Transitions between aquatic and terrestrial environments can be recognized as biogeochemically active ecotones that support high CH4 release. We studied the links between littoral CH4 fluxes and aquatic vegetation, hydrologic conditions, and sediment quality, and integrated the CH4 fluxes into a whole-lake assessment. Methane fluxes were measured using a closed chamber method in the littoral and pelagic zones of three Finnish mid-boreal lakes from May to October. The cumulative CH4 fluxes were spatially integrated over the lake relative to the vegetation coverage in the littoral, and to depth zones in the pelagic regions. During the ice-free period, 66–77% of the CH4 was released from the littoral zone, and the mean CH4 effluxes from these lakes were 0.08–0.42 mol m−2 ice-free season−1. Littoral and pelagic productivity was reflected in CH4 release from the lakes. Our results show that estimates of lake CH4 release should include an assessment of the vegetated littoral zone.
 Methanogenesis is one of the major pathways of organic matter decomposition in anoxic freshwater lake sediments. Methane (CH4), by-product of catabolism, is both consumed by microbial oxidation and emitted to the atmosphere [e.g., Rudd and Hamilton, 1978; Fallon et al., 1980; Capone and Kiene, 1988; Liikanen et al., 2002]. Because CH4 effectively absorbs infrared radiation in the atmosphere, being so-called greenhouse gas, sources and sinks of it have attracted much attention [Houghton et al., 2001]. Small lakes and ponds in the north have been shown to be important at the regional scale CH4 release [Bartlett et al., 1992; Bubier et al., 1993; Hamilton et al., 1994; Reeburgh et al., 1998]. According to Whiticar , global CH4 emissions from freshwater bodies are 5 Tg yr−1, representing 1% of the total CH4 emissions. However, there is still uncertainty in the estimates, because much less attention is paid to CH4 release from boreal lakes than to that from boreal peatlands. In order to understand better the role of lakes in regional carbon exchange, especially, the carbon dynamics of littoral zones of northern lakes should be evaluated. Yet littoral zone can be small in area; such transitions between aquatic and terrestrial environments, like pond margins and scars in permafrost wetlands, often support high CH4 release [e.g., Bartlett et al., 1992; Bubier et al., 1993, 1995]. Account of littoral zone for the whole lake CH4 release is seldom quantified, but it can be up to 90% [Smith and Lewis, 1992]. Littoral metabolism is greatly fuelled by terrestrial nutrient inputs, which are retained by the lake shore vegetation. Thus carbon binding and turnover is accelerated, and presumably some of the detritus is released to the pelagic zone [Wetzel, 1992].
 The arctic, subarctic, boreal, and north temperate zones contain the greatest number of lakes [Wetzel, 2001]. For example, they cover 10% and 7.6% of the total land area in northern countries such as Finland and Canada [Raatikainen and Kuusisto, 1990; Environment Canada, 1998]. Most of these northern lakes are of glacial origin, small and shallow, where the proportion of the littoral zone is high [Wetzel, 2001]. This becomes apparent when the total lake shore length and surface areas in Finland and in the Laurentian Great Lakes are compared. The shoreline length of Finnish lakes is over 7 times that of the Great Lakes, whereas the area of the Great Lakes is seven-fold that of the lakes in Finland [Karlsson, 1986; Raatikainen and Kuusisto, 1990] (see also Natural Resources Canada, 2003, available at http://atlas.gc.ca/site/english/facts/shoreline.html). The area of vegetated littoral zone varies considerably according to the depth profile, sediment and water characteristics, and shore exposure.
 In this study we evaluate the CH4 release from the lake littoral zone on a seasonal basis. We use unpublished data and data from our previous studies on the CH4 fluxes in boreal lakes [Juutinen et al., 2001, 2003; Huttunen et al., 2003a]. Links between seasonal littoral CH4 release and aboveground biomass, water level, and sediment characteristics were evaluated. Methane release was spatially integrated on the basis of the vegetation cover, and the relative importance of the littoral zone for the whole lake CH4 release was quantified. Extensive spatial and temporal sampling of CH4 fluxes was conducted in three mid-boreal lakes.
