The Ediacaran Period (635-539 Ma) witnessed the largest negative carbonate carbon isotope (δ13Ccarb) excursion in Earth’s history, known as the Shuram Excursion (SE, or Wonoka/DOUNCE/EN3) between 574.0 ± 4.7 and 567.3 ± 3.0 Ma1,2,3,4. Typically, the SE is characterized by a dramatic drop in the δ13Ccarb from as high as +5‰ to as low as −12‰ globally. The difficulty in explaining the large shift in the global carbon cycle and the often-positive correlation between δ13Ccarb and δ18Ocarb have led some researchers to argue that the SE is a product of later diagenesis (e.g., meteoric alteration5,6 or burial diagenesis7). But diagenesis is inherently a local process and seems inconsistent with the global occurrence of the SE1,3,4. Supportive evidence for the primary nature of the SE comes from petrography8, organic geochemistry9, and high-spatial-resolution isotope analysis10,11. Consequently, the SE is interpreted as a result of enhanced oxidation of 13C-depleted organic carbon reservoirs, such as the dissolved organic carbon (DOC) in seawater12,13,14, fossil organic matter exposed on land15, hydrocarbon-rich fluids from the subsurface9, and/or methane emitted from the sediments16. The onset of the SE is closely followed by the first appearance of large, unambiguous metazoan fossils17. Thus, constraining the origin of the SE can help elucidate the interaction between Earth system oxygenation and the radiation of early animals.

A sustained oxidation of organics over ~7 million years (Myrs)2,18 during the SE requires a consistent supply of oxidant, if it is indeed an oceanic oxygenation event. Multiple lines of evidence, including S-U-Tl isotope compositions of carbonate or shale13,14,15,19,20,21 and iron speciation in shale22, have supported an increased oceanic ventilation, i.e., rising marine concentrations of dissolved O2 and/or sulfate during the SE, although a numerical model has argued that seawater oxidants would not be sufficient in driving a global carbon perturbation for millions of years23. The dissolution of older continental evaporite may provide additional oxidant24, and/or the oxidation of DOC could be spatially heterogeneous due to local oxygenic photosynthesis25. Nevertheless, the oxidants available for organic matter remineralization during the SE are, (1) dissolved O2 in the ocean supplied from atmosphere and/or local photosynthesis, (2) newly formed sulfate from sulfide/sulfur oxidation on the continents and/or in ocean basins, and/or (3) weathered sulfate from the dissolution of older evaporite deposits on the continents. Regardless of whether O2 played a direct or indirect role in oxidizing organics, the role of sulfate as an oxidant of organics could be significantly enhanced due to a rising O2 concentration. This may occur through enhanced oxidative weathering of sulfide minerals on the continents26,27 or ventilation of deeper sulfidic seawater. Therefore, the origin of the sulfate is key for understanding the nature of the SE as an oceanic oxygenation event.

Sulfur and triple oxygen isotope compositions (i.e., δ34S, δ18O, and δ17O), especially 17O anomalies [i.e.,\({\Delta ^{\prime 17} }{{{{{\rm{O}}}}}}\equiv ({{{{{\rm{ln}}}}}}\frac{{{{{{{\rm{\delta }}}}}}}^{17}{{{{{\rm{O}}}}}}+1000}{1000}-0.5305\times {{{{{\rm{ln}}}}}}\frac{{{{{{{\rm{\delta }}}}}}}^{18}{{{{{\rm{O}}}}}}+1000}{1000})\times 1000\textperthousand\)], provide a powerful tool to disentangle the origin of sulfate28. If atmospheric O2 was indeed the main oxidant responsible for the SE, an increase in sulfate concentration during the SE would occur through enhanced sulfide/sulfur oxidation as a result of a deep-water invasion of atmospheric O2 or an expansion of ventilated ocean volume. Importantly, the increased sulfate would be accompanied by a negative shift in the sulfate Δ′17O value of the ocean because newly-formed sulfate would inherit the atmospheric O2 signature that bears a unique mass-independent 17O depletion28,29,30,31,32,33,34,35. We note that such a negative shift in sulfate Δ′17O value is unlikely to be generated by diagenesis, as no diagenetic process has been found to be capable of bringing more atmospheric O2 into the sediments or rocks. In contrast, if the dominant oxidant is evaporite sulfate dissolved from the continents24, it would be highly unlikely to detect distinctly negative Δ′17O values for sulfate extracted from the SE carbonates because most of the pre-Ediacaran evaporites do not exhibit large 17O depletion31,32. Sulfate generated from sulfide/sulfur oxidation by photosynthetic O2 in the water column at local photic zones would also not have significantly negative 17O anomalies36. Moreover, further insights into the trigger and dynamics of the SE oxygenation event can be obtained when the temporal trends and spatial heterogeneity of sulfate’s sulfur and triple oxygen isotope compositions are combined with associated carbon isotope trends.

