Old carbon reservoirs were not important in the deglacial methane budget

See allHide authors and affiliations

Science  21 Feb 2020:
Vol. 367, Issue 6480, pp. 907-910
DOI: 10.1126/science.aax0504

Small burden from old sources

Methane is a potent greenhouse gas with large natural sources, reservoirs, and sinks. Dyonisius et al. found that methane emissions from old, cold-region carbon reservoirs like permafrost and methane hydrates were minor during the last deglaciation (see the Perspective by Dean). They analyzed the carbon isotopic composition of atmospheric methane trapped in bubbles in Antarctic ice and found that methane emissions from those old carbon sources during the warming interval were small. They argue that this finding suggests that methane emissions in response to future warming likely will not be as large as some have suggested.

Science, this issue p. 907; see also p. 846


Permafrost and methane hydrates are large, climate-sensitive old carbon reservoirs that have the potential to emit large quantities of methane, a potent greenhouse gas, as the Earth continues to warm. We present ice core isotopic measurements of methane (Δ14C, δ13C, and δD) from the last deglaciation, which is a partial analog for modern warming. Our results show that methane emissions from old carbon reservoirs in response to deglacial warming were small (<19 teragrams of methane per year, 95% confidence interval) and argue against similar methane emissions in response to future warming. Our results also indicate that methane emissions from biomass burning in the pre-Industrial Holocene were 22 to 56 teragrams of methane per year (95% confidence interval), which is comparable to today.

Methane (CH4) is an important contributor to the greenhouse effect, with a global warming potential ~28 times higher than that of carbon dioxide (CO2) on a 100-year time scale (1). Natural CH4 emissions currently account for ~40% of total emissions (2) and there are considerable uncertainties in their response to future warming (3). Although wetlands are the dominant natural source of CH4, increased emissions from large, climate-sensitive old carbon reservoirs such as permafrost (4) and hydrates under ice sheets (5) might become important in the coming century. Marine hydrates may also have the potential to emit a substantial amount of CH4 into the atmosphere in response to warming (6), but the time scale of marine hydrate dissociation is relatively long (on the order of hundreds to thousands of years). Furthermore, there is a growing consensus that CH4 release to the atmosphere from dissociating marine hydrates will be buffered by efficient CH4 oxidation in the sediments and water column (3, 7).

The last deglaciation [18 to 8 kilo-annum before present (ka BP)] provides the opportunity for evaluating the long-term sensitivity of these old carbon reservoirs (marine hydrates, permafrost, and hydrates under ice sheets) to a changing climate. There is abundant evidence of the destabilization of marine hydrates (8, 9), land permafrost degradation (10), and thermokarst lake (permafrost thaw lake) formation (11) during the last deglaciation. However, CH4 emissions from these old carbon reservoirs into the atmosphere are not well constrained. The paleoatmospheric CH4 mole fraction and its isotopic composition from trapped air in ice cores provide a historical perspective on how natural CH4 sources respond to climate change (e.g., 12, 13). Measurements of carbon-14 (14C) of CH4 (14CH4) from ice cores specifically provide an unambiguous top-down constraint on the globally integrated 14C-free CH4 emissions from all old carbon reservoirs.

14C decays radioactively and is thus strongly depleted in carbon reservoirs that have been isolated from the atmosphere for time periods longer than its half-life of ~5730 years. Because of the low abundance of 14C (on the order of 10−12 compared with 12C), measurements of 14CH4 in ice cores are challenging, requiring ~1000 kg of ice per sample. We collected ice cores from a well-dated ice ablation site on Taylor Glacier, Antarctica (14), which provides easy access to large volumes of old ice at shallow depths. Petrenko et al. (15) recently presented measurements of paleoatmospheric 14CH4 from Taylor Glacier for the Younger Dryas–Preboreal (YD-PB) transition (11.7 to 11.3 ka BP) and concluded that 14C-free CH4 emissions were small [<7.7% of total CH4 emissions, 95% confidence interval (CI)]. However, their results only spanned a brief time interval within the deglacial transition. In this study, we present 11 additional measurements of paleoatmospheric 14CH4 (Fig. 1A) combined with stable isotope measurements (δ13CH4 and δD-CH4) (Fig. 1, C and D) in the 15- to 8-ka BP time interval, providing a more complete picture of the deglacial CH4 budget.

Fig. 1 CH4 isotopes, mole fraction, NH temperature reconstruction, and relative sea level (RSL) during the last deglaciation.

