Introduction

The Ediacaran Period (635–539 million years ago, or Ma) was a pivotal timeframe in the evolution of life on Earth, evidenced by the initial appearance of macroscopic multicellular eukaryotes in the fossil record (i.e., the “Ediacara biota”) and its subsequent decline1. The rise and decline of the Ediacara biota are among the biggest mysteries in the evolution of life. Numerous studies have linked the rise of multicellular eukaryotes and the diversification of the Ediacara biota to ocean oxygenation2,3. Likewise, the decline of the Ediacara biota has been linked to a drop in ocean oxygen levels during the terminal Ediacaran Period4,5. However, the causal relationships between biological innovation and ocean redox during this period remain debated because of disparate ocean redox conditions reconstructed by different paleoredox proxies6. Redox-sensitive element (RSE) enrichments and carbon and sulfur isotope variations found in the sedimentary record are interpreted as evidence of widespread oxic conditions, and perhaps even ventilation of the deep ocean, at least episodically during the Ediacaran2,3,7,8. Yet, Fe speciation data indicate persistently anoxic deep waters throughout most of the Ediacaran9,10. In recent years, a growing body of evidence supports a redox-stratified Ediacaran ocean, with some continental shelf margins overlain by a dynamic mid-depth euxinic water mass (“euxinic wedge”)11,12. In this scenario, early animals could have thrived in oxic shelf settings even as deep oceans remained anoxic13,14. It was even suggested that surface ocean environments that met the oxygen requirements of the earliest metazoans (0.5–4.0% of present atmospheric levels) were already established well before the Ediacaran15, leading to the proposal that the radiation of animals was not prohibited by oxygen availability. Thus, additional information about ocean redox conditions and their changes could help to resolve these complexities.

In this study, we refine the redox evolution in the Ediacaran ocean using mercury (Hg) stable isotopes. In particular, we focus on evidence for the presence of anoxic and H2S-rich waters in the marine photic zone, a condition referred to as photic zone euxinia (PZE)16,17. Different from the previously proposed “euxinic wedge” that refers to a mid-depth euxinic water mass11,12, PZE emphasizes the shallowing of the euxinic water mass into the photic zone, which is particularly detrimental to shallow marine inhabitants and has been proposed as a potent kill mechanism during almost all Phanerozoic mass extinction events17 as well as a key factor responsible for the evolutionary stasis during the mid-Proterozoic18. The shelf areas are believed to have been major habitats for early animals19, and thus the development of PZE would have shrunk the oxygenated habitats and possibly contributed to the delayed rise and eventual decline of the Ediacara biota. Moreover, PZE is typically found to be driven by enhanced nutrient influx to the ocean20, which is exactly the mechanism proposed to trigger the oxygenation events in the Ediacaran ocean7,8. Therefore, there could be a potential link between ocean oxygenation and PZE, and thus the investigation on PZE may provide insights into the cause and consequence of ocean oxygenation in the Ediacaran Period.

Mercury (Hg) isotopes in sedimentary rocks are an emerging proxy for tracing paleoenvironmental changes such as large volcanic emission21, as well as PZE16. A previous study has shown that marine sediments deposited under PZE display distinct negative excursions of mass independent isotope fractionation (MIF) of Hg relative to non-PZE conditions, demonstrating that Hg isotopes are a promising proxy for changes in ocean redox conditions16. Here we report Hg concentration and isotope ratios of the Ediacaran Doushantuo Formation in South China (Fig. 1). We studied four black-shale dominated sections, including Weng’an (WA, shallow shelf), Taoying (TY, upper slope), Wuhe (WH, lower slope), and Yuanjia (YJ, basin) sections, which represent a shelf-to-basin transect. The Doushantuo Formation (~635–560 Ma) is known for its exceptionally well-preserved fossil and geochemical record, and has been extensively studied to reconstruct the Ediacaran ocean environments22. Previous studies on the same suite of black shale sections reported evidence for three ocean oxygenation events (OOEs) at ~635 Ma, ~580 Ma, ~565 Ma based on a variety of redox proxies (i.e., Fe speciation, RSE and δ34Spyrite)7,8. Various metal isotopes (δ98Mo, δ53Cr, and ε205Tl) have also been measured in these samples23,24,25. In the context of the published data, our Hg isotope data provide strong evidence for recurrent PZE as well as its impact on the oceanic redox evolution and biological innovation during this critical period.

Fig. 1: Paleogeographic location of the study sections in the Yangtze Platform during the deposition of the Doushantuo Formation in South China.
figure 1

The region marked by the red rectangle in panel a corresponds to panel b. The numbers in yellow circles are the study sections: Weng’an (WA), Taoying (TY), Wuhe (WH), Yuanjia (YJ). Reprinted from Jiang, G., Shi, X., Zhang, S., Wang, Y. & Xiao, S. Stratigraphy and paleogeography of the Ediacaran Doushantuo Formation (ca. 635–551 Ma) in South China. Gondwana Res. 19, 831–849 (2011), with permission from Elsevier.

