The Greenland Ice Sheet (GIS) contains the equivalent of 7.4 metres of global sea-level rise1. Its stability in our warming climate is therefore a pressing concern. However, the sparse proxy evidence of the palaeo-stability of the GIS means that its history is controversial (compare refs 2 and 3 to ref. 4). Here we show that Greenland was deglaciated for extended periods during the Pleistocene epoch (from 2.6 million years ago to 11,700 years ago), based on new measurements of cosmic-ray-produced beryllium and aluminium isotopes (10Be and 26Al) in a bedrock core from beneath an ice core near the GIS summit. Models indicate that when this bedrock site is ice-free, any remaining ice is concentrated in the eastern Greenland highlands and the GIS is reduced to less than ten per cent of its current volume. Our results narrow the spectrum of possible GIS histories: the longest period of stability of the present ice sheet that is consistent with the measurements is 1.1 million years, assuming that this was preceded by more than 280,000 years of ice-free conditions. Other scenarios, in which Greenland was ice-free during any or all Pleistocene interglacials, may be more realistic. Our observations are incompatible with most existing model simulations that present a continuously existing Pleistocene GIS. Future simulations of the GIS should take into account that Greenland was nearly ice-free for extended periods under Pleistocene climate forcing.
Subscribe to Journal
Get full journal access for 1 year
only $3.90 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Tax calculation will be finalised during checkout.
Rent or Buy article
Get time limited or full article access on ReadCube.
All prices are NET prices.
Houghton, J. T. et al. (eds) Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change (IPCC) (Cambridge Univ. Press, 2001)
Bennike, O. et al. Early Pleistocene sediments on Store Koldewey, northeast Greenland. Boreas 39, 603–619 (2010)
Funder, S., Bennike, O., Bocher, J., Israelson, C. & Petersen, K. S. Late Pliocene Greenland—the Kap Kobenhavn Formation in North Greenland. Bull. Geol. Soc. Den. 48, 117–134 (2001)
Bierman, P. R. et al. Preservation of a preglacial landscape under the center of the Greenland Ice Sheet. Science 344, 402–405 (2014)
de Vernal, A. & Hillaire-Marcel, C. Natural variability of Greenland climate, vegetation, and ice volume during the past million years. Science 320, 1622–1625 (2008)
Reyes, A. V. et al. South Greenland ice-sheet collapse during Marine Isotope Stage 11. Nature 510, 525–528 (2014)
Willerslev, E. et al. Ancient biomolecules from deep ice cores reveal a forested Southern Greenland. Science 317, 111–114 (2007)
Alley, R. B. et al. History of the Greenland Ice Sheet: paleoclimatic insights. Quat. Sci. Rev. 29, 1728–1756 (2010)
NEEM Community Members. Eemian interglacial reconstructed from a Greenland folded ice core. Nature 493, 489–494 (2013)
Yau, A. M., Bender, M. L., Blunier, T. & Jouzel, J. Setting a chronology for the basal ice at Dye-3 and GRIP: implications for the long-term stability of the Greenland Ice Sheet. Earth Planet. Sci. Lett. 451, 1–9 (2016)
Calov, R., Robinson, A., Perrette, M. & Ganopolski, A. Simulating the Greenland ice sheet under present-day and palaeo constraints including a new discharge parameterization. Cryosphere 9, 179–196 (2015)
Otto-Bliesner, B. L. & Brady, E. C. The sensitivity of the climate response to the magnitude and location of freshwater forcing: last glacial maximum experiments. Quat. Sci. Rev. 29, 56–73 (2010)
Applegate, P. J., Parizek, B. R., Nicholas, R. E., Alley, R. B. & Keller, K. Increasing temperature forcing reduces the Greenland Ice Sheet’s response time scale. Clim. Dyn. 45, 2001–2011 (2015)
Stone, E. J., Lunt, D. J., Rutt, I. C. & Hanna, E. Investigating the sensitivity of numerical model simulations of the modern state of the Greenland ice-sheet and its future response to climate change. Cryosphere 4, 397–417 (2010)
Gow, A. J. & Meese, D. A. Nature of basal debris in the GISP2 and Byrd ice cores and its relevance to bed processes. In Proceedings of the International Symposium on Glacial Erosion and Sedimentation (ed. Collins, D.) 134–140 (International Glaciological Society, 1996)
Chmeleff, J., von Blanckenburg, F., Kossert, K. & Jakob, D. Determination of the Be-10 half-life by multicollector ICP-MS and liquid scintillation counting. Nucl. Instrum. Methods Phys. Res. B 268, 192–199 (2010)
Korschinek, G. et al. A new value for the half-life of Be-10 by heavy-ion elastic recoil detection and liquid scintillation counting. Nucl. Instrum. Methods Phys. Res. B 268, 187–191 (2010)
