Algal δ13C reveals climate changes during the Cambrian Explosion Authors: Chao Chang, Mingyu Zhao, Pengcheng Ju, Dongjing Fu, Ruiyun Li, Junfeng Guo, Kangjun Huang, Mingkun Wang, Linhao Cui, Yuning Yang, Mengyin Wu, Elena Yu Golubkova, Dmitrii V Grazhdankin, Robert R Gaines, Xingliang Zhang Published: 2026 ABSTRACT AbstractCarbon dioxide is a fundamental atmospheric component that regulates Earth’s climate. However, atmospheric CO2 levels (pCO2) during the Cambrian Explosion remain poorly constrained. Here we report pCO2 variations reconstructed via carbon isotope signatures of algal fossils from eight fossil deposits across the Ediacaran–Cambrian transition (~553 to ~508 Ma). Results reveal an increase in pCO2 (±1σ) from ~3–9 to ~9–21 PAL during ~553–529 Ma, followed by a progressive decline to ~5–17 PAL and ~3–8 PAL at ~517 Ma and ~508 Ma, respectively. These temporal pCO2 changes coincide with shifts in marine strontium isotope values, which, together, are most consistent with tectonically driven changes in the carbon and strontium cycles, as supported by the results of Earth system biogeochemical box modelling. Tectonic modulation over a three-stage greenhouse–hypergreenhouse–greenhouse climate could have profoundly influenced marine environments in ways that impacted the early diversification of animals.Subject terms: Carbon cycle, Geochemistry Introduction The Cambrian Explosion represents a pivotal bio-evolutionary event in Earth’s history, characterized by the rapid diversification of animals, the emergence of widespread biomineralization, and the establishment of metazoan-dominated marine ecosystems during the Ediacaran–Cambrian transition (~560–520 Ma)1. Meanwhile, the Earth system experienced significant perturbations of the atmosphere, hydrosphere, and lithosphere. Elucidating the environmental context of the Cambrian Explosion has long been a research frontier that has attracted extensive attention and multidisciplinary study. In the last decade, significant progress has been made in resolving marine redox evolution during the Ediacaran–Cambrian transition2,3, yet other important environmental factors, such as global climate and atmospheric levels of CO2 (pCO2), remain relatively poorly understood. Atmospheric pCO2 regulates global surface temperatures, which exert a significant influence on terrestrial weathering intensity, seawater pH, and marine redox states, among other effects4. Moreover, increasing evidence demonstrates that atmospheric pCO2 could have played a crucial role in geological events characterized by significant perturbations of the biosphere5–11. Reconstruction of atmospheric pCO2 across the Ediacaran–Cambrian transition is critical to elucidating the possible cause-and-effect relationships among global tectonics and environmental responses to changes in atmospheric pCO2, such as climate, marine redox states, and weathering intensity during the rapid diversification of metazoans. However, the current understanding of atmospheric CO2 levels during the Ediacaran–Cambrian transition is based almost exclusively on models of the long-term carbon cycle12–17, results from which differ significantly in both the magnitude and temporal trend of pCO2 reconstructed for this time interval. Direct estimates from analysis of the rock record (e.g., ref. 17) are necessary. Reconstruction of pre-Silurian atmospheric pCO2 is very challenging because materials required for most available proxies, such as stomata of land plants and fossils of phytoplankton or liverworts, are either absent or rarely preserved in the geological record4. Nevertheless, dissolved CO2 concentrations in the ocean ([CO2]aq) are a major factor influencing the magnitude of photosynthetic carbon-isotope fractionation (Ɛp) via marine primary producers. Therefore, Ɛp provides an opportunity to understand variations in pCO2 in deep time via empirical equations which have been established from cultivation experiments and natural observations of phytoplanktonic algae, making quantitative [CO2]aq and pCO2 reconstruction via geological records of photosynthetic Ɛp possible18–21. Photosynthetic Ɛp has been calculated from Mesoproterozoic–recent records and has proven to be useful in constraining [CO2]aq and pCO2, in particular for pre-Silurian strata, even though large uncertainties often arise from poor constraints on potential diagenetic alteration