Incorporating connectivity into reserve selection procedures

Methods for selecting sites to be included in reserve networks generally neglect the spatial location of sites, often resulting in highly fragmented networks. This restricts the possibility of dispersal between sites, which for many species may be essential for long-term persistence. Here I describe iterative reserve selection algorithms which incorporate considerations of reserve connectivity and evaluate their performance using a data set for macroinvertebrates in ponds. Methods where spatial criteria were only invoked when ties between sites occurred did not perform significantly better than a simple greedy algorithm in terms of reserve connectivity. An algorithm based on a composite measure of species added and changes in reserve connectivity produced a reserve network with higher connectivity, but needed more sites to represent all species. A trade-off between connectivity and efficiency may be inevitable, but the costs in terms of efficiency may be justified if long-term persistence of species is more likely.