Agent Based Modeling (ABM) is a spatially explicit oriented paradigm that models space as a heterogeneous environment (e.g. crop, grassland, forest patches), where individuals (e.g. predators, preys, etc.) move, all represented as “agents”. The latter are autonomous entities, which move on the virtual space and interact with each other and the environmental patches. During simulation, dynamics emerge and can be observed and quantified. Agent based modeling is versatile and keeps proximity with the real system: data such as Geographical Information System (GIS) could be easily included in the model; model architecture is close to the ecological expert knowledge; operational outputs can be tracked and explored for, e.g., decision making. Agent based modeling permits to produce large-scale models composed of several hundred thousand interacting agents. These models bring to light microscopic and spatial dynamics that could not be observed with Equation Based Model (EBM). Nevertheless, computational limits are often overflowed. This problem is frequent in ecology where giant populations (e.g. hundreds of thousand millions of individuals) are at stake. Here we propose to couple ABM to EBM in a multiscale model to overcome this limit and simulate the spreading of a montane water vole population in the Haute-Romanche valley (France). The colonization of an Alpine valley where the species was absent before was observed from 1998 to 2010 (Halliez et al. 2015) reaching peaks of 500-1000 voles/ha, and a total population estimated at more than 2,500,000 individuals. EBM allow quantifying the growth of such a large population but cannot take spatial heterogeneity (land use, soil, elevation) and individual vole dispersal into account. Vole are usually sedentary and establish themselves in the near vicinity (some 10 meters) where they are born, but sometimes, when juvenile individuals suffer of neighborhood stress, they move away on long distance (some hundred meters or kilometers). This specific behavior can be modeled using ABM. A grid was built from a GIS, altitude data imported and slopes computed. Each cell (30m X 30m) describes an age-structured population of voles, based on a logistic model. When the population exceeds a threshold, agents are created to model short distance and long distance individual dispersers. Each agent moves according to the topology within a distance range as long as they do not find a cell with a population below the threshold. There, the agent is aggregated in the population. Additionally, survival as a function of age in the population, of status (e.g. moving agent) and reproduction season length can be parametrized. The next step will be to simulate density dependent mortality (e.g. predators and diseases) in this system in order to explore the conditions by which travelling waves can occur.