Cultivation of a perennial grass for bioenergy on a boreal organic soil - carbon sink or source

Cultivation of a perennial grass for bioenergy on a borealorganic soil – carbon sink or source?

N A R A S I N H A J . S H U R PA L I *, N I I N A P. H Y V O N E N *, J A R I T . H U T T U N E N *, R O B E R T J .

C L E M E N T w , M A R K U S R E I C H S T E I N z, H A N N U N Y K A N E N *, C H R I S T I N A B I A S I * and

P E R T T I J . M A R T I K A I N E N *

*Department of Environmental Science, University of Kuopio, PO Box 1627, FIN-70211 Kuopio, Finland, wRoom No. 202 Crew

Building, West Mains Road, Edinburgh, EH9 3JN, UK, zBiogeochemical Model-Data Integration Group, Max-Planck-Institute for

Biogeochemistry, Hans-Knoll-Strasse 10, D-07745 Jena, Germany

Abstract

The area under the cultivation of perennial bioenergy crops on organic soils in the

northern countries is fast increasing. To understand the impact of reed canary grass

(RCG, Phalaris arundinaceae L.) cultivation on the carbon dioxide (CO2) balance of an

organic soil, net ecosystem CO2 exchange (NEE) was measured for four years in a RCG

cultivated cutover peatland in eastern Finland using the eddy covariance technique.

There were striking differences among the years in the annual precipitation. The annual

precipitation was higher during 2004 and 2007 and lower during 2005 and 2006 than the

1971–2000 regional mean. During wet growing seasons, moderate temperatures, high

surface soil moisture and low evaporative demand favoured high CO2 uptake. During

dry seasons, owing to soil moisture and atmospheric stress, photosynthetic activity was

severely restricted. The CO2 uptake [gross primary productivity (GPP)] was positively

correlated with soil moisture, air temperature and inversely with vapour pressure deficit.

Total ecosystem respiration (TER) increased with increasing soil temperature but

decreased with increasing soil moisture. The relative responses of GPP and TER to

moisture stress were different. While changes in TER for a given change in soil moisture

were moderate, variations in GPP were drastic. Also, the seasonal variations in TER were

not as conspicuous as those in GPP implying that GPP is the primary regulator of the

interannual variability in NEE in this ecosystem. The ecosystem accumulated a total of

398 g C m�2 from the beginning of 2004 until the end of 2007. It retained some carbon

during a wet year such as 2004 even after accounting for the loss of carbon in the form of

harvested biomass. Based on this CO2 balance analysis, RCG cultivation is found to be a

promising after-use option on an organic soil.

Keywords: bioenergy, carbon balance, drained peatlands, eddy covariance, greenhouses gases, net

ecosystem exchange, organic soil, perennial crop, Phalaris arundinaceae, reed canary grass

Received 14 November 2008; revised version received 15 December 2008 and accepted 20 December 2008

Introduction

Carbon dioxide (CO2) is the principal greenhouse gas

driving global warming. Current CO2 levels in the

atmosphere stand at 387 ppm, up almost 40% since

the industrial revolution and are the highest for at least

the last 650 000 years. From 1970 to 2000, the concentra-

tion rose by about 1.5 ppm each year, but since 2000 the

annual rise has surged to an average 2.1 ppm. The

annual mean CO2 growth rate for 2007 was 2.14 ppm.

According to IPCC (2007), this increase is anthropo-

genic and is brought about primarily by the combustion

of fossils fuels and deforestation. It is mandatory, there-

fore, for all signatory members of the Kyoto Protocol to

reduce their CO2 emissions by at least 5% below the

1990 levels by the 2008–2012 commitment period. One

of the strategies to reduce these emissions is to increase

the use of renewable energy. Bioenergy production from

biomass has been proposed to be one of the ways to

increase the share of renewable energy in the total

energy consumption. Bioenergy is considered to pro-

duce ‘carbon neutral’ energy, as the biomass releases theCorrespondence: Narasinha J. Shurpali, tel. 1 358 17 163 321,

fax 1 358 17 163 750, e-mail: [email protected]

GCB Bioenergy (2009) 1, 35–50, doi: 10.1111/j.1757-1707.2009.01003.x

r 2009 The AuthorsJournal compilation r 2009 Blackwell Publishing Ltd 35

same amount of carbon upon combustion as the

amount of atmospheric CO2 captured during the photo-

synthetic process. Furthermore, bioenergy use contri-

butes to the security of energy supply and enhances the

goals of local employment.

Northern wetlands store vast amounts of carbon. In

their natural state, they have been shown to be sinks for

atmospheric CO2 (Turunen et al., 2002) and sources of

methane on a long-term basis. However, when they are

drained for forestry, agriculture or peat extraction, the

thickness of their aerobic soil layer increases. This leads

to an increase in the rate of organic matter decomposi-

tion. As a result, peatlands, which were once a sink for

carbon, are turned into atmospheric carbon sources

after drainage (Minkkinen et al., 2002).

Irrespective of being cultivated with agricultural

crops such as barley, wheat or potato or grasses, organic

soils in the boreal region have been reported to be

emitting large amounts of CO2 (2.9–30.4 ton CO2 ha�1

yr�1) into the atmosphere (Nykanen et al., 1995; Kasi-

mir-Klemedtsson et al., 1997; Maljanen et al., 2002, 2003,

2004; Lohila et al., 2004). Drainage of peatlands for peat

extraction for energy is a common practice in the Fenno-

Scandinavian region, Canada and Russia. These peat-

land areas abandoned after peat extraction are a threat

to the environment (Silvola et al., 1996; Sundh et al.,

2000; Waddington et al., 2002). Significant CO2 emis-

sions have been reported also from afforested organic

soils (Makiranta et al., 2007). Therefore, organic soils in

general have been included among the areas with high

risk of significant soil carbon losses and hence have

been recommended to be kept out of biomass produc-

tion for bioenergy (OECD, 2007).

In the boreal region, the area under the cultivation of

reed canary grass (RCG) (Phalaris arundinacea L.) as a

bioenergy crop is fast increasing. For example, the area

under RCG in Finland during 2008 was about 19 000 ha

and it is projected to increase to 100 000 ha by 2012.

Although RCG can be grown on most soil types, hu-

mus-rich, wet soils have been recommended for its

cultivation. The crop thrives well with the highest

biomass yields on such soil types (Venendaal et al.,

1997; Lewandowski et al., 2003). As a result, RCG is

generally cultivated in the region (e.g., in Finland and

Sweden) on organic soils (e.g., on cutover peatlands

abandoned after peat extraction). While several studies

have documented the carbon balance of forage grasses

and cereal crops on organic soils, there is hardly any

information on the carbon balance of a perennial bioe-

nergy crop such as RCG on these risky soil types. Most

of the studies on RCG cultivation conducted so far have

focussed mainly on quantifying the yield potential,

genetic characters and fuel quality aspects of RCG as

a bioenergy crop (Venendaal et al., 1997; Sahramaa, 2004;

Finell & Nilsson, 2005; Larson, 2006). However, the green-

house gas dynamics and overall benefits of RCG cultiva-

tion for bioenergy use are yet unknown. As discussed

above, there is a risk that organic soils lose large amounts

of CO2 to the atmosphere thereby reducing the benefits of

bioenergy production. However, the data on perennial

crops on organic soils are lacking. With this in view, we

measured carbon balance of RCG cultivation on an

organic soil (a cutover peatland) in eastern Finland over

four growing seasons (2004–2007) employing the eddy

covariance (EC) technique. The main objectives of this

paper are to quantify the net ecosystem CO2 exchange

during 4 years with varying climatic conditions and

investigate the variables controlling CO2 exchange from

RCG cultivation on an organic soil.

