Introduction

Blue Ice Areas (BIAs) cover approximately 235,000 square kilometres of the Antarctic fringe1. They are regions of light-blue ice scoured by katabatic winds, which remove snow from ice surface to create fields of polished ice ripples. They are typically found in the vicinity of mountain ranges or nunataks (even when they are entirely covered by ice), in the lee side of obstacles that restrict snowdrift, or upon steep slopes and valley glaciers that enhance katabatic air drainage2,3,4 (Fig. 1). BIAs exist in various sizes and shapes, and they can persist over glacial/interglacial time scales5,6,7. The age of ice in BIAs described in the literature varies from tens of thousands8,9,10,11 to millions of years12,13,14. Scientific study of BIAs was initially slow after their accidental discovery in the 1940’s, but gained momentum after they were found to contain approximately 25% of meteorites worldwide, making them important meteorite hotspots15,16. More recently, research on BIAs has developed by addressing fundamental questions about their origin, properties, and potential significance for understanding past climates3.

Fig. 1: The extent of Blue Ice Areas in Antarctica.
figure 1

a Blue Ice Areas shown in red, layered over elevation data ≤1 km a.s.l. represented in green (RAMP2 DEM). The dendritic pattern in dark blue shows modelled subglacial meltwater routes56,57; b Blue Ice Areas (in white for better visibility) layered over ice velocity map (MEaSUREs InSAR-based 450 m resolution, units: m/y57), to highlight the location of BIAs on/near the fast-moving ice streams58; All data provided by Quantarctica59.

Full size image

Blue Ice Areas have a characteristic, often striking, blue colour that results from the stronger absorption of light by ice in the red part of the spectrum17. Areas where old, wind-polished ice is exposed, appear dark blue on account of the increased absorption of these wavelengths following long travel paths through the ice. BIAs with a high content of air bubbles appear pale blue or even white/grey, due to the scattering of light back to the observer17. The air bubbles also reduce the density of blue ice compared to solid glacier ice18,19. BIAs were originally defined as areas of blue ice present only above 1 km a.s.l. (hereafter high-altitude), “where ice is at the surface, and where negative mass balance (ablation) is not influenced by melt”3. Areas at lower elevations (hereafter low-altitude) and located at the coast were disregarded, hindering BIAs scientific understanding. Today, researchers agree that BIAs occur at both high and low altitudes (Fig. 1) however, their location is a key factor determining their formation and properties.

Low-altitude BIAs are formed by melt-driven ablation that is solely caused by incoming solar radiation, while high-altitude BIAs (≥1 km a.s.l.) are formed by sublimation-driven ablation that is dependent on air temperature, relative humidity, and wind speed. This difference in the formation of the blue ice occurs because at low altitudes near the coast, the snow and ice surfaces easily reach the melting temperature in the summer, at which point any excess energy input to the surface causes melt-driven ablation. However, at high altitudes, inland, the surface temperature typically remains below the melting point, so excess energy input to the surface can either warm the ice or lead to sublimation-driven ablation18,19. In addition, downslope katabatic winds provide a steady source of dry air to allow sublimation. Nevertheless, surface ablation at low altitudes is significantly greater than at high altitudes (e.g., 52 cm year−1 at 30 m a.s.l. vs 12 cm year−1 at 1200 m a.s.l.)20,21 because it takes more than 8 times as much energy input to sublimate a given mass of ice than to melt it. Until now, this significant elevation dependence of blue ice ablation was thought to indicate that high-altitude BIAs are dry and therefore devoid of life. However, that assumption disregarded the fact that at high altitudes, optical properties of BIAs become critically important. Here, visible light penetrates the ice and creates a solid-state greenhouse beneath the sublimating surface22. This radiative warming effect causes sub-surface melting at rates that are sufficient to account for approximately 15% of all surface and near-surface meltwater production in Antarctica22,23,24. The presence of dark debris within the ice (see below) further enhances this effect, allowing even high-altitude BIAs to produce meltwater beneath the cold ice surface when exposed to sunlight. In turn, this process provides the basis for an aquatic ecosystem that can thrive in one of the harshest environments on the planet25,26,27, usually in close association with the debris, which provides nutrients following weathering processes28.

Debris present in BIA and typically referred to in the literature as “cryoconite” is said in most studies to be supplied from ice-marginal sources by local winds26,29, following the same processes as those observed on glaciers elsewhere. However, on the Antarctic Ice Sheet, subglacial and englacial debris are transported with the ice over vast distances before accumulating at the surface of the BIA due to sublimation, melting and the upward compressive flow of ice5,30 (see Fig. 2). In addition, debris can also accumulate on the blue ice surface from exposed nunataks and weathered moraines located on the mountain ranges (Fig. 2). Most often, that debris (near or on the surface) absorbs sufficient solar radiation to create a local melt pool known as a cryoconite hole (“CH”, Fig. 3). The water collected in the hole then freezes at the surface creating an ice lid over the CH. The refrozen ice lid and ice walls maintain the greenhouse environment within the hole until the entire system most likely freezes during winter. To date, CHs in Antarctica have been found to vary in size from a few centimetres to a few metres in diameter, and have depth of up to ca. 50 cm26,28,29. Although important, the distribution of CHs across BIAs is poorly understood, not least because they can be difficult to detect using coarser remote sensing techniques on account of their perennial and thick ice lid.

