Nitrous oxide flux to the atmosphere from the littoral zone of a boreal lake



[1] The surface-atmospheric exchange of nitrous oxide (N2O) was investigated in the vegetated littoral zone of a eutrophied midboreal lake (Lake Kevätön, Finland) with a static chamber technique. During a dry summer (three to six samplings per site), the meadow site and two marsh sites in the temporarily flooded eulittoral zone and the Phragmites australis-dominated site in the continuously flooded infralittoral zone had mean daytime N2O-N emissions from 11 ± 7 to 22 ± 7 μg m−2 h−1, whereas the Nuphar lutea-dominated site in the infralittoral zone had a mean N2O flux close to zero. During a wet summer (13–14 samplings per site), the mean daytime N2O-N fluxes ranged from 4 ± 1 to 15 ± 5 μg m−2 h−1 at the three eulittoral sites and were negligible at the two infralittoral sites. The littoral zone occupied 26% of the lake area but was estimated to account for most of the N2O emissions from the lake. The studied eulittoral zone, which did not have adjacent nitrogen fertilization, exhibited higher N2O emissions during the summer than seen in northern natural ecosystems in general, including peatlands, forests, and the pelagic regions of lakes. Thus in lake-rich landscapes the littoral zone and other lake-associated wetlands must be considered as potential sources of atmospheric N2O. An assessment of their atmospheric importance requires further data on the N2O fluxes and their regulation in different littoral areas and on the total littoral coverage, neither of which is yet available.

1. Introduction

[2] Agriculture, fossil fuel combustion, and other human activities have disturbed the global nitrogen (N) cycle by increasing the availability and mobility of N in ecosystems [Vitousek et al., 1997]. As a result, certain aerobic and anaerobic microbial processes in the biosphere, mainly nitrification and denitrification [Davidson and Schimel, 1995], produce more nitrous oxide (N2O), which is then emitted into the atmosphere. N2O is an important atmospheric greenhouse gas [Khalil, 1999] and it may contribute to the loss of stratospheric ozone, especially as the emissions of chlorofluorocarbons (CFCs) are reduced.

[3] About one third of the global terrestrial and aquatic N2O emission is considered to be anthropogenic [Seitzinger et al., 2000]. In aquatic ecosystems, N loading can lead to eutrophication and can affect the exchange of N2O between water and the atmosphere. Significant N2O emissions have been measured from N-enriched rivers, as well as estuarine and coastal waters [Seitzinger and Kroeze, 1998; Seitzinger et al., 2000]. Seitzinger et al. [2000] estimated that the annual N2O release from rivers, estuaries, and continental shelves totals 1.9 Tg N. According to Seitzinger et al., over 90% of the estuarine and riverine N2O emission (1.2 Tg N yr−1) may be anthropogenic, and 90% of this originates from the Northern midlatitudes, because of heavy N fertilization and high atmospheric N deposition in these regions. Before it enters the streams and rivers, N leached from terrestrial ecosystems comes into contact with the riparian (streamside) ecosystems [Lowrance et al., 1997], where part of the N load is processed to N2O and released to the atmosphere. A recent review by Groffman et al. [2000] showed that the riverine ecosystems are probably “regional hot spots” in N2O production but their global N2O release is unknown. The N-enriched rivers have been included in the recent global estimates of the aquatic N2O emissions, whereas the N2O emissions from inland freshwater lakes are still excluded [Seitzinger and Kroeze, 1998; Seitzinger et al., 2000]. The neglect of lakes and their littoral zones in the ecosystem N2O exchange studies may raise serious uncertainties over estimates of the regional N2O emissions, especially in northern, lake-rich landscapes. For example, in Finland lakes occupy about 10% of the country's surface area [Raatikainen and Kuusisto, 1990] and in Canada the corresponding figure is 7.6% [Environment Canada, 1998]. The pelagic regions of freshwater lakes and reservoirs are considered only to be minor sources of N2O, although their N2O fluxes have shown extensive variability [Mengis et al., 1997; Huttunen et al., 2000, 2001, 2003]. Instead, similar to the streamside ecosystems [Groffman et al., 2000] and wetlands receiving a high N load [Merbach et al., 2001; Silvan et al., 2002], the lake littoral zones with accelerated N cycling represent potential sites for substantial N2O release. In this study, the seasonal dynamics of N2O fluxes were investigated in the littoral zone of a eutrophied boreal lake, and the fluxes were compared to those previously presented for other natural and managed boreal and temperate ecosystems.

