Selection of a minimum-boundary reserve network using integer programming

In the conservation literature, heuristic procedures have been employed to incorporate spatial considerations in reserve network selection with the presumption that computationally convenient optimization models would be too difficult or impossible to formulate. This paper extends the standard set-covering formulation to incorporate a particular spatial selection criterion, namely reducing the reserve boundary to the extent possible, when selecting a reserve network that represents a set of target species at least once. Applying the model to a dataset on the occurrence of breeding birds in Berkshire, UK, demonstrated that the technique resulted in significant reductions in reserve boundary length relative to solutions produced by the standard set-covering formulation. Computational results showed that moderately large reserve network selection problems could be solved without issue. Alternative solutions may be produced to explore trade-offs between boundary length, number of sites required or alternative criteria.