The ca. 183 Ma Toarcian Oceanic Anoxic Event (T-OAE) represents a significant perturbation to Earth’s atmosphere–ocean system and was associated with a global mass extinction1,2. The 300–500 kyr event3–5 coincides with and is thought to have been triggered by emplacement of the subaerial Karoo–Ferrar large igneous province (LIP) in southern Pangea1,3,6. The environmental perturbances triggered by LIP emplacement included a dynamic climate7–9, enhanced hydrological cycle and continental weathering9–11, increased organic productivity3, and a significant expansion of seafloor anoxia and euxinia12,13. Despite the OAE naming convention which emphasizes the importance of expanded seafloor anoxia, the global extent of bottom water anoxia ([O2]aq = 0 ml/l)14 across the T-OAE remains poorly constrained, as is the relative extent of ferruginous anoxia (Fe2+-rich bottom-waters) and euxinic anoxia (H2S-rich bottom-waters).
The global extent of ocean euxinia during the T-OAE has been inferred from molybdenum isotope compositions (δ98Mo) in organic-rich mudrock (ORM). When the oceans are primarily well-oxygenated, manganese oxides enriched in lighter-mass Mo isotopes are buried over a larger portion of the seafloor, resulting in global seawater with a higher δ98Mo. By contrast, the low δ98Mo of locally euxinic ORM from multiple sections in Europe suggest that global seafloor euxinia expanded and oxic seafloor contracted during the early stages of the T-OAE15–17. Euxinic waters have been estimated to cover 2–10% of the Toarcian seafloor, which is one to two orders of magnitude greater than the modern ocean (~0.1–0.2%)15,17,18. However, suggestions that the Toarcian sections of Europe were deposited in sedimentary basins highly restricted from global ocean circulation have called these interpretations into question8,19−21. Importantly, Mo isotope data do not reliably constrain the extent of ferruginous seafloor because the magnitude of isotope fractionation during Mo removal to sediments under Fe2+-rich bottom waters overlaps with fractionations observed for Mo burial in weakly euxinic to mildly oxygenated settings22.
Thallium isotope compositions (ε205Tl) from ORM have also been applied to reconstruct the global ocean redox landscape during the T-OAE. Preferential burial of isotopically heavy Tl adsorbed to Mn oxides deposited in well-oxygenated seafloor sediments shifts seawater ε205Tl towards lighter values, whereas limited areas of oxic seafloor will shift seawater to heavier values. In the early Toarcian, two distinctive shifts towards heavier ε205Tl are captured in ORM from western Canada, one ~500 kyr before a negative carbon isotope excursion (N-CIE) and another at the onset of the N-CIE, which are thought to indicate two expansions of ocean anoxia23. The highest ε205Tl values at the onset of the N-CIE indicate that burial of Mn oxides decreased by at least 50% during the T-OAE, suggesting an equivalent increase of weakly-oxygenated, ferruginous, and/or euxinic bottom waters23. However, the Tl isotope proxy does not deconvolve the relative seafloor areas covered by these three O2-deficient redox settings.
An alternative approach to quantify the overall extent of oceanic anoxia—including ferruginous seafloor—during the T-OAE is required. Here, we apply a recently developed elemental mass balance model24 that uses rhenium and Mo enrichments from Pliensbachian to Toarcian ORM (Gordondale Member, Fernie Formation, British Columbia, Canada; Figure 1) to quantitatively estimate total anoxic seafloor area (euxinic plus ferruginous) and total euxinic seafloor area, respectively, before and during the T-OAE. Both Re and Mo are ideal trace metals to quantify the global redox conditions in the oceans due to their conservative behaviour in oxygenated seawater, long ocean residence time (130–440 kyr today)25, and low detrital background in organic-rich marine sediments18,24. High Re burial rates, and thus Re enrichments, in marine sediments occur in both ferruginous and euxinic environments via complexation of Re with organic matter and sulfide minerals26, but negligible Re enrichment occurs in oxygenated sediments. Molybdenum preferentially forms thiomolybdate complexes in the presence of dissolved H2S and the burial efficiency of sulfurized Mo species in euxinic sediments is significantly elevated compared to Mo burial in non-euxinic settings.
