From the samples 34 spore, 32 pollen, and 8 algae genera have been identified (Supplementary Table 1, Figure 2, and Supplementary Figures 1, 2) that are typical of Late Triassic and Early Jurassic palynological assemblages (see section “The Stratigraphic Position of the TJB in the Jiyuan Basin”). They are assigned to six palynological assemblage zones (AZ) based on the palynomorph abundance variations and CONISS: Psophosphaera—Chasmatosporites—Cycadopites assemblage zone (AZ–I; samples #JY 23—#JY 20), the Punctatisporites—Cyathidites—Psophosphaera assemblage zone (AZ–II, #JY 19—#JY 18), the Punctatisporites—Verrucosisporites—Triquitrites assemblage zone (AZ–III, #JY 17—#JY 14), the Classopollis—Cyathidites—Cycadopites assemblage zone (AZ–IV, #JY 13—#JY 12), the Cyathidites—Classopollis—Pseudopicea assemblage zone (AZ–V, #JY 11—#JY 7), and the Pinuspollenites—Cycadopites—Classopollis assemblage zone (AZ–VI, #JY 6—#JY 1) (Figure 2 and Supplementary Figure 1). The composition of AZ-I is dominated by gymnosperm pollen (mean (x̄) = 56.5%, including Psophosphaera and Chasmstosporites), followed by fern spores (x̄ = 43.5%, including Punctatisporites and Crassispora). The compositions of AZ-II and AZ-III are noticeably different, although fern spores (including Cyathidites and Punctatisporites) dominate followed by gymnosperm pollen (including Psophosphaera and Cycadopites), but in AZ-II algae dominate (x̄ = 57.6%) and the proportion of spores and pollens decreases by ca. 45% from 38 to 21 genera (Figure 2). Nine of the fern spore genera (dominated by Verrucosisporites and Laevigatosporites) and eight gymnosperm pollen genera (mainly Rotundipollis, Podocarpidites and Callialasporites) decreased in abundance (Supplementary Table 1 and Figure 2). The compositions of AZ-IV, AZ-V, and AZ-VI are generally similar and are dominated by gymnosperm pollen (x̄ = 72.0, 71.7, and 72.8%, respectively) including Classopollis, Cycadopites, and Pseudopicea. AZ-IV, AZ-V, and AZ-VI have low proportions of ferns (x̄ = 28.0, 28.3, and 27.2%, respectively) including Cyathidites and Osmundacidtes, while algal composition is uniformly low (x̄ = 7.1, 3.4, and 1.6%, respectively). In AZ-IV, the proportion of spores and pollens decreases by ca. 44% and from 50 to 28 genera. In this assemblage 14 fern spore genera (mainly Verrucosisporites, Punctatisporites, Triquitrites, and Triporoletes) and eight gymnosperm pollen genera (mainly Paleoconiferus, Protopinus, and Callialasporites) each decreased in abundance (Supplementary Table 1 and Figure 2). In addition, AZ-IV (sample #JY 13) sees a decrease in diversity but a rapid increase in the abundance of Classopollis and Cythidites, and includes the first appearance of Cerbropollenites (Figure 2).

The parent plants of the palynological taxa have been evaluated to reconstruct the floral successions of the Jiyuan Basin (Supplementary Tables 1, 2 and Figure 2). Diverse plant groups are recognized in the study area from the Late Triassic to Early Jurassic including mosses (x̄ = 0.6%), lycopsids (x̄ = 3.2%), horsetails (x̄ = 0.7%), “filicalean” ferns (x̄ = 34.0%), conifers (x̄ = 39.9%), cycads (x̄ = 14.7%), and pteridosperms (seed ferns; x̄ = 0.4%) (Supplementary Tables 1, 2 and Figure 2). These taxa mainly grow in temperate to subtropical, warm and humid climates (Liu et al., 2015; Li et al., 2020; Lu et al., 2021a). In AZ-I, AZ-II, and AZ-III, ferns are dominant (x̄ = 34.8, 52.9, and 52.1%, including the Osmundaceae, Dicksoniaceae/Cyatheaceae, and Dipteridaceae/Matoniaceae), followed by conifers (x̄ = 31.4, 28.8, and 24.1%, including Taxodiaceae and Podocarpaceae) and cycads. Horsetails and lycopsids are less abundant, while pteridosperms appear sporadically (Supplementary Tables 1, 2, Figure 2, and Supplementary Figure 1). In AZ-IV, AZ-V, and AZ-VI, conifers are dominant (x̄ = 47.3, 49.4, and 47.3%, including Cheirolepidiaceae, Taxodiaceae, and Pinaceae), followed by ferns (x̄ = 28.0, 25.9, and 23.8%, including Dicksoniaceae/Cyatheaceae and Osmundaceae), while cycads, horsetails, lycopsids, and seed ferns appear sporadically and mosses are absent (Supplementary Tables 1, 2, Figure 2, and Supplementary Figure 1).