2. Materials and Methods
2.1. Study Sites
 The study was conducted in three mid-boreal lakes in eastern Finland (Table 1). Methane fluxes were measured at a total of 12 littoral sites, traversing the whole littoral zone or only a number of subzones (Table 2). In this study, the littoral zone denotes the area between the highest shoreline and the outer limit of floating-leaved vegetation. It can consist of several subzones having distinctive vegetation and moisture conditions. The terminology used follows that of Wetzel : upper eulittoral (UE) denotes temporally flooded mainly grass-dominated zone, lower eulittoral (LE) denotes mainly sedge-dominated temporally flooded zone, upper infralittoral (UI) denotes continuously inundated and usually reed dominated, and lower infralittoral (LI) denotes stands of floating-leaved vegetation. The zone of submersed vegetation is not considered here. The measurement plots were representative of the above littoral subzones. Vascular species composition and percentage cover were determined for each plot. In the pelagic (open water area), three replicate CH4 flux measurements were made at a time. The pelagic sites represented areas of different depth. In Lake Kevätön, there were three pelagic sites [Huttunen et al., 2003a] and in Lake Mekrijärvi, there were four sites. In Lake Heposelkä, one open-water site was located in front of the littoral sites.
Littoral Subzone, Study Year, and Dominant Species
Different sites are numbered and lakes are indicated by letters: K, Lake Kevätön; H, Lake Heposelkä; M, Lake Mekrijärvi. Sub-zones are: UE, upper eulittoral; LE, lower eulittoral; UI, upper infralittoral; LI, lower infralittoral. Here n indicates the number of measurement plots, and LOI indicates the loss in weight on ignition (%). Where not determined, the approximate soil quality is given. C:N indicates carbon (%) to nitrogen (%) ratio in the sediment. Nd denotes not determined.
2.2. Gas Flux Measurements and Temporal Integration
 The methane flux was measured using dark, closed static chambers (volumes from 72 to 216 L). Aluminium collars (area 0.36 m2), with water grooves to ensure airtight closure of the chambers, were installed on the study sites before the start of the study. The chambers had battery-driven fans, and tubing passed through a rubber septum for sampling. A 40-mL gas sample of the chamber air was collected four times over 20 min with polypropylene syringes. The CH4 concentrations were determined within 24 hours after sampling using a gas chromatograph equipped with a FID [Juutinen et al., 2001, 2003]. The CH4 flux was calculated from linear regression of the CH4 concentration in the chamber headspace versus time. The fluxes were rejected if the coefficient of determination was below 0.90. Measurement plots (collars) were located beside boardwalks, built in order to minimize any disturbance to the sediment and vegetation. The collars were suspended so that their levels could be adjusted according to the fluctuation in water level. Floating chambers were used in the pelagic sites (measurement period 30–60 min). Ebullition (bubble fluxes) was determined by means of submerged funnel gas collectors in Lake Kevätön and Lake Mekrijärvi. For Lake Kevätön, CH4 ebullition was observed and was assumed to be included in the flux estimate based on the chamber measurements [Huttunen et al., 2003a]. In Lake Mekrijärvi the ebullition was negligible.
 Methane fluxes were measured over the ice-free season during the period 1995 to 2000 (Table 2). In most cases the measurements lasted from early May to late October, with a minimum from June to October. The frequency of the measurements varied from site to site: semiweekly to biweekly in the littoral sites, and monthly in the pelagic sites. The cumulative CH4 flux estimate over the ice-free period (May–October) was generated from the series of observed values. Missing values between the observed values were generated by means of linear interpolation, after which the daily values were summed. Owing to the daytime maxima in CH4 fluxes observed in Phragmites australis (Cav.) Trin. ex Steudel [Alm et al., 1996], the use of CH4 fluxes measured only during the working day would lead to an overestimate for the whole day. Therefore the cumulative ice-free period CH4 flux estimate for the Phragmites plots was corrected with a coefficient derived from the series (4–6 weeks) of measurements conducted on an hourly basis. This was done using an automated system [Silvola et al., 1992] in two separate Phragmites stands (six plots). No corrections were made for sedge-type vegetation because it proved to have only little diurnal pattern.