Full size image

Results and discussion

Hg concentration and isotope compositions

In WH, the total Hg concentrations (THg) are >100 ppb except for only two samples. Member IV, which correlates with the previously proposed OOE interval, shows notably higher THg than Member II and III (Fig. 2). To constrain the mechanism of Hg enrichment in sediments, THg is often normalized to major host phases of Hg, including organic matter (proxied by total organic carbon, TOC), sulfide (proxied by pyrite S, Spy) and clay minerals (proxied by Al concentration)26; the ratios of Hg/TOC, Hg/Spy and Hg/Al are all significantly elevated in the Member IV OOE interval of WH (Fig. 2). The TY section, where only the three OOE intervals were analyzed, shows in general the highest THg among all sections and an increasing trend of THg and Hg/TOC from Member II to IV. YJ and WA sections were only analyzed for the Member II OOE interval. In YJ, THg and normalized ratios (Hg/TOC, Hg/Spy and Hg/Al) show a transient elevation at the lowermost of the section, and decreases up-section. In WA, the variations of THg and normalized ratios show no apparent trend. The THg of different intervals is correlated with different host phases, and these correlations are described in details in Supplementary Text S1 and Supplementary Fig. S1.

Fig. 2: Chemostratigraphy of all study sections of the Doushantuo Formation in South China.
figure 2

a Wuhe (WH), b Yuanjia (YJ), c Taoying (TY) and d Weng’an (WA) sections. Fe speciation, redox sensitive elements (U, Mo), δ34Spyrite, P concentration and P/Al are from ref. 8. “TOC” stands for total organic carbon. The horizontal blue bands mark the intervals of previously proposed ocean oxygenation event (corresponding to the E intervals). The gray band in the Δ199Hg and Δ201Hg column of panel a represents the analytical uncertainty of Hg odd-mass independent fractionation (±0.04‰). The error bars for Hg isotopes are analytical uncertainties defined in the method section.

Full size image

The Δ199Hg values (representing MIF of odd isotopes) of WH show cyclic variability, with distinct negative excursions at the lowermost Member II and III, as well as Member IV (down to −0.10‰ in the lowermost Member II), coinciding with three previously proposed OOE intervals8, and gradual positive shifts (up to 0.19‰) between OOEs (Fig. 2). To facilitate discussion, we label the three intervals that show negative excursions of Δ199Hg in Member II, III and IV of WH and the corresponding intervals in other sections as “E1”, “E2” and “E3”, respectively (“E” standards for excursion). In WH, E1 shows the strongest negative excursion (with Δ199Hg decreasing by as much as −0.29‰ relative to the highest value between the E intervals), whereas both E2 and E3 show smaller but still notable negative excursions (with Δ199Hg dropping by −0.19‰). The other deep-water section, YJ, also shows negative Δ199Hg values (down to −0.15‰) at the bottom of E1, which then gradually increase to 0.01‰ up section. In contrast, the two shallower sections, WA and TY, show mostly positive Δ199Hg in the E intervals (up to 0.24‰ and 0.19‰, respectively), which is similar to the positive values in WH between the E intervals.

The δ202Hg values (representing mass dependent fractionation, MDF) of WH show a strong negative correlation with Δ199Hg (the Δ199Hg/δ202Hg slope = −0.09 ± 0.02, R2 = 0.47, P < 0.001), with positive excursions (up to 0.09‰) concurrent with negative shifts in Δ199Hg, and gradual negative shifts (down to −2.60‰) concurrent with positive shifts in Δ199Hg (Figs. 2 and 3). The WA, TY and YJ sections also show negative correlations between Δ199Hg and δ202Hg with similar Δ199Hg/δ202Hg slopes as in WH (Supplementary Fig. S2). Collectively, all four sections reveal a coherent linear Δ199Hg/δ202Hg correlation with a slope of −0.10 ± 0.01 (R2 = 0.62, P < 0.001) (Fig. 3a).

Fig. 3: Stable isotope compositions of Hg in all study sections.
figure 3

a Δ199Hg vs. δ202Hg and b Δ200Hg vs. Δ199Hg for Weng’an (WA), Taoying (TY), Wuhe (WH), and Yuanjia (YJ) sections. Common Hg sources based on modern samples (average ±1 SD) are also plotted to show their possible contributions to Hg isotope variations of the Doushantuo samples. See Supplementary Fig. S5 and Supplementary Text S3 for details and references for modern data. The “Mesoproterozoic PZE” and “Late Devonian PZE” are the Hg isotope data in sedimentary rocks deposited under photic zone euxinia (PZE) from the Mesoproterozoic Atar and El Mreiti groups16, and from the Late Devonian Chattanooga Shale39, respectively. The black short-dash lines are the linear regressions for all Doushantuo samples (in panel a, slope = −0.10 ± 0.01, R2 = 0.62, P < 0.001; in panel b, 0.10 ± 0.01, R2 = 0.55, P < 0.001). The red dash lines are reference lines of experimental results of dark abiotic oxidation of Hg(0) by thiol compounds (in panel a, slope = −0.12; in panel b, slope = 0.13)46. The error bars are analytical uncertainties defined in the method section.

Full size image

In addition to Δ199Hg/δ202Hg, the relationships between different MIF values (Δ200Hg/Δ199Hg and Δ199Hg/Δ201Hg) are also diagnostic of the mechanism of Hg isotope fractionation. The variation of Δ200Hg (representing MIF of even isotopes) in each single section is within the analytical uncertainty, but when all sections are combined, significant positive relationships between Δ200Hg and Δ199Hg, and hence negative correlations between Δ200Hg and δ202Hg can be observed, with a Δ200Hg/Δ199Hg slope of 0.10 ± 0.01 (1SE, R2 = 0.55, P < 0.001) (Fig. 3b) and a Δ200Hg/δ202Hg slope of −0.01 ± 0.00 (1SE, R2 = 0.46, P < 0.001) (Supplementary Fig. S3). Δ199Hg and Δ201Hg also show a significant positive relationship with a Δ199Hg/Δ201Hg slope of 1.49 ± 0.09 (R2 = 0.96, P < 0.001) for all sections (Supplementary Fig. S4).

