Lacustrine sedimentation patterns at the Northern Antarctic Peninsula and surroundings as a response to late Holocene and Modern Climate changes

Abstract

The Northern Antarctic Peninsula (NAP) and the West Antarctic Ice Sheet (WAIS) are likely to respond rapidly to climate changes by increasing the collapse of peripheral ice shelves and the number of days above 0 °C. These facts make this region a representative hotspot of the global sea level rise and the location of one of the global climate tipping points (thresholds in the Earth system whose changes may become irreversible, if exceeded). Understanding the climate evolution of the NAP, based on past evidences, may help infer its future scenario. Herein, from a comprehensive survey of lacustrine sedimentation in proglacial and periglacial lakes/ponds, we investigated the impact of climate changes on the terrestrial environment in two complementary time scales (Late Holocene and contemporary age). For the longer time scale, regional climate database and biogeochemical properties of Lake Long/NAP sediment core, suggest warming between 4.0 and 2.0 kyr BP following a cooling phase towards the present, that endorse previously suggested Late Holocene Neoglacial (LHN). We attribute the LHN phase to a combined action of long-term decline in total solar irradiance, the Andean volcanism and the El Niño Southern Oscillation. For the contemporary age, we found a rapid coupled response of atmosphere/cryosphere/lithosphere to present warming levels.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-83754-0.

Introduction

The Northern Antarctic Peninsula (NAP) and the West Antarctic Ice Sheet (WAIS) are among the regions of the most rapidly warming of the planet¹ since the mid-20th century, playing a pivotal role in global sea level rise^2,3. WAIS has been steadily witnessing unprecedented surface snow melting events⁴ and increasing ice sheet destabilization⁵. In recent decades, ground-based meteorological stations at NAP, and the surrounding region, have documented great variability in mean annual air temperature trends, including a steep warming of 0.32 ± 0.20 °C between 1979 and 1997 and a relative cooling period of − 0.47 ± 0.25 °C between 1999 and 2014⁶. In February 2020, Esperanza Base located at the NAP/Graham Land (Lat. 63 °S) recorded a maximum annual air temperature of + 18.3 °C (SMN-Servico Meteorológico Nacional from Argentina - https://www.smn.gob.ar/), officially the warmest temperature ever instrumentally recorded in Antarctica as announced by the World Meteorological Organization (WMO). These short-term successive temperature increases and reversals illustrate how unpredictable the climate at the NAP has become in recent decades.

Part of the complexity in describing the climate of the NAP is due to its location in a climatic transition zone influenced by the temperate North, when air masses derive from the South Pacific or Southern South America, and the polar South, when coming from the Weddell Sea or Bellingshausen Sea⁷. Several controls (or combined effects of them) have been proposed to explain the climate variability at NAP (and WAIS), which includes atmosphere-ocean coupling linked to the Tropical Pacific⁸, El Niño Southern Ocean (ENSO)⁹, anthropogenic amplification of the circumpolar westerlies via greenhouse gases (GHG) and ozone depletion¹⁰, the Southern Hemisphere Annular Mode (SAM)¹¹, phase coupled ENSO-SAM¹², the action of atmospheric rivers¹³ and intrusions of Warm Circumpolar Deep Water¹⁴. For longer time scales, retrieving the Holocene climate history of the NAP based on proxy records, has been a challenging issue.

Attempts to recover air temperature records from ice cores based on isotope analysis were rarely successful at the NAP because of intense summer melting detected at locations below 1,000 m. At these altitudes, the glacial-chemical record is modified due to vertical ionic elusion along the core profiles¹⁵ and the seasonal and interannual variability of oxygen and hydrogen isotopes are not well preserved due to water percolation and refreezing and therefore they do not allow the regional climatic history. Few works (e.g.: Fernandoy et al., 2012¹⁶) were successful at altitudes above 1,000 m although they could not retrieve more than a decade of climatic history. In contrast, analysis of near-shore glaciomarine sediment deposits may reveal consistent interplay among the glaciogenic, oceanographic, climatic, and physiographic compartments¹⁷. However, at shallow sites, ice scouring from icebergs or other floating ice masses may gouge the seafloor and disturb the surface sediment distribution.

In the context of the NAP (extensible to all Maritime Antarctica), the lacustrine environment emerges as an alternative to provide insights into past climate changes that can extend back far beyond the instrumental epoch. As depositional systems, they receive fluxes of mineral dust and chemical compounds from the atmospheric fallout. Throughout the Holocene, these sites witnessed the transition from an underneath glacier environment into an ice-free region becoming genuine proglacial lakes. As a result, the bottom lake sediments preserve physical properties and biogeochemical compounds from which we can assess past productivity, temperature, runoff, and atmospheric circulation.

In this work, we investigate the past climate changes at NAP, and the surrounding region, from two sedimentary records dated for different (complementary) time scales: (1) modern sedimentation enclosing approximately the last 120 years retrieved from 8 (eight) lakes/ponds in the proglacial environment¹⁸ of King George Island (KGI)/South Shetland Islands (these lakes are located close to the ice front of glaciers, Fig. 1d); and (2) from 15 (fifteen) Holocene periglacial/isolated lakes (located away from the glacier front) Fig. 1a, c. Combining these two databases, we can better recover the sedimentary history (and associated climate/environmental history) for the study region. This approach may also allow the comparison of sedimentary records with ice core and deep ice borehole data, as well as determine the role of natural climate forcing (total solar irradiance, insolation, ENSO, volcanism and CO₂).

Fig. 1.