Results of PCA analysis based on the relative abundance of spore-pollen genera are shown in Figures 2–4. Two main ordination axes representing the largest variance in palynological composition are used to explain the two most dominant environmental gradients that control the dataset (e.g., Bonis et al., 2010; Li et al., 2020). In this study, axis 1 and axis 2 separately account for 42.28 and 16.75% of the sporomorph spectra difference (Figure 3). On the positive side of axis 1, the xerophytic Classopollis, Pinuspollenites, and Quadraeculina pollen and hygro-mesophytic Cyathidites spores (e.g., Lindström, 2016; Li et al., 2020) have high scores, and the hygrophytic Punctatisporites spores (e.g., Lindström, 2016; Li et al., 2020) show high scores on the negative side of axis 1. Relatively cool, temperate conifer pollen including Pinuspollenites and Pseudopicea have high scores on the negative side of axis 2, and the relatively warm condition spores Cyathidites and Punctatosporites (e.g., Li et al., 2020) and the pollen Classopollis (e.g., Lindström, 2016) have high scores on the positive side of axis 2. Thus, the first and the second axes are interpreted to reflect the changes in relative humidity and temperature, respectively (Figures 2, 3). The PCA results indicate stable cool-humid conditions prevailed during AZ-I, and cool-dry paleoclimate during AZ-V and AZ-VI, while mixed or fluctuating climatic conditions prevailed from AZ-II to AZ-IV, especially during two warming intervals in AZ-II and AZ-IV (Figures 2, 3).

Results for TOC and δ¹³C_org are shown in Figure 2 and Supplementary Table 3. TOC values vary from 0.24 to 1.01 wt% (x̄ = 0.46 wt%). At the base of the succession TOC values are low and have a gradually decreasing trend in the mudstones and yellow siltstones (samples 50–43; x̄ = 0.39 wt%), followed by an interval of fluctuating values and including two TOC peaks of 1.01 and 0.85 wt% in the latest Triassic-earliest Jurassic mudstones and gray siltstones (samples 42–25; x̄ = 0.56 wt%). TOC values show a generally increasing trend through the Jurassic mudstones, siltstones, and sandstone to the top of our succession (samples 24–1; x̄ = 0.44 wt%) (Figure 2). From samples through the complete succession, peak temperature of rock pyrolysis (T_max) values vary from 442 to 446°C (x̄ = 444.1°C) (Supplementary Table 3) and indicate that kerogen maturity varies from low-maturity to mature as defined by the China National Standard (SY/T 5477-2003) (Supplementary Table 5).

δ¹³C_org values vary from −27.5 to −20.8‰ (x̄ = −22.8‰; Figure 2) and stratigraphically have two heavier plateaus either side of a phase of lighter values around the T-J transition (Figure 2). In the earlier plateau (samples 50–43, roughly corresponding to AZ-I), δ¹³C_org values are stable and average −22.9‰. In the phase of lighter values between the two plateaus (samples 42–25, roughly corresponding to AZ-II to AZ-IV) values vary from −29.0 to −22.5‰ (x̄ = −24.3‰) and this part includes two negative δ¹³C_org excursions (CIEs) with magnitudes of −4.7‰ (CIE-I) and −2.9‰ (CIE-II) relative to the background mean. These CIEs are near-synchronous with two falls in plant diversity in AZ-II and AZ-IV (Figure 2). Above CIE-II (samples 24–1, corresponding to AZ-V to AZ-VI), δ¹³C_org values return to a heavier and stable plateau and average −22.0‰, including a slight decrease between samples 24 and 10 (Figure 2).

