Does the temperature sensitivity of decomposition of soilorganic matter depend upon water content, soil horizon,or incubation time?
M A R K U S R E I C H S T E I N 1 , J E N S - A R N E S U B K E 2 , A N D R E W C . A N G E L I 3 and
J O H N D . T E N H U N E N 1
Department of Plant Ecology, University of Bayreuth, D-95440 Bayreuth, Germany
Abstract
Several studies have shown multiple confounding factors influencing soil respiration in
the field, which often hampers a correct separation and interpretation of the different
environmental effects on respiration. Here, we present a controlled laboratory
experiment on undisturbed organic and mineral soil cores separating the effects of
temperature, drying–rewetting and decomposition dynamics on soil respiration.
Specifically, we address the following questions:
(1) Is the temperature sensitivity of soil respiration (Q10) dependent on soil moisture or
soil organic matter age (incubation time) and does it differ for organic and mineral
soil as suggested by recent field studies.
(2) How much do organic and mineral soil layers contribute to total soil respiration?
(3) Is there potential to improve soil flux models of soil introducing a multilayer source
model for soil respiration?
Eight organic soil and eight mineral soil cores were taken from a Norway spruce (Piceaabies) stand in southern Germany, and incubated for 90 days in a climate chamber with a
diurnal temperature regime between 7 and 23 1C. Half of the samples were rewetted
daily, while the other half were left to dry and rewetted thereafter. Soil respiration was
measured with a continuously operating open dynamic soil respiration chamber system.
The Q10 was stable at around 2.7, independent of soil horizon and incubation time,
decreasing only slightly when the soil dried. We suggest that recent findings of the Q10
dependency on several factors are emergent properties at the ecosystem level, that
should be analysed further e.g. with regard to rhizosphere effects. Most of the soil CO2
efflux was released from the organic samples. Initially, it averaged 4.0 lmol m�2 s�1 and
declined to 1.8 lmol m�2 s�1 at the end of the experiment. In terms of the third question,
we show that models using only one temperature as predictor of soil respiration fail to
explain more than 80% of the diurnal variability, are biased with a hysteresis effect, and
slightly underestimate the temperature sensitivity of respiration. In contrast, consis-
tently more than 95% of the diurnal variability is explained by a dual-source model,
with one CO2 source related to the surface temperature and another CO2 source related to
the central temperature, highlighting the role of soil surface processes for ecosystem
carbon balances.
Keywords: drying–rewetting, dual-source model, incubation experiment, Q10 , soil moisture, soil
respiration
Received 22 July 2004; received in revised form 4 March 2005; accepted 9 March 2005
1Present address: Department of Forest Environment and Resources, University of Tuscia, Via S. Camillo de Lellis 01100 Viterbo, Italy and
Potsdam Institute of Climate Impact Research, Telegrafenberg C4, D-14473 Potsdam, Germany.
Correspondence: Markus Reichstein, [email protected] and Jens-Arne Subke, e-mail: [email protected]
2Present address: Stockholm Environment Institute at York, Department of Biology, University of York, York, YO10 5DD, UK.3Present address: University of Wisconsin-Madison, Madison, WI 53706, USA.
Global Change Biology (2005) 11, 1754–1767, doi: 10.1111/j.1365-2486.2005.01010.x
1754 r 2005 Blackwell Publishing Ltd
Introduction
Considering that terrestrial ecosystems take up about
one-third of the CO2 emissions from anthropogenic
fossil fuel burning and cement manufacturing (Schimel
et al., 2001), it is critical to improve our knowledge and
understanding of the carbon exchange between terres-
trial ecosystems and the atmosphere. At 68–80 Pg C yr�1,
soil respiration represents the second largest global
carbon flux between ecosystems and the atmosphere
(Raich & Schlesinger, 1992; Raich & Potter, 1995; Raich
et al., 2002). This amount is more than 10 times the
current rate of fossil fuel combustion and indicates that
each year around 10% of the atmosphere’s CO2 cycles
through the soil. Thus, even a small change in soil
respiration could significantly intensify – or mitigate –
current atmospheric increases of CO2, with potential
feedbacks to climate change. Despite this global
significance as well as considerable scientific commit-
ment to its study over the last decades, there is still
only limited understanding of the factors controll-
ing temporal and interecosystem variability of soil
respiration.
It is clear that the most important factors influencing
soil respiration are soil temperature, soil water avail-
ability, and substrate quality and availability, the latter
being influenced by supply strength through vegetation
(photosynthate transport, root growth and exudation).
