Our sampling strategy was designed to characterise the range of sources contributing to sediment accumulation in the most distal lake, Isfallssjön, and encompasses cryoconite from the snow-free surface of the ablation zone (n=14), sediments within the two braided proglacial outlet streams (n=11), surface sediments from the overridden inner slopes of the north and south moraine (n=10), and sediments from the central foreland (n=2) (Fig. 1). Combined, these samples span 1150 to 1340 m a.s.l. Rockfall also contributes as a sediment source within the catchment; however this material was not sampled due to the active nature of the rockfall, which would make sampling dangerous. All samples were collected between 7 and 17 August 2017, towards the end of the ablation season when snow cover was minimal outside of the accumulation zone. We conducted cryoconite sampling in two transverse profiles to investigate the effects of aspect and distance from the valley side on the accumulation of FRNs and other materials, and we sampled in the proglacial outlets at intervals along the reaches of the two streams to investigate any possible downstream changes in FRN activity concentrations. Each sample was collected in a spatially integrated manner by sampling from five sites within a metre of a central point and stored within new, clean plastic sampling bags or 50 mL tubes. We also retrieved a 38 cm lake sediment core from Isfallssjön using a HTH 90 mm diameter gravity corer, and we extruded the core on-site at 1 cm intervals. Samples were subsequently oven-dried at 100 ∘C for the minimum time required to reach a constant weight and the <75 µm component retrieved for subsequent geochemical analyses. Due to the limited amount of cryoconite available for sampling at each supraglacial site, which varied in mass based on cryoconite type (e.g. in a cryoconite hole, submerged in supraglacial water, distributed on the ice surface), we reserved the entire bulk sample to ensure we had sufficient material for gamma spectrometry. Lake core sections were also preserved in bulk for particle size analysis. Particle size analysis was performed in triplicate on all samples taken from the catchment, using laser diffraction. We use the surface-area-weighted mean particle size, or D [3, 2], for subsequent data analysis presented here as this is the most sensitive measure where fine particulates are common within the size distribution (Malvern, 2015), and relevant where reactivity and bioavailability are of potential importance.

Radioactive analyses were carried out in the ISO9001 accredited Consolidated Radio-isotope Facility (CoRIF) at the University of Plymouth, applying an established methodology (e.g. Wynants et al., 2020). Particulate samples for the well detector were packed and sealed into 4 mL plastic vials, and samples for analysis on the planar detector were packed and sealed into 90 mm plastic petri dishes. Sample weighing was conducted on a calibrated balance. Samples were packed within 14 d of return to the laboratory to ensure rapid analysis and were incubated for 22 d prior to analysis to allow the development of secular equilibrium along the 238U decay chain. Gamma counting was conducted using well (GWL-170-15-S; N-type) and planar (GEM-FX8530-S; N-type) spectrometers, both consisting of liquid-nitrogen-cooled, high-purity germanium semiconductor detectors (EG & G ORTEC, Wokingham, UK). The well detector had a full width at half maximum (FWHM) for the 1330 keV line of 60Co of 2.17 keV, and the planar detector had a FWHM of 1.76 keV and a relative efficiency of 50.9 %. The energies, peak widths, and efficiencies of the gamma spectrometers were calibrated using a natural, homogenised soil, with low background activity, which had been spiked with a certified, traceable mixed radioactive solution (80717-669 supplied by Eckert & Ziegler Analytics, Georgia, USA). Calibration relationships were derived using ORTEC GammaVision © software. After incubation, the spiked soils and the samples from Isfallsglaciären, and empty sample containers as blanks, were counted for at least 24 h, and all activities were decay-corrected to the sample collection date. The uncertainties were estimated from the counting statistics and are quoted with a 2σ counting error. Unsupported 210Pb (210Pbun) activities were obtained by the subtraction of 226Ra activity, deduced from the gamma emissions of 214Pb, from the measured total activity of 210Pb (210PbT). Quality control analyses were carried out regularly using soils from the IAEA world-wide proficiency tests, including a moss soil (IAEA-CU-2009-03) and another soil (IAEA-TEL-2012-03) (Table A1).

