Atmospheric Methane and Nitrous Oxide of the Late Pleistocene from Antarctic Ice Cores

See allHide authors and affiliations

Science  25 Nov 2005:
Vol. 310, Issue 5752, pp. 1317-1321
DOI: 10.1126/science.1120132


The European Project for Ice Coring in Antarctica Dome C ice core enables us to extend existing records of atmospheric methane (CH4) and nitrous oxide (N2O) back to 650,000 years before the present. A combined record of CH4 measured along the Dome C and the Vostok ice cores demonstrates, within the resolution of our measurements, that preindustrial concentrations over Antarctica have not exceeded 773 ± 15 ppbv (parts per billion by volume) during the past 650,000 years. Before 420,000 years ago, when interglacials were cooler, maximum CH4 concentrations were only about 600 ppbv, similar to lower Holocene values. In contrast, the N2O record shows maximum concentrations of 278 ± 7 ppbv, slightly higher than early Holocene values.

Earth's climate during the late Pleistocene was characterized by ice age cycles with relatively short warm periods (interglacials) and longer cold periods (glacials) (1). The Vostok ice core provided an archive of climate and atmospheric composition over the past four climatic cycles back to marine isotope stage (MIS) 11, about 420 thousand years before the present (420 kyr B.P.) (2). That record demonstrated the high correlation of temperature changes with greenhouse gas concentration changes in the atmosphere in the past. The European Project for Ice Coring in Antarctica (EPICA) Dome Concordia (Dome C) ice core (75°06′S, 123°21′E, 3233 m above sea level) provides an ice core archive much longer, spanning eight climatic cycles over the past 740 thousand years (ky) (3). It demonstrates that the oldest four interglacials were cooler but lasted longer than the younger interglacials. Such findings raise the question whether the greenhouse gases CH4 and N2O behaved differently before MIS 11. Here, we present CH4 and N2O records derived from the EPICA Dome C ice cores reaching back to 650 kyr B.P.

Extracting air from bubbles trapped in polar ice enables us to reconstruct directly the past composition of the atmosphere. The records of CH4 and N2O over the past thousand years (46), the Holocene (7, 8), the transition from the last glacial maximum (LGM) to the Holocene (6, 9), parts of the last glacial period (1014), and the last four glacial-interglacial cycles (2, 1517) have provided important information about environmental changes in response to regional and global climate variations. Variations in the concentration of globally well-mixed atmospheric CH4 are attributed to variations in the extent and the productivity of natural wetlands, the main natural sources (18). Similarly, variations in the atmospheric N2O burden are thought to be dominated by variations in the global source strength. About two-thirds of the total preanthropogenic N2O sources are terrestrial soils, and one-third are nitrification and denitrification processes in the ocean (13).

We present high-resolution records of CH4 and N2O covering MISs 2 to 7 (Fig. 1), to compare with existing records, and MISs 11 to 16 (Fig. 2), to extend the records back to an age of 650 kyr B.P. Measurements were performed along the EPICA Dome C ice cores (EDC96 and EDC99) at 553 and 476 different depth levels for CH4 and N2O, respectively, at the University of Bern and at Laboratoire de Glaciologie et Géophysique de l'Environnement (LGGE) (19). The mean CH4 time resolution is 770 years for MISs 2 to 7 and 840 years for MISs 11 to 16. All ages and their uncertainties are based on the EPICA EDC2 gas and ice age scales (3). N2O analyses were performed only at Bern. Their resolution is similar for MISs 2 to 7 (760 years) but lower for MISs 12 to 16 (1110 years). To be consistent with existing Dome C data, offset corrections for both gases and both laboratories have been applied (19).

Fig. 1.

Dome C CH4 (purple line), N2O (red line), and δD (black line) (3) records over the past 220 ky. Also shown are Vostok CH4 data (blue line) (2, 15) and δD data (gray line) (2). Vostok δD data were increased by 42‰ for better comparison. The Vostok CH4 and δD records were individually synchronized to Dome C using wiggle matching (35). N2O artifacts (open red diamonds) are separated from the N2O record (red filled diamonds) with the aid of the dust record (figs. S1 to S3). Additionally, N2O records published earlier (gray lines and crosses) (6, 13, 17), matched by CH4 to Dome C, are shown for comparison. Error bars represent the 1 SD measurement uncertainty. Dome C CH4 and N2O data over the Holocene (8), the transition (9, 36), and the LGM (20, 36) are included in this data set. Dome C measurements covering the time period 0 to 40 kyr B.P. (depth interval 99.5 to 783.8 m) were performed along the EDC96 core, whereas older samples are from the EDC99 core. Numbers 0 to 25 above the CH4 record denote D/O events (23), and A1 to A7 denote Antarctic warming events (37). Data are plotted on the EDC2 time scale (3).

