Laboratory study on nitrate removal and nitrous oxide emission in intact soil columns collected from nitrogenous loaded riparian wetland, Northeast China

Nitrate (NO3−) pollution of surface and groundwater systems is a major problem globally. For some time now wetlands have been considered potential systems for improving water quality. Nitrate dissolved in water moving through wetlands can be removed through different processes, such as the denitrification process, where heterotrophic facultative anaerobic bacteria use NO3− for respiration, leading to the production of nitrogen (N2) and nitrous oxide (N2O) gases. Nitrate removal and emission of N2O in wetlands can vary spatially, depending on factors such as vegetation, hydrology and soil structure. This study intended to provide a better understanding of the spatial variability and processes involved in NO3− removal and emission of N2O in riparian wetland soils. We designed a laboratory experiment simulating surface water flow through soil columns collected from different sites dominated by different plant species within a wetland. Water and gas samples for NO3−,NH4+ and N2O analyses were collected every 5 days for a period of 30 days. The results revealed significant removal of NO3− in all the soil columns, supporting the role of riparian wetland soils in removing nitrogen from surface runoff. Nitrate removal at 0 and 10cm depths in sites dominated by Phragmites australis and Carex schnimdtii was significantly higher than in the site dominated by Calamagrostis epigeio. Nitrous oxide emissions varied spatially and temporally with negative flux observed in sites dominated by P. australis and C. schnimdtii. These results reveal that in addition to the ability of wetlands to remove NO3−, some sites within wetlands are also capable of consuming N2O, hence mitigating not only agricultural nitrate pollution but also climate change.


Introduction
Nitrate ðNO À 3 Þ pollution of surface and groundwater systems has become a major problem globally. Both human activities and natural cycles are sources of NO À 3 ; however, human activities, such as the use of organic and inorganic fertilizers, poor placement of livestock waste, and would vary spatially and with depth because of the high environmental heterogeneity creating different micro-environments within the vegetation types. To unravel this, soil columns collected from sites dominated by different vegetation types (C. epigeios, P. australis and C. schnimdtii) were incubated in the laboratory, treated with NO À 3 enriched water and measured for the concentrations and fluxes of different N species at varying soil depth, which could eventually infiltrate and become groundwater in a natural setting.

Study area
The study was conducted in Qixing River Wetland National Nature Reserve, with permission from the National Nature Reserve administration, which is responsible for the protection of the reserve. Qixing River Wetland National Nature Reserve is a typical and representative riparian natural freshwater marsh type located in Sanjiang Plain, Northeastern China (463 9'45"-46˚48'24"N, 132˚00'22"-132˚24'46"E) at an average elevation of 80m (Fig 1). The reserve is approximately 20,000ha in size and it is divided into a buffer zone (3,600ha), core zone (7,960ha) and experimental zone (8,440ha). This study area is under the influence of temperate humid monsoon climate with an average yearly temperature and rainfall of about 1.9˚C and 550 mm, respectively. Qixing River Wetland National Nature Reserve soil type is characterized by mires soils, including peat-boggy soils, humus, meadow and peat soils.
The reserve is one of the most preserved, almost virgin and important ecosystems in the Sanjiang Plain, supporting a diversity of flora and fauna. It provides a habitat for about 29 threatened species of which 26 are birds and 3 are mammals. Moreover, the wetland is endowed with diverse species of plants and the most dominant ones are C. epigeios, P. australis and C. schnimdtii. Other companion species include Quercus mongolica, Glyceria spiculosa, Betulla platyphylla among many others. Apart from providing ecological functions, Qixing River Wetland National Nature Reserve also maintains good water quality and provides water storage services, thereby allowing infiltration to occur, hence recharging aquifers and groundwater. In addition the wetland provides significant socio-economic and cultural benefits.
In the recent years, wetlands that were once widely distributed in the Qixing catchment has become patches embedded in agricultural landscape due to long-term agricultural development [17]. The remaining Qixing River Wetland National Nature Reserve is surrounded by agricultural farmlands (Fig 1). The farmlands have been converted from drylands to paddy field. Due to agricultural activities, the wetland is strained in offering its environmental functions and services.

