Introduction

Methanogenesis in anoxic marine and lacustrine systems produces substantial amounts of methane (~370 Tg yr−1) during organic matter fermentation1,2,3. Sulphate is available in high concentrations in modern seawater (28 mM). In methane-rich marine environments, most methane is consumed at the expense of sulphate through sulphate-driven anaerobic oxidation of methane (SD-AOM)2,4, which greatly reduces seabed methane release. In contrast, lacustrine sedimentary environments (e.g., lakes) typically have low-sulphate concentrations5 and, thus, represent the largest (~90%) natural atmospheric methane source3. This is because sulphate is the most dominant electron acceptor for AOM6. Although it has been suggested that AOM in low-sulphate freshwater environments is coupled to nitrate, manganese or iron reduction7,8,9, methane consumption via these pathways is largely regarded as inconsequential3. Experimental evidence indicates that high SD-AOM rates can occur under low-sulphate levels, rendering possible high methane consumption even in low-sulphate lacustrine sediments10,11. Sulphate availability is, thus, a key factor that affects methane consumption in lacustrine sediments10. Methane-rich lacustrine environments are commonly located along coastlines, resulting in changes of sulphate availability caused by sea-level change that affect the mode and degree of methane consumption12,13. In marine environments, SD-AOM has been recognised frequently in the geological record from highly 13C-depleted seep carbonates, for example, in the middle Palaeozoic ocean14. However, geological evidence for methane consumption from lacustrine environments is scarce15, which limits reconstruction of ancient methane carbon cycling.

Lake Pannon was a long-lived, large lake on the southern margin of the European plate and was the largest European late Cenozoic non-marine biodiversity hotspot16,17. The lake formed after a glacio-eustatic sea-level drop at ~11.6 Ma (Fig. 1), which led to final disintegration of the central and southeastern European Paratethys Sea17,18,19. During its maximum extent between 10.5 and 10.0 Ma, it covered around 233,500 km2, attaining more than half the size of the modern Black Sea with a greatest depth of around 1000 m17. Biota and sedimentary features indicate that Lake Pannon was initially brackish and that it freshened gradually following isolation from the Paratethys Sea16,18,19. Its marine heritage might have impacted lake water sulphate concentrations and methane-related carbon cycling during early-stage lake evolution. However, little is known about methane formation and consumption in the lake due to the scarcity of suitable geochemical data20,21,22. As an authigenic sulphide mineral, pyrite derived from SD-AOM commonly carries characteristic geochemical signatures23,24,25, which can constrain the sediment biogeochemistry of Lake Pannon.

Fig. 1: Location of the study site and summary stratigraphy of the studied cores.
figure 1

a Position of the studied cores on the western margin of Vienna Basin; b Lake Pannon during the early Pannonian around 11.3 Ma18; c lithological logs of Pannonian lower PA1 parasequence from three drill cores; arrows indicate distances between drill sites26. ba. barren, m. molluscs, os. ostracods.

Full size image

Recently, an up to 4-m-thick pelite unit containing abundant tubular pyrite aggregates was detected in three adjacent boreholes in the Vienna Basin (Fig. 1). These formed at ~11.3 Ma during early-stage Lake Pannon26,27, when the lake surface area was smaller with large islands18 (Fig. 1a). During this period, bottom waters in Lake Pannon were well oxygenated down to the storm wave base in coastal sublittoral regions28. Deeper sublittoral regions of Lake Pannon had low oxygenation, which was only interrupted on millennial scales by increased bottom water oxygenation, as indicated by rapid but short-lived dreissenid bivalve settlements29. The tubular pyrite structures have not been reported from the well-studied younger lake sediments30,31, and likely record early lake-stage biogeochemical processes. We present geochemical, mineralogical, and magnetic evidence to assess the origin of the pyrite aggregates and to test for SD-AOM signals to evaluate sulphur and carbon cycling in Lake Pannon.

Geological background and samples

The study area is located within the Vienna Basin, which is a rhombic extensional basin that strikes roughly SW-NE from Lower Austria in the SSW to the Czech Republic in the NNE (Fig. 1a). The Vienna Basin basement is formed by Alpine-Carpathian nappes, with Neogene basin fill that reaches a maximum thickness of 5500 m32,33. Three 60-m long cores, RKB 5260 (N 48.221543°, E 16.345415°), RKB 5300 (N 48.220493°, E 16.341733°) and RKB 7260 (N 48.196074°, E 16.352512°) (Fig. 1) were studied27, which penetrated lower Pannonian lacustrine sediments (20.5–55.5 m in RKB 5260; 19.5–31.5 m in RKB 5300; and 9.5–60.0 m in RKB 7260, respectively). The cored sediments have uniform lithology and comprise clay and silty clay with a few intercalations of silty sand. Based on sedimentological, palaeontological, and geophysical data for these cores (Fig. 1c), two parasequences (Pannonian PA1 and PA2) were identified26. The lower PA1 parasequence formed in medium-deep sublittoral muddy lake bottom environments below the storm wave base during the first rapid transgression of Lake Pannon (Fig. 1b). Barren intervals indicate periods of severe bottom water anoxia at the boundary between the PA1 and PA2 parasequences, which were interrupted by short phases of biotic resettlement by Ostracoda and thin-shelled lymnocardiid bivalves26. Within lower parasequence PA1, a 3–4-m-thick interval contains abundant iron sulphide aggregates in the cores (RKB 5260: 46–43 m, RKB 5300: 25–22 m, RKB: 7260 45–41 m; Fig. 2). Pyrite tubes are accompanied by ostracod mass occurrences, with subordinate molluscs26 (Fig. 1c). Based on the sedimentation rates26, the 3–4-m-thick interval with tubular pyrite aggregates likely formed over ~4–6 kyr.

