Predictive modelling of protected habitats in riparian areas from catchment characteristics

Wetland habitats are among the most threatened of all ecosystems today and still face an on-going threat despite several international agreements and national policies. In Europe, the Habitats Directive (HD) plays an important role in the protection of habitats and species of European importance. In the present study we use statistical modelling techniques and geographic information systems to explore linkages between HD Annex 1 listed habitats in wetlands and catchment characteristics, e.g. size, geology and land uses at various spatial scales (50m, 100m, 500m and whole catchment). Specifically we test if we can predict the spatial distribution of protected wetland habitats from catchment characteristics and additionally that we can identify critical variables and their spatial scale. We find that we are able to use catchment characteristics to predict the occurrence of protected habitats in riparian areas with 76% correctly classified habitats. At the catchment scale a low percentage of anthropogenic drainage together with a high percentage of forest increase the probability of having protected habitats in riparian areas, whereas at the local scale a low percentage of arable land and a high percentage of natural vegetation increase the probability of having protected habitats. We believe that the model approach used can have important implications. Firstly, the model can be used as a screening tool for identifying areas with potential conservation value. Secondly, the model can also be used as a management planning tool. Riparian areas are increasingly being used as mitigation measures to reduce loads of nutrients and toxic compounds to freshwater ecosystems. These measures may interfere with the hydrological and biochemical settings in riparian areas and threaten communities that are sensitive to eutrophication, e.g. HD Annex 1 habitats. The model can with a relatively high predictability point to areas where mitigation measures should be avoided because of conservation interests. Similarly the model can be used to identify areas that potentially can be restored successfully.