Kerogen macerals mainly comprise exinite (x̄ = 29.1%), sapropelinite (x̄ = 28.5%) and vitrinite (x̄ = 26.4%), followed by fusinite (x̄ = 16.1%) (Supplementary Table 4 and Figure 2). Exinite content varies from 19.2 to 39.3% (x̄ = 29.1%) and mainly comprises sporopollenite, cutinite, and subertinite (Figures 2, 5). Sapropelinite content varies from 21.9 to 38.9% (x̄ = 28.5%) (Figures 2, 5). Vitrinite content varies from 12.4 to 37.8% (x̄ = 26.4%) and comprises non-fluorescent telinite and collinite (Figures 2, 5). Fusinite content varies from 6.7 to 39.6% (x̄ = 16.1%) and is entirely fragmental fusinite which is opaque, pure black, does not fluoresce under fluorescence illumination, and is mostly long and thin or fragmental in shape with sharp edges (Figures 2, 5). Fusinite enrichment in the studied strata (Figure 2) corresponds to the emergence of a large number of charcoal particles in sediments (Figure 5), with fusinite representing the product of incomplete combustion from wildfire (e.g., Goodarzi, 1985; Bustin and Guo, 1999; Lu et al., 2020a). We therefore use fusinite content as a proxy for wildfire. In this study, two obvious fusinite peaks with the values of 37.0 and 39.6% in the uppermost Triassic and earliest Jurassic strata (samples 38 and 30, respectively) correspond to the falls in plant diversity in AZ-II and AZ-IV (Figure 2).

Results for Hg concentrations are shown in Figure 2 and Supplementary Table 3. Hg concentrations show considerable variation, ranging from 2.23 to 183.42 ppb (x̄ = 34.33 ppb) and have a distribution that is broadly anticorrelated to δ¹³C_org values (Figure 2). The Hg concentrations show stronger covariation with TOC (r = + 0.67) than with Al (r = + 0.17) or TS (r = −0.03) (Supplementary Figure 3), suggesting Hg is mostly hosted by organic matter (OM) (e.g., Shen et al., 2020; Lu et al., 2021a,b,c). As a result, we present Hg/TOC values as a proxy for volcanism. Hg/TOC varies from 5.19 to 234.72 ppb/wt% (x̄ = 66.16 ppb/wt%) with distribution broadly similar to the raw Hg concentrations (Figure 2). There are two Hg/TOC peaks with values of 234.72 ppb/wt% in AZ-II (from which we infer pulses of volcanic activity = VA-I) and 144.93 ppb/wt% in AZ-IV (VA-II; Figure 2). The intervals of Hg/TOC enrichment roughly correspond to the position of the CIEs, enhanced wildfire, and decreases in plant diversity (Figure 2).

The TJB has been tentatively placed in the upper part of the Anyao Formation in the Jiyuan Basin (see Lu et al., 2021a and section “Geological Setting”). Palynological composition can be effective in determining stratigraphic age (e.g., Lu and Deng, 2005; Sha et al., 2015; Lindström et al., 2017; Li et al., 2020; Wignall and Atkinson, 2020). In western Tethys, floral changes near the TJB are characterized by: (1) an end-Triassic fern peak (including Polypodiisporites polymicroforatus, and trilete megaspores in some areas); (2) the earliest Jurassic first appearance (FO) of Cerbropollenites thiergartii and the last occurrence (LO) of Lunatisporites rhaeticus; and (3) a brief earliest Jurassic proliferation of Classopollis pollen (Cheirolepidiaceae) and the subsequent restoration and dominance of conifers (e.g., Olsen et al., 2002; Whiteside et al., 2007; Bonis et al., 2009, 2010; van de Schootbrugge et al., 2009; Pieńkowski et al., 2012; Vajda et al., 2013; Lindström et al., 2017; Wignall and Atkinson, 2020; Boomer et al., 2021). In eastern Tethys, a fern spore spike occurs in the Haojiaggou section from NW China during the T-J transition and is followed by the FO of Cerbropollenites thiergartii, the LO of Lunatisporites rhaeticus, and the co-dominance of Classopollis and Cythidites in Lower Jurassic strata (Lu and Deng, 2005; Deng et al., 2010; Sha et al., 2015; Fang et al., 2021); a fern spore spike also occurs in the Qilixia section of South China during T-J transition, followed by the co-dominance of Classopollis and Cythidites in the overlying Lower Jurassic strata (Wang et al., 2010; Li et al., 2020).

