Report

Ice Record of δ13C for Atmospheric CH4 Across the Younger Dryas-Preboreal Transition

See allHide authors and affiliations

Science  25 Aug 2006:
Vol. 313, Issue 5790, pp. 1109-1112
DOI: 10.1126/science.1126562

Abstract

We report atmospheric methane carbon isotope ratios (δ13CH4) from the Western Greenland ice margin spanning the Younger Dryas–to–Preboreal (YD-PB) transition. Over the recorded ∼800 years, δ13CH4 was around –46 per mil (‰); that is, ∼1‰ higher than in the modern atmosphere and ∼5.5‰ higher than would be expected from budgets without 13C-rich anthropogenic emissions. This requires higher natural 13C-rich emissions or stronger sink fractionation than conventionally assumed. Constant δ13CH4 during the rise in methane concentration at the YD-PB transition is consistent with additional emissions from tropical wetlands, or aerobic plant CH4 production, or with a multisource scenario. A marine clathrate source is unlikely.

Ice core records reveal prominent changes in atmospheric methane concentration [CH4] associated with abrupt climate change (1) but the causes, including source and sink changes, remain controversial (1, 2). Modern contributions from individual sources or sinks have been constrained by the 13C/12C ratio of atmospheric methane (δ13CH4) (3, 4). New analytical techniques extend this approach to air samples from gas occlusions in polar ice. Using ice samples from the Pakitsoq outcrop (Western Greenland) (5), we measured δ13CH4 in air dating between 11,360 and 12,220 years before the present (yr B.P.) (6). The record covers the transition between the Younger Dryas (YD) and Preboreal Holocene (PB), when temperature (7) and [CH4] (1) increased rapidly at the termination of the last ice age (Fig. 1).

Fig. 1.

Methane concentration and δ13CH4 in Pakitsoq ice samples. (A) [CH4] measured at the University of Victoria (V-2001 to V-2003), at NIWA (N-2003 and N-2004), and from the GISP2 core, together with [CH4] model results used to constrain the transition source (8). The age spread of the samples is indicated by symbol size (V samples, 25 to 35 years) or horizontal bars (N samples, 100 to 240 years). The bottom axis shows sample age according to the Pakitsoq gas age scale (8); the top axis shows sample location within the profile (nonlinearity between the axes is due to differential thinning of ice layers). In (A) to (C), vertical shading marks the transition period. (B) Pakitsoq and GISP2 δ15N data (5, 7) show the onset of atmospheric warming (11,570 yr B.P.). (C) δ13CH4 versus sample age. The slight depression of modeled δ13CH4 (8) during the transition (∼0.1‰) is due to a temporary imbalance between 13C-depleted emissions and 13C-enriching sinks (22). Data have been corrected for gravitational, thermal, and diffusion fractionation (8). Colored error bars show the standard error of multiple samples. Black error bars on single measurements show analytical precision (±1σ) derived from standard ice [0.44‰ (n = 11 replicates) in 2001, 0.37‰ (n = 33) in 2002, and 0.48‰ (n = 12) in 2003].

The suitability of Pakitsoq ice for paleostudies has been demonstrated by the agreement of [CH4] and other geochemical tracers with records from the Greenland Ice Sheet Project 2 (GISP2) ice core (5). Samples were collected during three field campaigns (2001 to 2003) by means of oil-free chainsaws and shipped frozen to the University of Victoria. The main data set was measured after wet extraction by gas chromatography–isotope ratio mass spectrometry (GC-IRMS) (8). [CH4] measurements were duplicated at Washington State University. Six samples from three time periods were analyzed for δ13CH4 at the National Institute of Water and Atmospheric Research (NIWA) using ∼100-liter air samples extracted in the field (8) (Fig. 1). All samples were dated with a gas age scale derived by comparison of geochemical records from Pakitsoq and GISP2 (8). Results are consistent throughout the three field seasons and form a composite data set (Fig. 1).

Our δ13CH4 data reveal several interesting features, two of which we discuss here. First, the YD-PB methane is 13C-enriched by ∼1 per mil (‰) relative to modern atmospheric δ13CH4 (–47.1‰)(9) and ∼5.5‰ higher than expected from previously proposed natural CH4 budget scenarios (table S1) (3, 10). Second, there is no significant change in δ13CH4 across the YD-PB transition. In the Pakitsoq record, [CH4] rises from 490 to 750 parts per billion by volume (ppbv) at the transition, which is consistent with GISP2 data (1, 5) (Fig. 1A). During the YD, δ13CH4 has a mean of –46.0 ± 0.5‰ (1σ) (Fig. 1C). Slight variations fall within the envelope of uncertainty. The PB mean δ13CH4 is –45.7 ± 1.2‰. Surprisingly, there is no significant difference in δ13CH4 between the two climatic intervals, nor is there an isotope shift during the ∼250-ppbv [CH4] increase.