Fig. 2: Composite sketch representing distribution of ice flow, ice ages (isochrones in shades of blue), debris and path of particles (dashed lines) in blue ice areas (BIA in the ablation zone).
figure 2

a in the open zone, b in the closed zone restricted by a geological formation (a mountain chain or a nunatak)12,30,60,61,62.

Full size image
Fig. 3: Cryoconite hole (CH) schematic and examples of CHs present in blue ice in Jutulsessen, Droning Maud Land.
figure 3

a Schematic of a CH with marked dimensions based on field measurements, b an example of cryoconite debris being disintegrated at the edge of an ice cliff (also bottom f); c water sample from one of the CHs; d cryoconite debris present inside CH; e looking inside a cryoconite hole; f–h drone images of cryoconite holes in the Blue Ice Area, people and a snow scooter are present for scale. Picture f also shows the presence of liquid water at the base of the ice cliff.

Full size image

Hidden inside CHs are microbial communities that can harbour highly diverse taxa, including bacteria, eukarya, and archaea capable of photosynthesis and heterotrophic production26,31,32,33,34,35,36. The origin of microorganisms inside CHs is not clear, although they may be inoculated by local environments via atmospheric deposition36 or by microorganisms advected into BIAs with ice streams by ice flow. It is so far uncertain for how long these refugia can persist, although it has been shown that individual holes can be sealed from the atmosphere for an entire season, decades37 or perhaps thousands of years38 at a time.

Whether all BIAs in Antarctica presented in Fig. 1 host productive microbial ecosystems is presently unclear, especially with respect to inland high-altitude Blue Ice Areas (that are the focus of this study). These areas remain far less known than their low-altitude counterparts26,35,38,39,40, yet they offer important insights into the capacity of BIAs to act as microbial refugia at environments previously thought to be barren, as they experience conditions closer to the limits of life. Here we show through an example of blue ice in Jutulsessen, Dronning Maud Land, that high-altitude BIAs can function as critically important “powerplants” that produce water, nutrients, major ions, as well as organic and inorganic carbon. They support regionally significant ecosystem that is dominated by cryoconite holes originated by processes different to those observed on glaciers. Furthermore, we demonstrate that the dense network of cryoconite holes present in the high-altitude blue ice is of great potential significance for supporting diverse microbial life as well as producing, storing and releasing water, nutrients and carbon to downstream ecosystems in Antarctica’s changing coastal environment.

Results and discussion

Cryoconite hole distribution and cover

This study produced the largest and most comprehensive map of CHs to date, encompassing the entire Jutulsessen BIA to reveal as many as 856,000 cryoconite holes within just 62 km2 (see Fig. 4 and Methods—Field mapping and sampling). The mapping also identified seven distinct zones with variable CH size and distribution ranging from 0.01% to 14% coverage of the blue ice surface (see Fig. 4 and Table 1). The CH concentration zones corresponded to the ice flow and their boundaries appeared to rest on the ice flow bands. The total coverage of CHs accounted for 4.2 km2 or 6.7% of the entire BIA, which is marginally greater than the average 4–6% coverage reported at small spatial scales elsewhere in Antarctica26,29,33. The CHs were filled with meltwater, and, depending on their location, had variable size (see below), depth (20–80 cm), and ice lid thickness (10–40 cm). The smallest and shallowest detectable holes (approx. 10 cm in diameter according to ground-truthing observations) were found at the edge of BIA (see Zones 2 and 3: Table 1), while holes exceeding 3 m in diameter were measured in the terminal basin Sætet (Zone 7). Our mapping revealed that CHs may reach over 8 m in diameter in the case of large hole conglomerates, formed in areas where the sheer volume of holes caused them to merge.

Fig. 4: A complete digitized map of cryoconite holes in high-altitude BIA of Jutulsessen, Dronning Maud Land.
figure 4

a In black and white, showing variability in CH distribution and density across the blue ice area. Each black dot represents one cryoconite hole; b A composite sketch of digitized and colorized field of CHs presented in a, with marked mass balance stakes used to measure ice ablation, the direction of ice flow and its speed (arrows), outline of CH Zones in red, and the Troll Airfield (dashed blue line). Dashed brown line indicates a likely existence of geological formation underneath the ice surface and suspected source of debris for CH in Zones 3 and 6. It is also the location of a heavily crevassed area and the ice cliff. This figure is a composite of digitized derivative of a November 2019 satellite image (World View Stereo ©2013–2017 Digital Globe, Maxar Technologies) and Sentinel2 (source: Quantarctica59); c Cross-section of the blue ice in Jutulsessen62 with marked locations of change in ice thickness that can also be visible on b.