The combined records from the Haojiagou and Qilixia sections allow the development of an integrated stratigraphic TJB framework for the terrestrial eastern Tethys region (e.g., Lu and Deng, 2005; Li et al., 2020; Fang et al., 2021). In this study, similar changes in palynological assemblages allows us to correlate these two sections (see Supplementary Material). Significantly, the fern spore spike occurs in AZ-II and AZ-III in our study, but it lacks some of the usual fern elements including trilete type spores and Polypodiisporites polymorphoratus, which are key palynostratigraphic markers for the TJB in western Tethys (e.g., Lindström et al., 2017; Wignall and Atkinson, 2020). Similar patterns are also seen in the Haojiagou and Qilixia sections, which have been interpreted as the fern peak associated with global records from the T-J transition interval (see Supplementary Material). In our study, the spore and pollen composition of AZ-IV is very similar to that from the Badaowan Formation of NW China and the Zhenzhuchong Foamation of South China, with the co-dominance of Cyathidites (Dicksoniaceae/Cyatheaceae) and Classopollis (Cheirolepidiaceae), the common occurrence of Cycadopites, and the presence of Dictyophyllidites, Chasmatosporites, and Quadraeculina, which indicate an Early Jurassic age (e.g., Lu and Deng, 2005; Li et al., 2020). This is further supported by the disappearance at the base of AZ-IV of the Triassic taxa Aratrisporites, Taeniaesporites, and Kraeuselisporites (e.g., Fu and Yuan, 1998; Li et al., 2020). Furthermore, AZ-IV in the study area records the rapid proliferation of Classopollis pollen, and the subsequent restoration and dominance of conifers at the top of the assemblage zone (Supplementary Table 1, Figure 2, and Supplementary Figure 1). Similar records are known from the Early Jurassic of western Tethys (see above; e.g., Wignall and Atkinson, 2020). Collectively, these observations are consistent with the placement of the TJB at the base of AZ-IV (Figure 2 and Supplementary Figure 1). As a result, we consider that the fern spike recorded in the study area is representative of the global TJB record.

The T-J transition, beginning with the onset of the ETME, is one of the most ecologically significant intervals of the past 200 Ma and is characterized by dramatic global C cycle perturbations together with major changes in climate and ecosystem composition on land and in the oceans. Across the T-J transition, in particular where fossils or other diagnostic age markers are absent, chemostratigraphy and in particular C isotope stratigraphy can be used to correlate strata deposited in a diverse range of depositional environments (e.g., Korte et al., 2019; Pieńkowski et al., 2020; Ruhl et al., 2020; Fang et al., 2021).

