700,000 years of tropical Andean glaciation

The study addresses the lack of continuous, well-dated tropical records of glaciation spanning multiple glacial–interglacial cycles, which has limited understanding of interhemispheric teleconnections and the pacing of ice-age climate. Marine benthic foraminiferal δ18O stacks (for example, LR04) provide a global ice-volume framework dominated by Northern Hemisphere signals, offering limited insight into Southern Hemisphere and tropical mountain glaciation timing and extent. Moraine chronologies in the Southern Hemisphere confirm broad synchronicity with global ice maxima but are discontinuous and incomplete across multiple cycles. Lakes in glaciated catchments can archive continuous signals of up-valley glaciation, yet few tropical records extend beyond the last glacial cycle or have robust independent age control. Here, the authors present a continuous, independently dated, approximately 700 ka lacustrine record from Lake Junín (central Peruvian Andes) to test whether tropical Andean glaciation tracked global ice volume on orbital timescales, assess amplitude differences across glacial cycles, and evaluate millennial-scale teleconnections with high-latitude climate variability.

Foundational late Cenozoic ice-age pacing is derived from benthic δ18O stacks (for example, LR04). However, ~80% of global ice-volume change occurred in mid- to high-latitude Northern Hemisphere ice sheets, limiting these marine records' ability to resolve tropical and Southern Hemisphere glaciation timing. Radiometrically dated moraines in the Southern Hemisphere show the last local glacial maximum broadly aligned with MIS 2 but are inherently discontinuous due to overprinting by younger advances. Previous tropical lacustrine records (for example, Lake Titicaca) have been hindered by insufficient age control or ambiguous glacial signals. Regional paleoecological records such as the Sabana de Bogotá pollen sequence document tropical Andean environmental responses at orbital scales. Speleothem δ18O records (for example, Pacupahuain, Huagapo) and Greenland ice-core data reveal millennial-scale variability (DO events) with known teleconnections to the South American summer monsoon (SASM) and Altiplano hydrology. Modeling and empirical studies suggest greenhouse gases (CO2, CH4) played key roles in deglacial glacier retreat and global synchronicity despite insolation heterogeneity.

Study site and coring: Lake Junín (11° S, 76° W, 4,100 masl) is a seasonally closed-basin lake in the uppermost Amazon basin between the eastern and western cordilleras of the Peruvian Andes. Glacial outwash fans and lateral moraines border the basin; exposure ages indicate multiple glacial cycles and that the lake was not overridden by ice during the past 700 ka. Seismic reflection identified a main reflector at ~100 m below lake floor, corresponding to the base (~95 m) of the composite lacustrine section, marking a transition (dated to MIS 16) from fluvial sand/gravel to lacustrine silt/clay and the onset of continuous lacustrine sedimentation. Piston cores totaling ~95 m composite length were recovered from the depocentre under the ICDP program. Age-depth model: The upper 88 m were dated with a multiproxy approach: (1) 80 AMS radiocarbon ages from the upper ~17 m; (2) 12 U–Th-dated intervals of authigenic calcite from five carbonate-rich intervals between ~21 and 71 m; (3) 17 geomagnetic relative paleointensity (RPI) tie points between 24 and 88 m. This yielded an age of 677 ± 20 ka at 88 m. Four samples from 93–95 m show normal polarity directions, consistent with deposition within the Brunhes chron and a basal age younger than 773 ka (the Brunhes–Matuyama boundary). The chronology is not orbitally tuned, allowing independent comparison with external records. Sedimentology and proxies: The composite section alternates between siliciclastic-rich intervals and authigenic carbonate-dominated intervals. Siliciclastic intervals are characterized by lower carbonate (<40%) and organic carbon (<5%), higher bulk density (>1.3 g cm−3), higher magnetic susceptibility (MS), and elevated concentrations of terrigenous elements (Ti, Si, K). The Junín glaciation index (GI) integrates MS and Ti/Ca as a proxy for clastic sediment flux and regional ice extent. Elevated GI during MIS 2–3 corresponds with lowered regional snowlines (by ~300–600 m) and moraine evidence for expanded ice east of the lake. Siliciclastic flux during MIS 2–3 was 1–2 orders of magnitude higher than the late glacial and Holocene. Time-series analyses and correlations: Wavelet analysis of the GI assessed orbital-scale periodicities. Cross-correlations compared GI to Antarctic EPICA Dome δD/δ18O temperature proxies and to the LR04 benthic δ18O stack (global ice volume proxy). To compare relative amplitudes, the team synchronized Junín GI and EPICA δD using nine tie points aligning prominent features, supplementing the 80 14C ages. They then computed z-scores of GI and LR04 on the EPICA-aligned timescale, defined a tropical glacier enhancement (TGE) index as GI minus LR04 (both normalized), and examined relationships with lake-level indicators (weight % organic carbon) and cross-equatorial January insolation gradients (11° S - 11° N). Atmospheric GHG concentrations (CO2, CH4) from Antarctic ice cores were compared to evaluate forcing mechanisms. Event-scale comparisons: Millennial-scale changes (abrupt decreases in Ti, MS, bulk density and lake level indicated by organic-rich sediments) were compared against regional speleothem δ18O records (Pacupahuain, Huagapo) and Greenland ice-core DO events to assess teleconnections via SASM variability and ITCZ shifts.

