Understanding the mechanisms of microbial metabolism beneath the seafloor in anoxic sediments remains a widely studied research area due to its importance in understanding planetary evolution, astrobiology, and life in extreme environments. In marine sediments, organic carbon is the dominant electron donor due to its availability and high metabolic energy yield (Orcutt et al., 2011). The oxygen penetration depth in marine sediments is usually only a few millimeters or centimeters (Glud, 2008; Boyko et al., 2018), while oxygen can penetrate meters into the sediments of extremely oligotrophic marine systems such as the South Pacific gyre (Fischer et al., 2009). Nitrate is consumed immediately below this depth and in the deeper anoxic sediments sulfate, Fe(III) (hyrdr)oxides and Mn(IV) oxides are the quantitatively-important terminal electron acceptors utilized for microbial metabolism (Thamdrup, 2000; Jørgensen, 2021). Mineralization of organic matter coupled to the reduction of Mn(IV) phases produces the highest amount of metabolic energy in anoxic zones, but sulfate and Fe(III) phases are more abundant in modern marine sediments. Although reduction of poorly crystallized Fe(III) (hydr)oxides produces more metabolic energy than reduction of sulfate, reduction of more stable Fe(III) phases often yields less metabolic energy than sulfate reduction (Bethke et al., 2011; Postma and Jakobsen, 1996). Depending on the content and reactivity of sedimentary organic matter and terminal electron acceptors, biogeochemical zones which are characterized by the consumption of various electron acceptors, often partially overlap (Postma and Jakobsen, 1996; Blonder et al., 2017).

The sulfur cycle in marine sediments starts with microbial sulfate reduction to hydrogen sulfide coupled to the oxidation of organic matter (Jørgensen, 1982). The fate of this produced hydrogen sulfide includes re-oxidation (Yao and Millero, 1993, 1996; Avetisyan et al., 2021), precipitation (Berner, 1970), and incorporation into organic matter (Amrani, 2014 and references therein). The oxidation of H₂S in natural aquatic systems results in the formation of elemental sulfur and sulfate, as well as polysulfides, sulfite and thiosulfate (Yao and Millero, 1993, 1996; Zopfi et al., 2004; Kamyshny and Ferdelman, 2010; Findlay and Kamyshny, 2017). The sulfur and iron cycles in sediments are coupled through the oxidation of hydrogen sulfide by reactive iron(III) (hydr)oxides (Yao and Millero, 1996; Zopfi et al., 2004); and Mn(IV) oxides (Burdige and Nealson, 1986; Yao and Millero, 1993; Avetisyan et al., 2021) as well as the formation of pyrite (FeS₂) (Berner, 1970; Raiswell and Canfield, 1998). Iron(III) (hydr)oxides minerals also represent the most bioavailable form of iron, although there is no direct correlation between the reactivity of iron minerals toward hydrogen sulfide and their bioavailability (Laufer et al., 2020).

In the case where the flux of iron to a sedimentary environment is higher than the flux of reduced sulfur, two situations may arise. First, microbial sulfate reduction may be - to some extent - suppressed due to ecological competition with iron-reducing microorganisms (Lovley and Phillips, 1987). Secondly, hydrogen sulfide, which is formed by microbial sulfate reduction, will be immediately oxidized by the Fe(III) phases (Yao and Millero, 1996). Combination of these processes may lead to formation of so-called “cryptic” sulfur cycling, e.g., the absence of hydrogen sulfide or a decrease in sulfate concentrations in the pore waters despite the presence of microbial sulfate reduction (Wehrmann et al., 2014, 2017; Mills et al., 2016; Blonder et al., 2017; Boyko et al., 2022). One of the lines of evidence for a cryptic sulfur cycle is the presence of sulfide oxidation intermediates. Isotopic composition of sulfur species, especially oxygen isotopes of sulfate, may also serve as evidence for cryptic sulfur cycling in the sediments (Mills et al., 2016; Blonder et al., 2017). Trace amounts of pyrite in the sediments may provide an additional line of evidence for the existence of microbial sulfate reduction (Blonder et al., 2017). The ultimate proof of presence of the active sulfur cycling is a direct quantification of microbial sulfate reduction (Holmkvist et al., 2011).

