Cryoconite: an efficient accumulator of radioactive fallout in glacial environments

. Cryoconite is rich in natural and artificial radioactivity, but a discussion about its ability to accumulate 20 radionuclides is lacking. A characterization of cryoconite from two Alpine glaciers is here presented. Results confirm that cryoconite is significantly more radioactive than the matrices usually adopted for the environmental monitoring of radioactivity, such as lichens and mosses, with activity concentrations exceeding 10,000 Bq kg -1 for single radionuclides. This makes cryoconite an ideal matrix to investigate the deposition and occurrence of radioactive species in glacial environments. In addition, cryoconite can be used to track


Introduction
Radioecological research is primarily focused on Earth surface, where continuous atmospheric deposition of fallout radionuclides (FRN), both natural and artificial, are accumulated. The most common FRNs are cosmogenic nuclides, 222 Rn progeny and artificial products. The last ones of which have been released into the environment since the second half of the twentieth century, as a consequence of nuclear test explosions and accidents. Hundreds of thousands of PBq were spread in 40 the high troposphere and stratosphere from the 1940s to the 1960s, allowing for global dispersion and contamination at the Earth surface. The extent and impact of FRN deposition on the Earth surface is monitored through the analysis of different environmental matrices, which are used to reconstruct FRN deposition (Steinhauser et al., 2013) and understand their environmental mobility and distribution (Avery et al., 1996;Yasunari et al., 2011). Among the matrices used in the study of FRNs, those that receive the greatest attention share common features: their composition is directly influenced by airborne 45 deposited impurities; the contribution from environmental compartments other than atmosphere is limited; they are widespread, accessible and preferably easy to sample. Given these attributes, lichens, mosses and peat are commonly used to study the distribution of FRNs and establish depositional inventories (Nifontova, 1995;Kirchner and Daillant, 2002).
In recent years cryoconite began drawing attention in the field of radioactive monitoring as an alternative environmental matrix. Cryoconite is the dark, unconsolidated sediment that is found on the surface of glaciers worldwide (Takeuchi et al., 50 2001). It forms exclusively at the ice-atmosphere interface and in presence of meltwater. It can be found as a dispersed material on the ice surface or as a deposit accumulated on the bottom of characteristic water-filled holes melted into ice, usually aggregated into granules (cryoconite holes, see Fig. 1). Cryoconite forms out of the interaction between the mineral particles present on the ice surface (both allochthonous and autochthonous), and the complex microbial communities that develop on the surface of glaciers (Cook et al., 2015). Among the microbes that are present on glaciers, cyanobacteria play a 55 major structural role (Langford et al., 2010). During the ablation season, when liquid water is available at the surface of glaciers, cyanobacteria develop films and filaments that promote the formation of aggregates composed of mineral sediments and organic matter, resulting in cryoconite granules (Takeuchi et al., 2001). The composition of cryoconite is dominated by a mineral component accounting for 85-95 % of its mass, whereas the remnant fraction is comprised of both living and dead organic matter and is responsible for its dark colour (Cook et al., 2015). The formation of cryoconite holes is attributable to 60 the dark colour, and thus low albedo, of cryoconite, which enhances the absorption of solar radiation and locally increases ice melting to foster the development of holes in the ice surface. Due to its contribution to ice melting, its diverse composition, and the role in biodiversity, cryoconite has been studied by a range of disciplines, including glaciology, microbiology, biogeochemistry and ecology (Takeuchi et al., 2001;Langford et al., 2010;Cook et al., 2015;Ferrario et al.,  Similar observations were successively reported by Cota and co-authors (2006) who founded unexpectedly high radioactivity burdens (related to 137 Cs, 210 Pb and 7 Be) in cryoconite from the surface of multi-year sea ice in the Canadian Arctic. Since then no other studies (to our knowledge) have focused on the radioactivity of cryoconite until the presentation of an extensive characterization of samples from an Austrian glacier in 2009 (Tieber et al., 2009). That study has showed that 80 cryoconite is contaminated not only by 137 Cs, a common artificial radionuclide spread into the environment, but also by several other species, artificial and natural in origin. The reported activity levels in this study are extremely high, and comparable to soil samples from nuclear incident and explosion sites (Steinhauser et al., 2014;Abella et al., 2019).
Subsequent studies, carried out in different regions of global cryosphere (Alps, Caucasus, Svalbard and Canada), corroborated the ability of cryoconite to accumulate radionuclides Łokas et al., 2016Łokas et al., , 2018Łokas et al., , Owens et 85 al., 2019, with radioecological consequences concerning the presence of FRNs that extend to the pro-glacial areas (Łokas et al., 2017(Łokas et al., , Owens et al., 2019. In addition to FRNs, cryoconite has also been shown to accumulate other anthropogenic contaminants including heavy metals (Nagatsuka et al., 2010;Łokas et al., 2016;Baccolo et al., 2017;Singh et al., 2017, Huang et al., 2019, artificial organic compounds (Ferrario et al., 2017;Weiland-Bräuer et al., 2017), and microplastics (Ambrosini et al., 2019). However, among the species found in cryoconite, radionuclides show by far the highest 90 concentration with respect to other environmental matrices. The considerable activity concentrations of artificial FRNs found in cryoconite, has allowed for the application of cutting-edge radiological techniques, offering important insight to the sources and distribution of radioactivity deposited within the glaciated regions of the Earth (Łokas et al., 2018, 2019), and making novel contributions to the emerging field of environmental nuclear forensics (Steinhauser, 2019).
Even though the link between cryoconite and environmental radioactivity is now indisputable, several important questions 95 remain. Firstly, is not clear why and how cryoconite accumulates radionuclides and other anthropogenic impurities and to what extent this accumulation is related to processes specific to glacial environments. Previous studies (Osburn Jr., 1963;Pourcelot et al., 2003) have focused on the role of snowmelt in accumulating radionuclides in residual snow patches during summer, suggesting that nival and melt processes could encourage local accumulation of radioactivity in a similar fashion to what has been observed in cryoconite on glaciers. Secondly, it is not fully understood where this radioactivity comes from. 100 This paper aims to present cryoconite as a promising tool for radioecological monitoring in high latitude and high-altitude areas and sheds light on some of the open issues related to the themes explored above. Data concerning cryoconite from two Alpine glaciers are presented and compared to data from previous studies on both cryoconite and other environmental matrices used for radioactive monitoring.