Materials and methods

The site

The study site (referred to as the Linnansuo cutover

peatland site – 621300N, 301300E) located in eastern

Finland lies on the border of southern and mid boreal

climatic zones. It is 15 ha in area and is cultivated with

RCG, a perennial grassland crop. Based on the 30-year

normal (1971–2000) climatic data, the mean annual

temperature and precipitation in the region are 2.1 1C

and 669 mm, respectively (Drebs et al., 2002). The max-

imum and minimum temperatures of �10.6 and 16.0 1C

occur in January and July, respectively. July and August

are the wettest months with 80–90 mm of rainfall.

February, March and April, on an average, receive the

lowest amount of precipitation mostly in the form of

snow with about 35 mm during each month.

The preparation of the site for peat extraction, includ-

ing its drainage by ditching, was initiated in 1976. Peat

extraction began in 1978 (Vapo Ltd. Energy, 2003). In

2001, however, when the residual peat thickness ranged

between 20 and 85 cm, peat extraction ceased and

cultivation of bioenergy crops began. The drainage

ditches divide the area into 20-m-wide strips. The strips

with the residual peat layer were tilled, limed and

fertilized with a per hectare surface application of

59.5 kg N, 14.0 kg P and 45.5 kg K. These strips were

then sown with the seeds of a low-alkaloid, hardy

RCG variety ‘Palaton’. As a part of the agronomic

practice in the region, the cultivation of RCG on such

organic soils generally follows a 10-year rotation cycle.

Except in the beginning of the rotation cycle, the soil is

not tilled at all. Fertilizers are applied every year.

Whether the soil needs additional lime is decided

depending upon the soil pH and crop performance in

the preceding year. The general RCG cultivation practice

in the Fenno-Scandinavian region is to harvest the crop in

36 N . J . S H U R PA L I et al.

r 2009 The AuthorsJournal compilation r 2009 Blackwell Publishing Ltd, GCB Bioenergy, 1, 35–50

the spring of the following year. According to Landstrom

et al. (1996) and Burvall (1997), harvesting RCG in the

following spring, as opposed to autumn harvesting,

improves its fuel quality for both combustion and gasi-

fication. The fresh weight crop yield during 2004, 2005,

2006 and 2007 was 3.7, 2.0, 3.6 and 4.7 t ha�1, respectively.

Micrometeorological measurements

The EC measurements of CO2, momentum, latent and

sensible heat fluxes initiated in late March of 2004 are

still continuing. The EC system consists of a 3D sonic

anemometer (Model CSAT-3: Campbell Scientific,

Logan, UT, USA) and an open path infrared CO2/H2O

analyzer (Model LI-7500: LI-Cor Inc., Lincoln, NE, USA).

The instruments are mounted on a tower at a height of

3.7 m above the ground and aligned at an angle of 2251.

The predominant winds in the region are from south-

west and northeast. Constrained by the size of the site,

the instrument tower is located on the northern border

of the site with the instruments pointing towards the

south-westerly direction so that the fetch within the

acceptable range of sonic wind angles (225 � 851) is

representative of the RCG cultivation site. The EC

signals are measured at a frequency of 10 Hz using a

datalogger (Model CR5000: Campbell Scientific) and the

raw data are stored in a binary format on 1 Gb PCMCA

cards. Real-time flux calculations are performed to ob-

tain 30 min averaged flux estimates.

Postprocessing of the raw data was performed using

the ‘EDIRE’ software (available at http://www.geos.ed.

ac.uk/abs/research/micromet/EdiRe) developed at

the University of Edinburgh (Mauder et al., 2007). To

begin with, sonic tilt correction was applied to the data

to orient the vertical axis of the sonic anemometer to the

‘true’ vertical direction. The signals were then despiked

with an upper limit of 20 SDs of the differential of the

raw signal within a signal width of eight scans. Rele-

vant statistics such as means, SDs, variances and covar-

iances were calculated. Finally, frequency response

corrections were applied to account for sensor separa-

tion and high frequency flux losses due to poor instru-

mental responses (Moore, 1986) and density (WPL)

corrections for simultaneous transfer of heat and water

vapour (Webb et al., 1980). Initial quality of the 30 min

averaged flux values and associated micrometeorologi-

cal data was performed using the ‘EDIVAQ’ data quality

control software package developed at the University of

Edinburgh. The various quality control tests such as

range test, despiking, stasis and difference tests were

used to set the bad quality data apart. The data were

then further quality controlled to include only those data

collection periods during when the prevailing winds

were from an acceptable range. Missing or rejected data

amounted to nearly 46% of the data measured during

periods with favourable wind directions.

EC data gap filling and flux partitioning

Data gaps were filled with the marginal distribution

sampling method described in Reichstein et al. (2005)

and further evaluated in Moffat et al. (2007). In this

approach, the gap filling of the EC and meteorological

data is performed through methods that consider both

the covariation of fluxes with meteorological variables

[global radiation, temperature and vapour pressure def-

icit (VPD)] and the temporal autocorrelation of the fluxes.

The gap filling results are more reliable if there are no

gaps in the supporting micrometeorological data. There-

fore, to ensure that we had a complete set of supporting

data to begin with, we first filled the gaps in these data

(u* not included) with the relevant meteorological data

monitored at Mekrijarvi Research Station, University of

Joensuu, situated within 30 km of our study site.

Only original data (not gap filled) were used for flux

partitioning according to Reichstein et al. (2005). The

underlying assumption employed here is that at night,

in the absence of gross primary productivity (GPP), net

ecosystem CO2 exchange (NEE) values as measured by

the EC technique represent total ecosystem respiration

(TER). With this assumption we examined the relation-

ship between night-time NEE and u* within and outside

the growing season in different measurement years. We

found that NEE was apparently not correlated with u*

in this ecosystem (Fig. 1). Nevertheless, we used a

default u* filter of 0.1 m s�1 and night-time regressions

of NEE and air temperature (T) were calculated using

an exponential regression model (Lloyd & Taylor, 1994)

R Tð Þ ¼ Rref eE0

1Tref�T0

� 1T�T0

� �;

u ∗ (friction velocity) (m s–1)

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

NE

E (

mg

m–2

s–1

)

0.2

0.4

0.6

0.8

1.0

1.2

Fig. 1 Medians ( � standard errors) of the half-hourly night-

time CO2 flux plotted against the respective friction velocity class

during 2004–2007 measurement years.

C O 2 B A L A N C E O F A B I O E N E R G Y C R O P O N A N O R G A N I C S O I L 37


where T0 5�46.021, Tref 5 10 1C, and Rref, the tempera-

ture-independent respiration rate and temperature sen-

sitivity, E0, were free parameters. The temperature

sensitivity E0 was estimated on a 15-day time scale,

but then averaged and assumed to be constant in time.

A temporally varying estimate of Rref was generated

thus linking variation in NEE between shorter and

longer time scales. Once Rref(T) and E0 were deter-

mined, half-hourly daytime TER was estimated from

measured air temperature, and GPP was calculated as

the difference between NEE and TER. For further

details (http://www.bgc-jena.mpg.de/bgc-mdi/html/

eddyproc/index.html), refer to Reichstein et al. (2005).

Supporting measurements

Aboveground plant and root biomass of RCG were

sampled twice a month during the growing season.

The plants were completely harvested from 25 cm�25 cm plots (n 5 9) by clipping at the stem base, and

separated into living, green biomass and dead standing

biomass. From six out of the nine plots a soil core was

additionally taken for belowground root biomass de-

termination. The soil core was sampled to a depth of

15 cm depth, thus up to the maximum rooting depth of

the plants (in general, roots from the 10 to 15 cm deep

layer contributed o2% to the total (0–15 cm depth) root

biomass, by using a metal soil corer (of 7 cm diameter).

In the laboratory, roots were collected and carefully

washed over a 0.2 mm sieve. Only the white, living

roots were sampled. Both aboveground plant parts

and roots were then oven dried at 60 1C for 24 h to

obtain a constant weight.

