Optimal selection of a connected reserve network

Spatial considerations are important in conservation reserve design. A particularly important spatial requirement is the connectivity of selected sites. Direct connections between reserve sites increase the likelihood of species persistence by allowing dispersal and colonization of other areas within the network without species having to leave the reserve. The conventional set-covering and maximal-covering formulations of the reserve selection problem assume that species representation is the only criterion in site selection. This approach usually results in a small but highly fragmented reserve, which may not be desirable. We present a linear integer programming framework incorporating spatial contiguity as an additional site selection criterion. An empirical application to a data set on the occurrence of breeding birds in Berkshire, United Kingdom, demonstrates that site connectivity requires a significantly larger reserve. Incorporation of spatial criteria increases the computational complexity of the problem. To overcome this, we use a two-stage procedure where the original sites are aggregated first and an optimum solution is determined for the aggregate sites. Then, site selection is restricted to original sites included in the aggregate solution and a connected reserve is determined. In this particular application the above procedure generated a significantly more efficient reserve than a heuristic selection.