Rhenium isotopes reveal enhanced rock organic carbon oxidation over the Toarcian Oceanic Anoxic Event Authors: Madeleine Stow, Alexander J. Dickson, Julie Prytulak, Mathieu Dellinger, Victoria Alcock, Stephen P. Hesselbo, Geoff M. Nowell, Robert G. Hilton Published: 2026 ABSTRACT Weathering plays a central role in the geological carbon cycle. Silicate mineral weathering is invoked as a stabilizing feedback on CO2 emissions, for example from volcanism during the emplacement of Large Igneous Provinces. However, modern-day studies show weathering can emit CO2 during oxidation of rock organic carbon (OCpetro) in sedimentary rocks and function as a positive feedback on climate warming. Here we measure the rhenium isotope composition (δ187Re) of Early Jurassic marine sediments to explore how OCpetro oxidation rates changed during warming across the Toarcian Ocean Anoxic Event (T-OAE). We find a 0.22 ± 0.10‰ decrease in δ187Re values during the T-OAE, with mass balance modeling showing this can be explained by increased OCpetro weathering intensity on land associated with 6–7 °C of global warming. We estimate this could have delivered 7600–20,490 PgC to the oceans and atmosphere, demonstrating that chemical weathering does not simply act as a stabilizing feedback during hyperthermal events. Introduction Earth’s climate is moderated by variations in the concentration of atmospheric CO2 set by the transfer of carbon between the solid Earth, oceans, biosphere and atmosphere. Over geological timescales, atmospheric sources of CO2 from volcanism and metamorphism1,2 are joined by the oxidative weathering of rock-derived organic carbon (OCpetro) and sulfide minerals3–6. These CO2 inputs are removed from the atmosphere-ocean system by silicate mineral weathering coupled to carbonate precipitation7 and organic carbon burial8. The CO2 inputs are crucial to forcing changes in the surface carbon reservoirs9. In the case of volcanism, pulsed CO2 inputs have been linked to warming events over 105–106 year timescales10 with associated global environmental and ecological impacts11. Secular changes in volcanic CO2 emissions are proposed as an important control on the carbon cycle over 10 s to 100 s of millions of years12. In contrast, past rates of OCpetro oxidation and the associated CO2 emissions are uncertain13,14, even though modern global rates of OCpetro oxidation15 of 68 +16/−6 MtC yr−1 are similar to those from volcanism2. Indeed, based on studies of modern weathering systems, we might expect OCpetro oxidation fluxes to change over geological time due to changes in global climate, tectonic regime, crustal lithology, and erosion rate, but OCpetro oxidation remains challenging to quantify. Modern weathering studies suggest that OCpetro weathering and oxidation may act as a previously unrecognized climatic feedback, with increased CO2 release linked to enhanced erosion and warming6,16. This appears to counter the well-studied negative feedback offered by the weathering of silicate minerals by carbonic acid7,17. Empirical data and models suggest that physical weathering and erosion can supply large amounts of OCpetro to near-surface environments where chemical weathering and organic oxidation rates can be high18. Direct measurements of CO2 release from OCpetro oxidation have shown that rates increase > 2-fold for a 10 °C rise in local atmospheric temperatures16,19, thus acting as a potential positive feedback on warming. This process could play out at a global scale because there are large areas of the weathering zone, particularly in erosive settings, where OCpetro oxidation is incomplete16,20. Modern global OCpetro oxidation during weathering15 of 68 +16/−6 MtC yr−1 occurs alongside the export of 43 +61/−25 MtC yr−1 as incompletely weathered OCpetro contained in river sediments20. This implies an apparent OCpetro weathering intensity of 0.61 today3,15, meaning that changes in the rates of OCpetro oxidation linked to warming could increase global emission fluxes. Thus, it is critical to constrain how these feedbacks operated in the geological past to reach a quantitative understanding of Earth-system behavior under different equilibrium climate states, especially those that may be analogous to a warmer world in the near future. Whilst quantifying the solid residue of weathering using OCpetro across stratigraphic sections holds much promise21,22, we require other tracers of OCpetro oxidation to help provide a regional to global view of changes in oxidative weathering. Rhenium isotopes (185Re and 187Re, with abundances of 37.4% and 62.6% respectively) could provide quantitative insight on oxidative weathering in the geological past23–25. While 187Re is a radiogenic isotope and decays to 187Os, the half-life of this decay is ~4 ×1010 years and therefore the isotopes of Re can be considered ‘stable’ on the timescales of weathering processes23. The association of Re with organic matter in sedimentary rocks26,27, its coupled loss during OCpetro oxidation24,28, and its solubility and mobility upon oxidation and transport in the dissolved loads of rivers29–31, provide rationale for using Re isotopes as a paleo-weathering proxy25. It has been demonstrated by several studies on modern weathering environments that Re isotopes are fractionated during oxidative weathering23,32,33. The δ187Re values (relative to the NIST 3143 standard, δ187Re = ((187Re/185Re)sample/(187Re/185Re)NIST 3143 − 1) × 1000) decrease as Re is lost during oxidative weathering, which is suggested to be caused by the preferential oxidative weathering of phases enriched in the heavy 187Re isotope23,33. Low weathering intensity results in the largest fractionation between river waters and weathering residues, while higher weathering intensity leads to river waters with lower δ187Re values that approach the starting rock composition23,32,33. Since rivers are the main source of Re to the oceans25,29,34, changes in oxidative weathering intensity and flux over geological time should change river δ187Re values and shift seawater Re isotopic compositions that may be recorded in marine sediments. Here we assess oxidative weathering during the Early Jurassic Toarcian Oceanic Anoxic Event, T-OAE (also known as the Jenkyns Event, ~183 Ma). The T-OAE is characterized by a global perturbation in the carbon cycle, expressed as a positive carbon isotope excursion interrupted by a negative carbon isotope excursion (CIE) in marine and terrestrial sedimentary sections35–39. The duration40 of the CIE has been estimated between 0.3–1 Myr. The carbon cycle perturbation and negative CIE have variously been explained by the emplacement of the Karoo-Ferrar Large Igneous Province that released CO2 directly through degassing41 or by thermogenic alteration following magmatic intrusion into organic rich