Small-scale distribution of species richness in a grassland (Bile Karpaty Mts, Czech Republic)

Variation in the number of species was studied in a subthermophilous grassland at a scale of 0.05 x 0.05 m during a 5-year period. The observed variance of species richness (VSR) was compared with a null model based on random distribution of species over a set of squares. It was found that distribution of species richness had more values than expected around the mean and less values at the ''shoulders''. Both tails fell within the predicted limits. Application of the procedures removing spatial dependence (random shifts, rotation/reflection method by PALMER & VAN DER MAAREL 1995) and environmental heterogeneity (patch model by WATKINS & WILSON 1992) did not change the observed pattern. Using simulations in which the number of clumps and clumping intensity were manipulated it was found that the effect of the clumped spatial pattern on VSR results in a wide range of variances. Both variance excess and variance deficit were found more frequently than expected under the null model. To test the effect of the limitation to the number of individuals per square a null model was developed which included the observed number of plant shoots per square, the observed distribution of the number of shoots belonging to individual species per square and the observed spatial distribution of the shoots. The observed VSR was still lower than that produced by the null model. Therefore, it is concluded that at a scale of 0.05 x 0.05 m plant species combine in a non-random way in the studied grassland. It is suggested that the shape of left and right ''shoulders'' of the species richness distribution may be caused by different factors, such as positive and negative covariance between species, respectively. Their simultaneous impact can generate the observed pattern in species richness.