The influence of Hg sources on Hg isotope compositions in Doushantuo shales

The Hg isotope compositions in the study sections are unlikely modified by diagenetic or metamorphic alterations (see a detailed discussion in Supplementary Text S2), and thus their variations should represent changes in the primary Hg isotope signals of contemporaneous seawater, which could be affected by Hg sources or fractionation processes in ocean. It is important to consider all patterns of Hg isotope fractionation (including the direction and extent of the shifts in both MIF and MDF values as well as their relationships), because together they provide “multi-dimensional” constraints on the mechanism of Hg isotope fractionation and thus allow us to constrain the processes driving these fractionations. Since WH comprises a continuous deposition of black shales from Member II to IV, most discussion below focuses on WH.

The main Hg sources to the modern ocean include atmospheric deposition and terrestrial input21, which have distinguishable Hg isotope signatures (Supplementary Fig. S5 and Supplementary Text S3). Enhanced input of terrestrial Hg was often invoked to interpret negative shifts in Δ199Hg of sedimentary rocks21, based on the fact that modern terrestrial Hg associated with organic matter (OM) derived from soil and biomass typically exhibits negative Δ199Hg27. However, enhanced terrestrial Hg input is unlikely the main mechanism for the negative Δ199Hg shift in the E intervals. Precambrian terrestrial biomass was very scarce compared to today, except fungus-like microorganism found in cryptic karstic environments28, and thus the modern-type OM-associated terrestrial Hg was unlikely a major source of Hg. Before the emergence of land plants, the dominant form of terrestrial Hg was likely geogenic Hg released by weathering and erosion of continental rocks, which is characterized by near-zero MIF29 and thus could not explain the significantly negative Δ199Hg values in E1 of WH and YJ sections. Although the near-zero Δ199Hg in E2 and E3 of WH seem to be consistent with that of geogenic Hg, the correlations between THg and Al (a proxy of continental weathering input) are insignificant in the entire WH section as well as in YJ and two E intervals of TY sections (Supplementary Text S1 and Supplementary Fig. S1), which strongly argue against enhanced terrestrial Hg input as the cause of the negative shift in Δ199Hg during E intervals. In addition, the Hg concentration in crustal igneous and metamorphic rocks is typically <10ppb30,31, much lower than the >100ppb (and frequently >1000ppb) THg in almost all Doushantuo shales, which also argues against terrestrial Hg from continental weathering as a dominant source of Hg. Moreover, we simulated the shift in Δ199Hg caused by enhanced terrestrial weathering using a simplified Hg isotope box-model (see details of the model in Supplementary Text S4)32, and found that even under a scenario of 50× increase in terrestrial Hg input, the Δ199Hg of the ocean can only be decreased by 0.02‰ (Fig. 4), which is an order of magnitude lower than the extent of Δ199Hg excursions during the E intervals (−0.19‰ to −0.29‰). The reason for the muted Δ199Hg shift under high terrestrial inputs is that the enhanced terrestrial input would increase the size of the oceanic Hg reservoir, which would also lead to enhanced re-emission of Hg from ocean to the atmosphere as observed in modern ocean33,34. The re-emitted Hg would undergo atmospheric redox transformations that produce a net positive Δ199Hg signal for atmospheric Hg species35, which then deposit back to surface ocean. Thus, the enhanced re-emission and re-deposition of Hg with positive Δ199Hg following enhanced terrestrial weathering eventually counteracted the negative shift in Δ199Hg caused by terrestrial input. A full investigation on the effects of Hg re-emission and re-deposition on marine Hg isotope signatures is beyond the scope of the current study and is being undertaken in a separate study. Nevertheless, the modeling results further strengthen our argument that terrestrial Hg could not account for the negative Δ199Hg shift during E intervals.

Fig. 4: Numerical simulation results using a simplified Hg isotope box-model.
figure 4

Simulated marine Hg enrichment (b) and Δ199Hg (c) in response to an increase in terrestrial weathering Hg flux by 10–50× during E1 (~635 Ma), E2 (~580 Ma), and E3 (~565 Ma), and a further increase in local volcanic Hg emission by 100× during E3 (a). The “enrichment factor_ocean” value is defined as the marine Hg reservoir during E intervals relative to the reservoir size between E intervals.

Full size image

Atmospheric Hg deposition is the dominant Hg source to global ocean in modern environment36, and thus was likely a major Hg source in the Ediacaran ocean as well. The Hg isotope compositions of modern open ocean seawater are proposed to reflect mixing between atmospheric Hg(II) and Hg(0) deposition36, which have opposite isotope signatures, with positive Δ199Hg, positive Δ200Hg and negative δ202Hg for the former, and negative Δ199Hg, negative Δ200Hg and positive δ202Hg for the latter (Supplementary Fig. S5 and reference therein). Therefore, a simple explanation for the cyclic variation of Hg isotopes in WH is a change in the proportion of atmospheric depositions, i.e., with a higher proportion of Hg(0) deposition during intervals with relatively more negative MIF (the E intervals) and a higher proportion of Hg(II) deposition during intervals with more positive MIF (between the E intervals). However, this simple interpretation is inconsistent with several observations. First, the Δ199Hg/Δ201Hg of both atmospheric Hg(II) and Hg(0) is ~1.0 (typically between 1.0 and 1.1)29, whereas the Δ199Hg/Δ201Hg of Doushantuo shales is ~1.5 (Supplementary Fig. S4), which cannot be solely explained by the mixing of atmospheric Hg(II) and Hg(0). Second, changes in the proportions of atmospheric deposition to modern ocean have been shown to cause corresponding variation of Δ200Hg by more than 0.1‰ in marine sediments36, because of the opposite Δ200Hg for atmospheric Hg(0) and Hg(II) (Supplementary Fig. S5), but this is inconsistent with the invariant Δ200Hg in WH (Fig. 2).

