Evidence for rapid weathering response to climatic warming during the Toarcian Oceanic Anoxic Event

Chemical weathering consumes atmospheric carbon dioxide through the breakdown of silicate minerals and is thought to stabilize Earth’s long-term climate. However, the potential influence of silicate weathering on atmospheric pCO2 levels on geologically short timescales (103–105 years) remains poorly constrained. Here we focus on the record of a transient interval of severe climatic warming across the Toarcian Oceanic Anoxic Event or T-OAE from an open ocean sedimentary succession from western North America. Paired osmium isotope data and numerical modelling results suggest that weathering rates may have increased by 215% and potentially up to 530% compared to the pre-event baseline, which would have resulted in the sequestration of significant amounts of atmospheric CO2. This process would have also led to increased delivery of nutrients to the oceans and lakes stimulating bioproductivity and leading to the subsequent development of shallow-water anoxia, the hallmark of the T-OAE. This enhanced bioproductivity and anoxia would have resulted in elevated rates of organic matter burial that would have acted as an additional negative feedback on atmospheric pCO2 levels. Therefore, the enhanced weathering modulated by initially increased pCO2 levels would have operated as both a direct and indirect negative feedback to end the T-OAE.

. Global palaeogeography of the Early Toarcian (modified from ref. 71). Star represents this study's location. Arrows point to the UK study locations 20,23 , which are geographically close to one another.  23 . The Yorkshire dataset was originally interpreted to represent a 400-800% increase in continental weathering rates 20 ; however, other interpretations suggests that the radiogenic values during the exaratum ammonite subzone were caused by hydrographic restriction 21,22 . The close palaeogeographic proximity between these two sites, coupled with their significantly different 187 Os/ 188 Os i values suggests a regional influence on 187 Os/ 188 Os sw values in the European epicontinental seaway during the T-OAE. the record (see Fig. 3 and SI Dataset 1), which suggests a local increase in the contribution of continentally derived materials during the event. However, their concentrations remain high as 187 Os/ 188 Os i values decrease after the Toarcian CIE, which suggests a minimal influence of a detrital component of rhenium and osmium to the osmium isotopic signature (see Fig. 3, Methods, and SI Dataset 1).

Discussion
Comparison of Early Jurassic 187 Os/ 188 Os i records. Other marine 187 Os/ 188 Os i records from the Lower Jurassic (Hettangian through Toarcian stages) generally show unradiogenic values 20,23,29,30 . These are likely related to relatively elevated inputs of unradiogenic osmium from the weathering of the Central Atlantic Magmatic Province (CAMP) and the alteration of juvenile oceanic lithosphere or direct injection of mantle-derived osmium from initial opening of the North Atlantic 31 (Fig. 2). The Mochras data show higher 187 Os/ 188 Os i values just before the T-OAE CIE (~0.4), which increase to an acme of 0.8 during the T-OAE, and decrease to ~0.3 after the event 23 (Fig. 2). While the absolute 187 Os/ 188 Os i values differ between the sites, the magnitude of the excursions at East Tributary and Mochras are similar at 0.4, and are almost half the magnitude observed at Yorkshire (0.7).
The differences observed between the 187 Os/ 188 Os i records at East Tributary, Mochras, and Yorkshire suggest there were regional differences in 187 Os/ 188 Os sw during the studied interval. These differences likely represent local processes such as differing degrees of hydrographic restriction from the open ocean and the amounts of local runoff and its 187 Os/ 188 Os composition. However, the similarity in the magnitude of the excursions recorded at East Tributary and Mochras suggest this likely represents the global record of change during the T-OAE. This observation, coupled with the more extreme 187 Os/ 188 Os i excursion record at Yorkshire, supports the suggestion that the Yorkshire 187 Os/ 188 Os sw record was influenced by a local riverine input of radiogenic osmium during the T-OAE 21 , and the East Tributary and Mochras records are more representative of global osmium seawater chemistry With these observations in mind, we advocate, when possible, analyzing osmium isotope records from coeval stratigraphic successions deposited in different sedimentary and ocean basins 18, 24-26 before attempting to interpret them as a global signal. This methodology is especially important regarding palaeoceanographic studies on intervals older than the Cretaceous since the preserved records are predominantly from continental margin and epicontinental successions, where geochemical signatures have a greater potential to be modified by local processes.