We analysed all samples for a full suite of major and minor elements using wavelength-dispersive X-ray fluorescence (WD XRF) spectrometry. For the proglacial area and lake core, each sample was milled using a Fritsch pulverisette, mixed with a Ceridust 6050M S1000 polypropylene wax binder (Clariant, Switzerland), and pressed into a pellet. The dried cryoconite samples were powdered by hand using a pestle and mortar prior to being packed into 40 mm diameter cups fitted with 6 µm polypropylene spectromembrane (Chemplex, USA). All samples were packed to the same volume and left to settle for 24 h prior to analysis. Analyses were undertaken in the CoRIF lab by WD XRF spectrometry (Axios Max, PANalytical, the Netherlands). The instrument was operated at 4 kW using a Rh target X-ray tube. During sequential analysis of elements, tube settings ranged from 25 kV, 160 mA for low-atomic-weight elements up to 60 kV, 66 mA for higher-atomic-weight elements. All analyses were undertaken using the Omnian analysis application (PANalytical, Netherlands) under a medium of He. This approach offers a rapid and non-destructive means of determining a wide range of elemental concentrations in cryoconite. Repeatability of the approach was assessed by repacking and analysing cryoconite samples in triplicate with the relative standard deviation found to be <10 % across triplicates. Cross comparison to results obtained from a validated inductively coupled plasma optical emission spectrometry (ICP-OES) procedure showed XRF-derived concentrations were in close agreement (within 15 % relative to ICP-OES) for the elements of interest.

The dried cryoconite samples were ground by hand using a pestle and mortar. Particulate N and C were determined via elemental analysis (Carlo Erba, EA1110, Italy). The instrument was calibrated using acetanilide, empty pre-combusted capsules were analysed as blanks, and the accuracy of the analyses were checked using the certified reference material PACS-2 (National Research Council of Canada). The results showed that the analyses were accurate to within 10 % of certified values. The ground samples for 13C/12C and 15N/14N analysis were packed into tin capsules and weighed using a calibrated balance. The 13C/12C and 15N/14N ratios were determined at the Isotope Bioscience Laboratory at Ghent University using an elemental analyser (ANCA-SL, Sercon, UK) coupled to an isotope ratio mass spectrometer (20–22, Sercon, UK). The measured δ13C and δ15N values were given relative to the international standards, Vienna PeeDee Belemnite (V-PDB) and AIR, respectively. This calibration was done using the IA-R001 15N/13C wheat flour laboratory standard (δ13CV-PDB=-26.43±0.08 ‰ and δ15NAIR=+2.55±0.22 ‰) and an in-house quality assurance organic reference. The average standard deviation on the δ value was determined by measuring five randomly selected samples in triplicate, giving a standard deviation of 0.32 ‰ for 13C and 0.14 ‰ for 15N. For the final analysis of cryoconite, each sample was analysed in duplicate, and an average was taken.

The sedimentary archive from the proglacial lake core was used to construct a sedimentation chronology using fallout 210Pb supported by known 137Cs date horizons. Reference dates from Chernobyl (1986) and weapons testing (1963 peak and 1952 onset) were used to constrain the chronology following principles outlined in Appleby (2002) in two phases. Due to known disruption of sediment flux from the glacier to Isfallssjön in 1959 with formation of Frontsjön, the constant rate of supply (CRS) model was applied to the core in two sections. This was done to account for any potential change in secondary 210Pbun supply. The CRS model was initially run fitting the 210Pb profile to the lowermost measurement of 241Am, wherein 241Am is known to have been predominantly supplied by global weapons testing fallout (Olszewski et al., 2018). The core was then split at the dated horizon 1959, and the lower section was analysed with a separate CRS model benchmarked to the 1952 onset of 137Cs fallout. Due to low activity concentrations and detection challenges in the lowermost section, the profile tail was modelled using an exponential function fitted (r2=0.99) to three high-precision measurements derived from extended count times, noting the tail represents a small overall proportion of total inventory. Horizon dates and sediment accumulation rates were derived as outlined by Appleby (2002) for each separate model application.

