Cross-effect of wetland substrates properties on anammox process in three single-substrate anammox constructed wetlands for treating high nitrogen sewage with low C/N

Constructing a stable and efficient anammox-driven constructed wetlands (CWs) system for efficiently treating high-nitrogen wastewater with low C/N remains a challenge, due to slow growth rate and high sensitivity of anammox bacteria to changing environmental conditions. Notably, sensitive anammox bacteria is still affected by the physicochemical properties of wetland substrates and their effects are still unknown. Therefore, three single-substrate (gravel, zeolite, and oyster shell) CWs were constructed with the goal of enhancing total nitrogen (TN) removal by anammox-driven/dominant process and determining the effect of substrate on anammox process. The gravel, zeolite and oyster shell systems achieved desired TN removal rates of 20.50, 14.25 and 22.15 g·(m2·d)-1 when influent TN load was 32.57 g·(m2·d)-1 without carbon source and costly aeration, respectively. Oyster shell system exhibited the highest removal ability and better capacity for resistance to influent nitrogen load, followed by gravel and zeolite systems (p < 0.05). Integrated analyses indicated anammox-driven/dominant process was the foremost reason accounted for the enhanced nitrogen treatment performance in all systems. The abundance of anammox gene was higher than the total abundance of denitrifying genes in the three CWs when influent TN load reached 14.85 g·(m2·d)-1. Path analysis further demonstrated anammox process was the foremost nitrogen removal pathway. [anammox] had a highest positive direct contribution (97.3%) on TN transformation rate in gravel system; [anammox/(napA+narG+nirK+nirS+nosZ)] showed highest positive direct contribution (92.4% and 97.4%) on that in zeolite and oyster shell systems, respectively. Substrate configurations significantly affected nitrogen transformation pathway and microbial communities, particularly those of anammox bacteria. Anammox genera of Candidatus Brocadia (primary anammox genera) and Candidatus Kuenenia exhibited different evolutions among the three CWs. Machine learning of Least absolute shrinkage and selection operator (LASSO) analyses showed pH, Ca, Mg, EC, and K were the key physicochemical properties of wetland substrates affecting anammox gene and anammox genera. In conclusion, Oyster shell was the optimal substrate for anammox bacteria growth.