Full size image
Table 1 Cryoconite hole (CH) cover, water volume, and water storage in the different zones of the Jutulsessen Blue Ice Area
Full size table

Our data show that the origin and distribution of CHs in Jutulsessen are most likely not governed by the prevailing katabatic winds as reported elsewhere in glacier-based CHs26,35, although wind contribution to debris supply is not disputed here. Instead, they are governed by the geology of Dronning Maud Land and the direction of ice stream flow in the area that redistributes the debris (Figs. 4 and 5). This is because in Jutulsessen katabatic winds blow westward opposing most of the ice flow, and their path does not correspond to the biogeochemical signature of CHs described below. The eastward flowing ice, quite possibly from Fjellimellom and most likely Fimbulheimen, bends almost 90° southward after passing Trollisen and Mimebrønnen, only to be advected towards Sætet and rest against Jutulsessen bedrock (see Figs. 4 and 5). The second ice stream contributing to BIA formation is flowing from the opposite direction, westward from Hellehallet, through Stabbkleiva, turning around Stabben also spilling out into Sætet (Fig. 5). Both bring debris eastwards and southwards into the BIA to form a high density of large CHs in Zones 6 and 7 (Fig. 4a). The long transport path, slow movement of ice (17–25 cm year−1) and thus high concentration of debris is most likely responsible for large size and large volume of the CHs present in Zones 5–7. We also observed that Zones 2–3 holding much smaller CHs, and in markedly smaller quantity, were located only on faster flowing ice (205–342 cm y−1, ice flow measurements based on 2009–2018 surveys performed by the Norwegian Polar Institute).

Fig. 5: Jutulsessen Blue Ice Area (Sentinel2) with geology overlay63, and estimates of the total mass loads of nutrients, carbon, and major ions in kilograms, from all cryoconite holes in each Zone.
figure 5

White outlines indicate boundaries of distinct CH Zones. Detailed information on CHs in each Zone as well as ice movement is presented in Table 1, Fig. 4 and Supplementary Table 1.

Full size image

In addition, the eastward flowing ice is highly crevassed at Andrefallet, before passing Trollisen and Mimebrønnen (dashed brown lines in Fig. 4b). This ice flow pattern, the distribution pattern of CHs, and the presence of crevasses also suggest that subglacial erosion of a yet to be mapped rock ridge provides an additional debris source for CH formation in originally Zones 1, 3 and further, others located in Seætet.

Since the ice flow in the area has various directions, it erodes different lithologies (see Fig. 5). Below we show how the chemical composition of the CH waters reflects those different mineral assemblages being weathered within them, and thus support our statement that unlike on glaciers, bedrock geology and ice movement are the key parameters responsible for shaping CH ecosystems on the continental ice sheet.

Hydrological significance of the cryoconite hole network

Field measurements performed within each of the seven zones allowed for the assessment of the water content within CHs, enabling estimation of the total water storage across the entire BIA (Table 1). We estimate that the total water storage is equivalent to 19.6 mm of water across the BIA, which may be compared to observations of total annual ablation from the stake network (presented in Fig. 4) of 50–80 mm year−1. The blue ice in Jutulsessen is also able to maintain subsurface drainage networks. These are well-known anecdotally to residents of Troll Research Station, especially those given the responsibility of maintaining the blue ice runway. Subsurface drainage has been described scientifically in snow-free valley glaciers of the McMurdo Dry Valleys22,29 and other BIAs41,42. In Jutulsessen, we also observed hydrological connections between the CHs through ice cracks or fractures, even resulting in open channel flow, as well as accumulation of water and cryoconite debris in pools across Zone 7 during mid-summer (Fig. 6). Furthermore, after sampling, water in holes was often quickly replenished to the pre-sampling level, or further in some cases, causing it to discharge onto the blue ice surface. Our observations suggest that CHs residing in high-altitude blue ice of Jutulsessen are not isolated refugia, as previously assumed, but rather potentially complex interconnected hydrological systems with the capacity to flow towards low elevation basins, or even connect to glacial drainage systems, as is increasingly noticed in the region41,42.

Fig. 6: Aerial image over Jutulsessen blue ice in the Blåfallet area, taken with a drone DJI Mavic Mini, showing an active hydrological system. The colors of the image were not altered.
figure 6

Looking down from 30 m altitude above the surface of the ice. The edge of the ice along the bottom left corner is the ice cliff, on the side of which cryoconite holes are being disintegrated and supplying debris material (visible in dark brown and black) to the ice below. The greenish colour is water on the surface of blue ice and covering newly supplied debris. The greenish color of water indicates a potential biological activity occurring on the ice surface.

Full size image

Light penetration through blue ice and subsurface melting

Monitoring incident radiation within a representative CH in Zone 1 showed that subsurface debris at a depth of 40 cm receives between 11 and 18% of the solar irradiance under steady state conditions. This demonstrates the importance of the ice lid and ice walls in reducing exposure of debris-hosted microbes to potentially harmful light levels, including UV light. We show that high-altitude CHs in Jutulsessen receive similar levels of irradiance to CHs in low- altitude BIA26 and larger “cryolakes”43, where photosynthesis in the debris layer has been shown to be adapted to the low light conditions26.