Other supporting meteorological measurements

include air temperature and relative humidity

(Model MP101A: Rotronic Instruments UK Ltd, West

Sussex, UK), net radiation and its components (Model

CNR1: Campbell Scientific, Leicestershire, UK), solar

(Model LI-200SA: LI-Cor Inc.) and photosynthetically

active radiation (Model LI-190SA: Li-Cor Inc.), wind

speed and direction (Model QMV101: Vaisala

Oyj, Vantaa, Finland), soil heat flux (Model HFP01SC,

Campbell Scientific), soil temperature profile

(Model 107: Campbell Scientific), precipitation (MOD-

EL 52202/52203: Campbell Scientific), soil moisture

profile (Model CS616: Campbell Scientific) and snow

depth (Model SR 50A: Campbell Scientific). We used

MATLAB [Version 7.5.0.0342 (R2007b)] and associated

Curve Fitting Toolbox 1.2 for curve fitting and other

data analysis requirements, SPSS package (SPSS for

Windows, Release 14.0.1, 18 November 2005) for statis-

tical analyses.

Results and discussion

Climatic conditions during the measurement period

The climatic conditions during May, June and July are

crucial for the growth and phenology of the RCG crop.

The final crop yield is primarily dependent upon how

vigorous is the growth during these months as they

represent the RCG linear growth phase. Compared with

a 30-year regional average (1971–2000), monthly mean

air temperatures in May and June in 2004, May 2005,

June 2007 were lower than normal and higher in June

2005, May and June 2006 and May 2007. Except in 2005,

July monthly mean temperatures were similar to the

30-year average in all other years. August and Septem-

ber experienced temperatures warmer than the normal

in all years (Fig. 2).

Striking differences existed in the amount of rainfall

received during the various growing seasons (Fig. 2).

The growing season in 2004 was the wettest (554 mm

during 1 May–30 September), relatively drier in 2007

(423 mm) and the driest during 2005 and 2006 (246 and

249 mm, respectively). Compared with a regional 30-

year (1971–2000) average value (338 mm), the May–

September rain amount was higher by 64% in 2004, by

25% in 2007 and 26% lower in 2005 and 2006. Thus, 2004

and 2007 growing seasons were wetter and 2005 and

2006 seasons were drier than normal.

The volumetric soil water content at 30 cm depth (y30

was consistently high and it showed little variation

among the years. The volumetric soil water content

Mon

thly

air

tem

pera

ture

devi

atio

n fr

om th

e no

rmal

(°C

)

–10

0

10

2004200520062007

Mon

thly

pre

cipi

tatio

n de

viat

ion

from

the

norm

al r

ainf

all (

mm

)

–60

0

60

(a)

(b)

Fig. 2 Monthly patterns of deviations from the normal in (a) air

temperature and (b) precipitation at the study site in eastern

Finland during the 2004–2007 measurement period.



measured at 2.5 cm below the surface (y2.5) being the

most sensitive to the changes in surface energy balance

and peatland hydrology varied the most, followed by

the water content at 10 cm soil depth (y). The effect of

frequent rain events during 2004 and 2007 and longer,

frequent dry spells during 2005 and 2006 is clearly

evident in the upper layer soil moisture patterns (Fig. 3).

GPP: photosynthetically active radiation (PAR) andseasonality

As photosynthesis is driven by light and the structure

and function of the RCG cultivation changes markedly

over the growing season, we have examined in the

following the relation between GPP and PAR (Fig. 4)

over two types of growing seasons – one over a wet

growing season (May–September 2004) and the other

over a dry season (similar period in 2005). The response

of GPP to PAR during each month is fitted with a

rectangular hyperbolic model. The results of the non-

linear regression analysis along with prevailing air

temperature, VPD and soil moisture conditions are

shown in Table 1.

During May, the GPP was low (Fig. 4). Under similar

conditions of air temperature, VPD and soil moisture,

GPP values during May 2004 were higher than during

May 2005. GPmax estimated from the May relationship

was nearly twice higher during 2004 than during 2005

(Table 1). GPP increased from May to July. June marks

the period with a linear increase in the RCG plant

growth. Peak vegetation growth occurs during late June

to mid July. During June 2004, with moderate tempera-

Pre

cipi

tatio

n (m

m)

0

30

60

90

120

150

0.0

0.2

0.4

0.6

0.8

Precipitation�2.5

�10

Pre

cipi

tatio

n (m

m)

0

30

60

90

120

150

0.0

0.2

0.4

0.6

0.8

Pre

cipi

tatio

n (m

m)

0

30

60

90

120

150

0.0

0.2

0.4

0.6

0.8

1/1/07 1/3/07 1/5/07 1/7/07 1/9/07 1/11/07 1/1/08

1/1/06 1/3/06 1/5/06 1/7/06 1/9/06 1/11/06 1/1/07

1/1/05 1/3/05 1/5/05 1/7/05 1/9/05 1/11/05 1/1/06

1/1/04 1/3/04 1/5/04 1/7/04 1/9/04 1/11/04 1/1/05P

reci

pita

tion

(mm

)

0

30

60

90

120

150

Vol

umet

ric w

ater

con

tent

(m

3 m

–3)

Vol

umet

ric w

ater

con

tent

(m

3 m

–3)

Vol

umet

ric w

ater

con

tent

(m

3 m

–3)

Vol

umet

ric w

ater

con

tent

(m

3 m

–3)

0.0

0.2

0.4

0.6

0.8

(a)

(b)

(c)

(d)

Fig. 3 Growing season daily distribution of volumetric soil moisture content measured at 2.5 and 10 cm below the soil surface and

precipitation during (a) 2004, (b) 2005, (c) 2006 and (d) 2007.



ture and VPD and high surface soil moisture, GPP was

high. The GPmax value estimated for June 2004 has a

high standard error associated with it and the rectan-

gular hyperbolic relationship between GPP and PAR

during this period yielded a low R2 value (Table 1).

Also, the June 2004 data in Fig. 4 provide no hint of light

saturation implying probably that a hyperbolic relation

may not necessarily be the best-fit option to describe the

relationship. Therefore, we performed a linear regres-

sion between GPP and PAR for this period. The regres-

sion analysis resulted in a highly significant positive

correlation between the two variables (R2 5 0.41,

P 5 0.0001). During June 2005, temperature and VPD

were moderate to high and surface soil moisture con-

tent was severely depleted. The effect of a severe

drought stress is clearly evident from the GPmax value,

which was reduced to nearly half the corresponding

value in June 2004. Similar drought stress conditions

existed in July 2005. As a result, the photosynthetic

activity was restricted to half the potential in July 2004.

August marks the onset of plant senescence and GPP

values showed a declining trend during this month.

Under similar temperature and VPD conditions, GPP

values during August and September of a wet and dry

year were comparable. This seasonal GPP analysis

suggests that June and July are the two crucial months

during which the maximum photosynthetic activity

occurs. The range of GPmax values from different

months with contrasting rainfall distribution patterns

and associated soil moisture regimes presented here are

similar to the values reported for grassland species

studied in Finland and elsewhere (Flanagan et al.,

2002; Lohila et al., 2004). Associated with these GPmax

values, the apparent quantum yields also varied mark-

edly among different years (Table 1). The surface soil

moisture conditions and therefore, the soil hydrological

balance (as evidenced through soil moisture variations)

during these two crucial months largely decide the

extent of seasonal/annual atmospheric carbon uptake

by RCG in this cutover peatland ecosystem.

PAR (μmol m–2 s–1)

0 500 1000 1500 2000 2500

GP

P (

μmol

m–2

s–1

)

GP

P (

μmol

m–2

s–1

)

–30

–24

–18

–12

–6

0


0 500 1000 1500 2000


0 500 1000 1500 2000


0 500 1000 1500 2000–30

–24

–18

–12

–6

0


0 500 1000 1500 2000

May June

July

August September

Fig. 4 Monthly reed canary grass gross primary productivity (GPP) responses to incident photosynthetically active radiation (PAR)

under different soil moisture conditions. Solid circles represent the GPP–PAR response under high soil moisture and open circles under

dry conditions. The fitted curves are the best-fit lines constructed from rectangular hyperbolic nonlinear regression analyses. The results

of the regression analysis are given in Table 1. Only measured data are used in this analysis.