and other parameters included in calculation22–28. Typically, either body fossils of phytoplankton or specific biomolecules such as alkenones and phytane are used for Ɛp calculation and pCO2 reconstruction23,25–28. However, the geological record of alkenones can only be reliably traced to the Early Cretaceous27,29,30, while phytane is not preserved in Cambrian and older strata due to generally high thermal maturity (e.g., refs. 25,31–33). Alternatively, bulk-rock δ13Corg signatures have been used to reconstruct pCO2 in deep time in previous studies22,24. However, bulk-rock δ13Corg may not accurately represent the primary carbon-isotope signature of primary producers because organic contributions from anaerobic chemoautotrophs in sediments can be significant, particularly under the lower oxygen conditions of the Ediacaran and early Cambrian oceans34,35. In this study, we target another important group of photoautotrophs in marine ecosystems that are common in some Ediacaran to Cambrian fossil deposits, macroalgae36. Here, we report δ13Corg data of macroalgal fossils collected from eight fossil deposits spanning the late Ediacaran to the early middle Cambrian, including seven fossil deposits from China and one from Russia: the Miaohe biota (~553 Ma), the Dongpo biota (~545 Ma), the Vasilevsky Ostrov biota (~530 Ma), the Yanjiahe biota (~529 Ma), the Chengjiang biota (~517 Ma), the Qingjiang biota (~517 Ma), the Balang biota (~514 Ma), and the Kaili biota (~508 Ma) (Fig. S1; Supplementary Information 1). In these fossil deposits, macroalgae are preserved as organic compression fossils and contain organic remains that are sufficient to readily isolate and analyze (Fig. 1)34. Ɛp for each fossil deposit was calculated based on the algal δ13Corg data. Carbon-isotope analysis of multiple carbonaceous fossil specimens of the same type from each of the eight formations maximizes the possibility that the observed variation in photosynthetic Ɛp is primarily controlled by temporal changes in marine [CO2]aq and atmospheric pCO2. Therefore, tentative quantitative calculation of marine [CO2]aq and atmospheric pCO2 for each interval was carried out using empirical equations. These data reveal significant pCO2 variations from 553 Ma to 508 Ma, results that fit well with observed changes in surface environmental factors that are closely associated with atmospheric pCO2 levels, such as continental weathering intensity and marine redox state. Together, these results are consistent with tectonic control over atmospheric pCO2 through degassing during this critical interval. These findings have profound implications for understanding the carbon cycle and environmental changes during the Ediacaran–Cambrian transition. Carbonaceous algae fossil Qingjiangthallus cystocarpium from the Qingjiang biota. A Reflected light image of the fossil specimen. B EPMA image of (A) showing carbonaceous composition of the fossil. The color gradient scale in (B) denotes the relative abundance of carbon. Results All algal fossils analyzed are dark in color (Fig. S1), and Raman spectroscopic analyses confirmed that the fossils are composed of carbonaceous material (Fig. S2). Algal δ13Corg was measured using carbonaceous material recovered directly from the algal fossils, and at least 14 measurements were carried out for each of the fossil deposits analyzed (Supplementary Data 1). Reported algal δ13Corg data for the Miaohe biota were also included in our calculation34. Analyzed algal fossils yield a wide range of δ13Corg values from <−38‰ to >−24‰, with a relatively narrow range of isotopic signatures for samples from within each fossil deposit (4–6‰) (Fig. 2A; Supplementary Data 1). Algal δ13Corg is systematically higher than bulk δ13Corg in all the studied fossil deposits (Fig. S3; Supplementary Data 1), suggesting that the bulk δ13Corg was likely affected by organic contributions from highly 13C-depleted anaerobic chemoautotrophs in anoxic sediments34,35 (Supplementary Information 2.1). Variations in key parameters and calculated pCO2 for the studied fossil deposits. A algal δ13Corg, B algal Δ13Ccarb–org, C algal Ɛp, and D atmospheric pCO2 estimated based on algal δ13Corg. Error bars of the x and y axes indicate the uncertainty range of estimated ages for the fossil deposits and the 1 SD uncertainty range of the data points estimated by Monte Carlo simulations. For (A–C), gray circles correspond to values generated from the analytical measurements of algal samples, while black circles correspond to the mean values of the sample suite for each individual deposit. For (D), circles indicate pCO2 values computed from the analytical data from algal samples, whereas large diamonds indicate the mean of the estimated atmospheric pCO2 for each corresponding deposit. Based on algal δ13Corg values from the fossil deposits, the photosynthetic fractionation factor Ɛp was determined using the equation where Δ13Ccarb–org is the isotopic difference between inorganic carbon from sedimentary carbonates and organic matter, Δ2 is the potential isotopic shift of residual organic matter relative to primary biomass, and Δcarb is the isotopic depletion of dissolved CO2 relative to carbonates22. As all algal fossils were collected from mudstones, coeval background δ13Ccarb for each fossil deposit was estimated based on δ13Ccarb data obtained from carbonate cements of the fossil-bearing mudstones or from coeval carbonate rocks in adjacent areas; for Δ2, a recommended value of 1.5‰ is adopted for all the fossil deposits following previous works22,24; and Δcarb is calculated individually for each fossil deposit based on constraints on seawater temperature provided by previous studies (Supplementary Information 2.2–2.4). By evaluating algal and bulk-rock carbon isotopic signatures of the fossil deposits, a constant carbon isotopic difference (Δ13Calgae–phy) of 0.9‰ between the coexisting phytoplankton and macroalgae was assumed for all the fossil deposits (Supplementary Information 2.6). Corresponding marine [CO2]aq and atmospheric pCO2 were then tentatively calculated using Monte Carlo simulations by combining Eq. (1) and the following Eqs. (2) and (3): Equation (2) was developed by refs. 20, 22, where b accounts for physiological factors that can influence the isotopic fractionation associated with photosynthesis by phytoplankton, including cell geometry, growth rate, cell carbon allocation, and bicarbonate uptake, etc.21; Ɛf is the maximum isotopic fractionation induced by photosynthetic carbon fixation; and Ɛp is the estimated isotopic fractionation associated with photosynthesis for phytoplanktonic algae. Equation (3) is Henry’s law, which allows estimation of pCO2 from [CO2]aq by assuming that the ocean and atmosphere were in equilibrium. k0 in the equation is the Henry constant and can be obtained using a function of temperature (T) and salinity (S) as follows: ln(k0) = 9345.17/T − 60.2409 + 23.3585 ln(T/100) + S [0.023517 − 0.00023656 T + 0.0047036 (T/100)2]37. Aside from the algal δ13Corg and δ13Ccarb data, other parameters used in Eqs. (1)–(3) were estimated based on available data reported in previous works (Supplementary Information 2). The adopted values and uncertainty ranges of the parameters are provided in Supplementary Data 2. With all the parameters set, the concentration and uncertainty range of pCO2 for the fossil deposits were estimated sample by sample using Monte Carlo simulations by combining the results of Eqs. (1) and (2) directly into Eq. (3). The results overall reveal an atmospheric pCO2 range of 4–17 PAL (pre-Industrial atmospheric level: 280 p.p.m.v.), with possible minimum and maximum concentrations reaching 3 PAL and 21 PAL, respectively, within 1σ (68%) uncertainty (Fig. 2D and Supplementary Data 2). Our data suggest that average atmospheric CO2 levels for the fossil deposits increased from 6 PAL at ~553 Ma to 14 PAL at ~529 Ma, and then progressively decreased to 7 PAL at ~517 Ma, and 5 PAL at ~508 Ma (Fig. 2D). Discussion Our results provide a novel record of atmospheric pCO2 from the terminal Ediacaran to the Cambrian Miaolingian (553–508 Ma) based on analysis of geological materials. It is important to acknowledge, however, that the results of each fossil deposit span a fairly wide range of pCO2 values and involve large uncertainties (Fig. 2D). For each of the parameters in Eqs. (1)–(3), values and uncertainties are adopted conservatively following previous studies and using Monte Carlo simulations (Supplementary Information 2). However, the possibility remains that the values we adopted diverge from the true ancient parameters, as available constraints on most of the parameters are limited and difficult to accurately quantify. Changes in the adopted parameter values