This study evaluated the nitrogen compound removal efficiency of a hybrid subsurface constructed wetland, which began treating milking parlor wastewater in Hokkaido, northern Japan, in 2006. The wetland's overall removal rates of total nitrogen (TN) and ammonium (NH4+-N) improved after the second year of operation, and its rate of organic nitrogen (Org-N) removal was stable at 90% efficiency. Only nitrate (NO3-N) levels were increased following the treatment. Despite increased NO3-N (maximum of 3 mg-N/L) levels, TN removal rates were only slightly affected. Removal rates of TN and Org-N were highest in the first vertical bed. NH4+-N removal rates were highest in the second vertical bed, presumably due to water recirculation and pH adjustment. Concentrations of NO3-N appeared when total carbon (TC) levels were low, which suggests that low TC prevented complete denitrification in the second vertical bed and the final horizontal bed. In practice, the beds removed more nitrogen than the amount theoretically removed by denitrification, as calculated by the amount of carbon removed from the system. This carbon-nitrogen imbalance may be due to other nitrogen transformation mechanisms, which require less carbon.

INTRODUCTION

The livestock industry is a major source of nitrogen pollution. In Japan, the livestock industry annually generates about 83 million tons of waste and 70 thousand million tons of nitrogen (MAFF 2009). Annual nitrogen waste was reduced from approximately 95 million tons in 1997 to about 83 million tons in 2013, using methods such as composting, converting to manure, treatment, and incineration. In spite of this achievement, the livestock industry faces pressure to reduce waste further. Roughly one-quarter of complaints against the livestock industry concern water pollution (MAFF 2010). Currently, the inorganic nitrogen level in waste effluent is limited to 700 mg-N/L (). This level might be reduced even further by regulators in 2016. However, stricter standards (total nitrogen (TN) limit of 120 mg-N/L, daily average of 60 mg-N/L) are only applied to facilities that release more than 50 m3/day. Since many small to medium scale farmers release less than that amount, these wastewater discharges are not strictly regulated. Farmers should be aware that regulation of nitrogen pollution is likely to become more stringent in the future.

Where adequate space exists, constructed wetlands offer an alternative to conventional activated sludge treatment. A constructed wetland mimics a natural wetland to treat different kinds of wastewater, including domestic and industrial sewage, agricultural wastewater, and acid-mine drainage wastewater (IWA 2000). Constructed wetlands have several benefits; they require little to no energy input, have low operation and maintenance cost, and can be designed to resemble the natural environment so as to benefit the local ecosystem (IWA 2000). There are two types of constructed wetlands: one uses free surface flow and the other uses subsurface flow. Among subsurface flow constructed wetlands, some flow vertically and others horizontally. In a vertical flow subsurface constructed wetland, water flows perpendicular to the bed; in a horizontal flow subsurface constructed wetland, water flows parallel to the bed. A hybrid subsurface constructed wetland (HSCW) uses both vertical and horizontal flows.

In a constructed wetland system, nitrogen undergoes many transformations before it is removed (IWA 2000; Vymazal 2007). Often some nitrogen compounds are not removed due to the complex route they must take. In a classic nitrogen removal pass-way for example, high organic nitrogen (Org-N) wastewater is first ammonified or mineralized. Next, ammonium () is nitrified into nitrite (), and then to nitrate (). Finally, is denitrified into nitrogen gas (N2). In addition to this classical route, there are absorption, adsorption, assimilation, and routes recently discovered by researchers. These new routes are partial nitrification-denitrification, Anammox, and Canon. Engineers are attempting to incorporate these processes in their wastewater systems to remove nitrogen because they require less oxygen and carbon compared with classical route processes. All of these processes, including those of the classical route, are sensitive to environmental conditions (such as temperature, pH, dissolved organic and carbon sources) (Vymazal 2007) and system maturity. Hence, all nitrogen compounds and their removal rates must be carefully followed and quantified. However, there are very few examples of constructed wetlands that work with high organic wastewater. Furthermore, the monitoring periods for many of them are very short. Therefore, a thorough investigation of these systems, over a prolonged period, is crucial.

