Flexibility, efficiency, and accountability: adapting reserve selection algorithms to more complex conservation problems

Abstract

Flexibility, efficiency and accountability are considered key attributes of good reserve selection methods. Flexibility, the ability to incorporate all the diversity of considerations, concerns and information that typically impinge on real conservation problems, is fundamental if the particulars of any given situation are to be addressed and land use conflicts are to be effectively resolved. High efficiency, the representation of the maximum diversity of the relevant features (e.g. species) at the minimum cost, is important because reserves will commonly be in direct competition with other forms of land use. Accountability means that the solutions are obtained in a transparent way. allowing others to understand why and how the result was arrived at. Because of the robustness of the general integer linear model, a remarkably rich variety of problems concerning the management and efficient use of scarce resources can be represented as problems of this type. This study starts by analysing a simple representation problem and then develops more general problems that can be applied to a variety of conservation planning exercises. It is illustrated how high flexibility can be attained, while simultaneously addressing efficiency and accountability, by modelling reserve selection questions as integer linear problems.

Conservation planning in a subdivided world

“…Originally developed for operations research, this mathematical selection method aims to represent all natural features (e.g. species or habitats) a given number of times in the smallest possible area, fewest numbers of sites, or with the lowest overall cost (Rodrigues et al 2000). Typically, analyses of this type have concentrated on the identiWcation of the minimum set of sites required to represent all species at least once.…”

“…Consequently, 25 optimal solutions were obtained for each representation target. In all cases, successive optimal solutions (where possible) were found by adding a new constraint to the site selection algorithm to exclude the preceding optimal solution (Rodrigues et al 2000).…”

A review of applications of the six-step method of systematic conservation planning

“…Therefore, by being more efficient, more land is available for other types of use. Early work used heuristic algorithms to maximize network efficiency (Pressey and Nicholls 1989), but later advances in computing power gave rise to the use of optimization algorithms (Rodrigues et al 2000c, Williams et al 2004, Moilanen et al 2009). The main debate in the literature on the issue of tools to apply to SCP has (until recently) hinged on the merits of using heuristic vs. optimization algorithms.…”

“…Proponents of optimization algorithms (also known as integer or linear programming techniques) argue that heuristic algorithms can only achieve a near-optimal solution (Rodrigues et al 2000c) and thus are not maximally efficient. Optimization is either based on finding the minimum number of sites for a given number of occurrences (the "location set covering problem", Church et al 1996; also known as the "species set covering problem", ReVelle et al 2002) or on using a fixed amount of area, and optimizing the location of the set of sites to maximize the number of species or features captured (the "maximal covering location problem", Church et al 1996; also known as the "maximal covering species problem", ReVelle et al 2002).…”

Complementarity and the selection of nature reserves: algorithms and the origins of conservation planning, 1980–1995

“…These included WorldMap (which had been developed by 1991) (Vane-Wright et al 1991), C-Plan (Pressey 1998), Target(which included cost trade-offs in a new definition of complementarity) (Walker and Faith 1998), ResNet (Kelley et al 2002;) (based on the original MNP algorithm), and Zonation (Moilanen et al 2005). Exact algorithms continued to have advocates (Rodrigues et al 2000;Revelle et al 2002;Rodrigues and Gaston 2002) but it became clear that complex spatial problems with large data sets could not be solved in reasonable time using exact algorithms. 23 To the extent that issues of spatial configuration mattered, for the decade after 1995, metaheuristic algorithms had come to stay.…”