Introduction

Methane (CH4) is the third-most important greenhouse gas (after water vapor and carbon dioxide) and a significant contributor to global climate change1,2,3,4,5,6,7,8. Its globally averaged marine surface annual mean mole fraction in the atmosphere steadily increased from ~1600 parts per billion (ppb) to ~1775 ppb in the 1980–1990s, stabilized around ~1775 ppb during the period 1999–2006, and then returned to the earlier pattern of increases leading to ~1860 ppb in 20181,3,9,10. There are anthropogenic (e.g., agriculture, wastes, fossil fuels, biomass burning) and natural (e.g., wetlands, freshwaters, geological seepage, wild fires) sources of CH4 to the atmosphere. The carbon-isotopic composition of CH4 (the ratio of stable isotopes 12C and 13C expressed as δ13C (‰) relative to the Vienna Pee Dee Belemnite standard), together with estimates of flux from each source-type, can be used to infer the relative contributions of various CH4 emitters to the global budget by matching co-constrained global observations of CH4 and its δ13C3,4. Furthermore, temporal variations in δ13C of atmospheric CH4 are a highly useful indicator of changes in the trends of various emitters and/or CH4 sinks. The global mean atmospheric δ13CCH4 trended upward between 1980 (approximately −47.7‰) and 1997 (approximately −47.1‰), and remained relatively stable until 2008, before decreasing towards the most recent values of approximately −47.3‰1,6.

This recent increase in the atmospheric CH4 burden, coincident with the depletion of 13C, has been attributed to the increasing contribution of biogenic CH4 from wetlands and from agricultural activities such as cattle husbandry that produce CH4 with δ13C usually more negative than −55‰1,4,9. While atmospheric CH4 sinks are less extensively studied, changes in the sink strength may at least partially explain some of the long-term observed trends in δ13CCH411. Fossil fuel production also contributes to increasing atmospheric CH45,12,13. However, CH4 in fossil fuels is, on average, enriched in 13C (δ13C = −44‰4,14) relative to globally-averaged atmospheric CH4. Decreasing δ13C of atmospheric CH4 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 developments5 and from shale gas/oil developments in particular2 (see Text S1 in Supplementary Information) have been the greatest single cause of the recent global increase of atmospheric CH4. Here, we use a large global dataset of δ13CCH4 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 CH4 since 2008.

Materials and Methods

Global isotopic dataset

We analyzed δ13CCH4 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 Etiope15 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 gas17. 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 δ13CCH4 values

Values of production volume-weighted δ13CCH4 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 δ13CCH4 for the corresponding shale formation, and then summing up the results. Emission volume-weighted δ13C-CH4 values were derived by first calculating the proportion of CH4 emissions from each shale formation in total emissions, multiplying that value by average δ13CCH4 for corresponding shale formation, and then summing up the results.

Results

The arithmetic mean δ13CCH4 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 13C than CH4 produced from conventional oil and gas reservoirs (mean δ13CCH4 = −44.0 ± 0.1‰, n = 6079 in the study of Sherwood et al.14; mean δ13CCH4 = −42.8 ± 0.1‰, n = 12,697 in the study of Milkov et al.16) and significantly more enriched in 13C than the modern atmospheric δ13CCH4 (−47.3‰1,6). We note that shale gas is even more enriched in 13C relative to the global average δ13CCH4 (about −54‰4) of all atmospheric sources prior to isotopic fractionation of atmospheric CH4 by all sinks resulting in the modern atmospheric value above.

Figure 1
figure 1

δ13CCH4 values from a global dataset of 1619 samples of produced shale gases from around the world. The data are displayed using a box plot, which shows distribution of values as histogram, average (mean) value (−41.3‰) as black star, median value (−41.4‰) as dotted line, first quartile (Q1), third quartile (Q3), lower adjacent value, upper adjacent value, and outliers. The first quartile (Q1) is the median of the lower half of the data set. 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.

Global shale gas production increased from about 31 billion cubic meters (bcm) in 2005 to about 434 bcm in 201518. 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 δ13CCH4 data on gases produced from these and other principal shale formations in the USA.

Figure 2
figure 2

Cumulative production (in billion cubic meters or bcm) of dry gas from shale plays in the USA. Data are from Energy Information Administration17.

Figure 3
figure 3

Main statistics on δ13CCH4 for gases produced from main shale formations in the USA, Canada and China. The data are displayed using box plots (see Fig. 1 for legend). Original data are in Table S1.

In this study, we use the global δ13CCH4 dataset to derive δ13CCH4 representative of both produced gas (volume-weighted average) and the δ13CCH4 signature when weighted for measured emissions across plays (emission-weighted average). Table 1 presents average δ13CCH4 for the main producing shale plays in the USA. The 1002 available gas samples with δ13CCH4 data are from plays that account for 94% of cumulative US shale gas production. The average δ13CCH4, 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 CH4 is significantly enriched in 13C (mean δ13CCH4 is −32.0‰, n = 98). This latter source significantly influences the average volume-weighted isotope signature of CH4 produced from shales in the USA.

Table 1 Data used to calculate the average δ13CCH4 in shale gases produced in the USA since 2008.

The average shale δ13CCH4 weighted by the amount of emissions measured in 2015 from the main USA shale plays19 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 δ13CCH4 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 δ13CCH4 (e.g., Marcellus and Haynesville formations) to both production and emissions increased in that period.

