Silicate weathering represents a major feedback mechanism in the Earth’s climate system, helping to stabilize atmospheric CO2 levels and temperature on million-year time scales. On shorter time scales of greater relevance to understanding the fate of anthropogenic CO2, the efficacy and responsiveness of weathering is less clear. Here, we present high-resolution osmium-isotope data that reflect global chemical weathering from a stratigraphically thick record of the early Toarcian oceanic anoxic event (T-OAE; ca. 182 Ma). A pronounced decrease in the carbon-isotope composition of exogenic carbon reservoirs during this event has been linked to the large-scale release of 12C-enriched carbon. Our data indicate that the flux of radiogenic osmium to the oceans increased in lockstep with the decrease in carbon-isotope values, demonstrating a geologically synchronous coupling between massive carbon release and enhanced global continental crust weathering. We show that abrupt shifts in carbon isotopes, previously interpreted as millennial-scale methane hydrate melting or terrestrial carbon-release events, are coeval with rapid increases in weathering. Global weathering may have increased by >40% across each of these intervals of rapid carbon injection. Our results help to reconcile previous estimates of weathering change during the T-OAE, and support the view that, overall, global weathering rates may have increased six-fold through the entire event.

Determining the response time and mechanistic links between rapid, large-scale climate change and weathering in deep time is challenging because of the typically low temporal resolution of geological records of climate change and the need for proxies that can unambiguously track globally significant changes in weathering on short time scales. The early Toarcian oceanic anoxic event (T-OAE; ca. 182 Ma) represents an ideal interval for studying weathering responses to inferred large-scale atmospheric CO2 and temperature change. A pronounced decrease in the carbon-isotope (δ13C) composition of exogenic reservoirs through this event, coupled with associated evidence for seawater warming and atmospheric CO2 increase, has been linked to the release of thousands of gigatons of 12C-rich carbon to the biosphere (Hesselbo et al., 2000; McElwain et al., 2005; Kemp et al., 2005; Ruebsam et al., 2019).

Osmium-isotope (187Os/188Os) data from a handful of previous studies have demonstrated that the flux of radiogenic 187Os derived from continental crust weathering increased significantly during the T-OAE (Cohen et al., 2004; Percival et al., 2016; Them et al., 2017a; van Acken et al., 2019). Nevertheless, the precise timing and pattern of this increase, and its exact relationship to inferred carbon release, are uncertain. In particular, the decrease to minimum δ13C values has been shown to comprise a series of abrupt shifts, each likely reflecting millennial-scale carbon injection from methane hydrate or terrestrial wetland and/or permafrost sources (Kemp et al., 2005; Hesselbo and Pieńkowski, 2011; Them et al., 2017b; Izumi et al., 2018a; Ruebsam et al., 2019). The weathering response to these rapid injections is unclear. Furthermore, the overall magnitude of weathering change across the T-OAE is uncertain because Os-isotope values presented in the previous studies differ. In this study, we address these issues with a new high-resolution Os-isotope record through an expanded sedimentary succession of the T-OAE from Japan.

To reconstruct the Os-isotope composition of contemporaneous seawater through the T-OAE (187Os/188Osi ; i—initial), we analyzed lower Toarcian organic-rich mudrocks exposed at Sakuraguchi-dani, Japan (34°08ʹN, 131°03ʹE), that were deposited on the margin of Panthalassa (Fig. 1). Previous organic carbon-isotope (δ13Corg) analysis at Sakuraguchi-dani has revealed a >35 m record of the T-OAE negative carbon-isotope excursion (CIE; Izumi et al., 2012, 2018a; Kemp and Izumi, 2014; Fig. 2). We selected 24 samples for Os-isotope analysis spanning the T-OAE and extending into the Pliensbachian. Abundances of detrital elements Al, Ti, and Zr were also quantified on these samples using X-ray fluorescence (XRF) analysis. For 11 samples, δ13Corg was measured to supplement previously published data. Full details of the methods used for rhenium and osmium extraction, purification, and analysis and XRF and δ13C analysis, and all data, are in the Supplemental Material1.

The lower part of the Sakuraguchi-dani succession (−27.12 m to −9.47 m) is below the T-OAE negative CIE, and is characterized by extremely unradiogenic (188Os-enriched) 187Os/188Osi (∼0.17), except for the lowermost sample (187Os/188Osi = 0.47; Fig. 2A). Above −9.47 m and through the onset of the CIE, 187Os/188Osi increases, reaching a maximum of 0.58 at 1.45 m height (Fig. 2A). Above this level, 187Os/188Osi varies between 0.34 and 0.53 over the rest of the CIE (to ∼30 m; Fig. 2A). Between ∼40 m and 90 m, 187Os/188Osi is broadly constant at ∼0.34 (Fig. 2A). In detail, the increase in 187Os/188Osi from unradiogenic values close to the CIE onset (−14.5 m) to the highest (more radiogenic) values at 1.45 m closely mirrors changes in δ13Corg (Fig. 2B). Notably, δ13Corg and 187Os/188Osi strongly negatively correlate over this interval (r2 = 0.93, p-value <0.0001; Fig. 2C). Below and above this interval, there is no correlation between δ13Corg and 187Os/188Osi (Fig. 2C).

