ARTICLES PUBLISHED ONLINE: 19 JUNE 2011 | DOI: 10.1038/NGEO1169 Chinese stalagmite δ18O controlled by changes in the Indian monsoon during a simulated Heinrich event Francesco S. R. Pausata1,2*(, David S. Battisti2,3, Kerim H. Nisancioglu1,4 and Cecilia M. Bitz3 Carbonate cave deposits in India and China are assumed to record the intensity of monsoon precipitation, because the δ18O of the carbonate tracks the isotopic signature of precipitation. These records show spatially coherent variability throughout the last ice age and suggest that monsoon strength was altered during the millennial-scale climate variations known as Dansgaard–Oeschger events and during the Heinrich cooling events. Here we use a numerical climate model with an embedded oxygen-isotope model to assess what caused the shifts in the oxygen-isotope signature of precipitation during a climate perturbation designed to mimic a Heinrich event. Our simulations show that a sudden increase in North Atlantic sea-ice extent during the last glacial period leads to cooling in the Northern Hemisphere, reduced precipitation over the Indian basin and weakening of the Indian monsoon. The precipitation is isotopically heavier over India and the water vapour exported to China is isotopically enriched. Our model broadly reproduces the enrichment of δ18O over Northern India and East Asia evident in speleothem records during Heinrich events. We therefore conclude that changes in the δ18O of cave carbonates associated with Heinrich events reflect changes in the intensity of Indian rather than East Asian monsoon precipitation. uring the last glacial period (∼110–10 kyr bp), millennial- and the isotopic composition of the source (for example, owing to scale climate variability was characterized by abrupt changes in sea surface temperature (SST) or river runoff17). Dtransitions between cold stadial and warm interstadial states. Oxygen-isotope records in speleothems throughout Asia are Several of these cold stadial periods are interrupted by extreme often interpreted as an index for the `strength' or `intensity' of 1 18 ice-rafting events, the so-called Heinrich events (H-events) . Asian monsoons, because they feature large oscillations in δ Oc H-events, as well as the more recent Younger Dryas (YD, sometimes that are highly correlated with changes in local summer insolation referred to as H0), are large freshwater discharges from the North owing to the precession of the equinoxes (see, for example, American ice sheet into the North Atlantic occurring irregularly refs 11,13). Several studies have suggested that orbital variations 2 18 throughout the ice age, causing long-lived cold states ; seven H-like cause changes in δ Oc by changing local precipitation processes events occurred during the last glacial period3. The climate changes and/or amounts13,15,18,19, including, for example, variations in the coincident with H-events are not restricted to the North Atlantic seasonality of precipitation. Recent studies, however, suggest that basin, but are communicated over large parts of the globe4,5 through it is difficult to explain the influence of orbital variations on changes in atmospheric and ocean circulation in response to rapid isotopes in the Asian cave records through local climate changes (for changes in Nordic Sea sea-ice extent6–8. example refs 20,21) and that the cave records more probably reflect The rapid climate changes associated with the most recent changes in the δ18O of water vapour owing to non-local climate H-event (H1) and the YD are faithfully recorded in the oxygen- processes, including changes in the processing of the vapour en isotopic composition of stalagmites (speleothems) through- route to the caves and changes in the fractionation at the source out South and East Asia (see, for example, refs 9–12). The (land or ocean)20–23. 18 18 oxygen-isotopic composition of the stalagmite calcite (δ Oc; see The millennial-scale variations in δ Oc captured in cave records equation (1) in Methods) reflects the temperature of the cave throughout southern and eastern Asia and associated with abrupt 18 18 and the precipitation-weighted δ O of precipitation (δ Op; see climate changes, such as H-events, are about one-third of the 18 13 equation (2) in Methods) at the cave site. A change in δ Op may amplitude of the precessional changes . This is remarkable because result from a change in several local and non-local processes, it indicates that variability in the `monsoon strength' on millennial including: the ratio of summer-to-spring precipitation (season- timescales, which is owing to dynamics internal to the Earth's 18 ality of precipitation affects δ Op because summer rainfall has climate system, is of the same order as that associated with the lighter—that is, more negative—δ18O values compared with spring orbitally forced changes. In this study, we shall use models to precipitation13–15); the intensity of precipitation falling at a given lo- examine the changes in climate and in the isotopic composition of cation (amount effect16) owing to changes in the strength of the con- precipitation that result from a sudden change in sea-ice extent in vection; the origin of the water vapour delivered to the site owing to the northern North Atlantic, which is thought to be the causal agent changes in circulation; the isotopic composition (not the amount) of global climate changes associated with H-events. Our particular of water vapour that is arriving from one or more regions owing to goal is to understand the climatological significance of the signals 18 changes in the processing of water vapour en route from the source recorded in the δ Oc of Asian speleothems. 