Fig. 2.

Dome C CH4 (purple line) and N2O (red line) records over the period of MISs 11 to 16, together with high-resolution δD data from Dome C (24) (black line). CH4 measurements performed at Bern are shown as purple diamonds, whereas LGGE measurements are shown as purple circles. N2O artifacts (red open diamonds) are separated from the N2O record (red filled diamonds) with the aid of the dust record (figs. S1 to S3). Error bars represent the 1 SD measurement uncertainty. Dome C CH4 data over termination V (3) is included in this data set. Parts of the CH4 record marked with a star (*) are regular events whose amplitude could be strongly influenced by the precession signal as seen in low-to mid-northern latitude summer insolation. MISs are labeled on the bottom (cold stages) and top (warm stages) of the δD record. Data are shown on the EDC2 time scale (3).

The Dome C N2O record is disturbed by artifacts in certain depth intervals as in other ice cores (6, 13, 17, 20). High scattering of N2O values is observed in depth intervals with elevated dust concentrations (19) (figs. S1 and S2) during parts of the cold periods MISs 2, 4, 6, 12, 14, and 16 (fig. S3). The EPICA Dome C dust record (19) is used to define depth intervals where the dust concentration exceeds 300 ppbw (parts per billion by weight). N2O measurements in these depths are considered to be disturbed by artifacts and excluded from the record (fig. S3). The good agreement of the remaining record with records measured along other ice cores (Fig. 1) with different characteristics in the concentration of chemical impurities (6, 13, 14, 17) supports the assumption that this record shows atmospheric concentrations of the past. Although all samples might be contaminated at some level, no outliers are observed during the highly resolved Holocene. This supports our assumption that other interglacials are also likely free from artifacts.

The Dome C CH4 data over the last two glacial cycles (Fig. 1) are in good agreement with the Vostok CH4 record (15, 19), as well as Greenland CH4 records, when taking into account the interpolar difference (7, 21) and the signal attenuation at low accumulation sites (22). The Dome C CH4 data over the last glacial period confirm the close relation with the 25 Dansgaard/Oeschger (D/O) events as recorded in the North Greenland Ice-Core Project (NGRIP) temperature proxies (23). In addition, we find that N2O also varied during the last glacial period in parallel with the most prominent D/O events (Fig. 1), in agreement with previous results (6, 13, 14, 16). N2O reaches Holocene concentrations of 253 to 272 ppbv (parts per billion by volume) (single values) (8) during the maximum of some of these events and during the last interglacial (14, 17). The lowest values found, about 200 ppbv, are similar to those in records published earlier (13, 14). Because the CH4 and N2O measurements were performed on the same extracted air samples, a direct comparison without relative time uncertainty is possible between the two gases. The start and end points of single D/O events in CH4 and N2O do not necessarily coincide (13); e.g., the N2O concentrations at D/O event 21 start to rise more than 1000 years earlier than CH4 concentrations. Furthermore, N2O remains at interglacial values for this specific event, when CH4 has already dropped to glacial values at the end of the event. The amplitude of some events differs substantially between the two gases as demonstrated by D/O events 19 and 20, also shown in the GRIP record (13) (Fig. 1).

The depth interval from 2700 to 3060 m in the EPICA Dome C core permits us to make reconstructions of climate and atmospheric composition for the interval 390 to 650 kyr B.P., most of which precedes the Vostok record (2). CH4 and N2O measurements performed over this depth interval are shown in Fig. 2, together with high-resolution δD measurements (24).

To better characterize warm and cold intervals during the past 650 ky, we refer to “interglacial” as a period during which δD exceeds –403‰ (per mil) (3). The use of this definition marks MISs 1, 5.5, 7.3, 7.5, 9.3, and 11.3 as interglacials, which is consistent with the marine records (1). In the following, we describe the individual counterparts of the marine stages (25) from MISs 16.2 to 11.2.