Fig. 2: Morphologies, structures, and mineralogical data for authigenic minerals.
figure 2

a, b Tubular pyrite aggregates. c Aggregate with pyrite framboids. d, e Resin-impregnated polished sections of pyrite framboids (Py) with overgrowths (O). f Goethite layer (Go) surrounding pyrite framboids. g, h Irregular granular greigite aggregates (Gr) and coexisting goethite. i, j Resin-impregnated polished sections of greigite aggregates. The large range of BSE intensities was caused by the occurrence of goethite. k Co-occurring framboidal pyrite and greigite nanocrystals. Fine particles are greigite and coarser/brighter particles are pyrite. (a, g, h, k) in optical microscope mode and (b–f, i, j) in SEM-BSE mode. l X-ray diffraction analyses of selected authigenic minerals from different depths.

Full size image

Results

Mineralogy and morphology of pyrite and greigite aggregates

To explore the nature of the iron sulphide aggregates, samples from core RKB 5260 were collected here for further analysis. Pyrite is the main phase in iron sulphide aggregates picked from sediment in core RKB 5260 (mainly from 43.0 to 47.2 m; Fig. 2 and Supplementary Figs. 1 and 2). Most aggregates are black and tubular, with similar habits and textures (Fig. 2a, b). The pyrite aggregates consist of clustered framboids with variable sizes (Figs. 2c and 3b). In many parts of the studied interval, framboidal pyrite aggregates are overgrown by secondary pyrite. These overgrowth layers fill the space between framboids (Fig. 2d, e). In some aggregates, oxidation products (e.g., goethite) occur as secondary layers around pyrite aggregates (Fig. 2f). Magnetic greigite aggregates occur mainly from 41.0 to 42.8 m in core RKB 5260. Most of the greigite occurs as black, irregular granular particles that commonly coexist with brownish goethite (Fig. 2g, h). Clusters of (i) fine greigite microcrystals and (ii) acicular crystals occur commonly within the aggregates (Fig. 2i, j). Greigite nanocrystals also occasionally coexist with pyrite framboids (Fig. 2k). X-ray diffraction analysis further indicates the occurrence of minor mackinawite in the greigite aggregates (Fig. 2).

Fig. 3: First-order reversal curve (FORC) diagrams for typical bulk sediments from different depths in core RKB 5260.
figure 3

a–d Strongly magnetostatically interacting medium-coercivity single domain component, which is typical of greigite-bearing sediments from sulphidic and methanic diagenetic environments36. e, f Lower coercivity authigenic SP/SD greigite35,36 in pyrite-dominated intervals. FORC diagrams were processed using the FORCsensei algorithm88, which searches 1350 FORC models using all combinations of VARIFORC smoothing parameters89 to produce optimal FORC distributions in which noise is smoothed without over-smoothing the underlying signal. VARIFORC parameters for each diagram are: a sc,0 = 2, sc,1 = 2, su,0 = 2, su,1 = 2, λ = 0.00, and ψ = 0.64; b sc,0 = 2, sc,1 = 2, su,0 = 2, su,1 = 2, λ = 0.00, and ψ = 0.66; c sc,0 = 2, sc,1 = 2, su,0 = 2, su,1 = 2, λ = 0.04, and ψ = 0.51; d sc,0 = 2, sc,1 = 2, su,0 = 2, su,1 = 2, λ = 0.00, and ψ = 0.61; e sc,0 = 2, sc,1 = 2, su,0 = 2, su,1 = 2, λ = 0.00, and ψ = 0.54; f sc,0 = 3, sc,1 = 5, su,0 = 3, su,1 = 5, λ = 0.16, and ψ = 0.57.

Full size image

Magnetic susceptibility (χ) and FORC diagrams

In core RKB 5260, higher χ values (up to 0.30 × 10−6 m3 kg−1) coincide with the presence of greigite aggregates (above 42.8 m), particularly from 41.0 to 41.6 m. In contrast, χ has low and constant values of ~0.09 m3 kg−1 in intervals with pyrite aggregates (from 43.0 to 47.2 m). To characterise the magnetic domain state and magnetostatic interactions among magnetic particles, first-order reversal curves (FORCs)34 were measured for bulk samples from intervals with different χ values (Fig. 3). FORC diagrams are similar for greigite-bearing sediments with high χ values (>0.10 × 10−6 m3 kg−1; Fig. 3). They reveal a magnetostatically interacting stable SD greigite component with high coercivity (Fig. 3a–d), which is typical of greigite-bearing sediments from sulphidic and methanic diagenetic environments35,36 and are consistent with the presence of greigite aggregates. In pyrite-dominated intervals with low χ values, the FORC diagrams (Fig. 3e, f) indicate the presence of lower contents of authigenic SP/SD greigite with lower coercivity35,36.