sediments42,43. Alternatively, increasing temperatures could have led to the destabilization and release of methane from gas hydrates44, enhanced fungal decomposition of organic matter45, or thawing permafrost46. Increased hydrological cycling47,48 is inferred to have driven increases in continental weathering rates, as indicated by radiogenic and stable isotope systems such as osmium and calcium49–52. Recent work on the T-OAE carbon budget highlighted an apparent missing source of carbon10, with the amount of CO2 released by the Karoo-Ferrar LIP during the T-OAE smaller than the amount required to drive associated changes in the partial pressure of atmospheric carbon dioxide (pCO2) and temperature in Earth System models. We hypothesize that CO2 release through oxidative weathering could have provided an additional source of CO2 during the T-OAE. We present a Re isotope record from the Llanbedr (Mochras Farm) borehole, Cardigan Bay Basin, U.K., hereafter ‘Mochras’. Our data evidence a negative Re isotope excursion, which we evaluate using a mass balance model of the Re cycle. The only coherent explanation for the Re isotope excursion is that there was a notable increase in OCpetro oxidation rates during the transient warming event, confirming the presence of an unrecognized weathering feedback operating in the geological past. Results and discussion Sedimentary δ187Re records of the Toarcian OAE The Mochras borehole captures continuous sedimentation from the Upper Triassic to Lower Jurassic in the Cardigan Bay Basin, an unrestricted, open marine setting53. The major lithologies are mudstone and limestone, which vary in proportion throughout the core48. The lack of major facies variations throughout the studied core interval supports the interpretation that there were no extreme variations in bottom water redox state or depositional environment that might have affected the local speciation of Re burial. We focus on the time period spanning the T-OAE, between core depths of ~850–775 m. The low abundance of Re in the Mochras core deposits makes isotopic measurements challenging, and we focus on 12 samples that span the T-OAE interval (Supplementary Data 1). Rhenium concentrations decrease below the T-OAE negative CIE, from a value of 6.7 ng g−1 at 970 m depth in the Upper Pliensbachian margaritatus Zone to ~1 ng g−1 at 845 m depth, near the base of the T-OAE interval (Fig. 1). Concentrations remain at ~1 ng g−1 Re through the lower part of the OAE and through the negative-CIE interval, until ~800 m depth, where there is a progressive increase to ~4 ng g−1 Re. Decreases in Re concentrations before the main CIE have been documented in other sedimentary successions, and are thought to reflect an increase in the global proportion of anoxic seafloor, and consequent drawdown of the global seawater Re inventory prior to the T-OAE54.Fig. 1Rhenium isotopes in the Mochras core sedimentary record of the Lower Jurassic.The Toarcian Oceanic Anoxic Event (T-OAE, shaded red and the associated negative carbon isotope excursion (n-CIE) in blue) is shown with ammonite Zones alongside lithology as defined previously55,57. The stable carbon isotope composition of organic matter (δ13Corg, green) is shown48,53 alongside mercury (Hg) enrichment (gray) which provides a proxy for global volcanism10, while the initial Os isotope ratio of seawater (187Os/188Osi) (orange) is sensitive to changes in continental weathering70. Published Re concentration data are shown by the blue line and diamond symbols70, with values measured in this study shown as circles. The new δ187Re record is shown as black circles, with duplicate measurements shown as open gray circles. Whiskers are either the 2SE uncertainties propagated from the measurements and bracketing zero delta standards or the 2 SD of duplicate or triplicate measurements, whichever is larger. Rhenium isotopes in the Mochras core sedimentary record of the Lower Jurassic. The Toarcian Oceanic Anoxic Event (T-OAE, shaded red and the associated negative carbon isotope excursion (n-CIE) in blue) is shown with ammonite Zones alongside lithology as defined previously55,57. The stable carbon isotope composition of organic matter (δ13Corg, green) is shown48,53 alongside mercury (Hg) enrichment (gray) which provides a proxy for global volcanism10, while the initial Os isotope ratio of seawater (187Os/188Osi) (orange) is sensitive to changes in continental weathering70. Published Re concentration data are shown by the blue line and diamond symbols70, with values measured in this study shown as circles. The new δ187Re record is shown as black circles, with duplicate measurements shown as open gray circles. Whiskers are either the 2SE uncertainties propagated from the measurements and bracketing zero delta standards or the 2 SD of duplicate or triplicate measurements, whichever is larger. There is notable variability in δ187Re values in our studied samples, from −0.03 ± 0.09‰ to −0.51 ± 0.11‰, ~4 times greater than typical analytical uncertainty (Methods, Supplementary Data 1). δ187Re values decrease from −0.03 ± 0.08‰ at 894.3 m depth to −0.26 ± 0.12‰ at 845.0 m in the upper Pliensbachian spinatum Zone, before decreasing further to a minimum value of −0.51 ± 0.11‰ within the lower Toarcian exaratum Subzone of the serpentinum Zone (at 779.2 m), coincident with the negative CIE of the T-OAE. The δ187Re isotope excursion into the negative CIE and during the main period of Ferrar volcanism41 was −0.22 ± 0.10‰, expressed as the difference between the average of measurements below the T-OAE interval (average δ187Re = −0.16 ± 0.08‰, n = 7, ±2SE, 970.8 m − 845.0 m) and within the negative CIE interval (−0.39 ± 0.07‰, n = 7, ±2SE, 833.0 m–808.2 m). Above the T-OAE interval, δ187Re values were −0.10 ± 0.08‰, a similar value as before the event. The variability in δ187Re values observed at Mochras likely reflects some mixture of changes in seawater δ187Re values at the time of deposition, alongside potential fractionation during the burial and formation of the sedimentary deposits. We note that additional sedimentary records of δ187Re during the T-OAE from open marine sites are not available, leading to inherent uncertainty on the global average seawater Re isotope composition. Nevertheless, to explore the role of fractionation during sedimentary burial and diagenesis, the magnitude of Re isotope fractionation between the sediments and seawater (Δ187Resed-sw) can be explored. An advantage of the Mochras core is that the sedimentary composition of the core remains fairly constant throughout the Lower Jurassic interval48,53, consisting largely of gray siltstone and silty mudstone containing ~1–3 wt.