Seasonal variations of pollutants removal and microbial activity in integrated constructed wetland – microbial fuel cell systems

This study investigated the seasonal variations of pollutants removal and microbial activity in constructed wetland – microbial fuel cell systems (CW – MFCs). The results showed that the atmospheric temperature signi ﬁ cantly in ﬂ uenced the bioelectricity generation and removal of organics and nitrogen in CW – MFCs by primarily in ﬂ uencing the microbial enzymatic activity. The electricity output of CW-MFCs was extremely low below 5 (cid:1) C, and reached the maximum above 25 (cid:1) C. The organics and nitrogen removal of closed-circuit CW – MFC reached the highest in summer and autumn, followed by spring, and decreased by an average of 10.5% COD, 14.2% NH 3 -N and 10.7% TN in winter, demonstrating smaller seasonal ﬂ uctuations compared to open-circuit CW – MFC in which the difference between summer and winter was 13.4% COD, 15.1% NH 3 -N and 15.1% TN. Even at low temperatures, the MFC current could enhance the enzymatic activity and stabilize the growth of microorganisms on the electrodes, moreover, the closed circuit operation can promote the bacteria diversity on CW – MFC anodes as well as the abundance of electrogens on CW – MFC anodes and cathodes, and thus reduce the adverse effect of cooling on organics and nitrogen removal in CWs. However, neither MFC nor temperature had a signi ﬁ cant in ﬂ uence on phosphorus removal in CW – MFCs. MFC reduced adverse effects of cooling on organics and nitrogen removal in CWs.

layer cover on the CW could keep the temperature inside the wetland constantly above 6 C, even when the atmospheric temperature dropped to À8 C during winter, which provided effective system thermal insulation and maintained high pollutants removal (95.0% BOD 5 , 84.6% NH 3 -N, and 88.2% TP) in the freezing winter period. Kadlec & Wallace () reported that during the winter icing period, wetlands could be operated under a frozen layer by adjusting the water level, and a relatively high water temperature would be maintained inside the wetlands. Wang et al. () summarized that hybrid CWs consisting of various types of CWs arranged in series possess higher treatment performance than a single CW in a cold climate.
The interest in integrated constructed wetland-microbial fuel cell (CW-MFC) systems has increased due to their ability to produce electricity and enhance the wastewater treatment efficiency (Doherty et

Experimental setup
Two parallel integrated up-flow CW-MFC reactors (closed-

Water sampling and analysis
Water samples were collected between 9.00 and 10.00 am every 3 days from the effluent, and were immediately analyzed in the lab for COD, NH 3 -N, TN, PO 3À 4 -P, and TP using a Digital Reactor Block 200 and a HACH DR 2800 spectrophotometer, according to the standard procedure provided by HACH Company, USA. Specifically, COD was measured by the quick digestion spectrophotometry method, NH 3 -N was measured by the salicylate method, TN was measured by the persulfate digestion method, PO 3À 4 -P was measured by the molybdovanadate method, and TP The atmospheric temperature was monitored using a temperature recorder (provided by Hangzhou Sinomeasure Automation Technology Co., Ltd, China), and the data were collected every 30 minutes. The pollutants removal rate (R) was calculated as follows: where C i and C e are the mean influent and effluent concentration (mg/L), respectively.

Bioelectricity measurement and analysis
The output voltage (U) was collected using a multi-channel data logger (Model CT-4008-5v10 mA-164, Shenzhen Neware Electronics Co., Ltd, China) and recorded by a computer at intervals of 30 min. The output current (I ) was calculated by Ohm's law. The volumetric power density (P d ) was calculated as follows: where P is the power (mW), V is the total volume of the CW-MFC reactor (m 3 ), U is the output voltage (V), and R ex is the external resistance (Ω).