Sediment Coring Locations. (a) Surveyed isolated lakes at the North Antarctic Peninsula, NAP, and the surrounding region (blue dots). Lakes at King George Island, KGI, are denoted by a red dot (this issue) and a green dot; (b) KGI and 2 boxes indicating the location of the two largest ice-free areas (Fildes Peninsula and Admiralty Bay) and Potter Peninsula; (c) Detail of Fildes Peninsula with locations of isolated lakes investigated in this study (red squares). Exception for point 8 (proglacial lake); (d) Detail of Admiralty Bay with locations of periglacial lakes/ponds (white squares from 1 to 7); BD: Bellingshausen Dome, WI: Warszawa Icefield and KI: Krakow Icefield.

Source: (a) (b) (c) - adapted from the Natural Environment Research Council (NERC)-BAS (http://www.antarctica.ac.uk) after editions at Roberts et al. (2017); (d) – modified from the original SPOT satellite image 23fev2000 from the archives of the Brazilian Antarctic Program, PROANTAR/CNPq.

Site description

King George Island (KGI) climate, proglacial/periglacial environment

King George Island (KGI) is the largest island of the South Shetland archipelago, located at the northwestern end of the NAP, between 61°54’S-62°06’S and 057°35’W-059°02’W. Most of the air masses acting at KGI derive from the Pacific sector of the Southern Ocean (classified as maritime polar; mP), from advection north and northwest of 60°S latitude (classified as maritime tropical; mT)¹⁹, and cold air incursions from continental Antarctica that bring extremely low temperatures to the region. The western NAP is warmer and more humid and is characterized by a smaller sea ice seasonal extent, whereas the eastern sector is colder and more susceptible to Antarctic inland air flow with lower air temperatures and humidity⁶.

Albeit infrequent, liquid precipitation has been reported at almost all stations in the South Shetland Islands. Liquid precipitation, besides air temperature, has been considered a factor that directly impacts regional mass balance through modulation of the surface albedo and the acceleration of glaciers‘ retreat by eroding the ice or through hydrofracturing when associated with surface melting. At King George Island, Bellingshausen station has reported an average summer precipitation of 165 mm yr^− 1 between 1969 and 2017 (this issue).

Proglacial lakes investigated are located at Admiralty Bay/KGI ice-free areas (Fig. 1b) while periglacial lakes at Fildes Peninsula/KGI and the surrounding region, Fig. 1. Admiralty Bay is formed by irregular fjords of tectonic origin containing inlets and coves and a deep main channel reaching a maximum depth of 535 m. Fildes Peninsula is the largest ice-free area in extent and is located in the Southwestern sector of KGI bordered by the Drake Passage, the Maxwell Bay and the Bellingshausen ice dome. Due to depressions created during moraine formation and the action of fluvial drainage networks in ice-free areas from melting freshwater discharge during deglaciation, a lake sequence is found across the peninsula. This process was relatively common in most domains of the South Shetland Islands²⁰.

Results

Modern sedimentation: an atmosphere-cryosphere-lithosphere coupled system

The composite of the air temperature curve used in this work is derived from 3 meteorological stations at South Shetland Islands (Bellingshausen, March and Deception) (Mean annual data is presented in Supplementary information 2). The mean annual air temperature curve shows that from the 1970s to the first decade of the 21st century, regional levels increased by 2 °C with a peak around the year 2000 following a drop of 1 °C in the subsequent years, Fig. 2. Following this pattern, Equilibrium Line Altitude (ELA)²¹(the altitude where accumulation equals ablation over a 1-year period) presented a similar behavior, i.e., changing more than 100 m in the same period at Bellingshausen Dome/KGI. The method for ELA estimation at KGI is fully described at Mavlyudov, 2023²¹. It should be noted that the sequential climatic response presented in Fig. 2 was inferred from Maritime Antarctic glaciers, which are notably more sensitive to changes in air temperatures as demonstrated in previous studies²² and cannot be generalized to drier continental-type glaciers.

Fig. 2.

Atmosphere-Cryosphere-Proglacial environment climatic responses. Sequential climatic response at North Antarctic Peninsula (from right to left): instrumental mean annual air temperature (South Shetland Islands air temperature composite is formed by data from the stations: Bellingshausen, March, Comandante Ferraz and Deception - Supplementary information 2), Equilibrium Line Altitude for Bellingshausen Dome (Mavlyudov, 2023), mean lacustrine sedimentation for eight proglacial lakes (Fig. 1), (this issue), and Lake Profound XRF data (García-Rodríguez et al., 2021) as a proxy for catchment erosion/runoff. Sedimentation rate time series was obtained from the chronological model CRS that provided the age of each sediment depositional layer and then arranged vertically. Three events of sedimentation rate peaks observed in this work (light blue rectangles) corresponded to changes in Ti/Ca and Ti/Al ratios from sediment layers presented in García-Rodríguez et al. (2021) work for Lake Profound at Fildes Peninsula. The time scale for Ti/Ca and Ti/Al series is enlarged to fit the upper layers sedimentation peaks.

The time scale of these observations comprises an interval compatible with the dating of modern sedimentation processes at lakes/ponds close to glacier fronts in rapid retraction processes. They receive melting water during the summer season or during rapid warming episodes along the year. ²¹⁰Pb excess technique allow an age-depth estimation for sediment core profiles of these periglacial shallow lakes. From the Constant Rate of Supply (CRS) model (see methods) applied to the ²¹⁰Pb excess activities measured at each stratification of the sediment cores, we obtained the sedimentation rates from eight lakes at proglacial sites in KGI (Supplementary information 3). In this case, Ti/Ca and Ti/Al ratios, obtained by XRF at Lake Profound sediment core²³, widely known to be very sensitive to changes in terrigenous input (e.g.: Mesa-Fernández et al., 2022²⁴) can be here interpreted as proxies for catchment erosion/runoff, since their variability corresponded to peaks at the sedimentation curve, as shown in Fig. 2.