GPP: temperature, VPD and soil moisture

Among the environmental variables, temperature, VPD

and soil moisture constitute an important group of

factors governing CO2 exchange, in addition to incident

PAR. It is difficult to assess the impact of these indivi-

dual factors on GPP as these explaining variables are

often found to exhibit collinearity under field condi-

tions. Therefore, we examined the impact of air tem-

perature and VPD on GPP separately under different

soil moisture classes. For the purpose, we subdivided

the GPP and controlling variable data from each grow-

ing season into May to September monthly datasets,

monthly data were further grouped into four moisture

classes covering a 0.0–0.8 m3 m�3 range in volumetric

surface layer moisture content. Finally, the various

responses presented in Figs 5 and 6 are from periods

when PAR exceeded 1000 mmol m�2 s�1. Note that the

statistically significant GPP–temperature response

curves presented in Fig. 5 are from periods when GPP

did not correlate with VPD. Similarly, VPD–GPP re-

sponse curves in Fig. 6 are from periods when GPP did

not correlate significantly with air temperature.

GPP increased with increasing temperatures. The

relationship between temperature and GPP was ex-

plained by linear and quadratic functions. When soil

moisture was not limiting, GPP increased linearly with-

in a narrower range of temperature variation. However,

when soil moisture was limiting, the air temperature

varied over a wider range. Under these circumstances,

GPP was related to air temperature in a quadratic

relationship, hinting at a peak in GPP at optimum

temperatures within 20–25 1C range. GPP tended to

decrease with any further increase in temperature

above the optimum range (Fig. 6). GPP is known to

decrease with increasing VPD (note the negative sign

convention used for GPP throughout this paper). With

increasing VPD, the plants attempt to restrict moisture

loss through transpiration, especially under soil moist-

ure stress, by closing their stomata. As CO2 also uses the

same pathway for diffusion into the leaves, the rate of

CO2 uptake is drastically reduced. The GPP–VPD re-

sponse curves shown in Fig. 6 show this classical

phenomenon in action in this ecosystem. Similar GPP,

temperature, VPD and soil moisture interactions have

been reported in several ecosystems in previous inves-

tigations [e.g., Wang et al. (2004) in a larch forest in

Japan, Luyssaert et al. (2007) in European pine forests].

TER: temperature, soil moisture, GPP

Although soil temperature is a good indicator of annual

ecosystem respiration, the parameter values derived

from an analysis involving the complete annual data

may not represent the true response to temperature.

TER, which shows large seasonal variations, is a func-

tion of not only temperature but also of plant photo-

synthetic activity, leaf area development, root biomass

and, soil moisture conditions, substrate quality and

other unknown variables. Therefore, with an intention

to examine TER responses to governing variables and

minimize the influence of confounding factors, we

subdivided the annual night-time TER data (estimated

from measured NEE values only) of each year into

growing season (May–September) data and data

outside the growing season (January–April and

October–December). The growing season data were

further subdivided into monthly subsets from May to

Table 1 Seasonal variation in the responses of gross primary productivity (GPP) of reed canary grass on an organic soil to incident

photosynthetically active radiation (PAR) during a wet (2004) and a dry (2005) year

Year Month T VPD y2.5 GPmax a R2

2004 (wet) May 10–15 0.2–1.2 0.6–0.8 �3.1 (0.4) �0.02 (0.06) 0.51

June 15–20 0.1–1.6 0.3–0.7 �29.4 (5.9) �0.03 (0.05) 0.35

July 15–20 0.1–1.0 0.6–0.8 �26.4 (1.9) �0.07 (0.01) 0.56

August 10–15 0.1–0.7 0.5–0.6 �7.1 (0.5) �0.06 (0.02) 0.51

September 10–15 0.1–0.6 0.4–0.6 �5.1 (0.7) �0.03 (0.01) 0.62

2005 (dry) May 10–15 0.1–1.1 0.5–0.7 �1.2 (0.1) �0.08 (0.003) 0.33

June 20–25 1.0–2.0 0.1–0.2 �16.1 (2.9) �0.02 (0.05) 0.40

July 15–20 0.1–1.5 0.0–0.2 �13.5 (0.9) �0.13 (0.03) 0.66

August 10–16 0.1–0.7 0.2–0.4 �6.3 (0.9) �0.08 (0.024) 0.27

September 10–15 0.1–1.2 0.2–0.4 �5.2 (0.4) �0.02 (0.004) 0.60

Results of the nonlinear regression between GPP and PAR are presented. GPP was related to PAR in a rectangular hyperbolic

relation of the form: GPP 5 (GPmax�PAR� a/(GPmax 1 PAR� a)), where GPmax (mmol m�2 s�1) and a are the fitted parameters. Also

shown are the relevant prevalent ranges in environmental variables such as air temperature (T, 1C), vapour pressure deficit (VPD,

kPa) and soil moisture measured at 2.5 cm below the soil surface (Y2.5, m3 m�3). Standard errors of the fitted parameters are shown

brackets. All relationships were statistically significant at Po0.001. Adjusted R2 of regressions are also presented.



September and within each month, TER response to soil

temperature (measured at 2 cm) was examined with the

Q10 model within different soil moisture classes (four

classes with a class interval of 0.2 m3 m�3 covering a soil

moisture range of 0–0.8 m3 m�3). Subdividing the data

and parameterising the Q10 model separately for each

subset did reveal a large variation in both Q10 and R10

values (Fig. 7 and Table 2). The monthly Q10 values

VPD (kPa)0.5 1.0 1.5 2.0 0.5 1.0 1.5 2.0

GP

P (

μmol

m–2

s–1

)

–3

–2

–1

0

Wet

VPD (kPa)0.0 0.5 1.0 1.5 2.0 2.5

–30

–20

–10

0

DryWet

VPD (kPa)

–18

–15

–12

–9

Wet

Fig. 6 Monthly reed canary grass gross primary productivity (GPP) responses to vapour pressure deficit (VPD) under different soil moisture

conditions. Solid circles represent the GPP–VPD response under high soil moisture and open circles under dry conditions. The fitted curves are

the best-fit lines constructed from linear regressions between GPP and VPD. See text for more details. The data used in this analysis are from

periods when air temperature had no significant relation with GPP and photosynthetically active radiation values were 41000mmol m�2 s�1.

Temperature (°C)0 5 10 15

GP

P (

μmol

m–2

s–1

)

GP

P (

μmol

m–2

s–1

)

–5

–4

–3

–2

–1

0

WetDry

10 15 20 25 30 10 15 20 25 30–25

–20

–15

–10

–5

0WetDry

–20

–15

–10

–5

0

Dry

–20

–15

–10

–5

DryWet

10 15 20 25 5 10 15 20–10

–8

–6

–4

–2

0

WetDry

Temperature (°C) Temperature (°C)

May Jun

Jul

Aug Sep


Fig. 5 Monthly reed canary grass gross primary productivity (GPP) responses to air temperature under different soil moisture

conditions. Solid circles represent the GPP–air temperature response under high soil moisture and open circles under dry conditions. The

fitted curves are the best-fit lines constructed from different regression analyses (linear, exponential and quadratic). See text for more

details. The data used in this analysis are from periods when vapour pressure deficit (VPD) had no significant relation with GPP and

photosynthetically active radiation values were 41000mmol m�2 s�1.



ranged from about 2.0 to 5.4. The highest Q10 values

were estimated during May. Low Q10 values were

associated with low soil moisture conditions implying

higher temperature sensitivity when the soil moisture is

high. These observations are consistent with the results

from previous investigations (e.g., Silvola et al., 1996;

Janssens & Pilegaard, 2003).

We investigated the relationship between soil tem-

perature and night-time TER data from periods outside

the growing season (winter period – January–April and

October–December, Fig. 8). A nonlinear regression ana-

lysis employing the Q10 model resulted in high winter-

time Q10 values. The values ranged from 6.2 in 2005 to

12.3 in 2007. These values are two to three times higher

than the July–August values presented above (Fig. 8).