Chromium reducible sulphur (CRS) content and iron speciation

Total CRS contents in core RKB 5260 are generally low (<0.25 wt.%) except for four distinct peaks at 43.6 m, 44.4 m, 45.2 m and 47.2 m (Fig. 4). Among the three non-sulphur-bound reactive iron components (Feacetate, Fedithionite, Feoxalate), Feacetate is the main iron pool in all samples. These iron components have similar trends despite variable contents, with distinctive peaks at depths of 42.2 m, 44.4 m, 45.2 m and 47.2 m (Fig. 5). Assuming that CRS is derived totally from pyrite (1:2 stoichiometry of Fe:S), pyrite-bound iron (Fepy) is obtained from the CRS content of each sample. The total reactive iron (FeHR) is the sum of Feacetate, Fedithionite, Feoxalate and Fepy. The extent of pyritization (Fepy/FeHR ratio) is typically lower than 0.4 throughout the studied interval (Fig. 5).

Fig. 4: Geochemical data for core RKB 5260.
figure 4

a Chromium reducible sulphur (CRS) content. b Zr/Ti and c Zr/Ti ratios. d Organic matter content. e Extent of pyritization (Fepy/FeHR ratio; FeHR = Feacetate + Fedithionite + Feoxalate + Fepy).

Full size image
Fig. 5: Magnetic susceptibility, framboid size distribution and geochemical data for core RKB 5260.
figure 5

a Mass magnetic susceptibility of bulk sediment. b Framboid size distributions in box-and-whisker plots, where boxes extend from quartile Q = 0.25 to 0.75 and include 50% of data. The vertical line in the box is the median; lines at the left- and right-hand sides are minimum (Q = 0.00) and maximum (Q = 1.00) values. c Chromium reducible sulphur (CRS) content. d Ba/Al and e Mo/Al ratios. f Iron species distribution in the sediment. g δ13C of total inorganic carbon in bulk sediments. The shaded zone refers to greigite-dominated sediments.

Full size image

Sulphur isotopic compositions

Within greigite-bearing sediments (i.e., 41.0–42.8 m, core RKB 5260), δ34SCRS values are relatively constant at +3.5‰ (n = 10) (Fig. 6). Corresponding Δ33SCRS values vary over a narrow range close to 0‰ (Fig. 7). From 43.0 to 47.2 m, the multiple sulphur isotope compositions of CRS (pyrite) are characterised by large δ34SCRS (from −15.2 to +7.4‰) and Δ33SCRS variations (−0.04 to +0.03‰, n = 22) (Fig. 7). δ34Spyrite values of hand-picked pyrite aggregates range from −11.7 to +8.6‰ (Fig. 6).

Fig. 6: Sulphur isotope data for core RKB 5260.
figure 6

a Sulphur isotopic composition of CRS (δ34SCRS) and hand-picked pyrite tubes (δ34SPyrite) in bulk sediments and in-situ pyrite SIMS δ34SSIMS. b Distribution pattern of δ34SSIMS variation of each pyrite aggregate from different depths.

Full size image
Fig. 7: Variation of δ34S and Δ33S values for chromium reducible sulphur (CRS) in pyrite- and greigite-dominated depths from core RKB 5260.
figure 7

The yellow oval (labelled ‘OSR-pyrite’) represents typical sulphur isotope compositions of pyrite derived from organoclastic sulphate reduction (OSR) in shallow sediments of modern marine seep environments45,49. All data for pyrite-dominated (orange squares; 43.0–47.2 m) and greigite-dominated samples (green circles; 41.0–42.8 m) align along the mixing zone between OSR-pyrite and modern seawater sulphate end members. Error bars are from standard deviation (SD) for Δ33S values.

Full size image

Two pyrite varieties—framboids and overgrowth layers—from various depths were targeted for SIMS analysis (Fig. 6). However, complex composite pyrite aggregate textures proved difficult to resolve even for SIMS analysis with 15-μm spot size. To avoid mixed signals from different paragenetic phases during analysis, only larger pyrite aggregates (>20 μm) were measured (Supplementary Figs. 3 and 4). In some cases, possible admixture between overgrowths and framboids cannot be excluded24; such slightly mixed spot analyses are classified here as pyrite overgrowths (Supplementary Figs. 3 and 4).