2.3. Measurement of Environmental Variables
 Continuous series of water level and temperature data were collected at the study sites. Aboveground vascular biomass was harvested during its seasonal maximum in five randomly selected plots (0.25 m2) per littoral sub zone along transects 1, 4, and 5 (only UI + LI) to 10 (Table 2). The harvested biomass was oven dried at 105°C for 24 hours. In site 2, the aboveground biomass was estimated for each collar using the number and length of culms on the plot and the shoot length-weight ratio determined by Kansanen et al. . Sediment organic matter content was determined from dry sediment as loss in weight on ignition (LOI%). Sediment dry weight and LOI% were determined according to SFS standard 3008 [Finnish Standards Association SFS, 1990]. The carbon and nitrogen content of the dry sediment organic matter was determined with an elemental analyzer [CHN-S/O 1106 m-E, Carlo Erba Strumentazione, Italy]. The 0–20 sediment layer was analyzed. Lake water chemistry data were supplied by the North Karelia and North Savo Regional Environment Centres (Table 1).
 Relationships between cumulative CH4 flux and median water level, median depth of aerated layer (equal to water level, if water level below ground; equal to 0, if water level at or above ground), LOI, C:N ratio, and median sediment temperature, aboveground vascular biomass and species growth form were studied using multiple regression approach. Effect of growth form (grasses, sedges, reeds, floating-leaved) was tested by creating dummy variables [Zar, 1999]. Calculations and regression analyses were made using SPSS for Windows statistical package (release 10.1.0, SPSS Inc.).
2.4. Spatial Integration
 Methane emissions determined for specific vegetation communities (subzones of the study sites) were integrated over the whole lake assuming that externally alike sites have similar CH4 fluxes as our study sites. Mean values of cumulative CH4 fluxes calculated for specific stand types were used for the extrapolation. For the pelagic, the mean value weighted by the area of the depth zones was used. For Lake Kevätön the value was obtained from the work of Huttunen et al. [2003a]. Bathymetric data were provided by the North Karelia and North Savo Regional Environment Centres. The vegetation map for Lake Mekrijärvi was drawn on the basis of aerial photos supported by a field survey to the lake. The vegetation areas were digitized with coordinate data (topographic map, scale 1:20,000) using MapInfo Professional version 6.0 (MapInfo Corporation), and the surface area of the different types of cover was calculated by the spherical method. Vegetation covers for Lake Kevätön were calculated from the vegetation map of Nybom . For Lake Heposelkä the area of Phragmites stands and other littoral wetlands were calculated from the topographic map (scale 1:20,000) with a precision of 0.25 ha. The coverage of Phragmites on the map and on an aerial photograph covering an area of 5 km2 were compared and found to be in good agreement.
3.1. Spatial Variability in Littoral CH4 Fluxes
 Methane fluxes in the studied littoral subsites varied from a low influx of atmospheric CH4 to a large efflux to the atmosphere. The largest cumulative efflux (about 19 mol m−2) was estimated for the plot at the landward margin of a Phragmites stand (Figure 1). The second largest efflux (2.8 mol m−2) was found within a few meters distance in the same stand. Generally, the highest CH4 release occurred in the Phragmites and Carex dominated stands of the upper infralittoral and lower eulittoral zones, while those from the open water, dry graminid dominated stands in the upper eulittoral zone, and the floating-leaved stands, were the smallest (Figures 1 and 2a). The mean CH4 flux from the pelagic of Lake Kevätön was, however, clearly higher than the fluxes from the other pelagic sites. Within- and among-site variation in CH4 release was high, but mostly among-site variation exceeded within-site variation (Table 3). Within-site variation was the largest in the Lake Heposelkä Phragmites stand where one plot showed very high CH4 efflux. Variation was large also in the ephemerally flooded zones, greatly due to the moisture gradient [see Juutinen et al., 2001, 2003]. There was also considerable differences in amount of vegetation within a community (Table 3), and similar differences can assumed to be found between the measurement plots, which can contribute to the within-site variability in CH4 release.