Microbial sampling and analysis
Microbial sampling was conducted once during spring, summer, autumn and winter, respectively, between 9.00 and 10.00 am. The anode samples were collected by cutting a small amount of carbon fiber at a depth of 10-15 cm from the substrate surface, and samples from the five carbon fiber brushes were uniformly mixed as one sample. The cathode samples were collected from the biomass attached on the graphite plates, and samples from the four graphite plates were uniformly mixed as one sample. In order to minimize the disturbance of sampling to the cathodes, only 1.0 × 1.0 cm biomass was gently scraped away from the graphite plate, accounting for 1% of the total area of the plate. The substrate (lava) samples were collected at a depth of 10-25 and 25-40 cm, and were uniformly mixed as one sample.
During substrate sampling, the influents dosing was stopped and the water in the reactor was drained, and then the substrate sampling holes (10 cm in diameter, Figure 1) were opened to collect substrate samples. Once the sampling was completed, the influents dosing was restarted.
All microbial sampling was conducted between 9.00 and 10.00 am.
The enzymatic activity was determined by the analysis of dehydrogenase and catalase activity. Samples were freeze-dried first, and then ground and screened through a 16-mesh sieve for determination of enzymatic activity.
Dehydrogenase activity (DHA) was measured with the triphenyl-tetrazolium chloride (TTC) method, for which 1 g (dry basis) of sample was cultured using 1 mL TTC solution

Data analysis
One-way analysis of variance (ANOVA) was conducted to detect significant differences in the treatment efficiency between the two CW-MFC reactors, followed by a Duncan post hoc test (P < 0.05). All of the statistical analyses were conducted using SPSS and Origin software.

RESULTS AND DISCUSSION
As illustrated in Figure  found that the biodegradation rates of a petroleum hydrocarbon mix (i.e. phenanthrene and benzene) and maximum power density in MFCs were both two times higher at 40 C (97.10% and 1.15 mW/m 2 anode, respectively) than those at 30 C, but were four times lower when the operating temperature was raised to 50 C. It has been reported that the most suitable temperature for the electro-  Pollutants removal in the CW-MFC systems under different seasons is provided in Figure 5. The average removal rates in closed-circuit CW-MFCc under spring, summer, autumn and winter were, respectively, higher than that in open-circuit CW-MFCo by approximately 11.2, 9.4, 10.0 and 12.4% for COD,4.8,4.8,and 12.42,11.19,11.60 and 15.6% for TN. It can be found that the gap between CW-MFCc and CW-MFCo in terms of organics and nitrogen removal widened in winter. For closed-circuit CW-MFCc, the organics and nitrogen removal in summer did not significantly (P > 0.05) differ from that in autumn, but was significantly (P < 0.05) higher than that in spring by an average of 3.2% COD, 4.2% NH 3 -N, 3.5% TN, and that in winter by an aver- Many studies have found that the physio-chemical processes of substrates were mainly responsible for P removal in CWs, including sedimentation, filtration, interception, adsorption, absorption, precipitation, ion exchange, and complexation reactions (Lan et al. ). From the results of this study, it can be summarized that the primary method of P removal in CW-MFCs was also the physiochemical processes of substrates.
Overall, the MFC not only significantly (P < 0.05) improved the organics and nitrogen removal but also mitigated the negative effects of lower temperature on organics and nitrogen removal in CWs. However, both MFC and temperature had no significant influence on phosphorus removal in CW-MFCs.

Enzymatic activity
As an intermediate carrier of hydrogen, DHA can reflect the microbial oxidative capability during the organics degradation, therefore it is often used to measure microbial activity (Barrena et al. ). As shown in Figure 6(a) and 6(b), DHA of the anodes was higher than that of the cath- winter. This result indicated that the closed circuit mode promoted the bacteria diversity on MFC anodes, which was consistent with the study by Li et al. (). For the cathodes, it was only in summer that the bacterial community diversity in CW-MFCc was higher than that in CW-MFCo. This may be because the effect of temperature on microbes was stronger than the stimulation of MFC, since