A constructed wetland combines the functions of the substrate, plants, and microorganisms, such as filtration, adsorption, ion exchange, plant absorption, and microbial degradation, to realize the efficient purification of aquaculture tail water. Such a system has the advantages of simple processes, less investment costs, low energy consumption, high anti-shock load capacity, effective water treatment, and remarkable ecological benefits. However, due to the influence of salinity, such systems are rarely used to treat mariculture tail water. High-throughput sequencing technology provides a rapid and affordable method for sequencing millions of DNA molecules simultaneously and has become an important technique for studying the characteristics of a microbial community. To determine the NO₃ ^– -N, NO₂ ^– -N, NH₄ ⁺ -N, and TN removal efficiencies in a constructed wetland for the treatment of mariculture tail water and to evaluate the microbial community characteristics in the plant rhizosphere and different substrate layers, we conducted experiments with an integrated vertical flow constructed wetland containing Spartina alterniflora, fine sand, coal cinder, and crushed stone. We determined the nitrogen removal efficiency of a vertical flow composite constructed wetland using the tail water of grouper culture. The characteristics of the microbial community in the plant rhizosphere and different substrate layers were determined using high-throughput sequencing. In this experiment, water samples were collected at four locations every day from September to November, and the concentrations of COD, NO₃ ^– -N, NO₂ ^– -N, NH₄ ⁺ -N and TN were determined using the national standard method. Subsequently, microbial samples were collected from 16 locations and sequenced using high-throughput sequencing technology. The results of water quality analysis show that the average concentrations of COD, NO3 – -N, NO2 – -N, NH4 + -N, and TN in the outlet water were 4.00 mg/L, 0.15 mg/L, 0.16 mg/L, 0.04 mg/L, and 0.64 mg/L, respectively, yielding removal rates of 48.98%, 92.10%, 94.49% and 78.36%, respectively. High-throughput sequencing revealed a highly abundant and diverse microbial community in the plant rhizosphere and fine sand layer, in contrast to other substrates. This is because plants in a constructed wetland form an oxidative microenvironment in the reducing medium of the root zone through photosynthesis. The co-existence of aerobic and anaerobic zones provides suitable niches for aerobic, facultative anaerobic, and anaerobic microorganisms in the root zone and promotes the growth and reproduction of microorganisms. Additionally, the contact between the fine sand layer and the plant rhizosphere increases the community richness and diversity of the fine sand layer compared to those in other substrate layers. The predominant microbes were in the phyla Proteobacteria, Bacteroidetes, Actinobacteria, Chloroflexi, and Firmicutes with relative abundances of 53.7%, 11.5%, 11.9%, 6.4%, and 3.7%, respectively. The predominant microbes were of the classes Alphaproteobacteria, Gammaproteobacteria, Actinobacteria, and Bacterodia with relative abundance of 30.1%, 20.9%, 11.9%, and 10.3%, respectively. The predominant functional microbes related to nitrogen removal included those in the genera Nitrosomonas, Nitrospira, Bacillus, Pseudomonas, and Acinetobacter. Microbial metabolism function was abundant in the constructed wetland, and all samples had a similar functional composition. Microbial community composition showed little variation within the same substrate layer, indicating that microbial composition remains relatively stable within a consistent environment. The degree of difference between microbial communities in each substrate layer was smaller in the secondary wetland unit than in the primary wetland unit. This is because the majority of tail water purification occurs in the primary wetland, where the change gradient in the concentration of nitrogen, phosphorus, and other pollutants, dissolved oxygen, salinity, and other related factors is greater along the flow direction, resulting in larger changes in the microbial growth environment. Consequently, variation in the microbial community composition is more pronounced. However, the gradient of various influencing factors along the flow direction is smaller in the secondary wetland, and the variation of the microbial growth environment is also limited, resulting in less variation in the microbial community composition in the secondary wetland than in the primary wetland. These results reveal the relationship between nitrogen removal efficiency and microbial community structure. Moreover, they provide a theoretical basis for elucidating the nitrogen removal mechanism of each substrate layer in a constructed wetland, to enable the construction of wetland systems for the efficient treatment of tail water.