In addition to terrestrial input and atmospheric deposition, upwelling of Hg associated with OM (Hg-OM) from the deep ocean has also been proposed to drive the Hg isotope variation in the Doushantuo Formation by a previous study on coeval shallow shelf lagoon sections (e.g., Jiulongwan section)37. This study found primarily positive Δ199Hg shifts and negative δ202Hg shifts that are roughly in sync with the negative carbonate carbon isotope (δ13Ccarb) excursion in the cap dolostone (Member I) and Member IV in shelf lagoon sections. These Hg isotope signals were proposed to reflect the Hg released during the oxidation of deep ocean OM that is enriched in light carbon isotopes37. The isotopic signatures of Hg associated with deep ocean OM are hypothesized to resemble open ocean sediments38. Since the oxidation of OM is oxygen-demanding, this source should be more prominent during the proposed OOE intervals and in shallow water sections where the upwelling DOC meets the chemocline and is oxidized. In our study sections, Hg-OM is unlikely a dominant source for deep-water sections (WH and YJ), because they actually show opposite shifts in Δ199Hg and δ202Hg during the E intervals than the purported Hg-OM isotope signature (Fig. 2). E1 of the two shallower sections (WA and TY) shows primarily positive Δ199Hg that is similar to modern open ocean sediments, and strong correlations between THg and TOC (Supplementary Text S1 and Supplementary Fig. S1), indicating possible contributions of Hg-OM. However, Hg-OM alone still cannot explain the high Δ199Hg/Δ201Hg slope for all sections (~1.5) (Supplementary Fig. S4), which is significantly higher than those of open ocean sediments (close to 1.0)38.

Therefore, while terrestrial Hg, atmospheric Hg, and deep ocean Hg-OM may all constitute the sources of Hg in the Doushantuo shales, the variations of Hg isotopes could not be fully explained by changes in or mixing of these Hg sources, but require fractionation of Hg isotopes in ocean.

Hg isotope evidence for PZE in the Ediacaran ocean

The distinct negative excursions of Δ199Hg with concurrent positive excursions of δ202Hg during the E intervals were most likely driven by Hg isotope fractionation in seawater under PZE conditions, and this hypothesis is supported by abundant and compelling evidence. First, there exists a strong positive correlation between Δ199Hg and pyrite S isotopes (δ34Spyrite) in WH section (R2 = 0.83, P < 0.001) (Fig. 5). Very negative δ34Spyrite values (as low as −34.9‰) are found during the proposed OOE intervals at WH and have been attributed to an increased availability of marine sulfate, which permits the expression of large-magnitudes of S isotope fractionation during bacterial sulfate reduction7. Increased marine sulfate availability would also increase H2S production, fueling the expansion of euxinia. Thus, the highly synchronized shifts in Δ199Hg and δ34Spyrite suggest a first-order control of redox conditions on the variation of Hg isotopes. Furthermore, the THg of WH shows the strongest correlation with Spy among all host phases in at least two E intervals (corresponding to Member II and Member IV OOEs) (Supplementary Text S1 and Supplementary Fig. S1), suggesting that Hg was primarily scavenged to sediments by sulfide minerals. The strong correlation between THg and sulfide is typically interpreted as a sign of euxinic depositional environment26. In contrast, between the E intervals, THg shows a stronger correlation with TOC than with Spy (Supplementary Text S1 and Supplementary Fig. S1), suggesting that the sedimentation of Hg was controlled by OM, like in modern open oceans21. Thus, the change in host phases from sulfide to OM likely indicates a transition from locally euxinic to non- or less euxinic conditions.

Fig. 5: Positive correlation between Δ199Hg and δ34Spyrite in Wuhe section.
figure 5

The black solid line represents the linear regression between Δ199Hg and δ34Spyrite (R2 = 0.83, P < 0.001).

Full size image

Second, the patterns of Hg isotope excursions (including the direction and extent of the shifts in both Δ199Hg and δ202Hg values as well as Δ199Hg/δ202Hg) in the E intervals are strikingly similar to the Hg isotope systematics from two previously reported successions where PZE was validated independently. Zheng et al. reported Hg isotopic data from Mesoproterozoic black shales (~1.1 Ga) of the Atar and El Mreiti groups, West Africa16, and from the Late Devonian Chattanooga Shale, North America39, and both display comparable negative shifts in Δ199Hg (by as much as ~ −0.2‰) during PZE with similar negative Δ199Hg/δ202Hg of −0.15 and −0.10, respectively (Fig. 3). The PZE condition in the two above-mentioned studies was independently validated by organic biomarkers produced by anoxygenic phototrophic bacteria that metabolize H2S40,41. While indigenous biomarkers are not yet available for the Doushantuo shales42, the highly similar Hg isotope signatures between these successions serves as compelling evidence of a shared driving mechanism.

As proposed previously, PZE can result in concurrent negative Δ199Hg shift and positive δ202Hg shift via two mechanisms16. The first is photoreduction of Hg(II) bound to reduced organic sulfur ligands [i.e., Hg(II)-S] in a sulfide-rich photic zone. Under sulfidic conditions, dissolved organic matter in seawater will likely have a high fraction of thiols due to sulfurization43, favoring the formation of Hg(II)-S complexes. Multiple experimental studies have demonstrated that the photoreduction of Hg(II)-S increases δ202Hg and decreases Δ199Hg in the remaining Hg(II) phase44,45, which is consistent with the pattern of Hg isotope shifts observed in the E intervals. One caveat is that the experimental Δ199Hg/δ202Hg slope of this photoreduction process (~−0.7) is different from that of the Doushantuo shales (−0.10)44, suggesting that the photoreduction of Hg(II)-S is not the only process involved.

