Most theories of plant species coexistence assume the presence of diversity maintenance mechanisms, i.e. mechanisms enhancing species richness in a community. We wished to determine whether such a mechanism was operating by establishing a field experiment in the species-rich wooded meadow of Laelatu, western Estonia. Ten to seventeen subordinate species were removed periodically (for 4 years) from 10 permanent plots of 1 x 1 m (each plot had its specific list of excluded species; 10 plots served as control). Since the removed species were all subordinate- ones, very little biomass was removed, but at the same time richness was reduced by 25-33%. If some diversity maintenance mechanism was operating, we would expect that immigration of other subordinate species would restore the original species richness. It was not possible to reject the null hypothesis of an identical immigration rate of new species into manipulated and control plots. The rate of small-scale species turnover was not affected by the removal of subordinate species. Interrelations of five richness characteristics were studied, by comparing empirical correlations among them, with those expected from a null model of random migration of species. The immigration rate of new species *appeared to be related to the number of constant species, and immigration/extinction balance related to initial richness, more strongly than predicted by the null model. In the manipulated series these relationships matched the expectation from the null model. While the results generally support the so-called species pool hypothesis (and the carousel model), it seems that species small-scale turnover depends on the richness pattem in the studied grassland. In the case of plots with artificially reduced richness no departures can be detected from the random migration hypothesis.