In human-dominated environments, efforts to preserve ecological connectivity between animal species habitats is an important issue. These areas are indeed concern by anthropogenic pressures leading to rapid land use and land cover changes (LUCC). This is particularly the case of urban fringes, affected by many changes (e.g. urban sprawl) specifically for several decades. Of all the methods to model the ecological networks connectivity, landscape graphs provide a relevant approach, which interest for decision support in land-use planning has already been highlighted (Foltête et al., 2014). This approach allows to represent ecological networks in a simplified way as graphs composed of nodes (species habitat patches) and links representing potential connections between these nodes according to species movement capacities. From this simplified representation, it is possible to measure the degree of networks functional connectivity from spatial metrics, among which the PC index (Saura et al., 2011) used to characterize long-distance connectivity with regard to the overall network. Analyses are typically conducted at one spatial scale. However, including multiple scales may be necessary to model ecological networks. We introduce a methodology based on landscape graphs in a multi-scale approach to measure ecological impact of LUCC in a context of urban fringes. Two input data are required, (1) a multi-date high-resolution cartography of land-use, and (2) information about life traits (habitat and movement comportments) of several species representing biodiversity. From these inputs, ecological network modelling integrates a multi-scale approach by the construction of metapatches (Zetterberg et al. 2010). Decomposing iteratively LUCC by transition type, the core of the analysis consists to compute in each case the variation of the PC index. We discuss how this methodology allows to synthesize ecological impact of LUCC and have potential application in land-use planning.