Table 3. Variation in CH4 Fluxes and Amount of Biomass Between and Within Subsites Expressed as Coefficient of Variation (CV%)
Between Sites, CV%
Within Sites, CV%
 Differences in the mean cumulative CH4 release between the littoral vegetation communities were related to the differences in their mean aboveground biomass and in the median depth of aerated sediment layer (Figures 2 and 3). Amount of biomass explained much of the variation in the gas fluxes, coefficient of determination being 0.56. The driest communities, i.e., graminoids of the upper eulittoral (median water level 20 cm below the soil surface) but also floating-leaved beds of the lower infralittoral, had a smaller CH4 release than sedge vegetation with a similar amount of biomass. Thus, after inclusion of depth of aerated sediment layer into the model, coefficient of determination increased to 0.68 (y = −0.01 + 0.006 × biomass + 0.02 × aerated layer, F = 30, p < 0.001). Abundance of certain species group, however, improved the model only slightly showing no significance. Sediment temperature (Figure 2a), the sediment organic content (LOI%), or sediment C:N ratio were not significant variables determining the spatial variability of CH4 release.
 In Lake Kevätön site (1) and in two Lake Mekrijärvi sites (7 and 8) measurements were conducted during 2 years with contrasting hydrology (exceptionally high and close to normal “low” water levels), and significant seasonal and interannual variability in CH4 release was related to these changes in water level. Sedge communities had lower CH4 release under exceptionally long inundation than in the following year under lower water level, whereas the opposite pattern was observed in upper and drier grass communities. In the sedge communities, smaller CH4 release during the year with high water level could be explained by long submergence and diminished plant mediated transport of CH4, but partly also by concurrent reduction in amount of biomass (Figure 2b). In the dry communities, the amount of biomass was also higher in the drier year, but CH4 release, in contrast, was higher in the year with high water levels supporting anoxia. The greatest CH4 flux of this study, measured in a P. australis stand, cannot be explained by the amount of biomass on the plot. However, this plot was located in the inner margin of a very productive stand, collecting litter near the shore.
3.2. Integration of CH4 Fluxes on the Lake-Wide Scale
 The vegetated littoral zone covered 7.8, 14.2, and 24.5% of the total surface area of Lakes Heposelkä, Mekrijärvi, and Kevätön, respectively. The larger the lake, the smaller was the proportion of the vegetated area. This was also the order of lakes in respect to phosphorus, nitrogen, and chlorophyll a concentrations in the water (see Table 1). Phragmites australis stands covered the largest areas and were the strongest littoral CH4 sources in Lake Kevätön and Lake Heposelkä (Figure 4). In Lake Mekrijärvi, in turn, the temporally flooded Carex and other sedge species dominated sites, and the floating-leaved stands, had the largest coverage and were major CH4 sources.
 The area weighted mean of cumulative CH4 release was greatest in the most eutrophied lake, second largest in the large mesotrophic lake, and smallest in the polyhumic mesotrophic lake. The mean release for the littoral zones alone was 1.13, 0.93, and 0.44 mol m−2, and that for the whole lake (littoral zone included) was 0.42, 0.09, and 0.08 mol m−2 in Lake Kevätön, Lake Heposelkä, and Lake Mekrijärvi, respectively. There was a slight trend toward increasing CH4 release with increasing mean littoral biomass and chlorophyll a concentration in the water, although the humic Lake Mekrijärvi with a comparatively high chlorophyll a concentration had a low CH4 release (Figure 5). The contribution of the vegetated littoral to the total CH4 release during the ice-free season was 66% in Lake Kevätön, 77% in Lake Heposelkä, and 76% in Lake Mekrijärvi. Relative extent of the littoral CH4 release was lowest in the most eutrophic lake, which had clearly the highest proportion of the vegetated littoral zone, but also the highest pelagic CH4 fluxes.