There are, however, ambiguous results showing the
extent to which soil respiration factors interact. This
lack of understanding partly stems from the fact that
many studies were performed under uncontrolled field
conditions where several co-varying factors are difficult
to isolate. Controlled laboratory experiments, on the
other hand, have often been performed on disturbed
soil samples (mixed and/or sieved, roots removed),
substantially altering the environment such that a
transfer of results to the ecosystem level seems dubious
(e.g. Winkler et al., 1996; Lomander et al., 1998; Reich-
stein et al., 2000). Moreover, soils are often incubated at
different and constant temperatures, which introduces
the confounding effect that with time substrate avail-
ability (and potentially microbial community) changes
among samples, as labile pool sizes decline more
rapidly in the higher temperature treatment (see
Reichstein et al., 2000 for a detailed discussion). For
example, Fang & Moncrieff (2001) form an exception in
that they tested the temperature dependence on
undisturbed soil samples, and exposed all samples to
identical temperature regimes. However, their investi-
gation into interdependencies with soil moisture only
considered broad moisture classes with ‘relatively dry’
as the driest of three categories, and accordingly no
moisture effects were found.
In the context of global warming, it is particularly
important to understand the temperature sensitivity of
soil respiration, as it is anticipated that in a warmer
world ecosystems provide a positive feedback to the
greenhouse effect because of the stronger response of
respiratory processes to temperature, compared with
assimilatory processes, which are easily limited by light
and nutrient supply (e.g. Kirschbaum, 2000). Recently
developed coupled climate-vegetation models predict
that the current biospheric CO2 sink will cease during
this century, partly because of enhanced respiration in a
warmer climate (Cox et al., 2000), assuming an
exponential relationship between temperature and
respiratory processes with a doubling per 10 1C (Q10
relationship). On the contrary, a number of field studies
at varying scales have cast some doubt on the
correctness of such an invariable Q10 relationship. For
instance, Giardina & Ryan (2000) did not find a clear
trend of turnover rates with mean annual temperature,
Liski et al. (1999) observed decomposition rates of old
organic matter to be temperature insensitive, and a
number of studies found a varying temperature
sensitivity of soil and ecosystem respiration (Q10
between 1 and 4), depending on the soil water
availability (e.g. Carlyle & Ba Than, 1988; Xu & Qi,
2001; Reichstein et al., 2002, 2003).
To better understand this critical issue, we continu-
ously observed CO2 efflux in a controlled experiment
from minimally disturbed forest soil monoliths. We
specifically addressed the following questions concern-
ing the temperature sensitivity of soil respiration. (1) Is
the temperature sensitivity of soil respiration (Q10)
dependent on soil moisture, drying/rewetting events,
or soil organic matter (SOM) age? (2) How much do
organic and mineral soil layers contribute to total
soil respiration and do their temperature sensitivities
differ? (3) How do estimates of the temperature
sensitivity change when fluxes from two layers are
accounted for compared with single-layer models?
Materials and methods
Soil sampling
Intact soil monoliths, each from the organic and the
upper mineral horizon, were sampled randomly in a
mature spruce forest (‘Weidenbrunnen’, Picea abies; 114
years old) in the Fichtelgebirge, a small mountain range
in SE Germany in April 2001. The local soil type is a
cambic podzol over granitic bedrock with low pH
values in the mineral (2.9–4.3) and organic (2.6–3.6)
layers. The average depth of the organic layer (includ-
ing surface litter) was 11 � 1 cm, and forest ground
vegetation at the sampling locations was dominated by
T E M P E R AT U R E S E N S I T I V I T Y O F D E C O M P O S I T I O N 1755
r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 1754–1767
the two grass species Deschampsia flexuosa and Calama-
grostis villosa. The main properties of the soil horizons
are described in Table 1. Field measurements of soil
CO2 efflux have been carried out during the growing
season in 1999. A more complete site description, as
well as, analytical results and flux observations are
reported in Subke et al. (2003).
The sampling took place 2 days after a period of
considerable rainfall, implying that the soils were near
field capacity. Cylindrical soil cores were taken of the
organic layer using a custom built steel cutter of 30 cm
diameter. After inserting the cutter between 15 and
20 cm into the forest floor, the monolith was carefully
retrieved from the cutter (having removed remaining
mineral soil from the base of the sample), and
transferred to custom-built polyvinyl chloride (PVC)
containers. Mineral soil samples were taken after
removing the organic horizons by inserting the cutter
into the mineral soil to a similar depth to the organic
soil samples in close proximity to the location where the
organic layer sample was taken. The strong build of the
cutter allowed it to be forced into the soil using a
sledgehammer. It was, therefore, readily possible to cut
through roots of up to approximately 2 cm in diameter,
while sampling attempts were aborted if larger roots
were encountered, as it would have resulted in
considerable sample disturbance to force the cutter
under these conditions. Mean sample volumes of both
mineral and organic soil were 8.1 � 1.6 dm3 with no
significant difference between mean volumes of the two
soil layers. The properties of the soil samples harvested
after the incubation are shown in Table 1.
