JEQ Grow Your Career With ASA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 27 October 2006
Published in J Environ Qual 35:2360-2373 (2006)
DOI: 10.2134/jeq2006.0038
© 2006 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Søvik, A. K.
Right arrow Articles by Wachniew, P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Søvik, A. K.
Right arrow Articles by Wachniew, P.
Agricola
Right arrow Articles by Søvik, A. K.
Right arrow Articles by Wachniew, P.
Related Collections
Right arrow Wetlands and Aquatic Processes
Right arrow Air Pollution

TECHNICAL REPORTS

Wetlands and Aquatic Processes

Emission of the Greenhouse Gases Nitrous Oxide and Methane from Constructed Wetlands in Europe

A. K. Søvika,*, J. Augustinb, K. Heikkinenc, J. T. Huttunend, J. M. Neckie, S. M. Karjalainenc, B. Kløvea,f, A. Liikanend, Ü. Manderg, M. Puustinenh, S. Teiterg and P. Wachniewe

a Bioforsk-Norwegian Institute for Agricultural and Environmental Research–Soil and Environment Division, Frederik A. Dahls vei 20, 1432 Ås, Norway
b Institute of Primary Production and Microbial Ecology, The Leibniz-Centre for Agricultural Landscape and Land Use Research (ZALF), Eberswalder Str. 84, D-15374 Müncheberg, Germany
c North Ostrobothnia Regional Environment Centre (NOREC), P.O. Box 124, 90101 Oulu, Finland
d Department of Environmental Sciences, Bioteknia 2, University of Kuopio, P.O. Box 1627, 70211 Kuopio, Finland
e AGH-University of Science and Technology, Faculty of Physics and Applied Computer Science, Department of Environmental Physics, al. Mickiewicza 30, 30-059 Kraków, Poland
f Water Resources and Environmental Engineering Laboratory, Department of Process and Environmental Engineering, PL 4300, 90014 University of Oulu, Finland
g University of Tartu, Institute of Geography, 46 Vanemuise Street, 51014 Tartu, Estonia
h Finnish Environmental Institute (SYKE), P.O. BOX 140, 00251 Helsinki, Finland

* Corresponding author (anne.sovik{at}bioforsk.no)

Received for publication January 26, 2006.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIAL AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
The potential atmospheric impact of constructed wetlands (CWs) should be examined as there is a worldwide increase in the development of these systems. Fluxes of N2O, CH4, and CO2 have been measured from CWs in Estonia, Finland, Norway, and Poland during winter and summer in horizontal and vertical subsurface flow (HSSF and VSSF), free surface water (FSW), and overland and groundwater flow (OGF) wetlands. The fluxes of N2O–N, CH4–C, and CO2–C ranged from –2.1 to 1000, –32 to 38 000, and –840 to 93 000 mg m–2 d–1, respectively. Emissions of N2O and CH4 were significantly higher during summer than during winter. The VSSF wetlands had the highest fluxes of N2O during both summer and winter. Methane emissions were highest from the FSW wetlands during wintertime. In the HSSF wetlands, the emissions of N2O and CH4 were in general highest in the inlet section. The vegetated ponds in the FSW wetlands released more N2O than the nonvegetated ponds. The global warming potential (GWP), summarizing the mean N2O and CH4 emissions, ranged from 5700 to 26000 and 830 to 5100 mg CO2 equivalents m–2 d–1 for the four CW types in summer and winter, respectively. The wintertime GWP was 8.5 to 89.5% of the corresponding summertime GWP, which highlights the importance of the cold season in the annual greenhouse gas release from north temperate and boreal CWs. However, due to their generally small area North European CWs were suggested to represent only a minor source for atmospheric N2O and CH4.

Abbreviations: CW, constructed wetland • FSW, free surface water • GWP, global warming potential • HSSF, horizontal subsurface flow • OGF, overland and groundwater flow • SSF, subsurface flow • VSSF, vertical subsurface flow


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIAL AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
THERE IS A GENERAL AGREEMENT that the observed increase in the atmospheric concentrations of greenhouse gases like nitrous oxide (N2O), methane (CH4), and carbon dioxide (CO2) due to use of fossil fuel, land-use changes, and intensive agriculture has led to a warming of the earth's climate. Production of N2O in the soil and subsequent emission accounts for about 70% of both anthropogenic and natural N2O sources (Smith, 1997). Other sources are the oceans, biomass burning, and emissions from industry and automobiles. Nitrous oxide is increasing in the atmosphere at a rate of 0.3% a year, and the gas is anticipated to be responsible for about 5% of the global warming (IPCC, 2001a). Nitrous oxide has a global warming potential of 296 relative to CO2 (over a 100-yr time horizon) (IPCC, 2001b). Methane arises from burning of fossil fuel, enteric fermentation, as well as terrestrial and aquatic systems (Mosier, 1998). Terrestrial production of methane is mainly associated with wetlands, flooded rice paddies, and landfills (Bartlett and Harriss, 1993). Methane has a global warming potential of about 23 relative to CO2 (100-yr time horizon) (IPCC, 2001b) and the gas contributes to 25% of the global warming (Mosier, 1998).

A substantial amount of work has been devoted to measure fluxes of N2O and CH4 from various types of ecosystems such as agricultural soils (Baggs et al., 2000; Machefert et al., 2002), rice paddies (Yang et al., 2003; Jain et al., 2004), riparian buffer zones (Hefting et al., 2003), and pristine and managed wetlands (Bubier et al., 1993; Glenn et al., 1993; Martikainen et al., 1995; Augustin et al., 1998; Maljanen et al., 2001; Kang and Freeman, 2002). Natural wetlands (e.g., peatlands) are characterized by anaerobic conditions in their water-saturated soils and low turnover rates for organic matter, and are thus important terrestrial sinks for C and N (Augustin et al., 1998). At the same time, these systems may emit substantial amounts of methane. The production of N2O may be of minor importance in natural wetlands as lack of oxygen restricts nitrification and the reduced availability of NO3– limits denitrification (Regina et al., 1996).

