Properties of permafrost soils
All sites contained syngenetic permafrost in which the active layer and the uppermost permafrost have experienced numerous freeze-thaw cycles since formation during the Holocene22. Permafrost organic matter radiocarbon ages ranged from 7850 ± 30 to 830 ± 20 y B.P. (Supplementary Data), with the western sites containing the oldest SOM and the northern Hudson Bay peatlands containing the youngest SOM.
For organic layers, permafrost soil C:N atomic ratios (14.50 [12.61–19.67], median [25th–75th]) were lower and H:C atomic ratios (0.13 [0.13–0.14]) were greater relative to the active layer, 23.89 [19.33–29.50] and 0.14 [0.12–0.15], respectively (Supplementary Fig. 1). Similarly for mineral layers, permafrost C:N ratios were lower (12.0 [3.23–18.09]) and H:C ratios higher (0.24 [0.15–0.86]) compared to the active layer, 16.21 [13.67–18.87] and 0.17 [0.15–0.21], respectively. For both thermal layers, in organic layers C:N ratios were higher and H:C ratios were lower than in mineral layers.
These stoichiometric properties are typical of boreal and tundra soils (Supplementary Fig. 1)23. The higher C:N and very low H:C ratios of the organic layers relative to mineral layers suggest higher contents of condensed aromatic structures originating from peat24. Permafrost layers displayed lower C:N properties suggesting different SOM composition (e.g., lignin, tannins, lipids, sugars or amino acids) and an enrichment in microbial biomass relative to the active layer24. The absence of a downward trend of C:N and H:C within the permafrost (Supplementary Fig. 1), except at Daring Lake, indicates that soil development and microbial processing were effectively halted soon after permafrost aggradation23.
Active layer and permafrost yields of DOC and nitrogen
DOC content correlated with soil C content in both active layer (r2 [log–log] = 0.748, P < 0.001) and permafrost (r2 [log–log] = 0.834, P < 0.001, Supplementary Fig. 2). DOC and TDN pools were larger in organic layers relative to mineral soil layers (Fig. 1a, b). Organic permafrost yielded greater amounts of DOC and TDN than the active layer and mineral permafrost (Fig. 1a, b). We estimate that organic permafrost layers contained 0.396 [0.178–0.644] kg DOC m−3, and 37.6 [15.7–59.8] 10−3 kg TDN m−3, potentially releasing 5.3 times more DOC and 7.2 times more TDN upon thaw than either of the organic active layer, the mineral active layer, and the mineral permafrost layer (DOC: Kruskal–Wallis chi-squared = 103.22, P < 0.001, TDN: Kruskal–Wallis chi-squared = 85.3, P < 0.001, Fig. 1a, b), respectively. DOC pools in both active layer and permafrost were similar to values reported for mineral layers in Pleistocene Yedoma permafrost5 and in Orthels and Turbels from interior Alaska6, as well as for organic layers in Histels from Alaska and Siberia6,25.
Samples are grouped by layer type with the following classification: organic layers (soil C ≥ 12%) and mineral layers (soil C < 12%). Letters indicate significant differences between thermal layers and soil types indicated by Mann-Whitney U test. The box plots summarize the distribution of a the amounts of dissolved organic carbon (DOC, kg C m−3) and b total dissolved nitrogen (TDN, kg N m−3) and the c water extractable yields in DOC (g DOC 100 g soil C−1) and d TDN (g TDN 100 g soil N−1) for each layer type and in the active layer and underlying permafrost (calculated from multiple 25 cm increments). In each box plot, the diamond represents the mean, the horizontal line represents the median, the end of the box the 25th and 75th percentiles, and the lines extending from the box are 1.5 interquartile ranges from the median. Data points outside of the 1.5 interquartile ranges are represented as dots.
