Introduction

Major mass extinction events during the last 500 Ma of Earth’s history coincide with the eruptions of Large Igneous Provinces (LIPs)1,2,3,4. They have been attributed to a combination of magmatic activities and greenhouse gas release2,3,5,6. During the eruptions, massive quantities of greenhouse gases (e.g., CO2 and CH4) were emitted into the atmosphere, leading to rapid global warming, which then contributed to the widespread demise of both aquatic and terrestrial ecosystems1,6,7,8,9,10. The Late Permian mass extinctions (LPME), the most severe biosphere crisis in Earth’s history, eliminated more than 90% of the Earth’s species. Two independent extinction events during the LPME, the Guadalupian-Lopingian extinction (GLE) and the Permian-Triassic extinction (PTE), occurred within a fairly short period (~10 Ma), which, in timing, could be attributed to the eruption of the Emeishan Large Igneous Province (ELIP; ~260 Ma) and Siberian Traps Large Igneous Province (STLIP; ~252 Ma) respectively5,8,9,11,12,13,14,15. High-resolution stratigraphy and paleo-biodiversity studies also support the strong correlation between the LPME and the LIPs7,15,16. The links among the global carbon cycle, climate change, and mass extinctions have been recorded in the 5 to 8 per mil (‰) negative shift in stable carbon isotopes of both carbonate and organic carbon (δ13Ccarb and δ13Corg) through the LPME7,10,12.

Although an association between global warming and the LPME has been widely accepted, temporal emission mechanisms of greenhouse gases are not entirely clear and remain a topic of discussion. The volcanism associated with the LPME triggered greenhouse gas outbursts and extreme climate changes2,5,6,7,8,9,12, primarily due to CO2 from magma degassing, thermal metamorphism, and combustion of coal, carbonates, and shales1,9,11. In contrast, CH4 released by volcanic intrusion into coal, destabilization of submarine and permafrost clathrates, and enhanced microbial methanogenesis is thought to have had a secondary effect1,7,11. However, large quantities of solid bitumen (or termed as pyrobitumen), the by-product of the pyrolysis of oils into CH417,18, can be found in areas within the ELIP and STLIP regions19,20,21. This additional source of methane, generated underground by oil cracking and subsequently released into the atmosphere, may play a much more important role during the LPME interval than previously thought. CH4 is a potent greenhouse gas, and its global warming potential is approximately 28 times relative to CO2 for a 100-year time horizon without considering climate feedback (e.g., stratospheric ozone depletion)22,23,24.

Pyrolysis of paleo-oil and related CH4 emissions induced by the volcanic activity of both the ELIP and STLIP have been investigated previously based on petrological recordings and gas-venting pipes25,26,27. It was suggested that carbon gases (CO2 and CH4) and even halocarbons might have been released into the atmosphere through pipes in both Sichuan Basin, China25 and Tunguska Basin, Siberia26 during the Late Permian. Nevertheless, previous work was based on indirect evidence in the petrological recordings to infer the possibility of gas emissions. They did not quantify the extensiveness of the pyrolysis process and had insufficient evidence of its strong correlation with the LIPs. In this study, we focus on natural gases by combining clumped isotopes for methane and isotope tracers for noble gases together with basin modeling techniques, to investigate the link between volcanism and CH4 generated by large-scale oil cracking, and evaluate its impact on the LPME.

The Sichuan Basin (~260,000 km2) in the southwestern China is an ideal case study, because (a) it is located in the outer zone of the ELIP in the Upper Yangtze platform of South China continental block (Fig. 1a); (b) abnormal heating events induced by the ELIP have been identified in the basin8,13,28 and (c) widespread pyrobitumen and natural gas pools have been discovered within the Sinian-Cambrian dolostones in the basin, indicating that massive in-situ paleo-oils have been completely pyrolyzed into CH4 in its geological history18,21,29. The ELIP is partly occupied by the Sichuan Basin and its magmas from the Emeishan mantle plume (EMP) have intruded through the Upper Yangtze sedimentary sequences (i.e., Precambrian to Silurian dolostones, marls, and shales) (Fig. 1a). Large basaltic eruptions, crustal melting, hydrothermal activity, and abnormal heating events occurred in the basin and may have initially covered >500,000 km2 from inner to outer zones1. A thick succession of marine carbonates overlies the ~800 Ma old basement, where a large-scale paleo-uplift (Leshan-Longnvsi) formed in the center of the basin and became a large petroleum system over the early Paleozoic29,30 (Fig. 1b). The sedimentary cover of the paleo-uplift includes Sinian-Ordovician marine carbonates, Permian-Triassic carbonate-clastic rocks and Triassic-Quaternary clastic rocks (Fig. 1c). The Anyue gas field, with an area of 22,000 km2, is located on the high point (Moxi-Gaoshiti Bulge) of the paleo-uplift with proven gas reserves of over one trillion cubic meters within the Sinian Dengying (Z2dn) and Cambrian Longwangmiao (Є1l) Formations. The thick black shale of the Qiongzhusi Formation was deposited within the depression on the west of the high point. As the major source rocks (dominated by Type-I kerogen), the shale generated and expelled oils which migrated into the traps within the high points during the oil generation window from Ordovician to Devonian31,32,33. Subsequently, the accumulated oils were pyrolyzed into pyrobitumen and methane gas, which were the most common fluids filling the pore space in the Sinian-Cambrian reservoirs30,31,32. Geochemical evidence from pyrobitumen indicated that the gas pools were formed by in-situ thermal pyrolysis of paleo-oil pools18,30.