Q10 values as high as 23 have been reported under low

temperatures in a beech forest by Janssens & Pilegaard

(2003). Higher sensitivity of TER at lower temperatures

has also been reported by Kirschbaum (1996) and Ray-

ment & Jarvis (2000). It is speculated that the strong

temperature response during winters could be asso-

ciated with microbial species richness and the strong

temperature effect on microbial populations and their

activity under low-temperature conditions (Janssens &

Temperature (°C)0 4 8 12

TE

R (

μmol

m–2

s–1

)

TE

R (

μmol

m–2

s–1

)

0

2

4

6DryWet

8 10 12 14 16

DryWet

10 12 14 16 18

Dry

0

2

4

6DryWet

8 10 12 14 16 4 6 8 10 12 14 16

DryWet


JuneMay

July

August September


Fig. 7 Growing season monthly reed canary grass total ecosystem respiration (TER) responses to soil temperature (measured at 2 cm

below the surface) moisture under different soil moisture conditions. Solid circles represent the TER–soil temperature response under

dry conditions and open circles under wet conditions. The fitted curves are the best-fit lines constructed from exponential nonlinear

regressions between TER and soil temperature. The results of the regression analyses are given in Table 3.

Table 2 Seasonal responses of total ecosystem respiration

(TER) of reed canary grass on an organic soil-to-soil tempera-

ture measured at 2 cm below the soil surface (Ts – 1C)

Month Year y2.5 R10 Q10 R2 P value

May 2006 0.4–0.6 0.32 (0.04) 5.25 (0.78) 0.91 (0.04) o0.0001

May 2005 0.0–0.2 0.24 (0.02) 5.42 (0.37) 0.99 (0.07) o0.0001

June 2007 0.2–0.4 0.99 (0.28) 2.97 (0.62) 0.76 (0.34) 0.007

June 2006 0.4–0.6 0.41 (0.04) 4.78 (0.37) 0.97 (0.06) o0.0001

July 2005 0.0–0.2 1.39 (0.25) 2.27 (0.25) 0.85 (0.35) o0.0001

August 2006 0.0–0.2 0.91 (0.13) 2.62 (0.38) 0.56 (0.38) o0.0001

August 2005 0.2–0.4 0.79 (0.22) 2.42 (0.46) 0.23 (0.34) o0.0001

September 2005 0.2–0.4 0.92 (0.05) 1.98 (0.09) 0.71 (0.14) o0.0001

September 2004 0.4–0.6 0.69 (0.06) 2.91 (0.24) 0.76 (0.24) o0.0001

TER was related to Ts by the Q10 model: TER ¼ R10 �QðTS=T10Þ10 ,

where R10 and Q10 are the fitted parameters. Details of the

regression analysis are shown in the table.



Pilegaard, 2003). The amount of microbial biomass

carbon in the surface (0–5 cm) soil layer was found to

be significantly higher during spring than during the

rest of the growing season at this study site (N. Pekkar-

inen, personal communication).

Ecosystem respiration as measured by the EC techni-

que includes autotrophic and heterotrophic respiration.

The autotrophic respiration in turn is a combined

process resulting from respiratory losses associated

with aboveground plant parts, roots and rhizosphere.

Gross primary production which is responsible for the

growth and development of plant structural elements

has been hypothesized to be one of the factors driving

ecosystem respiration (Janssens et al., 2002). With this in

view, we examined how the daily rates of TER and GPP

correlate with each other in this ecosystem during

different measurement years. A highly significant

(Po0.0001) positive correlation existed between TER

and GPP during each of the four measurement years.

GPP explained 70% or more of the variation in TER

during 2005, 2006 and 2007, while it could explain only

45% of the variation during 2004. These results seem to

suggest that TER is closely coupled to GPP during drier

years, whereas under high soil moisture conditions, the

two processes are decoupled. This could be attributed

to the differences in how the two processes respond to

soil moisture and temperature. As explained in the

sections above, high photosynthetic rates are favoured

and respiration is inhibited under high soil moisture

conditions. On the contrary, respiratory losses are high-

er and uptake is restricted during drier and warmer

conditions. The relative responses of the individual

processes to the prevailing environmental conditions

are different. The change in respiration rates to a given

variation in moisture level in this ecosystem is moder-

ate. However, change in GPP to a similar change in

moisture level is drastic.

Diurnal patterns of net ecosystem CO2 exchange

In an attempt to understand the interannual variation in

CO2 exchange patterns on a diurnal scale from the RCG

cultivation system, we generated mean diurnal patterns

during peak RCG photosynthetic activity in each mea-

surement year (2004–2007). A uniform averaging period

(from June 15 to July 20) to encompass the maximum

uptake and loss rates from different years was consid-

ered for the purpose. There were large differences

among years in mean diurnal NEE patterns (Fig. 9).

The mean maximum NEE value in 2004 was about

�18.0 mmol m�2 s�1, while it was �8.8, �11.1 and �15.0

mmol m�2 s�1 in 2005, 2006 and 2007, respectively. Also,

the timing of peak NEE during the day was different in

different years. On a mean diurnal basis, NEE peaked at

midday in 2004, around 1400 h in 2007 and at 800 and

900 h in 2005 and 2006, respectively (Fig. 9). There were

no clear differences among years in the mean maximum

night-time NEE values. On an average, the mean max-

imum CO2 losses during 2004–2007 ranged from 3.9 to

7.2mmol m�2 s�1.

Lohila et al. (2004) measured using EC technique CO2

exchange from perennial forage grasses on peat soils in

southern Finland. They reported maximum C uptake

and loss values of approximately �17 and 8mmol m�2

s�1, respectively, during the active growing season in

2002. Multiyear C exchange data from a Canadian

grassland reported by Flanagan et al. (2002) are within

the range of values observed in this study. The mean

maximum NEE values reported by them ranged from

�14 mmol m�2 s�1 in a wet year to �5 mmol m�2 s�1 in a

dry year. While the maximum ecosystem respiration

(7 mmol m�2 s�1) measured using chamber method by

Maljanen et al. (2001) from a grassland on an organic

soil in eastern Finland is comparable to peak TER rates

at our study site, the peak uptake rate (2.5mmol m�2 s�1)

reported in their study is lower compared with RCG

peak daytime NEE ported above. Nieveen et al. (2005)

measured carbon exchange from a grazed pasture on a

drained peat soil in New Zealand and recorded a peak

uptake of �17.2mmol m�2 s�1 and a maximum night-

time respiration value of 8mmol m�2 s�1 during the 2002

growing season. Compared with an agroecosystem, the

uptake and respiration rates measured at our study site

are substantially smaller (e.g., irrigated and rain fed

Temperature (°C)–2 0 2 4 6 8 10 12

Win

ter

TE

R (

μmol

m–2

s–1

)

0.0

0.5

1.0

1.5

2.0

2.5

2004200520062007

Fig. 8 Winter time (January–April and October–December)

reed canary grass total ecosystem respiration (TER) responses

to soil temperature (measured at 2 cm below the surface) during

2004, 2005, 2006 and 2007. The fitted curves are the best-fit lines

constructed from nonlinear regressions between TER and soil

temperature: TER ¼ R10 �QðTS=T10Þ10 , where R10 and Q10 are the

fitted parameters. Details of the regression analysis are also

shown in the figure. The presented regression analysis were

significant at Po0.0001.



Net

eco

syst

em C

O2

exch

ange

(NE

E)

(mg

m–2

s–1

)

–20

–10

0

10

2004200520062007

Pho

tosy

nthe

tical

ly a

ctiv

e ra

diat

ion

(PA

R)

(μm

ol m

–2 s

–1)

0

400

800

1200

1600

Air

tem

pera

ture

(°C

)

0

5

10

15

20

25

Time of the day0 4 8 12 16 20 24

Vap

or p

ress

ure

defic

it (k

Pa)

0.0

0.4

0.8

1.2

1.6

Fig. 9 Interannual variation in the mean diurnal pattern of net ecosystem CO2 exchange (NEE) during peak photosynthetic activity

(June 15–July 20) of reed canary grass cultivated on a cutover peatland in eastern Finland. Also shown are the prevalent environmental

variables when the NEE measurements were made. Values represent the mean � SE, n 5 35.



maize-based cropping systems studied by Verma et al.,

2005). The peak day and night-time NEE values pre-

sented in this paper are comparable to CO2 uptake and

loss rates from spruce forests on podzols in Sweden

(Lindroth et al., 2008).