The cylindrical sample containers were built from
large PVC pipes (30 cm diameter, 1 cm wall thickness),
and measured 25 cm in height (Fig. 1a). The base of the
containers was formed by a 1 cm PVC disk, which was
perforated with about 20 holes of 1 cm diameter
allowing excess water to drain during the course of
Table 1 Description of the soil profile and the soil samples
Object
Thickness
(cm)
Total organic
C (g kg�1)
C/N ratio
(g g�1)
Texture
(sand/silt/clay)
(g 100 g�1)
Bulk density
(g cm�3)
pH
(CaCl2)
Soil horizons*
L 1 478 24.8 nd nd 3.6
Of 5 372 20.7 nd nd 2.9
Oh 7 376 22.6 nd nd 2.6
Ahe 10 38.9 22.9 52/38/10 0.97 2.9
Bh 2 90.5 22.6 34/50/16 nd 3.3
Bhs 18 53.6 14.1 45/45/10 0.73 3.9
Bv-Cv 25 8.4 16.8 46/43/11 1.36 4.3
Incubated samplesw
Organic
0–2 cm 2 488 (34) 27.7 nd 0.11 (0.02) nd
2–6 cm 4 381 (23) 22.3 nd 0.41 (0.12) nd
6–12 cm 6 371 (21) 21.7 nd 0.52 (0.11) nd
Mineral 12 45.2 (4.2) 22.6 nd 0.92 (0.21) nd
Values in parentheses represent one standard deviation (n 5 8).
*(Subke et al., 2004a), (Kalbitz, 2004, neighbouring stand with same soil type).wThis study.
nd, not determined.
25 cm
30 cm
Sample container (PVC)
Chamber lid adapter (PVC)
Sample airReference air
Ambient air inlet
Chamber lid (Perspex)
to IRGA
Fig. 1 Cross section and dimensions of soil sample containers.
The base of the containers was perforated and covered with a
1 mm mesh gauze to allow water drainage. The containers were
placed on PVC saucers to ensure a gas seal around the base. All
containers were fitted with adapters to accommodate the
difference in size between the container and the Perspex
chamber lids (20 cm diameter).
1756 M . R E I C H S T E I N et al.
r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 1754–1767
the experiment, preventing water logging. The perfo-
rated base was covered with 1 mm mesh nylon gauze to
contain soil samples. The container base was raised by
approximately 3 cm above the lower end of the cylinder
to provide drainage space for excess water, and
containers were placed in PVC saucers in the lab to
create a gas seal.
Laboratory set-up, soil moisture and temperaturetreatments
After about 4 h of sampling in the field, the soil cores
were immediately taken to a climate chamber (1.5 h
transport). The eight samples of each layer were
randomly split into two different water treatments,
one with a constant moisture treatment and one with a
drying–rewetting treatment, resulting in four types of
soil cores with four replicates each: Organic soil/
constant moisture (OC), organic soil/dry (OD), mineral
soil/constant moisture (MC), and mineral soil, dry
(MD). For conciseness, the ‘constant moisture’ and the
‘drying–rewetting’ treatment are called wet and dry
treatment, respectively. The samples were set up within
the climate chamber in a randomized design to
eliminate possible (undetected) gradients in tempera-
ture or air circulation within the chamber. Above-
ground plant parts of the ground vegetation were
removed at the beginning of the experiment, as was any
regrowth during the course of the experiment.
All soil samples were weighed upon arrival in the
lab, and the recorded mass was considered as reference
conditions for soil moisture, under which no limitations
of soil CO2 production exist. Over an initial period of 10
days, the mass of all monoliths was recorded daily.
Based on the loss of mass in 24 h, water was added to
each of the samples in the wet treatments by spraying
the given amount of water onto the surface of each
sample. After this initial period of 10 days, the wet
treatments were sprayed with water every day accord-
ing to the average loss previously experienced and the
mass of all containers was controlled every 3–4 days.
The samples of the dry treatments were watered to
initial water status within 24 h towards the end of the
experiment (OD samples on day 76, MD samples on
day 86). At the end of the experiment, the dry weight
and the volume of the samples was determined, so that
volumetric soil moisture could be calculated from the
recorded masses. To allow for comparison between the
different soil horizons, these water contents were
converted into soil water matric potential using Van
Genuchten parameters (Van Genuchten, 1980). The
parameters were optimized to describe the observed
field capacity and permanent wilting point of the
retention curve well and were previously used success-
fully to model the soil water dynamics at the site (Subke
et al., 2003).