Fig. 1: Tectonic and stratigraphic background of the study area and adjacent regions.
figure 1

a Map showing the geographic distribution of the Emeishan Large Igneous Province (ELIP) and Sichuan Basin in the Upper Yangtze platform, southwestern China13,28. The administration boundaries in the map are originated from map products of National Geomatics Center of China (http://www.webmap.cn). b Tectonic setting of the study area and adjacent regions. The map of burial depth depicts the Cambrian bottom and constraints the boundary of Leshan-Longnvsi paleo-uplift29. Nine gas samples for methane clumped isotope and noble gas analysis are indicated, samples not shown in the map are also collected from the same area in Sinian-Cambrian reservoirs. They are located on the Moxi-Gaoshiti Bulge as part of the Leshan-Longnvsi paleo-uplift, which is the center of the basin as well as the outer zone of ELIP. c Generalized stratigraphic column of the Sichuan Basin65. Marine environments persisted in the basement from the Sinian to the Middle Triassic controlled by the Yangtze, Caledonian, and Hercynian movements, occurring unconformities of the Sinian-Cambrian by the deformation, the Devonian-Carboniferous by the late Paleozoic lifting, and the Permian-Triassic by the ELIP-induced dome and lifting13,28,65. After the tectonic evolution during the middle Triassic, the terrestrial succession had become the main sedimentary facies until Late Cretaceous controlled by the Indosinian and Yanshannian movements33,66. The Yanshannian-Himalayan lifting led to extensive absence of the Cenozoic in the basin32,66. Symbols used in the figure include Є – Cambrian; O – Ordovician; C – Carboniferous; K – Cretaceous; E – Paleogene; N – Neogene; Q – Quaternary.

Full size image

Although previous studies have proposed that the formation of pyrobitumen within the strata may have been affected by occasional hydrothermal activity associated with the ELIP18,25,30, their evidence is not sufficient to directly link the massive generation of oil-cracked CH4 within the ELIP to the EMP26,27,31,32,33,34. It remains uncertain whether the ELIP could have acted as a widespread-impact “coking furnace” for promoting the massive generation of oil-cracked CH4 and pyrobitumen underground. Here, we examine 20 natural gas samples (dominantly CH4) collected from the Sinian-Cambrian dolostones in the Anyue gas field in the central Sichuan Basin within the outer zone of ELIP (Fig. 1b, c). We measure methane-clumped isotopes (9 samples) to obtain methane formation temperatures for investigation of the abnormal geothermal activity and determine noble gas isotopic compositions (11 samples) to understand the mantle influence on ELIP region. Furthermore, we conduct numerical simulation of basin evolution and hydrocarbon generation for the gas field to assist in constraining methane formation temperature and genesis (see Methods). This work quantifies the ELIP-induced CH4 generation within the reservoirs. The total volume of methane released into the atmosphere from the whole basin during the gas formation period is estimated as 1440 Gt.

Results and discussion

Gas composition and geochemical characteristics

The natural gas appears as a typical dry gas with a dryness index (C1/ΣC2–5) ranging from 583 to 3019. It contains CH4 ranging from 90.11% to 99.87%, C2+ ranging from 0.03% to 0.17%, CO2 ranging from undetectable to 8.36%, H2S ranging from undetectable to 1.51%, N2 ranging from undetectable to ~3.23%, and trace amounts of other gases (Supplementary Table 1). Methane shares similar δ13C values ranging from −32.2‰ to −34.2‰ and similar δD values ranging from −134‰ to −147‰, while ethane shares similar δ13C values ranging from −27.5‰ to −33.3‰ (Supplementary Table 1). Low H2S and CO2 content is likely related to minor thermochemical sulfate reduction (TSR) that took place in the reservoirs18,32. Almost all gases from the Sinian and Cambrian reservoirs were generated from oil precursors accompanied with abundant pyrobitumen from 0.1 to over 8.0%wt (average approximately 1.0%wt)17,21. In addition to extremely high dryness index (≫10), δ13C and δD values (much higher than −60‰ and −300‰, respectively) also show that gases from the Sinian and Cambrian reservoirs have reached thermal equilibrium. They fall within the equilibrium thermogenic field35 defined by using δ13C and δD values (Supplementary Fig. 6a), indicating that there are no influences from biogenic or abiotic gases35,36,37. Extremely high dryness index, a single in-situ gas source, and equilibrium conditions permit the use of the Δ13CH3D thermometer (defined by equation 4 in the Supplementary Information) for deriving formation temperatures of methane.