Measurements of the sulfur isotope composition of sulfate provide only very limited information on transformations of sulfur species during the cryptic sulfur cycling (Mills et al., 2016; Blonder et al., 2017). In such systems, the oxygen isotopic composition of sulfate (δ¹⁸O_SO4) may serve as the best isotopic tool for understanding redox transformations of sulfur. The enrichment of sulfate in ¹⁸O isotope during microbial sulfate reduction is a result of incorporation of oxygen isotopes from water into sulfate when it is in intracellular intermediate valence state sulfur species (Turchyn et al., 2006; Antler et al., 2013; Blonder et al., 2017). Cryptic sulfur cycling in iron-rich sediments has been reported in various environments including salt marshes (Mills et al., 2016) and coastal sediments influenced by high iron input from fluvial and aeolian sources (Wehrmann et al., 2014, 2017; Blonder et al., 2017; Boyko et al., 2018, 2019, 2022).

One sedimentary system that hosts a cryptic sulfur cycle due to high fluxes of highly reactive iron to the sediments is Svalbard fjords. A source of iron to the sediments in the Svalbard fjords is glacial runoff and the source of organic matter is primary productivity in the water column (Michaud et al., 2020). According to Wehrmann et al. (2017), in these sediments low concentrations of “easily” reducible particulate iron phases coincides with low porewater Fe²⁺_(aq) and Mn²⁺_(aq), and high concentrations of reduced sulfur. In contrast at high concentrations Fe²⁺_(aq) (> 850 μM) and Mn⁺²_(aq) (>650 μM), FeS and FeS₂ sedimentary contents are low, and sulfur cycle may be defined as cryptic. Another sedimentary system with a similar sedimentary cryptic sulfur cycle is the Red Sea and especially its North-Eastern extension, the Gulf of Aqaba (Blonder et al., 2017; Boyko et al., 2018, 2019, 2022). Two sources of reactive iron to these environments include direct deposition of dust from adjacent deserts, especially the Sahara, and seasonal flash floods transferring dust that was deposited on the shores to the gulf (Katz et al., 2015; Torfstein et al., 2017; Blonder et al., 2017; Boyko et al., 2019). At 700 m water depth, the concentrations of hydrogen sulfide, its oxidation intermediates and pyrite are extremely low, the presence of active sulfur cycling is evident from the oxygen isotope composition of sulfate (Blonder et al., 2017).

The Red Sea and especially the Gulf of Aqaba are enclosed basins with limited water exchange with the open ocean. Thus, it is not clear if the presence of a cryptic sulfur cycle in the sediments, which are strongly affected by aeolian dust deposition from Sahara, is a common phenomenon or rather restricted to the specific condition of high inputs of highly reactive iron, which prevail in the Red Sea and the Gulf of Aqaba (Boyko et al., 2022). The main goal of this research was to clarify if the sediments of the north-eastern Atlantic Ocean which are strongly affected by the aeolian deposition of Saharan dust are also hosting a cryptic sulfur cycling. In order to answer this question, we studied sulfur, iron, and manganese speciation as well as the isotopic composition of sulfur species in the ODP and IODP cores from the Atlantic coast of Africa in sediment during the last ~100 ka.

The Sahara Desert is the largest source of dust on Earth and accounts for up to half of all aeolian material deposited at the surface of the world’s oceans (Goudie and Middleton, 2001 and references therein). Up to 700 Tg of crustal material is mobilized from the Sahara per year and about 1/3 of this amount is deposited in the North Atlantic Ocean (Duce et al., 1991). Total iron content in the Saharan dust is 240–330 mmol kg^-1 (Goudie and Middleton, 2001 and references therein). Sediments in the North Atlantic primarily originate from hemipelagic particle flux due to coastal upwelling and dust input, with limited fluvial contribution in northern Morocco (Henrich et al., 2010; Holz et al., 2004). The area between 18°N and 20°N along the NW-African margin features well-developed canyon systems and gullies that influence sediment distribution and transport on the continental slope by serving as path for turbidity currents (Henrich et al., 2010).

Over the last 100 ka, the period with the highest amount of precipitation in Sahara occurred at c.a. 5.5 – 15.0 ka and is known as the “Green Sahara” period, due to the abundant vegetation (Cheddadi et al., 2021). Between 15.0–100 ka, the amount of precipitation was lower than during the Green Sahara period with periods of enhanced dust transport to the Atlantic Ocean (Skonieczny et al., 2019).