Pre-anthropogenic δ13CH4 was expected to be depleted in 13C (relative to modern atmospheric CH4) because of the absence of fossil fuel combustion, slash-and-burn agriculture, and landfills, all of which emit 13C-enriched CH4 (3). Such 13C depletions are observed in ice from 100 to 300 yr B.P. (1113), whereas 400 to 2000 yr B.P. values from the late preindustrial Holocene (LPIH) are unexpectedly 13C-enriched, similar to our YD-PB δ13CH4 (12).

An initial explanation for our high δ13CH4 was enrichment during postocclusion microbial oxidation of CH4; that is, a storage artifact. However, this is ruled out because the amount of oxidation required for the observed isotopic shift would be between 15 and 48% (8). This would be readily detected as discrepancies between the Pakitsoq and GISP2 [CH4] records (Fig. 1A).

The difference between δ13CH4 measured at the YD-PB and LPIH is not an artifact and must result from changes in CH4 sources or sinks. The multitude of variables affecting atmospheric δ13CH4 and uncertainties in paleoenvironmental data make it difficult to reconstruct a definitive CH4 budget for the YD-PB. However, we discuss five possible explanations for the 13C enrichment.

First, biomass burning is the most 13C-enriched (δ13CH4 ∼–25‰) of all sources (3). For the LPIH, elevated pyrogenic emissions of 25 Tg/year have been inferred from ice core δ13CH4 (12). Even with fire emissions of this magnitude, the YD isotope scenario would still be too 13C-depleted (table S1), whereas charcoal records indicate less burning before the LPIH (14). Nevertheless, literature estimates (10) of late glacial fire emissions (5 Tg/year) may be too low, and pyrogenic CH4 could possibly contribute to 13C enrichment. In the LPIH, high δ13CH4 values have been partly attributed to human-made fires (12). Interestingly, we observe even higher 13CH4 values in the YD-PB, when anthropogenic burning is expected to have been negligible.

Second, geologic (natural thermogenic) sources are usually not included in CH4 budgets, even though they emit enough 13C-enriched methane today to significantly affect atmospheric δ13CH4 (15). Because of lower overall emissions during the YD, geologic sources probably constituted a larger fraction of the budget than today, leading to higher atmospheric δ13CH4. Additionally, lower YD sea levels may have increased this source (16).

Third, previous CH4 budget calculations have grouped tropical wetlands with other wetlands, resulting in integrated wetland δ13CH4 of –58 to –59‰ (3, 4). However, for tropical wetlands alone, flux-weighted mean δ13CH4 as high as –53 to –55‰ has been reported (17). We suggest that tropical wetlands must be treated separately from boreal and temperate wetlands because of differences in climatic response and δ13CH4. Tropical wetlands could partly account for the measured 13C enrichment (table S1).

Fourth, the recent discovery of aerobic methane production (AMP) from plant material (18) has revealed a previously unknown source with high δ13CH4 (∼–50‰). At estimated emission rates of 150 ± 90 Tg/year (18), it may be as important for the natural CH4 cycle as wetlands (∼140 Tg/year) (19) and could contribute to higher atmospheric δ13CH4.

Fifth, there is evidence that Cl in the marine boundary layer (MBL) acts as a CH4 sink, with a large isotope effect that increases atmospheric δ13CH4 (20).

The potential impact of these five processes on δ13CH4 can be estimated with isotope mass balance calculations. Although we recognize the considerable uncertainty introduced by the range of reported values and incomplete knowledge of YD conditions, we estimate that none of these processes alone can explain the 13C enrichment. Conversely, the cumulative effect of all five would be too large (table S1). Several processes together could balance the YD isotope budget, but the exact combination cannot be determined without further evidence.

For completeness, one must also consider changes in the ratio of C3 to C4 vegetation, temperature-dependent fractionation coefficients, the relative amounts of CH4 oxidation and production in wetland soils, and weighted total fractionation (αWT) of the overall sink (21). These all influence atmospheric δ13CH4, depending on climatic and anthropogenic factors that were probably different during the YD than today. Preliminary studies lead us to assume that neither the individual (±1‰), nor the combined impact of such changes (13C depletion of 0.7‰), accounts for the δ13CH4 enrichment.

A primary objective of our study is to understand why [CH4] increased abruptly during the YD termination. This rise was not associated with a sustained shift between steady states or with an episodic δ13CH4 excursion (within measurement error) (Fig. 1). The latter observation suggests that the [CH4] increase was not caused by a short-lived perturbation, such as a release of stored CH4.