Away from the debris layer, the subsurface melt zone can penetrate far deeper than the base of the CHs, reaching even several metres or more (see Methods—Optical properties of blue ice and radiative transfer modelling). Our modelling shows that the attainable depths of this melt-zone are primarily governed by cloud thickness and solar radiation, as has been observed on other BIAs24. Liquid water may even be perennial under the scenario of sustained cloudless or thin cloud cover conditions (0.022 mm of precipitable water). However, if cloud cover is constant and thick (typically 1 mm of precipitable water), sub-surface melting may be prevented even in summer, and ice temperatures may rarely exceed −10 °C (see Methods—Optical properties of blue ice and radiative transfer modelling).

The thickness and depth of the sub-surface melt-zone produced under meteorological forcing conditions are also influenced strongly by the presence of air bubbles in the ice, whose scattering effect reduces the depth of meltwater production. Our analysis of air bubbles in six ice cores, revealed their concentrations between 0.80% and 11.5% by volume, (average 3.5%, coefficient of variation 114%). The size of air bubbles was predominantly small (average diameter 0.71 mm, coefficient of variation 24%). Variations in the distribution and concentration of air bubbles within the blue ice are therefore significant sources of uncertainty in understanding the extent of sub-surface melting within BIAs44, and should be taken into consideration when estimating sub-surface meltwater content.

Addressing this issue along with improving our understanding of the impact of climate change on cloudiness in the region, will be crucial for answering fundamental questions about how Blue Ice Areas sustain the ecosystems they support, and how they facilitate hydrological connections with the rest of the Antarctic Ice Sheet.

Biogeochemistry of the blue ice ecosystem

Water inside the CHs was generally acidic and, on some occasions, more acidic than the snowpack (CH water average pH = 5.5, min = 3.99 vs snow average pH = 4.6, min = 4.33: Supplementary Table 1). The water in the CHs was also in most cases an order of magnitude more concentrated in dissolved ions than the surrounding snowpack, the ice lids and the blue ice itself. This demonstrates effective weathering processes and solute exclusion during freezing of water, and vapour deposition onto the underside of the lid during its formation. Variable concentration of all solutes observed across CHs may also reflect variable rates of flushing by  waters produced after sub-surface melting and when the holes get connected.

In general, water inside CHs was rich in nutrients such as nitrogen (NO3−, NH4+), phosphorus (PO43-) and dissolved organic carbon (DOC). Highly heterogeneous spatial distribution amongst those ions, suggests that described below microbial communities present within the holes, also play a role in shaping CH hydrochemistry. For example, average concentrations of nitrate (NO3−) in Jutulsessen were in comparison double to concentrations found in CHs of the Vestfold Hills. What is more, dissolved organic carbon (DOC) was five times higher26. In addition, CH waters in Jutulsessen also contained 230 times more NO3− and 17 times more PO43- than CH waters in Sør-Rondane Mountains, in the eastern Dronning Maud Land38.

On average, across the entire BIA, CH water enrichment (indicated by the average ratios in the CH water to the maximum values in snow or ice) in NH4+ and Na+, although still predominant, was the least common among the nutrients and cations (77% and 83% respectively). Both DOC and DIC were more often depleted than enriched (29% and 17%, respectively). The degree of enrichment was greater for the cations (19 times or more) than either the nutrients or Cl− (up to a factor of five). These differences indicate that simple freeze-concentration effects are supplemented in CHs by more selective microbes – rock – water interaction. This is especially true for ions derived from mineral weathering.

Therefore, heterogeneity in the chemical signature of CHs in Jutulsessen also results from the distribution of different rock types, especially in solutes demonstrating high enrichment (Si, K+, Ca2+, Mg2+, Na+, and PO43-). For example, Zones 1-6 have high contents of Si and K+, but relatively low PO43- and Ca2+, which can be assigned to ice-flow interacting with felsic lithologies, like granitic gneiss (see Fig. 5 and Supplementary Table 1b). Supply of felsic mineral assemblages, consisting of the dominant mineral assemblage quartz (SiO2) and feldspar (Na,K)Al3SiO8 favour enrichment of K, Si, and perhaps Mg as observed in Zones 1, 2, 3, 5, and 6. Conversely, Zone 4 is dominated by mafic assemblages consisting of dominantly Ca-feldspar (CaAl2Si2O8), ferromagnesian silicates (amphibole, pyroxene), and apatite ((Ca5(PO4)3(FOH), leading to the enrichment of Ca2+, Mg2+ and PO43- relative to felsic lithologies (see Fig. 5 and Supplementary Table 1b).