Constraining methane formation temperatures in the Anyue field using Δ13CH3D values

Δ13CH3D values indicate methane formation temperatures in the Z2dn samples vary from 248−16/+17 °C to 269−20/+22 °C (average of 256−20/+22 °C), while the Δ13CH3D-based temperature values of methane in Є1l samples vary from 246−15/+17 to 250−17/+19 (average of 249−17/+19 °C) (Fig. 2b; Supplementary Table 2). These temperatures are significantly higher than the present reservoir temperatures (140–165 °C)21,38 and the peak temperatures of either TSR (~140 °C)34,39 or oil-cracking (160–180 °C)40,41 (Fig. 2b; Supplementary Fig. 3). Also, they are significantly higher than the modeled reservoir temperatures (200–220 °C) of the maximum burial during Late Cretaceous (Fig. 2a, b). In contrast, clumped methane isotope temperatures are closer to the highest trapping temperatures (249–319 °C) of the quartz inclusions that have been reported in the same gas reservoirs in the Anyue gas field30, representing the invasion of deep-to-epizonogenic hydrothermal fluids corresponding to the ELIP, EMP18,25 (Fig. 2b). These generally higher temperatures derived from Δ13CH3D values indicate significant hydrothermal control in addition to the thermal effect associated with burial process.

Fig. 2: History of geological evolution matches modeled geological temperatures.
figure 2

a The burial history and modeled reservoir temperatures (Є1l and Z2dn) in the Anyue gas field, central Sichuan Basin recovered by model simulations on well-MX and well-GS (see Supplementary Fig. 1 and Supplementary Fig. 2). The reservoirs with dolomite diagenesis have allowed storage of the pre-existing oils and subsequent oil-cracked gases30. The modeled temperatures correspond to the geological evolution history in the successive order of early cementation, surface dolomitization, syngenetic dissolution, meteoric karstification, burial dissolution, burial dolomite precipitation, hydrothermal invasion, and deepest burial dissolution30,66. b Diagram of Δ13CH3D-based temperatures (Δ18-based T) vs. C1/ΣC2-5 ratios. Dashed lines specify possible formation temperatures at which methane can be generated under different conditions21,30,34,38,39,40,41. The error bars for Δ13CH3D-based temperatures are dominantly derived from ±0.1‰ standard deviation for a constant offset against the stochastic distribution (see Supplementary Information).

Full size image

The interpretation of high-temperature methane formation in the Anyue gas field is supported by the gas compositions. Studies have shown that temperature differences can generate completely different end products especially in organic reactions42. The kinetics of the oil-cracking process has two distinct stages with significant difference in gas composition of methane (C1) and heavy hydrocarbon gases (C2–5)40,41. The first stage is characterized by dominant production of C2–5 wet gases and pyrobitumen, whereas the second stage is characterized by re-cracking of the C2–5 wet gases to methane. This process leads to a progressive increasing dryness index of the gas17,40. Kinetic modeling of the C2-5 gases showed that the maximum yield of the C2–5 gases was from 168 °C to 185 °C at geological heating rates from 2 °C/Ma to 10 °C/Ma and they were completely pyrolyzed into methane at a temperature of ~265 °C40,42. Our methane generation model (Fig. 2 and also see Methods and Supplementary Fig. 3) which is based on the kinetic model and a geological heating rate of 5.0 °C/Ma for the Sichuan basin indicates that such extremely dry gases (C1/ΣC2–5 = 583–3019) in the Sinian-Cambrian reservoirs would require a formation temperature beyond 250 °C. In contrast, the formation temperature at the maximum burial depth during Late Cretaceous had not exceeded 220 °C (Fig. 2a, b). At this temperature, it was unlikely for the oils in the reservoirs to form such extremely dry gases. The gas products would have been characterized by high content of wet gases, which contradicts the gas compositions observed. Therefore, evidence from gas composition, kinetic modeling and the Δ13CH3D-based temperatures support the impact of the ELIP on the formation of methane in the Anyue gas field, by triggering abnormal heating and rapid oil cracking.

High-temperature methane formation in the Anyue gas field can also be supported by petrological evidence. Optical characteristics of pyrobitumen in the Є1l and Z2dn reservoirs18,25,30 were observed to be similar to mesophase pitch, a liquid crystal material produced at high temperatures in a coking furnace, indicating that organic matter had been transferred to a graphite crystal. Honeycomb micropores, generally observed in carbon foams, appeared in the pyrobitumen18, suggesting that the reservoirs had undergone a coking process by hydrothermal fluid invasion with rapid heating rather than gradual burial in the geothermal history. These anomalous temperatures suggest that methane generated in the Anyue gas field was mainly controlled by the invasion of hydrothermal fluids18,30 during the ELIP interval rather than the maximum burial at the Late Cretaceous, as such high temperatures in the region are only available during the ELIP period (Fig. 2). Further investigation on the mantle involvement is carried out by using noble gas tracers.

