4. DISCUSSION
Permafrost regions have experienced dynamic environmental changes induced by climate change, especially in the north-eastern Siberia (Nitzbon et al. , 2020). Recent climate-induced increases in thaw propagation have triggered changes in local relief in the Yedoma uplands, including soil subsidence (Günther et al. , 2015), activation of thermokarst and thermoerosion processes (Grigorievet al. , 2009; Morgenstern et al. , 2021), and the expansion of pond and thermokarst lake areas (Nitze et al. , 2017; Veremeevaet al. , 2021). In the large rivers, an increase in runoff is both expected and already observed: in the Kolyma, mean annual discharge has increased over the 2010–2020 by 27.7% (94.6 to 120.7 km3 year-1) compared to a baseline period of 1971–2000 (Mann et al. , 2022). The same authors estimated an increase of 50% (±25%) in Kolyma’s mean annual runoff by 2100, in line with climate projections. Climate warming resulted also in an increase in Kolyma’s sediment transport rates from 20 to 100 t km-2 year-1 in 1941-2000 (Chalovet al. , 2018), which can be also attributed to mining impact (Walling & Fang, 2003) or changes due to permafrost thaw (Chalovet al. , 2021).
The data presented here shows the hydrochemical patterns of different waters in an environment shaped by permafrost seasonal and long-term changes. The ionic composition of all sampled waters is likely impacted by both suprapermafrost waters and an inflow of deeper, intra- and subpermafrost waters; especially the permafrost creeks P1 and P2 are likely connected to unfrozen groundwater (Cochand et al. , 2019).
Among the studied samples, the permafrost creeks differ from river waters by the high content of DOC, major cations and anions, and selected metals and metalloids, including heavy metals. The pathways of OC transport and transformation in Siberian freshwater have been thoroughly analysed recently, e.g. by Wild et al. (2019), Mannet al. (2022), or Jong et al. , (2022). Mann et al.(2022) stated that permafrost-derived OC additions will significantly enhance inland OC turnover over the upcoming decades, however the fate of OC is uncertain due to additional transformation factors (Keskitaloet al. , 2022). In the large rivers, OC occurs mainly as DOC: for Kolyma, the dissolved share has been calculated by Mann et al.(2022) as 80%. Similar values are reported by others (Jong et al. , 2022; Keskitalo et al. , 2022). Keskitalo et al.(2022) have concluded, based on isotopic evidence, that the smaller fraction by mass, the POC, is more affected by the legacy Yedoma-derived OC, than is DOC. Similarly, Wild et al. (2019) claim that POC in the large Siberian rivers may originate from the direct erosion of the old Yedoma layers, although they also estimate the legacy OC share in the DOC and POC fractions in the Kolyma to be approximately equal. Bröder et al. (2020) remarks that Kolyma transports more suspended POC at higher discharges, and both the total POC and legacy POC concentrations are higher there than in a small stream (the average POC ages in Kolyma and a small stream were estimated at ~2,840 and ~590 years, respectively). These claims are supported by Jong et al. (2022), who observed an increased contribution of POC in the TOC transport in large rivers and connected it to the increased direct erosion of the river banks, while a smaller stream exhibited a higher share of DOC. Therefore, the POC in Kolyma likely originates from recent vegetation and old permafrost erosion combined. Mann et al. (2022) suggest also that the OC flux and total export will rapidly increase due to climate change and permafrost degradation.
