Poyang Lake is an extremely important wintering ground for the Siberian crane (Leucogeranus leucogeranus, IUCN Critically Endangered), supporting about 85.7% of its global population. Previously, Siberian cranes primarily fed on tubers of Vallisneria spp. at shallow waters of nature wetlands at Poyang Lake. However, our previous studies indicated that, in recent years, a sharp decline of Vallisneria tubers has led to food composition shift of Siberian cranes. Rice grains (Oryza sativa), lotus root (Nelumbo nucifera) and Persicaria criopolitana have become important foods of Siberian cranes. Due to the temporal variation of food resources at Poyang Lake throughout the winter, the food compositions of Siberian cranes might also show temporal variation. Here, based on 45 fecal samples collected in the winters of 2020/2021 and 2021/2022, we used DNA metabarcoding to explore the temporal variation of food compositions of Siberian cranes during different wintering periods. The fecal samples were collected from five sites, covering the major habitats of Siberian cranes. We also surveyed the habitat use of Siberian cranes. We identified 29 food items in the feces of Siberian cranes, including 22 species, 6 genera, and 1 family. Our analysis results revealed different food compositions in different wintering periods. In early winter (October-November), Siberian cranes primarily fed on lotus ((47.93±45.15)% SD) and Persicaria ((47.63±43.43)%). In mid-winter (December-January), Siberian cranes primarily fed on Persicaria ((49.22±33.83)%), lotus ((26.20±32.58)%), and rice ((19.08±35.18)%). In late winter (February-March), Siberian cranes primarily fed on Persicaria criopolitana ((23.94±16.75)%), rice ((20.82±17.09)%), lotus ((19.08±37.42)%), and Poaceae ((13.07±15.18)%). The NMDS (Nonmetric multidimensional scaling) analysis separated the diet compositions of different wintering periods into different clusters. The ANOSIM (Analysis of similarities) analysis indicated that the diet composition differed significantly between each wintering periods. The numbers of food items, Shannon-Wiener diversity, evenness, and dietary niche breadth were highest in late winter and lowest in early winter. The temporal variation of food composition was similar to that of habitat use. Siberian cranes mainly used lotus ponds in early winter, rice paddies and lotus ponds in mid-winter, and the utilized proportion of natural wetlands increased in late winter. The abundance variation of food resources in environment might drive the change of food composition of Siberian cranes. Our study could improve our understanding about the temporal variation of food composition of Siberian cranes and its influencing factors, and provide a scientific basis for the formulation of conservation policies.