In this paper, we argue that integrated modeling is about integrating disciplinary discourses reflecting various points of view on a given system rather than only setting up a very complex model from the start, or putting together independently designed models. In our approach, we propose an integrated modeling environment, based on ontologies, to represent and articulate the various points of view. Accordingly the proposed framework, called MIMOSA, is based on two components: 1) a declarative component based on ontologies to describe the concepts, the modeled system using the described concepts, and its dynamics using an extensible set of formalisms, 2) an executive component based on DEVS (Discrete EVent System), used to map the declarative component into. Additionally, we describe the architecture to initialize and to observe the simulation. The originality lies in the way we are using ontologies and declarative modeling to design the model, the use of DEVS, as well as in supporting the whole process within a software platform. The approach is illustrated with a socio-ecological model for assessing the sustainability of community-based regulations in Madagascar.