The overall decrease in δ13Corg between −14.5 m and 1.45 m comprises two abrupt shifts in δ13Corg of −2.3‰ and −2.9‰ that each occur over a few centimeters of strata (a and b in Fig. 2B). These two shifts are associated with pronounced increases in 187Os/188Osi of 0.09 (0.31 to 0.40) and 0.15 (0.40 to 0.55), respectively (Fig. 2B). Shift a is coincident with a thin (∼18 cm) brecciated interval likely associated with a minor fault (strata on either side of the interval belong to the same ammonite zone; Izumi et al., 2018a). Shift b spans ∼5 cm, and there is no sedimentological evidence for a fault or hiatus. Previous work has demonstrated that both shift b and the stratigraphic interval encompassed by shift a can be correlated with abrupt shifts through the CIE onset recognized globally (Izumi et al., 2018a).

The 187Os/188Osi record at Sakuraguchi-dani is similar to the record from the East Tributary of Bighorn Creek, Canada (hereafter ‘East Tributary’; Them et al., 2017a; Fig. 3), which was situated on the opposite margin of Panthalassa (Fig. 1). There is no evidence for water-mass restriction in either succession, and the close match in the pattern and magnitude of large-scale 187Os/188Osi changes in both sections strongly suggests that these variations reflect changes in the composition of global seawater Os. No correlation exists between detrital proxies (Zr, Al, Ti) and 187Os/188Osi at Sakuraguchi-dani, further emphasizing the predominantly hydrogenous nature of the measured Os (Fig. S1 in the Supplemental Material). Nevertheless, the occurrence of volcanic ashes at Sakuraguchi-dani (Izumi et al., 2012) means that we cannot rule out possible local influence of dissolved unradiogenic Os entering the basin, perhaps accounting for the intermittent low 187Os/188Osi values between 5.15 m and 25.1 m that are not apparent in the East Tributary data (Them et al., 2017a; Fig. 3). This possibility notwithstanding, our data and the data from East Tributary are only weakly similar to 187Os/188Osi data from Europe (Cohen et al., 2004; Percival et al., 2016; van Acken et al., 2019; Fig. 3). Post-CIE 187Os/188Osi values of ∼0.3–0.4 occur in all sections, but variable hydrographic restriction in European epicontinental basins may have caused 187Os/188Os to evolve to values partly reflective of local Os inputs, particularly during the CIE (McArthur et al., 2008; Them et al., 2017a).

The high resolution of our sampling and the thickness of the Sakuraguchi-dani section allows us to infer a geologically synchronous coupling between carbon release and weathering. Through the onset of the CIE, which may have spanned ∼200 k.y. (e.g., Ruebsam et al., 2019), ∼93% of the variance in 187Os/188Osi can be explained by changes in δ13Corg driven by carbon release (Fig. 2C). As noted above, abrupt negative shifts in δ13Corg during the CIE onset at Sakuraguchi-dani (shifts a and b in Fig. 2B) and recognized globally have been interpreted as being due to astronomically controlled hydrate melting or terrestrial carbon-release events that occurred over just a few thousand years each (Kemp et al., 2005; Them et al., 2017b; Izumi et al., 2018a; Ruebsam et al., 2019). The tight correlation and direct correspondence between these carbon-release events and increases in 187Os/188Osi suggest that global weathering responded to carbon inputs on similar time scales. The rapid response of seawater 187Os/188Os to flux changes suggests that the seawater residence time of Os was comparable to the time scale of carbon release. Modern Os residence-time estimates are variable (103–104 yr; e.g., Georg et al., 2013; Rooney et al., 2016); we note that a short Toarcian residence time is consistent with an increase in the burial flux of Os related to expansion of anoxia during the T-OAE (Them et al., 2017a).

Quantifying the magnitude of the weathering increase through the T-OAE, both overall and across abrupt δ13C shifts a and b, is complicated because the change in 187Os/188Osi (as controlled by changes in the flux and/or isotopic value of Os delivered to the ocean) has no unique solution. Mass-balance calculations based purely on Os fluxes by Them et al. (2017a) indicated that a 500% increase in continental crust weathering rate would have been required to drive the overall 187Os/188Osi shift at East Tributary (0.25–0.6), assuming modern mantle and continental crust 187Os/188Os values of 0.12 and 1.4, respectively. Our pre-CIE 187Os/188Osi value of ∼0.2 (Fig. 2), which is also consistent with the lowest pre-CIE values in East Tributary (Fig. 3), adjusts this estimate to ∼800% (scenario 1 in Fig. 4A). An increase of this magnitude is unlikely (Percival et al., 2016; Them et al., 2017a) and conflicts with an independent estimate of a ∼500% increase in global weathering rate based on Ca isotopes (Brazier et al., 2015).