2. Materials and Methods

2.1. Site Description

[4] The study area was located on a relatively exposed vegetated shore of Lake Kevätön (63°06′N; 27°37′E), a shallow and highly eutrophic freshwater lake [see Huttunen et al., 2001] in the middle boreal zone in Finland. Lake Kevätön has received nutrients from intensive agriculture around its catchment, and additionally in the sewage load from a hospital during the period from the 1930s to 1975. The average concentrations of chlorophyll a were 20 and 31 μg l−1, total N was 850 and 930 μg l−1 and total P was 45 and 51 μg l−1 at a depth of 1 m in the water column in the pelagic zone of the lake during the open water periods of 1997 and 1998, respectively (J. T. Huttunen, unpublished data, 1997–1998). According to Forsberg and Ryding [1980], the concentrations of 7–40 μg l−1 for chlorophyll a, 600–1500 μg l−1 for total N, and 25–100 μg l−1 for total P during the open water period indicate a eutrophic state of lakes. Lake Kevätön is ice-covered from mid-November to early May. It has an area of 407 ha, including the infralittoral (continuously flooded littoral) vegetation, and a catchment area of 2400 ha. The lake volume is 0.00932 km3 with mean and maximum depths of 2.3 and 10 m, respectively. Lake Kevätön has an extensive littoral zone, and the study area formed a 60-m wide transition along its moisture and vegetation gradient. The five study sites were located within four different vegetation zones: (1) a meadow (meadow site) and (2) a sedge marsh (marsh-1 and marsh-2 sites, i.e., upper and lower marsh) in the eulittoral zone (temporarily flooded littoral), and the stands of (3) emergent common reed (reed site) in the upper infralittoral zone and of (4) floating-leaved yellow water lily (water lily site) in the middle infralittoral zone. In the entire lake, the area of flood meadows (meadow and marsh sites) is 39 ha, and the area of emergent aquatic vegetation (reed site) is 41 ha. The floating-leaved vegetation (water lily site) covers 26 ha of the total lake area. The dominating vegetation and the carbon and nitrogen contents of soil/sediment of the sites are presented in Table 1.

Table 1. Dominating Vegetation and the Carbon (C) and Nitrogen (N) Contents and C/N-Ratios in the Soil/Sediment in the Littoral Zone of Lake Kevätön
Vegetation ZoneDominating SpeciesSoil/Sediment C and Na
Depth, cmC, %N, %C/N
  • a

    On 15 September 1999, one soil/sediment sample (diameter 7.5 × 7.5 cm) per site was taken for the measurements of the soil/sediment C and N contents. The C and N contents were analyzed from the dry soil/sediment with a model 1106 Carlo Erba elemental analyzer (Carlo Erba, Milano, Italy).

  • b

    Not measured.

Eulittoral Zone
MeadowCalamagrostis canescens (F.H. Wigg) Roth, Carex acuta L.0–242221
MarshCarex aquatilis Wahlenb., Calla palustris L., Potentilla palustris (L.) Scop.0–225118
Infralittoral Zone
ReedPhragmites australis (Cav.) Trin. ex Steudel0–2120.621
Water lilyNuphar lutea (L) Sibth. and Sm.NMbNMNM

[5] Soil moisture differed at the study sites between the two studied summer periods. Summer 1997 was extremely dry, whereas summer 1998 was extremely wet. The mean air temperatures in May–October were nearly similar in 1997 and 1998. According to the recordings of precipitation and air temperature between May and October in 1997 and 1998 at the Rissala Airport, located 14 km southeast from Lake Kevätön, summertime precipitation was 246 and 530 mm, and the mean air temperature was 11.9° and 11.3°C, respectively [Finnish Meteorological Institute, 1998, 1999]. At the same location, the long-term (1961–1990) precipitation and mean air temperature between May and October were 352 mm and 11.3°C [Finnish Meteorological Institute, 1991].

