Effectiveness of alternative heuristic algorithms for identifying indicative minimum requirements for conservation reserves Original Research Article

We compared the results of 30 heuristic reserve selection algorithms on the same large data set. Twelve of the algorithms were for presence-absence representation goals, designed to find a set of sites to represent all the land types in the study region at least once. Eighteen algorithms were intended to represent a minimum percentage of the total area of each land type. We varied the rules of the algorithms systematically to find the influence of individual rules or sequences of rules on efficiency of representation. Rankings of the algorithms according to relative numbers or areas of selected sites needed to achieve a specified representation target varied between the full data set and a subset and so appear to be data-dependent. We also ran optimizing algorithms to indicate the degree of suboptimality of the heuristics. For the presence-absence problems, the optimizing algorithms had the advantage of guaranteeing an optimal solution but had much longer running times than the heuristics. They showed that the solutions from good heuristics were 5–10% larger than optimal. The optimizing algorithms failed to solve the proportional area problems, although heuristics solved them quickly. Both heuristics and optimizing algorithms have important roles to play in conservation planning. The choice of method will depend on the size of data sets, the representation goal, the required time for analysis, and the importance of a guaranteed optimal solution.