In our study, Rock-Eval pyrolysis indicates that OM Rock-Eval pyrolysis data suggests that diagenetic processes are unlikely to be responsible for the observed shifts in the δ¹³C signature of the OM, because such changes occur in late diagenetic to metamorphic burial stages that our material has not been subjected to (Dal Corso et al., 2018; Lu et al., 2021a). The two CIEs in the studied strata are also unlikely to have occurred through proportionate changes in terrestrial and algal organic carbon sources, which can be isotopically distinct from one another (Cloern et al., 2002). Variations in kerogen macerals reveal that the OM in the studied strata was from mixed sources of terrestrial plants and lacustrine plankton (see Supplementary Material). However, the proportion of OM sources did not change significantly during the two CIEs. Similarly, relatively high C/N ratios (see Supplementary Material) suggest that the OM throughout the succession is predominantly terrestrial with lower C/N ratios supporting a proportional increase in algal production during AZ-II and AZ-III (the shift to lower C/N ratios occurs between samples 42 and 41) (Cloern et al., 2002). Crucially no apparent relationship exists between C/N and δ¹³C values across the CIEs. Furthermore, the δ¹³C_org is largely limited by the pathway of plant photosynthesis and varies with plant types (e.g., Collister et al., 1994; Diefendorf and Freimuth, 2017). The two CIEs in the studied strata are synchronous with changes in fern spore and gymnosperm pollen abundances and the decreased plant diversity, indicating that whatever drove the CIEs might have affected the plant composition (e.g., van de Schootbrugge et al., 2008). However, changes in plant types is unlikely to be the driver of the observed CIEs in this study because those CIEs are remarkably similar to those in T-J transition δ¹³C records in carbonates and in bulk marine and terrestrial OM (e.g., Bacon et al., 2011; Figure 6 and Supplementary Figure 4). Compilation of global carbon isotope data has revealed a pattern of three isotopic excursions in both marine and continental strata. These comprise an initial excursion, then a small (unnamed) secondary excursion after a positive plateau, and finally a globally widespread main excursion (e.g., Hesselbo et al., 2002; Ruhl et al., 2009, 2010, 2020; Kovács et al., 2020; Wignall and Atkinson, 2020; Fang et al., 2021; Shen et al., 2022; Figure 6). In this study, the C isotope record exhibits similar features with two negative excursions comprising an initial CIE and an unnamed small CIE (Figure 6), that permit correlation with other TJB sections globally. The amplitude of the two CIEs in this study is consistent with those recorded in marine (–4 to –6.5‰) and continental (–2 to –4‰) strata in the Western Tethys Ocean in the middle and low latitudes of the northern hemisphere (Figure 6 and Supplementary Figure 4). Therefore, we consider that CIE-I and CIE-II in this study reflect the changes to the global carbon cycle, namely the massive input carbon from an isotopically depleted carbon source.

In our study, CIE-I (−4.7‰) is characterized by a 45% reduction in plant diversity, the proliferation of fern spores, and an Hg/TOC peak (234.72 ppb/wt%, VA-I) during a phase of climatic warming (Figures 2, 6). Similar features also characterize the former candidate global stratotype section and point (GSSP) at St. Audrie’s section in the United Kingdom and the GSSP at Kuhjoch in Austria (Figures 1A, 6). Here we use the candidate GSSP St. Audrie’s section and the GSSP Kuhjoch section records as reference points (Figures 1A, 6) since almost all other global T-J sections have been correlated with these (e.g., Ruhl et al., 2009, 2010, 2020; Lindström et al., 2017; Percival et al., 2017; Zaffani et al., 2018; Kovács et al., 2020; Pieńkowski et al., 2020; Wignall and Atkinson, 2020; Fang et al., 2021). However, numerous studies provide a range of interpretations in stratigraphic correlation, although in general the global C isotope curves are similar (Hesselbo et al., 2002; Ruhl et al., 2009, 2010, 2020; Lindström et al., 2017; Percival et al., 2017; Zaffani et al., 2018; Kovács et al., 2020; Wignall and Atkinson, 2020; Fang et al., 2021). The interpretation of Wignall and Atkinson (2020) is used here as it is the most representative and up to date. In their scheme, the initial CIE corresponds to the first extinction pulse that includes plant extinctions (see below) and Hg enrichment. A secondary unnamed CIE corresponds to a secondary extinction pulse that saw further losses within the plant record and Hg enrichment (e.g., Percival et al., 2017; Wignall and Atkinson, 2020). This indicates that the mass extinction history in Europe is pulsed and synchronous with C isotope excursions, facilitating global correlation of these phenomena (Wignall and Atkinson, 2020). Furthermore, the fern spore peaks in our study coincide with the initial CIE and the ETME. This observation has been reported in numerous sections (see references above) and further supports the placement of the onset of the ETME in the study area around CIE-I.

