Abstract. Vegetation succession in riparian zones is traditionally considered to be mainly driven by allogenic hydrogeomorphological factors, while the importance of autogenic, plant-induced processes may become more significant as landform stability is achieved. This conventional notion was tested at selected point bars and cutbank edges along a salt marsh creek system in southwestern Denmark. It was predicted that, within each narrow habitat (ca. 5 m width), sites closer to creeks experience more dynamic changes in hydrogeomorphology, and hence, more changes in plant species composition than those farther away from creeks. These sites were compared in terms of the rate of vegetation succession between 2006 and 2011, using nonmetric multidimensional scaling. In the resulting two-dimensional diagram, the Euclidean distance between any two samples representing the same quadrat of different time periods was interpreted as the degree of change in species composition. The predicted differences in succession rates among sites with varying distances from the channel were not observed. Locations adjacent to the creek have experienced the most intensive allogenic fluvial-geomorphic processes. However, the rate of allogenic succession under such physical dynamism has not necessarily been greater than that of the autogenic succession that was highly dominant at locations slightly away from creeks. Allogenic and autogenic factors are probably equally, or at least simultaneously, important to vegetation dynamics even under dynamic hydrogeomorphology. Such a view is in disagreement with the conventional belief in fluvial ecology, calling for a more explicit inclusion of autogenic processes in modeling the evolution of vegetation-landform complexes in the riparian zone. In other words, a true biogeomorphic nonlinearity exists in highly dynamic fluvial systems in that output (i.e., vegetation and landform dynamics) is not necessarily a direct product of input (i.e., hydrogeomorphic effects), due to an unexpectedly significant intervening variable, autogenesis.