These distinctive redox characteristics mean that a predominantly well-oxygenated global ocean will have large dissolved reservoirs of Re and Mo sourced from rivers, and thus pronounced enrichments will occur in ORM that cover small areas of the open ocean within oxygen minimum zones on continental margins. By contrast, an expansion of anoxic and euxinic marine sediment sinks will draw down the dissolved oceanic Re and Mo reservoirs, respectively, leading to muted enrichments in ORM deposited in unrestricted marine settings18,24. Using these principles, Re and Mo oceanic mass balance models can be used to infer the extent of seafloor anoxia and euxinia, respectively, using their enrichments in locally anoxic or euxinic ORM deposited in unrestricted marine settings. The extent of ferruginous seafloor can be calculated as the difference between the total anoxic and euxinic seafloor areas.
Local paleoenvironmental setting of the T-OAE in British Columbia
The T-OAE is expressed globally by synchronous deposits of ORM recording a N-CIE in organic and carbonate carbon, embedded within a broader positive CIE27–33. The T-OAE was identified in the Red Deer Member (Fernie Formation) of Alberta, Canada in cores (1-35-62-5W6 and 6-32-75-20W5) and outcrop (Bighorn Creek East Tributary) by a N-CIE with a magnitude of –3‰ to –4‰27,34. We report new elemental and organic carbon isotope data (δ13Corg) from a drill core (c-B6-A/94-B-8) in British Columbia, Canada, that hosts equivalent Fernie Formation strata (Gordondale Member). This core contains the N-CIE across a vertical interval of 7.1 m and has a magnitude of –2.0 to –2.4‰ (Figure 2). Using this δ13Corg profile, the Gordondale Member can be subdivided into Pre N-CIE (>1582.10 m), N-CIE (1582.10–1575.00 m) and Post N-CIE (< 1575.00 m) intervals. A second N-CIE is observed below 1587.00 m but may represent an older Sinemurian or Pliensbachian event34,35. Direct age constraints are not available within the study core, however, detailed litho- and chemostratigraphic correlations with Early Jurassic sections in Alberta containing ammonite biostratigraphy, Re–Os (ORM) and U–Pb (bentonite) ages constrain the age of the Toarcian N-CIE in the core to between ~185 and 182 Ma (see SI).
Local bottom water redox conditions during deposition of the Gordondale Member in our study section can be constrained using redox-sensitive metals like uranium and vanadium. Like Re and Mo, these metals are soluble in oxygenated seawater and have low burial rates into oxygenated sediments36. Both Mo and V require bottom water euxinia for large authigenic enrichments (Xauth; Equation S1) in sediments, whereas Re and U do not37–44. Thus, low or mild Vauth and Moauth enrichments but elevated Uauth (> 10 µg/g)43 and Reauth (> 42 ng/g)40 enrichments in sediments suggest deposition from locally ferruginous bottom waters. Elevated enrichment of all four metals (with Vauth > 320–500 µg/g; Moauth > 100 µg/g)39,42,44 suggests deposition from locally euxinic bottom waters. Sediments deposited from euxinic bottom-waters have lower Re/Moauth (< 4)45 and U/Moauth (< 0.13)40,45−47 ratios whereas higher ratios are expected in sediments deposited from ferruginous (Re/Moauth = 4–10)40,45 or suboxic (Re/Moauth > 10)40,45 bottom-waters.
