CO2-driven vegetation responses reduce projected rice irrigation water requirements across Asia

Accurate projections of irrigation water requirement are central to food security and water allocation planning in Asia’s rice systems. Most large-scale assessments of irrigation water use rely on the conventional FAO-56 reference-crop framework, which accounts for atmospheric water demand but typically treats vegetation-related coefficients as constants, thereby omitting potential changes under rising atmospheric CO 2 . Here we quantify how vegetation responses to rising CO 2 reshape rice irrigation water requirements using a process-based analytical approach that explicitly represents CO 2 -driven vegetation effects. We estimate changes of irrigation water requirements across major Asian rice-growing regions from 1982 to 2020–2020–2100 under SSP5–8.5, using both this dynamic-vegetation estimate ( I VEG ) and the conventional FAO-56 approach ( I RC ). While I RC and I VEG show broadly similar spatiotemporal variability during 1982–2020, they show diverging trajectories under rising CO 2 . Over 2020–2100, I RC increases by 6.7 mm yr⁻¹ per decade, whereas I VEG decreases by 3.6 mm yr⁻¹ per decade, yielding an end-of-century gap of ~155 mm yr⁻¹ . Attribution analyses show that I VEG additionally captures eco-physiological feedbacks that partially offset the climate-driven increase of irrigation demand as projected for I RC . Through factorial experiments, we further decompose vegetation effects into two opposing components: a CO 2 -driven physiological water-saving effect (−2.55 mm yr⁻¹ yr⁻¹) and an LAI-driven structural water-consuming effect (1.28 mm yr⁻¹ yr⁻¹), with stronger modeled net reductions associated with higher rice harvested-area density. These results highlight that vegetation representation is an important source of structural uncertainty in future rice irrigation-demand projections, and that neglecting CO 2 -driven vegetation responses may lead to higher projected net irrigation requirements under high-CO 2 conditions.