can significantly affect the reconstructed pCO2 levels, particularly the maximum concentrations. For instance, when the assumed temperatures are increased by 2 °C, the estimated average atmospheric pCO2 range for the fossil deposits shifts from 5–14 PAL to 5–17 PAL (Fig. S10A). Conversely, a decrease of 2 °C in the adopted temperatures leads to lower average atmospheric pCO2 estimates, ranging from 4 to 12 PAL (Fig. S10B). Therefore, caution is warranted when interpreting the absolute pCO2 values estimated from fossil deposits. It should be emphasized, however, that relatively large uncertainties were assigned to all parameters in the Monte Carlo simulations. For example, as robust seawater temperature reconstructions for the late Ediacaran–early Cambrian are lacking, a relatively warm seawater temperature range of 27–31 °C was tentatively adopted for pCO2 calculations of the fossil deposits, based on data from published modeling and analytical studies (Supplementary Information 2.4 for details). To mitigate the uncertainty introduced by this tentative temperature assignment, a conservative, large uncertainty margin of ±4 °C was implemented for the temperature parameter (T) in the Monte Carlo simulations. This conservative approach to parameterization ensures that the full range of adopted temperatures, combined with the associated uncertainty, has a high probability of fully encompassing the actual seawater temperatures relevant to the fossil deposits. Moreover, pCO2 estimates for each fossil deposit were calculated on a sample-by-sample basis. Most deposits exhibit a variation in algal δ13Corg values exceeding 2‰ (Fig. S3), which further increases the likelihood that the true atmospheric pCO2 levels fall within the reconstructed 1σ (68%) uncertainty range. From this perspective, the pCO2 results for the Terreneuvian Yanjiahe Formation are less representative than those for other fossil deposits, as only four samples were included in the Monte Carlo simulations after excluding samples potentially affected by matrix organic matter (Supplementary Information 2.6). Nevertheless, the reconstructed pCO2 levels for the Yangjiahe biota are comparable to the maximum pCO2 estimates for the broadly coeval Vasilevsky Ostrov Formation (Fig. 2D), supporting the conclusion that these results also lie within reasonable ranges. On the other hand, several lines of evidence lend additional support to the reliability of the pCO2 results recovered here. First, the temporal variation of pCO2 mirrors that of algal Δ13Ccarb–org, suggesting that this trend is driven by the δ13Corg and δ13Ccarb data rather than by values of other parameters adopted for Monte Carlo simulations. In addition, the estimated average pCO2 levels (~5–14 PAL) are consistent with pCO2 constraints provided by coeval climatic and biological changes (4–64 PAL)16 and similar pCO2 results are revealed from the coeval Chengjiang and Qingjiang biotas (Fig. 2D; Supplementary Data 2). Although the reconstructed pCO2 levels for Cambrian Series 2 are lower than those reconstructed by available models (Fig. 2D; refs. 12–15), pCO2 concentrations <10 PAL during this interval are in line with observations of widespread calcifying cyanobacteria, suggesting that pCO2 was possibly below ~10 PAL38–40. Moreover, a significant increase in continental weathering intensity during the Terreneuvian was recently demonstrated by lithium isotope records from carbonate and siliciclastic successions, consistent with the effects of elevated pCO2 reconstructed from our algal δ13Corg data41,42. Elevated pCO2 levels during the Terreneuvian are expected due to increased seafloor spreading associated with the breakup of Pannotia43, an increasing mantle depletion rate44, and an expansion in subduction-zone length6, which would have been accompanied by enhanced mantle degassing of CO2 associated with arc volcanism and seafloor spreading at mid-ocean ridges45. Our findings are also consistent with a compilation of detrital zircon U–Pb ages that suggests a peak in active arc magmatism during the early Cambrian5, and direct evidence for active continental arc volcanism and hydrothermal activity during the late Terreneuvian has been well established from the South China Block42,46–48. The close association between atmospheric pCO2 variation and deep degassing rates is further corroborated by the strong correlation