Storage of dissolved nutrients, major ions, and organic carbon in various zones (calculated from weighted mean concentration and the volume of water measured in CHs, after up-scaling with the digitized map of CHs, see Figs. 4 and 5), showed that the largest resource of dissolved ions is stored in Zone 7 (Sætet), where ice is presumably the oldest and CHs cover 14% of the ice surface. In contrast, the smallest resource was recorded in the north-westernmost edge of the blue ice area, Zone 2, where ice was fast-flowing, CH density was low and CH size was small (Fig. 5, Table 1). Although our very conservative estimates do not include conglomerate holes, nor holes that are smaller than 25 cm in diameter (due to uncertainty in their volume), our calculations show that the entire high-altitude blue ice area of Jutulsessen holds at minimum ca. 420 kg km−2 of dissolved nutrients (or 26 tons), 660 kg km−2 (or 41 tons) of dissolved organic carbon, 663 kg km−2 (or 41 tons) of dissolved inorganic carbon and 6,700 kg km−2 (or 410 tons) of dissolved major ions.

Highly significant resources are therefore available in the high-altitude blue ice for biological assimilation or transport downstream through ice fractures and existing flowpaths discussed above, as well as other developing ones. These quantities clearly demonstrate that the high-altitude BIA, previously thought to be barren, is a biogeochemical hotspot on the otherwise frozen ice-sheet surface. It also has the capacity to force downstream change as the ice sheet margins become more hydrologically active45,46.

Microbial life in the high-altitude blue ice

Bacterial communities were inferred from the DNA in the CH debris and in shallow glacier ice cores recovered from the sub-surface melt zone. While the CH waters were found to have a significant level of heterogeneity and  microbial biodiversity, no DNA was recoverable from any of the glacier ice samples (which did not yield fluorometrically-quantifiable levels of DNA using techniques optimised for glacial samples47). CH communities were therefore dominant in the blue ice, with twenty-two bacterial phyla found in eighteen samples. In all but two samples, Cyanobacteria, Proteobacteria, Actinobacteriota and Bacterioidota collectively represented 80-97% of the phylum-level diversity. In the remaining two, their summative abundance was 65-67%. Cyanobacteria were the most dominant in 39% of the CHs. Actinobacteriota dominated in 33% of the samples, Proteobacteria in 16.7% and Bacterioida in 11.1%. Planctomycetota, Patescibacteria and Verrucomicrobiota were also notably abundant, although the latter was only observed in a third of the samples. Together with fifteen other phyla, the abundances of these last three phyla ranged between 3 and 33%, depending on the CH and Zone sampled (see Methods – Cryoconite microbiology).

The four most abundant bacterial groups present in Jutulsessen occupy CHs in other areas of Antarctica and at similar relative abundances26,38,48 However, we have also observed that less abundant phyla (Verrucomicrobiota, Bdellovibrionota and Myxococcota) rarely reported in Antarctic CHs are present in Jutulsessen. These phyla are commonly seen in dramatically different environments, yet they were observed in all zones in Jutulsessen. In addition, Zone 7, with the densest network of CHs, and most likely the oldest ice, also hosted Firmicuta, Desulfobacterota, Nitrospinota, Nitrospirota, Fibrobacterota and Methylomirabilota (see Methods – Cryoconite microbiology).

DNA levels that were below detection limit for conventional techniques in the glacier ice samples may raise uncertainty about the importance of the release of cells when sub-surface melting occurs. Furthermore, whilst the bulk debris content of the ice cores in the sub-surface melt zone adjacent to the CHs was low (1.9 ± 1.1 mg kg−1, n = 6) it was not negligible, making the absence of easily recoverable DNA even more surprising. This is because viable cells are known to be associated with debris within Antarctic glacial ice49. This suggests that the CHs provide a refuge that blue ice itself cannot. For example, once they reach a certain diameter, shading effects protect the cells within both the debris and the underlying ice from high levels of UV light.

The CH refugia may also offer more time and energy for biological production, protection from harsh conditions on the ice surface, as well as a reliable water supply. More research on the source and resilience of microorganisms in high-altitude blue ice habitats is needed to broaden the knowledge on these extreme-environment ecosystem dynamics, and to increase our understanding of the limits of life.

Conclusions

At 235 000 km2, BIAs may only constitute just 1.67% of the Antarctic ice sheet surface however, their location, capacity for water, nutrient and microbial storage suggest they play a disproportionately large role in ice sheet ecosystems. Furthermore, being located on the periphery of the Antarctic ice sheet makes them critical environments for gaining insights into the response of the Antarctic Ice Sheet to climate changes, from hydrological, biogeochemical and ecosystem perspectives.

Our study of high-altitude Blue Ice Area in Jutulsessen, in Dronning Maud Land, the largest to date, shows that unlike on glaciers, the key parameters responsible for creation of cryoconite holes and their chemical signature on the continental ice sheet, are ice flow supplying debris and bedrock geology supplying chemical species.

We show that contrary to previous belief, high-altitude, inland Blue Ice Areas operate as “biogeochemical powerplants”, producing and storing large quantities of nutrients, carbon (organic and inorganic), and viable microorganisms in a network of cryoconite holes. Some of the organisms observed in Jutulsessen have rarely been reported in cryoconite holes elsewhere, and are usually observed in dramatically different environments.