The interannual variations in diurnal patterns are

associated with changes in precipitation, soil moisture,

air temperature, PAR levels and VPD during 06:00 and

18:00 hours. During wet years, due to sustained cloudy

conditions and well-distributed rain events during the

growing season, the soil moisture was generally non-

limiting, the PAR levels were low, air temperatures

were moderate and the evaporative demand was low

(Figs 2 and 3). These conditions seemed to have fa-

voured a high C uptake by RCG. RCG is known to be a

flood and cold tolerant species capable of accumulating

large biomass on organic soils. Coops et al. (1996) found

in a water-depth gradient experiment that RCG grew

well when the water table was close to the surface and

that the biomass of RCG was significantly reduced

when the water table was 30 cm below the soil surface.

Kellogg et al. (2003) observed a high rate of RCG

germination under flooded conditions than under other

hydrological treatments (leading to lower soil moisture

conditions) in their experiment. They also found that

the RCG seedling mass was higher in soils with high

organic content.

In contrast, due to long dry spells and generally clear

sky conditions during dry years, soil moisture was

limiting, PAR levels, air temperature and VPD after

late-morning hours were generally higher. Soil moisture

at both 2.5 and 10 cm depths were low during the peak

growing seasons in 2005 and 2006 (Fig. 3). As RCG

prefers high soil moisture conditions, it develops a

shallow root system. Based on RCG biomass measure-

ments made at our site during 2006 and 2007 seasons,

we found that nearly 95% of the RCG roots are found in

the upper 15 cm soil depth. Therefore, owing to a

shallow root system and higher evaporative demand

during most of the daytime in dry years, RCG closes its

stomata to restrict the excessive water loss and in the

process, C uptake is drastically reduced. The peak

NEE during 2005 was less than half the peak rate in

2004 (Fig. 10). The soil moisture at 30 cm below the soil

Date (d/m/yy)1/1/04 1/7/04 1/1/05 1/7/05 1/1/06 1/7/06 1/1/07 1/7/07 1/1/08

Net

eco

syst

em C

O2

exch

ange

(NE

E, g

C m

–2 d

ay–1

)

–12

–8

–4

0

4

CO

2 ex

chan

ge (

g C

m–2

day

–1)

–20

–15

–10

–5

0

5

10

GPPTER

(a)

(b)

Fig. 10 Daily distribution of (A) gross primary productivity (GPP) and total ecosystem respiration (TER) and (B) net ecosystem CO2

exchange (NEE) during 2004–2007 from reed canary grass cultivation on a cutover peatland in eastern Finland.



surface was near saturation even during dry years.

However, it appears that the water at this depth re-

mained inaccessible to the shallow roots of RCG.

Daily, seasonal and annual CO2 exchange

The daily variation in CO2 exchange during the grow-

ing season was regulated by variations in soil moisture

in the uppermost soil layers as affected by rainfall

amount and its distribution, incident light, air and soil

temperature and VPD (see the sections above for a

discussion on the factors controlling C exchange). The

number of days during which the ecosystem was a net

sink for atmospheric CO2 varied from year to year. It

was 71 days in 2004, 88 days in 2005, 66 days in 2006

and 96 days in 2007. Despite the smaller number of days

of sink strength in 2004, the net uptake during this

season was the highest. This is indicative of the high

potential of RCG as a bioenergy crop in fixing the

atmospheric C, given the high soil moisture regime

and proper atmospheric conditions (such as low to

moderate temperatures, low VPD and high PAR levels).

Peak daily rates of GPP were high during wet years

than during drier ones. The maximum daily GPP value

in 2004 was�15.8,�11.9 g C m�2 day�1 in 2007, whereas

the peak values were �8.5, �9.8 g C m�2 day�1 in 2005

and 2006, respectively (Fig. 10). The peak GPP occurred

during mid July in 2004, early July in 2005 and 2007 and

late June in 2006. Maximum daily ecosystem respiration

(TER) rates in 2004, 2005, 2006 and 2007 were 7.6, 7.1, 5.3

and 6.3 g C m�2 day�1, respectively (Fig. 10). Corre-

sponding peak NEE rates were �9.6, �4.5, �5.5 and

�7.5 g C m�2 day�1. Maximum NEE occurred during

mid July in 2004, early July in 2005 and 2007 and late

June in 2006. Flanagan et al. (2002) observed in their

study that the year with peak daytime NEE was also the

year with peak night-time NEE. However, we did not

find such consistent pattern in CO2 uptake and its loss.

While the peak daytime NEE was the highest in 2004,

the maximum night-time NEE was the highest in 2006.

This might be a typical characteristic of cultivations on

organic soils, because a lot of respiration might come

from old carbon, not related to recent assimilation.

For an insight into the seasonal exchange patterns, we

divided each calendar year into three parts – January–

April, May–September and October–December and

calculated the total GPP, TER and NEE for each of these

periods in each year (Table 3). The ecosystem accumu-

lated 398 g C m�2 during the period from the beginning

of January 2004 until the end of 2007. On an annual

basis, the ecosystem was a net sink for atmospheric CO2

in all years. During wet years, the ecosystem was a

strong sink and a weaker one during dry years. Most of

the variation in annual NEE values is a result of the

large variation in NEE during the May–September

period [with a coefficient of variation (CV) of nearly

53% over the 4 years]. The CV of NEE during January–

April and October–December periods were 20% and

24%, respectively. And among the May–September per-

iods, GPP was relatively more variable (CV of 14.4%)

than TER (CV of 10.5%), implying that GPP was the

stronger driver of the interannual variability in CO2

exchange in this ecosystem.

During the wettest year of 2004, there was a net gain

of 211 g C m�2. The ecosystem acted as a strong sink for

atmospheric CO2 during this year. Soil and atmospheric

conditions during this period were favourable for high

GPP and low TER resulting in a net C sink. The 2005

and 2006 seasons were drier than normal, the photo-

synthetic activity was restricted and therefore, they

were weaker sinks for carbon. The 2005 year accumu-

lated about 9 g C m�2, while the 2006 year had a net

uptake of 52 g C m�2. The total rainfall during the May–

September period in these 2 years was nearly the same

(about 245 mm). Nevertheless, the net uptake during

the 2006 season was relatively higher compared with

2005. Considering the poor crop performance during

the 2005 season, lime was added to the soil in the

beginning of the 2006 season to improve soil pH. This

may have resulted in better growth conditions for the

RCG plants. The soil moisture conditions were better

during the May–September of 2007 with above normal

rainfall. The GPP was the highest during the period

Table 3 Interannual variation in the seasonal and annual

sums of gross primary productivity (GPP), total ecosystem

respiration (TER) and net ecosystem CO2 exchange (NEE) of

reed canary grass on an organic soil in eastern Finland

Year Jan.–Apr. May–Sep. Oct.–Dec. Annual

GPP

2004 �7.5 �624.5 �4.6 �636.6

2005 �4.6 �496.8 �6.1 �507.5

2006 �3.5 �521.8 �6.0 �531.3

2007 �6.2 �672.0 �11.5 �689.7

TER

2004 34.5 349.3 34.1 417.9

2005 25.1 413.6 53.1 491.8

2006 16.6 405.9 46.4 468.9

2007 21.0 452.6 81.2 554.8

NEE

2004 26.9 �267.2 28.9 �211.4

2005 22.8 �79.0 47.5 �8.7

2006 17.4 �110.0 41.1 �51.5

2007 18.8 �215.1 69.7 �126.6

Values are in g C m�2 per averaging period.

