Circum-Arctic release of terrestrial carbon varies between regions and sources

Arctic change is expected to destabilize terrestrial carbon (terrOC) in soils and permafrost, leading to fluvial release, greenhouse gas emission and climate feedback. However, landscape heterogeneity and location-specific observations complicate large-scale assessments of terrOC mobilization. Here we reveal differences in terrOC release, deduced from the Circum-Arctic Sediment Carbon Database (CASCADE) using source-diagnostic (δ13C-Δ14C) and carbon accumulation data. The results show five-times larger terrOC release from the Eurasian than from the American Arctic. Most of the circum-Arctic terrOC originates from near-surface soils (61%); 30% stems from Pleistocene-age permafrost. TerrOC translocation, relative to land-based terrOC stocks, varies by a factor of five between circum-Arctic regions. Shelf seas with higher relative terrOC translocation follow the spatial pattern of recent Arctic warming, while such with lower translocation reflect long-distance lateral transport with efficient remineralization of terrOC. This study provides a receptor-based perspective for how terrOC release varies across the circum-Arctic.


Supplementary Methods 1: End member definition for the Circum-Arctic shelf seas
To distinguish between terrOC and marine OC this study applied a large database of 13 C and 14 C measurements of end member materials within the circum-Arctic watershed from the published literature. For all regions, the source apportionment assumes mixing from two terrestrial end members and marine phytoplankton as major OC sources. Overall, terrOC end members represent i) OC in surface soils with 0-100 cm depth, with an age of up to a few thousand years (incl. permafrost soils; except for the Beaufort Sea where surface soil was combined with peat), and ii) a strongly pre-aged terrOC source from deeper layers (ICD, petrogenic C, or deep peat OC; depending on occurrence in the catchment).

Surface soil
The circum-Arctic end member for surface soil builds on a large in-house data collection employed by previous studies 9, 10 , which was further amended by a large soil radiocarbon database 11 , and only contains data of samples from inside the circum-Arctic watershed. The surface soil end member contains data from the permafrost active layer in permafrost regions to a maximum depth of 100 cm, and data for nonpermafrost soils to a depth of 100 cm. The δ 13 C value of the surface soil end member is -26.6 ± 1.7‰ (n=219) and the Δ 14 C is -201.1 ± 229.3‰ (n=304).

Marine organic carbon
This study applied two different marine OC end members. The open-marine OC end member is based on marine phytoplankton/algae sampled from open ocean settings under influence of water mass inflow from the Atlantic or the Pacific Ocean (-50 ± 12‰ for Δ 14 C; n=5; -21 ± 2.6‰ for δ 13 C; n=31) and was applied in the Canadian Arctic Archipelago, Beaufort Sea, Chukchi Sea, ESAS east of 160°E, and the Barents Sea. The open marine end member excludes any measurements of phytoplankton from the Laptev and East Siberian seas west of the frontal zone at 160°E 12 and the Kara Sea, since these areas are strongly affected by terrOC and which influences the δ 13 C ratios of marine phytoplankton 13,14 . A wider marine end member was consequently applied for the Laptev and Kara Seas, as well as the East Siberian Sea west of 160°E. The wider marine endmember contains all data of marine phytoplankton available for the Arctic Ocean (-16 ± 53‰ for Δ 14 C; n=12; -23.2 ± 3.5‰ for δ 13 C; n=52) and thereby accounts for the possible influence of terrOC on the isotopic composition of marine phytoplankton in this region.

ICD
The end member for ICD in Eastern Siberia, Alaska and northwestern Canada applied a large in-house database of 14 C measurements from coastal exposures in Siberia, Alaska and northwestern Canada that was published and updated throughout previous studies 2, 9,10 . The Δ 14 C of the ICD end member is -960.3 ± 60.7‰ (n=487). For the δ 13 C, this study applied an average value reported in a literature review (-26.3 ± 0.7‰; n=374) 15 .

Peat
This study also accounted for peat as a potential terrOC source to Arctic Ocean sediments. For the Kara Sea catchment, which hosts the world's largest peatland, this study distinguished an end member for peat deposits deeper than 100 cm depth using a large collection of 14 C-dated peat cores 16 from the Circum-Arctic (-503.2 ± 158.9‰; n=263). For the Beaufort Sea, this study accounted for the large peatlands in the drainage basin by using a combined end member for surface soil and peat OC, which is based on published 13 C and 14 C data from peat core samples (incl. 0-100 cm depth) within the watershed of the Beaufort Sea (-377.9 ± 201.3‰; n=191). Data for δ 13 C in peat is rare but is expected to resemble the isotopic characteristics of terrOC that was produced by C3 plant photosynthesis (-27‰). This study thus used the same δ 13 C for peat as for surface soil OC (-26.6 ± 1.7‰; n=219).

Petrogenic C
A petrogenic carbon end member was applied for the Canadian Arctic Archipelago, the Canadian Beaufort Sea and the Barents Sea to account for ancient C reservoirs in the form of petrogenic sources in these catchments 17 . The petrogenic carbon end member is based on previous studies of sediment OC sources in this region and builds on δ 13 C measurements of kerogen in sedimentary rocks around -26‰±2 18,19 while the Δ 14 C was assumed to be 14 C-dead/undetectable at -998‰.  Fig. 1: Comparative analysis between the I-CRI of the different terrestrial organic carbon (terrOC) compartments and environmental factors for each drainage basin. Shown are a) the relationship between the I-CRISurfSoil and the catchment area, and b) the I-CRIICD and average rates of coastal erosion 24 . Further shown are the I-CRI for c) terrOC and d) for surface soil (SurfSoil) with warming trends of the warm period (May-Nov) for 1960-2015 25,26 . Panel e) shows the ratio of estimated inland CO2 efflux 27 to the total OC discharge for the major river systems (denoted as grey text) of each drainage basin 4,28,29 . Panel f) shows the I-CRISurfSoil and the distribution of continous permafrost 30 . Panels g) and h) show the correlation between the I-CRISurfSoil and I-CRIterr with the southward extent of the drainage basins. Panel i) shows the correlation between the % of the total terrOC stock and the % of total terrOC accumulation for each circum-Arctic shelf sea. Solid black lines indicate significant correlations, for which the coefficients of determination (R 2 ) and p-values are shown. The abbreviations of the shelf seas are CAA -Canadian Arctic Archipelago, BFS -Beaufort Sea, CS -Chukchi Sea, ESS -East Siberian Sea, LS -Laptev Sea, KS -Kara Sea, BS -Barents Sea, CAO -Central Arctic Ocean.