Constructed wetlands used for wastewater treatment can be seen as combinations of natural wetlands and conventional wastewater treatment plants and are constructed to reduce input of nutrients to water bodies and thus prevent eutrophication. Constructed wetlands are working at the same level of efficiency as conventional systems and the management costs are significantly lower than in the conventional prototypes (Kadlec and Knight, 1996). Constructed wetlands have been successfully used to remove nutrients and organic matter from different types of wastewater such as municipal wastewater (Harris and Mæhlum, 2003), agricultural runoff (Braskerud, 2002; Koskiaho et al., 2003; Koskiaho and Puustinen, 2005), landfill leachate (Mæhlum, 1995), and peat harvesting runoff (Heikkinen et al., 1995). In conventional municipal wastewater treatment plants, P and organic matter are most often removed in a biochemical activated sludge process. As the process removes little N, the effluents are rich in ammonium and occasionally also nitrate depending on the aeration efficiency of the treatment plant. Only recently has N reduction been required in conventional systems (Kadlec and Knight, 1996). Constructed wetlands have been used as a low-cost option to remove part of the N. In CWs the N is removed in the nitrification–denitrification process, which converts nitrate to N2 and N2O. The limiting factors vary due to differences in hydraulic loading rate, and N and C content in individual CWs. For municipal wastewater rich in ammonium, the limiting factor for the N removal is usually oxygen. For agricultural runoff, where N mainly occurs as nitrate, the limiting factor is the denitrification step controlled either by C or by the reaction time. Peat mining waters contain about equal portions of organic and mineral N (ammonium and nitrate) (Kløve, 2001), and the process could be controlled by reaction time when the hydraulic load is high.

When wetlands are used for purification of wastewaters, their hydrological and microbial processes will likely be changed. The wetlands will be constantly waterlogged and the temperature variation will partly be controlled by the wastewater temperature. Increased input of nutrients and organic matter will increase the productivity of the ecosystem and probably increase the production of gaseous compounds such as N2O and CH4. The studies conducted so far have indicated that CWs have high N2O and CH4 emissions (Freeman et al., 1997; Tanner et al., 1997; Fey et al., 1999; Gui et al., 2001; Johansson et al., 2003, 2004; Mander et al., 2003, 2005a, 2005b; Karjalainen et al., 2005; Stadmark and Leonardson, 2005; Teiter and Mander, 2005; Liikanen et al., 2006). The question then arises if CWs, used to protect freshwater ecosystems, are a solution to an environmental problem or if they substitute one problem with another, i.e., reducing water pollution but increasing the emission of greenhouse gases. The total area of CWs worldwide is small compared to that of all natural wetlands and agricultural areas, but the worldwide increase in the development of CWs necessitates an understanding of their potential atmospheric impact, in light of the trend that natural wetlands in many countries are decreasing (e.g., USA) while environmental regulatory agencies are trying to stimulate an increase in CW acreage.

This work summarizes the greenhouse gas fluxes measured in CWs in four countries in Europe, i.e., Estonia, Finland, Norway, and Poland. The fluxes have been measured both during winter and summer in the different types of CWs. The fluxes are compared to those previously reported for other temperate and boreal ecosystems to assess the importance of north temperate and boreal CWs as sources of greenhouse gases to the atmosphere. For comparison, reference greenhouse gas flux data are available from conventional municipal wastewater treatment plants (e.g., Hanaki et al., 1992; Zheng et al., 1994; Czepiel et al., 1995).


   MATERIAL AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIAL AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
This work summarizes results from 10 CWs: two Estonian (Kodijärve, Kõo), four Finnish (Hovi, Kompsasuo, Lakeus, Ruka), three Norwegian (Skallstuggu, Ski, Skjønhaug), and one Polish (Nowa Slupia) CW (Table 1). The systems studied include the most commonly used constructed wetland treatment systems in the northern temperate and boreal zones in Europe: subsurface flow (SSF) systems, ponds with a free water surface, and peatlands with overland flow and groundwater flow. Fluxes of greenhouse gases and some water quality characteristics (Table 2) were measured in the selected systems from 2001 to 2003 in the research consortium ‘Process-based integrated management of constructed and riverine wetlands for optimal control of wastewater at catchment scale (PRIMROSE)’ (EVK1-CT-2000–00065) of the EU fifth framework program. Detailed site-specific studies on the N2O, CH4, and CO2 fluxes in this summary have been previously presented for the Estonian sites by Mander et al. (2003, 2005a, 2005b) and Teiter and Mander (2005), and for the Kompsasuo wetland in Finland by Liikanen et al. (2006). Nitrous oxide exchanges in the Lakeus and Ruka wetlands in Finland have been earlier described by Karjalainen et al. (2005), whereas the details of the N2O and CH4 fluxes in the Norwegian wetlands Ski and Skjønhaug have been studied by Kløve et al. (2005) and Søvik and Kløve (unpublished data, 2006), respectively.


View this table:
[in this window]
[in a new window]
  		Table 1. Characteristics of the constructed wetlands presented in this work. OGF, overland and groundwater flow; HSSF, horizontal subsurface flow; VSSF, vertical subsurface flow; FSW, free surface water flow.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Chemical characteristics (mean ± standard error of the mean (SE) during the experimental period) of the water in the inlet and the outlet from the various constructed wetlands.