1Bjerknes Centre for Climate Research, NO-5007 Bergen, Norway, 2Geophysical Institute, University of Bergen, NO-5007 Bergen, Norway, 3Department of Atmospheric Sciences, University of Washington, Seattle, Washington 98195-1640, USA, 4UNI Research, NO-5007 Bergen, Norway. (Present address: Joint Research Center, Institute for Environment and Sustainability, I-21027 Ispra (VA), Italy. *e-mail: [email protected]. 474 NATURE GEOSCIENCE j VOL 4 j JULY 2011 j www.nature.com/naturegeoscience © 2011 Macmillan Publishers Limited. All rights reserved. NATURE GEOSCIENCE DOI: 10.1038/NGEO1169 ARTICLES a 90° N 75° N 60° N 45° N Latitude 30° N 15° N 0° 180° 150° W 120° W90° W60° W30° W0° 30° E60° E90° E 120° E 150° E 180° Longitude ¬16 ¬10 ¬7 ¬5 ¬3 ¬2 ¬1 ¬0.5 0 0.5 1 2 Temperature difference (H1¬LGM) (˚C) b 90° N 75° N 60° N 45° N Latitude 30° N 15° N 0° 180° 150° W 120° W90° W60° W30° W0° 30° E60° E90° E 120° E 150° E 180° Longitude ¬100 ¬80 ¬60 ¬40 ¬20 0 20 40 60 80 100 Precipitation difference (H1¬LGM) (%) Figure 1 j Annual averaged temperature and precipitation difference between the H1 and LGM. a, Surface temperature difference (◦C). b, Precipitation difference (%). Markers indicate the locations of the following caves: Hulu (circle), Songjia (square), Dongge (star) and Timta (diamond). The lines in a indicate the annual climatological 50% sea-ice extent for H1 (white) and LGM (red) in the North Atlantic sector. Simulated isotopic change owing to an archetypal H-event of sea ice over the northern North Atlantic, causing a strong cooling We take as a starting point the climate of the Last Glacial extending throughout the Northern Hemisphere, as also shown in Maximum (LGM) as simulated by a fully coupled climate model previous studies (for example refs 5,7,29; the simulated warming (Community Climate System Model version 3, CCSM3) using over northeast India is discussed in Supplementary Information). insolation, carbon dioxide, ice sheets and continental geometry Precipitation is greatly reduced throughout the North Atlantic from 21 kyr bp (ref. 24). A second set of experiments (H1) was and northern Indian Ocean in the H1 experiment. Note that carried out in which freshwater is abruptly added to the North there are no significant changes in the annual average or seasonal Atlantic to mimic an H-event (see Methods), causing an extensive distribution of precipitation at the location of the Chinese caves expansion of sea ice in the northern North Atlantic25. We call these (Fig. 1, Table 1). The atmospheric circulation and the spatial and experiments `H1' because they feature an extension of sea ice in the seasonal changes in surface temperature and precipitation in the North Atlantic that is consistent with the extension of ice-covered uncoupled experiments are very similar to those from the coupled area in a typical H-event1 and because the sedimentary chronology experiments (compare Fig. 1 and Supplementary Fig. S1). This is unambiguously tied to the cave chronologies for the H1 event26. enables us to use the uncoupled LGM and H1 experiments to The coupled climate model used in the LGM and H1 experiments determine changes in the isotopic composition of precipitation does not contain an isotope module. Hence, to examine the isotopic associated with abrupt changes in the North Atlantic sea ice. 18 changes associated with the sudden expansion of sea ice in the The simulated change (H1 minus LGM) in δ Op over south 18 North Atlantic, we ran two further off-line experiments using the Asia is shown in Fig. 2. The increase in δ Op in southern and same atmospheric model (Community Atmosphere Model version eastern Asia is primarily owing to changes in the δ18O and 3, CAM3; ref. 27) as is used in CCSM3, but with an embedded amount of precipitation during the monsoon season—May to module for stable water isotopes28. These experiments use the same August (MJJA; Supplementary Fig. S2). The cooler H1 climate 18 (21 kyr bp) boundary conditions as the coupled experiments. The features an increase in δ Op throughout southern and eastern annual cycle in SST and sea-ice concentration is prescribed to be Asia, including all cave sites, ranging from C0:9 at Hulu to identical to that from the coupled LGM experiment and from the C1:7 at Timta.