The younger CIE-II in the Jiyuan Basin succession is also widely known from marine and terrestrial records as the unnamed minor CIE in other areas (Hesselbo et al., 2002; Ruhl et al., 2009, 2010, 2020; Percival et al., 2017; Pieńkowski et al., 2020; Wignall and Atkinson, 2020; Fang et al., 2021; Figure 6). Like CIE-I, CIE-II is accompanied by a significant reduction in plant diversity in which 44% of genera are lost, and similar losses are known from other terrestrial records near the TJB (e.g., Wignall and Atkinson, 2020; Lindström, 2021). CIE-II is also accompanied by a minor Hg/TOC peak (144.93 ppb/wt%, VA-II). Warmer and drier climates persisted at this level in the study area indicated by the occurrence of calcareous nodules that formed through groundwater evaporation (e.g., Lu et al., 2020a and references therein) (Figures 2, 6). Therefore, we consider it reasonable to place the TJB during the CIE-II interval based on (i) changes in biotic composition as reflected in biostratigraphy and palynofloral assemblage composition, and (ii) C isotope stratigraphy. This is supported by the ammonite-defined TJB at the GSSP at Kuhjoch, Austria (Ruhl et al., 2009) and other well-studied sections (e.g., Hesselbo et al., 2002; Percival et al., 2017; Pieńkowski et al., 2020; Ruhl et al., 2020; Wignall and Atkinson, 2020; Boomer et al., 2021; Fang et al., 2021; Shen et al., 2022) in which the TJB is below the main CIE and corresponds with a unnamed small negative CIE above the initial CIE (Figure 6).

Carbon cycle perturbations in the Jiyuan succession appear to have ended in the latter part of AZ-IV and the AZ-V interval is characterized by relatively stable C isotopic composition. This interval also saw stability return to climates and floral compositions (Figures 2, 6), marking the end of the perturbations of the T-J transition. This can be correlated with records such as the Qilixia section in South China and the Haojiagou section in NW China, suggesting that the T-J transition approximately corresponds to the end of the secondary minor CIE, after the initial CIE (Fang et al., 2021; Shen et al., 2022).

The timing and causal mechanism(s) of the ETME are controversial (e.g., van de Schootbrugge et al., 2009; Barbacka et al., 2017; Lindström et al., 2019; Wignall and Atkinson, 2020; Lindström, 2021; Lucas, 2021). Wignall and Atkinson (2020) document two pulses of extinction in marine and terrestrial settings, with pulses occurring during the initial and the minor (unnamed) negative CIEs during the T-J transition. On land, the first pulse of extinction is manifest as a decline in conifer pollen abundance and an overall plant diversity decline. This was followed by an interlude during which terrestrial assemblages were dominated by ferns, whose spores dominate the record from that interval. The second pulse of extinction caused another abrupt, palynological change, resulting in the loss of further pollen taxa and a brief proliferation of Classopollis pollen (Wignall and Atkinson, 2020). Our study also identifies a bimodal reduction in plant diversity, suggesting that the two-step extinction pattern is a globally widespread phenomenon (e.g., Wignall and Atkinson, 2020; Lindström, 2021). A rapid recovery of fern spores and their floral dominance (reaching 68.2% of all palynomorphs) (Figure 2) occurred in the interlude between ETME-I and ETME-II in the Jiyuan Basin. A similar pattern is known from both the southern and northern hemisphere including the Western Tethys ocean and the southern margins of Pangea (van de Schootbrugge et al., 2009; Lindström, 2016; Wignall and Atkinson, 2020).

The two pulses of plant diversity loss in the Jiyuan Basin (ETME-I and ETME-II) occurred during warming intervals indicated by the PCA analysis (Figures 2, 6). Warming during the main CIE is supported by oxygen isotope data from oysters at Lavernock Point (United Kingdom) in the northwestern Tethys (Korte et al., 2009) and pCO₂ increases are recorded at both CIE levels are documented elsewhere in the record of pedogenic carbonates from the Newark Basin (Schaller et al., 2011) and the stomatal index (plant fossil cuticles) in Greenland and Sweden (McElwain et al., 1999; Slodownik et al., 2021), Germany (Bonis et al., 2010), and Northern Ireland (Steinthorsdottir et al., 2011; Figure 6).

