Late Quaternary Atmospheric CH4 Isotope Record Suggests Marine Clathrates Are Stable
One explanation for the abrupt increases in atmospheric CH4, that occurred repeatedly during the last glacial cycle involves clathrate destabalization events. Because marine clathrates have a distinct deuterium/hydrogen (D/H) isotope ratio, any such destabilization event should cause the D/H ratio of atmospheric CH4 (δDCH4) to increase. Analyses of air trapped in the ice from the second Greenland ice sheet project show stable and/or decreasing δDCH4 values during the end of the Younger and Older Dryas periods and one stadial period, suggesting that marine clathrates were stable during these abrupt warming episodes. Elevated glacial δDCH4 values may be the result of a lower ratio of net to gross wetland CH4 emissions and an increase in petroleum-based emissions.
3. P. S. Berloff, J. C. McWilliams, A. Bracco, J. Phys.
Oceanogr. 32, 764 (2002).
4. J. B. Joergensen, J. Mann, S. Ott, H. L. Pe´cseli, J. Trulsen,
Phys. Fluids 17, 035111 (2005).
5. M. C. Wolf, R. Voigt, P. A. Moore, J. Chem. Ecol. 30, 501
6. P. Moore, J. Crimaldi, J. Mar. Syst. 49, 55 (2004).
7. S. Edouard, B. Legras, F. Lefe`vre, R. Eymard, Nature 384,
8. M. Nelkin, R. M. Kerr, Phys. Fluids 24, 1754 (1981).
9. B. Sawford, Annu. Rev. Fluid Mech. 33, 289 (2001).
10. L. F. Richardson, Proc. R. Soc. London Ser. A 110, 709
11. G. K. Batchelor, Q. J. R. Meterol. Soc. 76, 133 (1950).
12. P. K. Yeung, Phys. Fluids 6, 3416 (1994).
13. S. Ott, J. Mann, J. Fluid Mech. 422, 207 (2000).
14. G. Boffetta, I. M. Sokolov, Phys. Rev. Lett. 88, 094501
15. P. K. Yeung, M. S. Borgas, J. Fluid Mech. 503, 93 (2004).
16. F. Nicolleau, G. Yu, Phys. Fluids 16, 2309 (2004).
17. L. Biferale et al., Phys. Fluids 17, 115101 (2005).
18. A. La Porta, G. A. Voth, A. M. Crawford, J. Alexander,
E. Bodenschatz, Nature 409, 1017 (2001).
19. G. A. Voth, A. La Porta, A. M. Crawford, J. Alexander,
E. Bodenschatz, J. Fluid Mech. 469, 121 (2002).
20. M. Virant, T. Dracos, Meas. Sci. Technol. 8, 1539 (1997).
21. N. T. Ouellette, H. Xu, E. Bodenschatz, Exp. Fluids 40,
22. A. N. Kolmogorov, Dokl. Akad. Nauk SSSR 30, 301 (1941).
23. A. M. Obukhov, Izv. Akad. Nauk SSSR 5, 453 (1941).
24. K. R. Sreenivasan, Phys. Fluids 7, 2778 (1995).
25. P. K. Yeung, Annu. Rev. Fluid Mech. 34, 115 (2002).
26. D. J. Thomson, B. J. Devenish, J. Fluid Mech. 526, 277
27. S. Kurien, K. R. Sreenivasan, Phys. Rev. E 64, 056302
28. R. A. Shaw, Annu. Rev. Fluid Mech. 35, 183 (2003).
29. Z. Warhaft, J. L . Lumle y, J. Fluid Mech. 88,659
30. This research was supported by the Physics Division of
NSF and by the Max Planck Society. We thank L. Collins,
G. Falkovich, J. Hunt, J. Schumacher, D. Vincenzi, and
Z. Warhaft for helpful discussions and suggestions over
the course of this work.