2.2. N2O Fluxes

[6] A boardwalk system was constructed into the study area in June 1997 to avoid the disturbance of soil/sediment during sampling. Aluminum collars (60 × 60 × 30 cm) were installed permanently in the ground (at a depth of 20–25 cm below the soil surface) at the meadow and marsh sites for daytime N2O flux measurements (between 0900 and 1900 UT) with the static, nonsteady state chamber technique [Nykänen et al., 1995, 1998]. At the infralittoral sites, similar collars were hung by height-adjustable wooden frames keeping the lower parts of the collars about 15 cm below the water surface. Three replicate collars were used at each site. The collars had water grooves to ensure a gas-tight connection to the chambers during the measurements. Additional collars with a height of 30 cm were settled upon the permanently installed collars to raise the chambers at the meadow, marsh, and reed sites in order to prevent any possible damage to the vegetation. From each site, the fluxes of N2O were measured 3–6 times during the snow-free period in 1997 and 13–14 times in 1998 (Table 2). The fluxes of N2O were also measured during the winter, on 12 December 1997 and on 17 March and 15 April 1998, using four to six chambers at the meadow site and two chambers at the reed site. The chamber methods used in trace gas exchange studies are described and discussed by Livingston and Hutchinson [1995].

Table 2. N2O-N Flux Measurements in the Littoral Zone of Lake Kevätöna
DateMeadowMarsh-1Marsh-2ReedWater LilyTotal Littoral
  • a

    Values are the number of individual successful flux measurements (see the text).

1997 Season
3 July 1997332311
21 July 199711226
30 July 19972223312
25 Aug. 19973332314
16 Sept. 19971333313
27 Oct. 19973339
Total 199713151513965
1998 Season
9 June 1997333312
24 June 1998325
30 June 199822
7 July 19983339
8 July 1998325
13 July 19983332314
20 July 19983233314
30 July 19983339
3 Aug. 1998336
10 Aug. 19983333214
17 Aug. 1998325
19 Aug. 199833
24 Aug. 1998333312
1 Sept. 19982237
7 Sept. 19982233112
14 Sept. 1998222319
28 Sept. 19982123311
6 Oct. 199811
12 Oct. 19983232313
26 Oct. 19983332314
Total 19983633363536176

[7] During the N2O flux measurements, aluminum chambers (60 × 60 × 20 cm) were inserted into the water-filled grooves of the collars. In winter, snow was removed from the ground and the chambers were placed onto the ground and padded with snow [Alm et al., 1999]. Four gas samples were taken at 5–8-min intervals from the headspace of the chambers through PVC tubes into 50-ml polypropylene syringes (Terumo Europe, Leuven, Belgium) during the 25–30-min measuring period in summer and during the 60-min measuring period in winter. Air was mixed inside the chambers by electrical, brushless fans. The shorter measuring period in summer than in winter was chosen to avoid warming of soils/sediments during the measurements. An average increase of 2°C was detected in the chamber air temperatures during the summertime measuring periods, the importance of which for the precision of the flux determination was suggested to be minor compared with that which could be induced if the temperature-dependent microbial activities in the soil were disturbed by a longer measuring period [see Livingston and Hutchinson, 1995].

