Reservoirs have altered the flow regime of most rivers on the globe. To simulate the natural flow regime, experimental floods are being implemented on regulated rivers throughout the world to improve their ecological integrity. As a large‐scale disturbance, the long‐term sequential use of floods provides an excellent empirical approach to examine ecosystem regime shifts in rivers. This study evaluated the long‐term effects of floods (15 floods over eight years) on a regulated river. We hypothesized that sequential floods over time would cause a regime shift in the ecosystem. The floods resulted in little change in the physicochemistry of the river, although particulate organic carbon and particulate phosphorus were lower after the floods. The floods eliminated moss cover on bed sediments within the first year of flooding and maintained low periphyton biomass and benthic organic matter after the third year of flooding. Organic matter in transport was reduced after the third year of flooding, although peaks were still observed during rain events due to tributary inputs and side slopes. The floods reduced macroinvertebrate richness and biomass after the first year of floods, but density was not reduced until the third year. The individual mass of invertebrates decreased by about one‐half after the floods. Specific taxa displayed either a loss in abundance, or an increase in abundance, or an increase followed by a loss after the third year. The first three flood years were periods of nonequilibrium with coefficients of variation in all measured parameters increasing two to five times from those before the floods. Coefficients of variation decreased after the third year, although they were still higher than before the floods. Analysis of concordance using Kendall's W confirmed the temporal changes observed in macroinvertebrate assemblage structure. An assessment of individual flood effects showed that later floods had ∼30% less effect on macroinvertebrates than early floods of similar magnitude, suggesting that the new assemblage structure is more resilient to flood disturbance. We conclude that the floods caused an ecosystem regime shift that took three years to unfold. Additional long‐term changes or shifts are expected as new taxa colonize the river from other sources.

This study examined the long‐term changes in community assembly, resistance, and resilience of macroinvertebrates following 10 years of experimental floods in a flow regulated river. Physico‐chemistry, macroinvertebrates, and periphyton biomass were monitored before and sequentially after each of 22 floods, and drift/seston was collected during six separate floods over the study period. The floods reduced the density and taxon richness of macroinvertebrates, and a nonmetric dimensional scaling (NMDS) analysis distinguished temporal shifts in community assembly. Resistance (measured as the relative lack of loss in density) to floods varied among taxa, and the abundance of resistant taxa was related to the temporal changes in community assembly. Community resistance was inversely related to flood magnitude with all larger floods (>25 m³/s, >16‐fold over baseflow) reducing densities by >75% regardless of flood year, whereas smaller floods (<20 m³/s) reduced taxon richness approximately twofold less than larger floods. No relationship was found between flood magnitude and the relative loss in periphyton biomass. Resilience was defined as the recovery slope (positive slope of a parameter with time following each flood) and was unrelated to shifts in community assembly or resistance. Macroinvertebrate drift and seston demonstrated hysteresis (i.e., a temporal response in parameter quantity with change in discharge) during each flood, although larger floods typically had two peaks in both parameters. The first peak was a response to the initial increases in flow, whereas the second peak was associated with streambed disturbance (substrate mobility) and side‐slope failure causing increased scour. Drift density was 3–9 times greater and that of seston 3–30 times greater during larger floods than smaller floods. These results demonstrate temporal shifts in macroinvertebrate community assembly toward a pre‐dam assemblage following sequential floods in this flow regulated river, thus confirming the ecological role of habitat filtering in organism distribution and abundance. Community resistance and resilience were unrelated to shifts in community assembly, suggesting that they are mostly evolutionary properties of ecosystems as populations adapt to changing environmental (disturbance regimes) and biotic (novel colonists) conditions. As these systems show behaviors similar to dispersal‐limited ecosystems, a long‐term perspective is required for management actions targeted toward regulated and fragmented rivers.