Carbon Isotope Chemostratigraphy Across the Permian-Triassic Boundary at Chaotian, China: Implications for the Global Methane Cycle in the Aftermath of the Extinction

During the end-Permian extinction, a substantial amount of methane (CH4) was likely released into the ocean-atmosphere system associated with the Siberian Traps volcanism, although fluctuations in the global CH4 cycle in the aftermath of the extinction remain poorly understood. The carbon (C) isotopic composition of carbonate (δ13Ccarb) across the Permian-Triassic boundary (P-TB) was analyzed at Chaotian, South China. The δ13Ccarb values decrease from ca. +1 to –2‰ across the P-TB, possibly caused by a collapse of primary productivity associated with the shallow-marine extinction. The frequent intercalation of felsic tuff layers around the P-TB suggests that a volcanogenic carbon dioxide (CO2) input to the surface oceans may also have contributed to the δ13Ccarb decline. The magnitude of the δ13Ccarb decrease (∼3‰) is substantially smaller than the magnitude of a decrease in C isotopic composition of organic matter (δ13Corg) in the same P-TB interval (∼7‰). This apparent δ13Ccarb-δ13Corg decoupling could be explained by proliferation of methanogen (“methanogenic burst”) in the sediments. A global δ13C compilation shows a large variation in marine δ13Corg records, implying that the “methanogenic burst” according to the Siberian Traps volcanism may have contributed, at least in part, to the δ13Corg variability and to the elevated CH4 levels in the atmosphere. The present and previous observations allow us to infer that the global CH4 cycle may have fluctuated substantially in the aftermath of the extinction.

The global carbon (C) cycle was likely perturbed during the Permian-Triassic transition in association with those geologic events that potentially contributed to the extinction (e.g., Kump and Arthur, 1999). Stable C isotope geochemistry is useful to correlate sections in different regions and to reveal the changes in the global C cycle (e.g., Hayes et al., 1999). Plenty of studies analyzed the C isotopic composition of carbonate (δ 13 C carb ) of P-TB rocks and documented a negative δ 13 C carb shift by ∼5‰ during the extinction in various sections around the world, including South China, Iran, Armenia, northern Italy, Austria, Slovenia, and Pakistan (e.g., Magaritz et al., 1988;Baud et al., 1989;Holser et al., 1989;Jin et al., 2000;Richoz, 2006;Horacek et al., 2007a;Horacek et al., 2007b;Korte and Kozur, 2010 and references therein; Shen et al., 2013;Schobben et al., 2016;Joachimski et al., 2020). Although several studies pointed out a substantial diagenetic overprint and/or erosional unconformity on the P-TB δ 13 C carb records (e.g., Heydari et al., 2001;Heydari and Hassanzadeh, 2003;Grasby and Beauchamp, 2008;Yin et al., 2014;Schobben et al., 2016;Li and Jones, 2017), the widely recognized P-TB δ 13 C carb decrease has been generally regarded as an original isotopic signal of seawater (Bagherpour et al., 2019).
Considering a simple box model of the surface-ocean C pool, two principle mechanisms could explain the P-TB δ 13 C decrease in marine carbonates: 1) a decrease in the output and 2) an increase in the input of 13 C-depleted C (e.g., Korte and Kozur, 2010). A collapse in primary productivity and reduction in biological pump ("strangelove oceans") corresponds to the former mechanism (e.g., Kump, 1991), although it may have produced only a ∼3‰ negative δ 13 C carb shift (Rampino and Caldeira, 2005) and might not be sufficient to explain larger P-TB δ 13 C declines commonly observed around the world. The increase in the input of 13 C-depleted C to the surface oceans seems to be more important (e.g., Payne and Clapham, 2012), and several geologic events have been proposed for the C injection, including volcanogenic CO 2 emission involved in the Siberian Traps volcanism (e.g. Renne et al., 1995;Hansen, 2006), methane (CH 4 ) release during destabilization of submarine and permafrost clathrates or thermal alteration of coal by volcanic intrusion (e.g. Erwin, 1993;Morante, 1996;Krull and Retallack, 2000;Berner, 2002;Sarkar et al., 2003;Retallack and Jahren, 2008), enhanced erosion and reoxidation of sedimentary organic matter or terrestrial soil (e.g. Baud et al., 1989;Holser et al., 1989;Ward et al., 2000;Sephton et al., 2005), and the oceanic overturn or shoaling of deep-water (e.g., Kajiwara et al., 1994;Knoll et al., 1996;Algeo et al., 2007a). Korte and Kozur (2010) comprehensively correlated the P-TB δ 13 C carb records of shelf carbonates on a global scale. Based on a gradually decreasing trend toward the P-TB, the authors suggested that the δ 13 C carb decrease was caused by a combination of the Siberian Traps volcanism and a shoaling of anoxic deep-waters onto shelves.
A negative shift of the C isotopic composition of organic C (δ 13 C org ) across the P-TB has been reported in marine strata in many sections such as Arctic Canada (e.g., Grasby and Beauchamp, 2008;Algeo et al., 2012), Greenland (e.g., Twitchett et al., 2001), Spitsbergen (e.g., Wignall et al., 1998;Zuchuat et al., 2020), South China (e.g., Cao et al., 2002), Japan (e.g., Takahashi et al., 2010), and Australia (e.g., Morante, 1996). The marine δ 13 C org decrease has been correlated normally with the negative δ 13 C carb shift in shelf carbonates by assuming that the δ 13 C carb and δ 13 C org trends were parallel and the difference between the δ 13 C carb and δ 13 C carb values (Δ 13 C) was consistent during the Permian-Triassic transition, although the Δ 13 C value is generally controlled by several factors such as carbon dioxide (CO 2 )-fixing enzymes and atmospheric CO 2 levels (pCO 2 ) (e.g., Hayes et al., 1999). A parallel δ 13 C carb and δ 13 C org decline has been reported in the P-TB intervals in Austria (Magaritz et al., 1992), Italy (Sephton et al., 2002), mid-Panthalassa (Musashi et al., 2001), and Kashmir (Algeo et al., 2007b). However, some studies pointed out a δ 13 C carb -δ 13 C org decoupling across the P-TB. The Δ 13 C values apparently decrease in some sections in South China, Slovenia, and Iran , and increase in some other Chinese sections (e.g., Shen et al., 2012).
The δ 13 C org decrease in a terrestrial P-TB succession has also been reported in South China (e.g., Shen et al., 2011a), South Africa (e.g., Ward et al., 2005;Gastaldo et al., 2020), Antarctica (e.g., Krull and Retallack, 2000), Australia (Morante, 1996), Madagascar (de Wit et al., 2002), and Germany (Scholze et al., 2017). However, the apparent δ 13 C org trends on land are generally variable compared to those of marine sediments, and the P-TB δ 13 C org decline is not clearly recognized in many terrestrial sections (e.g., Retallack et al., 2005;Fielding et al., 2019). Moreover, the P-TB is generally not well assigned in the terrestrial successions due to their poorer biostratigraphic constraints. Under these circumstances, it is still difficult to correlate the terrestrial δ 13 C org records well with the marine δ 13 C org and δ 13 C carb records.
Previous studies particularly suggested that a substantial amount of CH 4 was released into the ocean/atmosphere during the Siberian Traps volcanism via several processes, including volcanic intrusion into coal (e.g., Retallack and Jahren, 2008;Shen et al., 2012;Rampino et al., 2017), destabilization of submarine and permafrost clathrates (e.g., Krull et al., 2000;Krull et al., 2004;Brand et al., 2016), and enhanced microbial methanogenesis ("methanogenic burst") (Rothman et al., 2014). As CH 4 is a potent greenhouse gas, the huge CH 4 input may have contributed to climate warming in the aftermath of the extinction (e.g., Hallam and Wignall, 1997;Joachimski et al., 2012;Sun et al., 2012;Cui and Kump, 2015). However, the global CH 4 cycle during the Permian-Triassic transition has been poorly examined and constrained. In this study, we analyzed the δ 13 C carb records of the P-TB carbonates at Chaotian in northern Sichuan, South China. Together with the previously reported δ 13 C org records of the same interval, we examined the sedimentary C cycle in eastern Paleotethys during the Permian-Triassic transition. Moreover, we compiled the δ 13 C carb and δ 13 C org records of P-TB successions in various marine and terrestrial environments around the world, to examine whether the isotopic signal detected at Chaotian was a global one. Based on those results, we suggest fluctuations in the global CH 4 cycle in the aftermath of the end-Permian extinction.