This paper outlines how the nitrogen compounds were transformed and removed in a HSCW, which has treated dairy farm wastewater for more than 7 years.

MATERIALS AND METHODS

The studied HSCW is located in the Town of Embetsu, Hokkaido, northern Japan (N44 °45.0′ latitude and E141 °48.4′ longitude). The town's average annual temperature is 6.45 °C and total annual precipitation is 1,112 mm. This HCSW treats milking parlor effluent, discharged from a free stall-milking parlor that holds about 120 dairy cows.

The HCSW is constructed of two vertical beds (VB) and one horizontal bed (HB). The areas of the VBs are 160 m2 each and the HB is 336 m2, as shown in Figure 1 and Table 1. The main vegetation is Phragmites australis. Feces, urine, non-shippable milk, and detergent are sent from the milking parlor, then mixed and stored in the sedimentation tank (underground storage tank). The stored sewage flows into the first vertical bed (VB1), and then into the second vertical bed (VB2). Part of the VB2 effluent is recycled back to the inlet of VB2 to increase removal efficiency (RE). This recycling is done by a recirculation pump equipped with a timer and it is activated every other hour. The remaining water from VB2 flows into the HB and then finally water is discharged into a nearby stream. Inlets of VB1 and VB2 are each equipped with a pump and a French based self-priming siphon (Molle et al. 2005) to supply water intermittently. The inlet of HB is also equipped with a pump. Details of the bed media are listed in Table 1. Porous recycled glass material called Supersol© is used on the surface beds of VB1 and VB2, and autoclaved aerated lightweight concrete (ALC) is used on the bed of HB. Both these materials insulate the beds to prevent water from freezing in winter (Kato et al. 2013). These cover materials and reinforced bypass tubes are used in all of the beds to prevent clogging, which is a problem for all constructed wetlands particularly during the startup period (Kato et al. 2013). These improvements enabled this system to treat wastewater with high loads of organic pollutants, even in extreme cold conditions. VB1 and VB2 are partitioned into two long strips, so one of the sections can be in a dry phase to facilitate oxidation and drying of the organic layer. The sections are used alternately during the growing seasons.
Table 1

Bed sizing and media composition

BedArea (m2)Average depth (m)Volume (m3)Bed materialCover material
VB1 160 0.75 120 River gravel Supersol 
VB2 160 0.71 114 Clinker ash River gravel Supersol 
HB 336 0.72 242 Sand ALC 
Total 656 – 476 – – 
BedArea (m2)Average depth (m)Volume (m3)Bed materialCover material
VB1 160 0.75 120 River gravel Supersol 
VB2 160 0.71 114 Clinker ash River gravel Supersol 
HB 336 0.72 242 Sand ALC 
Total 656 – 476 – – 
Figure 1

Schematic diagram of the constructed wetland. HB and VB represent horizontal bed and vertical bed, respectively.

Figure 1

Schematic diagram of the constructed wetland. HB and VB represent horizontal bed and vertical bed, respectively.

Samples were collected on a monthly to bimonthly basis at the inlet of the system and the outlet of each bed. The samples were analyzed for TN, , , , and total carbon (TC). TN and TC were measured using an elemental analyzer (Elementar vario MAX; Elementar Analysensysteme GmbH, Hanau, Germany). Ammonium nitrogen was measured using a segmented-flow analysis system (QuAAtro; SEAL Analytical GmbH, Norderstedt, Germany). Org-N was calculated by subtracting the inorganic nitrogen () from TN.

