Abstract: The lack of organic carbon in constructed wetlands leads to low efficiency of nitrogen removal. The aim of this research is to find a more efficient plant carbon source to improve the C/N ratio in constructed wetland. The optimum plant carbon source was selected according to the static carbon release of the fringe of reed, stem of reed and root of reed subject to different pretreatments, and the effect of plant carbon source on nitrogen removal was analysed by bench-scale constructed wetland experiment. Compared with other pretreatment methods, alkali-heat treatment (AHT) could boost Chemical Oxygen demand(COD) release by 10%, and release some nitrogen and phosphorus compounds ahead of time; the highest COD release of fringe of reed reached 182.00 mg/L in the early stage of AHT. Moreover, the average TN release in 144 hours of fringe of reed after AHT (6.16 mg/L) was larger than those of root of reed after AHT (4.21 mg/L) and stem of reed after AHT (2.99 mg/L) in sequence. The TN release of stem of reed was the smallest, more than half smaller than that of fringe. The releases of TN, TP, NH⁺₄-N, NO^-₃-N and NO^-₂-N of reed treated by alkali soaking and AHT were all smaller than those by simple treatment.Alkali-heat treated stem of reed has strong persistent carbon releasing capability and minimal impact on nitrogen and phosphorus accumulation, hence was chosen as the optimum plant carbon source. By adding alkali-heat treated stem of reed, the denitrification and dephosphorization capabilities of constructed wetland both enhanced markedly in the meantime of increasing inconspicuously the COD in system effluent. The average removing rates of TN and TP augmented by 61.14% and 32.53%, respectively, compared with those of control group.

Key words: plant carbon source, nitrogen removal, Chemical Oxygen demand(COD), constructed wetland, alkali-heat treatment, denitrification