Choosing reserve sites probabilistically: A Colombian Amazon case study

This study demonstrates a method for modelling species habitats and selecting reserves for their conservation. The method has a number of advantages: It makes use of well-known techniques, is straightforward to implement, does not require species absence data, produces georeferenced digital maps for visual analysis and geographical identification, and can be adapted to any scale of analysis or data resolution. Using existing presence data for the Colombian Amazon and standard linear optimization techniques, the study models landscape level probabilities of reptiles and amphibian habitats and then uses this probabilistic habitat data to prioritize reserves for their protection. The first stage of the study uses an ecological niche factor approach to produce a series of spatially explicit probabilistic habitat suitability maps. The second stage implements an objective function that chooses appropriate sites for protection according to the suitability of these modelled habitats to support focal reptiles and amphibians. On the assumption that more suitable habitats (expressed as a probability between 0 and 1) will contain more individual numbers of amphibians and reptiles than those that are unsuitable, any objective function used with this approach will implicitly choose sites that maximize the expected number of individual animals comprising a taxa. This is in contrast to many standard selection algorithms that focus directly on species occurrences, usually seeking to cover a representative taxa at least once somewhere on the landscape.