Nishiizumi, K. et al. Absolute calibration of 10Be AMS standards. Nucl. Instrum. Methods Phys. Res. B 258, 403–413 (2007)
Nishiizumi, K. et al. In situ produced cosmogenic nuclides in GISP2 rock core from Greenland Summit. Eos 77, F428, abstr. OS41B-10 (1996)
Heisinger, B. et al. Production of selected cosmogenic radionuclides by muons: 2. Capture of negative muons. Earth Planet. Sci. Lett. 200, 357–369 (2002)
Heisinger, B. et al. Production of selected cosmogenic radionuclides by muons. 1. Fast muons. Earth Planet. Sci. Lett. 200, 345–355 (2002)
Balco, G. & Rovey, C. W. An isochron method for cosmogenic-nuclide dating of buried soils and sediments. Am. J. Sci. 308, 1083–1114 (2008)
Granger, D. E. & Muzikar, P. F. Dating sediment burial with in situ-produced cosmogenic nuclides: theory, techniques, and limitations. Earth Planet. Sci. Lett. 188, 269–281 (2001)
Marrero, S. M. et al. Cosmogenic nuclide systematics and the CRONUScalc program. Quat. Geochronol. 31, 160–187 (2016)
Bender, M. L., Burgess, E., Alley, R. B., Barnett, B. & Clow, G. D. On the nature of the dirty ice at the bottom of the GISP2 ice core. Earth Planet. Sci. Lett. 299, 466–473 (2010)
Alley, R. B., Clark, P. U., Huybrechts, P. & Joughin, I. Ice-sheet and sea-level changes. Science 310, 456–460 (2005)
Fyke, J., Eby, M., Mackintosh, A. & Weaver, A. Impact of climate sensitivity and polar amplification on projections of Greenland Ice Sheet loss. Clim. Dyn. 43, 2249–2260 (2014)
Dutton, A. et al. Sea-level rise due to polar ice-sheet mass loss during past warm periods. Science 349, (2015)
Granger, D. E. in In-Situ Produced Cosmogenic Nuclides And Quantification Of Geological Processes (eds Siame, L., Bourles, D. & Brown, E. T. ) Spec. Pap. 415, 1–16 (The Geological Society of America, 2006)
Lisiecki, L. E. & Raymo, M. E. A Pliocene-Pleistocene stack of 57 globally distributed benthic delta O-18 records. Paleoceanography 20, PA1003 (2005)
Schaefer, J. M. et al. High frequency Holocene glacier fluctuations in New Zealand differ from the northern signature. Science 324, 622–625 (2009)
Davis, J. C. et al. LLNL/UC AMS Facility and Research-Program. Nucl. Instrum. Methods Phys. Res. B 52, 269–272 (1990)
Granger, D. E. et al. New cosmogenic burial ages for Sterkfontein Member 2 Australopithecus and Member 5 Oldowan. Nature 522, 85–88 (2015)
Paul, M. Separation of isobars with a gas-filled magnet. Nucl. Instrum. Methods Phys. Res. B 52, 315–321 (1990)
Timmers, H., Weijers, T. D. M. & Elliman, R. G. Unique capabilities of heavy ion elastic recoil detection with gas ionization detectors. Nucl. Instrum. Methods Phys. Res. B 190, 393–396 (2002)
Fifield, L. K., Tims, S. G., Gladkis, L. G. & Morton, C. R. 26Al measurements with 10Be counting statistics. Nucl. Instrum. Methods Phys. Res. B 259, 178–183 (2007)
Balco, G., Stone, J. O., Lifton, N. A. & Dunai, T. J. A complete and easily accessible means of calculating surface exposure ages or erosion rates from Be-10 and Al-26 measurements. Quat. Geochronol. 3, 174–195 (2008)
Stone, J. Air pressure and cosmogenic isotope production. J. Geophys. Res. 105, 23753–23759 (2000)
Nishiizumi, K. Preparation of Al-26 AMS standards. Nucl. Instrum. Methods Phys. Res. B 223/224, 388–392 (2004)
Borchers, B. et al. Geological calibration of spallation production rates in the CRONUS-Earth project. Quat. Geochronol. 31, 188–198 (2016)
Phillips, F. M. et al. The CRONUS-Earth project: a synthesis. Quat. Geochronol. 31, 119–154 (2016)
Rovey, C. W. & Balco, G. Paleoclimatic interpretations of buried paleosols within the pre-Illinoian till sequence in northern Missouri, USA. Palaeogeogr. Palaeoclimatol. Palaeoecol. 417, 44–56 (2015)
Bierman, P. R., Marsella, K. A., Patterson, C., Davis, P. T. & Caffee, M. Mid-Pleistocene cosmogenic minimum-age limits for pre-Wisconsinan glacial surfaces in southwestern Minnesota and southern Baffin island: a multiple nuclide approach. Geomorphology 27, 25–39 (1999)
We gratefully acknowledge help from the National Ice Core Laboratory (NICL) and thank the GISP2 steering committee for providing the bedrock core samples. J.M.S. acknowledges support by the Lamont Climate Center and the Comer Family Foundation. R.B.A. acknowledges support by the NSF (AGS 1338832), as do J.M.S., J.P.B. and N.E.Y. (PLR Arctic System Science Program number 1503959). We thank J. Fyke for assistance in producing Fig. 1b. G.B. acknowledges support from the Ann and Gordon Getty Foundation. M.W.C. acknowledges support from the US National Science Foundation, grant EAR-1153689. The pioneering cosmogenic-nuclide study of the GISP2 bedrock core under the lead of Kuni Nishiizumi in the late 1990s motivated our study. This is LDEO publication number 8068.