Air temperature within the climate chamber cycled
between 7 and 23 1C every day, while the dew point
within each chamber was kept constant at 4 1C.
Following a ‘cold’ period, the temperature was set to
23 1C for 6 h, before moderation to 20 1C for another 6 h.
Similarly, temperatures were reduced drastically to 7 1C
following the warm period, before moderation to 10 1C.
As one part of this study involved the analysis of the
soil respiration dynamics in relation to vertical tem-
perature gradients (with the dual-source model) we
executed a prior experiment to understand if the
temperature conduction in our set up was mainly in
the vertical direction as in field conditions. Here, we
installed each five temperature sensors 2 cm below the
surface, at the centre and 2 cm above the bottom of the
soil core in a radial symmetric manner in the form of a
cross (each one sensor at the center and four sensors at
2 cm from the container wall). This prior experiment
showed that radial temperature gradients were negli-
gible, an order of magnitude lower than vertical
gradients (Fig. 2). We explain this quasiradial insulation
of the soil core as a combination of an insulation effect
of the containers, and much stronger convective heat
transport at the surface of the core because of
continually circulating air in the headspace of the open
dynamic chamber and a self-insulation effect, as the
containers were placed densely next to each other. The
results show further that changes in temperature are
attenuated within the first few centimetres of a
monolith. Therefore, we regard the central temperature
as a reasonable approximation of the bulk monolith
temperature, while the temperature of the surface layer
−10−8−6−4−202468
10
0 5 10 15 20 25 30Time (h)
Radial difference: 2 cm–central
Vertical difference: 2 cm–central
Vertical difference: surface–central
Tem
pera
ture
(°C
)
Fig. 2 Analysis of vertical and radial temperature differences
in an organic soil core during the pre-experiment. Time series of
temperature differences between the central temperature and
the surface temperature as well as between central temperature
and the average of temperatures 2 cm below the surface or 2 cm
inside the container (at the depth of the central temperature),
respectively, are shown.
T E M P E R AT U R E S E N S I T I V I T Y O F D E C O M P O S I T I O N 1757
r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 1754–1767
is best described by direct measurements. For the CO2
efflux experiments, temperature probes were hence
installed at the centre of each of the soil monoliths, and
each of the five chamber lids used for soil CO2 efflux
also carried a temperature probe that made contact
with the soil surface of the monolith where efflux
measurements were conducted.
CO2 efflux measurements
A soil CO2 efflux system designed for continuous
measurements from soil chambers in the field (Subke
et al., 2003) was used to monitor CO2 evolution from the
monoliths. A simple PVC collar allowed the field
chamber lids to fit onto the monolith container ensuring
air-tight seals at all connecting points (Fig. 1). The
measuring system allows sequential CO2 flux measure-
ments from five collars and an additional intake for
zero calibration, with sampling intervals set to 5 min
between collars, so that two readings were obtained per
collar per hour. The lids were moved between soil
containers every day at about the same time (between
13:00 and 14:00 hours), in a way that at any time, each
soil treatment was included in the measuring cycle.
Upon restoring the initial water status of the samples in
the dry treatments, the allocation of lids was altered
with four lids monitoring only rewatered samples, for
10 days, while one lid continued to rotate between
other treatments.
Field measurements
Parallel to the lab incubation (May–June), we measured
soil CO2 efflux in the field at the same forest stand
where soil samples had been taken. The field set-up
was the same as described in detail for measurements
in 1999 in the same forest stand (Subke et al., 2003).
Briefly, 15 soil collars were installed in five groups of
three collars. These five groups were located within the
same part of the forest stand where samples had been
taken, but a minimum distance of 2 m was kept
between sampling sites and collar locations. Soil CO2
efflux was measured sequentially from one of five
chamber lids which were placed on one collar in each
group, forming an open dynamic soil CO2 efflux
chamber. The automated set-up produced hourly efflux
measurements from each of the lids. Lids were moved
between collars in each group every 2–3 days to prevent
measuring artefacts from prolonged measuring peri-
ods. Hourly averages of measurements from all collars
were used to estimate the spatial mean of the stand soil
CO2 efflux.