The Gordondale Member hosts elevated Reauth (207 ± 91 ng/g, 1 s.d.) and Uauth (13 ± 8 µg/g, 1 s.d.), like many Phanerozoic ORM deposited from anoxic bottom waters24,43. By contrast, Vauth (764 ± 568 µg/g, 1 s.d.), Moauth (96 ± 88 µg/g, 1 s.d.), Re/Moauth (4.7 ± 4.4, 1 s.d.) and U/Moauth (0.30 ± 0.36, 1 s.d.) are variable. The divergence between anoxic (Uauth, Reauth) and euxinic (Moauth, Vauth, Re/Moauth, U/Moauth) indicators reveals that deposition occurred under both ferruginous and euxinic bottom water conditions, and rarely under suboxic conditions. Large variance (>±70% of the mean) remains when comparing the euxinic indicators in the Pre N-CIE and N-CIE intervals. Hence, we further subdivide the Gordondale Member into Lower Pre N-CIE (euxinic), Upper Pre N-CIE (ferruginous), Lower N-CIE (ferruginous) and Upper N-CIE (euxinic) intervals. The greatest enrichments occur in the Lower Pre N-CIE (Reauth = 215 ± 88 ng/g; Moauth = 168 ± 91 µg/g; 1 s.d.) and Upper N-CIE (Reauth = 277 ± 74 ng/g; Moauth = 108 ± 22 µg/g; 1 s.d.) intervals and the smallest enrichments occur in the Upper Pre N-CIE (Reauth = 125 ± 68 ng/g; Moauth = 23 ± 11 µg/g; 1 s.d.) and Lower N-CIE (Reauth = 192 ± 95 ng/g; Moauth = 33 ± 18 µg/g; 1 s.d.) intervals. Samples that were likely deposited under suboxic conditions (Re/Moauth > 10) were not included in the mean calculations. The low Moauth in the Upper Pre N-CIE and Lower N-CIE is probably not due to an expansion of global euxinia drawing down the oceanic Mo reservoir18 because Reauth (and Uauth) should show a similar drawdown, which is not observed. Instead, the decline in Moauth and increased U/Moauth and Re/Moauth in the middle portion of the core are interpreted as a transition from euxinic to ferruginous anoxia in the local depositional environment.
Covariations of Mo versus U and cadmium versus Mo can be used to determine paleohydrographic conditions during deposition of the Gordondale Member47,48 (see discussion in SI). The Mo–U covariation trend for the Gordondale ORM suggests vigorous water-mass exchange with the open ocean47. The Cd–Mo covariation corroborates this interpretation, and the elevated Cd content further suggests an environment where upwelling of deep-water nutrients enhanced primary productivity48. Based on the local hydrographic regime and its deposition from ferruginous or euxinic bottom waters, the Gordondale Member core is ideal for estimating global ocean redox conditions from redox-sensitive metal concentrations.
Mass Balance Modeling of Global Seafloor Total Anoxia and Euxinia
First-order approximations of global anoxic and euxinic seafloor area can be inferred from Reauth and Moauth enrichments in ORM deposited in areas well-connected to global ocean circulation18,24. Modern seawater Re and Mo concentrations are regulated primarily by rivers with minor seafloor hydrothermal inputs25, and burial into sediments deposited from oxic, suboxic, and anoxic (Re) or euxinic (Mo) bottom waters18,24. Rhenium is removed most efficiently to the anoxic sink as indicated by an average burial rate of 1.3 ng/cm2yr, which is one to three orders of magnitude greater than suboxic (4.2 × 10−1 ng/cm2yr) and oxic (1.6 × 10−3 ng/cm2yr) settings24. Burial rates of Re into sediments beneath ferruginous and euxinic bottom waters are not significantly different because Re removal to organic-rich sediments depends on O2 deficiency rather than H2S availability24,40. In contrast, Mo removal to these sediments hinges on the presence of free H2S in bottom waters42 such that there is a large offset between average Mo burial rates in oxic (2.75 × 10−3 µg/cm2yr) and suboxic (0.27 µg/cm2yr) versus euxinic (1.53 µg/cm2yr) sediments18. The high Re and Mo burial rates in anoxic and euxinic settings, respectively, means that a relatively minor expansion of anoxic or euxinic seafloor will result in substantial drawdown of these metals from seawater until a new steady state is reached. A smaller oceanic metal reservoir should lead to lower enrichments in marine ORM deposited from locally ferruginous or euxinic waters. Hence, Re and Mo enrichments in ORM, coupled with a recently developed mass balance model24 (Equations S2–S5) can be used to track changes in the extent of global ocean anoxia and euxinia associated with the T-OAE.