GEOLOGICAL SETTING AND STRATIGRAPHY
During the Permian to early Triassic, South China was isolated from other continental blocks and located at low latitudes on the eastern side of Pangea ( Figure 1B; Muttoni et al., 2009). Shallowmarine carbonates and terrigenous clastics with diverse fossils accumulated extensively on its platform (e.g., Zhao et al., 1981;Yang et al., 1987;Jin et al., 1998). In northern Sichuan along the northwestern edge of South China, carbonates and mudstones of deep-water facies accumulated on a slope/basin that faced on the eastern Paleotethys ( Figure 1C; Zhu et al., 1999;Wang and Jin, 2000). The Chaotian section is located ca. 20 km to the north of Guangyuan city in northern Sichuan (Figure 1; 32°37′N, 105°51′E; Isozaki et al. 2004). At Chaotian, Guadalupian (middle Permian) to lowermost Triassic carbonates are continuously exposed along the bank of the Jialingjiang River in a narrow gorge called Mingyuexia. We mapped the eastern bank of the gorge on the southern limb of an E-W trending anticline.
The Permo-Triassic rocks at Chaotian (>300 m thick in total) are composed of the Guadalupian Maokou Formation, the Lopingian Wujiaping and Dalong formations, and the lowermost Triassic Feixianguan Formation, in ascending order ( Figure 2; Isozaki et al., 2004;Isozaki et al., 2007;Saitoh et al., 2013a;Saitoh et al., 2013b;Saitoh et al., 2014a). The Maokou Formation (>150 m thick) is composed mainly of massive dark gray bioclastic limestone with diverse shallow-marine fossils, such as calcareous algae, brachiopods, ostracodes, crinoids, rugosa corals, and fusulines. However, the uppermost part (∼11 m thick) of the Maokou Formation is composed of thinly bedded black calcareous mudstone and black chert/siliceous mudstone with abundant radiolarians, conodonts, and ammonoids. The Wujiaping Formation (∼70 m thick) is composed mainly of massive dark gray bioclastic limestone with black chert nodules/lenses, containing shallow-marine fossils such as fusulines, smaller foraminifer, rugosa corals, calcareous algae and brachiopods. At the base of the Wujiaping Formation, a ca. 2 m thick tuffaceous Wangpo bed occurs. The Dalong Formation (∼25 m thick) is composed mainly of thinly bedded black mudstone, black siliceous mudstone and dark gray limestone with radiolarians, ammonoids, bivalves, and brachiopods. The Feixianguan Formation (>30 m thick) is composed mainly of thinly bedded light gray micritic limestone containing few conodonts, ammonoids, and brachiopods. Zhao et al. (1978) and Yang et al. (1987) originally described the overall biostratigraphy of the Chaotian section based on fusulines, conodonts, and ammonoids. Isozaki et al. (2004) and Isozaki et al. (2007) re-examined the stratigraphy of the section that spans across the two mass extinction intervals; i.e. the Guadalupian-Lopingian boundary (G-LB) and the P-TB in higher resolution. More analyses on the litho-, bio-, and chemo-stratigraphy of the Chaotian section added further information (Isozaki et al., 2008;Lai et al., 2008;Saitoh et al., 2013a;Saitoh et al., 2013b;Jost et al., 2014;Saitoh et al., 2014b;Saitoh et al., 2015;Saitoh et al., 2017). Isozaki et al. (2007) and Ji et al. (2007) constructed the detailed lithostratigraphy and conodont zonation for the ∼12 m thick interval across the P-TB, and suggested that intermittent felsic volcanism may have contributed to the extinction. Cao et al. (2010) analyzed the δ 13 C carb values around the P-TB and found a negative δ 13 C excursion around the extinction horizon. Saitoh et al. (2014a) analyzed the nitrogen and organic C isotopic composition of the ∼40 m thick P-TB interval at Chaotian and suggested enhanced nitrogen fixation in the anoxic oceans throughout the Changhsingian.
In the present study, we focused on the ∼40 m thick P-TB interval at Chaotian (Figure 2). This interval is identical to that analyzed in Saitoh et al. (2014a), which contains the ∼12 m thick carbonates analyzed in Isozaki et al. (2007). Fresh rock samples, collected from outcrops and from core samples by scientific drilling to a depth of >30 m, were prepared as polished slabs and thin sections for describing microtextures by petrographical observations. The analyzed P-TB interval is composed of three stratigraphic units: 1) the upper Wujiaping Formation, 2) the Dalong Formation, and 3) the lowermost Feixianguan Formation, in ascending order. The upper Wujiaping Formation (∼10 m thick) is composed mainly of massive dark gray limestone (lime mudstone/wackestone) with some sandy/muddy limestones ( Figure 3D). Black chert nodules (<10 cm in diameter) occur in the uppermost part of the Wujiaping Formation. The upper Wujiaping limestones contain diverse shallow-marine fossils such as calcareous algae, crinoids, brachiopods, radiolarians, and ostracodes. Burrows frequently occur in the upper Wujiaping limestones.
The Dalong Formation (∼25 m thick) is composed mainly of thinly bedded black calcareous mudstone, black siliceous mudstone, dark gray muddy limestone, and bedded gray limestone (lime mudstone/wackestone) (Figures 2, 3). The Dalong Formation contains abundant radiolarians with a minor amount of ostracodes, brachiopods, ammonoids, and conodonts, bivalves and small foraminifers. The uppermost (∼3.5 m thick) part of the Dalong Formation mostly consists of bedded gray limestone (lime mudstone/wackestone) ('Unit C and D' in Isozaki et al., 2007). Thin (<10 cm thick) acidic tuff layers frequently occur in these limestones. Burrows are observed in the lower and upper Dalong Formation, although bioturbation is generally absent in the middle Dalong Formation. Small pyrite framboids (mostly 3-7 µm in diameter) occur abundantly throughout the Dalong Formation. The lowermost Feixianguan Formation (∼5 m thick) is composed of thinly bedded gray marl and light gray micritic limestone with some sandy/muddy limestones. In particular, the lowermost (1.4 m thick) part ('Unit E' in Isozaki et al., 2007) is composed of gray marl. This marl unit is almost barren of fossil, although few ammonoids and bivalves occur from the basal part. The upper part of the lowermost Feixianguan Formation consists of light gray micritic limestone containing few conodonts and brachiopods. Few trace fossils are recognized in the lowermost Feixianguan Formation.
Based on index fossils (fusulines, conodonts, ammonoids, corals, and brachiopods), the analyzed P-TB interval at Chaotian is dated as follows ( Figure 2; Zhao et al., 1978;Yang et al., 1987;Isozaki et al., 2004;Isozaki et al., 2007): The upper Wujiaping Formation is correlated with the Wuchiapingian (early Lopingian). The lower Dalong Formation is correlated with the late Wuchiapingian, whereas the middle Dalong Formation belongs to the late Wuchiapingian to early Changhsingian. The upper Dalong Formation is correlated with the late Changhsingian. The lowermost Feixianguan Formation is correlated with the latest Changhsingian to early Induan (early Early Triassic). The main extinction horizon is assigned at the base of the Feixianguan Formation ('Unit D/E boundary' in Isozaki et al., 2007), whereas the biostratigraphically defined P-TB is assigned at the base of the overlying micritic limestones ('Unit E/F boundary' in Isozaki et al., 2007). Based on the litho-and bio-facies, the sedimentary environments of the three stratigraphic units of the analyzed P-TB interval were reconstructed ( Figure 2; Saitoh et al., 2014a). The upper Wujiaping limestones were deposited on the shallow euphotic shelf under oxic conditions. In contrast, the lower and middle Dalong Formation was deposited on the relatively deep slope/basin under anoxic conditions. The upper Dalong limestones were deposited on the relatively shallow slope below the storm wave base under oxic conditions. The lowermost Feixianguan formations were deposited on a relatively shallow slope under anoxic conditions.

