<p>In recirculating aquaculture systems, dissolved oxygen and water temperature are two critical water quality indicators that must be maintained within suitable ranges. However, due to the strong coupling between their control processes, regulating only one of these variables makes it difficult to ensure an optimal water environment. To address this issue, a state-space equation-based model predictive controller (MPC) was designed for decoupling control of dissolved oxygen and water temperature. This method first establishes differential equations that can describe the dynamic response characteristics of dissolved oxygen and water temperature, thereby exploring the coupling relationship between them. Subsequently, system identification is employed to transform the established differential equations into a set of state-space equations, which serve as the internal predictive model of the MPC, thereby reducing the computational load while simultaneously enhancing control accuracy and decoupling performance. To verify the reliability of the proposed MPC, tracking control experiments were conducted in both simulated and real aquaculture environments, with the results demonstrating that MPC outperforms the proportional integral differential (PID) controller in both scenarios. Specifically, MPC can maintain a tracking error range of ± 0.4&#xa0;mg/L for dissolved oxygen and ± 0.15&#xa0;°C for water temperature in real aquaculture environments, and its control outputs (airflow and hot water flow) are relatively small compared to traditional PID controller. These results further demonstrate that the proposed MPC offers the advantages of strong stability and significant energy savings, making it particularly well suited for multivariable water quality regulation in recirculating aquaculture systems.</p>