Data analysis
Each diurnal cycle of soil CO2 efflux was analysed by
two different models relating CO2 efflux to soil temper-
ature. Model I was the commonly applied approach
where soil respiration is modelled as a function of soil
temperature at a certain location in the soil:
Rsoil ¼ Rref QðTsoil�TrefÞ=10�C10 : ð1Þ
Rsoil is the instantaneous soil CO2 efflux (mmol m�2 s�1),
Tsoil is the soil temperature ( 1C) at the centre of the soil
core, Rref is the reference efflux (mmol m�2 s�1) at
Tref 5 15 1C, and Q10 the parameter that determines
the temperature response of the soil CO2 efflux (i.e. the
factor by which efflux increases for an increase in
temperature by 10 1C). We also analysed the data with
the model by Lloyd & Taylor (1994), but results were
indistinguishable within the temperature range of this
study, in agreement with (Katterer et al., 1998). The
Lloyd-and-Taylor and the Q10 models only differ
substantially, when larger temperature ranges are
considered. We present here the results from the Q10
model for the intuitiveness of the Q10 parameter and
because a majority of ecosystem models are (still) using
this function (cf. reviews by Rodrigo et al., 1997; Cramer
et al., 2001). For a given temperature, the Q10 and E0 (the
temperature sensitivity parameter of the Lloyd-and-
Taylor model) can be converted into each other (see
Reichstein et al. (2003) for details).
While still very simplified, model II partly accounts
for the fact that the soil temperature distribution cannot
be represented by a single temperature measurement. It
is assumed that the total soil respiration can be
attributed to two major sources, a surface layer and
the bulk soil (Fig. 2), expressed as:
Rsoil ¼ Rsurf;ref QðTsurf�TrefÞ=10�C10
þ Rcent;ref QðTcent�TrefÞ=10�C10 ; ð2Þ
where symbols are as in the previous equation, but the
subscripts ‘surf’ and ‘cent’ mean ‘of the surface
compartment’ and ‘of the bulk soil compartment’,
characterized by the temperature at the soil surface
and at the centre of soil core, respectively. For clarity,
Tcent in model II and Tsoil in model I, are physically the
same; we only differentiate for conceptual reasons.
Mathematically, model II represents a rough discre-
tization of the volume integral
RsoilðtÞ ¼1
A
Z Z Z
x;y;z
Rðx; y; z; tÞdx dy dz; ð3Þ
where R(x, y, z, t) is the respiration density (per volume;
mmol m�3 s�1) at the specified location and time, and A
is the area over which the flux is observed, neglecting
physicochemical processes (diffusion, solution, etc.).
1758 M . R E I C H S T E I N et al.
r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 1754–1767
Apart from the discretization another simplification of
the model formulation is that the Q10 is assumed to be
the same in both soil compartments. This model
described the data equally as well as more complex
formulations (e.g. with different Q10 by compartment,
and/or more layers), which were found to be over-
parameterized (high parameter correlations, unidentifi-
able parameters). In the following, models I and II will
be referred to as ‘single-source’ and ‘dual-source’
models, respectively.
The model parameters were estimated by a common
nonlinear regression algorithm using the least-sum-of-
residual-squares criterion (Levenberg–Marquardt algo-
rithm, implemented in the PV-WAVE 7.0 advantage
package, function nlinlsq, Visual Numerics Inc., 2001).
Standard errors of parameter estimates were calculated
according to Draper & Smith (1981), using standard
assumptions (e.g. normality and independence of the
residuals).
Results
Soil temperature and moisture
While the surface temperature responded fastest to
changes in air temperature, the temperature at the
centre of a sample never reached the value of the air
temperature, even at the end of respective ‘cold’ and
‘warm’ periods (Fig. 3). Soil water availability de-
creased in all samples of the dry treatment, as is shown
by the decrease in matric potential throughout the
experiment (Fig. 4). As water was primarily lost via
passive evaporation (no water extraction by transpira-
tion) the samples did not dry down near the permanent
wilting point but rather to matric potentials around
�1000 to �2000 hPa, even after 80 days in relatively dry
air (dew point 4 1C).
Soil CO2 efflux in relation to temperature
Soil CO2 efflux was positively related to central
soil temperature, but with a clear hysteresis effect
(Fig. 5a–c), (i.e. at the same central soil temperature the
observed soil efflux is higher during the warming
phase than during the cooling phase). This occurs
because the central temperature lags behind the
temperature at a more superficial layer (cf. Fig. 3).
Consequently, the single-source model typically leaves
10–30% of the efflux variance unexplained (Fig. 5d). In
contrast, the dual-source model, which used Tsurf and
Tcent as predictors, consistently explained more than
95% of the variance in the organic samples leaving
less than 5% unexplained (cf. also Table 2). Also in
the mineral soils the dual-source model worked
−3−2−10123456
133 134 135 136 137 138 139 140 141Julian day
Rso
il (µ
mol
m−2
s−1
)
5
10
15
20
25
30
35
Tem
pera
ture
(°C
)
RsoilTsurfTcent
Fig. 3 Time series for soil CO2 efflux (open circles, left axis), as
well as surface and central temperature (diamonds and crosses,
right axis) of one of the organic wet samples.