% TOC and with variable carbonate content48,55. Without large swings to very high TOC, it is reasonable to assume that similar Re species were buried, thus minimizing local artifacts on our putative seawater Re reconstruction from a single core site. The decoupling of our isotope and concentration data also indicate that changes in Re burial in anoxic sediments across the T-OAE54 was not a major driver of our record. Deposition of the Mochras sediments occurred dominantly in open marine conditions, with sustained bioturbation48 and a decrease but not complete disappearance of macrofauna in the lower Toarcian suggesting the bottom waters in the Cardigan Bay Basin became oxygen depleted, but not fully anoxic56,57. In this scenario, soluble ReO4- would have been scavenged from seawater and incorporated into sediments following reduction or thiolation reactions in porewaters25,33, with some exchange with an oxygenated water column above, meaning that sediment δ187Re can be fractionated from the seawater value. The only Re isotope fractionation factors determined thus far are from ab initio calculations32 and predict that reduced and thiolated species should be isotopically lighter than the soluble ReO4- ion in seawater. Therefore, the Mochras core sediment δ187Re values are likely lower than the δ187Re of coeval seawater. Since there is no sedimentological evidence to suggest a parallel change in Re speciation during the T-OAE, we infer that the δ187Re excursion recorded in the Mochras core reflects a change in seawater δ187Re of a similar magnitude over this interval. Oxidative weathering as a driver of δ187Re during the T-OAE In order to investigate the drivers of the negative δ187Re shift across the T-OAE, we developed an isotope mass balance model to reconstruct temporal variations in seawater δ187Re (Methods, Eqs. 2 and 5). Seawater δ187Re values (δ187Resw) are controlled by the balance between the input and output fluxes to the ocean (Fig. 2)25,29,34. At present, Re is largely delivered to the oceans via rivers (Friv, 4.3 ×105 mol yr−1) with an isotopic composition δ187Reriv = −0.24 ± 0.10‰ (Methods). Hydrothermal vents are a net removal pathway for Re from the oceans32,34 with an isotopic fractionation similar in direction to Re burial in marine sediments58. Rhenium behaves conservatively in the oceans with concentrations of ~7.4 pg g−1 Re throughout the water column and a well-mixed seawater isotopic composition (δ187Resw)59. We thus describe the changes in δ187Resw through time as a function of riverine inputs and sedimentary burial (Eqs. 2–6). The magnitude of Re isotope fractionation during burial (Δ187Resed-sw) is a function of fluxes and fractionation factors associated with Re transformation to thiolated and reduced species, and adsorption on or co-precipitation with a range of other mineral phases (Eq. 4). We assume that the vast majority of Re burial occurs as ReO3S in oxygen-deficient marine sediments, with a Δ187Resed-sw of −0.33 ‰ (Methods). The model is applied for a duration of 0.3 Myr to provide a conservative estimate on the excursion magnitude, but durations of up to 1 Myr40,53 do not impact the model outcomes compared to the measured T-OAE excursion (Methods).Fig. 2The modern rhenium cycle and the influence of weathering.A The major source of Re to the oceans is from rivers25,29,34 (Flux of rhenium into the ocean Fin ~ 4.3 ×105 mol yr−1; δ187Reriv ~ −0.2 to −0.4‰) contributing to the total oceanic Re mass (NRe) with seawater Re isotope composition (δ187Resw). Re is buried in sediments under low oxygen conditions. Most Re burial occurs under anoxic bottom waters which may form in both restricted [1] and unrestricted [2] basins. In addition, Re burial can occur in settings with oxic bottom waters [3], where high sedimentation rates occur and organic matter respiration promotes the formation of anoxic pore waters, e.g., continental margins. The overall Re burial flux Fout ∼ 4.3 × 105 mol yr−1, but the Re isotope fractionation factors (Δ187Resed-sw) during burial are still unknown. The schematic figures display the effect of increasing temperature on the weathering flux (B) weathering intensity (C) and δ187Reriv (D) as a function of the erosion rate, where red lines indicate warmer temperatures in the weathering zone. The modern rhenium cycle and the influence of weathering. A The major source of Re to the oceans is from rivers25,29,34 (Flux of rhenium into the ocean Fin ~ 4.3 ×105 mol yr−1; δ187Reriv ~ −0.2 to −0.4‰) contributing to the total oceanic Re mass (NRe) with seawater Re isotope composition (δ187Resw). Re is buried in sediments under low oxygen conditions. Most Re burial occurs under anoxic bottom waters which may form in both restricted [1] and unrestricted [2] basins. In addition, Re burial can occur in settings with oxic bottom waters [3], where high sedimentation rates occur and organic matter respiration promotes the formation of anoxic pore waters, e.g., continental margins. The overall Re burial flux Fout ∼ 4.3 × 105 mol yr−1, but the Re isotope fractionation factors (Δ187Resed-sw) during burial are still unknown. The schematic figures display the effect of increasing temperature on the weathering flux (B) weathering intensity (C) and δ187Reriv (D) as a function of the erosion rate, where red lines indicate warmer temperatures in the weathering zone. First, we explore how changes in the sedimentary Re sink from the oceans could impact δ187Resw through time. While the burial fractionation factors are uncertain, their sign and magnitude mean that it is unlikely that changes in the Re outputs from seawater during the T-OAE could drive a negative shift in δ187Resw values. The extent of global euxinia is thought to increase during the T-OAE54. Using a negative value of Δ187Resed-sw as suggested from ab initio calculations32, an increase in the proportion of Re buried as thiolated S species would lead to an increase in δ187Resw. This is opposite to the shift we observe (Fig. 3A).Fig. 3Mass balance models of the Re isotopic composition of seawater following perturbations to a steady state.Perturbations occur over 0.3 Millions of years (Myr) from 1 Myr, considering: (A) increasing the proportion of burial of thiolated Re species in low O2 sediments with Δ187Resed-sw values of −0.33‰; B increase in the riverine flux of dissolved Re; and (C) changing isotopic composition of river water. The dashed black line and blue shading show the magnitude of the δ187Re excursion into the T-OAE of −0.22 ± 0.10‰ (Fig. 1), assuming its deviation from seawater as shown in the figure. D Shows the modeled river δ187Re values (δ187Reriv) as a function of the oxidative weathering intensity of Re, ωRe. The star shows the ωRe value of the modern day15, and the shift in δ187Reriv reported relative to that initial condition. The blue box shows the range of δ187Reriv values which would be consistent with the Toarcian Oceanic Anoxic Event (T-OAE) δ187Re excursion from the