ANALYTICAL METHODS
Fresh rock samples were carefully chosen based on detailed observations of polished slabs and thin sections. Powdered sample was reacted with 100% phosphoric acid at 28°C for 24 h using a GasBench (Thermo Fisher Scientific). The extracted CO 2 was separated in a chromatography line with a helium flow, and the carbon and oxygen isotope ratios were measured with a DELTA V PLUS mass spectrometer. The carbonate carbon and oxygen isotopic compositions are presented using the delta notation δ 13 C (( 13 C/ 12 C) sample / ( 13 C/ 12 C) standard −1) and δ 18 O (( 18 O/ 16 O) sample /( 18 O/ 16 O) standard −1), respectively. The δ 13 C and δ 18 O values are reported in ‰ relative to the Vienna Peedee Belemnite (V-PDB) standard. The analytical reproducibility of the δ 13 C carb and δ 18 O carb values was better than 0.1 and 0.1‰, respectively. Table 1 lists all the measured δ 13 C carb and δ 18 O carb values of the P-TB interval. Figure 4 shows chemostratigraphic profiles of the δ 13 C carb , δ 18 O carb , δ 13 C org , and Δ 13 C values and TOC contents. The δ 13 C org values and TOC contents were previously reported in Saitoh et al. (2014a). Figure 5 shows a δ 13 C carb -δ 18 O carb cross plot. The δ 13 C carb values range from −6.9 to +4.3‰, with an average value of ca. 0‰. The δ 18 O carb values range from −7.9 to −2.9‰, with an average value of ca. −5.9‰.

