Summary Thesis ‘Studying biosphere-atmosphere exchange of CO2 through carbon-13 stable isotopes’ Ivar van der Velde Making predictions of future climate is difficult, mainly due to large uncertainties in the carbon cycle. The rate at which carbon is stored in the oceans and terrestrial biosphere is not keeping pace with the rapid increase in fossil fuel combustion and deforestation, resulting in an increase of atmospheric carbon dioxide (CO2). To gain a better understanding of the global carbon cycle we need to combine multiple sources of data into one consistent analysis, such as, forest and agricultural statistics, satellite data, atmospheric and ecological observations, and mechanistic models. This thesis describes fundamental research on some of the key components of the terrestrial carbon cycle, i.e., gross primary production (GPP) and terrestrial ecosystem respiration (TER) of CO2, which forms the key to improved prediction of net exchange. Droughts have been shown to strongly influence this exchange, and to interpret these responses adequately we have turned to a large collection of new atmospheric observations of CO2, and its 13C isotope (13CO2), to constrain key model components. In Chapter 2 we studied the global budget of atmospheric CO2 and the ratio of 13CO2/12CO2 (δ13C) and investigated the main terrestrial drivers of interannual variability (IAV) responsible for the observed atmospheric δ13C variations. In this chapter we introduced the SiBCASA biogeochemical model that we provided with a detailed isotopic discrimination scheme (to calculate the natural preference of 12CO2 over 13CO2 in uptake processes), separate 12C and 13C biogeochemical pools, and satellite-observed fire disturbances. This model was able to calculate uptake of 13CO2 and 12CO2 and produced return fluxes from its differently aged carbon pools, contributing to the so-called disequilibrium flux. Our simulated terrestrial isotope processes, plant discrimination and disequilibrium, closely resembled previously published values and similarly suggested that discrimination variations in C3 type plants and year-to-year variations in C3 and C4 productivity are the main drivers of IAV. The year-to-year variability in the terrestrial disequilibrium flux was much lower than required to match variations in atmospheric observations, under the common assumption of low variability in net ocean CO2 exchange, constant discrimination, and a closed CO2 budget. It was unclear how to increase IAV in the terrestrial biosphere, which suggested that SiBCASA missed adequate drought responses resulting in a latent isotope discrimination and variability in C3/C4 plant productivity. Implementation of carbon isotope cycling, biomass burning, and SiBCASA’s drought response were closely studied in Chapter 3. Our biomass burning emissions were similar as in CASA-GFED; both in magnitude and spatial patterns, and the implementation of isotope exchange gave a global mean discrimination value of approximately 15 ‰, and varied spatially depending on the photosynthetic pathway in the plant. These values compared well (annually and seasonally) with other published results. Similarly, the size of the terrestrial isotopic disequilibrium was close to that of other studies. As plants experience drought stress, they respond by closing their stomata to prevent the loss of water. This process also inhibits the uptake of CO2 and reduces the isotope discrimination against 13CO2 molecules. We found that the amplitude of drought response in SiBCASA was smaller than suggested by the measured isotope signatures. We also found that a slight increase in stomatal closure for large vapor pressure deficits amplified the variations in the respired isotope signature. Finally, we saw the need for modified starch/sugar storage pools to improve the propagation of isotopic discrimination anomalies to respiration on short-term time scales. In Chapter 4 we developed a multi-tracer inversion system to interpret signals in atmospheric CO2 and δ13C observations simultaneously. We wanted to know whether drought stress in plants can induce changes in atmospheric δ13C and whether they are interpretable. Using inverse modeling we were able to refine the discrimination parameter for plants as it reflected detectable variations in atmospheric δ13C. The results showed that the isotope discrimination values were consistently smaller during large severe droughts in the Northern Hemisphere, exceeding the estimates from SiBCASA (i.e., a larger reduction). Decreased discrimination suggested an increase in the regional intrinsic water use efficiency, which was also recorded at a large number of measurement sites. The IAV in net ecosystem exchange was relatively insensitive as we allowed the variability of the discrimination parameter to increase more than 8-fold, but it also allowed significant correlation between annual net exchange and discrimination. This study suggested a larger effect of droughts on discrimination than previously thought and that the treatment of drought response in biosphere models needs to be improved. Carbon cycle research is far from complete as many components are still largely uncertain, which prevents us from making better predictions of future climate. This thesis, however, highlights the importance of isotope observations to assess and improve biogeochemical models, especially with regard to the allocation and turnover of carbon, and responses to droughts.