 
Study Sites
Estonia
The horizontal subsurface flow wetland of Kodijärve is a planted sand filter (constructed in October 1996; Tartu County, northern temperate zone) that purifies wastewater from a hospital for about 40 persons (Mander et al., 2001). The system consists of two beds of coarse sand, each 156.3 m2. The long-term (1997 to 2004) average discharges are 2.2 and 2.3 m3 d–1 and the average retention times are 6 to 7 and 7 to 8 d in the left and right bed, respectively.

The hybrid treatment wetland system in Kõo (constructed in 2000; Viljandi County, northern temperate zone) consists of a two-bed VSSF filter (VSSF; 2 by 64 m2, filled with crushed limestone, planted with common reed, Phragmites australis), a HSSF (365 m2, filled with 15 to 20 mm crushed limestone, planted with common cattail [Typha latifolia] and reed), and two free surface water beds (FSW; 3600 and 5500 m2, planted with T. latifolia). The Kõo hybrid system purifies raw municipal wastewater generated by about 300 population equivalents (PE). The average discharge for VSSF + HSSF is about 60 to 70 m3 d–1 and the average retention time in the HSSF is 3 to 4 d.

Finland
The Hovi wetland (60°25' N, 24°22' E) (6000 m2) was constructed in 1998 on arable land for the treatment of highly loaded agricultural runoff water. The wetland consists of an area with deep open water (1 to 2 m deep) and an area with shallow water (0.3 to 0.6 m deep) vegetated with cattail (Typha latifolia), club-rush (Scirpus sylvaticus), common water plantain (Alisma plantago-aquatica), reed canary grass (Phalaris arundinacea), meadowsweet (Filipendula ulmaria), yellow flag (Iris pseudacorus), and compact rush (Juncus conglomeratus). The water residence time during high water conditions in springtime is on average 33 h and the mean discharge is 78.6 m3 d–1 (Koskiaho, 2003; Koskiaho et al., 2003; Liikanen et al., 2004).

The Kompsasuo wetland (2.4 ha) with a combined OGF is situated in the southern aapa mire zone in northern Finland (65°44'43'' N, 25°57'80'' E). It was constructed in 1987 on a natural peatland at Kompsasuo to purify draining water from the adjacent peat harvesting area. The mire is surrounded by coniferous forests of the mid-boreal type, and the prevailing peat type is Spaghnum-Carex. Vegetation consists mainly of Menyanthes trifoliata, Carex lasiocarpa, and Potentilla palustris. The average discharge and retention time during the experimental period were 1857 m3 d–1 and 6 to 17 d, respectively.

The Lakeus wetland of the Lakeus Central Treatment Plant in Kempele is a constructed FSW wetland (44000 m2) in the mid-boreal zone (64°53'82'' N, 25°27'76'' E). The wetland was constructed in 1996 on natural reed vegetation and it polishes chemically and biologically purified municipal wastewater (Vääräniemi and Lakso, 2000). The mean discharge to the wetland was 3624 m3 d–1 from 2002 to 2003. The average retention time during the experimental period was 1.5 d.

The Ruka wetland is located in the northern boreal zone (66°10'10'' N, 29°07'28'' E) and has combined OGF (8200 m2). The wetland is part of the Ruka Sewage Treatment Plant in Kuusamo. It was constructed in 1995 on a natural pine mire to polish mechanically and chemically purified municipal wastewater from a skiing resort (Pirttijoki, 1996). The prevailing peat type is Carex-Sphagnum peat and the present vegetation is dominated by sedges. The average retention time and discharge during the experimental period were 1.9 d and 289 m3 d–1, respectively.

Norway
The Skallstuggu wetland (65 m2) was constructed in 1999 and treats wastewater coming from a tourist resort in Nord-Trøndelag (middle of Norway, northern temperate zone). The system consists of an unsaturated pre-filter, a saturated filter with shell sand, and infiltration ditches in peat soil that are filled with shell sand. The wetland is a combined OGF. The discharge and retention times are difficult to estimate as the use of the tourist resort varies considerably.

The meso-scale SSF wetland at Ski (southeastern Norway) was constructed in 1999 inside a ventilated greenhouse (Kløve et al., 2005). The wetland receives wastewater from a single domestic household. The system consists of a 50 cm high pre-filter unit (0.05 m2) with unsaturated vertical flow and a 3 m long saturated unit with horizontal flow (0.9 m2). Both filters are filled with coral sand and no vegetation was planted on the filter surfaces. The pre-filter unit may be removed and the system operated solely with the HSSF unit. The discharge varied between 0.012 and 0.072 m3 d–1, giving retention times of 18 and 3 d, respectively.

The Skjønhaug wetland was constructed in 1999 and is a FSW wetland polishing chemically treated municipal wastewater in southeastern Norway (Søvik and Kløve, unpublished data, 2006). The wetland consists of three ponds as well as unsaturated filters with light weight aggregates, and covers an area of 4000 m2. The second and third ponds were planted with yellow flag (Iris pseudacorus), cattail (Typha latifolia), and common clubrush (Scirpus lacustris). The average discharge is 600 m3 d–1 and the average retention time was 4 d during summer times.