ocean mass balance model (C). Mass balance models of the Re isotopic composition of seawater following perturbations to a steady state. Perturbations occur over 0.3 Millions of years (Myr) from 1 Myr, considering: (A) increasing the proportion of burial of thiolated Re species in low O2 sediments with Δ187Resed-sw values of −0.33‰; B increase in the riverine flux of dissolved Re; and (C) changing isotopic composition of river water. The dashed black line and blue shading show the magnitude of the δ187Re excursion into the T-OAE of −0.22 ± 0.10‰ (Fig. 1), assuming its deviation from seawater as shown in the figure. D Shows the modeled river δ187Re values (δ187Reriv) as a function of the oxidative weathering intensity of Re, ωRe. The star shows the ωRe value of the modern day15, and the shift in δ187Reriv reported relative to that initial condition. The blue box shows the range of δ187Reriv values which would be consistent with the Toarcian Oceanic Anoxic Event (T-OAE) δ187Re excursion from the ocean mass balance model (C). Having found that changes in the burial rates of Re are unlikely to explain a negative shift in δ187Resw, we instead explore changes in the flux and isotopic composition of dissolved Re delivered to the oceans via rivers (Friv and δ187Reriv) which reflect oxidative weathering processes on land. Weathering reactions are a trade-off between the rate of supply of minerals to the weathering zone, linked to erosion, versus the kinetics of the reactions which can be controlled by temperature and hydrology17,31. At lower erosion rates, OCpetro oxidation is thought to be supply limited18,60 with very high weathering intensity while fluxes are low. At higher erosion rates, OCpetro oxidation rate increases with temperature16,19. In these settings, OCpetro is supplied in excess and thus the overall weathering intensity is lower and climatically sensitive (Fig. 2B, C). These shifts in weathering intensity will impact δ187Reriv (Fig. 2D): as weathering intensity increases, δ187Reriv values should trend towards the value of unweathered bedrock, following more intensive oxidative weathering and near-complete Re loss23,33. In the context of the T-OAE, the timescales are too short for a wholesale change in continental erosion patterns linked to tectonics which could shift OCpetro weathering intensity15. However, the increases in temperature globally61 should increase the intensity of OCpetro weathering in landscapes that are kinetically-limited16,19 (Fig. 2B and C). We model changes in Friv from +2x to +10x and δ187Reriv between the modern value of −0.24‰ (Methods) and a minimum value of −0.60‰ (Fig. 3B, C). A change in the riverine flux alone leads to a decrease in δ187Resw, but this shift is small even for an unrealistically high 10-fold increase in global OCpetro oxidation (Fig. 3B). In contrast, a shift in δ187Reriv of −0.24 ± 0.11‰ (to a δ187Reriv value of −0.48 ± 0.11‰) can explain the −0.22 ± 0.10‰ shift in δ187Re observed in the Mochras record over the T-OAE (Fig. 3C, Methods). Lowering of the Re isotope composition of river waters can be achieved by increasing the weathering intensity of sedimentary rock containing organic matter23,32,33. To determine whether these shifts are feasible in terms of predicted changes in weathering intensity of OCpetro, we turn to a Rayleigh model to describe isotopic patterns as a function of weathering intensity33 (Methods, Eqs. 8 and 9). Using a range of values based on our knowledge of the modern-day Re cycle, we calculate that a decrease in δ187Reriv values by −0.24‰ can be achieved by increasing the proportion of oxidized Re (ReO4-) leaving the weathering zone in the dissolved load of rivers from 0.61 (equivalent to modern day15) to 0.96 (Fig. 3D). Assuming that the single core record of δ187Re captures global swings in seawater chemistry during the T-OAE, for a constant rate of Re denudation, this corresponds to an increase in the global oxidized Re flux by ~1.6 times. These changes in oxidative weathering are plausible given that previous assessments have suggested a 1.6-fold51 to 8-fold49 increase in silicate weathering during the T-OAE in response to increased temperature61 and an enhanced hydrological cycle48. A positive feedback via OCpetro oxidation during warming events Our Re isotope data indicate that there was an increase in continental oxidative weathering that peaked during the T-OAE. This increase in weathering would have released CO2 into the oceans and atmosphere and acted as an additional driver of global temperature change. The CO2 released by OCpetro oxidation could account for an apparent discrepancy between the CO2 flux from volcanism and the overall change in atmospheric pCO2 predicted by climate models. Fendley et al.10 used Hg concentrations in the Mochras core and Hg/CO2 ratios to calculate that a total of 6200 Pg C was emitted by LIP volcanism during the T-OAE interval. However, the GEOCLIM and COPSE climate models predict that additional emissions of between 8500 Pg C and 69,000 Pg C respectively are required to explain δ13C, pCO2 and temperature proxy data10. To explore whether oxidative weathering of OCpetro could contribute to these “missing” CO2 emissions, we use recent empirical data from weathering studies to provide additional constraints on OCpetro weathering independent from the Re isotope record (Methods). This approach requires us to estimate: i) the apparent temperature sensitivity of OCpetro oxidation16,19; ii) the proportion of the global land surface that experiences kinetically-limited weathering15,20 (i.e., the supply to the weathering zone is very high); and iii) global OCpetro weathering flux and overall OCpetro weathering intensity. The apparent temperature sensitivity parameter (Q10, the increase in flux over a 10 °C temperature change) has been determined from modern day field based measurements in two shale weathering locations16,19 of 2.0 ± 0.1 and 2.7 ± 1.0. An approximate T-OAE warming61,62 of ~6–7 °C could thus cause a 1.6 to 2 times increase in the oxidative CO2 flux if weathering occurs in kinetically limited regimes (Methods). Today, ~60% of the OCpetro weathering flux is attributed to highly erosive regions where oxidative weathering may be temperature sensitive (kinetically limited)15. Global OCpetro oxidation is estimated to be 68 +18/-6 Mt C yr−1, with 43 Mt C yr−1 of incompletely weathered OCpetro exported to the ocean15,20, which equates to an average OCpetro weathering intensity of 0.61. Assuming that oxidative weathering on the Early Jurassic land surface was not completely supply-limited and the distribution of sedimentary rocks was not vastly different than today, we calculate that the T-OAE warming would equate to an additional 25–41 Mt C yr−1 release from OCpetro oxidation (Methods, Eqs. 12 and 13). This corresponds