Quantifying the Early Jurassic marine osmium cycle. To gain a more quantitative measure of the changes in the marine osmium cycle during the Toarcian we employed a numerical box model that simulates the osmium inventory of the ocean and its isotopic composition (see Supplemental Information). Specifically, we test whether the osmium isotope excursion associated with the T-OAE (~300-500 kyr in duration) 31,32 can be reproduced by a transient increase in the weathering input of radiogenic osmium to the ocean. We also explored other situations that may have potentially driven the observed T-OAE osmium isotope record, but are likely implausible, such as decreasing the input flux of mantle-derived osmium to zero (see Table 1 for values explored and Supplemental Information for a discussion of these cases). Overall, the numerical model results show that the osmium isotope excursion can be reproduced by a transient three-to six-fold increase in the input of continental-derived osmium to the oceans over 100 to 200 kyr 31, 32 ( Fig. 4; more details of the modelling results including sensitivity tests can be found in the Supplemental Information).
Changes in the 187 Os/ 188 Os cont to more radiogenic values through the differential weathering of lithologies such as shales and cratonic rocks [33][34][35] could have played a role in the T-OAE osmium isotope record. We investigated the potential effect this change would have on the osmium budget during the event by running simulations where we elevated 187 Os/ 188 Os cont from 1.4 to 2 (see Supplemental Information for a discussion of the choice of the maximum 187 Os/ 188 Os cont value). In these simulations, a nearly three-fold increase of the input of continental-derived osmium to the oceans was still necessary to reproduce the excursion (Fig. 4), regardless of timescale used, and solely increasing 187 Os/ 188 Os cont to reasonable values cannot reproduce the observed excursion (see Supplemental Information). Given the plausible proposition of the changing composition of the continental weathering flux, we conservatively suggest that T-OAE weathering rates increased by as much as three-fold.
A potential source of radiogenic, continentally derived osmium was the remnants of the Central Pangaean Mountains, a Himalayan-scale mountain belt in eastern North America and northwestern Africa. This mountain belt was positioned at tropical and subtropical latitudes in the Early Jurassic (Fig. 1). The rifting of Pangaea during the Late Triassic and Early Jurassic would have exposed the core of the mountain range leaving this material open to weathering or erosion. General circulation models predict large increases in the air temperature and runoff during the T-OAE in the geographic region that contained these mountains 36 . These regional climatic changes would have facilitated enhanced chemical weathering, and makes this mountain belt a plausible source of the enhanced input of osmium to the oceans advocated here.
The weathering of organic-rich rocks and sediments would be another plausible way to raise the isotopic composition of the continental weathering flux, but also results in a net release of CO 2 to the atmosphere 37  enhanced continental runoff would also have increased nutrient delivery and stimulated primary productivity in aquatic environments leading to increased hypoxia, anoxia, and potentially euxinia 5 . Elevated burial of organic matter in these environments would have sequestered much more atmospheric CO 2 than that associated with any black shale weathering, which we suggest represent only a fraction of the continental materials that were predominantly weathered during the event.
Differences in the osmium isotope response between OAE events. A striking feature of the 187 Os/ 188 Os records during the Mesozoic OAEs is the directionality of their excursions. The T-OAE records show a positive 187 Os/ 188 Os excursion, whereas the onset of the Cretaceous OAE 1a and OAE 2 both display negative excursions. The difference in the 187 Os/ 188 Os response to these events most likely lies in the environment where the LIPs were emplaced. The Cretaceous events are associated with subaqueous emplacements of the Ontong Java Plateau (OAE 1a) and the Caribbean and High Arctic LIPs (OAE 2). Emplacement of these LIPs would have supplied large amounts of unradiogenic, mantle-derived osmium directly into the oceans from weathering of basalts on the seafloor, resulting in osmium isotope excursions to nonradiogenic values 25,[38][39][40] .
The T-OAE, on the other hand, is associated with a subaerial emplacement of the Karoo-Ferrar LIP at high latitudes (Fig. 1), where the semi-arid climate would have made the relative weathering potential of this material low. In contrast to the younger OAEs, the Toarcian 187 Os/ 188 Os i records reflect enhancement of the weathering of continental materials facilitated by the injection of greenhouse gases into the atmosphere and subsequent climate changes. Notably, delivery of osmium from the Karoo-Ferrar LIP would have also been delayed, as compared to the Cretaceous LIPs. However, if weathering of the Karoo-Ferrar LIP was a significant source of osmium to the oceans during the T-OAE, then its lower 187 Os/ 188 Os compositions [41][42][43][44][45] would necessitate an even greater contribution of continental material to generate the observed 187 Os/ 188 Os i excursion.

