The enigma of a massive carbon imbalance – Two decades of cropland eddy flux measurements at Gebesee, Thuringia, Germany

Agriculture plays an important role in the land surface-atmosphere exchange of greenhouse gases and terrestrial carbon cycling. The determination of carbon budgets with varying system boundaries has been in the focus of environmental research, particularly at agricultural sites, which occupy roughly 50% of the global land surface. Accurate carbon balancing at crop sites is subjected to a number of uncertainties, for example through variations in vegetation cover development, carbon imports and exports by fertilization, harvest, and recurring disturbance by soil and plant management. Since 2001, continuous observations of CO₂ and water vapour net ecosystem exchange, meteorological parameters, and field operations have been carried out at the ICOS Class-1 station Gebesee (DE-Geb), located in the Thuringian Basin, Germany, making it the longest operating agricultural eddy-covariance site in Europe. In this contribution, we present two decades of flux measurements under the given land management and bioclimatic conditions. We found two opposing trends between net ecosystem and net biome production, called NEP and NBP, respectively. While the net CO₂ exchange resulted in an average annual ecosystem carbon gain of ca. 2.2 t C ha^-1 (NEP), including lateral fluxes from harvest and fertilization turned the site into a substantial deficit, indicating a loss of ca. 1.6 t C ha^-1 (NBP) per year. The mean values in repeated measurements of soil organic carbon (SOC) stocks in 2004 and 2019, however, did not show any significant changes in any of the soil horizons. Taking conservative uncertainty estimations from all components into account, the carbon loss revealed in the NBP trend over 21 years barely matched the outer ranges of the SOC uncertainty in the 0-60 cm layer. As the soil type at DE-Geb is a black earth with high carbonate content, we speculate that a certain amount of the emitted CO₂ is of inorganic origin and may partly explain the site’s carbon imbalance. An additional experiment investigating the δ¹³C signature of the emitted CO₂ to potentially prove this assumption is still ongoing. Our study highlights the importance of long-term continuous biosphere-atmosphere observations within research infrastructures like ICOS for understanding carbon cycling in agricultural landscapes, but also emphasizes the importance of site-specific knowledge to improve sustainable land management.