Contrary to their low-altitude counterparts, surface ice melt (ablation) in high-altitude Blue Ice Areas is driven by sublimation and so is influenced by air-temperature, relative humidity and wind speed.

However, solar radiation drives sub-surface melt across much of the blue ice, creating an expansive sub-surface hydrological network also connected to cryoconite holes via cracks and incipient channels. Water production and storage is most pronounced in terminal zones of the BIA, where debris advected by ice flow and transported for hundreds or thousands of kilometres, eroded from the bed or deposited on the surface create cryoconite holes covering up to 14% of the ice surface.

High concentrations of nutrients evolve in these holes, suggesting possibly old age and slow rates of flushing by the sub-surface melting allow the aqueous products of rock-water-microbe interactions to accumulate. These are further concentrated by solute rejection during refreezing and vapour deposition onto the underside of the ice lid that forms above the water stored in the holes during summer.

Enhanced melting taking place even in high-altitude BIAs associated with debris-rich cryoconite holes and sub-surface aquifers associated with them has yet to be accommodated into estimates of meltwater production in blue ice across the ice sheet surface. The volume of water storage in the holes themselves, equivalent to a 19.6 mm water layer across the entire Jutulsessen BIA, suggests that this process should not be overlooked.

Similarly, sub-surface aquifers created by solar radiation that can develop and persist throughout the winter at depths even down to 10 m should also be included in calculation of meltwater production across the ice sheet surface. Their presence, extent, capacity and possible connections within the ice should be taken into account.

Understanding the wider impact of high-altitude BIAs on the generation and distribution of water, production of nutrients and carbon, presence of microorganisms, as well as their distribution to downstream environments is essential. Especially because there are clear signs of an increasingly active drainage system developing upon the Antarctic Ice Sheet in recent decades42.

Consequently, understanding the wider significance of all BIAs for enabling meltwater production, nutrient transformation and allowing for microbial life to thrive within the most hostile surface conditions requires our urgent attention.

Methods

A field campaign in Jutulsessen was undertaken in the 2019/2020 Antarctic summer season between mid-November 2019 and the end of January 2020 (Fig. 7) producing data on cryoconite hole characteristics, their distribution, water content, their chemical signature and microbiology, characteristics of shallow ice cores, ice ablation as well as optical properties of the blue ice. Data from this campaign are supplemented by data describing ice flow and ice ablation, provided from the Norwegian Polar Institute’s monitoring program. These data were used in conjunction with the analysis of ice core samples and meteorological monitoring to model heat production with depth using a radiative transfer model.

Fig. 7: Study location.
figure 7

a: Satellite image of Jutulsessen in Dronning Maud Land (LIMA Landsat, 15 m resolution55); b: profiles of the bed in the western and central Dronning Maud Land, from Veseskarvet nunatak (SANAE IV research station on Fig. 1) to Jutulsessen (Troll Research Station), with a close-up on Jutulsessen blue ice62; c: drone image of cryoconite holes in Sætet, with people and snow scooters for scale; d: overview of cryoconite holes in Jutulsessen blue ice from 30 m elevation, view towards Troll Research Station.

Full size image

Field mapping and sampling

During the fieldwork eight primary sites were selected in each of the zones of the Blue Ice Area (Figs. 3, and 7). After arbitrarily locating suitable holes, measurements of CHs were performed in each zone (internal dimensions, ice lid thickness, water depth) and samples for biogeochemical and microbial analyses collected. In Total 53 CHs were sampled after crossing each zone and 81 measured. The ice lids of the holes were drilled through with a 10% HCl cleaned Kovacs drill, ice chips produced through drilling were collected in sterile Whirlpack bags for later processing. Water from CHs was collected with a sterile 60 ml syringe and also transferred into sterile Whirlpack bags for later processing. In situ measurements of electrical conductivity, pH, and oxygen saturation were performed with a HQD40d handheld metre and respective electrodes. Low ionic strength electrodes were used for electrical conductivity and the pH, while an optical sensor was used for oxygen. Alongside the cryoconite holes chosen for sampling, the upper 1 m of blue ice was also sampled for chemical analyses using the ice drill technique employed for the ice lid samples. In addition, six 10 L snow samples were collected from arbitrarily selected sites in Mimelia, Grjotlia and Stabben area, where small amount of snow could be found. Snow and glacier ice samples were melted at room temperature and processed with the water samples in a clean facility at Troll Research Station.

A further six ice cores of up to 2.4 m were collected in 0.8 m sections using a pre-cleaned 14 cm Kovacs Ice Corer with electrical drill. These samples were wrapped in sterile Whirlpak bags at the place of extraction and then dispatched at −18 °C by ship to Aberystwyth University for molecular analysis (see below).