Our data suggests that the two pulses of plant diversity collapse in the Jiyuan Basin are associated with significantly different atmospheric humidities. ETME-I and the subsequent recovery occurred during humid conditions indicated by the proliferation of hygrophytic floral elements (including all spores, Alisporites, and Cycadopites) (Figure 2 and Supplementary Figure 1). Similar records have been observed in Western Tethys (e.g., Hungary, Britain, Denmark, Greenland, and Canada) and southern Pangea (e.g., New Zealand, Western Australia, and Eastern Australia) and this has been interpreted as a result of increased humidity (van de Schootbrugge et al., 2009; Lindström, 2016). In contrast, ETME-II is associated with the development of an arid paleoclimate with high evaporation, as evidenced by decreased fern spore content, increasing Cheirolepidiaceae conifer content (Classopollis pollen), and the widespread occurrence of calcareous nodules in the core and in terrestrial outcrops (Figures 2, 7 and Supplementary Figure 1). Similar declines in fern spores and increases in gymnosperm pollen (e.g., Classopollis pollen) have been described for South China where they were interpreted as marking a shift toward warmer and drier conditions during the earliest Jurassic (Srivastava, 1976; Huang, 2001; Wang et al., 2010; Zhou et al., 2021).

In the Jiyuan Basin, ETME-I and ETME-II are also accompanied by enhanced paleo-wildfire indicated by the enrichment of fusinite (charcoal), as well as the appearance of a large number of charcoal particles in sediments (Figures 3, 7). Enhanced wildfires during the T-J transition appears to be a global phenomenon (Supplementary Figure 4) as similar observations are known for other regions including North America (Jones et al., 2002), East Greenland (Belcher et al., 2010), Poland (Marynowski and Simoneit, 2009), Denmark and Sweden (Petersen and Lindström, 2012), and South China (Song et al., 2020).

The emplacement of the CAMP is widely implicated as the ultimate driver of the ETME (Hesselbo et al., 2002; Schaltegger et al., 2008; Clémence et al., 2010; Bartolini et al., 2012; Percival et al., 2017; Lindström et al., 2019; Heimdal et al., 2020; Ruhl et al., 2020; Capriolo et al., 2021a,b; Shen et al., 2022). We have identified two Hg and Hg/TOC peaks in AZ-II and AZ-IV that are significantly higher than background levels seen in the other Assemblage Zones (Figure 2 and Supplementary Figure 5). The behavior of Hg is complicated by depositional environment (Yager et al., 2021) and in terrestrial lake settings Hg can be sourced from the atmosphere (i.e., that associated with LIP volcanism) as well as from hydrological runoff (e.g., Shen et al., 2020; Lu et al., 2021a,b). The Hg preserved in the core material in this study is associated with OM derived from a mixture of terrestrial higher plants and lacustrine plankton in AZ-II and AZ-IV. Minor increases in TOC pre- and post-ETME are not accompanied by Hg peaks (Figure 2), supporting the notion that where terrestrial OM is enriched in Hg this was due to the effects of contemporaneous large-scale volcanism. Previous studies have shown that two or three Hg or Hg/TOC enrichment anomalies with likely origins in CAMP can be correlated across various sedimentary facies, from marine to terrestrial, during the T-J transition (Thibodeau et al., 2016; Percival et al., 2017; Lindström et al., 2019; Ruhl et al., 2020; Yager et al., 2021; Shen et al., 2022). Although the values of Hg and Hg/TOC enrichment anomalies vary spatially across different studied sections (Figure 6), Hg isotope data further support a CAMP source for elevated Hg contents across the ETME interval (Yager et al., 2021). Furthermore, two Hg enrichment anomalies in our study are synchronous with two pulses in terrestrial plant diversity, carbon cycle, enhanced wildfire and elevated temperature (Figures 2, 6). Therefore, we consider the Hg/TOC ratios in this study as a reliable proxy for CAMP volcanism.