Soil sampling
Three sites were selected based on vegetation types; C. epigeios, P. australis and C. schnimdtii within the wetland experimental zone (area reserved for research) to allow for comparison. During the sampling period, the standing water depth in sites dominated by P. australis and C. schnimdtii ranged approximately from 0.4 to 0.7m and from 0 to 0.2m in the site dominated by C. epigeios. At every sampling site, three intact soil cores, excluding the plants, were collected using polyvinylchloride (PVC) pipes (4.5 cm internal diameter, 100cm in length) drilled with holes on its side to be used for collecting water samples in the laboratory. The holes were equally spaced at a 10cm interval (0, 10, 20 and 30 cm). The PVC pipes were sharpened at the bottom for easy penetration into the sediment during drilling, as described by [9]. The pipes were manually drilled to a depth of 40cm into the soil. Immediately after retrieving every PVC pipe with the soil content inside, both ends were sealed with cork and the sampling holes were covered with waterproof tape to avoid water leakage. Soil samples for determination of total nitrogen (TN), total organic carbon (TOC) and bulk density were collected at every sampling site following the same procedure. The samples were then transported to the laboratory and stored at room temperature for three to five days before the experiment was set up and sampling occurred. Soil samples for determination of TN and TOC were sectioned to get nutrients for each level.

Laboratory experiment setup
In the laboratory, the columns were set up to simulate downward surface water flow through the soil columns collected at different sites. The experiment consisted of two reservoirs A and B positioned at different heights on the same stand (Fig 2). Reservoir A was filled with nitrate solution (60 mg/L NO À 3 À N) prepared in the laboratory by dissolving potassium nitrate in distilled water. The solution from reservoir A flowed to reservoir B due to the air pressure generated when the solution level in B reduced below the nitrate solution level of the tube. The tube in reservoir B was used to regulate the nitrate solution level. The waterproof tape seal on the holes was replaced with Rhizon soil moisture samplers that were used to collect water samples from the soil using a nylon syringe. All the nine PVC pipes (three from each site) were connected to a common supplier of nitrate solution from the smaller reservoir B, and the flow rate was set at 0.042mL/second using a valve. The inlet from reservoir B into the columns was placed 5cm above the 0cm depth (soil-water surface) in all the cores (Fig 2). To ensure a constant water head in each soil column during the experiment, the water level in reservoir B was the same as in the inlet at any given time to allow the nitrate solution to percolate through each soil column.

Gas sampling
Prior to running the experiment, initial gas samples were collected from the column headspace and injected in aluminum-multi-layer foil composite film gas sampling bags. This was done by closing the PVC pipe ends with rubber stoppers and the gas was collected from the head space volume at 0, 10, 20 and 30 minutes interval using 20 ml nylon syringes. The gas samples were then stored into the dark at room temperature for 3 days before analysis. After 30 minutes, the rubber stoppers were removed and the experiment was left to run. Subsequent sampling was done every 5 days following the same procedure. The N 2 O gas concentration was measured using a gas chromatograph (Agilent 7890A, Agilent Co., Santa Clara, CA, USA) equipped with an electron capture detector (ECD). The slope of the N 2 O gas concentration over time (μmol/ mol/h) during sampling was used to compute the N 2 O flux(g/m2/h) between soils and the atmosphere according to the equation from [18] and [19] Flux ¼ dc dt  Nitrate removal and nitrous oxide emission

