[Objective] Rural ditches are major pathways through which agricultural nitrogen and phosphorus pollutants enter surface waters. Ditch ecological landscape systems typically consist of multiple interconnected units, including buffer zones, constructed wetlands, and ecological ponds. The overall spatial allocation of these systems critically determines their performance. This study addresses this issue by developing a methodological framework for optimizing the spatial allocation of ditch ecological units in plain regions, thereby overcoming subjectivity and the lack of theoretical and data-driven approaches in traditional designs. [Methods] Representative rural ditches in the Chezhegou watershed of the Huaibei Plain were selected for the spatial optimization of ecological units. A coupled modeling framework was developed in MATLAB, integrating a spatial optimization model of rural ditch ecological units with pollutant proxy functions. The computational procedure involved three key steps: (1) spatial parameters of suitable construction areas for ecological units at each node were input together with hydrological and water quality data under rainfall scenarios; (2) proxy functions were developed using multiple linear regression based on simulation data from 13 ecological units; and (3) a cost-constrained optimization model was developed to maximize pollutant removal efficiency under practical budget limitations. Using TN and TP as key performance indicators, the model aimed to optimize multiple pollutant reduction in rural ditch systems while satisfying prescribed cost constraints. Optimization was conducted using a bi-level particle swarm optimization (BPSO) algorithm to obtain the optimal allocation solution. [Results] (1) Implementation of ecological landscape systems effectively reduced the total pollutant load within rural ditch systems. Although higher construction costs improved pollutant removal efficiency, diminishing marginal returns on investment were observed. Specifically, the incremental improvement in purification efficiency per unit investment decreased progressively with additional spending.(2) Comprehensive analysis under different budget scenarios indicated that buffer zones B and C, together with wetlands D, C, G, and E, achieved superior pollutant reduction performance in the Huaibei Plain, providing valuable guidance for future rural ditch management strategies.(3) This study proposed a novel optimization framework for the spatial allocation of rural ditch ecological units. By integrating optimization algorithms with proxy functions, the framework reduces the subjectivity and empirical limitations of traditional design approaches. The method improved the scientific rigor of spatial allocation schemes and enhanced pollutant removal efficiency across multiple scenarios, demonstrating its feasibility and effectiveness. [Conclusion] The effectiveness of integrating the BPSO algorithm with proxy functions for optimizing the spatial allocation of ditch ecological units is demonstrated. This approach overcomes the limitations of previous designs, including subjectivity and limited theoretical and data support. Nevertheless, several challenges remain, including (1) high data demands, as the model requires extensive preprocessing and detailed input parameters for the target area; (2) uncertainty-related limitations, since simplifications are still necessary to accommodate uncertainties in pollutant removal simulations and surface runoff dynamics in plain regions; and (3) scalability constraints, as large-scale applications may encounter data scarcity and computational challenges due to the extensive and complex nature of rural landscapes. Future research should prioritize the development of high-resolution terrain and land-use databases tailored to rural environments, aim to streamline data requirements for the ecological unit spatial allocation module, and enhance the computational efficiency and generalizability of the optimization framework.