Global variability and controls on the accumulation of fallout radionuclides in cryoconite

⁎ Corresponding author. E-mail address: caroline.clason@durham.ac.uk (C.C. Clas http://dx.doi.org/10.1016/j.scitotenv.2023.164902 Received 28 March 2023; Received in revised form 1 Available online 19 June 2023 0048-9697/© 2023 The Authors. Published by Elsevi • Accumulation of fallout radionuclides is commonplace in cryoconite on glaciers


H I G H L I G H T S G R A P H I C A L A B S T R A C T
• Accumulation of fallout radionuclides is commonplace in cryoconite on glaciers around the world. • There is large inter-regional variability in fallout radionuclide activity concentrations in cryoconite. • Fallout radionuclides concentrations can be orders of magnitude higher than found in other environmental matrices. • The organic content of cryoconite plays a central role in its ability to accumulate fallout radionuclides.
Science of the Total Environment 894 (2023) 164902

Introduction
The cryosphere is currently undergoing significant change in response to climatic warming, posing a threat to glacier-fed freshwater resources with associated impacts for communities and ecosystems in polar and mountain environments (IPCC, 2019;Clason et al., 2023). Glaciers are important archives of atmospheric constituents, including contaminants, whose deposition history can sometimes be reconstructed through analysis of ice cores (e.g. Barbante et al., 2004). However, glaciers are open systems and materials preserved within their ice and snow will eventually be released into the downstream environment, including legacy contaminants first emitted to the atmosphere decades ago (Bogdal et al., 2009;Ferrario et al., 2017;Beard et al., 2022a, b). As such, glaciers act as secondary sources of contaminants discharged from both snow and ice, and from proglacial areas (Schmid et al., 2011;Pavlova et al., 2016), under present-day accelerated melting.
Cryoconite, a supraglacial material comprising both organic and inorganic components (Cook et al., 2016a), plays an important role as a sink for nutrients and contaminants (Bagshaw et al., 2013;Baccolo et al., 2017;Li et al., 2017;Huang et al., 2019), and its ability to accumulate fallout radionuclides (FRNs) has received attention in several regions of the cryosphere (e.g. Łokas et al., 2018, 2022Huang et al., 2019;Owens et al., 2019Owens et al., , 2023Baccolo et al., 2020aBaccolo et al., , 2020bClason et al., 2021;Cao et al., 2023). The activities of FRNs found in some cryoconite deposits are among the highest ever observed in environmental matrices outside of nuclear exclusion zones, with activities exceeding 10,000 Bq kg −1 for a single radionuclide (Tieber et al., 2009;Baccolo et al., 2017;Owens et al., 2019). FRNs are also present within glacial ice and snow, but at concentrations in the range of mBq kg −1 (Olivier et al., 2004;Gäggeler et al., 2020), several orders of magnitude lower than found in cryoconite. Such efficient accumulation of FRNs in cryoconite might be partly explained by its high organic matter content (Cook et al., 2016a), which has geochemical affinity for radionuclides and other particle-bound contaminants (Gadd, 1996). Cryoconite can often be in contact with supraglacial meltwater for prolonged periods, and this interaction with meltwater has also been proposed as a potential mechanism for accumulation of FRNs in cryoconite, via remobilisation, transport, and scavenging of legacy contaminants (Łokas et al., 2014Owens et al., 2019;Baccolo et al., 2020a). Analysis of FRNs with different half-lives demonstrates that accumulation of FRNs in cryoconite from meltwater is a continuous process, and that the radioactivity burden of cryoconite increases with time before being released from glaciers to the downstream environment (Baccolo et al., 2020a(Baccolo et al., , 2020b. Once released, cryoconite and sorbed radionuclides are diluted in proglacial waters and sediments, although it has been shown that before being diluted, radionuclides can accumulate in proglacial areas, including lake sediments, posing potential ecological risks (Łokas et al., 2017Owens et al., 2019Owens et al., , 2023Clason et al., 2021).
