Résumé
Landscape connectivity is considered a key issue for biodiversity conservation and for the maintenance of natural ecosystems stability and integrity. Landscape connectivity defines the degree to which the landscape facilitates or impedes movement among resource patches. A wide range of methodological approaches can be involved in such studies. Spatial distribution analyses are common tools but can hardly integrate connectivity. We do here suggestions to apply graph theory and least cost path approaches in a specific application related to common frog habitats connectivity. Amphibian's life cycle involve seasonal migrations between terrestrial and aquatic habitats which constrain them to regularly cross an inhospitable fragmented landscape matrix. Thus, there is a growing need for maintaining and restoring landscape connectivity between their habitat patches. This is especially the case for the common frog Rana temporaria, a widespread amphibian in Europe occurring in various habitat types and migrating between forest and aquatic habitats for breeding. The aim of preliminary study is to explore a method based on habitat suitability modeling and graph theory in order to analyze an ecological network. In order to assess in which manner habitat patches distribution can affect landscape connectivity between ponds, we use both configuration and distributions of suitable forest patches as model inputs. The link between common frog occurrence and forest patches configuration and distribution is defined with a probabilistic model from sampled data and relevant indices. Especially, elevation, land use distribution, distances to forest patches, distance to rivers, and landscape indices computed from forest patches distribution were shown being the main significant environmental variables influencing habitat patches distribution. In our application, we obtained then a suitable habitat patches distribution map by the use of ponds occupancy location data and maximum entropy modelling. Then, we applied least cost path modelling and graph theory approach in order to highlight the connected ponds and their importance for regional connectivity. These results emphasize the potential of maximum entropy modelling, and graph theory approach for integrating connectivity in landscape planning. The quantification of landscape matrix permeability in relation with the common frog dispersion patterns appears as limited in order to quantify edge between nodes for the design of a graph integrating ponds as nodes for a regional perspective. Nevertheless, this method combined with the use of genetic markers may be useful to assess main barriers and corridors for the common frog from a regional to a local perspective for planning. In this context, the use of genetic distances could be considered as a good surrogate to the use of least cost path as edges in a graph theoretical approach for studying connectivity.