effect of catchment characteristics on aquatic carbon export from a boreal catchment and its...
TRANSCRIPT
Effect of catchment characteristics on aquatic carbonexport from a boreal catchment and its importance inregional carbon cyclingJ U S S I HUOTAR I * , HANNU NYKÄNEN † , MART IN FORS IUS ‡ and LAURI ARVOLA*
*Lammi Biological Station, University of Helsinki, Paajarventie 320, Lammi FI-16900, Finland, †Department of Biological and
Environmental Science, University of Jyvaskyla, P.O. Box 35, Jyvaskyla FI-40014, Finland, ‡Finnish Environment Institute, P.O.
Box 140, Helsinki FI-00251, Finland
Abstract
Inland waters transport and emit into the atmosphere large amounts of carbon (C), which originates from terrestrial
ecosystems. The effect of land cover and land-use practises on C export from terrestrial ecosystems to inland waters
is not fully understood, especially in heterogeneous landscapes under human influence. We sampled for dissolved C
species in five tributaries with well-determined subcatchments (total size 174.5 km2), as well as in various points of
two of the subcatchments draining to a boreal lake in southern Finland over a full year. Our aim was to find out how
land cover and land-use affect C export from the catchments, as well as CH4 and CO2 concentrations of the streams,
and if the origin of C in stream water can be determined from proxies for quality of dissolved organic matter (DOM).
We further estimated the gas evasion from stream surfaces and the role of aquatic fluxes in regional C cycling. The
export rate of C from the terrestrial system through an aquatic conduit was 19.3 g C m�2(catchment) yr�1, which cor-
responds to 19% of the estimated terrestrial net ecosystem exchange of the catchment. Most of the C load to the recipi-
ent lake consisted of dissolved organic carbon (DOC, 6.1 � 1.0 g C m�2 yr�1); the share of dissolved inorganic
carbon (DIC) was much smaller (1.0 � 0.2 g C m�2 yr�1). CO2 and CH4 emissions from stream and ditch surfaces
were 7.0 � 2.4 g C m�2 yr�1 and 0.1 � 0.04 g C m�2 yr�1, respectively, C emissions being thus equal with C load to
the lake. The proportion of peatland in the catchment and the drainage density of peatland increased DOC in streams,
whereas the proportion of agricultural land in the catchment decreased it. The opposite was true for DIC. Drained
peatlands were an important CH4 source for streams.
Keywords: boreal catchment, CH4, CO2, dissolved inorganic carbon, dissolved organic carbon, flux
Received 12 December 2012 and accepted 5 July 2013
Introduction
Carbon (C) transport from land to sea with en route
processing and degassing into the atmosphere has
recently been emphasized to be of central importance
in estimates for regional and global C cycling (Cole
et al., 2007; Battin et al., 2008; Aufdenkampe et al., 2011;
Butman & Raymond, 2011). The carbon dioxide (CO2)
emissions from inland waters, for example, have shown
to be in the range 1–50% of net ecosystem production
or C accumulation rates of the surrounding catchments
(Kling et al., 1991; Hanson et al., 2004; Jonsson et al.,
2007; Huotari et al., 2011; Ojala et al., 2011). When all
transported C species have been considered, terrestrial
C transport to streams and subsequent emissions to the
atmosphere have been shown to possibly reverse the
terrestrial CO2 sink of a temperate peatland to become
neutral or a source (Billett et al., 2004).
Northern soils store large quantities of C, however,
the susceptibility of C to be released due to changes in
land-use, as well as in hydrology and temperature as
climate changes, is uncertain and poses a potential
threat to further increase the concentrations of atmo-
spheric CO2 and CH4. Butman & Raymond (2011) have
highlighted the importance of the smallest streams in C
cycling due to their contribution to the out-gassing of
CO2. In C cycling, the role of CH4 emissions may be
small but due to its high radiative forcing (Forster et al.,
2007), freshwater CH4 emissions as CO2 equivalents
has been estimated to offset 25% of the terrestrial green-
house gas sink (Bastviken et al., 2011). Some previous C
budget studies in northern areas have incorporated sur-
face waters in larger watersheds and mostly in remote
regions with minimal human disturbance (Christensen
et al., 2007; Jonsson et al., 2007; Buffam et al., 2011; Stri-
egl et al., 2012), while less is known about aquatic
export of terrestrial C from heterogeneous boreal catch-
ments under human disturbance, including agriculture.
Human induced perturbations within the catchments
affect C export into aquatic systems and may also haveCorrespondence: Jussi Huotari, tel. +358 40 576 2675,
fax +358 91 91 40746, e-mail: [email protected]
© 2013 John Wiley & Sons Ltd 3607
Global Change Biology (2013) 19, 3607–3620, doi: 10.1111/gcb.12333
large implications in regional C cycling by changing the
terrestrial net ecosystem exchange (NEE). Contempo-
rary forestry practises with harvesting and site prepara-
tion initially enhances the C export followed by
effective C storing in fast growing stand biomass and
finally declining productivity in the maturing forests
(Ryan et al., 1997; Kolari et al., 2008; Schelker et al.,
2012). Peatland drainages for forestry purposes, most of
which are situated in Nordic countries and Russia
(Minkkinen et al., 2008), trigger the same kind of suc-
cession in C cycling and C export to aquatic systems
(Armentano & Menges, 1986; Minkkinen et al., 2002).
Clearing a forest or peatland for agriculture enhances
the export of C and reduces C accumulation in the soil
(Armentano & Menges, 1986). In the Mississippi River
catchment, for example, increase in agriculture over the
last century has enhanced inorganic C transport from
soils, which, however, may act as a CO2 sink due to
sequestration in carbonate weathering (Raymond et al.,
2008). Changes in land cover affect water retention in
the system, whether it is lost by evapotranspiration or
flushed through in near surface or ground water runoff,
subsequently affecting C escape rates (Mayorga, 2008).
As the quality of dissolved organic matter (DOM)
varies depending on the source, it serves as a tool to
estimate the origin of stream water C (Battin, 1998;
Helms et al., 2008) and may reveal the effect of human
disturbance on the C transport as well. The quality of
DOM can be estimated by absorption spectrum of the
water sample, which varies depending on the molecu-
lar weight and the degree of aromaticity of DOM (e.g.
Hernes et al., 2008). Knowing the source of C would
help to improve the assessments of the role of inland
waters in large scale estimates of C cycling (Cole et al.,
2007; Tranvik et al., 2009).
Our goal was to find out how heterogeneous land-
scapes, which include forests, peatlands (mostly
drained), agriculture and lakes, affect terrestrial C loss
as well as the CO2 and CH4 concentrations of streams
in a boreal catchment. We sampled the main tributaries
of Lake P€a€aj€arvi, southern Finland, for dissolved C spe-
cies as well as in several sites along two of the tributar-
ies with varied land cover/land-use types. Data over a
full year allows us to estimate the total dissolved C
export as well as the CH4 and CO2 evasion from
streams of the whole catchment. Furthermore, we dis-
cuss the importance of aquatic C fluxes in regional C
cycling including lakes in the region, of which CO2 and
CH4 emissions have recently been estimated (Kankaala
et al., 2013). The specific research questions we wanted
to answer were: (i) How does peatland and agricultural
land affect C export from the catchment, i.e. the lateral
transport and gas emissions from inland water sur-
faces? (ii) What is the role of the streams and lakes in
regional C budget within this well determined and rep-
resentative boreal catchment with heterogeneous land-
use?
Material and methods
Data were gathered from five main subcatchments of Lake
P€a€aj€arvi (61°04′N, 25°08′E; Fig. 1; Fig. S1 in colour) that
together make up 174.5 km2 and cover 83% of the whole
catchment area (210.1 km2). Two of the subcatchments are
rather small in area: 12 L€oytynoja (8.2 km2) and 13 Ko-
iransuolenoja (6.5 km2), two are larger 6 Mustajoki (79.3 km2)
and 14 Haarajoki (57.8 km2) and one is intermediate: 7 Luh-
danjoki (22.8 km2). There is a higher percentage of agricultural
land in 13 Koiransuolenoja and 7 Luhdanjoki subcatchments,
while the Mustajoki subcatchment has a higher proportion of
peatland relative to others (Table 1). Practically all peatlands
in the area are drained for forestry with only small areas of
undrained natural peatland remaining.
To better understand the effect of peatland and agricultural
land on stream water chemistry, two of the subcatchments
were sampled additionally at different locations along the
main stream channel: the Mustajoki subcatchment (sample
points 1–6) was sampled at six locations of which one (no. 5)
was a tributary to Mustajoki; the L€oytynoja subcatchment (9–
12), which originates from a groundwater spring, had four
sampling points, one of which is a tributary to L€oytynoja from
a periodically overflowing drained peatland (no. 11) (Fig. 1).
