a 3 to 4 fold SO42− concentration increase in porewater with depth (reaching up to 235 mM). Continuously dissolved SO42− can be used as substrate for BSR and therefore, it prevents the occurrence of methanogenesis. The extremely high SO42− content in the system constrains not only sulphur cycling but also other biogeochemical processes. Coupling integrated geochemical and isotopic data from different lake compartments with field observations is valuable to understand ecosystem functioning.

Materials and methods

The study involved two sampling campaigns to collect water and sediment samples from the study site for the subsequent chemical and isotopic characterization. The first was performed on November 2020 and the second on June 2021, to check seasonal variations. Two sampling points were stablished in the margins of the lake and named “A” and “B” and one closer to the centre of it “C”, to check spatial differences (Fig. 1). In November 2020, triplicate water samples were obtained from the middle of the water column following a diurnal cycle from sites A and B. In June 2021, water samples were obtained from two different water depths (subsurface and bottom of the water column) also following a diurnal cycle for sites A, B and C. In both campaigns, sediment cores were obtained early in the morning. Furthermore, water samples were also obtained from several sources identified nearby the lake during both sampling campaigns (Fig. 1). Details are reported in Table 1.

Table 1 Sampling campaigns.
Full size table

Prior to each water sample collection, temperature, pH, Eh, O2 and conductivity were determined in situ by using a multiparametric probe (Aquaread AP-5000). Water samples were collected in 500 mL plastic flasks after three rinses. Collection at different water column depths was achieved by using 250 mL syringes connected to tygon tubes. Sediment cores were collected by using sealed PVC tubes (10 cm diameter, 40 cm height). Water and sediment samples were immediately treated after collection as follows.

An aliquot for each surface water sample was filtered through pre-ashed GFF filters and HCl acidified for NPDOC determination immediately after collection. The remaining water sample was filtered through 0.2 µm Sterivex Millipore® filters. Aliquots for SO42− determination were HCl acidified while for dissolved major elements determination it was HNO3 acidified, they were all preserved at 4 °C. For the SO42− isotopic analysis, the dissolved SO42− was precipitated as BaSO4 by adding BaCl2·2H2O after acidifying the sample with HCl in order to prevent precipitation of BaCO355.

Sediment cores were sliced inside an anaerobic chamber under a N2 atmosphere at different depth intervals, each slice was introduced into tubes and subsequently centrifuged to separate porewater from the solid fraction except for one aliquot that was preserved to determine the water content after lyophilization. After centrifugation, porewater samples were prepared with the same methods used for surface water samples. The solid fraction was frozen to determine the content and isotopic composition of C and S after lyophilization and milling in the laboratory.

The SO42− content was determined by nephelometry and DOC by organic matter combustion56. Concentration of major ions (Ca, Mg, K and Na) was determined by ICP-OES (iCAP 6000 series, Thermo Scientific). The amount of C and S in lyophilized sediment samples was measured with an elemental analyser (EA, Flash 2000, Thermo Scientific). The δ34S-SO42− was analysed with a Carlo Erba EA coupled in continuous flow to a Finnigan Delta XP Plus isotope ratio mass spectrometer (IRMS). The bulk carbon and sulphur content and isotopic composition (δ13C-Cbulk and δ34S-Sbulk) of the sediment samples were determined by EA-IRMS (Flash 2000 EA and Delta V plus IRMS, Thermo Scientific). The content and δ13C of organic carbon in the sediment samples after decarbonation was also determined by EA-IRMS (Elementar-Isoprime)57. The standard deviation for δ34S and δ13C analyses was below ± 0.1‰. The isotopic notation is expressed in terms of δ per mil relative to international standards Vienna Canyon-Diabolo-Troilite (VCDT) for δ34S and Pee Dee Belemnite (VPDB) for δ13C, following:

$$\delta \, = \text{ } \frac{{\text{R}}_{\text{sample}}-{\text{R}}_{\text{standard}}}{{\text{R}}_{\text{standard}}},$$

where R = 34S/32S and 13C/12C, respectively.