−2500
−2000
−1500
−1000
−500
0
100 120 140 160 180 200 220 240Julian day
So
il w
ater
mat
ric
po
ten
tial
(h
Pa)
MC
OC
MD
OD
Fig. 4 Average soil water matric potential during the drying–rewetting experiment for each treatment. Error bars indicate one standard
deviation. OC, OD, organic constantly wet, organic drying/rewetting; MC, MD, mineral constantly wet, mineral drying/rewetting.
T E M P E R AT U R E S E N S I T I V I T Y O F D E C O M P O S I T I O N 1759
r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 1754–1767
5 10 15 20 250
2
4
Rso
il (µ
mo
l m−2
s−1
)
Rso
il (µ
mo
l m−2
s−1
)
Rso
il (µ
mo
l m−2
s−1
)
25 120 140 160 180 200Julian day
Dual-source modelSingle-source model
Tsurf (°C)Tcent (°C) 5
1015
202525
2015
105
4
2
6
50
2
4
6
0.5
0.6
0.7
0.8
0.9
1
r2
10
8
6
(a) (c)
(b)
(d)
20Tsurf (°C)
Tcent (°C)
10 15
Fig. 5 Scatter plots of soil CO2 efflux vs. (a) central soil temperature and (b) soil surface temperature for the measurement cycles of
Fig. 3. Labels indicate the integer hour of the respective day. (c) Soil CO2 efflux as a function of surface and central temperature. The
mesh shows the prediction from the dual-source model (fitted to these 3 days of data), data points (label as in a, b) are observed
data (Rsoil, ref 5 1.22 � 0.05 mmol m�2 s�1; Rcent, ref 5 1.03 � 0.04 mmol m�2 s�1; Q10 5 2.72 � 0.06; r2 5 0.96, RMSE 5 0.25). Arrows in a–c
underline the trajectory with time showing the hysteresis effect. (d) Time series of coefficients of determination obtained with the dual-
and the single-source model fitted to each diurnal cycle for one replicate of the organic wet treatment (OC). RMSE, root mean
square error.
Table 2 Average coefficient of determination and NRMSE when modelling the diurnal courses of soil chamber CO2 efflux using
the single- or dual-source model for each treatment at the beginning and at the end of the experiment as well after the rewetting
OC OD MC MD Mean
Single Dual Single Dual Single Dual Single Dual Single Dual
Coefficient of determination (r2)
Begin 0.65 0.97 0.80 0.96 0.51 0.77 0.43 0.63 0.60 0.83
End 0.83 0.98 0.84 0.98 0.73 0.91 0.29 0.66 0.67 0.88
Rewet na na 0.76 0.88 na na 0.67 0.83 0.71 0.85
NRMSE
Begin 0.124 0.048 0.070 0.030 0.124 0.082 0.121 0.070 0.110 0.057
End 0.051 0.017 0.062 0.031 0.079 0.040 0.106 0.064 0.074 0.038
Rewet na na 0.088 0.079 na na 0.195 0.032 0.142 0.055
OC, OD: organic constantly wet, organic drying/rewetting; MC, MD: mineral constantly wet, mineral drying/rewetting; NRMSE,
normalized root mean squared error. The column ‘mean’ contains the mean quantity over all treatments.
1760 M . R E I C H S T E I N et al.
r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 1754–1767
considerably better explaining around 20% more of the
variance than the single-source model (Table 2). In both
soil types, the dual-source model also roughly halved
the model error compared with the single-source model
to below 10% in the mineral soils and to below 5% in
the organic soils (except after rewetting).
Standardized respiration rates (Rref)
The time courses of the soil CO2 efflux standardized to
15 1C (Rref), that were yielded from the regression
analysis performed for each diurnal cycle are depicted
in Fig. 6. Soil CO2 efflux decreased throughout the
experiment for all treatments. The decrease in the wet
treatments indicates the effect of incubation time
during which substrate for decomposition diminishes.
Initial rates of CO2 efflux were around 3.5–
4.5mmol m�2 s�1 for organic soil, and 1–2 mmol m�2 s�1
for mineral soil. Soil CO2 efflux rates for all treatments
decreased rapidly during the first week, but subse-
quently declined at a slower rate. This behaviour is
indicative of labile pool depletion, and was particularly
pronounced in the mineral soil. Consequently, the
proportion mineral Rref to organic Rref declined rapidly
and stabilized at approximately 20% after 2 weeks of
incubation.