to an increase in global OCpetro weathering intensity to 0.83-0.98 if total OCpetro denudation remained constant. This shift in weathering intensities appears plausible based on modern observations of near-complete oxidation of OCpetro in slowly eroding, temperate climates3,4,28, alongside efficient OCpetro weathering in warm and humid tropical river floodplains63. While uncertain, these changes in global weathering flux calculated from T-OAE warming61,62 and an empirical temperature sensitivity of OCpetro oxidation16,19 are comparable to those suggested by modeling the decrease in δ187Reriv values that are consistent with the measured T-OAE δ187Re excursion (Fig. 3D, Methods). With an approximate duration of between 0.3 and 0.5 Myr, increased OCpetro oxidation of 25–41 Mt C yr−1 would have released a total of 7600–20,490 Pg C across the T-OAE. The magnitude of this emission flux is in the range of the ‘missing’ carbon emissions that appear not to be supplied from volcanism during the event10. The increase in oxidative weathering of OCpetro that explains the new Re isotope record and its apparent temperature sensitivity are consistent with other geochemical records. Published shifts in the Os isotope composition of seawater have been interpreted to reflect changes in silicate weathering flux49,51. However, sulfide minerals and OCpetro are enriched in Re (and 187Re), and other workers have suggested that enhanced oxidative weathering could increase 187Os supply and increase the Os isotope initial of seawater64,65, thus making the shifts seen in the T-OAE sedimentary record consistent with the δ187Re excursion (Fig. 1). An additional proxy which could reflect OCpetro oxidation is the stable carbon isotope record of carbonates (δ13Ccarb) and organic matter (δ13Corg)13,14. In isolation, increased OCpetro oxidation would result in a lowering of seawater δ13Ccarb and δ13Corg values as 13C-depleted organic matter is oxidized and supplied to the atmosphere and oceans. Unfortunately, the carbon isotope mass balance at this period of time is complicated by several factors14,66, most importantly: i) the unknown δ13C values of the volcanic inputs, particularly if rocks intrude sedimentary units43; ii) changes in the size of the DIC reservoir of seawater; iii) changes in δ13Corg due to changes in pCO2; iv) variation in the fluxes and proportion of carbon leaving seawater as carbonate minerals vs organic matter. Recent geological carbon cycle models have started to incorporate a dynamic organic carbon cycle and variable rates of OCpetro oxidation, showing these processes change our understanding of the geological regulation of the carbon cycle67. It is clear that future insight can come from new δ187Re records (Fig. 1), detrital proxies of OCpetro oxidation21,22 alongside coupled isotopic modeling of δ187Re, Os and C. Our findings suggest that the weathering carbon-cycle feedbacks for the T-OAE are more complex than previously recognized. Silicate mineral weathering by carbonic acid draws down CO2 through increased alkalinity flux to the oceans (and when coupled to carbonate burial in the oceans) and these reactions can increase under warmer, wetter and higher pCO2 conditions, as a negative feedback. However, warmer temperatures and increased erosion under an enhanced hydrological cycle may also lead to more shale weathering and CO2 release by OCpetro oxidation. Here, we show that a negative shift in δ187Re values can be explained by a global increase in the weathering intensity of OCpetro during the T-OAE (Fig. 1), which thus acted as a positive feedback that increased the total amount of CO2 transferred to the ocean-atmosphere system. These conclusions support those from other major environmental perturbations, including OCpetro oxidation being hypothesized as an additional source of CO2 during hyperthermal events of the Cenozoic21,22. Our work suggests that oxidative weathering needs to be better accounted for to understand the complex carbon cycle interactions that operate during climatic perturbations. Methods Sample digestion Chemical separation and Re isotope ratio measurements were conducted in the clean laboratories at the University of Oxford, Royal Holloway University of London, and Durham University following published methods23,33,58,68. Unless specified, all acids were distilled in either quartz or Teflon stills, and diluted with ultrapure MQ water (18.2 Ωcm resistivity). The measurement of Re isotope ratios is challenging due to the low concentrations of Re in geological samples. To make precise and accurate Re isotope measurements, at least 1 ng Re is required, while higher Re masses allow for improved analytical precision68. Based on the measured Re concentrations, we aimed to digest at least 2–3 g of material for each sample. Sample powders (masses up to 500 mg) were initially digested in Savillex Teflon beakers using 3 ml 16 M HNO3 and 3 ml 28 M HF, on a hotplate at 160 °C for 48 h. This step digested the silicate component of the samples, and started to digest the organic matter. Samples were evaporated at 120 °C to incipient dryness and repeatedly covered in 16 M HNO3 and evaporated at 180 °C to destroy fluoride phases. Samples were then transferred to Digest n-vessels purchased from QMX, specifically to break down the remaining organic matter. The vessels are composed of a digestion vessel base and collection tube lid, where acid vapors evaporating from the base vessel condense and collect. As the acid evaporates, the pressure drops in the base, allowing the acid condensing in the collection tube to fall back into the base. This process allows samples to be continually refluxed, promoting faster, more effective digestion of organic matter. 2 ml 10 M HCl and 2 ml 16 M HNO3 was added to the base of the vessel, and 3 ml 16 M HNO3 to the top collection tube. The vessels were refluxed on a hotplate at 180 °C for 48 h. After this stage, if there was no solid material left, samples were evaporated and refluxed in 5 ml 5 M HCl at 120 °C, before being finally dissolved in 20 ml 1 M HCl loading solution for column chromatography. If samples appeared sticky or cloudy at this stage, the step in the digest-n vessels was repeated to ensure complete digestion. Two samples (SSK74550 and SSK49509) were digested using an adapted procedure to assess the approximate proportion of Re associated with the silicate versus non-silicate fraction of the sediments. Samples were initially digested using HNO3 and HCl in the Digest n-vessels as described above, in order to digest the non-silicate fraction. The remaining solid silicate material was separated by centrifugation and then digested using 3 ml 16 M HNO3 and 3 ml 28 M HF on a hotplate at 160 °C for 48 h. Re concentrations were measured on a Perkin Elmer NexION 5000 ICP-MS as discussed below. In these samples, only 6% of the total Re was associated with the silicate fraction, which demonstrates that there is a negligible contribution of Re from detrital silicate material. 