Within the active layer, organic and mineral soil samples displayed similar DOC (W = 48, P = 0.941) and TDN yields (W = 35, P = 0.963). Both DOC and TDN yields increased in the permafrost relative to the active layer (Fig. 1c, d), with mineral layers displaying greater DOC yields than organic layers in the permafrost (W = 1077, P = 0.019), extracting 1.3% [0.6–3.1] of soil C and 0.9% [0.3–1.4] of soil N, respectively. Our yields were in the range of values reported for permafrost soils in North America25,26.
In the active layer, dissolved organic nitrogen (DON = TDN − DIN) dominated the TDN pool representing ~78% of TDN in both mineral and organic layers. The DIN contribution increased with depth and ammonium-N dominated the DIN pool, with its concentration reaching 0.410 ± 0.326 mg N-NH4+ g−1 soil at 1–2 m depth in organic permafrost layers (Supplementary Fig. 3). In contrast, nitrate concentrations remained very low in both mineral and organic soils with 67 of 220 samples having NO3− concentrations below our instrument detection limit (0.006 mg N L−1, Fig. 2, Supplementary Fig. 3). The TDN, NH4+, and NO3−concentrations measured in this study were within the range of values reported for permafrost soils6,27,28,29.
Samples are grouped by layer type with the following classification: organic layers (soil C ≥ 12%) and mineral layers (soil C < 12%). Letters indicate significant differences between thermal layers and soil types indicated by Mann–Whitney U test. The box plots summarize the distribution of a the amounts of ammonium (NH4+) (kg N m−3) and b nitrate (NO3−) (g N m−3) for each layer nature and in the active layer and permafrost. In each box plot, the diamond represents the mean, the horizontal line represents the median, the end of the box the 25th and 75th percentiles, and the lines extending from the box are 1.5 interquartile ranges from the median. Data points outside of the 1.5 interquartile ranges are represented as dots.
While the active layer and mineral permafrost layers contained relatively little DIN, the pool of NH4+ in the organic permafrost layers (32.8 [11.3–66.2] g N–NH4+ m−3, Fig. 2) was very large compared to the general low nitrogen availability in Arctic environments13. Similar enrichment in NH4+ were reported in various permafrost settings in Sweden, Siberia, and Greenland28,29,30 as being a particular feature of permafrost soils in the Arctic. Saturated and oxygen-limited conditions in peatlands and syngenetic permafrost aggradation likely prevented microbial processing and/or leaching of NH4+ out of the soil profile. Seasonal NH4+ leaching down through the active layer at the end of the growing season, leading to NH4+ accumulation at the permafrost table during the winter freeze-up, likely contributed to the accumulation of a large pool of NH4+ 28 in the solute-enriched permafrost transition zone31 on decadal to centennial time scales. In addition, as microorganisms in permafrost soils are active under freezing conditions32, and some pore water remains within the permafrost, at least some of the large NH4+ pool likely originates from in-situ mineralization below 0 °C over thousands of years6,28. A thaw-induced release of permafrost NH4+ of this magnitude will likely affect Arctic terrestrial and aquatic ecosystems changing trophic structures and providing new habitat niches13 while altering landscape-wide carbon cycling processes15.
Optical properties of DOM in permafrost soils
The PARAFAC modeling for DOM fluorescence was conducted in all permafrost soils, spanning a wide range of surficial geology and climate conditions. We related six identified fluorescent components to fractions previously reported in surface waters in northern latitudes (Fig. 3, Table 1). Components C2, C3, C4, and C5 were previously related to humic-like or fulvic-like components associated with high molecular weight (HMW), aromatic organic compounds originating from terrestrial sources such as plant inputs and SOM33,34,35. Components C1 and C6 are ubiquitous across a wide range of terrestrial and marine environments and correspond, respectively, to tyrosine-like (C1) and tryptophan-like (C6) components, representing proteinaceous compounds from microbial activity such as amino acids, peptide materials and free or bound proteins34.
Excitation and emission peak positions of the three-dimensional excitation-emission matrices of the independent components are indicated alongside descriptions in Table 1.