Mapping of cryoconite holes (CHs)

A complete high-resolution map of CHs greater than 25 cm diameter (Fig. 8) was created by processing high-resolution satellite images (World View Stereo © 2013 – 2017 DigitalGlobe, Maxar Technologies, 5 m resolution, 25 cm pixel size) with QGIS Semi-automatic Classification Plugin to digitize CHs field (ISODATA, 8 classes, 20 iterations). The processing was manually validated by in-situ measurements of CH dimensions, aerial imagery from drone flybys (DJI Mavic Mini at 30 m altitude), and cross-examination of high-resolution satellite images with the classification results. The accuracy of QGIS processing was manually validated against DigitalGlobe image, DJI Mavic Mini images from ground-truthing campaign between November 2019 and January 2020, Fiji image processing software, and manual measurements of the holes’ dimensions. A comparison of satellite images of the BIA and its processed, digitized map is presented in Fig. 8.

Fig. 8: Comparison of the satellite image (Sentinel2) of the high-altitude Blue Ice Area in Jutulsessen with its complete digitized twin.
figure 8

a: Sentinel2 image of the BIA (source: Quantarctica59), blue rectangle was selected for a closeup presented in b; b: a close-up of the ice surface showing cryoconite holes extracted from a high-resolution satellite image taken in November 2019 (World View Stereo © 2013 – 2017 DigitalGlobe, Maxar Technologies); c: digitized version of a; created in QGiS with Semi-automatic classification plugin. Cryoconite holes in the digitized version are visible as dark blue dots. Areas with white outline represent Zones with distinct concentration of the holes; d: The close-up of the rectangle in c.

Full size image

CH Zones were delineated in QGIS using high-resolution Digital Globe satellite images and were based on the distribution of visible ice flow lines and the distribution of CHs across the BIA after a semi-automatic classification. Areas excluded from Zones were discarded due to either absence of CHs or high concentration of surface debris preventing identification of CHs and manual validation. It is difficult to quantify and assign individual errors to CHs mapping because total uncertainty is comprised of satellite image pixel resolution uncertainty, uncertainty in the empirical retrieval algorithm (which has not yet been explicitly quantified for Antarctica) and uncertainty in CH boundaries derived from pixel reflectance thresholding.

Optical properties of blue ice and radiative transfer modelling

Fieldwork

Incoming solar irradiance at the blue ice surface and within a cryoconite hole upon the debris surface (both in Zone 3) were used to supplement more comprehensive radiation measurements at Troll Station (Fig. 9). These employed Apogee SP110 pyranometers with a Campbell Scientific Cr800 data logger to construct a record of average values every 15 minutes from 30 s measurement interval. The debris surface sensor was emplaced after drilling through the cryoconite hole ice lid and disturbing the hole as little as possible.

Fig. 9: Light penetration through high-altitude blue ice in Jutulsessen, and modelled development of sub-surface aquifers in that area.
figure 9

a: Graph of solar radiation (irradiance) measured on the ice surface and inside cryoconite hole just above the debris in Zone 1, during December 2019 field season, b: Transmission (upwelling radiance divided by incoming irradiance at the surface) as a function of depth at 450 (green) and 650 nm (red) measured in blue ice (points). The lines are fitted to the measurements from 10 to 87 cm, as the measurement at 109 cm was affected by variability in incoming light, c: Results of radiative transfer model for the period 2018-2022, for two ice samples collected in Zone 1. Meteorological data used in the model were obtained from a weather station at Troll. The model considered variable solar zenith angle and different scenarios depending on cloud cover, concentration of air bubbles and impurities. White contours indicate −30, −20 and −10C isotherms, the black line indicates 0 C. Sample 1 has contained less air bubbles and they are also smaller. Sample 1 has bubble volume fraction 5.8 × 10−5 (L/L) and effective bubble radius of 35 µm. Sample 2 has bubble volume fraction of 8.0 × 10−3 (L/L) and effective bubble radius of 255 µm.

Full size image

Modelling

We used a multi-stream radiative transfer model (AccuRT), to compute the light field for several different coupled layers simultaneously50. In the development of AccuRT, particular attention has been given to the parameterizations of optical properties of sea-ice and snow51. Glacier ice differs optically from sea-ice in that it lacks brine pockets, but it can contain considerable amounts of bubbles and impurities. In the model, scattering was assumed to be dominated by non-absorbing bubbles, whilst impurities were assumed to be absorbing only. Eight atmospheric streams and 40 log-spaced detector depths from the ice surface down to 20 m were used. For the spectral resolution, 64 log-spaced wavelengths between 300 and 2500 nm were used, and the total scalar irradiance was calculated and extrapolated to additional depths. Gershun’s law was then used to compute the radiative heating of the ice (Fig. 9). Finally, a polynomial fit was used to approximate the effect of the changing solar zenith angle on the radiative heating. The polynomial was developed using solar zenith angles 50-89.9˚, and 60˚ was set as a reference zenith angle in all following simulations.