The authors declare no competing financial interests.
Nature thanks P.-H. Blard, G. Milne and the other anonymous reviewer(s) for their contribution to the peer review of this work.
Extended data figures and tables
10Be (n = 5) and 26Al (n = 2) concentration–depth profiles compared to the depth dependence of nuclide production rates (including both spallogenic and muon production). In both a and b, the red and blue boxes represent 10Be (red) and 26Al (blue) measurements. The vertical dimension of each box represents a distinct segment of core, and vertical lines connect multiple core segments that were amalgamated for each 10Be or 26Al analysis. In most cases amalgamated segments were adjacent to each other, but in some cases (for example, the uppermost two core segments), they were separated by gaps. The width of the boxes shows measurement uncertainty (1σ; see Extended Data Table 1) on nuclide concentrations. In a, we attempt to fit the observations by assuming that the bedrock surface is the land surface, the erosion rate is zero, and by allowing the duration of a single period of exposure to vary. For this exercise we treat 10Be and 26Al separately, that is, the predictions are not forced to obey the production ratio. The continuous thin black lines show predicted nuclide concentrations for this model, and the discontinuous, darker, black bars show predicted nuclide concentrations averaged over depth ranges corresponding to each analysis. The black bars, therefore, are the model predictions that we compare to the measurements. This model cannot be fitted to the data, because observed nuclide concentrations do not decrease as rapidly as they would if the bedrock surface were the land surface during exposure. In b, we perform the same fitting exercise, but also include a thickness of shielding mass above the bedrock surface as an additional fitting parameter. This removes the systematic misfit shown in a and makes it possible to fit the observations. A cover thickness of 350 g cm−2 best fits the observations.
In a and d, the red and blue boxes represent 10Be (n = 5; red) and 26Al (n = 2; blue) measurements. The vertical dimension of each box represents a distinct segment of core, and vertical lines connect multiple core segments that were amalgamated for each 10Be or 26Al analysis. In most cases amalgamated segments were adjacent to each other, but in some cases (for example, the uppermost two core segments), they were separated by gaps. The width of the boxes shows measurement uncertainty (1σ; see Extended Data Tables 2 and 3) on nuclide concentrations. The thick black lines are nuclide concentrations predicted for each core segment by the best-fitting parameters of each model. b and e show the exposure history implied by the best-fitting parameters of each model compared to the LR04 oxygen isotope stack from ref. 30. Red bars represent periods of surface exposure and blue bars periods of cover by the ice sheet. c and f show observed nuclide concentrations compared to model predictions for samples in which both 10Be and 26Al were measured, normalized to production rates at their respective depths implied by each model, on a two-nuclide diagram29. Red ellipses are 68% confidence regions for the nuclide concentrations including measurement uncertainties only, and black dots are nuclide concentrations predicted by best-fitting model parameters. The solid black lines show the simple exposure region; darker dashed lines are isolines of burial in increments of 1 Myr, and lighter dotted lines are isolines of exposure time in increments of 0.1 Myr. a, b and c show the fit of model 1, the simplest possible model that fits the data, which includes a single period of surface exposure, a single period of burial, 350 g cm−2 of cover thickness above the bedrock surface during exposure, and zero surface erosion. In c, nuclide concentrations are normalized to production rates at sample depths below this additional cover thickness. This model provides a good fit to the measurements. The exposure history implied by the best-fitting parameters for model 1 (280-kyr exposure, 1.1 Myr burial) provides a maximum limiting constraint on the length of time the present ice sheet has been continuously present at the core site. d, e and f show the fit of model 1B, which includes a long period of continuous exposure with steady surface erosion, a single period of burial, and zero additional cover thickness above the bedrock surface. In f, nuclide concentrations are normalized to production rates at sample depths below the bedrock surface. This model cannot be adequately fitted to the observations.