4.1. Littoral Vegetation and CH4 Fluxes
 This study emphasized the role of a vegetated littoral zone in CH4 release from lakes, and supports the use of vegetation for predicting CH4 release from the lake shore vegetation communities (Figures 3 and 5). Vegetation is recognized to be a key factor determining CH4 release in wetlands by numerous studies, attributed to primary production, which supplies organic matter to the sediment, thus promoting methanogenesis, and to the plant-mediated transportation of CH4. A positive relationship between the amount of biomass or net CO2 assimilation and CH4 release has been shown in various wetlands and rice paddies [e.g., Whiting et al., 1991; Morrissey and Livingston, 1992; Whiting and Chanton, 1993; Huang et al., 1997; Alm et al., 1997; Bellisario et al., 1999; Kankaala et al., 2003]. This study shows a reasonable relationship between the easily measurable aboveground biomass and the release of CH4 on a seasonal basis. A positive correlation was also found between the seasonal CH4 flux and the net ecosystem CO2 exchange, which was determined for some of the study sites [see Larmola et al., 2003]. Net ecosystem CO2 exchange was proposed to be a better predictor of CH4 release than the amount of biomass by Whiting and Chanton .
 Methane flux values relative to the amount of biomass in this study compare well with those obtained in temperate and boreal areas over the same measurement period of 6 months (Figure 3) [Saarnio et al., 1997; Saarnio, 1999; Hyvönen et al., 1998; Kim et al., 1998; Arkebauer et al., 2001; Brix et al., 2001; Kankaala et al., 2003]. However, the relationship presented for various mire sites in discontinuous permafrost zone [see Bellisario et al., 1999] (Figure 3) deviate much from our results. This might be explained by omission of biomass of other species than sedge or features associated with peat component and permafrost degradation. Moreover, the outer margin of Phragmites/Typha latifolia L. stand in a Finnish lake (Figure 3) [see Kankaala et al., 2003] had much higher CH4 release relative to the amount of shoot biomass than most stands of our study. Still, other parts of the same stand agree well with our data. Within-stand differences in the distinguished outer margin were explained by an intense alga and Lemna trisulca L. growth and litter accumulation [Kankaala et al., 2003]. Equally, accumulation of litter near the shore bank probably caused the greatest CH4 flux of this study, about 19 mol m−2 ice-free season−1, in the inner margin of Phragmites stand. This may be the case also with the other outliers, one in the upper margin of flood and one in another two Phragmites plots (Figure 2b). Thus, in addition to the standing biomass, organic matter can be relocated to the site by floods and ice thrust that form litter banks and floating mats; this is especially great within the Phragmites stands. Littoral production may also be displaced to the pelagic for subsequent decomposition there.
 Littoral zone can contain belts that are only ephemerally flooded and subsequent aeration of the soil there both diminishes CH4 production and enhances the CH4 oxidation, causing a lower net release of CH4 [e.g., Moore and Dalva, 1993; Nykänen et al., 1998]. By definition, biomass-CH4 relationship did not hold within those communities with low water tables, but when depth of aerated layer was included, the model was markedly improved in prediction of CH4 fluxes. Within the floating-leaved stand, in turn, one possible explanation for the lower CH4 release compared to that among Carex is the relatively low number of efflux sites in the former stand because only a small part of each plant is connected to the atmosphere. Tidally flooded stands have also shown small CH4 release [van der Nat and Middelburg, 2000] (see Figure 3). In many cases the highest productivity occurs in the shoreline region, where conditions are ideal also for the formation and release of CH4 (Figure 2a).
 Linkages between vegetation and CH4 fluxes suggest that comparatively recently formed and labile organic compounds are the preferential source of CH4-C. This is proved by isotope analyses showing that most of the CH4 originates from organic matter produced within a time span of a few years in many cases [e.g., Chanton et al., 1995; Bellisario et al., 1999; Chasar et al., 2000; Huttunen et al., 2002b]. Therefore, not finding a correlation between sediment organic content and the CH4 flux is logical. Bottom quality (peat or mineral material) also did not affect release of CH4 from hydroelectric reservoirs in the study of Huttunen et al. [2002b]. Sometimes, however, CH4 with an age ranging from hundreds to thousands of years has been detected in some small bog lakes [Martens et al., 1992; Zimov et al., 1997]. In Lake Mekrijärvi, the non-vegetated peat bottom located beneath an eroding peat bank had a CH4 efflux up to 0.26 mmol m−2h−1 when it was exposed dried and shrunk after a lowering of the water level (unpublished data).