Concurrently, data were downloaded from the pressure sensor (water level meter) with a data logger (DL/N70; Sensor TechnikSirnach (STS) AG, Sirnach, Switzerland, or S&DL Mini; Oyo Corp., Tokyo, Japan) placed inside each siphon pit and pump pit (Figure 1). The data logger recorded changes in water level at a 10-min interval. The system's daily inflow (unit of m3/day) was calculated by multiplying the base area of the siphon tank by the changes in water level of the pit or tank and the number of times the siphon was activated. The daily flow was calculated by adding all water volumes that entered the pit per day.

Rainfall data were obtained using a tipping-bucket type rain gauge installed on site or from the Japan Meteorological Agency's local station (AMeDAS Embetsu Station).

The annual input loads were calculated by multiplying the concentrations of the samples by the total volume of the sampling periods. One sampling period was defined as the time from the sampling date to the day prior to the next sampling date. The total annual load was calculated by adding all the loads throughout the year. The average daily loads were calculated by dividing the annual load by total number of days, and the average annual concentration was calculated by dividing the total annual load by the total annual volume. Performance of the system was evaluated based on concentration and loads. The two different methods of evaluation were used. These methods are purification efficiency (PE; based on concentrations) and removal efficiency (RE; based on loads). The equations to calculate these rates are as follows:
formula
1
formula
2
where Cin is the influent concentration, Cout is the effluent concentration, Min is the influent load, and Mout is the effluent load.

RESULTS AND DISCUSSION

Influent and effluent concentrations and loads

Figure 2 shows influent and effluent concentrations and loads for TN, , , and Org-N. TN concentrations and loads were the highest in Year 1. This is likely due to the large amount of unshippable milk disposed into the constructed wetland in that year. Influent concentrations and loads fluctuated, but the effluent concentrations and loads remained stable. The PE and RE of TN improved during the first 4 years of operation and remained stable thereafter. Since PE and RE of Org-N were stable around 90% during all operational years, enhanced PE and RE of TN are due to improved removal. The removal rate was high in Year 1 and then decreased significantly in Year 2. However, the rate increased steadily after Year 2. This pattern occurred because was removed via adsorption to bed media during Year 1. After Year 2, the growth of plants and their root systems facilitated removal by bacterial colonies in the rhizomes. Also, increased soil biota and other macro-organisms, such as earthworms, presumably contributed to nutrient recycling. However, simple reduction of (transformation to through nitrification) did not contribute to TN reduction. This indicates that was further transformed into N2 gas by the denitrification process or through other mechanisms. Other mechanisms may include assimilation by biomass, plant uptake and ammonia volatilization. However, removal by these mechanisms should have been minimal compared with the overall nitrogen removal (IWA 2000). The influent concentrations and loads of were very low and they increased significantly in the effluent. Even with the increased (maximum of 3 mg-N/L), PE and RE of TN were only slightly affected. The effect was minimal because the average TN concentration of the effluent was 20.17 ± 10.69 mg-N/L and the average load was 118.24 ± 54.83g-N/day. Thus, even the highest NO3-N concentration was responsible for just 15% of TN concentration.
Figure 2

Inflow and outflow concentrations, loads and removal efficiencies for nitrogen pollutants.

Figure 2

Inflow and outflow concentrations, loads and removal efficiencies for nitrogen pollutants.

Composition of wastewater in the system

Figure 3 shows nitrogen composition, based on the initial load at each sampling point. For the initial inflow, and Org-N concentrations were about 43 and 56% of the initial TN, and concentration was almost zero. In the effluent, Org-N and concentrations were approximately 6 and 10% of the original TN load, respectively. Nitrate concentration drastically increased after passing through VB1, and then decreased gradually to about 1% of the initial TN concentration.
Figure 3

Average composition of nitrogen compounds in the wastewater based on the initial TN load (initial TN = 100%).

Figure 3

Average composition of nitrogen compounds in the wastewater based on the initial TN load (initial TN = 100%).