To test the hypotheses above, we selected SE-containing units from three paleogeographically different continents1,3: the Doushantuo Formation of South China, the Wonoka Formation of South Australia, and the Shuiquan Formation of Tarim, all of which have well-constrained carbon isotope records (Fig. 1; Supplementary Discussion), and we extracted carbonate-associated sulfate (CAS) and analyzed the NaCl-leached and the HCl-leached fractions for their respective δ34S, δ18O, and ∆′17O values (see Methods).

Fig. 1: C-S-O isotope compositions of the Shuram Excursion.
figure 1

a South China, b South Australia, and c Tarim. The light-grey shaded areas refer to intervals of the largest negative carbon isotope excursion (defined here by δ13Ccarb lower than −6‰). The solid lines represent LOWESS curves. Analytical uncertainty (1σ) for isotope compositions is smaller than the sizes of the symbols. Carbon isotope data in grey-filled cycles represent the samples used in this study, while those in open cycles are from refs. 25,69,70.

Results and discussion

Sulfur and triple oxygen isotope data

The Δ′17O values of the HCl-leached CAS range from −0.53 to −0.14‰, −0.51 to −0.07‰, and −0.35 to +0.06‰ in South China, South Australia, and Tarim, respectively (see Supplementary Data 1). The Δ′17O nadirs are distinctly lower than −0.07 ± 0.09 ‰ (1σ), the average value for modern and Phanerozoic sulfate minerals (also see compiled data in refs. 31,37). Moreover, the Δ′17O records in all three paleocontinents examined display a remarkably similar stratigraphic trend, with an initial decrease followed by a recovery over the SE intervals (Fig. 1). This trend is also observed for their corresponding δ34S, δ18O, and δ13Ccarb records. The δ34S values of the HCl-leached CAS range from 16.3 to 42.0‰, 15.2 to 29.1‰, and 10.2 to 18.4‰ and their corresponding δ18O values from 8.0 to 23.1‰, 8.8 to 15.3‰, and 10.4 to 15.0‰ in South China, South Australia, and Tarim, respectively.

Sulfate 17O anomaly in geological records for tracking paleoatmospheric O2

The 17O anomaly signals found in geological sulfate minerals are a powerful tool to investigate biogeochemical cycles of carbon, sulfur and oxygen over Earth’s history, especially with regard to processes related directly to atmospheric O228,29,30,31,32,33. Atmospheric O2 is the only known source compound that bears a negative Δ′17O value, which originates from mass-independent fractionation during photochemical reactions involving O2, O3, and CO2 in the stratosphere34,35. The Δ′17O value of atmospheric O2 is primarily determined by pO2/pCO2 ratios and the rates of gross primary production34,38. Lower pO2/pCO2 ratios or lower gross primary production would lead to more 17O-depleted atmospheric O238. The 17O anomaly in atmospheric O2 could be passed onto newly formed sulfate through sulfide/sulfur oxidation, as demonstrated by laboratory experiments in which approximately 8 to 30% of the oxygen atoms in the produced sulfate originated from O239,40. However, the 17O anomaly formed in the sulfate can potentially be erased by sulfur redox processes, as these processes enable sulfate to exchange oxygen atoms with the surrounding water through a backward exchange between sulfite and sulfate41 and/or via anaerobic oxidation of H2S/S0 by nitrate or Fe(III) in water or sediments42. Nevertheless, the preserved 17O anomaly signals in geological sulfate minerals establish a direct association between the isotopic composition of sedimentary sulfate and paleoatmospheric O2. Significantly negative Δ′17O values in geological sulfate provide strong evidence for the partial incorporation of paleoatmospheric O2 into sulfate, usually through sulfide/sulfur oxidation28,29,30,31,32,33.

