The Arctic is a low precipitation environment where an increase in the mean summer temperature has the potential to increase evapotranspiration and thawing of the permafrost, reducing water availability and lowering the water tables. Several studies have suggested that the temperature in the Arctic is increasing and thus drying is already occurring. Therefore, understanding how arctic ecosystems will respond to changes in climate is highly relevant. The coastal plant communities in Barrow, Alaska are strongly tied to the fine-scale variation in topography. Consequently, the position relative to the water table plays an important role in plant community composition. At the ecosystem level small changes in topography combined with water availability and increased temperatures have the potential to increase ecosystem respiration (ER). An increase in temperature would represent a positive feedback to ER by allowing more oxygen to diffuse deeper into the soil and favoring the microbial breakdown of the organic matter preserved for centuries within the permafrost. Here we report on fine-scale variation in chamber-level net CO₂ and component fluxes in randomly stratified plots situated across a naturally drained basin in Barrow, Alaska during the growing seasons of 2005, 2006, and 2007 and the proportion of ecosystem gross primary productivity (GPP) contributed by moss. The study area is a peatland dominated by wet sedges and mosses, where anoxic conditions prevail at shallow depths as result of permafrost and high water tables favoring peat accumulation (carbon storage). The season of 2006 presented the lowest mean temperature followed by 2005; 2007 had significantly higher mean temperature (Figure 1). Both GPP and ER were lower in 2006 than 2005; 2007 had similar ER to 2005, but warm temperatures throughout the season translated into carbon losses in areas dominated by mosses. The decrease in ER was particularly strong in 2006. As a result, 2006 was a stronger CO₂ sink than 2005 and 2007 (Figure 2). Moss-dominated plots located at the top of polygon rims had higher GPP and significantly higher ER in 2005 and 2006 seasons. In 2007 the moss-dominated areas experienced lower GPP and higher ER than the other vegetation types, resulting in a net carbon loss for the season of 10.4 CO₂-C g/m² season. Intermediate plots were the strongest seasonal CO₂ sink for 2005 and 2006 seasons, followed by vascular and moss-dominated plots (Figure 3). In 2007, vascular-dominated areas had the strongest carbon uptake, followed by intermediate plots. We analyzed the plant cover and found that mosses cover >50% of the ground in 89% in of the plots. LAI (m²/m²) of vascular plants was greater than 0.5 in only 44% of the plots. The lowest cover found for mosses was 30% and 0.2 LAI for vascular plants. Additionally, we measured CO₂ fluxes including only the highest light periods (0800 to 1700h ADT, Figure 4) to assess moss contribution to the ecosystem GPP. The moss layer contributed up to 48% of the daily GPP and 33% of the seasonal GPP for each vegetation type. However, mosses (106 g of CO₂ /m²/season) can contribute with up to 37% of the seasonal ecosystem GPP (289 g of CO₂ /m²/season) for all vegetations types included. These results suggest that microtopography interacts with moss cover to significantly affect the ecosystem GPP, and that short term drying can have a significant negative effect on the carbon uptake capacity of the arctic ecosystems in particular those dominated by mosses.