We suggest that the distance from shore and latitude are potential controls on the source of iron, transport and delivery to this area west of the Sahara. Thus, the sediments studied in this work can be divided into three groups (Table 1; Figure 1). Group 1: cores from DSDP 79 (Site 546), ODP 157 (Site 954), and DSDP 41 (Site 368) are located < 550 km from the African shore and in the 17°N to 34°N area, e.g. at the same latitude as the Sahara Desert (12°N to 36°N). Group two: core ODP 157 (Site 952) is in the same latitude interval but is 1114 km from the African shore. Group 3: cores DSDP 41 (Site 366) and ODP 108 (Site 665) are located 790–930 km from the shore of Africa, but to the south of Sahara, at 2°N to 6°N. Sediments lithology and composition are summarized in Table 2.

Although we did not measure the ages of the individual samples in this study, a combination of the reported sediment accumulation rates and an assumption that up to one meter of surface sediment was lost during sampling, allows us to place all our samples in the interval between 0 and 115 ka (Lancelot et al., 1978a, b; Hinz et al., 1984; Ruddiman et al., 1988; Schmincke et al., 1995a, b). This period corresponds to two types of climatic conditions: (1) the “Green Sahara” period, 14.5 to 5.5 ka before present and (2) more arid periods.

The samples were received from the IODP Bremen Core Repository (MARUM) in sealed plastic bags. The samples were weighed, dried to the constant weight at 50°C for 5 days and stored in hermetically closed vials at ambient conditions prior to analysis.

The same sub-samples were used for analysis of zero-valent sulfur (ZVS), acid volatile sulfide (AVS), and chromium-reducible sulfur (CRS). Around 1.0 g of sediment was transferred into a 50 mL plastic centrifuge tube. Methanol (HPLC-grade) was added to the sample at a sample-to-methanol ratio of 1:20 [w/v], following the methodology of Zopfi et al. (2004). Extraction of ZVS was performed on a rotary shaker at 120 rpm for 16 hours. Quantification of ZVS was carried out on the supernatant after centrifugation of the sample. Determination of ZVS in the methanol extract was done with HPLC with UV-VIS detector and a reverse-phase column. An Agilent Technologies 1260 Infinity High-Performance Liquid Chromatography (HPLC) system was utilized for chromatographic analyses, employing a Grace Prevail C18 reversed-phase column (5µm, 250 x 4.6 mm I.D). This setup was complemented with an Agilent G1365D UV-VIS detector and an Agilent G1321B fluorescence detector (FLD). The mobile phase consisted of HPLC-grade methanol and the flow rate was 1 mL min^-1. Spectrophotometric quantification was performed at a 230 nm wavelength. Spectrometric analyses were carried out using the Lambda 365 UV-vis spectrometer by PerkinElmer^®. The method detection limit (MDL) was 10 µmol kg^-1 of wet sediment. Following the methanol extraction, analyses of AVS and CRS were performed on the same sample by distillation with HCl and with an acidic solution of CrCl₂, respectively, as described by Fossing and Jørgensen (1989), with a 15 mL of 50 g L^-1 zinc acetate as the trapping solution. The quantification of zinc sulfide in the trapping solution was performed spectrophotometrically at 665 nm. The MDL for the method was 10 µmol kg^-1.

Total manganese (Mn_TOT) and total iron contents (Fe_TOT) were quantified in c.a. 0.5 g of sediment sample, which was ashed at 450°C for 8 hours, followed by a 24-hour digestion in near-boiling 6 mol L^-1 HCl as described by Aller et al. (1986). Iron contents were determined by the ferrozine method outlined by Stookey (1970) with one amendment: an ascorbic acid solution (20 mg per 10 mL) was added, and the solution was shaken for 1 min by hand in order to convert all iron to the Fe²⁺ form. Total manganese contents were measured by the photometric (2-pyridylazo)-2-naphthol (PAN) technique according to Goto et al. (1977) after adjusting pH of the sample to 7 with a 6 mol L^-1 NaOH solution. Both manganese and total iron were quantified spectrophotometrically at a wavelength of 562 nm, with the MDLs 20 µmol kg^-1 for both metals.