This work focuses on Caesium-137 ( 137 Cs), Americium-241 ( 241 Am), and Lead-210 ( 210 Pb) as previous studies have demonstrated that they are the most abundant gamma-emitting FRNs present in cryoconite, with activities often exceeding 1000 Bq kg −1 for 137 Cs and 210 Pb (e.g., Baccolo et al., 2020b;Clason et al., 2021;Łokas et al., 2022). 137 Cs was introduced into the environment via atmospheric nuclear weapons testing from 1945 onwards, with the majority of test activity occurring in the 1950s and 1960s (UNSCEAR, 2000), prior to the decline of above-ground tests following the introduction of the Nuclear Test Ban Treaty in 1963. Accidental releases of 137 Cs from nuclear power and reprocessing plants have also been major contributors to global fallout, including the 1986 Chernobyl reactor fire and the Fukushima accident in 2011 (Bunzl and Kracke, 1990;Chino et al., 2011;Steinhauser et al., 2014), in addition to a range of other local and regional sources (Foucher et al., 2021). 137 Cs is a high-yield fission product of 235 U with a half-life (t 1/2 ) of 30.1 years, and due to this relatively long half-life, and its biogeochemical properties, 137 Cs has been able to spread across the globe, becoming one of the most studied radioactive contaminants in the environment (Cambray et al., 1990;Foucher et al., 2021). 241 Am (t 1/2 = 432 years) emits both alpha and gamma radiation. It is present in small quantities in spent nuclear fuel, but its environmental occurrence mostly depends on the decay of its parent radionuclide, 241 Pu (t 1/ 2 = 14.3 years), an important component of global radioactive fallout from atmospheric nuclear tests (Thakur and Ward, 2018). As a consequence of the decay of 241 Pu, the activity of 241 Am in the environment is increasing and is expected to peak by the end of the 21st century (Muravitsky et al., 2005). The geographic distribution of 137 Cs and 241 Am, and indeed all artificial FRNs, is not uniform, and notably the intensity of artificial global fallout in the Southern Hemisphere is lower than in the Northern Hemisphere, partly due to the greater number of above-ground bomb tests and nuclear accidents in the latter (UNSCEAR, 2000). Indeed,~77 % of total worldwide fallout of 137 Cs is estimated to have occurred in the Northern Hemisphere (UNSCEAR, 2000;Evrard et al., 2020). 210 Pb (t 1/2 = 22.3 years) is a natural product of the 238 U series, produced from the decay of gaseous 222 Rn. In terrestrial soils and sediments, total 210 Pb comprises a supported 210 Pb component, which is continuously generated in situ from the decay of 226 Ra, and an unsupported or excess 210 Pb component ( 210 Pb-ex), which is produced by radioactive decay of atmospheric 222 Rn. Both wet and dry deposition returns the 210 Pb-ex to the Earth's surface. Global distribution of 210 Pb is variable and dependent upon air mass sources, while also demonstrating some latitudinal variability (Baskaran, 2011). From a global dataset of atmospheric 210 Pb, Zhang et al. (2021) identify the highest concentrations in the mid-latitudes of the Northern Hemisphere, with higher depositional fluxes generally found in the Northern Hemisphere than the Southern Hemisphere due to its larger landmass and associated radon emanation rates (Baskaran, 2011).
To date, studies on the accumulation of FRNs in cryoconite have focussed on individual glaciers or regions. The objectives of this study are thus: 1) to provide the first global database of FRNs in cryoconite and assess geographical variability of activity concentrations; 2) to evaluate controls on the accumulation of natural and anthropogenic FRNs in cryoconite; and 3) to assess the extent to which FRNs accumulate more efficiently in cryoconite than proglacial sediments. Here we synthesize and evaluate both new and previously published data documenting the presence of FRNs in cryoconite and sediments from proglacial settings, and address the influence of key environmental variables on FRN activity concentrations. We also review current gaps in understanding of the downstream consequences of legacy contaminant release in glaciated catchments, and identify priority areas for future research in this field.

Study sites
The data presented here span five continents, including 32 individual sites -20 of which are previously unpublishedand a total of 520 individual cryoconite samples. We also include data from a further 209 sediment samples from the proglacial areas of 11 glaciers. The spatial coverage of these data provides insight into the variability of fallout radionuclide activity in cryoconite from icefield, ice cap, ice sheet and mountain glacier settings, spanning~70 to 5500 m a.s.l., and from 79 o N to 79 o S. These sites represent cool temperate, boreal, and polar temperature domains, and both moist and dry moisture domains ( Fig. 1; Sayre et al., 2020). Details of individual sites, and associated FRN data, are provided in Tables A1 and A2.

