Carbon isotopic evidence for rapid methane clathrate release recorded in coals at the terminus of the Late Palaeozoic Ice Age

The end of the Late Palaeozoic Ice Age (LPIA) ushered in a period of significant change in Earth’s carbon cycle, demonstrated by the widespread occurrence of coals worldwide. In this study, we present stratigraphically constrained organic stable carbon isotope (δ13Corg) data for Early Permian coals (312 vitrain samples) from the Moatize Basin, Mozambique, which record the transition from global icehouse to greenhouse conditions. These coals exhibit a three-stage evolution in atmospheric δ13C from the Artinskian to the Kungurian. Early Kungurian coals effectively record the presence of the short-lived Kungurian Carbon Isotopic Excursion (KCIE), associated with the proposed rapid release of methane clathrates during deglaciation at the terminus of the Late Palaeozoic Ice Age (LPIA), with no observed disruption to peat-forming and terrestrial plant communities. δ13Corg variations in coals from the Moatize Basin are cyclic in nature on the order of 103–105 years and reflect changes in δ13Corg of ~±1‰ during periods of stable peat accumulation, supporting observations from Palaeozoic coals elsewhere. These cyclic variations express palaeoenvironmental factors constraining peat growth and deposition, associated with changes in base level. This study also demonstrates the effectiveness of vitrain in coal as a geochemical tool for recording global atmospheric change during the Late Palaeozoic.


Results
The range of δ 13 C org for the coals of the Moatize Formation falls within the typical range of C3 plant organic matter 17 , ranging from an absolute maximum of −20.0‰, to an absolute minimum of −26.9‰. The data were collected across all locations domained by each respective ply within the Bananeiras, Chipanga, and Sousa Pinto seams ( Fig. 2A). From observing the range of data within these domains, three distinctive ply-domained stages are interpreted.
Stage 1, Artinskian coals of the Sousa Pinto seam exhibit variable ranges of δ 13 C org , suggesting some variation in palaeoenvironmental factors controlling low-magnitude (~±1‰) δ 13 C cycling. A weak (~1.5‰) positive shift in δ 13 C org of Stage 1 coals suggests a more long-lived change in atmospheric CO 2 concentrations and δ 13 C. These changes are concurrent with the development of widespread peat deposits, resulting in increased rates of carbon burial coincident with the Artinskian 6,8 . Stage 2 -Terminal deglaciation. Stage 2 coals, encapsulate the basal Chipanga seam ply only (BCB). The mean δ 13 C org value for this stage is −24.7‰, with a high standard deviation (σ = 1.5‰, n = 15). Stage 2 coals have the most negative δ 13 C org of all the data domains. A striking, high-magnitude (~3.5‰) negative excursion is observed δ 13 C org in Stage 2, coincident with the base of the Chipanga seam in the early Kungurian. This negative excursion is relatively short-lived compared to smaller-scale δ 13 C org cycles (~±1‰) in Stage 1 coals of the Artinskian, and Stage 3 coals of the Kungurian. The compiled δ 13 C org record from the Moatize Formation coals is time equivalent to other, continuous δ 13 C org records from sediments in both low and high-latitude sediments (Fig. 3), suggesting the observed negative carbon shift may be the globally recorded KCIE.   www.nature.com/scientificreports www.nature.com/scientificreports/ Stage 3 -Cyclic pluvials. Stage 3 coals, encapsulate the lower Chipanga seam, through to the upper Bananeiras seam (plys LUCB, LUCT, MCM, UCB, UCT, BNL and BNU). The average δ 13 C org value for this stage is −22.6‰ (σ = 0.6‰, n = 222), and remains stable throughout each ply domain, regardless of seam and age distribution. When the total sample set (excluding statistical outliers, n = 305) is normalised for each ply domain, the cyclic variation of δ 13 C org within these coals can be observed (Fig. 2B). Mechanisms for these δ 13 C org cycles are further discussed below.