[8] Gas samples were analyzed for their N2O concentration with a Hewlett-Packard Series II (Hewlett-Packard, Palo Alto, CA) or a Shimadzu GC-14B (Shimadzu Corp., Kyoto, Japan) gas chromatograph, equipped with electron capture detectors (ECD), within 24 hours of sampling. The Hewlett-Packard system has been described by Nykänen et al. [1995, 1998]. In the Shimadzu gas chromatograph, samples were first passed through a glass tube filled with P2O5 to remove water before their loading into a 2-ml loop in a 10-port Shimadzu VLA-1 valve. A column (1.8 m × 1/8 inch) was packed with HayeSep Q 80/100 mesh (Hayes Separations, Bandera, TX). Nitrogen (32 ml min−1) was used as the carrier gas. The sensitivity of ECD was enhanced by doping it with Ar/CH4 (5% Ar, 95% CH4) at a flow of 2 ml min−1. Oxygen was omitted from ECD by bypassing it via an automatic valve. The temperature of ECD was 325°C and the temperature of the oven was 40°C. The retention time was 2.2 min. Peak areas were integrated with a Shimadzu GLASS-CR 10 program. The Hewlett-Packard and Shimadzu systems had a precision of 0.7 and 0.4% for N2O, respectively. The accuracy of the analyses was maintained by calibrating the gas chromatographs against a standard gas mixture after every 12 samples, which kept the coefficient of variation of the replicated concentration determinations below 1%. The N2O fluxes were calculated from the linear change in the N2O concentrations in the chamber headspace during the measuring periods. The least squares regression lines “headspace N2O concentration versus time” were first visually inspected for abrupt changes in the direction of the flux, resulting from disturbances such as the leakage of the chamber or disturbances of soils/sediments during sampling. For the rest of the data, the linear concentration change (three to four samples) over time was assumed if the r2-value of the regression line was above 0.60. This low limit for r2 was selected to avoid discrimination against low N2O fluxes. Altogether, 26 and 16% of the fluxes were rejected in 1997 and 1998, respectively. Examples of the regression lines from two flux determinations are shown in Figure 1. The detection limit of the chamber measurement for N2O-N was 2–4 μg m−2 h−1.

Figure 1.

The least squares regression lines “chamber headspace N2O concentration versus time” from two N2O flux determinations in September 1998.

2.3. Environmental Variables

[9] Air temperature inside and outside the chambers was measured with a Fluke 52 K/J thermometer (Fluke Corp., Everett, WA). Soil temperature was determined at the soil/sediment surface. Soil temperature was not measured during the winter when the soil was frozen.

2.4. Statistical Analyses

[10] Nonparametric Spearman correlation coefficients were calculated to study the relations between the N2O fluxes and site moisture and temperature conditions in summer. The correlation analyses were performed separately using the daily values for each site (N = 13–36) and the seasonal mean values for each chamber (N = 15). An SPSS statistical package (SPSS Inc., release 9.0.1) was used for the statistics.

3. Results

3.1. Temporal and Spatial Patterns of N2O Fluxes and Environmental Variables

[11] The summertime littoral N2O-N fluxes had frequency distributions skewed to high values (Figure 2), reflecting seasonal and spatial "hot spots" for N2O release in the study area (see below). The fluxes in the individual measurements ranged from −26 to 140 μg m−2 h−1 in 1997 (N = 65) and from −19 to 120 μg m−2 h−1 in 1998 (N = 176). In summer 1997, the mean, standard deviation, and median N2O-N fluxes were 15, 32, and 8 μg m−2 h−1, respectively. In summer 1998, the corresponding values were 6, 15, and 2 μg m−2 h−1, respectively.

Figure 2.

Frequency distributions of N2O fluxes measured during the snow-free seasons of (a) 1997 and (b) 1998 in the littoral zone of Lake Kevätön.

[12] Precipitation was lower in July–August in 1997 than in 1998, being also lower than the long-term (1961–1990) average precipitation in these months (Figure 3). This could be seen in lower water table levels in the littoral zone in 1997 (Figure 4 and Table 3). During the dry summer of 1997, the meadow, marsh, and reed sites had their maximum monthly N2O-N emissions (mean ± SE) up to 46 ± 23 μg m−2 h−1 between July and August (Figure 4), but the spatial variation within all of the sites was large during the entire study (Figure 5). At the eulittoral meadow and marsh sites, and at the reed site in the infralittoral zone, the seasonal mean N2O-N emissions (mean ± SE) ranged from 11 ± 7 to 22 ± 5 μg m−2 h−1 in 1997, whereas the N2O flux was negligible at the infralittoral water lily site (Table 3).

Figure 3.