Currently, there are only small scale restocking of forests, no
large scale clearing for agriculture, and peatland drainages
were excavated mainly in 1960–1970s with some drainages
reopened in 1990–2000s. The outlet of Lake P€a€aj€arvi (8
P€a€aj€arvi) was sampled 3 km downstream from the lake. The
study area with the demarcated subcatchments as well as all
the sample points is shown in Fig. 1. Tables 1 and 2 give the
characteristics of the subcatchments. Definitions of subcatch-
ment areas as well as land-cover proportions used are based
on Topographic Database of National Land Survey of Finland
(2010) and analysis with ArcMap 10. The bedrock of the catch-
ment belongs to “the Proterozoic Svecokarelidic mountain
range and consists of mica gneisses, amphibolites, granites,
granodiorites and minor bodies of basic plutonic rocks” (Lai-
takari, 1980). Mean annual temperature and precipitation in
the area measured over the period of 1981–2010 are 4.2 °C and
645 mm, respectively (Pirinen et al., 2012), while our study
year was slightly warmer (5.7 °C) and wetter (694 mm) (May
2011–April 2012).
Weekly sampling began on 4 May 2011 in the five main trib-
utaries and the outlet stream, and ended on 30 May 2012. The
streams were sampled for pH, specific conductance, dissolved
organic carbon (DOC) concentration and absorbance measure-
ments as well as for dissolved CH4 and CO2 concentrations.
At the same time water temperature was recorded in situ. The
sampling for CO2 and CH4 was conducted every second week
in September and October and monthly from December to
March. We started additional sampling in Mustajoki and L€oyt-
ynoja subcatchments on 10 June 2011 and ended on 23 May
2012 with the same intensity except for that monthly sampling
© 2013 John Wiley & Sons Ltd, Global Change Biology, 19, 3607–3620
3608 J . HUOTARI et al.
continued from December until May. The main tributaries
were sampled during 1 day and additional sampling was con-
ducted during the next day. Altogether 57 samples were taken
from the main tributaries except for CH4 and CO2 (N = 38).
Additional points were sampled 26 times (N = 20 for gases)
except for sample points 11, because it dried out during sum-
mer, and 1, 5 and 11 because of ice cover during the winter.
Samples for DIC, CO2 and CH4 were taken into 60-ml poly-
propylene syringes closed with three-way stopcocks. Below
the stream surface triplicate 30 ml samples were drawn into
the syringes avoiding air bubbles. The samples were kept in
crushed ice until the measurements were made in the labora-
tory within 24 h. pH of the samples were lowered with 3.0 ml
of 1% HNO3 to transform all DIC to CO2. Headspace of 30 ml
N2 was then added into syringes followed by a gas chromato-
graphic determination of CO2 and CH4 concentrations of well-
equilibrated headspace at 20 °C as explained by Ojala et al.
(2011). CO2 concentration was calculated from DIC, pH and
temperature as described by Butler (1982).
The samples for pH (Thermo Scientific Orion 3-star pH
Benchtop) and specific conductance (YSI 3200) were mea-
sured in the laboratory. Duplicate DOC samples from fil-
trated water (Millex-HA 0.45 lm) were stored at �22 °C in
plastic vials until determinations. They were measured with
Pt-catalysed high-temperature combustion, using a Shimadzu
TOC-5000 total carbon analyser. Absorbance measurements
were conducted between 200 and 750 nm with 0.5 nm inter-
vals using a 0.01 m quartz cuvette (Shimadzu UV-1800) and
sample absorbance spectra were referenced to a blank
spectrum of MQ-water. For DOM quality proxies, we used
wavelength specific UV-absorbance at 254 nm (SUVA254) cal-
culated as the absorbance divided by DOC concentration,
which correlates with the aromaticity of the DOM and it is
inversely related to its biodegradability (Weishaar et al.,
2003; Wickland et al., 2012). We used the ratio of the absor-
bances at 254 and 365 nm (A254/365) for the proxy of molecu-
lar weight of DOM, which is negatively correlated with
molecular weight and biodegradability of DOM (�Agren et al.,
2008). We did not verify the molecular weight or biodegrad-
ability of DOM with further analyses and thus these
measures should be considered as suggestive.
Discharge (Q) data for 6 Mustajoki, 12 L€oytynoja and 14 Ha-
arajoki as well as for the outlet stream of Lake P€a€aj€arvi (no. 8)
were taken from a database of the Finnish Environment Insti-
tute (OIVA service on environment and geographic informa-
tion, 1 June 2012). Instantaneous values of Q based on our own
measurements of the cross-section area of the stream channel
and water velocity (MiniAir�2; Schiltknecht Messtechnik AG,
Gossau, Switzerland) for 13 Koiransuolenoja were plotted
against the respective daily Q values of Mustajoki, Haarajoki
Fig. 1 Lake P€a€aj€arvi and the studied subcatchments. Numbered circles represent the sample points of which catchment characteristics
are given in Table 1 and stream water characteristics given in Table 2.
© 2013 John Wiley & Sons Ltd, Global Change Biology, 19, 3607–3620
AQUATIC C EXPORT FROM A BOREAL CATCHMENT 3609
and L€oytynoja and the best attained regression was used to
produce a continuous Q time series. The daily Q of Haarajoki
best explained the variation of the measured instantaneous
Q of Koiransuolenoja and the form of the regression was a
second-order polynomial function (r2 = 0.822, N = 47). For
the four subcatchments of the tributaries, we had continuous
discharge data which covered 72.2% of the whole catchment
area of Lake P€a€aj€arvi. For 7 Luhdanjoki we do not have data
on Q. Flow velocity for sample points 1–5 and 9–11 were mea-
sured at every sampling occasion and for other sample points
it was calculated from Q and the measured cross-section area
of the stream channel.
Table 2 Stream water characteristics (mean � SD) measured at the sample points shown in Fig. 1
t, °C pH DIC, g m�3
Specific
conductance,
lS cm�1 (25 °C) Total N, mg m�3 Total P, mg m�3
1 8.4 � 7.0 5.5 � 0.4 2.5 � 1.0 36 � 4.4 ND ND
2 5.6 � 5.7 6.2 � 0.6 2.6 � 1.0 57 � 5.1 ND ND
3 5.9 � 6.1 6.3 � 0.6 2.7 � 1.1 65 � 6.2 ND ND
4 6.2 � 6.3 6.4 � 0.5 3.1 � 1.4 71 � 8.5 ND ND
5 7.1 � 5.8 6.7 � 0.4 4.0 � 1.1 105 � 17 ND ND
6 Mustajoki 7.0 � 6.6 6.6 � 0.3 3.8 � 1.7 91 � 10 1610 � 800 30 � 5
7 Luhdanjoki 6.8 � 5.9 7.1 � 0.3 7.8 � 1.4 156 � 18 3710 � 1930 35 � 15
8 P€a€aj€arvi 8.9 � 8.3 7.1 � 0.1 4.0 � 0.4 94 � 2.8 1400 � 180 13 � 4
9 5.4 � 2.2 6.6 � 0.2 4.3 � 0.3 90 � 6.9 ND ND
10 5.7 � 3.1 6.7 � 0.2 4.5 � 0.4 105 � 7.1 ND ND
11 7.4 � 5.5 3.9 � 0.3 5.0 � 4.1 73 � 29 ND ND
12 L€oytynoja 6.1 � 4.0 6.9 � 0.3 3.8 � 1.6 98 � 8.8 1120 � 400 20 � 8
13 Koiransuolenoja 7.5 � 6.5 7.2 � 0.2 6.1 � 1.2 214 � 16 3240 � 1850 40 � 89
14 Haarajoki 7.6 � 6.6 6.6 � 0.2 5.2 � 1.8 95 � 17 1360 � 890 35 � 12
ND, not determined.
Table 1 Characteristics of the studied subcatchments. Most important mineral soil and peat types of the subcatchments are
according to Quaternary deposits 1 : 20 000 of Geological Survey of Finland (2008)
Forest% Peatland% Agriculture% Lake%
Total
Area,
km2
Discharge,
m3 s�1
Ditch m/
Peatland ha Mineral soil Peat type
1 83.5 13.9 0 2.5 5.1 – 273
2 70.1 28.3 0.8 0.8 20.8 – 262
3 57.6 26.7 2.7 1.1 40.9 – 255
4 67.6 25.4 6.1 1.0 45.5 – 254
5 63.0 15.6 21.1 0.3 21.5 – 276
6 Mustajoki 65.0 21.3 13.1 0.7 79.3 0.5 255 Moraine,
fine sand
Sedge
peat 68%
7 Luhdanjoki 50.4 12.9 35.7 1.1 22.8 – 191 Silt, moraine Sedge
peat 77%
8 P€a€aj€arvi 62.5 13.5 16.2 7.8 231.7 1.9 ND
9 87.4 3.3 9.3 0 1.2 – 66
10 76.8 2.7 20.5 0 2.1 – 131
11 60.6 37.8 1.7 0 2.8 – 254
12 L€oytynoja 70.4 16.3 13.4 0 8.2 0.07 243 Sand, fine
sand, gravel
Sphagnum
peat 66%
13 Koiransuolenoja 67.4 4.7 27.9 0.01 6.5 0.07 222 Fine sand,
moraine
Sedge
peat 100%
14 Haarajoki 69.8 13.1 11.6 5.4 57.8 0.7 240 Moraine,
fine sand,
silt
Sedge
peat 76%
ND, not determined.
