Effects of Partial Saturation on Nitrogen Removal and Bacterial Community in Vertical-Flow Constructed Wetlands

The laboratory-scale pilot of constructed wetlands has been in operation for six months; (1) an unsaturated vertical �ow constructed wetland (UVF-CW), this system was used to represent the classic vertical constructed wetlands, (2) a saturated vertical �ow constructed wetland (SVF-CW), to evaluate the effects of the saturated condition on nitrogen removal and composition of the microbial community. The results showed that the saturation condition positiveley in�uenced the removal e�ciencies of the nitrogen,, the aeverage removal rate of the total kjeldahl nitrogen increased from 56% in unsaturated vertical �ow constructed wetland (UVF-CW) to 63% in saturated vertical �ow constructed wetland ( SVF-CW). In addition, the microbial communities also was affected by the saturation condition, the relative abundances of nitrifying bacterium in UVF-CW are 13.8% (Nitrosomonas), 7.2% (Nitrosospira), 18.1% (Nitrospira) and 15.3% (Nitrobacter). In contrast, in SVF-CW, Nitrosomonas, Nitrosospira, Nitrospira and Nitrobacter only accounted for 6.8%, 5.6%, 7.4% and 10.6% respectively. However, the saturation condition seemed to increase denitrifying bacterium more than three times, in unsaturated vertical �ow constructed wetland, only Pseudomonas (6.5%) and Paracoccus (4.85%) were detected, but in saturated vertical �ow constructed wetland (SVF-CW), the abundance of Pseudomonas (13.08%) and Paracoccus (9.74%) were increased, and three other groups of denitrifying bacteria were also detected as Zoogloea (3.32%), Thauera (5.41%) and Thiobacillus (3).


Introduction
Excess nitrogen in aquatic ecosystems can cause many problems such as eutrophication,,which have negatively affect on biodiversity,climate and human health (Stevens et al. 2019).Therefore, the remove of nitrogen from waste water it is necessary.Traditional activated sludge technology has been used for nitrogen removal in multiple wastewater treatment plants (WWTPs) due to its high e ciency (EPA, 1993).
However, these conventional technologies require high construction cost and consume more energy (Liu et al. 2016).
In recent years, constructed wetland (CW) is an ecological technology.Due to its various advantages over traditional wastewater treatment technologies, it has been rapidly developed in wastewater treatment in scattered areas in industrialized counties and low-income countries (Liu et al. 2016 ; Zhang et al. 2019).
These include low construction and operating cost, easier maintenance and good integration into the landscape and promotion of biodiversity (Álvarez et al. 2017;Paing et al. 2015).However, the removal of nitrogen in constructed wetlands (CWs) exhibits large uctuations and is often unsatisfactory ( Wang et al. 2017).The total nitrogen removal in vertical ow or in horizontal ow constructed wetlands is usually not completely removal, but instead converts it to various nitrogen compounds (Pelissari et al. 2017).
In fact, the vertical ow constructed wetlands is an unsaturated systems, which often fed with several pulses intermittently throughout the day, resulting in a high oxygen transmission capacity, that is bene cial to the nitrifying bacteria in the bed (Platzer et al. 1999).However, horizontal ow constructed wetlands are mainly operated under anoxic/anaerobic conditions, which makes it a suitable environment for denitri cation process (Vymazal et al. 2007).Generally, the nitrogen removal limitation in constructed wetlands can be explained by the competition for oxygen by autotrophic and heterotrophic microorganisms heterotrophic (Saeed et al. 2012), and the limit of organic carbon available for the process of denitri cation (Lavrova et al. 2010).Vertical ow constructed wetlands have more attention than horizontal ow constructed wetlands due to its less demand for land ( Meng et al. 2014).On the other hand, traditional vertical ow constructed wetlands cannot remove total nitrogen (TN) satisfactorily due to the lack of appropriate hypoxic conditions for denitri cation (Pelissari et al. 2017) In order to completely remove nitrogen, various types of enhanced vertical ow constructed wetlands have been studied, such as arti cial aeration, tidal ow and integrated vertical ow constructed wetlands (Pelissari et al. 2018;Hu et al. 2016).There are other operational conditions to enhance the removal e ciency of total nitrogen in vertical ow constructed wetlands, such as recirculation of e uent to improve nitri cation e ciency ( Wu et al. 2016) and stepwise feed to enhance carbon source supply to promote denitri cation ( Li et al. 2017).However, these modi cations have increased operating costs and maintenance complexity.The latter strategy uses a partially saturated vertical ow constructed wetland con guration to create anoxic/anaerobic conditions at the bottom of the bed and aerobic conditions at the top of the bed to promote adequate condition for simultaneous nitri cation and denitri cation (Torrijos et al. 2017;Kim et al. 2014).This model is more e cient in total nitrogen removal than traditional unsaturated vertical ow constructed wetlands with sequential nitri cation and denitri cation (Silveira et al. 2015;Dong et al. 2007;Kim et al. 2015a).
In general, all efforts aimed at maximizing the nitrogen removal in constructed wetlands are directly related to the activities of enhancing microbial communities, which are responsible for the conversion of various nutrients in the lter media and rhizosphere biomass (Mayo et al. 2005).Many studies have evaluated the dynamics of bacterial communities in constructed wetlands.Foladori et al. (2015) showed that the number of viable bacteria in the surface layer is 3.7 times that of the deep layer.Adrados et al.
(2014) characterized the prokaryotic microbial communities of vertical ow constructed wetlands, horizontal constructed wetlands and biological lter sand, and reported higher bacterial activity than archaea in all research systems.Other studies have shown that the diversity of bacterial communities in constructed wetlands may affect the quality of the nal e uent (Calheiros et al. 2009).Button et al.
(2015) indicated that microbial metabolic functions identi ed in different constructed wetlands types are related to the design of each system, spatial position within the bed, and especially with levels of pretreatment.
There is no rigorous research on the knowledge of nitrogen-transforming bacteria in vertical constructed wetlands.Therefore, it is clearly necessary to further understand the kinetics of nitrogen conversion bacteria in unsaturated vertical ow constructed wetlands, especially in partially saturated vertical ow constructed wetlands, in order to increase the total nitrogen removal in a single vertical ow constructed wetland.The purpose of this study is to: (1) compare the conversion and removal e ciency of nitrogen in saturated and unsaturated vertical ow constructed wetlands; (2) identify nitrifying and denitrifying bacteria covered with gravel in unsaturated and saturated zones.

