From Mountains to Molecules: Insights into Soil Carbon Dynamics

Soil organic carbon (SOM) constitutes the largest terrestrial reservoir of organic carbon, and in a time of unprecedented climate change it is crucial to attain a better understanding of the stability and potential vulnerability of this reservoir. However, how soil organic matter dynamics (carbon stocks, turnover and fluxes) vary across different spatial scales is poorly understood. Additionally, most studies have focused on the top soils, overlooking the large deep soil carbon stocks. Furthermore, SOM is composed of a multitude of different compounds (e.g. sugars, lipids and lignin) which can be stabilized by various mechanisms (e.g. organo-mineral interactions, aggregation). Therefore, significant insights can be gained by analyzing specific carbon pools and different compounds classes within the SOM pool. Until now, knowledge of the turnover of specific carbon pools and their potential susceptibility to environmental change is limited. In this framework, radiocarbon constitutes a powerful tool for unraveling soil carbon dynamics on both decadal as well as centennial and millennial timescales. This thesis addresses these issues by investigating soil carbon dynamics in the top and deep soils across spatial scales (Chapter 2) and over time (Chapter 3), coupled to a more in-depth approach on a subset of samples. For this subset, dissolved organic carbon (DOC) provides insights into a highly dynamic SOM pool (Chapter 4) whilst fraction and lipid compound specific radiocarbon data provides insights into highly stabilized SOM pools (Chapter 5). Climatological and soil textural drivers were not found to correlate with the 14C signature changes except with mean annual temperature (MAT) with 14C in the top five centimeters of soil (MAT 1.3-9.2 C, mean annual precipitation (MAP) 600 to 2100 mm m-2y-1). Furthermore, radiocarbon signature of deep soils are surprisingly homogeneous despite highly dissimilar climatological settings. Plot-scale variability was found to be significant and of the same order of magnitude as between different geographic regions. These results hail the importance of representative (grid-based) sampling campaigns of soil-based or carbon-cycle and biomarker studies. Time-series radiocarbon dating on a subset of sites (sampled in the 1990’ and 2014) provides insights into the carbon dynamics (stocks, turnover, fluxes). Using a novel numerical modeling approach the turnover time was estimated in both the top and deep soil, resulting an unprecedentedly detailed understanding of soil carbon turnover in a variety of ecosystems (temperate to alpine). Turnover time increases and fluxes decrease exponentially with depth in all non-waterlogged soils. This implies that although deep carbon stocks are non-negligible, fluxes may be. No correlation of environmental drivers (MAT, MAP, net primary production (NPP)) or soil texture was found with carbon turnover or fluxes, implying that no single factor dominates. Turnover modeling could be used to used to pinpoint the input of petrogenic carbon, i.e. the interface between the active and passive (sedimentary) carbon cycle. In order to attain a better understanding of separate compartments of the complex SOM pool, two approaches are taken (i) dissolved organic carbon (DOC) was collected and (ii) operationallydefined fractions and compounds (lipids) were isolated from the bulk SOM. During the 2015 drought the DOC aged which implies either a net loss of stabilized carbon or a different carbon source. Considering the projected increase in drought re-occurrence, this implies that increased amounts of older carbon could be destabilized. The DOC radiocarbon measurements coupled to the time-series data revealed that DOC could be a key contributor to rapid (decadal) overprinting of SOM. Additionally, a non-negligible interaction between deep bulk SOM and DOC was found, implying that DOC is continuously in contact with bulk DOC. The compound-specific radiocarbon analyses of lipid biomarkers indicate that they could be apt markers of the most stable pool of carbon and their predominant association with the mineral-bound pool implies they could be relatively impervious to environmental change. Furthermore, lipids encompass an equal or larger spread in 14C ages than operationally-defined fractions. Lastly, biomarker isotopic values unequivocally indicate the contribution of geogenic or petrogenic carbon to soil organic matter. Overall, this thesis contributes to the understanding of the spatial and temporal patterns of soil carbon dynamics and pinpoints potentially vulnerable carbon pools and potentially highly resilient pools by exploring novel methods to assess SOM dynamics