Abstract
Mean juvenile fish abundance and fish frequency in a large lowland river during low discharge largely differed among the unvegetated and three morphologically contrasted macrophyte habitats. Single separate models revealed that juvenile fish distribution was largely influenced by trophic variables. With the exception of Leuciscus cephalus , which responded mainly to physical variables (depth and substratum), multiple regression models emphasized the importance of trophic variables for fish distribution. For Blicca bjoerkna , L. cephalus and Lepomis gibbosus , habitat shifts with respect to prey size were apparent; small juvenile fishes mainly responded to small zooplankton abundance, whereas large individuals were more influenced by the abundance of large zooplankton. Whatever the species, predictions from multiple regression models were always better for large individuals. Small juvenile fishes appeared to be less affected by the habitat variables measured, and exhibited more uniform spatial distribution. The relative importance of trophic resources and habitat physical structure among macrophyte types for fish‐habitat relationships is discussed, and the necessity of quantifying habitat structural complexity is emphasized.