Sample collection and preparation
The cryoconite and proglacial sediments described here were sampled in the field between 1992 and 2020. Cryoconite sampling was conducted on the ablation zones of various glaciers, with the samples collected using plastic or stainless-steel tools (disposable pipettes, spatulas, or trowels) and stored in sterile plastic tubes. In some cases, the samples were preserved in 96 % ethyl alcohol. In the laboratory, the majority of samples were stored at -20°C and were subsequently evaporated and dried at 70-105°C. Most of the samples were fine-grained and sufficiently homogeneous that no further treatment (e.g. sieving) was required, and previous work has shown that particle size of cryoconite is less important for FRN accumulation than other physical variables such as organic content (Owens et al., 2019). Samples from Canada were sieved to <63 μm in the laboratory prior to analysis. Proglacial soils and sediments from Svalbard were collected using a stainless steel/PVC cylinder. Proglacial samples from Sweden and Iceland were collected using a plastic trowel, oven/freezedried, and sieved to <75 μm prior to analysis. A sediment core from a proglacial lake was also collected in Sweden using a HTH 90 mm diameter gravity corer and was extruded and sliced into 1 cm sections in the field, oven dried at 100°C, and sieved to <75 μm. Further details on specific sampling protocols for previously published data can be found in published sources (see Tables A1 and A2).
All FRN data reported here are based on samples collected by authors of this study, with the exception of cryoconite samples from the Hallstatter/ Schladminger and Stubacher Sonnblickkees glaciers which were collected and first reported by Tieber et al. (2009)  Scott Glacier catchment in Svalbard. Most of the data presented here were initially collected for studies of individual glaciers or regions, so it is important to acknowledge that there is variability in both sample collection and preparation methods. From our collective experience, we recommend the following for future cryoconite sampling and preparation for analysis of FRNs: that at least 6 g of cryoconite (dry weight) is collected per sample; that cryoconite is collected using plastic utensils and stored in sterile, water-tight plastic containers; that cryoconite is freeze-dried where possible; and that cryoconite is analysed in bulk unless containing obviously coarse particles or large granules, in which case it should be sieved to <63 μm.

Gamma spectrometry
Analysis of FRNs, supported by analytical quality assurance, was carried out by gamma spectrometry in three laboratories in Europe (University Milano-Bicocca, Italy; Institute of Nuclear Physics PAS, Poland; University of Plymouth, UK) and one in Canada (University of Manitoba, Canada). Table A3 provides further information on the analytical capabilities and quality assurance for these facilities. Gamma analysis using a well detector requires the dried cryoconite samples to be packed and sealed in 4 ml plastic/polypropylene vials, and the quantity of material for each of our samples varied between 0.5 g and 22 g. The sealed samples were incubated for a minimum of 24 days to allow establishment of radioactive equilibrium with 222 Rn and its progenies, prior to counting. Activity concentrations were determined using different detectors with a counting time of at least 24 h (up to 72 h for samples with lower weights). The isotopes 210 Pb (t ½ = 22.3 years), 241 Am (t ½ = 432 years), and 137 Cs (t ½ = 30.1 years) were determined by their gamma emissions at 46.52, 59.54, and 661.6 keV, respectively, for the majority of samples, with 241 Am in some samples from Svalbard, the Caucasus, and Ecology Glacier, Antarctica analysed by alpha spectrometry (Łokas et al., 2018). The data were verified by inter-laboratory comparisons with soil/sediment and reference materials (IAEA, Vienna, Austria; see Table A4 for exemplar quality assurance data). All values given here are correct to the date of sampling where sampling was conducted between 2015 and the present day, with other values decay-corrected to 2020 to optimise comparison between sites. We acknowledge that use of different detectors and run times introduces uncertainty when making direct comparisons, however the nature of this study and inclusion of previously published data precluded the use of a single, standardised approach. We also acknowledge that there is uncertainty around the extent to which cryoconite will adsorb/absorb and leach radionuclides, but decay-correcting values from prior to 2015 allows us to represent where radioactivity is likely to have decreased over longer timescales. We report 210 Pb-ex unless otherwise stated, which is estimated by subtracting 214 Pb from total 210 Pb (Zaborska et al., 2007).

Loss on ignition
A subset of 229 cryoconite samples from 11 regions were analysed for loss-on-ignition (LOI) to provide an estimate of the organic content of the samples. Ceramic or aluminium crucibles were preconditioned at 105°C overnight prior to analysis and were allowed to cool in a desiccator before having their weight recorded. All samples were dried at 105°C for a minimum of 2 h prior to being placed in a furnace. 1 g (± 0.9 g) of each sample of dried cryoconite was placed into a preconditioned crucible and the weight recorded. Each sample was heated at either 600°C or 1050°C for 1 h before being allowed to cool for 10 min and re-weighed. LOI was calculated, as a percentage, as follows: where W s is the crucible plus sample dried at 105°C (g), W A is the crucible plus ignited sample (g), and W c is the preconditioned crucible (g). The difference in heating temperature and crucible material used here reflects that samples were analysed for LOI in two different laboratories (University of Plymouth, UK; Institute of Nuclear Physics PAS, Poland).

