Physiological models for scaling plot measurements of CO2 flux across an arctic tundra landscape

Regional estimates of arctic ecosystem CO2 exchange are required because of the large soil carbon stocks located in arctic regions, the potentially large global-scale feedbacks associated with climate-change-induced alterations in arctic ecosystem C sequestration, and the substantial small-scale (1-10 m(2)) heterogeneity of arctic vegetation and hydrology. Because the majority of CO2 flux data for arctic ecosystems are derived from plot-scale studies, a scaling routine that can provide reliable estimates of regional CO2 flux is required. This study combined data collected from chamber measurements of CO2 exchange, meteorology, hydrology, and surface reflectance with simple physiological models to quantify the diurnal and seasonal dynamics of whole-ecosystem respiration (R), gross primary production (GPP), and net CO2 exchange (F) of wet- and moist-sedge tundra ecosystems of arctic Alaska. Diurnal fluctuations in R were expressed as exponential functions of air temperature, whereas diurnal fluctuations in GPP were described as hyperbolic functions of diurnal photosynthetic photon flux density (PPFD). Daily integrated rates of R were expressed as an exponential function of average daily water table depth and temperature, whereas daily fluctuations in GPP were described as a hyperbolic function of average daily PPFD and a sigmoidal function of the normalized difference vegetation index (NDVI) calculated from satellite imagery. These models described, on average, 75-97% of the variance in diurnal R and GPP, and 78-95% of the variance in total daily R and GPP. Model results suggest that diurnal F can be reliably predicted from meteorology (radiation and temperature), but over seasonal time scales, information on hydrology and phenology is required to constrain the response of GPP and R to variations in temperature and radiation. Using these physiological relationships and information about the spatial variance in surface features across the landscape, measurements of CO2 exchange in 0.5-m(2) plots were extrapolated to the hectare scale. Compared to direct measurements of hectare-scale F made using eddy covariance, the scaled estimate of seasonally integrated F was within 20% of the observed value. With a minimum of input data, these models allowed plot measurements of arctic ecosystem CO2 exchange to be confidently scaled in spacr and time.