Discussion
The Early Permian coals of the Moatize Formation exhibit a three-stage evolution in atmospheric δ 13 C from the Artinskian to the Kungurian. In this study, δ 13 C org cycles (particularly striking in Stage 3, see Fig. 2B), indicate a ~±1‰ shift in δ 13 C org , over discrete, regular spacing at normalised depths, from which time intervals may be estimated.
Cyclic variation of δ 13 C org in coal at similar scales has been previously observed in high-resolution isotopic studies from Eastern Australia [18][19][20] . In these works, the primary control on the distribution of δ 13 C cycles within coal is attributed to palaeoenvironmental factors controlling peat accumulation, including water availability, salinity, pH and atmospheric temperature 21 . However, the timescales over which these cycles occur have not yet been addressed.
The accumulation rates of peat are dependent on both depositional environment, and biological productivity, often genetically linked with peat-forming plant communities 22 . In the Moatize Formation coals of Stage 3, it is demonstrated that both the plant community, and depositional environment controlling peat distribution remained temporally stable to be able to preserve these δ 13 C org cycles. This also implies a state of atmospheric δ 13 C equilibrium, with no significant injections of isotopically heavy or light carbon, nor major changes in CO 2 concentrations, to disrupt δ 13 C org cycling.
By assuming a relatively constant rate of peat deposition, similar to modern rates of high latitude peat accumulation 12,23 , with peat-to-coal compaction ratios sourced from literature 22,24-26 , a range of potential time scales for each δ 13 C cycle may be calculated (Table 1).
From these calculations, it is likely that δ 13 C org cycling occurs on a similar scale to short-term (10 3 -10 5 years) trends inferred from Palaeozoic palaeosol development 27 , and in palaeofloral communities 28 . These short-term changes observed in low-latitude sediments, attributed to pluvials, result in changes in base level. These base-level changes, coincident with δ 13 C org cycling, are also observed in coals from Eastern Australia [18][19][20] , and the 10 3 -10 5 year time-frame is coincident with Milankovitch-scale orbital frequencies 28 , also observed in Mesozoic and Cenozoic coals 29,30 .
The observed KCIE is equivalent to the duration of a 10 3 -10 5 year cycle. The short-lived nature of this isotopic excursion suggests the rapid injection of 13 C-depleted carbon into the atmosphere, rather than any relatively long-lived changes in CO 2 concentration. It is possible that this negative carbon isotopic shift is due to the release of methane clathrates (CH 4 ) into the atmosphere during terminal deglaciation. Furthermore, the contribution of deep soil organic carbon (SOC) loss and CH 4 from terrestrial permafrost may also have contributed to widespread δ 13 C perturbation 31 .
The timing of this rapid CH 4 release is equivalent to the development of euxinic lake deposits across Southern Africa as a result of deglaciation marking the end of the LPIA [13][14][15] . The stratigraphic equivalent of these euxinic lacustrine deposits is represented by organic rich black shale separating the Sousa Pinto and Chipanga seams, at variable thickness at each sample location (Fig. 1B).
The accumulation of peat, evidenced by the occurrence of the Sousa Pinto seam during Stage 1, implies that more gradual global scale warming and glacial retreat resulting in base-level rise had initiated in the Artinskian, prior to evidence of any catastrophic CH 4 release. Furthermore, atmospheric CH 4 injection indicated by the KCIE seems to have little to no observable effect on peat accumulation subsequent to the ultimate terminus of the LPIA, suggesting peat-forming terrestrial ecosystems remained relatively stable during this period.
This estimated time-frame of carbon cycle perturbation during the KCIE is relatively short lived, corresponding to the short residence time of CH 4 in the atmosphere 32 . This brief time-period of potential methane clathrate release, and subsequently rapid oxidation to CO 2 , is not accompanied by any known mass extinctions, or terrestrial ecosystem catastrophe during the Early Permian 33 .
These observations suggest that whilst CH 4 release may have contributed to enhanced global warming during the terminus of the Late Palaeozoic Ice Age, the proposed effects of continental weathering and organic carbon burial linked with uplift and subsequent erosion of the Hercynian range demonstrate what maybe a more profound, and long-lived impact on global climate 8 . Additionally, the lack of observable effects on land plant communities despite significant carbon cycle perturbation during the KCIE event further supports the resilience of terrestrial flora to the effects of global scale atmospheric perturbation 34 .
The authors suggest an understanding of the global carbon cycle across geological time may greatly benefit from further research into δ 13 C org from coals.

Methods
Samples were taken from plys of the Bananeiras (n = 62, average seam thickness = 8.5 m), Chipanga (n = 175, average seam thickness = 31.3 m) and Sousa Pinto (n = 75, average seam thickness = 13.1 m) coal seams (n total = 312). Great care was taken to only sample bright (vitrain) bands from coals, as to minimise δ 13 C org variation with coal lithotype or biochemical composition 18,21,35 . Vitrains were hand-picked at a millimetre scale to avoid any potential carbonate contamination from mineralised cleats. The typical low taphonomic diversity of peat-forming ecosystems 28 , minimises the likelihood of δ 13 C org variation dependent on taxa 9 .
The δ 13 C org values were determined in the Stable Isotope Geochemistry Laboratory (SIGL) at the University of Queensland using a stable isotope ratio mass spectrometer (Isoprime), coupled in continuous flow mode with an elemental analyser (Elementar Cube) (EA-CF-IRMS). Calibration was performed by use of two standards, USGS24 (−16.1‰ δ 13 CPDB) and NAT76H (−29.26‰ δ 13 CPDB), interspersed throughout analytical runs. Each sample was analysed in duplicate, using 50-200 μg of concentrate combusted at 1020 °C in 3.5 mm × 5 mm tin capsules. Any sample with a beam size outside the working range of 1 × 10 −9 to 9 × 10 −9 Å, or with a δ 13 C org result variation between duplicates of >0.4‰, was re-analysed, in accordance with laboratory quality control practices.
Final data values were normalised and are reported in ‰ VPDB.