RESULTS
The δ 13 C carb values are consistently ca. +4‰ in the upper Wujiaping Formation and decrease from ca. +4‰ to −2‰ across the Wujiaping/Dalong formation boundary. In the Dalong Formation, the δ 13 C carb values increase slightly from ca. −2 to +1‰ upward except for some anomalously low (<−3‰) values (open symbols in Figure 4). The δ 13 C carb values decrease from ca. +1‰ to −2‰ across the Dalong/ Feixianguan formation boundary (i.e., the P-TB) and are consistent around −2‰ in the lowermost Feixianguan carbonates. The present δ 13 C carb results around the P-TB are generally identical to the results in Cao et al. (2010). The δ 18 O carb values are somewhat variable in in the upper Wujiaping limestones and are mainly around −7‰ in the overlying Dalong Formation. Above the P-TB, the δ 18 O carb values are mostly around −6‰ in the lowermost Feixianguan Formation. The Δ 13 C values are mainly between 27 and 29‰ in the upper Wujiaping Formation and decrease slightly to ca. 26‰ in the Dalong Formation. The Δ 13 C values increase to ca. 30‰ in the overlying lowermost Feixianguan carbonates.
No linear correlation is observed between the δ 13 C carb and δ 18 O carb values ( Figure 5). It indicates that secondary overprinting on the Chaotian carbonates is not significant (Knauth and Kennedy, 2009). In the chemostratigraphic profile, however, some δ 13 C carb values in the Dalong Formation are anomalously low (<−3‰; open symbols in Figure 4) and these samples were likely affected at least partly by the secondary addition of 13 C-depleted C. Except for these anomalous values, the present δ 13 C carb results represent a smooth chemostratigraphic trend in the analyzed interval. This trend may record secular changes in the δ 13 C value of dissolved inorganic carbon (DIC) in seawater (δ 13 C DIC ) in the Lopingian to earliest Triassic. Based on the C isotopic analyses of the P-TB carbonates in Iran, Schobben et al. (2016) suggested that the first-order δ 13 C carb trend in the bulk carbonates is robust, although small-scale isotopic fluctuations may be due to secondary alteration. In the following discussion, we focus solely on the first-order δ 13 C carb trend in the analyzed interval at Chaotian. The δ 13 C carb values unidirectionally decrease from ca. +4 to −2‰ across the Wujiaping/Dalong formation boundary at Chaotian (Figure 4). Although the timing of this negative δ 13 C carb shift is not well constrained because of the poor occurrence of index fossils, the δ 13 C carb shift likely occurred around the Wuchiapingian-Changhsingian boundary. This δ 13 C carb decrease apparently coincides with the deepening of the sedimentary setting from a shallow shelf to relatively deep slope/basin floor. However, we emphasize that carbonates in the Dalong Formation of deep-water facies comprise mainly finely fragmented bioclasts and few secondary dolomite ( Figures 3E, G-J). Thus, regardless of the lithofacies change, this prominent δ 13 C carb decrease likely records the Lopingian secular change in the δ 13 C DIC value in the surface oceans in the eastern Paleotethys. Two possible mechanisms could explain the δ 13 C carb decrease: 1) collapse of primary productivity in the surface oceans and 2) addition of isotopically light C into the DIC pool of the surface oceans. As certain shallow-marine taxa went extinct across the Wuchiapingian-Changhsingian boundary (e.g., Knoll et al., 1996;Bambach, 2006), extinction-related productivity deficiency may have contributed to the δ 13 C carb decrease at Chaotian. However, the magnitude of the δ 13 C carb drop (∼6‰) seems to be too large to be caused solely by the collapse of primary productivity (e.g., Berner, 2002); thus the addition of 13 C-depleted C to the DIC pool in the surface oceans likely contributed to the δ 13 C carb drop. We emphasize that the sedimentary setting deepened from oxic shelf to anoxic slope/basin floor. The ubiquitous occurrence of small pyrite framboids in the Dalong Formation suggests the dominance of sulfate reduction in an anoxic deep-water mass on the slope/basin. The DIC in the deep-water mass would have become enriched in 12 C due to the anaerobic respiration; therefore, the injection of 13 C-depleted C into the surface oceans via shoaling of the deep-water might have contributed to the negative δ 13 C carb shift. It is noteworthy that no eruption of a large igneous province or substantial sea-level change occurred on a global scale around the Wuchiapingian-Changhsingian boundary (e.g., Haq and Schutter, 2008;Bond and Wignall, 2014). An input of volcanogenic excess CO 2 and/or reoxidized sedimentary organic C during a eustatic sea-level fall is unlikely for the cause of the δ 13 C carb drop at Chaotian.
Regional and global correlations would be useful to constrain the cause of the δ 13 C carb decrease at Chaotian. Around the Wuchiapingian-Changhsingian boundary, a ca. 6‰ negative δ 13 C carb shift was reported at Shangsi, ca. 60 km southwest of Chaotian, in northern Sichuan (Bai et al., 2008;Shen et al. 2013), which is almost identical in magnitude to that at Chaotian. Therefore, the deep-water carbonates in northwestern South China probably record the common δ 13 C DIC decrease in eastern Paleotethys. Similar δ 13 C carb declines were also reported from other sections in South China, including Dukou in Sichuan, Shiligou in Chongqing, Zhuqiao in Hubei, and Heshan and Matan in Guangxi (Shao et al., 2000;Shen et al., 2013;Yang et al., 2019); nonetheless, the magnitude of the δ 13 C carb decreases (∼2-5‰) is variable and smaller than that at Chaotian. At Abadeh in central Iran, Richoz (2006) found a negative δ 13 C carb shift in the corresponding interval in the Hambast Formation, which was later re-confirmed by Liu et al. (2013). A similar negative δ 13 C carb excursion (∼4‰) was also reported from the equivalent horizons of Julfa beds/Alibashi Formation boundary at Kuh-e-Ali Bashi in northwestern Iran . These almost co-eval negative δ 13 C carb shifts around the Wuchiapingian-Changhsingian boundary in Iran are comparable to that at Chaotian. Moreover, a large negative δ 13 C carb excursion (∼7‰) was demonstrated in the Bellerophon Formation at the Reppwand section in the Carnic Alps, Austria (Buggisch et al., 2015), which is also correlative to the negative shift at Chaotian. A negative δ 13 C org shift, as well as the δ 13 C carb drop, around the Wuchiapingian-Changhsingian boundary has been documented in several sections in South China and Arctic Canada (e.g., Bai et al., 2008;Beauchamp et al., 2009;Wei et al., 2015;Liao et al., 2016). These δ 13 C org decreases may have likewise recorded fluctuations in the global C cycle; however, a negative δ 13 C carb /δ 13 C org shift around the Wuchiapingian-Changhsingian boundary is not clear in the rest of the Permian world (e.g., Baud et al., 1996;Korte et al., 2004;Richoz, 2006;Baud et al., 2012). Even at the abovementioned sections in which a negative δ 13 C shift is recognized, the magnitude of and the trend in the δ 13 C decrease vary substantially . Under the circumstances, we infer that the C cycle in the surface oceans around the Wuchiapingian-Changhsingian boundary was not globally uniform but rather widely variable possibly owing to local factors, such as primary productivity and oceanic circulation along the continental margins. More studies with high chemoand bio-stratigraphic resolutions in various sections around the world are necessary to reveal the fluctuations in the global C cycle around the Wuchiapingian-Changhsingian boundary.
δ 13 C carb Decrease Across the P-TB The δ 13 C carb values decrease from ca. +1 to −2‰ across the Dalong/Feixianguan formation boundary (P-TB) at Chaotian (Figure 4). This negative δ 13 C carb shift is identical to that previously reported at Chaotian in Cao et al. (2010). Likewise, two possible mechanisms should be considered for the cause of this δ 13 C carb decrease: 1) collapse of primary productivity in the surface oceans and 2) addition of 13 C-depleted C to the DIC pool in the surface oceans. The negative δ 13 C carb shift occurs across the major extinction horizon, suggesting that the shift was driven by the collapse of primary productivity during the extinction. The addition of isotopically light C to the shallow-marine DIC pool is another plausible candidate for the cause of the δ 13 C carb decrease. Isozaki et al. (2007) found frequent intercalations of thin felsic tuff layers across the extinction horizon at Chaotian and suggested volcanic stress on the shallow-marine biota during the extinction. An excess input of volcanogenic CO 2 (δ 13 C: ∼−5‰) may also have contributed to the δ 13 C carb decrease. Weathering of sedimentary organic matter due to a large regression is another possible mechanism for the δ 13 C carb decrease (e.g., Yin et al., 2014). However, the sea-level drop occurred significantly before the δ 13 C carb shift at Chaotian Saitoh et al., 2014a), without any evidence for shallowing across the extinction horizon ( Figure 4). This suggests that a contribution of reoxidized organic C to the δ 13 C carb drop was negligible. Song et al. (2012a) proposed the existence of a large vertical δ 13 C carb gradient in the end-Permian oceans, along which a deepening of the sedimentary setting possibly caused the observed δ 13 C carb decrease. At Chaotian, however, the upper Dalong and lowermost Feixianguan carbonates across the P-TB were deposited on a relatively shallow slope without abrupt sea-level changes. Thus the δ 13 C carb decrease cannot be attributed to the assumed vertical δ 13 C carb gradient in the water column.
Δ 13 C Increase Across the P-TB Saitoh et al. (2014a) reported the δ 13 C org chemostratigraphy of the P-TB interval at Chaotian and revealed that the δ 13 C org values drop abruptly by 7‰ (from −25 to −32‰) immediately above the extinction horizon (Figure 4). This clear δ 13 C org decrease is consistent apparently with the δ 13 C carb shift documented in the present study, although the magnitude of the δ 13 C org decrease (∼7‰) is substantially larger than that of the δ 13 C carb decrease (∼3‰). The Δ 13 C ( δ 13 C carb −δ 13 C org ) values increase by 4‰ from +26‰ to +30‰ in the aftermath of the extinction at Chaotian. Two possible mechanisms may explain this Δ 13 C increase: 1) an increase in C isotopic fractionation of biological C fixation and 2) additional input of 13 C-depleted C to the sedimentary organic matter. The high aqueous CO 2 concentration ([CO 2 (aq)]) generally promotes the large C isotopic fractionation during photosynthetic C fixation (ε p ) in the surface oceans (e.g., Rau et al., 1992, Rau et al., 1997Kump and Arthur, 1999). The Δ 13 C values are ∼26‰ in the middle to upper Dalong Formation at Chaotian, which are consistent with the typical Calvin cycle (e.g., Schidlowski et al., 1983), although the values increase substantially to ca. 30‰ above the P-TB (Figure 4). It should be noted that a substantial amount of CO 2 was released during the Siberian Traps volcanism around and mostly after the P-TB (Supplementary Information) (e.g. Renne et al., 1995;Hansen, 2006;Cui et al., 2013;Cui and Kump, 2015). The increased [CO 2 (aq)] may have promoted greater ε p (e.g., Kump and Arthur, 1999).
The previous estimates of the amount of emitted CO 2 and of the elevated pCO 2 during the Permian-Triassic transition are useful to constrain the influence of ε p changes on the δ 13 C records. Cui and Kump (2015) estimated that pCO 2 rose from 500 to 4,000 ppm to ∼8,000 ppm during the extinction (Supplementary Information). Kump and Arthur (1999) derived a simplified relationship between pCO 2 and ε p from Frontiers in Earth Science | www.frontiersin.org February 2021 | Volume 8 | Article 596178 isotopic data of modern marine algae, with some assumptions: ε p ≈ 25-2,301/pCO 2 . According to this relationship, ε p with pCO 2 of 500, 4,000, and 8,000 ppm is calculated apparently to be ∼20.4, ∼24.4, and ∼24.7‰, respectively. Thus, the relatively large δ 13 C org decline (∼7‰) compared to the δ 13 C carb decrease (∼3‰) at Chaotian can be explained by the enlarged ε p according to the elevated pCO 2 . However, it is obviously difficult to extrapolate the relationship formula between pCO 2 and ε p in Kump and Arthur (1999) directly to the end-Permian system, due to several reasons. First, several assumptions with large uncertainties are included in this simplified formula, such as the dissolved phosphate concentration in the surface seawater. Moreover, ε p is controlled, not only by pCO 2 , but also by temperature and the growth rate of phytoplankton, in general (e.g., Rau et al., 1997;Kump and Arthur, 1999). Second, this relationship formula was obtained empirically on the basis of data of modern haptophyte algae and seawater (Bidigare et al., 1997). Different coefficients on the formula would be more appropriate for the end-Permian phytoplankton communities with unknown physiology. Third, the relationship was obtained on the basis of observations of modern environments with relatively low [CO 2 (aq)] (calculated pCO 2 < 850 ppm, assuming equilibration according to Henry's Law at 25°C). It is uncertain whether the formula can be extrapolated to environments with substantially high [CO 2 (aq)], which are particularly supposed in the aftermath of the end-Permian extinction (Cui et al., 2013;Cui and Kump, 2015). Nonetheless, the low sensitivity of ε p to pCO 2 under high-pCO 2 conditions is probably essential (Rau et al., 1997;Kump and Arthur, 1999). Because the reciprocal dependence of ε p on pCO 2 is attributed theoretically to isotopic discrimination by diffusion of CO 2 from the ambient seawater to the phytoplankton cell (Laws et al., 1995). It is also suggested that the sensitivity of ε p to pCO 2 of land plants is lower than that of marine phytoplankton (e.g., Popp et al., 1989). The estimated high pCO 2 up to 4,000 ppm before the end-Permian extinction (Brand et al., 2012;Cui et al., 2013;Cui and Kump, 2015) allows us to postulate that the influence of ε p change on the Chaotian δ 13 C records was not significant.
Additional input of 13 C-depleted C to the sedimentary organic C pool is the other possible mechanism for the Chaotian P-TB Δ 13 C increase. Rothman et al. (2014) suggested that increased Ni input to the ocean/atmosphere during the Siberian Traps volcanism promoted methanogen proliferation in the oceans ("methanogenic burst"). Methanogens generally fix C via the reductive acetyl-CoA pathway, which can produce greater C isotopic fractionation than the Calvin cycle (e.g., Preuss et al., 1989). The organic C from methanogenic biomass may have contributed, at least partly, to the Δ 13 C increase. Terrestrial plants are another possible C source for sedimentary organic matter in the Chaotian carbonates. In general, the δ 13 C org value of Permian terrestrial plant is thought to be higher than the value of marine phytoplankton (e.g., Faure et al., 1995;Foster et al., 1997;Korte et al., 2001). Thus, if the terrestrial C flux decreased across the P-TB, the bulk δ 13 C org and Δ 13 C values of the sediments would have decreased and increased, respectively. However, two lines of evidence exclude the possibility of the decreased terrestrial flux at Chaotian. First, as discussed above, the water depth of depositional site did not change substantially across the P-TB, and a large change in terrestrial flux owing to sea-level changes is unlikely. Second, previous studies suggested enhanced chemical weathering and increased continental flux during the Permian-Triassic transition (Algeo and Twitchett, 2010;Algeo et al., 2011;Cao et al., 2019). At Chaotian, the enhanced chemical weathering is supported by the increased supply of clay minerals/micas around the P-TB based on bulk nitrogen isotope records (Saitoh et al., 2014a). The terrestrial flux may have increased (rather than decreased) in the aftermath of the extinction, and thus the observed Δ 13 C increase cannot be attributed to the decreased terrestrial C flux. Also, the putative vertical δ 13 C org gradient (Luo et al., 2014) cannot explain the Δ 13 C increase because the water depth of depositional site did not change substantially across the P-TB at Chaotian.
In summary, the P-TB Δ 13 C increase at Chaotian was most likely due to the proliferation of methanogen in the sediments during the Siberian Traps volcanism (Figure 4), corresponding to the "methanogenic burst" event at a local scale (Rothman et al., 2014).
Global δ 13 C Correlations Around the P-TB Regional δ 13 C Differences The δ 13 C profile across the P-TB at Chaotian is here relatively perceived through global correlation, in the context of global C cycle during the Permian-Triassic transition. We focused particularly on the apparent magnitude of the δ 13 C decrease across the P-TB in previous studies and examined its frequency distribution on a global scale (Figures 6, 7). All the reference sections in the current compilation are summarized in Table 2. We categorize the compiled sections geographically into seven realms; i.e., Boreal, eastern Paleotethys, western Paleotethys, western Pangea, mid-Panthalassa, Neotethys, and Gondwana ( Figure 6).

