Using global isotopic data to constrain the role of shale gas production in recent increases in atmospheric methane

The accelerated increase in global methane (CH4) in the atmosphere, accompanied by a decrease in its 13C/12C isotopic ratio (δ13CCH4) from −47.1‰ to −47.3‰ observed since 2008, has been attributed to increased emissions from wetlands and cattle, as well as from shale gas and shale oil developments. To date both explanations have relied on poorly constrained δ13CCH4 source signatures. We use a dataset of δ13CCH4 from >1600 produced shale gas samples from regions that account for >97% of global shale gas production to constrain the contribution of shale gas emissions to observed atmospheric increases in the global methane burden. We find that US shale gas extracted since 2008 has volume-weighted-average δ13CCH4 of −39.6‰. The average δ13CCH4 weighted by US basin-level measured emissions in 2015 was −41.8‰. Therefore, emission increases from shale gas would contribute to an opposite atmospheric δ13CCH4 signal in the observed decrease since 2008 (while noting that the global isotopic trend is the net of all dynamic source and sink processes). This observation strongly suggests that changing emissions of other (isotopically-lighter) CH4 source terms is dominating the increase in global CH4 emissions. Although production of shale gas has increased rapidly since 2008, and CH4 emissions associated with this increased production are expected to have increased overall in that timeframe, the simultaneously-observed increase in global atmospheric CH4 is not dominated by emissions from shale gas and shale oil developments.

www.nature.com/scientificreports www.nature.com/scientificreports/ in fossil fuels is, on average, enriched in 13 C (δ 13 C = −44‰ 4,14 ) relative to globally-averaged atmospheric CH 4 . Decreasing δ 13 C of atmospheric CH 4 since 2008 implies that emissions from biogenic sources are therefore increasing at a greater rate relative to emissions from fossil fuels. However, recent studies have suggested that emissions from conventional petroleum developments 5 and from shale gas/oil developments in particular 2 (see Text S1 in Supplementary Information) have been the greatest single cause of the recent global increase of atmospheric CH 4 . Here, we use a large global dataset of δ 13 C CH4 from produced shale formations, which leads us to conclude that emissions from shale gas and oil production have not played a dominant role in the increase in atmospheric CH 4 since 2008.

Materials and Methods
Global isotopic dataset. We analyzed δ 13 C CH4 data for 1619 samples of produced natural gas from 38 shale formations around the world originally presented in 73 studies (Table S1). This shale gas dataset is a subset of a larger global inventory of gas samples from conventional reservoirs, shales, coals, seeps and other geological settings originally published by Sherwood et al. 14 and further expanded and discussed by Milkov and Etiope 15 and Milkov et al. 16 . Although most gas samples come from formations dominated by true shale lithology (e.g., the Marcellus Formation, USA), we also include samples collected from unconventional low-permeability (tight) reservoirs dominated by very fine-grained sandstone or siltstone (e.g., the Montney Formation, Canada) or mixed clastic/carbonate lithologies (e.g., the Niobrara Formation, USA) developed through hydraulic fracturing and commonly included in the inventories of produced shale gas 17 . The produced gas may be free gas associated with relatively little condensate liquids (e.g., in the Haynesville Formation) or oil-dissolved gas (e.g., in the Eagle Ford Formation). Most samples come from the USA (n = 1238), followed by China (n = 252), Canada (n = 124), United Kingdom (n = 2), Sweden (n = 2) and Australia (n = 1). calculation of weighted δ 13 c CH4 values. Values of production volume-weighted δ 13 C CH4 for shale gases were derived by first calculating the proportion of gas production from each shale formation in the total production, then multiplying that value by the average δ 13 C CH4 for the corresponding shale formation, and then summing up the results. Emission volume-weighted δ 13 C-CH 4 values were derived by first calculating the proportion of CH 4 emissions from each shale formation in total emissions, multiplying that value by average δ 13 C CH4 for corresponding shale formation, and then summing up the results.