Late Holocene sedimentation

For the Late Holocene period, the sedimentary history was inferred from the sediment core-top ages enclosing 15 surveyed lakes in NAP and the surrounding region (ice-free areas at Fildes Peninsula/KGI, Potter Peninsula/KGI, James Ross Island, Signy Island, South Georgia Island, and Livingston Island) (Table 1). In this work, we added 4 new core surface datings to the regional database for Holocene periglacia (isolated) lakes at Fildes Peninsula: Lakes Profound, Long (Supplementar information 3), Slamnoe (AMS ID D-MAS 051940), Trigorskoe (AMS ID D-AMS 051938) and Geographensee (AMS ID D-MAS 051934). All ages presented in Table 1 were performed by radiocarbon, expressed as¹⁴C cal yr BP, and are related to top-surface sediment sections of the cores (mostly within the top 5 cm as presented in the literature or performed in the present work). The ancient dates found for sediment core − tops of the surveyed Holocene periglacial (isolated) lakes mean that they have not experienced sedimentation since that time. We interpret this finding as representing a regional cooling climate phase when no significant sedimentation took place after so. Geological surveys at King George Island indicate that limestone occurrences are basically formed by erratic fossiliferous boulders²⁵, and therefore might not represent significant sources of “old carbon” to the bottom surface sediments. Additionally, some references employed small fragments of plants to date the core-top sediment sections (as mosses) in a stand of bulk sediments.

Table 1.

Literature survey of ages of sediment core − tops from North Antarctic Peninsula and surround region.

Lake name	Location	Core-top age	References
		14C cal. yr BP
Bolder	James Ross Island	≈ 1530	Bjorck et al., 199626
Yanou	Fildes Peninsula, King George Island	≈ 768	Watcham et al., 20112
Belén	Fildes Peninsula, King George Island	≈ 382	Watcham et al., 20112
Gaoshan	Fildes Peninsula, King George Island	≈ 1087	Watcham et al., 20112
L15-H16	Potter Peninsula, King George Island	≈ 540	Barión et al., 202327
Midge	Livingston Island	≈ 605	Bjorck et al., 199128
Asa	Bayers Peninsula, Livingston Island	≈ 450	Bjorck et al., 199329
Sombre	Signy Island	≈ 1420	Jones et al., 200030
Heywood	Signy Island	≈ 340	Jones et al., 200030
Fan	South Georgia Island	≈ 781	Strother et al., 201531
Profound	Fildes Peninsula, King George Island	≈ 1074	Piccini et al., 202432
Long	Fildes Peninsula, King George Island	≈ 1700	this issue, 2024
Slamnoe	Fildes Peninsula, King George Island	≈ 3883	this issue, 2024
Trigorskoe	Fildes Peninsula, King George Island	≈ 758	this issue, 2024
Geographensee	Fildes Peninsula, King George Island	≈ 600	this issue, 2024

The variability in core-top ages in Table 1 may reflect the action of different processes as sampling, lake geomorphology, lake depth, the amount of fine sediments existing around the lakes (available to runoff), the amount of winter snow accumulated in its borders and possibly other factors. Besides that, different age-depth models were applied to the database by the original authors. Finally, the survey was taken in a regional context, from a northern limit on South Georgia Island (latitude 54°–55°S) to a southern limit on James Ross Island (latitude 64°S), from the Western and Eastern sides of the Antarctic Peninsula, influenced by the Bellingshausen Sea and Weddell Sea, respectively, and further north where the marine influence is very high.

Late holocene climate imprint in sediment core

Long Lake exhibited laminated sedimentary features (from the X-ray image) with roughly even-spaced light and dark density band formation. This structure reveals sequential deposition of organic clay-rich material (dark-colored) interbedded with silty to sandy material (light-colored). The low-energy hypoxic to anoxic bottom water conditions and the low inner circulation strength, typical of Sub-Antarctic lakes, may promote the laminated sedimentary profile and its preservation. The formation of these laminated structures commonly found in temperate, subpolar, and proglacial environments represents successive cold-to-warm climatic phases. Relationships between laminated sediment layers and climatic variability are well documented in the literature through several locations³³ mostly indicating annual resolution. The light-colored deposits correspond to the melting (warmer) periods, and the dark-colored laminae correspond to colder periods, as supported by the covariability detected in the upper part of the record for the C(%) and N(%) data. TEX₈₆ data (recovered only from the top 23 cm) indicated colder temperatures in this phase, probably due to cold waters entering the lake via runoff of melting water. For KGI, sequential light-and-dark sedimentary bands point to vigorous centennial-scale climate processes taking place at NAP during the Late Holocene which could be more significant than the annual climate swing.

The laminated pattern is observed throughout the core and is more evident within the upper 50 cm depth. The physical and biogeochemical properties of the sediment core indicate both marine and terrestrial lacustrine sedimentation histories. The detection of salt-intolerant diatoms in the uppermost units of the core (top 50 cm), Fig. 3a, as the species Achnanthidium australexiguum, suggests a terrestrial phase of the lake, as previously described for islands located at the NAP and the surrounding region³⁴. This is corroborated by the abrupt change in total C and N in the sediment core. They were enhanced substantially when the terrestrial phase (approximately 4.2 cal kyr BP) was established. Long Lake is located approximately 300 m from the shoreline at an elevation of only 14 m a.s.l. Its first isolation is estimated to have occurred at 6.445 ± 40 ¹⁴C yr BP(7.360 cal yr BP)³⁵. From the base of the core, dated from 6.764 cal yr BP, to approximately 4.0 cal kyr BP the X-ray radiograph depicts a pattern of less intense sedimentary oscillations compared to the above sections until the core-top. This marks a transition from marine to terrestrial environment at Lake Long site during the mid-Holocene isostatic uplift of the South Shetland Islands, due to deglaciation and tectonics³⁵. Nearly 4.0 kyr cal BP the lake is definitively isolated what is supported by the geochemical analysis and the abundance of the diatom species Achnanthidium australexiguum. Looking in a detailed sedimentary section (5–10 cm from top to bottom) of the dark-and-light sequence (grey scale/density profile), TEX₈₆ (a temperature proxy related to the cyclization of isoprenoidal GDGT (glycerol dialkyl glycerol tetraether) lipids produced by archaea), light layers correspond to colder phases. TEX₈₆ correlates positively and linearly with sea surface and freshwaters temperature³⁶. TEX₈₆ reduction in light-colored deposits, Fig. 3b, may indicate an inflow of colder meltwaters into the lake. Darker layers correspond to more biologically productive periods when total C and N increase in the lake. However, around the top-12 cm, which corresponds to 2.3 kyr BP (according to our age-depth model in Supplementary information 4), C and N started to decline nearly coincident with the onset of the neo-glacial phase of NAP and the surrounding region.

