Geostatistical modelling of biocenosis distribution in the Strait of Sicily

As a result of increasing resource management and conservation needs, considerable attention has recently been given to the development of marine habitat classification and mapping systems. The Strait of Sicily, like other Mediterranean regions, still lacks a comprehensive classification of marine habitats and communities. The present study is a first attempt to develop a large-scale thematic mapping of marine benthic biocenosis in this area, based on scientific trawl surveys. Hauls with indicator benthic species and substrate-type records were selected from a large data set collected over a ten-year period from 1990 to 2000. Based on this information along with analysis of commercial catch composition, a biocenosis category was assigned to each sampling site. In all, nine biocenosis/facies types were identified. The indicator kriging approach was used for the spatial interpolation of data, to model the distribution of categorical variables such as biocenosis type. Each biocenosis category was codified into an indicator variable in the geostatistical model. Each indicator was studied individually with respect to its spatial variability, and a map of conditional probability of occurrence was produced for each category of biocenosis. Predictions at unsampled locations were obtained by overlaying different maps and selecting the category with the largest conditional probability. The classification procedure resulted in a univariate distribution map of the biocenosis as well as a map of local uncertainty. An independent data set was used to validate the model and to estimate the misclassification rate.