Water sampling
In order to allow the system to attain equilibrium with the water flow rate, the experiment was run for 5 days before the initial sampling took place. After that, sampling was done at 5 days intervals for a period of 30 days from 4 different depths (0 (soil-water surface), 10, 20 and 30cm (lower outlet)) using a 20 ml nylon syringe attached to the Rhizon soil moisture sampler. Water samples were also collected at the inlet (5cm above the soil-water surface) for every sampling occasion. Only 12 ml of water was sampled for every depth on every sampling occasion. The water samples for NO À 3 and ammonium (NH þ 4 ) were immediately filtered through 0.45μm GD/X Whatman filters and stored at -18˚C until analysis. Nitrate and NH þ 4 were analyzed using standard methods as described in [20] and [21] respectively. Nitrate removal in percentage (N r ) was calculated using the following equation where IC is the concentration of NO À 3 (mg N/L) at the inlet and Cd is the concentration of NO À 3 (mg N/L) at a given depth.
To determine nutrients fluxes, the formula proposed by [9] was used. First, the mean values of nutrient concentrations for the replicate columns were determined from each depth as follows: where � y = the average of the nutrient concentration X (mg N/L) determined from the site j in the soil column c at depth d on day i, while n is the number of soil columns collected from each site. Then the average flux (F) of nutrients in the soil columns and at different depths was calculated using the measured nutrient concentrations and flow rates.
Flow rateðmL=secondÞ Soil surface areaðsq mÞ ð4Þ Where C � yðoutÞ is the average nutrient concentration (mg/L) in the outflow and C � yðinletÞ is the average nutrient concentration (mg/L) in the inlet (5cm above soil-water surface) [22].

Statistical analyses
Before analysis, all the variables were tested for normality using the Kolmogorov-Smirnov test and homogeneity of the data was tested using the Bartlett test. With exception of pH all the variables were log transformed to satisfy the normality and variance assumption before conducting analysis. For the different soil profiles (0, 10, 20 and 30cm depth) repeated measure ANOVA was carried out to determine the differences in NO À 3 concentration among the sampling sites and depth. For multiple comparisons in case of a significant ANOVA result, post hoc comparisons using the Tukey HSD test were applied. Multiple regression analyses were performed using the forward stepwise selection procedure to select those predictor variables with a significant F-value that significantly increased the regression sum of squares. The predictor and response variables used are pH, TOC, TN, bulk density and C/N and NO À 3 and NH þ 4 , respectively. Figures were drawn using R software (version 3.4.1) [23]. Note that the results were reported as mean value ± standard error (SE).

Soil characteristics
The soil collected from the three sites within the Qixing River Wetland National Nature Reserve had distinct characteristics as indicated in Table 1. Overall, total nitrogen (TN) and total organic carbon (TOC) concentration decreased with increasing depth (from 0-40cm) while bulk density increased with increasing depth. Average soil total organic carbon computed between 0-40cm depths was 27.14, 18.29 and 17.60 mg C/g at the site dominated by C. epigeios, P. australis and C. schnimdtii, respectively. Similarly, mean total nitrogen was 2.41, 2.26 and 2.65 mg N/g at sites respectively. Mean soil bulk density for the sites dominated by C. epigeios, P. australis and C. schnimdtii was 1.12, 1.04 and 1.03 g cm -3 respectively and could be categorized as fine silt and clay soil. Notable are the low values of bulk density less than 0.5 g cm -3 from 0-10cm depth at the sites dominated by P. australis and C. schnimdtii indicating that the soil is very rich in organic matter. Soil pH was relatively low in the site dominated by C. epigeios.

