Genetically Improved Farmed Tilapia (GIFT tilapia, Oreochromis niloticus), a selectively bred strain of Oreochromis niloticus is a globally important aquaculture species known for its rapid growth and stress resistance. As a tropical fish species, GIFT tilapia is extremely sensitive to low temperatures. The overwintering period has become a major challenge for its aquaculture in most regions of China. To improve overwintering survival rates and ensure sustainable aquaculture, closed aquaculture systems have gained increasing attention. Closed aquaculture systems mainly include recirculating aquaculture systems (RAS) and biofloc technology (BFT). They offer significant advantages in water conservation, temperature control, and environmental sustainability. However, the microbial community dynamics and ecological interactions in RAS and BFT systems under overwintering conditions remain unclear. This study investigated the structural evolution of microbial communities in water and fish tissues under RAS and BFT systems during overwintering. High-throughput 16S rDNA sequencing was employed to characterize these communities. In this study, two treatment groups were established: RAS groups and BFT groups. Each group included three replicates. Both groups were stocked with juvenile GIFT tilapia. They were maintained under identical stocking densities and environmental management. At the beginning and end of the culture period, samples were collected from the water, biofilter media (RAS groups), biofloc particles (BFT groups), as well as from the intestinal and gill tissues of the fish. These samples were used for microbial DNA extraction and subsequent diversity analysis. In addition, key water quality parameters, including ammonia nitrogen, nitrite, and nitrate were monitored to evaluate environmental changes during the culture process. Alpha diversity analysis revealed that the Chao1 and Simpson indices of RAS water, biofilter media, and BFT water samples increased during the overwintering period. In contrast, the microbial diversity in the intestinal and gill tissues of the fish showed a decreasing trend. At the early stage of culture, Proteobacteria (75%) dominated the microbial community in the RAS groups, with Paracoccus (34%) being the predominant genus. Meanwhile, the BFT groups was mainly composed of Firmicutes (86%) and Bacillus (85%). By the end of the culture period, the proportion of Proteobacteria in the RAS system had decreased to 26%. Chloroflexi (43%) became dominant in the BFT groups, and the relative abundance of Bacillus dropped to 3%. The dominant phyla in the biofilter shifted from Proteobacteria in the early stage to Chloroflexi and Actinobacteria later on. Notably, at the end of the culture period, the microbial composition of the water and GIFT tilapia intestines in the BFT groups showed strong consistency. They shared dominant phyla including Fusobacteria, Proteobacteria, Chloroflexi, and Bacteroidota. In contrast, the microbial correlation between the water and fish intestines in the RAS groups was relatively weak. In summary, different overwintering aquaculture models significantly altered the microbial communities in the rearing water and the intestinal tract of GIFT tilapia. This study provides some theoretical foundation for the development of low-temperature overwintering strategies for GIFT tilapia, and offers practical guidance for the optimization and application of RAS and BFT systems. For overwintering aquaculture, it is recommended to choose the BFT systems as the preferred model for GIFT tilapia.