A method of statistical modelling of coral reef fish distribution: can it aid conservation planning in data poor regions?

Abstract

Conservation planning for biodiversity requires spatially continuous species data that encompasses the area of interest. Because of the lack of such data in most coral reef settings, we developed a statistical model to expand site-specific coral reef fish species counts to continuous spatial data. Biotic and abiotic reef characteristics were evaluated in their suitability to serve as predictor variables. These variables were then applied to our species distribution model. We used a Generalised Linear Modelling approach to determine the probability of occurrence of seven focal fish species in an unsurveyed patch or habitat. For six out of seven species we derived suitable models and predicted their presences across the whole study area of Kimbe Bay, Papua New Guinea. The resulting data were entered into the reserve selection software MARXAN, comparing the modelled data with the original species dataset in their ability to identify priority sites for conservation in irreplaceability maps. The species distribution model constituted a powerful tool to expand limited site records to be used in conservation planning in conjunction with abiotic and biotic reef information for a larger planning area. At the same time, sites with high conservation priority identified by species data were also identified by modelled data.