China is the largest producer of marine aquaculture, and in terms of the output of sea water products, China has ranked first in the world for many years. In the traditional aquaculture process, residual bait and feces produce ammonia nitrogen (NH4+-N) and nitrite nitrogen (NO2–-N), which are toxic to aquaculture organisms. Recirculation in aquaculture systems can be performed using biological filters to purify water and convert NH4+-N and NO2–-N into nitrate nitrogen (NO3–-N) which has a lower toxicity; however, its accumulation leads to chronic adverse effects on aquaculture organisms. Heterotrophic denitrification is a highly efficient biological denitrification technique that converts NO3–-N into harmless nitrogen. Because of the low carbon-to-nitrogen ratio of mariculture wastewater, additional carbon sources are required. Liquid carbon sources have the disadvantages of difficulty in controlling the dosage and the ease of the production of N2O. Solid carbon sources can effectively solve these problems. Biodegradable polymers have a high denitrification efficiency and long duration of action; but their cost is high. Agricultural-waste carbon sources have the advantages of low cost and large quantity, but they are associated with problems of slow start-up and high water color and turbidity in the early stages of operation. A mixture of polycaprolactone (PCL) and corn cob (CC) can overcome the problem of high cost and has excellent denitrification performance, so it is an excellent externally added carbon source. In this study, a denitrification system consisting of inlet tanks (180 L), peristaltic pumps, column reactors and outlet tanks (180 L) was constructed, and a composite carbon source comprising PCL mixed with CC at a mass ratio of 1:1 was used as the denitrification carbon source, and three types of influent nitrate concentrations (INC), 20 mg/L (N-20), 30 mg/L (N-30), and 40 mg/L (N-40), and three temperature conditions, 20℃ (T-20), 25℃ (T-25), and 30℃ (T-30), were set. A 90-d experiment was conducted to investigate the effects of these two factors on denitrification, including a 60-d start-up phase and a 30-d continuous phase. During this period, the inlet and outlet water samples were taken daily to determine the NO3–-N, NO2–-N, NH4+-N, and total nitrogen content using an automatic nutrient salt analyzer (QuAAtro, SEAL, Germany). The inlet and outlet water samples were taken at 61 d, 75 d, and 90 d to determine the chemical oxygen demand (COD), dissolved organic carbon (DOC), and short chain fatty acids (SCFAs). In addition, the biofilm on the carbon source was sampled at 90 d for Illumina MiSeq high-throughput sequencing. The experimental results showed that INC affected denitrification by changing the C/N ratio. In the pre-startup period of the denitrification system (0~10 d), a very high INC (40 mg/L) was detrimental to the operation of the denitrification system, and the nitrate removal efficiencies (NREs) were all lower at 19.52% (N-20), 32.67% (N-30), and 25.28% (N-40). After biofilm maturation, an increase in the INC resulted in a higher denitrification rate but not a significant increase in NRE. The effluent DOC concentration tended to decrease as the INC increased. The optimal INC was 30 mg/L, and the corresponding NRE reached 99.12% in the last 30 days. No obvious accumulation of NH4+-N and NO2–-N was observed during the process, and the decrease in its DOC and DOC concentrations was also stable and rapid. Temperature also had an important effect on denitrification. In a certain interval, the increase in temperature enhanced the denitrification performance and accelerated the nitrogen removal efficiency. During the reactor startup stage, the effluent NO3–-N concentration of each reactor decreased gradually, and the nitrate removal rate increased with an increase in temperature. The NRE of reactor T-30 (70.20%) was higher than that of reactors T-25 (69.96%) and T-20 (28.63%). The effluent DOC concentration of the system at T-30 was significantly higher than that of the other two groups, and the temperatures of 20℃ and 25℃ were more suitable. The optimum temperature was 25℃ when the microbial enzyme activity was higher, and the NRE of the system reached 99.21% in the last 30 days. In contrast, the NH4+-N and NO2–-N produced were also lower, and the organic matter utilization was high. The SCFAs produced by each system had the largest proportion of acetic acid (AC) and no detectable butyric acid, and the AC/PA(propanoic acid) were all > 1, which was favorable for denitrification to proceed efficiently. The dominant phylum at different INCs was Proteobacteria, and its abundance decreased with the increasing of INC, which was 54.46%, 39.96% and 24.77% in N-20, N-30 and N-40 groups, respectively. Temperature had a significant influence on the microbial community, and the dominant phylum was Proteobacteria at T= 30℃ and 25℃, with abundances of 55.86% and 38.85%, respectively, whereas the dominant phylum was Bacteroidota (28.87%) at T = 20℃. The dominant genus in all systems was Rhodobacter. In addition, several other active phyla in the denitrification process existed in the system. The influent NO3–-N concentration of 30 mg/L and T = 25℃ were the optimal conditions for the denitrification system using CC+PCL as an additional carbon source. The combination of CC+PCL as a composite carbon source has excellent denitrification performance, which can provide a theoretical basis for the process optimization of solid phase denitrification of mariculture wastewater.