δ34SSIMS values for the two pyrite varieties are highly variable for each pyrite aggregate from the same depth and through the studied interval (Fig. 6). A total spread of 106‰ was obtained for pyrite δ34SSIMS values, ranging from −58.7 to +45.3‰ (Fig. 6). Larger pyrite framboids and extensive pyrite overgrowths are abundant in pyrite-dominated intervals (from 43.0 to 47.2 m). In these intervals, δ34SSIMS values of intimate pyrite overgrowth layers tend to be higher than for early pyrite framboids. The δ34SSIMS difference between framboids and overgrowths can be up to 30‰ in a single aggregate. Only a few pyrite aggregates were chosen for sulphur isotope measurements from each sample; therefore, the obtained δ34SSIMS values might not reflect the entire variability of δ34SSIMS values. This circumstance explains some mismatches between δ34SCRS and δ34SSIMS values.

Trace element contents of bulk sediments

Zr/Ti, K/Al, Ba/Al, Mo/Al and P/Al ratios are shown in Figs. 4 and 5. Zr/Ti and K/Al ratios are constant through the studied horizons corresponding to around 185 (*10−4) and 0.28, respectively. In contrast, Ba/Al, Mo/Al and P/Al ratios vary with peaks in some CRS rich layers (e.g., 45.6 and 47.2 m).

Total inorganic carbon contents and carbon isotopic compositions

Throughout the studied stratigraphic interval, sedimentary total inorganic carbonate (TIC) has low contents of 1.3–2.8 wt.%. δ13CTIC values for bulk sediments range from −0.2 to +3.4‰ (n = 32; Fig. 5), with some peaks coinciding with layers enriched in Feacetate.

Discussion

Evidence for SD-AOM in Lake Pannon sediments

Sedimentary pyrite formation is impacted by sulphate reduction processes such as organoclastic sulphate reduction (OSR)37 and sulphate-driven anaerobic oxidation of methane (SD-AOM), and by the availability of reactive iron compounds38,39. Pyrite formation is thought to be controlled either by (i) the polysulphide pathway40 or (ii) the hydrogen sulphide pathway41. For both pathways, aqueous iron (II) monosulphide or surface-bound species on minerals like mackinawite can react with dissolved sulphide or polysulphide to promote pyrite formation (see Rickard39 for a review). It has also been suggested that pyrite can form directly under Fe(II)-excess conditions, such as in sedimentary environments with abundant ferric (oxhydr)oxides42.

The degree of pyritization (Fepy/FeHR) throughout the studied interval ranges mainly from 0.1 to 0.5, which is below the threshold (0.8) for euxinic conditions43. This indicates that reactive iron is abundant throughout the studied interval. Therefore, pyrite enrichment in multiple horizons was unlikely to be controlled by reactive iron availability under sulphidic conditions. Moreover, pyrite tubes are accompanied by massive ostracod occurrences26 (Fig. 1c), which suggest oxic water column conditions. Input of detritus was constant (indicated by Zr/Ti and K/Al)44 and organic matter contents do not vary much (41.0–47.2 m; Fig. 4). This cumulative evidence argues against a mode of pyrite formation that reflects changes of water geochemistry resulting from either a change of detrital input or productivity.

Pyrite aggregates in the studied sediments mostly have tubular shapes (Fig. 2) with late-stage overgrowths, which is reminiscent of aggregates forming in modern methane seeps and gas-hydrate environments24,45,46,47. The tubular shape likely represents sedimentary micro-channels created by pressurised advection or diffusion of methane-rich fluids24. The tubular void spaces were later filled by pyrite. Intervals with high pyrite contents (Fig. 4a), therefore, likely result from enhanced SD-AOM and point to the positions of palaeo-sulphate methane transition zones (palaeo-SMTZs)24,45. Extensive overgrowths on framboids suggest that SD-AOM led to fluids supersaturated in dissolved sulphide, which favoured pyrite framboid and overgrowth formation24,39.

Throughout the studied interval (41.0–47.2 m, core RKB 5260), framboids commonly have low δ34SSIMS values (as low as −59‰; Fig. 6). This points to OSR with preferential microbial 32S-sulphate uptake24,45,48,49. In contrast, higher bulk sediment δ34SCRS (from −15.2 to +7.4‰) suggests admixture of 34S-enriched pyrite, probably derived from SD-AOM24,45,46,50. Considering the presence of two pyrite end members derived from OSR and SD-AOM, the pyrite sulphur isotopic compositions should represent mixing between these end members45,46,49. Pore water sulphate is thought to be converted quantitatively to sulphide within the SMTZ during SD-AOM51, with isotope mass balance resulting in the δ34S values of hydrogen sulphide leaning toward the original isotopic composition of dissolved sulphate50. All analysed pyrite samples have intermediate values between those of OSR-derived pyrite and modern seawater sulphate (Fig. 7). Thus, pyrite derived from SD-AOM in Lake Pannon likely reflects an aggregated sulphate sulphur isotopic composition similar to that of modern seawater sulphate (δ34S = +21.24‰ and Δ33S = +0.050‰)52. This mixing scenario is also supported by the multiple sulphur isotope pattern obtained by SIMS analysis (Supplementary Fig. 5). Lake Pannon pyrite has extremely wide-ranging δ34SSIMS values (Fig. 6). Pyrite δ34SSIMS commonly increases from framboids to overgrowth layers within the same aggregate. For example, for the 43.4 m and 45.2 m intervals, the wide range of pyrite δ34SSIMS values indicates continuous and finally near-complete dissolved sulphate consumption near the SMTZ following Rayleigh-type distillation24,45. However, pyrite derived from SD-AOM can also have low δ34S due to enhanced diffusive sulphate replenishment derived from seawater or oxidative sulphur cycling53. Authigenic 13C-depleted seep carbonates record SD-AOM54 and can co-occur with 34S-enriched pyrite at SMTZs49,55; yet no 13CTIC depletion is observed at the identified palaeo-SMTZs in Lake Pannon. Such absence of 13C-depleted carbonate is common at the SMTZ24,56,57, and is a function of the local sedimentary environment (e.g., pH). Methanogenesis typically occurs below the SMTZ and results in higher δ13C values of dissolved bicarbonate. Therefore, positive δ13CTIC values likely reflect upward diffusion of dissolved inorganic carbon from methanogenesis below the SMTZ58. In addition to high δ13CTIC values, the same horizons with relatively high δ13CTIC values have elevated Feacetate contents (Fig. 5), which suggest Fe(II)-rich environments. These patterns are consistent with upward diffusion of Fe(II) derived from iron reduction in the methanic zone7,59. Finally, given the succession of biogeochemical processes in space and time and the presence of steep geochemical gradients in sedimentary environments where methane is present, some of the observed geochemical signatures are likely to reflect mixed signals observed in the same horizon.

