As mentioned above, ammonia removal performance was similar to that AZD1283 of TOC. The removal efficiencies of ammonia were maintained above 90%, with average effluent concentrations both below 0.2 mg L−1. However, TN removal efficiencies declined after operation for 2 months in phase II, which was mainly attributed to the deficiency of carbon source because few nutrients released from P. communis. Meanwhile, the removal efficiencies of organics in both reactors increased to above 70%. It is well accepted that using plant as carbon sources for denitrification is economic and effective. However, the addition of plant carbon sources may introduce secondary pollution because of the unstable and uneven releasing of nutrients. It has been reported that some kinds of organics are not suitable to be used to denitrification ( Gibert et al., 2008), even caused some harmful effects, such as incomplete nitrogen removal (mixture of wood chips, shredded bark and topsoil; willow wood chips), nitrogen releasing (mixture of wood chips, shredded bark and topsoil; willow wood chips), excessive reduction of nitrate to ammonium rather than nitrogen gas (compost obtained from the biological decomposition, organic wastes of wood trimmings, leaves, rotten vegetables and food scraps) ( Gibert et al., 2008), nitrite accumulation (glucose and methanol) ( Srinandan et al., 2012). However, the addition of P. communis did not cause these problems.