Results
The arithmetic mean δ 13 C CH4 for all shale gas samples is −41.3 ± 0.2‰ (n = 1619, range from −70‰ to −23.3‰, median −41.4‰) (Fig. 1). The mean value is slightly more positive than −42.5 ± 0.3‰ reported by Sherwood et al. 14 based on a smaller dataset of 647 samples. Methane from produced shales is, on average, more enriched in 13 C than CH 4 produced from conventional oil and gas reservoirs (mean δ 13 C CH4 = −44.0 ± 0.1‰, n = 6079 in the study of Sherwood et al. 14 ; mean δ 13 C CH4 = −42.8 ± 0.1‰, n = 12,697 in the study of Milkov et al. 16 ) and significantly more enriched in 13 C than the modern atmospheric δ 13 C CH4 (−47.3‰ 1,6 ). We note that shale gas is even more enriched in 13 C relative to the global average δ 13 C CH4 (about −54‰ 4 ) of all atmospheric sources prior to isotopic fractionation of atmospheric CH 4 by all sinks resulting in the modern atmospheric value above.
Global shale gas production increased from about 31 billion cubic meters (bcm) in 2005 to about 434 bcm in 2015 18 . In the USA, the cumulative production of shale gas from 2000 to mid-2019 reached approximately 4.5 trillion cubic meters (tcm), including about 4.1 tcm produced since 2008 (Fig. 2, based on dry gas production). Half of the cumulative shale gas was produced from the Marcellus, Barnett and Haynesville formations. Figure 3 summarizes δ 13 C CH4 data on gases produced from these and other principal shale formations in the USA.
In this study, we use the global δ 13 C CH4 dataset to derive δ 13 C CH4 representative of both produced gas (volume-weighted average) and the δ 13 C CH4 signature when weighted for measured emissions across plays (emission-weighted average). Table 1 presents average δ 13 C CH4 for the main producing shale plays in the USA. The 1002 available gas samples with δ 13 C CH4 data are from plays that account for 94% of cumulative US shale gas production. The average δ 13 C CH4 , when weighted by the amount of cumulative production from each shale play during 2008-2019, is −39.6‰. A large proportion (28%) of cumulative shale gas production comes from the Marcellus Formation where CH 4 is significantly enriched in 13 C (mean δ 13 C CH4 is −32.0‰, n = 98). This latter source significantly influences the average volume-weighted isotope signature of CH 4 produced from shales in the USA.
The average shale δ 13 C CH4 weighted by the amount of emissions measured in 2015 from the main USA shale plays 19 is −41.8‰ (Table 2). Sensitivity analysis suggests that this value changes little when emission measurements from other years are considered (see Table S2, Text S3). We also calculated how the average δ 13 C CH4 signature of shale-emitted gas changed over time. When weighted by production or emissions, the US average signature becomes heavier (thus, opposite to the direction of the atmospheric trend) by about 4-7‰ from 2000 to mid-2019 (Fig. 4). This is because the relative contribution of shales with relatively more positive δ 13 C CH4 (e.g., Marcellus and Haynesville formations) to both production and emissions increased in that period.
Gas samples from three other countries currently producing shale gas commercially (Canada, China, Argentina) indicate somewhat more positive δ 13 C CH4 values than the USA (Fig. 3), resulting in a global volume-weighted δ 13 C CH4 signature of −38.8‰ (Table S3, Text S4). The volume-and emission-weighted δ 13 C CH4 values calculated here do not account for shale plays for which production has become negligible after the 1990s (see Text S4).
The US shale δ 13 C CH4 weighted by the amount of cumulative production from 2000 to mid-2019 for each shale play is −40.0‰ (Table S4). The mean δ 13 C CH4 of produced shale gas in the USA since 2008 is −39.6‰ ( Table 1). The slight (by 0.4‰) enrichment in 13 C during 2008-2019, relative to 2000-2019, is due to a relatively larger contribution of production from the Marcellus and Haynesville formations in 2008-2019, and a smaller www.nature.com/scientificreports www.nature.com/scientificreports/ contribution from the Barnett Formation during that period. The mean δ 13 C CH4 of globally produced shale gas in 2018 is −38.‰ (Table S3). We also estimated an emission-weighted δ 13 C CH4 of −41.8‰ for the principal US shale plays in 2015 (Table 2). These values are appropriate for utilization in models that constrain CH 4 emissions from shale developments to the atmosphere based on matching modelled global δ 13 C to observed δ 13 C. Methane from produced shales is, on average, significantly enriched in 13 C relative to atmospheric CH 4 (δ 13 C CH4 ~−47‰).

Discussion
Recently, atmospheric CH 4 became more abundant but also depleted in 13 C, as δ 13 C decreased from about −47.1‰ in 2007 to −47.3‰ in 2017. If shale gas (with δ 13 C CH4 around −40‰ as documented in this study) and conventional oil and gas (with δ 13 C CH4 around −43‰ 16 ) were conceived to collectively dominate recent emissions of CH 4 to the atmosphere, then atmospheric CH 4 would very simply become more enriched in 13 C relative to the current global mean δ 13 C, which is not consistent with global observations. While we agree that This means that about 25% of the values in the data set lie below Q1 and about 75% lie above Q1. The third quartile (Q3) is the median of the upper half of the data set. This means that about 75% of the values in the data set lie below Q3 and about 25% lie above Q3. The lower adjacent value is the smallest observation that is greater than or equal to the lower inner fence, which is the first quartile minus 1.5 × IQR, where IQR stands for the interquartile range. The upper adjacent value is the largest observation that is less than or equal to the upper inner fence, which is the third quartile plus 1.5 × IQR. Outliers are all values that fall outside of either of the fences. Original data are in Table S1. www.nature.com/scientificreports www.nature.com/scientificreports/ shale developments (and fossil fuel in general) represent an important CH 4 source, and that emissions from those sources have been likely increasing due to growing production, we conclude that the increases in global atmospheric CH 4 concentrations since 2008 are not as strongly attributable to shale gas and conventional oil and gas emissions as some studies claim 2,5 , based on our global observations of isotopic fractionation.
Additionally, we must emphasize that the measured atmospheric δ 13 C CH4 signal is the sum-total of all CH 4 source and sink terms. For example, a decrease in biomass burning emissions (significantly enriched in 13 C (δ 13 C CH4 −22.3 ± 1.9‰ 4 ), and an increase in fossil fuel emissions (including shale gas), could in principle result in the same global average atmospheric δ 13 C CH4 signal over time as if both sources had no trend 4,5 . The biomass burning category includes fires and solid biofuels (e.g., for use in cook stoves). Data on global CH 4 emissions from fires is not entirely conclusive. Remote sensing data of CH 4 and CO (and assuming (i) biomass burning CH 4 /CO emission ratios and (ii) a partitioning of CO emissions across sectors) suggests decreased fire CH 4 emissions of  Table S1.