Tracing ELIP-related hydrothermal activity in the Anyue gas field using noble gases

Mantle-derived hydrothermal fluids would be expected to retain a mantle signature in He isotopes (Supplementary Table 3 and Supplementary Table 4). Measured 3He/4He ratios (R/Ra, where Ra is the atmospheric value of 3He/4He) range from 0.0115 to 0.0256 in Є1l samples showing crustal dominance of helium isotopes43, while the R/Ra values in samples from Z2dn-G1 and Z2dn-G2 range from 1.37 to 2.36 and 0.300 to 0.418, respectively, suggesting a strong mantle signature44. Here G1 and G2 are two subgroups of samples from the Z2dn reservoir, taking 1Ra as the separation standard. Because the ELIP induced by the EMP was a mafic continental large igneous province developed in a typical non-rifted continental margin13,28, we can use He ratios in the subcontinental lithospheric mantle (SCLM) as the mantle endmember of He for further study. Therefore, the mantle contribution of helium can be resolved. The results show that He in Є1l samples is almost entirely crustal derived (>99.7%), with the mantle contribution <0.30%. However, He in Z2dn-G1 and Z2dn-G2 samples are a mixture of mantle and crustal He in various proportions: The mantle 3He contributions in Z2dn-G2 samples vary from 4.80% to 6.74%. For Z2dn-G1 samples, the mantle 3He contributions are ranging from 22.43% to 38.54% (Fig. 3a). He signatures in all samples suggest that, although long-term cratonic stability of the Sichuan basin basement prevented significant volatile contributions from the mantle45,46, isolated 3He “hotspots” do exist and are likely associated with the ELIP18,30.

Fig. 3: Noble gas signatures support the link of methane with the mantle influence during the Emeishan igneous period.
figure 3

a Diagram of 4He/20Ne vs. 3He/4He (R/Ra) ratios. The 3He/4He ratio R is normalized to the atmospheric 3He/4He ratio (Ra = 1.384 × 10−6)67. Measured 4He/20Ne ratios in all samples ranged from 16.5 to 9.96 × 106. They are much higher than the ratio in air-saturated water (ASW = 0.288)68 or air (0.318)67. Therefore, the atmospheric or ASW-derived gas has negligible contribution to the He concentrations. To estimate mantle He contributions, a simple two-component mixing model44 was used between an upper crustal endmember (0.008Ra)48 and a mantle endmember relative to the sub-continental lithospheric mantle (SCLM) (6.1 ± 0.9Ra)69. b Diagram of 40Ar*/4He vs. 3He/4He (R/Ra) ratios. 40Ar* represents the resolved non-atmosphere derived excess 40Ar. The extrapolated mixing line between the crust and mantle endmembers was defined by Stuart et al. using unfractionated cases measured in Dae Hwa (South Korea) W-Mo deposit fluid inclusions44,49. The mantle 40Ar*/4He = 0.69 ± 0.06 is typical of unfractionated samples from the mantle50. In contrast, the crustal 40Ar*/4He = 0.007 is far lower than the crustal production ratio of ~0.2, typically representing a fluid derived from shallow cool regions of the crust48, in which crustal fluids often mix with a He-rich endmember due to preferential addition of 4He to the gas phase44.

Full size image

Measured 36Ar concentrations among all samples are diverse and 40Ar/36Ar ratios show clear deviation from the atmospheric 40Ar/36Ar ratio of 298.647. Measured 40Ar/36Ar ratios vary from 2168 to 5973 in Є1l samples, from 288 to 348 in Z2dn-G1 samples, and from 323 to 13558 in Z2dn-G2 samples. Resolved non-atmosphere derived excess 40Ar (40Ar*) contributes from 86.2% to 95.0% of the measured 40Ar concentrations in Є1l samples, from 0.3% to 14.7% in Z2dn-G1 samples, and from 7.5% to 97.8% in Z2dn-G2 samples. All 40Ar*/4He values in Є1l samples range from 0.148 to 0.181, which is much higher than those found in a fluid derived from shallow cool regions of the crust, where 40Ar*/4He = 0.007, but close to the crustal production ratio of 0.248 (Fig. 3b), implying the contribution from hotter (or deeper) regions of the crust where the diffusivities of noble gases are sufficiently high to permit the release of both He and Ar. From an extrapolated mixing line of 40Ar*/4He vs 3He/4He (R/Ra) between the crust and mantle endmembers44,49, the 40Ar*/4He values of Z2dn-G1 (from 0.020 to 4.140) and Z2dn-G2 (from 0.002 to 0.044) samples indicate the mixing of the crustal and mantle components, implying diverse 40Ar contributions from the mantle or possible fractionation between He and Ar by gas, water, and rock interactions (Fig. 3b). A natural logarithmic fitting curve for 3He/4He and 40Ar*/4He ratios in the Z2dn samples (R2 = 0.73) shows that higher 3He/4He ratios correlate with higher 40Ar*/4He values. However, this curve is different from the extrapolated mixing line. Sample 40Ar*/4He values are either lower than the value for the shallow cool regions of the crust (40Ar*/4He = 0.007)48 or beyond the mantle production ratio (40Ar*/4He = 0.69 ± 0.06)50 (Fig. 3b).