between the trajectory of estimated pCO2 and oceanic 87Sr/86Sr signatures during the Ediacaran–Cambrian transition (Fig. 3A, C)49,50. Diagram showing the results of the CHEES model during the Ediacaran–Cambrian transition. A Oceanic 87Sr/86Sr signatures constrained by sedimentary 87Sr/86Sr data reported in previous studies (black data points, compiled in ref. 50) and modeling results of seawater 87Sr/86Sr variation (black line) forced by hydrothermal input through mid-ridge spreading; B Temporal changes in mid-ridge spreading rates; C atmospheric pCO2 variation induced by CO2 degassing through mid-ridge spreading; D timeline of major evolutionary events, with color intensity or band width indicating the rate of taxonomic diversification (adapted from refs. 1,59). Shaded areas in (A–C) represent 1 SD uncertainty ranges of modeled seawater 87Sr/86Sr, mid-ridge spreading rates, and atmospheric pCO2, respectively. The density bar indicates the relative confidence of the modeled results. Data points in (C) represent estimated atmospheric pCO2 in this study. Modeling results of MAGic, geoclimTEC, GEOCARBSULFOR, and COPSE in (C) are from refs. 12,13,14,15, respectively. Seawater 87Sr/86Sr is determined by a balance of mantle input to seawater via mid-ocean ridges and riverine input to seawater from continental weathering51. Mantle input is consistently less radiogenic with an 87Sr/86Sr value of ~0.703, whereas 87Sr/86Sr of the riverine input is complicated by weathering of different lithologies: ~0.705 for basalt, ~0.715 for felsic rocks, and values for carbonate rocks lying generally close to seawater at the time of their formation15,51,52. Recent osmium isotope data demonstrated that enhanced erosion of continental arc-related basaltic rocks could have contributed to the decrease of oceanic 87Sr/86Sr signatures during the early Terreneuvian42. Siliciclastic lithium isotope records, however, demonstrated both enhanced continental weathering and a shift to more felsic weathering terranes during the late Terreneuvian41, while the oceanic 87Sr/86Sr signatures continued to decrease (Fig. 3A). In the context of elevated terrestrial input of radiogenic Sr, a significant increase of mantle Sr input through mid-ocean ridge hydrothermal activity is required to account for the observed decrease in seawater 87Sr/86Sr, an interpretation that matches well with a rapid increase in seawater Sr concentrations during the Ediacaran–Cambrian transition recently revealed from analysis of fluid inclusions53. Concomitant mantle degassing from mid-ocean ridges could have contributed to a further increase in atmospheric CO2 concentrations during the late Terreneuvian (Fig. 2D). The subsequent increase in seawater 87Sr/86Sr suggests that hydrothermal input of mantle-derived Sr declined relative to more-radiogenic continental sources that progressively dominated the oceanic Sr reservoir (Fig. 3A), as supported by geological evidence for weakening hydrothermal activity during the Cambrian Epoch 246,48. Consequently, the amount of CO2 contributed by deep degassing would have dropped significantly, while the consumption of atmospheric CO2 remained high as a result of strong continental weathering41, leading to a sustained decrease of pCO2 during Cambrian Epoch 2 (Fig. 2D). To further explore this scenario, a modified version of the CHEES (CHemical Evolution of Earth Surface Spheres) Earth system biogeochemical box model54 was employed to investigate coupled variations in the carbon and strontium cycles across the Ediacaran–Cambrian transition. CHEES is a long-timescale forward model configured with five oceanic boxes, which resolves the first-order mass exchanges between the ocean, atmosphere, and sediments, and incorporates all major source–sink processes governing the long-term carbon cycle (see details in “Methods”). We analyzed the predicted variations in Sr isotopic composition and atmospheric pCO2 resulting from changes in ridge spreading rate and the consequent flux of high-temperature hydrothermal input, while terrestrial riverine Sr fluxes were held constant. Within the modeling framework, the CO2 degassing rate was scaled with the ridge spreading rate and inverted during the model–data fitting exercise to reproduce the reconstructed pCO2 curve. Driven solely by this mechanism, the model reproduced strontium isotopic and pCO2 trends consistent