Evaluating the extracted CAS

Extracted CAS from carbonate outcrops could be contaminated by present-day atmospheric sulfate43 and/or newly generated sulfate from pyrite oxidation during outcrop weathering or laboratory experiments44,45. Prior to the HCl-solution extraction, repeated NaCl-solution leaching can effectively remove most, if not all, of the syn-sedimentary evaporites (if present), sulfate generated by post-depositional processes (e.g., diagenesis, laboratory treatment), and modern atmospheric deposition. Often, sulfate produced via the oxidation of sulfide minerals has lower δ34S and δ18O values than the original seawater sulfate45,46.

To evaluate the degree of contamination of the non-original sulfate in the HCl-leached sulfate, we compared the sulfur and oxygen isotope compositions of the NaCl-leached and the HCl-leached sulfate. The Δ′17O values of the HCl-leached sulfate are more negative (i.e., more distinct) than those of the corresponding NaCl-leached sulfate, while both the δ34S and δ18O values of the HCl-leached sulfate are generally higher than those of the NaCl-leached ones (Fig. 2, Supplementary Figs. 1, 2). This pattern holds true for samples from all three paleocontinents. Meanwhile, the stratigraphic trends of these S- and O-isotopes are drastically different between the HCl-leached and the NaCl-leached sulfate (Supplementary Fig. 2). These observations indicate that the NaCl-leached fraction contains a significant amount of sulfate originating from post-depositional oxidation of sulfide minerals due to their low δ34S and low δ18O values, and/or of present-day atmospheric sulfate due to the positive Δ′17O values (up to +0.51‰)35,47,48. In contrast, the presence of non-original sulfate in the HCl-leached CAS is minimal. It is notable that the Tarim CAS may still have small fractions of the present-day atmospheric sulfate, as indicated by their slightly positive Δ′17O values (up to +0.06‰). Thus, the Δ′17O values of the original sulfate during the SE in Tarim should be more negative than the data shown here. Nevertheless, the magnitudes of the Δ′17O negative shift of −0.39‰, −0.44‰, and −0.41‰ in the HCl-leached CAS during the SE in South China, South Australia, and Tarim, respectively, are almost identical, strongly favoring the observed sulfate Δ′17O negative shifts being a credible primary and global signature (Fig. 1).

Fig. 2: Crossplots of Δ′17O, δ18O and δ34S.
figure 2

a δ18O vs. δ34S. b Δ′17O vs. δ34S. c Δ′17O vs. δ18O. Arrows filled with a gradual color denote the ranges of δ34S and δ18O of the HCl-leached sulfate converge to a point when the Δ′17O values are the most negative.

Depending on the formation condition, the original sulfate residing as HCl-leached CAS can be entirely of seawater sulfate origin if the carbonate rocks were precipitated from seawater and had a negligible contribution from carbonate cements formed during early diagenesis. The SE carbonates were most likely precipitated in the water column or at the sediment–water interface, as independently supported by their calcium and magnesium isotope composition49,50. Furthermore, the SE is recorded in a variety of depositional facies, ranging from shallow water peritidal to open deep slope settings1,3. Therefore, the HCl-leached CAS we measured primarily reflects seawater sulfate of the SE if later diagenetic alteration is excluded (see discussion followed).

Evaluation of diagenetic alteration

In addition to the near-identical sulfate Δ′17O negative shift in the three separate paleocontinents, our data display tight stratigraphic co-variations of their C–S–O isotope composition (Fig. 1), supporting an original, coupled perturbation to marine carbon and sulfur cycling during the SE. Indeed, no diagenetic processes have been found to yield significantly negative Δ′17O values in sulfate, a signal unique to sulfate that carries atmospheric O2 signature of geological times28,29,30,31,32.