The Cambrian reservoirs (Є1l) contain significant crustal noble gas components. This may be related to its much thicker underlying depositional systems (carbonates and shales), resulting in higher lateral fluxes of crustal radiogenic He, but not permitting efficient vertical flux of mantle-derived gases (e.g., 3He) to the present gas reservoir (Fig. 4d, e). Compared to the Cambrian reservoir, the Sinian reservoirs (Z2dn) which recorded significant mantle-derived He, directly overlie the basement and have thinner underlying depositional formations which facilitate easier transit of mantle-derived gas and lower contributions of He from crust. However, samples with intermediate 3He/4He ratios in the Z2dn-G2 have highest 4He concentrations. This demonstrates variable contributions of both mantle and crustal He in the Sinian formations close to the basement, likely indicating a heterogeneous upwelling of mantle fluids associated with a deeper thermal pulse. Specifically, the Sinian reservoirs (Z2dn) exhibit high 40Ar*/4He values in the Z2dn-G1 samples and low 40Ar*/4He values in the Z2dn-G2 samples (Fig. 3b). The closure temperature for 40Ar* in minerals is much higher than that of He48,50,51. Therefore, He may be preferentially released from minerals at low temperature in the Z2dn-G2 reservoir and Ar may be preferentially released from minerals at high temperature in the Z2dn-G1 reservoir. High 40Ar*/4He ratios suggest that the Z2dn-G1 reservoir has received high 3He and 40Ar* fluxes close to fractures or vents, and experienced more intense hydrothermal activity, leading to higher temperatures of surrounding rocks, which facilitated the release of 40Ar* from minerals. Variation of 40Ar*/4He ratios also suggests the presence of abnormal thermal events and the heterogeneity of associated hydrothermal activity. Generally, fluid inclusions in minerals (e.g., carbonates) cannot trap gases at temperatures greater than 250 °C51. At high temperatures, noble gases would diffuse quantitatively into free gas phases producing high 40Ar*/4He ratios48. Based on the 40Ar*/4He ratios in Cambrian reservoirs (Fig. 3b), the maximum temperature in this isolated formation would be higher than 250 °C, which is consistent with the clumped isotope temperatures.

Fig. 4: The pattern of methane production and emission in the central Sichuan Basin, China.
figure 4

a Global paleogeographic map during the Late Permian-Early Triassic period16,70 and the locations of Sichuan Basin and mantle plume. b Surface uplift, generation of Emeishan basalts, crustal accretion in context of upwell of mantle plume28, and greenhouse gas emission patterns. c The formation of paleo-oil pools prior to late Permian29. The Qiongzhusi Formation (Є1q) mainly distributed at the intra-cratonic sag is the main source rock to provide hydrocarbon sources18,29,38. d The hydrothermal fluids associated with the magma events invaded the oil reservoirs during the Late Permian-Early Triassic through hydrothermal channels caused by pull-apart function of strike-slip faults53, which caused the oil-cracking, precipitation of hydrothermal minerals18,30, methane emissions to ancient atmosphere, and mantle-degassing. e The basin experienced continuous deposition until late Cretaceous. Then the extensive tectonic uplift occurred due to the Himalayan orogeny. The differential extent of uplift created a geological pattern with west of the basin higher than the east of basin. This geological adjustment further isolated the reservoirs resulting in good traps. It provided unique conditions for long-term retention of natural gases and mantle-derived gases especially in the Z2dn reservoirs. Symbols used in the figure include AnZ – Pre-Sinian; Z1 – Doushantuo Formation; Z2dn – Dengying Formation; Є1q – Qiongzhusi Formation; Є1c – Canglangpu Formation; Є1l – Longwangmiao Formation; Є2+3 – Middle-Upper Cambrian; O – Ordovician; P1 – Lower Permian; T2+3 – Middle-Upper Triassic; J1+2 – Lower-Middle Jurassic.

