CARBON SEQUESTRATION UNDER CONSERVATION AGRICULTURE STUDY AND MODELLING OF CARBON DYNAMIC

This work aims to improve the existing modelling tools that allow quantifying and evaluating the CA impact on SOC sequestration with a specific link to the ARMOSA cropping system model. This model is a versatile tool to represent the carbon and nitrogen fluxes and the influence of high levels of agroecosystem processes varying in response to agricultural management and pedoclimatic conditions. To define which conservation agriculture practices impact the most on SOC sequestration and to quantify their single impact I reviewed the previous scientific research published between 1998 and 2020 with a meta-analytical approach. The results described that CA performance dramatically depends on the initial SOC stock amount, superficial crop residue retention, soil clay content and duration of the management application. Based on these initial results and on the need to get reliable model outputs in the SOC simulation, I defined which were the ARMOSA requirements that would improve the general model reliability. For this reason, I developed a specific module that accounts for the surface crop residue degradation that was not previously considered. This new module resulted highly dependent on the soil temperature and water content variation. Therefore, the model's capability to react to a variation of these conditions is a key improvement due to the rising temperatures and lack of water that will affect agriculture under the future climate change scenario. On the other hand, besides the carbon input from the surface, the core of the SOC dynamic representation occurs at the bulk soil level. Again, even though many carbon-oriented models represent in detail the bulk soil carbon dynamic, only the full cropping system model has the reliability to be identified as a decision support tool. In addition, the very last scientific modelling guides suggested that these models should ideally be verifiable using physically defined and measurable pools (namely DOM, dissolved organic matter, POM, particulate organic matter and MAOM, mineral associated organic matter) rather than only with conceptual pools as for most historical ecosystem models. For this reason, I developed a new ARMOSA 2.0 release that gathers the robustness of the classical ARMOSA version, with a new SOM dynamic conceptualization accounting for these last scientific achievements. In this last release, the central role of the microbe is worth mentioning as a “microbially explicit” approach has been integrated into the ARMOSA 2.0 version. Thus, microbial biomass now directly leads the decomposition process of the SOM pools. Finally, I tested the ARMOSA 2.0 release compared to the previous ARMOSA 1.0 version and the SALUS model. This comparison was based on measured carbon data collected across different countries and allowed me to test the performance of the new release in the simulation of conventional, minimum and no-till management. The RMSE coefficients (5.3 for ARMOSA 1.0, 5.2 for ARMOSA 2.0 and 4.3 for SALUS, on average from all the simulations) retrieved from the three models are promising since ARMOSA 2.0 performed equal or, in specific cases, even better than the other two competitors. The specific behaviour of the different pools allowed to captured specific characteristics of the CA management. The capability of this new model release to capture the SOC pathways across different soil management practices will be extremely useful in predicting how conservation agriculture can impact SOC across different climates and locations.