RESUMO
Removal of planktivorous fish is used extensively in northern temperate lakes to reduce phytoplankton abundance via enhanced zooplankton grazing. However, whether this method would work also in large subtropical highland lakes to alleviate cyanobacterial blooms is unknown. We conducted a one-year pilot in situ experiment where we removed a substantial biomass of fish in a fenced-in area, followed by a 3-year whole-lake experiment where the dominant fish species (Japanese smelt) was removed in Lake Erhai in southwest China. The fencing experiments showed that between July and November, when the biomass of the removed stock reached 4 g/m2, the zooplankton biomass inside the fence increased significantly compared to a control fence. In the full-lake experiment, we found that sustained removal of Japanese smelt led to an increase in the biomass of cladocerans (Daphnia spp. but especially of Bosmina spp.) and a significant decrease in the biomass of Cyanobacteria and Chlorophyta. Additionally, a marked increase in the ratio of zooplankton to phytoplankton biomass, as well as an increase in the body size of cladocerans, emphasising the importance of enhanced top-down control for mitigating cyanobacterial blooms following extensive fish removal. Our results reveal that removal of small fish (here Japanese smelt) can lead to a reduction of the phytoplankton and cyanobacteria biomass through a trophic cascade in highland deep subtropical lakes. Thus fish removal may be a feasible additional restoration tool to external nutrient loading reduction in such lakes.