Several lines of evidence suggest that increased emissions triggered by climate change caused the [CH4] rise (1, 2). We used an atmospheric box model (22) to find the δ13CH4 of these additional emissions or “transition source” required to explain our measured [CH4] and δ13CH4 histories (8) (Fig. 1). Results depend on whether a MBL sink is considered and other uncertainties in αWT. The model predicts a transition source δ13CH4 between –53.6 ± 1.5‰ (including a MBL sink) and –50.7 ± 0.7‰ (without a MBL sink) (23).

The transition source could have been a combination of two or more sources. If those had different δ13CH4 values, then their relative emissions would have to fortuitously match in order not to affect atmospheric δ13CH4. For example, a 13C-enriching component such as wildfire CH4 or the marine Cl sink could be balanced by a 13C-depleted microbial source. In the absence of unequivocal geologic evidence, these scenarios remain unresolved.

A sink decrease caused by higher volatile organic carbon emissions from forests (24) could have increased [CH4] without significant impact on αWT and δ13CH4. This scenario requires a large expansion of forests on the short time scale of the [CH4] increase. This may be compatible with paleobotanical data (25, 26), but seems less likely when rates of ecosystem reorganization are considered. However, a slight sink decrease caused by feedbacks in atmospheric chemistry due to higher [CH4] is probable (27).

Most microbial CH4 sources, such as boreal and temperate wetlands, animals, and clathrates, have δ13CH4 of –60‰ or lower (3). The calculated impact of each of these sources on the transition mass balance would be –1.3 to –2.2‰ (depending on the assumed sink configuration). This exceeds the least significant difference of 1.1‰ between the YD and PB that would be discernible from our data at the 90% confidence level (based on a two-sided t test result of ±0.54‰). Therefore, microbial CH4 can be ruled out as a sole transition source. However, CH4 from marine clathrates can become 13Cenriched by microbial oxidation as it migrates through sediment and water columns (28). Isotopic fractionation associated with the oxidation progressively enriches the remaining CH4 in 13C relative to the clathrate. In order to match the transition source δ13CH4, only 30 to 40% of the gas initially released from clathrates could have reached the atmosphere, according to the Rayleigh equation (8).

It has been proposed that marine clathrates drove the YD-PB [CH4] rise and maintained high [CH4] until mature wetlands developed (∼8000 yr B.P.) (2). The total amount of clathrate dissociation required by this scenario can be calculated from (i) the magnitude of additional emissions (62 × 1012 g/year) (1), (ii) the fact that the latter represent only 30 to 40% of the destabilized clathrate gas, and (iii) the postulated time scale (∼3000 years) (2), giving a result of 410 × 1015 to 540 × 1015 g of dissociated clathrate carbon. For comparison, only 175 × 1015 g potentially became unstable within the last 100,000-year glacial cycle (29), a period with more than 20 abrupt [CH4] rises (1). If clathrate release had driven all 20, then the potentially unstable reservoir for each event would have been exhausted within only ∼50 years. In conclusion, our δ13CH4 record supports neither catastrophic nor gradual clathrate emissions at the YD-PB transition, as is also indicated by CH4 deuterium (δD-CH4) records (30).

Hydrological proxies point to tropical wetlands as a driver of the YD-PB [CH4] increase (31, 32). The higher of the δ13CH4 values (–53 to –55‰) reported for tropical wetlands (17) closely matches the modeled transition source (especially if the MBL sink is included). The impact of tropical wetlands (0.1 to 0.7‰) on the mass balance during the transition would not be detectable. However, controversy remains about whether wetlands were a source of sufficient magnitude (19) and responded quickly enough to climate change (2).

In contrast, vegetation-derived AMP (18) could have changed quickly and strongly as indicated by paleoreconstructions of net primary productivity (NPP) (33). Scaling emission estimates (18) to NPP results in a potential increase of AMP by 50% between glacial and interglacial conditions. Terrestrial carbon stocks of vegetation, a possible proxy for AMP, reach equilibrium after climatic change within 250 years (25), and a first vegetation response occurs within decades (26). These time scales compare well with the observed lag time and duration of the YD-PB [CH4] increase (7). Also, the match between transition-source δ13CH4 in the non-MBL sink scenario and that of AMP is good. In that case, the mass balance impact would be 0.7‰, which is within the uncertainty of our data. Whether the vegetation response was sufficient to sustain the [CH4] increase at the YD-PB should be investigated with vegetation models, once estimates of emission rates are confirmed. Our δ13CH4 data are therefore consistent with a fast-responding AMP source and with the hypothesis that low-latitude wetlands caused [CH4] rises during the last glacial cycle (1).

Further insights into the atmospheric CH4 budget require a reduction in the uncertainties of methane source and sink strengths and signals, and the combination of higher-precision δ13CH4 and δD-CH4 data.

Supporting Online Material

www.sciencemag.org/cgi/content/full/313/5790/1109/DC1

Materials and Methods

Table S1

References

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article