The largest stocks of soil organic carbon can be found in cold regions such as Arctic, subarctic and alpine biomes, which are warming faster than the global average. Discriminating between particulate and mineral-associated organic carbon can constrain the uncertainty of projected changes in global soil organic carbon stocks. Yet carbon fractions are not considered when assessing the contribution of cold regions to land carbon–climate feedbacks. Here we synthesize field paired observations of particulate and mineral-associated organic carbon in the mineral layer, along with experimental warming data, to investigate whether the particulate fraction dominates in cold regions and whether this relates to higher soil organic carbon losses with warming than in other (milder) biomes. We show that soil organic carbon in the first 30 cm of mineral soil is dominated or co-dominated by particulate carbon in both permafrost and non-permafrost soils, and in Arctic and alpine ecosystems but not in subarctic environments. Our findings indicate that soil organic carbon is most vulnerable to warming in cold regions compared with milder biomes, with this vulnerability mediated by higher warming-induced losses of particulate carbon. The massive soil carbon accumulation in cold regions appears distributed predominantly in the more vulnerable particulate fraction rather than in the more persistent mineral-associated fraction, supporting the likelihood of a strong, positive land carbon–climate feedback.

P.G.-P. acknowledges support from the Spanish Ministry of Science and Innovation via the I+D+i project PID2020-113021RA-I00 and the TED project TED2021-130908A-C42 (funded by European Union—NextGenerationEU). Work at Lawrence Livermore National Laboratory by N.W.S. was performed under the auspices of the US DOE OBER, under contract DE-AC52-07NA27344 award #SCW1632. M.P. acknowledges financial support by the Comunidad de Madrid and the Spanish National Council of Scientific Researches research grant Atracción de Talento (grant number 2019T1/AMB14503).

18 páginas.- 4 figuras.- 63 referencias.

Peer reviewed