Application of the mass balance model to data from the Gordondale Member before and during the T-OAE requires samples selected from locally anoxic (Re) or euxinic (Mo) ORM18,24. Thus, the euxinic seafloor area from the Mo mass balance is not calculated for the locally non-euxinic Upper Pre N-CIE and Lower N-CIE intervals, but the anoxic seafloor area can be deduced from all four defined Gordondale Member intervals. To produce a representative model of the change in global anoxic and euxinic seafloor area before and during the T-OAE, local spatiotemporal parameters were applied to these stratigraphic intervals. We assess the impact on mass-balance model solutions by local bulk sediment mass accumulation rate (Equation S6), thermal maturity (Equation S7), and Early Jurassic Re and Mo input fluxes from rivers and hydrothermal sources. We present an environmentally realistic scenario, with further sensitivity analysis described in the SI.
The modern Re and Mo riverine fluxes to the oceans are 4.29 × 105 mol/yr and 3.00 × 108 mol/yr, respectively24,25, which is the assumed baseline (pre N-CIE) flux given broadly similar atmospheric O2 levels at that time compared to modern49. Riverine fluxes scale with the rate of continental weathering, which is estimated to have increased by 215–530% over a 100–200 kyr period at the onset of the T-OAE based on Os and Ca isotope data10,11. The Os isotope data is from the Red Deer Member10, which is closely related to the Gordondale Member in the c-B6-A/94-B-8 core, however, a sill between the units may have created minor basin restriction during Red Deer Member deposition50. The c-B6-A/94-B-8 core contains consistently elevated authigenic Re (208 ± 91 ng/g, 1 s.d.) and total organic carbon (TOC) content (6.4 ± 2.2%, 1 s.d.), while the East Tributary section hosting the Red Deer Member contains variably low Re concentrations (52 ± 49 ng/g, 1 s.d.)10 and moderate TOC content (3.9 ± 1.0%, 1 s.d.)27. The lower Re contents at East Tributary is not a result of more oxygenated conditions because sedimentary Fe speciation data indicates deposition from locally euxinic bottom waters51. Hence, the minor hydrographic restriction at the Red Deer Member depositional locality could have caused local seawater 187Os/188Os to be higher than global seawater such that the weathering rate evaluation (215–530% increase) from the East Tributary section10 may be a mild overestimation. Hence, we apply a conservative threefold increase of Re and Mo riverine flux at the onset of the N-CIE and maintain this flux through both Lower and Upper N-CIE intervals.
Hydrothermal Re and Mo fluxes to modern seawater are poorly constrained but are likely minor compared to riverine flux24,25. Sensitivity analyses of hydrothermal fluxes demonstrate that its subordinate contribution compared to river flux obviate its impact on the Re and Mo mass balance (see SI), thus the hydrothermal flux was excluded from the mass-balance model.
Maximum Extent of Anoxia at the Onset of the T-OAE
The Re model yields the proportion of total seafloor area covered by anoxic sediments (Aanoxic). The results fall within a narrow range for the Lower Pre N-CIE (0.14–2.5%), Upper Pre N-CIE (1.1–8.2%) and Upper N-CIE (2.9–6.4%) intervals (Figure 3A,B). The Lower N-CIE interval has the greatest uncertainty in Aanoxic (3.9–100%) due to the asymptotic nature of the model when Aanoxic is less than 1% or greater than 10% total seafloor area. Variation in Aanoxic is rooted in the Reauth variance for each interval, which likely reflects, at least in part, changes in local depositional conditions such as fluctuating redox conditions, host phase availability/composition or sedimentation rates52, rather than global redox changes. It is important to note that lower Re concentrations, caused by local depositional factors, do not actually convey meaningful information about the global extent of ocean anoxia. Hence, mean Re concentrations are a more robust global redox indicator and were used to calculate mean anoxic seafloor areas of 0.84%, 2.6%, 6.9% and 4.1% of the global seafloor area for the respective stratigraphic intervals. These estimates may reflect maximum constraints for the extent of seafloor anoxia because Re concentrations higher than the means suggest lower extents of seafloor anoxia.
