Two hypotheses of bottom-up control that predict that the species richness of Carabidae will depend either on the taxonomic diversity of plants ("taxonomic diversity hypothesis") or on the structural heterogeneity of the vegetation ("structural heterogeneity hypothesis") were tested. Plant species were classified into nine plant structural groups through cluster analysis of morphological traits (e.g. total height) at 30 early successional temporary wetlands in the east-German agricultural landscape. In a linear regression analysis, the heterogeneity of vegetation structures explained 55% of the variation in carabid beetle diversity. According to a partial correlation analysis, plant taxonomic diversity did not have a significant effect, consistent with the "structural heterogeneity hypothesis," and contradicting previous studies which concluded that plant taxonomic diversity would be the most important factor in early successional habitats. An experimental study was used to test hypotheses on the processes underlying this bottom-up control by vegetation structure: the "hunting efficiency hypothesis," the "enemy-free space hypothesis," and the "microhabitat specialization hypothesis." The composition of plant structural groups in 15 vegetation plots (1 m(2)) was manipulated, creating a gradient from dense vegetation to open plots. Subsequent pitfall catches revealed significant differences in the activity-abundances of the carabid species. Large species preferred dense vegetation plots, consistent with the enemy-free space hypothesis that large species are more vulnerable to predation on the open plots and prefer dense vegetation to escape from natural enemies. The results indicate that bottom-up control is not mediated only by plant taxonomic or functional group diversity and that vegetation structures may be more important than previously suggested.
