Predicting plant species richness in a managed forest.

This paper describes an attempt to predict ground flora species richness under various forest management scenarios. The approach is based on a geographic information system (GIS) and uses three standard map layers of topography, soils and stands to derive environmental gradients of light, nutrients, water and disturbance. A simple floristic survey provides the data necessary to relate plant distribution with environmental variables. The potential distribution of 60 understorey plant species is modelled based on the four derived gradients. The sum of these maps, i.e., the total potential diversity, is used as a proxy for the prediction of actual species richness. The model predicts high species diversity along roads and in relatively disturbed areas and low species diversity in stands with coniferous species and in stands of old, deciduous trees (mainly beech). The overall predicted pattern of species diversity corresponds well with observations made in the forests. However, the model explained only a fraction of the variation in the data set on the plot level. Dispersal effects, demographic stochasticity and biological factors are the probable causes of this. The combination of GIS-based spatial operations and fuzzy cognitive mapping proved to be an efficient way of producing gradient maps based on standard forestry maps and expert knowledge. [Copyright &y& Elsevier]