Geographical variability in FRN activities in cryoconite
The sites included in our database span a range of glaciological settings and a wide range of climatic regimes, from the polar desert of the Dry Valleys of Antarctica to the moist, coastal region of Vatnajökull in SE Iceland. The highest 137 Cs activities for individual samples exceed 100,000 Bq kg −1 and are found in cryoconite from glaciers in the Eastern European Alps (Fig. 2; Table A1): the Hallstatter / Schladminger glaciers and Stubacher Sonnblickkees. Outside of the Alps, the highest value for individual samples (24,500 Bq kg −1 ) was found in cryoconite from Norway, with lower but still elevated activities (>4000 Bq kg −1 ) recorded for other sites in the Alps, Alaska, Sweden, and the Caucasus. The Eastern Alps and the Scandinavian Peninsula were among the regions most impacted by 137 Cs deposition following the 1986 Chernobyl event (European Commission et al., 2009;Meusburger et al., 2020), as reflected in their FRN activities. The lowest 137 Cs activities for individual cryoconite samples (<10 Bq kg −1 ) were found in samples from Iceland, the Chilean Andes, Alaska, and the Antarctic Dry Valleys, with the lowest maximum values also recorded in the Dry Valleys. The highest individual value of 210 Pb-ex was found in a sample from the Ladakh Himalaya, at >27,500 Bq kg −1 , while elevated values (>10,000 Bq kg −1 ) were also found in cryoconite from the Alps, Alaska, Sweden, Norway, Svalbard, and the Caucasus. The lowest values of 210 Pb-ex in individual samples (<100 Bq kg −1 ) were found in the Antarctic Dry Valleys, Iceland, the Chilean Andes, Alaska, and the Greenland Ice Sheet.
There is considerable inter-and intra-regional variability in 241 Am in cryoconite, which was not detected / reported for sites in the Dry Valleys, some sites in Svalbard, and one site in Greenland. Detector sensitivity and analytical conditions, such as sample size, run times, and backgrounds, preclude the detection of low activities of 241 Am in cryoconite above the MDAs (minimum detectable activities) at some sites, while the low activity concentrations reported for other sites are detectable but often return high uncertainties (Table A1). For sites where 241 Am is reported, the highest maximum values (>50 Bq kg −1 ) for individual samples were recorded in cryoconite from Svalbard, the Caucasus, Sweden, Norway, and the European Alps, with the highest value of 120 Bq kg −1 recorded for the Morteratsch Glacier in Switzerland.
The highest mean activities of 137 Cs in cryoconite (>2000 Bq kg −1 ) were found in the European Alps, the Caucasus, Sweden, and Norway, up to a maximum of~37,400 Bq kg −1 on the Hallstatter / Schladminger glaciers. Mean 137 Cs values at some sites in Greenland, Iceland, Antarctica, the Alps, and the Chilean Andes are up to three orders of magnitude lower at <100 Bq kg −1 (Figs. 2 and 3; Table A1). A similar spatial variability is also evident for 210 Pb-ex, with up to four orders of magnitude difference in mean activities between regions, e.g. <10 Bq kg −1 in the Antarctic Dry Valleys and~11,800 Bq kg −1 in Norway. The highest mean 241 Am values (>40 Bq kg −1 ) were reported for sites in the Austrian Alps and Norway, and the lowest (<2 Bq kg −1 ) for sites in Iceland, the Chilean Andes, and the Swiss Alps. Mean activities demonstrate clear regional patterns, however the activity ranges found for most glaciers are high and partly overlap between regions (Figs. 2 and 3). While we focus on inter-regional variability here, it has also been observed in the Alps that for a single glacier FRN activities in cryoconite can range by up to three orders of magnitude because of differing degrees of interaction between cryoconite deposits and meltwater transporting legacy nuclides (Baccolo et al., 2020a).

