Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Evolution and forcing mechanisms of El Niño over the past 21,000 years


The El Niño Southern Oscillation (ENSO) is Earth’s dominant source of interannual climate variability, but its response to global warming remains highly uncertain1. To improve our understanding of ENSO’s sensitivity to external climate forcing, it is paramount to determine its past behaviour by using palaeoclimate data and model simulations. Palaeoclimate records show that ENSO has varied considerably since the Last Glacial Maximum (21,000 years ago)2,3,4,5,6,7,8,9, and some data sets suggest a gradual intensification of ENSO over the past 6,000 years2,5,7,8. Previous attempts to simulate the transient evolution of ENSO have relied on simplified models10 or snapshot11,12,13 experiments. Here we analyse a series of transient Coupled General Circulation Model simulations forced by changes in greenhouse gasses, orbital forcing, the meltwater discharge and the ice-sheet history throughout the past 21,000 years. Consistent with most palaeo-ENSO reconstructions, our model simulates an orbitally induced strengthening of ENSO during the Holocene epoch, which is caused by increasing positive ocean–atmosphere feedbacks. During the early deglaciation, ENSO characteristics change drastically in response to meltwater discharges and the resulting changes in the Atlantic Meridional Overturning Circulation and equatorial annual cycle. Increasing deglacial atmospheric CO2 concentrations tend to weaken ENSO, whereas retreating glacial ice sheets intensify ENSO. The complex evolution of forcings and ENSO feedbacks and the uncertainties in the reconstruction further highlight the challenge and opportunity for constraining future ENSO responses.

This is a preview of subscription content, access via your institution

Relevant articles

Open Access articles citing this article.

Access options

Buy article

Get time limited or full article access on ReadCube.


All prices are NET prices.

Figure 1: TRACE simulation and observation.
Figure 2: ENSO, BJ index and annual cycle.
Figure 3: ENSO in single forcing experiments.


  1. Collins, M. et al. The impact of global warming on the tropical Pacific ocean and El Nino. Nature Geosci. 3, 391–397 (2010)

    Article  ADS  CAS  Google Scholar 

  2. Moy, C. et al. Variability of El Niño/Southern Oscillation activity at millennial timescales during the Holocene epoch. Nature 420, 162–166 (2002)

    Article  ADS  CAS  Google Scholar 

  3. Tudhope, A. et al. Variability in the El Niño–Southern Oscillation through a glacial–interglacial cycle. Science 291, 1511–1517 (2001)

    Article  ADS  CAS  Google Scholar 

  4. Riedinger, M. et al. A 6100 14C yr record of El Niño activity from the Galapagos islands. J. Paleolimnol. 27, 1–7 (2002)

    Article  ADS  Google Scholar 

  5. Koutavas, A. & Joanides, S. El Niño–Southern Oscillation extrema in the Holocene and Last Glacial Maximum. Paleoceanography 27 PA4208 (2012)

    Article  ADS  Google Scholar 

  6. Cobb, K. et al. Highly variable El Niño–Southern Oscillation throughout the Holocene. Science 339, 67–70 (2013)

    Article  ADS  CAS  Google Scholar 

  7. Conroy, J. et al. Holocene changes in eastern tropical Pacific climate inferred from a Galapagos lake sediment record. Quat. Sci. Rev. 27, 1166–1180 (2008)

    Article  ADS  Google Scholar 

  8. Rein, B. et al. El Niño variability off Peru during the last 20,000 years. Paleoceanography 20 PA4003 (2005)

    Article  ADS  Google Scholar 

  9. Sadekov, A. et al. Paleoclimate reconstructions reveal a strong link between El Niño-Southern Oscillation and tropical Pacific mean state. Nature Commun. 4 2692 (2013)

    Article  ADS  Google Scholar 

  10. Clement, A., Seager, R. & Cane, M. Suppression of El Niño during the Mid-Holocene by changes in the Earth’s orbit. Paleoceanography 15, 731–737 (2000)