Marine Records
Previous δ 13 C carb studies are concentrated mostly in eastern (China) and western Tethys realms (Iran to Italy) (Figures 6, 7; Table 2), where extensive carbonate platforms developed under warm tropical climate. These two realms share almost the same frequency distribution of the magnitude of the P-TB δ 13 C carb decrease (Figure 7), suggesting a common δ 13 C DIC decline throughout Paleotethys during the Permian-Triassic transition. The magnitude of the δ 13 C carb decrease ranges mostly between 3 and 6‰ in those tropical regions. In the Boreal realm, the magnitude of the P-TB δ 13 C carb decrease in few marine records is substantially large (to 19‰), which indicates a diagenetic overprint onto the original isotopic signal from seawater (Mii et al., 1997). In contrast, the magnitude of the P-TB δ 13 C org decrease varies substantially around the world (Figures 6, 7). In particular, a relatively large (∼7‰) δ 13 C org decline has been reported in several marine sections in high latitudes such as Greenland, Spitsbergen, Australia, and Antarctica (e.g., Retallack and Jahren, 2008;Nabbefeld et al., 2010). It is consistent with the previous notion that the magnitude of the δ 13 C org decrease in high latitudes was substantially larger than that of the δ 13 C carb decrease in equatorial regions (e.g., Krull et al., 2000;Krull et al., 2004;Korte and Kozur, 2010;Payne and Clapham, 2012;Saltzman and Sedlacek, 2013;Yuan et al., 2015).