The optical properties of DOM differed between thermal layers and between organic and mineral soil leachates (Fig. 4). Permafrost DOM exhibited a distinctive signature with a greater proportion of low molecular weight (LMW) proteinaceous organic compounds than found in the active layer (Fig. 4). In the active layer, the HMW aromatic fluorophores derived from terrestrial sources were the most abundant in both mineral and organic layers (94.7% [88.3–96.7] of the total fluorescence). These fluorophores contributed less to the fluorescent DOM, representing 77.0% [57.9–88.6] and 40.7% [23.9–70.8] of the fluorescence in permafrost mineral and organic layers, respectively (Supplementary Fig. 4). The tyrosine-like component (C1) was the most abundant fluorophore in the organic permafrost layers accounting for 51.2% [22.4–62.9] of the total fluorescence, and the tryptophan-like component (C6) representing 7.2% [5.0–10.1]. In the mineral permafrost, the tyrosine-like and tryptophan-like components represented 12.8% [2.7–34.4] and 7.6% [5.5–11.0] of the total fluorescence, respectively (Supplementary Fig. 4). These components contributed less in the active layer than in the permafrost representing 5.3% [4.1–22.8] in organic and 5.3% [3.3–10.7] in mineral layers of the fluorescent DOM (Supplementary Fig. 4).
The analyses takes into account the relative contribution of PARAFAC components and optical indices: the absorption coefficient at 350 (E350), the specific UV absorbance (SUVA254, L mg C−1 m−1), the absorption spectral slope over the spectral band 275–295 nm (S295), the slope ratio between S295 and the spectral slope over 350–400 nm (SR), the fluorescence index (FI), the freshness index (BIX), the humification index (HIX). a Explanatory variable loadings are shown as black dots and scores across the first and second principal components for all samples labeled by layer type—organic layers are shown as circles and mineral as diamonds—and colored by location in the active layer or permafrost. b Box plot of the scores of the first principal component for organic (soil C ≥ 12%) and mineral layers (soil C < 12%) in the active layer and in the permafrost. c Box plot of the scores of the second principal component for organic (soil C ≥ 12%) and mineral layers (soil C < 12%) in the active layer and in the permafrost. In each box plot, the diamond represents the mean, the horizontal line represents the median, the end of the box the 25th and 75th percentiles, and the lines extending from the box are 1.5 interquartile ranges from the median. Data points outside of the 1.5 interquartile ranges are represented as dots. Letters indicate significant differences between thermal layers and soil types indicated by Mann–Whitney U test.
The multivariate statistical analyses based on absorbance and fluorescence indices demonstrated that permafrost DOM exhibited a distinct optical signature shared by both mineral and organic layers (Fig. 4a, b). The permafrost leachates were enriched in LMW proteinaceous compounds with low aromaticity while the active layer DOM was comprised of HMW aromatic condensed organic compounds. The fluorescence index FI (a common proxy for the relative contribution between microbial and terrestrial DOM) and the ratio of absorption spectral slopes SR, which decreases with molecular weight and aromaticity, increased with depth (Fig. 4a, b, Supplementary Fig. 5) indicating more protein-derived substrate with a lower average molecular weight in the permafrost34. HIX and SUVA254 values, which are known to increase with bulk aromaticity and condensing, decreased with depth and in the permafrost (Supplementary Fig. 5) suggesting that permafrost stores less condensed and aromatic DOM than in the active layer20,34. Similar values of HIX and SUVA254 in active layer mineral and organic layers illustrate the similar long-term microbial processing of organic matter in the seasonally thawed zone Supplementary Fig. 5). In both thermal layers, DOM in organic and mineral layers only differed by the absorption at 350 nm, with organic layers displaying higher absorption values due to greater concentrations of aromatic DOM33 (Fig. 4a, c).