Volcanic activity can release copious amounts of isotopically light CO₂ into the reservoirs of the exogenic carbon cycle and can drive major global warming. As shown here and elsewhere (Figure 6), the close correspondence of Hg or Hg/TOC enrichment anomalies and CIEs during the T-J transition suggests that a large amount of isotopically light CO₂ entered the land atmosphere-ocean system during this interval. CLIMBER-3α + C Earth System model simulations suggest that pulses of CAMP volcanism potentially increased the global temperature by more than 4°C and might have triggered the disruption of the global end-Triassic carbon cycle (Landwehrs et al., 2020). Conservative estimates suggest that the emplacement of CAMP released between ∼1.4 × 10³ Pg and 2.1 × 10⁴ Pg of mantle C (Heimdal et al., 2020 and references therein). However, the isotope composition of volcanogenic CO₂ (mantle carbon ∼−6‰) means that the mass derived from CAMP emplacement is insufficient to drive the large negative CIEs associated with the ETME (e.g., Paris et al., 2016; Lindström et al., 2021). cGENIE model simulations (Vervoort et al., 2019) indicate that a negative CIE of ≥ 3‰ (i.e., of similar magnitudes to those associated with the ETME), lasting ∼10–100 s of kyr, would require > 3 × 10⁴ PgC if the C mantle source had a δ¹³C composition of −6‰. Similarly, GEOCLIM model simulations suggest that a CAMP-related negative CIE up to ∼−6‰ could be achieved by the release of repeated pulses of volcanic CO₂ if its isotopic signature was significantly light (–20‰; Paris et al., 2016). As a result, the ETME CIEs were likely in part driven by the addition of strongly ¹³C-depleted C to the land atmosphere-ocean system from sources other than volcanic eruptions themselves. There is growing consensus that thermogenic C release was a contributing factor for the ETME CIEs (Svensen et al., 2009; van de Schootbrugge et al., 2009; Dal Corso et al., 2014; Heimdal et al., 2018, 2020). In addition, terrestrial or rock-bound organic C oxidation and methane release may have further contributed to the negative carbon isotope excursions (e.g., Ruhl et al., 2020; Capriolo et al., 2021a). In our study, two CIEs were contemporaneous with wildfires (Figures 2, 6), raising the possibility that increased wildfire associated with volcanic activity provided a positive feedback in the global carbon cycle that further increased the magnitude of CIEs by releasing isotopically light carbon through combustion of biomass (e.g., Ivany and Salawitch, 1993; Belcher et al., 2010). Indeed, these hypotheses are not mutually exclusive and the different contributory factors probably acted in tandem to drive two major negative CIEs in the T-J interval. As a result, we favor a scenario in which repeated episodic light ¹³C CO₂ release related to CAMP volcanism and its direct effects was responsible for the CIEs (e.g., Capriolo et al., 2021b), and drove contemporaneous global warming, as suggested by the PCA results.

Though global warming as a result of large-scale volcanism could be expected to have a major impact on plants, it appears that the development of intense wildfires—an indirect function of volcanism—seems to have the most profound impact on terrestrial plant ecosystems (Figure 8 and Supplementary Figure 6). We suggest that the warming climate resulting from pulsed CAMP eruptions increased the frequency of lightning and wildfires (see Supplementary Material), resulting in the deterioration of terrestrial ecosystems (e.g., soil erosion) and associated reductions in plant diversity (Figures 2, 6, 8 and Supplementary Figure 6). Frequent and intense wildfires were not only the direct cause of changes in terrestrial plant ecosystems and species diversity but also represent a vital link between the lake and terrestrial ecological crisis (e.g., Lu et al., 2020a; Mays et al., 2021). Thus, wildfire disruption to land surface vegetation would cause increased soil erosion, itself increasing nutrient runoff, but also further exposing bedrock and increasing continental weathering leading to siltation (Glasspool et al., 2015; Lu et al., 2020a). These processes can lead to large amounts of OM (including charcoal and un-charred material), and other substances entering the lake system through surface runoff (Supplementary Figure 6), promoting eutrophication and blooms of cyanobacteria and algae in freshwater environments, just as shown in the Jiyuan Basin. In this case, oxygen circulation between freshwater and atmosphere would have been inhibited by floating inert organic particles, cyanobacteria and algae in surface waters (e.g., Lu et al., 2020a). Cyanobacterial and algal blooms would likely have contributed to the consumption of dissolved oxygen through decomposition post-mortem. It is possible that they produced secondary metabolites toxic to animals, impeding the recovery of lake freshwater ecosystems (Mays et al., 2021) in a scenario, similar to that proposed for the Carnian Pluvial Episode (Lu et al., 2021a) and the end-Permian extinction (Mays et al., 2021).