Terrestrial Records
The present δ 13 C compilation also demonstrates that the P-TB δ 13 C org decrease is not clearly recognized in a number of terrestrial sections, especially around the Neotethys and in Gondwana realms ( Figure 6). The δ 13 C org decline is not clearly recognized in previous studies at 14 out of 49 terrestrial sections ( Table 2). In these sections, the δ 13 C org values are sometimes substantially scattered and no smooth δ 13 C org trend is observed (e.g., de Wit et al., 2002;Retallack et al., 2005;Coney et al., 2007). In other Gondwanan sections, the δ 13 C org decline is recognized and the apparent magnitude of the δ 13 C org decrease is mostly not large (Figure 7). However, the δ 13 C org decline is obscured frequently by sharp δ 13 C org drops to −45‰ (e.g., Krull and Retallack, 2000;Retallack et al., 2005).
Variable Δ 13 C Records Magaritz et al. (1992) originally proposed a parallel δ 13 C carb and δ 13 C org trend on the basis of the P-TB isotope profile of the Gartnerkofel Core from the Carnic Alps, Austria, and later studies confirmed similar parallel trends not only in the Tethyan but also in the Panthalassan realms (e.g., Musashi et al., 2001;Algeo et al., 2007b;Luo et al., 2014). However, other studies pointed out a decoupling between the δ 13 C carb and δ 13 C org trends around the P-TB. For example, Riccardi et al. (2007) analyzed the δ 13 C carb and δ 13 C org values of the P-TB carbonates at Meishan and Shangsi in South China, and compiled the Δ 13 C changes during the extinction in Iran, Slovenia, Japan, Austria, FIGURE 6 | Paleolocation map of the P-TB sections examined in inorganic (upper) and organic (lower) carbon isotope studies. Marine and terrestrial sections are shown as circles and squares, respectively. Filled and opened symbols represent sections in which the reported magnitude of the P-TB δ 13 C carb decrease is large (>5‰) and small (<5‰), respectively. A half-filled symbol represents a section in which both large and small δ 13 C decreases were reported. A terrestrial section at which the P-TB δ 13 C org decrease is not clearly recognized is shown as a red star. The paleolocation of mid-Panthalassan sections is not precisely specified. Paleoclimate classification is after Schneebeli-Hermann (2012) and Benton and Newell (2014 and South China. They found a Δ 13 C decrease across the extinction horizon in these regions and suggested that it was due to proliferation of green sulfur bacteria with photic zone euxinia (e.g., Grice et al., 2005;Zhang et al., 2017). Kaiho et al. (2009) confirmed the Δ 13 C decrease at Meishan. In contrast, the apparent Δ 13 C increase was reported at several sections in Italy (Siegert et al., 2011), South China this study), and Pakistan (Schneebeli-Hermann et al., 2013). It is difficult to reconstruct the Δ 13 C changes in high latitudes as carbonates are generally scarce in those regions. Nonetheless, the present compilation illustrates that the δ 13 C org records are substantially variable on a global scale particularly in high latitudes, in marked contrast to the rather consistent δ 13 C carb records in equatorial Paleotethys (Figures 6, 7).

Potential Causes of the Marine δ 13 C org Variability
Several possible mechanisms may explain the observed δ 13 C org variability in marine records on a global scale during the Permian-Triassic transition: 1) eustatic sea-level changes, 2) intense continental weathering, 3) proliferation of green sulfur bacteria, 4) elevated pCO 2 , and 5) "methanogenic burst".

Eustatic Sea-Level Changes
The δ 13 C org value of terrestrial plant was generally thought to be higher than the value of marine phytoplankton in the Permian (e.g., Faure et al., 1995;Foster et al., 1997;Korte et al., 2001). Eustatic sealevel changes during the Permian-Triassic transition may have been responsible for the δ 13 C org variability of shallow-marine records, controlling the mixing ratio of 13 C-enriched terrestrial and 13 C-depleted marine C in the shelf sediments. The Permian-Triassic transition is marked by a major transgression on a global scale (e.g., Hallam and Wignall, 1999;Erwin et al., 2002; but also see; Yin et al., 2014). It is therefore likely that shelf sediments of proximal facies shifted to be of more distal facies during the transgression, and that the terrestrial C flux to the depositional setting reduced, which resulted in the apparent δ 13 C org decrease in the bulk sediments. Although the eustatic sea-level changes could have exerted a firstorder control on the global δ 13 C org variability of shallow-marine records, they do not fully explain the regionally variable δ 13 C org records. For example, the δ 13 C org records of terrestrial to marine transitional sections in western South China do not record a simple mixing of terrestrial and marine C, despite of a regional transgression, but rather the atmospheric C isotopic signal (Cui et al., 2017). Moreover, the P-TB δ 13 C org drop and Δ 13 C increase at the present Chaotian cannot be attributed to the sea-level changes, because the water depth of depositional site did not change substantially across the P-TB, as discussed above (Figure 4).