A difference in efflux rates between wet and dry
treatments developed during the first 25 days of the
experiment, but did not increase (in absolute terms)
after this period, although soil moisture continued to
decline. The rapid rewetting of the dry samples at the
end of the experiment led to an immediate increase in
soil CO2 efflux, which was initially higher than that of
the corresponding wet samples, but lower than the
MC
MD
0
1
2
3
4
5
120 140 160 180 200 220 240Julian day
Rre
f (µm
ol m
−2 s
−1)
Rre
f (µm
ol m
−2 s
−1)
0
1
2
3
4
5
120 140 160 180 200 220 240Julian day
OCOD
Organic layer
Mineral layer
Beg End Rewet
Beg End Rewet
(a)
(b)
Fig. 6 Time series of Rref as fitted to each diurnal cycle in the
laboratory incubation, for the organic layer (a) and mineral layer
(b) soil cores. Filled symbols denote the constantly wet
treatment; open symbols the drying–rewetting treatment. For
clarity, estimation errors are summarized as � 1 average
standard error of estimate for each treatment (OC, OD: organic
constantly wet, organic drying/rewetting; MC, MD: mineral
constantly wet, mineral drying/rewetting). Rref is derived from
the dual-source model as Rref 5 Rsurf,ref 1 Rcent,ref (cf. Eqn (2)).
All, the incubation time effect, the drying–rewetting effect and
the effect of the soil horizon are significant.
0
1
2
3
4
5
Rre
f (µm
ol m
−2 s
−1)
"Surface"
Organic
Mineral0
1
2
3
4
5
FieldBeg End Beg End Rewet
Const. wet Drying–rewetting
(a) (b)
Fig. 7 (a) Partitioning of fluxes at reference temperature (Rref; 15 1C) between three different compartments in the constantly wet
treatment (left) and the drying–rewetting treatment (right) at the beginning (Beg), end (End) of the drying period and after the rewetting
(Rewet; see shaded areas in Fig. 6). (b) Rref estimate from parallel field measurements. The difference in respiration between the two
treatments at the beginning of the incubation is because of the fact that the soil respiration was already reduced in the dry treatment over
the averaging period. Error bars denote standard errors.
T E M P E R AT U R E S E N S I T I V I T Y O F D E C O M P O S I T I O N 1761
r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 1754–1767
efflux observed in the wet samples at the beginning of
the treatment.
The dual-source model allowed us to tentatively
separate the organic layer into a ‘surface’ layer and a
‘bulk organic’ layer, that could be added to fluxes
measured on the mineral soil monoliths, resulting in
three compartments (Fig. 7). At the beginning of the
experiment, the Rref summed up for the three compart-
ments in the wet treatment compared favourably to the
respective Rref estimate from the field measurements,
both at around 4 mmol m�2 s�1, with overlapping con-
fidence intervals. However, after the 60–80 days of
incubation, the total soil CO2 efflux was nearly halved,
and the relative contribution of the surface compart-
ment declined slightly (from 36% to 29%), while the
bulk organic compartments relative contribution in-
creased (from 45% to 55%).
Total CO2 efflux in the dry treatment was already
reduced within the first 10 days of incubation, and by
the end of the dry period total CO2 efflux was reduced
to about one-third of initial (wet) values. The surface
compartment experienced a more than proportional
decrease over this period, and eventually only con-
stituted of 0.25mmol m�2 s�1, or 20% of the total efflux.
The difference between the two columns labelled ‘End’
in Fig. 7 document the effect of soil drying alone. Soil
drying reduced the surface flux in absolute, but also
relative terms. The reduction in flux from the bulk
organic compartment is less severe, resulting in a larger
relative contribution of 70% in the dry treatment,
compared with 55% in the wet treatment. Rewetting
approximately doubled the total flux, reaching slightly
higher rates than the wet treatment suggesting a small
rewetting effect. Here the surface compartment re-
sponded most strongly, raising its relative contribution
from 20% to 36%.
Temperature sensitivity (Q10)
While the overall level of the soil CO2 efflux (expressed
as Rref) exhibited clear treatment effects (incubation
time effect, drying effect, mineral vs. organic soil effect),
the temperature sensitivity (expressed as Q10) was
nearly invariant, regardless of the treatment (Fig. 8).
