While much research has been done on predicting the occurrence of waterlogged soils in small catchments, we need to improve our knowledge of the extension of these soils in large catchments. The aim of this study is to analyse the extent of soils with redoximorphic features in the valley bottomland domain as a function of the stream order. This study concerns a 10,000 km2 catchment (River Vilaine, Armorican massif, north-western France) where valley bottomland soils commonly associated with hydromorphic characteristics cover up to 20% of the basin area. To describe the catchment organisation, we used the stream order classification of Strahler to test the behaviour of the topographic index in a large range of landscape morphological settings. Two methods were used to define the extent of the hydromorphic zone (HZ) in each valley bottomland according to stream order: (i) A field study based on mapping the HZ according to the occurrence of redoximorphic features along 60 transects; (ii) a modelling approach linking a DEM-derived topographic index to the digitized stream network of the River Vilaine. In view of the topographic factors, progressive valley widening may represent an enhancing factor of HZ extent. Thus, simple topographic index modelling predicts an increase in waterlogging in high-order channel settings (orders 6-7). By contrast, field mapping suggests that HZ extent remains stable with increasing order and decreases significantly for high-order settings (orders 6-7). Therefore, topographic index modelling appears effective in upper catchment settings (1st, 2nd and 3rd order). On the contrary, modelling efficiency is limited in high-order settings where the indices prove to be inappropriate: in such contexts, interactions between adjacent hillslope and HZ are of secondary importance. Along the longitudinal profile of the catchment, soil material near the streams shifts from having a colluvial origin in low-order to an alluvial origin in high-order settings. In high-order settings, the fine-scale valley bottomland topography and the spatial organisation of deposits control waterlogging duration and possibly play a major role in HZ extent. Finally, the integration of stream order data should considerably improve the efficiency of modelling the spatial distribution of soils over large catchments.