© 2013 John Wiley & Sons Ltd, Global Change Biology, 19, 3607–3620
3610 J . HUOTARI et al.
C export was calculated from discharge and concentration
data using the Load Estimator (LOADEST) program (Runkel
et al., 2004). CH4 and CO2 exchange between stream water and
the atmosphere was calculated as a product of gas exchange
coefficient and the gas concentration difference between the
surface water and the equilibrium concentration. The gas
exchange coefficient at 20 °C (k600) was calculated according to
the equation by Raymond et al. (2012), where the empirically
determined k600 was approximated from the slope of the
stream segment (S) and the flow velocity (V, m s�1), i.e. k600(m d�1) = S 9 V 9 2841.6 + 2.03. The gas exchange (Fc) was
then calculated as Fc = k (Caq-Ceq), where Caq and Ceq are gas
concentrations of the surface water and that which the surface
water would have if it was in equilibriumwith the atmosphere,
respectively. Gas transfer coefficient was calculated from k600as k = k600 (Sc/600)�0.5 using gas specific Schmidt numbers
(Sc) taken from J€ahne et al. (1987). To attain area specific
annual efflux rates for streams with daily V values we assumed
linear change in gas concentrations and integrated the attained
daily exchange rates over the year, whereas for streams with
sporadic V data we calculated monthly average efflux rates
and multiplied them with the respective number of open water
days in a month and further summed them up. No gas efflux
was assumed during ice cover. The attained efflux rates were
grouped according to the stream width classes used in the
Topographic Database of National Land Survey of Finland and
based on field determination, i.e. stream width <2 m, 2–5 m
and >5 m. N for efflux estimates of each class were 1, 9 and 2,
respectively. Stream surface areas were estimated by assuming
mean widths of 0.5 m and 3.5 m for classes of <2 m and 2–
5 m, respectively, while the stream areas for class >5 m were
given in the database, and multiplying the mean widths by
lengths of the streams of the respective width classes. Effluxes
were extrapolated over the whole stream surface area of the
catchment assuming attained stream width specific evasion
rates and multiplying by the respective surface area of the
stream width classes. For the estimate of the CH4 and CO2
fluxes from all water surfaces of the study area, we adopted
the areal gas flux estimates from Kankaala et al. (2013), esti-
mated for different sized lakes especially for this area.
Possible relationships between measured values were tested
with correlation analysis for each sampling point individually.
We calculated monthly averages, which were further averaged
to attain a single yearly average for all variables. These num-
bers were analysed against land cover information. In addi-
tional sampling points variables had only a single value
during winter months that represented the respective month.
If some point could not be sampled each month due to drought
or ice cover, we did not use any gap filling to attain the
monthly value, except for gas effluxes which were assumed as
zero, but only months with measured information were used
for yearly averages. All variables were log10-transformed
prior to analysis. We used Spearman’s correlation because all
the variables were not normally distributed (Shapiro–Wilk,
P < 0.05). All analyses were performed with PASW Statistics
20 (IBM Corp., Armonk, NY, USA). Uncertainties of the
estimated loads and fluxes as well as the choice of the gas
exchange model are discussed in Supporting Information.
Results
Time series for load estimates and dissolved C concen-
trations are shown in Figs S2–11.
Stream discharge
Discharge patterns of the streams followed each other
and only the discharge of the regulated outlet stream (8
P€a€aj€arvi) diverged from the other streams (Fig. 2). The
discharge was highest in late April at the time of snow
melt, and the second highest peak was found in late
December. Smaller peaks were recorded in November,
September and June. Concurrent discharge peaks con-
firmed the uniform climate of the study area. Winter
base flow was higher than summer base flow except in
12 L€oytynoja.
CH4 and CO2 concentrations and fluxes
Gas concentrations were always above atmospheric
equilibrium (Fig. 3a and b). CH4 concentrations were
highest in the summer and in September. The highest
concentrations in general were in 8 P€a€aj€arvi and 14
Haarajoki, i.e. 9.3 � 7.6 (mean � SD; range 1.6–27.7)and 7.8 � 4.9 (1.5–19.4) mg C m�3, respectively, but
there was an exceptionally high CH4 concentration of
Fig. 2 Discharge (mm d�1) of the studied tributaries draining
to Lake P€a€aj€arvi and of its outlet stream.
© 2013 John Wiley & Sons Ltd, Global Change Biology, 19, 3607–3620
AQUATIC C EXPORT FROM A BOREAL CATCHMENT 3611
232 mg C m�3 in 6 Mustajoki in early August (Fig. 3a).
CH4 concentrations were low during the winter and
they slightly increased in late winter or early spring
except in 7 Luhdanjoki and in 12 L€oytynoja, of which
concentration remained relatively low all the time, i.e.
1.3 � 0.5 (0.6–2.8) mg C m�3. CH4 concentration corre-
lated negatively with Q in 6 Mustajoki, 8 P€a€aj€arvi and
14 Haarajoki (Table 3).
CO2 concentration was close to equilibrium
(<2 9 equilibrium) only twice in 7 Luhdanjoki in June
and once in 13 Koiransuolenoja in July (Fig. 3b). In 6
Mustajoki and 14 Haarajoki there was a clear two-
peaked pattern with the first peak occurring in late
summer–autumn with the highest concentration in
Haarajoki of 6.3 g C m�3, and a second peak in the
winter. The winter peak was distinct also in the other
streams except in 8 P€a€aj€arvi, which had rather stable
CO2 concentration throughout the study, i.e. 0.8 � 0.2
(0.4–1.3) g C m�3. CO2 correlated with Q in 12 L€oyt-
ynoja and 13 Koiransuolenoja (Table 3).
Within theMustajoki subcatchment (1–6) CH4 concen-
trations were almost constantly highest in its topmost
sample point 1 (Fig. 3a), which peaked in late summer–early autumn. CH4 concentration in general decreased
downstream, but there was a CH4 source to the stream
between the sample points 3 and 4, in an area with peat-
land, agriculture and forest cover of 13%, 36% and 51%,
respectively. As with CH4, the lowest CO2 concentra-
tions were often measured in sample point 3 after which
the concentrations increased, i.e. 4.0 � 1.3 (2.7–8.2) mg
C m�3 as CH4 and 1.2 � 0.7 (0.6–3.4) g C m�3 as CO2
vs. 5.8 � 3.0 (2.0–11.2) mg C m�3 and 1.4 � 0.7 (0.8–3.8) g C m�3 in sample points 3 and 4, respectively.
In the L€oytynoja subcatchment the effect of the
groundwater spring was clear in gas concentrations.
The uppermost sample point (no. 9) always had low
CH4 concentrations, i.e. 0.6 � 0.3 (0.3–1.6) mg C m�3,
as it is situated 560 m downstream from the spring and
with only minor effects from small areas of peatland
and agriculture. The pattern in both gases between the
sample points 9 and 10 was similar and although the
CO2 concentrations were always almost identical, the
area between the sample points (distance ca. 800 m)
was a clear source of CH4 (Fig. 3a). The area (0.9 km2)
has peatland, agriculture and forest cover of 2%, 37%
and 61%, respectively. Sample point 11, discharging the
drained peatland to L€oytynoja showed high concentra-
tions of both CH4 and CO2 in the summer but its influ-
ence on gas concentrations had already disappeared at
the sample point 12. CO2 concentration decreased along
the 3.6 km route from sample point 9 to 12 L€oytynoja,
whereas CH4 increased (Fig. 3a and b), i.e. more CH4 was
produced to the stream water en route than was evaded.
Areal effluxes of CH4 were highest in the smallest
stream width class reflecting high CH4 concentration
from drained peatlands, whereas the highest areal CO2
emissions occurred in the stream width class of 2–5 m
(Table 4). This was due to high gas transfer velocity
(k600) of the width class, i.e. 7.7 � 5.4 m d�1, compared
with that of 1.3 m d�1 of the smallest width class and
2.3 � 0.8 m d�1 of the width class of >5 m. The surface
area of the smallest width class was almost double that
of class 2–5 m, resulting in an order of magnitude
higher CH4 emission and 67% higher CO2 emission. C
gas emissions from streams were 65% higher than that
of lakes and ponds (Table 4).
DOC concentration and DOM quality
Stream discharge and DOC concentrations were posi-
tively correlated (Table 3). The highest concentrations
(a)
(b)
(c)
Fig. 3 CH4 concentration (mg C m�3) (a), CO2 concentration (g
C m�3) (b), and DOC (g C m�3) (c) for each sample points
shown in Fig. 1. Dashed lines represent the atmospheric equilib-
rium concentration of CH4 (a) and CO2 (b) at 20 °C.