• The Lake Junín record provides, to the authors' knowledge, the first continuous, independently dated tropical archive of glaciation spanning ~700 ka.
• Tropical Andean glaciation tracked global ice volume closely on orbital timescales, with seven clear glacial–interglacial cycles and a dominant ~100-kyr periodicity in GI, consistent with LR04 and Antarctic δD/δ18O. The 41-kyr and 23-kyr bands are also present.
• The basal lacustrine age (younger than 773 ka; 677 ± 20 ka at 88 m) indicates lake formation near the end of the Middle Pleistocene Transition (MIS 16), when global ice volume shifted to 100-kyr pacing, implying pre-transition tropical glaciers were too limited to dam the basin.
• Correlation statistics: Junín GI and LR04 have r = 0.62 over the full record and r = 0.74 over the past ~140 ka. Without synchronization, GI lags EPICA δD by −18,750 years; with nine tie points, the lag is reduced to +500 years, indicating near-synchronous behavior within age-model uncertainties.
• A 400-kyr modulation in the relative magnitude of tropical glaciation (TGE) is evident, with elevated TGE between ~200–400 ka (MIS 7–11), suggesting enhanced tropical Andean glacier extent relative to global ice volume. During interglacials MIS 7, 9, and 11, glaciers appear to have persisted in the catchment, reflected by millennial-scale GI variability and intervals of elevated hydrologic balance (low organic carbon).
• Lake level (inferred from organic carbon minima) is tightly coupled to the cross-equatorial insolation gradient following precessional (23-kyr) pacing, modulated by eccentricity (400-kyr). This coupling weakens during ~200–400 ka, when hydrologic variability amplitude is reduced and 50–75 kyr intervals of elevated hydrologic balance occur, implying supplemental non-SASM precipitation and links to reduced atmospheric CH4.
• Millennial-scale events: Abrupt glacier retreat and lake-level lowering during the last glacial cycle align with weakening of the SASM (increased speleothem δ18O at Pacupahuain) and Greenland DO warmings, indicating a robust interhemispheric teleconnection via ITCZ shifts and moisture transport changes. Similar patterns likely occurred in MIS 6 (Huagapo speleothem vs synthetic Greenland δ18O), although precise correlations are limited by age uncertainties.
• The Junín GI does not record the Antarctic Cold Reversal readvance observed at higher Andean elevations, consistent with insufficient elevation for local ice persistence at that time.

The results demonstrate that tropical Andean glaciers responded in phase with global ice-volume changes on orbital timescales, supporting the hypothesis that global greenhouse gases, themselves modulated by Northern Hemisphere ice-sheet dynamics and orbital forcing, were key drivers synchronizing tropical glacier mass balance with extratropical climate. The dominant 100-kyr pacing after the Middle Pleistocene Transition and the minimal lag with Antarctic temperature proxies suggest fast-acting mechanisms coupling greenhouse forcing to tropical atmospheric temperatures. The enhanced relative glaciation (TGE) in MIS 7–11 indicates that hydroclimate, particularly precipitation linked to monsoon dynamics and possibly additional moisture sources, modulated glacier extent in the tropics beyond the global ice-volume signal, with a 400-kyr eccentricity modulation. On millennial timescales, coherent glacier retreats with Greenland DO warmings and SASM weakening underscore strong interhemispheric teleconnections: North Atlantic temperature instabilities shift the ITCZ northward, reduce SASM strength, decrease snowfall in the outer tropical Andes, and drive glacier mass-balance changes. The findings reconcile multi-scale forcing: greenhouse gas–driven temperature changes set the primary orbital-scale pacing, whereas regional precipitation variability linked to high-latitude North Atlantic processes modulates glacier extent on shorter timescales.

This study presents a continuous, independently dated ~700 ka record from Lake Junín that captures tropical Andean glaciation across seven glacial–interglacial cycles. Tropical glaciers largely followed the global 100-kyr beat established during the Middle Pleistocene Transition, with minimal lag relative to Antarctic temperatures, implicating greenhouse gases as rapid, global drivers of tropical glacier change. Periods of enhanced tropical glaciation relative to global ice volume, notably ~200–400 ka, reflect sustained intervals of elevated hydrologic balance and modulated monsoon dynamics. Millennial-scale glacier fluctuations coherently align with Greenland DO events via SASM teleconnections, indicating persistent interhemispheric linkages. These insights highlight the sensitivity of outer tropical Andean glaciers to both temperature and precipitation. Future work should refine age control during glacial intervals (beyond U–Th constraints), quantify the relative contributions of temperature vs precipitation to mass balance, assess basin-evolution effects on sediment proxy sensitivity, expand similar continuous records across the tropics, and employ transient climate–ice models to disentangle mechanisms across orbital and millennial scales.

• Age-depth uncertainties increase beyond ~50 ka (±5–10%), limiting precise alignment of millennial-scale events across records.
• U–Th dating is restricted to carbonate-rich, low-sedimentation interglacial intervals; glacial intervals with higher sedimentation and low carbonate lack direct U–Th constraints, potentially biasing phase estimates (for example, the initial −18.75 kyr lag vs EPICA).
• The GI is a proxy for clastic flux and glacier extent that can be influenced by basin ontogeny (for example, progradation of outwash fans), potentially introducing long-period, unidirectional sensitivity changes.
• The Lake Junín catchment elevation is insufficient to register certain regional glacial events (for example, Antarctic Cold Reversal readvance), limiting spatial representativeness for high-elevation glacier dynamics.
• Synchronization via tie points to Antarctic records introduces correlation assumptions that may affect phase assessments.