Sustained and intensified lacustrine methane cycling during Early Permian climate warming – Nature.com
Age and stratigraphic constraints
Samples were collected from the Lucaogou Formation at the Jingjingzigou section along the southern margin of the Junggar Basin (Fig. 1 and Supplementary Fig. 1) for zircon U-Pb and geochemical analyses. The Lucaogou Formation can be subdivided into two members. The lower member mainly consists of mudstone, shale, dolomitic siltstone, and dolomite, with minor amounts of gypsum in some layers. The upper member is composed of organic-rich shale interbedded with dolomite beds and nodules without evaporite minerals (Fig. 2a, c; Supplementary Figs. 1 and 2). This sequence reflects the evolution from a relatively shallow evaporative lake to a persistently deep brackish-to-freshwater lacustrine environment (see Supplementary Note 1). Despite decades of sedimentological and geochemical/hydrocarbon research, due to the economic importance of the Lucaogou Formation11,12,13,22,23, the succession lacks any reliable age constraints in the absence of datable volcanic ash beds and biostratigraphically useful fossils24,25. Previous detrital zircon U-Pb geochronology obtained by in situ laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) assigned broad maximum depositional ages of ca. 270–268 Ma24 or ca. 261 Ma26 to the Lucaogou Formation. However, limited accuracy due to reworked zircons and/or post-crystallization Pb loss can lead to statistically biased results. In addition, a previously published high-precision U-Pb CA-ID-TIMS age of 281.39 ± 0.10 Ma27 from the overlying Hongyanchi Formation from the southern Bogda Mountains (Figs. 1 and 2c) resulted in a contradictory stratigraphic framework.
Here we present high-precision U-Pb zircon age from a volcanic ash bed in the upper, organic-rich member of the Lucaogou Formation (Fig. 2a, b). The 4 cm-thick ash layer interbedded within shales occurs ~925 m above the base of the Lucaogou Formation (Figs. 2c and 3). The sample (VA-1) contains zircons that are small, equant or prismatic, and euhedral, with oscillatory zoning under cathodoluminescence (Supplementary Fig. 3a). The Th/U ratios of the zircon crystals vary from 0.26 to 1.27 (Supplementary Dataset 1). The U-Pb ages determined by LA-ICP-MS have an average 2σ uncertainty of ±5.63 million years (Myr) and are distributed around a well-defined peak, with a weighted mean 206Pb/238U age of 286.14 ± 0.65 Ma (2σ internal error only; mean-squared weighted deviation [MSWD] = 1.01, n = 53; Supplementary Fig. 3a). For further verification, four single zircon grains from this sample were analyzed independently by the CA-ID-TIMS method (average 2σ uncertainty of ±0.55 Myr), with the three youngest analyses constituting a coherent cluster with a weighted mean 206Pb/238U age of 286.39 ± 0.25/0.30/0.43 Ma (2σ; MSWD = 2.0; Fig. 2b, c and Supplementary Dataset 2). Furthermore, one tuffaceous siltstone (sample TS-1) from the uppermost part of the underlying Jingjingzigou Formation was analyzed using the LA-ICP-MS method. Ninety-four zircon analyses from this sample yielded a wide range of ages, with a weighted mean 206Pb/238U age of 294.1 ± 1.4 Ma (2σ; MSWD = 1.5; Fig. 2c and Supplementary Fig. 3b) based on 13 youngest analyses (YC2σ[3+]28), and this is interpreted as being the maximum constraint on the depositional age. The available radioisotope geochronology collectively places the lower and upper boundaries of the Lucaogou Formation at ca. 294 Ma and ca. 285 Ma, respectively, and constrains its upper shale member to the Artinskian Stage; therefore, the age is significantly older than previous estimates24,25,26.
Organic carbon isotope excursions (CIEs) have been proven to be an effective global stratigraphic correlation proxy29. Our results show that a prominent negative δ13Corg excursion occurs at the upper part of the succession, with a total CIE magnitude of ~3.5‰ (above ~725 m; Fig. 3a and Supplementary Dataset 3). The observed parallel δ13CTLE and δ13CAsph records from the total lipid extract (TLE) and asphaltene (Asph), after extraction and separation, also exhibit a largely similar negative CIE in shape and magnitude to that of the bulk δ13Corg record. In addition, all the δ13Cn-alkane records of short-chain n-C19, mid-chain n-C21, and long-chain n-C27 alkanes (Supplementary Dataset 4) display a negative CIE with a magnitude of ~4‰ (Supplementary Fig. 4a). It is considered unlikely that the thermal maturity (early oil window) and proportional changes in the organic matter resulted in the observed CIE in the Lucaogou shales (see Supplementary Note 2; Supplementary Fig. 5). Importantly, under our high-precision CA-ID-TIMS age constraint, the negative CIE is comparable to that recorded in coeval marine brachiopod shells (USA and Russia)30, carbonate (South China)31, and in coastal strata (North China;29 Supplementary Fig. 4). Therefore, the observed parallel CIE signatures in bulk δ13Corg and δ13Cn-alkane reflect a perturbation of the global carbon cycle during the Artinskian (see Supplementary Note 2).
