Understanding the impact of past climate variability on glacier fluctuations is crucial for discerning the roles of natural and anthropogenic factors in current and future glacier retreat. While modern observations provide high-frequency forcing data, paleoglacial extents from moraine records offer insights into millennial-scale behavior. Small mountain glaciers are particularly useful due to their rapid response times. Holocene glacier evolution has generally been linked to summer insolation, with Northern Hemisphere mid-latitude glaciers expanding during the mid-to-late Holocene, reaching a maximum during the Little Ice Age (LIA), while Southern Hemisphere glaciers retreated. However, recent moraine dating in the North Atlantic sector reveals early Holocene extents larger than those of the LIA, contradicting the insolation trend. Similarly, in the tropical Andes, the large-scale drivers of glacier behavior remain unclear, showing an irregular retreat trend despite increasing austral summer insolation. The absence of mid-Holocene moraines in some tropical areas might be due to their obliteration by later advances, suggesting additional forcing mechanisms beyond insolation. This study aims to address these discrepancies by creating more comprehensive regional and global glacier reconstructions using new and published cosmic-ray exposure (CRE) ages. The study also explores potential influences of other forcings, including greenhouse gas concentrations, ice-sheet disintegration, volcanic activity, land use changes, and variations in the Atlantic Meridional Overturning Circulation (AMOC). The AMOC's significant influence on global climate, through heat redistribution, is well-known, and recent research has documented its Holocene variations, but the impact on mountain glaciers remains unclear.

Existing research on Holocene glacier fluctuations highlights the complexity of the relationship between climate and glacier response. Studies focusing on summer insolation as the primary driver generally show a distinct pattern for the Northern and Southern Hemispheres, with Northern Hemisphere glaciers advancing during the mid-to-late Holocene and Southern Hemisphere glaciers retreating (Schaefer et al., 2009; Kaplan et al., 2013; Reynhout et al., 2019; Putnam et al., 2012). However, these studies often neglect other potential factors and regional variations. Recent studies focusing on the North Atlantic sector have demonstrated a more complex pattern with early Holocene glacier extents larger than LIA maxima (Schimmelpfennig et al., 2014; Protin et al., 2021; O’Hara et al., 2017; Schweinsberg et al., 2019), challenging the straightforward insolation-driven model. In the tropical Andes, the situation is further complicated by the sensitivity of glaciers to both temperature and precipitation (Jomelli et al., 2011; Favier et al., 2004; Jomelli et al., 2014). Previous studies have attempted to connect glacier variations to El Niño Southern Oscillation (ENSO) variability, but this relationship is inconsistent across the Holocene (Carré et al., 2014; Cobb et al., 2013; Zhang et al., 2014). The role of AMOC variations in shaping Holocene climate and glacier behavior, especially in the tropics and North Atlantic regions, has received comparatively less attention.

This study integrates new in situ 10Be and 14C CRE ages from glacial valleys in the tropical Andes (Bolivia), French Alps, Greenland, and the Cascade Range (USA), along with published glacier records from around the globe. The researchers employed established techniques for cosmogenic nuclide extraction and analysis at multiple laboratories. Moraine samples were processed for 10Be extraction, and 10Be/9Be ratio measurements were performed using Accelerator Mass Spectrometry (AMS). For Charquini North glacier, a numerical model was used to simulate nuclide production, decay, and erosion to reconstruct glacier history based on combined in situ 10Be and 14C concentrations in deglaciated bedrock. This model explores a range of plausible exposure and burial scenarios to find those that match the measured nuclide concentrations, providing insights into the timing and extent of past ice cover. The study also leveraged published CRE data and other glacier proxies (e.g., lake sediments and plant remains) for a comprehensive analysis of Holocene glacier fluctuations. To investigate the potential influence of the AMOC on glacier behavior, researchers compared glacier chronologies with recent AMOC reconstructions from proxies and transient climate model simulations (TraCE and LOVECLIM). They developed a semi-empirical model to estimate the "AMOC-corrected" temperature and precipitation impacts at regional scales. This model combines simulated responses to external forcing with the impacts of AMOC variations using three independent AMOC reconstructions and results from freshwater hosing experiments from five Atmosphere-Ocean General Circulation Models (AOGCMs). The calibration of AMOC changes is based on a hemispheric-scale temperature reconstruction to ensure consistency with observed data. The model then produces adjusted temperature and precipitation estimates, which are then compared to the observed glacier changes.