Study site and sampling strategy 105
The Morteratsch and Forni Glaciers are located in the European Alps and are situated ~50 km apart ( Fig. ). Both glaciers are among the biggest and most studied in the Alps Di Mauro et al., 2019). The Morteratsch is a Swiss glacier and the largest in the Bernina range, spanning an altitudinal range of roughly 2000 m and with a terminus located at 2100 m a.s.l.. It has an area of ~7.5 km 2 , but until 2015 the value exceeded 15 km 2 due to its connection with a tributary glacier that has since detached. A similar setting characterizes the Forni Glacier, the biggest one in the Ortles-Cevedale range 110 in Italy. Forni is a north-facing valley glacier presenting a glacial tongue that was, until recently, fed by three accumulation basins. Owing to retreat of the glacier, the connections between the basins have become weaker and the glacier is now fragmented in two parts . The Forni glacier ranges between 2500 to 3750 m a.s.l. and is characterized by a surface area of ~10 km 2 considering both ice bodies. From a climatic perspective, the glaciers are similar; they are characterized by a continental climate and in the last years experienced a darkening of their ablation zones because of the 115 accumulation of impurities on the surface and the progressive emergence of detritus from medial and lateral moraines (Di Fugazza et al., 2019). The abundance of supra-glacial debris is favourable for the formation of cryoconite, and in fact the two glaciers are well studied in terms of cryoconite and its components Di Mauro et al., 2017;Pittino et al., 2018a;Ambrosini et al., 2019;Fugazza et al., 2019). We have considered these glaciers also because of the relatively easy access, which makes them ideal sites for annual sampling campaigns. 120 Samples considered here were collected in summer 2015 (July and September) and 2016 (July) from the ablation zones of the glaciers (Fig. 2). Each sample represents a distinct cryoconite hole. The sampling was carried out using sterile disposable 130 pipettes and ethanol cleaned spoons; samples were kept in sterile tubes at 0°C during the field campaign and stored at -20°C in the EUROCOLD laboratory of the University Milano-Bicocca, until preparation for the geochemical analyses. We selected the most abundant cryoconite deposits, so as to have material available for other analyses also: twelve samples have been gathered on the Morteratsch Glacier (between 2100 and 2300 m a.s.l.) and ten on the Forni Glacier (between 2600 and 2800 m a.s.l.), each one consisting in 10-40 g of wet cryoconite. Where possible, multiple analyses have been carried out on 135 the same samples, but this has not been always possible. Part of the data concerning gamma spectrometry applied to cryoconite from the Morteratsch samples has already been published .

Radioactivity measurements
The activity of natural and artificial radionuclides in cryoconite, has been measured using a number of techniques. 137 Cs, 140 241 Am, 207 Bi, 40 K and 238 U and 232 Th decay chain nuclides were analysed through γ-spectrometry about six months after sampling, full details are found in the supplementary material. Aliquots dedicated to γ-spectrometry consist in ~1 g of dry material (dried until constant weight at 50°C, 2 mm sieved) sealed in polyethylene vials. The acquisition of the γ-spectra took place at least two weeks after the sealing, to allow the secular equilibrium between 222 Rn and its progenies to be attained. Each sample have been counted for one week using a customized high purity germanium well detector (Ortec). 145 Details about the instrument, calibration and analytical performances are presented elsewhere .
Aliquots dedicated to Pu analyses (~1 g of dry material) have been ashed at 600° C to remove organic matter. The ash has been dissolved with mineral acids and the resultant liquid samples have undergone radiochemical separation and concentration of Pu. The procedure is extensively described elsewhere by Łokas and colleagues (2016). Activities of 239+240 Pu and 238 Pu have been determined through α-spectrometry after Pu co-precipitation with NdF3, using Canberra 7401 150 and Ortec Alpha Duo spectrometers. After further radiochemical purification, the 240 Pu/ 239 Pu atomic ratio was measured through MC-ICP-MS (Thermo Fisher Scientific Neptune spectrometer), in accordance with Łokas et al. (2018). 238 U and 232 Th activities have not been directly measured but have been estimated considering the total content of U and Th.