Controls on the variability of FRN activity
Inter-regional patterns in FRN activity are controlled by a combination of climatological, environmental, and biogeochemical factors. Comparison of 137 Cs and 210 Pb-ex activities in cryoconite with well-documented C.C. Clason et al. Science of the Total Environment 894 (2023) 164902 latitudinal global fallout proportions (UNSCEAR, 2000;Evrard et al., 2020) demonstrate a general accordance for sites in the Southern Hemisphere and the mid-latitudes (Fig. A1). However, a number of sites do not follow this trend, with disproportionately high values for some sites above 60°N, most notably certain sites in Norway, Sweden and Svalbard. Reported FRN activities in cryoconite are lower in the Southern Hemisphere, increasing from very low values in the arid Antarctic Dry Valleys and Andean sites to higher values in more humid areas of the Andes and King George Island. The disproportionately elevated values at high northerly latitudes may be explained in-part by the influence of long-range atmospheric transport during positive phases of the North Atlantic Oscillation, which transports contaminants to the Arctic from distal sources (Stohl, 2006). However, the large variation in activities for Northern Hemisphere sites illustrates that latitude alone cannot account for FRN variability. Similarly, we find that although proximity to major FRN sources such as Chernobyl may explain some of the most extreme values when comparing artificial radionuclides in cryoconite, we find no significant correlation between 137 Cs and distance from Chernobyl based on the Pearson's Correlation Coefficient (Fig. A2), and a moderate negative correlation for 241 Am and which deteriorates to no significant correlation when only Northern Hemisphere sites are considered. In comparison with activities from meta-analyses of radionuclides in soils and sediments globally (e.g. Evrard et al., 2020; Foucher  Table A1 for data. Note that logarithmic scales and different scale ranges are used due to the large variability in activity values. Data are presented in latitudinal order from north to south.  Fig. A1), we find that 137 Cs activities in cryoconite are up to three orders of magnitude higher, demonstrating the unique affinity between FRNs and cryoconite. We explored the correlation between FRN values and a range of other environmental variables (Fig. A2) to investigate the potential influence of factors such as precipitation, continentality, and elevation. We use distance from the sea as a proxy for continentality, and find no correlation with artificial radionuclides, but a moderate positive correlation with 210 Pb-ex. We also find a moderate positive correlation between 210 Pb-ex and mean elevation of sampling sites when Northern Hemisphere sites only are considered. No correlation was found between mean annual precipitation and 210 Pb-ex activity concentrations, however there is a moderate positive correlation between precipitation and both 137 Cs and 241 Am (Fig. A2), highlighting the importance of atmospheric contaminant scavenging by precipitation (Franz and Eisenreich, 1998;Herbert et al., 2006). Artificial FRN activities increase with annual mean precipitation from the lowest values in Antarctica (e.g. 137 Cs = 3 Bq kg −1 ; 214 Am = 0 Bq kg −1 for Taylor Glacier), to the high values reported for Scandinavian and Alpine glaciers, such as Blåisen Glacier ( 137 Cs = 18,000 Bq kg −1 ; 214 Am = 46 Bq kg −1 ) and the Hallstatter/Schladminger glaciers ( 137 Cs = 37,438 Bq kg −1 ; 214 Am = 40 Bq kg −1 ), with an exception of Iceland where the activities are among the lowest despite moderate precipitation (e.g. 137 Cs = 11 Bq kg −1 ; 214 Am = 1 Bq kg −1 at Virkisjökull). There is high variability of 210 Pb-ex in terms of relationship to precipitation, and the strong degree of continentality at Ladakh might explain the high 210 Pb-ex concentration in cryoconite at this site which is >1400 km from the sea. Unsupported 210 Pb is often more abundant in the atmosphere at continental sites (Rangarajan et al., 1986;Baskaran, 2011), although enrichment of air masses with 210 Pb-ex is also dependent on patterns of air mass circulation (Zhang et al., 2021). Principal component analysis (PCA) of standardised geographical and environmental variables was conducted to further explore variance within the data, revealing clustering of regions ( Fig. A3; Table A5). There are large positive loadings from mean 137 Cs, 241 Am and 210 Pb-ex for principal component 1, explaining 43 % of the data variance, with latitude, mean annual precipitation, and mean elevation exerting the strongest influence on principal component 2, explaining a further 20 % of the variance. The PCA further supports that no single environmental factor is the primary control on variability of FRN activities.