    Article  ADS  Google Scholar 

  11. Roberts, W. An Investigation into the Causes for the Reduction in the Variability of the El Niño-Southern Oscillation in the Early Holocene in a Global Climate Model. PhD thesis, Univ. Washington. (2007)

  12. Liu, Z., Kutzbach, J. & Wu, L. Modeling climatic shift of El Niño variability in the Holocene. Geophys. Res. Lett. 27, 2265–2268 (2000)

    Article  ADS  Google Scholar 

  13. Otto-Bliesner, B. et al. Modeling El Niño and its tropical teleconnections during the glacial–interglacial cycle. Geophys. Res. Lett. 30 10.1029/2003GL08553 (2003)

    Article  Google Scholar 

  14. Liu, Z. et al. Transient simulation of last deglaciation with a new mechanism for Bølling–Allerød warming. Science 325, 310–314 (2009)

    Article  ADS  CAS  Google Scholar 

  15. Shakun, J. et al. Global warming preceded by increasing CO2 during the last deglaciation. Nature 484, 49–54 (2012)

    Article  ADS  CAS  Google Scholar 

  16. Wittenberg, A. Are historical records sufficient to constrain ENSO simulations. Geophys. Res. Lett. 36, L12702 (2009)

    Article  ADS  Google Scholar 

  17. Wolff, C. et al. Reduced interannual rainfall variability in East Africa during the Last Ice Age. Science 333, 743–747 (2011)

    Article  ADS  CAS  Google Scholar 

  18. Penland, C. & Sardeshmukh, P. The optimal growth of tropical sea surface temperature anomalies. J. Clim. 8, 1999–2024 (1995)

    Article  ADS  Google Scholar 

  19. Chiang, J., Fang, Y. & Chang, P. Pacific climate change and ENSO activity in the mid-Holocene. J. Clim. 22, 923–939 (2009)

    Article  ADS  Google Scholar 

  20. Masson-Delmotte, V. et al. In Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds Stocker, T. F. et al.) 383–464 (Cambridge Univ. Press, 2013)

  21. Carre, M. et al. Holocene history of ENSO variance and asymmetry in the eastern tropical Pacific. Science 345, 1045–1048 (2014)

    Article  ADS  CAS  Google Scholar 

  22. Felis, T. et al. Pronounced interannual variability in tropical South Pacific temperatures during Heinrich Stadial 1. Nature Commun. 3 965 (2012)

    Article  ADS  Google Scholar 

  23. Kim, S. & Jin, F. An ENSO stability analysis. Part I: results from a hybrid coupled model. Clim. Dyn. 36, 1593–1607 (2011)

    Article  Google Scholar 

  24. Liu, Z., Brady, E. & Lynch-Steiglitz, J. Global ocean response to orbital forcing in the Holocene. Paleoceanography 18, 1041 (2003)

    Article  ADS  Google Scholar 

  25. Liu, Z. & Xie, S. Equatorward propagation of coupled air–sea disturbances with application to the annual cycle of the eastern tropical Pacific. J. Atmos. Sci. 51, 3807–3822 (1994)

    Article  ADS  Google Scholar 

  26. Timmermann, A. et al. The influence of a weakening of the Atlantic Meridional Overturning Circulation on ENSO. J. Clim. 20, 4899–4919 (2007)

    Article  ADS  Google Scholar 

  27. Liu, Z. A simple model study of the forced response of ENSO to an external periodic forcing. J. Clim. 15, 1088–1098 (2002)

    Article  ADS  Google Scholar 

  28. Timmermann, A. et al. The effect of orbital forcing on the mean climate and variability of the tropical Pacific. J. Clim. 20, 4147–4159 (2007)

    Article  ADS  Google Scholar 

  29. Meehl, G., Teng, H. & Branstator, G. Future changes of El Niño in two global climate models. Clim. Dyn. 26, 549–566 (2006)

    Article  Google Scholar 

  30. Timmermann, A., Jin, F. & Collins, M. Intensification of the annual cycle in the tropical Pacific due to greenhouse warming. Geophys. Res. Lett. 31, L12208 (2004)