Reduction of nitrate in different soil depth
The mean concentration of NO À 3 for the 30 days study period at the inlet (5cm above soilwater surface) in the soil columns ranged between 59.32 and 59.37 mg N/L while the mean lower outlet (30cm) concentration ranged between 0.59 and 0.97 mg N/L ( Table 2). This indicates that NO À 3 concentrations were reduced with increasing depth in all the soil columns. At the 0cm (soil-water surface) depth, sites dominated by P. australis and C. schnimdtii had a higher NO À 3 reduction of about 60% which is five times higher compared to that of C. epigeios at the same depth. At 10cm depth mean NO À 3 concentration had reduced to 20.06, 4.26 and 0.84 mg N/L for sites dominated by C. epigeios, P. australis and C. schnimdtii, respectively which corresponded to 66.20, 92.82 and 98.58% NO À 3 removal respectively (Fig 3). With the exception of the site dominated by C. epigeios, the NO À 3 concentration for the other two sites was almost constant at 20 and 30cm (lower outlet) depths (Fig 3). Ammonium concentrations measured at the inlet ranged between 0.59 and 0.60 NH þ 4 mg N/L and increased between inlet and outlet for all soil columns (Fig 4). The rate of NH þ 4 mg N/L increase was remarkably higher at 0cm depth in soil columns collected from the site dominated by P. australis followed by C. schnimdtii. Table 1. Soil characteristics of the three studied wetland vegetation sites (C. epigeios, P. australis and C. schnimdtii) in Qixing River National Nature Reserve wetland. Results are in mean ± SE. In view of the fluxes of NO À 3 (g N/m 2 /h), negative values (nitrate consumption) were recorded in all different depths of the three soils columns revealing that the soils are net sink (Fig 5). There was statistical significantly difference in NO À 3 influxes between soil depth in all sites as determined by one-way ANOVA (p<0.05). A post hoc test revealed that NO À 3 influxes at the 0cm depth in sites dominated by P. australis and C. schnimdtii differed significantly from the other depths (i.e. 10, 20 and 30cm). On the other hand, NO À 3 fluxes at 20 and 30cm depths did not show statistical significant difference at the site dominated by C. epigeios. The pattern of NH þ 4 fluxes (mg N/m 2 /h) differed from NO À 3 fluxes in all depths of all the soil columns. Positive fluxes of NH þ 4 were recorded in all depths of all three soils indicating that the soils are net sources (Fig 6). Mean fluxes of NH þ 4 in sites dominated by P. australis and C. schnimdtii differed significantly within soil depths p = 0.014 and p = 0.034, respectively. Post hoc test revealed that at the site dominated by P. australis, the mean NH þ 4 fluxes were significantly lower at 20cm depth (21.67±3.30 mg N/m 2 /h, p = 0.027) and at 30 cm depth (63.09± 17.46 mg N/m 2 /h, p = 0.041) compared to 0cm depth (500.49±170.64 mg N/m 2 /h). Similarly, post hoc test indicated that the mean efflux of NH þ 4 at 0 cm depth in site dominated by C. schnimdtii was statistically significantly higher than at 30cm depth (p = 0.047). NH þ 4 fluxes were not statistically significantly from the different depths in site dominated by C. epigeios.

Temporal variation of NO À 3 concentration in different depth
Figs 7-9 shows the temporal variation of NO À 3 concentration from the different depths of the soil column experiment over the 30 day period. Generally, the concentration of NO À 3 decreased remarkably in the first 5 days of the experiment. For the soil column from the site dominated by C. epigeios, NO À 3 concentration at the 0cm depth (soil-water surface) was 43.22 mg/L corresponding to 27.19% NO À 3 removal in the first 5 days from the initial concentration at the inlet of 59.36 mg/L (Fig 7). Then it increased from 43.22 mg/L to an average of 51.67 mg/L for the remaining days of the experiment corresponding to 12.95% of NO À 3 removed. For sites dominated by P. australis and C. schnimdtii 60% NO À 3 was removed in the first 5 days at the 0cm depth (soil-water surface) (Figs 8 and 9). Further rapid decrease in NO À 3 concentration was observed from the site dominated by C. schnimdtii on day 10 followed by a gradual increase in NO À 3 concentration up to day 30. Unlike at the 0cm depth (soil-water surface) which showed a significantly temporal variation of NO À 3 concentration among the soil columns, the NO À 3 concentration patterns in the other depths (10, 20 and 30cm (outlet)) were almost similar for the C. schnimdtii (Fig 9). After the first 5 days of the experiment, the concentration of NO À 3 for C. schnimdtii increased gradually at 0cm (soil-water surface) depth until it reached a peak on the 20 day, followed by a slight decrease. This clearly shows that NO À 3 removal was very high during the first 5 days in the soils collected from the site dominated by C. schnimdtii.

