Modeling complex systems like soil functioning turns out to be a challenge due to its heterogeneous structure, the number of interactions and processes entangled with each other at various levels and scales. Its heterogeneous structure requires a spatially explicit account, by coupling models through the space by using its resources, as proposed by the spatial model coupling approaches. However, these approaches raise the problem of resource conflicts : interacting models can modify resources simultaneously while the amount available cannot supply models demands. As spatial model coupling approaches do not provide proper resource distribution methods to handle that matter, inconsistencies might occur if resource conflicts are poorly managed. In order to solve this problem, we propose a novel coupling approach based on agent paradigm in which the space, associated with a coordination unit supports model exchange and conflict management. This architecture is illustrated by an UML Class Diagram that we implemented into the GAMA platform. To prove its effectiveness, we test the approach on a simple soil model in which numerous models compete for resources scattered in the space. Comparative results suggest that handling conflicts using the coordination unit shows more precise indications of resource use, compared to the model in which conflicts are mismanaged.