Generally, Org-N is removed by ammonification, nitrification, and denitrification. However, the Org-N load decreased drastically without an increase in load in VB1. There should have been two removal mechanisms. The first should have been the classical route, which involves adsorption, ammonification, nitrification, and denitrification. The increase in load could not be observed because of the spontaneous reaction or the adsorption. The second mechanism should have been physical filtering of particulate Org-N.

Figure 4 shows nitrogen removal performed by each bed. More than half the removal of TN and Org-N was done by VB1. Since solid waste accumulation mainly occurred in VB1 (organic layer of about 25 cm in depth) compared with VB2 (organic layer of about 5 cm), it is likely that removal of particulate nitrogen, including Org-N, occurred in VB1. Therefore it is likely that the highest removal of TN occurred in VB1 due to the physical removal of particulate Org-N. Ammonium removal was the same in VB1 and VB2. The high removal rate in VB2 was probably due to recirculation of the effluent and neutralization of pH from clinker ash. In many studies, removal efficiencies were improved by recirculation (Brix & Arias 2005; Ayaz et al. 2012). Since denitrifying and nitrifying bacteria favor a neutral pH, it is possible the clinker ash neutralized the wastewater and helped the system to maintain high removal rate. Unlike TN, , and Org-N, concentrations increased rapidly in VB1 and were removed by the following beds.
Figure 4

Contribution of each bed for removal of different nitrogen pollutants.

Figure 4

Contribution of each bed for removal of different nitrogen pollutants.

Relationship between nitrogen and carbon

Figure 5 shows the concentrations of and in VB2in and HBin, plotted against corresponding TC concentrations. Concentrations of and TC were positively correlated (R2 = 0.76). In contrast, concentrations had a completely different relationship. concentrations were low at high TC concentrations but started to appear at low TC. This may have occurred because the denitrifying bacteria did not have enough carbon to denitrify all . Therefore, was able to be nitrified to , but was not denitrified and NO3-N was released. A similar trend was observed in HB, but with a lower peak. If further denitrification is required in the future, one approach could be the addition of external carbon. However, current TN concentrations in the effluent are far below the limit so further reduction may not be necessary.
Figure 5

Nitrogen versus TC concentration for VB2in (left) and HBin (right).

Figure 5

Nitrogen versus TC concentration for VB2in (left) and HBin (right).

The amount of theoretical carbon used for denitrification was calculated using the assumption that all was denitrified, and none was lost due to ammonia volatilization, plant assimilation or biomass uptake. Under this very simplified assumption, every gram of consumed 1.07 g of carbon for denitrification (Vymazal 2007). Therefore, the theoretical carbon consumption could be calculated as follows:
formula
3
where Carbonremoved, denitrification represents the mass of carbon used if all the (and transformed ) were denitrified into N2. is the total amount of removed from that bed, and .

When Carbonremoved, denitrification was compared with the actual amount of TC removed from the same bed, there were several instances where Carbonremoved, denitrification exceeded the actual amount of TC removal. This indicates more was removed than projected by the theoretical model. Therefore in at least some instances, it is assumed that TN was reduced due to other inorganic nitrogen removal mechanisms, which require less carbon. In other studies, there are cases where Anammox bacteria were possibly involved in treating wastewater from the dairy industry or at least the theoretical potential of Anammox was recognized (Tao et al. 2012; Waki et al. 2013).

Both partial nitrification-denitrification and Anammox require nitrite () (van Dongen et al. 2001). Many samples had undetectable levels of but some had more than 1 mg-N/L. Concentrations of exceeding 1 mg-N/L (high as 6.1 mg-N/L) were observed at all sites except in the outflow. It is commonly understood that the Anammox process requires the ratio of to be 1/1.32, but this system never reached that ratio. However, readily decomposes so this compound may have decomposed within the bed and during transportation of the samples.