Fourteen samples of cryoconite were retrieved from the surface of Isfallsglaciären, which are characterised by the range of radionuclide concentrations described in Table A2. The mean activity concentrations of 137Cs, 210Pbun, and 241Am in cryoconite are 3069±941, 9777±780, and 25.8±16.7 Bq kg-1 respectively, reaching a maximum of 4533±350, 14663±1167, and 74.0±10.2 Bq kg-1. While 210Pb is a natural radioisotope derived from the decay of 222Rn in the atmosphere (Gäggeler et al., 2020), 137Cs and 241Am are anthropogenic FRNs, distributed via atmospheric transport, and are common fission by-products from nuclear reactors and weapons testing (Lindblom, 1969). The anthropogenic radionuclide 137Cs is partially soluble in water, and is a radionuclide of concern in terms of both animal and human health (Van Oostdam et al., 1999). With a half-life of 30.17 years, it is relatively short-lived in the environment, and the 137Cs deposited globally through long-range atmospheric deposition following the Chernobyl accident has decayed by 50 % since 1986 (Olszewski et al., 2018). The half-life of 241Am is considerably longer at 432.2 years and is increasing in the environment due to the short half-life (14 years) of its parent radionuclide 241Pu. 241Am is an alpha emitter and less exchangeable (acid–water soluble) than 137Cs (Kovacheva et al., 2014); however it is potentially harmful if ingested (e.g. Harrison et al., 1994). The primordial radionuclide 40K is also detected in our cryoconite samples at relatively high activities (an average of 1839±168 and a maximum of 2054±207 Bq kg-1). The 40K mean activities found in cryoconite on Isfallsglaciären exceed the maximum activities found on both the Forni and Morteratsch glaciers in the Italian and Swiss Alps (770±200 and 810±55 Bq kg-1 respectively) as reported by Baccolo et al. (2020a), and the considerably higher maximum of 1440±40 Bq kg-1 recorded on the Stubacher Sonnblickkees glacier of the Austrian Alps by Wilfinger et al. (2018). Since 40K is a natural component of rock, the high activities found at Isfallsglaciären are likely related to the geochemical signature of the surrounding catchment geology, and in particular to the abundance of potassium in the rock and sediment.

By considering both the activities of radionuclides and the content of C and N, it is possible to explore the relationship between the organic content of cryoconite and the distribution of radioactivity. The spatial variability of activities for selected natural and anthropogenic radionuclides is illustrated in Fig. 2, with the mass fraction of C and N (% C and % N) measured in the cryoconite samples through bulk stable isotope analysis shown in Fig. 2C. The relationship between organic content and accumulation of some radionuclides is illustrated by the relatively low values of both % N and % C and natural radionuclides 210Pb and 7Be, in the southernmost samples from the upper glacier transect. Higher relative values are found in samples collected to the north. Values of % C and % N cover a wide range of 6.4 %–24.5 % and 0.4 %–1.2 % respectively, with an average C/N ratio of 12.1±1.6 %, which may be attributable to the complex microbial community present in the cryoconite. Typically, micro-organisms mineralise N from the organic matrix to support plant uptake of the nutrient. This is because soil micro-organisms require a cellular C/N ratio of about 8, which is maintained via the N mineralisation. Material in cryoconite holes in Antarctica was found to have a relatively low carbon content of 0.06 %–0.35 %, contributing to C/N ratios in the range 3.5–8.2 (Bagshaw et al., 2013), while samples from the Morteratsch and Forni glaciers had mass ratios of 9.4±1.4 % and 7.2±0.8 % for organic matter respectively and 0.5±0.25 % and 0.2±0.2 % for elemental carbon (Baccolo et al., 2020a). Thus, there are considerable differences in the C and N contents of particulate matter found in cryoconite holes across the cryosphere.