Monthly precipitation and mean air temperature in June–October in 1997 and 1998 at the Rissala Airport, located 14 km southeast from Lake Kevätön [Finnish Meteorological Institute, 1998, 1999], and corresponding long-term averages for the period 1961–1990 [Finnish Meteorological Institute, 1991].

Figure 4.

Monthly average soil/sediment temperatures, water table levels (negative below the soil surface), and N2O fluxes in the littoral zone of Lake Kevätön. The error bars are standard errors of the mean.

Figure 5.

Variation in the monthly average N2O fluxes between 15 chambers in the littoral zone of Lake Kevätön during the snow-free periods of 1997 and 1998. Individual bars at each month correspond to the monthly N2O fluxes from chambers 1 to 15. The chambers with accepted fluxes are presented in parentheses for each month. ND, not determined.

Table 3. Statistics of the N2O-N Fluxes, Water Table Levels, and Soil/Sediment Surface Temperatures in the Snow-Free Periods of 1997 and 1998 in the Littoral Zone of Lake Kevätön
SiteN2O-N Flux, μg m−2 h−1Water Table, cmaSoil/Sediment Temperature, °C
Mean ± SEbMin.Max.NcMean ± SEMin.Max.NMean ± SEMin.Max.N
  • a

    Negative below the soil surface.

  • b

    SE, the standard error of the mean.

  • c

    N, the number of individual measurements.

Meadow21 ± 12−1814013−19 ± 4−53−21414 ± 3−13013
Marsh-119 ± 10−1814015−2 ± 1−811615 ± 203016
Marsh-211 ± 7−239315−2 ± 1−701515 ± 303015
Reed22 ± 5−3591318 ± 30321319 ± 2132710
Water lily−1 ± 5−2620947 ± 62360918 ± 113239
Meadow15 ± 4−111203610 ± 10253714 ± 142237
Marsh-18 ± 2−8543325 ± 211503314 ± 132230
Marsh-24 ± 1−17153622 ± 27453312 ± 152834
Reed1 ± 1−19163564 ± 148763616 ± 152225
Water lily1 ± 2−143736100 ± 1831133914 ± 151818

[13] In summer 1998, the monthly rainfall was more than twice the long-term monthly precipitation in June–August but remained below the long-term average in September (Figure 3). During this summer the water tables were continuously at or above the soil surface at the eulittoral sites (Figure 4 and Table 3). In 1998, the monthly average N2O-N fluxes were mostly below 5 μg N2O-N m−2 h−1, and peak emissions were found only at the meadow and marsh-1 sites (Figure 4), coinciding with the lowest monthly precipitation in September (Figure 3). The seasonal average N2O-N fluxes (mean ± SE) were 15 ± 4, 8 ± 2, and 4 ± 1 μg m−2 h−1 in 1998 at the meadow, marsh-1, and marsh-2 sites, respectively, and were negligible in the infralittoral zone (Table 3). The monthly soil/sediment surface temperatures were higher in July in 1997 than in 1998, but the decreases in the temperature during rest of the summer season were larger in 1997 (Figure 4). The high mean soil/sediment temperatures in the infralittoral sites in 1997 (Table 3) were due to the absence of measurements at these sites in October (Table 2 and Figure 4).

3.2. Relations Between N2O Fluxes and Environmental Variables

[14] The importance of the temporarily varying soil moisture and temperature conditions in the N2O fluxes was studied by correlating the daily fluxes with the corresponding water table levels and soil/sediment temperatures. These correlations were calculated separately for the five littoral study sites. The daily N2O flux correlated significantly (p < 0.05) with the water table level or the soil/sediment surface temperature only at the marsh-1 site. There the daily N2O flux was positively related to soil temperature (Spearman Rank correlation coefficient ρ = 0.713, N = 15) in 1997 and negatively to the water table level (ρ = −0.573, N = 33) in 1998. The effects of these factors on the spatial variation in the N2O fluxes within the littoral zone were assessed using the seasonal average fluxes, water table levels, and soil/sediment temperatures from each chamber locations (i.e., 15 microsites) in the correlation analyses. In 1998, the mean summertime N2O flux correlated negatively with the mean water table level (ρ = −0.713, p = 0.003, N = 15), but in 1997 this relation was not statistically significant (ρ = −0.493, p = 0.062, N = 15). The seasonal mean N2O flux did not correlate significantly with the mean soil/sediment surface temperature (not shown).

