Trade-offs between blue water use and greenhouse gas emissions related to food systems : An optimization study for French adults ; Compromis entre l'utilisation de l'eau bleue et les émissions de gaz à effet de serre liées aux systèmes alimentaires : Une étude d'optimisation pour les adultes français

International audience ; Water use (WU) and greenhouse gas emissions (GHGe) are two main issues facing food systems. Still, they have rarely been studied together even though they are potentially conflicting because their levers for improvement are not necessarily the same. Data on food-related environmental pressures suggest that GHGe and WU can be improved jointly, but their potential conflicts and trade-offs have not been explored. This is what we studied here by a compromise programming approach, using multi-criteria non-linear optimization under a set of nutritional and epidemiological constraints. We used food consumption data of adults aged 18-64 years (n=1,456) from the French representative study INCA 3 (2014-2015) coupled with food environmental impact data from the Agribalyse ® database. A full range of scenarios was identified by prioritizing the two objectives differently, giving weight from 0% to 100%, by 5-% steps, to GHGe improvement over WU improvement.Overall, we showed that it is possible to achieve significant joint reductions in WU and GHGe relative to their observed values: across the prioritization scenarios, WU reduction ranged from -36% to -14% as its prioritization decreased, while GHGe reduction varied less, from -44 to -52% as its prioritization increased. These joint important reductions in GHGe and WU required the removal of meat consumption (beef, pork, poultry and processed meat), while the consumptions of offal and dairy products remained moderate in order to meet nutrient reference values. However, the consumption of some foods varied according to the priority given to WU over GHGe reductions (namely, vegetables, fruit juice, dairy products, eggs, refined cereal, substitutes, offal and potatoes). Fish, whole grains, and fruit remained more constant due to the epidemiological constraints used. Whatever the scenario, the modeled diets were more plant-based than the observed diet from which they differed significantly (only 23-31% of common food consumptions), and were therefore ...