Rapid northern hemisphere ice sheet melting during the penultimate deglaciation

Retreating ice sheets during glacial terminations release large meltwater fluxes that raise sea level and can trigger strong ocean–atmosphere feedbacks, especially if freshwater reaches North Atlantic deep-water formation regions. While Termination I (TI) is well studied, the timing, rates, and feedback sequence of earlier terminations are poorly constrained due to limited absolute chronologies. The penultimate deglaciation (Termination II, TII) is particularly important because orbital forcing differed from TI and the subsequent interglacial reached a higher sea-level stand, implying greater ice-sheet retreat despite similar atmospheric CO2. This study asks: when did Northern Hemisphere ice sheets melt during TII, how rapidly did they retreat, and how did associated freshwater forcing influence regional climate and AMOC? The authors propose that coastal NW Iberian speleothems capture the δ18O of the proximal eastern North Atlantic moisture source, enabling a direct, absolutely dated reconstruction of surface-ocean freshening and regional air-temperature changes to elucidate deglacial feedback mechanisms during TII and to compare them with TI.

Prior work has extensively characterized TI timing and feedbacks, identifying meltwater-induced AMOC slowdowns and associated climate responses, but absolute chronologies for earlier terminations remain scarce. Some studies suggest millennial feedback sequences differed in prior terminations. TII followed different orbital boundary conditions and led to an interglacial sea-level highstand (+1.2 to 5.3 m) indicating larger ice-sheet retreat (Greenland and Antarctica) despite similar CO2 highs. Marine records and modeling have been used to infer meltwater routes and δ18O_sw evolution, but Mediterranean and inland European speleothems show additional hydrological effects that complicate direct source-water δ18O reconstructions. Reconstructions of Eurasian Ice Sheet (EIS) extent suggest larger marine-based sectors during the penultimate glacial than at the LGM, implying greater sensitivity to ocean forcing. Benthic δ18O stacks tuned to insolation have been used to estimate global ice volume and sea level, but such tuning may bias deglacial rate estimates for TII. AMOC proxies (e.g., Pa/Th) and IRD records indicate complex deglacial variability; however, quantitative AMOC reconstructions for TII are limited and the relative roles of freshwater forcing versus circulation thresholds remain debated.

- Archive and site: The North Iberian Speleothem Archive (NISA) from caves within 10 km of the Atlantic coast at <70 m elevation in NW Iberia. Multiple stalagmites spanning the last 25 kyr (TI) and TII were analyzed to build δ18O and δ13C records. - Proxies and rationale: δ18O_NISA reflects the δ18O of the proximal eastern North Atlantic surface waters (δ18O_sw) transferred to rainfall and dripwater due to dominant oceanic moisture sourcing and minimal inland/elevation distillation. δ13C_NISA is used as a temperature-sensitive proxy via vegetation productivity and soil CO2 effects on dripwater carbon, supported by inverse correlations with regional SST. - Modern process validation: Back-trajectory moisture source analysis (HYSPLIT) for 104 rain events (2015–2016) near the cave site identified dominant moisture uptake in the eastern North Atlantic north of Iberia, confirming proximal source dominance. - Chronology: U–Th (230Th) dating on MC-ICP-MS at Xi'an Jiaotong University and University of Minnesota; additional 14C dating (ETH Zurich) where needed. Age models developed with Bchron and linear interpolation for short phases. Annual fluorescent laminae counted using confocal laser scanning microscopy to refine growth rates and resolve centennial-scale changes, especially in stalagmite Garth spanning 135–112 ka. - Geochemistry: High-resolution micromilling/drilling for δ18O and δ13C; analytical precision ~0.08‰. Mg/Ca measured by ICP to derive a PCP index, screening for in-cave processes (CO2 degassing, prior calcite precipitation) and isolating robust climate signals. Sections with evident PCP influence were excluded or complemented by other stalagmites. - Marine–speleothem integration: Demonstrated strong correlation over 25–5 ka between δ18O_NISA and δ18O_sw from Irish, W Iberian, and S Iberian margins (best r2=0.91 with Irish Margin), validating δ18O_NISA as a surface freshening proxy. For TII, marine core age models (ODP 977, 976; MD01-2444, etc.) were tuned to the speleothem chronology by aligning major freshening and cooling events. A derived metric δ18O_NISA-BC (surface δ18O anomaly relative to Iberian Margin benthic δ18O) was used to quantify regional freshening while minimizing global ice-volume uncertainty. - Comparative analysis: Constructed TI vs TII syntheses of δ18O_NISA, δ13C_NISA, benthic δ18O, alkenone SST, sea-ice-sensitive alkenone (C37:4) abundances, and melt-rate estimates from sea-level reconstructions. Examined IRD records (ODP 983, 984) and subpolar hydrographic indicators to infer AMOC behavior. - Modeling/context: Considered HadCM3 simulations of early TI meltwater routing/salinity anomaly and dynamical ice-sheet model scenarios suggesting a larger marine-grounded EIS during the penultimate glacial.