3.3. Winter N2O Fluxes

[15] The N2O-N fluxes varied from −7 to 4 μg m−2 h−1 in the individual wintertime chamber measurements. The N2O-N fluxes were negligible in winter, averaging (mean ± SE) 1 ± 0.4 and 0 ± 2 μg m−2 h−1 at the marsh and reed sites, respectively.

4. Discussion

4.1. Littoral N2O Emissions and Their Atmospheric Importance

[16] The present sparse data hint at the potential importance of boreal littoral regions and lake-associated wetlands in the total atmospheric N2O load. In this study, the summer N2O-N fluxes in the driest part of the littoral zone (meadow and marsh sites), and at the reed site, ranged from 11 ± 7 to 22 ± 5 μg m−2 h−1 during the dry summer 1997, and even higher mean summertime N2O-N fluxes, 16-27 μg m−2 h−1, have been reported for temporarily flooded meadow and swamp sites in midboreal eutrophied Lake Postilampi, Finland [Huttunen et al., 2000]. These emission rates match or are higher than the N2O emissions from various wetlands, forest and aquatic ecosystems in the boreal and temperate regions (Table 4). The littoral sites, however, showed lower N2O emissions than those reported for fertilized agricultural soils (Table 4).

Table 4. Mean N2O-N Fluxes From Some Boreal and Temperate Ecosystems
EcosystemsLocationMean Flux, μg N2O N m−2 h−1NaSeasonbReferencesc
Peatlands, naturalFinland−0.8–5.39summer1
Peatlands, drainedFinland−0.1–258summer1
Peatlands, naturalFinland−0.8–2510summer2
Various wetlands, naturalOntario, Canada−0.4–1.315annual3
Restored peatland buffer zoneFinland−2.1–2701summer4
Drained fen grassland, controlGermany9.1–472annual5
Drained fen grassland, fertilizedGermany14–1802annual5
Lakes and Rivers
Freshwater lakes and reservoirsFinland−2.1–8.29summer6
Freshwater lakes, oligo-mesotrophicSwitzerland0.3–244annual7
Freshwater lakes, eutrophic and aeratedSwitzerland9.5–192annual7
Hudson RiverNew York, USA6.41annual8
Platte RiverColorado, USA62d1annual9
Mixed forestSweden0.8–1.41annual10
Mixed forestOntario, Canada0.41summer11
Deciduous forestSaskatchewan, Canada4.4–5.71summer12
Deciduous forestsSaskatchewan, Canada0.03–0.42annual13
Deciduous forestNew York, USA0.8–4.11summer14
Coniferous forestsNortheastern USA−1.1–2.15summer15
Coniferous forestGermany161annual16
Deciduous forestGermany581annual16
Coniferous forestSwitzerland0.8–6.71summer17
Deciduous forestSwitzerland9.0–101summer17
Agricultural Land
Organic agricultural soilsFinland42–4202annual18
Mineral agricultural soilsFinland17–902annual18
Afforested organic soilsFinland21–1203summer19
Cropped soils, fertilizedSaskatchewan, Canada1.9–273annual13
Pasture/hay landSaskatchewan, Canada0.52annual13
Bare agricultural soilOntario, Canada421annual20
Bare agricultural soil, manuredOntario, Canada831annual20
Agricultural soils, cultivatedOntario, Canada16–722annual20