The determination of highly reactive iron (Fe_HR) in ~0.5 g of sediment samples was performed following the procedure published by Poulton and Canfield (2005), in which Fe_HR is defined as the iron extracted in the five steps of sequential extraction and sulfur-bound iron (Fe_SB), which is calculated as a sum of AVS and CRS divided by 2. The first step of the sequential procedure was extraction of ion-exchangeable iron (Fe²⁺) using a magnesium chloride solution (Fe_chl). No iron was released in this step from all analyzed samples. This step was followed by the treatment with a sodium acetate solution to extract carbonate-associated iron (Fe_ac) typically in the form of siderite and ankerite. Subsequently, “easily reducible” iron (hydr)oxides (Fe_hydr) including ferrihydrite and lepidocrocite were extracted with hydroxylamine hydrochloride solution. The fourth step employed a sodium dithionite solution (Fe_dith) to remove “reducible” oxides, which include goethite, hematite, and akaganéite. The final extraction step involved the use of ammonium oxalate solution (Fe_ox) to dissolve poorly crystalline iron oxides, e.g., magnetite. Concentrations of the dissolved iron were measured by spectrophotometric technique according to Stookey (1970). The MDL was 20 µmol kg^-1 for each fraction.

For the analysis of isotopic composition of sulfur in CRS, zinc sulfide formed during hydrogen sulfide distillation was converted to silver sulfide by the addition of an excess of silver nitrate solution. Silver sulfide samples were stored for at least one week in the dark, washed three times with 10 mL of Milli-Q water, stored overnight in 1 mL of 1 mol L^-1 NH₄OH solution, and washed three times with 10 mL of Milli-Q water. The samples were dried at 50°C to a constant weight.

For the oxygen isotope analysis of sulfate, 2 g of the sediment was shaken with 20 mL of 10% NaCl solution for 24 hours. After centrifugation and filtration through a 0.22 µm filter, 3 mL of solution was mixed with 1 mL of 1 mol L^-1 BaCl₂ solution stored at 4°C for 48 hours, centrifuged, washed with MQ water and dried at 70°C for five days. Barium sulfate was purified according to Bao (2006), e.g., 14 mg of the sample was dissolved in 7 mL of diethylenetriaminepentaacetic acid solution (40 g NaOH and 19.7 g diethylenetriaminepentaacetic acid in 1 L solution) followed by acidification with 10.2 M HCl to pH 3-4. After one hour, the samples were centrifuged and decantated to remove excess liquid. The samples were washed three times with Milli-Q water and dried at 70°C for five days.

Approximately 400 μg of barite was crushed and placed in tin capsules for analysis. The sulfur isotope composition of the barite was measured after combustion at 1020°C in an EA Isolink (Thermo Fisher) and continuous helium flow to a Delta V mass spectrometer in the Godwin Laboratory for Paleoclimate Research at the University of Cambridge. The samples were bracketed with blank boats due to extensive memory effects in the EA Isolink. Samples were bracketed with repeat analysis of NBS127 (20.3‰) and SO-6 (-34‰) which was used for drift correction. Analytical precision based on repeat analysis of the standards and replicates of several samples was 0.2‰ (2σ).

The oxygen isotope composition of sulfate (as barite) was measured with a high temperature elemental analyzer coupled to Sercon 2022 isotope ratio mass spectrometer. Analytical standards IAEA-601 (23‰ VSMOW), IAEA-SO-5 (12‰ VSMOW), and IAEA-SO-6 (-11‰ VSMOW) were used for calibration, and the analytical error was less than 1.0‰ for δ¹⁸O values.

The prolonged storage of sediment cores at the Bremen Core Repository, at temperatures 4-6°C, can lead to the oxidation of pyrite and Fe(II) minerals, resulting in the formation of fresh, highly reactive Fe(III) (hydr)oxides. In order to quantify fresh reactive ferrihydrite and to evaluate the quality of the samples a method proposed by Boyko et al. (2022) was applied. Dry samples (0.5 g) were mixed with 10 mL of a 0.5 mol L^-1 HCl solution, a condition noted for neither reducing ferric ions nor oxidizing ferrous ions (Lovley and Phillips, 1986). The samples were shaken for one hour followed by centrifugation at 4000 RPM for 10 minutes. Two 1 mL subsamples of the supernatant were used for iron quantification: the first subsample was treated with 100 µL of ferrozine, and the second was treated with both 100 µL of ferrozine and 100 µL of 1.4 mol L^-1 hydroxylamine hydrochloride in 2 mol L^-1 HCl, for Fe²⁺ and total dissolved iron quantification, respectively. The concentration of “fresh” ferrihydrite in the samples was calculated by subtracting the iron content measured in the first subsample from that in the second. However, it is crucial to acknowledge, as noted by Boyko et al. (2022), that this methodology does not distinguish between ferrihydrite which was present in the core during its retrieval and ferrihydrite which could be developed during its subsequent storage.