In our research, the DOC concentration in a permafrost ice sample amounted to 9.13 mg L-1 and it was approximately 11 times less than the concentrations in the creeks draining the permafrost cliff. Similar DOC values of 154-336 mg L-1 and 103.4 mg L-1, for creeks draining the permafrost cliff at Duvanny Yar, were obtained by Vonk et al. (2013) and Jonget al. (2022), respectively, and the concentrations of POC in the same samples amounted to 6700-9240 mg L-1, eclipsing the DOC contribution. (This is also consistent with the additionally sampled POC in P1 and P2 samples, amounting to 8500 and 12000 mg L-1, respectively – methods are reported in theSupp. Mat. 1. ) Conversely, Jong et al. (2022) obtained DOC concentrations (2.76 to 4.97 mg L-1) at approximately double the POC concentrations (1.49 to 2.73 mg L-1) for Kolyma below Duvanny Yar, and similar results for Maly and Bolshoy Anyui (3.16-4.43 mg L-1 and 1.29-1.7 mg L-1, respectively). In our research, thermokarst lakes were studied also, with DOC concentrations similar to a small stream studied by Jong et al. (2022), representing soil leaching and active layer drainage ([DOC] = 21.5 mg L‑1). The legacy DOC from permafrost can be approximated by the DOC concentration in permafrost ice (9.13 mg L-1), yet the extra 90% of DOC in permafrost creeks may originate both from permafrost active layer (i.e. relatively recent atmospheric deposition and microbial processes) and the exclusion of various organic compounds during groundwater freeze. POC from the ice may also be a source of DOC in thaw waters. The cluster analysis of our data has shown similarities in the water chemistry between the permafrost ice sample (P3), the bottom layer of water in Lake 2 and a sample from Kolyma below Duvanny Yar (K31, sampled mid-depth away from the cliff side, where the main river current flows), which suggests the influence of legacy permafrost thaw upon the Kolyma river in this cross section. At the same time, the connection between the other lake samples and the samples from the permafrost creeks suggests a general permafrost waters influence over the lake waters, although it is probably a combined influence of the active layer drainage and deeper talik-type waters.
In the collected permafrost ice, a contrastingly high level of aluminium concentration has been noted as compared to the other samples; a relatively high concentration of Al was also found in the permafrost creeks and the sampled lakes. Aluminium co-occurs with K+, Be, but also with Co and As, frequently classified as heavy metals. Arsenic occurred at elevated (against other samples) concentrations in the lake bottom waters of this study. In the lake waters, permafrost creeks and permafrost ice, we found also elevated concentrations of other heavy metals (Co, Mn, Ni, Zn, Cu and Sb). Zinc, copper and cadmium showed elevated values also in the Kolyma, while the bottom layer waters of the thermokarst Lake 2 exhibited also a very high concentration of Pb. Monhonval et al. (2021) characterised the chemical composition of Yedoma waters, by exploring the mineral element stock in sediments of the Yedoma. They found a relatively high concentration for Si, followed by Al, Fe, K, Ca, Ti, Mn, Zr, Sr, and Zn. Moreover, the stock of Al and Fe (598 ± 213 and 288 ±104 Gt) was at the same order of magnitude as the OC stock (327–466 Gt). Our surface water and ice samples from the Kolyma watershed exhibited similarly high concentrations of Si and Fe, as well as Al. The cluster analysis of all the collected samples shows that in cluster C3, the elements Si and Fe were grouped together with Ni and Mn (Figure 6a). However, in river waters they correlated with Zn, Sb, Cu, V, Co, As, and DOC (cluster c1, Figure 7). Both clusters may be connected to the suprapermafrost waters of the active layer, which are flowing intensely into the thermokarst lakes and rivers. Ji et al. (2021) studied elements in permafrost soils and concluded that Mn, Ca, Mg, Al, and Ti showed the highest mobility from soil to suprapermafrost water and further into the ponds and flowing waters. Furthermore, they considered Mn a proxy for predicting the processes occurring in the active layer during summer-autumn thaw. Conversely, the long-term research in the Yukon River basin (1982-2014), underlain by discontinuous permafrost, shows a relationship between the deepening of permafrost active layer and significant increases in the Ca, Mg, and Na annual flux in the Yukon River and its tributary Tanana, and in the SO4 and P annual flux in the Yukon (Toohey et al. , 2016). The authors described increased active layer expansion, weathering, and sulfide oxidation due to permafrost degradation throughout the studied basin. Frey & Mcclelland (2009) forecast also a significant increase in the concentrations of major ions due to permafrost degradation and lowering water tables (except for nitrate in Siberia). In the studied samples from the Kolyma watershed, clusters C5 and c3 (in the all sample and river sample analysis, respectively; Figure 6a, Figure 7) indicate the likely influence of rock weathering on water composition. The visible division within the c3 cluster into c3a (Sr, Se, K+, Mg2+, Ca2+ and SO42-) and c3b (Cl-, Na+ and F-) may represent two separate water supply factors, namely the shallow and deep groundwater flow (Douglas et al. , 2013). An extra contribution to the total sulphate load in the rivers may also result from the wet deposition connected to wildfires, which have influenced considerably the deposition of sulphur (and nitrogen) compounds in the regions of Siberia and the Russian Far East (Berezin et al. , 2013). On the other hand, the lower contribution of Ca2+ in thermokarst lake waters may result, according to Monhonval et al. (2021), from the leaching of soluble elements such as Ca during former thaw periods.