Nitrous oxide emission from soil column
The N 2 O fluxes measured from the soil columns are presented in Fig 10. Overall, the N 2 O emission depicted temporal and spatial variation during the sampling period. Hourly, mean N 2 O emission from the site dominated by C. epigeios, P. australis and C. schnimdtii ranged between 0.76 to 13.93 μg m -2 h -1 , -1.25 to 12.14 μg m -2 h -1 and -3.40 to -0.35 μg m -2 h -1 respectively (Fig 10). Assuming a constant N 2 O fluxe over the year, the annual fluxes would be between -297.84 to 1220.27 g N 2 O ha -1 y -1 for the three sites dominated by different vegetations within the wetland. The soil from the site dominated by C. epigeios was a net source of N 2 O, while soils from sites dominated by P. australis and C. schnimdtii were net sinks of N 2 O. Nitrous oxide emission from the site dominated by C. epigeios increased gradual from day 1 until it reached on a peak on day 20 and decrease afterward (Fig 10A). Similar but reverse pattern to that from the site dominated by C. epigeios was observed in soil from the site dominated by C. schnimdtii reaching a peak of N 2 O consumption on day 20 (Fig 10C). Table 3 shows the correlation results between pH, TOC, TN, C/N, NO À 3 concentration and NH þ 4 concentration. From the results NO À 3 depicted a significantly positive relationship with TOC and TN. A multiple forward stepwise regression revealed that TOC was the most important variable related to NO À 3 removal (NO À 3 removal = -9.310+0.984TOC, p = 0.000, R 2 =

Spatial dynamics of nitrate and ammonium in wetland soil
Wetlands have the ability to improve surface water quality by reducing NO À 3 pollution. Previous studies have revealed plant uptake, soil retention, denitrification, microbial immobilization and nitrification as the major process of nitrogen retention in wetlands [24][25][26]. In wetlands which receive high nitrate loading from agricultural runoff or sewage treatment plant discharge, denitrification has been found as the most dominant process of reducing NO À 3 [26,27]. The results of our experiment revealed remarkable reductions in NO À 3 concentrations from between 59.32 and 59.37 mg N/L at the inlet (5cm above soil-water surface) to concentrations below 1 mg N/L at the lower outlet of the soil columns. Therefore, almost 98% of the NO À 3 disappeared as the water flowed down through the soil columns indicating an efficient nitrate removal capacity of the wetland soils. These findings are in agreement with other Nitrate removal and nitrous oxide emission studies which have confirmed nitrate removal efficiency as high as 90% in nitrate-loaded riparian wetlands [9,28]. Similar observations were also made in constructed wetlands systems. For instance, [29] reported significantly nitrate removal efficiencies of about 70-99% in a small scale flow constructed wetland system. Denitrification which has been documented as a NO À 3 sink in riparian areas by researchers [9,30], was probably the major pathway of nitrate removal; however, dissimilatory nitrate reduction to ammonium (DNRA) also might have played a role. Actually, the increasing concentration of NH þ 4 between the inlet and outlet implied the existence of DNRA or mineralization of organic matter. Studies have shown that in the presence of high soil organic carbon, DNRA can occur jointly with denitrification under reducing environments [9,[31][32][33]. This concur with the findings of this study as the production of NH þ 4 was high at 0cm (soil-water surface) and 10cm depths where the total organic carbon was also high. Note, however, that NH þ 4 concentrations do not support DNRA as the dominant process responsible for reducing NO À 3 in the soil columns as the amount of N accumulation as NH þ 4 does not account for the N lost as NO À 3 .