a–d show parameter values for two-stage exposure-burial models with four free parameters (model 1C: free parameters are exposure time texp, burial time, surface erosion rate during exposure, and additional cover thickness above the bedrock surface during exposure periods) that yield acceptable fits to the observations. All panels show the same set of results; only the axes differ. Note that model 1 with best-fitting parameters (280-kyr exposure, 1.1 Myr burial, zero surface erosion, 350 g cm−2 cover thickness) is an endmember of this distribution.
In a, c, e and f, the red and blue boxes represent 10Be (red) and 26Al (blue) measurements. The vertical dimension of each box represents a distinct segment of core, and vertical lines connect multiple core segments that were amalgamated for each 10Be or 26Al analysis. In most cases amalgamated segments were adjacent to each other, but in some cases (for example, the uppermost two core segments), they were separated by gaps. The width of the boxes shows measurement uncertainty (1σ; see Extended Data Tables 2 and 3) on nuclide concentrations. The thick black lines are nuclide concentrations predicted for each core segment by the best-fitting parameters of each model. b, d, f and h show the exposure history implied by the best-fitting parameters of each model compared to the LR04 oxygen isotope stack from ref. 30. Red bars represent periods of surface exposure and blue bars periods of cover by the ice sheet. a and b show model 2, which is a dynamic steady state model with 100-kyr cycles. This model has one free parameter, which is the fraction of each cycle during which the site is ice-free, and assumes zero surface erosion and 350 g cm−2 cover thickness above the bedrock surface. The best-fitting length of ice-free periods for this model is 8,000 years. All other panels show models with ice-free conditions during some or all middle and late Pleistocene interglaciations, and have two free parameters: the duration of ice-free conditions during interglaciations, and the length of an initial ice-free period in the middle or early Pleistocene. c and d show a model in which the core site is ice-free during all interglacials within the period of 100-kyr-long glacial–interglacial cycles after 1.1 Myr ago, during an (arbitrarily long) series of short interglacials during 41-kyr-long glacial–interglacial cycles before 1.1 Myr ago, and during an initial longer period of exposure in the middle Pleistocene. The best-fitting duration of ice-free conditions during interglacials for this model is 4,200 years. e and f show ice-free conditions during MIS 9, MIS 11 and MIS 13, an (arbitrarily long) period of short ice-free interglaciations during 41-kyr-long cycles before 1.1 Myr ago, and an initial period of continuous exposure in the middle Pleistocene. The best-fitting duration of ice-free conditions during interglacials for this model is 7,400 years. g and h show ice-free conditions during MIS 9, MIS 11 and MIS 13, with (arbitrarily located) occasional ice-free periods in the early and middle Pleistocene. The best-fitting duration of ice-free conditions during interglacials for this model is 18,200 years.
About this article
Cite this article
Schaefer, J., Finkel, R., Balco, G. et al. Greenland was nearly ice-free for extended periods during the Pleistocene. Nature 540, 252–255 (2016). https://doi.org/10.1038/nature20146
Earth and Planetary Science Letters (2021)
Measuring multiple cosmogenic nuclides in glacial cobbles sheds light on Greenland Ice Sheet processes
Earth and Planetary Science Letters (2021)
In situ cosmogenic <sup>10</sup>Be–<sup>14</sup>C–<sup>26</sup>Al measurements from recently deglaciated bedrock as a new tool to decipher changes in Greenland Ice Sheet size
Climate of the Past (2021)
Quaternary Science Reviews (2021)
Limited glacial erosion during the last glaciation in mid-latitude cirques (Retezat Mts, Southern Carpathians, Romania)