Fig. 3.

Biogeochemistry of Long Lake core. (a) X-ray radiograph of the Long Lake core/KGI accompanied by the grayscale/density profile and biogeochemical profiles of C (%), N (%) and diatom species Achnanthidium australexiguum; (b) detail of section. (5–10 cm from top to bottom) of dark-and-light stratification with additional data of TEX₈₆, δ¹³C and δ¹⁵N.

Discussion

For the modern period, peaks of sedimentation rate at KGI occurred around 1973, 1999, and 2003/2004 corresponding to increases in terrigenous Ti/Ca and Ti/Al ratios detected in Profound Lake (Glubokoe) sediment core retrieved in the vicinity of the Bellingshausen dome²³ (Fig. 2). This suggests a high sensitivity of the proglacial/glacial system to warming events recorded in ground stations. Therefore, the lacustrine sedimentation rates and the elemental ratios of Ti/Ca and Ti/Al can be used as potential proxies for climate variability and meltwater discharge at the NAP. In the terrestrial environment of Maritime Antarctica, Ti sources are associated to sediments in the catchments, volcanism and from local wall rock erosion (the aeolian component can be only greatly enhanced during the glacial period). Ti is related to siliciclastic input of lithological origin and Ca reflects authigenic carbonate content, therefore Ti/Ca ratio can be used as a proxy for the influx of terrestrial material³⁷.

In the geographical context, the NAP and the surrounding region are subjected to different synoptic climatic regimes (i.e. a western sector strongly influenced by air mass advection from the Pacific subtropical ocean and the westerly winds, and an eastern sector influenced by the Weddell Gyre, Southern Atlantic Ocean and polar advection). Despite these differences, a comparison between the 20th century modeled glacier surface snow melting, calculated for KGI (see Supplementary information 3), and the annual (July–June) melt percentage at JRI (melting events)⁴ showed good agreement (Fig. 4a), thus evidencing the regional impact of modern warming at that site. The geomorphological feature of the Antarctic Peninsula constraints the westerly winds in a way that air masses while crossing the Peninsula relieve warm adiabatically as they descend towards the eastern Antarctic Peninsula^38,39 where JRI is located (Fig. 4b). This process also weakens the colder easterly wind, due to a stronger westerly wind acting in NAP since the 70´s decade⁴⁰.

Fig. 4.

Melting events around the North Antarctic Peninsula. (a) Annual (July–June) melt percentage at JRI (James Ross Island), East Antractic Peninsula, Abram et al. (2013)⁴, and modeled surface melting at KGI (this issue), West Antarctic Peninsula, at 4 drainage basins (details in Supplemtary information 1). The shaded box highlights the melting peak event; (b) location of KGI (west side) and JRI (east side).

Considering the fact that liquid precipitation can strongly enhance glaciers’ ice mass loss by eroding the ice surface at the ablation zone and through hydrofracturing during intense surface melting⁴¹, we have also considered this parameter as a potential driver in the sedimentation issue. Melting water may carry sediments eroded from the glacier bed, sediment-filled surface depressions and the local terrain ahead of the glaciers front into proglacial lakes (an example is illustrated in Fig. 5f in Adimiralty Bay). In Fig. 5 we compared the proglacial sedimentation since 1946 at KGI (Fig. 5a) with the regional summer temperature, DJF, (composite of data from Bellingshausen-March-Condandante Ferraz-Deception stations), Fig. 5b, with the annual number of days with rain and rain + drizzle⁴² from Rothera and Faraday-Vernadsky Stations, Fig. 5d, and summer precipitation from Bellingshausen Station at Fildes Peninsula, Fig. 5e.

Fig. 5.

Liquid precipitation (rain), air temperature and sedimentation rate at North Antarctic Peninsula. (a) Mean proglacial sedimentation at KGI; (b) regional summer air temperature, DJF, composite of data from Bellingshausen-March-Condandante Ferraz-Deception stations; (c) Summer SAM (Marshall et al., 2006); (d) annual number of days with rain and drizze data from Rothera and Faraday-Vernadsky Stations (Vignon et al., 2021)⁴²; (e) Bellingshausen Station summer precipitation (this issue); (f) sediment plume formation at proglacial site at Admiralty Bay/KGI; (g) lag-correlation between air summer temperature and sedimentation. The blue curve depicts the correlation curve with indication of the time interval in a significance level at 0.05. Dotted and dashed curves are polinomial fits at boxes (d) and (e).

The Summer Southern Annular Mode (SAM) Fig. 5c, was also included in the analysis since it is the leading mode of extratropical Southern Hemisphere climate variability, associated with the positioning and strength of the polar jet that modulates the inflow of heat and moisture to high latitudes⁴³. None of the observed precipitation data could explain the mean proglacial sedimentation pattern presented in Fig. 5a. Precipitation showed slight increases during the 90´s decade, contrasting with the peak of sedimentation rates that occurred from the mid-90s to the first decade of the 21st century, Fig. 5b/d/e.