Jan., January; Apr., April; Sep., September; Oct., October; Dec.,

December.



among all years. However, TER was also the highest

during this period. Higher rates of CO2 uptake were

offset by high rates of CO2 loss and therefore, this

period in 2007 accumulated less C than the 2004 season.

While TER is more or less similar during January–April

and October–December period over the years, TER

during the May–September period showed an increas-

ing trend from 2004 to 2007. RCG is cultivated in the

region as a perennial crop and therefore has the ten-

dency to develop increasing rootstocks over the succes-

sive cultivation years. Autotrophic respiration and root

respiration, in particular, seems to be a dominant com-

ponent in TER at our study. In view of this, increasing

TER trend from 2004 to 2007 is not unexpected.

To our knowledge, this is the first study that reports

interannual variations in C exchange from a perennial

bioenergy crop cultivation. Therefore, we are not aware

of any other studies with which we could compare our

observed C exchange rates. Nevertheless, we can com-

pare the values reported in this paper with C exchange

rates from grasslands, for example on organic and other

soil types in Finland and elsewhere. Lohila et al. (2004)

made EC measurements of CO2 exchange from barley

and perennial grass on an organic soil in southern

Finland and reported net annual CO2 losses of the order

of 210 and 79 g C m�2, respectively. Maljanen et al. (2004)

measured NEE using chambers in organic agricultural

soils in western Finland cultivated with forage grass

and barley with a fertilization rate of 100 kg N ha�1 (and

an additional application of farmyard manure for the

grassland site). They found that both sites were net

sources of CO2 to the atmosphere (830 and 330 g C m�2

ha�1). Such high net emissions on an annual basis as

opposed to net C uptake by RCG at our study site could

be attributed to high fertilizer input, regular soil tillage

and lower photosynthetic capacity of the vegetation at

their site. It is noteworthy that for a complete green-

house gas balance of the RCG cultivation system, not

only is the quantification of CO2 balance necessary but

also of the CH4 and N2O emissions.

Net biome productivity (NBP)

NBP in the context of this study refers to the net

ecosystem CO2 exchange after the CO2 losses due to

removal of the biomass from the ecosystem in harvest

have been accounted for. The RCG crop yield for the

2004 season (harvested in spring of 2005) was 3.7 tonne

ha�1. The yield values for the 2005, 2006 and 2007

seasons were 2.0, 3.6 and 4.7 tonne ha�1, respectively.

Accounting for an average moisture content of 17.5%

and carbon content of 45.8% (Vapo Ltd. Energy, 2003),

the corresponding seasonal biomass yields amount to

140, 76, 136 and 181 g C m�2. An annual NEE of �211 g

C m�2 in 2004 reduces to an NBP of �71 g C m�2 after

subtracting the biomass yield value of 140 g C m�2 from

the NEE value (Fig. 11). Similarly, NBP values for the

2005, 2006 and 2007 seasons were estimated to be 56, 85

and 31 g C m�2, respectively. It is interesting to note that

NBP in 2004 is negative. This implies that even after

accounting for the carbon in the harvested biomass,

there is still some amount of fixed C left in the ecosys-

tem and is sequestered in the soil during a wet year

such as 2004. NBP values are positive during 2005, 2006

and 2007 seasons. During such dry years, the ecosystem

becomes a source of carbon to the atmosphere after

accounting for the C losses due to biomass harvest.

Based on the NBP analysis, 2005 was a smaller source

compared with 2006. NEE was less negative during

2005 and the harvested biomass was also low during

this season. Despite a high net uptake during 2006, it

was a bigger C source because of the high biomass

yield.

Concluding remarks

On the one hand, global organic soils have been con-

sidered to be too risky to grow bioenergy crops. And on

the other, in the northern countries such as Sweden and

Finland, perennial grass species are recommended to be

cultivated on drained organic soils for biomass produc-

tion with clear advantages. Therefore, our objective in

initiating this pioneering study was to understand how

the cultivation of a perennial bioenergy crop such as

RCG impacts the net CO2 balance of a drained organic

soil (a cutover peatland site abandoned after peat

extraction). Based on the four years of continuous CO2

balance measurements, we conclude that during wet

years, the RCG cultivation system acts as a distinct sink

Year2004 2004 2005 2005 2006 2006 2007 2007 2008

C e

xcha

nge

(kg

ha–1

a–1

)

–6000

–4000

–2000

0

2000

4000

GPPTERNEEYieldNBP

Fig. 11 Interannual variation in the annual carbon uptake

(GPP), loss (TER), net ecosystem CO2 exchange, economic yield

and net biome productivity (NBP) of reed canary grass cultiva-

ted on an organic soil in eastern Finland during 2004–2007. GPP,

gross primary productivity; TER, total ecosystem respiration.



for atmospheric C. During dry years, however, the

ecosystem acts as a weaker sink bordering the sink-

source line (considering a general uncertainty level of

� 50 g C m�2 in EC measurements – Baldocchi, 2003).

We also found that the net carbon balance after account-

ing for the harvested biomass was still a net carbon sink

during wet years, implying that a part of the captured

carbon is left over in the ecosystem. Thus, the RCG

cultivation was more than a ‘carbon-neutral’ system.

According to the 2007 IPCC report, regions in the high

latitudes have been projected to receive over 20% higher

annual precipitation and therefore, a higher frequency

of wet years under the changed climate (Meehl et al.,

2007). Therefore, the capacity of the RCG cultivation to

act as a carbon sink would increase in future climatic

conditions.

The results from this study have positive implications

for the use of RCG as a bioenergy crop on organic soils.

The results based on the net CO2 balance alone provide

a strong evidence to suggest that the cultivation of RCG

on such problematic soils is a promising land use

option. The annual NEE of the peat soil growing RCG

during the entire study period ranged from �9 to �211

g Cm�2yr�1. Furthermore, during wet years, the eco-

system was still a C sink even after accounting for the

harvested biomass C. These results highlight for the

first time that a perennial bioenergy crop such as reed

canary grass can be cultivated successfully on an or-

ganic soil to mitigate CO2 loss.

Acknowledgements

We dedicate this article to the fond memory of Jari T. Huttu-nen (3 April 1970–28 December 2008) who was not only acolleague but also a great friend. He will be ever rememberedfor his keen sense of observation, attention to detail and a greatsense of humour. This work was made possible by the fundingfrom the Finnish Funding Agency for Technology and Innova-tion (Tekes) and Eastern Finland University. Excellent supportfrom Alpo Hassinen and Matti Lemettinen of Mekrijarvi Re-search Station, Ilomantsi, eastern Finland is highly appreciated.We thank Jorma Nuutinen and many other field staff for theirhelp in field data collection. Finally, we express our thanks to thetwo anonymous reviewers and editor for an expeditious reviewprocess.

References

Baldocchi DD (2003) Assessing ecosystem carbon balance: pro-

blems and prospects of the eddy covariance technique. Global

Change Biology, 9, 479–492.

Burvall J (1997) Influence of harvest time and soil type on fuel

quality in reed canary grass (Phalaris arundinacea L.). Biomass

and Bioenergy, 12, 149–154.

Coops H, van den Brink FWB, van der Velde G (1996) Growth

and morphological responses of four helophyte species in an

experimental water-depth gradient. Aquatic Botany, 54, 11–24.

Drebs A, Nordlund A, Karlsson P, Helminen J, Rissanen P (2002)

Climatological Statistics of Finland 1971–2000 (pp. 1–95. Meteor-

ological Research and Basic Weather Services, Finnish Meteor-

ological Institute, Helsinki.

Finell M, Nilsson C (2005) Variations in ash content, pulp yield,

and fibre properties of reed canarygrass. Industrial Crops and

Products, 22, 157–167.

Flanagan LB, Wever LA, Carson PJ (2002) Seasonal and inter-

annual variation in carbon dioxide exchange and carbon

balance in a northern temperate grassland. Global Change

Biology, 8, 599–615.

IPCC (2007) Climate change 2007: the physical science basis. In:

Contribution of Working Group I to the Fourth Assessment Report

of the Intergovernmental Panel on Climate Change (eds Solomon S,

Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M,

Miller HL), Cambridge University Press, Cambridge, UK,

pp. 1–996.