Greigite is a metastable ferrimagnetic iron sulphide mineral that is common in lake sediments and is considered to record lake geochemistry60,61,62. Greigite is also present in marine sediments and has been reported widely from methane seeps and gas-hydrate-bearing environments63,64,65,66. Greigite occurrences can represent sulphidization fronts around the SMTZ, where it transforms into pyrite upon supply of excess hydrogen sulphide63,65. In the Lake Pannon sediments, greigite tends to occur in horizons with limited pyrite (i.e., above 43.0 m). In a Δ33S/δ34S plot, all greigite-dominated samples have more negative Δ33S values than expected for OSR-derived sulphide forming in a seawater sulphate-dominated environment (quadrant II). Values for greigite are closer to those for seawater sulphate (Fig. 7) and are more consistent with the pattern for iron sulphides derived from SD-AOM45,49,53,67. In contrast to the wide range of δ34SCRS values for pyrite-dominated depths, δ34SCRS values from greigite-dominated sediments fall within a narrow range (Fig. 7), which suggests that greigite formation was rapid and associated with hydrogen sulphide diffusion from the SMTZ. FORC diagrams for sediments from greigite-bearing intervals (Fig. 3) are typical of authigenic greigite from methane-rich sedimentary environments36, which is consistent with an origin from SD-AOM.

Variable redox conditions controlled by seepage dynamics

In modern marine seep environments, spatiotemporal methane flux variations commonly lead to vertical SMTZ movement, as evident from sedimentary geochemical signatures49,68. Elevated Mo/Al ratios occur at some of the inferred palaeo-SMTZs of early Lake Pannon sediments (e.g., 47.2 m; Fig. 5e). Along modern continental margins, dissolved Mo in pore water is mainly derived from seawater and commonly decreases with depth due to scavenging by authigenic iron sulphide formation in the sediment. Thus, Mo enrichment at methane seeps typically reflects enhanced Mo fixation in authigenic iron sulphides at shallow SMTZs25,68,69. Therefore, pyrite-independent Mo/Al ratios of the background sediment at the inferred SMTZs suggests changes in the depth of the SMTZ, probably controlled by methane flux. Likewise, Ba/Al peaks at some SMTZs of today’s marine sediments imply authigenic barite formation just above the SMTZ, which is also impacted by SMTZ mobility70. Overall, in addition to the presence of tubular pyrite aggregates and greigite formation, evidence for multiple palaeo-SMTZs (Fig. 5) suggests the former prominent presence of methane seepage in Lake Pannon. Although mechanisms controlling methane flux at ~11.3 Ma are not well-defined, it is likely that the low lake water level (~200 m depth; Harzhauser et al., unpublished seismic data) triggered a methane seepage increase in the lake71.

Within the SMTZ, hydrogen sulphide released from SD-AOM commonly leads to sulphidic dissolution of reactive iron72. Thus, iron (oxyhydr)oxide contents and magnetisation can be markedly reduced at the SMTZ35,72. However, we find no such reactive iron decrease (e.g., Feacetate, Fedithionite, Feoxalate) at the inferred palaeo-SMTZs (SMTZ1-4; Fig. 5). Na-acetate solution can extract substantial Fe(III) from amorphous iron (oxyhydr)oxides73,74, where Feacetate is from iron carbonates and poorly crystalline Fe(III) minerals, whereas Fedithionite and Feoxalate include crystalline haematite, goethite, or magnetite73,75. High reactive iron contents at the inferred palaeo-SMTZs can, therefore, be caused by iron sulphide oxidation after burial. However, this fails to explain low reactive iron contents at SMTZ1 (i.e., 43.4 m) where pyrite is abundant (Fig. 5). Alternatively, vertical SMTZ movement at dynamic seep sites and resultant changing redox conditions could have caused iron sulphide oxidation within a former SMTZ and secondary iron (oxyhydr)oxide formation76.