Instrumental neutron activation analysis
The Th and U composition of cryoconite has been assessed through instrumental neutron activation. Samples have been 155 irradiated at the LENA laboratory, where a TRIGA Mark II nuclear reactor is available. The irradiation have lasted for six hours under a thermal neutron flux of 2.4 ± 0.2 · 10 12 neutron s -1 cm -2 . To determine the concentration of Th and U in the samples, the following nuclear reactions and γ-emissions have been exploited: 232 Th (n,γ) 233 Th → 233 Pa (analyzed emissions: 300.3 and 312.2 keV) and 238 U (n,γ) 239 U → 239 Np (analyzed emissions: 228.2 and 277.6 keV). Irradiated sample have been counted for six hours a few days after the irradiation, using the same well detector applied for γ-spectrometry. The 160 quantification of concentrations have been carried out comparing samples and reference materials .

Carbonaceous content
A thermo-optical analyzer (Sunset Lab Inc. analyzer) have been used for the determination of organic and elemental carbon content (OC and EC respectively), following the protocol adopted in Baccolo et al. (2017). Cryoconite samples have been suspended on clean quartz fiber filters and analyzed. The mass concentration of OC and EC have been obtained combining 165 the information relative to filter superficial concentrations and the mass of cryoconite deposited on the latter, determined using an analytical microbalance (precision 1 µg) which has been operated inside an air-conditioned room (T = 20 ± 1 °C; relative humidity = 50 ± 5 %). Mass concentration of OC has been converted into organic matter content (Pribyl, 2010).

Statistics
To evaluate the degree of correlation between variables and samples, two multivariate statistical tools have been applied. 170 Multidimensional scaling (MDS) has been used to appreciate the degree of correlation between the radionuclides (Diaconis et al., 2008). MDS has been applied to a similarity metric derived from the correlation matrix (Pearson's correlation coefficient) of the original data, following Eq. 1, where the distance d between variables v1 and v2 is obtained considering their Pearson correlation coefficient (r) (van Dongen and Enright, 2012).

Eq. 1
The correlation between samples and the differences between the two glaciers have been evaluated applying the principal component method to standardized data (PCA). The first two components (which explain 65 % of the total variance) have been taken into consideration.

Cryoconite natural radioactivity
Our results confirm the ability of cryoconite to accumulate radioactivity and in particular FRNs. In fact only FRNs, whose environmental occurrence is related to atmospheric transport, are actually accumulated in cryoconite from the Morteratsch and Forni glaciers, not the lithogenic ones. This can be observed in Fig. 3, where the activity of lithogenic radionuclides is presented. A substantial secular equilibrium characterizes the nuclide belonging to the 238 U and 232 Th decay chains, except 185 for 210 Pb (t1/ 2 = 22.3 yr), which presents an excess with respect to the other 238 U-related nuclides. Excluding it, the average 238 U and 232 Th chain activities are 70 ± 15 and 52 ± 8 Bq kg -1 respectively for the 238 U chain (Morteratsch and Forni samples, ± standard deviation) and 50 ± 10 and 55 ± 10 Bq kg -1 for the 232 Th chain. These values, as seen in Fig. 3, are slightly higher than the average 238 U and 232 Th radioactivity of upper continental crust (UCC) reference (Rudnick and Gao, 2003), which is 34 and 43 Bq kg -1 for 238 U and 232 Th respectively. The difference is probably related to the accumulation in cryoconite of 190 heavy minerals, where U and Th are typically enriched, because of hydraulic sorting related to meltwater flow ). The activity of the primordial radioactive nuclide 40 K (t1/2 = 1.28·10 9 yr) in the samples from the Morteratsch and Forni glaciers (810 ± 55, 770 ± 200 Bq kg -1 ) is of the same order of magnitude of 40 K UCC activity, i.e. 720 Bq kg -1 (Rudnick and Gao, 2003). Such results point to a typical crustal origin for the lithogenic radionuclides measured in cryoconite. An exception to this is 210 Pb, which, although being a decay product of 238 U progeny, shows activity levels two 195 orders of magnitude higher than the other 238 U-chain nuclides. The average activities in the samples from the Morteratsch and Forni glaciers are 2,800 ± 800 and 6,200 ± 1,900 Bq kg -1 respectively and are statistically different (Student's t test: t20 = 5.9; p-value < 0.001) within the two glaciers. Finding such high 210 Pb activities in samples collected on the surface of glaciers is not completely unexpected. It is common to observe an excess of 210 Pb in Earth surface environments, due to its dual source. A fraction of 210 Pb is present in materials of geologic origin because of the internal decay of 238 U progeny 205 (supported 210 Pb); a second fraction (unsupported 210 Pb) is found in samples exposed to the atmosphere and is attributable to the scavenging by precipitation of atmospheric 210 Pb, produced from the decay of the gaseous 222 Rn released into the atmosphere from rocks and soils. Given its relatively long half-life (22.3 yr), precipitated 210 Pb concentrates in surficial environments, but typically its activity doesn't exceed tens or a few hundreds of Bq kg -1 in matrices strongly influenced by atmospheric deposition and rich in organic matter, for which Pb is particularly affine (Strawn and Spark, 2000).