Poland
The CW is located in the village Nowa Slupia in central Poland (21°05' E, 50°52' N). The treatment system was built in 1995 to clean municipal wastewaters from the Nowa Slupia sewage system as well as wastewaters from septic tanks, which are not connected to the sewage system. The treatment system serves about 1500 village inhabitants. The designed average diurnal flow of this system is equal to 150 to 200 m3 d–1. The HSSF system consists of three parallel gravel cells (total area of 5616 m2) overgrown with common reed (Phragmites australis). The cells are filled with one layer of sand and three layers of gravel of different granular sizes. A sedimentation pond and an aeration chamber provide primary treatment.

Gas Measurements
Chamber locations and measuring periods for all CWs in all countries are listed in Table 3. The measurements were performed with dark chambers, thus the CO2 fluxes did not include photosynthesis. The presented CO2 fluxes represent dark respiration of plants and soil biota, nominated hereafter as community CO2 production. The concentration of gases was determined by gas chromatography according to Sitaula et al. (1992), Nykänen et al. (1995), and Loftfield et al. (1997).


View this table:
[in this window]
[in a new window]
  		Table 3. The dimensions and locations of the chambers, the measuring period, the months covering the summer and winter periods of the year as well as the number of measurements for each specific wetland.

 
Estonia
The fluxes of N2O, CH4, and CO2 were measured with the static chamber method (Hutchinson and Livingston, 1993). The chambers had a cover made from PVC, were painted white to avoid heating, and were sealed with a water-filled ring at the soil surface. At the end of each 1-h measuring period, gas samples were taken from the chambers with previously evacuated gas bottles (100 mL).

Finland
The fluxes of N2O, CH4, and CO2 were measured with the static chamber method, i.e., aluminum frames were inserted into the soil at each measuring point and an aluminum chamber was placed over the frame (Nykänen et al., 1995). Gas tightness was ensured by filling the groove of the frame with water. In winter, the chambers were placed directly into the soil and tightened with snow, or directly on the ice in the case of Hovi. Floating aluminum chambers (Huttunen et al., 2002) were used for flux measurements in the pools and ponds. Gas samples (50 mL) were collected from the chambers with polypropylene syringes (Terumo Europe) equipped with 3-way stopcocks (Codan Steritex) at 4- to 7-min intervals over 24- to 30-min measuring periods. In winter gas samples were collected at 15-min intervals over 60-min measuring periods. Gas ebullition was measured in the Hovi and Lakeus wetlands at 5 sampling points in each wetland (3 in the deep part and 2 in the vegetation zone of Hovi wetland, and 5 in the pond near the outlet of Lakeus wetland) with submerged bubble traps from July to October 2002.

Norway
The fluxes of N2O, CH4, and CO2 were estimated by use of closed cylindrical metal chambers, which were either pressed into the filter material to ensure a water lock by the groundwater table or placed on top of metal bars driven into the sediments in the ponds. Gas samples were taken from the chambers after 0, 30, and 60 min. In the Ski CW, samples were collected during eight different experimental runs in 2001, i.e., the system was run alternately with and without the pre-filter unit and with high and low hydraulic load (each experimental condition was tested twice).

Poland
Stainless steel chambers with the form of truncated pyramids were used for gas flux measurements (CH4 and CO2). The chambers were sealed with a water-filled ring at the soil surface. Samples were collected every 10 and 20 min (summer and winter campaigns, respectively) with gas-tight syringes (50 mL).

Data Analysis
The wetlands were divided into four groups: (i) HSSF, (ii) VSSF, (iii) FSW, and (iv) wetlands with a combination of OGF. In both the FSW and OGF wetlands, the sampling locations were divided into three sections from the inlet to the outlet along the flow direction (see Table 3).

The year has been divided into a summer period and a winter period for all wetlands. These periods cover different months for the various wetlands (see details in Table 3). The wetland at Ski was located inside a greenhouse where the temperature was always above zero. The measurements at Ski were divided into a "summer" (June to the beginning of November) and a "winter" period (late November and December) based on the temperature in the porewater. The summer period had temperatures above 10°C, whereas during the winter period the temperatures were below 8°C.

In this work the term flux is used for both positive and negative gas exchanges. For fluxes of N2O and CH4, the term emission is used for positive fluxes, while the term consumption is used for negative fluxes. The gas flux data were found to be skewed. As statistical tests require data that are normally distributed, the populations were log-transformed to obtain normal distribution.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIAL AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Fluxes of N2O–N
The large temporal and spatial variation of N2O–N fluxes covered a range of –2.1 to 1000 mg m–2 d–1. The average flux of all data from the summer season was 91 mg m–2 d–1, whereas in the winter season it was 4.6 mg m–2 d–1 (the total average from the two seasons was calculated from the average of each CW, n = 12). During the summer season the average flux for each of the four wetland types was in the range of 0.90 to 440 mg m–2 d–1, whereas during the winter, the average flux ranged from 0.08 to 5.2 mg m–2 d–1. Both during the summer and the winter, OGF and VSSF wetlands had the lowest and the highest average flux, respectively (Table 4). For HSSF wetlands, the flux both during the summer season and the winter season was higher in the inlet section of the wetland filters compared to the outlet section (Fig. 1e and 1f). In the FSW wetlands Skjønhaug (Norway) and Lakeus and Hovi (Finland), the fluxes were higher from the vegetated areas than from the nonvegetated areas during the summer season (Sections 2 and 3 in the Skjønhaug, and Section 2 in the Lakeus and Hovi wetlands had growing vegetation) (Fig. 1a).


View this table:
[in this window]
[in a new window]
  		Table 4. Fluxes of N2O–N, CH4–C, and CO2–C (mean ± SE, number of samples in parenthesis) from the various constructed wetlands. HSSF, horizontal subsurface flow wetland; VSSF, vertical subsurface flow wetland; FSW, free surface water wetland; OGF, overland and groundwater flow wetland.

