摘要:Recent evidence suggests that biogeochemical processes in the Arctic during late winter and spring-thaw strongly affect the annual cycling of carbon and nutrients, indicating high susceptibility to climate change. We therefore examined the carbon and nutrient dynamics in a sub-arctic heath and a birch forest with high temporal resolution from March until snowmelt at both ambient and experimentally increased snow depths. Ecosystem respiration (ER) from mid-March to snowmelt at ambient snow was high, reaching 99 ± 19 (birch) and 67 ± 1.4 g C m −2 (heath). Enhanced snow depth by about 20–30 cm increased ER by 77–157% during late winter but had no effects during spring-thaw. ER rates at the birch site were poorly described by classic first-order exponential models ( R 2 = 0.06–0.10) with temperature as a single variable, but model fit improved considerably by including the supply of dissolved organic carbon (DOC) or nitrogen (DON) in the model ( R 2 = 0.40–0.47). At the heath, model fit with temperature as the single variable was better ( R 2 = 0.38–0.52), yet it improved when the supply of DOC or DON was included ( R 2 = 0.65–0.72). Microbial carbon decreased by 43% within a few days after the first soil freeze-thaw event, while microbial nitrogen and phosphorus decreased more slowly. Because soil inorganic nitrogen and phosphorus concentrations were low, nutrients released from lysed microbial cells may have been sequestered by surviving microbes or by plants resuming growth. The fast change in microbial biomass and the dependence of ER on substrate availability stress the need for high temporal resolution in future research on ecosystem carbon and nutrient dynamics at snowmelt in order to make robust models of their turnover.