24 October 2005; accepted 5 January 2006
Late Quaternary Atmospheric CH
Isotope Record Suggests Marine
Clathrates Are Stable
One explanation for the abrupt increases in atmospheric CH
, that occurred repeatedly during
the last glacial cycle involves clathrate destabalization events. Because marine clathrates have
a distinct deuterium/hydrogen (D/H) isotope ratio, any such destabilization event should cause the
D/H ratio of atmospheric CH
) to increase. Analyses of air trapped in the ice from the second
Greenland ice sheet project show stable and/or decreasing dD
values during the end of the
Younger and Older Dryas periods and one stadial period, suggesting that marine clathrates were
stable during these abrupt warming episodes. Elevated glacial dD
values may be the result of a
lower ratio of net to gross wetland CH
emissions and an increase in petroleum-based emissions.
he ice core record of atmospheric CH
changes covering the past 650,000
years exhibits two primary frequencies.
Over long time scales (greater than 10,000
years) atmospheric CH
changes have a sub-
stantial amount of variance concentrated in
the precessional bandwidth (19,000 and 23,000
years) (1, 2) that is considered to be an integral
part of tropical climate throughout the late
Pleistocene. One hypothesis that accounts for
this observation involves an energized hydro-
logic cycle during periods of elevated low-
latitude insolation. The invigorated hydrologic
cycle promotes an increase in wetland extent
driving a concomitant increase in CH
sions that raise atmospheric CH
warm periods. Embedded within the precession
signal are millennial- and century-scale varia-
tions that are tightly coupled to Greenland
temperature (3, 4). In general, increasing at-
levels are synchronous with, or
slightly lag (by a few decades), the surface tem-
perature increase over Greenland (5). Assessing
the nature of these abrupt CH
events is im-
portant for understanding how ecosystems and
climate are connected and in estimating the
degree to which future CH
levels may con-
tribute to changes in Earth_s radiation budget.
There are two competing explanations for the
increases. One hypothesis holds that
the terrestrial biosphere is capable of rapidly
emissions in response to abrupt
changes in the hydrologic cycle that are tele-
connected to surface temperatures over Green-
land (3, 4). The other explanation involves the
sudden release of marine clathrates situated
along the continental margin where episodic
destabilization events may have been triggered
by enhanced ventilation (warming) of upper
thermocline waters (6). The majority of the
ultimately travels across the air-
sea interface leading to atmospheric CH
Model estimates of changes in the primary
sink (tropospheric hydroxyl radical) dur-
ing the last glacial termination suggest that
the observed CH
variations must be due in
large part to changes in the sources as opposed
to changes in the rate of removal (7). The
isotopic composition of atmospheric CH
therefore provides additional information on
the relative contribution of the various sources.
Variations in the D/H ratio of atmospheric CH
) can be used to infer variable clathrate
contributions on the basis of their elevated dD
values compared with all terrestrial CH
sources (Fig. 1). Methane clathrates within the
continental margin sediments are formed al-
most exclusively by CO
reduction or thermal
cracking of longer chain hydrocarbons, whereas
emissions are primarily acetic-
lastic in nature (8, 9). During CO
the methyl hydrogen atoms come directly from
that is in isotopic equilibrium
with the porewater (10). The resulting dD
values are lower than the porewater dD
to a È180 per mil (°) biologically induced iso-
tope effect associated with CO
reduction (9, 11).
Marine clathrate dD
values from 13 near-
shore sites scattered throughout the Northern
Hemisphere are surprisingly constant (–189 T
27°; error is SD) given the diverse nature of
the geologic and sedimentologic settings and
the varying propor tions of microbial and
at each site (12, 13). In con-
production in terrestrial ecosystems
is dominated by acetogenesis (acetate fer-
mentation) where three-fourths of the hydro-
gen atoms in the emitted methane originate
from the methyl group associated with the
acetate substrate. The remaining hydrogen
comes from the local water with the resulting
values generally ranging from
–250 to –380°, with the local dD
strongly influenced by the dD of precipitation
An atmospheric dD
record (Fig. 2) was
generated from the second Greenland ice sheet
project (GISP II) ice core using a previously
described technique with an external precision
of T4.2° (14). The general picture of dD
variations associated with the deglaciation shows
a progressive decrease in dD
as the con-
centration of CH
increases, opposite to that
predicted by increasing clathrate contributions
due to warming associated with the termina-
tion. During the last glacial maximum (LGM),
values were generally È5° higher than
the Bolling/Allerod values E15 to 13 thousand
years ago (ka)^ and È20° higher than early
Holocene values. There are three factors that
can be reasonably constrained as contributing
to the elevated dD
values during the LGM.
