Coupled atmosphere-ice-ocean dynamics during Heinrich Stadial 2

The study addresses how high- and low-latitude climate systems were coupled during Heinrich Stadial 2 (HS2) in the Last Glacial Maximum, a period with distinct boundary conditions including low greenhouse gas concentrations and sea level. Prior work identifies Dansgaard–Oeschger oscillations and Heinrich events, with AMOC slowdowns producing a bipolar seesaw between Greenland and Antarctica. However, precise interhemispheric phasing during the LGM remains unresolved due to large age uncertainties in Greenland ice cores (±500–800 years) and a scarcity of precisely dated, high-resolution low-latitude records. The authors aim to build a robust chronology linking tropical monsoon changes to high-latitude signals, test whether atmospheric teleconnections at HS2 onset were synchronous across hemispheres, quantify phase relationships at onset and termination of AHP2/SAHP2, and assess the low-latitude hydroclimate’s role in initiating and terminating HS2-related changes.

Previous research has documented repeated millennial-scale stadials and interstadials in Greenland ice cores (Dansgaard–Oeschger events) and six Heinrich events marked by iceberg discharges in the North Atlantic. Marine and modeling studies implicate AMOC weakening or shutdown during Heinrich stadials, producing reduced northward heat transport and out-of-phase temperature changes between Greenland and Antarctica (the bipolar seesaw). Evidence suggests rapid atmospheric teleconnections, with signals propagating from northern high latitudes to the tropics and southern high latitudes on sub-centennial timescales. Speleothem records from Asia and South America indicate monsoon weakening (AHP) and strengthening (SAHP), consistent with southward ITCZ shifts, and some studies point to tropical hydroclimate actively influencing AMOC recovery via Amazon runoff and Atlantic salinity anomalies. Nonetheless, Greenland ice δ18O often shows muted Heinrich signatures, and LGM ice-core chronologies bear substantial uncertainties. This motivates the need for precisely dated low-latitude records to refine interhemispheric phasing and assess tropical-atmospheric contributions during HS2.

• Data acquisition: Nine speleothem δ18O records spanning 27–22 ka BP from monsoon domains were produced, including two high-resolution, annually laminated records from Cherrapunji Cave (India) and additional records from Mawmluh (India), Yongxing and Dongqinghe (China), and Marota, Paixão, Botuverá (Brazil), plus updated 230Th dates for NAR-C (Peru).
• Dating and age models: 127 230Th dates underpin chronologies. Cherrapunji-2 and Cherrapunji-2017-1 age models combine annual lamina counting (validated via confocal laser fluorescence microscopy) with 230Th dates, yielding <50-year age uncertainty and ~4-year resolution in key intervals. Other records use the StalAge algorithm, with Monte Carlo ensembles defining uncertainties; alternative age-modeling schemes tested yielded consistent results.
• Stable isotope analyses: ~2810 δ18O (and δ13C for Cherrapunji) measurements using a Thermo-Finnigan MAT253 IRMS; analytical precision better than 0.1‰. Spatial sampling resolutions ranged from 0.05 to 1 mm.
• Proxy interpretation: In ASM/SASM regions, speleothem δ18O reflects large-scale monsoon circulation intensity and ITCZ shifts at millennial-to-orbital scales. Greenland ice-core [Ca2+] is used as a dust proxy reflecting Asian westerly winds and hydroclimate in dust source regions, which are dynamically linked to ASM circulation.
• Chronology alignment: Greenland [Ca2+] time series on GICC05 were tuned to the precisely dated Cherrapunji δ18O using six prominent, statistically robust tie points identified by BREAKFIT and supported by Trend-fitting; this implies a uniform +320-year shift of GICC05 between 27–23 ka BP. Sensitivity tests and z-score analyses corroborate tie point robustness.
• Antarctic synchronization: A unique volcanic triplet spike, attributed to the Oruanui eruption, identified in both Greenland and Antarctic ice cores, was used to align chronologies. With GICC05 shifted by +320 years, the volcanic triplet at 24,939 ± 90 y BP in Greenland requires shifting WD2014 by +400 years (including an earlier suggested +80-year adjustment). The adjusted WD2014 aligns WDC tephra at 25,718 y BP with radiocarbon-calibrated ages (25,675 ± 90 cal y BP).
• Change-point detection: Applied Ramp-fitting (MCMC-based two-change-point model) and BREAKFIT (single breakpoint with bootstrap uncertainties) to identify onsets, rebounds, and terminations in speleothem and ice-core series. Combined uncertainties were computed by quadratically adding method-derived change-point and age-model uncertainties. Trend-fitting served as an independent check. Sensitivity tests varied search windows; most change points were robust.
• Additional analyses: Probability density functions of spatial age offsets assessed inter-regional phasing. Marine records previously tied to GICC05 were uniformly shifted by +320 years to compare with speleothem chronologies across HS2. Data and MATLAB code for trend fitting are provided in supplementary materials and NOAA repository.

