Professor Virginia Tech Blacksburg, Virginia, United States
Abstract: Changing climate is anticipated to alter Antarctic soil ecosystems. The McMurdo Dry Valleys are of particular interest because of the long-term data available to examine responses to such changes. We are currently working on systematic and scalable measurements of microbial mats and biocrusts outside of stream and lake margins. These organisms are critical to the productivity of this ecosystem because they are the base of the food web. Thus, understanding controls over their current distribution is an essential first step in understanding the impacts of climate change on Dry Valley soil ecosystems. We aim to assess habitat suitability of biocrusts by modeling their distribution in the Lake Fryxell basin region of Taylor Valley using a combination of remotely sensed multispectral reflectance measurements predictive of biocrust occurrence and a variety of physical, chemical, and climate data derived from geospatial products, field surveys, and in-situ sampling. During the 2022/2023 austral summer, we ground truthed 64 locations covering a range of predicted moisture conditions and biocrust abundance. We collected samples for analysis of soil water content, pigments, organic matter, soil chemistry, invertebrate communities, and hyperspectral surface reflectance measurements. Our sampling sites that contained visually conspicuous biocrust (n = 17) were on average wetter (gravimetric water content of 0.082 g/g), less alkaline (pH of 8.43), and less saline (electrical conductivity of 216 µS), compared with sites that lacked biocrust (n = 47) which were on average drier (gravimetric water content of 0.035 g/g), more alkaline (pH of 9.24), and more saline (electrical conductivity of 1610 µS). The majority of sites supporting biocrusts also had seasonal snow cover in WorldView-2 and -3 multispectral satellite imagery. It is noteworthy that among this typically arid region, water from snowpacks and groundwater seeps appear to provide enough moisture to sustain biocrust presence. We are using a random forest algorithm to determine the main predictors of biocrust presence including physical (elevation, slope, aspect, water content, distance to snow), chemical (pH, electrical conductivity, soil nutrient content), and climate (air temperature, relative humidity, solar flux) drivers derived from multispectral satellite data, other geospatial products, field surveys and in-situ sampling data. This work brings us closer in our efforts to understand controls over the distribution and activity of these critical soil communities using remote sensing, an ideal technology for measuring ecosystem dynamics in Antarctic ecosystems which are particularly climate-sensitive and difficult to access.