Janssens IA, Lankreijer H, Matteucci G et al. (2002) Productivity

overshadows temperature in determining soil and ecosystem

respiration across European forests. Global Change Biology, 7,

269–278.

Janssens IA, Pilegaard K (2003) Large seasonal changes in Q10 of

soil respiration in a beech forest. Global Change Biology, 9, 911–

918.

Kasimir-Klemedtsson A, Klemedtsson L, Berglund K, Martikai-

nen PJ, Silvola J, Oenema O (1997) Greenhouse gas emissions

from farmed organic soils: a review. Soil Use and Management,

13, 245–250.

Kellogg CH, Bridgham SD, Leicht SA (2003) Effects of water

level, shade and time on germination and growth of fresh-

water marsh plants along a simulated successional gradient.

Journal of Ecology, 91, 274–282.

Kirschbaum MUF (1996) The temperature dependence of soil

organic matter decomposition, and the effect of global warm-

ing on soil organic C storage. Soil Biology and Biochemistry, 27,

753–760.

Landstrom S, Lomakka L, Andersson S (1996) Harvest in spring

improves yield and quality of reed canary grass as a bioenergy

crop. Biomass and Bioenergy, 11, 333–341.

Larson ED (2006) A review of life-cycle analysis studies on liquid

biofuel systems for the transport sector. Energy for Sustainable

Development, 10, 109–126.

Lewandowski I, Scurlock JMO, Lindvall E, Christou M (2003)

The development and current status of perennial rhizomatous

grasses as energy crops in the US and Europe. Biomass and

Bioenergy, 25, 335–361.

Lindroth A, Klemedtsson A, Grelle A, Weslien P, Langvall O

(2008) Measurement of net ecosystem exchange, productivity

and respiration in three spruce forests in Sweden shows

unexpectedly large soil carbon losses. Biogeochemistry, 89,

43–60.

Lloyd J, Taylor JA (1994) On the temperature dependence of soil

respiration. Functional Ecology, 8, 315–323.

Lohila A, Aurela M, Tuovinen JP, Laurila T (2004) Annual CO2

exchange of a peat field growing spring barleys or perennial



forage grass. Journal of Geophysical Research, 109, D18116,

doi: 10.1029/2004JD004715.

Luyssaert S, Janssens IA, Sulkava M et al. (2007) Photosynthesis

drives anomalies in net carbon-exchange of pine forests at

different latitudes. Global Change Biology, 13, 2110–2127.

Makiranta P, Hytonen J, Aro L et al. (2007) Soil greenhouse gas

emissions from afforested organic soil croplands and cutaway

peatlands. Boreal Environmental Research, 12, 159–175.

Maljanen M, Komulainen V-M, Hytonen J, Martikainen PJ, Laine

J (2004) Carbon dioxide, nitrous oxide and methane dynamics

in boreal organic agricultural soils with different soil manage-

ment. Soil Biology and Biochemistry, 36, 1801–1808.

Maljanen M, Liikanen A, Silvola J, Martikainen PJ (2003) Mea-

suring N2O emissions from organic soils with closed chamber

or soil/snow N2O gradient methods. European Journal of Soil

Science, 54, 625–631.

Maljanen M, Martikainen PJ, Aaltonen H, Silvola J (2002) Short-

term variation in fluxes of carbon dioxide, nitrous oxide and

methane in cultivated and forested organic boreal soils. Soil

Biology and Biochemistry, 34, 577–584.

Maljanen M, Martikainen PJ, Walden J, Silvola J (2001) CO2

exchange in an organic field growing barley or grass in eastern

Finland. Global Change Biology, 7, 679–692.

Mauder M, Foken T, Bernhofer C et al. (2007) Quality control of

CarboEurope-IP flux data. Part II: intercomparison of eddy-

covariance software. Biogeosciences Discussion, 4, 4067–4099.

Meehl GA, Stocker TF, Collins WD et al. (2007) Global climate

projections. In: Climate Change 2007: The Physical Science Basis.

Contribution of Working Group I to the Fourth Assessment Report

of the Intergovernmental Panel on Climate Change (eds Solomon S,

Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M,

Miller HL), Cambridge University Press, Cambridge, United

Kingdom and New York, NY, USA.

Minkkinen K, Korhonen R, Savolainen T, Laine J (2002) Carbon

balance and radiative forcing of Finnish peatlands 1900–2100 – the

impact of forestry drainage. Global Change Biology, 8, 785–799.

Moffat AM, Papale D, Reichstein M et al. (2007) Comprehensive

comparison of gap filling techniques for eddy covariance net

carbon fluxes. Agricultural and Forest Meteorology, 147, 209–232.

Moore CJ (1986) Frequency response corrections for eddy corre-

lation systems. Boundary-Layer Meteorology, 37, 17–35.

Nieveen JP, Campbell DI, Schipper LA, Blair IJ (2005) Carbon

exchange of grazed pasture on a drained peat soil. Global

Change Biology, 11, 607–618.

Nykanen H, Alm J, Lang K, Silvola J, Martikainen PJ (1995) Emis-

sions of CH4, N2O and CO2 from a virgin fen and a fen drained for

grassland in Finland. Journal of Biogeography, 22, 351–357.

OECD (2007) Biofuels: Is the Cure Worse than the Disease? OECD

Sustainable Development Studies. OECD Publications, Paris.

Rayment MB, Jarvis PG (2000) Temporal and spatial variation of

soil CO2 efflux in a Canadian boreal forest. Soil Biology and

Biochemistry, 32, 35–45.

Reichstein M, Falge E, Baldocchi D et al. (2005) On the separation

of net ecosystem exchange into assimilation and ecosystem

respiration: review and improved algorithm. Global Change

Biology, 11, 1–16.

Sahramaa M (2004) Evaluating germplasm of reed canarygrass,

Phalaris arundinacea L. Unpublished PhD thesis, University of

Helsinki, Helsinki, 47 pp.

Silvola J, Alm J, Ahlholm U, Nykanen H, Martikainen PJ (1996)

The contribution of plant roots to CO2 fluxes from organic

soils. Biology and Fertility of Soils, 23, 126–131.

Sundh I, Nilsson M, Mikkela C, Granberg G, Svensson BH (2000)

Fluxes of methane and carbon dioxide on peat-mining areas in

Sweden. Ambio, 29, 499–503.

Turunen J, Tomppo E, Tolonen K, Reinikainen A (2002) Estimat-

ing carbon accumulation rates of undrained mires in Finland:

application to boreal and subarctic regions. Holocene, 12, 69–80.

Vapo Ltd. Energy (2003) Monitoring programme for the use, treat-

ment and leaching of elements from peat extraction areas of the

North Karelian Regional Environment Centre, Vapo Ltd. Energy,

Eastern Finland, 19 pp. (in Finnish).

Venendaal R, J�rgensen U, Foster CA (1997) European energy

crops: a synthesis. Biomass and Bioenergy, 13, 147–185.

Verma SB, Dobermann A, Cassman KG et al. (2005) Annual

carbon dioxide exchange in irrigated and rainfed maize-based

agroecosystems. Agricultural and Forest Meteorology, 131,

77–96.

Waddington JM, Warner KD, Kennedy GW (2002) Cutover peat-

lands: a persistent source of atmospheric CO2. Global Biogeo-

chemical Cycles, 16, doi: 10.1029/2001GB001398.

Wang H, Saigusa N, Yamamoto S, Kondo H, Horano T, Toriyamo

A, Fujinuma Y (2004) Net ecosystem CO2 exchange over a

larch forest in Hokkaido, Japan. Atmospheric Environment, 38,

7021–7032.

Webb EK, Pearman GI, Leuning R (1980) Correction of flux

measurements for density effects due to heat and water

vapour transfer. Quarterly Journal of Royal Meteorological

Society, 106, 85–100.



Cultivation of a perennial grass for bioenergy on a boreal organic soil - carbon sink or source

Documents