Hypothetically, O2-rich meteoric water circulating through pyrite-rich carbonates may oxidize the pyrite to form sulfate with potentially negative Δ′17O values if the O2 had a negative Δ′17O value. Such a scenario would result in the redistribution of sulfate in these carbonates and a homogenized CAS multi-isotope range in the stratigraphic level, which is not observed. In fact, a consistent C–S–O isotope variation is observed within tens of meters of carbonate formations and across three separate paleocontinents (Fig. 1). In addition, if the CAS originated from pyrite oxidation, the δ34S value of the CAS would be close to that of co-existing pyrite in carbonates, and its δ18O value would be much lower, often close to that of the solution water51. However, neither scenario is observed13,14,20,52. For example, the average δ34S value of the HCl-leached CAS from South China is ~22.8‰, substantially higher than the mean value of −7.4‰ in the co-existing pyrite20. The δ18O value of the meteoric water is most likely lower than 0‰53, but the measured CAS exhibits significantly higher δ18O values, averaging 12.2–12.4‰ (Fig. 2, Supplementary Table 1).

Some researchers argue in favor of a diagenetic origin of the SE, primarily based on the co-variation between δ13Ccarb and δ18Ocarb in carbonates5,7, which is commonly observed in typical SE sections (although not universally1,3). While late meteoric diagenesis can generate a positive correlation between δ13Ccarb and δ18Ocarb, the existence of this correlation does not necessarily confirm meteoric diagenesis. The δ18Ocarb is sensitive to changes in ambient temperature, the δ18O of solution water, or late diagenesis4,11. Often, the δ18Ocarb is more susceptible to resetting than the corresponding δ13Ccarb due to the greater abundance of external oxygen compared to carbon during later fluid-rock interactions3. Therefore, the δ13Ccarb could be of primary origin, as supported by our data, even if the δ18Ocarb is influenced by diagenesis. Furthermore, recent studies, based on in-situ carbon isotope analysis10,11, Ca and Mg isotopes and carbonate-associated phosphate concentrations over different SE sections49,50,54, and other geochemical and stratigraphic observations1,8,9, have argued against meteoric or late burial diagenesis as the cause of the SE. Our results cannot rule out the possibility of a diagenetic origin for the δ18Ocarb in the SE carbonates, thus the discussion will primarily focus on sulfate multi-isotopes and their relationships with the δ13Ccarb values.

Marine sulfate evolution

The spatial heterogeneity of sulfur and oxygen isotope compositions of the pre- and post-SE sulfate supports a scenario of low and fluctuating marine sulfate concentration. The δ34S and δ18O values of the pre-SE HCl-leached CAS from different paleocontinents vary widely, with average values ranging from 12.3 to 38.8‰ and from 14.2 to 19.3‰ (the South Australia section lacks a pre-SE record), respectively, whereas those during the SE cluster around 13.8–22.8‰ and 12.2–12.4‰, respectively (Fig. 2; Supplementary Table 1). These patterns are consistent with a low marine sulfate concentration preceding the SE when the sulfur and oxygen isotope compositions were susceptible to local perturbations, while during the SE an enhanced sulfate flux was supplied to continental shelves. The added sulfate appears to have the same origin worldwide and was overwhelming in quantity because its sulfur and oxygen isotope compositions are converging to the same set of values across three different continental shelves (Fig. 2).

What then, is the source of the enhanced sulfate flux? Our CAS data from three examined paleocontinents all show distinctly negative Δ′17O values and similar δ34S and δ18O values during the SE. This observation argues against evaporite dissolution being the main source of the added sulfate input, as contribution from evaporite dissolution should be local and it is unlikely that evaporites with the same multi-isotope composition would exist on different continents. The cross-plots of Δ′17O and δ18O show that the pre-Ediacaran evaporites do not fall on the mixing lines and are therefore unlikely a component of the SE sulfate (see Supplementary Fig. 3). In addition, pre-Ediacaran evaporites are found in small scales, with the current volume of ~4.7 × 105 km3 in total55. If being all gypsum and all dissolved, they could only supply ~0.6 Myr of the sulfate flux (i.e., ~1 × 1013 mol yr−1) needed in Shields et al’s model estimate24. Even considering dissolution of pre-Ediacaran evaporites after deposition, these numbers require an unreasonably large volume of evaporites exposed during pre-Ediacaran time, and no evidence for massive pre-Ediacaran evaporite dissolution was found.