    ADS  Google Scholar 

  31. Collins, W. et al. The community climate system model version 3 (CCSM3). J. Clim. 19, 2122–2143 (2006)

    Article  ADS  Google Scholar 

  32. Berger, A. Long-term variations of daily insolation and quaternary climatic changes. J. Atmos. Sci. 35, 2362–2367 (1978)

    Article  ADS  Google Scholar 

  33. Joos, F. & Spahni, R. Rates of change in natural and anthropogenic radiative forcing over the past 20,000 years. Proc. Natl Acad. Sci. USA 105, 1425–1430 (2008)

    Article  ADS  CAS  Google Scholar 

  34. Peltier, W. Global glacial isostasy and the surface of the ice-age earth: the ICE-5G (VM2) model and GRACE. Annu. Rev. Earth Planet. Sci. 32, 111–149 (2004)

    Article  ADS  CAS  Google Scholar 

  35. He, F. Simulating Transient Climate Evolution of the Last Deglaciation with CCSM3. PhD thesis, Univ. Wisconsin-Madison. (2011)

  36. McManus, J. et al. Collapse and rapid resumption of Atlantic meridional circulation linked to deglacial climate changes. Nature 428, 834–837 (2004)

    Article  ADS  CAS  Google Scholar 

  37. He, F. et al. Northern Hemisphere forcing of Southern Hemisphere climate during the last deglaciation. Nature 494, 81–85 (2013)

    Article  ADS  CAS  Google Scholar 

  38. Deser, C. et al. Tropical Pacific and Atlantic climate variability in CCSM3. J. Clim. 19, 2451–2481 (2006)

    Article  ADS  Google Scholar 

  39. Rittenour, T., Brigham-Grette, J. & Mann, M. El Niño-like climate teleconnections in New England during the late Pleistocene. Science 288, 1039–1042 (2000)

    Article  ADS  CAS  Google Scholar 

  40. McGregor, S. et al. Inferred changes in El Niño-Southern Oscillation variation over the past six centuries. Clim. Past 9, 2269–2284 (2013)

    Article  Google Scholar 

  41. Brown, J. et al. Issues in quantitative model–proxy data comparisons. Paleoceanography 23, PA3202 (2008)

    Article  ADS  Google Scholar 

  42. Kim, S. & Jin, F. An ENSO stability analysis. Part II: results from the twentieth and twenty-first century simulations of the CMIP3 models. Clim. Dyn. 36, 1609–1627 (2010)

    Article  Google Scholar 

Download references


This work is supported by the US National Science Foundation (NSF)/P2C2 Program, Chinese NSFC41130105, the US Department of Energy/Office of Science (BER), Chinese MOST2012CB955200, NSF1049219 and 1204011. The computation is carried out at Oak Ridge National Laboratory of the Department of Energy and the National Center for Atmospheric Research supercomputing facility.

Author information

Authors and Affiliations



Z. Liu conceived the study and wrote the paper. Z. Lu and XW performed the analysis. Z. Liu and B.O.B. contributed to the simulations. A.T. and K.C. contributed to the interpretation. All authors discussed the results and provided inputs to the paper.

Corresponding author

Correspondence to Zhengyu Liu.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Extended data figures and tables

Extended Data Figure 1 Power spectrum of ENSO.

Power spectra of Niño3.4 monthly SST variability in TRACE (after removing the annual cycle) in five 1,000-year windows: 1–2 kyr ago, 5–6 kyr ago, 10–11 kyr ago, 15–16 kyr ago and 19–20 kyr ago. The spectral peak remains at 2 years, but with a different intensity. For each spectrum, the 95% cut-off level and the corresponding red noise curve are also plotted (in dotted lines). The black bar at the bottom shows the 1.5–7-year band used for the calculation of ENSO variance.

Extended Data Figure 2 Evolution of ENSO and annual cycle along the Equator.