• Chronology refinements:
  • Greenland: GICC05 requires a +320-year shift during 27–23 ka BP, based on six robust tie points between Greenland [Ca2+] and Cherrapunji δ18O; residual uncertainty ~90 years (2σ).
  • Antarctica: WD2014 requires a +400-year shift near 26–23 ka BP, consistent with a bipolar volcanic triplet alignment and radiocarbon-dated Oruanui tephra; gas chronology shifted similarly given small Δage.
• Synchronous global onset: The HS2/AHP2/SAHP2 onset is synchronous globally within sub-centennial uncertainties. In Cherrapunji, AHP2 onset begins at 24.42 ± 0.04 ka BP with an abrupt ~+2‰ δ18O increase lasting 76 ± 5 years, followed by a rapid rebound starting at 24.33 ± 0.02 ka BP lasting 139 ± 2 years (−1‰), then a stable Stage II (24.19 ± 0.02 to 23.87 ± 0.02 ka BP), and a gradual termination (Stage I) starting 23.87 ± 0.02 ka BP over ~350 years.
• Tropical atmospheric seesaw: A pronounced, rapid rebound (Stage III) occurs within decades (lamina-count constrained) in low-latitude monsoon records (ISM and SASM), with anti-phase δ18O excursions indicative of ITCZ meridional shifts. This feature is absent in NH mid-to-high-latitude proxies (Greenland [Ca2+], EASM δ18O, NGRIP d-excess) and in SH mid-high-latitude records, evidencing a low-latitude atmospheric tipping-point behavior superimposed on the bipolar seesaw.
• Methane linkage: A −40 ppb CH4 increase coincides with Stage III and the DO-2.2 peak on the adjusted chronologies, consistent with enhanced SH tropical wetland emissions early in AHP2.
• Termination phasing: SASM shows a prolonged drying trend through Stage II, with SAHP2 terminating hundreds of years before AHP2 (SASM termination precedes ASM by ~300 years). After SAHP2 termination (~23.85–24.05 ka BP), SASM records indicate peak aridity, whereas ASM remains weak through Stages I–II.
• Ocean–atmosphere coupling in termination: SASM drying implies reduced Amazon discharge, fostering positive sea-surface salinity anomalies in the Amazon Plume and downstream advection to North Atlantic deep-water formation regions, aiding AMOC strengthening. Independent marine SSS reconstructions in the eastern subpolar North Atlantic show increasing trends during AHP2 termination, consistent with this mechanism and subsequent ASM intensification and Greenland warming.
• Conceptual advance: Results highlight an active tropical atmospheric role during HS2, requiring an expansion of the bipolar seesaw concept to include rapid tropical atmospheric feedbacks and low-latitude processes under LGM boundary conditions.

The refined, tightly constrained chronologies directly address long-standing phasing uncertainties during HS2 by aligning high-precision tropical speleothem records with ice-core and marine proxies. The synchronous onset across hemispheres within sub-centennial uncertainty supports a rapid atmospheric teleconnection driving immediate monsoon weakening (ASM) and strengthening (SASM) with a southward ITCZ shift. The discovery of a decadal-scale, tropical-only rebound (tropical atmospheric seesaw) indicates that low-latitude atmospheric processes can amplify or modulate millennial-scale perturbations from the North Atlantic independent of mid-high-latitude fingerprints, revealing a tipping-point behavior in the tropical hydroclimate system. During termination, the earlier SASM drying and SAHP2 end relative to AHP2, alongside increasing North Atlantic SSS and subsequent AMOC recovery, identify a plausible causal pathway where tropical rainfall/runoff changes precondition ocean circulation resumption. Together, these findings reconcile disparate proxy responses, refine the temporal order of events, and emphasize that atmospheric and oceanic processes in the tropics and Southern Hemisphere actively shape millennial climate variability beyond a simple bipolar thermal seesaw.

This study constructs a high-precision interhemispheric chronology for HS2 by coupling annually laminated speleothem records with Greenland dust and Antarctic volcanic tie points, requiring +320-year (GICC05) and +400-year (WD2014) shifts with ~90-year (2σ) precision. It reveals a globally synchronous onset of HS2 and identifies a rapid, low-latitude “tropical atmospheric seesaw” at the beginning of AHP2/SAHP2 that is not expressed in mid-to-high-latitude proxies. The termination sequence shows SASM leading ASM by several centuries, consistent with an Amazon runoff–Atlantic salinity–AMOC feedback facilitating recovery. These results extend the bipolar seesaw framework to include critical low-latitude atmospheric dynamics under LGM conditions. Future work should target higher-resolution Antarctic water isotope records to resolve AIM2 phase relationships, additional independent volcanic/radiometric tie points for chronology validation, expanded tropical speleothem networks, and coupled climate model experiments isolating atmospheric versus oceanic contributions to abrupt events.

• Antarctic ice δ18O records near AIM2 have low signal-to-noise, preventing robust identification of some change points despite improved chronology.
• Chronology alignment uses tuning via tie points (Greenland [Ca2+] to speleothem δ18O, volcanic triplet), which, while cross-validated, carries potential for circular reasoning; uncertainties are conservatively estimated (~90 years, 2σ).
• Some SASM records lack dense 230Th control around SAHP2, limiting local precision; a large sub-centennial dry event in MAG required cropping for trend identification.
• Greenland δ18O has muted HS2 imprint, necessitating reliance on [Ca2+] as a dust/westerly proxy and the assumption of negligible lag between dust emission and Greenland deposition at decadal scales.
• Differences in transition sharpness between speleothem δ18O and [Ca2+] indicate unresolved process or archive response discrepancies requiring further study.