 

Figure 1
View larger version (41K):
[in this window]
[in a new window]
  		Fig. 1. Fluxes of N2O–N (mg m–2 d–1) (mean and standard error of the mean [SE]) from various sections within each examined constructed wetland. (a) and (b) Free surface water (FSW) wetlands; (c) and (d) overland and groundwater flow (OGF) wetlands; (e) and (f) horizontal subsurface flow (HSSF) wetlands; (g) and (h) vertical subsurface flow (VSSF) wetlands for the summer and winter seasons, respectively.

 
The fluxes of N2O–N were significantly higher during the summer season than during the winter season (P = 0.0115, Student's t test). The fluxes of N2O–N from the different types of wetlands were significantly different both during the summer and the winter (P < 0.0001, Oneway ANOVA) (Table 4). In CWs with municipal wastewater there was no significant difference between fluxes from wetlands with peat and wetlands with gravel, sand, or clay as soil type (P = 0.5567 and P = 0.8932 for the summer and winter seasons, respectively, Student's t test). In CWs with sand, gravel, or clay as soil type, the emission from wetlands treating agricultural runoff had significantly lower fluxes of N2O–N than CWs treating wastewater during the summer season (P = 0.001, Student's t test).

The N2O–N flux given as percentage of the total N load ranged from 0.0037% in the Lakeus FSW wetland to 1.6% in the Hovi wetland. The N2O–N flux given as percentage of the N retention ranged from 0.17% in the Lakeus and Kõo HSSF wetlands to 16% in the Ski VSSF wetland (Table 5).


View this table:
[in this window]
[in a new window]
  		Table 5. Yearly average flux of N2O–N and CH4–C (mg m–2 d–1) as a percentage of the yearly average loading of total N and total C (mg m–2 d–1). Yearly average flux of N2O–N and CH4–C (mg m–2 d–1) as a percentage of yearly average amount of total N and total C removed (mg m–2 d–1). All data given as mean ± SE. VSSF, vertical subsurface flow wetland; HSSF, horizontal subsurface flow wetland.

 
Fluxes of CH4–C
The large temporal and spatial variation in CH4–C fluxes covered a range of –32 to 38 000 mg m–2 d–1. The average flux of all the data from the summer season was 290 mg m–2 d–1, whereas in the winter season it was 63 mg m–2 d–1. During the summer season the average flux for each of the four wetland types was in the range of 110 to 300 mg m–2 d–1, with VSSF wetlands having the lowest mean and HSSF wetlands having the highest mean. During the winter, the average flux ranged from 6.0 to 190 mg m–2 d–1, with the lowest mean for the HSSF wetlands, and the highest mean for the FSW wetlands (Table 4). Within most of the HSSF and OGF wetlands, the fluxes both during the summer season and the winter season were higher in the inlet section compared to the outlet section (Fig. 2). No specific pattern was observed with regard to vegetation in the FSW wetlands (Fig. 2a).


Figure 2
View larger version (44K):
[in this window]
[in a new window]
  		Fig. 2. Fluxes of CH4–C (mg m–2 d–1) (mean and standard error of the mean [SE]) from various sections within each examined constructed wetland. (a) and (b) Free surface water (FSW) wetlands; (c) and (d) overland and groundwater flow (OGF) wetlands; (e) and (f) horizontal subsurface flow (HSSF) wetlands; (g) and (h) vertical subsurface flow (VSSF) wetlands for the summer and winter seasons, respectively.

 
The fluxes of CH4–C were significantly higher during the summer season than during the winter season (P < 0.0001, Student's t test). During summer, there was no significant difference in methane fluxes from the various types of wetlands (P = 0.766). During winter, the fluxes of methane were significantly higher in FSW wetlands than in the other types of wetlands (Table 4). There was no significant difference between fluxes from wetlands with peat and wetlands with gravel, sand, or clay as soil type (loaded with municipal wastewater), neither for summer nor the winter (P = 0.5717 and P = 0.3115, respectively, Student's t test). There was no significant difference between fluxes from wetlands treating agricultural runoff compared to wetlands treating municipal wastewater (summer season) (P = 0.4202, Student's t test).

The CH4–C flux given as percentage of the total C load ranged from 0.39% in the Ski VSSF to 26% in the Kompsasuo wetland. The CH4–C flux given as percentage of the C retention ranged from –250% in Hovi to 390% in Kompsasuo wetland (Table 5). The negative value for the Hovi wetland is due to a slight increase in the concentration of total organic carbon (TOC) from the inlet to the outlet (Table 2), while the very high value for Kompsasuo is due to a higher yearly average flux of methane than the yearly average retention of TOC.

Fluxes of CO2–C
The range of CO2–C fluxes was –840 to 93 000 mg m–2 d–1. The flux was higher during the summer season with an average flux of 4500 mg m–2 d–1 than during the winter season with an average flux of 1200 mg m–2 d–1. During the summer the average flux value for the various wetland types varied between 1900 and 8700 mg m–2 d–1 with the highest mean value in the VSSF wetlands and the lowest mean value in the FSW wetlands. During the winter the average flux ranged from 690 to 1600 mg m–2 d–1, with the lowest mean value for the HSSF wetlands, and the highest mean for the VSSF wetlands (Table 4). The CO2 fluxes from the HSSF wetlands were clearly higher in the inlet section compared to the outlet section (Fig. 3e and 3f). No such pattern was seen for the other wetlands. In the FSW wetlands the CO2–C fluxes were higher in the vegetated areas than in the nonvegetated areas during the summer season (Fig. 3a).