4.2. Spatial Integration
 The methane fluxes measured in the study sites were extrapolated over the whole lake areas according to depth zones in the pelagic and to the vegetation cover in the littoral zone. Lake Mekrijärvi was studied in the most detail because it had 13 sites representing the most common vegetation types and depth zones. In Lake Kevätön, our littoral site was well representative of the different vegetation zones in the lake, but the Phragmites stand studied was of low density and had low productivity compared to some other stands in the same lake. Thus our estimate for Lake Kevätön is probably on the low side. In Lake Heposelkä the two separate Phragmites sites had a representative density gradient and therefore should give a reliable estimate of CH4 release in the Phragmites stands on the lake. However, the plot with an extremely high CH4 emission was omitted from the extrapolation. Pelagic measurements were insufficient in Lake Heposelkä, but the low pelagic CH4 effluxes recorded there are reliable on the basis of the low fluxes and negligible ebullition also measured in the pelagic of Lake Mekrijärvi. Similar low CH4 effluxes have also been found in the mesotrophic reservoir Porttipahta, and in two north boreal ponds in Finland [Huttunen et al., 2002a, 2002b].
 Overall, the spatial error should be small, because we were able to use relatively detailed data of the vegetation cover. In this respect, the data was weakest for Lake Heposelkä. Littoral estimate are generally assumed to show diffusive flux, but episodic CH4 release also in the form of ebullition can contribute substantially to total CH4 release in some places [e.g., Christensen et al., 2003; Kankaala et al., 2003]. In the site (H3) the amount of observations rejected due to bubbling was large, 40%, and possibly the cumulative fluxes were underestimated, but in Lake Mekrijärvi, data rejections were rare. Discrimination between natural and disturbance related ebullition is difficult. Episodics associated to the water level changes [e.g., Windsor et al., 1992] should be rather well included in the estimates of this study due to comparatively frequent measurements, made at its best semiweekly [Juutinen et al., 2001]. The estimates given here represent a situation considered as an average year, but substantial changes (10–40%) occur in annual littoral CH4 release following changes in hydrology [Juutinen et al., 2003].
4.3. Littoral Zone and Lake-Wide CH4 Release
 The vegetated littoral zone was the major source (66–77%) of CH4 during the ice-free period. This agrees with the values presented by Smith and Lewis . In their study, Nuphar lutea beds accounted for 51 and 99% of the release in a small Coloradan lake, the percentage depending on the amount of vegetation cover. High contribution of littoral zone was also pointed out in small temperate lakes [Striegl and Michmerhuizen, 1998]. A sum of attributes works in the littoral zone: warm inundated sediments, high productivity, and effective venting to the atmosphere [see also Bartlett et al., 1992; Bubier et al., 1993, 1995]. In our study, the relative importance of the littoral zone was emphasized with decreasing trophic status and correspondingly lower pelagic emissions, although the mean littoral CH4 release was still the highest in the most eutrophied lake. Arvola et al.  reported for a set of Finnish lakes that the littoral CH4 release rate correlated with lake trophic status (phosphorus concentration) and the presence of Phragmites australis. Our study is generally in line with this as lakes with Phragmites dominated and productive vegetation (Lake Kevätön and Lake Heposelkä) had a higher mean CH4 release than Lake Mekrijärvi, where the ephemerally flooded littoral zone and floating-leaved species dominated the vegetation (Figure 4). In those lakes where light extinction is reduced due to humic content, or water level fluctuation is high, for example, in Lake Mekrijärvi, the macrophyte abundance and productivity in deeper parts of the littoral might be restricted and subsequently CH4 release remain lower despite comparatively high nutrient content.