with geological observations (Fig. 3). A sensitivity analysis was conducted by varying the climate sensitivity parameter. A higher climate sensitivity necessitates greater CO2 degassing to match the pCO2 record. For example, increasing climate sensitivity from 3 to 5 °C requires a 70% increase in peak degassing to reproduce the observed pCO2 (Fig. S10). However, the two- to three-fold increase in the degassing rate during the studied period is likely still within a reasonable range. When taking enhanced continental arc volcanism into consideration42, the model produces a lower intensity of mid-ocean ridge hydrothermal activity during the early Terreneuvian, but the close association between deep degassing and elevated atmospheric pCO2 remains. It is worth noting that the model predicts a peak in pCO2 at ~520 Ma based on seawater 87Sr/86Sr, which postdates the highest pCO2 values reconstructed from algal δ13Corg data (Fig. 3). However, precise confirmation of this prediction lies beyond the scope of this study because of the patchy distribution of fossil deposits bearing carbonaceous algal fossils. Nevertheless, recent geochronological studies have identified supereruptions and associated CO2 release during the same interval, lending support to the possibility that a peak in pCO2 indeed occurred55. Taken together, the modeling results lend strong support to deep degassing as the primary control on pCO2 variation during the Ediacaran–Cambrian transition and suggest the possible trajectory of a more continuous pCO2 record. Carbon dioxide has exerted a primary control on variations of Earth’s surface temperature and planetary climate over geologic time4,6,56. Despite uncertainty in the absolute values of pCO2 estimated here, our results reveal that Earth’s planetary climate could have experienced considerable variation during the Ediacaran–Cambrian transition, with a terminal Ediacaran greenhouse followed by a Terreneuvian hypergreenhouse that in turn was followed by greenhouse conditions in Cambrian Epoch 2 (Fig. 3C)57. Mild greenhouse climatic conditions also prevailed during the middle Ordovician and are suggested to have increased seawater carbonate saturation, paving the way for the Great Ordovician Biodiversification Event58. Consistently, the advent of carbonate biomineralization in metazoans during the terminal Ediacaran also occurred under greenhouse conditions (Fig. 3D; ref. 1). Moreover, the peak in Archaeocyathid diversity occurred primarily during the Cambrian Age 3, following the transition from hypergreenhouse to greenhouse conditions (Fig. 3C, D; ref. 59). These correlations suggest that the development of greenhouse climate may have played a critical role in modulating carbonate biomineralization during the Ediacaran–Cambrian transition58. In contrast, the Terreneuvian hypergreenhouse climate could have suppressed oceanic mixing and thermal circulation56, facilitating the expansion of marine anoxia and promoting a shallow chemocline in the ocean2,41,60–62. Reduced O2 solubility at elevated temperatures likely contributed to the development of low-oxygen shallow marine environments (e.g., ref. 63), providing an ecological window for dominance of metazoans with low O2 demand during Phase 2 of the Cambrian Explosion, while contributing to the delayed diversification of metazoans with higher O2 requirements during Phase 3 (Fig. 3D; refs. 62,64). In summary, isotopic data from carbonaceous algal fossils across eight Ediacaran to Cambrian fossil deposits allows for exploratory reconstruction of atmospheric pCO2 from ~553 to ~508 Ma at a temporal resolution of ~10-million years. These results suggest that tectonic controls possibly regulated the balance of CO2 consumption via continental weathering and CO2 input via arc/ridge degassing during the Ediacaran–Cambrian transition. These new data reveal a three-stage greenhouse–hypergreenhouse–greenhouse trajectory of planetary climate, which likely influenced contemporaneous seawater chemistry and biological evolution during the Cambrian Explosion. Overall, our results and interpretations align well with existing geological evidence. It should be noted, however, that the pCO2 estimates presented here are subject to considerable uncertainty and are at a temporally low resolution. While the general trend of pCO2 variation appears plausible, more complex and higher-frequency