© 2013 John Wiley & Sons Ltd, Global Change Biology, 19, 3607–3620
3612 J . HUOTARI et al.
were recorded in November–December while the
peak during the highest Q in April–May was slightly
lower. The lowest DOC concentrations occurred in
August–September and February–March. All of the
main tributaries had a similar pattern in DOC,
whereas outlet stream 8 P€a€aj€arvi had a rather constant
DOC concentration throughout the study (range 8.9–13.5 g m�3). The range in DOC was quite large in all
other sample points and e.g. in 6 Mustajoki and 12
L€oytynoja the concentrations fluctuated between 6.3–34.5 g m�3 and 2.7–26.9 g m�3, respectively (Fig. 3c).
In the Mustajoki subcatchment (1–6), DOC concen-
tration was the most stable, i.e.15.5 � 3.1 (11.9–22.5) g m�3 in sample point 1. DOC concentrations
correlated with Q and the correlations were much
stronger downstream, i.e. concentrations were higher
during high discharge and lower during low discharge
conditions than in the uppermost sample point (no. 1).
However, on average DOC concentration increased
downstream, although in the late winter–spring the
main-stem DOC concentration was diluted by the
waters from tributary (no. 5).
In the L€oytynoja subcatchment (9–12) DOC concen-
tration did not correlate with Q in sample point 9, and
DOC was low (<5 g m�3) all the time in sample points
9 and 10, closest to the spring. The only exception was
the snow melt peak when the DOC concentrations rose
close to 10 g m�3. DOC concentration in sample point
11 varied between 33 and 108 g m�3 and was lowest
during snow melt and highest in September right after
the summer drought. DOC concentration was clearly
higher in sample point 12 than in 9 and 10 except dur-
ing summer drought and in midwinter when sample
point 11 was frozen solid, implying that there was no
stream connection between the peatland area (no. 11)
and the 12 L€oytynoja stream.
Also dissolved organic matter (DOM) quality seemed
to be related to discharge. However, the relationship
varied between the streams, i.e. there was a negative
correlation between Q and aromaticity of DOM in 6
Mustajoki and 13 Koiransuolenoja, while in other
streams the correlation was positive (Table 3). Since
DOC positively correlated with Q in all streams, it indi-
cated that 6 Mustajoki and 13 Koiransuolenoja received
a large amount of less aromatic DOM during high flow
compared with low flow, while in the other streams the
DOM during the high flow was probably more aro-
matic than during low flow. However, there was no
connection between changes in SUVA254 and DOC in
13 Koiransuolenoja, but we found SUVA254 on average
to decrease with increasing DOC consistently in sample
points most influenced by peatlands (>15%), except in
12 L€oytynoja (16.3%), which is groundwater influenced
and where SUVA254 increased together with DOC.
SUVA254 increased together with DOC also in sample
points 9 and 10, as well as in 7 Luhdanjoki. Proxy for
Table 3 Spearman’s correlation coefficients and p-values (2-tailed) for stream log10-transformed discharge (log Q) and log10-trans-
formed DOC, SUVA254, A254/365, CH4 and CO2
6 Mustajoki 8 P€a€aj€arvi 12 L€oytynoja 13 Koiransuolenoja 14 Haarajoki
DOC
r 0.789 0.437 0.813 0.592 0.673
P 0.000 0.001 0.000 0.000 0.000
N 57 57 57 54 54
SUVA254
r �0.663 0.347 0.619 �0.503 0.379
P 0.000 0.009 0.000 0.000 0.005
N 56 56 56 53 53
A254/365
r 0.848 �0.579 0.071 0.847 0.194
P 0.000 0.000 0.605 0.000 0.163
N 56 56 56 53 53
CH4
r �0.791 �0.922 0.003 �0.049 �0.836
P 0.000 0.000 0.984 0.779 0.000
N 38 38 38 35 35
CO2
r 0.001 �0.163 0.582 0.826 �0.175
P 0.994 0.329 0.000 0.000 0.319
N 38 38 38 35 35
Statistically significant correlations are highlighted by italics.
© 2013 John Wiley & Sons Ltd, Global Change Biology, 19, 3607–3620
AQUATIC C EXPORT FROM A BOREAL CATCHMENT 3613
molecular weight of DOM was negatively correlated
with Q in 6 Mustajoki whereas the correlation was posi-
tive in 8 P€a€aj€arvi (Table 3). When monthly averages of
SUVA254 and A254/365 were plotted against each other,
streams formed distinct but slightly overlapping
groups implying the most biodegradable DOM in the
outlet stream of Lake P€a€aj€arvi while DOM from Mustaj-
oki catchment with 21% of peatland coverage appar-
ently had the lowest biodegradability (Fig. 4).
DOM quality was rather stable within the Mustajoki
subcatchment with the aromaticity of DOM slightly
decreasing along the water course (Fig. 5). Instead, in
the L€oytynoja subcatchment DOM quality proxies var-
ied a lot and reflected the influence of the groundwater,
as well as of the drained peatland into stream water
DOM, as aromaticity of DOM was always much lower
in sample points 9 and 10 than in 12 (Fig. 5a). This was
even during the drought and the midwinter when the
connection of the drained peatland through the sample
point 11 to the L€oytynoja stream was ceased based on
the DOC data. The variation of DOM quality was great-
est close to the spring (sample point 9) and lowest in
the peatland site (sample point 11) (Fig. 5). The rela-
tionship between the proxy for molecular weight and
DOC in sample point 11 diverged that from other sites
with moderate peatland coverage possibly implying
differences due to sphagnum peat when other sites
consisted mainly of sedge peat (Table 1).
Land-cover effect
Peatland proportion correlated with DOC concentra-
tion and the quality of DOM of the streams (Table 5).
Moreover, the extent of which the peatlands were
drained within the catchment, measured as ditch length
per peatland area, was a good predictor of the DOM
quality and it was related to DOC concentrations as
well (Table 5). Peatland proportion and drainage den-
Table
4W
ater
surfaceareasan
dsp
ecificeffluxes.Areal
CH
4an
dCO
2fluxes
forlakes
aread
optedfrom
Kan
kaa
laet
al.(20
13)
Streams/
ditch
esLak
es
Totalarea
or
flux(km
2ortC
yr�
1)
<2m
2–5m
>5m
Total
(t)
<0.01
km
2
0.01–0.1
km
2
0.1–1
km
2
1–10
km
2
>10
km
2Total(t)
Stream
length
(m)/N
lakes
9917
0475
143
ND
3321
81
1
Areaha
49.6
26.3
8.7
8.5
72.2
174.5
199.7
1348
.818
.8
CO
2flux
(gC
m�2yr�
1)
1710
1929
849
159
7761
4146
CH
4flux
(gC
m�2yr�
1)
40.1
7.17
4.9
2.96
1.11
0.33
0.25
0.11
TotalCO
2flux(kgC
yr�
1)
8479
0750
7297
7348
514
28.7
1350
155
613
1064
4281
897
6204
5287
7.9
2306
.6
TotalCH
4flux(kgC
yr�
1)
1988
418
85.7
424.0
22.2
251.3
801.7
575.8
499.4
1483
.73.6
25.8
ND,notdetermined
.
Fig. 4 Monthly means of A254/365 vs. SUVA254 [l (mg C)�1 m�1]
for the five main tributaries of Lake P€a€aj€arvi and its outlet
stream.
© 2013 John Wiley & Sons Ltd, Global Change Biology, 19, 3607–3620
3614 J . HUOTARI et al.
sity both correlated with CH4 concentration, although
the relationship with peatland proportion and CH4 was
only close to significant (P = 0.051). Agricultural land
proportion positively correlated with pH and specific
conductance as expected, and it positively correlated
with DIC as well. The correlation of agricultural land
proportion with CH4 was negative, although there was
an indication of a summertime source of CH4 from agri-
cultural land in sample points 4 and 10. However, their
CH4 concentrations correlated differently with Q, i.e.
negatively in 4 and positively in 10. DOC/DIC ratio
was highly dependent on peatland proportion in the
catchment, also the ratio decreased together with a pro-
portion of agricultural land. However, DOC/DIC ratio
was below unity only due to high groundwater
influence in sample points 9 and 10.
Dissolved C load to Lake P€a€aj€arvi and regional CH4 andCO2 effluxes
The load of dissolved carbon, including DOC, DIC and
CH4 to Lake P€a€aj€arvi from the four main subcatchments
with continuous Q records and covering 72.2% of the
catchment of P€a€aj€arvi summed up to 1050 � 168
t C yr�1. Assuming the same load per area from the
whole catchment, the total load estimate is 1455 � 233 t
C yr�1 [7.1 � 1.1 g C m�2(catchment) yr�1]. DOC
dominated the load at 86.5% and DIC made up practi-
(a) (b)
Fig. 5 SUVA254 [l (mg C)�1 m�1] (a), and A254/365 (b) for each sample points shown in Fig. 1.
