Millennial-scale changes in the northern boundary of the South Atlantic Subtropical Gyre
The decreases in the relative abundance of G. truncatulinoides occurred simultaneously with increases in precipitation over NE Brazil that were, in turn, associated to southward displacements in the Intertropical Convergence Zone (ITCZ) during HS (Fig. 2c-d)28,39. The meridional position of the ITCZ determines the location of the equatorial ascending branch of the Hadley Cells. The Hadley Cells from both hemispheres have an important role in the interhemispheric atmospheric heat transport40. Under a weak Atlantic Meridional Overturning Circulation (AMOC) (e.g., during HS) (Fig. 2a-b)35–37 the decreased northward oceanic heat transport warmed the South Atlantic13,41. This resulted in a southward migration of the equatorial ascending branch of the Hadley Cells, partially compensating the decrease in northward oceanic heat transport via an increase in the northward atmospheric heat transport42,43.
Changes in the Hadley Cells directly affect the oceanic Subtropical Cells (STC) by changing the trade winds stress on the surface. Indeed, the wind-driven oceanic STC can be described as the upper ocean counter-part of the Hadley Cell44. Therefore, changes in the meridional position of the equatorial branch of the STC are linked to the ITCZ position via the Hadley Cell45. During HS, (ref. 43) described a southward shift of the ascending branch of the South Atlantic STC that followed the ITCZ43. The southward shift of the STC in the South Atlantic should lead to the southward displacement of the nSASG during HS. We suggest that southward migrations of the nSASG during HS increased the upper water column stratification (i.e., shallower thermocline) at our core site, decreasing the abundance of G. truncatulinoides (Figs. 2c, 3a). An increased stratification in the upper water column of the western tropical South Atlantic during HS has been confirmed by (ref. 46) and (ref. 47) (Fig. S1d), supporting our suggestion.
In line with our results, in the tropical North Atlantic at the southern boundary of the North Atlantic Subtropical Gyre (sNASG), increases in the abundance of G. truncatulinoides (Fig. S1e)27 during HS suggest southward migrations of the sNASG. This is supported by upper water column temperature and salinity data48, as well as model experiments49, suggesting a tight coupling between the ascending branches of both STC and subtropical gyres in the Atlantic during HS.
The suggested southward migrations of the nSASG during HS were also accompanied by decreases in the strength of the SE trade winds50, a consequence of the decreased meridional SST gradient in the South Atlantic13,41. The reduced strength of the SE trade winds was thus co-responsible for the increases in upper water column stratification in the western tropical South Atlantic. At our core site, however, the large amplitude decreases in the abundance of G. truncatulinoides (Figs. 2c, 3a) point to the occurrence of changes in upper water column structure. Such changes would be accomplished by the nSASG crossing southwards our core site (Fig. 1).
Moreover, our core site is located at the modern bifurcation of the South Equatorial Current (SEC) in the upper 100 m of the water column (Fig. 1d). Our suggestion of southward migrations of the nSASG to be dynamically linked to southward shifts of the ITCZ position during HS contrasts to the seasonal mode of changes in the SEC bifurcation (i.e., during austral summer, a southward migration of the ITCZ occurs simultaneously to a northward migration of the SEC bifurcation)51. Thus, our results highlight the need to consider the timescale while investigating the processes responsible for changes in western tropical South Atlantic upper water column stratification.
Impacts of changes in the South Atlantic Subtropical Gyre
The G. truncatulinoides abundance records from the nSASG (core M125-95-3) and the sSASG (core MD07-3076Q) reveal an antiphase pattern during HS6-4 and HS1 (Fig. 3a–b). Notably, in both G. truncatulinoides records, sinistral and dextral morphotypes were quantified together. While in the nSASG G. truncatulinoides abundance decrease during HS6-1 (Fig. 3a), in the sSASG G. truncatulinodes abundance increase during the HS6-4 and HS1 with no clear trend during HS3 and HS2 (Fig. 3b)24. The antiphase pattern suggests that the whole SASG was displaced southwards during HS6-4 and HS1. In contrast, the reduction in G. truncatulinoides in the nSASG together with constant values in the sSASG suggests a meridional contraction of the SASG during HS3 and HS2 (Fig. 3a–b).
The Southern Hemisphere westerly winds control the position of the STF in the South Atlantic (e.g.,56). A southward displacement of the STF during HS has been suggested57. The concurrent HS increases in the abundance of G. truncatulinoides in the sSASG (Fig. 3b) and the decreases in dust flux around Antarctica (a proxy for the Southern Hemisphere westerly winds intensity) suggest a link between the southward displacements of the sSASG and the Southern Hemisphere westerly winds58. The southward (northward) displacement of the STF has commonly been correlated to the increased (reduced) water inflow from the Indian to the Atlantic Ocean through the Agulhas Leakage59,60. A SST record under the influence of the Agulhas Leakage indeed shows systematic millennial-scale increases during HS, indicating southward shifts of the STF (Fig. 3c)15. Also, a planktonic foraminiferal index for the relative position of the STF in the South Atlantic (% Neogloboquadrina pachyderma (sinistral) / N. pachyderma total) corroborates the southward migrations of the STF during HS (Fig. 3d)14. The strong correlation between the G. truncatulinoides record from the sSASG, Agulhas Leakage SST and the STF index (Fig. 3b-d) suggest that the position of the sSASG was closely related to the Southern Hemisphere westerly winds and the STF. We suggest that the extratropical atmospheric circulation accompanied the southward displacement of the ITCZ and the nSASG during the HS6-4 and HS1, as indicated by model experiments43.
A southward-displaced extratropical atmospheric circulation during HS6-4 and HS1, shifted the sSASG polewards. Model simulations of a collapsed AMOC show a positive temperature anomaly (ca. 4ºC) at ca. 500 m water depth in the SASG47. Concurrently, in the Antarctic Circumpolar Current the increased eddy heat transport together with a southern position of the westerlies, allowed for more heat to reach high southern latitudes causing a retreat in Antarctic sea ice47. The increased SST around Antarctica61 resulted in a strengthening of the Southern Ocean deep-water upwelling (e.g.,62). Increased upwelling around Antarctica, in turn, fostered CO2 release to the atmosphere62. A weakened dust-driven biological pump in the Southern Ocean also contributed to the rise in atmospheric CO2 during HS (e.g.,63) (Fig. 3g).
Decreases in G. truncatulinoides abundance at the nSASG and the absence of major changes in the abundance of this species at the sSASG (Fig. 3a–b) suggest a meridional contraction of the SASG during HS3 and HS2. At the end of Marine Isotope Stage (MIS) 3 and during most of MIS2, the abundance of G. truncatulinoides at the sSASG24 shows nearly constant values between 2-4% (Fig. 3b). This is the period (i.e., ca. 30-19 ka) when full glacial boundary conditions were reached. This period encompasses HS3 and HS2, which were not related to southward shifts of the sSASG (Fig. 3b). We suggest that the full glacial boundary conditions hindered the sSASG to migrate southwards even under HS forcing. Under full glacial boundary conditions, the northward migration of the Polar and Subantarctic Fronts together with extensive sea-ice around Antarctica probably hampered southward displacements of the sSASG. The striking increase in sea-ice in the Atlantic and Indian sectors of the Southern Ocean under full glacial boundary conditions corroborates this suggestion (Fig. 3e)52,53. The long-term expansion of sea-ice equatorwards was fostered by low obliquity64 that reached minimum value at ca. 30 ka (Fig. 3f)54. Changes in sea-ice extent should be accompanied by changes in the oceanic Polar and Subantarctic Fronts (e.g.,12,16,65,66). Records on millennial-scale temporal resolution of the Agulhas Leakage SST, the position of the STF and dust flux to the Southern Ocean (Fig. 3c–d)14,15,58 confirm the presence of full glacial boundary conditions during HS3 and HS2. Importantly, full glacial boundary conditions were associated to a significant northward displacement of the Southern Hemisphere westerly winds and a marked decrease in Southern Ocean deep-water upwelling, that hindered CO2 to be released from the Southern Ocean to the atmosphere, as recorded in ice-cores during the HS3 and HS2 (Fig. 3e–g)55.
In summary, enhanced poleward heat fluxes occurred during HS6-4 and HS1 and were favored by southward shifts of the SASG. Such meridional migrations of the SASG may have played a central role on oceanic carbon storage or release during the last glacial period on millennial timescales by controlling heat delivery to the Southern Ocean. Regarding the ongoing poleward displacement of the SASG5, our results suggest that an increase in heat transport to the Southern Ocean may strengthens deep-water upwelling and CO2 release to the atmosphere, constituting a positive feedback for global warming.