From Fig. 5g, air temperature and sedimentation rate are significatively correlated (p < 0.05) with lag of 0–10 years. Despite the several parameters that may define the ice mass loss, air temperature play a major role in the dynamics of the ELA, glacier mass balance and consequently the sedimentation at the proglacial environment. Previous works conducted at the Ecology Glacier/King George Island concluded that an increase in mean annual air temperature of 1 °C may enhances glacier ablation as high as 15%⁴⁴. The temporal lag observed between the two parameters (warming and sedimentation) can be attributed to uncertainties of the dating method for the sedimentation the use of several proglacial lake data to calculate the mean sedimentation rate (Fig. 2) and hydrofracturing of the glacial body that allow the surface melting water to percolate, refreeze in subglacial cavities and then discharge the freshwater into proglacial channels. Our observations are in accordance with conclusions of Zhang et al. (2023)⁴⁵ from a broad investigation of erosion and sediment transport in cold regions in which peak sediment yield is reached with or after peak meltwater. Concurrently, the impact of SAM on the change in sedimentation rate is evident, since its positive phase is related to warmer conditions at NAP.

For the longer time scale considering the Late Holocene, our survey of isolated lakes at periglacial environments shows that they have experienced successive sedimentation hiatuses in time, as air temperature became progressively lower, characterizing a Neoglacial phase. This result is consistent with the last 2.0 kyr air temperature inferred from mean borehole temperature (black thick line in Fig. 6) at WAIS (West Antarctic Ice Sheet) Divide (79^o28′S, 112 ^o05′W, 1766 m a.s.l.), to be on average 0.52 ± 0.28^oC colder than the last 100-year average⁴⁶. Gathering the glaciological data and the fact that 64% of the surveyed periglacial lakes have experienced sedimentation hiatuses between 200 and 800 year BP and 100% up to 340 year BP, it is consistent with the premise that the Little Ice Age (LIA) period did take place at a large portion of western Antarctica and that the LIA was not a seesaw-type climate event. LIA was a period of unusually short time and wide-spread cooling that lasted from about 1250 to 1860 AD⁴⁷ when glaciers across the North Hemisphere were significantly larger with mean annual air temperatures 1 °C lower than today.

Fig. 6.

Sediment core-top ages. Grouping of core-top ages in sediment profiles of periglacial lakes used to identify sedimentation hiatus (horizontal bars), corresponding to Table 1, from North Antarctic Peninsula and the surrounding region. The black thick line represents the mean borehole temperature obtained in WAIS Divide at the center of West Antarctica (Orsi et al., 2012); the dashed line identifies the colder phase in the last 2 kyr.

In the NAP the hiatus in sedimentation is mantained until nearly mid-19th century and rapidly increases towards the 20th century until the great peak of the beginning of 21st century, as depicted in Fig. 2. At the Northeastern Antarctic Peninsula sector, glaciers advance also corroborate our findings where repeatedly expansions occurred around 2.4 and 1.0 kyr BP and over the last few centuries⁴⁸. The sequence of sedimentation hiatuses depicted in Fig. 6 suggests that the Late Holocene Neoglacial phase in NAP/WAIS can be divided into 2 periods at the terrestrial periglacial environment: the first being characterized by an initial cooling phase, between 1.4 and 1.0 kyr BP, (corresponding to the Medieval Warm Period/Medieval Climate Anomaly- MWP) and a second by a deep cooling phase, between 800 and 200 year BP (corresponding to the North Hemisphere Little Ice Age - LIA). The MWP is a warm climate period revealed by archaeological evidence, botanic and paleoclimatic records coming from parts of the North Atlantic and North America.

At Maxwell Bay, a site in the vicinity of Fildes Peninsula (Fig. 1), where 8 of the periglacial lakes were investigated, grain size data from a Holocene sediment core¹⁷ allowed distinguish a relatively warm phase dated from the MWP (when fine grains are predominated in the sediment record) from a preceding cooling phase, dated from the LIA. Maxwell Bay receives melting water from several drainage systems of KGI glaciers. Therefore, fine sediments at Maxwell Bay may reflect erosive processes at Fildes Peninsula and other adjacent ice-free areas. From this observation, and differently from the record of isolated periglacial lakes, during the on set of the cooling phase at NAP, glaciomarine shore sites still received some sediments indicating a high sensitivity to summer air temperatures.