Figure 3
View larger version (47K):
[in this window]
[in a new window]
  		Fig. 3. Fluxes of CO2–C (mg m–2 d–1) (mean and standard error of the mean [SE]) from various sections within each examined constructed wetland. (a) and (b) Free surface water (FSW) wetlands; (c) and (d) overland and groundwater flow (OGF) wetlands; (e) and (f) horizontal subsurface flow (HSSF) wetlands; (g) and (h) vertical subsurface flow (VSSF) wetlands for the summer and winter seasons, respectively.

 
The flux of CO2–C was significantly higher during the summer season than during the winter season (P < 0.0001, Student's t test). The flux of CO2–C was significantly higher in the VSSF wetlands than in the other types of wetlands during the summer season (P < 0.0001, Oneway ANOVA, Tukey-Kramer test). During the wintertime there was also a significant difference between the various wetlands (P < 0.0001, Oneway ANOVA and Tukey-Kramer test) (Table 4). The flux of CO2–C was not significantly different in wetlands with peat and wetlands with gravel, sand, or clay as soil type (P = 0.2844 and P = 0.2807 for the summer and winter seasons, respectively). The flux of CO2–C was significantly higher in wetlands treating municipal wastewater than in wetlands treating agricultural runoff (P < 0.0001).

Ebullition Data from Finland
Ebullition of gases was only measured in the Hovi and Lakeus FSW wetlands in Finland (Table 6). It is seen that for methane, the fluxes are in the same range as the fluxes measured with the static chamber method (Table 4). The gas ebullition of CO2 and N2O are, on the other hand, much lower than the fluxes measured with the static chamber method (Table 4 and 6). The concentrations of CH4 and CO2 (%) in the ebullition traps were 39, 60, and 55 as well as 0.89, 0.72, and 1.49 in the Hovi vegetated area, Hovi deep pond, and Lakeus pond, respectively. The N2O concentrations in the traps in the same areas were 660, 1375, and 717 ng L–1, respectively.


View this table:
[in this window]
[in a new window]
  		Table 6. Gas ebullition data from the constructed wetlands Hovi and Lakeus in Finland.

 
Global Warming Potential
For both seasons, fluxes of N2O and CH4 from the entire data set as well as from the various wetland systems have been converted into CO2 equivalents (mg m–2 d–1) using the calculation factors given by the IPCC (2001b) (Table 7). The contributions from N2O and CH4 to the GWP were compared, both for the entire data set as well as for the various wetland systems (Table 7). Regarding all data, the contribution from N2O to the GWP during winter was significantly higher than the contribution from CH4 (t test, P = 0.002). During the summer the contribution from CH4 to the GWP was higher than the contribution from N2O (t test, P < 0.0001) (Table 7). The results for the various wetland systems varied (see Table 7).


View this table:
[in this window]
[in a new window]
  		Table 7. CO2 equivalents (mean ± SE) from emissions of CH4, N2O, as well as the sum of CH4 and N2O. The conversion of the fluxes into CO2 equivalents is given with 296 for N2O and 23 for CH4 (IPCC, 2001b). OGF, overland and groundwater flow; HSSF, horizontal subsurface flow; VSSF, vertical subsurface flow; FSW, free surface water flow.

 
To examine which type of wetland contributes the most to the GWP, the CO2 equivalents (contribution from both N2O and CH4) were compared for the winter and the summer seasons. It was found that during both the summer and winter seasons, there was a significant difference between the types of wetland (P < 0.0001, Oneway ANOVA) (Table 7).

During wintertime we may assume that no photosynthesis is taking place, thus at this time of the year, the measured CO2 flux may give an estimate of the emission of CO2 from the various wetland systems. For HSSF and OGF wetlands, the flux of CO2 was significantly higher than the flux of CO2 equivalents (contribution from both N2O and CH4), whereas for VSSF and FSW wetlands, there was no significant difference between CO2 and CO2 equivalents (Table 7).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIAL AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
Nitrous Oxide
The emission of N2O–N was found to be significantly higher during the summer than during the winter. The low emission rates in wintertime are due to a slowdown of the denitrification and nitrification processes at low temperatures. Biological processes respond positively to higher temperatures within a range where the enzymes are stable and retain full activity; this has also been shown for the denitrification process (Holtan-Hartwig et al., 2002). The product ratio (N2O/N2) has, however, been found to increase with decreasing temperatures (Keeney et al., 1979; Avalakki et al., 1995). Most factors that slow down the overall rate of denitrification appear to cause an accumulation of N2O as the major end product (Firestone and Davidson, 1989). As the product ratio was not measured for the wetlands in this study, it is unknown whether an increase in this ratio occurred during wintertime.

In general the wetlands receiving water with high concentrations of total N (i.e., the wetlands receiving municipal wastewater) are also the wetlands with highest emissions of N2O–N (Table 2 and 4), with Lakeus as the only exception. However, at Lakeus the N removal is low, which explains the low N2O emission. The VSSF wetlands were found to have the highest emission rates of N2O among all the studied CWs both during winter and summer, whereas OGF wetlands had the lowest emission rates in both seasons. Vertical SSF wetlands are aerobic filters constructed to oxidize ammonium to nitrate via the nitrification process before the water is led into the anaerobic HSSF filter. Probably both the nitrification and the denitrification processes are causing the high emission rates from VSSF wetlands. Nitrification has been found to release N2O to a large degree in micro-aerobic conditions (Mosier, 1998; Dundee and Hopkins, 2001), whereas high loading rates of wastewater may give partial anaerobic conditions where nitrate may be reduced to N2O and N2 gases. However, the presence of O2 may have increased the product ratio N2O/N2 during the denitrification process (Mørkved et al., 2005). For HSSF wetlands, the emission rates of N2O were higher in the inlet part of the wetland and decreased toward the outlet. The water coming from the VSSF systems is rich in NO3, thus the first part of the HSSF systems have high NO3 concentrations probably promoting high N2O emission by denitrification. The HSSF systems seem to be more anaerobic systems than FSW and OGF CWs, giving restricted nitrification in the latter part of the filter. Free surface water and OGF systems may, on the other hand, have higher oxygen concentrations leading to nitrification throughout the wetlands. The nitrate produced is then further reduced to N2O in the anaerobic bottom sediments in FSW or in anaerobic layers of OGF wetlands.