(A) ∆14CH4 from Taylor Glacier (blue diamonds; this study), ∆14C of contemporaneous CO2 from IntCal13 [green line (19)], IntCal13 raw data [gray crosses (19)], and earlier ∆14CH4 results [light blue diamonds (15)]. Two ∆14CH4 samples from the 2014–2015 field season (at 17.8 and 12.8 ka BP) were rejected because of suspected addition of extraneous 14C [see section 3 of the materials and methods (20)]. (B) CH4 mole fraction from discrete WAIS Divide ice core measurements [red dots (39)], Taylor Glacier (blue diamonds; this study), and an earlier Taylor Glacier study [light blue diamonds (15)]. (C) δ13CH4 from TALDICE (red squares), EDML [yellow squares (13)], and Taylor Glacier (blue squares; this study). (D) δD-CH4 from EDML [green triangles (13)] and Taylor Glacier (blue triangles; this study). (E) Composite NH temperature stack (red line) and its 95% CI (shaded orange area) (16). (F) Global RSL inferred from coral data (32). All ice core data are plotted with respect to the WD2014 age scale (40); IntCal13, RSL, and NH temperature stacks are plotted on their respective age scales. All error bars represent the 95% CI.

The Oldest Dryas–Bølling (OD-B) transition (14.6 to 14.45 ka BP) represents the first large and abrupt CH4 rise during the last deglacial sequence of events (Fig. 1B) at the time when sea level was ~100 m lower than today. This abrupt CH4 rise was synchronous with the acceleration of Northern Hemisphere (NH) warming (16) (Fig. 1E), ice sheet retreat, and rapid sea-level rise (17). This climate transition may have also coincided with the first instance of marine hydrate destabilization during the last deglaciation caused by hydrostatic pressure relief from NH ice sheet retreat and incursion of warm intermediate ocean water into shallow, hydrate-bearing Arctic sediments (8). During the destabilization of marine hydrate reservoirs, abrupt events such as submarine landslides (18) or collapse of marine hydrate pingos (8) could result in large and rapid CH4 expulsions that may have contributed to the rapid atmospheric CH4 rise (9) if they were capable of bypassing oxidation in the water column.

In contrast to old carbon reservoirs, contemporaneous CH4 sources such as wetlands and biomass burning emit CH4 with a 14C signature that reflects the contemporaneous Δ14CO2 at the time (15). Our Δ14CH4 measurements for the OD-B transition are all within 1σ uncertainty of the contemporaneous atmospheric Δ14CO2 (19) (Fig. 1A), indicating a dominant role of contemporaneous CH4 sources. We used a one-box model (see section 4.2 of the materials and methods) (20) to calculate the amount of 14C-free CH4 emission into the atmosphere (Table 1, fig. S9, and table S10) (20). Our box model shows that the total 14C-free CH4 emissions during the OD-B transition were small [on average, <13 teragrams (Tg) of CH4 per year, 95% CI upper limit]. Combined with earlier Δ14CH4 data from the YD-PB transition (15), our results argue strongly against the hypothesis regarding old carbon reservoirs being important contributors to the rapid CH4 increases associated with abrupt warming events (Dansgaard–Oeschger events) (9). This conclusion is consistent with previous studies (13) showing no major enrichment in the CH4 deuterium/hydrogen ratio (δD-CH4) concurrent with the abrupt CH4 transitions (CH4 from marine hydrates is relatively enriched in δD). It has been shown that even at a relatively shallow water depth of ~30 m, ~90% of the 14C-free CH4 released from thawing subsea permafrost was oxidized in the water column (21). We hypothesize that during the OD-B transition, relatively rapid sea-level rise associated with meltwater pulse 1-A (17), combined with CH4 oxidation in the water column (22), may have prevented CH4 emissions from disintegrating marine hydrates and sub-sea permafrost from reaching the atmosphere.

Table 1 CH4 source strength estimates (95% CI) for the time intervals of our samples.

Sample ages were determined by value matching of globally well-mixed gases (CH4 and δ18O of atmospheric oxygen) to WD2014 chronology [see section 1 of the materials and methods (20)]. The sample ages given in this table represent the “best” (maximum probability) age on the probability distribution (fig. S3) (20) with respect to WD2014 chronology (40).

View this table:

Our measurements of 14CH4 during the Bølling–Allerød interstadial (14.45 to 13 ka BP) and the early Holocene (10 to 8 ka BP) warm period (Fig. 1A) provide an opportunity to assess the likelihood of delayed CH4 emissions from old carbon reservoirs in response to warming. The onset of marine hydrate dissociation might lag the initial warming signal on decadal (23), centennial, or even millennial (18) time scales. Permafrost degradation could also lag a warming signal on decadal and centennial time scales (24) depending on local environmental conditions such as permafrost depth, soil types, and moisture content (4). During parts of the early Holocene, Arctic temperatures were likely warmer than today (25), providing a good analog for Arctic conditions in the coming decades. Proxy reconstructions of thermokarst lake initiation (11) and land permafrost degradation (10, 24) suggested a potential increase of CH4 emissions from these processes during both the Bølling–Allerød interstadial and the early Holocene warm period. However, our Δ14CH4 measurements (Fig. 1A and Table 1) show no evidence of delayed 14C-free CH4 emissions after warming. These results are consistent with present-day observations that carbon from thermokarst lakes and permafrost is predominantly emitted in the form of CO2 rather than CH4 (4, 26), and that CH4 emissions from permafrost systems are dominated by relatively contemporaneous carbon (26, 27).