 The order of lakes in respect to sole littoral or whole lake CH4 release rates was the same, which suggests that both are controlled by lake trophic status. However, the whole lake estimates differed more than littoral estimates. Phytoplankton and macrophytic vegetation show generally a similar trend in respect to the nutrient content in water body, but in some cases they deviate; the study of both groups is recommended for environmental monitoring, and for assessing the trophic status [Ilmavirta and Toivonen, 1986; Wetzel, 2001]. This could be the case also when assessing CH4 emissions. Moreover, although pelagic primary production and CH4 production or release are related [e.g. Casper, 1992; Huttunen et al., 2002b], there are many other factors that must be considered in the case of the pelagic. For example, owing to the topography of the bottom, there may be sites of intense sedimentation, and correspondingly of intense CH4 formation, resulting in bubble fluxes. This was true in Lake Kevätön [Huttunen et al., 2003a]. Methane released in bubbles is only little exposed to methanotrophic oxidation, which can enhance the net flux. The concentration of O2, oxidation and net release of CH4 are also dependent on several factors affecting mixing of the water column [Rudd and Hamilton, 1978; Fallon et al., 1980; Riera et al., 1999]. Large and shallow Lake Mekrijärvi is well mixed and thus oxygenated.
 In a study on a large set of lakes, high pelagic release of CH4 was found in lakes with a small size and intense littoral vegetation [Michmerhuizen et al., 1996; Striegl and Michmerhuizen, 1998]. Thus littoral vegetation reflects conditions in the pelagic, or there is C input from the littoral zone [Wetzel, 1992], or both. Our data along with the results of Striegl and Michmerhuizen  suggest that whole lake adjusted CH4 release correlates with standing biomass of the littoral zone, although observations are too few for a qualified proof (Figure 5). Moreover, whole lake estimates of this study and pelagic CH4 release by Riera et al.  and Huttunen et al. [2002a] show increasing trend with increasing pelagic productivity (chlorophyll a concentration in the water) (Figure 5). Amount of chlorophyll a relative to the productivity can be high in humic Lake Mekrijärvi, for example, because of great abundance of alga Gonyostomum semen Ehrenberg. The vegetated littoral zone, as in this study, was not included in the earlier estimates (if there was any vegetation present). As an experiment, we used the biomass-CH4 ratio established in Figure 5 for two Finnish lakes: for the mesohumic oligotrophic Lake Suomunjärvi (biomass data from Toivonen and Lappalainen ) and for the oligomesotrophic Lake Pääjärvi (biomass data from Kansanen et al. ). Rough approximations resulted in ice-free season CH4 release <0.05 mol m−2 and about 1 mol m−2, respectively. The latter is 2.4 times that estimated for eutrophic Lake Kevätön, which, however, we consider to be too low. An Equisetum stand in Lake Pääjärvi has showed high CH4 release [Hyvönen et al., 1998] (see Figure 3). Again, this oversimplified approach may be applied to limited sets of lakes, where the total amount of littoral vegetation is also assessed.
 Methane release estimates, including the littoral zone, presented for these three lakes over the ice-free period of 6 months fit into the range of CH4 emissions reported for lakes, reservoirs, and ponds in the north (Table 4). The lakes we studied are relatively large in the inspected set of lakes, and this may be the reason why their CH4 emissions are moderate even though we included the littoral zone in our estimates. Small, shallow lakes have, in many cases, proved to have higher CH4 emissions than larger lakes with respect to summertime or ice-out CH4 release [Bartlett et al., 1992; Michmerhuizen et al., 1996]. For a comparison, mires edging onto and close to Lake Mekrijärvi had a mean CH4 release of from 1.1 to 2.6 mol m−2 snow free season−1 [Saarnio et al., 1997; Nykänen et al., 1998], i.e., 3 to 6 times higher than our whole lake estimates.
