Alterations to a river’s flow regime can change habitat quality and quantity at multiple temporal scales. Decreased flood frequency can, over several years, gradually change the substrate composition of riverbeds. Within a year, water abstraction for irrigation can reduce reach wetted area and change reach hydraulics. Flow-mediated impacts on physical habitat may affect fish abundance. In several parts of the world, models used to forecast how such changes to physical habitat affect fishes are not process-based and are built on several assumptions that are unlikely to be valid. These flaws lower confidence in the models, hampering their effective use in decision making. In this talk we describe the problem with certain “physical habitat” models, and present a possible solution: an ecosystem model that links (a) effects of reach substrate composition and hydraulics (depth and velocity distribution) on macroinvertebrate composition and biomass, and (b) macroinvertebrate preferences of fishes, to (c) forecast how flow-mediated habitat change might impact fish carrying capacity. The model is currently in development after a few years of data collection for parameter estimation. Some preliminary results will be presented, and we will discuss the model’s advantages and disadvantages, as well as possible next steps.