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 la rge global-scale feedbacks associated with climate-change-induced alteratio ns in arctic ecosystem C sequestration, and the substantial small-scale (1- 10 m(2)) heterogeneity of arctic vegetation and hydrology. Because the majo rity of CO2 flux data for arctic ecosystems are derived from plot-scale stu dies, a scaling routine that can provide reliable estimates of regional CO2 flux is required. This study combined data collected from chamber measurem ents of CO2 exchange, meteorology, hydrology, and surface reflectance with simple physiological models to quantify the diurnal and seasonal dynamics o f 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 hyperbol ic functions of diurnal photosynthetic photon flux density (PPFD). Daily in tegrated rates of R were expressed as an exponential function of average da ily water table depth and temperature, whereas daily fluctuations in GPP we re described as a hyperbolic function of average daily PPFD and a sigmoidal function of the normalized difference vegetation index (NDVI) calculated f rom satellite imagery. These models described, on average, 75-97% of the va riance in diurnal R and GPP, and 78-95% of the variance in total daily R an d GPP. Model results suggest that diurnal F can be reliably predicted from meteorology (radiation and temperature), but over seasonal time scales, inf ormation on hydrology and phenology is required to constrain the response o f GPP and R to variations in temperature and radiation. Using these physiological relationships and information about the spatial v ariance in surface features across the landscape, measurements of CO2 excha nge in 0.5-m(2) plots were extrapolated to the hectare scale. Compared to d irect measurements of hectare-scale F made using eddy covariance, the scale d estimate of seasonally integrated F was within 20% of the observed value. With a minimum of input data, these models allowed plot measurements of ar ctic ecosystem CO2 exchange to be confidently scaled in spacr and time.