700,000 years of tropical Andean glaciation

Abstract

Our understanding of the climatic teleconnections that drove ice-age cycles has been limited by a paucity of well-dated tropical records of glaciation that span several glacial-interglacial intervals. Glacial deposits offer discrete snapshots of glacier extent but cannot provide the continuous records required for detailed interhemispheric comparisons. By contrast, lakes located within glaciated catchments can provide continuous archives of upstream glacial activity, but few such records extend beyond the last glacial cycle. Here a piston core from Lake Junín in the uppermost Amazon basin provides the first, to our knowledge, continuous, independently dated archive of tropical glaciation spanning 700,000 years. We find that tropical glaciers tracked changes in global ice volume and followed a clear approximately 100,000-year periodicity. An enhancement in the extent of tropical Andean glaciers relative to global ice volume occurred between 200,000 and 400,000 years ago, during sustained intervals of regionally elevated hydrologic balance that modified the regular approximately 23,000-year pacing of monsoon-driven precipitation. Millennial-scale variations in the extent of tropical Andean glaciers during the last glacial cycle were driven by variations in regional monsoon strength that were linked to temperature perturbations in Greenland ice cores¹; these interhemispheric connections may have existed during previous glacial cycles.

Fig. 1. Location of Lake Junín and drainage basin in the central Peruvian Andes.

Contour interval 200 m; stippled blue pattern indicates wetlands; orange arcuate lines are toes of proglacial outwash fans; red arcuate lines are ¹⁰Be-dated MIS 2 and 3 moraines; purple lines are ¹⁰Be-dated pre-MIS 3 moraines; grey outline is the drainage basin (see Methods). Inset map: black line demarcates the Amazon drainage basin; solid green fill is Perú;  black circles indicate Andean records discussed in the text: 1, Lake Junín, Perú; 2, Sabana de Bogotá; 3, Lake Titicaca.

Fig. 2. Age–depth model for the Junín piston core.

The age–depth relationship is based on 80 radiocarbon dates (<17 metres composite depth (mcd)), 12 U-Th-dated intervals of authigenic calcite from five carbonate intervals between about 21 and 71 mcd (ref. ) and 17 palaeomagnetic tie points. The red line is the mean age model; purple and black dashes represent the 1 sigma and 2 sigma uncertainties around the mean, respectively. Four arrows mark the depth of four samples that yielded normal polarity (depths shown in the inset along the age of the Brunhes–Matuyama (B/M) reversal boundary) and are younger than 773 ka (ref. ) (see Methods). Numbers 1–9 are tie points (Fig. 3b,c) used as the age model for Fig. 4a–c; tie points are for illustration only and were not used in the generation of this radiometric and palaeomagnetic age model. Source data

Fig. 3. Comparison of the Junín record of tropical Andean glaciation with extratropical records of glaciation and temperature change.

a–f, Benthic δ¹⁸O proxy of global ice volume and MIS boundaries (a), Junín GI based on the radiometric and palaeomagnetic age model (b), EPICA Dome (Antarctica) δD temperature proxy (c), wavelet analysis of the GI record based on the radiometric and palaeomagnetic age model (d), and cross-correlation plots of b and c using the radiometric and palaeomagnetic age model (e) and the age model based on 80 radiocarbon ages and tie points in b and c and circled in Fig. 2 (f). SMOW, standard mean ocean water. Source data

Fig. 4. Comparison of the amplitude of glacial cycles in the tropical Andes and global ice volume.

a–e, Junín GI and benthic δ¹⁸O proxy of global ice volume (z-scores) using 80 radiocarbon ages and tie points in Figs. 2 and 3b,c for the Junín GI age model (a), index of tropical glacier enhancement from the difference between the Junín GI and the LR04 proxy of global ice volume (b), weight percentage organic carbon in Junín drill core (c), cross-equatorial January insolation gradient between 11° S and 11° N (d) and atmospheric GHG concentrations from Antarctic ice cores (e; CO₂ (refs. ^,); CH₄ (ref. )). Dashed vertical lines in c and d highlight the tight coupling between the Junín lake level and the regional monsoon strength. Source data

Extended Data Fig. 1. Lithostratigraphy and physical properties of the Junín piston core.

Stratigraphic column (a) of main lithotypes in the drill core from the depocentre of Lake Junín and downcore plots illustrating variations in percentage siliciclastic sediment (b), magnetic susceptibility (MS; c), Ti/Ca (d), CaCO₃ (e), b* sediment colour index (f), percentage total organic carbon (TOC; g) and digital composite core image (h). Source data

Extended Data Fig. 2. Proxy indicators of clastic sediment concentration in the Lake Junín piston core.

Data are plotted on the age model tuned to EPICA Dome (Antarctica) δD temperature proxy (Fig. 4a). Ti/Ca (a), magnetic susceptibility (b), Junín GI (see Methods; c), siliciclastic sediment flux (d) and carbonate content (e). Source data

Extended Data Fig. 3. Tuning points applied to the radiometric age model for the Junín piston core.

Age offset between the age model derived by tuning Junín GI to EPICA Dome (Antarctica) δD temperature proxy and the radiometric and RPI-based age model (black line) using the nine tie points in Figs. 2 and 3 (numbers). Positive (negative) values denote that the tuned age is older (younger) than the radiometric and RPI-based age model, with the differences largely falling within ±2 sigma uncertainty of the radiometric and RPI-based age model (grey shading). Source data

Extended Data Fig. 4. Comparison of millennial-scale variability in the Junín piston core with regional cave records of hydroclimate and Greenland air temperature during MIS 2, 3 and 6.

High-resolution MS is a proxy for siliciclastic flux and regional ice extent (see Methods and Extended Data Fig. 2) and organic matter content is an inverse proxy for the hydrologic balance of Lake Junín. a–e, Expanded views of the 20–50-ka (MIS 2 and 3) interval. f–j, Expanded views of the 130–190-ka (MIS 6) window. d, Pacupahuain Cave (Fig. 1) speleothem record. e, Greenland δ¹⁸O record. i, Huagapo Cave (Fig. 1) speleothem record. j, Synthetic Greenland δ¹⁸O record. Dashed lines in a–e illustrate the temporal connection between Dansgaard–Oeschger (DO) events in Greenland ice cores and intervals of drying and deglaciation in the tropical Andes. Source data