The second mechanism is enhanced oxidation of atmospheric Hg(0) in seawater induced by the abundant thiol compounds under PZE conditions. This mechanism is fully compatible with the Hg isotope signatures of Doushantuo shales. Our previous experimental study shows that thiol-induced aqueous Hg(0) oxidation leads to negative shifts in Δ199Hg (by as much as −0.20‰), positive shifts in δ202Hg and no changes in Δ200Hg in the oxidized Hg(II) phase, with a characteristic Δ199Hg/δ202Hg of −0.12 and a Δ199Hg/Δ201Hg of 1.646, which are all fully consistent with the pattern of Hg isotope variation in the E intervals (Fig. 3 and Supplementary Fig. S2). Although it may seem counter-intuitive that Hg(0) oxidation is enhanced in sulfidic water that is typically considered as a reducing environment, multiple experimental studies indeed show that aqueous Hg(0) oxidation can be enhanced by either free thiol or thiol ligands in organic matter46,47. Sulfide also inhibits the re-emission of Hg(0) from seawaters by complexing and scavenging the oxidized Hg to sediments. Thus, the presence of PZE could promote the marine uptake of atmospheric Hg(0) by enhancing its oxidation in the surface ocean and subsequent sequestration to sediments. Moreover, the rate of Hg(0) oxidation in sulfidic waters determined by experiments is comparable to many other major redox pathways of Hg in natural aquatic systems (e.g., photooxidation and photoreduction)47, making the thiol-induced Hg(0) oxidation an environmentally relevant process and potentially the key mechanism that caused the observed shifts in Hg isotopes during periods of PZE.

In contrast to the E intervals, the gradual positive shifts in Δ199Hg with concurrent negative shifts in δ202Hg between the E intervals indicate the contraction of PZE and the transition to modern-like oxygenated surface ocean, which is supported by the change in Hg host phases from sulfide to OM (Supplementary Fig. S1), as well as the similarly positive Δ199Hg and negative δ202Hg in modern ocean sediments36,38,48 (Supplementary Text S3 and Supplementary Fig. S5). Note that this interpretation does not conflict with the Fe speciation data between the E intervals (FeHR/FeT > 0.38, Fepy/FeHR > 0.8) (Fig. 2), which is typically interpreted to reflect local bottom water euxinia49. Sahoo et al. who first reported these Fe speciation data suggested that the high Fepy/FeHR could potentially reflect sulfidic conditions in sediment porewater rather than the water column7. In contrast, Hg isotopes specifically track euxinia in the photic zone, because the two proposed mechanisms controlling Hg isotope fractionation under PZE both operate in surface water where light can penetrate or atmospheric gaseous Hg(0) can dissolve. There is currently no known mechanism that produces characteristic Hg isotope signatures in euxinic (but non-PZE) bottom water or sediment porewater, where sulfide only acts as a ligand that binds with Hg. In fact, the study by Zheng et al.16 found that the Mesoproterozoic deeper-water section deposited under euxinic (but not necessarily PZE) conditions shows no or a much weaker negative shift in Δ199Hg than the shallower section where PZE occurred, suggesting that Hg isotopes track specifically PZE rather than euxinia in general.

Spatial variation of PZE in the South China Basin

The Hg isotope chemostratigraphic data are incomplete for three of the four study sections (YJ, TY, and WA) because they are not completely preserved, thus we are unable to perform a “full profile” reconstruction for redox conditions in these sections as for WH. However, all four sections have Hg isotope data for E1 and they show clear variations of Hg isotopes within E1. By comparing the variation patterns of Hg isotopes during E1 across the four sections, which cover a transect from the shelf to the basin, we are able to evaluate the spatial extent of PZE and the dynamics of the euxinic water mass within the Ediacaran South China Basin. The E1 of WH also shows the strongest negative shift in Δ199Hg among all E intervals, possibly indicating the most extensive PZE event. Thus, our discussion below focuses on E1.

Among the four study sections, the basinal section (YJ) shows the most negative Δ199Hg and most positive δ202Hg at the lowermost E1, and it also exhibits an up-section positive shift in Δ199Hg and concurrent negative shift in δ202Hg with a Δ199Hg/δ202Hg ratio similar to that in WH, suggesting that PZE was likely the most extensive at the onset of E1 but weakened towards the end of E1 at YJ (Fig. 2). In contrast, both the lower-slope (WH) and upper-slope (TY) sections show clear up-section negative shifts in Δ199Hg and concurrent positive shifts in δ202Hg with similar Δ199Hg/δ202Hg ratios, suggesting a gradual expansion of PZE at these two sites. The WA section located on the shallow shelf margin has the most positive Δ199Hg and the most negative δ202Hg among the four sections, and its THg correlates with TOC and Al, but not with sulfide (Supplementary Text S1 and Supplementary Fig. S1), suggesting that WA was likely located above the euxinic water mass or the PZE at WA was too sporadic to be captured by the current data. This is in agreement with previous paleogeographic reconstruction of the Doushantuo Formation in South China, which suggests that WA was likely exposed to oxic/suboxic water above the chemocline22.