The presence of the cosmogenic radionuclide 7Be in these samples also provides insight into the process of accumulation of radionuclides in cryoconite. Despite being one of the most stable beryllium radioisotopes, the half-life of 7Be is relatively short at 53 d (cf. 1.39 million years for 10Be), yet it is present in all but one of the cryoconite samples, with a mean activity of 1014±599 Bq kg-1. This is at least an order of magnitude higher than the activities of 7Be typically observed in surface soils and fine river sediments in the mid-latitudes (e.g. Smith et al., 2014; Ryken et al., 2016). 7Be demonstrates rapid sorption to sediment particles and has been shown to have an affinity for reducible (e.g. Fe/Mn oxides) and oxidisable (e.g. organic) fractions (Taylor et al., 2012). Finding high activities of 7Be in cryoconite implies a recent accumulation history and supports the role of meltwater – likely sourced from recent snowfall – in providing a crucial link between the radionuclides stored in glacier ice and cryoconite (Baccolo et al., 2020b). The atmospheric deposition of 7Be is affected by a number of factors, including its availability in surface air for scavenging by precipitation (Aldahan et al., 2001). Concentrations of 7Be are generally higher in mid-latitude surface air masses (Kulan et al., 2006), with atmospheric circulation driving the downward transport of 7Be-rich air from the upper troposphere (Aldahan et al., 2001). However, in the polar regions, descent of upper troposphere air is less owing to the stability of the air masses; thus, surface air of polar origin is typically found to have relatively low 7Be activities. It is generally accepted that 7Be is largely transported to the Arctic from the mid-latitudes, with a strong seasonal variation (higher in late winter–spring) that can correspond with transport of contaminants (Feely et al., 1989). Potential 7Be-rich air masses in late winter–spring with corresponding deposition, coupled with summer meltwater production and the concentrating effect of radionuclide exchange at the water–sediment interface described above, may help to explain the relatively high activities found in these cryoconite samples from Arctic Sweden. Routine monitoring of 7Be in precipitation can identify temporal variability linked to atmospheric processes (Taylor et al., 2016). In this regard 7Be could be a useful proxy for transfer of contaminants to the cryosphere in the context of seasonal dynamics of atmospheric circulation (Terzi et al., 2020).

The average inorganic composition of cryoconite based on XRF analysis is shown in Fig. 3. Not unexpectedly, SiO2 is by far the most abundant element in the samples, averaging ∼357900 ppm, followed by Al2O3 (∼111400 ppm) and Fe2O3 (∼105400 ppm). Based on calculating the normalised standard deviation for each element, the element with the highest variance between the 14 cryoconite samples is Cu (0.376), while the least variance between samples (0.034) is found for Fe2O3, one of the most abundant elements found in cryoconite on Isfallsglaciären. The sum of the concentrations of major and trace element oxides detected via XRF spectrometry for the cryoconite samples is between 63.7 % and 79.5 %, and a further 6.8 %–13.5 % can be attributed to C and N based on bulk stable isotope analysis.

The means of selected metal concentrations in cryoconite samples are shown in Table A3. The concentrations for Cu, Fe, Pb, and Zn are elevated over those associated with cryoconite granules found in a glacier in Svalbard (Łokas et al., 2016). These values together with the Al concentrations were used to determine enrichment factors (EFs) from Eq. (1): 1EF=(M)(Al)(M)Ref(Al)Ref, where (M) and (Al) are the total concentrations of a metal and of aluminium, and (M)Ref and (Al)Ref are the reference values of the metal and aluminium for the upper continental crust (Wedepohl, 1995). In these samples from Isfallsglaciären the mean of Al concentrations (n=14) was 57930±3250 mg kg-1. Internationally, the Canadian sediment guidelines for risk to aquatic life (CCME, 1995) can be used to evaluate whether the particulate matter at a site is contaminated or not. In general terms, a sample with an EF falling in the range 1<EF<3 has minor enrichment, the range 3<EF<5 indicates moderate enrichment, the range 5<EF<10 is assessed as moderate to severe enrichment, and EF>10 is classed as severe enrichment. The cryoconite samples have elevated metal concentrations, but only Cr and Pb have concentrations above the probable effect level (PEL). The EF values are clustered in three groups with the highest for Cu and Pb, followed by Cr and Ni, while Fe, Ti, and Zn have the lowest EFs.