Smaller weathering-rate increases can be invoked if unradiogenic Os fluxes decreased or if crustal 187Os/188Os increased (e.g., Them et al., 2017a). A pre-CIE 187Os/188Osi value of ∼0.2 indicates that the flux of mantle-derived unradiogenic Os overwhelmed any radiogenic Os flux. Some of this unradiogenic Os could have derived from basalts erupted as part of the Karoo large igneous province (present-day southern Africa) across the Pliensbachian-Toarcian boundary and leading up to the CIE (Moulin et al., 2017; Ruebsam et al., 2019; Xu et al., 2018a; Fig. S2). Karoo volcanism was multi-phased and pulsed (Moulin et al., 2017), and basalts weather rapidly after eruption (e.g., Li et al., 2016). It is thus possible that a decrease in basalt weathering–derived unradiogenic Os flux contributed to the overall 187Os/188Osi rise through the T-OAE. Any decrease was likely only modest given the persistence of Karoo and Ferrar (present-day Antarctica) volcanism through the T-OAE (e.g., Moulin et al., 2017) and the low weatherability of these high-latitude basalts (Them et al., 2017a). Nevertheless, an overall decrease in unradiogenic Os flux across the T-OAE is consistent with data from all sections indicating that 187Os/188Osi remains high after the event, and maxima in 187Os/188Osi occur after the end of the CIE onset, and thus perhaps after cessation of carbon release (Fig. 3). A ∼30% decrease in unradiogenic Os flux would be needed to reconcile the Sakuraguchi-dani 187Os/188Osi record with the ∼500% weathering-increase estimate of Brazier et al. (2015) (scenario 2 in Fig. 4A). A smaller or negligible decrease in unradiogenic flux, and smaller weathering increases, can be invoked if the onset of the T-OAE was accompanied by weathering of 187Os-enriched organic-rich rocks and/or sulfides (e.g., Them et al., 2017a). A rise in crustal 187Os/188Os from 1.4 to 2 at the start of the CIE would necessitate an ∼410% increase in continental weathering to explain the Sakuraguchi-dani data, or ∼260% if a coeval decrease in unradiogenic Os flux occurred (scenarios 3 and 4, respectively, in Fig. 4B). A value of ∼2 is a probable upper limit on global continental crust 187Os/188Os (Cohen et al., 2004; Them et al., 2017a).

Regardless of the overall magnitude of the weathering increase across the T-OAE, the marked increases in 187Os/188Osi across the inferred rapid carbon-release events at δ13Corg shifts a and b imply abrupt increases in weathering. The broad mass-balance constraints explored above suggest that these carbon-release events were associated with increases in radiogenic Os inputs of between ∼40% (shift a, scenario 4) and ∼80% (shift b, scenario 1), assuming instantaneous equilibration of seawater 187Os/188Os and no change in mantle or crust 187Os/188Os across the shifts themselves (Fig. 4). The larger of the two δ13Corg shifts (shift b) spans 5 cm of strata and has been attributed to a carbon-release event with an estimated time scale of <2 k.y. (Izumi et al., 2018a). The maximum of the coeval 187Os/188Osi increase (0.55) occurs 16 cm above the end of this δ13Corg shift. This suggests that the weathering increase associated with this carbon-release event (∼50% to ∼80%; Fig. 4) occurred over a comparable time scale (an inference also supported by the short residence time of seawater Os relative to carbon). On such short time scales, runoff may be a key limiting factor controlling rapid and large-scale Os flux to the ocean (Dosseto et al., 2015; Eiriksdottir et al., 2011; Bastian et al., 2017), and there is widespread evidence that tropical precipitation, runoff, and fluvial power all increased contemporaneously with carbon release and/or warming during the T-OAE (Izumi et al., 2018a; Xu et al., 2018b; Kemp et al., 2019).

Kemp was supported by the National Natural Science Foundation of China (grant 41888101), the National Recruitment Program for Young Professionals (P.R. China), the UK Natural Environment Research Council (NERC) (grant NE/I02089X/1), and the Great Britain Sasakawa and Daiwa Anglo-Japanese Foundations. Selby acknowledges A. Hofmann, G. Nowell, and C. Ottley, a TOTAL Endowment Fund, and a China University of Geosciences (Wuhan) Dida Scholarship. Izumi acknowledges the Japan Society for the Promotion of Science (JSPS) (grants KAKENHI 12J08818 and 15J08821). This study contributes to International Geoscience Programme (IGCP) Project 655 (IUGS-UNESCO). We thank three reviewers for their careful assessment of our work.

1Supplemental Material. Description of methods, Table S1, and Figures S1 and S2. Please visit to access the supplemental material, and contact with any questions.
Gold Open Access: This paper is published under the terms of the CC-BY license.