Based on the above observations, we can reasonably propose that PZE was widespread during the deposition of the basal Doushantuo shale in South China, extending at least from the upper slope to the basin, which is consistent with the previous proposal for a mid-depth euxinic wedge on continental shelves of the South China Basin11,12. Furthermore, the euxinic wedge probably became shallower approaching the end of E1, and this is evident from the different patterns of Hg isotope variations within E1 across different sections. As discussed above, PZE was weakened at the deepest YJ section towards the end of E1, while PZE intensified at the shallower WH and TY sections, suggesting an up-slope migration of the euxinic water mass toward the end of E1 as depicted by Fig. 6. It is worth noting that E1 coincided with a sea level highstand23. Thus, the up-slope migration of the euxinic wedge towards the end of E1 could be associated with a global or local sea level change, although this hypothesis needs to be tested further.

Fig. 6: Photic zone euxinia (PZE) in the Ediacaran ocean and its impact on ocean oxygenation and the Ediacara biota.
figure 6

Conceptual shifts in Δ199Hg and δ202Hg under the depicted redox conditions are shown in inserted boxes. The numbers in text boxes represent the sequence of events. During intervals of expanded PZE (panel a), PZE developed on continental margins due to increased marine nutrients and sulfate availability (likely due to enhanced postglacial continental weathering, particularly during E1), and was further promoted by the positive feedback from P regeneration under euxinic water. PZE would have enhanced the photoreduction of Hg(II)-S (“S” represents reduced sulfur ligands) and the oxidation of atmospherically deposited Hg(0), leading to the negative shift in Δ199Hg with concurrent positive shift in δ202Hg. Expanded PZE also promoted anoxygenic photosynthesis and inhibited oxygenic photosynthesis, resulting in relatively rapid termination of the previously proposed ocean oxygenation event. Spatial variations in Hg isotopes indicate that PZE may have migrated upslope toward the end of E1, as depicted by the upper euxinic wedge in panel a. During intervals of contracted PZE (panel b), the postglacial surge of weathering may have waned down, leading to the contraction of PZE and resulting in the negative shift in δ202Hg with concurrent positive shift in Δ199Hg. PZE may have also limited the diversification of complex eukaryotic life. At Weng’an (WA) section that was not affected by PZE, the Weng’an biota was found in the upper phosphorite. At Taoying (TY) section where PZE was likely sporadic, the Wenghui biota was found in the upper Doushantuo black shales. In contrast, at Wuhe (WH) and Yuanjia (YJ) sections where PZE or anoxic water persisted, no benthic fossils have been reported.

Full size image

Plausible drivers of PZE

The occurrence of PZE during the previously proposed OOE intervals was likely a result of increased marine sulfate and nutrient availability, which has been previously proposed to trigger the Member II OOE (~635 Ma, E1) due to enhanced continental weathering in the aftermath of the Marinoan glaciation7. During the process of deglaciation, it is hypothesized that a surge of nutrients from freshly exposed land surfaces and finely ground glacial tills was delivered to the ocean, which increased primary productivity and thus oxygen consumption by newly produced organic matter, leading to anoxia on a timescale of ~104 years2,50. The enhanced primary productivity would also increase the organic matter available for microbial sulfate reduction and hence seawater H2S accumulation in near-shore environments, leading to PZE. On a longer timescale (~106 years), increased nutrient influx would increase organic carbon burial and hence atmospheric oxygen50. A similar sequence-of-events may have followed the Gaskiers glaciation, which occurred broadly coeval with the Member III OOE (~580 Ma, E2)8. However, this link is speculative given the uncertain timing of the Member III OOE, and it merits further examination.

The increased availability of nutrients in global oceans after the Marinoan glaciation is well-documented by the apparent expansion of the marine phosphorous (P) reservoir51,52,53,54. Phosphorous is a key limiting nutrient and increased P influxes to the ocean have been proposed to account for the development of PZE in many cases across Earth history, particularly during various Phanerozoic mass extinction events55,56, and in modern meromictic lakes (e.g., the Mahoney Lake)57. Both P and P/Al (to account for changes in sedimentation rate) in the WH section show a cyclic variation similar to Hg isotopes and other redox proxies, with lower P during the E intervals and higher P in between (Fig. 2). The accumulation of P in sediments is controlled by local redox conditions, and euxinic conditions would re-dissolve and recycle P in sediments back to the water column due to rapid reduction of P-bearing Fe oxides by H2S and preferential release of P from remineralization of organic matter by sulfate-reducing bacteria58. Thus, the variation of sedimentary P is consistent with the local redox conditions at WH, and suggests intensive recycling and regeneration of P modulated by a rising marine sulfate reservoir54, which may have stimulated PZE via positive feedback.

The Member IV OOE (E3) of both WH and TY sections shows higher Hg enrichments and different patterns of Hg isotopes compared to other OOEs, which we ascribe to Hg input from local volcanic emission (for WH) or upwelling of deep ocean Hg-OM (for TY) (see detailed discussion in Supplementary Text S5). Despite the influences of volcanic or upwelling Hg, the E3 interval of WH still shows a strong correlation between THg and Spy (Supplementary Fig. S1). Furthermore, we also simulated the shift in Δ199Hg caused by enhanced local volcanic Hg input using the Hg isotope box-model (more details in Supplementary Text S4), and found that an increase in local volcanic Hg flux by 100× could only decrease Δ199Hg by <0.03‰ (Fig. 4), which is similar to the muted Δ199Hg shift caused by terrestrial Hg input. As explained earlier, the reason for the muted Δ199Hg shift might be due to the re-emission of volcanically-sourced Hg from ocean and the re-deposition of atmospheric Hg with positive Δ199Hg back to ocean. Thus, the negative excursion E3 with a drop in Δ199Hg by −0.19‰ may also suggest PZE. This hypothesis is consistent with the previous proposal for widespread euxinia on continental shelves based on V isotopes in black shales of Member IV59. Volcanism and upwelling may also lead to enhanced nutrient availability, which may trigger PZE, although this hypothesis requires further verification.