94% of the Re was associated with the non-silicate fraction. Chemical separation The procedure for Re separation from sample matrix was adapted from methods described in Dellinger et al.68. Biorad AG1-X8 anion exchange resin (200–400 mesh) was pre-cleaned by shaking with 5 M HCl, 8 M HNO3 and MQ H2O. 1 ml of pre-cleaned AG1-X8 resin was loaded into polypropylene Biorad columns, and cleaned with 10 ml 10 M HCl, 10 ml 1 M HCl and 30 ml 8 M HNO3. The resin was conditioned with 5 ml 0.5 M HNO3 and 5 ml 1 M HCl. Samples were loaded in ~20 ml 1 M HCl, but this volume could be increased if samples had higher mass or appeared more viscous. A further 10 ml 1 M HCl and 15 ml 0.5 HNO3 was used to elute the sample matrix, and then Re was collected in 16.5 ml 4 M HNO3. This is a higher collection volume than used by Dellinger et al.68, but we found this was necessary to ensure quantitative Re recovery. The Re fractions were evaporated overnight at 100 °C, refluxed in 16 M HNO3 for 24 h and evaporated at 150 °C to destroy any resin particles which have bled through the column. Samples were then dissolved in 2 ml 2% HNO3 and a 100 µl aliquot (5%) was taken for Re concentration measurements by multi-quadrupole ICP-MS. This column procedure was repeated twice more in order to remove any remaining sample matrix. The column was conducted as described above, except samples were loaded in 1 ml 1 M HCl. Following the third column, samples were refluxed in 16 M HNO3 to destroy resin-derived organics, and then dissolved in 0.4 ml 3% HNO3 for isotopic analysis. Following three passes through the column, Re yields were typically >85%. Re concentration and isotope ratio measurements Re concentrations of samples after the first pass through the anion exchange column were measured on a Perkin Elmer NexION 5000 ICP-MS at the University of Oxford. These concentration measurements were used to calculate how much sample mass was required for isotopic analysis, and to confirm that there was quantitative recovery of Re through the three column chromatography procedures. The ICP-MS was operated in standard (no gas collision cell) mode. Calibration standards of known Re concentration were run at the start of each measurement session. The typical range of calibration standards span concentrations from 1 pg g−1 to 100 pg g−1 Re. Samples were doped with 1 ng g−1 Rh internal standard to monitor machine drift. Certified reference materials AQUA-1 and SLRS-6 from the Canadian Geological Survey were measured throughout runs at 1x and 10x dilution to assess the accuracy of measurements. Errors are reported as the within run RSD of individual analyses, and for Re these are typically <5%. Re isotope ratio measurements were performed following the method in Dellinger et al.68 using a Neptune Plus MC-ICP-MS at Durham University and Royal Holloway University of London. There are minor differences between the machine setup and running conditions, as described below. At Durham University, measurements were made in low resolution mode, with masses 183W, 184W, 185Re, 186W, 187Re and 189Os measured on cups L3, L2, L1, C, H1 and H3 respectively. 1013Ω resistors were connected to cups L1 and H1 to measure the small Re beams, which are typically <100 mV. The sample introduction system consisted of a concentric flow nebulizer with flow rate of 50 µl/min coupled to a quartz cyclonic spray chamber. Typical sensitivity was ~20 mV/ppb on 187Re using this setup. Standards and samples were typically run at concentrations between 1 – 5 ppb, with the aim of obtaining the maximum Re signal possible for the Re concentration of that sample. Each individual measurement was composed of one block of 25 cycles with 16.7 s integration time. The interference of 187Os on 187Re was corrected by monitoring 189O, and assuming a 189Os/187Os ratio of 8.219440. Mass bias and instrument drift was corrected by standard sample bracketing with the NIST 3143 standard, and doping with W, to a W/Re ratio of 20 (i.e., if the Re concentration was 5 ppb, we doped with 100 ppb W). The concentration of Re and W/Re ratio was matched within 10% for samples and bracketing standards. The correction assumes a 186W/184W ratio of 0.927672 and uses an exponential law. At Royal Holloway University of London, the cup configuration is slightly different than that used at Durham. Measurements were made in low resolution mode, with masses 184W, 185Re, 186W, 187Re, 188Os and 190Os measured in cups L3, L2, L1, C, H1 and H2 respectively. 1013Ω resistors were connected to cups L2 and C to measure the small Re beams. The sample introduction system consisted of a PFA nebulizer with flow rate of 100 µl/min coupled to a quartz SIS spray chamber. Typical sensitivity was ~80 mV/ppb on 187Re using this setup. Standards and samples were typically run at Re concentrations of 2ppb. Each individual measurement was composed of one block of 3 cycles with 8.4 s integration of a 3% HNO3 blank solution, followed by one block of 40 8.4 s integrations of the sample. The average signal of the blank 3% HNO3 was then subtracted from the sample signal prior to carrying out the interference and mass bias corrections. The interference of 187Os on 187Re was corrected by monitoring 188Os, and assuming a 188Os/187Os ratio of 5.7504. Mass bias and instrument drift was corrected by standard sample bracketing with the NIST 3143 standard, and doping with W, to a W/Re ratio of 20, following the same procedure described above for Durham. Quality control and assurance Re isotope ratios are reported in delta notation relative to NIST 3143 (Supplementary Data 1 and 2):1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$({{\rm{\delta }}}^{187}{{Re}}=({({}^{187}{{R}}{{e}}/{}^{185}{{R}}{{e}})}_{{\rm{sample}}}/{({}^{187}{{R}}{{e}}/{}^{185}{{R}}{{e}})}_{\rm{NIST\,3143}}-1)\times 1000)$$\end{document}(δ187Re=((R187e/R185e)sample/(R187e/R185e)NIST 3143−1)×1000) Total procedural blanks ranged from 0.5–20 pg Re (n = 6), which was typically <1% of the total Re processed. The precision and accuracy of both methods was assessed by repeat analysis of the NIST 989 Re solution during measurement sequences, and analysis of USGS reference materials and other standards which had been processed alongside each batch of samples (e.g., BCR-2, SDO-1, 00N118 shale, MAG-1 and MESS 4). Over a year long period, NIST 989 had an average isotopic composition of δ187ReNIST 3143 = −0.27 ± 0.07‰ (2 SD, n = 16) at Durham and δ187ReNIST 3143 = −0.29 ± 0.10‰ (2 SD, n = 32) at Royal Holloway University of London, agreeing with previous studies (e.g., −0.28 ± 0.04‰ (2 SD, n = 26)68 and −0.27 ± 0.10‰ (2 SD, n = 14)58. Values