Rice–crab coculture, as China’s third-largest integrated farming model, is pivotal for sustainable Chinese mitten crab aquaculture. This study conducted untargeted metabolomics and 16S rRNA gene sequencing on gut contents of crabs from rice fields and ponds, integrating metabolic and microbial profiles. We aimed to reveal the chemical traits of rice-field Chinese mitten crab linked to gut microbiota, providing scientific guidance for optimizing culture practices and developing microbial additives. Both groups were dominated by the phyla Firmicutes, Proteobacteria, and Bacteroidota, but the phylum Bdellovibrionota was not detected in group R. A total of 1271 distinct amplicon sequence variants (ASVs) were identified, which were annotated to 649 genera. At the ASV level, the Chao1 index for the R group (197.12 ± 17.88) was notably lower compared to the P group (288.75 ± 30.59) (p < 0.01). In contrast, the Shannon index for the R group (3.90 ± 0.06) was significantly greater than that of the P group (3.70 ± 0.06) (p < 0.01). The PCA plot demonstrated a distinct discrimination between the groups. The P group had more microbial species but was dominated by Candidatus_Bacilloplasma, resulting in uneven distribution. In contrast, the R group had fewer species but a more balanced distribution. Among 3531 metabolites identified in both groups, 865 differed significantly. Compared to P, 736 metabolites were significantly upregulated and 129 were significantly downregulated in R. Key metabolic pathways included amino acid, carbohydrate, cofactor and vitamin metabolism, signaling, and xenobiotics biodegradation. Group R had higher levels of L-leucine, L-phenylalanine, L-tyrosine, 2-amino-1-phenylethanol, choline, and pyrophaeophorbide a, which correlated with genera like Candidatus_Hepatoplasma and Aeromonas (p < 0.05), suggesting better nutritional value, flavor, and metabolic health in rice-field crabs.

This study investigated the effects of culture systems and feed types on the gut microbiota of mandarin fish (Siniperca chuatsi), using three production culture modes: pond–live bait (CE), pond–artificial feed (CS), and RAS–artificial feed (GS). Under identical pond culture conditions, CS fish fed compound feed had higher relative abundances of carbohydrate-metabolizing Bacteroidota and Firmicutes than CE. The CS group showed higher α-diversity, but β-diversity analysis indicated no significant community difference (p = 0.36), demonstrating that feed replacement only induced mild microbial shifts. LEfSe analysis uncovered 17 differential taxa: homeostasis-related genera enriched in CS, while the lipid-adapted Macellibacteroides prevailed in CE. Under identical artificial feed conditions, switching culture mode from pond (CS) to industrial RAS (GS) drastically reduced gut microbial diversity and simplified community structure. GS exhibited a sharp rise in Fusobacteriota, with Cetobacterium (64.94%) becoming the single dominant genus. α-diversity was lower in GS than CS, and β-diversity confirmed highly distinct community profiles between the two groups (p = 0.004). LEfSe detected 43 differential taxa with diverse metabolic genera abundant in CS. No PICRUSt2-inferred KEGG pathways remained significant after FDR correction in the above two comparison groups, so these functional trends are only for reference.