Ice Core Records of Atmospheric N2O Covering the Last 106,000 Years

See allHide authors and affiliations

Science  15 Aug 2003:
Vol. 301, Issue 5635, pp. 945-948
DOI: 10.1126/science.1085293


Paleoatmospheric records of trace-gas concentrations recovered from ice cores provide important sources of information on many biogeochemical cycles involving carbon, nitrogen, and oxygen. Here, we present a 106,000-year record of atmospheric nitrous oxide (N2O) along with corresponding isotopic records spanning the last 30,000 years, which together suggest minimal changes in the ratio of marine to terrestrial N2O production. During the last glacial termination, both marine and oceanic N2O emissions increased by 40 ± 8%. We speculate that our records do not support those hypotheses that invoke enhanced export production to explain low carbon dioxide values during glacial periods.

Nitrous oxide (N2O) is an important component of the atmosphere that behaves as a “greenhouse” gas and participates in many atmospheric chemical reactions involving O3, NOy, and OH. Tropospheric N2O levels are controlled by bacterial N2O emissions from the terrestrial and marine biospheres, stratospheric photolysis and photooxidation of N2O, and the stratosphere-troposphere air exchange rate (1, 2).

During the last glacial-interglacial transition, atmospheric N2O values increased from ∼190 to 265 parts per billion (ppb) (35). Model studies suggest that the lifetime of N2O in the atmosphere has remained effectively constant over the last 18,000 years (18 ky) (6, 7). This implies that the observed N2O concentration ([N2O]) variations have been the result of changes in the source strength. N2O oscillations have been documented during the cold Younger Dryas event and the one Dansgaard/Oeschger (D/O) event (number 8) that has been studied in detail. These events also have large documented fluctuations in atmospheric methane (CH4) concentration that are likely the result of changes in CH4 emissions from terrestrial wetlands. Wetland CH4 emissions increase with temperature, primary productivity, and soil saturation (810). Terrestrial N2O emissions behave similarly, but are highest for soils with between 60 and 90% water-filled pore spaces; at higher saturation, the low-redox conditions promote complete reduction of NO3 to N2 gas, whereas in drier areas NO emissions dominate (11, 12).

Here, we extend the ice core N2O record back to 106,000 years ago (106 ka) and provide atmospheric N2O isotope results spanning the last glacial termination. Concentration measurements were made on 106 samples of Greenland Ice Sheet Project 2 (GISP II), central Greenland (73°N, 38°W), ice with gas ages ranging from 0.5 to 106 ka (10) by means of a dry extraction technique (13, 14). Samples integrating fewer than 30 years of atmospheric history are spaced on average 1.1 ky apart. The overall external precision and blank associated with the technique are ±5 and 4 ± 4 ppb, respectively (1σ errors). In addition, we measured 24 samples of ice from the Taylor Dome, Antarctica, ice core (77°47′S, 158° 43′E) for [N2O], δ15N, and δ18O of N2O using a previously described technique (15). The external precision for the δ15N and δ18O measurements is ±0.5‰ and ±2‰, respectively, based on six samples of ice with gas ages between 1785 and 1820 A.D. (1σ errors). The ages of the Taylor Dome samples were previously determined by correlating the CH4 record from Taylor Dome into the GISP II record (16). All data (concentration and isotopic) have been corrected for all mass-dependent fractionation processes (including gravitational fractionation) based on δ15N of N2 data that were measured on each gas sample (17, 18).

The N2O concentration data from the GISP II and Greenland Ice Core Project (GRIP), Greenland, and Dome C, Antarctica, cores are plotted in Fig. 1 (3). The ages for GRIP N2O data have been transferred to the GISP II time scale for direct comparison (19). The N2O data sets were generated with different extraction techniques that yield very similar N2O records during overlapping periods, suggesting that, in general, the ice core records are likely to reflect paleoatmospheric N2O fluctuations within experimental uncertainty. There are two major exceptions. First, the N2O values from GISP II are low in the depth interval where air bubbles are transformed to solid clathrates [900 to 1400 m, gas ages of 4 to 8 ka (20)]. These results are readily explained by preferential incorporation of N2O into, but incomplete gas recovery from, the solid air-hydrate phase [supporting online material (SOM) text]. Second, a few N2O results from the GRIP core (11.8 and 38.9 ka) are anomalously high relative to results from neighboring samples (3). These high values are thought to be related to an unidentified artifact that may involve in situ N2O that was liberated by microbial processes (15).

Fig. 1.