The negative impacts of PZE on ocean oxygenation and the Ediacara biota

PZE may have exerted a negative impact on the oxygenation of the Ediacaran ocean by promoting anoxygenic photosynthesis and inhibiting oxygenic photosynthesis (Fig. 6). This may be an underappreciated mechanism for the transient nature of the Ediacaran OOEs. While oxygenic photosynthesis is the overwhelming source of free O2, it can be outcompeted by anoxygenic photosynthesis under conditions that favor the latter, such as extensive PZE60,61 or ferruginous conditions62. Compelling experimental evidence has shown that even μM levels of H2S can completely shut down oxygenic photosynthesis63. It has been suggested that the competition for light and nutrients between anoxygenic and oxygenic photosynthetic organisms was a key mechanism responsible for the delayed oxygenation of Earth’s atmosphere62, and for the relatively low level of oxygen throughout the mid-Proterozoic18. Although our Hg isotope data only cover South China, lipid biomarkers for PZE have been reported in late-Ediacaran petroleum from Eastern Siberia64, suggesting that PZE may have been common in various continental margins during the Ediacaran Period. Continental margins are the major habitats for early primary producers such as cyanobacteria and ancestral photosynthetic eukaryotes61. Thus, we propose that the widespread PZE in continental margins initiated a negative feedback on oxygenic photosynthesis, which may be partially responsible for the relatively rapid termination of OOE. This argument is consistent with a recent hypothesis based on theoretical models that found the cyclic and oscillating oxygenation and deoxygenation events during the Ediacaran Period can be explained by internal feedbacks in the biogeochemical cycles of carbon, oxygen and phosphorous, without the need for specific external forcing65. As an important step forward, our finding of PZE provides an independent validation and mechanistic insights into this theoretical hypothesis.

Recurrent and widespread PZE in the Ediacaran ocean may have also inhibited the diversification and ecological expansion of early macroscopic eukaryotic life by limiting their habitable space via H2S poisoning. H2S toxification during PZE was a potent kill mechanism for some of the largest mass extinctions, such as the end-Permian20,32,66 and end-Triassic mass extinctions67. Emerging geochemical data suggest a redox control on the rise and fall of the Ediacara biota4,5,68. The unstable redox conditions and repeated development of PZE in early Ediacaran oceans as revealed by Hg isotopes likely limited the ecological expansion of macroscopic multicellular eukaryotes. For example, the lower slope and basinal facies, as represented by the WH and YJ sections, were strongly influenced by PZE during the OOEs (the E intervals) and were consistently anoxic between OOEs, excluding the establishment of benthic multicellular eukaryotes that require free oxygen. In relatively shallow-water shelf and upper slope facies, as represented by the WA and TY sections, eukaryotes were able to colonize the benthic realm where and when PZE was absent. For example, multicellular eukaryote fossils of the Weng’an biota at WA occur in phosphorite strata correlated with Member II above the E1 interval69. Similarly, macroalgal fossils of the Wenghui biota at TY occur in black shales correlated to the Member III above the E2 interval (Fig. 2). These fossiliferous intervals represent locally oxic environments (as indicated by negative Ce anomalies in fossiliferous phosphorite70) that were habitable for early Ediacaran eukaryotes. A limitation of this study is that we did not directly measure samples from fossiliferous intervals. Thus, future studies that compare fossiliferous and non-fossiliferous intervals are needed to further verify the proposed impact of PZE on the Ediacara biota.

In summary, our study reveals recurrent, widespread and spatially dynamic PZE on the continental margin of South China during previously proposed OOE intervals of the Ediacaran Period. Although the presence of a dynamic euxinic wedge at South China during the Ediacaran Period has been proposed previously, our study shows strong evidence for the invasion of euxinic water mass into the marine photic zone, which has important implications for understanding its impacts on the redox evolution and biological innovation in the Ediacaran ocean. We suggest that widespread PZE at continental margins was a consequence of increased marine nutrient and sulfate reservoirs due to their positive feedback on primary productivity, but PZE may have also initiated a series of negative feedbacks that inhibited oxygen production (and thus decelerated ocean oxygenation) by promoting anoxygenic photosynthesis and limiting the habitable space for eukaryotes carrying out oxygenic photosynthesis. Therefore, PZE could be one of the key reasons for the transient nature of ocean oxygenation during the Ediacaran Period and the delayed rise of the Ediacara biota.

Methods

Study sections

The locations, stratigraphy, and paleogeography of the study sections are described in detail in previous work7,8,22, and thus only a brief overview is presented here. The Doushantuo Formation is best known in the Yangtze Gorges area where it is divided into four lithostratigraphic members. Member I serves as a regional stratigraphic marker bed and corresponds to the cap carbonate overlying the terminal Cryogenian-age glacial diamictite of the Nantuo Formation. A zircon U-Pb date from a volcaniclastic layer atop Member I constrains its deposition age to be 635 ± 0.6 Ma71. Member II consists primarily of black shales with subordinate carbonate beds. Member III is dominated by carbonate although black shale becomes increasingly abundant toward deep-water facies outside the Yangtze Gorges area. The Member II-III boundary is correlated with a regional stratigraphic discontinuity in shallow-water facies22, and if this regional discontinuity tracks a widespread marine regression, it may be time-equivalent with the Gaskiers glaciation (~580 Ma)8. Member IV is dominated by organic-rich black shales, and can also be used as a regional stratigraphic marker bed for the Doushantuo Formation in South China. The age of Member IV is a topic of ongoing debate, but it is probably ~565 Ma72. The shelf and upper slope sections (WA and TY) contain Ediacaran microfossils of the Weng’an biota14 and macrofossils of the Wenghui biota73, respectively. No fossils have been reported from the WH and YJ sections that represent lower slope and basinal facies.