At 624 kyr B.P., at the end of MIS 16.2, CH4 falls to a value of 368 ppbv (mean over 7 ky), comparable to the low CH4 concentration during MIS 6 and MIS 2 in the Dome C record. The increase (termination VII) into MIS 15.5 by about 250 ppbv is smaller than at the younger terminations I, II, IV, and V (15). While CH4 only reaches the mean Holocene (0.3 to 10 kyr B.P.) level of 608 ppbv during MIS 15.1 (600 ppbv, mean over 27 ky) and 15.5 (617 ppbv, mean over 7 ky), the corresponding mean N2O concentrations of 272 ppbv and 274 ppbv are higher than the Holocene mean (262 ppbv) but similar to early Holocene values. Both records, CH4 and N2O, show the distinct separation of MISs 15.1 and 15.5 (Fig. 2) by lower concentrations and three shorter events each during the colder near-glacial periods of MISs 15.2 to 15.4. MIS 15.1 is a surprisingly stable climatic period (1 SD: ± 15 ppbv CH4, ± 6 ppbv N2O). Its duration is about 27 ky on the EDC2 age scale. A similar duration is seen in the records of δD (24) and CO2 (26); related uncertainties are discussed in (26).

In Fig. 3A, δD is plotted together with CO2 (26), CH4, and N2O, showing the transition from MISs 14 to 13. During MIS 14, the three gases are closely linked in terms of their timing: Relative maxima (dashed lines) occur at the same time within the data resolution. One of these maxima corresponds to MIS 14.3. CO2, CH4, and N2O seem to lead the δD record and therefore, Antarctic temperature, by about 2000 ± 500 years. However, this could be due to an artifact in the EDC2 time scale, as discussed in Siegenthaler et al. (26). During MIS 13, the relative CO2, CH4, and N2O maxima are slightly out of phase. In contrast to younger glacial terminations (2), the temperatures do not rise directly to a local maximum. Instead, the δD, CO2, and CH4 records reveal a transition to the warmest phase (MIS 13.1), which evolves in steps with relapses in between. Nevertheless, sequences of the CO2 and CH4 increases to MISs 13.3 and 13.1 look similar to those of the last deglaciation (9): CO2 rises first, CH4 is delayed, and CO2 stops rising when CH4 rises abruptly. In comparison to the last deglaciation, the CH4 rise at 513 kyr B.P. is rather small. This may be explained by the colder interglacial temperatures that would have favored a relative increase in methanotrophy over methanogenesis, thereby contributing to lowered net CH4 emissions in the northern extratropics and/or the subtropical and tropical regions. Additionally, the Northern Hemisphere ice sheets may not have retreated as far north during MIS 13. As a consequence, the uncovered areal extent of additional CH4 source regions was smaller and therefore, the CH4 overshoot, usually found at terminations, may not have occurred.

Fig. 3.

(A) Dome C CO2 (26) (green line), CH4 (purple line), and N2O (red line) records over the period of MISs 13 to 14 as shown in Fig. 2, together with high-resolution δD data from Dome C (24) (black line). Vertical dashed lines highlight the coincidence of local CO2, CH4, and N2O maxima. Data are plotted on the EDC2 time scale (3). (B) CH4 records from Dome C (purple line) and Vostok (15, 30) (blue line), together with high-resolution δDdatafrom Dome C (24) (black line) covering MIS 11 (labeled at the bottom). The data from MIS 1 are shown as gray lines (time axis on top) for comparison; the curves are aligned by synchronizing the two glacial-interglacial CH increases. Error bars represent the 1 SD measurement uncertainty. Data are plotted on the EDC2 age scale (3).

At this point, it is difficult to characterize MISs 13 and 15 as interglacials, because the δD values barely exceed –403‰. We therefore refer to these periods as intermediate warm periods (IWP) (27). This and their specific evolution of CH4 suggests that they are similar to MISs 5.1 to 5.3 and MISs 7.1 to 7.3 (2) (Fig. 4). In all cases, regular events with CH4 amplitudes of 50 to 140 ppbv occurred, similar to events marked by stars in Fig. 2. The average return time of the events during MIS 13 is approximately 20 ky, and they last for 5 to 10 ky. The origin of these fluctuations is probably related to the precession cycle (19 to 23 ky) of summer insolation in mid-low to low northern latitudes, as observed for the CH4 amplitude during the last glacial period (13). The N2O amplitude appears to be modulated by different mechanisms from those responsible for CH4, as seen in MISs 13 and 15 (e.g., at 498 and 596 kyr B.P.), where N2O peaks are quite large despite the small CH4 peaks. Similar discrepancies are in fact found in the last glacial period, e.g., at D/O events 19 and 20 (13).