VFCW systems and operation design
Two laboratory-scale vertical ow constructed wetlands (VFCW) were evaluated in duplicate with the same size and substrate.They operate in parallel, where unit 1 is composed of a partial vertical ow wetland (SVF-CW) (50%), and unit 2 is composed of a traditional unsaturated vertical ow wetland (UVF-CW).A schematic diagram of the laboratory scale of VFCWs is shown in Fig. 1.The two laboratory-scale constructed wetlands are made of polyvinyl chloride tube, each having a diameter of 41 cm, a total height of 100 cm, and a media bed height of 80 cm.These units are equipped with a 10 cm coarse gravel (25 to 40 mm) drainage layer at the bottom and a 70 cm main lter layer composed of 2-4 mm gravel.The downwardly applied saturation zone is maintained by the siphon structure at the outlet.Young reed species are collected from local valleys and planted in all vertical ow wetlands.
In order to the growth of plants and bio lms in the medium, the wetland has accumulated water for up to 2 months (50% of rural wastewater and 50% of tap water).After the adaptation period, these systems intermittently fed with rural wastewater at a ow rate of 20 L/d ( ve batches per day).The operation cycle is divided into a feeding period (3 days) and a rest period (4 days).Table 1 shows the chemical characteristics of in uent wastewatercycle was divided into a feeding period (3 days) and a rest period (4 days).Table 1 shows the chemical characteristics of the in uent wastewater.