fluctuations almost certainly occurred across the Ediacaran–Cambrian transition. Therefore, future studies—employing newly developed pCO2 proxies and more advanced Earth system models (e.g., refs. 25,28,65–67)—will be essential to reconstruct more continuous and precise atmospheric CO2 records throughout this interval, and better clarify the interactions among tectonic, climatic, and biological processes during the rise of complex life on Earth. Methods Organic carbon-isotope analysis of algal fossils Algal fossils from the Vasilevsky Ostrov Formation were extracted directly through acid dissolution of the rock matrix using ref. 68. Carbonaceous remains of algal fossils from all other deposits were observed under a magnifying glass and then scraped from the rock matrix using a ceramic knife. Recovered fossil materials were acidified using 1 M HCl to remove any carbonate component. The residues were washed with ultra-pure water and dried in an oven at 60°C. Organic carbon isotopic analysis of algal fossils was carried out at the Shaanxi Key Laboratory of Early Life and Environment, Xi’an, China. The decarbonated fossils were loaded into tin capsules, combusted at 1000 °C and converted to CO2 gas in a Thermal Fisher elemental analyzer. The coupled Conflo Ⅳ interface automatically transfers CO2 into the Delta V Advantage isotope ratio mass spectrometer for isotopic measurement. The organic carbon-isotope ratio was expressed in δ notations as permil (‰) relative to the Vienna Pee Dee Belemnite (V-PDB) standard. An international reference material (UREA, No. IVA33802174: δ13C = −37.02‰) and an in-house reference standard (Acetanilide: δ13C = −29.53‰) were added to the run after every 12 samples in order to monitor analytical error. Precision (1 SD) is better than 0.2‰ for δ13Corg. For each fossil deposit, at least one sample was selected and analyzed three times with an average standard deviation (1 SD) of 0.25‰ for δ13Corg. Inorganic carbon-isotope analysis Inorganic carbon and oxygen isotope measurements were carried out at the Shaanxi Key Laboratory of Early Life and Environment, Xi’an, China. Fresh interiors of carbonate samples were chipped and ground to 200 mesh. Approximately 0.2 mg powdered samples were loaded into a 96-position Gasbench Ⅱ headspace autosampler and reacted with concentrated H3PO4 (density = 1.92 g/mL) at 70 °C for 3 h to extract CO2 from carbonate minerals. The CO2 generated was measured on a Delta V Advantage isotope ratio mass spectrometer coupled with a Gasbench Ⅱ interface. Delta values of inorganic carbon and oxygen isotopes were reported relative to the V-PDB standard. Two working standards were analyzed for every 10 samples, with δ13C and δ18O values of +0.13 ‰ and –8.18‰ (VPDB), respectively. This working standard was calibrated by the international standard IAEA CO8 with δ13C and δ18O values of –5.76 ‰ and –22.7‰ (VPDB), respectively. The external precision (1 SD) of working standards is better than 0.1 ‰ for δ13C and 0.2 ‰ for δ18O. Box modeling The model used in this study is an updated version of the CHEES (CHemical Evolution of Earth Surface Spheres) box model, originally developed by ref. 54 as a coupled long-timescale ocean–atmosphere–sediment biogeochemical Earth system model. CHEES belongs to a well-established family of forward Earth system box models, which are widely applied to interrogate the coupled evolution of carbon, oxygen, nutrient, and redox cycles across geological timescales. Notably, this model has been previously validated and applied to simulate continental weathering processes, the phosphorus (P) cycle, and the carbon cycle during the early Cambrian69,70. For this study, we configured the model with five discrete oceanic boxes: the shallow margin, deep margin, low-latitude surface ocean, high-latitude surface ocean, and deep ocean. We explicitly simulated all major source–sink processes that regulate the long-term carbon cycle, including air–sea CO2 exchange, volcanic degassing, silicate and carbonate weathering, marine primary productivity, organic carbon burial, and carbonate burial. Full governing equations, key parameters, present-day reservoir sizes, and climate sensitivity analysis are documented in the Supplementary Information. In our simulation, CO2 degassing is implemented as a time-dependent tectonic forcing scaled linearly to mid-ocean ridge spreading