Previous studies have documented the climatic phases of the NAP and the surrounding region in the Holocene, using multiproxy approaches²⁷. However, our understanding of the effective role of main climate drivers remain highly uncertain. In Fig. 7 we compared glacial record, an ocean record and lacustrine sedimentation histories, together with climate forcings (TSI, ENSO, CO₂ concentrations, summer insolation, and volcanic episodes) to investigate how these variables covariate. Volcanism used here refers to Simple Dome A (SDM-A ice core record)⁵³ (Fig. 7h) located in West Antarctica. From Fig. 7 we identified two distinct climatic phases at NAP and the surrounding region during the Late Holocene: (1) a warm phase covering the period approx. 2.0 kyr BP – 5.5 kyr BP imprinted in the three compartments ocean-cryosphere-lithosphere (lacustrine environment).  This period accompanied lower ENSO activity, lower volcanism, lower summer insolation (Fig. 7j) and increased CO₂ (Fig. 7). Despite this condition, during the warm phase, KGI periglacial environment experienced regular swings of temperature and primary productivity as evidenced by laminated structures at Long Lake sediment core, as shown in Fig. 3.; (2) an important drop in the regional average air temperature from approx. 1.7–2.0 kyr BP to nearly mid-20th century characterizing a Neoglacial phase in the regional paleoclimate history. This was evidenced by the deuterium analysis at JRI ice core⁴⁹ (Fig. 7f), the increase of sea ice-related diatoms (e.g.: A. actinochilus and F. cylindrus), a decrease of open water taxa (e.g.: R. styliformis) and decrease of total organic carbon (TOC) (Fig. 7e) from a sediment core retrieved at South Scotia Sea⁵⁰. Along the Antarctic coast, the discharge of highly δ¹⁸O-depleted glacial ice (from both melting water and/or solid ice) is the sole process that can significantly reduce δ¹⁸O in seawater. Consequently, δ¹⁸O in diatoms may reflect that process. A record from an ODP sediment core at site 1098 reported from the Western Antarctic Peninsula showed a steady decrease in the isotopic signal since the last approx. 4.0 kyr BP with a steeper trend after 2.0 kyr BP ¹². This behavior accompanied the periglacial sedimentation hiatus throughout the investigated region (Fig. 7a, b). When comparing the natural climate archives with parameters in Fig. 7 we identified a concomitant increase in volcanic events and decrease of total solar irradiance (TSI) during the Late Holocene Neoglacial phase. Such a combination of volcanism and TSI resulting in cooling was previously documented for the North Hemisphere during the LIA epoch when solar minima Wolf, Spörer, Maunder, and Dalton were nearly coincident with volcanic eruption sequences⁵¹, although alternative theories have also been proposed for LIA⁵². LIA in the North Hemisphere started around 1,300 CE⁵³ lasting around 550 years.   Our database for the NAP and the surrounding region suggests that a possible volcanic trigger of the cooling is better explained by the volcanic record of WAIS/SDM-A ice core, which depicts a steady increase in activity throughout the Late Holocene. Nonetheless, this volcanism sequence is mostly of moderate VEI (Volcanic Explosive Index) and derived from two main sources: the Southern Andes volcanic sector and from the Antarctic continent, corresponding to a regional origin. Moderate but recurring volcanism, to a lesser extent, may also act as climate forcing since the aerosol loading may reach the lower stratosphere, as experimentally demonstrated in previous studies⁵⁴. Moreover, it has been claimed that decadal-paced eruptions could generate more cooling than single large eruptions if the recurrence interval is shorter than the upper ocean temperature relaxation time of decades⁵⁵. Another point is the proximity of these volcanic sites to the NAP that could induce a direct impact. The SDM-A location is of relatively lower elevation, 974 m a.s.l., compared to other sites on the Antarctic Plateau but very representative of the atmospheric transport and circulation pattern enclosing the continental regions closer to the Southern Ocean and the WAIS. The constant action of cyclonic systems around 60°S favors northerly air mass incursions to West Antarctica and the advection of air parcels and terrigenous aerosols from meridional South America⁵⁶. Concomitant with the increased volcanism since the last 2,000 years we observed a progressive decrease of the TSI (Fig. 7c). In Fig. 7 we show that the synchronism in total solar irradiance and volcanism is consistent with the timing when air temperature decreases and sedimentation ceases at NAP and the surrounding region. The fact that during the LIA at the NAP and the surrounding region a cooling condition was in progress, reinforces the concept of a global effect of this climatic episode. However, an issue that remains relatively unclear is the feedback amplifications involved since air temperature estimated in several places of the North Hemisphere cannot be solely explained by a lowering in solar irradiance and an increased volcanism⁴⁶. In addition to solar and volcanic climatic forcings, an intensified Niño 3 + 4 index accompanied these trends (Fig. 7i). The impact of El Niño events over West Antarctica has been widely discussed⁵⁷. However, such impacts are not spatially uniform. For the Amundsen and Bellingshausen Sea sectors that border the Antarctic Peninsula, intense El Niño events may lead to ice mass gain near shelves and marginal coastal zones⁵⁷. Our correlation maps of Niño 3 + 4 and air temperature (Fig. 7c) and Niño 3 + 4 and SST (sea surface temperature) (Fig. 7d), using the NCEP-NCAR reanalysis II, suggest that more ENSO events can strengthen the cooling at the NAP and surround region (Fig. 7c, d). Therefore, in addition to the decrease in TSI and increase in volcanism, ENSO events could also play a role in the NAP Neoglacial phase despite the influence of the slight increase in CO₂ (on the order of 10 ppm, Fig. 7g) and the increase in austral summer insolation (maximum of 20 W m^-2, Fig. 7f). The climate mechanism behind the recovery to a warming phase at the NAP and the surrounding region since mid-20th century is an issue not yet fully explained. Important controls postulated for this process are (a) more warm advection from the Subtropical Pacific Ocean, (b) an increased influence of the westerly winds from a combined positive phase of the SAM and an amplification due to the Antarctic ozone depletion^58,59 and (c) a higher frequency of northerly warm air mass advection, as revealed by the increasing amount of mineral dust particles reaching the NAP⁶⁰ and WAIS⁵⁶. All these mechanisms resulting from the southward shift of the westerlies. Considering SAM reconstructions, modern values of SAM fall outside the 2σ range and such increase can only be explained if an anthropogenic component is placed as an additional driver⁶¹. The rapid air temperature increases since the late 20th century documented at the NAP and the surrounding region are also observed in southern Patagonia, reconstructed from intra-annual bands in the wood of Nothofagus pumilio⁶².

Fig. 7.