The proportion of loaded N emitted as N2O–N in this work (0.0037 to 1.6%) was comparable to another FSW CW (0.02 to 0.53%) (Johansson et al., 2003). In a study by Hanaki et al. (1992) concerning N2O production during denitrification of wastewater, it was found that up to 8% of influent N was converted to N2O, which is substantially higher than the numbers for CWs. When the emission of N2O–N is compared to removed N, it is seen that most of the wetlands treating municipal wastewater emit only a very small amount of the removed N as N2O–N (Table 5). The CW at Ski is the only exception where especially the Ski VSSF wetland emits substantial amounts of the removed N as N2O–N (16%, Table 5). Suggested reasons for this phenomenon are given in the paragraph above. Similar results have been found for nitrification of wastewater in a laboratory scale experiment, i.e., up to 16% of the nitrified N was converted to N2O, and lower contents of dissolved oxygen resulted in higher N2O production (Zheng et al., 1994). Regarding the Ski HSSF wetland it should be noted that the portion of retained N emitted as N2O was low for experimental runs when the N removal was above 10%. It was only during the experimental runs when the removal of N was close to 0% that the portion of retained N emitted as N2O was very high (Kløve et al., 2005). This underlines the fact that when the environmental factors are unfavorable for the denitrification process, an accumulation of N2O as the major end product occurs (Firestone and Davidson, 1989). In the Hovi wetland, 7.6% of the retained N was emitted as N2O; this high number was probably caused by the low retention of nitrate in this wetland (Table 2). The amount of N removed, which is not released as N2O, is most probably released as N2 (e.g., Teiter and Mander, 2005).

Denitrification is also regulated by the availability of organic C. Most denitrifiers are heterotrophs, and gain their electron donors from organic C (Zumft, 1997). In FSW wetlands it was found that ponds with vegetation had a higher emission rate of N2O than ponds with no vegetation. This was presumably due to increased microbial surface provided by plants and perhaps also by the beneficial effect of released organic substrates from the plants. It has also been reported that the rhizosphere in a CW may enhance nitrification by oxygen release (Armstrong et al., 1990). A part of the released N2O from vegetated ponds might thus derive from the nitrification process, as emissions of N2O from the nitrification process may be substantial in micro-aerobic soils as previously noted.

Apart from the studies summarized in this work, three other studies have examined the release of N2O–N from FSW CWs treating wastewater, and the reported figures are in the range –5.3 to 28 mg N2O–N m–2 d–1 (Fey et al., 1999; Gui et al., 2001; Johansson et al., 2003) (Table 8). The fluxes from the FSW wetlands reported in this study are in the same range. Emission rates of N2O–N from conventional sewage treatment plants have been reported to be as high as 1100 mg N2O–N m–2 d–1 (Czepiel et al., 1995) (Table 8). Such high emission rates have also been measured in the Ski VSSF CW. However, due to the low number of measurements for the Ski VSSF CW, these fluxes are a bit uncertain. The range of fluxes reported for a minerotrophic fen and riparian buffer zones (Paludan and Blicher-Mathiesen, 1996; Augustin et al., 1998; Teiter and Mander, 2005) (Table 8) are in the lower range of what has been reported for the CWs in this study. In the Netherlands, however, high fluxes were measured in a forested riparian buffer zone (local spots exceeding rates of 100 mg N2O–N m–2 d–1) (Hefting et al., 2003). European forest ecosystems have yearly averages in the range 12 to 730 mg N2O–N m–2 yr–1, whereas agricultural land in Europe has yearly averages from 210 to 2400 mg N2O–N m–2 yr–1 (Machefert et al., 2002) (Table 8). The yearly averages of the FSW at Skjønhaug, the HSSF at Kodijärve, the VSSF at Kõo, the FSW at Lakeus, the OGF at Ruka, and in the CW in Sweden (Johansson et al., 2003) are 1000, 1600, 3700, 1100, 1100, and 730 mg N2O–N m–2 yr–1, respectively (assuming a year consists of six months of summer and winter, respectively). Thus, CWs treating wastewater have in general higher emission rates than European forested ecosystems, but seem to have rates comparable to most European agricultural ecosystems.


View this table:
[in this window]
[in a new window]
  		Table 8. Literature data of N2O–N emissions from wetlands, sewage treatment plants, and forested and agricultural ecosystems.

 
Methane and Carbon Dioxide
Temperature, substrate supply, and degree of oxidation are the principal controls on methane fluxes from all soils and wetlands. The presence of vascular plants may also be important in various kinds of wetlands. Supply of organic matter is essential for methane production, even though only a small portion of the organic substrate pool is directly utilized by the methanogens, mainly acetate and CO2 (Oremland, 1988).

