By changing competition regimes and dominance hierarchies between plant species, invertebrate herbivores may impose strong influences on the structure of terrestrial plant communities (Brown and Gange 1989a, Carson and Root 1999, Schädler et al. 2003). These changes are commonly explained by selective feeding of herbivores on particular plant species. Selective feeding by generalist herbivores is supposed to be closely related to functional traits of plants, e.g. growth rates, resource allocation patterns, nutrient uptake and reten-

We investigated the effects of insect herbivory on a plant community of a productive old-field community by applying foliar and soil insecticides in a full factorial design. During the first 3 years of succession, insecticide treatments had only minor effects on total cover abundance and species richness. However, species ranking within the plant community was strongly affected by soil insecticide but not by foliar insecticide. Creeping thistle, Cirsium arvense, dominated the experimental plots with reduced root herbivory, while square-stemmed willow-herb, Epilobium adnatum, dominated the control and the plots with foliar insecticide. When soil insecticide was applied, cover abundance of monocarpic forbs increased and cover abundance of polycarpic herbs decreased compared to the control. However, this effect was due to a few abundant plant species and is not based on a consistent difference between life history groups. Instead, application of soil insecticide promoted persistence of species that established at the start of succession, and suppressed species that established in the following years. We conclude that below-ground herbivory reduces competitive ability of resident species and, thus, facilitates colonization by late-successional species. Hence, soil insects can exert strong top-down effects on the vegetation of productive sites by affecting dominant plant species and altering competitive balances.

Despite the myriads of herbivores in terrestrial ecosystems, the world is green (Hairston et al. 1960). Nevertheless, all these animals consume plant material and they should therefore have some influence on plant individuals, populations and communites. Hence, in order to understand community and ecosystem processes, we need a quantitative understanding of the impact of herbivores on plant communities and of its variability across ecosystems.Many ecologists will agree, that invertebrates have a lower impact on plant communities than vertebrates (Crawley 1989). However, exclusion experiments revealed that invertebrates influence biomass, species richness, competition regimes and nutrient cycling in plant communities (Bach 1994, Brown 1994, Brown and Gange 1989, Carson and Root 1999, Fraser and Grime 1997, Gibson et al. 1990). Furthermore, the results of these experiments showed considerable variability of the herbivore effect across plant communities (Hendrix et al. 1988, Fraser andGrime 1997). One explanation for this variability is the so-called Fretwell-Oksanen model. Fretwell (1977, 1987), Oksanen (1990) and Oksanen et al. (1981) suggested that the relative importance of top-down and bottom-up forces within communities changes with productivity. A low primary productivity should not be sufficient to sustain appreciable populations of herbivores (bottomup control of herbivores), whereas at high levels of productivity predators control herbivores (top-down control of herbivores). Hence, on the community level herbivore impact should show a hump at intermediate levels of productivity. Oksanen et al. (1981) and Oksanen (1990) emphasised that their model applies only to vertebrates. Invertebrates should have limited possibilities to respond on primary productivity, due to their low mobility, narrow dietary niche breadth and low costs of dormancy. Furthermore, invertebrate carni-vores exploit also detritus-feeding prey and need little energy to run their physiological machinery throughout the year. Nevertheless, the Fretwell-Oksanen model has been applied to invertebrate grazing systems. The results of some recent studies provided some support for a hump-shaped pattern along nutrient gradients (Fraser 1998, Fraser andGrime 1997). Therefore, the objective of our study was to approach the following question: Do the available experimental results suggest a general hump-shaped relationship between productivity and the impact of invertebrate phytophages on plant communities?