Our values of optical indices are consistent with the few available characterizations of DOM in the active layer and permafrost5,6,7,8,25,36. Due to different organic matter sources, DOM in mineral permafrost layers has been previously characterized by lower SUVA254 values ([0.6–1.2] L mg C−1 m−1)5,8 than organic active layer materials ([1–4.5] L mg C−1 m−1)6,8,25,37 due to a relative enrichment in microbial exometabolites and root exudates37 in mineral soils compared to peaty organic layers. Last, molecular techniques also support our findings by showing enrichment of LMW compounds such as carbohydrates8, acetate and butyrate5 in the permafrost in Alaska and Siberia.
Controls of the pan-Canadian permafrost DOM pool
Even though permafrost DOM displays contrasting optical properties from active layer DOM, the signature of permafrost DOM is weakly detected within Arctic catchments. This is mainly due to the high reactivity of DOM, which experiences sorption and desorption when percolating in the soil25, photo-oxidation of HMW aromatic components in streams and rivers, and microbial processing of less aromatic compounds10,18,25. Indeed, while our results show the high contribution of tyrosine-like component to the fluorescent DOM in both mineral and organic permafrost in the Canadian Arctic region, studies that have conducted PARAFAC modeling on DOM in Arctic surface waters have only detected a low contribution of this component in headwater catchments38,39, with no contribution at all in Arctic rivers33,35. The major tyrosine contribution to the permafrost fluorescent DOM and its disappearance within the fluvial continuum provides additional evidence of its high reactivity5,10,19.
Following permafrost thaw, three main processes lead to the decrease in the proteinaceous contribution during lateral flow downslope along the permafrost table, prior to reaching water bodies. First, soil microbes could preferentially mineralize tyrosine-like DOM components, such as amino acids and oligopeptides40, as they utilize aliphatic compounds41 almost immediately following permafrost thaw5. Recent studies in Arctic permafrost environments highlight that low SUVA25419, high SR42 values and a high contribution of tyrosine-like43 correlated with greater lability of bulk DOM6,36,43,44. Therefore, we suggest that Canadian permafrost likely contains a high proportion of biodegradable DOM. Secondly, in northern ecosystems plants take up amino acids and oligopeptides directly as a source of N, leading to a decrease in proteinaceous compounds in exported waters45. Finally, the retention of less aromatic compounds and protein-like fluorophores in mineral subsoils and the preferential release of HMW aromatic fluorophores lead to the depletion of tyrosine-like components in riverine fluorescent DOM and the relative enrichment in aromatic compounds25.
Our results demonstrate robust relationships between SOM stoichiometry (C:N and H:C ratios) and chromophoric and fluorescent DOM properties and DOC yield highlighting the control of permafrost organic matter on the DOM pool (Fig. 5). E350 decreased with increasing H:C ratios (Fig. 5a), suggesting that chromophoric DOM abundance was greater for C-rich SOM24. The C:N ratio was negatively correlated with SR and FI (Fig. 5b, c) demonstrating that N-rich SOM (i.e., enriched in microbial biomass) yields LMW DOM with low aromaticity. In addition, SOM properties (i.e., C:N ratio), the degree of aromaticity (i.e., SUVA254) and the microbial origin of fluorescent DOM (i.e., FI) was correlated with the water extractable DOC yield (Fig. 5d–f). Therefore, permafrost samples with lower C:N ratio yield proportionally more DOM that contains a higher proportion of less aromatic and more proteinaceous compounds, which are likely labile44 relative to the active layer. As soil C:N and H:C ratios are widely measured in permafrost soils23, these findings allow for upscaling of site or plot level analyses to a circumpolar characterization of the DOM pool.
Relationships between E350 (a), SR (b), FI (c, e), SUVA254 (L mg C−1 m−1) (e)) and soil properties (H:C (a), C:N (b, c, f), and DOC yield (g DOC 100 g soil C−1) (d, e, f)) for both active layer and permafrost samples. Each point represents one sample. The linear regression lines and equations are for active layer and permafrost samples.