Intense Continental Weathering and Increased Terrestrial C Flux
The Permian-Triassic transition is characterized by extensive vegetation collapse on lands (e.g., Retallack et al., 1996;Benton and Newell, 2014), massive soil erosion (e.g., Retallack, 2005;FIGURE 7 | Frequency histogram of the magnitude of the P-TB δ 13 C decrease in various regions. Frontiers in Earth Science | www.frontiersin.org February 2021 | Volume 8 | Article 596178 Sephton et al., 2005), and intense continental weathering and increased continental flux to the oceans on a global scale (e.g., Algeo and Twitchett, 2010;Algeo et al., 2011;Cao et al., 2019). An increased flux of terrestrial plant C (with relatively high δ 13 C org value) to the shelf sediments via the vegetation collapse and intense weathering may have increased the bulk δ 13 C org values of sediments. However, this process generally decreases the bulk Δ 13 C value of shelf carbonates and cannot explain the P-TB Δ 13 C increase observed in several sections, including Chaotian, as discussed in the former section ( Figure 4). Aged refractory organic matters may also have been derived into the sediments via the enhanced continental weathering and/or soil erosion , though it is difficult to constrain their δ 13 C org values. On the other hand, it is most likely that a terrestrial nutrient flux to the oceans also increased simultaneously via the intense continental weathering, which stimulated eutrophication and algal blooms in the coastal oceans (e.g., Xie et al., 2007;Algeo et al., 2011;Kaiho et al., 2016). The algal blooms may have resulted in an increased flux of marine C (with relatively low δ 13 C org value) to the shelf sediments, offsetting the influence of increased terrestrial C flux on the bulk δ 13 C org value of the sediments. Because of this offset effect, the net influence of the intense continental weathering on the marine δ 13 C org records is highly uncertain.

Proliferation of Green Sulfur Bacteria With Photic Zone Euxinia
Characteristic green sulfur bacteria (GSB) may have proliferated under photic-zone euxinic conditions during the Permian-Triassic transition (e.g., Grice et al., 2005). They generally fix C via the reverse tricarboxylic acid (TCA) cycle, which can produce smaller C isotopic fractionation (∼12. 5‰;van Breugel et al., 2005), than the Calvin cycle (variable but mostly 25-35‰) (e.g., Schidlowski et al., 1983). Thus, the proliferation of GSB and an increased contribution of GSB biomass to the bulk organic-C pool in the shelf sediments would have decreased the Δ 13 C value of the sediments. Riccardi et al. (2007) reported the P-TB Δ 13 C decrease in various sections in the peri-Tethyan realm and attributed it to the GSB proliferation. Nonetheless, the proliferation of GSB cannot explain the apparent Δ 13 C increase observed in several sections, including Chaotian ( Figure 4).

Enlarged C Isotopic Fractionation During Photosynthesis Under the Elevated pCO 2
The increased [CO 2 (aq)] and ε p in the surface oceans is another potential mechanism for the observed δ 13 C carb -δ 13 C org decoupling (e.g., Rau et al., 1992, Rau et al., 1997. However, as discussed above for the Chaotian record, the influence of ε p change on the P-TB δ 13 C records may not have been significant.

Methanogenic Burst
In addition to the several potential mechanisms discussed above, we emphasize here that the "methanogenic burst" may also have contributed to the variable δ 13 C org records on a global scale (Figures 6, 7). A substantial amount of Ni was presumably released into the atmosphere during the Siberian Traps volcanism (Le Vaillant et al., 2017;Rampino et al., 2017).
Methanogen generally fixes C via the reductive acetyl-CoA pathway, which can produce larger C isotopic fractionation up to 40‰ (e.g., Preuss et al., 1989), compared to the Calvin cycle (e.g., Schidlowski et al., 1983). Thus, according to the "methanogenic burst", the increased organic-C flux from expanded methanogen biomass to the bulk organic-C pool in the local sediments may have caused the large δ 13 C org decrease (Figures 6, 7). However, the activity of methanogen is generally regulated, not only by the Ni availability, but also by a number of environmental factors, such as temperature, CO 2 levels, and availability of organic substrates (Supplementary Information; e.g., Singh et al., 2010;Nazaries et al., 2013). Although the excess Ni input during the Siberian Traps volcanism likely promoted methanogenesis, the variable activity of methanogen in the local sediments may have been responsible for the observed δ 13 C org variability on a global scale. The elevated temperature and pCO 2 may also have stimulated the "methanogenic burst", because these factors generally increase the biogenic CH 4 emissions in various environments (Supplementary Information; e.g., van Groenigen et al., 2011;Yvon-Durocher et al., 2014;Aben et al., 2017). The "methanogenic burst" may have occurred not only in marine sediments but also in terrestrial wetlands ( Figure 8). Nonetheless, it is still difficult to estimate the total amount of released Ni during the Siberian Traps volcanism (Le Vaillant et al., 2017), and to evaluate the influence of the "methanogenic burst" on the global δ 13 C org records quantitatively.
In summary, the P-TB δ 13 C org variability in marine records on a global scale may have been attributed to several potential mechanisms, including the eustatic sea-level changes, intense continental weathering, and GSB proliferation. We infer that the "methanogenic burst" was also involved, at least in part, in the substantial δ 13 C org decrease in several sections, including the present Chaotian (Figure 4).