Because carbon stored in permafrost is not expected to be 14C free (28), we also attempted to use our 14CH4 results to calculate the possible magnitude of CH4 emissions from thawing old carbon in permafrost (Section 4.3) (20). This calculation assumed that the 14C activity of permafrost CH4 emissions follows the predepositional age of terrigenous biomarkers released from thawing permafrost (7500 ± 2500 years old relative to our sample age) (10). Resulting CH4 emissions from old permafrost carbon range from 0 to 53 Tg CH4 per year (table S10) (20) throughout the last deglaciation and may have contributed up to 27% of the total CH4 emissions to the atmosphere (95% CI upper limit) at the end of the OD-B transition (14.42 ka BP). However, we consider this calculation speculative (see section 4.3 of the materials and methods) (20).

When the global sea level was lower, exposure of continental shelves may have resulted in higher CH4 emissions from natural geologic seeps (29). A recent study also inferred the existence of CH4 hydrate deposits underneath ice sheets and suggested that the proglacial meltwater discharge is likely an important source of CH4 to the atmosphere (5). Ice sheet retreat during the last deglaciation may have destabilized the subglacial hydrate deposits, which contain old, 14C-depleted CH4. However, our data, which span most of the deglacial ice retreat and sea-level rise (Fig. 1F), argue strongly against both hypotheses. The 14C-free CH4 emissions were small throughout the last deglaciation (Table 1) and appear to be insensitive to both global sea level and ice volume.

Biomass burning is an important component of the global carbon cycle and is tightly coupled with emissions of carbon monoxide (CO), nitrogen oxides (NOx), nonmethane hydrocarbons, and aerosols that have substantial effects on atmospheric chemistry and radiative energy fluxes. Compared with other proxies of past biomass burning, CH4 has an advantage because it is a well-mixed gas in the atmosphere and can represent the globally integrated biomass-burning emissions. Bock et al. (13) provided the most recent stable isotope–based (δ13C and δD) study of the glacial–interglacial CH4 budget, but they were unable to separate the relative contributions from CH4 sources that are enriched in heavier isotopes (biomass burning and natural geologic emissions). With improved estimates of natural geologic emissions, our results allow for better constraints on the overall CH4 budget. We used the stable isotope data (Fig. 1, C and D) in a one-box model (see section 5 of the materials and methods) (20) to calculate CH4 emissions from biomass burning (CH4 bb) and microbial sources (CH4 mic, composed of emissions from wetlands, ruminants, and termites) for the Early Holocene (Table 1 and fig. S11) (20). We extended our calculation to the late Holocene (~2 ka BP) (Table 1) to directly compare our CH4 source strength estimates with those of earlier studies (30, 31). This assumption can be justified because a large change in the natural geologic emissions between the early Holocene and 2 ka BP seems unlikely because global sea level and ice volume did not change appreciably after 8 ka BP (32). However, we did not perform this calculation for the pre-Holocene samples because estimates of the CH4 interpolar difference, atmospheric global average CH4 stable isotope values, and stable isotopic signatures of the sources are more uncertain (Section 5) (20).

We calculated relatively high CH4 bb emissions in the early Holocene (33 to 56 Tg CH4 per year, 95% CI) at 10 ka BP and a slight decrease of CH4 bb emissions (22 to 42 Tg CH4 per year, 95% CI) toward the late Holocene (Table 1). However, the magnitude of the decrease in biomass-burning emissions (~7 Tg CH4 per year) is small relative to the uncertainties for both the CH4 bb and CH4 mic emissions (±11 and ±18 Tg CH4 per year, respectively, 95% CI uncertainties). Our estimate of 22 to 42 Tg CH4 per year (95% CI) CH4 bb emissions for the late Holocene period (~2 ka BP) is within the upper range of estimates from previous ice core studies (13, 30, 31). Considering the large downward revision of natural geologic emissions inferred from our 14C data, an upward revision in pyrogenic CH4 emissions is expected to balance the CH4 stable isotope budget. The increase in CH4 bb expected from a reduction in natural geologic emissions is partly offset by a –0.5 to –1‰ revision in atmospheric δ13CH4 values (12, 30, 31) because the δ13CH4 values from earlier studies (30, 31) were likely biased because of krypton (Kr) interference (33). Our CH4 bb estimates are also reduced because, unlike previous studies, we accounted for temporal shifts in the isotopic signatures of CH4 bb and CH4 mic between the pre-Industrial Holocene and the modern period expected from anthropogenically driven changes in the δ13CO2 precursor material and land use (see section 5.2 of the materials and methods) (20). Our best CH4 bb estimates for the late Holocene (22 to 42 Tg CH4 per year, 95% CI) are comparable to the present-day estimates of combined pyrogenic CH4 emissions from anthropogenic biomass burning and wildfires (2). This result is supported by some (34, 35), but not all (36), independent paleoproxies of biomass burning.