Implications and Conclusions.
Based on the osmium isotope records and our modelling results, the transient increase in continental weathering rates during the T-OAE may be one of the largest observed during the Phanerozoic. Chemical weathering rates are also suggested to have significantly increased across the Permian-Triassic boundary 46 , Triassic-Jurassic boundary 47,48 , and the Paleocene-Eocene Thermal Maximum 49 , all of which are associated with intervals of global warming, environmental deterioration, and extinction events 50 . The rapid response of the osmium isotope system during the T-OAE, as well as during other OAEs 38-40 , indicates that chemical weathering feedbacks may respond to episodes of rapid climatic warming on short timescales (10 3 -10 6 years) and lead to a net drawdown of atmospheric CO 2 5 . Enhanced continental runoff would also have increased nutrient delivery and stimulated primary productivity in nearshore environments, leading to increased marine hypoxia, anoxia, and potentially euxinia 5 . CO 2 would also have been sequestered through the deposition of organic-rich sediments in marine and lacustrine settings 5, 6, 51 .
In the case of the Toarcian OAE, increased weathering likely played a critical role in reversing the enhanced greenhouse state induced by Karoo-Ferrar magmatism. As atmospheric CO 2 was consumed through these mechanisms, global temperatures would have declined 5,20 . As modern atmospheric CO 2 levels continue to increase at rates much higher than any point during the Cenozoic 52 , increased weathering, through the chemical and physical weathering feedbacks and stimulation of primary production and subsequent organic matter burial, may eventually act as a negative feedback to global warming, although on timescales much longer than what is necessary to mitigate the immediate environmental and ecological deterioration due to this warming 53 .
Methods δ 13 C and total organic carbon analysis. δ 13 C and total organic carbon (TOC) were measured from each sample for rhenium, osmium, and trace metals (see below). The samples were prepared and analysed using the same methods from ref 15. Rhenium and osmium isotopic analysis. In order to isolate primarily the hydrogenous rhenium and osmium from our samples, and minimize the removal of detrital rhenium and osmium, we followed the procedures of ref. 54. Between ~0.25 and 1 g of sample powder (dependent upon previously measured rhenium abundances via inductively-coupled plasma mass spectrometry) were digested with a known amount of 185 Re and 190 Os tracer (spike) solutions in 8 mL of a CrO 3 -H 2 SO 4 solution; this reaction occurred in sealed Carius tubes, which were heated incrementally to 220 °C for 48 hours. The tubes were allowed to cool before opening. The osmium was immediately isolated and purified from the acid medium by solvent extraction using chloroform. This step was followed by the back reduction of Os from the chloroform into HBr. The Os fraction was further purified by micro-distillation. Rhenium was purified from the remaining CrO 3 -H 2 SO 4 solution by a NaOH-Acetone solvent extraction 55 and further purified using anion exchange chromatography. The purified Re and Os fractions were then loaded onto Ni and Pt filaments, respectively, and analysed for their isotopic composition using negative thermal-ionization mass spectrometry (NTIMS) 56, 57 using a Thermo Scientific TRITON mass spectrometer with static Faraday collection for Re and ion-counting using a secondary electron multiplier in peak-hopping mode for Os. In-house Re and Os solutions were continuously analysed during the course of this study to ensure and monitor long-term mass spectrometry reproducibility. A 125 pg aliquot of the Re std solution and a 50 pg aliquot of DROsS yield 185 Re/ 187 Re values of 0.5983 ± 0.002 (1 SD, n = 6) and 187 Os/ 188 Os values of 0.16089 ± 0.0005 (1 SD, n = 8), respectively; both are identical to previously reported values 57 . The measured difference in 185 Re/ 187 Re values for the Re std solution and the accepted 185 Re/ 187 Re value (0.5974) 58 is used for mass fractionation correction of the Re sample data. All Re and Os data are oxide and blank corrected. Procedural blanks for Re and Os in this study were 12 ± 3 pg/g and 0.07 ± 0.05 pg/g, respectively, with an 187 Os/ 188 Os value of 0.25 ± 0.15 (n = 4). The 187 Re/ 188 Os and 187 Os/ 188 Os uncertainties are determined through full propagation of uncertainties, including those in weighing, mass spectrometer measurements, spike calibrations, blank abundances and reproducibility of standard values.

