China's diets are undergoing a profound transformation, and carbon storage loss generated by dietary structure change is increasingly serious, accompanied by concerns about sustainable diets and food supply-demand conflicts. This paper designed a dual 'material-environmental' framework based on the dietary structure classification, around which we conducted the heterogeneity analysis of food supply-demand patterns and land carbon loss, and then tracked hotspot diets of trade-generated carbon loss through the multi-regional input-output (MRIO) model. The research found that 31 provinces could be classified into five clusters with obvious spatial agglomeration patterns, most of which still had a high proportion of staches, but also showed an evolutionary trend towards diversified compositions. Driven by dietary structure, the national food supply-demand match under the material accessibility dimension showed barely balanced overall, while in terms of environmental effects, ecosystems lost significant amounts of carbon stocks, with losses of 331 and 950 Tg1 of vegetation carbon and soil organic carbon, respectively, which together accounted for 14 percent of the total carbon stocks in the vegetation and topsoil carbon pools. The clusters to which Henan, Jiangsu, Zhejiang and Guangdong belong were tracked as hot dietary clusters exhibiting hotspots in inter-regional trade, contributing to carbon loss in external provinces through trade linkages. Heilongjiang and Inner Mongolia were at a low level of supply-demand balance but played the role of a carbon storage loser, while Guangdong, Zhejiang, and Jiangsu not only possessed higher balance levels but also pulled more external carbon loss. China's food system should optimise internal cropping structures and monitor external land use conversion. It is equally important to track food consumption side rights and responsibilities.