Biomass sampling and bacteria detection
For bacterial analysis of nitrifying and denitrifying communities, 20 grams of bed media were collected in all pilot plants at two different depths (-20) cm and (− 60) cm, and 100 milliliters of deionized water was added to it.The samples are mixed and sonicated in ultrasound for 5 minutes to loosen the bio lm from the medium.Then, the sample was centrifuged at 1500 g for 5 minutes.All biomass samples were prepared according to the standard method of uorescence in situ hybridization (FISH) analysis (Zwirglmaier et al. 2005).To x the samples, 1ml of each sample (mixed water and bio lm suspension) was centrifuged at 16,000 g for 5 minutes, and then, by adding paraformaldehyde solution (4%) and phosphate buffered saline (PBS) 3:1, According to Amann (1995), the recovered particles were xed at 4°C for 3 hours.Then, the xed samples were washed twice with 1 mL of PBS and nally suspended in a solution of PBS and absolute ethanol and stored at − 20°C.
For the quanti cation of total bacterial cells, considering the Eubacterial domain, a volume of 10 µl of each sample was xed on a PTFE-printed microscope slide and covered with 1µg.mL − 1 of 4', 6-diamidino-2-benzene Based indole (DAPI) solution.Table 2 shows the speci c probes for nitrifying and denitrifying bacterial communities used in this study.All probes are labeled with the uorescent pigment Cy3 at the 5'position.The quanti cation of speci c bacteria was performed by directly counting 20 random elds in each well using an epi uorescence microscope (OlympusBX41, Tokyo, Japan).In order to estimate the abundance of cells hybridized with the probe EUB mixture, the cells stained with DAPI were considered to represent 100% of all microorganisms identi ed by digital images.For the remaining probes, 20 regions were randomly selected, and the cells stained with the probe EUB mixture were considered to be 100% of all bacteria identi ed by the digital image.The "Microbial Ecology Digital Image Analysis (DAIME)" software was used to determine the relative abundance of nitrifying and denitrifying bacteria from the total DAPI staining (Daims et al. 2006) The unsaturated vertical ow (UVF-CW) had higher ammonium removal e ciencies than partially saturated vertical ow (SVF-CW).This result can be attributed to the higher diffusion of oxygen to the partially unsaturated layer, which is bene cial to the nitri er development (Kraiem et al. 2019).
In contrary, increasing saturation level seemed to have a positive effect on Total Kjeldahl Nitrogen (TKN) removal e ciency.In fact, vertical ow constructed wetlands with saturated layer (SVF-CW) had higher NTK removal than unsaturated bed (UVF-CW) (Table 4).This higher removal e ciency resulted from the development of denitri cation process in saturated layer (Pelissari et al. 2017;Del Toro et al. 2019).
With regard to NO 2 --N, this form of nitrogen showed that an increase at both con gurations of the systems with respect to the in uent, NO2 -N concentration in the e uents remained above the values in the in uent.Due to the fact, this oxidized form of nitrogen is only an intermediate in different nitrogen reaction transformations; it is understandable to nd such very low concentrations in this study.
There are signi cant differences in the removal effect of NO 3 − -N between the two vertical ow constructed wetlands (p < 0.05).Table 4, shows that the removal e ciency of SVF-CW on NO 3 − N is signi cantly higher than that of UVF-CW (P < 0.05),indicating that denitri cation capacity in the vertical ow constructed wetlands was promoted signi cantly with the saturated layer.This can be explained by the fact that the saturation layer is produced the required hypoxic condition and enhanced denitri cation activity.This nding is consistent with the research conducted in the subsurface vertical ow constructed wetlands (Pelissari et al. 2018).In addition, although autotrophic nitri cation-heterotrophic denitri cation is the main process of SVF-CW nitrogen removal, other mechanisms probably took place in this system.
Other possible mechanisms are ANAMMOX, which could NO 3 − N produce (Dong et al. 2007;Kraiem et al. 2019).The increased of nitrate concentration in UVF-CW con rms a lightly higher capacity of UVF-CW for nitri cation due to supplementary oxygenation through the unsaturated layer.

In uence of Partially Saturated on Nitrifying and Denitrifying Bacterial Communities
In constructed wetlands, the diversity and abundance of microbial communities depends on environmental factors, wastewater types, media types, operating conditions and plant species (Meng et al. 2014).Aerobic conditions (inside the media) often support autotrophic nitri cation in vertical ow constructed wetlands (Wu et al. 2016;Kizito et al. 2017).In order to better understand the remove of nitrogen obtained in two units with different saturation levels (UVF-CW and SVF-CW), (FISH) analysis were used in this study.Figure 2 shows that the detection of nitrifying bacteria by FISH technology, which shows that higher relative abundance of nitrifying bacteria and higher ratio of nitrite oxidizing bacteria (NOB) to Ammonium oxidizing bacteria (AOB) followed by higher ammonium removal performance were obtained in UVF-CW compared to SVF-CW.
The relative abundances of nitrifying bacteria were signi cantly different among the two different vertical ow constructed wetlands (p < 0.05).The abundances in UVF-CWwere13.8%(Nitrosomonas),7.2% (Nitrosospira),18.1% (Nitrospira) and15.3%(Nitrobacter).On the contrary, Nitrosomonas, Nitrosospira,Nitrospira and Nitrobacter merely accounted for 6.8%, 5.6%,7.4% and10.6% of the total bacteria in SVF respectively (Fig. 3).These abundances are higher than that obtained by Guan et al. (2015), Wang et al. (2016) and Pelissari et al. (2017) The presence of saturated media, resulting the reduction of the aerobic zone (entering the media) (Vymazal and Kröpfelová, 2015) which could have played a major role in reducing nitrifying bacteria.This result con rmed the nitri cation performance obtained in Sect.3.4.These results also indicated that Nitrosomonas in the two vertical ow constructed wetlands ((UVF-CW and SVF-CW) was more abundant than Nitrosospira.This outcome can be explained by the high in uent concentration of ammonium (Table 4), forming a community dominated by Nitrosomonas, with low substrate a nity but a high maximum activity than Nitrosospira ( Wang et al. 2016).
In contrary to nitrifying bacteria which deceased with increasing saturation level, the denitrifying bacteria increased with increasing of saturation depth.The relative abundances and variety of denitrifying microorganisms were signi cantly different among the two different vertical ow constructed wetlands (p < 0.05), Zhang et al. (2015) also observed similar ndings.Furthermore, under saturated vertical constructed wetland (SVF-CW) denitrifying bacteria increased more than 3 times in comparison to unsaturated vertical constructed wetland (UVF-CW).. Foladori et al. (2015) showed that due to the different environmental condition of the bed mesdia along the vertical section, the depth has a great in uence on the distribution of bacteria.In fact, in (UVF-CW), only Pseudomonas (6.5%), and Paracoccus (4.85%) were detected in this study (Fig. 4).These two bacteria are associated to the classic pathways of denitri cation that carried out mainly by Pseudomonas spp.(Ahn, 2006) and with aerobic denitri cation and heterotrophic nitri cation by Paracoccus denitri cans ( Richardson, 2000).
However, in (SVF-CW), Pseudomonas (7%) and Paracoccus (5%) were increased in abundance but three other groups were also detected as Zoogloea (1.8%), Thauera (2%) and Thiobacillus (3%), owing to low availability of oxygen seems to favor denitrifying bacteria (Fig. 4).These results indicated that, besides the classical pathways of denitri cation, other pathways of denitri cation also participated in nitrogen transformation in saturated zone.In fact, the detection of Thiobacillus, which can use inorganic compounds as source of carbon and compounds such as nitrates or nitrites as electron donors was associated to autotrophic denitri cation (Pelissari et al. 2017).