We propose the following scenario to explain the increased reactive iron content at three palaeo-SMTZs (Fig. 8). Sulphide formation during SD-AOM caused iron sulphide accumulation at or near a shallow SMTZ when methane fluxes were high, which was possibly caused by reduced hydrostatic pressure during lake-level lowstands. When seepage diminished, downward moving, seawater-derived fluids promoted iron sulphide oxidation at the former SMTZ, and secondary iron (oxyhydr)oxide formed76. Remaining iron sulphide would maintain the primary δ34S patterns because sulphur isotope effects for the involved oxidative processes are small77. Resurgent high methane fluxes and positioning of the new SMTZ at shallower depths, possibly in combination with rapid burial, subsequently promoted iron (oxyhydr)oxide preservation without further alteration by sulphidization76.

Fig. 8: Simplified scenario for dynamic sulphur and iron cycling in sediments under the influence of variable seepage intensities.
figure 8

a During high seepage activity, the sulphate methane transition zone (SMTZ) is located near the sediment surface. Within the SMTZ, sulphate-driven anaerobic oxidation of methane (SD-AOM) results in hydrogen sulphide release and pyrite formation. Upward- and downward-diffusing hydrogen sulphide leads to greigite formation. b After a decline of seepage intensity, downward moving oxidising fluids promote iron sulphide oxidation at the former SMTZ, which leads to secondary iron (oxyhydr)oxide formation. c Resurgence of high methane flux and/or rapid sediment burial promote iron (oxyhydr)oxide preservation during burial into a methanic environment without further alteration by sulphidization.

Full size image

Sulphate-rich water in Lake Pannon

In marine sulphate-rich environments, most upward-diffusing methane is consumed at the expense of sulphate during SD-AOM2. Although low-sulphate freshwater environments might have high AOM rates10, sulphur-bearing mineral formation (e.g., pyrite, gypsum) is limited in lake deposits. Multiple horizons with abundant tubular pyrite, as documented here, have also been observed in modern marine methane seep areas49. Sulphur isotopic compositions of pyrite at palaeo-SMTZs from Lake Pannon resemble observations from authigenic, SD-AOM-derived pyrite in modern seep environments, where seawater sulphate is the main sulphur source for pyrite formation45,46,49. It is, thus, likely that lake water sulphate in Lake Pannon had a similar sulphur isotopic composition as contemporary seawater sulphate. Although the initial sulphate concentration remains unknown, abundant tubular pyrite derived from enhanced SD-AOM indicates that dissolved sulphate was a major lake water anion. Therefore, AOM coupled to nitrate, iron, and manganese reduction can be excluded as main pathways for methane consumption7,8,9. Tubular pyrite aggregates have been reported only from lower Pannonian deposits and not from the well-studied younger Pannonian sediments. This suggests that Lake Pannon initially had a marine sulphate signature, reflecting its Central Paratethys Sea heritage. Likewise, during isolation of Lake Pannon from the Paratethys Sea16,18,19, Vienna Basin sediment could still have contained sufficient organic matter for methane production at depth. The still high sulphate pool and enhanced methane flow would have, thus, stimulated SD-AOM and enhanced sedimentary methane consumption. Subsequent lake freshening led to a dissolved sulphate decrease in the water column. Although SD-AOM can also occur at lower sulphate levels, pyrite formation and preservation might be impeded by the reduced dissolved sulphate concentration. We, therefore, make a case for use of authigenic pyrite to constrain carbon cycling in methane-rich environments and to assess its effects on climate.

Conclusions

Distinct horizons with abundant tubular iron sulphide aggregates are reported here from upper Miocene deposits of the early stage of the ancient European mega-lake, Lake Pannon. Geochemical, mineralogical, and magnetic analyses indicate that sedimentary iron sulphide formation resulted largely from SD-AOM.

  1. (i)

    Multiple sediment layers from early-phase Lake Pannon with high pyrite contents represent palaeo-SMTZs, as supported by multiple proxies (e.g., Ba/Al, Mo/Al). δ34SSIMS values of pyrite are highly variable (−58.7 to +45.3‰; n = 223), and pyrite overgrowths on framboids typically have higher δ34SSIMS values than framboids. Sulphur isotopic compositions of bulk pyrite range from −15.2 to +7.4‰ for δ34SSIMS, and from −0.04 to +0.03‰ for Δ33S, which represent signatures of the combined effects of OSR and SD-AOM.

  2. (ii)

    Sulphur isotopic compositions of bulk greigite range from +1.3 to +5.8‰ for δ34SCRS, and from −0.03 to 0‰ for Δ33SCRS. These narrow ranges suggest rapid greigite formation, likely caused by hydrogen sulphide diffusion from SMTZs. FORC diagrams for authigenic greigite-bearing samples are consistent with those reported from marine methane seeps and gas-hydrate-bearing environments.