All three factors are temperature dependent, so
Department of Geosciences and the Earth and Environ-
mental Systems Institute, Pennsylvania State University,
University Park, PA 16802, USA. E-mail: sowers@geosc.
10 FEBRUARY 2006 VOL 311 SCIENCE www.sciencemag.org
estimates have been made on the basis of two
different tropical LGM temperature estimates
(Table 1). First, colder temperatures during this
period would have increased atmospheric dD
through the temperature-dependent kinetic iso-
tope effect (KIE) associated with the primary
removal process, OH oxidation in the tropo-
sphere. The magnitude of this effect is þ3.4°
on the basis of the laboratory-determined
temperature dependence (15), assuming tropo-
spheric temperatures were 5-C colder during
the LGM. Secondly, dD changes in mean ocean
water arising from changes in continental ice
volume impart a direct effect on atmospheric
by altering the dD of porewater H
utilized by CO
reducing methanogens (oceanic
and terrestrial). A less direct effect occurs as
the oceanic dD
change is propagated through
the hydrologic cycle and incorporated in ter-
restrial organic hydrocarbons (16, 17), the pri-
mary substrate for the fermentative methanogens.
The mean ocean dD
change E7.5° (18)^ is
transferred through the LGM global hydro-
logic cycle (19, 20), causing changes in dD
values for terrestrial CH
that range from –5 to
5°, depending on the assumed LGM tropical
temperatures (8, 9). Finally, a 10% decrease in
the ratio of C
-type plants during the
LGM (21) would have lowered atmospheric
valuesby0to1.9° relative to Holo-
cene values. Together these three factors
account for a small portion of the observed
shift between the LGM and early
Holocene, implying that other factors must be
There are at least three additional factors
associated with the termination that are difficult
to quantify. First, elevated dD
the LGM may be the result of a decrease in the
ratio of net to gross (N/G) CH
has been fairly well documented through in-
hibitor studies that as much as 50% of the CH
produced at depth in soils is consumed by
microbia lly mediated methane oxidation near
the soil-atmosphere interface (22). The dD
values for the emitted CH
are strongly de-
pendent on the N/G ratio because of the large
KIE associated with methane oxidation E–95
to –285° (23)^. For example, lowering N/G
by 11% during the LGM (with the dD
assigned as –300° and the KIE for
methane oxidation as –95°) would raise
by 10°. The sense of this
change is consistent with observations that the
methane-producing communities are more sen-
sitive to temperature changes than methane-
oxidizing communities (24).
Two additional factors contributing to the
values during the LGM in-
volve an increase in the relative proportion
Fig. 2. Results from the last glacial termination as recorded in the GISP II ice core. The upper
purple curve is the isotopic temperature (30). The red curve is from the current dD
1s (4.2°) error bars. The bottom CH
concentration curve (green) is from Brook et al.(3). ppb,
parts per billion. The increased sample resolution associated with the abrupt CH
increases associated with the onset of the Bolling/Aller od and the end of the Younger Dryas periods are
shown in expanded view in Fig. 3. The present-day dD
value for the Southern Hemisphere (SH) is
shown as a horizontal dashed line for reference.
Table 1. Constrainable factors influencing dD
during the LGM.
(LGM to Holocene) (°)
Factor CLIMAP SST* 5-C Tropical cooling
10% decrease in C
ratio during LGM† –1.9 0
KIE for OH oxidation 0 3.4
DSea level‡ 5.1 0 to –5
Total change þ3.2 3.4 to –1.6
*CLIMAP Climate: Long-Range Investigation, Mapping, and Prediction; SST, sea surface temperature. †Assuming the D/H
ratio in C
plants differs by È15° [C
plants have higher dD values (31)], then a 10% reduction in the C
wetland plants during the LGM (21) would have raised atmospheric dD
values by È1.9° relative to Holocene values. Assume
additional 5-C cooling during LGM yields no change in C
ratio. ‡Seawater dD
during LGM 0 7.5° standard mean
ocean water (18). General circulation model simulations suggest little change in dD
precipitation using CLIMAP SST (19, 20)buta
slight decr ease in dD
precipitation for 5-C tropical cooling (19). Finally, assume dD
0 0:675 (8).