[17] The average winter N2O fluxes were negligible, corresponding to natural peatlands in Finland where the wintertime N2O emissions have been found to be small [Alm et al., 1999]. In drained Finnish peatlands the N2O release during wintertime has contributed up to 28% of the annual emissions [Alm et al., 1999]. However, our measurements did not include the freezing-thawing periods in spring and autumn, when high N2O emissions have been measured from various terrestrial environments [e.g., Wagner-Riddle et al., 1996; Huttunen et al., 2002; Martikainen et al., 2002; Schürmann et al., 2002]. For example, in farmed mineral and organic soils in Finland, the winter N2O release has accounted for an average of 57% of the annual emissions [Martikainen et al., 2002]. In an alfalfa field in Ontario, Canada, the N2O emissions between October and February were approximated to account for 50% of the annual emissions [Wagner-Riddle et al., 1996], whereas in some alpine soils in Switzerland [Schürmann et al., 2002] and in spruce and beech plantations in Germany [Papen and Butterbach-Bahl, 1999] even more N2O was released during the winter than in summer. The possible episodic N2O emissions during freezing and thawing of soils [e.g., Papen and Butterbach-Bahl, 1999] or diel short-term variations in the flux [e.g., Maljanen et al., 2002], neither of which is easily obtained with the manual chamber techniques, must be further assessed in the littoral wetlands with automated chambers [Maljanen et al., 2002]. The micrometeorological techniques, although being nonintrusive compared to the use of the chambers, are not applicable in the narrow littoral zones such as our study area (width 60 m). The fetch in the littoral zone between the land and the open water area would be too small, and on the other hand, information of the spatial flux heterogeneity (see above) cannot be attained by micrometeorological measurements. In fact, in the simultaneous N2O flux measurements by the chamber and micrometeorological techniques, the results have agreed satisfactorily [Smith et al., 1994; Christensen et al., 1996; Laville et al., 1999].

4.2. Regulation of N2O Fluxes

[18] Aerobic nitrification and anaerobic denitrification are the major processes producing N2O in soils/sediments, whereas N2O can be consumed by denitrification in highly anoxic conditions [Davidson and Schimel, 1995]. Our data, however, could not differentiate between the importance of nitrification and denitrification in the littoral N2O production, and thus, a following speculation of possible reasons for the high N2O emissions during summer droughts must be viewed with care. The high N2O emissions during the dry summer 1997 could be a result of enhanced N turnover and nitrification rates in the aerobic soil/sediment surface, and/or of enhanced denitrification due to increased nitrate supply into anaerobic, denitrifying microsites in the soils. In the wet summer 1998, nitrification in the surface soil/sediment layers could be limited by the lack of oxygen, which then would limit the nitrate supply for denitrification. On the other hand, the wetter conditions in 1998 could favor anoxia and reduction of nitrate and N2O to N2, limiting the N2O release. However, the proportion of produced N2O increases both in nitrification and denitrification at low oxygen concentrations [Davidson and Schimel, 1995], thus the summertime lowering of the water table levels increasing the volume of modestly aerated soil could increase the proportion of N2O production in either of these processes. Nonetheless, the N2O production was most probably limited at our littoral study sites by nitrate availability. In the rural areas of the boreal zone, atmospheric N deposition is usually low and the leaching of N forms a major source of N into aquatic ecosystems, especially in agricultural and forested catchments [Rekolainen, 1989; Cooke and Prepas, 1998; Mannio et al., 2000]. At the meadow and swamp sites in Lake Postilampi, where there are higher N2O emissions [Huttunen et al., 2000], nitrogen leaching was probably higher as a result of more intensive fertilization in the surrounding area. Laboratory studies with littoral sediments taken close to the meadow and marsh sites in Lake Kevätön have suggested that the littoral N2O production is limited by low nitrate availability, due to low nitrification activity [Liikanen et al., 2003]. The vegetated sediment cores incubated in this glasshouse experiment with the water table levels at 0 or −15 cm did not show increases in the N2O fluxes following ammonium fertilization (3 g of NH4+-N m−2), but exhibited from ten- to hundredfold increases in the N2O release after nitrate addition (3 g of NO3-N m−2). The mean N2O-N flux from the cores before the extra N was added was 16 μg m−2 h−1, being similar to those we found from the eulittoral zone in this field study. The laboratory experiment, however, could not judge that nitrification was not changed in the littoral zone during summer drought, since the sediment temperature (average 16°C) was lower in the laboratory than in the field during the large emissions in 1997.