Fig. 4.

CH4 record over the past 650 ky, composed of Dome C CH4 (purple line) [(8, 9) and new data] and Vostok CH4 (blue line) (2, 15). Also shown are the N2O data measured along the Dome C ice cores (red line) [(8, 20, 36) and new data] and δD records from Dome C (black line) (24) as well as those from Vostok +42‰ (gray line) (2). N2O artifacts are not shown in this figure. Gray shaded areas highlight interglacial periods with a δD value >–403‰ as defined in (3). Numbers of MISs are given at the bottom of the figure (25). Data are shown on the EDC2 time scale (3).

MIS 12 is believed to be a very cold period showing millennial time-scale variations similar to those observed in the last glacial period (28). The Dome C record shows CH4 variations lasting about 1 to 3 ky, with amplitudes of 40 to 120 ppbv (Fig. 2). In addition, Antarctic warming events of 7 to 19‰ are present in the δD record. Their independent evolution on millennial time scales is an indication that the bipolar seesaw may have been operational also during MIS 12 (29), but the resolution of our measurements and the uncertainty in their timing prevent us from drawing final conclusions.

Our CH4 measurements provide an undisturbed record for the entire MIS 11 (Fig. 3B) and support, except for a small offset (19), the recently reconstructed MIS 11 record of Vostok (30). This is important for two reasons. First, MIS 11 is the longest warm phase of the Antarctic temperature record over the past 740 ky (24), with a mean temperature comparable to the Holocene. CH4 exceeded the minimum Holocene value of 560 ppbv for more than 28 ky. Second, CH4 increased to 689 ppbv at termination V, followed by an early decrease for about 5000 years, which is very similar to the early Holocene CH4 record (gray curves in Fig. 3B). It then rose to the high concentration of about 700 ppbv. Ruddiman (31) argued that such a behavior of the CH4 during the early Holocene is due to early anthropogenic interference and that this would be unique. Our data provide a crucial counterexample to this postulated behavior: The early temporary reduction was clearly not an indication of an impending ice age, which started some 20,000 years later in MIS 11, and the increase after this reduction has been established without human influence.

The good agreement between the Dome C and the Vostok (15, 30) record over MISs 1 to 7 and MIS 11 increases our confidence in the fidelity of a composite record of atmospheric CH4 over the past 650 ky (Fig. 4). The composite record, established by wiggle matching (32), demonstrates, within the resolution of our measurements, that preindustrial atmospheric concentrations of CH4 reconstructed from two Antarctic ice cores have not exceeded 773 ± 15 ppbv (single values at MIS 9.3) during the past 650 ky. The highest mean N2O values (278 ppbv ± 7 ppbv) over at least a period of 7 ky covered by our records are found during MIS 15.1, slightly higher than early Holocene values. The differing response to climate change of CH4 and N2O during MISs 15.1 and 15.5 suggests that the main source regions and/or strengths may strongly differ during interglacials and IWP. High levels are sustained longer for N2O than for CH4 during MISs 5.5, 15.1, and 15.3 (Fig. 4), either due to the release of oceanic N2O or because N2O soil sources are also productive under semiarid conditions (13).

In general, CH4 is well correlated with δD on glacial-interglacial time scales (>40 ky), including lower CH4 concentrations during the IWP than during interglacials of the past 420 ky. At terminations, the amplitude of the CH4 increase is highly correlated with the corresponding Antarctic temperature (r2 = 0.80). CH4 has mainly tropical and Northern Hemisphere sources, but only very small austral sources (mainly open-ocean contributions), which suggests that the generally smaller glacial-interglacial temperature change before 440 kyr B.P. revealed by Dome C δD was also of global importance at that time (33). The precessional insolation signal in low- to mid-northern latitudes has a strong influence on the amplitude of CH4 variations for glacial periods and IWP, but the Northern wetlands, episodically covered by boreal ice sheets and exposed to different precipitation patterns, are likely to contribute to this D/O-like variability of the CH4 concentration (21). Furthermore, sink feedbacks, through changes of hydrocarbon emissions from the terrestrial biosphere, could have been considerable (34).

Supporting Online Material

Materials and Methods

Figs. S1 to S3


References and Notes

View Abstract

Navigate This Article