Fig. 1. Characteristic dD
values for various
sources. All the data except the
value for the marine clathrates (13)arefrom
(26–28 ). Presen t day atmospheric dD
mates are È–90 T 5° (29). The enriched
atmospheric value is the result of a large KIE
(þ250°) associated with the primary sink
(tropospheric OH) (14). Error bars for each source
correspond to the tabulated range of values.
SMOW , standard mean ocean water.
www.sciencemag.org SCIENCE VOL 311 10 FEBRUARY 2006
of petroleum-based and/or biomass-burning
emissions, both of which have elevated
D/H ratios (Fig. 1). Model simulations of bio-
mass burning, however, suggest lowered CH
emissions during the LGM (25). If, as recently
suggested (12, 13), CH
from petroleum seeps
contributed a larger proportion of global
sources during the LGM compared with early
Holocene periods, then we would expect higher
values during the LGM.
Assuming global CH
emissions during the
LGM were 111 Tg/year (3) and the character-
value for the terrestrial biosphere
was –300°, then a 10° dD
signal can be
accounted for by increasing the fraction of CH
emissions based on petroleum and/or biomass
burning by 9% during the LGM (compared
with early Holocene emissions).
The high-resolution d D
the end of the Younger and Older Dryas
periods (11.5 and 14.7 ka, respectively) and
the onset of interstadial 8 (IS8) (38.5 to 38 ka)
provide important constraints for assessing
clathrate stability during these periods (Fig. 3).
With the exception of one short period of
between 15.2 and 15.0 ka, the
record from GISP II shows
relatively stable or slightly decreasing dD
values during periods of increasing CH
centration. This trend in not consistent with either
suggesting that marine clathrates were stable
throughout the last glacial termination as well as
during periods of abrupt warming.
To estimate the magnitude of the atmospher-
shift associated with a hypothetical
clathrate destabilization event, a simple one-
box model of the atmosphere was developed
using the CH
concentration history from the
end of the Younger Dryas period to constrain
emissions (3). The model consists of
two sources and a single sink term. Terre s tr ia l
emissions, the lifetime of atmospheric CH
value for terrestrial CH
(–300°), and the KIE associated with the
sink (þ165°) were all held constant through-
out the simulation. Then, beginning at model
year 11.64 ka, we introduced clathrate-derived
0 j189°) at a rate of 0.8 Tg/year
for the next 100 model years, after which
clathrate emi ssi ons were held constant at 80
Tg/year. The model predicted evolution of
concentration and dD
are included in
Fig. 3 for comparison with the GISP II data
from the Younger Dryas. Assuming clathrate
was the only new CH
source at the end
of the Younger Dryas, the predicted dD
change was þ21°. The relatively constant
values throughout the transition to ele-
levels suggest little change in the
relative proportion of all individual emissions
with near-constant characteristic dD
The transition from the Older Dryas to
Bolling period (15 to 14 ka) provides a very
different view of the factors influencing dD
(Fig. 3). During the 300-year period immedi-
ately preceding the abrupt increase in atmo-
by 15° followed by a rapid 10° increase,
during which time atmospheric CH
remained effectively constant. Obviously, many
more data are needed to document this os-
cillation but, with the limited data in hand, it
appears that a rapid shift in the characteristic
values of various sources is needed in
the absence of substantial global emission
changes during a period of relative climate
The general trend of decreasing dD
throughout the termination, combined with rel-
atively stable dD
values during periods of
rapidly increasing CH
, suggests that marine
clathrates are stable during this period and
specifically during abrupt warming events. The
elevated LGM dD
values are likely to be
related to a number of factors, the most im-
portant being decreased N/G ratios and an
increase in petroleum-based CH
during the glacial period. Further insight into
these factors will derive from future measure-
ments of d
that are in progress.