Trace metal analysis. In order to compare the changes in [Re] and [Os] to sedimentation patterns across
the T-OAE, we also analysed the concentrations of aluminum and titanium in each sample, which are used to estimate the contribution of terrigenous input to a sedimentary basin 59, 60 (see Fig. 3 and SI dataset). Approximately 0.05 g of powder was added to a teflon beaker, followed by the addition of 4 mL of a 50:50 mixture of concentrated HCl and concentrated HNO 3 . This solution was placed inside a (CEM MARS 5) microwave assisted digestion system and run until all organic material had broken down at a temperature of 150 °C. The samples were then dried down and the silicates were dissolved using 4:1 HNO 3 to HF, dried down, and re-dissolved in 5% HNO 3 solution. A 100 μL solution split was spiked with an internal standard to measure elemental abundances using an Agilent 7500cs inductively-coupled plasma mass spectrometer in He and H mode. Internal standard was used to correct the samples for machine drift. International standards USGS SCO-1 and SDO-1 were also measured and had a reproducibility of ± 5%. . HF and nitric acids in a 10:1 ratio, respectively, are added to the liner, which is then placed in a Parr high pressure device and dissolution is achieved at 220 °C for 40 hours. The resulting solutions are dried on a hotplate at 130 °C, 50 μL 6 N HCl is added to microcapsules and fluorides are dissolved in high-pressure Parr devices for 12 hours at 180 °C. HCl solutions are transferred into clean 7 mL PFA beakers and dried with 2 μL of 0.5 N H 3 PO 4 . Samples are loaded onto degassed, zone-refined Re filaments in 2 μL of silicic acid emitter 64 .

U-
Isotopic ratios are measured with a modified single collector 354S (with Sector 54 electronics) thermal ionization mass spectrometer equipped with analogue Daly photomultipliers. Analytical blanks are 0.2 pg for U and up to 1.9 pg for Pb. U fractionation was determined directly on individual runs using the EARTHTIME ET535 mixed 233-235 U-205 Pb isotopic tracer and Pb isotopic ratios were corrected for fractionation of 0.25 ± 0.03%/amu, based on replicate analyses of NBS-982 reference material and the values recommended by ref. 65. Data reduction employed the excel-based program of ref. 66. Standard concordia diagrams were constructed and regression intercepts, weighted averages calculated with Isoplot 67 . Unless otherwise noted all errors are quoted at the 2-sigma or 95% level of confidence. Isotopic dates are calculated with the decay constants λ 238 = 1.55125E-10 and λ 235 = 9.8485E-10 (ref. 68) and a 238 U/ 235 U ratio of 137.88. EARTHTIME U-Pb synthetic solutions are analysed on an on-going basis to monitor the accuracy of results.
Five single zircon grains from the bentonite at −1.9 meters in the East Tributary section (see Fig. 3) were analysed by the uranium-lead chemical abrasion isotope dilution thermal ionization mass spectrometry technique (U-Pb CA-ID-TIMS). A weighted mean 206 Pb/ 238 U age of 188.58 ± 0.17 (0.25) [0.32] Ma, (MSWD = 0.89) is based on concordant and overlapping results for three of the analysed grains (see SI Dataset 2). Older results for the other two grains suggest that they are xenocrysts and/or contain inherited cores. It is important to note that this bentonite has a previously published multigrain U-Pb ID-TIMS age of 188.3 + 1.5/−1 Ma 69 .
Five single zircon grains from the bentonite at 2.35 meters in the East Tributary section (see Fig. 3 Fig. 3). Linear interpolation was used to calculate ages between the bentonites layers and between the age assigned for the Toarcian CIE. The onset of the CIE is placed at 183.1 Ma, with a total duration of 300 kyr 31 . Sedimentation rates are also assumed to remain constant after the Toarcian CIE. The initial osmium isotopic composition of the oceans ( 187 Os/ 188 Os i ) was calculated using the following equation and the 187 Re decay constant from ref. 70 11 1 This equation accounts for the 187 Os produced after deposition by the decay of 187 Re. As stated above, the age component was derived from U-Pb ages from this succession (this study) and previously published dates for the age and estimated duration of the Toarcian CIE 31 . Furthermore, if a longer 500-kyr duration 32 is assigned to the T-OAE CIE, the calculated 187 Os/ 188 Os i values do not change significantly and our interpretations do not change (see Supplemental Information).