The microbial community structure is an important indicator for evaluating the water quality of the aquaculture environment. In this study, V4 regions of 16S rRNA gene of pond (PC) and greenhouse cultured (GC) C. reevesii were sequenced. Results showed that a total of 1,993,090 high quality counts and 105,159 observed OUTs were obtained; and the Chao1 richness estimator of PC was significantly higher than that of GC groups. Beta-diversity showed that the microbiota of two groups were isolated

more »

... om each other. In addition, the correlation analysis of environmental factors showed that NO2-N, PH, PO4-P, and stocking density played significant roles in the bacterial community composition. The dominant phyla in PC groups were cyanobacteria, proteobacteria, actinobacteria, bacteroidetes, verrucomicrobia, planctomycetes; and in GC groups were proteobacteria, bacteroidetes, firmicutes, cyanobacteria, chloroflexi, actinobacteria. The functional prediction showed that the top5 Picrust prediction gene functions were protein processing in endoplasmic reticulum, retinol metabolism, proteasome, glycan binding proteins, and stilbenoid, diarylheptanoid and gingerol biosynthesis. Meanwhile, the numbers and types of KEGG pathway annotations showed a significant difference between the two cultivation environments. The prediction of bacterial phenotype implied that the GC environment is more likely to deteriorate, and turtles are more susceptible to pathogens than those of PC environment. This is the first report to explore and understand the difference of microbiota characteristics between different cultivation environments in different growth stages of C. reevesii, which will provide basic data for water quality adjustment, disease prevention, and healthy breeding of turtle.