Hg concentration analysis

Mercury concentrations in solid samples were analyzed on a Lumex R-915F Hg Analyzer at Tianjin University. Mercury in samples was released as Hg(0) vapor by combustion at a temperature of ~750 °C, and then measured by a cold vapor atomic absorption spectroscopy (CV-AAS) with a detection limit lower than 0.5 ng/g. A certified reference material GBW07311 (GSD-11, freshwater sediment) was measured repeatedly alongside samples to monitor analytical accuracy and reproducibility. The GSD-11 yielded an average Hg concentration of 72.2 ± 5.8 ng/g (2 SD, n = 8), which is consistent with its certified value 72 ± 9 ng/g.

Hg isotope analysis

Mercury isotopes were analyzed using multi-collector inductively coupled plasma mass spectrometry (MC-ICPMS, Neptune Plus, Thermo Fisher Scientific) at School of Earth System Science, Tianjin University based on published methods74. Prior to isotopic analysis, Hg in samples was first extracted by acid digestion and then purified by ion-exchange chromatography (see details in Supplementary Text S6)16. To monitor Hg yields and the accuracy of Hg isotope measurement, procedural blanks and three standard reference materials (SRM), including GBW07405 (Yellow/red soil), NIST 2702 (Inorganics in marine sediment), USGS SBC-1 (Brush Creek Shale), and the NIST SRM 3133 Hg isotope standard, were processed alongside samples. Recoveries for all SRM were 100 ± 19% (2 SD, n = 19) (Supplementary Table S1). Mercury concentrations in procedural blanks were typically less than 1% of the Hg present in samples.

Solutions eluted from the chromatographic procedure were diluted to 1.5 ng/g of Hg using a matrix solution containing 5% HCl. Thereafter, Hg in the diluted solutions was reduced by SnCl2 (3%, w/v) to gaseous Hg(0), which was then carried into the plasma of MC-ICPMS by Hg-free argon gas. Simultaneously, thallium (Tl) aerosol (NIST SRM 997) generated by Aridus II desolvator was introduced together with Hg(0) vapor into the plasma. Five Hg isotopes (198Hg, 199Hg, 200Hg, 201Hg, and 202Hg) and two Tl isotopes (203Tl, 205Tl) were simultaneously measured via Faraday cups. Instrumental mass bias was corrected for using a combination of internal calibration with measured 205Tl/203Tl ratios and standard-sample-standard bracketing (relative to the NIST SRM 3133 Hg standard). The bracketing standard was matched to samples in terms of both matrix and Hg concentration (less than 10% difference). On-peak zero corrections were applied to all measured Hg masses. Mercury isotope compositions are reported using δ notation defined by the following equation:

$${\delta }^{x}{{{{{\rm{Hg}}}}}}(\textperthousand )=\left[ \frac{{({\,}^{x}{{{{{\rm{H}}}}}}{{{{{\rm{g}}}}}}/{\,}^{198}{{{{{\rm{H}}}}}}{{{{{\rm{g}}}}}})}_{{{{{{\rm{sample}}}}}}}}{{({\,}^{x}{{{{{\rm{H}}}}}}{{{{{\rm{g}}}}}}/{\,}^{198}{{{{{\rm{H}}}}}}{{{{{\rm{g}}}}}})}_{{{{{{\rm{std}}}}}}}}-1 \right ] \times 1000$$
(1)

where XHg is 199Hg, 200Hg, 201Hg, or 202Hg, and “std” represents the NIST SRM 3133 standard. The MDF is reported as δ202Hg, and MIF is reported as the capital delta notation (Δ) according to the following equation:

$$\varDelta {}^{x}{{{{{\rm{H}}}}}}{{{{{\rm{g}}}}}}(\textperthousand )=\delta {}^{x}{{{{{\rm{H}}}}}}{{{{{\rm{g}}}}}}-(\delta {}^{202}{{{{{\rm{H}}}}}}{{{{{\rm{g}}}}}} \times \beta )$$
(2)

where x is the mass number of Hg isotope 199, 200, and 201. β is a scaling constant used to calculate the theoretical kinetic MDF, and it is 0.2520, 0.5024, and 0.7520 for 199Hg, 200Hg, and 201Hg, respectively75.

To ensure data quality, each sample was measured at least twice, and a commonly used reference standard NIST 8610 was measured every 6 − 7 samples to monitor instrument performance. The averages of all NIST 8610 are: δ202Hg = −0.54 ± 0.06‰, Δ199Hg = −0.03 ± 0.04‰, Δ200Hg = 0.00 ± 0.02‰ (2 SD, n = 39), consistent with the published values75. The NIST SRM 2702 yielded average δ202Hg, Δ199Hg and Δ200Hg values of −0.77 ± 0.04‰, −0.03 ± 0.06‰, and 0.00 ± 0.02‰ (2SE, n = 3), respectively (Supplementary Table S1). All isotope data are reported in Supplementary Table S2 and analytical uncertainties are reported as either 2 standard error (2SE) of sample replicates or 2 SD of all measurements of the NIST 8610, whichever is higher.