Species richness and environmental correlates in deciduous forests of Northwest Germany

The hump-backed curve is a well-known model to explain the local variation in plant species richness and predicts highest species numbers at intermediate productivity levels. To test the validity of this model we studied various types of deciduous forest in Northwest Germany in regard to the relation between species richness and environmental factors. We recorded species numbers at three different scales in totally 48 sites and determined productivity, values of soil parameters and light conditions in all sites. Our data showed that the hump-backed model in fact best explained the relationship between species richness and productivity on a local scale. Also the relationship between species richness and several soil factors (pH, P, Ca and C/N) was best described by a unimodal curve. The amount of variation explained decreased from the largest plot size (100 m(2)) to the smaller plot sizes (10 and 1 m(2)). We developed multivariate models for the prediction of species richness by dividing the data set into a training and a test data set for generating and testing of the models. The most important variables in the multiple models were C/N, pH and productivity. The predictive abilities of the best models were fairly high. but there was still much unexplained variation in the relation between observed and expected species richness. The remaining variation could partly be, explained by factors related to the habitat configuration of the sites (patch size and isolation). (C) 2004 Elsevier B.V. All rights reserved.