The last deglaciation serves only as a partial analog to current anthropogenic warming, with the most important differences being the much colder baseline temperature, lower sea level, and the presence of large ice sheets covering a large part of what are currently permafrost regions in the NH. Although Arctic temperatures during the peak early Holocene warmth were likely warmer than today (25), they were still lower than the Arctic temperature projections by the end of this century under most warming scenarios (37). However, there are also many similarities between the last deglaciation and current anthropogenic warming. Both deglacial and modern warming include strong Arctic amplification, and the magnitude of global warming (~4°C) (16) during the last deglaciation was comparable to the expected magnitude of equilibrium global temperature change under midrange anthropogenic emission scenarios (37). Because the relatively large global warming of the last deglaciation (which included periods of large and rapid regional warming in the high latitudes) did not trigger CH4 emissions from old carbon reservoirs, such CH4 emissions in response to anthropogenic warming also appear to be unlikely. Our results instead support the hypothesis that natural CH4 emissions involving contemporaneous carbon from wetlands are likely to increase as warming continues (38). We also estimated relatively high CH4 bb emissions for the pre-Industrial Holocene that were comparable to present-day combined pyrogenic CH4 emissions from natural and anthropogenic sources. This result suggests either an underestimation of present-day CH4 bb or a two-way anthropogenic influence on fire activity during the Industrial Revolution: reduction in wildfires from active fire suppression and landscape fragmentation balanced by increased fire emissions from land-use change (deforestation) and traditional biofuel use (burning of plant materials for cooking and heating).

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S12

Tables S1 to S11

References (4197)

References and Notes

  1. Materials and methods are available as supplementary materials.
Acknowledgments: We thank M. Jayred and J. Jetson for ice drilling; camp managers K. Schroeder and C. Llewelyn and field team members A. Palardy and J. Ward for assistance; the U.S. Antarctic Program for logistical support; M. Sigl and F. Adolphi for assistance with age-scale transfer between IntCal13 and WD2014; H. Schaefer for suggestions regarding the isotope box model; G. Mollenhauer for discussions regarding the 14C-age of terrigenous biomarkers; and P. F. Place Jr., M. Diaz, and M. Paccico for help with field gear preparations. Funding: This work was supported by NSF awards PLR-1245659 (V.V.P.), PLR-1245821 (E.J.B), and PLR-1246148 (J.P.S.); the Packard Fellowship for Science and Engineering (V.V.P.); the European Research Council (ERC) under the European Union’s Seventh Framework Programme FP7/2007-2013 ERC Grant 226172 [ERC Advanced Grant Modern Approaches to Temperature Reconstructions in Polar Ice Cores (MATRICs); H.F.]; the Swiss National Science Foundation 200020_172506 (H.F.); Australian Government for the Centre for Accelerator Science at ANSTO through the National Collaborative Research Infrastructure Strategy (A.M.S, Q.H., B.Y.); and the National Institute of Water and Atmospheric Research through the Greenhouse Gases, Emissions and Carbon Cycle Science Program (P.S.). Author contributions: V.V.P., E.J.B., and J.P.S. designed the study. M.N.D., V.V.P., and B.H. conducted field logistical preparations. M.N.D., V.V.P., J.A.M., S.A.S., B.H., I.V., D.B., T.K.B., P.S., E.J.B., J.P.S., and R.H.R. conducted the field sampling and reconnaissance. M.N.D. extracted CH4 and CO from air samples. Q.H. and B.Y. graphitized the 14C samples. A.M.S. conducted the 14C measurements. J.B. and B.S. made the CH4 stable isotopes measurements (δ13C and δD-CH4) under the supervision of J.S., M.B., and H.F. D.B. and J.P.S. made the Xe/Kr, Kr/N2, and Xe/N2 measurements. R.B. made the δ15N2, δ18Oatm, 40Ar/36Ar, O2/N2, and Ar/N2 measurements. C.H. made the CH4 mole fraction and halogenated trace gas measurements under the supervision of R.F.W. M.K. made the CH4 mole fraction and total air content measurements on the ice samples under the supervision of E.J.B. I.V. made the δ13CO measurement for the CO dilution gas. S.A.S and M.N.D. developed the age scale for the samples. M.N.D. and V.V.P. analyzed the results and wrote the manuscript with input from all authors. Competing interests: The authors declare no competing interests. Data and materials availability: Data from this work are available through the USAP Data Center:

Stay Connected to Science

Navigate This Article