National audience. Spatial modelling is essential in ecology for decision-making and management. It involves modelling the heterogeneity of the landscape and the fact that different species of agents may behave at different scales, which increases the need for data, and requires their precise integration into a intertwined hierarchy of spatial and temporal scales (local/regional, seasonal/multiannual, etc.). Multi-agent modelling (MMA is a “geographically explicit” paradigm that includes heterogeneity of space (e.g. crops, meadows, forest plots), where individuals (e.g. prey, predators, etc.) move as “agents”. The latter are autonomous entities, moving in virtual space and interacting with each other and environmental patches. During the simulation, the dynamics are emerging and can be observed and quantified. These models highlight microscopic and spatial dynamics that cannot be observed with classical mathematical equation models (MME). However, coupling between MMA and MME is necessary to avoid calculation limits when giant populations (e.g. hundreds of millions of individuals) are studied to simulate ecological phenomena as treated in the Arc Jassien workshop area concerning campia-pullings. The preliminary methods and results of this approach are presented here.