Table 2 Data used to calculate the emission-weighted average δ13C of CH4 emitted from, mostly, shale gas production in selected plays and areas in the USA in 2015.
Figure 4
figure 4

Increasing δ13CCH4 signature of produced and emitted gas from shale developments in the USA from 2000 to mid-2019. Only formations with sufficient isotopic data (Marcellus, Barnett, Haynesville, Permian, Eagle Ford, Fayetteville, Woodford, Niobrara-Codell and Bakken) and emission data (Marcellus, Barnett, Haynesville, Eagle Ford, Fayetteville, Niobrara-Codell and Bakken) are used to construct this figure.

Gas samples from three other countries currently producing shale gas commercially (Canada, China, Argentina) indicate somewhat more positive δ13CCH4 values than the USA (Fig. 3), resulting in a global volume-weighted δ13CCH4 signature of −38.8‰ (Table S3, Text S4). The volume- and emission-weighted δ13CCH4 values calculated here do not account for shale plays for which production has become negligible after the 1990s (see Text S4).

The US shale δ13CCH4 weighted by the amount of cumulative production from 2000 to mid-2019 for each shale play is −40.0‰ (Table S4). The mean δ13CCH4 of produced shale gas in the USA since 2008 is −39.6‰ (Table 1). The slight (by 0.4‰) enrichment in 13C 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 contribution from the Barnett Formation during that period. The mean δ13CCH4 of globally produced shale gas in 2018 is −38.‰ (Table S3). We also estimated an emission-weighted δ13CCH4 of −41.8‰ for the principal US shale plays in 2015 (Table 2). These values are appropriate for utilization in models that constrain CH4 emissions from shale developments to the atmosphere based on matching modelled global δ13C to observed δ13C. Methane from produced shales is, on average, significantly enriched in 13C relative to atmospheric CH4 (δ13CCH4 ~−47‰).

Discussion

Recently, atmospheric CH4 became more abundant but also depleted in 13C, as δ13C decreased from about −47.1‰ in 2007 to −47.3‰ in 2017. If shale gas (with δ13CCH4 around −40‰ as documented in this study) and conventional oil and gas (with δ13CCH4 around −43‰16) were conceived to collectively dominate recent emissions of CH4 to the atmosphere, then atmospheric CH4 would very simply become more enriched in 13C relative to the current global mean δ13C, which is not consistent with global observations. While we agree that shale developments (and fossil fuel in general) represent an important CH4 source, and that emissions from those sources have been likely increasing due to growing production, we conclude that the increases in global atmospheric CH4 concentrations since 2008 are not as strongly attributable to shale gas and conventional oil and gas emissions as some studies claim2,5, based on our global observations of isotopic fractionation.

Additionally, we must emphasize that the measured atmospheric δ13CCH4 signal is the sum-total of all CH4 source and sink terms. For example, a decrease in biomass burning emissions (significantly enriched in 13C (δ13CCH4 −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 δ13CCH4 signal over time as if both sources had no trend4,5. The biomass burning category includes fires and solid biofuels (e.g., for use in cook stoves). Data on global CH4 emissions from fires is not entirely conclusive. Remote sensing data of CH4 and CO (and assuming (i) biomass burning CH4/CO emission ratios and (ii) a partitioning of CO emissions across sectors) suggests decreased fire CH4 emissions of ~3.7 Tg/yr from the 2001–2007 to the 2008–2014 periods5. In contrast, remote sensing of burned fire area suggests no such trend20 (no trend over this period apart from inter-annual variation; Fig. S1). Furthermore, CH4 emissions from solid biofuels are reported to have increased from 12.2 to 13.6 Tg/yr from 2000–201221 (latest time series available). While this data does not indicate an immediately apparent decrease in global biomass burning CH4 emissions, more research is needed. Potential trends in the various CH4 sink processes such as the soil sink22 and the tropospheric OH sink11 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 CH4 trends, when attributing the global signal1,9.

From the above, it follows that attributing ~1/3 of the global CH4 increase to North American shale gas production and another ~1/3 to conventional gas and oil with a simple mass balance approach2 is not supported by observations because of unconstrained uncertainties. Based on long-term airborne CH4 measurements over the US, previous analysis concludes that oil and gas industry CH4 emissions (shale and conventional) over the past decade have increased at about the same rate as natural gas production volume7. 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 USA23, and additional international studies paint a similar picture24,25, 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 CH4 source and sink terms and isotopic signatures, additional CH4 emissions associated with increased shale gas development in the USA cannot account for a large fraction of the recent increase in atmospheric CH4. Yet, oil and gas industry expansion remains a significant factor in the complex patterns of global atmospheric CH4 emissions and concentrations4,23,24,25. And, of equal importance, fossil fuel CH4 sources may be mitigated with policy and best (or better) industrial practice that can effectively reduce emissions. We suggest that the rise in global CH4 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 CH4 concentration1. 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 CH4 since 2008, we should not lose sight of the powerful impact of interventions to reduce emissions from sources we have.

Conclusions

CH4 recently increased in the atmosphere and simultaneously became more depleted in 13C. In this study, we compiled a large global dataset of isotopic composition of CH4 produced from shale formations that account for most global shale gas production. Developments of shale gas and oil on average emit CH4 significantly more enriched in 13C than the atmospheric CH4 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 CH4 to the atmosphere. It is important to understand all sources of CH4 that collectively contribute to recent atmospheric increases, and isotopic data provide key constraints for this.