Fig 5. Hourly fluxes of surface water NO À
From our results, it is quite clear that the soils collected from different sites within the wetland acted as a sink for NO À 3 . On average, the NO À 3 fluxes were 4.95, 4.91 and 3.78 g N/m 2 in the sites dominated by C. schnimdtii, P. australis and C. epigeio, respectively. These values are almost in the same range as those reported from laboratory studies in Danish peat soil (4.6 g N /m 2 ) by [34], but much higher than those reported in laboratory microcosms in central and northern Jutland, Denmark (1.1 g N/m 2 ) by [9].
Although the average NO À 3 removal rate from the inlet to the lower outlet was almost the same from the three sites, a significant vertical gradients of NO À 3 removal rates along the soil profiles were evident with the highest rates occurring in the upper layers. The differences in NO À 3 removal rates along the soil profiles were most likely the result of differences in total organic carbon. The fact that there was no significant difference between NO À 3 fluxes at 20 and 30cm depth in the soil columns suggests that most of NO À 3 removal in this system occurred at depths above 20 cm where total organic carbon was also very high. As indicated in Table 1, all the soils used in this study contained less than 5% TOC while the NO À 3 removal rate was very high. This agrees with previous studies showing that NO À 3 removal was not C limited in a nitrogen loaded riparian wetlands with a TOC content less than 7% [9]. Upper soil profiles Nitrate removal and nitrous oxide emission contained the highest amounts of organic carbon (Table 1). This observation was further supported by the result of multiple linear regression which singled out TOC as important variable related to NO À 3 removal (Table 3). [27] and [30] revealed that NO À 3 removal through the process of denitrification in riparian wetland soil is very high in the upper layers and depends on microbial activity and soil organic carbon level. [35] further observed a significant relationship between NO À 3 reduction and extractable carbon. Interestingly, the results of our study further revealed variation in NO À 3 removal at 0cm (soil-water surface) and 10cm depths. In the sites dominated by P. australis and C. schnimdtii, NO À 3 removal was significantly higher than in the site dominated by C. epigeio (Fig 3). This is contrary to the expectation, since the upper soil profiles of the site dominated by C. epigeio had higher amounts of total organic carbon compared to the other sites (Table 1).
This spatial variability in NO À 3 may be explained by organic carbon quality. Scores of studies reported that removal of NO À 3 through denitrification process is highly correlated with available carbon (e.g. microbially labile) rather than total organic carbon [27,36,37]. Thus, although C. epigeio had higher concentrations of total organic carbon, the available carbon was Nitrate removal and nitrous oxide emission potentially low hence reducing NO À 3 removal. Note, however, that available carbon (e.g. microbially labile) was not measured in this study. While assessing the dynamics of microbial communities during decomposition of plant litter from soil ecosystems, [38] observed that high C/N ratios of C. epigeio litter is less attractive for microbial degraders. Therefore it is also possible that the high C/N ratio observed in the upper layers (0-10cm) of the site dominated by C. epigeio may have affected the removal rate of NO À 3 . The variability in NO À 3 removal could also be due to soil pH and denitrifying enzymes sensitivity. Soil pH is one of the essential factors that can control denitrification process through enzyme sensitivity [39,40]. Low soil pH, as observed in the site dominated by C. epigeio could limit the availability of organic carbon and available mineral N to denitrifying bacteria [39,41]. However, this should be interpreted with caution because scientists have argued differently on the relationship between pH and the denitrification process. For instance [42] observed that below pH 6 denitrification was somewhat retarded but even at pH 4.9 more than 70% of added nitrate was lost within 2 weeks. On the other hand, [43] pointed out that the rate of carbon mineralization (measured as CO 2 production), rather than pH, controlled the rate of denitrification in a system.