A principal component analysis (PCA) was conducted for the cryoconite samples to help explore the variance between the samples, based on gamma spectrometry, particle size analysis, and stable isotope analysis. The PCA scores and loadings for principal components 1 and 2 are depicted in Fig. 4, and no outliers were identified in the data based on the Mahalanobis distance. The PCA loadings (Fig. 3B; Table A4) show that the content of C and N in cryoconite has large positive loadings on principal component 1, while area-weighted particle size (D [3, 2]) has a large positive loading on principal component 2, closely followed by strong negative loadings from δ13C and δ15N. These first two components explain 70 % of the variance in the data, with principal component 1 explaining 47 % and component 2 explaining a further 23 %. Principal components 3, 4, and 5 explain 10 %, 6.4 %, and 5.9 % of the variance respectively, and combined the first five principal components explain 93 % of the variance in the data. Table A4 contains the eigenvector values for principal components 1 to 5, highlighting the influence of 7Be, 40K, and 210Pb on components 3 to 5. While sample numbers are limited, the PCA nevertheless reveals a clustering in cryoconite samples collected from the north side of the southern glacier terminus (Fig. 4a; see Fig. 1C), suggesting that exposure to sunlight may be an influencing factor on the accumulation of FRNs, due to increased melting and/or available energy (see Fig. 2). This in turn may play an important role in organic content of cryoconite by providing energy for flora and microbial life and warrants further investigation in the future. It has previously been shown that a higher proportion of bioavailable C, N, and P is present in cryoconite in comparison to source materials (Bagshaw et al., 2013), highlighting the significant role it plays within biogeochemical cycling. As demonstrated here, other elements are also found in higher concentrations where energy availability is greater, likely due to the microbial activity within cryoconite.

A second PCA was performed to investigate the role of inorganic composition of cryoconite by including major and minor elements detected through XRF analysis (PCA scores and loadings for principal components 1 and 2 are depicted in Fig. 4c and d). In this case the first three components explain 70 % of the variance, and the first six 90 %, with the first principal component explaining 45 % and components 2 to 6 explaining 14 %, 12 %, 9 %, 6.6 %, and 4.3 % of the variance respectively. Examination of the eigenvector values for this PCA illustrate the influence of a negative loading from CaO and Na2O for principal component 1, followed closely by MgO and Cl, with area-weighted particle size (D [3, 2]) again showing a strong negative loading for component 2. CaO, Na2O, MgO, and Cl are all relatively soluble elements, which may explain the clustering of these elements and their role in explaining variance within the samples.

The activities of radionuclides detected within the proglacial area of the Isfallsglaciären catchment are significantly lower than those in cryoconite (Table A1; Fig. 5). This supports the fact that cryoconite is a highly efficient accumulator of radionuclides to the extent that activities are orders of magnitude above those which were deposited and accumulated “off ice”. The deglaciated central forefield has very low levels of FRNs despite having been exposed by the ice before the weapons testing era and Chernobyl; by comparison, the samples of proglacial outwash, which are fed by a regular supply of meltwater and sediment from the glacier, are characterised by much higher activity concentrations of natural radionuclides, particularly for 210Pb and 7Be. Indeed, of the proglacial samples, 7Be is only found in proglacial outwash and lake core sediments, suggesting that the transport of sediment and radionuclides in meltwater is important for their downstream accumulation (Fig. 5) in addition to their accumulation in cryoconite (Baccolo et al., 2020b). The importance of interaction with meltwater, and possible enrichment by runoff of supraglacial sediments such as cryoconite, is further supported by the elevated levels of radionuclides present in the upper portions of the proglacial lake core which are in excess of all other off-ice sediment sources sampled here. Correlation analysis also supports the importance of sediment for accumulation of some FRNs, particularly 40K, in both proglacial and lake core sediments via a negative relationship between FRN activity concentrations and particle size (Fig. 6). This relationship is not present for moraine sediments and thus likely reflects the importance of hydrological sorting of sediments and the presence of fine particles. 241Am is found only in the middle portions of the lake core, corresponding to known dates of nuclear activity which will be discussed further below. These results may reflect a more continuous flux of natural radionuclides from the glacier to the proglacial area, while FRNs, deposited during temporally restricted events, melt out more sporadically due to their storage in defined layers within the snow, firn, and ice. While the activity concentrations of FRNs in moraine sediments are generally low in comparison to cryoconite and proglacial outwash, there is a clear anomaly in the radionuclide concentrations from one sample which contained 74 Bq kg-1 of 210Pb and 207 Bq kg-1 of 137Cs (Fig. 5). This anomaly suggests the possible presence of localised off-ice hotspots, which may be representative of efficient accumulation of FRNs via lichens and mosses from direct atmospheric deposition, as has been reported in other environments (e.g. Sumerling, 1984; Paatero et al., 1998; Kirchner and Daillant, 2002); however we are cautious in our interpretation of this single anomalous sample.