Evolution of the amplitudes (standard deviation in 100-year window) of interannual (1.5–7 years) variability (a), the annual cycle of SST (b), total variability (<7 years) (c) and the ratio of the amplitudes of the interannual over the annual cycle (d) along the equatorial Pacific (5° S–5° N) in TRACE. The total variability is dominated by the annual cycle, except in the central-eastern Pacific, where ENSO becomes dominant. This occurs because ENSO variability shows a broad pattern from the central to the eastern Pacific, whereas the annual cycle is strong along the eastern boundary and decays rapidly towards the central Pacific.

Extended Data Figure 3 Evolution of variability amplitude.

Evolution of the amplitude of interannual (1.5–7-year) variability of precipitation (blue) in TRACE in Ecuador2 (a), the Galapagos islands (Lake El Juno)7 (b), Niño1 + Niño2 (c), Niño3 (d) and Niño4 (e) (locations marked in Extended Data Figs 4 and 5). All the anomalies are the monthly data with a 3-month running mean filtered to the 1.5–7-year band. The amplitude is calculated as the standard deviation of the 1.5–7-year band-passed monthly time series in succeeding 300-year windows. For reference, the amplitude of Niño3.4 SST interannual variability is also plotted in each panel (red). The correlation between each curve and the amplitude of ENSO is also calculated. All correlations are highly significant (P < 0.01, with a sample size of 70).

Extended Data Figure 4 Correlation map between ENSO and precipitation.

ad, Map of time series correlation of interannual (1.5–7 yr) variability between monthly Niño3.4 SST and precipitation during 21–20 kyr ago (a), 16–15 kyr ago (b) and 1–0 kyr ago (c) in TRACE, and the present observation (1981–2005) (d). eh, Map of interdecadal amplitude correlation between interannual ENSO (Niño3.4 SST) and precipitation variability during 21–20 kyr ago (e), 15–10 kyr ago (f), 10–5 kyr ago (g) and 5–0 kyr ago (h). All the anomalies are the monthly data with a 3-month running mean after filtered to the 1.5–7-year band. For eh the amplitude is calculated as the standard deviation in a 40-year window, and the detrended amplitude in the 1,000-year period is used to calculate the amplitude correlation. The two black crosses indicate the region of proxy observation in the Galapagos islands7 and on the Ecuador coast2, respectively, and the three black boxes denote the regions of Niño1 + Niño2, Niño3 and Niño4 as discussed in Extended Data Fig. 3. Colours in correlations indicate regions where the correlation is significant at more than the 99% level.

Extended Data Figure 5 Correlation between the amplitudes of ENSO and precipitation variability.

Map of the correlation between the 200-year running amplitudes of interannual ENSO and precipitation variability during 21–15 kyr ago (a), 15–10 kyr ago (b) and 10–0 kyr ago (c). The two black crosses indicate the region of proxy observation in the Galapagos islands7 and on the Ecuador coast2, respectively, and the three black boxes denote the regions of Niño1 + Niño2, Niño3 and Niño4 as discussed in Extended Data Fig. 3. The result will be similar if the amplitude is calculated directly using a 300-year window as in Extended Data Fig. 3. Colours in model correlations indicate regions where the correlations are significant at more than the 99% level.

Extended Data Figure 6 Uncertainty for ENSO model–data comparison.