Earlier studies used simple two-stream models to estimate the blue ice radiative heating24,52,53. More importantly, the cloud cover was assumed to be constant year-round at a coverage of about 50%. In this study, simulations were run for different values of cloud thickness, and the numerical scheme used for the subsequent one-dimensional heat modelling uses an explicit finite-difference method, similar to previous studies, including a radiating bottom boundary at 20 m, similar to that used by Rasmus53. The implementation of the ice-surface boundary was based on Liston et al.24, except for meteorological measurements, which was retrieved from the Troll meteorological station for the years 2018-2022 (https://henry.ub.uit.no/Record/npolardata:oai:npolar.no:dataset%2F79d4b7c7-3bee-40a6-975f-f9ed4d253c44).

The treatment of melt-freeze processes was also similar to the treatment by Liston et al.24. The blue ice was assumed to have an initial temperature of -23 °C for all depths, which was determined from subsurface ice temperature profiles under a snow-covered surface close to the Troll airstrip. The time resolution was 120 s with a 2-cm spatial resolution. To test the sensitivity of both ice properties and cloud cover, four-year periods were simulated based on available meteorological data with a selection of log-spaced cloud thicknesses.

Biogeochemistry

All samples were syringe-filtered through 0.45 µm Whatman Puradisc Aqua 30 filters and stored in +4 °C until analyses in sterile 60 ml Corning centrifuge tubes prewashed with filtrate. DOC and TIC samples were stored in 40 mL Sievers-certified sterilized glass vials. Ion analyses were performed at the University of Sheffield, UK (Dionex ICS-1500 for cations, Dionex ICS-1100 for anions). Precision errors for these ions were all <2%. Si and NH4+ were analysed on Skalar San + + Continuous Flow Analyzer. Carbon analyses were performed on Sievers 5310 with UV and persulphate digestion method, at the University Centre in Svalbard (detection limit 0.010 mg L−1, Precision errors <3%).

Cryoconite microbiology

All pre-amplification steps for the analysis of bacterial communities within cryoconite holes were performed aseptically in a dedicated clean laboratory, with manipulations conducted in a laminar flow hood using aerosol-resistant tips and certified DNA-free plasticware by operators in clean suits, masks, and gloves. All equipment and surfaces were regularly disinfected by 1000 ppm available chlorine solutions. Extraction and no-template controls were analysed in parallel with field samples. Community genomic DNA was extracted from 250 mg fresh weight of cryoconite debris using a Qiagen, Inc. Powersoil DNA kit as specified by the manufacturer.

The DNA yield was quantified using a Qubit high sensitivity DNA fluorometry kit (Invitrogen, Inc.) V1–V9 16 S ribosomal RNA gene PCR was conducted in single cap tubes using 7 µL of each DNA extract as a template in 30 µL reactions primed by 27F-1389R54 where 27 F was modified by the presence of a sample-specific Oxford Nanopore barcode, with reaction cocktails comprising 2× OneTaq Quick-Load 2X Master Mix with Standard Buffer (NEB: #M0486) supplemented with 2 µg molecular biology grade bovine serum albumin (NEB: #B9000). Thermocycling conditions for the PCR comprised 35 cycles of one minute denaturation (94 °C), annealing (55 °C), and extension (68 °C) followed by a ten-minute terminal extension (68 °C). Pooled products were cleaned using a gel extraction kit (Qiagen, Inc) and used for ligation sequencing (SQK-LSK109) on R9.4.1 flow cells (Oxford Nanopore Technologies, Ltd) on a MinION Mk1C as instructed by the respective manufacturers. High Accuracy basecalled reads were demultiplexed using Porechop v0.2.4 (https://github.com/rrwick/Porechop) and analysed using SituSeq55 with the default settings for stream 1, and alignment to SILVA 138.1. Relative abundance of bacterial phyla in randomly selected holes (H) across different Zones of the blue ice in Jutulsessen is presented in Fig. 10.

Fig. 10: Relative abundance of bacterial phyla in randomly selected holes (H) across different Zones of the blue ice in Jutulsessen.
figure 10

Two versions of the graph a, b were used to clearly show recorded phyla. a: relative abundance showcasing phyla with lower abundance; b: relative abundance showcasing phyla with high abundance. V1–V9 16 S ribosomal RNA gene sequencing performed on community genomic DNA. 1,222,427 reads were assigned to barcodes representing each of the twenty cryoconite debris samples with an average of 64,496 reads per sample and a range of 41,010–91,516 reads per sample. SituSeq55 was used to assign taxonomy to the sequences generated.

Full size image

Ice core analysis: air bubbles

End samples from the upper core sub-section were collected from each of the six sites just prior to their processing for microbiological analysis in the form of thin (ca. 1 cm) disks. In all cases, the disks broke into irregular fragments. Each fragment was therefore weighed and then photographed with a 16 MP APS-C sensor on a petri dish with scale bar. A higher-than-expected bubble density meant that overlapping bubbles precluded automated image analysis. Therefore, all bubbles were counted manually and then the bubble size distribution was established through sub-sampling and measurement of 108–145 bubbles from each core.