Shallow lakes process large amounts of carbon (C) via gross primary production (GPP) and respiration (R), but C fluxes are highly variable among lakes. We used a two-prong approach to determine whether C fluxes differed between two alternative stable states observed in shallow lakes. First, we used a replicated whole-lake experiment where we manipulated fish densities in four experimental lakes to induce shifts from the phytoplankton-dominated state (turbid state) to a submersed macrophyte-dominated state (clear state), and determined whether whole-lake GPP, R, and net aquatic production (NAP) changed in response to the manipulation. We also compared lake metabolism in the four experimental lakes to four lakes in a turbid state and four lakes in a clear state. Second, we used sediment cores from 68 shallow lakes to test whether modern burial rates of organic C differed between lakes in clear and turbid states. Biomanipulation in the experimental lakes reduced abundance of fish and phytoplankton and increased abundance of aquatic invertebrates and submerged macrophytes. However, there was no significant change in GPP, R, or NAP. Similarly, GPP, R, and NAP did not differ among experimental lakes, turbid-state lakes, or clear-state lakes. Lastly, organic C burial in sediments did not differ between lakes in clear vs. turbid states, though variability among sites was high. High light and nutrient availability facilitate rapid transitions between two alternative groups of competing, rapidly growing primary producers in shallow lakes. These characteristics facilitate relatively uniform C fluxes at the ecosystem scale despite substantial differences in community structure.