Nitrous oxide emission from wetland soil
Previous studies have reported temporal and spatial variability in N 2 O fluxes from natural riparian wetlands [44,45]. Some riparian wetland soils have been found to act as sources of N 2 O [46] and others as sinks of N 2 O [45]. In the present study, the emission of N 2 O varied spatially and temporally. The average N 2 O fluxes measured in the sites dominated by C. epigeios, P. australis and C. schnimdtii of 0.76 to 13.93 μg m -2 h -1 , -1.25 to 12.14 μg m -2 h -1 and -3.40 to -0.35 μg m -2 h -1 , respectively are almost in the same range to those reported by [45] from four Danish riparian wetlands of -44 to 122 μg N 2 O -N m -2 h -1 . We hypothesized that N 2 O production from the riparian wetland would vary spatially because of the high environmental heterogeneity that created different micro-environments within the vegetation types. Our findings clearly supported our hypothesis. Positive N 2 O fluxes were observed from soil collected from the site dominated by C. epigeio, implying production of N 2 O by the soil to the atmosphere. On the other hand, soils collected from sites dominated by P. australis and C. schnimdtii had negative fluxes of N 2 O implying consumption of the gas by the riparian wetland soil. These differences could potentially be ascribed to: (i) differences in total organic matter. In their study, [47] revealed that sites loaded with high total organic carbon in boreal upland soils were net sources of N 2 O while low total organic carbon sites were net sinks of N 2 O. This is in agreement with our findings, since the average soil organic carbon from the site dominated by C. epigeio was higher compared to that of sites dominated by P. australis   [45,48,49]. On the other hand, the high NO À 3 concentration on the upper layers of C. epigeio could have hindered N 2 O reduction during denitrification [45,50]. Note that there is no substantial increase in NH þ 4 under C. epigeio to attribute N 2 O production to DNRA. Soil pH is also one of the most important factors that affect denitrifier community composition, denitrification rates and denitrification end products [39]. Our results revealed relatively high positive fluxes in the site dominated by C. epigeio where the soil pH values were relatively low (Table 1 and Fig 10A). These findings are in agreement with [51] and [52] who reported that low soil pH inhibits di-nitrogenoxide reductase thus increasing the amount of N 2 O production. On the contrary, [39] and [45] found that low soil pH (<5) was associated with low N 2 O production. It is quite clear that the influence of low soil pH on N 2 O production is still debatable and possibly inconsistent. While assessing the relationship between N 2 O production and soil pH, [53] noted that the net effect of low pH on N 2 O is not straightforward since multiple environmental factors can have influence on denitrification rates. Note, also, that the generally low N 2 O emission observed in this study could also be explained by the use of laboratory soil columns which allows small-size areas to be sampled and monitored hence possibly missing hotspots. Moreover, the sampling time and frequency may have missed peak hot moments of N 2 O production [54].

Conclusion
This laboratory experimental study was intended to provide a better understanding of the spatial variability and processes involved in the attenuation of NO À 3 and the emission of the greenhouse gas N 2 O in riparian wetland soils located in the middle of farmlands. The riparian wetland receives agricultural runoff rich in nitrogenous fertilizers from all directions. In the laboratory, the experiment was designed to simulate downward surface water flow through the soil columns collected at different sites dominated by different vegetation types (C. epigeios, P. australis and C. schnimdtii). Our results revealed significant and rapid removal of NO À 3 in all of the soil columns, supporting the crucial role of riparian wetland soils in removing N from surface runoff. Our study further revealed that NO À 3 removal at 0cm (soil-water surface) and 10cm depths in the sites dominated by P. australis and C. schnimdtii was significantly higher than in the site dominated by C. epigeio. This could be attributed to organic carbon quality and low pH values in the site dominated by C. epigeio. Moreover, our study showed that N 2 O emissions varied spatially and temporally with negative flux observed in soils from the sites dominated by P. australis and C. schnimdtii. It is clear from this study that in addition to riparian wetland soil being able to remove NO À 3 , some sites within the wetland are capable of consuming N 2 O hence mitigating not only health problems (e.g. methemoglobinemia in infants and young children) and environmental problems (e.g. eutrophication) related to NO À 3 but also atmospheric N 2 O pollution.