4.2. Nitrogen Mass Balance and Transformation Pathways
Nutrient budgets (N and P) for the wetland were based on the water budget study and on information drawn from detailed studies of denitrification, plant uptake and soil accumulation [
1,
10]. To better understand the processes involved, transport and transformation pathways in nutrient removal in the pilot wetland, the differences in concentrations and loading of influent and effluent data were used.
Principal processes transforming nitrogen in aquatic systems that can lead to mass loss include ammonification (mineralization), nitrification, bacterial denitrification (carbon dependent), plant uptake, assimilation, decomposition and burial [
1]. A nitrogen mass balance was estimated by quantifying wetland input, output and storage in plant biomass, and estimating the denitrification rates based on the water balance and water quality monitoring.
Table 6 presents nitrogen mass loading rates for NH
4+, NO
3− + NO
2− and organic nitrogen (ON) summed from the six-month data.
Equation (2) presents major components of nitrogen mass balance approach:
Figure 11 shows the main components of nitrogen transformation pathways known to occur within the wetlands, and the system inflow and outflow forms measured in this study. Total loading of NH
4+, ON and (NO
3− + NO
2−) was computed from the monthly loading from October 2008 through March 2009. During the pilot study, 61% of the nitrogen mass was removed, of which 57% was estimated as loss by gasification. Based on literature estimates [
1] of biomass nitrogen content, system storage was estimated to be approximately 0.2 kg N/m
2. The mass loading rate for nitrogen for the six-month period totaled 5.21 kg N/m
2 and the loss in terms of storage and gasification combined was 3.21 kg N/m
2 and the total mass leaving from the system was 2.01 kg N/m
2 which totals to 5.21 kg N/m
2.
In a heavily loaded ammonia rich source water like the RO concentrate used in this pilot study, production of NH
4+ from decomposition of biomass is present by relatively insignificant compared to the external loading. Volatilization of NH
4+ along with denitrification of NO
3-N and NO
2-N contributed to the loss of nitrogen through gasification. Another loss of nitrogen from the water column is in the form of burial of particulate organic nitrogen which becomes storage in sediments [
1]. Ammonification of organic nitrogen further complicates the system interpretation since the input ROC contained a high percentage of ammonia. The large loading of TKN, i.e., ammonia and organic nitrogen indicates this CW is a microbially dominated system [
1]. For the demonstration wetland, a primary stage of aerated subsurface flow wetlands was designed to provide microorganisms the conditions to further nitrification, biodegradation and enhance the system efficiency for organic matter and nitrogen removal [
39].
As shown in
Figure 11, major processes for nutrient cycling include (1) particulate settling and resuspension; (2) diffusion of dissolved forms; (3) plant translocation; (4) litterfall; (5) ammonia (un-ionized) volatilization (gasification); (6) anaerobic ammonia oxidation (Anammox); and (7) sorption of soluble nitrogen on substrates (detritus and sediment).
The limited availability of labile carbon (carbon forms that are more easily broken down) in this pilot study (
Figure 9) may have restricted denitrification. Denitrification requires an organic carbon substrate at a stoichiometric ratio of approximately one labile carbon per NO
3-N [
42]. Less than required available carbon is one of the reasons for incomplete denitrification in the CW.
Recalcitrant carbon may not be a reliable source in many nutrient transformation processes. At lower carbon-to-nitrate ratios, as was observed for this pilot study, denitrification may be incomplete. To treat low C/N ratio concentrate with nitrate rich influent and low labile carbon as in this study, the carbon source from the root exudates of macrophytes is not sufficient to maintain a high performance of nitrate removal [
37]. Also, the available oxygen (electron acceptor) in this study may have a limited contribution to the nitrification of NH
4+ into NO
3−. Microbially mediated processes involves the dissimilatory transformation of NO
3− to NH
4+ via NO
2−. However, this process would be favored in NO
3−-limited environments rich in labile carbon [
43]. In this study, this pathway of NO
3− transformation would be limited since the wetland was rich in NO
3− and low in labile carbon. Given the low C/N condition observed in this study, the demonstration wetland design addressed denitrification as a secondary stage, where additional organic carbon would be contributed by decomposition of an anaerobic substrate media (peat, compost). For future consideration, the CW could be enhanced by an external supply of electron donors via direct organic carbon addition or through the pathway of microbial anammox [
38].
Anammox is another important nutrient transformation pathway of nitrogen cycle with limited labile carbon or an excess of nitrogen relative to carbon input [
43], as was found in this study. In this oxygen limited HSSF study, the path of nitrogen loss due to anammox process could be significant since the process requires less oxygen than the nitrification/denitrification process [
13].
Gasification resulting from processes of denitrification, volatilization of ammonia and anaerobic ammonia oxidation (annamox) contributed to the removal of about 57% of the total input loading. Treatment under anoxic conditions could be substantial given that the pilot wetland had a relatively oxygen limited environment with relatively lower nitrification and plant uptake rates. Generally, NH
4+ volatilization occurs at pH > 8 [
13,
44]. The water pH of the CW in this pilot study was always < 7.5 (
Table 3). Therefore, the contribution of NH
4+ volatilization in the nitrogen output of this CW is assumed to be low or negligible.