The study reveals a previously unrecognized in-phase millennial-scale glacier behavior in the tropical Andes and North Atlantic regions (TANAR) throughout the Holocene, distinct from other regions. CRE dating indicates that TANAR glaciers experienced a maximum Holocene extent during the early Holocene, followed by a significant mid-Holocene retreat and a late Holocene re-advance, peaking during or before the LIA. The early Holocene maximum extent is evident in the tropical Andes (Charquini North and South glaciers in Bolivia, showing early Holocene moraines at approximately 10 ka), the French Alps (Saint-Sorlin glacier, with early Holocene moraines around 9.8-11.4 ka), and northeastern Greenland (Clavering glacier, with early Holocene moraines at 9.5-10.5 ka). These findings contrast with the generally observed mid-to-late Holocene expansion of mid-latitude Northern Hemisphere glaciers. The mid-Holocene retreat is also observed in these regions. The analysis shows that this pattern doesn't fully align with the external forcings (orbital variations, greenhouse gases, ice-sheet extent, and volcanic activity) included in the TraCE and LOVECLIM transient climate simulations. The analysis further explores the role of internal climate variability, focusing on ENSO. However, the data reveals inconsistencies between ENSO proxy records and glacier fluctuations, suggesting ENSO's influence is likely secondary. Analysis reveals a significant relationship between AMOC strength and TANAR glacier changes. A strong AMOC is associated with warmer North and tropical Atlantic waters, while weaker AMOC is linked to cooler temperatures in these areas, affecting the ITCZ position. A stronger AMOC is linked to drier conditions in the tropical Andes and leads to reduced glacier extent, while a weaker AMOC enhances precipitation in this region. The semi-empirical model, integrating AMOC variations into transient climate simulations, shows that incorporating AMOC changes produces climate responses consistent with the observed glacier evolution. This model suggests that AMOC variations may have caused up to a 0.5°C temperature decrease and a 1 mm/day precipitation increase in the tropical Andes between the mid-Holocene and the last millennium. The findings highlight the significance of AMOC variations as a driving force of glacier fluctuations, especially in the tropical Andes.

The study's findings challenge the existing paradigm of summer insolation as the sole primary driver of Holocene glacier fluctuations. The in-phase behavior of TANAR glaciers, which is not replicated by transient climate simulations focusing only on external forcings, highlights the importance of internal climate variability. The strong correlation between AMOC strength and glacier fluctuations across the TANAR supports the hypothesis that AMOC played a significant role in regulating glacier extent throughout the Holocene. While the precise mechanisms linking AMOC variations to glacier changes require further investigation, the ITCZ migration driven by AMOC-induced changes in temperature and precipitation provides a plausible explanation for the observed synchronicity of glacier fluctuations in the tropics and North Atlantic. The observed discrepancies between AMOC variations in proxy records and transient model simulations highlight a critical need to improve the representation of AMOC dynamics in climate models. Future research should concentrate on refining our understanding of the drivers of AMOC variability, potentially including factors such as meltwater input and changes in Mediterranean outflow. The study’s conceptual model reveals how AMOC weakening could cause regional cooling and increased precipitation in the TANAR, potentially counteracting some aspects of current warming-induced glacier melting, but the model is limited and more sophisticated models are needed to better understand the AMOC’s impact.

This study demonstrates the importance of the Atlantic Meridional Overturning Circulation (AMOC) as a potential key driver of Holocene glacier fluctuations in the tropics and North Atlantic regions. The in-phase millennial-scale glacier changes observed across the TANAR, and their lack of correlation with external forcings alone, highlight the need to incorporate internal climate variability, specifically AMOC dynamics, into future climate models. The significant inconsistencies between AMOC variations in proxy records and climate simulations underscore the limitations of current climate models, suggesting a critical need for improved understanding of past AMOC behavior and its impact on regional climates. Future research should focus on refining AMOC representation in climate models, potentially by incorporating factors like meltwater inputs and changes in Mediterranean outflow, and on improving the resolution of regional-scale climate responses within general circulation models.

The study acknowledges several limitations. The semi-empirical model used to incorporate AMOC variations into transient climate simulations is relatively simple, and more sophisticated approaches might provide a more robust quantification of AMOC-induced climate changes. The reliance on moraine dating also assumes that moraines accurately reflect glacier equilibrium with climate, potentially ignoring nuances in glacial response times and other short-term forcings. Regional-scale climate responses to AMOC variations are complex and difficult to capture fully with current GCMs. Additional uncertainty arises from the discrepancies between AMOC reconstructions from proxies and climate simulations. Ultimately, the findings should be interpreted cautiously, emphasizing the necessity of further investigation using advanced modeling techniques and higher-resolution datasets.