The Last Interglacial period (LIG), approximately 129,000 to 116,000 years ago (ka), serves as a valuable analogue for understanding future warmer conditions on Earth. Previous research in central Europe suggests summer temperatures during the LIG were 1–2 °C warmer than present, with alpine regions potentially experiencing even greater warming. However, the exact magnitude, timing, and spatial variability of LIG climate remain debated due to a scarcity of high-resolution, independently dated records spanning the entire period. Most existing temperature reconstructions are based on biogenic archives, potentially introducing biases and uncertainties in age control. This study aims to address these limitations by reconstructing temperatures using fluid inclusions extracted from precisely dated speleothems. Fluid inclusions, trapped within speleothems during their growth, provide a reliable record of past environmental conditions, and the analysis of hydrogen isotopes (δD) in the inclusion water offers a precise means of calculating paleotemperatures. The research utilizes δD data from speleothems collected from two subalpine caves in Switzerland to provide a centennial to millennial-scale temperature reconstruction.

Existing studies on the LIG climate in central Europe have yielded varying results regarding the magnitude of temperature changes. Some studies suggest summer temperatures were approximately 1–2 °C warmer than present during the LIG optimum, while others indicate potentially higher increases in alpine regions (+4.3 ± 1.6 °C). There's a growing consensus on the presence of millennial- to centennial-scale climate swings during this period, but the overall spatial variability and the magnitude of temperature changes remain highly debated. The lack of independently dated records covering the entire LIG, especially those with sufficient resolution to capture subtle climate oscillations, has hampered a comprehensive understanding. Existing temperature reconstructions primarily rely on biogenic archives (e.g., pollen data, chironomids), which might introduce biases. Furthermore, the lack of precise age control in many biogenic proxies leads to poorly constrained temperature fluctuations. This study directly addresses these limitations through the use of precisely dated speleothems and their fluid inclusion water isotopes.

The study focuses on four speleothems collected from two caves in the Schrattenkarst region of central Switzerland. These caves are located at high altitudes (1716-1727 m a.s.l.), offering a unique perspective on alpine climate. The speleothems were precisely dated using U-Th dating techniques. A total of 38 powdered calcite samples were analyzed using a ThermoFisher Neptune Plus multi-collector inductively coupled plasma mass spectrometer. The age data were processed using OxCal 4.3 software using a Bayesian approach. A time-depth model was constructed to determine the age of each layer within the speleothems and to identify any hiatuses in growth. Stable isotope analysis (δ¹³C and δ¹⁸O) was conducted using a ThermoFisher Delta V isotope ratio mass spectrometer. A total of 2905 stable isotope samples were analyzed, providing a high temporal resolution record. Fluid inclusions were extracted from the speleothems, and the hydrogen isotope ratios (δD) in the water from these inclusions were analyzed using continuous-flow analysis. A total of 28 calcite blocks were analyzed. The precision of replicate measurements of the in-house calcite standard was typically 1.5% for δD. Paleotemperatures were reconstructed using the modern-day regional water isotope-temperature relationship from the Grimsel Pass GNIP station. This station is close geographically and shares a similar elevation with the cave site. The δ¹⁸O/annual air temperature slope was determined from the Grimsel Pass GNIP station for the period spanning 1971–1990. A range of possible transfer function scenarios (0.6-0.7‰/°C), accounting for the error in the Grimsel Pass transfer function, were considered. A δD/T transfer function was determined by converting the annual air temperature slope using a factor of eight, based on the global meteoric water line relationship. The δD values were corrected for the ice volume effect. The modern uplift rates in the study area were considered and incorporated in the calculation to adjust the SKR-FIT values accordingly. The study also includes the analysis of a Holocene stalagmite from the same cave to test the validity of the method.

The study's primary finding is a reconstruction of the temperature record (SKR-FIT) from the Swiss Alps during the LIG, showcasing temperatures that were significantly higher than present-day values. The SKR-FIT record indicates temperatures up to 4.3 ± 1.4 °C higher than the 1971–1990 average. This high-temperature period occurred in two distinct intervals: 127.3–125.9 ka and 124.6–124.1 ka. During these periods, median temperatures were 3.6 ± 1.4 °C above the modern average. The early LIG thermal maximum was disrupted by a prominent cooling event between 125.8 ± 0.5 and 124.6 ± 1.0 ka. This cooling episode, characterized by a 1.9 ± 1.4 °C decrease centered at 125.5 ± 0.5 ka, is consistent with evidence from the North Atlantic and represents the first well-constrained evidence of such cooling in central Europe, suggesting the impact of this cold event was more widespread than previously thought. Despite this cooling, temperatures remained above modern values (0.5–1.8 ± 0.8 °C) even during the glacial inception (122–115 ka). This higher-than-modern temperature persists into the late MIS 5e, consistent with evidence of continued heat transport into the North Atlantic. A gradual climate deterioration is revealed through cooling conditions between 124.2 ± 0.5 and 115.8 ± 0.4 ka.

The exceptionally warm conditions during the LIG, along with the observed climate instability, have significant implications for understanding future climate change. The significantly higher temperatures at high elevations compared to lower elevation pollen-based reconstructions suggest that high-altitude sites are more sensitive to warming. This finding aligns with previous research from other high-elevation sites and suggests that future warming could be amplified at higher altitudes. The study’s evidence of a widespread cooling event (C27) comparable in magnitude to the 8.2 ka BP event highlights potential future climate instability associated with increasing temperatures and a potential reduction in NADW formation due to accelerated melting of the Greenland Ice Sheet. While direct comparisons between the two warm periods (LIG and pre-industrial Holocene) are limited by differing orbital settings, the findings suggest a pattern of high temperatures and climate instability. These factors are particularly alarming in the context of current anthropogenic greenhouse gas emissions and their potential to exacerbate warming and associated climate variability.

This study provides the first well-constrained, high-resolution temperature reconstruction for the European Alps during the Last Interglacial period, demonstrating exceptionally warm conditions and significant climate instability. The finding of amplified warming at higher elevations highlights the sensitivity of alpine regions to climate change and carries implications for future climate projections. The identified widespread cooling event further emphasizes potential for future climate instability linked to changes in North Atlantic Deep Water circulation. Future research should focus on refining the understanding of regional climate dynamics and exploring the potential for similar climate instabilities under future greenhouse gas forcing scenarios.

The study is limited by the spatial extent of the temperature reconstruction; it is specific to the Swiss Alps and may not be fully representative of the broader European Alps. The paleotemperature reconstruction relies on the assumption of a constant lapse rate between the LIG and present day, and changes in seasonality and atmospheric circulation are not incorporated. While the study attempts to address uncertainties by considering a range of transfer functions, additional sources of error might exist due to the complexities of reconstructing past climate from proxy data.