Habitat models and their transfer for single and multi species groups:: a case study of carabids in an alluvial forest

Environmental factors influencing the occurrence of single species and multi species groups of carabids in alluvial forest at the River Elbe were determined with habitat suitability models. Two representative species for different ecotypes present in the investigated community, Agonum livens and Pterostichus oblongopunctatus, were defined by means of a discriminant analysis. The two species differed greatly in their microhabitat distribution. Agonum livens was chosen as target species for a multi species group of wetland species that inhabited the fringes of temporary waters in the forest. In contrast, P. oblongopunctatus should represent species of deciduous forests. Using stepwise multiple logistic regression statistically significant habitat suitability models were estimated, reliably predicting the species' occurrence. A subsequent evaluation by cross-validations and receiver operating characteristic (ROC) curves indicated a high discriminatory power. A transfer of the model onto different data sets in time was applied to validate the model. Alternatively, a multi species habitat model, taking into account weighted occurrence data of the wetland species group, was estimated. In order to show that the chosen target species truly holds an umbrella effect upon the represented multi species group, we used ROC curves to indicate the transferability of the target species' habitat model for the multi species group and vice versa.