This study investigates how environmental flows (e-flows) can be designed as dynamic operating policies to optimize long-term economic and ecosystem performance in reservoir systems. The main goal is to provide e-flow solutions that contribute to better preparedness and flexibility of hydro-systems to face multiyear stress periods, reducing the impact of water crises. The methodology framework combines a fish-flow model with a multi-objective evolutionary algorithm to construct multiple environmental water demand curves and capture the opportunity cost of different levels of ecosystem preservation. The water demand curves applied to a stochastic dynamic hydro-economic model then derive dynamic e-flow policies that balance immediate and future water use tradeoffs. The approach, termed dynamically adaptive environmental flows (DAE-flows), is demonstrated on the Paraná River Basin, Brazil, a large-scale hydropower system. Results show that the approach can adjust e-flows (coordinated with other hydro-system releases) over the time horizon, sacrificing them at certain times at the expense of some ecosystem loss, but improving long-term ecosystem functioning. A long-term approach to adaptation also yields better results for the environment without imposing a hard constraint to hydropower during droughts. Even under a drier climate change scenario, this allowed maintenance and improvement of environmental performance in most years, so during severe droughts the water could still be reallocated to hydropower but at a lesser cost to the environment.