While not assessed here, variations in FRN activities between glaciers will likely reflect the different inventories of FRNs within snow and ice layers, in addition to the residence time of water within glaciers and their accumulation area ratios, further controlled by glacier hypsometry and mass balance. Intra-glacier variations in cryoconite activities also likely reflect their different histories of exposure to FRNs, with cryoconite in cryoconite holes capturing radionuclides from both ice and meltwater interactions (Baccolo et al., 2020a). Heterogeneity in melt rates both between glacier sites and within the same site could thus have ramifications for quantification of supraglacial FRN inventories, with samples retrieved from high melt areas near the margin versus samples from lower melt areas further up-glacier yielding different FRN accumulation rates. Supraglacial meltwater pathways will also likely act as a control on intra-glacier FRN variability by determining the extent to which meltwater interacts with cryoconite deposits. Supraglacial pathways can be relatively stable at seasonal or even multi-year timescales, but are likely to be dynamic at decadal scales, influencing patterns of FRN mobilisation, scavenging, and accumulation. Furthermore, near-surface hydrology and the weathering crust can play an important role in regulating supraglacial discharge (Stevens et al., 2018), and in areas like the Antarctic Dry Valleys water in cryoconite holes can be relatively isolated yet contribute a significant portion of total runoff (Fountain et al., 2004). While no study has yet attempted to predict the fluxes of FRNs to proglacial areas from the supraglacial environment and through mobilisation of cryoconite, future research might incorporate modelling of melt rates and hydrological pathways to improve understanding of cryoconite-meltwater interactions (Cook et al., 2016b).
A crucial factor controlling FRN accumulation in cryoconite is its ability to capture and retain radionuclides. The capacity of cryoconite to bind radionuclides partly depends on the extracellular polymeric substances (EPS) produced mainly by cyanobacteria (Pereira et al., 2011;d'Abzac et al., 2013), which have been demonstrated to be important for biogeochemical cycling (Nagar et al., 2021), and the very high FRN activities found in some samples indicate that this capacity is not easily saturated. Individual cryoconite granules are composed of recently accumulated biomass, although they may contain fragments of older granules that disintegrate after reaching a threshold size (Takeuchi et al., 2010). The granules sampled from a glacier thus represent a spectrum of life histories spanning different times of exposure to FRNs, reflecting highly variable radioactivity levels even on individual glaciers. Analysis of organic content via loss-on-ignition (LOI) for a subset of individual cryoconite samples in this study (Fig. 4) supports that both 137 Cs and 210 Pb-ex activities are positively correlated with LOI and significant above the 95 % confidence level. This relationship is strongly correlated when considering all sites, or only those in the Northern Hemisphere, and demonstrates a moderate, but still significant correlation when samples from Norway -which have particularly high organic content and 137 Cs concentrations -are excluded. The relative depletion of 137 Cs and 210 Pb in cryoconite for some sites might thus reflect a lack of older cryoconite material in high-ablation regions or highly dynamic supra-glacial environments, preventing prolonged periods of cryoconite-meltwater interaction and limiting accumulation of FRNs. When excluding Norwegian samples, the degree of correlation between FRN content and organic matter is higher for 210 Pb-ex than for 137 Cs, with a coefficient of determination exceeding 0.6 for 210 Pb-ex compared with 0.37 for 137 Cs. This might be explained by differences in the deposition mechanisms of the two radionuclides: continuous deposition for 210 Pb-ex, but event-related for 137 Cs. 137 Cs accumulation in cryoconite is also more influenced by local factors than 210 Pb-ex, whose deposition is more uniform both temporally and spatially (Persson and Holm, 2011). Moreover, it has been shown that the degree of affinity for organic matter is higher for 210 Pb-ex than for 137 Cs (Teramage et al., 2013), which is compatible with our data.
Contemporary volcanic activity in Iceland and Chile could also affect the geochemical composition of cryoconite in these regions, in turn affecting its ability to accumulate FRNs, however further research is required to assess spatial variability in cryoconite geochemistry and the role of mineralogy, which has been shown to play a role in 137 Cs sorption in other environments (e.g. Nakao et al., 2015). The high clay content of cryoconite may also play a role in accumulation of 137 Cs, as demonstrated for soils in Japan following the Fukushima accident (Fujii et al., 2014;Baccolo et al., 2017), although some studies have shown that the particle size composition of cryoconite is less important than organic matter content (Owens et al., 2019). The controls on FRN accumulation at individual glacier sites most likely vary considerably between regions, reflecting a combination of intrinsic and extrinsic factors, such that while proximity to FRN sources may override other factors at some sites, for other sites precipitation amount, glacial melt rate, or cryoconite organic matter composition and geochemistry may play a much more central role.