Full size image

The extent of methane production and emission induced by ELIP

Figure 4 shows the pattern of methane production and emission in the central Sichuan Basin induced by ELIP during its geological evolution. It suggests close association with mass extinction caused by global warming. Prior to late Permian (Fig. 4c), the high-quality source rocks of Qiongzhusi Formation (Є1q) developed in the intra-cratonic sag entered the oil generation window of formation temperatures over 120 °C during the Ordovician-Devonian period31,32,33. This supplied sufficient hydrocarbons which then formed large oil pools within high-quality Є1l and Z2dn reservoirs at the high points of the Moxi-Gaoshiti Bulge, central Sichuan Basin. Due to stable geological conditions and good seals, these oil pools were well-preserved until the late Permian ELIP. During late Permian (Fig. 4d), corresponding to the period of ELIP activity (Fig. 4a–c), high-temperature hydrothermal fluids (~319 °C)30 invaded the basin through faults, unconformities, and high-porosity reservoirs, and then efficiently pyrolyzed the oils into methane and pyrobitumen within the reservoirs31,32,52. This process has been recorded in in-situ pyrobitumen, hydrothermal veins, quartz inclusions and other hydrothermal-related minerals18,21,30,33. Because the abnormal thermal event caused by the ELIP (259.1 ± 0.5 Ma)14 might have peaked within 1−5 Ma4, the timing of oil pyrolysis coincided with the GLE (~260 Ma). The abnormally high clumped temperatures strongly support this scenario, and high 3He/4He isotopic ratios record the link between the basin and mantle during the period of ELIP.

We propose total methane generation (TMG) by oil cracking based on the total content of pyrobitumen and its yield (see Methods and Supplementary Table 5) is up to 2.01 × 1014 m3 in the Anyue region. Excluding preserved methane (total gas reserves, TGR), the total methane emission (TME) into the paleo-atmosphere could have been 2.00 × 1014 m3 or 144 Gt (>99% of total gas generated) (see Methods and Supplementary Table 5). Such a large amount of methane could be emitted through hydrothermal channels and fracture systems triggered by late Permian volcanic and seismic activities, and its release could be driven by rapid gas expansion and escape (Fig. 4d). Such channels and faults have been revealed in previous research based on seismic data. For example, the Sinian Dengying Formation contains widely developed hydrothermal channels that formed on the basis of tiny-grabens caused by the pull-apart function of strike-slip faults53, which likely provided potential faults and pipes contemporaneous with the ELIP for gas venting25. Such rapid expansion and extreme high pressures have been preserved in CH4-dominated gaseous inclusions within the quartz formed in the reservoirs52. Paleo-pressure coefficients of larger than 3.0 were recorded under Late Permian trapping depths compared to the values of the gas pool pressure coefficients of less than 1.2 under current reservoir conditions21,30. This indicates that the Anyue gas field has experienced rapid oil cracking, gas expansion, and escape processes induced by the ELIP. Gas generation by burial only played a minor role in the Anyue region.

The total volume of pyrobitumen in the Sichuan Basin and its surrounding area is over ten times that of the Anyue region17,18. Therefore, the entire Sichuan Basin could have released 1440 Gt CH4, which is equivalent to at least 40,410 Gt CO2, as the global warming potential of CH4 is at least 28 times that of CO2 in a 100-year period22,24. This is comparable to previous work which linked the GLE event with 16,800 Gt CO2 (11,200 Gt from magma release and 5600 Gt from magma intrusions) emitted by the ELIP volcanism1. The mass of oil-cracked CH4 equivalent to 40,410 Gt CO2 suggested in this work is more than twice as large as the CO2 released by the ELIP itself, or over one thousand times the annual global carbon emission by human beings (e.g., 36 Gt CO2 emission in 2019)54. This indicates that CH4 induced by the ELIP could be a key driver of global warming and climate change during the GLE event, without taking climate feedbacks into account, such as stratospheric ozone depletion by methane chemical loss22,23.

Significance of high-temperature methane emissions to the paleoclimate, LPME, and global carbon cycle

The PTE event is the largest known extinction event, coincident with the greatest STLIP volcanism, which happened 8 million years after the ELIP1,7. The STLIP volcanism may have had a magma volume 6–10 times of the ELIP55,56 in a region that had accumulated large-scale Precambrian-Paleozoic sediments containing abundant pyrobitumen19,20, in a geological setting and sedimentary framework similar to the Upper Yangtze plate (e.g., Sichuan Basin). Therefore, significantly more CH4 (probably over 104 Gt) could have been generated by the STLIP, implying CH4 emissions may directly have contributed to global warming and the greatest mass extinction event at the end of the Permian.