Various climate records covering the later portion of the last glacial period. Top curve (black): June insolation for 20°N (44). The isotopic temperature record for Greenland is shown with gray symbols, with the resampled record denoted by the solid black line (45); the green curve is the N2O record from GISP II, with blue crosses representing N2O data from Dome C and GRIP (3, 4); the red curve is the resampled CH4 record from GISP II, with red symbols representing the original record (10); and the blue curve is the resampled CO2 record (46) based on the Bender et al. Vostok gas age model (24). Also included are the CO2 data (filled blue circles) covering the last 60 kyfrom the Taylor Dome ice core (2123). The three vertical black arrows above the N2O data represent anomalously high N2O data obtained from the GRIP core (3).

In the lower portion of Fig. 1 we plot the CO2 records from the Taylor Dome and Vostok ice cores to complete the picture of changes in the concentration of major long-lived bioactive trace-gas species over the last 106 ky (2123). We have used a Vostok gas chronology that had previously been tied to the GISP II gas chronology (24). Presently, the N2O record is of lower resolution than either the CH4 or the δ18Oice records; for comparison, we have resampled the CH4, CO2, and δ18Oice records at times for which we have N2O data. The resampled curves are shown with solid lines in Fig. 1.

In Fig. 2 we present results from 21 measurements of the δ15N and δ18O of N2O from the Taylor Dome ice core between 2.6 and 32.6 ka covering the last glacial termination and the Holocene. Unexpectedly, the isotope data show little variability over this period of rapid concentration change, despite the large differences in the characteristic isotope values for the marine and terrestrial N2O [δ15N values ∼8 ± 3‰ and –15 ± 15‰, δ18O values ∼45 ± 3‰ and 35 ± 15‰ (standard mean ocean water, SMOW) for marine and terrestrial sources, respectively] (25). The average δ15N and δ18O values from Taylor Dome over the last 33 ky are 11.1 ± 0.8‰ and 46.5 ± 1.7‰, respectively. These average values are statistically indistinguishable from the preanthropogenic (1785 to 1819 A.D.) values obtained from the GISP II core (10.1 ± 0.6‰ and 47.4 ± 1‰, number of samples n = 6) and Eemian (115 to 130 ka) values from the Vostok core (10.1 ± 1.0‰ and 45.8 ± 0.8‰, n = 5) (15, 26).

Fig. 2.

Measurements of [N2O] and δ15N and δ18O of N2O from the Taylor Dome ice core are shown in red with 1σ error bars (except [N2O]). Also included are recent isotope data from a shallow core from the GISP II site (26) representing the anthropogenic perturbation over the past two centuries. The GISP II [N2O] data from Fig. 1 are reproduced here in blue for comparison with the Taylor Dome [N2O] data. Shown for reference (top) is the isotopic temperature record from GISP II, likely representing the mean climate state of the circum–North Atlantic region. Note the small amount of variance in the Taylor Dome isotope data before the anthropogenic era. Present-day δ15N and δ18O of N2O values (8.2 and 44.5‰) are indicated with horizontal lines for reference.

The data sets in Figs. 1 and 2 can be used to test models of global biogeochemical cycles involving N2O across a variety of time scales. With one exception (the Younger Dryas), we suggest that throughout the last 33 ky the ratio of terrestrial to marine N2O emissions (T/MN2O) has remained fairly constant. We obviously cannot exclude the possibility that there were substantial changes in T/MN2O that were offset by tightly coupled changes in the characteristic isotope ratios of the sources and/or sinks that resulted in the nearly constant atmospheric values. However, a more parsimonious explanation would be that marine and terrestrial N2O emissions have varied together, as suggested by a recent modeling study of the marine N2O emissions during the Younger Dryas period (27).

We can place an upper limit on T/MN2O variability based on the variability of the N2O isotope measurements. Under steady conditions, the isotopic composition of atmospheric N2O is dictated by the fraction of each source or sink multiplied by the characteristic isotope value. Assuming that the primary sink (stratospheric photolysis), its characteristic isotopic effect, and the isotopic values of the sources have remained constant, we can calculate the maximum change in T/MN2O that is consistent with our measurements. The variability of all δ15N and δ18O values about the mean (±0.8‰ and 1.7‰, respectively) is similar to our external analytical precision [±0.5‰ and ±2‰ (15)]. This implies that the calculated change in T/MN2O based on the measured isotope variability is likely to be larger than the true variability. The δ15N data are more precise and therefore provide the stronger constraint, indicating that T/MN2O varied by less than 16% throughout the last 33 ky.

Model studies designed to simulate changes in the chemistry of the atmosphere during the last glacial period suggest that the primary sink for N2O has remained effectively constant (±15%) over the last 18 ky (6, 7). Assuming that the lifetime of N2O has remained constant (118 years), we can quantitatively assign changes in global sources based on the measured concentration changes. N2O levels during the LGM were ∼190 ppb, implying that global sources were 7.7 Tg N/year. N2O levels increased during the termination to ∼265 ppb because of a 40% (190 to 265 ppb) increase in global N2O sources. We can then use the δ15N values to calculate the magnitude of the change in marine and terrestrial N2O sources during the termination. Using δ15N values for terrestrial and oceanic N2O of –15 and 8‰, respectively, and the isotope effect associated with stratospheric photolysis of 1.023 (26), we calculate that LGM terrestrial and oceanic N2O emissions were 4.3 and 3.4 Tg N/year, respectively. Considering the concentration and isotope data together, we conclude that both terrestrial and oceanic N2O emissions increased by 40 ± 8% during the last termination, accounting for the 40% increase in tropospheric N2O loading.