- Validation of proxy system: Over the last 25 kyr, δ18O_NISA is tightly correlated with eastern North Atlantic δ18O_sw, strongest with the Irish Margin (r2=0.91), indicating NISA stalagmites robustly record proximal surface-ocean freshening. δ13C_NISA inversely tracks regional SST (r2≈0.82–0.87), supporting its use as a temperature proxy. - TII onset and structure: δ18O_NISA indicates deglacial freshening of the eastern North Atlantic beginning at 135.7 ka. Two large, rapid freshening pulses mark the onset (MWP TII-A and MWP TII-B). Each began during rapid warming and was followed by abrupt regional cooling, consistent with freshwater-induced AMOC slowdowns. The first slowdown lasted ~600 years (shorter than HE1 in TI); the second corresponds to HE11. - Persistence of freshening: Following MWP TII-B, δ18O_NISA remained more negative than the final interglacial baseline for ~3000 years (134.7–131.2 ka), implying sustained meltwater addition exceeding oceanic mixing capacity. Rapid reduction of the anomaly after ~131 ka coincides with decreased IRD and a slowdown in the rate of benthic δ18O decrease, indicating reduced melt rates. - Late deglacial events: A rapid negative δ18O_NISA shift at ~129.7 ka during warming indicates a melt acceleration. Continued meltwater flux through ~129.3 ka coincides with a distinctive Labrador margin sediment layer attributed to a NAIS outburst flood, suggesting most NAIS melting was complete by then. - Deglacial rates: On the W Iberian margin, benthic δ18O reached interglacial values within ~5 kyr of TII onset versus ~9 kyr in TI, implying more rapid deep-ocean freshening/warming during TII. Applying the speleothem-tuned chronology yields higher and earlier maximum rates of sea-level rise/freshwater forcing compared to chronologies tuned only to insolation or the original global benthic stack. - TI vs TII contrast: The eastern North Atlantic δ18O surface anomaly below −1‰ persisted ~3.5 kyr in TII but only ~1.2 kyr (16.4–15.2 ka) in TI. The TII anomaly is consistent with a larger contribution from a marine-based EIS, which produces a stronger, longer-lived freshening in the eastern North Atlantic than comparable NAIS meltwater routing. - AMOC dynamics: Annual-layered records show enhanced freshwater addition preceded and likely caused early TII AMOC reductions. A mid-termination AMOC recovery occurred after the first slowdown despite relatively low CO2 (on ice-core chronology), suggesting sensitivity to temporal variations in meltwater forcing and/or differing boundary conditions. Proxy indicators imply HE11 had a less extreme AMOC/sea-ice response than HE1, arguing against circulation differences alone explaining the longer TII freshening anomaly. - Carbon cycle timing: A marine boron-isotope CO2 proxy (western Caribbean) aligned to the speleothem chronology indicates a potentially earlier and rapid CO2 rise coincident with rapid EIS melting and the second AMOC slowdown, which may have sustained warming and melting during HE11. - Ice-sheet configuration: The magnitude and persistence of TII freshening, IRD peaks, and rapid benthic δ18O changes support scenarios of a substantially larger, more marine-grounded EIS during the penultimate glacial (60–71 m s.l.e. EIS), enhancing susceptibility to rapid retreat via marine ice-sheet processes.