a, Detecting trend of ENSO amplitude in ‘pseudo-corals’. Histogram of Holocene (7–0 kyr ago) linear trends of ENSO amplitude derived from 30-year ‘pseudo-coral’ records of the Niño3.4 SST in TRACE. A linear trend (regression coefficient) is derived from the ENSO amplitudes of a random set of 50 (red) ‘corals’, with each ENSO amplitude as the standard deviation of the interannual (1.5–7 years) SST variability of a 30-year section of ‘coral record’. The PDF on the right (marked with TRACE) is derived from the linear trends of 100,000 randomly formed sets of coral records, whereas the PDF on the left (marked with Null) is the null hypothesis of no trend in ENSO amplitude, and is derived from the linear trend of a time series after random scrambling of the Niño3.4 SST. Two additional PDFs are derived with the number of corals increased to 200 (blue) and 1,000 (black) in each set. The right-side one-tailed 95% significance levels are 0.106 (red), 0.052 (blue) and 0.024 (black) for the null hypothesis, and the left-side one-tailed 95% significance levels are 0.036 (red), 0.088 (blue) and 0.116 (black) for TRACE. With 50 corals, the trend in TRACE cannot be identified at the 95% level because the significance level in TRACE is below that of NULL (0.036 < 0.106); with 200 corals, the trend can be identified at the 95% level because the significance level of TRACE is beyond that of NULL (0.088 > 0.052); with 1,000 corals, the 95% significance levels are well beyond the NULL (0.116  0.024), implying a highly significant trend of ENSO strengthening in the Holocene. b, ENSO amplitude in TRACE in a 100-year window (red thick line, same as in Fig. 1e) and a 30-year window (blue thin line) (both on the left axis) as well as the ENSO variance reconstructed from corals in the central Pacific6 (dark green dots) and from the variance of annual SST range from mollusc shells along the Peru coast21 (black horizontal bars). The two data sets are plotted in changes relative to the present ENSO amplitude (on the right axis), which is then rescaled with the model ENSO amplitude such that the relative change in model ENSO amplitude can also be scaled on the right axis. All the model and proxy data are aligned and referenced to their last millennium average (1–0 kyr ago).

Extended Data Figure 7 Evolution of the BJ index and its components.

Evolution of ocean–atmosphere feedbacks in the eastern equatorial Pacific region (180° E–80° W, 5° S–5° N) for interannual (1.5–7-year band) ENSO variability in 100-year windows in TRACE. a, ENSO amplitude (red) and the BJ index (purple). b, The two negative (damping) feedbacks (surface heat flux feedback (cyan) and mean advection feedback (green, offset by +2)) and the three positive feedbacks (zonal advection feedback (blue), Ekman upwelling feedback (red) and thermocline feedback (brown)), as well as the (sum) total feedback (BJ index, purple, offset by +2). The ENSO amplitude largely follows the BJ index (an increasing trend) in the Holocene, but not in the deglaciation, suggesting the dominant role of ocean–atmosphere feedback for ENSO intensification in the Holocene, but not for millennial variability in deglaciation.

Extended Data Figure 8 Evolution of feedback coefficients.

Response sensitivity and mean state for interannual variability in the eastern equatorial Pacific region (170° E–80° W, 5° S–5° N) in TRACE. a, Atmospheric response sensitivity to SST. b, Zonal current response sensitivity to wind stress. c, Upwelling response sensitivity to wind stress. d, Thermocline response sensitivity to wind stress (βhah, brown), βh (cyan) and ah (grey). e, Mean zonal SST gradient. f, Mean stratification. g, Mean upwelling. In ag, each curve is plotted to the same relative scale such that their variation can be compared directly: the scale of a variable y ranges from median (y) − 0.6 × |mean (y)| to median (y) + 0.6 × |mean (y)|.

Extended Data Figure 9 Response of equatorial thermocline in the Holocene.

ac, The annual cycle of the difference between 10 kyr ago and 1 kyr ago of insolation (a), and in South Pacific temperature (°C) at the surface (b) and 149 m depth (c) for different latitudes in TRACE. Zonal mean annual mean temperature difference in the upper ocean of the South Pacific (180–100° W) (d). In the subtropical South Pacific (40–10° S), the insolation warming in austral winter (in a) leads to a SST warming in austral winter–spring (in b), which is then ventilated into the subsurface thermocline as a warming throughout the year (in c), and eventually into the equatorial thermocline (in d), decreasing the stratification there.

Extended Data Figure 10 Phase-locking of ENSO.

Evolution of phase locking of Niño3.4 SST interannual variability as a function of the calendar month in 100-year windows. The peak of ENSO anomaly is locked strongly to boreal winter in early deglaciation; this phase locking is weakened towards the early Holocene, but re-emerges towards the late Holocene.

PowerPoint slides

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Liu, Z., Lu, Z., Wen, X. et al. Evolution and forcing mechanisms of El Niño over the past 21,000 years. Nature 515, 550–553 (2014).

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:

This article is cited by


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing