Comparison of environmental changes between northeastern Brazil and western Patagonia during the last deglaciation reveals concomitant trends in moisture from the Intertropical Convergence Zone (ITCZ) and southern westerly winds (SWW). The data confirm an atmospheric teleconnection between the ITCZ and SWW, associated with Atlantic Meridional Overturning Circulation (AMOC) variations. When the AMOC decreases, both the ITCZ and the SWW shift southward; they shift northward when the AMOC increases. Climate simulations in which the AMOC is made to vary agree with this general pattern. Additional experiments performed with an atmosphere-only model show that the tropical Atlantic is a key area in promoting relationships between the AMOC, ITCZ, and SWW. Our data show that this mechanism, which transfers climate changes between low and middle latitudes to high latitudes in the Southern Hemisphere, acted throughout the abrupt climatic events of the last deglaciation.


The last glacial-interglacial transition (LGIT, ca. 21–10 ka) was punctuated by abrupt climatic events, characterized by opposite trends in temperature changes between hemispheres. This phenomenon, termed bipolar see-saw (Crowley, 1992), has been attributed to changes in the Atlantic Meridional Overturning Circulation (AMOC). However, the atmospheric circulation could also play a role in setting up the North Atlantic–Southern Hemisphere teleconnection (Lamy et al., 2007): a connection between the Intertropical Convergence Zone (ITCZ) and the southern westerly winds (SWW) has been proposed as a rapid mechanism acting on the Southern Hemisphere extratropical climate for the beginning of the LGIT (Anderson et al., 2009). Climate models have also shown that a southward shift of the ITCZ related to a North Atlantic cooling can strengthen the SWW via the weakening of the Hadley circulation and the Southern Hemisphere subtropical jet stream (Lee et al., 2011). However, little is known about the potential role of such a teleconnection in the climatic changes recorded in South America and, more generally, in the Southern Hemisphere.

Here we focus on two regions of South America, the northern part of northeastern Brazil (NNEB) and western Patagonia (WP), where the ITCZ and the SWW are the main precipitation drivers, respectively. In NNEB, close to the coast, most precipitation falls from February to May, when the ITCZ is in its southernmost position. In WP, the SWW generates strong precipitation, increasing southward along the coast (Garreaud et al., 2013). The SWW spread northward to 30°S in austral winter, but remain south of ∼47°S during the austral summer. The SWW intensity and associated precipitation reach their maxima at these latter latitudes. The NNEB and WP thus offer the possibility to compare and study atmospheric changes between the ITCZ and SWW.


We use the following paleoclimatic records from the NNEB and WP (Fig. 1): lacustrine core MA97-1 (coastal area of NNEB, 3°S; Ledru et al., 2006), and oceanic core MD07-3088 (offshore central WP, 46°S; Montade et al., 2013). Vegetation changes reflect precipitation changes directly influenced by the ITCZ in the NNEB and the SWW in WP. The well-constrained age models of both records provide a reliable age control for investigating multimillennial trends of ITCZ and SWW changes (see the GSA Data Repository1). We further analyze ITCZ and SWW variations and their links with changes in AMOC with numerical experiments performed with the Institut Pierre Simon Laplace climate model version 4 (IPSL-CM4), i.e., an atmosphere-ocean general circulation model (AOGCM) and with its atmospheric component, the Laboratoire de Météorologie Dynamique (LMDZ) atmospheric general circulation model (AGCM) (Marti et al., 2010). The resolution of the atmosphere model is 3.75° in longitude, 2.5° in latitude, and 19 vertical levels in both the AOGCM and AGCM. The AOGCM runs use the Paleoclimate Modeling Intercomparison Project Phase II protocol for the Last Glacial Maximum (LGM) (Braconnot et al., 2007). The reference experiment, characterized by an active AMOC, is labeled “LGM AMOC on” (Fig. 2A). From this experiment, we start the “+0.1 Sv” experiment, in which an additional freshwater flux of 0.1 Sv is added at the surface of the North Atlantic north of 40°N and of the Arctic Ocean during 420 yr (Kageyama et al., 2009). This flux is enough to make the AMOC collapse after ∼250 yr. At the end of the +0.1 Sv experiment, we stop the freshwater flux (“LGM AMOC off”) and the AMOC does not recover. From the LGM AMOC off run, we start four experiments in which we impose evaporative fluxes over the North Atlantic (50° to 70°N). These experiments, described by Marzin et al. (2013), are labeled “−0.1 Sv” and “−0.5 Sv” according to the value of the perturbation. Three −0.5 Sv experiments were performed, starting from initial conditions, 50 yr apart. To define the appropriate geographical domain corresponding to central WP, we accounted for the fact that this region, characterized by a large seasonality of precipitation, is shifted north by ∼6° in our model (Rojas et al., 2009).


Atmospheric Changes

From ca. 17.5 to 15 ka, during Heinrich Stadial 1 (HS1), the development of tropical rainforest in NNEB indicates a significant reduction of the dry season (Fig. 1A), consistent with speleothem records in the easternmost part of this region (Cruz et al., 2009). These results are interpreted as a southward shift of the ITCZ relative to its LGM position (Escobar et al., 2012). Simultaneously in central WP, palynological richness increase shows the colonization of vegetation, consistent with a sea-surface temperature (SST) increase recorded in core MD07-3088 (Figs. 1C and 1D). The benthic minus planktic foraminifera 14C age (Fig. 1G) and δ13C (Fig. 1F) from the same core decrease. This parallel decrease can only be explained by an increase in the vertical mixing in the Southern Ocean, i.e., an increase in the Southern Ocean upwelling (Siani et al., 2013). Regional vegetation change suggesting a rapid southward shift of the SWW south of central WP (Montade et al., 2013) explains the increased Southern Ocean upwelling, which is also supported by opal productivity changes recorded in the Southern Ocean (Anderson et al., 2009) and by simulations (Toggweiler et al., 2006).

After HS1, although the rainforest remains developed in NNEB, its composition reveals changes in precipitation regimes. The Moraceae percentages, characteristic of a rainforest developing under very humid conditions, decrease ca. 15.5 ka and remain low from 15 to 13 ka (Fig. 1B). This suggests a decrease of precipitation consistent with the speleothem records from NNEB (Cruz et al., 2009), while precipitation increases in the Northern Hemisphere neotropics (Peterson et al., 2000; Escobar et al., 2012). These results mark a northward shift of the ITCZ. In central WP, Astelia percentages increase abruptly at 14.5 ka and remain high until 13 ka (Fig. 1E). Astelia is characteristic of the humid Magellanic moorland vegetation that develops today along the coast in southern WP, corresponding to the core of the SWW belt, defined by strong SWW intensity and precipitation (Heusser, 1995). The development of Astelia thus indicates a strengthening of SWW during the Antarctic Cold Reversal (ACR) in central WP. Simultaneously, the increase of 14C age and δ13C (Figs. 1F and 1G) suggests a decrease of wind-driven upwelling in the Southern Ocean, as shown by Anderson et al. (2009). While SWW intensity seems to increase in central WP, intensity decreases over the Southern Ocean, suggesting a northward shift of the SWW during the ACR, also described by García et al. (2012) and Moreno et al. (2012).

From 12.8 ka, during the Younger Dryas (YD), a humidity increase is shown by a second Moraceae increase in NNEB (Fig. 1B) while dry conditions are observed in Cariaco Basin record (10°N), suggesting a southward shift of the ITCZ (Haug et al., 2001). However, other records in NNEB indicate either a slight precipitation increase (Arz et al., 1998; Wang et al., 2004) or no particular trend (Cruz et al., 2009), suggesting either less intense changes or a more complex pattern than for the HS1. After the ACR, the decrease of Astelia percentages points to a reduction of SWW in the central WP (Fig. 1E), while wind-driven upwelling in the Southern Ocean increased, as shown by 14C ages and δ13C records (Figs. 1F and 1G). These changes support a southward shift of the SWW, consistent with Pesce and Moreno (2014), and opal productivity increase recorded in the Southern Ocean during the YD (Anderson et al., 2009).

Climate Simulations

Our freshwater flux experiments have been designed to obtain large AMOC changes (Fig. 2A). These AMOC changes are associated with a cooling in the North Atlantic, at both extratropical and tropical latitudes, and warming in the southern Atlantic (see the Data Repository; Kageyama et al. 2009). Significant changes also appear in the atmospheric circulation in the tropics and in the high southern latitudes. When the AMOC and Atlantic Ocean northward heat transports decrease, the atmosphere compensates by expanding its northern Hadley cell southward (Kageyama et al., 2009). The ITCZ therefore shifts southward, which explains the precipitation variations over the NNEB (Fig. 2B). This shift of the tropical atmospheric circulation features is also found in the southern tropical jet stream (Figs. 2C and 2D; see the Data Repository). In our model, over the eastern South Pacific Ocean along South America, in annual mean and for glacial boundary conditions, there is only a single and broad westerly wind belt, which includes the subtropical jet stream and, near the surface and at middle latitudes, the SWW. These two features of the southern atmospheric circulation are therefore strongly connected. Because the upper level jet stream is also connected to the Hadley cell through angular momentum conservation, it is a useful diagnostic for analyzing tropical-extratropical circulation interactions. The latitude of the maximum 200 hPa zonal wind is anticorrelated with NNEB precipitation (Fig. 2C) and correlated with WP precipitation (Fig. 2D). Our ensemble of coupled atmosphere-ocean simulations therefore shows a plausible mechanism connecting AMOC variations to precipitation over NNEB and WP.


