The selection of reserves for biodiversity conservation involves the evaluation of multiple criteria, ranging from representativeness of ecological features to anthropogenic interests and spatial configuration. Among the principal spatial attributes to be considered, connectivity has received particular emphasis in response to the escalating threat of habitat loss and fragmentation. Connectivity is an intrinsic property of networks. Consequently, we have observed the gradual development of the concept of reserve networks, enlisting also tools from the mathematical branch of network theory. Here, we first outline three key aspects of reserve selection for connectivity conservation based on network analysis. 1) It may be based on the application of topological indices, which take into consideration only the geographical position of the habitat patches, or area-weighted indices, which add a premium to larger patches. 2) It may be done through single-node analysis, where the relative importance of patches is evaluated individually, or with the more efficient multi-node analysis, where we search for the optimal group of patches that best complement each other in the role of maintaining connectivity. 3) The goal of the selection may be to avoid fragmentation of the population into isolated portions, or to ensure that reachability is maintained to all habitat patches, including peripheral sites. In previous studies, we had introduced multi-node analysis to the prioritization of reserves, using fragmentation and reachability indices, but these were limited to topology only. Here, we present an improved approach where multi-node prioritization is performed with area-weighted fragmentation. We apply it to 20 bird species in Catalonia, Spain. In comparison with single-node and/or topological fragmentation, we observed here a decentralization of the selected reserve sets: they included not only the main core population, but also secondary clusters of well-connected habitat. This may potentially bring two added advantages to the reserve network: spreading of risk, and inclusion of a wider variety of local genetic profiles. We propose combining this approach with topological reachability, to account for peripheral populations and maximize accessibility to the entire network.