The rate and consequences of future high-latitude ice sheet retreat are major concerns given ongoing anthropogenic warming. Understanding past deglaciations, particularly the rapid ice sheet melting events, is crucial for predicting future climate change. While the last deglaciation (Termination I, TI) is well-studied, knowledge of previous terminations, like Termination II (TII), is limited due to the lack of precise dating. TII is particularly interesting because orbital forcing differed, and the subsequent interglacial had a higher sea level than TI, suggesting greater ice sheet retreat. This study aims to provide high-resolution records of rapid Northern Hemisphere ice sheet melting during TII using precisely dated stalagmite data from NW Iberia. The stalagmite δ¹⁸O and δ¹³C records serve as proxies for surface ocean δ¹⁸Osw (a measure of ocean freshening) and air temperature, respectively. By comparing these records with those from TI, this study seeks to understand the timing, rate, and feedbacks associated with ice sheet melting in these two deglaciations.

Previous research on the last deglaciation (TI) has extensively documented the timing and associated deglacial feedbacks. However, knowledge about previous terminations, especially TII, is less complete due to challenges in establishing accurate chronologies for ice retreat and warming. Studies suggest that the sequence of millennial-scale climate feedbacks may differ between TI and TII. The penultimate deglaciation (TII) is important because orbital boundary conditions were different and resulted in a higher interglacial sea level than TI, indicating extensive retreat of both Greenland and Antarctic ice sheets. To understand the higher sea level during the last interglacial, precise knowledge of ice sheet melting rates during TII is necessary. Studies have used changes in the δ¹⁸Osw of the North Atlantic to diagnose meltwater pathways during TI, and this study leverages a similar approach for TII. Speleothems from coastal European locations are expected to record the δ¹⁸Osw signal through transfer of the isotopic signal of the ocean moisture source to rainfall and drip water used for speleothem formation.

This study utilizes the North Iberian Speleothem Archive (NISA) from caves near the Atlantic coast. Six stalagmites spanning the last 25 ky were analyzed for their δ¹⁸O and δ¹³C isotopic composition. A composite splice speleothem δ¹⁸O record was created and compared to independent estimates of δ¹⁸Osw from North Atlantic marine sediment cores. The relationship between speleothem δ¹⁸O and δ¹⁸Osw was evaluated, particularly focusing on the strong correlation found between NISA δ¹⁸O and the δ¹⁸Osw of the eastern North Atlantic Ocean. The relationship between δ¹³CNISA and regional temperature records was also tested. For TII, three ²³⁰Th-dated stalagmites were used, with annual layer counting on one stalagmite (Garth) providing high-resolution temporal constraints. The potential effects of in-cave processes like Prior Calcite Precipitation (PCP) were evaluated using Mg/Ca ratios. To enhance the precision of the chronology of marine sediments, the marine sediment age models were tuned to the speleothem chronology by synchronizing major freshening events and key temperature events. This approach involved comparing the NISA records to the δ¹⁸Osw records from west and south Iberian marine sediments. The researchers employed HYSPLIT model to identify the dominant moisture source for precipitation near the study caves, ensuring that the speleothem δ¹⁸O records accurately reflected the marine δ¹⁸Osw signal. The study compared the speleothem chronologies for TI and TII to better understand the difference in melting rates and processes between these two events. The researchers used various statistical methods to identify and quantify correlations between different datasets (e.g. speleothem δ¹⁸O and marine δ¹⁸Osw). In addition, they used models (e.g., HadCM3) to simulate the spatial patterns of meltwater-induced salinity anomalies during the early phases of the last deglaciation. The team also assessed various climate indicators and considered potential alternative explanations for the observed differences in meltwater anomalies between TI and TII. The study carefully considered potential sources of error and uncertainty in the data and interpretations.

