Two PhD studies, driven by forestry issues mean to mitigate habitat fragmentation, namely retention forestry targets and green belts, are presented here. The Compiegne study, carried out by Gwendoline Percel, aims at developing guidelines to help with designing the saproxylic belt, i.e. the ecological network of saproxylic features (old-growth unharvested patches, habitat trees, forest reserves). The Cevennes study, led by Floriane Kondratow in Toulouse and supervised by Antoine Brin, relates to land use planning issue, and address the following question: « Could small ancient forest patches have a conservation interest over a wide territory? ». In the Compiegne study, the objective is to disentangle the effects on associated biodiversity (beetles, bats and bryophytes) of saproxylic habitat quantity and connectivity at the landscape scale. The sampling design focuses on two major saproxylic habitat features, whose network has been mapped ((i) cavities, and (ii) large deadwood and polypores). In the Cevennes study, ancient forest patches in the Cevennes national park have been selected to test the habitat-amount hypothesis (Fahrig, 2013), by crossing habitat quantity, isolation and matrix quality at the landscape scale. As a conclusion, difficulties to build a sampling design relevant for several taxa, contrasting in terms of dispersal ability and ecological requirements, are pointed out.