SUMMARY1. The impact of changes in subn\erged macrophyte abundance on fish-zooplanktonphytoplankton interactions was studied in eighteen large-scale (100 m-) enclosures in a shallow eutrophic lake. The submerged macrophytes comprised Potamotegon pectinatus L., P. pusillus L. and Callitriche hermaphroditica L. while the fish fry stock comprised threespined sticklebacks, Gasterosteus aculeatus L., and roach, Rutilus rutilus L. 2. In the absence of macrophytes zooplankton biomass was low and dominated by cyclopoid copepods regardless of fish density, while the phytoplankton biovolume was high (up to 38 mm' 1') and dominated by small pennate diatoms and chlorococcales. When the lake volume infested by submerged macrophytes (PVI) exceeded 15-20% and the fish density was below a catch per unit effort (CPUE) of 10 (approx. 2 fry m"^), planktonic cladoceran biomass was high and dominated by relatively large-sized specimens, while the phytoplankton biovolume was !ow and dominated by small fastgrowing flagellates. At higher fish densities, zooplankton biomass and average biomass of cladocerans decreased and a shift to cyclopoids occurred, while phytoplankton biovolume increased markedly and became dominated by cyanophytes and dinoflagellates. 3. Stepwise multiple linear regressions on log-trans formed data revealed that the biomass of Daphnia, Bosmina, Ceriodaphnia and Chydorus were all significantly positively related to PVI and negatively to the abundance of fish or PVI x fish. The average individual biomass of cladocerans was negatively related to fish, but unrelated to PVI. Calculated zooplankton grazing pressure on phytoplankton was positively related to PVI and negatively to PVI X fish. Accordingly the phytoplankton biovolume was negatively related to PVI and to PVI X zooplankton biomass. Cyanophytes and chryptophytes (% of biomass) were positively and Chlorococcales and diatoms negatively related to PVI, while cyanophytes and Chlorococcales were negatively related to PVI x zooplankton biomass. In contrast diatoms and cryptophytes were positively related to the zooplankton biomass or PVI x zooplankton. 4. The results suggest that fish predation has less impact on the zooplankton community in the more structured environment of macrophyte beds, particularly when the PVI exceeds 15-20%. They further suggest that the refuge capacity of macrophytes decreases markedly with increasing fish density (in our study above approximately 10 CPUE). Provided that the density of planktivorous fish is not high, even small improvements in submerged macrophyte abundance may have a substantial positive impact on the zooplankton, leading to a lower phytoplankton biovolume and higher water transparency. However, at high fish densities the refuge effect seems low and no major zooplankton mediated effects of enhanced growth of macrophytes are to be expected. 255 P. Schriver et alThis further emphasizes the usefulness of macrophyte refuges as a lake-restoration tool in shallow lakes, but also demonstrates the shortcomings of the method, if t...