The expanded use of water-intensive crops, such as maize, has exacerbated soil drying and water scarcity in watersheds with large irrigated surfaces. As we face climbing drought risks, water can be conserved in such areas by adopting agricultural systems that limit water deficits at the watershed level. In this study, integrated modeling was used to assess how different approaches to cultivating irrigated maize influenced water balance in France's Aveyron watershed, which experiences strong and recurrent water deficits. Ten-year simulations were performed using the MAELIA model, which can simulate at fine-scale spatial and temporal interactions among agricultural activities, the functioning of soil-crop systems, hydrological dynamics, and water resource management. Three specific scenarios were modeled in addition to the benchmark situation representing current management practices: 1) maize irrigation was carried out on an as-need basis, where all farms were equipped with a decision‑support tool for assessing water stress in real time; 2) maize monocultures were replaced by a wheat/maize succession; and 3) the two previous scenarios were combined. In the first two scenarios, irrigation withdrawal volumes declined substantially, eliminating approximately 30 % of the watershed's water deficit due to a better scheduling of the irrigation to the needs and by the replacement of maize by wheat with lower water requirements. There was also a slight increase in river flow rates in the summer and a slight decrease in annual drainage. The third scenario went even further in lowering withdrawal volumes and increasing flow rates thanks to additive effects: withdrawal volumes declined by 28 %, and river flow rates increased by 4 %. This study demonstrated that combining water-saving practices with crop diversification holds promise for limiting water deficits in irrigated watersheds while having a negligible impact on crop yields. It also demonstrated the usefulness of a multiagent integrated modeling platform that can conduct simulations at fine levels of spatial and temporal resolution on both agronomic and hydrologic aspects