for USGS reference materials and other in-house standards (Supplementary Data 2) agree with previous values (Supplementary Data 3). Rhenium isotope ocean mass balance model We describe the evolution over time of the Re isotopic composition of seawater (δ187Resw)59 as a function of the mass of Re in the seawater inventory (NRe, mol), the input of Re from rivers (Friv, mol yr−1) which dominates the input flux with an isotopic composition δ187Reriv, and the output of Re via sedimentary burial (Fout, mol yr−1) with an isotopic composition δ187Resed:2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{d({N}_{{Re}}\delta {}^{187}{Re}_{{sw}})}{{dt}}={F}_{{riv}}\times {\delta }^{187}{{Re}}_{{riv}}-{F}_{{out}}\times {\delta }^{187}{{Re}}_{{sed}}$$\end{document}d(NReδResw187)dt=Friv×δ187Reriv−Fout×δ187Resed Given the current state of knowledge of the Re mass balance it is convenient to describe:3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\delta }^{187}{{Re}}_{{sed}}=({\Delta }^{187}{{Re}}_{{sed}-{sw}}+{\delta }^{187}{{Re}}_{{sw}})$$\end{document}δ187Resed=(Δ187Resed−sw+δ187Resw)where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Delta }^{187}{\mathrm{Re}}_{{sed}-{sw}}$$\end{document}Δ187Resed−sw describes the fractionation factor between the sedimentary sink and seawater. This can be formulated as a mixture of a Re sink in low oxygen environments (e.g., anoxic, suboxic), and a Re sink in other sites which have higher oxygen contents25,65:4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Delta }^{187}{{Re}}_{{sed}-{sw}}=[(\,{f}_{{lowO}2}\times {\Delta }^{187}{{Re}}_{{lowO}2-{sw}})+({1-f}_{{lowO}2})\times {\Delta }^{187}{{Re}}_{{other}-{sw}}]$$\end{document}Δ187Resed−sw=[(flowO2×Δ187RelowO2−sw)+(1−flowO2)×Δ187Reother−sw]where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${f}_{{lowO}2}$$\end{document}flowO2 is the fraction of Re leaving seawater in low oxygen environments through transformation to thiolated and/or reduced species with a fractionation factor \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Delta }^{187}{\mathrm{Re}}_{{lowO}2-{sw}}$$\end{document}Δ187RelowO2−sw. The removal of Re via adsorption or co-precipitation to/with a range of other mineral phases is captured in the model by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Delta }^{187}{\mathrm{Re}}_{{other}-{sw}}$$\end{document}Δ187Reother−sw. We model this in incremental timesteps such that the δ187Resw value is calculated from the previous timestep (t-1) and the timestep duration Δt (yr) given by:5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\delta }^{187}{{Re}}_{{sw}}=\frac{\left[({\delta }^{187}{{Re}}_{{sw},t-1}\times {N}_{\mathrm{Re},t-1})+({{F}_{{riv}}\times \Delta t\times \delta }^{187}{{Re}}_{{riv}})-({{F}_{{out}}\times \Delta t\times \delta }^{187}{{Re}}_{{sed}})\right]}{{N}_{{Re},t}}$$\end{document}δ187Resw=(δ187Resw,t−1×NRe,t−1)+(Friv×Δt×δ187Reriv)−(Fout×Δt×δ187Resed)NRe,t When running perturbations to mass flux, we take account of the role of Re concentration on the output flux, and modify the output flux, Fout* based on the perturbation above the initial Re seawater inventory, NRe,06\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${F}_{{out}}^{*}={F}_{{out}}\times \Delta t\times \frac{{N}_{{Re}}}{{N}_{{Re},0}}$$\end{document}Fout*=Fout×Δt×NReNRe,0 The parameters used in the model are described herein. The scenarios were run over 5000-year timesteps and the perturbations were applied after 1 Myr for a duration of 0.3 Myr40,53. We use constraint on compositions from the modern Re cycle, with δ187Resw = −0.17 ± 0.13‰ (2 SD, n = 12)59 and a Re inventory of seawater, NRe = 5.48 x 1010 mol. For the riverine input, only data from the Mackenzie River and its tributaries in Northern Canada has been published so far23, with an average δ187Re of −0.27 ± 0.09‰ (2 SD, n = 10). This is similar to the average δ187Re of all large global rivers measured so far (−0.24 ± 0.10‰, 2 SD, n = 12; Dellinger, pers. comm.) which is used as the initial δ187Reriv value. The global pre-anthropogenic34 Friv = 4.3 × 105 mol yr−1 with the input of oxidized Re from rivers directly linked to oxidative weathering of OCpetro15,36. Fout assumed to balance the inputs at mass steady state and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${f}_{{lowO}2\,}$$\end{document}flowO2 = 0.94 (ref. 65). The isotope fractionation associated with burial of Re in the oceans remains poorly constrained by measurements, but can be determined from ab initio calculations and mass balance arguments. First, we can consider mass and isotopic steady state, which based on Eqs. 2 and 3, simplifies to:7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\delta }^{187}{{Re}}_{{sw}\,}=\delta \,{{}^{187}{Re}_{{riv}}}-\Delta^{187}{{Re}}_{{sed}-{sw}}$$\end{document}δ187Resw=δReriv187−Δ187Resed−sw For δ187Reriv = −0.24‰ and this suggests that the overall Re isotope fractionation factor between sediments and seawater during burial (Δ187Resed-sw) would be −0.07‰ to return the measured seawater composition given the input fluxes. Another constraint comes from ab initio calculations32. These suggest larger Δ187Resed-sw values of −0.33‰ and −0.64‰ for removal of Re as thiolated ReVIIO3S- and ReVIIO2S2- species respectively, with a value up to a maximum value of −1.52‰ for reduced ReIVCl62- species32. In the absence of any observational or experimental evidence, we set the initial \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Delta }^{187}{\mathrm{Re}}_{{lowO}2-{sw}}=-\!0.33{{\textperthousand}}$$\end{document}Δ187RelowO2−sw=−0.33‰, which would imply a \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Delta }^{187}{\mathrm{Re}}_{{other}-{sw}} > 0\textperthousand$$\end{document}Δ187Reother−sw>0‰ (Eq. 4). We note that the difference between seawater δ187Re values and modern and Holocene marine sediment reference materials (MAG-1, MESS-4) is ~−0.11‰ to −0.16‰ (Supplementary Data 2 and 3). While standard reference materials should not be used to interpret geological information, if these values were used for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\,\Delta }^{187}{\mathrm{Re}}_{{lowO}2-{sw}}$$\end{document}Δ187RelowO2−sw then \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Delta }^{187}{\mathrm{Re}}_{{other}-{sw}} \sim 0\textperthousand$$\end{document}Δ187Reother−sw~0‰. While considerable uncertainty remains, we suggest the burial fractionation factors are likely to be towards the lower end of the theoretical values32. While changes in the output flux cause positive excursions and cannot explain the data (Fig. 3A), future work should prioritize quantifying these fractionation factors. We perturb the model parameters Friv, δ187Reriv and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${F}_{{out}}\times {f}_{{lowO}2}$$\end{document}Fout×flowO2 