Abstract
Ecosystems are complex and data-intensive systems, and the ecologists still struggle to understand them in an integrated manner. Models that miss key dynamics can possibly lead to fallacious conclusions about the ecosystem fate. To address these limits and encompass whole and realistic ecosystems, we develop here a qualitative model with the help of discrete-event models. This model, based on formal Petri nets, was able to integrate biotic, abiotic and human-related components (e.g. grazing) along with their processes into the same interaction network. The model was also able to grasp ecosystem development, as defined by sharp changes of the interaction network structure itself. Furthermore, the model was possibilistic and thus rigorously computed all possible ecosystem states reached after a specific (present-day) initial state. This innovative approach in ecology then allows to rigorously and exhaustively identifying all possible ecosystem trajectories and to study their impacts and outcomes. For the first time in a realistic ecosystem, we illustrated such discrete and qualitative models in the case study of temporary marshes in the Mediterranean part of France, the Camargue delta. The model demonstrated that when marshes are exposed to extensive grazing the presence of marsh heritage species (i.e. with a conservation value) is facilitated by opening up the vegetation through various trajectories. This supports the commonly used management practices of extensive grazing to conserve certain protected habitats. The detailed analysis of the computed ecosystem trajectories allows exploring a range of recommendations for management strategies.