In this study the emission of methane was found to be significantly higher during the summer than during the winter. This is as expected as the activity of methanogens has commonly been found to fluctuate in response to temperature (Schütz et al., 1989). During the summer, there was no significant difference in the methane emission from the various types of wetlands, suggesting that the average degree of oxidation is similar for the various wetlands. The methane production during summer does not, however, seem to be related to the amount of organic C in the incoming wastewater, as the TOC in the inlet vary a lot between the various wetlands (Table 2). One reason for the observed lack between fluxes and incoming concentrations of C could be that the fluxes also reflect the quality of the sediment and soil that has accumulated on the site before our studies. During winter the emission of methane was significantly higher from FSW wetlands than from the other types of wetlands. This might be due to an increased decay of C from plants or ice covers preventing oxygen diffusion.

Regarding spatial variation of methane production within the wetlands, the methane emissions from the inlet section were reported to be higher than the emissions from the outlet section during both summer and winter (most of the HSSF and OGF wetlands, Fig. 2c, 2d, 2e, 2f). A similar pattern was also seen in the work of Tanner et al. (1997). This is most likely due to higher process rates at higher C concentrations in the inlet. The reason that this pattern was not seen within the FSW wetlands is probably that the C load to these systems was low and the potential for oxygen diffusion might be better in these systems compared to subsurface flow systems. Other studies have found higher emissions of methane from vegetated sites compared to sites without vegetation (e.g., Whiting and Chanton, 1993), as plants may act as conduits for the transport of methane. Such a pattern was not seen for the FSW wetlands in this study. This might be explained by high levels of ammonium (Table 2) combined with root-zone release of oxygen, thus stimulating the nitrification (Gersberg et al., 1986). Denitrifiers will then compete with the methanogens for the available organic C.

The fluxes of methane as percentage of C removed was very low for the Ski VSSF (0.63%), reflecting the aerobic conditions within the filter. The values for the other CWs in Norway and Finland ranged from 13% (Skjønhaug) to 59% (Ruka), and up to 390% for Kompsasuo (Table 5). These numbers suggest that a rather high amount of the removed C was emitted as methane. However, as previously noted, the fluxes of methane also reflect the quality of the sediment at the site. This is especially seen for the Kompsasuo wetland where the average flux of methane was higher than the average retention of C (Table 5). A large part of the methane emitted from the Kompsasuo CW is thus probably formed in the peat that has accumulated during the last thousands of years after the last deglaciation. The C load to the various CWs that was not released as methane may have been emitted as CO2, accumulated in the wetland sediment and plants or immobilized in the microorganisms.

For the Hovi wetland, a negative average CO2 flux was found during the summer season (Fig. 3a). One explanation could be the high pH of the site (pH up to 9) and the "chemical enhancement" of CO2 uptake (equilibrium favors carbonates and bicarbonates rather than free dissolved CO2 in alkaline waters).

The average methane emissions reported for the various wetlands in this study range from 29 to 980 mg CH4–C m–2 d–1during the summer (Table 4). These values are comparable to the average value reported by Johansson et al. (2004) for a FSW wetland and the range of median values reported by Tanner et al. (1997) for a HSSF wetland (Table 9). There has also been a substantial amount of work devoted to the measurement of fluxes of methane from northern natural wetlands. Common wetlands in the subarctic and boreal zones are peat-producing mires such as bogs and fens. The measurements from northern wetlands (altitude of 45 to 70° N) cover fluxes varying by three orders of magnitude from less than 1 to approximately 1500 mg CH4–C m–2 d–1 (Table 9). Most of the maximum fluxes reported for the CWs in this study are in the same range. The only exception is the Kodijärve wetland in Estonia with a flux of 38 000 mg m–2 d–1, which is about 25 times higher than the maximum flux measured in natural wetlands (Table 9). This high flux is most likely related to the high biological oxygen demand (BOD) in the inlet water (Table 2).


View this table:
[in this window]
[in a new window]
  		Table 9. Literature data of CH4–C emissions from northern wetlands.

 
Greenhouse Gas Emissions from Constructed Wetlands
Considering both CH4 and N2O, FSW and VSSF wetlands had significantly higher contributions to the GWP than HSSF and OGF wetlands during wintertime. During the summer, when the emissions of both gases are significantly higher, VSSF and OGF wetlands contributed significantly more to the GWP than HSSF and FSW wetlands (Table 7). Thus, VSSF wetlands have significantly higher fluxes of N2O and CH4 in CO2 equivalents than the other types of wetlands. These wetlands do, however, require less surface area for the same water purification efficiency, and thus the total emission of greenhouse gases from these systems is probably of less interest compared to the other CW systems. For the other systems, there does not seem to be any difference regarding the contribution to the GWP.

When judging the effect of CWs on the global climate, it is of importance to compare emissions from CWs to the ones from conventional wastewater treatment plants. Work on gas emissions from conventional wastewater treatment plants referred to in this study suggest that at least emissions of N2O–N may be substantially higher than the emissions from CWs (Hanaki et al., 1992; Zheng et al., 1994; Czepiel et al., 1995). Nitrogen and C in nontreated sewage released directly into streams and rivers will also be degraded elsewhere and thus still contribute to the global climate through emissions of N2O and CH4 from rivers and lakes. Even if there is an increase in the emissions of greenhouse gases from CWs compared to natural wetland systems, the emissions seem to be lower than for conventional wastewater systems, and taken into account the generally small area of North European CWs, it is suggested that the studied systems only represent a minor source for atmospheric N2O and CH4.


   ACKNOWLEDGMENTS
 
This study was financed by the project ‘Process based integrated management of constructed and riverine wetlands for optimal control of wastewater at catchment scale, PRIMROSE’ (EVK1-CT-2000-00065) of the EU 5th framework programme.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIAL AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free