In this study, mercury was found only in permafrost creeks, permafrost ice and in the Maly Anyui. Recent research has shown that Hg may be released from continental permafrost with climate-change-induced thaw, and there are very different quantitative estimates of Hg fluxes connected to various local sources (permafrost is considered one source among the many; Campeau et al. , 2022; Ci et al. , 2020; Muet al. , 2020; Schuster et al. , 2018). Schuster et al. (2018) estimated that the entire Northern Hemisphere permafrost area contains 1,656 ± 962 Gg Hg, of which 793 ± 461 Gg Hg (~47%) is frozen in permafrost. Except in permafrost ice, our samples may contain mercury of various origin, both from permafrost thaw and from atmospheric deposition (Schuster et al. , 2018), and in the case of Maly Anyui it may also be connected to mining. (Gold mining occurs in parts of both the Maly and Bolshoy Anyui catchments, yet higher production comes from around the Maly Anyui tributary Karalveem River). However, the low Hg concentrations (<LOD) in the other collected samples (from Kolyma, Omolon and the lakes) indicate that any atmospheric sources of Hg would likely be only local and periodic (e.g., forest fires) (Francisco López et al. , 2022). Furthermore, it cannot be excluded that most Hg in the analysed samples occurred in the particulate form, which has not been investigated here. Lim et al. (2019), who have analysed the POC and Hg fluxes in the rivers of various permafrost zones, indicate minimal particulate Hg export at the sporadic to discontinuous permafrost zone. According to these authors, the current climate warming, the northward shift of permafrost boundaries and the increasing active layer depth in Western Siberia, may result in an enhanced particulate Hg export by small rivers to the Arctic Ocean by a factor of two over the next 10-50 years. The possible origin of the Hg may be also connected to mining areas (both abandoned and active) distributed over Maly Anyui river.
Besides mercury, also As and Cd are heavy metals sometimes mentioned as posing a new hazard due to permafrost thaw (Zhang et al. , 2021). In this study, a potential source of those is the erosion of a permafrost cliff, given the significant enrichment of As and Cd in permafrost ice, the extremely high EF in permafrost creeks, and very high to extremely high EF in the deep waters of thermokarst lakes (Table S4, Suppl. Mat. 1 ). The supply of Cd may be connected to the drainage of mixed suprapermafrost water, which may include atmospheric deposition, and deeper water both. With the active layer deepening, atmospheric pollutants deposited in the past, which have been excluded from the freeze-thaw cycle for a time (Ji et al. , 2021), may be remobilised (Edwards et al. , 2021; Li et al. , 2020; Limet al. , 2019; Rubino et al. , 2016). The notable increase in Cu, Ni, Pb and Zn concentrations in deep lake waters indicate the likely source of these elements in permafrost thaw, talik water, or enhanced mobility of suprapermafrost water eluting them from soil. Importantly, it is not only in the Lake 1 near the Cherskii research station and the settlement which experienced the elevated concentrations of these metals in the bottom waters, but also in the further from any human activity Lake 2, showing that such contamination is probably irrelevant of local infrastructure.
In the studied freshwaters, it was evident that thermokarst lakes form at least temporary traps for heavy metals connected to one or several types of permafrost influence. Thermokarst lakes form one of the more dynamic elements of the continental permafrost landscape, and their mass emergence is among the more frequently observed impacts of climate change in the Northern Hemisphere (in’t Zandt et al. , 2020; Karlsson et al. , 2012; Plug et al. , 2008), including the Kolyma region (Veremeeva et al. , 2021). Due to the dynamically increasing number of thermokarst lakes, and their predicted drainage with the advancing climate change and deepening active layer, the sedimentation in these lakes and the biogeochemical processes happening in them (in’t Zandt et al. , 2020) will likely be of high importance in studies of continental permafrost areas.