9
In Fig. 4, cryoconite radioactivity is compared to data concerning other environmental matrices. With respect to lichens and mosses, which are known to be efficient in accumulating radioactive atmospheric species (Kirchner and Daillant, 2002), cryoconite shows a 210 Pb activity that is, on average, higher by one order of magnitude. Two hypotheses are made to explain the excess found in cryoconite: 1) the glaciers considered here are located in areas where the atmospheric deposition of 210 Pb is enhanced; 2) cryoconite is more efficient at concentrating atmospherically derived radionuclides than lichens and mosses. 215 At the Morteratsch Glacier a comparison has been made between cryoconite and samples collected from the surface of the moraines surrounding the glacier . The moraine sediments have had a mean activity of 145 ± 30 Bq kg -1 for unsupported 210 Pb, while in cryoconite it has exceeded 2,500 Bq kg -1 . This evidence rejects the first hypothesis: if an anomaly of atmospheric 210 Pb deposition was present in the Morteratsch valley, it should impact both the the moraine and glacial surfaces. Several studies have reported high unsupported 210 Pb activity in cryoconite from different regions of the 220 Earth (Tieber et al., 2009;Baccolo et al., 2017;Łokas et al., 2016, suggesting that high 210 Pb activity is related to specific characteristics of cryoconite.

Anthropogenic radioactivity in cryoconite and other environmental matrices
In Fig. 4, the comparison between radioactive contamination of worldwide lichens, mosses, soils and sediments from other studies (full information in the supplementary material), is extended to all of the radionuclides that have been found in 225 excess in cryoconite in this study. Out of all of them, 210 Pb is the only natural occurring species while the others are anthropogenic in origin. In descending order of average activity in cryoconite, they are: 137 Cs (t1/2 = 30.1 yr), 239+240 Pu (t1/2 = 24,110 and 6,536 yr respectively), 241 Am (t1/2 = 432.2 yr), 207 Bi (t1/2 = 31.6 yr), 238 Pu (t1/2 = 87.7 yr). These nuclides have been released into the environment as a consequence of nuclear incidents and explosions and have been atmospherically transported and deposited globally. For all radionuclides, the activities measured in cryoconite are always higher than those 230 of other environmental matrices (see Fig. 4). To find samples with activities comparable to the ones found in cryoconite, it would be necessary to consider sites within the vicinity of nuclear tests or incidents. The mean ratios between the activity levels found in cryoconite and in lichens for 137 Cs, 239+240 Pu, 241 Am, 207 Bi, 238 Pu are 9.5, 58, 39, 35, 7 respectively, and the values are even higher when matrices less efficient in accumulating radionuclides are considered. This supports the hypothesis that cryoconite accumulates atmospherically derived artificial radionuclides more efficiently than other matrices, 235 as already suggested by exploring unsupported 210 Pb.
The accumulation ability of cryoconite can be observed not only for common artificial radionuclides, such as 137 Cs and 239,240 Pu, but also for less abundant species, such as 241 Am, 207 Bi and 238 Pu. 137 Cs is among the most common long-lived fission products from 235 U and has been released in the environment due to commercial reactor failures and fission bomb test explosions. The plutonium isotopes 239 and 240 also originate from atmospheric weapon tests and nuclear accidents, but the 240 relative contribution from atmospheric tests is larger, since 239 Pu has been the most common fissile material used in fission bombs and for igniting fusion devices. Because of their widespread dispersion, 137 Cs and 239,240 Pu are the most abundant artificial nuclides found in cryoconite from the two glaciers, confirming previous results (Tieber et al., 2009; and Forni cryoconite respectively). Less abundant nuclides are present in cryoconite with lower concentration, including 241 Am (30 ± 35 and 4 ± 1.5 Bq kg -1 ), 207 Bi (9 ± 7 and 6 ± 2 Bq kg -1 ) and 238 Pu (2.5 ± 2.5 and 0.22 ± 0.08 Bq kg -1 ). Despite being low, such activities are still significant and among the highest ever found in the environment. Typical environmental 255 activities usually do not exceed 1 Bq kg -1 for 241 Am and 207 Bi, and 0.1 Bq kg -1 for 238 Pu (Fig. 4), being their rarity related to their production mechanisms (Shabana & Al-Shammari, 2001;Bossew et al., 2006). The presence of 241 Am in the environment is not primarily related to direct deposition (it is present in nuclear power plant spent fuel); it is mostly produced in situ, from the decay of its parent nuclide ( 241 Pu, t1/2 = 14.3 yr), which has been released into the environment alongside other Pu isotopes. Thanks to 241 Pu decay, the environmental activity of 241 Am globally is increasing and will peak 260 around year 2100 (Thakur and Ward, 2018). 238 Pu is one of the rarest plutonium isotopes produced by commercial reactors and nuclear explosions, and its diffusion is mostly related to the atmospheric re-entry of satellites powered by pure 238 Pu thermoelectric generators (Łokas et al., 2019) and to a smaller degree by the release from nuclear fuel reprocessing plants into marine environment (Bryan et al., 2008). 207 Bi has been released as a consequence of a few high yield thermonuclear explosion tests (Noshkin et al., 2001) and has rarely been observed within the environment. Finding easily detectable 265 activities in cryoconite for these rare radionuclides, many of which were released decades ago, is both surprising and unprecedented. Studies focused on them usually require the application of pre-concentration and separation procedures, but for cryoconite a direct measure of activity was sufficient. These results highlight the potential of this environmental matrix for radioecological monitoring.
Looking in detail at the two Alpine glaciers considered here, only the activity of Pu isotopes is significantly different 270 between the two sites, with higher values found in the samples from the Morteratsch Glacier (Student's t test: t12 = 2.99; pvalue = 0.010 for both 238 Pu and 239+240 Pu). 241 Am is also more abundant in Morteratsch cryoconite, but not significantly because of the large standard deviation (Student's t test: t20 = 2.24; p-value = 0.018). The ratios between the mean activity of the Morteratsch and Forni samples are 15.9, 11.6 and 6.8 for 239+240 Pu, 238 Pu and 241 Am respectively.

