A new predictor of the irreplaceability of areas for achieving a conservation goal, its application to real-world planning, and a research agenda forfurther refinement

A new statistical approach is described for predicting the irreplaceability of areas (or 'sites') within a region, defined as the likelihood that a gi ven site will need to be protected to ensure achievement of a set of region al conservation targets. The paper begins by clarifying the relationship be tween irreplaceability and other conservation planning concepts such as fle xibility, rarity, endemism and complementarity. We explain why direct measu rement of irreplaceability is currently intractable for most real-world app lications, and hence the need for prediction. A new predictive approach is proposed which overcomes a number of major shortcomings of previous approac hes to predicting irreplaceability. The new approach employs the central li mit theorem to estimate the expected frequency distribution of the area of a feature protected by all possible combinations of a set of sites. This ex pected distribution is used to estimate the total number of site combinatio ns that would achieve target for the feature. The distribution is then used , for each site in turn, to estimate the number of these combinations for w hich the site of interest is a critical component. This latter number, expr essed as a proportion of the estimated total number of representative combi nations, provides a measure of the irreplaceability of a site for a single feature. Two techniques are presented for extending this approach to measur e irreplaceability in terms of multiple features. Recent application of the new predictor to regional conservation planning in eastern New South Wales and elsewhere is described, with examples. We then present results of a pr eliminary evaluation of the accuracy of the predictor. Finally, we outline a future research agenda for further validation and refinement of the new t echnique. (C) 2000 Elsevier Science Ltd. All rights reserved.