Table 5 Spearman’s correlation coefficients between log10-transformed land cover proportions and measured log10-transformed
variables
DOC SUVA254 A254/365 CO2 CH4 DIC Cond pH
Peatland%
r 0.873 0.736 �0.754 �0.055 0.530 �0.556 �0.640 �0.670
P 0.000 0.003 0.002 0.852 0.051 0.039 0.014 0.009
N 14 14 14 14 14 14 14 14
Forest%
r �0.367 0.024 �0.029 0.226 �0.327 �0.424 �0.323 �0.244
P 0.197 0.935 0.923 0.436 0.253 0.131 0.260 0.401
N 14 14 14 14 14 14 14 14
Agriculture%
r �0.609 �0.591 0.371 �0.420 �0.530 0.675 0.952 0.925
P 0.021 0.026 0.191 0.135 0.051 0.008 0.000 0.000
N 14 14 14 14 14 14 14 14
Lake%
r �0.082 �0.242 0.422 0.240 0.358 0.000 �0.182 �0.067
P 0.780 0.404 0.133 0.409 0.209 1.000 0.533 0.821
N 14 14 14 14 14 14 14 14
Drainage density
r 0.692 0.918 �0.692 �0.143 0.577 �0.670 �0.560 �0.538
P 0.009 0.000 0.009 0.642 0.039 0.012 0.046 0.058
N 13 13 13 13 13 13 13 13
Statistically significant correlations are highlighted by italics.
© 2013 John Wiley & Sons Ltd, Global Change Biology, 19, 3607–3620
AQUATIC C EXPORT FROM A BOREAL CATCHMENT 3615
cally the entire remainder 13.5%. CH4 load in terms of
C was negligible (0.19 � 0.03 t C yr�1, 0.01%).
Total CO2 and CH4 evasion from streams to the
atmosphere in the catchment was estimated at
1429 � 586 and 22 � 9.8 t C yr�1, respectively
(Table 4). Including both CH4 and CO2 effluxes of
all the water surfaces in the whole study area the ef-
fluxes summed up to 2332 t C yr�1. This is 60%
higher than dissolved C transported to lake P€a€aj€arvi
and together make up to 3787 t C yr�1, i.e. 18.5 g
C m�2 of the catchment, which can be regarded as
terrestrial loss without including sedimentation and
particulate organic carbon (POC) transport.
Discussion
Streams and ditches cover only 4.4% of the water sur-
faces of the study area but their C gas emissions were
65% higher than that of lakes and ponds, which cover
7.8% of the whole study area. The emission vs. lateral
transport ratio of the streams was 1 : 1 without POC
included in our estimate for transport. For comparison,
it is same with the Yukon River basin (Striegl et al.,
2012) whereas in Sweden emissions were much lower
in relation with lateral transport, i.e. 0.2 : 1 (Humborg
et al., 2010).
Our mean CO2 and CH4 efflux estimates of
1690 � 575 g C m�2 (stream surface) yr�1 and
26 � 10 g C m�2 yr�1, respectively, fall well within the
range of fluxes measured from a Scottish peatland
stream; 259–45878 and 0–345 mg C m�2 d�1 for CO2
and CH4, respectively (Hope et al., 2001). Our CO2
fluxes are also comparable to estimates for Swedish riv-
ers of 473–3032 g C m�2 yr�1 (Humborg et al., 2010),
and a little less than the estimated mean for streams
and rivers in the USA (i.e. 2370 � 800 g C m�2 yr�1;
Butman & Raymond, 2011), as well as that used by
Jonsson et al. (2007) for 1st to 5th order streams in their
C budget for boreal catchment in Sweden, i.e. 7686 mg
C m�2 d�1. The highest uncertainty in our flux esti-
mates remains in the estimate for peatland ditches. In
Finland, CH4 fluxes from ditches of peatlands drained
for forestry have been measured to be in autumn on
average 229 and 66 mg CH4 m�2 d�1 for minerotrophic
and ombrotrophic mires, respectively (Minkkinen et al.,
1997), with the annual average CH4 emissions ranging
7–164 g CH4 m�2 yr�1 (Minkkinen & Laine, 2006).
These are comparable with our estimate of 110 mg
C m�2 d�1 (40 g C m�2 yr�1) for our sample site 11 of
ombrotrophic peatland. Our estimate is also close to
that of Dinsmore et al. (2010) who measured an average
evasion rate at 107 mg C m�2 d�1 for Black Burn,
Scotland, the stream draining an ombrotrophic peat-
land affected by peat extraction area and overgrown
drainage ditches. Minkkinen & Laine (2006) measured
high CH4 fluxes from peatland drainage ditches even
through the ice cover in winter and thus, assuming zero
flux during ice cover may underestimate actual CH4
fluxes from streams. The role of the effluxes from
narrowest stream width class is large and prone to pro-
duce significant error if not taken into account (Butman
& Raymond, 2011; Wallin et al., 2013). Although the
role of CH4 was minor in C cycling, its radiative forcing
was 12% of the total GWP when calculated as CO2
equivalents (Forster et al., 2007).
Peatlands are considered as important sources for
CO2 and CH4 in streams and they have shown to corre-
late with stream water organic carbon concentrations
(Hope et al., 2001; Teodoru et al., 2009). We did not find
such a correlation either with peatland coverage and
CO2 or DOC and CO2 concentration, a finding similar
to Finnish lakes, where CO2 concentrations correlated
negatively with peatland coverage in the catchment
(Kortelainen et al., 2006). The difference may stem from
a high proportion of drained peatlands in Finnish
catchments. However, our most peatland affected
sample point showed very high CO2 concentrations in
the summer and it is possible that the effect of peatland
in stream CO2 concentration is diminished if sampling
takes place at a distance from the peatland from where
the CO2 is effectively emitted. €Oquist et al. (2009) esti-
mated that 90% of the groundwater DIC export to bor-
eal headwater streams is emitted as CO2 into the
atmosphere within 200 m of the water entering the
stream. However, peatlands were an important CH4
source and especially the drainage density correlated
with CH4 concentration, which is consistent with Dy-
son et al. (2011) who found stream water CH4 concen-
trations to be an order of magnitude higher in a
drained peatland site than in their undrained control
site in eastern Finland. This is counterbalancing the
decrease in CH4 fluxes from drained peatland surfaces
compared with natural ones (Nyk€anen et al., 1998).
Lake sediments are also hot spots for CH4 production
(e.g. Kankaala et al., 2006) from where it is released to
streams as well as to the atmosphere especially during
spring and autumn turnover periods. In our study area,
the increase in CH4 in streams in late summer–earlyautumn could be associated with turnover of lakes and
ponds. There was a clear summer time CH4 source
from agricultural land in both intensively studied
catchments, although in the annual scale the effect of
agricultural land on CH4 concentrations was negative.
It might be that agricultural soils were not the source
per se but easily degradable DOM was probably turned
into CH4 in anoxic sediments of the stream during low
flow (no. 4) or in a connected pond from where it was
released with increasing flow (no. 10). Otherwise, the
© 2013 John Wiley & Sons Ltd, Global Change Biology, 19, 3607–3620
3616 J . HUOTARI et al.
better degradability of DOM from agriculture could not
be seen in the gas concentrations.
In contrast, agricultural land clearly increased DIC
concentrations in streams. The water fluxes can be very
different between agricultural land and forests since in
agricultural land a larger amount of water leaves
through ground water runoff enabling DOC to be
absorbed by mineral soils, while in nonagricultural
land water flows in near surface runoff where DOC
concentrations are higher (Mayorga, 2008; Dalzell et al.,
2011). Furthermore, water and bicarbonate fluxes have
shown to be increased due to the runoff changes by
agriculture, which has led to increased weathering of
silicate and carbonate rocks and the consequent seques-
tering of CO2 (Raymond et al., 2008). Thus, runoff from
agricultural land dilutes the DOC in streams while DIC
is higher, which was also evident in our data set
(Table 5). Some of the increase in DIC may be due to
liming of agricultural land. In Sweden, Humborg et al.
(2010) estimated that 4% of the carbonate and bicarbon-
ate export to the sea in Sweden was from pollution,
mainly from liming of agricultural land, which covers
8% of Sweden. Liming probably increases stream water
DIC more in small agriculture-intensive catchments
discharging to lakes. However, it is difficult to assess
from our data the extent of how much liming of agri-
cultural land increases stream water pH and thus retain
more inorganic C dissolved in the water.
Positive correlations between DOC and Q, and rela-
tions between DOC quality and season and Q are well-
established in the literature (e.g. Striegl et al., 2007;
Spencer et al., 2008; Wickland et al., 2012). DOC concen-
tration showed a similar pattern in all streams draining
into Lake P€a€aj€arvi, reflecting the mixture of peatland
and forest in the catchments and diverging C source
from peatlands to forest soils with increasing runoff, as
suggested by Laudon et al. (2011). However, different
land cover and land-use properties clearly affected the
quality of DOM when the flow regime diverged. For
the Mustajoki catchment with moderate peatland cov-
erage, the decreasing aromaticity and molecular weight
with increasing DOC suggest the source of more biode-
gradable DOM during high flow, i.e. diverging source
from peatlands to forest soil (Laudon et al., 2011), as the
peatland DOM can be regarded as less biodegradable
(Tipping et al., 1999; Kalbitz et al., 2003). Also, that the
agricultural land produced DOM of lower aromaticity,
supposedly more labile to microbial degradation, is
consistent with the findings of Williams et al. (2010) in
southern Ontario, Canada. Opposite to the Mustajoki
catchment, in 7 Luhdanjoki and in the L€oytynoja catch-
ment excluding sample point 11, increasing DOC was
associated with the increase in aromaticity of DOM,
implying the higher influence of ground water during
the base flow. DOC concentrations from agricultural
land are usually low and tend to be diluted with
increasing flow (e.g. Ruark et al., 2009; Dalzell et al.,
2011). Thus, the effect of agriculture may be difficult to
discern from stream water chemistry data with even
moderate agricultural land cover. The data also suggest
that different peat types, i.e. sphagnum vs. sedge peat
may give the different DOM quality signature.
