The greenhouse gas balance of European grasslands

Abstract. The long-term carbon balance (NBP) of grasslands is estimated by combining scarce multi-year eddy-covariance observations at ecosystem observation sites where information on carbon inputs and harvesting removals is available. Following accounting for carbon leached to rivers, we estimated grasslands to be net carbon sinks of 74±10 g C m−2 yr−1. Uncertainties arise from the small number of sites and the short measurement period. Only 11 sites, out of a total of 20 grassland sites in Europe where eddy covariance systems are installed, were set-up for estimating NBP. These 11 selected sites are representative of intensive management practice and we lack information on disturbance history, such as plowing. This suggests that the grassland NBP estimate is likely biased towards overestimating the sink, compared to the European average. Direct measurements of Net Primary Productivity (NPP) are not possible in grasslands given permanent biomass removal by grazing and mowing, uncertainties in rhizodeposition and production of volatile organic carbon compounds lost to the atmosphere. Therefore, the grassland process-based ecosystem model PASIM was used to estimate the spatial-temporal distribution of NPP, providing a European average value of 750±150 g C across extensively grazed, intensively grazed pastures, and forage production systems. In Europe the NPP of grasslands seems higher than that of croplands and forests. The carbon sequestration efficiency of grasslands, defined as the ratio of NBP to NPP, amounts to 0.09±0.10. Therefore, per unit of carbon input, grasslands sequester 3–4 times more carbon in the soil than forests do, making them a good candidate for managing onsite carbon sinks. When using the 100 yr greenhouse warming potential for CH4 and N2O, their emissions due to management of grasslands together offset roughly 70–80% of the carbon sink. Uncertainties on the European grassland greenhouse gas balance, including CO2, CH4 and N2O fluxes are likely to be reduced in the near future, with data being collected from more sites, and improved up-scaling methods.