Sources of anthropogenic radioactivity in cryoconite 275
To infer the potential sources of the radioactivity found in cryoconite from Alpine glaciers, isotopic and activity ratios between Pu and Cs isotopes have been calculated (Fig. 5). The use of such ratios has been to estimate the provenance of environmental radioactivity, since specific signatures are associated to different sources (Steinhauser, 2019). The atomic ratio 240 Pu/ 239 Pu and activity ratio 238 Pu/ 239+240 Pu show that the plutonium-related radioactivity of Morteratsch and Forni cryoconite is compatible with the worldwide signal from global radioactive fallout ( Tab. 1 and Fig. 5a,b). The latter reflects 280 the composition of the stratospheric reservoir, established in the '60s as a consequence of atmospheric nuclear weapon testing. On average, more than 99 % of the Pu found in cryoconite from the Morteratsch glacier is from global fallout, while for the Forni glacier the average contribution is 95 %, suggesting a non-negligible influence from the Chernobyl accident (~5 %).

Łokas et al., 2019
Svalbard 93±4 56±25 By comparing the 240 Pu/ 239 Pu ratio of global fallout and of modern snow deposited in the Alps (Gückel et al., 2017), it is possible to further discuss the Pu sources in cryoconite. Modern Alpine snow has a slightly higher ratio than global fallout (0.21 vs. 0.18), probably because of the partial influence of re-suspended Chernobyl radioactive fallout, which is more 290 enriched in 240 Pu than global fallout (Ketterer and Szechenyi, 2008). Only two cryoconite samples (one from Forni and one from Morteratsch) show a Pu isotopic composition pointing to the Chernobyl influence. They show a ratio of 0.286 ± 0.006 and 0.24 ± 0.02 respectively, which is even higher than the that of modern Alpine snow. The occurrence of only two samples with a partial Chernobyl signature can be explained by the presence of fallout particles from Chernobyl nuclear fuel in these specimens. The non-volatile constituents of nuclear fuel, such as Pu, were not scattered into the environment homogenously, 295 as in the case of the more volatile 137 Cs, but as micrometric and highly radioactive particles (Sandalls et al., 1993). The presence of even one of such particles in the two samples could be sufficient to explain the anomalies. The other samples present a global signature fully compatible with global fallout and not with modern Alpine snow. This implies that Pu accumulated in Alpine cryoconite dates back to when the deposition of radionuclides was dominated by global stratospheric  Fig. 2. fallout, approximately from 1960 to 1980 (Hirose et al., 2008). The deposition of plutonium which is still occurring on the Alpine snowpack, is too weak to influence the isotopic fingerprint of cryoconite. Pu activity in fresh snow ranges from 0.4 to 310 fallout rather than the contemporary one, at least when considering Pu. The only source that can provide to cryoconite FRNs from past atmospheric deposition is ice accumulated within the period of maximum deposition of atmospheric radioactivity, when the latter was dominated by global stratospheric fallout. The presence of 207 Bi in cryoconite also supports this 315 hypothesis. In the northern hemisphere 207 Bi was produced during the explosion of the Tzar thermonuclear device in 1961 in Novaja Zemlya (Aarkrog and Dahlgaard, 1984). A few years after this event the 207 Bi atmospheric contamination decreased until reaching non-detectable levels (Kim et al., 1997). If a considerable amount of 207 Bi is present in cryoconite, it means that the cryoconite has had the possibility to interact with ice deposited shortly after 1961.