The comparison of paleoclimatic records suggests simultaneous shifts of the ITCZ and SWW, similar to the model results. To summarize these changes and compare with the AMOC, we calculate a normalized index (Fig. 3). A rainfall increase in NNEB indicates a southward shift of the ITCZ, while a rainfall increase in the central WP indicates a northward shift of the SWW, and conversely, this index is based on the ratio between the most sensitive moisture indicators of the ITCZ changes in core MA97-1 (Moraceae) and of the SWW changes in core MD07-3088 (Astelia). An index increase represents a southward shift of atmospheric structures that occurs when the AMOC is reduced during the HS1 and the YD (McManus et al., 2004) and when the climate cools over Greenland (Rasmussen et al., 2006) and warms over Antarctica (Lemieux-Dudon et al., 2010). However, during HS1, although the index mainly reflects ITCZ changes because Astelia do not yet show significant changes, the regional comparison above supports a southward shift of the SWW. An opposite scenario occurs during the ACR with a northward shift of atmospheric structures. Our results show that the atmospheric teleconnection between the ITCZ and SWW appears to be closely related to AMOC variations throughout the LGIT.

Such observed changes closely correspond to the model results, in which the meridional shifts of the ITCZ and SWW are related to the imposed AMOC variations. Lee et al. (2011), whose results were further interpreted by Chiang et al. (2014), proposed that ITCZ shifts modulate the strength of the subtropical jet stream, which in turns acts to strengthen the eddy-driven mid-latitude jet stream. Such a split jet is not visible in our simulations and our model shows a shift of the main atmospheric circulation features rather than a modulation of their strength. The differences in experimental designs between the studies of Lee et al. (2011) and Chiang et al. (2014) and our study make it difficult to further ascribe the reasons for these qualitatively different responses to AMOC changes. Common setups would be needed to improve our understanding of the model differences.

Using paleorecords or AOGCM results only, it remains impossible to determine whether the concomitant climate changes in NNEB and central WP are the sole result of the atmospheric teleconnection, or of other features related to AMOC changes, i.e., SST changes at southern extratropical latitudes. We therefore ran additional experiments with the LMDZ AGCM (Marzin et al., 2013) to test the importance of the tropics in transmitting a signal from the North Atlantic to WP. In the first experiment, we prescribe the global SST and sea-ice cover from 50 yr of the LGM AMOC on run, and the same other boundary conditions and forcings as for the coupled model. In a second run, we imposed the global SST and sea-ice cover from the last 50 yr of the +0.1 Sv coupled run (Fig. 4A). From these two simulations, we checked that the atmospheric response to these SSTs and sea-ice cover forcings are consistent with those obtained in the corresponding coupled simulations, which is the case (not shown). In a third simulation, we imposed the SSTs and sea ice from the LGM AMOC on everywhere but in the tropical Atlantic, over which we imposed the SSTs from the end of the +0.1 Sv simulation (Fig. 4B). We thus tested the impact of the tropical Atlantic SST changes related to an AMOC collapse. This response consists of a dipole, with cooling over the northern tropical Atlantic and warming over the southern tropical Atlantic (see the Data Repository). Most of the South American precipitation change obtained with global SST changes is obtained by only prescribing tropical Atlantic SST changes. In both cases, the southern subtropical jet stream moves southward. This gives us confidence that a teleconnection involving simultaneous movements of the ITCZ and the southern subtropical jet stream initially triggered by AMOC variations and mediated through tropical Atlantic SST changes is a major component in the explanation of the concomitant precipitation changes in NNEB and WP. From our experiments, we cannot rule out that the tropical Pacific could also play a role in tropical-extratropical interactions, because in our AOGCM experiments, SST variations over this region were much smaller than over the tropical Atlantic. Other models simulate SST changes over the Pacific in response to AMOC changes (Kageyama et al., 2013), and these could be used to compare the role of the Pacific versus Atlantic SST changes.


These results confirm the teleconnection between the ITCZ and the SWW related to the AMOC changes during the LGIT. We furthermore show that the tropical Atlantic appears to be a key area in promoting rapid north-south teleconnections, from the North Atlantic to South America. Such a teleconnection, which seems to be crucial to transfer abrupt climatic changes between low latitudes and middle to high latitudes over South America, could have played a key role on the Southern Ocean CO2-degassing via the southward intensification of SWW during the HS1 or the YD.

Financial support was provided by the Pachiderme project from the Institut National des Sciences de l’Univers (Les Enveloppes Fluides et l’Environnement), and Montade benefited from a postdoctoral position funded by Fundação Cearense de Apoio ao Desenvolvimento Científico and Ecole Pratique des Hautes Etudes. This is Institut des Sciences de l’Evolution de Montpellier publication #2015-115 and Laboratoire des Sciences du Climat et de l’Environnement contribution #5506.

1GSA Data Repository item 2015252, supplementary chronologic data and figures of climatic simulations, is available online at www.geosociety.org/pubs/ft2015.htm, or on request from editing@geosociety.org or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301, USA.