Statistische Modelle und räumlich explizite Vorhersage natürlicher Waldgesellschaften im Unteren Werraland (Nordhessen)

Abstract

Peppler-Lisbach, C.: Statistical models and spatially explicit prediction of natural woodland communities in the lower Werra region (northern Hessen, Germany). – Hercynia N.F. 40 (2007): 213 -234.Based on statistical models, the potential distribution of five natural woodland community types (Luzulo- Fagetum, Galio-Fagetum, Hordelymo-Fagetum, Carici-Fagetum, Stellario-Alnetum) is predicted for a landscape transect in northern Hessen, Germany. The logistic regression models use topographic variables derived from a digital elevation model and geological information as independent variables. Predicted probabilities of occurrence for the single vegetation types are combined to a predicted vegetation map. Model performances are evaluated and discussed. The general distribution pattern is well predicted by the models, but the performances of the Carici-Fagetum and Galio-Fagetum models are rather weak. This can be traced back to insufficient quality of some predictor variables and to lack of information about soil characteristics, respectively. In order to point out the relations between natural vegetation types and actual vegetation, the distribution of actual vegetation types with respect to the predicted areas of natural vegetation types are evaluated. The suitability of the approach for constructing potential natural vegetation maps is discussed.