Transport and accumulation of FRNs in glaciated catchments
In addition to considering the role of cryoconite in scavenging FRNs on the surface of glaciers, it is important to consider the transport and release of FRNs in cryoconite, and implications for downstream environments. Hydro-sedimentological processes are key for mobilising cryoconite and associated FRNs into the proglacial environment in response to both longer-term patterns of glacier melt and event-scale increases in melt and precipitation, in addition to erosion and re-mobilisation of sediments within proglacial areas (e.g. Leggat et al., 2015). Numerous studies have reported that glaciers are retreating rapidly, with a total mass loss of 267 ± 16 Gt per year between 2000 and 2019 (Hugonnet et al., 2021), exposing new land to subaerial processes. Few studies have determined FRN activities in the soils and sediments of proglacial areas, which can also be helpful tracers of glacier retreat (Navas et al., 2017(Navas et al., , 2020Lizaga et al., 2019), and in most cases concentrations are considerably lower than those recorded in cryoconite (e.g. Hasholt et al., 2000; Table A2). Comparison of cryoconite with proglacial material for selected sites (Fig. 5) demonstrates the ability of cryoconite to efficiently accumulate FRNs in comparison to proglacial sediments, with up to three orders of magnitude between activities in cryoconite and proglacial sediments for some sites. A study in British Columbia, Canada (Owens et al., 2019) documented declining FRN concentrations in fluvial sediments with increasing distance from the glacier terminus, while studies in Arctic Sweden (Clason et al., 2021) and Svalbard (Łokas et al., 2017) found that some radionuclides were higher in sediments from southern proglacial outlets than those from northern outlets, highlighting the importance of local variability in glacier topography and hydrology.
The lower activities found in proglacial sediments may be due to the dilution of cryoconite released from the glacier with sediment from other sources, such as eroding channel banks, hillslopes, and subglacial substrates, which typically contain much lower concentrations of FRNs (e.g. Evrard et al., 2020;Foucher et al., 2021). However, the activity of FRNs detected in soils and sediments in depositional environments can be relatively high, suggesting that downstream accumulation of FRNs, or localised off-ice 'hotspots' is possible for some sites (Łokas et al., 2017;Clason et al., 2021). Much of the total inventory of FRNs stored in and on glaciers is likely to be legacy in nature, having been deposited in the years following nuclear events. Given that it can take decades or longer for material to be transported from the accumulation zone to the termini of glaciers, we can expect the release of historical contaminants to increase under current and future warming. Inventories of FRNs in glacier ice remain very poorly constrained, however, so better understanding of contaminant load should be a focus for future studies to both contribute to environmental risk assessment and improve knowledge of FRN inventories in polar and mountain regions.

Downstream implications of legacy contaminant release
Here we have demonstrated the prevalence of high activities of FRNs in cryoconite throughout the global cryosphere. This highlights the importance of cryoconite within glacial systems for the accumulation of atmospherically deposited materials, including contaminants, and the role played by these unique biomes in biogeochemical transfers and cycling. What remains poorly understood are the consequences of legacy contaminant release downstream, enhanced by continued melting and downwasting of glaciers, for both ecosystem health and socio-economic impacts (Fig. 6), with very few studies exploring both supraglacial and proglacial FRN activities to date (e.g. Owens et al., 2019Owens et al., , 2023Clason et al., 2021). The behaviour and retention of FRNs in the environment, particularly in soils, depends on local geochemistry and organic content, including the inhabiting microbial community. Factors such as organic matter content, pH, and permeability of soils, in addition to reworking of proglacial deposits (Łokas et al., 2017), will affect the migration of FRNs once they reach the proglacial environment (Walther and Gupta, 2015). Proglacial soils are usually low in organic carbon (Bradley et al., 2014) with a low sorption capacity, which may result in relatively high mobility of FRNs in some regions. However, this mobility is limited in regions such as the High Arctic and Antarctica due to low bioturbation and long frost periods (Łokas et al., 2014). Further dispersal depends on the soil's microorganisms, which play a major role in controlling the cycling and toxicity of FRNs by direct reduction, uptake, and accumulation, as well as turnover of organic matter (Simonoff et al., 2007). The microbial community structure in glacier forefields differs significantly from the communities of cryoconite holes (Edwards et al., 2013) which could have a direct impact on the fate of FRNs released from glaciers. The freshly exposed soils in front of receding glaciers also undergo microbial community succession, gradually increasing in evenness, phylogenetic diversity, and number of phylotypes (Franzetti et al., 2020). FRN contamination has been shown to impact the microbial community structure and functionality of soil ecosystems in non-glaciated environments (Rogiers et al., 2021) which could negatively impact the establishment of soil communities in