Intensified ecosystem-level microbial CH4 cycling
Investigating microbial CH4 cycling in pre-Holocene environments is challenging owing to the scarcity of diagnostic proxy records; some lipid biomarkers (e.g., glycerol dialkyl glycerol tetraether and archaeol32,33) may be invalid with an increase in the thermal maturity of the organic matter. In this study, we present the distinctive δ13C records of authigenic dolomites and hopanes (bacterial-derived biomarkers) from the Lucaogou Formation, which provide new insights into the metabolic activities of methanogens and methanotrophs in the lake ecosystem during the Early Permian.
Dolomite beds and nodules in the upper member (Fig. 2a) have very positive δ13C values (+5.8 to +16.0‰) that are significantly higher than the δ13C values (+5.3 to +8.3‰) of the dolomite in the lower member (Fig. 3b and Supplementary Dataset 5). Several mechanisms have been proposed for 13C enrichment in inorganic C pools34,35,36. Of these, the Rayleigh distillation of volatile CO2 under highly evaporative conditions34 would not have been effective in paleo-Lake Junggar, as the upper member was not deposited in a hypersaline environment12,13. In addition, the photosynthetic fixation of CO2 during productivity blooms cannot explain the positive values, because this process usually only enriches δ13C values by +2 to +3‰35, and this Permian lake was not eutrophic13. Such positive δ13C signatures have been recently attributed to authigenic dolomite precipitation associated with microbial methanogenesis10, and it is likely that some dolomite samples with less positive δ13C and lower δ18O values in the upper member have been influenced by subsequent diagenesis36 (Supplementary Fig. 6). Microbial methanogenesis is geochemically characterized by significant C isotopic fractionation, generating 13C-depleted biogenic CH4 (δ13C as low as −60 to −110‰)37 and 13C-enriched CO2 (δ13C up to +15‰ or higher)36. Such isotopically heavy CO2 acted as a substantial C source and was incorporated into the authigenic dolomite. The closest modern analogues of these dolomites, commonly observed in organic carbon-rich continental margin sediments, have been documented in the Gulf of California38 and along the Peru Margin39, where methanogenesis is highly active in oceanic sediments. Thus, the 13C-enriched authigenic dolomites presented here are a fingerprint of biogenic CH4 production in lake sediments.
Putative methanogen microfossils have been found in these 13C-enriched dolomites from the adjacent Hongyanchi section, and their abundances show a positive correlation with δ13C values10. The elevated δ13C signature of the dolomites can therefore be used to trace changes in methanogenesis. In the current study, the dolomite δ13C values show an overall increasing trend from the bottom to the top within the succession (Fig. 3b). The dolomite beds and nodules that are marked by high δ13C values occur above ~610 m. Most importantly, the more abundant strongly 13C-enriched dolomites occur within the upper part of the Lucaogou Formation (Fig. 3b), indicating a higher methanogenic rate and/or an expanded methanogen community in the anoxic lake sediments during this period. In contrast, the absence of exceptionally 13C-depleted authigenic dolomite in the studied section (δ13Ccarbonate values typically <−30‰; C largely derived from biogenic CH4)40 suggests that the anaerobic oxidation of methane was not an important process occurring in the anoxic sediments at that time. Therefore, large amounts of CH4 produced in bottom sediments escaped substantial consumption and were emitted into the overlying water column; as such, they would have provided substrates for aerobic methanotrophs.
