Snow depth effects on vegetation dynamics and development of near-remote sensing techniques in high-Arctic tundra

Abstract

Snow exerts key controls on many aspects of plant ecology in the Arctic, including community composition. With climate predictions forecasting dramatic changes in winter climate and snow cover in the Arctic in the near future, it is important to improve our understanding of snow effects on plant communities in these regions. This study used a snow depth manipulation experiment established in 2006 in Adventdalen, Svalbard, Norway (78°10’N, 16°04’E) to investigate long-term effects of deepened snow on plant community composition. Two common tundra vegetation types were studied (Cassiope heath and mesic meadow) using data from three years (2015, 2020, and 2021). The study further used ‘near-remotely’ sensed vegetation indices (VIs; RGB-based indices, image based, and non-image based NDVI) to describe differences between snow regimes, years, and vegetation types. Green Chromatic Coordinate as well as image and non-image based NDVI were compared with cover of major plant groups in an initial step towards understanding the relationships between VIs and plant cover over several years and in different vegetation types. This study documented general decreases in the cover of live vascular plants, especially shrubs, and simultaneous increases in bryophytes and the forb Bistorta vivipara under deepened snow. Community changes were similar between the Heath and the Meadow vegetation types but changes were more pronounced in Heath. Near-remotely sensed VIs showed differences between snow regimes, possibly reflecting the documented vegetation change. However, relationships between VIs and plant cover were ambiguous when compared between years, vegetation types and snow regimes. The relationships generally differed in magnitude, but sometimes also direction, and were likely confounded by phenology and variations in maximum VI values between years. These findings highlight remaining challenges in the use of near-remote sensing as a tool for vegetation monitoring. Further studies should investigate the relationships between VIs and plant cover in a context of annual variations in maximum VI values, and phenological stages, as this may improve the usefulness of near-remote sensing in the future.