• floodplain grassland vegetation;
  • average water level;
  • flooding;
  • water level fluctuations;
  • dyke;
  • land use;
  • ordination


The effects of floodplain fragmentation by dykes on grassland vegetation were evaluated through field studies along the Middle Elbe River (federal states Saxony-Anhalt and Brandenburg, Germany). Plant species composition was examined in 206 sites between 1996 and 1998 in the entire floodplain, which can be divided into the floodplain types ‘recent floodplain’, ‘older floodplain’ (which is separated from the recent one by dykes) and the ‘margin of the floodplain’ (which is the part of the older floodplain that forms the boundary and is furthest from the river). Dynamics in hydrology were examined weekly between November 1996 and February 1999 with the help of 40 water level wells which were installed near the studied sites. The hydrological parameters ‘average water level’, ‘average groundwater level’, ‘flooding duration’, ‘flooding depth higher 50 cm above soil surface’ and ‘standard deviation of the water level line’ were calculated to characterize the considered floodplain types and to relate species composition to hydrology. Furthermore several parameters of current management of the vegetation were recorded to evaluate the importance of land use versus hydrology for floodplain grasslands.

Detrended and canonical correspondence analysis (DCA, CCA) were used to identify major environmental gradients governing the vegetation and to determine if there is a relationship between the different locations within the floodplain, variation in species composition, and gradients of measured environmental variables.

The results indicate that the vegetation is closely related to a combination of water level fluctuations, which are different among the floodplain types, and soil moisture, while type and intensity of current management are not important in this context.

The results of contingency tables underline the significance of dykes for the occurrence and absence of individual species among the floodplain types. The observed patterns can also be explained by the different hydrological properties of the recent and older floodplain as the results of logistic regression reveal. Furthermore, disturbance and dispersal processes and their alteration by dykes have to be taken into account to explain the pattern of species occurrence.

Partial ordination detected residual differences in vegetation among the different floodplain types after accounting for the effects of the measured environmental variables. Grain size distribution is discussed as a further factor that might influence species composition. Copyright © 2004 John Wiley & Sons, Ltd.