ABSTRACT
A measure of β-diversity based on the distribution of Kullback's information-theoretical divergence of single plots from the pooled sample is proposed for data on species relative abundances. Unlike other indices explicitly developed to summarise plot-to-plot variability along an environmental gradient, the proposed measure provides meaningful estimates of β-diversity independently of the presence of coenoclines. In the present study, a classical artificial data set is used to illustrate its application. Further, some analogies between the proposed measure and Whittaker's index of β-diversity are discussed.