Natural archives and parameters related to climate at North Antarctica Peninsula. (a) Summary of paleoclimate records related to Late Holocene Neoglacial: Strother et al., 2014³¹, 2. This issue, 3. Glasser et al., 2004⁶⁴, 4. Jones et al., 2000³⁰; (b) locations of records of “a”; (c) correlation map for surface mean annual air temperature and Niño 3 + 4 (this issue); (d) correlation map for Sea Surface Temperature and Niño 3 + 4 (this issue); (e) TOC-Total Organic Carbon (Bak et al., 2007)⁵⁰; (f) air temperature reconstructed from JRI ice core (Mulvaney et al., 2012)⁴⁹; (g) TSI reconstructed (Vieira et al., 2011); (h) volcanic events recorded at Simple Dome A ice core (81°39.530 S, 148°48.720 W; 621 m) and trend curve (Kurbatov et al., 2006); (i) Niño 3 + 4 reconstructed from speleothems (Zhu et al., 2017); (j) austral summer insolation at NAP; (l) Composite Antarctic Holocene CO₂ (Law Dome - blue curve and Dome C - red curve) Antarctic Holocene CO₂ (Monnin et al. 2004; MacFarling Meure et al. 2006).

The Late Holocene Neoglacial phase extended further north from the NAP to sub-Antarctic and temperate sites as in South Georgia at 54°S⁶³ and in South and North Patagonian Ice field at 48°S and 47°S⁶⁴. In South Georgia, GDGT-derived temperatures for the last 2,000 year exhibited a prevalent cold phase without an obvious MWP, which was represented only as a short-term event in comparison to its Northern Hemisphere signal recorded in the GISP2 ice core⁶⁵. In Patagonia, a prolonged cold phase has been recognized from glacier advances at 2.7-2.0 kyr BP; 2.3 kyr BP; 1.6–1.4 kyr BP, and during the Little Ice Age⁶⁴.The sedimentation response to cooling and the climatic characteristics presented in Fig. 7, representing the Neoglacial phase at NAP and the surrounding region, is corroborated by the negative trend in reconstructed air temperatures at 2 m over the past two millennia acquired from tree climate model simulations (CMIP6, ECHAM, and NB2014)(Supplementary information 7).

Conclusions

From a new multi-proxy approach combining proglacial/periglacial lake sedimentation, ice core and marine records, we give new insights on the climate changes and their sedimentary implications at NAP (and the surrounding region) from Late Holocene to the contemporary era. Our results and regional data compilation recognized a scenario of high climate variability with transitions from a relatively warm climate during the onset of the Late Holocene to a Neoglacial phase comprising the 2 kyr BP until nearly the beginning of the 20th century. This cold phase extended throughout the LIA and MWP, differently from the Northern Hemisphere when these two phases were well identified separately. Although the mechanism explaining these climate swings is not yet fully described, we found concomitant changes with the decrease of total solar irradiance, increase of the moderate Andean volcanism and ENSO variability. For the contemporary age, we found that present warming period is the major controlling factor of the regional lacustrine sedimentation and that elemental Ti/Ca and Ti/Al ratios are reliable geochemical proxies of such process. Finally, considering the high number of still unexplored proglacial/periglacial lakes at NAP, our work reinforces the importance to improve the regional lacustrine paleoclimate surveys as a tool to clarify the relationship between climate changes and the evolution of the terrestrial environment in Maritime Antarctica.

Methods

Sediment coring and sample preparation

Sediment coring of periglacial lakes investigated in this work (Long lake, Slamnoe lake, Trigorskoe lake and Geographensee lake) at KGI started during the austral summer of 2015 in the context of the UN/FAO/IAEA project “Assessing the Impact of Climate Change and its Effects on Soil and Water Resources in Polar and Mountainous Regions, INT/5/153”, with the support of the Russian Antarctic Program and the Brazilian Antarctic Program. Sediment cores were retrieved using two types of corer, one a hammer-gravity version of a UWITEC corer equipped with a 1.2 m plastic tube installed on a fixed tripod over a floating platform and a second a Russian Sediment/Peat Borer used from the frozen surface of the lakes. Cores from proglacial shallow lakes were sampled manually with PVC tubes to minimize disturbances. Sediment cores were stored at 4 °C until analysis and transported to Brazil where non-destructive analyses were first conducted (lithological description, X-ray, and magnetic susceptibility). After that the cores were longitudinally sectioned at 1 cm resolution, under clean room conditions, and each subsample was weighed, dried at 50^oC for 24 h, powdered, and sealed in plastic 47 mm-petri dishes for radiometric dating.

Radiometric dating and210Pb core chronology

Radiometric dating using ²¹⁰Pb_excess was conducted using a hyper pure germanium (GeHP) detector of coaxial geometry with a diameter of 56 mm and a length of 38.5 mm. The gamma system (detector and electronics) was installed in the Laboratory of Radioecology and Global Change (LARAMG / Rio de Janeiro State University). The relative efficiency is 20%, with a resolution of 1.8 keV at 1.33 MeV and 0.850 keV at 122 keV. A very low background lead shield of 12 cm with an internal copper lining on all walls was used. Detector efficiency curves were constructed from a NIST cocktail of radionuclides composed of ¹³³Ba⁵⁷, Co, ¹³⁹Ce, ⁸⁵Sr, ¹³⁷Cs⁵⁴, Mn, ⁸⁸Y and⁶⁵Zn in a solution of 0.5 M HCl. In addition to the natural radionuclides, we searched for ¹³⁷Cs trace activities in all samples.