The findings establish that NW Iberian speleothems provide an absolute, high-resolution record directly linking North Atlantic freshening to regional air-temperature changes and AMOC variability during TII. Deglaciation began with rapid, century-scale meltwater pulses that triggered AMOC slowdowns, coupling freshening and cooling. The prolonged TII surface freshening relative to TI is best explained by a larger marine-based Eurasian Ice Sheet, whose geometry would both amplify the local eastern North Atlantic δ18O anomaly and predispose the system to rapid retreat via ocean-forced instability. Although reduced AMOC can amplify and prolong freshening by slowing vertical mixing, proxy comparisons suggest HE11 was not more extreme than HE1 in circulation/sea-ice response, indicating source-region meltwater effects dominated the TII anomaly. The sequence of AMOC reductions and a recovery after the first slowdown implies that temporal variability in meltwater flux, and possibly evolving thresholds due to greenhouse forcing and boundary conditions, controlled AMOC state. Potentially earlier, rapid CO2 rise aligned with major melt phases could have contributed to sustained interstadial warming and continued melt despite AMOC weakening. Collectively, these results refine TII deglacial rates and mechanisms, underscore the key role of ice-sheet anatomy (marine grounding) in pacing melt and feedbacks, and provide a framework to reassess sea-level and carbon-cycle chronologies for the last interglacial onset.

This study delivers the first precisely dated, high-resolution record of Northern Hemisphere meltwater pulses during Termination II from NW Iberian speleothems, directly tying rapid surface-ocean freshening to regional air-temperature changes and AMOC variability. Key contributions include: (1) identification of two initial, century-scale meltwater pulses and associated AMOC slowdowns (HE11′ and HE11), with the first lasting ~600 years; (2) documentation of a ~3 kyr-long extreme freshening anomaly unique to TII; (3) evidence that TII deep-ocean/benthic δ18O reached interglacial values ~4 kyr earlier than in TI; and (4) support for a substantially larger, marine-based Eurasian Ice Sheet driving more rapid and sustained freshwater input. These results imply that ice-sheet configuration exerts strong control on deglacial rates and feedbacks, with implications for future marine-based ice-sheet stability under warming. Future work should: (a) integrate coupled ocean–atmosphere–ice-sheet models with realistic meltwater routing to quantify AMOC thresholds and recovery under varying forcings; (b) assimilate near-field geophysical constraints to refine PGM EIS reconstructions; (c) obtain higher-precision TII CO2 chronologies (ice and marine proxies) to resolve phasing with melt and AMOC; and (d) expand speleothem networks along Atlantic margins to map spatial gradients in meltwater signals and temperature responses.

- AMOC intensity during TII is inferred indirectly from multiple proxies (e.g., IRD, foraminiferal assemblages, C37:4 alkenones); quantitative, continuous AMOC reconstructions are lacking. - Marine age models for TII were tuned to speleothem chronology; while justified by shared events, tuning introduces uncertainties in absolute phasing and amplitude preservation among different proxies and sites. - Some stalagmite intervals show evidence of prior calcite precipitation (PCP) and condensed growth, necessitating splicing between records; although screened by Mg/Ca and growth-layer analyses, residual in-cave effects cannot be entirely excluded for specific intervals (e.g., Garth 128–122 ka). - The Neith stalagmite has low U content and relies on tuning to Garth for detailed structure, limiting its independent chronological constraint. - Ice-core CO2 chronology during TII may underestimate the rate/timing of CO2 rise; alignment with marine proxies carries additional uncertainties. - The study cannot confirm or rule out contributions from a hypothesized Arctic ice shelf; its likely modest volume suggests it is not the dominant cause of the prolonged TII freshening anomaly.