as described in the main text and Fig. 3. We report values for a 0.3 Myr duration to provide a conservative estimate of response, given the likely range of durations from 0.3-1.0 Myr40,53. Longer durations will cause a slight increase in the observed excursion, but will not meaningfully impact the outcomes and interpretations in comparison to the measured δ187Re in the Mochras record. Model constraints on oxidative weathering The oceanic Re isotopic mass balance model outputs suggest that a change in δ187Reriv values provide the most plausible way to explain the T-OAE data (Fig. 3), with a decrease in δ187Reriv values of −0.24 ± 0.11‰ to explain the decrease in δ187Re values across the T-OAE interval of −0.22 ± 0.10‰ (Fig. 3C) To provide an independent test on whether these shifts in δ187Reriv are reasonable in terms of weathering reactions, we calculate the paired isotopic evolution of river runoff (δ187Reriv) and a weathered solid residue (δ187Rew) as a function of weathering intensity. We use a Rayleigh distillation model which has been shown to fit patterns in δ187Re during weathering33:8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\delta }^{187}{Re}}_{w}={\delta }_{0}+{\Delta }_{{riv}-w}\times {ln}({f}_{{Re}-w})$$\end{document}δ187Rew=δ0+Δriv−w×ln(fRe−w)9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\delta }_{0}={{\delta }^{187}{Re}}_{w}\times {f}_{{Re}-w}+{{\delta }^{187}{Re}}_{{riv}}\times (1-{f}_{{Re}-w})$$\end{document}δ0=δ187Rew×fRe−w+δ187Reriv×(1−fRe−w)where δ0 is the initial isotopic composition of weathered rocks, fRe-w is the fraction of Re remaining in the weathered residue and Δriv-w is the fractionation factor between the solution (δ187Reriv) and the residual solid (δ187Rew). The term (1 – fRe-w) thus describes the proportion of oxidized Re leaving the weathering zone in solution. This is a metric for the overall oxidative weathering intensity of Re, ωRe = (1 – fRe-w) (Fig. 3D). Given the close link between OCpetro oxidation and Re loss during weathering of shales15,28, ωRe is likely to be proportional to the weathering intensity of OCpetro (see discussion below and Eq. 11). Given the importance of black and gray shales to OCpetro oxidation and Re mass balance, we take δ0 = −0.52‰ defined by the average of 16 shale samples from Dickson et al.33 and Δriv-w = 0.5‰ based on the fit to weathering profile and river data. The global average OCpetro weathering intensity15,20 is estimated to be 0.61, and if this equates to a fRe = 0.39 then Rayleigh model predicts a δ187Reriv = −0.22‰ (Eqs. 4 and 5). This is close to the published values of river δ187Re23, suggesting this approach captures the first order features of the weathering mass balance well. From the modern day initial condition, we then consider how much δriv could change from this initial condition as a function of changing fRe (Fig. 3D). OCpetro weathering scenarios under warming climate First, we consider that total OCpetro denudation (mass transfer, Mt Cyr−1) is the sum of an oxidized flux and an un-weathered flux:10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${J}_{{OCpetro}}={J}_{{OCpetro}-{ox}}+{J}_{{OCpetro}-{uw}}$$\end{document}JOCpetro=JOCpetro−ox+JOCpetro−uw Present day estimates of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${J}_{{OCpetro}-{ox}}=68$$\end{document}JOCpetro−ox=68 MtC yr−1 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${J}_{{OCpetro}-{uw}}=43$$\end{document}JOCpetro−uw=43 MtC yr−1. A metric of global weathering intensity, ωOCpetro, can thus be described as:11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\omega }_{{OCpetro}}=\frac{{J}_{{OCpetro}-{ox}}}{{J}_{{OCpetro}}}$$\end{document}ωOCpetro=JOCpetro−oxJOCpetro We can estimate the change in JOCpetro-ox during a warming event by considering that the global oxidative weathering flux and CO2 release is a combination of landscapes where OCpetro oxidation is “supply-limited”, versus those “kinetically-limited”. In areas where supply-limitation occurs, OCpetro oxidation is sensitive to erosion, whereas areas of kinetic-limitation describe landscapes where supply of OCpetro is high and temperature could influence fluxes. The kinetically-limited portion of this flux, JKL is then:12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${J}_{{KL}}={J}_{{OCpetro}-{ox}}\times {f}_{{KL}}$$\end{document}JKL=JOCpetro−ox×fKLwhere fKL is the fraction of the total flux delivered by kinetically limited landscapes. Based on global analysis of OCpetro weathering fluxes, we take the starting point that fKL = 0.6, reflecting the dominance of high erosion landscapes to the fluxes15. For a global flux of 68 MtC yr−1, this corresponds to JKL = 41 MtC yr−1. The weathering flux from kinetically limited terrains can then be modified by the apparent temperature sensitivity:13\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${J}_{{KL}-T}={J}_{{KL}-0}\times exp \,(\alpha T)$$\end{document}JKL−T=JKL−0×exp(αT)where JKL-0 is the initial OCpetro weathering flux before the warming event, JKL-T (Mt Cyr−1) is after a change in temperature, T (°C), as described by the growth rate parameter α (°C−1). Based on field measurements of OCpetro oxidation in erosive settings in France16 and New Zealand19, we define the growth parameters as between 0.069 °C−1 and 0.099 °C−1. These correspond to an apparent temperature sensitivity of OCpetro oxidation described by the Q10 parameter (increase in flux over 10 °C) of 2.0 to 2.7, respectively. The global patterns of denudation and lithological exposure remain poorly constrained in the geological record. As such, to estimate the impact of T-OAE warming, we use the modern day OCpetro weathering constraints to provide an order of magnitude estimate. For 6-7 °C of warming, JKL-0 = 41 MtC yr−1 could increase by an additional 25 to 41 MtC yr−1. If total OCpetro denudation does not change, this reflects an increase in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\omega }_{{OCpetro}}$$\end{document}ωOCpetro from an initial value of 0.61 to 0.83 to 0.98. Over the duration of a 0.3 Myr event, these fluxes suggest that OCpetro oxidation could have released an additional 7600–12300 PgC over the initial values. If the duration of the event was 0.5 Myr40,53, these totals would be 12700–20490 PgC. These simplified calculations reflect the current state of knowledge of OCpetro oxidation. The magnitude of the fluxes show the need for future work to not only include OCpetro oxidation as a major C input, but also to include supply and kinetically-limited controls on oxidation weathering into geological carbon cycle assessments. Supplementary information Description of Additional Supplementary Files Supplementary Data 1–3 Transparent Peer Review file Description of Additional Supplementary Files Supplementary Data 1–3 Transparent Peer Review file