Although the processes behind the observed temporal
and spatial changes in DOM quality are not fully
understood, the data suggest that DOM quality proxies
can be useful in interpretation of C sources (Battin,
1998; Helms et al., 2008) and warrant further investiga-
tions.
Although the coverage of the peatlands in the catch-
ment was the most important factor affecting stream
water DOC concentrations, also the extent of anthropo-
logical disturbance on peatlands, i.e. the drainage den-
sity, clearly affected the DOC concentration and DOM
quality. The positive correlation between peatland/
wetland coverage and stream water DOC concentration
have been shown in various studies in the Northern
Hemisphere (e.g. Dillon & Molot, 1997; Aitkenhead
et al., 1999; Arvola et al., 2004; Mattsson et al., 2009;
Laudon et al., 2011). Studies on the effects of peatland
drainages on water DOC concentration have shown
contradictory results (Holden et al., 2004 and references
therein), but e.g. Mitchell & McDonald (1995) have
shown, consistent with our results, that drainage den-
sity, determined as total channel length of 1st, 2nd and
3rd order stream per area, correlates with DOC concen-
tration in runoff. Also DOC concentration in the soil
water of a drained peatland was twice as high as in a
natural peatland within the same peat-complex (Laine
MPP, Str€ommer R & Arvola L, unpublished manu-
script).
The peatland coverage in boreal forest region is 24%
(Wieder et al., 2006), which is similar to the average
peatland percentage of the catchments in Finland, i.e.
22%, determined in a study that covered 88% of total
area of Finland (Mattsson et al., 2005). This emphasizes
the important role of peatlands in DOC export, espe-
cially in Finland where a large portion (55%) of peat-
lands are drained for forestry (Turunen, 2008). For
comparison, agricultural land forms about 8% of the
land area in Finland and e.g. in Canada, where one-
third of the world’s boreal forests are situated, at least
6% of the former boreal forests have been cleared for
farmland and settlements (Smith et al., 2000). C export
from catchments is largely controlled by water runoff
of which human induced disturbance in catchments
may increase. The climate change scenarios for Finland
predict warming and increase in precipitation espe-
cially in winter (Jylh€a et al., 2009) resulting in an
© 2013 John Wiley & Sons Ltd, Global Change Biology, 19, 3607–3620
AQUATIC C EXPORT FROM A BOREAL CATCHMENT 3617
increase in winter time Q. This has already shown to be
taken place in many catchments in Finland during the
period of 1912–2004 (Korhonen & Kuusisto, 2010). We
measured similar or higher DOC concentrations in late
autumn/early winter than during spring snow melt Q
peak (Figs S6–11). This, connected with projected
increase in Q would imply enhanced C loss from catch-
ments in winter. The increase in DOC export from Mus-
tajoki catchment has been predicted for future winters
while the spring load decreases (Naden et al., 2010).
Similar results were predicted for the nearby Valkea-
Kotinen catchment using the INCA-C model (Futter
et al., 2009),
In concordance with Wallin et al. (2013) CO2 evasion
from streams was a dominant component of the C
export and constituted 49% of the total lateral dissolved
C and evasive C flux of the streams. Our measurements
did not include sedimentation or transport of POC, but
the emission per accumulation ratio has been estimated
on average as 21 : 1 for Finnish lakes (Kortelainen et al.,
2006) and the DOC:POC ratio was 10 : 1 e.g. in a humic
river in northern Finland (Heikkinen, 1989). Adding
estimates according to these ratios for C accumulation
in lakes and POC transport in streams to our estimates
of emission and transport of dissolved C, together they
make up to 3955 t C yr�1, which is 19.3 g C m�2 that
can be estimated to escape from the terrestrial system.
This is 0.15% of the estimated mean soil C stock (132.1
Mg ha�1) and 0.28% of the mean standing forest bio-
mass (68.9 Mg ha�1) in the catchment of Lake P€a€aj€arvi
(Hartman et al., 2012). We do not have measurements
of the terrestrial net ecosystem exchange (NEE) of CO2
of the catchment, but when we use average values from
the literature for different aged forest from a clear-cut
to 75-year old pine forest (Kolari et al., 2004), for peat-
land drained for forestry (Lohila et al., 2011) and for
agricultural land (Lohila et al., 2004) in Finland and
take account of their areal extent, we end up with an
estimate of 100 g C m�2 for the whole catchment. Thus,
our estimated C export from the terrestrial system
would be 19% of the terrestrial NEE. For comparison,
Humborg et al. (2010) estimated that in Sweden aquatic
DIC fluxes alone would correspond to 10% of the NEE.
Our results clearly indicate the importance of consid-
ering different land-use patterns and both streams/
ditches and lake surface areas when determining areal
budgets of different C species. High peatland propor-
tion in the catchment as well as high drainage density
of the peatlands clearly increased the DOC in streams
whereas agricultural land increased DIC. Aquatic trans-
port of terrestrial C is regionally important and espe-
cially the gas evasion from the smallest streams must
be taken into account in C budgets. The ultimate fate
of terrestrial C transported by streams, whether it is
sedimented or returned to the atmosphere en route to
the sea, determines the net effect of the terrestrial
system to the global atmospheric C budget. Whatever
the fate is, the aquatic C escape from terrestrial system
is regionally and globally important and should be
taken into account as an integral part of the terrestrial
C cycle.
Acknowledgements
This study was supported by the Academy of Finland throughprojects Pro-DOC and CLIMES (Grants 127922 and 256231 toMartin Forsius) and Academy Research Fellow Post to HannuNyk€anen (Grants 136455, 140964), and by the Lammi BiologicalStation of University of Helsinki. Leena Vitie, Aki Virtanen andMartti Yl€atupa are acknowledged for their efforts in the fieldsampling and in the laboratory, as well as Jaakko Vainionp€a€aand Riitta Ilola for measurements and laboratory analyses. Wethank John Loehr for correction of the language of themanuscript.
References�Agren A, Buffam I, Berggren M, Bishop K, Jansson M, Laudon H (2008) Dissolved
organic characteristics in boreal streams in a forest-wetland gradient during the
transition between winter and summer. Journal of Geophysical Research, 113,
G03031. doi:10.1029/2007JG000674.
Aitkenhead JA, Hope D, Billett MF (1999) The relationship between dissolved organic
carbon in stream water and soil organic carbon pools at different spatial scales.
Hydrological Processes, 13, 1289–1302.
Armentano TV, Menges ES (1986) Patterns of change in the carbon balance of organic
soil-wetlands of the temperate zone. Journal of Ecology, 74, 755–774.
Arvola L, R€aike A, Kortelainen P, J€arvinen M (2004) The effect of climate and landuse
on TOC concentrations and loads in Finnish rivers. Boreal Environment Research, 9,
381–387.
Aufdenkampe AK, Mayorga E, Raymond PA et al. (2011) Riverine coupling of bio-
geochemical cycles between land, oceans, and atmosphere. Frontiers in Ecology and
the Environment, 9, 53–60. doi:10.1890/100014.
Bastviken D, Tranvik LJ, Downing JA, Crill PM, Enrich-Prast A (2011) Freshwater
methane emissions offset the continental carbon sink. Science, 331, 50.
Battin TJ (1998) Dissolved organic matter and its optical properties in a blackwater
tributary of the upper Orinoco river, Venezuela. Organic Geochemistry, 28, 561–569.
Battin TJ, Luyssaert S, Kaplan LA, Aufdenkampe AK, Richter A, Tranvik LJ (2008)
The boundless carbon cycle. Nature Geoscience, 1, 95–100. doi:10.1038/ngeo101.
Billett MF, Palmer SM, Hope D et al. (2004) Linking land-atmosphere-stream carbon
fluxes in a lowland peatland system. Global Biogeochemical Cycles, 18, GB1024. doi:
10.1029/2003GB002058.
Buffam I, Turner MG, Desai AR et al. (2011) Integrating aquatic and terrestrial com-
ponents to construct a complete carbon budget for a north temperate lake district.
Global Change Biology, 17, 1193–1211. doi:10.1111/j.1365-2486.2010.02313.x.
Butler JN (1982) Carbon Dioxide Equilibria and Their Applications. Addison-Wesley,
Boston.
Butman D, Raymond PA (2011) Significant efflux of carbon dioxide from streams and
rivers in the United States. Nature Geoscience, doi:10.1028/NGEO1294.