The areas of euxinic seafloor from the Mo mass balance model (Aeuxinic) also fall within a narrow range for the Lower Pre N-CIE (0.15–1.4%) and Upper N-CIE (3.9–6.2%) intervals and are independently stratigraphically consistent with Aanoxic (Figure 3C,D). Mean Aeuxinic estimates are 0.47% and 4.8% for the Lower Pre N-CIE and Upper N-CIE intervals, respectively. As a subclass of anoxia, euxinia should cover a seafloor area equal to or less than total seafloor anoxia. The range of Aeuxinic falls within the range of Aanoxic for both Lower Pre N-CIE and Upper N-CIE intervals, and the mean Aeuxinic for both intervals is not appreciably different considering model uncertainties (p-values > 0.16 for paired t-tests).
The Re mass balance model yields a maximum area of seafloor anoxia at the onset of the N-CIE and a contraction of anoxic seafloor area towards the conclusion of the N-CIE. An areal decrease of globally anoxic bottom waters through the T-OAE is consistent with other global redox reconstructions from Europe and Canada16,51. The Pre N-CIE interval exhibits global ocean redox conditions similar to modern, however, a minor expansion of seafloor anoxia directly preceding the N-CIE suggests that the T-OAE began prior to the classically recognized δ13C expression of the event. A similar observation was noted in the Tl isotope record from another Canadian T-OAE section51. This early expansion was not documented in global redox reconstructions from Europe because pre-N-CIE bottom water conditions recorded in those sections are thought to have been generally oxic15,16.
Combining the findings from the Re and Mo models illuminates the redox structure of the global ocean before and during the T-OAE. The offset between mean Aanoxic and Aeuxinic in the Lower Pre N-CIE and Upper N-CIE intervals are not statistically different, suggesting euxinic rather than ferruginous conditions covered most of the anoxic seafloor, as observed in modern anoxic environments.
The increase in seafloor area covered by euxinic conditions likely played a role in the extinctions that occurred during the T-OAE. Several studies have invoked anoxia, and specifically sulfidic anoxia, as a likely kill mechanism for marine fauna during other mass extinctions53–55. The modern continental shelf and slope where biota is most heavily concentrated account for 3.6% and 5.6% of the global ocean area, respectively, totalling 9.2%56. If the area of Toarcian continental shelf and slope were not substantially different from the present, then our model suggests that the maximum expansion of anoxia (primarily as euxinia) in the early portion of the T-OAE did not extend into the deep ocean but remained within the continental margins (e.g., Panthalassa ocean margins)27,30−33 or epicontinental seas (e.g., Tethys sea sub-basins)15–17,19−21,28,29. The Re mass-balance model results suggest that the open deep oceans away from continental margins likely remained predominantly oxygenated during the T-OAE, although potentially below modern levels57, contrasting with past statements suggesting “widespread” anoxia1,15. Our results align with estimates from previous Mo isotope studies which suggested euxinia covered < 10% of the global seafloor16,17. Our novel approach of using Re and Mo mass-balance models to independently estimate the total anoxic seafloor and euxinic seafloor areas show that even modest expansions of anoxia expressed as euxinia can be associated with catastrophic marine die-offs, as reported for the Toarcian2.
Our approach towards inferring global ocean redox conditions may be applicable to other intervals of the Phanerozoic characterized by severe mass extinctions, or in forecasting changes to the modern world. The modern Earth is likely experiencing dramatic surface changes such as an enhanced greenhouse atmosphere spurred by industrial, rather than volcanic, activity. The resulting environmental feedbacks associated with the accumulation of anthropogenic greenhouse gases is leading to species loss akin to those of past climate events58. Application of the coupled Re–Mo paleoredox proxy may improve understanding of the magnitude of ocean redox changes during ancient expansions of oceanic anoxia and euxinia and their effects on biodiversity, and thus may enable prediction of the results of human impacts on the biosphere. The coupled Re-Mo paleoredox proxy may also help track longer-term changes in the distribution of euxinic versus ferruginous seafloor throughout the Phanerozoic Eon.