For many conservation scientists, interdisciplinarity and participation can be efficient in the management of biodiversity. For both methods, new tools and new participative processes such as the so-called "co-modeling process" are required. The key questions addressed in this article are how group dynamics shape the model and why certain perspectives dominate in a process designed to be democratic. It is necessary, therefore, in order to appreciate the design and the legitimacy of the model that has been co-constructed, to address the questions of both the stakeholders' interests and their status in the process. Our case study is a co-modeling program based in a French biosphere reserve. It enabled us to highlight the key role of the mediator who had to govern social relationships and translate disciplinary jargon into a common technical language through a list of co-modeling rules.