Christensen TR, Johansson T, Olsrud M et al. (2007) A catchment-scale carbon and
greenhouse gas budget of a subarctic landscape. Philosophical Transactions of the
Royal Society A, 365, 1643–1656. doi: 10.1098/rsta.2007.2035.
Cole JJ, Prairie YT, Caraco NF et al. (2007) Plumbing the global carbon cycle: integrat-
ing inland waters into the terrestrial carbon budget. Ecosystems, 10, 171–184.
doi:10.1007/s10021-0069013-8.
Dalzell BJ, King JY, Mulla DJ, Finlay JC, Sands GR (2011) Influence of subsurface
drainage on quantity and quality of dissolved organic matter export from agricul-
tural landscapes. Journal of Geophysical Research, 116, G02023. doi:10.1029/
2010JG001540.
Dillon PJ, Molot LA (1997) Effect of landscape form on export of dissolved organic
carbon, iron, and phosphorus from forested stream catchments. Water Resources
Research, 33, 2591–2600.
© 2013 John Wiley & Sons Ltd, Global Change Biology, 19, 3607–3620
3618 J . HUOTARI et al.
Dinsmore KJ, Billett MF, Skiba UM, Rees RM, Drewer J, Helfter C (2010) Role of aqua-
tic pathway in the carbon and greenhouse gas budgets of a peatland catchment.
Global Change Biology, 16, 2750–2762. doi:10.1111/j.13652486.2009.02119.x.
Dyson KE, Billett MF, Dinsmore KJ, Harvey F, Thomson AM, Piirainen S, Kortelainen
P (2011) Release of aquatic carbon from two peatland catchments in E. Finland
during the spring snowmelt period. Biogeochemistry, 103, 125–142.
Forster PF, Artaxo P, Lowe DC et al. (2007) Changes in atmospheric constituents and
in radiative forcing. In: Climate Change 2007: The Physical Science Basis chapter 2.
(eds Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M,
Miller HL), pp. 129–234. Cambridge University Press, Cambridge.
Futter M, Forsius M, Holmberg M, Starr M (2009) A long-term simulation of the
effects of acidic deposition and climate change on surface water dissolved
organic carbon concentrations in a boreal catchment. Hydrology Research, 40,
291–305.
Hanson PC, Pollard AI, Bad DL, Predick K, Carpenter SR, Foley JA (2004) A model of
carbon evasion and sedimentation in temperate lakes. Global Change Biology, 100,
1285–1298.
Hartman M, J€arnstedt J, Starr M, Ukonmaanaho L (2012) Estimation of carbon stores
in forest biomass and soil in a catchment in south Finland BIOGEOMON2012. In:
BIOGEOMON 2012, The 7th International Symposium on Ecosystem Behaviour (eds
Fernandez IJ, Norton SA, Wilson T), p. 229. Villanova University, Villanova.
Heikkinen K (1989) Organic carbon transport in an undisturbed boreal humic river in
northern Finland. Archiv F€ur Hydrobiologie, 117, 1–19.
Helms JR, Stubbins A, Ritchie JD, Minor EC, Kieber DJ, Mopper K (2008) Absorption
spectral slopes and slope ratios as indicators of molecular weight, source, and
photobleaching of chromophoric dissolved organic matter. Limnology and Oceanog-
raphy, 53, 955–969.
Hernes PJ, Spencer RGM, Dyda RY, Pellerin BA, Bachand PAM, Bergamaschi BA
(2008) The role of hydrologic regimes on dissolved organic composition in an agri-
cultural watershed. Geochimica Cosmochimica Acta, 72, 5266–5277.
Holden J, Chapman PJ, Labadz JC (2004) Artificial drainage of peatlands: hydrologi-
cal and hydrochemical process and wetland restoration. Progress in Physical Geog-
raphy, 28, 95–123.
Hope D, Palmer SM, Billet MF, Dawson JJC (2001) Carbon dioxide and methane eva-
sion from a temperate peatland stream. Limnology and Oceanography, 46, 847–857.
Humborg C, M€orth C-M, Sundbom M, Borg H, Blenckner T, Giesler R, Ittekot V
(2010) CO2 supersaturation along the aquatic conduit in Swedish watersheds as
constrained by terrestrial respiration, aquatic respiration and weathering. Global
Change Biology, 16, 1966–1978. doi:10.1111/j.1365-2486.2009.02092.x.
Huotari J, Ojala A, Peltomaa E et al. (2011) Long-term direct CO2 flux measurements
over a boreal lake: five years of eddy covariance data. Geophysical Research Letters,
38, L18401. doi:10.1029/2011GL048753.
J€ahne B, Heinz G, Dietrich W (1987) Measurement of the diffusion coefficient of spar-
ingly soluble gases in water. Journal of Geophysical Research, 92, 10767–10776.
Jonsson A, Algesten G, Bergstr€om A-K, Bishop K, Sobek S, Tranvik LJ, Jansson M
(2007) Integrating aquatic carbon fluxes in a boreal catchment carbon budget. Jour-
nal of Hydrology, 334, 141–150. doi:10.1016/j.jhydrol.2006.10.003.
Jylh€a K, Ruosteenoja K, R€ais€anen J et al. (2009) Arvioita Suomen muuttuvasta ilmastosta
sopeutumistutkimuksia varten. ACCLIM-hankkeen raportti 2009. Reports 2009:4, Finn-
ish Meteorological Institute, Helsinki [The changing climate in Finland: estimates
for adaptation studies. ACCLIM project report 2009], pp. 104. [in Finnish with
English summary].
Kalbitz K, Schmerwitz J, Schwesig D, Matzner E (2003) Biodegradation of soil-derived
dissolved organic matter as related to its properties. Geoderma, 113, 273–291.
Kankaala P, Huotari J, Peltomaa E, Saloranta T, Ojala A (2006) Methanotrophic activ-
ity in relation to methane efflux and total heterotrophic bacterial production in a
stratified, humic, boreal lake. Limnology and Oceanography, 51, 1195–1204.
Kankaala P, Huotari J, Tulonen T, Ojala A (2013) Lake-size dependent physical forc-
ing drives carbon dioxide and methane effluxes from lakes in a boreal landscape.
Limnology and Oceanography, in press.
Kling GW, Kipphut GW, Miller MC (1991) Arctic lakes and streams as gas conduits
to the atmosphere: implications for tundra carbon budgets. Science, 251, 298–301.
Kolari P, Pumpanen J, Rannik €U, Ilvesniemi H, Hari P, Berninger F (2004) Carbon bal-
ance of different aged Scots pine forests in Southern Finland. Global Change Biology,
10, 1106–1119. doi:10.1111/j.1365-2486.2004.00797.x.
Kolari P, Pulkkinen M, Kulmala L et al. (2008) Forest ecosystem CO2 exchange. In:
Boreal Forest and Climate Change (eds Hari P, Kulmala L), pp. 351–382. Springer,
Dordrecht. doi: 10.1007/978-1-4020-8718-9_10.
Korhonen J, Kuusisto E (2010) Long-term changes in the discharge regime in Finland.
Hydrology Research, 41, 253–268. doi:10.2166/nh.2010.112.
Kortelainen P, Rantakari M, Huttunen JT et al. (2006) Sediment respiration and lake
trophic state are important predictors of large CO2 evasion from small boreal
lakes. Global Change Biology, 12, 1554–1567. doi:10.1111/j.1365-2486.2006.01167.x.
Laitakari I (1980) Lammin kartta-alueen kallioper€a. Suomen geologinen kartta 1: 100 000.
Kallioper€a karttojen selitykset, Sheet 2134, Geological Survey of Finland, Espoo [Pre-
Quaternary rocks of the Lammi map-sheet area.], pp. 29.
Laudon H, Berggren M, �Agren A et al. (2011) Patterns and dynamics of dissolved
organic carbon (DOC) in boreal streams: the role of processes, connectivity, and
scaling. Ecosystems, 14, 880–893. doi:10.1007/s10021-011-9452-8.
Lohila A, Aurela M, Tuovinen J-P, Laurila T (2004) Annual CO2 exchange of a peat
field growing spring barley or perennial forage grass. Journal of Geophysical
Research, 109, D18116. doi:10.1029/2004JD004715.
Lohila A, Minkkinen K, Aurela M, Tuovinen J-P, Penttil€a T, Ojanen P, Laurila T
(2011) Greenhouse gas flux measurements in a forestry-drained peatland indicate
a large carbon sink. Biogeosciences, 8, 3203–3218. doi:10.5194/bg-8-3203-2011.
Mattsson T, Kortelainen P, R€aike A (2005) Export of DOM from boreal catchments:
impacts of land use cover and climate. Biogeochemistry, 76, 373–394. doi:10.1007/
s10533-005-6897-x.
Mattsson T, Kortelainen P, Laubel A, Evans D, Pujo-Pay M, R€aike A, Conan P (2009)
Export of dissolved organic matter in relation to land use along a European cli-
matic gradient. Science of the Total Environment, 407, 1967–1976.