An earlier comprehensive study13 revealed that the saturated hydrocarbon fraction from Lucaogou shales was depleted in 13C, and this possibly indicates the presence of hopanes derived from methanotrophic bacteria. In this study, we conducted a compound-specific C isotope analysis of hopanes. Hopanoids are not exclusive to methanotrophs, but their stable C isotopic compositions can be used to assess specific methanotroph contributions33,41,42,43,44,45. Methanotrophic bacteria use biogenic CH4 as a carbon source for the biosynthesis of membrane lipids (e.g., hopanoids) that are highly 13C-depleted. Our results show that the hopanoids in Lucaogou shales are dominated by C30 17α,21β-hopane and C29 17α,21β-norhopane (Supplementary Fig. 7 and Supplementary Note 1). The hopane δ13C values remain low in all samples analyzed, ranging from −44.1 to −62.6‰ for C30 17α,21β-hopane and −41.6 to −53.6‰ for C29 17α,21β-hopane (Fig. 3c). The δ13C values in these two compounds yield a positive correlation (Supplementary Fig. 5d), indicating that they have a similar bacterial community source. Their corresponding 17β,21α(H) isomers are also characterized by similar low isotopic signatures (Supplementary Dataset 4). The δ13C values of the hopanes are markedly lower than those observed in the co-occurring bulk organic matter (−24.0 to −32.0‰; Fig. 3a, c) and n-alkanes (n-C21: −33.0 to −38.1‰). Such 13C-depleted hopanoids also appear in some modern/Holocene (e.g., Lake Rotsee, Switzerland33) and Eocene (e.g., Green River Formation, USA46) lake systems, where aerobic CH4 oxidation by methanotrophic bacteria was prevalent in the water column. Here we conducted a survey of hopanoid δ13C values from 19 lakes (283 data points; δ13Chopanoid ranging from −22.2 to −71.9‰; Supplementary Fig. 8 and Supplementary Dataset 6). The data compilation (see Supplementary Note 3 for data overview) suggests that hopanoid δ13C values below −40‰ are indicative of a pronounced aerobic methanotroph contribution to these compounds (>10–20%; calculated from a C isotopic mass-balance approach;33 see Methods).
Hopanoid δ13C values can be used to trace the temporal changes in aerobic CH4 oxidation33,42,44,45. In this study, the consistently low hopane δ13C values (<−40‰) throughout the section indicate that methanotrophic activity within the lake was sustained and vigorous, particularly during the late depositional stage of the Lucaogou Formation (Fig. 3c). Specifically, in the lower member, the suitably low hopane δ13C values (ca. −50‰) and the less positive dolomite C isotopic signatures (<+8.5‰; Fig. 3b) indicate mild-to-moderately active methanotrophy and methanogenesis. In the upper member, however, the coupling between highly 13C-depleted hopanes and 13C-enriched authigenic dolomites suggests that both methanotrophs and methanogens thrived in the lake biosphere. In particular, above ~750 m, there is a persistent and obvious decrease in hopane δ13C values of >10‰ (from ca. −46 to −63‰ for C30 hopane and from ca. −43 to −54‰ for C29 hopane; Fig. 3c). The lowest values within the uppermost stratigraphic interval are among the most 13C-depleted reported in the C30 and C29 hopanes for lacustrine systems (Supplementary Fig. 8). These isotopic signatures indicate that substantially intensified CH4 oxidation occurred in the water column, which closely coincided with elevated CH4 production in the sediments, as indicated by a temporal increase in dolomite δ13C values (Fig. 3b, c). The combined evidence from both authigenic dolomite and molecular fossil (hopane) suggests that an intensification of the microbial CH4 cycling occurred during the Artinskian age. Furthermore, active CH4 cycling had a wide geographical distribution in paleo-Lake Junggar, with evidence of similar 13C-depleted hopanes also documented in the Lucaogou shales from the adjacent Sangonghe section22 and the Santanghu Basin23, hundreds of kilometers from the studied area (Fig. 1). Based upon our estimated depositional duration (ca. 9 Myr) of the Lucaogou Formation, the intensified microbial CH4 cycling persisted for at least ca. 3–5 Myr. To our knowledge, such a long-term dynamic of lacustrine CH4 cycling in the Earth’s history has not been previously and directly revealed.
Positive feedback to Artinskian climate warming
To investigate the relationship between temperature and microbial CH4 cycling, we used the chemical index of alteration (CIA)47 to reconstruct changes in the land surface temperature (LST;18,48 see Methods). The collected samples were not affected by K-metasomatism, and their uniform Ti/Al ratios indicate no changes in provenance48 (Supplementary Fig. 9 and Supplementary Dataset 7), and they thus provide a reliable record of climate variation. The CIA profiles show an increase from 50–55 in the lower member to 65–75 in the upper member, suggesting a rapid rise in the estimated LSTs from ~4 °C (Sakmarian) to ~14 °C (Artinskian; Fig. 3d). Overall, the pronounced progression toward higher CIA values, combined with the alternative chemical index of weathering (CIW;49 Supplementary Fig. 9), indicates a shift toward warmer conditions14,18,48. This record is consistent (within age uncertainties) with an independently derived CIA trend in a contemporaneous succession from the Karoo Basin of South Africa19,21 (Fig. 3e). A cross-basin correlation revealed that a significant increase in CIA (temperature) globally began near the Sakmarian–Artinskian boundary (ca. 290 Ma)18. This major climate transition can be further corroborated by a coincident decrease in δ18O values from both low- and high-latitudinal fossil shells composed of low-Mg calcite50,51 (Fig. 3e). Therefore, the elevated continental weathering in this study reflects a global climate warming signal (i.e., the Artinskian Warming Event14), which developed contemporaneously with the intensification of CH4 cycling in paleo-Lake Junggar (Fig. 3).