Instead, there must be an enhanced influx of sulfate derived from sulfide/sulfur oxidation via atmospheric O2 during the SE. However, the sulfate source could come from either pyrite weathering on the continents15,25 or oxidation of sulfur compounds (e.g., H2S and S0) in the deeper ocean. While the latter is less known, H2S and S0 can be oxidized directly by O256,57, and this process may occur in natural environments where both H2S and O2 are present58,59,60. The two cases depict different scenarios of the SE oxygenation event. The former may suggest that the threshold of atmospheric pO2 (i.e., ~0.4% of the present-day27) for the sensitivity of oxidative pyrite weathering had not been reached in the Ediacaran, or there was an increase in the exposed surface area of sulfide-rich sediments on the continents during the SE15. The latter requires the ventilation of a deeper, sulfide-rich marine water body and ocean itself being a source of sulfate via sulfur oxidation. To determine the plausibility of each hypothesis, we resort to quantitative modeling.

Marine S-cycle modeling and the origin of the 17O-depleted sulfate

The continued stratigraphic sulfate S–O isotope records, along with the duration constrained by high-quality geochronological ages2,18, provide an unparalleled opportunity to place quantitative constraints on sulfur geochemical dynamics. Here we used a non-steady-state box model of global sulfur cycle to estimate the source and flux of the enhanced sulfate during the SE (see Supplementary Discussion for a full description of the model and Table 2 for all parameters used). Whereas such a model has been well developed for exploring sulfur cycling in ancient oceans61,62, this is the first time that the Δ′17O parameter has been incorporated into the model, which constrains processes such as H2S/S0 oxidation specifically.

Assuming that all of the increased sulfate was sourced from pyrite weathering on the continents, we adopted initial steady states (see Supplementary Table 3) established based on the pre-SE’s data of South China (δ34S ~ 40‰, δ18O ~ 20‰, and Δ′17O ~ −0.17‰) to drive changes and fit the observed temporal isotope trends. We selected South China data as a typical case for modeling because of its complete and continuous stratigraphic records, as well as its likely limited contamination to its S- and O-isotope compositions when compared to the other two sections. Our modeling results show that a rise in pyrite weathering flux (Fwp) alone cannot simultaneously reproduce the observed shifts in sulfur and oxygen isotope compositions (δ34S ~ 20‰, δ18O ~ 12‰, and Δ′17O ~ −0.5‰) during the SE (Supplementary Fig. 4a). The magnitude of the δ34S shift is much larger than that of the δ18O, which is likely due to (1) a buffering effect from the intense oxygen isotope exchange between intermediate sulfur species and the ambient H2O during microbial sulfate reduction (MSR) in seawater41,42, and (2) the significantly greater difference in δ34S ( − 17‰ vs. 40‰) than that of the δ18O (0‰ vs. 20‰) between the pyrite-derived and the pre-existing sulfate.

If we were to reproduce the full δ18O shift using only pyrite weathering, it would require an unrealistic increase of two orders of magnitude in its flux, as well as an additional condition that the δ34S of the pyrite-derived sulfate needs to be around 18‰ (Supplementary Fig. 4b), which is unreasonably higher than the sulfate δ34S (close to 4.8‰63) in modern global riverine systems. Furthermore, if this process were to last for ~7 Myrs, it would result in an unrealistic increase of more than two orders of magnitude in marine sulfate concentration (reaching up to ~300 mM; Supplementary Fig. 4b), which is much higher than the modern value of 28 mM42. Therefore, we conclude that a flux increase in the 17O-anomalous riverine sulfate alone cannot reasonably explain the temporal sulfate isotope trends during the SE.