Shale formation
Total dry shale gas production from 2008 to mid-2019 (bcm)

Portion (%) of total dry shale gas production from 2008 to mid-2019
Average δ 13 5 . In contrast, remote sensing of burned fire area suggests no such trend 20 (no trend over this period apart from inter-annual variation; Fig. S1). Furthermore, CH 4 emissions from solid biofuels are reported to have increased from 12.2 to 13.6 Tg/yr from 2000-2012 21 (latest time series available). While this data does not indicate an immediately apparent decrease in global biomass burning CH 4 emissions, more research is needed. Potential trends in the various CH 4 sink processes such as the soil sink 22 and the tropospheric OH sink 11 can further complicate the diagnosis of source trends. As a result, it is important to account for these processes, as well as other existing evidence such as latitudinal and seasonal CH 4 trends, when attributing the global signal 1,9 . From the above, it follows that attributing ~1/3 of the global CH 4 increase to North American shale gas production and another ~1/3 to conventional gas and oil with a simple mass balance approach 2 is not supported by observations because of unconstrained uncertainties. Based on long-term airborne CH 4 measurements over the US, previous analysis concludes that oil and gas industry CH 4 emissions (shale and conventional) over the past decade have increased at about the same rate as natural gas production volume 7 . The existence of unaccounted and poorly characterized emission sources within the oil and gas industry has also been demonstrated through intensive field studies in the USA 23 , and additional international studies paint a similar picture 24,25 , although little independent measurement data exist for many world regions including the Middle East, the Former Soviet Union, and Africa. Further research targeted for these areas, in addition to changing biogenic sources and sinks, will serve to further constrain the conclusions made in this work.
Based on existing knowledge of CH 4 source and sink terms and isotopic signatures, additional CH 4 emissions associated with increased shale gas development in the USA cannot account for a large fraction of the recent increase in atmospheric CH 4 . Yet, oil and gas industry expansion remains a significant factor in the complex patterns of global atmospheric CH 4 emissions and concentrations 4,[23][24][25] . And, of equal importance, fossil fuel CH 4 sources may be mitigated with policy and best (or better) industrial practice that can effectively reduce emissions. We suggest that the rise in global CH 4 concentrations is most effectively seen not through a lens of what is the most important or dominant source of emissions, but rather understanding all sources and how they can collectively explain the observed patterns of atmospheric increases. Indeed, a reduction in emissions from any major source (such as fossil fuels or cattle husbandry) would be expected to lead to a reduction in the global CH 4  Table 2. Data used to calculate the emission-weighted average δ 13 C of CH 4 emitted from, mostly, shale gas production in selected plays and areas in the USA in 2015. Gas production, percentage of emitted gas, and CH 4 emissions are from Peischl et al. 19 and references therein. These results account for ~60% of total USA shale gas production in 2015. www.nature.com/scientificreports www.nature.com/scientificreports/ concentration 1 . Therefore, although our analysis indicates that shale gas and conventional gas and oil production has not played a dominant role in the increase in atmospheric CH 4 since 2008, we should not lose sight of the powerful impact of interventions to reduce emissions from sources we have. conclusions CH 4 recently increased in the atmosphere and simultaneously became more depleted in 13 C. In this study, we compiled a large global dataset of isotopic composition of CH 4 produced from shale formations that account for most global shale gas production. Developments of shale gas and oil on average emit CH 4 significantly more enriched in 13 C than the atmospheric CH 4 signal. Given current knowledge of global isotopic data and processes, the increase in US shale oil and gas apparently does not dominate the recent increased emissions of global CH 4 to the atmosphere. It is important to understand all sources of CH 4 that collectively contribute to recent atmospheric increases, and isotopic data provide key constraints for this.

Data availability
The dataset used in this study is available as Supplementary information.