  3. (iii)

    Iron (oxyhydr)oxides are found to be enriched in palaeo-SMTZs, along with abundant iron sulphides. The former are interpreted to have resulted from iron sulphide oxidation at shallow depths at times when methane flux was reduced.

Our data are consistent with intensified methane consumption at the expense of sulphate in early Lake Pannon. Although precise sulphate levels during early lake stages are unknown, its marine Paratethyan heritage favoured SD-AOM as a key biogeochemical process that governed carbon cycling in the methane-rich sediments of Lake Pannon.

Methods

Sediment samples from a 6-m-thick interval of core RKB 5260 (41.0–47.2 m) were collected at 20-cm intervals for this study. Sediment aliquots were treated with diluted hydrogen peroxide for several hours and were then washed with tap water and sieved through a set of standard sieves. Authigenic black iron sulphide aggregates (pyrite and greigite) were hand-picked from the coarse fraction under a binocular microscope. Greigite-bearing aggregates were extracted with a rare earth hand-magnet. For petrographic observations, representative iron sulphide aggregates were examined with a ΣIGMA field emission SEM after carbon coating at the School of Earth Science and Engineering, Sun Yat-Sen University, China.

Mineral composition identification was performed at the Norwegian Center for Mineralogy, Oslo. Mineral compositions were analysed via micro-XRD after morphological observations. Targeted particles were scraped from epoxy-mounted samples with a knife or were picked from sediments. Targeted ~100 µm particles were picked for morphological observations. Samples were attached to a fibre loop stage and aligned with the X-ray beam of a Rigaku XtaLAB Synergy-S single-crystal X-ray diffractometer (Mo Kα). X-ray diffraction patterns were measured by Gandolfi move mode for powders at 50 kV and 1 mA. The detector distance was 65 mm and the exposure time was 30 s for each detector theta position. Major phases were identified with the Bruker DIFFRAC.EVA software.

Magnetic measurements were performed on freeze-dried bulk sediment at the School of Earth Sciences and Engineering, Sun Yat-Sen University. Sediment magnetic susceptibility (χ) was measured at 976 Hz in a 200 Am−1 field with an AGICO MFK1-FA Kappabridge system. First-order reversal curve (FORC) diagrams were measured at 0.2 mT field steps with 0.5 s averaging time.

Sulphide sulphur from bulk sediment was liberated as hydrogen sulphide via wet chemical sequential extraction78,79. Acid-volatile (mono)sulphides (AVS) were liberated with HCl (25%) for 1 h at room temperature, but no AVS was detected during extraction. The remaining samples were reacted with 1 M CrCl2 solution at sub-boiling temperatures for 2 h in an inert nitrogen atmosphere to liberate CRS pyrite, including greigite. Released hydrogen sulphide from CRS was precipitated as zinc sulphide using a 3% acetic acid zinc acetate solution. Zinc sulphide precipitates were subsequently converted to silver sulphide (Ag2S) using a 0.1 M AgNO3 solution. Ag2S precipitates were collected by filtration (<0.45 µm) and dried at 40 °C overnight. The CRS content was determined gravimetrically based on the Ag2S yield.

For δ34S analysis, Ag2S precipitates were measured as SO2 via combustion of a ~ 200 μg sample mixed with an equal amount of V2O5 using a ThermoScientific Delta V mass spectrometer linked to an Elemental Analyzer (EA-IRMS) at the Institut für Geologie und Paläontologie, Westfälische Wilhelms-Universität Münster. The isotope ratio is expressed in relation to the Vienna Canyon Diablo Troilite (V-CDT) standard with analytical precision better than ±0.3‰:

$${{{{{{\rm{\delta }}}}}}}^{34}{{\mbox{S}}} \, \left(\textperthousand,{{\mbox{V}}}-{{\mbox{CDT}}}\right)=\left[\left(\frac{{\left({\,\!}^{34}{{\mbox{S}}}/{\,\!}^{32}{{\mbox{S}}}\right)}_{{{\mbox{sample}}}}}{{\left({\,\!}^{34}{{\mbox{S}}}/{\,\!}^{32}{{\mbox{S}}}\right)}_{{{\mbox{V}}}-{{\mbox{CDT}}}}}\right)-1\right]\times 1000$$

δ34S measurements were calibrated with international reference materials IAEA-S1 (δ34S = −0.30‰), IAEA-S2 (δ34S = 22.62‰), IAEA-S3 (δ34S = −32.49) and NBS 127 (δ34S = 20.3‰)80.

For multiple sulphur isotope measurements (i.e., 32S, 33S, 34S, and 36S), ~2.5 mg of Ag2S was fluorinated to sulphur hexafluoride (SF6) by reaction with 5 times excess of F2 at 300 °C for 8 h in nickel reactors81. Following cryogenic and gas chromatographic purification, the multiple sulphur isotopes were measured using a ThermoScientific MAT 253 mass spectrometer at the Institut für Geologie und Paläontologie, Westfälische Wilhelms-Universität Münster. Results are calculated from δ33S, δ34S and δ36S and expressed as Δ33S values82:

$${\Delta }^{33}{{\mbox{S}}} \, \left(\textperthousand\right)={\Delta }^{33}{{\mbox{S}}}-1000\times \left[{\left(1+\frac{{{{{{{\rm{\delta }}}}}}}^{34}{{\mbox{S}}}}{1000}\right)}^{0{{\mbox{.}}}515}-1\right]$$

The multiple sulphur isotope measurements were calibrated with international reference material IAEA-S1 (δ34S = − 0.30‰ and δ33S = − 0.055‰)81. Analytical precision is better than ±0.02‰ (2 SD) for Δ33S.