Our results further indicate that the uppermost permafrost degradation has the potential to release a larger amount of DOC and TDN (mostly as DON and NH4+) than the active layer currently releases to high-latitude terrestrial and aquatic ecosystems. These findings are in agreement with observations of thawing permafrost leading to shifts in the supply of carbon and nutrients to surface waters46,47. Using the NCSCDv2 dataset3 for the pan-Canadian permafrost area, we estimated the DOC pool available right upon thaw increasing from 0.2 [0.2–0.5] Pg C in 0–1 m, to 1.1 [0.7–2.1] Pg C in 1–2 m and decreasing to 0.9 [0.7–1.5] Pg C in 2–3 m depth interval. Although DOC is a relatively small component of the total permafrost carbon pool, its depletion through hydrological mobilization will likely contribute relatively fast to soil carbon loss in Arctic landscapes48.
In addition to an increase in DOC availability in the large areas of peatlands located in southern boundaries of the permafrost region (i.e., the Mackenzie River region, the Hudson Bay lowlands), which have been subject to rapid thaw49 (Fig. 6a–c), the DOC pool increased with volumetric water content (Fig. 6d). While being more prone to thermokarst processes2, the degradation of ice-rich permafrost could potentially export more carbon to surface waters. Our data further show that the DOC pool was positively correlated with the NH4+ pool and the large DOM pools are characterized by lower aromaticity (i.e., low SUVA254) (Fig. 6e, f). This finding supports reported observations of the accumulation of NH4+ and biodegradable DOC in Arctic permafrost soils47. While proteins compose most of the DON that are thereafter degraded to amino acids and mineralized to NH4+, we demonstrate here the buildup of both NH4+ and tyrosine-like fluorophores in the permafrost, which then halted their decomposition and export. As reported in glacier ice, the dominant contribution of the tyrosine-like component to fluorescent DOM is evidence of microbial metabolism in permafrost50. The distribution of ground ice has been proven to be the main factor controlling the fate of permafrost carbon (i.e., thermokarst activity and methane production)2. Our data demonstrate that ice-rich permafrost soils have the potential to release great amount of NH4+ and DOM characterized by a specific low aromaticity protein-rich optical signature that suggests rapid mineralization potential.
a–c Maps of pool of DOC (kg C m−3) that would be available upon permafrost thaw for three soil layers: 0–1 m, 1–2 m, 2–3 m. The DOC pool in permafrost soils was estimated by multiplying soil carbon pool of the three different gelisol types from the NCSCDv2 by the median of water extractable DOC yield of mineral and organic soils. Blue dots represent the locations of the coring sites. The dark gray area represents the non-permafrost soils within the Canadian permafrost distribution zone (isolated, sporadic, discontinuous and continuous). d Volumetric water content (%), e pool of NH4+ (g N m−3), and f SUVA254 (L mg C−1 m−1) as a function of the pool of DOC (kg C m−3) for both active layer and permafrost samples. Each point represents one sample. The regression lines (dashed lines) and equations are for active layer and permafrost samples combined.
The uppermost layers of Canadian permafrost, found immediately below the seasonally thawing active layer, currently store substantial DOC and NH4+ pools, particularly in organic permafrost soils. This suggests that permafrost thaw—in addition to the release of reactive DOM—is likely to enhance nutrient inputs and availability into terrestrial, as well as aquatic systems. This nutrient priming mechanism will almost certainly interact with and affect microbial processing of organic matter and primary productivity, altering the net carbon balance of Arctic ecosystems15. Here, we show that permafrost DOM exhibits a highly consistent and distinctive chemical composition across a very large region of northern Canada and spanning numerous distinct environments and climates. In conclusion, our results support the premise that degradation of organic rich permafrost will increase carbon release and potentially affect aquatic systems through carbon and nutrient additions.
Source: Ecology - nature.com