Uptake of NH
4+, NO
2− and NO
3− by plant biomass and subsequent decomposition to organic nitrogen and accretion or burial plant material as sediments is another nutrient transformation pathway in microbial process dominated system like the CW in this study [
1]. Bacteria grow in biofilms attached to aggregate and plants and becomes part of upper soil strata through plant translocation and litterfall processes. Other than residence time, a number of other factors may explain the wide range of NO
3− removal rates reported in the literature such as, temperature and substrate in bioreactor [
45].
Storage into sediments is an important nutrient transformation pathway in terms of loss of nutrient from the water column. Storage in aboveground biomass through the settling process and removal by harvesting and resuspension of particulate matter transport resulted in an insignificant export of nutrients out of the CW [
46]. The removal of nutrients in the form of storage is a result of many nitrogen fixation processes, such as temperature, soil carbon content, soil pH and ammonium concentration in soil water [
10]. Ammonium concentrations in the ROC were mostly higher than 100 mg/L (
Figure 4). Therefore, the N fixation was negligible compared with the N import associated with the loading of concentrate. It is likely that some of the ammonia is converted to dissolved organic nitrogen, which could be regenerated to inorganic N downstream of the CW. Increasing DO can enhance the nitrification reaction rate and the growth rate of the nitrifying bacteria, but inhibit the denitrifying bacteria activities and constrain total nitrogen removal. Thus, a certain volume of DO is necessary in a treatment wetland [
46]. Based on DO level in CW, it was decided that forced aeration would be designed in the demonstration wetland for enhanced nitrification.
Adsorption capacity of soluble nitrogen could be low due to prolonged earlier loading [
10]. Also, denitrification could be low because of limited nitrification due to the anoxic conditions in the sediment and with the availability of labile carbon. The availability of labile carbon may have affected the adsorption rate. In summary, partially oxygen limited condition and low availability of labile carbon in the pilot wetland restricted many processes for nitrogen transformation.
4.3. Phosphorus Mass Budget and Transformation Pathways
Average concentrations of orthophosphate decreased by 16% and 29% during HRT1 and HRT2, respectively (
Table 3). Much of the orthophosphate was removed during early stages of HRT2 when temperature was relatively higher [
15], implying a biological role in the removal process.
The development of multiple linear regression models represents a simple and useful tool to understand, manage and design CWs where the goodness of fit represents the degree of correlation of key parameters [
47]. In this study, the influent and effluent concentrations of orthophosphate were found to be directly related as:
The difference between inflow concentration and outflow concentration is the loss of phosphorus from the water column (i.e., removal rate), or that assimilated by plants, microorganisms, and soil with phosphorus load. Reddy et al. [
48] found C
Effluent = 0.34 × C
inflow0.96; R
2 = 0.73; n = 373 for phosphorus loads ranging from 0.2 g phosphorus/m
2-yr to 1000 g phosphorus/m
2-yr. The areal phosphorus loading rate for this pilot study was 12.5 g /m
2-yr for HRT1 and 6.2 g /m
2-yr for HRT2. These values were at the lower end of the previously reported range [
48].
Significant removal of phosphorus is not normally expected in subsurface flow wetlands because phosphorus removal in this type of wetland is due to bacterial and plant uptake and precipitation of various phosphate salts including calcium phosphate (apatite or hydroxyapatite) [
1,
13,
46]. Phosphorus assimilation in vegetation was found to be short-term and dependent upon plant species, P loading, and wetland hydrology. Decomposition of detrital tissue resulted in rapid release of P into the water column. The modest removal of orthophosphate as found in this study using a relatively small sized wetland is consistent with the expectation of fully-grown wetland vegetation and the possibility of some export of organic matter in the form of bacterial biomass, root exudates and material, compounded by an evaporative increase in parameter concentrations.
Figure 12 shows the main components of the phosphorus transfer and transformation pathways in the wetland. The phosphorus mass transformation in the CW was quantified based on the influent and effluent loading data. Phosphorus retention mechanisms include uptake and release by vegetation, periphyton and microorganisms; sorption and exchange reactions with soils and sediments; chemical precipitation in the water column; and sedimentation and entrainment. Some export of organic matter in the form of bacterial biomass, root exudates and material are likely, compounded by an evaporative increase in parameter concentrations. In addition, adsorption to the gravel substrate and plant root surface provide sorption sites which are not saturated.
Vegetation, periphyton and microorganisms influence the phosphorus assimilation capacity by acting as transformers of phosphorus between biologically available and unavailable forms. Phosphorus assimilation could be both short-term storage (assimilation into vegetation, microorganisms, periphyton, and detritus) and long-term storage (assimilation by soil and accretion of organic matter). Under the short-term assimilation much of the phosphorus is released back into water upon vegetative decomposition [
49].
