Southern Ocean drives multidecadal atmospheric CO2 rise during Heinrich Stadials

Abstract

The last glacial period was punctuated by cold intervals in the North Atlantic region that culminated in extensive iceberg discharge events. These cold intervals, known as Heinrich Stadials, are associated with abrupt climate shifts worldwide. Here, we present CO₂ measurements from the West Antarctic Ice Sheet Divide ice core across Heinrich Stadials 2 to 5 at decadal-scale resolution. Our results reveal multi-decadal-scale jumps in atmospheric CO₂ concentrations within each Heinrich Stadial. The largest magnitude of change (14.0 ± 0.8 ppm within 55 ± 10 y) occurred during Heinrich Stadial 4. Abrupt rises in atmospheric CO₂ are concurrent with jumps in atmospheric CH₄ and abrupt changes in the water isotopologs in multiple Antarctic ice cores, the latter of which suggest rapid warming of both Antarctica and Southern Ocean vapor source regions. The synchroneity of these rapid shifts points to wind-driven upwelling of relatively warm, carbon-rich waters in the Southern Ocean, likely linked to a poleward intensification of the Southern Hemisphere westerly winds. Using an isotope-enabled atmospheric circulation model, we show that observed changes in Antarctic water isotopologs can be explained by abrupt and widespread Southern Ocean warming. Our work presents evidence for a multi-decadal- to century-scale response of the Southern Ocean to changes in atmospheric circulation, demonstrating the potential for dynamic changes in Southern Ocean biogeochemistry and circulation on human timescales. Furthermore, it suggests that anthropogenic CO₂ uptake in the Southern Ocean may weaken with poleward strengthening westerlies today and into the future.

Keywords: Heinrich Stadials; carbon cycle; carbon dioxide; ice core; paleoclimate.

Fig. 1.

Paleoclimate records of polar climate, greenhouse gas variability, and East Asian monsoon intensity reflecting meridional displacement of the Intertropical Convergence Zone. (A) Averaged δ¹⁸O from Greenland Ice Core Project (GRIP) and Greenland Ice Sheet 2 (GISP2) (black) (32, 33) on the GICC05 timescale (34) multiplied by 1.0063 after 31 ka (35), (B) WAIS Divide (WD) CH₄ (gray) (15), (C) Chinese speleothem δ¹⁸O (blue) (11), (D) Antarctic six-core average d_ln (dark green) (36), (E) Antarctic six-core average δ¹⁸O_ice (light green) (36), (F) WD CO₂ (orange dots) with binomial smoothed average (dark orange line), and pooled 1σ standard uncertainty (orange shading) (this study; 18, 21). Gray bars show the timing of Heinrich Stadials (HS) 1 to 5 (SI Appendix, Table S1). Dashed lines represent timing of decadal- to centennial-scale rises and decreases in WD CO₂ and CH₄, respectively.

Fig. 2.

Centennial-scale CO₂ and CH₄ variability during Heinrich Stadials 1 to 5. Upper panel: WD CH₄ (15), WD CO₂ from this study (closed circles), and previous publications (open circles) (18, 21) and replicated from previous publications (closed black diamonds). Light orange shading represents pooled 1σ standard uncertainty of individual CO₂ measurements. Shaded vertical bars highlight centennial-scale jumps in WD CO₂ and CH₄. Green lines indicate breakfit determinations (see text). Lower panel: WD CO₂ as in Upper panel sections A–E, with the magnitude and duration of abrupt CO₂ transitions shown in green.

Fig. 3.

Atmospheric δ ¹³C–CO₂, CO₂, and Southern Ocean marine sediment proxies during HS-4 and HS-1. Left panel: Taylor Glacier δ¹³C–CO₂ (light blue) (20), WD CO₂ (orange) (this study; 21), South Atlantic sediment proxies for deep-water oxygenation (blue) (43), and deep-water temperature (44). Right panel: δ¹³C–CO₂ (light blue) (29), WD CO₂ (orange) (this study; 21), simulated atmospheric CO₂ (pink) (;LH1-SO-SHW simulation), and difference in Southern Ocean and atmosphere ¹⁴C ages from shallow (yellow), intermediate (green), and deep (dark green) corals (27, 45) including 2σ U-Th dating uncertainties. Respectively, 160 and 150 y were subtracted from the Taylor Glacier chronology to align Taylor Glacier and WD CO₂ peaks at HS4 and HS1.

Fig. 4.

Centennial-scale Antarctic temperature response during Heinrich Stadials. (A) Stacked average change in WD CH₄ (gray) (15), WD CO₂ (orange) (this study), Antarctic δ¹⁸O (yellow) (36), Antarctic d_ln (green) (36), vapor source temperature (blue) (67), and site temperature (purple) (67) during HSs 1 to 5. Time “0” represents the mid-point of the associated WD CH₄ jump in HSs 1 to 5 (see text). (B) SST anomaly forcing based on Ferreira et al. (64) in response to a poleward displacement of the SH westerlies. The SST response from Ferreira et al. (64) was multiplied by two to better match Antarctic water isotopologs. Spatial pattern of the differences in (C) modeled surface temperature, (D) δ¹⁸O (D), and (E) d_ln simulated in iCAM5 before and after imposing SST temperature forcing as shown in (B). Location, abbreviation, and HSs proxy value of each ice core are indicated on each map by the colored dots. (F) Scatter plots showing the magnitude of change in WD CO₂ concentrations and HSs proxies associated with HSs 1 to 5 versus background values for CO_2, d_ln, and δ¹⁸O).