[19] In general, the N2O release from wetlands depends on the complex interactions of the N transformation processes [Davidson and Schimel, 1995; Groffman et al., 2000]. For example, in nutrient-rich boreal peatlands the long-term lowering of the water table by ditching has increased the N2O emissions, and this has been associated with the enhanced oxygen availability and the increase in N mineralization in the surface peat, whereas in nutrient-poor peatlands draining did not increase the N2O fluxes [Martikainen et al., 1993; Regina et al., 1996]. The short-term lowering of the water table in laboratory, simulating a summer drought, also has increased N2O release from peat soils [Aerts and Ludwig, 1997; Regina et al., 1998; Dowrick et al., 1999]. In peat cores from a mid-Welsh mire, UK, the increased N2O emissions following a drought were suggested to have been due to the increased N2O:N2 production ratio during denitrification in the saturated, but not totally water-logged conditions [Dowrick et al., 1999]. There could be several possible reasons for the high summertime N2O emissions from the littoral sites in our study, and the further assessment of the littoral N2O exchange should include studies on the nitrogen transformations along the littoral vegetation gradients. As suggested for the assessment of N2O fluxes in natural and constructed riparian ecosystems [Groffman et al., 2000], the N2O:N2 production ratio during denitrification could also be among the key factors to understand the littoral N2O dynamics and their atmospheric importance. The N in leaching and runoff that enters the littoral zone and is not removed from the system to the atmosphere as N2 or N2O, nor is retained into the vegetation, microbes, or soils and sediments, can fuel the N2O production in the N-receiving aquatic ecosystems, including lakes, rivers, and estuarine areas.

4.3. Contribution of the Littoral Zone to Total N2O Emissions From Lakes

[20] In Lake Kevätön, where the vegetated littoral zone covers 26% of the lake area, the littoral zone is the main source of N2O to the atmosphere. The eulittoral meadow and marsh sites released N2O-N from 0.3 ± 0.1 to 0.5 ± 0.3 and from 0.09 ± 0.02 to 0.4 ± 0.2 kg ha−1, respectively, during the summer seasons (150 days assumed). These constitute a substantial summertime N2O-N emissions, ranging from 21 ± 6 to 30 ± 17 kg in the flood meadows and from 6 ± 2 to 28 ± 15 kg in the marsh vegetation, in contrast to the negligible N2O fluxes in the pelagic zone of the lake. Nitrous oxide accumulates in the water column of Lake Kevätön during the winter, leading to N2O-N release of 0.003–0.015 kg ha−1 following the spring ice melt [Huttunen et al., 2001, also unpublished data, 1997–1999]. During two open water periods, minor total N2O-N uptake, averaging 0.006 and 0.025 kg ha−1, can be estimated for the pelagic zone based on the monthly measurements at three sites with floating static chambers [Huttunen et al., 2003]. The pelagic N2O-N emissions from the open water area of the lake (324 ha) can then be approximated to range from 1 to 5 kg at spring ice melt, and from −2 to −8 kg during the open water period.

[21] It is not currently possible to estimate the total N2O emissions from the littoral regions of boreal lakes, because there are no data, which encompass the total littoral wetland area. However, since there is a shoreline of 190,000 km in the Finnish lakes [Kuusisto, 1987], this implies that these littoral N2O emissions could have at least a regional importance in the N2O emissions in the boreal zone. Because the N2O emissions are consistently higher from wetlands receiving a high nitrate load (Table 4) [Merbach et al., 2001; Silvan et al., 2002] than at the presented littoral sites, the increasing eutrophication of northern lakes [Intergovernmental Panel on Climate Change, 1996] possibly will increase their littoral N2O emissions in the future.


[22] Eija Konttinen (†), Kaisa Mäntynen, Irma Nihtilä-Mäkelä, Heikki Päivinen, Raimo Asikainen, Markku Virnes, Kalle Maaranen, Tarja Niskanen, and Henna Harju are thanked for technical assistance. We would like to thank the Tiihonen family for all their support during the field campaigns. The valuable suggestions of the reviewers are greatly appreciated. This work was financially supported by the Maj and Tor Nessling Foundation, the Finnish Cultural Foundation, the Niilo Helander Foundation, and the Academy of Finland.