Conclusion
Partially saturated (SVF-CW) con gurations were developed to optimize the nitrogen removal in the same lter with low operation costs and low required area.This research highlighted that the application of saturation zone at the bottom can ensure nitrogen removal performance better than classic vertical ow constructed wetlands.FISH analysis in this study indicated that the diversity and the abundance of nitrifying and denitrifying bacteria were affected by the depth of saturated layer: 1.The average abundance of nitrifying bacteria in (UVF-CW) is higher than that in (SVF-CW).These results also showed that Nitrosomonas in the two systems was more abundant than Nitrosospira , this might be related to the high in uent ammonium concentration, which leads, to form a community dominated by Nitrosomonas, and its substrate a nity lower, but the activity is higher than Nitrosospira.
2. A saturated zone in uence was observed on relative abundances and diversity of denitrifying bacteria.In UVF-CW only Pseudomonas and Paracoccus were detected.However, in (SVF-CW), these bacteria were increased in abundance but three other groups were also detected as Zoogloea,Thauera and Thiobacillus.
3. In addition to the classic denitri cation and nitri cation, the occurrence of species as Paracoccus,Thiobacillus denitri cans and Thiobacillus thioparus have also been discovered, which are related to heterotrophic nitri cation and aerobic and autotrophic denitri cation.
This research is only carried out on a the laboratory scale, so, it is worthwhile to further study the removal effect and application of pilot scale and full scale to obtain more conclusive results.Other studies can focus on the effects of different factors that can improve the performance of nitrogen remove, such as plant species, substrate types.Our group is conducting further research on this topic, such as using a combination of llers of natural materials (e.g.olive seeds, olive pomace, compost, clay, biochar) to improve the e ciency of nitrogen remove in the constructed wetland wetlands.

Figure 2 FISH
Figure 2

Table 1
Characteristics of in uent wastewater.

Table 2
(Saeed et al. 2012;Chang et al. 2013;Huang et al. 2013)ationand bacterial abundance were compared by the measured value ± standard error.Use the statistical program STATISTICA software (http://www.statsoft.com)forstatisticalanalysis.The average value was compared by Fisher'shsd test (< 0.05).During the experiment period, no signi cant correlation was observed between e uent temperature and saturated zone depth.The average e uent temperature was 23.5 and 25.2°C for SFV-CW and UFV-CW respectively.The temperature range in two VFCWs was suitable for nitrogen removal through nitri cation and denitri cation process(Kotti et al. 2010;Chang et al. 2013;Huang et al. 2013).The e uent pH was not signi cantly different in two VFCWs (p > 0.05).The pH value of VFCWs is 6.98-7.12,which is slightly lower than the pH value of the in uent 7.31 (Table3).The pH range of 6.87-7.31wasfavorablefor microbial nitri cation and denitri cation(Saeed et al. 2012). The

Table 4
Nitrogen concentration at In uent and e uent and removal e ciency of the partially saturated vertical constructed wetland (SVF-CW) and unsaturated vertical constructed wetland (UVF-CW)