The spatial distribution of radionuclide activity concentrations in the proglacial area of Isfallsglaciären is shown in Fig. 7. 7Be is only present in proglacial outwash; however there is no clear spatial pattern of which samples this has been detected in (Fig. 7a). There is little variation in 137Cs between moraine, forefield, and proglacial outwash sediments; however the anomalous value described above is clearly visible in the moraine sample closest to the northern glacier terminus. There is much more obvious variation in activity concentrations for both 210Pb and 40K, which illustrate a clear difference between the sediments transported in the northern and southern proglacial outlet streams (Fig. 7b). 40K is, for the most part, present in higher levels in moraine sediments (particularly the northern moraine) than the proglacial outwash, in addition to the central forefield sample furthest from the present-day terminus (and most isolated from the braided stream system). 40K is a common element in the Earth's crust, and the relative stability of moraines in comparison to proglacial outwash may allow for increased accumulation of this radionuclide, which has a very long half-life of 1.251 billion years. Potassium is also a relatively soluble element, and thus it may be expected that activity levels would be lowest in areas influenced by a dynamic hydrological system, while levels in areas isolated from water (aside from precipitation) accumulate more 40K. The spatial distribution of 210Pb in the proglacial area of Isfallsglaciären is more complex, with values in sediments from the southern proglacial outlet stream being notably higher than those from the north, and those in moraine sediments. This may be influenced by the glacier surface topography, as the northern terminus lobe is considerably steeper and more crevassed than the southern lobe, which may restrict the ability for cryoconite to accumulate on the surface and meltwater to flow and transfer materials uninterrupted, thus leading to decreased FRN enrichment of supraglacial sediments. The southern proglacial outlet may also have a higher discharge, or be more dynamic in its flow pathways, allowing for accumulation of sediment from a larger sediment source pool.

The downcore profile of fallout 210Pb against mass depth (reflecting accumulation rates) in lake sediments from Isfallssjön (Fig. 8a) departs from exponential decline, implying periods of enhanced sedimentation. From mass depth ca. 40 g cm-2 (true depth 23 cm) downward, activity concentrations approached the limit of detection wherein selected samples were counted for a longer duration to achieve measurable values to model the tail for CRS modelling (not shown). The 137Cs profile (Fig. 8b) shows a first detectable activity concentration at mass depth 62 g cm-2 (true depth 35 cm). Following Lindblom (1969), this is inferred to represent the onset of early weapons testing in 1952. Subsequent peaks, moving upward, are inferred to represent (i) a mid-1950s spike in atmospheric fallout (mass depth 42–45 g cm-2); (ii) the 1963 peak in fallout (mass depth 35 g cm-2), which is also the first detectable activity concentration of 241Am linked predominantly to global fallout (Bunzl et al., 1995); and finally (iii), at mass depth ca 10–12 g cm-2, increased 137Cs activity concentration associated with fallout from the Chernobyl nuclear accident (Olszewski et al., 2018). Within this well-constrained geochronological framework (Fig. 8c), it can be seen that sedimentation rates (Fig. 8d) were significantly reduced c. 50 years ago, reflecting the post-1959 formation of Frontsjön following terminus retreat (Karlén, 1973), and subsequent “piracy” of proglacial waters from Isfallssjön. A critical question remains about deposition and release of FRNs from the glacial ice during seasonal melt and more recently accelerating retreat due to global warming. The sedimentary data imply a degree of lagged release during the 1960s, with a protracted detection of 241Am (0.4 to 0.5 Bq kg-1) in sediment from mass depth of 33 to 26 g cm-2 representing the period 1963 to 1970. This might relate to release by ice or erosion of surficial sediment from the forefield, i.e. a secondary source. 241Am activity concentration was below detectable limits (< ca. 1.5 Bq kg-1) after 1970, implying that any FRN activity associated with cryoconite has been diluted by other sediment sources during melt, release, and transportation, despite the limited distance between the lake coring site and the glacier.