Mayorga E (2008) Harvest of the century. Nature, 451, 405.
Minkkinen K, Laine J (2006) Vegetation heterogeneity and ditches create spatial vari-
ability in methane fluxes from peatlands drained for forestry. Plant and Soil, 285,
289–304.
Minkkinen K, Laine J, Nyk€anen H, Martikainen PJ (1997) Importance of drainage
ditches in emissions of methane from mires drained for forestry. Canadian Journal
of Forest Research, 27, 949–952.
Minkkinen K, Korhonen R, Savolainen I, Laine J (2002) Carbon balance and radiative
forcing of Finnish peatlands 1900-2100 – the impact of forestry drainage. Global
Change Biology, 8, 785–799.
Minkkinen K, Byrne KA, Trettin C (2008) Climate impacts of peatland forestry. In:
Peatlands and Climate Change ((ed Strack M), pp. 98-122. International Peat Society,
Saarij€arvi.
Mitchell G, McDonald AT (1995) Catchment characterization as a tool for upland
water quality management. Journal of Environmental Management, 44, 83–95.
Naden PS, Allott N, Arvola L et al. (2010) Modelling the impacts of climate change on
dissolved organic carbon. In: The Impact of Climate Change on European Lakes (ed.
George G), pp. 221–252. Springer, Dordrecht. doi: 10.1007/978-90-481-2945-4_13.
Nyk€anen H, Alm J, Silvola J, Tolonen K, Martikainen PJ (1998) Methane fluxes on bor-
eal peatlands of different fertility and the effect of long-term experimental lowering
of the water table on flux rates.Global Biogeochemical Cycles, 12, 53–69.
Ojala A, L�opez Bellido J, Tulonen T, Kankaala P, Huotari J (2011) Carbon gas fluxes
from a brown-water and a clear-water lake in the Boreal Zone during a summer
with extreme rain events. Limnology and Oceanography, 56, 61–76.€Oquist MG, Wallin M, Seibert J, Bishop K, Laudon H (2009) Dissolved organic carbon
export across the soil/stream interface and its fate in a boreal headwater stream.
Environmental Science & Technology, 43, 7364–7369. doi:10.10212/es90041h.
Pirinen P, Simola H, Aalto J, Kaukoranta J-P, Karlsson P, Ruuhela R (2012) Climatolog-
ical Statistics of Finland 1981–2010. Reports 2012:1, Finnish Meteorological Institute,
Helsinki, pp. 83.
Raymond PA, Oh NH, Turner RE, Broussard W (2008) Anthropogenically enhanced
fluxes of water and carbon from the Mississippi River. Nature, 451, 449–452.
Raymond PA, Zappa CJ, Butman D et al. (2012) Scaling the gas transfer velocity and
hydraulic geometry in streams and small rivers. Limnology and Oceanography – Flu-
ids & Environments, 2, 41–53. doi: 10.1215/21573689-1597669.
Ruark MD, Brouder SM, Turco RF (2009) Dissolved organic carbon losses from tile
drained agroecosystems. Journal of Environmental Quality, 38, 1205–1215. doi:10.
2134/jeq2008.0121.
Runkel RL, Crawford CG, Cohn TA (2004) Load Estimator (LOADEST): A FORTRAN
Program for Estimating Constituent Loads in Streams and Rivers, Tech. Meth., 4, chap.
A5, 69 pp, U. S. Geol. Surv., Deston, Virginia.
Ryan MG, Binkley D, Fownes JH (1997) Age-related decline in forest productivity:
pattern and process. Advances in Ecological Research, 27, 213–262.
Schelker J, Ekl€of K, Bishop K, Laudon H (2012) Effects of forestry operations on dis-
solved organic carbon concentrations and export in boreal first-order streams.
Journal of Geophysical Research, 117, G01011. doi:10.1029/2011JG001827.
Smith W, Evans B, Wilson T et al. (2000) Canada’s Forests at a Crossroads: An Assessment
in the Year 2000. A Global Forest Watch Canada ReportWorld Resources Institute,
Washington.
© 2013 John Wiley & Sons Ltd, Global Change Biology, 19, 3607–3620
AQUATIC C EXPORT FROM A BOREAL CATCHMENT 3619
Spencer RG, Aiken GR, Wickland KP, Striegl RG, Hernes PJ (2008) Seasonal and spa-
tial variability in dissolved organic matter quantity and composition from the
Yukon River basin, Alaska. Global Biogeochemical Cycles, 22, GB4002. doi: 10.1029/
2008GB003231.
Striegl RG, Dornblaser MM, Aiken GR, Wickland KP, Raymond PA (2007) Carbon
export and cycling by the Yukon, Tanana, and Porcupine rivers, Alaska, 2001–
2005. Water Resources Research, 43, W02411. doi:10.1029/2006WR005201.
Striegl RG, Dornblaser MM, McDonald CP, Rover JR, Stets EG (2012) Carbon dioxide
and methane emissions from the Yukon River system. Global Biogeochemical Cycles,
26, GB0E05. doi:10.1029/2012GB004306.
Teodoru CR, del Giorgio PA, Prairie YT, Camire M (2009) Patterns in pCO2 in boreal
streams and rivers of northern Quebec, Canada. Global Biogeochemical Cycles, 23,
GB2012. doi: 10.1029/2008GB003404.
Tipping E, Woof C, Rigg E et al. (1999) Climatic influences on the leaching of dis-
solved organic matter from upland UK moorland soils, investigated by a field
manipulation experiment. Environment International, 25, 83–95.
Tranvik LJ, Downing JA, Cotner JB et al. (2009) Lakes and reservoirs as regulators of
carbon cycling and climate. Limnology and Oceanography, 54, 2298–2314.
Turunen J (2008) Development of Finnish peatland area and carbon storage 1950–
2000. Boreal Environment Research, 13, 319–334.
Wallin MB, Grabs T, Buffam I, Laudon H, �Agren A, €Oquist MG, Bishop K (2013) Eva-
sion of CO2 from streams – The dominant carbon export through the aquatic conduit
in a boreal landscape.Global Change Biology, 19, 785–797. doi:10.1111/gcb.12083.
Weishaar JL, Aiken GR, Bergamaschi BA, Fram MS, Fujii R, Mopper K (2003) Evalua-
tion of specific ultraviolet absorbance as an indicator of the chemical composition
and reactivity of dissolved organic carbon. Environmental Science & Technology, 37,
4702–4708.
Wickland KP, Aiken GR, Butler K, Dornblaser MM, Spencer RGM, Striegl RG (2012)
Biodegradability of dissolved organic carbon in the Yukon River and its tributar-
ies: seasonality and importance of inorganic nitrogen. Global Biogeochemical Cycles,
26, GB0E03. doi: 10.1029/2012GB004342.
Wieder RK, Vitt DH, Benscoter BW (2006) Peatlands and the Boreal forest. In: Boreal
Peatland Ecosystems, Ecological Studies 188 (eds Wieder RK, Vitt DH), pp. 1–8.
Springer-Verlag, Heidelberg, Germany.
Williams CJ, Yamashita Y, Wilson HF, Jaff�e R, Xenopoulos MA (2010) Unraveling the
role of land use and microbial activity in shaping dissolved organic matter charac-
teristics in stream ecosystems. Limnology and Oceanography, 55, 1159–1171.
Supporting Information
Additional Supporting Information may be found in theonline version of this article:
Figure S1. Lake P€a€aj€arvi and the studied subcatchments.Figure S2. Modelled and measured load of DOC and DIC toLake P€a€aj€arvi from 6 Mustajoki catchment.Figure S3. Modelled and measured load of DOC and DIC toLake P€a€aj€arvi from 11 L€oytynoja catchment.Figure S4. Modelled and measured load of DOC and DIC toLake P€a€aj€arvi from 12 Koiransuolenoja catchment.Figure S5. Modelled and measured load of DOC and DIC toLake P€a€aj€arvi from 6 Haarajoki catchment.Figure S6. CH4 (mg C m�3), DIC (g C m�3) and DOC (gC m�3) concentrations within Mustajoki catchment, samplepoints 1–6.Figure S7. CH4 (mg C m�3), DIC (g C m�3) and DOC (gC m�3) concentrations in 7 Luhdanjoki.Figure S8. CH4 (mg C m�3), DIC (g C m�3) and DOC (gC m�3) concentrations in 8 P€a€aj€arvi.Figure S9. CH4 (mg C m�3), DIC (g C m�3) and DOC (gC m�3) concentrations within L€oytynoja catchment, samplepoints 9–12.Figure S10. CH4 (mg C m�3), DIC (g C m�3) and DOC (gC m�3) concentrations in 13 Koiransuolenoja.Figure S11. CH4 (mg C m�3), DIC (g C m�3) and DOC (gC m�3) concentrations in 14 Haarajoki.Table S1. Comparison of k600 values (mean of monthly val-ues � SE, m d�1) determined with different models fromRaymond et al. (2012).
© 2013 John Wiley & Sons Ltd, Global Change Biology, 19, 3607–3620
3620 J . HUOTARI et al.