Furthermore, more than one-third of the erupted volcanic rocks and the entire STLIP intrusive magmatism postdated the end-Permian mass extinction (EPME)2,7,57, reducing the likelihood that CO2 release is solely responsible for the mass extinction. Although elemental and isotopic signatures of Cu and Hg have verified that felsic volcanism (S-rich vapor release) in South China might have accelerated the process of extinction7, the oil-cracked source of CH4 proposed in this study should also be recognized as an important driver. LIPs can cause abnormal heating events and hydrothermal fluids before, during, and after their massive eruptions8,13,28. The relatively low temperature (<300 °C, see Supplementary Fig. 3) required for rapid oil cracking to form pyrobitumen and CH4, can be achieved before CO2 emission from massive basalt eruptions and magma intrusions with temperatures of ~400–1200 °C1. This implies that CH4 generation and emission could occur before CO2, and more likely be responsible for the EPME event, at least in the early stages. Similarly, the massive emission of CH4 (about 7200 Gt) that has been estimated to occur during the end-Triassic Central Atlantic Magmatic Province (CAMP)58, likely also contributed to global climate change which caused the end-Triassic mass extinction58,59, which is comparable to ELIP’s 1440 Gt and SLITP’s 104 Gt CH4 emissions. This further supports the concept that high-temperature CH4 emissions induced by LIPs are of major importance for understanding mass extinctions and the carbon cycle in Earth’s history.

In this study, we propose that high-temperature methane generated by magmatic activities is an important driver for climate change and carbon cycle. It is different from thermogenic methane produced from organic matter by geological burial processes. It is also different from abiotic methane associated with Fischer-Tropsch reactions60. This type of high-temperature methane is still thermogenic in origin, but it is produced by geothermal anomalies related to magmatic and hydrothermal activity rather than sedimentary processes. The unique feature of high-temperature methane is the extremely rapid cracking of organic matter caused by rapid and extreme heating events which contrasts with slow heating and therefore slow methane generation during sedimentary burial. Although the samples in this study are collected from the outer zone of the ELIP, they still show significant impact from the heat related to the ELIP. Samples closer to the intermediate and even inner zone (Fig. 1a; Fig. 4b), would exhibit even more rapid oil pyrolysis and high-temperature methane release would be greater still.

On a global scale, oil-cracked CH4 emissions likely have contributed significantly to elevated CH4 concentrations in the atmosphere in Earth’s history as an important part of the global carbon cycle. In addition to the effect of global warming, methane, as a reactive gas, is a precursor to other pollutants in the atmosphere (e.g., CH3C1), which have significant impact on the biosphere (e.g., ozone depletion)22,23. Therefore, high-temperature methane emissions require much more attention. In Earth’s history, large amounts of high-quality source rocks and thick carbonate rocks formed large numbers of paleo-oil reservoirs before the Permian17,18,19,20,21,30,31,32,33. The volcanic activity can then induce the rapid destruction and secondary cracking of the paleo-oil reservoirs, leading to methane releases that contributed to climate change and mass extinction events, particularly in the Late Permian. This potential source of methane is critical for our understanding of the evolution of the Earth’s history. Further understanding can be gained by focusing on methane emissions associated with LIP-induced oil cracking and investigating the global distribution and abundance of reservoir pyrobitumen in the relevant areas of global LIPs.

Methods

Samples and laboratory analysis

Natural gas samples were obtained from Z2dn and Є1l reservoirs via wellheads distributed in the Anyue gas field (Fig.1b, c). Samples for noble gas and CH4 clumped isotope analysis were collected using 10 mm diameter internally polished refrigeration grade copper tubes sealed with stainless steel pinch-off clamps on both ends61. Additional samples for the analysis of gas composition, carbon and hydrogen isotopes were collected using stainless steel cylinders61. Details of all samples are available in the Supplementary Information.

Bulk gas and stable isotope analysis were undertaken at the State Key Laboratory of Organic Geochemistry (SKLOG), Guangzhou Institute of Geochemistry, Chinese Academy of Sciences (GIGCAS), using previously described techniques62,63. Bulk gas content as a percentage was determined using an Agilent 6890 N gas chromatograph-mass spectrometer (GC-MS)63. Carbon and hydrogen isotope values were measured following established procedures by using GC-IRMS of Agilent 6890N-Isoprime 100 and Thermofisher 1310-Delta V, respectively62,63. Precision and reproducibility are typically better than ±0.2‰ for δ13C (PDB) and ±2‰ for δD (SMOW). Data used in this study are reported in Supplementary Table 1 and Supplementary Table 6.

Methane-clumped isotope and noble gas analysis were undertaken at the Subsurface Fluid Isotope Geochemistry Laboratory (SFIGL) at Lancaster University, UK. Methane clumped isotopologues (13CH4, 12CH3D, 13CH3D, and 12CH4) were measured by a tunable infrared laser direct absorption spectroscopy (TILDAS, Aerodyne) that houses two continuous wave quantum cascade lasers and directly connects a methane purification system (Inlet system, Protium MS, UK) (Supplementary Fig. 4). The experimental procedure is described in Supplementary Fig. 4. Spectral acquisition and data processing were carried out by using TDL Wintel software64. All measurements were made in reference to the lab’s working reference gas (LEC-1) taken from a high purity (>99.99%) methane gas cylinder after conducting systematic calibration (see Supplementary Information). Preparation for standard calibration gas and its gas characteristics are described in Supplementary Information (Supplementary Fig. 5, Supplementary Fig. 6, and Supplementary Table 6). All measurements reported in this paper were obtained at a nominal cell pressure of ca. 1.0 Torr. Data of reservoir samples used in this study are reported in Supplementary Table 2.