There has been a considerable research effort in understanding the uptake of FRNs within flora and fauna, particularly following the 1986 Chernobyl accident. Mosses, lichens, and fungi are environmental matrices known to efficiently accumulate FRNs (Heinrich, 1992; Steinnes and Njåstad, 1993) and play a crucial role in radionuclide uptake into the food chain, particularly for reindeer and other ruminants (MacDonald et al., 2007). The effect of deposition of 137Cs from atmospheric transport following the Chernobyl accident was considerable for reindeer herding in Sweden, Norway, and other northern countries, due to contamination of the lichen–reindeer–human food chain (Skuterud et al., 2016), and a number of studies have demonstrated the importance of origin (food source) for 137Cs transfer, highlighting the lichen content of diets as a key control on uptake (Ahman et al., 2001; Skuterud et al., 2004). Despite an initially rapid drop in 137Cs in reindeer tissue in Sweden post-Chernobyl (Ahman and Ahman, 1994), concerns around the long-term impacts of exposure to fallout remain. A recent testing campaign by the Swedish Radiation Safety Authority revealed levels of 137Cs of up to 39 706 Bq kg-1 in wild boar in 2017–2018, with as many as 30 % of 229 boar tested exceeding the Swedish limit for meat consumption of 1500 Bq kg-1 (Strålsäkerhetsmyndigheten, 2020). The uptake of 137Cs by lichens and fungi is likely contributing to this persistence of high levels of radioactivity in boar, in addition to the migration of boar into regions affected by Chernobyl. It has been suggested that areas with previous 137Cs contamination may augment 137Cs transfer within the food chain following future contamination events due to the existing 137Cs burden in soils (Ahman et al., 2001). The presence of radionuclides in proglacial sediments in response to ongoing glacial retreat and down-wasting could thus pose an emerging threat for ecosystem health, with the possibility of a knock-on socio-economic impact due to the health considerations of animal–human transfer and stringent controls on limits for sale of produce for human consumption (Kristersson et al., 2017).

The levels of 137Cs detected in cryoconite on Isfallsglaciären are exceptionally high. Indeed, to the best of our knowledge they are some of the highest activity concentrations found in natural environmental matrices outside of nuclear exclusion zones. In light of our findings, and the similarly elevated levels of FRNs detected in cryoconite in other regions of the cryosphere, we recommend an increased research focus on this poorly understood contributor to contamination in proglacial environments, particularly in light of a continued trend of glacier mass loss and meltwater production. We identify a need to establish the presence of FRNs in glacial sediments across a wider spatial range, with a particular focus on regions where glacial meltwater is crucial to downstream water and food security, including the Andes and Himalayas. To more fully understand any potential impact of secondary FRN contamination with glacier retreat, the total mass of cryoconite in glacier catchments must also be considered when assessing whether FRNs are likely to pose any threat to downstream ecosystems. Distribution of hotspots in proglacial environments, and the concentrations of FRNs in downstream sediment sinks, must also be better constrained in order to evaluate implications for both aquatic and terrestrial fauna. Furthermore, we recommend that the bioavailability of FRNs in glacial sediments, including cryoconite, is assessed to understand whether the presence of FRNs in these settings can be taken up in the food chain to levels which are potentially harmful to fauna or for human consumption, or whether downstream dilution and distribution render these harmless. While the risk to distal communities is almost certainly low due to the dilution in rivers, fragile, pioneering ecosystems in newly exposed proglacial zones are more likely to accumulate radionuclides from meltwater and cryoconite transfer and may be candidates for monitoring to evaluate risk to local fauna.