The study's key findings include: 1. A strong correlation (r² = 0.91) was found between the NISA speleothem δ¹⁸O and the δ¹⁸Osw of the eastern North Atlantic Ocean over the past 25,000 years. This robust relationship allowed the researchers to use the speleothem record as a proxy for changes in North Atlantic ocean salinity. 2. The NISA speleothem records revealed two major meltwater pulses (MWPTII-A and MWPTII-B) at the onset of TII, both followed by abrupt temperature coolings, consistent with freshening-induced AMOC slowdowns. 3. Following MWPTII-A, a local freshwater anomaly was maintained in the eastern North Atlantic for approximately 600 years before warming resumed and the anomaly dissipated. 4. After MWPTII-B, an even more sustained freshwater anomaly persisted for about 3000 years. This requires sustained meltwater input, exceeding the rate at which it could be dispersed through ocean mixing. 5. The TII deglaciation involved late phases of melting between 130 and 129 ka, with a rapid negative δ¹⁸O shift around 129.7 ka coinciding with regional warming and potentially representing a final NAIS meltwater outburst. 6. The rate of depletion in δ¹⁸O benthic during TII was much faster than during TI, indicating more rapid freshening of deep waters in the North Atlantic during TII. 7. Both TI and TII deglaciations started at a similar threshold of summer insolation, but the TII deglaciation reached interglacial values much more rapidly. The duration of the eastern North Atlantic δ¹⁸O anomaly was significantly longer in TII (3.5 ky) than in TI (1.2 ky). 8. The longer duration of the meltwater anomaly in TII is attributed to a larger proportion of meltwater from the Eurasian Ice Sheet (EIS) compared to TI. The EIS during the penultimate glacial maximum was substantially larger than its last glacial maximum counterpart. 9. The study found multiple phases of AMOC reduction and recovery during TII. The first AMOC slowdown was shorter (600 years) than Heinrich event 1 in TI (~3000 years), potentially due to a negative feedback between AMOC slowdown and melting rate. 10. The study raises questions about the role of CO2 in AMOC recovery between TI and TII, suggesting the need to reassess the rate of CO2 rise during TII using data from marine proxies. The observed data indicate more intense and sustained meltwater input into the North Atlantic during the penultimate deglaciation, compared to the last deglaciation, underscoring the key role of ice sheet geometry in the process.

The findings of this study address the research question by providing the first high-resolution, precisely dated records of rapid ice sheet melting during TII. The exceptionally strong correlation between speleothem δ¹⁸O and marine δ¹⁸Osw demonstrates the efficacy of using NISA speleothems as proxies for ocean freshening in the eastern North Atlantic. The differences in the duration and intensity of the meltwater anomalies between TI and TII highlight the significance of ice sheet configuration in modulating the deglacial response. The larger marine-based EIS during TII was evidently more susceptible to rapid melting than the largely land-based NAIS during TI, leading to a more sustained and extensive freshening of the North Atlantic. The observation of multiple AMOC slowdowns and recoveries in TII provides new insights into the mechanisms of AMOC instability during glacial terminations, suggesting the interplay of meltwater forcing and potential changes in AMOC sensitivity thresholds. The findings emphasize the complex interplay between ice sheet dynamics, ocean circulation, and atmospheric CO2 in driving millennial-scale climate variability during deglaciations. The study suggests a need for reevaluating the CO2 rise chronology during TII. The results have implications for improving climate models used to project future sea level rise.

This study presents the first direct high-resolution records of rapid Northern Hemisphere ice sheet melting during the penultimate deglaciation, using precisely dated speleothems from NW Iberia. The findings reveal distinct meltwater pulses and AMOC slowdowns, differing from those observed during the last deglaciation. The longer duration and intensity of the meltwater signal during the penultimate deglaciation are attributed to the larger size and marine-based nature of the Eurasian Ice Sheet. Further research is needed to refine ice sheet models and to investigate the coupled interactions between ice sheet dynamics, ocean circulation, and atmospheric CO2 using high-resolution coupled ocean-climate-ice sheet models.

The study primarily focuses on the eastern North Atlantic region, limiting the ability to fully characterize global climate responses. While the authors carefully considered the potential influence of in-cave processes, some uncertainties remain in precisely quantifying the influence of prior calcite precipitation on speleothem isotopic compositions. The precise timing of AMOC changes remains challenging to independently verify with other proxies in TII, although several different proxies support the timing reported here. The study relies on tuning of marine age models to the speleothem chronology, introducing potential uncertainties in the absolute age of marine events. While the study offers compelling evidence for a larger Eurasian Ice Sheet during the penultimate glacial maximum, independent confirmation through additional geophysical data is necessary.