Terrestrial Arctic ecosystems play a key role in the global carbon (C) cycle, as they store a large amount of organic matter in permafrost. Among regions with continuous permafrost, Svalbard has one of the warmest permafrost and may provide a template of the environmental responses of Arctic regions to future climate change. We analyze the CO2 fluxes at a polygonal tundra site in Adventdalen (Svalbard) during one full growing season across a vegetation and environmental gradient to understand how the interaction of different abiotic (thaw depth, ground surface temperature (GST), soil moisture, photosynthetic active radiation - PAR) and biotic (leaf area index (LAI), and plant phenology) factors affect the CO2 fluxes and identify the drivers of Net Ecosystem Exchange (NEE) and Ecosystem Respiration (ER). Three distinct periods (early, peak, and late) characterized the growing season based on plant phenology and the main environmental conditions. Comparing early, peak and late season, both NEE and ER exhibited specific patterns: ER shown high values since the early season, only slightly increased at peak, and then decreased drastically in the late season, with GST being the most important driver of ER. The drivers of NEE changed during the season: thaw depth, PAR and GST during the early season, LAI at peak, and PAR during the late season. These data allow to highlight that the thawing and freezing of the upper part of the active layer during the early and late season controls ER, possibly due to the response of microbial respiration in the upper part of the soil. Especially during the late season, despite the fully developed active layer (reaching its highest thawing depth), the freezing of the uppermost 2 cm of soil induced the drastic decrease of the respiratory efflux. In addition, the seasonal C balance of our plots indicated a seasonal source at our plots, in apparent contrast with previous eddy covariance (EC) measurements from a wetter area nearby. This difference implies that drier ecosystems act as sources while wetter ecosystems are sinks, suggesting that a drying trend in polygonal tundra could switch these ecosystems from CO2 sinks to sources in a feedback to future climate change.

Climate models project considerable ranges and uncertainties in future climatic changes. To assess the potential impacts of climatic changes on mountain permafrost within these ranges of uncertainty, this study presents a sensitivity analysis using a permafrost process model combined with climate input based on delta-change approaches. Delta values comprise a multitude of coupled air temperature and precipitation changes to analyse long-term, seasonal and seasonal extreme changes on a typical low-ice content mountain permafrost location in the Swiss Alps. The results show that seasonal changes in autumn (SON) have the largest impact on the near-surface permafrost thermal regime in the model, and lowest impacts in winter (DJF). For most of the variability, snow cover duration and timing are the most important factors, whereas maximum snow height only plays a secondary role unless maximum snow heights are very small. At least for the low-ice content site of this study, extreme events have only short-term effects and have less impact on permafrost than long-term air temperature trends.