Detailed inspection of the isotope results during the Younger Dryas interval (13 to 11.5 ka) suggests that the δ15N and δ18O values decreased, as one moves forward in time between 12.4 and 11.4 ka, by 1.8 and 5‰, respectively. This interval covers the end of the Younger Dryas period when atmospheric N2O values increased by ∼20 ppb. Although the measured isotope shift over this interval approximates our uncertainty envelope, the decrease in both δ15N and δ18O values over the same interval implies that the trends are probably significant. We believe that these anomalies may be related to enhanced terrestrial N2O emissions (relative to marine emissions) near the end of the Younger Dryas. This agrees with and strengthens the model results of Goldstein et al. (27), which require a large increase in terrestrial emissions to explain the concentration change at the end of the Younger Dryas.

Figure 1 shows much shared variance among the disparate indicators [correlation coefficients (r2) are greater than 0.6 and significantly different from zero between all pairs among N2O, CH4, CO2, and δ18Oice]. Because CO2 is primarily controlled by marine processes (on millennial time scales), CH4 by terrestrial processes, N2O by both terrestrial and marine processes, and δ18Oice can be viewed as a circum–North Atlantic climate record, the similarities and differences raise numerous fundamental questions about the climate system and the biogeochemical systems that cycle trace-gas species. We focus here on three questions.

First, the significant covariation between CO2 and N2O, and the covariation between marine and terrestrial sources of N2O over the last 30 ky, may provide new insights into the origin of the glacial-interglacial CO2 variations. Surface-water N2O levels are generally controlled by N mineralization in the waters above the thermocline, as reflected in the positive correlation between [N2O] and apparent oxygen utilization (AOU) in the modern ocean (28). If the positive N2O-AOU relation observed today applies for the glacial ocean, then lowering of the surface-water partial pressure of CO2 (pCO2) during the LGM through enhanced export production would have promoted elevated oceanic N2O levels that are not supported by our data. Our data are consistent with those scenarios that predict lowered pCO2 levels resulting from higher alkalinity, increased stratification, or both (29), 30).

Second, lowered oceanic N2O emissions during the LGM may reflect decreased emissions from the continental shelves when sea levels were much lower. Present-day estimates of global N2O emissions from continental shelves range from 0.6 to 2.7 Tg N/year (31, 32). These estimates are comparable to the additional oceanic N2O emissions that are required to satisfy our concentration and isotope records (1.4 Tg N/year).

Figure 1 shows that CH4 and N2O varied together not only on orbital time scales (the clear precessional signal and slower variations), but also on millennial time scales during the Younger Dryas, Dansgaard/Oeschger interstadial 8, and other events between about 40 and 90 ka. This observation is consistent with monsoonal coupling of terrestrial and marine realms. During periods of increased monsoonal activity (interstadial periods), enhanced upwelling of nutrient-rich water in the Arabian Sea and the Eastern Tropical Pacific supported increased marine productivity and thus increased organic-matter rain. The enhanced flux of organic matter caused a reduction in mid-water oxygen concentration, thereby increasing denitrification; a consequent increase in N2O production ultimately led to increased N2O emissions in upwelling areas modulated by monsoonal circulation patterns (33, 34). At the same time, enhanced monsoonal rainfall would have expanded the region of wet soils producing CH4 and N2O.

One notable exception to the general CH4/N2O correspondence occurs during the cooling event that began about 80 ka. N2O levels remained high for ∼7 ky after CH4 levels dropped precipitously. Perhaps during this period, the CH4/N2O temporal relation may reflect greater initial sensitivity of CH4 emissions to terrestrial drying. Complete soil saturation optimizes CH4 production, whereas reduced N2O emissions from drying of some regions below the 60 to 90% saturated window for optimal N2O production are partially offset by drying of wetter areas into the optimum N2O emission window. Additionally, because high northern latitudes were more important for CH4 [∼18% of preanthropogenic CH4 emissions in latitude bands north of 40°N (8)] than for N2O [∼9% (35)], icesheet advance may have impacted CH4 emissions more directly and quickly than N2O emissions. Additional isotopic analyses on samples before 30 ka might clarify the relative roles of marine and terrestrial sources of N2O in these cases, and why this behavior is not shown strongly at other times.

Supporting Online Material

SOM Text


References and Notes

View Abstract

Navigate This Article