Implications for the Global CH 4 Cycle in the Aftermath of the Extinction
A substantial amount of CH 4 was presumably released into the atmosphere during the Siberian Traps volcanism ( Figure 8B), via volcanic intrusion into coal (e.g., Retallack and Krull, 2006;Retallack and Jahren, 2008;Grasby et al., 2011;Shen et al., 2012;Rampino et al., 2017;Elkins-Tanton et al., 2020), and via destabilization of submarine and permafrost clathrates (e.g., Krull et al., 2000;Krull et al., 2004). Pyrogenic CH 4 was also produced by the incomplete combustion of organic carbon (e.g., Kirschke et al., 2013) and emitted during extensive wildfire events around the P-TB both in the northern and southern hemispheres (e.g., Shen et al., 2011b;Hudspith et al., 2014;Vajda et al., 2020). The "methanogenic burst" likely contributed to the elevated pCH 4 (Rothman et al., 2014). The claimed Araguainha impact event in Brazil (Tohver et al., 2013) may also have contributed to the CH 4 accumulation in the atmosphere, though its timing, magnitude, and effective time span for the global CH 4 cycle are not well-constrained. Together with other greenhouse gases like CO 2 , the elevated pCH 4 may have contributed to the climate warming during the earliest Triassic (e.g., Hallam and Wignall, 1997;Joachimski et al., 2012;Sun et al., 2012;Cui and Kump, 2015), although the long-term warming may have been disturbed intermittently by short-term SO 2 -induced cooling (Black et al., 2018).
Although there still remains large uncertainty, we infer fluctuations in the global CH 4 cycle in the aftermath of the extinction based on the present and previous observations ( Figure 8B). Firstly, aerobic methanotrophy may have prevailed in aerated terrestrial soils in the Gondwana and peri-Gondwanan realms ( Figure 8B; Krull and Retallack, 2000). The present compilation illustrates that the P-TB δ 13 C org decrease is not clearly recognized in a number of terrestrial sections, especially around the Neotethys and in Gondwana realms ( Figure 6; Table 2). The highly scattered and variable δ 13 C org records in the terrestrial successions were likely due to local-scale and short-term organic C dynamics in soils, including vegetation, selective microbial decomposition of sedimentary organic matters with C isotopic fractionation, and addition of microbial biomass to the sedimentary C pool (e.g., Krull and Retallack, 2000;Korte and Kozur, 2010). The locally enhanced methanotrophy might have been involved, at least in part, in the soil C dynamics and in the scattered δ 13 C org records in the Gondwana and peri-Gondwanan realms (Krull and Retallack, 2000), possibly according to the warming and permafrost thaw (Supplementary Information; e.g., Oh et al., 2020).
Secondly, the oceanic sediments may have been a significantly large source for atmospheric CH 4 at that time. In the modern oceans enriched in sulfate (the SO 4 2concentration 28 mM), almost all CH 4 , produced in deeper sediments, is consumed by anaerobic oxidation of methane (AOM) in the sulfate-methane transition zone (SMTZ) (Figure 8; Supplementary Information; e.g., Reeburgh, 2007). However, AOM would be substantially suppressed when the sulfate concentration is <0.5 mM (Knittel and Boetius, 2009). The Permian-Triassic transition interval is characterized by the substantially low FIGURE 8 | Schematic diagram of the global CH 4 cycle in the modern world (A) and during the Permian-Triassic transition (B). CH 4 influx to and efflux from the atmosphere are shown as red and yellow arrows, respectively. In the modern system, the global atmospheric CH 4 budget is 0.4-0.5 Gt C/year (Supplementary Information; e.g., Conrad, 2009;Nazaries et al., 2013). During the Permian-Triassic transition, several processes may have contributed to the elevated pCH 4 and climate warming, although terrestrial responses to the elevated pCH 4 were probably complex. The elevated pCH 4 enhanced aerobic methanotrophy in local soils whereas the intense soil erosion decreased a net CH 4 efflux in the global budget. AOM: anaerobic oxidation of methane; SMTZ: sulfate-methane transition zone. See Supplementary Information  sulfate concentration in the oceans (0.6-2.8 mM) (Luo et al., 2010;Schobben et al., 2017;Stebbins et al., 2019). This estimated SO 4 2range is slightly higher than the threshold of the AOM rate. Nonetheless, the porewater sulfate concentration in the sediments may have quickly become <0.5 mM at a very shallow depth, due to sulfate consumption via decomposition of other organic substrates. The sedimentary AOM was consequently suppressed and the oceanic sediments might have been a larger CH 4 source compared to in the modern oceans, further contributing to the elevated pCH 4 . The impact of "methanogenic burst" on the enhanced CH 4 emissions should have been significant under such sulfate-depleted conditions with less AOM. Prevailing oceanic anoxia (e.g., Wignall and Hallam, 1992;Isozaki, 1997;Song et al., 2012b) also helped CH 4 to escape from the oceans to the atmosphere (e.g., Ryskin, 2003).
Finally, the global CH 4 budgets may have been disturbed by the terrestrial devastation and intense continental weathering (e.g., Algeo and Twitchett, 2010;Cao et al., 2019). The Permian-Triassic transition is characterized by the extensive vegetation collapse on lands and massive soil erosion on a global scale ( Figure 8B; e.g., Retallack, 2005;Sephton et al., 2005;Benton and Newell, 2014). The destruction of terrestrial ecosystems and the decay of land plants in the aftermath of the extinction was claimed to bring the well-known Early Triassic "coal gap" (Retallack et al., 1996). Although the release of substantial amounts of Ni into the ocean-atmosphere during the Siberian Traps volcanism may have caused the "methanogenic burst", it may also have contributed to the vegetation collapse because excess Ni is generally toxic to plants (Fielding et al., 2019). The vegetation collapse could have stimulated the destabilization of permafrost and the CH 4 emissions in high latitudes according to the warming (Nauta et al., 2015), whereas the massive soil erosion might also have contributed to the atmospheric CH 4 accumulation as aerated terrestrial soils are a CH 4 sink ( Supplementary Information; Figure 8B).

CONCLUSIONS
The carbon isotopic composition of carbonate (δ 13 C carb ) across the Permian-Triassic boundary (P-TB) was analyzed at Chaotian, Sichuan, South China, and was correlated to the chemostratigraphy of the carbon isotopic composition of organic carbon (δ 13 C org ) of the same interval. The δ 13 C carb and δ 13 C org records at Chaotian were further integrated into the records from various marine and terrestrial environments all around the world, to examine fluctuations in the global methane (CH 4 ) cycle during the Permian-Triassic transition. The following results were obtained: (1) The δ 13 C carb values decrease from ca. +1 to −2‰ across the P-TB, possibly reflecting the shallow-marine extinction and the collapse of primary productivity in the oceans. The frequent intercalation of felsic tuff layers around the extinction horizon suggests that volcanic activity also contributed to the δ 13 C carb decrease. (2) The magnitude of the δ 13 C carb decline (∼3‰) is substantially smaller than the magnitude of the δ 13 C org decrease (∼7‰) across the P-TB. This δ 13 C carb -δ 13 C org decoupling could be explained by an elevated CO 2 concentration in the ocean/ atmosphere and/or proliferation of methanogen ("methanogenic burst") in the sediments, according to the Siberian Traps volcanism. (3) A global P-TB δ 13 C compilation shows a large variation in marine δ 13 C org records, which could be attributed to several potential mechanisms including eustatic sea-level changes and proliferation of green sulfur bacteria. We infer that the "methanogenic burst" may also have contributed, at least in part, to the δ 13 C org variability. The global CH 4 cycle might have fluctuated substantially in the aftermath of the extinction.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

AUTHOR CONTRIBUTIONS
MS designed the study. YI provided the samples for the analyses. MS and YI conducted the lithofacies description. MS conducted the isotopic analyses. MS and YI wrote the manuscript.