Chronologies of shallow lakes sediment cores were established using the Constant Rate of Supply (CRS)⁶⁶ model at 1 cm core depth resolution. Dating was corrected by the ¹³⁷Cs peak when detected, that is attributed to the North Hemisphere 1961/62 atomic bomb test of global impact⁶⁷. For the Antarctic margin, the nuclear peak corresponds to the year 1965, according to Clarke et al., 2012⁶⁸. For the CRS method, we used activity data for ²¹⁰Pb_excess⁶⁶ measured by a high-resolution GeHP gamma spectroscopy. The basic concept of the model is provided by the relationship ²¹⁰Pb_excess = ²¹⁰Pb_Total - ²¹⁰Pb_Supported. Successive layers of newly deposited sediments are enriched with ²¹⁰Pb_excess, which exponentially decays in the lake sediment profile. Therefore, the accumulation rate of the sediments being deposited can be estimated by a simple decay Eq. 6⁹ as: Activity_Pb-210(t) = Activity_Pb-210(t = 0)*e^{-(λ Pb-210)t}, where λ_Pb-210 is the radioactive decay constant of ²¹⁰Pb equal to 0.03114 year^-1. Sediment layer ages are obtained by the equation as follows:

where ΣA_m represents the integrated ²¹⁰Pb activity from the surface to depth (m) and ΣA_∞ is the total integrated concentration of unsupported ²¹⁰Pb. All activities of ²¹⁰Pb and ²²⁶Ra used for the CRS model were determined by high-resolution gamma spectrometry. In natural environments, factors influencing undisturbed downcore exponential decay profiles include bioturbation, diffusion, or slumping⁷⁰. The average annual sedimentation rate, considering all shallow lakes investigated in this work, was obtained after establishing the chronology of each deposition layer at each sediment core separately. Then, for each period of time all the corresponding sedimentation rates were averaged by aritmetic mean, obtaing the standard deviation.

The Long Lake sediment core (code: LL2-IAEA) chronology is based on¹⁴C analysis performed at Beta Analytic Laboratory AMS. Bulk sediments were acid-alkali-acid treated following standard protocols and results are reported in conventional¹⁴C ages according to Stuiver& Polach (1977). Conventional¹⁴C ages were calibrated using the SHCal20 calibration curve (Hogg et al., 2020). AMS results are presented in Table 3 of Supplementary information 4. AMS analysis of top sediment cores of Slamnoe lake, Trigorskoe lake and Geographensee lake were conducted at the International Chemical Analysis-US, Inc. (ICA).

Elemental carbon and nitrogen and isotope measurements

Samples were dried at 40 °C, and inorganic carbon was removed prior to analysis of %C and ¹³C/¹²C by acid fumigation method⁷¹. Moistened subsamples were exposed to the exhalation of HCl in a desiccator overnight; afterwards, the samples were dried at 40 °C before measurement. The N(%) and ¹⁵N/¹⁴N analyses were performed with unacidified samples. The stable N and C isotope ratios and the C and N concentrations were measured using an elemental analyser (Vario Isotope Select, Elementar, Langenselbold, Germany) coupled to an isotope ratio mass spectrometer (Isoprime 100, Elementar, Langenselbold, Germany) located at the Soil and Water Management and Crop Nutrition Laboratory of the Joint FAO/IAEA Centre of Nuclear Techniques in Food and Agriculture. The N isotope ratios were calibrated against the reference materials IAEA-N-1 and IAEA-N-2. Calibration of C isotope ratios was performed using two laboratory standards, a sugar beet standard with δ¹³C -26.07‰ and a sugar cane standard with δ¹³C -10.95‰. Both laboratory standards were calibrated against IAEA-CH-6 and IAEA-CH-7. The C and N concentrations were calibrated against the reference material NIST 1547. The analytical accuracies of the measured standards and reference materials were δ¹⁵N -0.05‰ and δ¹³C -0.03‰, and the analytical precisions were +/- 0.19‰ for¹⁵N and +/- 0.12‰ for¹³C isotopes. Stable isotope data were reported as delta values (‰) relative to air for¹⁵N and VDPB for δ¹³C.

XRF scanning

XRF Core Scanner data were collected every 2 mm downcore using generator settings of 20, 30 and 50 kV, a sampling time of 10 s directly on the split core surface of the archive half with XRF Core Scanner III (AVAATECH Serial No. 12) at MARUM, University of Bremen. The split core surface was covered with a 4-micron thin SPEXCerti Prep Ultralene1 foil to avoid contamination of the XRF measurement unit and desiccation of the sediment. The reported data were acquired by an SGX Sensortech Silicon Drift Detector (ModelSiriusSD^® D65133Be-INF with 133 eV X-ray resolution), a Topaz-X High-Resolution Digital MCA, and an Oxford Instruments 100 W Neptune X-ray tube with rhodium (Rh) target material. Raw data spectra were processed by the analysis of X-ray spectra by the iterative least square software (WIN AXIL) package from Canberra Eurisys. We selected Zr, Rb, Al, Si, K, Ca, Ti, Al, Fe and Mn as elements, from which we calculated several ratios as proxies for paleoenvironmental change. High-resolution XRF scanning techniques are useful for inferring interannual paleolimnological changes in Subantarctic proglacial lakes during the processes of centennial glacier retreat²³. XRF is especially appropriate for inferring events and increased catchment erosion inputs (i.e., Ti/Ca and Ti/Al^72–74).

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

H. E. General concept of the manuscript/Paleoclimate description and text writing; S.V. General concept of the manuscript/Paleolimnological description and text writing; M.P.E. Use of frequency decomposition methods, spectral analysis and text writing; B. M. Reconstruction and long term in situ measurements of Equilibrium Line Altitude at King George Island and text writing; S.K. Lipid analysis, 14 C model and text revision; A.B.J.O. Text and graphic formatting; M.V.L. 210Pb dating and melting model; G.D. Project PI in the IAEA/FAO/; C and N isotope analysis; Z.P. Diatom analysis; F.G-R. Elemental composition analysis in sediment cores by XRF and text writing; A. A. N. Magnetic susceptibility analysis and text comments; S. J. G. Jr. Sediment coring in Antarctica; R. C. A. Magnetic susceptibility analysis and text comments; M. H. C and N isotope analysis; J. S. General text revision and text comments; C. R. C and N isotope analysis; A. C. Additional gamma spectrometry for 210Pb; R. G. C and N isotope analysis; M. H S. Use of climate computational model for the study region.

Data availability

The authors declare that fundamental databases wich support the findings of this study are available within the paper and its supplementary information files. Complementary geochemical data used here are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.