Alternatively, an increased oxidation of reduced sulfur species in seawater linked to water-column ventilation may play a vital role in controlling the observed temporal multi-isotope trends. On one hand, water-column ventilation could increase the fraction of extracellular aerobic H2S reoxidation (i.e., higher freo and freo-aerobic) during the course of MSR64; on the other, as the concentration of dissolved O2 increases, it would become the dominant oxidant in organic matter remineralization, suppressing MSR in the water column (i.e., lower kmsr)65. These two processes both contribute to the decreases in sulfate sulfur and oxygen isotope compositions (Supplementary Fig. 4c–e). However, even when using parameters representing a well-oxygenated condition, i.e., a near-zero MSR flux in the water column (kmsr = 0; Fig. 3a) and a high fraction of aerobic H2S reoxidation (freo = 0.9; freo-aerobic = 0.95; Fig. 3b, c), the model still cannot reproduce the observed decreases in δ18O and Δ′17O during the SE (Fig. 3f, g), suggesting that other factors must be at play.

Fig. 3: Output of our S-cycling box-model.
figure 3

Driven forces: changes in a coefficient kmsr of MSR (microbial sulfate reduction) flux in the water-column, b fraction of H2S-reoxidized during the MSR, c proportion of aerobic oxidation during H2S reoxidation, and d added sulfate flux from oxidation of the pre-existing sulfur pool. Model results: e seawater sulfate δ34S, f seawater sulfate δ18O, g seawater sulfate ∆′17O, and h seawater sulfate concentration. The initial steady states with sulfate concentration of 0.5 mM (SS4-2, see Supplementary Table 3) and 1.1 mM (SS8-2) were adopted as suggested72,73.

Sulfidic water was expected in the shelf areas preceding the SE66, and this condition could contribute to organic carbon accumulation and provide a stable dissolved organic sulfur reservoir in the ocean through DOC sulfurization67. An enhanced oxidation of these pre-existing sulfur species (i.e., H2S and S0) could provide an additional sulfate source with lower δ18O and Δ′17O values (Fig. 4). When we also factor in the oxidation of the pre-existing sulfur species, the model is capable of reproducing all of the observed S–O isotope compositions and their temporal changes (Fig. 3d-g; Supplementary Figs. 5-7). The optimum condition requires the δ34S of the pre-existing H2S/S0 reservoir to be in the range of 16–25‰ [note that the pre-SE pyrite δ34S averages 18.7 ± 8.7 ‰ (1σ) in South China14], and the δ18O and Δ′17O values of its derived sulfate to be at 0–4‰ and −0.7 to −0.8‰, respectively. Additionally, the added sulfate flux from the oxidation of the pre-existing sulfur species needs to occur as a pulse of 1–2.5 × 1013 mol yr−1, with a duration of 1.5 Myrs (Fig. 3d). This scenario corresponds to an increase in the marine sulfate concentration from ~0.5–1 mM to ~20 mM (Fig. 3h).

Fig. 4: Schematic presentation of the origin of the 17O-depleted sulfate during the Shuram Excursion.
figure 4

Mass-independently 17O-depleted O2, generated during photochemical reactions of O2, O3 and CO2 in the stratosphere, mixes into the ocean through gas exchange between stratosphere and troposphere, and between atmosphere and ocean. The newly formed sulfate inherits the 17O-depleted signature from atmospheric O2 through oxidation of the reduced sulfur species in the ocean and on land. See text for more details. DOC dissolved organic carbon, DOS dissolved organic sulfur.

The sulfate 17O-depletion during the SE occurred ~60 Myrs after the basal Ediacaran 17O-depletion event2,68. This raises the question of if atmosphere O2 was consistently depleted in 17O throughout the Ediacaran Period and the high sulfate concentrations at the aftermath of Marinoan Snowball Earth and during the SE merely facilitated the preservation of the sulfate 17O record, or if atmospheric O2 was distinctly depleted in 17O only during these two episodes. A wider geographic coverage of a similar dataset is needed to further answer this question. Additionally, further research efforts are required to validate if sulfate derived from H2S oxidation in natural water bodies indeed incorporates the Δ′17O signal of atmospheric O2.