Higher temperatures may have stimulated methanogenesis in lake sediments, supporting a temperature control on CH4 cycling at the ecosystem level8,42,52. It has been proposed that the metabolic responses of methanogens are particularly sensitive to increases in temperature8,52. Since the predominant microbial methanogenesis occurred in the shallow sediment columns53, it would be expected that the increase in atmospheric temperature warmed the sediments and subsequently facilitated methanogenic activity. Additionally, under global warming, enhanced continental weathering (Fig. 3d) may have increased riverine nutrient influx and aquatic productivity in lakes, thereby resulting in increased substrate (e.g., acetate and H2/CO2; ref. 6) availability for methanogenesis8. However, methanotrophy is known to have a more positive effect on substrate (i.e., CH4) availability than temperature8, and the intensified CH4 consumption observed in the top part of the Lucaogou Formation (mid-Artinskian) was almost certainly a response to an increased CH4 substrate supply for methanotrophs (Fig. 3).
The balance between methanogenesis and methanotrophy ultimately controlled the amount of CH4 released into the atmosphere6,8,9. Nonetheless, if a warming-induced increase in CH4 production exceeds the increase in CH4 oxidation, an increase in net CH4 emissions is expected, and this provides potential positive feedback to climate warming. Indeed, owing to the different temperature sensitivities of methanogens and methanotrophs8,9, warming would increase CH4 emissions, which has been extensively observed in both modern freshwater ecosystems7,52,54 and laboratory incubations8,9,52. For example, experimental warming of artificial ponds has suggested a disproportionate increase in methanogenesis over methanotrophy9. Although aerobic methanotrophs did oxidize more CH4, but not enough to offset the greater warming-induced CH4 production9. Methane fluxes from lake ecosystems exhibit a temperature dependence8,52,54. The prevailing paradigm of the exponential response of CH4 emissions to temperature7,8,52,54 can be extrapolated to ancient lake systems, and the total CH4 emissions from paleo-Lake Junggar could potentially have increased by several-fold in response to Artinskian climate warming. Applying the average CH4 flux (total 31.6 Tg CH4 yr−1 in areas spanning 1,330,264 km2; i.e., 65 mg CH4 m–2 d–1)2 from modern lakes at similar latitudes to paleo-Lake Junggar (paleolatitude of 39–43°N)13, the flux was roughly estimated as 6.4 Tg CH4 yr–1 (accounting for 5–28% of annual lake CH4 emissions in the modern world4), and a total amount of ~19,200 Gt CH4 was emitted from this Early Permian lake (~270,000 km2; ref. 11; herein conservatively calculated using 3 Myr).
Although there is only evidence for intensified CH4 cycling in paleo-Lake Junggar (Fig. 1), this still provides a useful analogue for similar environments having responses to Artinskian (Early Permian) climate warming. In this respect, several contemporaneous lake systems (see Fig. 1 and Supplementary Dataset 8 for the locations of these lakes and associated essential information) may also be CH4 emission hotspots. However, accurate assessments of global CH4 emissions require clear constraints relating to the contemporaneous lake area, distribution, and environmental factors, and these are beyond the scope of this study. Nonetheless, large-scale lacustrine CH4 emissions would have acted as a positive feedback to Artinskian global warming and a critical mechanism for deriving carbon cycle perturbations. During this time period (after 290 Ma)15, the demise of the Late Paleozoic Ice Age (LPIA) was supported by a 6-fold drop in documented glacial deposits (Fig. 3f)20 and the full deglaciation in south-central Gondwana by 282 Ma19, representing one of the most prominent and enigmatic climate transitions in the Earth’s Phanerozoic history. Previous studies have demonstrated that widespread deglaciation was synchronous with an increase in atmospheric pCO2 (Fig. 3f)15 derived from volcanic eruptions (e.g., Tarim, Panjal, and Zaduo large igneous provinces)15,29, and this provides an evidence regarding the strong link between CO2 and glaciation. In addition to the contribution of CO2 (refs. 15,16,17) and potential methane clathrate release55, our results suggest that the injection of the terrestrial greenhouse gas CH4 into the atmosphere may have facilitated the demise of the LPIA and played a direct role in forcing the turnover from a long-lived icehouse to a greenhouse world.
In summary, this study investigated currently unexplored lacustrine ecosystem-level microbial CH4 cycling records, including methanogenesis and methanotrophy, in pre-Cenozoic sedimentary archives. Our results suggest that sustained and intensified CH4 cycling, as a response to Artinskian (Early Permian) climate warming, occurred in paleo-Lake Junggar. The release of the greenhouse gas CH4 from large paleo-lakes to the atmosphere could have provided a direct positive feedback to ancient global warming, at least during the Early Permian, which should improve our understanding of its role in near-future climate change within a warming-but-glaciated world.