Subsurface ocean warming preceded Heinrich Events

Heinrich Events, marked by layers of ice-rafted debris (IRD) in the glacial North Atlantic, reflect substantial freshwater releases linked to instabilities of the Laurentide Ice Sheet (LIS). While the climatic impacts of these events are well established, their triggers remain debated. The classic binge–purge hypothesis posits internally driven LIS oscillations, with freshwater discharge subsequently disrupting the Atlantic Meridional Overturning Circulation (AMOC). However, proxy evidence indicates that North Atlantic surface cooling and weakening of deep ocean circulation preceded IRD deposition, suggesting ocean circulation changes may prime ice-sheet instability. Model studies predict that reduced AMOC can cause strong subsurface warming and rapid retreat of marine-terminating ice margins around the Labrador Sea, but direct subsurface proxy records near this region were lacking. This study targets that gap by reconstructing subsurface temperature and salinity from a western subpolar North Atlantic core (GeoB18530-1) to test whether subsurface ocean warming systematically precedes Heinrich Events and to assess links to AMOC variability.

Prior work has extensively documented Heinrich Events and their global climatic effects, including freshwater-driven AMOC disruptions and surface cooling. The binge–purge mechanism has been proposed as an internal LIS driver, but timing mismatches between surface cooling and IRD deposition, and evidence for AMOC weakening prior to Heinrich Events, challenge a purely ice-internal trigger. Numerical models suggest reduced AMOC leads to subsurface warming that can destabilize marine ice margins (particularly around the Labrador Sea) and trigger Heinrich-like surges. Benthic temperature reconstructions indicate mid-depth North Atlantic warming prior to Heinrich Events. Yet, until now, subsurface proxy records near the LIS grounding line region were unavailable, preventing direct evaluation of the subsurface warming trigger mechanism.

Study site and material: Marine sediment core GeoB18530-1 (42°50′N, 49°14′W; 1888 m water depth) was collected east of Newfoundland at the southern boundary of the subpolar gyre, within the North Atlantic IRD belt and near the main iceberg trajectories from Hudson Strait. High temporal resolution (~250 years on average) records cover the last 27 ka. Proxies and measurements: Subsurface temperatures (~150 m) were reconstructed using Mg/Ca ratios of the subsurface-dwelling planktonic foraminifer Neogloboquadrina pachyderma sinistral (N. pachyderma sin.). Approximately 100 individuals per sample (>250 μm fraction) were cleaned (following Barker et al., 2003) and analyzed by ICP-OES (Varian 720 ES) at GEOMAR. Mg/Ca values were normalized to ECRM 752-1 and drift-corrected; analytical precision for Mg/Ca was ±0.01 mmol/mol. Fe and Al were measured to monitor silicate contamination; elevated Al/Ca in a few samples did not affect Mg/Ca trends. Stable oxygen isotopes (δ18O) of N. pachyderma sin. were measured on Thermo MAT 253 mass spectrometers with Kiel IV devices at MARUM and GEOMAR, calibrated to NBS19 and an in-house standard; long-term precision 0.06‰. Subsurface temperatures (subSSTMg/Ca) were calculated using a species-specific calibration deemed most suitable for the site (Kozdon et al., 2009): Mg/Ca = 0.13(±0.037)·T + 0.35(±0.17). Core-top comparison against instrumental profiles supported this choice. Salinity proxy: Ice-volume-corrected seawater oxygen isotopic composition (δ18Oic-sw) was calculated by removing global ice volume effects (relative sea-level curve of Waelbroeck et al., 2002) and subtracting the temperature component using the Shackleton (1974) temperature–δ18Ocalcite equation. δ18Ow values were converted between VPDB and VSMOW scales (Hut, 1987). Error propagation incorporating analytical uncertainties, calibration uncertainty, and δ18O–salinity relationship yielded an uncertainty of ~0.38‰ for δ18Oic-sw. IRD identification: XRF core scanning (Avaatech, MARUM) at 1 cm resolution quantified elemental intensities. Calcium-to-strontium (Ca/Sr) ratios were used as a proxy for detrital carbonate typical of Heinrich layers sourced from Paleozoic carbonates in Hudson Bay/Strait. Elevated Ca/Sr identified Heinrich Layers (HE1, HE2) and the Younger Dryas interval. Chronology: Twenty AMS 14C dates (19 on N. pachyderma sin., one mixed with G. bulloides) were obtained (Poznań; University of Bern). Ages were calibrated with IntCal20, using modelled local reservoir ages via PaleoDataView. The age model used BACON with 2000 age–depth realizations; proxy uncertainties were combined with age model ensembles to compute 95% confidence envelopes. Reported key tie points include calibrated median ages of ~17.1 ka BP for the onset of HE1 and ~25.4 ka BP near the HE2 warming peak. Comparative records were re-calibrated consistently where applicable (e.g., MD01-2461). Sensitivity tests: The potential influence of variable habitat depth (50–200 m) and seasonality shifts was assessed using modern hydrographic profiles near the site. The temperature gradient between 50 and 200 m was ~2.05°C (September), much smaller than observed subsurface anomalies (up to ~6°C). A maximal seasonality shift (April to September) would change 150 m temperature by ~0.25°C, also minor relative to signal magnitude. Comparative datasets: Subpolar NE Atlantic subsurface temperatures (MD01-2461), subtropical NE Atlantic alkenone SSTs (SU8118), AMOC strength proxies (231Pa/230Th from OCE326-GGC5 and ODP 1063), and the NGRIP δ18O record for Greenland air temperature were used for regional context and for assessing AMOC–subsurface temperature relationships.

