AbstractModels for the prediction of fishery production and/or harvest based on primary production, algal biomass, or nutrients and morphometry have been effective in many lakes and reservoirs. Lake Chapala, México's largest, is located on the Río Lerma, one of México's principal rivers. It was made a reservoir in 1903 by the construction of a hydroelectric dam on the out‐flowing Rio Santiago. For the first half of this century Lake Chapala was famous for its native white‐fish (Chirostoma lucius) fishery. This fishery has collapsed. The present fishery consists of smallChirostomaspecies and the introducedOreochromis.During the past 17 years the water level has fallen by over 3 meters. Now very shallow, clay resuspension creates low visibility (Secchi<1 m). Nutrients are abundant with total phosphorus exceeding 1 mg 1−1and inorganic nitrogen exceeding 0.5 mg 1−1. Photosynthesis is limited by light and especially the unfavorable mixing depth to photic depth ratio. Models based on phytoplankton production or biomass underestimate the fishery by about one order of magnitude while a morphoedaphic model overestimates the fishery to the same extent. We sought to explain alternate pathways to support the realized fishery. Experiments suggest that bacterial prodution, with a bypass of the microbial loop, may offer a partial explanation. Management practices to increase the fishery based on an increased autotrophic base to food chain would fail without consideration of bacterial processes in this highly turbid