Conclusions and implications for the SE event

Taken together, our sulfate S-O isotope data and modeling results support water-column ventilation during the SE, which led to enhanced sulfur oxidation by mass-anomalously 17O-depleted O2 dissolved in the oceans (Fig. 4). The positive co-variations of the C–S–O isotope compositions observed in three separate paleocontinents imply that the processes discussed above also apply to the carbon isotope records. It could be either an increase in atmospheric O2 concentration or increased ventilation of the deep oceans that facilitated the oxidation of marine organics, thus contributing to the negative carbon isotope excursion. In summary, our findings confirm the SE as a primary oceanic oxygenation event, rather than of a late diagenetic origin, and provide direct evidence for paleoatmospheric O2 being ultimately responsible for the oxidation of reduced sulfur and organics in the oceans. This resolves the long-standing debate on the origin of the largest negative C-isotope excursion in Earth’s history and establishes a link between environmental oxygenation and the rise of early animals.



We targeted the Ediacaran carbonates from three different paleocontinents which have recorded the Shuram Excursion (Fig. 1; Supplementary Discussion). A total of 117 samples were collected, including 51 samples from the Jiulongwan section (30°47′51″N, 110°59′32″E) in South China, 28 samples from the Parachilna Gorge section (31°9′51.6″S, 138°31′43.2″E) in South Australia, and 38 samples from the Mochia-Khutuk section (41°26′29″N, 87°51′47″E) in Tarim. Samples were cut, cleaned, and ground to ca. 200 mesh for bulk geochemical analyses. Stratigraphic details for these sections can be found in refs. 14,20,69,70.

CAS extraction and purification

Approximately 50 g sample powder was immersed in a 10% NaCl solution at least 4 times, with each immersion lasting at least 12-h. This process yielded NaCl-leached sulfate. After the final NaCl-leaching, a saturated BaCl2 solution was added to the filtered and acidified solution to check if there is BaSO4 precipitating. If precipitates are visible, one more NaCl leaching was conducted. Next, 4 M HCl solution was added to the NaCl-leached sample residue and let it sit in room temperature for less than 1 hour, resulting in HCl-leached sulfate. A fraction of the BaSO4 precipitate was purified using the DDARP method (DTPA dissolution and re-precipitation) for triple oxygen isotope measurement. Detailed processing protocols can be found in refs. 43,45,71. Pre-treatment experiments were conducted at both Louisiana State University and China University of Geosciences (Wuhan).

Isotope measurements

Sulfate δ34S was measured through in-line combustion of ~0.3 mg of powdered BaSO4 mixed with ~1.0 mg of V2O5 in a Flash elemental analyzer coupled to a Thermo Fisher Scientific Delta V Plus isotope-ratio mass spectrometer. Sulfate δ18O was measured via CO gas converted from ~0.20 mg of BaSO4 powder using a Thermal Conversion Elemental Analyzer at 1410 °C coupled to a Thermo Fisher Scientific Delta V Plus isotope-ratio mass spectrometer in continuous-flow mode. Sulfate Δ′17O was determined using a Thermo MAT253 Plus isotope-ratio mass spectrometer in dual-inlet mode via measuring O2 generated offline from ~8 to 12 mg of pure BaSO4 powder reacting with BrF5, utilizing a CO2 laser-fluorination system (further details refer to descriptions provided in refs. 32,47). The sulfur and oxygen isotope compositions are reported in δ-notation as per mil (‰) relative to Vienna Cañon Diablo Troilite (V-CDT) and Vienna Standard Mean Ocean Water (V-SMOW), respectively. The δ34S was calibrated using standards of two international (NBS127, 20.3‰, IAEA-SO-5, 0.5‰) and one inter-laboratory (OASIC-S, 14.5‰), while the δ18O using standards of one international (NBS127, 8.6‰) and one inter-laboratory (OASIC-O, 12.3‰). The analytical uncertainties (1σ) were better than ±0.1‰ and ±0.3‰ for δ34S and δ18O, respectively. The standard deviation (1σ) for the Δ′17O is ±0.02‰ based on multiple runs (N = 4) of the same BaSO4 sample. All isotope analyses were conducted in the International Center for Isotope Effects Research (ICIER) at Nanjing University.