CONCLUSIONS

A thorough analysis of the removal of nitrogen compounds indicates that the PEs and REs of TN improved during the first 4 years and remained stable thereafter. The system's PE and RE of Org-N were stable at around 90% and improved from 40% PE and 50% RE in Year 2 to about 90% in Year 6. This improvement was likely due to the growth of reed and their roots, flourishing of nitrifying and denitrifying bacteria in the rhizomes, development of soil biota, and ecosystem features like earthworms that enhance recycling of nutrients and stabilize infiltration rates. The removal rates of TN and Org-N were highest in VB1 and gradually decreased throughout the system. This is likely due to the physical filtration by VB, which was effective in removing particulate Org-N. Implementation of water recycling in VB2 most likely contributed to an increase in overall removal of . The TC concentration was positively, but moderately, correlated with concentration. On the other hand was not observed at high TC concentrations, but started to appear at low TC concentrations. This indicates there were occasions when insufficient carbon was available to completely denitrify all inorganic nitrogen. Finally, there were cases where more was being removed relative to the expected amount of removed TC from denitrification. Therefore, other mechanisms may have played a role in ammonia removal, such as partial nitrification-denitrification and Anammox.

ACKNOWLEDGEMENTS

We would like to express our sincere appreciation to the Ministry of Agriculture, Forestry and Fisheries, Japan for financial support. We are also thankful to the owner of the dairy farm in Embetsu for providing us with an opportunity to carry out this study in the farm. This work was also supported by JSPS KAKENHI Grant Number 26292185.

REFERENCES

REFERENCES
Ayaz
S. Ç.
Aktaş
Ö.
Findik
N.
Akça
L.
Kinaci
C.
2012
Effect of recirculation on nitrogen removal in a hybrid constructed wetland system
.
Ecological Engineering
40
,
1
5
. Doi:10.1016/j.ecoleng.2011.12.028.
Brix
H.
Arias
C. A.
2005
The use of vertical flow constructed wetlands for on-site treatment of domestic wastewater: new Danish guidelines
.
Ecological Engineering
25
,
491
500
. Doi:10.1016/j.ecoleng.2005.07.009.
IWA Specialist Group on use of Macrophytes in Water Pollution Control
2000
Constructed Wetlands for Pollution Control, Processes, Performance, Design and Operation
.
IWA Publishing
,
London
,
UK
.
Ministry of Agriculture, Forestry, Fisheries (MAFF)
2009
Situation of livestock industry (in Japanese)
. ).
Ministry of Agriculture, Forestry, Fisheries (MAFF)
2010
Status of number of complaints related to livestock industry (in Japanese). http://www.maff.go.jp/j/chikusan/kankyo/taisaku/pdf/kujou22.pdf
(
accessed 24 June 2014
).
Molle
P.
Liénard
A.
Boutin
C.
Merlin
G.
Iwema
A.
2005
How to treat raw sewage with constructed wetlands: an overview of the French systems
.
Water Science and Technology
51
(
9
),
11
21
.
Tao
W.
He
Y.
Wang
Z.
Smith
R.
Shayya
W.
Pei
Y.
2012
Effects of pH and temperature on coupling nitritation and anammox in biofilters treating dairy wastewater
.
Ecological Engineering
47
,
76
82
. Doi:10.1016/j.ecoleng.2012.06.035.
van Dongen
L. G. J. M.
Jetten
M. S. M.
van Loosdrecht
M. C. M.
2001
The combined Sharon/Anammox Process – A sustainable method for N-removal from sludge water. STOWA Report. IWA Publishing, London, UK.
Vymazal
J.
2007
Removal of nutrients in various types of constructed wetlands
.
Science of the Total Environment
380
,
48
65
. Doi:10.1016/j.scitotenv.2006.09.014.
Waki
M.
Yasuda
T.
Fukumoto
Y.
Kuroda
K.
Suzuki
K.
2013
Effect of electron donors on anammox coupling with nitrate reduction for removing nitrogen from nitrate and ammonium
.
Bioresource Technology
130
,
592
598
. Doi:10.1016/j.biortech.2012.12.101.