• Repeated, rapid subsurface warming in the western subpolar North Atlantic preceded each Heinrich Event identified in the same core over the last 27 ka. The warmest subsurface temperatures coincide with the onset of IRD deposition.
• Quantitatively, subsurface temperatures rose to ~8.4°C (HE2 onset) and ~12.5°C (HE1 onset), compared to modern ~7°C at ~150 m near the site.
• Calibrated ages constrain the onset of HE1 at ~17.1 ka BP and the HE2 warming peak at ~25.4 ka BP. Subsurface warming initiated during AMOC slowdowns at ~25.9 ka BP (HS2), ~18.6 ka BP (transition to HS1), and ~12.5 ka BP (onset of Younger Dryas).
• Following Heinrich Event onsets, subsurface waters rapidly cooled and freshened, consistent with massive meltwater input. During HE1, a secondary warming and salinification occurred later in the event, aligning with the recognized subdivision into HE1.1 (17.1–15.5 ka BP) and HE1.2 (15.9–14.3 ka BP).
• Subsurface warming is not reflected in Greenland air temperatures (NGRIP) or North Atlantic surface SSTs; instead, surface SSTs cool when subsurface warms, indicating strong surface–subsurface decoupling during stadials and prior to Heinrich Events.
• Early subsurface warming is also detected in the eastern subpolar North Atlantic (MD01-2461), most pronounced before HE1, suggesting basin-wide interior heat storage preceding IRD deposition.
• AMOC proxy records show a weakening during Heinrich Stadials 2 and 1, starting 1–2 kyr before Heinrich Events, temporally aligned with the build-up of subsurface heat at the study site.
• The Younger Dryas shows delayed and modest IRD deposition relative to subsurface warming, consistent with it being atypical (nonlinear, modest freshwater forcing and AMOC weakening).

The findings establish a robust temporal sequence: AMOC weakening during Heinrich Stadials led to subsurface heat accumulation in the subpolar North Atlantic, with peak subsurface temperatures coinciding with the onset of IRD deposition. This pattern supports a mechanism in which reduced deep-water convection redistributes heat within the Atlantic, cools and freshens the surface, and isolates subsurface waters, allowing interior warming via gyre-mediated advection. The resulting temperature inversion enables contact of anomalously warm waters with marine-terminating ice near grounding lines, promoting ice-shelf thinning, grounding-line retreat, acceleration of ice flow, and iceberg surges that produce Heinrich Layers. The absence of concurrent surface warming and the cross-basin expression of subsurface warming indicate strong stratification and interior heat storage. The sequence and magnitudes are consistent with model predictions of 4–6°C interior warming during AMOC slowdowns and with prior mid-depth warming reconstructions. The Younger Dryas deviations likely reflect different freshwater routing and weaker, non-linear AMOC perturbations. Overall, the results demonstrate that subsurface ocean warming acted as a proximate trigger for LIS instabilities during Heinrich Events, with AMOC variability as a key precursor.

This study provides the first direct subsurface (∼150 m) temperature and salinity reconstructions near the western subpolar North Atlantic that demonstrate rapid subsurface warming systematically precedes Heinrich Events over the last 27 ka. Peak warming aligns with the onset of IRD deposition, implicating subsurface heat as a trigger for marine-terminating LIS instabilities. The build-up of this interior heat reservoir coincides with AMOC slowdowns, verifying model-based mechanisms of stratification and interior warming during reduced overturning. These insights clarify the ocean–ice-sheet coupling underlying Heinrich Events and highlight that future AMOC weakening could amplify subsurface warming in the subpolar Atlantic, posing risks to modern Arctic marine-terminating glaciers and North Atlantic freshwater budgets. Future research should expand spatial coverage of subsurface reconstructions near critical ice-margin regions, refine age constraints, integrate multiproxy salinity and density reconstructions, and couple high-resolution proxy records with targeted ocean–ice-sheet model experiments to quantify thresholds and rates of ice-margin response to subsurface heat anomalies.

• Geographic and archive coverage: Conclusions are based on a single high-resolution core in the western subpolar North Atlantic, though comparisons to additional sites support basin-wide behavior. More spatially distributed subsurface records near the Labrador Sea and Hudson Strait would strengthen generality.
• Chronological uncertainty: While the onset of subsurface warming relative to IRD deposition is determined within the same core (minimizing relative dating offsets), absolute age uncertainties remain (hundreds of years). The AMOC lead time of 1–2 kyr exceeds typical age model uncertainties but still carries calibration and reservoir age uncertainties.
• Proxy and calibration uncertainties: Mg/Ca–temperature calibration for N. pachyderma sin. carries uncertainty; sensitivity tests for habitat depth (50–200 m) and seasonality shifts indicate small impacts relative to signal magnitude. δ18Oic-sw estimates include propagated errors (~0.38‰) from measurements and calibrations.
• Event variability: The Younger Dryas differs from classic Heinrich Events, with modest, nonlinear freshwater forcing and AMOC weakening, complicating direct comparison of mechanisms and timings.