Tab. 1 Data about the fraction of Pu and 137 Cs related to global fallout in cryoconite samples from the Morteratsch and Forni
Comparing our results with the data obtained for cryoconite collected in the Caucasus and in regions of the Arctic (Łokas et 320 al., 2018, 2019), it is possible to see that there are variations in the radioactive signatures, pointing to secondary regional influences, despite the general features are compatible with global stratospheric fallout (Fig. 5). Caucasian and Arctic samples are characterized by a lower 240 Pu/ 239 Pu atomic ratio than the Alpine samples ( Fig. 5a and b). Such a signature is compatible with the influence of weapon grade Pu, depleted in 240 Pu (Cagno et al., 2014). It has been argued that samples from the Caucasus were influenced by the debris spread from the Semipalatinsk (Kazakhstan) and Kapustin Yar (Russia) test 325 sites, where hundreds of nuclear explosions have been carried out (Łokas et al., 2018). The effects of high latitude nuclear polygons (Novaya Zemlya) and of the re-entry of 238 Pu powered satellites, explain the non-global fallout contribution observed in the Arctic cryoconite, which is enriched in both 239 Pu and 238 Pu with respect to the Morteratsch and Forni samples (Łokas et al., 2019). The latter, showing a good agreement with the global fallout reference, rule out the possibility that a fraction of the Pu found in Alpine cryoconite was produced during the Algerian atmospheric nuclear tests carried out 330 by France in 1960s.
By studying the 137 Cs and 239+249 Pu activity ratio ( Tab. 1 and Fig. 5c), it is possible to infer the potential sources of 137 Cs, whose activity is by far the highest among the artificial radionuclides found in cryoconite. While global fallout has been demonstrated as the main source of Pu, the same is not for 137 Cs. On average, the 137 Cs fraction found in the Morteratsch Glacier samples from global fallout is 41 % with respect to total 137 Cs. For the Forni samples the value is lower (12 %). The 335 non-global fraction of 137 Cs found in Alpine cryoconite is attributable to the radioactive contamination released during the Chernobyl event. The Alps, and in particular the Eastern Alps, were among the most heavily impacted areas by Chernobyl fallout, where 137 Cs was a dominant component (Steinhauser et al., 2014). This is confirmed by the radioactive signature of cryoconite from two Austrian Alpine glaciers in the Eastern Alps (Tieber et al., 2009;Wilflinger et al., 2018), whose 137 Cs content is dominated by Chernobyl contamination (more than 90 %). Samples from the Caucasus also show a dominant 340 Chernobyl contribution with respect to 137 Cs, while cryoconite from Svalbard is anomalous in being characterized by a primary influence from global fallout (56 %). This is, however, not unexpected since, among the glaciers considered in Fig.   5, the Waldemarbreen (Svaldbard) is the farthest from Chernobyl. While Pu has a dominant global source, 137 Cs is related both to global and Chernobyl-related fallouts. Pu, together with the other actinides, is highly non-volatile and its transport mostly takes place through the dispersion of micrometric particles 345 from nuclear fuel, fission and activation products (Sandalls et al., 1993), while Cs is volatile and its mobilization during nuclear accidents requires relatively low temperatures. The Pu contamination from Chernobyl was limited to few hundreds of km from the emission site and could not be efficiently transported for long distances, while 137 Cs transport was widespread, leaving a strong signature all over Europe (Steinhauser et al., 2014), as it is also supported by Alpine cryoconite.

Carbonaceous content 350
Results of carbon analyses are presented in Fig. 6. On average (± standard deviation), the carbonaceous composition of the Morteratsch Glacier samples is 9.4 ± 1.4 % m/m for organic matter and 0.50 ± 0.25% m/m for elemental carbon. Cryoconite from the Forni Glacier contains a lower concentration of both species: 7.2 ± 0.8 % for organic matter and 0.2 ± 0.2 % for elemental carbon. Values about organic matter are compatible with the wider literature, where organic matter in cryoconite has been reported to vary between 2 and 18 % (Cook et al., 2015). Very limited information is available about the elemental 355 and/or black carbon composition of cryoconite, despite a great deal of attention having been given to the carbonaceous impurities present in snow in relation to the effect on ice/snow darkening . Our results show that elemental carbon is accumulated in cryoconite with respect to Alpine snow, where typical concentrations are orders of magnitude lower (Jenk et al., 2006). Only contaminated urban soils present an elemental carbon concentration comparable to Alpine cryoconite samples (Lorenz et al., 2006). These findings support the hypotheses by Hodson (2014), who have 360 suggested that cryoconite plays a role in extending the residence time of black and elemental carbon on the surface of glaciers, with implications for the accumulation of hydrophobic contaminants and for ice darkening.
Cryoconite from the Morteratsch glacier presents a higher concentration of both organic and elemental carbon than the one from the Forni glacier (Student's t test: t19 = 3.80; p-value < 0.001 for organic carbon concentration -Student's t test t19 = 3.10; p-value = 0.003 for elemental carbon). We hypothesize that elevation has a role in explaining the difference.
Cryoconite from the Morteratsch glacier have been sampled at an elevation between 2100 and 2300 m a.s.l., while samples 370 from the Forni glacier have been collected between 2600 and 2800 m a.s.l. A higher elevation implies lower temperatures, a shorter summer season and thus a less pronounced biochemical activity, which is in accordance with the lower organic carbon content observed in cryoconite at the Forni glacier.