For SIMS analysis, selected pyrite and greigite samples were mounted on epoxy discs (25 mm diameter). The pyrite tubes were then cut, ground, and polished. SIMS analyses were made with a Cameca IMS1280-HR at the Guangzhou Institute of Geochemistry, Chinese Academy of Sciences83. A primary Cs+ ion beam (~2.0 nA current, 20 kV total impact energy) was focused at the sample surface with a 15-μm spot diameter. A 15-μm raster was applied during all analyses to slightly homogenise the Gaussian beam. Pre-sputtering for 20 s was applied to remove the Au coating, and a normal-incidence electron gun was used for charge compensation. The mass resolving power was set at ~5000 to avoid isobaric interference. A NMR field sensor was used to stabilise the magnetic field. 32S, 33S and 34S were measured simultaneously by three Faraday cups in the multi-collector system (L’2, L1 and H1, respectively). The total analysis time for each spot was ~4 minutes. An off-mount calibration procedure was used83. Pyrite standard UWPy-1 (δ34S = 16.04 ± 0.18‰)84 was used as the primary standard, and PPP-1 (δ34S = 5.3 ± 0.20‰)85 was used as a quality control standard. Pyrrhotite standard YP-136 (δ34S = 1.5 ± 0.1‰)83 was mounted together with the studied pyrite aggregates in epoxy discs to monitor analytical reliability. Standard YP-136 was measured at regular intervals between the analysis of every five samples. The isotope ratio is expressed as δ34S with the V-CDT standard. Analytical precision is better than ±0.2‰ for δ34S.

Sequential extraction of solid-phase sedimentary iron pools by different solutions was carried out following the procedure of Poulton and Canfield75 to quantify the following three iron mineral pools: (1) Na-acetate solution extracted iron (Feacetate): extracted using 10 ml of 1 M Na-acetate solution for 24 h, at solution pH 4.5 and ultrasonic vibration (UV) at 50 °C; (2) dithionite solution extracted iron (Fedithionite): 10 ml of 50 g l−1 Na dithionite and 0.2 mol l−1 tri-Na citrate, pH 4.8, with 2 h UV at room temperature; (3) oxalate solution extracted iron (Feoxalate): 10 ml of 0.2 mol l−1 ammonium oxalate and 0.17 mol l−1 oxalic acid, with 6 h at room temperature. Between each extraction step, residues were washed with 10 ml deionized water twice. Leachates from the above three steps were measured by solution inductively coupled plasma–atomic emission spectrometry (ICP-AES). Analytical reproducibility was typically better than ±5% (2 SD). Assuming that all CRS is from pyrite (1:2 stoichiometry of Fe/S), pyrite-bound iron (Fepy) is calculated from the measured CRS content of each sample.

Total carbon and total inorganic carbon (TIC) contents were measured via IR spectroscopy of CO2 using a CS-MAT 5500 carbon–sulphur analyzer. Sediment powder was combusted to CO2 at 1350 °C for total carbon analysis. HCl was added to liberate CO2 for total inorganic carbon contents. Total organic carbon (TOC) was calculated as the difference between total carbon and total inorganic carbon in the sediment. The reproducibility of the TOC analysis was typically better than ±2%.

For the analysis of δ13C values, CO2 was released from sediments using off-line phosphoric acid reaction method86 with phosphoric acid of a nominal concentration of 103%87 at 50 °C for 24 h and subsequent cryogenic distillation. Carbon isotope measurements were carried out with a Thermo-Finnigan DeltaPlus mass spectrometer at Institut für Geologie und Paläontologie, Westfälische Wilhelms-Universität Münster. Stable isotope results are expressed relative to the V-PDB standard. The carbon isotope measurements were calibrated with a carbonate standard (δ13C = 1.46‰). The analytical accuracy was better than ±0.1‰ for δ13C values.

Powdered samples were digested with a mixture of concentrated HNO3, HF, and HClO4 acids. The solution was evaporated on a hotplate to near dryness and then re-dissolved in 100 ml 7 M HCl. The final solution was diluted in 2% HNO3 solution and analysed with an Agilent 5110 inductively coupled plasma–atomic emission spectrometer (ICP-AES) for Al, Ti, K and P concentrations and analysed with an Agilent 7700 inductively coupled plasma mass spectrometer (ICP-MS) for Zr and Mo concentrations at the ALS Laboratory, Guangzhou. Analytical performance was monitored by repeatedly analysing certified reference materials (MRGeo08 and GBM908-10). Analytical precision was typically better than 7% for Zr, Ti, K, Ba, Mo and P, and better than 6% for Al.