1. Based on a comprehensive data set collected monthly during 8 years (1997–2004), we evaluated the effects of mechanical removal of Eichhornia crassipes on the limnological characteristics and algal biomass of a polymictic shallow tropical reservoir.
2. Interrupted time series analyses indicated that the limnological responses to macrophyte removal can be classified as an ‘abrupt permanent impact’ implying that the overall mean of the time-series shifted promptly after intervention. These analyses indicated a significant increase for pH, total phosphorus, total phytoplankton and cyanobacterial biomass, and a decrease in water transparency and CO2 concentrations in the surface water; also, the increase in water stability, increase of bottom soluble reactive phosphorus (SRP) and decrease in bottom oxygen levels.
3. Cyclic anoxic periods previously observed during springs and summers were replaced by a persistent period of anoxic conditions in the sediment overlying water. Anoxic conditions were suitable for SRP release from sediments. Heavy cyanobacterial blooms became more persistent, maximum biomass (4229 mm3 L−1) was 30 times larger, the blooms frequently reached 2 m and sometimes the bottom of the reservoir, contrasting to the preremoval period in which it reached at most 1 m deep.
4. The long-term P dynamics in the system, initially driven by allochthonous nutrient loadings were replaced by internal ecological processes. Water hyacinth removal markedly accelerated the process of eutrophication due to internal feedback mechanisms, leading to a switch to a more turbid state. Biological feedback mechanisms were driven by cyanobacterial blooms by enhancing water stability, oxygen anoxia at the bottom and by increasing suitable conditions for P internal loading. These data support the hypothesis of the role of cyanobacterial blooms as an important factor impairing water quality and driving the ecosystem towards a stable degraded state.
5. These findings have important implications for the restoration of shallow stratifying eutrophic lakes, as the alternative degraded state is most likely to occur when compared with their non-stratifying counterparts. Moreover, feedback mechanisms in tropical and subtropical shallow lakes seem to be stronger than in temperate ones, as stratification events are more likely to occur over the year, intensifying system resilience to restorative strategies.