Abiotic processes include sedimentation, adsorption by sediments/soils, precipitation, and exchange processes between soil/sediment and the overlying water column. The net effect of vegetation on phosphorus retention depends on type of vegetation, rootshoot ratio, turnover rates of detrital tissues, C/P (carbon to phosphorus) ratio of the detrital tissue, and physicochemical properties of the water column [
48]. The balance between mineralization (i.e., breakdown of organic P to inorganic P) and immobilization (i.e., assimilation of inorganic P into microbial biomass) depends on the C/P ratio of the organic matter and type of electron acceptors involved in the decomposition (i.e., aerobic vs. anaerobic). A net removal of 72 g P/m
2 by the wetlands was estimated for this study. A modest removal of 21% of input load is a result of C/P ratio of the detrital tissue including organic matter in this study.
Microorganisms play an important role in the transformation of organic phosphorus to inorganic P in soils and sediments [
50]. The size of the CW played a major role in P removal process as subsurface flow treatment wetlands have area-dependent abilities to remove phosphorus. The catabolic activities catalyze the mineralization of organic phosphorus, while during growth of microorganisms assimilate and transiently store phosphorus in their biomass. A bacterial biomass C/P ratio of <20 resulting from the redox condition and the presence of selected electron acceptors indicating phosphorus is not limiting the system.
Sorption of phosphorus is a limited process because the adsorption capacity is dependent on the quantity of calcium and iron in the soil and as soon as all sorption sites are occupied no further phosphorus removal due to adsorption can occur. Storage in sediments and detritus in this study was 21% due to combined processes of sorption and chemical precipitation. Recently, light weight clay aggregates (LECA) used for CW substrate were found to achieve better TP removal efficiencies (72% when CW planted with
P. australis and 88% with
T. domingensis plants) [
13].
Although complex numerical models are available to estimate N, P and C retention and transport, a simple understanding of retention at the process level is important, but the overall assessment provided by mass balance and kinetic evaluations are often more useful in estimating long-term nutrient retention.
4.5. Potential Improvements of the CW Treatments Based on Pilot Study
As shown and compared to the results with other studies, this pilot wetland performed relatively well in terms of removing nutrients from the heavily loaded ammonia-rich ROC [
1]. The mass-based removal efficiency by CW for NO
3-N, NO
2-N, NH
4-N was found to be 61%, 32% and 42%, respectively (
Table 4). These results may reasonably be expected for a small sized pilot CW treatment system receiving ROC with such a high HLR under a low HRT. The highly loaded, ammonia rich ROC source water supported plant growth and decomposition, contributing to internal loading and cycling of ammonia. Various potential improvements to the CW based on the current pilot study could achieve better removal efficiency in a full-scale plant operation.
As a potential CW performance improvement measure active microbial processes such as enhanced nitrification would be needed for removal of ammonia and organic nitrogen concentration as found in the high TKN concentrations noted here. The relatively low reductions of TKN and ammonia in this study could be enhanced by operational modifications such as forced aeration to provide dissolved oxygen for enhanced carbon and nitrogen assimilation. The deficiency of labile carbon and DO in the ROC and within the wetland affected the removal efficiency. Low available DO contributed to partial nitrification of NH4+ into NO3−. Aeration would enable microorganisms to more completely degrade and thus enhance system efficiency for removing organic matter and nitrogen. Injection of carbon into CW is another conceptually viable system improvement option which could improve removal of nitrogen compounds. Control of influent pH (with the target of pH > 8) could be another potential improvement for achieving better removal efficiency of nutrients. Introduction of ammonia oxidizing bacteria (Nitrosomonas sp.) could also improve removal in conjunction with addition of forced air.
In recognition of the complexity of treating the high-strength ROC tested in this pilot study, a 3500 m
2 (0.35 hectare) demonstration wetland has been constructed at the City’s Advanced Water Purification Facility (AWPF) to assess the performance of an anaerobic-aerobic wetland system designed to implement these types of improvements. A general layout and flow diagram of the AWPF demonstration CW is shown in
Figure 13. The initial stage of treatment includes planted horizontal flow gravel subsurface wetlands (HSSF) with forced aeration to provide an aerobic environment for enhanced nitrification. The second stage is a planted upflow vertical flow (VF) cell for microbial denitrification under anaerobic conditions supplemented by organic carbon leaching from an organic substrate. The final surface flow (SF) cell provides final nutrient removal through denitrification and biological assimilation in an aesthetic aquatic wetland habitat, useful as an environmental education component while providing additional contaminant polishing. This system began receiving ROC in 2018 and is currently being studied to assess treatment performance.
Various wetland substrate materials can be tailored so that the treatment wetland would act like a bioreactor and remove various forms of soluble phosphorus and nitrogen compounds from the ammonia rich source water by enhancing sorption and chemical reaction sites. For example, in the AWPF demonstration CW (
Figure 13) the HSSF cells are sand-and-gravel-based filter beds planted with
S. californicus plants for nitrifying ammonia. The upward VF cells have a lower layer of gravel and an upper layer of peat moss that supports a diverse list of brackish plant species.
Another potential improvement would be to lengthen the HRT beyond the duration tested in this study to support greater nutrient removal from the system, or to supplement with limiting constituents (e.g., labile carbon, dissolved oxygen) under favorable pH conditions as discussed here. The former approach may be limited by land area available. The latter approach is currently being implemented in the AWPF demonstration wetland.