rate. During model–data calibration, the ridge spreading history is iteratively adjusted to replicate the reconstructed atmospheric pCO2 record. Accordingly, the resulting degassing history is interpreted as an inferred time-dependent forcing within our adopted model framework, rather than an independently constrained boundary condition or a fully prognostic model output. This critical distinction must be accounted for when interpreting the model results and their associated uncertainties. To test whether the same tectonic–hydrothermal forcing can explain the observed evolution of seawater strontium isotopes, we incorporated an oceanic Sr reservoir and its associated source–sink fluxes into the model framework (modified from refs. 71,72). The model includes a major oceanic (OSr) Sr reservoir to track the effect of significant variations in Sr fluxes and associated isotope compositions. The main sources of Sr to the ocean are river influx (fSrriv), high-temperature hydrothermal (fSrHT), low-temperature hydrothermal (fSrLT), and diagenetic (fSrdia), while the main sink of Sr from the ocean is carbonate burial (fSrcarb). The contribution of Sr from high-temperature hydrothermal fluids is scaled to degassing (ccdeg) and spreading rate: Therefore, the differential equations for the Sr reservoirs are: An isotopic fractionation (δSrdia) of 0.00075 was applied for the diagenetic Sr sink71. Sr isotope compositions in ocean (δOSr) follow the mass balance: In Eq. 6, δ is used to represent 87Sr/86Sr ratios of different reservoirs. Modern values are incorporated for all fluxes except for fSrriv when constructing the mass balance model parameters (see Supplementary Data 3). Modern estimates of strontium flux and isotopic composition are utilized as references for parameterization71,73–75. In our simulation, riverine Sr input, low-temperature hydrothermal Sr flux, and diagenetic Sr release are held constant at modern values, while high-temperature hydrothermal Sr input is scaled linearly to ridge spreading rate and thus to CO2 degassing. The Sr module, therefore, serves as an independent internal consistency check on our tectonic forcing framework, rather than as a direct inverse constraint on degassing magnitude itself. Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Data 1-6 Transparent Peer Review file Acknowledgements The authors sincerely thank Zhenfei Wang, Yawen Lu, and Qingshan Wang for their technical assistance with the analytical work, and to Jun Xiao, Yunqiang Wang, Fei Zhang, Haidong Zhang, and Lijing Liu for stimulating and fruitful discussions. Author contributions Conceptualization: C.C., XZ. Methodology: C.C. Resources: C.C., D.F., J.G., K.H., M.W., L.C., Y.Y., M.W., X.Z., D.G., E.G. Investigation: C.C., R.L. Data curation: C.C., R.R.G. Modeling: M.Z., P.J. Supervision: X.Z. Writing—original draft: C.C. Writing—review and editing: C.C., D.G., R.R.G., X.Z. Peer review Peer review information Nature Communications thanks Swapan Sahoo and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Funding National Natural Science Foundation of China grant U24A20601; National Key Research and Development Program of China grant 2022YFF0802700; National Natural Science Foundation of China grant 41930319; National Natural Science Foundation of China grant 42272006; National Natural Science Foundation of China grant 42072127; 111 Project of China grant D17013; Russian Science Foundation grant 24-17-00142. Data availability All data are available in the main text or the Supplementary Materials. Code availability The code for the Monte Carlo simulation is uploaded to GitHub (https://github.com/jupc/Chang2024). For details of the box model, please contact M.Z. and P.J. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Chao Chang, Email: changchao@nwu.edu.cn. Robert R. Gaines, Email: robert.gaines@pomona.edu Xingliang Zhang, Email: xzhang69@nwu.edu.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-74516-9. References Associated Data Supplementary Materials Supplementary Information Description of Additional Supplementary Files Supplementary Data 1-6 Transparent Peer Review file Data Availability Statement All data are available in the main text or the Supplementary Materials. The code for the Monte Carlo simulation is uploaded to GitHub (https://github.com/jupc/Chang2024). For details of the box model, please contact M.Z. and P.J.