Predictive Benthic Habitat Mapping in the North Sea using GIS and Statistical Methods1

Abstract

The goal of this article is to introduce a methodological concept suitable to identify ecological sea floor provinces in the North Sea. The approach consists of two main working steps: (1) geostatistical analysis of biotic and abiotic measurement data and (2) calculation of sea floor provinces by means of multivariate statistics and GIS-techniques. From bottom water measurement data on salinity, temperature, silicate and nutrients as well as on grain size surface maps are calculated via geostatistical methods. At first, the autocorrelation structure is examined and modelled by variogram analysis. The variogram models are then used to calculate raster maps by applying ordinary point kriging. These raster maps are then intersected with data on benthic communities available at 182 sites within and near the German Exclusive Economic Zone (EEZ). Further, Classification and Regression Trees (CART) is applied to derive a decision tree for the occurrence of these communities. This decision tree is applied on the geostatistically estimated raster data to predict benthic habitats within the study area.