One objective of marine planning is the protection of biodiversity. The majority of planning strategies are based on descriptive methodologies, i.e. detecting and managing diversity hotspots. The disadvantage of such approaches is the lack of explanatory power in elucidating what facilitates the variation in diversity. The present study proposes an alternative approach: planning based on ecological processes. Correlations between 24 variables of the biophysical habitat and 4 ecological indices of fish assemblage composition were established. Variables strongly correlated to at least 1 index (adjusted r 2 > 0.6) were regarded as likely drivers of variation and used as a basis for cluster analyses to classify discrete, ecologically relevant habitat types. Variables were quantified at 2 spatial scales: individual transects (at scales of 10s of meters) and whole reefs (at scales of 100s of meters to kilometers). Habitat variables at the reef scale were not strongly correlated to ecological indices as often as variables at the transect scale. Aspects of vegetation and habitat complexity correlated strongly to most indices. In all approaches, high capacity habitat, i.e. habitat containing high values of habitat variables positively correlated to diversity indices, was defined by high abundance of Phaeophyta and high structural complexity of the reef itself. High capacity habitat was spatially confined. All cluster analyses yielded similar spatial patterns of habitat types. The statistical methodology used in the present study -identifying distinct and ecologically relevant habitat types and habitat with high capacity to support high diversity -be a useful approach for identifying candidate areas for protection in marine conservation planning.
KEY WORDS: Process-based management · Diversity · Fishes · Rocky reefs · HabitatResale or republication not permitted without written consent of the publisher Aquat Biol 12: 187-196, 2011 (Anderson & Millar 2004). Habitat variables operate at different spatial scales or grain sizes (Wiens 1989, Arias-Gonzalez et al. 2008, and it can be difficult to detect their respective effect when such variables are assessed at an inappropriate grain size (Underwood & Chapman 1998). The grain size for some habitat variables is the reef itself. Such variables include the size of the reef (MacNeil et al. 2009), effects of the surrounding matrix, e.g. distance to other reefs or other distinct habitats, or the reef perimeter, potentially facilitating edge effects (Dorenbosch et al. 2005). These habitat variables can affect fish assemblages at spatial scales of kilometres (Dorenbosch et al. 2006).Other habitat variables can be analysed at smaller grain sizes, i.e. 'patch' or 'transect' scales. Such variables, e.g. measures of structural complexity of the reef matrix (Steele 1999) or aspects of the vegetation (Bergman et al. 2001), vary substantially, on the order of metres (Flynn & Ritz 1999), and can affect abundance of fishes at similar spatial scales. Variables operating at smaller...