Considering radioactivity as a whole
To analyse possible relationships between the different radionuclides, MDS and PCA have been applied on our data. The 375 first tool has been used to represent the degree of similarity and dissimilarity between the radionuclides (Fig. 7a). In the twodimensional domain of MDS, the radionuclides are grouped within three clusters which are interpreted as: 1) artificial radionuclides; 2) 238 U-chain nuclides; 3) 232 Th-chain nuclides. Despite 40 K doesn't belong to any of these groups, its distance from the 238 U-and 232 Th chain clusters is limited, confirming that K, Th and U in cryoconite are all associated to lithogenic components. The most isolated of the nuclides is unsupported 210 Pb, in accordance to its peculiar biogeochemical cycle. 380 MDS is able to highlight the different sources of the radionuclides considered in this study: 1) the artificial radionuclides, whose presence on glaciers is mostly related to stratospheric fallout; 2) the lithogenic radionuclides which are present in the mineral fraction of cryoconite; 3) and 210 Pb which is deposited onto the glacier from the lower troposphere by precipitation.
This partitioning is useful for interpreting the differences observed between the two glaciers considered here. At the Morteratsch Glacier the activity of the stratospherically derived radionuclides (Pu, Am, Bi) is higher than on Forni Glacier; 385 for 210 Pb the opposite is true (Fig. 4 and Supplementary Material). This pattern may be related to the altitude of the glaciers.
The Morteratsch Glacier basin has a maximum altitude of 4,049 m a.s.l. and an average elevation higher than 3000 m a.s.l., while the Forni basin is delimited by peaks whose maximum altitude spans from 3200 to 3400 m a.s.l. and only occasionally exceed 3500 m a.s.l.. The lower altitude could explain the higher amount of 210 Pb found in cryoconite from the Forni Glacier, since the maximum atmospheric scavenging of 210 Pb occurs in the lower troposphere, below 4,000 m (Guelle et al., 1998). In 390 contrast, the Morteratsch basin, given its high elevation, is more exposed to stratospheric fallout, perhaps explaining why the cryoconite from this glacier is highly contaminated with Pu isotopes, 241 Am and 207 Bi. Another factor that should be considered to explain the stronger contamination of cryoconite from Morteratsch, is the higher concentration of carbonaceous compounds in cryoconite from this glacier, for which radionuclides are particularly affine (Gadd 1996;Fowler et al., 2010;Kim et al., 2011;Chuang et al., 2015). 395 Results from PCA allow for the distinction of cryoconite sampled from the two glaciers. As seen in Fig. 7b- is linked to the positive scores of Forni cryoconite. One sample from the Forni Glacier is an outlier with respect to the others, being characterized by low concentration activity for most of the radionuclides, in particular the artificial ones.

The age of cryoconite and its relationship with ice surface processes 405
Natural and artificial FRNs are widely used to constrain chronologies in sedimentary environments. Among the nuclides considered here, 210 Pb and 137 Cs are commonly applied for dating, while 207 Bi and 241 Am have been rarely used to mark the period of maximum FRN deposition from atmospheric weapon tests (Kim et al., 1997;Appleby, 2008). Given the high concentration of radionuclides in cryoconite, it would be interesting to assess if they could be used to estimate the age of cryoconite itself. The most important issue that makes any attempt at dating challenging, is the complete absence of a 410 stratigraphic record in cryoconite. In studying cryoconite it is only possible to obtain a set of distinct and uncorrelated samples. Wilflinger and colleagues (2018) used 210 Pb to infer the mixing age (intended as an approximate mean age) of cryoconite samples from an Austrian glacier, the Stubacher Sonnblickkees. To attempt the dating, an assumption was made: once cryoconite is formed, its radioactive content starts decreasing following the decay law, regardless of the aggregation and dissolution processes that affect cryoconite granules (Takeuchi et al., 2010). Based on this hypothesis, cryoconite is 415 viewed as a sort of pure concentrated airborne material which is rich in atmospheric derived contaminants, as FRNs, and maintains its composition despite the dynamism of the supra-glacial environment. In Wilflinger et al. (2018) a highly radioactive sample of airborne sediments extracted from fresh snow, was interpreted as a sort of time-zero reference (a primordial cryoconite material), however no further details were given about this specimen. Comparing the 210 Pb activity of cryoconite to the reference, the mixing age of the samples have been thus inferred. According to this conceptual model, older 420 cryoconite presents lower 210 Pb activity in the light of the fact that the more time has passed from its formation, the more profound should be 210 Pb depletion due to the exponential radioactive decay. The estimated ages ranged from a few years to more than a century (Wilflinger et al., 2018). The glacier considered by Wilflinger et al. (2018) is small (less than 1 km 2 ) and is undergoing significant retreat and fragmentation (Kaufmann et al., 2013). The distribution of cryoconite on glaciers is extremely dynamic and is influenced by meteorological processes, local ice morphology, and supraglacial melting and 425 runoff. It has been observed that within only a few days, single cryoconite holes can form, deepen and collapse, scattering cryoconite granules downstream on the glacier (Takeuchi et al., 2018). In addition, it is known that cryoconite is far from being a static sediment: its granules are in fact subjected to uninterrupted changes, such as aggregation and break-up, and their lifetime on glaciers don't exceed a few years (Takeuchi et al., 2010). In Antarctica, where cryoconite holes are usually covered by a permanent ice lid and supra-glacial hydrology is poor, the isolation age (i.e. the time period during which a 430 single cryoconite hole have remained isolated from glacial hydrology) of single cryoconite holes has been estimated through a biogeochemical method: it never exceeds a few years (Fountain et al., 2004;Bagshaw et al., 2007). The transience of surficial glacial environments is furtherly confirmed by glacier moss balls (conglomerations of mineral debris, moss and organic matter forming on the surface of glaciers), whose lifespan was observed not to exceed few years (Hotaling et al., 2019). Given these evidences, we find it unlikely that a fraction of cryoconite sampled on the surface of a small and steep 435 glacier as the Stubacher Sonnblickkees, could form at the end of the 19 th century and persist there since then without being subjected to significant compositional changes.
We present an alternative hypothesis to link the content of FRNs in cryoconite and its formation age. Our conceptual model arises from an assumption opposite to that of Wilflinger and coauthors (2018): cryoconite is not a static material, its composition changes with time because of the processes taking place on the surface of glaciers. In light of this, the 440 radioactive content of cryoconite is not only subjected to decay, but also to a build-up derived from continuous accumulation. Consequently, the older the cryoconite is, the higher is its 210 Pb content, because it has had a longer time within which to accumulate the radionuclide, which is continuously deposited on the glacier with snow and rain. We hypothesize that the build-up of radioactivity in cryoconite is derived from the interaction between ice, meltwater and cryoconite granules. During summer, the radionuclide content of ice and snow is mobilized through melting, including 445 unsupported 210 Pb, which is always present in relatively recent ice (given its lifetime, it is not present at detectable concentrations in ice older than 150-200 years). The interaction between cryoconite granules and meltwater containing 210 Pb, explains why the latter is always found at high concentrations in cryoconite, regardless of the geographic context. For artificial FRNs the case is different since they are not continuously deposited on the surface of glaciers; however, they are still present in cryoconite with high activities. Each year during the melting season part of the ice dating back to the peak of atmospheric nuclear tests and to major nuclear incidents, melts out, releasing its artificial nuclide burden which is transported by meltwater. As for unsupported 210 Pb, cryoconite granules retains such nuclides owing to their biogeochemical properties and accumulate a load of artificial radioactivity even if decades have passed since its original deposition on glaciers. The ability of cryoconite is likely related to the presence of organic matter and extracellular polymeric substances which are affine for heavy metals, including the radioactive ones (Gadd 1996;Fowler et al., 2010;Kim et al., 2011;Chuang 455 et al., 2015). An additional support for the importance of organic matter in this process is also given by previous studies showing that the organic fraction of cryoconite and snow algae accumulates heavy metals associated to anthropogenic atmospheric emissions (Fjerdingstad, 1973;Nagatsuka et al., 2010;Łokas et al., 2016;Baccolo et al., 2017;Owens et al., 2019;Huang et al., 2019).
One observation might corroborate our hypothesis. Wilflinger et al. (2018) reported about high activity of 7 Be (t1/2 = 53 d) in 460 their samples, of up to 34,000 Bq kg -1 . 7 Be is a short-lived cosmogenic radionuclide, deposited from the atmosphere with precipitation. We observed 7 Be within our samples, but we could not properly quantify it because months passed between sampling and γ-spectrometry. Finding an excess of 7 Be in cryoconite, implies that, given its lifetime, the absorption by cryoconite granules took place in the weeks just before sampling and not when the cryoconite originally formed. The presence of short-lived nuclides suggests that cryoconite granules continuously accumulates radioactive species through the 465 interaction not only with meltwater but also with rain, where 7 Be is always present.
According to our interpretation, cryoconite containing higher concentrations of radionuclides has formed on the glacier before cryoconite presenting lower activities. Beyond this, however, we believe it is difficult to attempt a more precise dating through radioactive decay, even if it remains an interesting task. Too many processes are poorly understood to make a rigorous attempt at present, we first should understand the relationships which exist between the formation of cryoconite and 470 of cryoconite granules, the geometry of the glacier, the age and displacement of ice, and in particular the exchanges between ice, meltwater and cryoconite granules.

