The relationships between soil seed bank, aboveground vegetation and disturbances in an eutrophic alluvial wetland of Western France

We studied the species composition of vegetation and seed bank in an experiment with grassland and oldfield plots in an eutrophic alluvial wetland (called Marais de Redon) of Western France. In this wetland, artificial disturbances (mowing) and natural disturbances (cattle, roebucks, boars, voles...) are very frequent. In order to mime these natural disturbances, experimental disturbances were generated in March 1996 after the end of the winter flooding and the seedlings counted three months later. The seed bank, the undisturbed vegetation and seedlings emerging in disturbed quadrats were sampled. Detrended Correspondence Analysis (DCA) of the undisturbed quadrats, disturbed quadrats and seed bank samples showed significant differences of species composition. Similarity between seed bank and undisturbed aboveground vegetation was low and not very different between grassland and oldfield. Very few seedlings emerged in undisturbed vegetation both in grassland and oldfield, which indicates the importance of gaps for seed bank expression. The great majority of seedlings emerging after experimental disturbances were mainly recruited from the soil seed bank. This result contrasts with other studies where the seed bank contributed very little to the seedling flora and vegetative regrowth clearly predominated recolonisation after disturbances. In the seed bank, few species lost after succession from grassland to oldfield vegetation were still present as seeds in the soil, but in most cases, species lost were not recorded in the seed bank. The two hypotheses about changes in the seed bank during secondary succession, predicting decrease in species richness or species diversity and seed density were only confirmed for seed density parameter.