This paper reports long-term metabolic shifts of a once eutrophic river based on a 32-year, high-frequency dataset from the Loire River, France. By comparing CO2 flux estimations from water chemistry data and metabolic modeling based on high-frequency oxygen measurements, the study demonstrates a metabolic shift from a eutrophic, phytoplankton-dominated regime to an oligotrophic, macrophyte-dominated regime around 2005. Building on a novel "trophlux states" framework, the analysis of this rare long-term dataset illustrates that riverine CO2 flux and its sources are dynamic within and across years, challenging the conventional understanding of simple relationships between riverine trophic status and carbon emissions.