Conclusions and future perspectives
We have described the ability of cryoconite to accumulate both artificial and natural FRNs. A comprehensive comparison against other environmental matrices revealed that cryoconite is, excluding samples from nuclear test and incident sites, one 475 of the most radioactive natural substance found in Earth surface environments. Our study is focused on cryoconite samples from the European Alps but results from other regions of the global cryosphere confirm our findings, proving that the accumulation of radioactivity is not a local phenomenon, but involves worldwide glaciated areas. The accumulation of FRNs in cryoconite is so efficient that it has even allowed for a relatively easy detection of not common FRNs. Cryoconite is a promising tool in the fields of radioecology and environmental nuclear forensics.

20
The use of diagnostic ratios has shed light on the sources of radioactivity found in cryoconite. Results show that multiple sources, both regional and global, influences the radioactive signature of Alpine cryoconite. Pu related nuclides reveal a dominant source of their presence to be the global stratospheric fallout from atmospheric nuclear tests carried out in the second half of the 20 th century. In contrast, the major contribution for 137 Cs is determined to have come from the 1986 Chernobyl accident. The ability of recording both planetary and more regional events, is also suggested by a comparison 485 with literature concerning other geographic contexts. Differences are observed and can be explained considering the impact of regional events. It is important to note that currently no information exist about the radioactivity of cryoconite from the Southern Hemisphere. To build a comprehensive picture of radioactivity in the global cryosphere this is a geographic gap that it would be valuable to close.
There is evidence to suggest that the fundamental process which makes cryoconite a "sponge" for impurities in glacial 490 environments, including radionuclides, is the interaction between ice and cryoconite itself, through the mediation of meltwater. When glaciers melt, they release and mobilize with meltwater the radionuclides originally preserved in snow and ice layers. Due to the organic matter content and its sticky properties, cryoconite efficiently binds and accumulates the impurities contained in meltwater, in particular those with an affinity for organic substances, including radionuclides.
This study has focused strictly on the glacial environment, ignoring the fate of cryoconite once it is released by glaciers and 495 transported into the downstream ecosystems. It is likely that owing to meltwater discharge, the radioactivity accumulated in cryoconite is promptly diluted, avoiding any health and ecotoxicological risk. However, caution should be taken considering those pro-glacial areas in close proximity to the ice, where the dilution could be limited, and some risks could exist. Given the global relevance of this phenomena, further research should focus on the extra-glacial fate of cryoconite and the contaminants contained within it. 500

Data availability
Full data are available as supplementary material.