rapidly exposed soils. Data on the impact of FRN activity on proglacial soils and adjacent communities remain scarce, and more research is necessary to predict the downstream fate and toxicity of FRNs in these environments. In addition to affecting soil microbial communities, the introduction of FRNs within downstream environments has the potential to influence the health and fertility of flora and fauna. Vegetation in glaciated areas tends to have less developed root systems as an adaption to frozen ground. It therefore relies on absorbing nutrients atmospherically and aquatically (Meyers, 2012), and is particularly susceptible to the accumulation of FRNs. Furthermore, biomagnification through higher trophic levels of grazing fauna can lead to even higher concentrations of FRNs (Hong et al., 2011;Stocki et al., 2016), with potential socio-economic and health implications for communities that rely on livestock for food and livelihoods. These populations are often indigenous communities who may have less access to standardised environmental quality testing (Stensrud, 2016). The impact of consuming local fauna, and by proxy the contaminated fauna and flora they have digested, as well as uptake from contaminated and eco-toxic water sources, could pose issues for both animal and human health (Donaldson et al., 2016). However, the extent to which the release of FRNs from glaciers poses a risk, if at all, is currently unknown, and there are notable research gaps around the processes involved in the accumulation and secondary deposition of FRNs downstream of glaciers, and associated effects on flora, fauna, human health and socio-cultural wellbeing. In addition to improving understanding of contaminant accumulation in cryoconite in under-represented regions of the cryosphere, these gaps should be addressed to assess and mitigate potential future risk to communities living within and downstream of glaciated environments.

Conclusions
Our global synthesis of FRN activities in 520 cryoconite samples from 32 different glacier environments shows, for the first time, that the accumulation of both artificial and natural FRNs in cryoconite is a commonplace and universal process, with FRN accumulation found in both polar and mountain glacier settings. Activities of 137 Cs, 241 Am, and 210 Pb in cryoconite are highly variable, both within individual glaciers and inter-regionally, but the activities we report are consistently, and significantly, above values typically reported in other environmental matrices such as soils and sediments. Indeed, observed activity concentrations in cryoconite can exceed common environmental values by up to 3 or even 4 orders of magnitude, with some activities for single FRNs exceeding 10,000 Bq kg −1 . Intra-glacier variability of FRN activity in cryoconite likely reflects differences in interaction between cryoconite and supraglacial meltwater, through which legacy contaminants stored in snow and ice are released and mobilized in the glaciohydrological environment. Between regions, the considerable variability in FRN activities is controlled by a combination of climatological, environmental, and biogeochemical factors, with a notable strong correlation between cryoconite organic content and radioactivity. The extent to which the release of FRNs from glaciers poses a downstream ecological threat is uncertain, but improved understanding of the behaviour of legacy contaminants within glacier catchments should be a future research priority to contribute to future environmental protection in ecologically vulnerable polar and mountain regions.
Our compilation and synthesis of a unique database of radionuclides in cryoconite samples was obtained using samples from a wide range of international sites analysed by non-systematic sample manipulation and instrumental approaches. Future progress in the understanding, and environmental significance, of FRN accumulation in cryoconite is currently hindered by a lack of analytical unity with respect to the coherence of data arising from a range of standard operating procedures. For example, various run times are used and, in some cases, but not all, radiological data may be supported by suitable Quality Control (QC) procedures. Therefore, in order to enhance the international comparability of future cryoconite datasets we propose the adoption of a standard operating procedure for FRN measurements in cryoconite, including sample preparation, run times, and QC, to ensure the continuity and quality of this database.
CRediT authorship contribution statement CCC devised and led this collaborative study. RA, EB, GB, CCC, JC, AF, RF, DL, EL, PNO, EP, FP NS, NT and PW conducted field sampling, GB, DB, CCC, EL, GEM, MN, PNO, MS and AT conducted sample analysis, and GB, CCC, EL, PNO and PW led interpretation of the collective data. All authors contributed to preparation of the manuscript.

Declaration of competing interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Caroline Clason reports financial support was provided by INTERACT. Caroline Clason reports financial support was provided by Quaternary Research Association. Ewa Poniecka reports financial support was provided by Natural Environment Research Council. Ewa Poniecka reports travel was provided by National Science Foundation. Philip Owens reports financial support was provided by Natural Sciences and Engineering Research Council of Canada. Philip Owens reports financial support was provided by Forest Renewal British Columbia. Edyta Lokas reports financial support was provided by National Science Centre Poland.