The contents of AVS in the sediments were extremely low, <0.22 mmol kg^-1 of dry sediment in all samples (Figures 2A, B) with the exception of one sample at Site 546 from DSDP Leg 79 which had a content of 0.55 mmol kg^-1 of dry sediment. The CRS contents were low as well. In all samples, the CRS content was <3 mmol kg^-1 of dry sediment except in the 1.33-1.45 m interval of ODP 157–954 in which the CRS content was up to 13 mmol kg^-1 (Figures 2C, D). The average Mn_TOT was relatively low in the four cores which were sampled closer to the shore, 52–86 mmol kg^-1 of dry sediment for DSDP 79-546, ODP 157-954, DSDP 41-368, and DSDP 41-366. For the two sites further from shore, ODP 108–665 and ODP 157-952, the average total Mn_TOT content was >100 mmol kg^-1 of dry sediment (Figures 2E, F). The average total iron (Fe_TOT) content was relatively high (>400 mmol kg^-1 of dry sediment) in three cores, which were sampled at distances closer to shore as well as ODP 108-665 (Figures 2G, H). In the two other cores, the average Fe_TOT content was <330 mmol kg^-1 of dry sediment. The average Fe_HR contents were higher in the cores with high Fe_TOT contents: >150 mmol kg^-1 of dry sediment for DSDP 79-546, ODP 157-954, DSDP 41-368, ODP 108-665, and < 128 mmol kg^-1 dry sediment in two other cores (Figures 2I, J). The Fe_SB pool consisted mostly of CRS (Figures 2C, D, K, L).

The most abundant of the Fe_HR pools in four of the six cores (ODP 157-954, DSDP 41-368, DSDP 41-366, ODP 108-665) was Fe_dith, which comprises goethite, hematite, and akaganéite. This is the most abundant group of minerals in the Saharan dust collected at the Gulf of Aqaba and in its sediments (Boyko et al., 2019, 2022) (Figure 3). On the other hand, in some sections of these cores, the contents of Fe_hydr, which accounts for the most reactive Fe(III) phases, ferrihydrite and lepidocrocite, were higher than those of Fe_dith. In cores DSDP 79–546 and ODP 157-952, Fe_hydr was the most abundant of the Fe_HR pools in the studied core. The third most abundant of the Fe_HR pools (and even the second in cores DSDP 79–546 and ODP 157-952) was the least reactive Fe_ox. In core ODP 157-952, sampled at the site with the furthest distance from the shore of Africa, Fe_ox fraction comprised 35 ± 11% (± 1σ) of Fe_HR.

The pyrite contents in all samples were too low for isotopic analysis. On the other hand, valuable information is gained by looking at the sulfur and oxygen isotopic composition of pore fluid sulfate. In Figures 4A, B, the sulfur and oxygen isotopic composition of sulfate are presented, and δ¹⁸O_SO4vs. δ³⁴S_SO4 are plotted in Figure 4C, with a schematic of microbial sulfate reduction in Figure 4C. The isotopic composition of sulfate does not show significant variability with depth for cores DSDP 41–366 and DSDP 79-546. In these cores, the δ³⁴S_SO4 is in the 19-22‰ range, close to the seawater value of 21.15 ± 0.15 ‰ (Johnston et al., 2014). In ODP 157–954 core, the δ³⁴S_SO4 is slightly lower and more variable – in the 15 to 20‰ range. The oxygen isotope composition of sulfate reveals the following trends: DSDP 41–366 and DSDP 79–546 cores do not show significant variability, with δ¹⁸O_SO4 ranging between 8-11‰ range, close to the value for the Atlantic Ocean, 9.06‰ (Longinelli and Craig, 1967). In ODP 157-954, the δ¹⁸O_SO4 are in the range of 12-15‰.