Vast gaps in available information on the spatial distribution of biodiversity pose a major challenge for regional conservation planning in many parts of the world. This problem is often addressed by basing such planning on various biodiversity surrogates. In some situations, distributional data for selected taxa may be used as surrogates for biodiversity as a whole. However, this approach is less effective in data-poor regions, where there may be little choice but to base conservation planning on some form of remote environmental mapping, derived, for example, from interpretation of satellite imagery or from numerical classification of abiotic environmental layers. Although this alternative approach confers obvious benefits in terms of cost-effectiveness and rapidity of application, problems may arise if congruence is poor between mapped land-classes and actual biological distributions. I propose three strategies for making more effective use of available biological data and knowledge to alleviate such problems by (1) more closely integrating biological and environmental data through predictive modeling, with increased emphasis on modeling collective properties of biodiversity rather than individual entities; (2) making more rigorous use of remotely mapped surrogates in conservation planning by incorporating knowledge of heterogeneity within land-classes, and of varying levels of distinctiveness between classes, into measures of conservation priority and achievement; and (3) using relatively data-rich regions as test-beds for evaluating the performance of surrogates that can be readily applied across data-poor regions.