Noble gas abundance and isotopic ratios were determined by an Isotopx NGX noble gas mass spectrometer at the SFIGL using previously described techniques43. The 4He, 20Ne, and 40Ar isotopes were measured using a Faraday detector while the remaining isotopes were counted on an electron multiplier. The noble gas elemental abundances for each sample were calculated by normalizing to those of air standards after blank correction. Due to low abundance of He in the atmosphere, the air standard for noble gas measurements at the lab was taken from a gas cylinder filled with a mixture of He spike and Air. The He spike is HESJ-standard with the estimated air normalizing 3He/4He ratio of 20.63 ± 0.1043. During Ne abundance and isotope analysis, appropriate mass peaks were monitored to correct for interferences caused by doubly charged ions of 40Ar++ and 44CO2 on 20Ne and 22Ne, respectively43. 4He, 20Ne, and 36Ar abundances had typical uncertainties of 1.6%, 1.7%, and 1.7%, respectively. Data used in this study are reported in Supplementary Tables 3 and 4.

Numerical simulation

Models for well-GS of Gaoshiti bulge and well-MX of Moxi bulge (Fig. 1b) were developed by using the software PetroMod version 2016.2, Schlumberger Company, which fundamentally recovered the burial-thermal history of Anyue gas field, central Sichuan Basin. In each well, the primary input parameters for building the model include the stratigraphy (lithology, thickness, and age), tectonic events (unconformities, erosion and age) and boundary conditions (paleo-water depth, history of heat flow and sediment-water interface temperature). Measured temperature and vitrinite reflectance values were used to validate the modeling results (Supplementary Fig. 1b; Supplementary Fig. 2b). The geological model of each well was completed once the modeled and measured results were consistent. Both burial-thermal diagrams are reported in the Supplementary Information (Supplementary Fig. 1a; Supplementary Fig. 2a) and target output is included in Fig. 2a, b and Supplementary Fig. 3a.

Generation models were calculated based on a kinetic model of Tang (2011) _ SARA_TI that comes with the software PetroMod version 2016.2 and can represent type-I kerogen (sapropel type organic matter) which dominated the Qiongzhusi (Є1q) source rocks31,32,33. The geological heating rate of 5.0 °C/Ma was taken from the maximum of the central basin (Supplementary Fig. 3a). The model results are reported in Supplementary Fig. 3b.

Methane emission model

Total methane emissions (TME) during the interval of LIPs are equal to total methane generations (TMG) minus total gas reserves (TGR). The TMG value can be estimated on the basis of methane and pyrobitumen yields of oil-cracking, because pyrobitumen can represent a potential bridge between gases and hydrocarbon precursors (e.g., oils)17. Previous work has investigated the relationship between pyrobitumen and methane yields by using artificial pyrolysis experiments of crude oils, showing a near-linear correlation with a slope of 1.09 between the yields from high to overhigh maturity (i.e., Easy%Ro > 1.5%)17. This slope is equal to the conversion ratio (R) between the volume of oil-cracked methane (in L) and the amount of oil-cracked pyrobitumen (in g), which is R = 1.09 L/g. Then, the gas generation intensity (GGI, i.e., amount of gas generation per unit area)17 can be calculated for assessing the scale of oil-cracked methane.

The GGI values can be determined as follows:

$${{{{{\rm{GGI}}}}}}=H\times C\times D\times R\times {10}^{7}\,{{{{{{\rm{m}}}}}}}^{3}/{{{{{{\rm{km}}}}}}}^{2}$$
(1)

Where H is reservoir thickness (m), C is pyrobitumen content (wt%), D is rock density (g/cm3), R is conversion ratio (i.e., the yield ratio of methane and pyrobitumen, L/g), and the coefficient 107 is for unit conversion.

The TMG value can be calculated by using the GGI value and the pyrobitumen distribution area (A) as follows:

$${{{{{\rm{TMG}}}}}}\,({{{{{{\rm{m}}}}}}}^{3})={{{{{\rm{GGI}}}}}}\,({{{{{{\rm{m}}}}}}}^{3}/{{{{{{\rm{km}}}}}}}^{2})\times A\,({{{{{{\rm{km}}}}}}}^{2})$$
(2)

Eventually, the TME values are estimated as follows:

$${{{{{\rm{TME}}}}}}\,({{{{{{\rm{m}}}}}}}^{3})={{{{{\rm{TMG}}}}}}\,({{{{{{\rm{m}}}}}}}^{3})-{{{{{\rm{TGR}}}}}}\,({{{{{{\rm{m}}}}}}}^{3})$$
(3)

Where the TGR value can be referred to in available results from geological survey. All data used in this study and results of calculations are listed and introduced in Supplementary Table 5.