Table 4. Examples of Seasonal CH4 Release From Some Northern Lakes, Ponds, and Reservoirs, Presented in Ascending Order
 Small features in landscape, ponds, and small lakes have a substantial influence on regional CH4 release in some northern areas. Ponds of the Hudson Bay lowlands, covering 8–12% of the area, accounted for 30% of the total CH4 release [Hamilton et al., 1994]. The small lakes in the Yukon-Kuskokwim Delta and Kuparuk River basin in the Alaskan tundra accounted for about 20% of the regional CH4 budgets, and the contribution of littoral vegetation was 2.4% in the first-mentioned area [Bartlett et al., 1992; Reeburgh et al., 1998]. Beaver bonds in the low boreal Canadian wetlands and Clay Belt region in Ontario accounted for 32% and 17% of the regional CH4 release [Roulet et al., 1992; Bubier et al., 1993]. Lakes of this study were larger than those above mentioned, and the effect of productive edges might thus be smaller. Extending the seasonal mean of three lakes in our study over Finland's lake surface results in total lake-associated CH4 release of approximately 105 Gg during a typical ice-free period. As an estimate for peatlands reaches 850 Gg of CH4 annually [Kuusisto et al., 1996], lakes, which represent 10% of the country's surface area, account for about 11% of the total release from freshwater wetlands and lakes. Our estimate is conservative as it covers only the six ice-free months and the potential winter fluxes of CH4 would increase the estimate. Especially in the pelagic, winter can contribute substantially in cases where CH4 accumulates in anoxic waters under the ice cover, and is released to the atmosphere after ice thaw. Ice-out fluxes of CH4 are reported to range from 13 to 48% of the annual CH4 budget [Michmerhuizen et al., 1996; Zimov et al., 1997; Phelps et al., 1998; Huttunen et al., 2003b]. The littoral zone also releases CH4 during the winter at rates similar to those in peatlands, about 1 mmol m−2d−1 (T. Larmola, personal communication, 2003).
 The study lakes represented size classes of 1–100 km2, which account for 39% of the Finnish lake area [Raatikainen and Kuusisto, 1990]. The large lakes (>100 km2), 44% of the lake area, as well as oligotrophic lakes, were absent from our data. However, oligotrophic lakes commonly have extensive Phragmites belts in their sheltered parts, and therefore these probably do not differ very much from the mesotrophic lakes of this study in respect of CH4 release. The final figure for the overall importance of boreal lakes and their littoral zones in the regional CH4 budgets cannot not be obtained without further data on the overall extension of the vegetated littoral zones and on CH4 release in larger lakes. The advances in the use of optical measurements in surveying biomass and productivity [McMichael et al., 1999; Boelman et al., 2003] would probably be helpful. The narrowness, and small-scale spatial variability in the hydrology and vegetation of some vegetated lake margins, however, might pose a challenge for a remote sensing.
 Importance of small water bodies in regional scale CH4 release has been shown in many cases. This study produced information about larger boreal lakes, with emphasis on the littoral zone. We found that the contribution of vegetated littoral areas, although accounting for less than 25% of the total lake area, can be as high as 60–80% of the whole lake CH4 release over the ice-free season. Thus, in lakes with vegetated fringe, seasonal estimates based on pelagic CH4 release alone should be multiplied by 2.5–5 in order to obtain a valid estimate for the whole lake area. In line with previous studies, the CH4 fluxes were dependent on the composition and productivity of the aquatic vegetation in a lake. Therefore nutrients in runoff from the catchment apparently accelerate plant productivity in the littoral zone, thereby boosting the CH4 fluxes. In our study lakes, lush littoral vegetation occurred near the outlets of drainage ditches and brooks.
 We are grateful to many fellow workers at the Mekrijärvi Research Station, the former Siikasalmi Research Station, and the University of Kuopio, who assisted us in the field and laboratory. We also thank the landowners, the Itkonen, Jyrkönen, Lintu, and Tiihonen families, the StoraEnso Company, and the municipality of Ilomantsi. The North Karelia and North Savo Regional Environment Centres kindly provided the data about the lake characteristics. This work was financed by the Academy of Finland, Graduate School in Forest Sciences, and the Maj and Tor Nessling Foundation.