References and Notes
1. J. Chappellaz, J.-M. Barnola, D. Raynaud, Y. S. Korotkevich,
C. Lorius, Nature 345, 127 (1990).
2. R. Spahni et al., Science 310, 1317 (2005).
3. E. Brook, T. Sowers, J. Orchardo, Science 273, 1087
4. J. Chappellaz et al., Nature 366, 443 (1993).
5. J. P. Severinghaus, T. Sowers, E. J. Brook, R. B. Alley,
M. L. Bender, Nature 391, 141 (1998).
6. J. P. Kennett, K. G. Cannariato, I. L. Hendy, R. J. Behl,
Methane Hydrates in Quaternary Climate Change
(American Geophysical Union, Washington, DC, 2003).
7. P. Martinerie, G. P. Brasseur, C. Granier, J. Geophys. Res.
100, 14291 (1995).
8. S. Waldron, J. M. Lansdown, E. M. Scott, A. E. Fallick,
A. J. Hall, Geochim. Cosmochim. Acta 63, 2237 (1999).
9. M. J. Whiticar, E. Faber, M. Schoell, Geochim. Cosmochim.
Acta 50, 693 (1986).
10. D. L. Valentine, A. Chidthaisong, A. Rice, W. S. Reeburgh,
S. C. Tyler, Geochim. Cosmochim. Acta 68, 1571 (2004).
11. M. Balabane, E. Galimov, M. Hermann, R. Letolle,
Org. Geochem. 11, 115 (1987).
12. B. Luyendyk, J. P. Kennett, J. F. Clark, Mar. Pet. Geol. 22,
13. A. V. Milkov, Org. Geochem. 36, 681 (2005).
14. Materials and methods are available as supporting
material on Science Online.
15. T. Gierczak, R. D. Talukdar, S. C. Herndon, G. L. Vaghjiani,
A. R. Ravishankara, J. Phys. Chem. 101, 3125 (1997).
16. D. Sachse, J. Radke, G. Gleixner, Geochim. Cosmochim.
Acta 69, 4877 (2004).
17. S. Epstein, C. J. Yapp, J. H. Hall, Earth Planet. Sci. Lett.
30, 241 (1976).
18. D. P. Schrag et al., Quaternary Sci. Rev. 21, 331 (2002).
19. C. D. Charles, D. Rind, R. Healy, R. S. Webb, Clim. Dyn.
17, 489 (2001).
20. J. Jouzel, G. Hoffmann, R. D. Koster, V. Masson,
Quaternary Sci. Rev. 19, 363 (2000).
21. G. J. Collatz, J. A. Berry, J. S. Clark, Oecologia 114, 441
22. W. S. Reeburgh, in The Atmosphere, R. F. Keeling, Ed.
(Elsevier-Pergamon, Oxford, 2004), vol. 4, pp. 65–90.
23. M. J. Whiticar, Chem. Geol. 161, 291 (1999).
24. S. C. Moosavi, P. M. Crill, J. Geophys. Res. 103, 29093
25. K. Thonicke, I. C. Prentice, C. Hewitt, Global Biogeochem.
Cycles 19, GB3008 (2005).
26. P. Quay et al., Global Biogeochem. Cycles 13, 445 (1999).
27. A. K. Snover, P. D. Quay, Global Biogeochem. Cycles 14,
28. A. K. Snover, P. D. Quay, W. M. Hao, Global Biogoechem.
Cycles 14, 11 (2000).
29. A. Rice, A. A. Gotoh, H. O. Aijie, S. C. Tyler, Anal. Chem.
73, 4104 (2001).
30. P. M. Grootes, M. Stuiver, J. W. C. White, S. Johnsen,
J. Jouzel, Nature 366, 552 (1993).
31. Y. Chikaraishi, H. Naraoka, Phytochemistry 63, 361
32. The research was supported by NSF (ATM-0117291
and OPP-0125900). I thank R. Alley, J. Kennett, and
three anonymous reviewers for carefully reviewing
Supporting Online Material
Materials and Methods
11 October 2005; accepted 17 January 2006
Fig. 3. Expanded views of three abrupt CH
concentration events recorded in the GISP II ice core.
The isotopic temperature record (30) and atmospheric CH
concentration record (3) are plotted for
reference. The red curves are from the current dD
analyses with 1s errors at each measured depth.
The results from the one-box clathrate-only model are shown for the Younger Dryas simulation with
black curves. All data in all three panels are plotted on the same y axes for comparison.
10 FEBRUARY 2006 VOL 311 SCIENCE www.sciencemag.org