Because of overall higher fluxes, the Q10 of organic
samples could be determined with less uncertainty,
fluctuating randomly between 2.5 and 3.0, while
mineral sample Q10 values fluctuated around a similar
mean but with higher variance. For both organic and
mineral soils, no pure incubation time effect on the Q10
of soil CO2 efflux could be detected. For both soils, a
minimal but significant (organic soils: P 5 0.049, miner-
al soils: P 5 0.006, t-test) effect of the dry treatment is
visible, resulting in slightly lower Q10s in the dry
treatment (Fig. 9). At the end of the dry period, the Q10
seems to drop more substantially, and significantly
(Po0.05) recovers after the rewetting. However, those
changes remain within 0.4 Q10 units. The situation
is summarized and compared with field estimates in
Fig. 9. The dual-source model consistently estimates
slightly higher Q10 values than the single-source model
(non-hatched vs. hatched bars). The Q10 value for the
field was derived from the same regression model and
parameters as described for Rref, and calculated as the
increase in CO2 efflux for a temperature change from
10 1C to 20 1C to allow the comparison with the lab data.
The Q10 estimates from the lab compare well with the
field estimate of Q10 that was derived with a single-
source model.
Discussion
A number of field and laboratory studies analysing the
temperature dependence of soil respiration suffered
1
2
3
4
120 140 160 180 200 220 240Julian day
Q10
Q10
1
2
3
4
120 140 160 180 200 220 240Julian day
OC
OD
MC
MD
(a)
(b)
Fig. 8 Time series of Q10 of soil CO2 efflux as fitted to each
diurnal cycle in the laboratory incubation, for the organic layer
(a) and mineral layer (b) soil cores. Filled symbols denote the
constantly wet treatment; open symbols the drying–rewetting
treatment. For clarity, estimation errors are summarized as � 1
average standard error of estimate for each treatment (OC, OD,
organic constantly wet, organic drying–rewetting; MC, MD,
mineral constantly wet, mineral drying–rewetting).
1762 M . R E I C H S T E I N et al.
r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 1754–1767
from the problem that the respiration response to
temperature is easily confounded by other factors
(Davidson et al., 1998, 2000; Russell & Voroney, 1998;
Koizumi et al., 1999; Moren & Lindroth, 2000; Savage
& Davidson, 2001; Reichstein et al., 2002). This is
most evident under uncontrolled field conditions,
where soil water availability, radiation, vegetation
production, and other factors co-vary with temper-
ature. This problem has most recently been shown by
Baath & Wallander (2003) who demonstrated that the
results by Boone et al. (1998) – a higher temperature
sensitivity of root compared the soil microbial respira-
tion – were partly caused by interaction with vegetation
activity.
In order to avoid confounding effects, laboratory
experiments need to be carefully designed, as relative
substrate availability changes during the experiment in
different temperature treatments as discussed in detail
by Reichstein et al. (2000). This problem can be
circumvented by using a decomposition model where
the rate constants are temperature dependent (e.g.
Katterer et al., 1998). However, decomposition models
already contain a number of assumptions, which one
might want to test independently. For instance, the
temperature sensitivity of decomposition is usually
assumed to be independent of SOM quality. This latter
assumption of invariant temperature sensitivity has
been criticized from very different perspectives. While
Liski et al. (1999) empirically estimated that deep SOM
(old low stabile carbon) decomposition is independent
of soil temperature, Bosatta & Agren (1999) argued that
the temperature sensitivity should theoretically in-
crease with decreasing carbon quality because the
activation energy increases. It has also been found that
the Q10 of soil and ecosystem respiration empirically
decreases with soil water deficit (Carlyle & Ba Than,
1988; Kutsch & Kappen, 1997; Reichstein et al., 2002).
Our laboratory experiment allowed these problems to
be studied on minimally disturbed soil cores under
controlled conditions without confounding effects by
varying the temperature at a much higher frequency
than other factors change, while continuously obser-
ving the soil CO2 efflux.
The observed decrease in soil respiration rate with
increasing incubation time is in accordance with earlier
studies, indicating a depletion of available substrate
(Witkamp, 1966; O’connell, 1990; Cotrufo et al., 1995;
Bottner et al., 1998; Bottner et al., 2000). Initial respira-
tion rates were very similar to those in the field, which
supports the fact that the internal structure of the
samples remained largely unaltered (minimally dis-
turbed samples). However, soil CO2 efflux in the field
includes root derived CO2, as well as truly hetero-
trophic respiration (i.e. CO2 derived from the decom-
position of litter and SOM). The soil cores in the lab did
contain roots, but as there was no new supply of
photosynthates from aboveground by trees or grasses,
this proportion soon diminished. The drop in respira-
tion rates within the first week of incubation observed
for all treatments is likely to be partly caused by the
rapid decline of CO2 flux from roots and labile root
derived organic material. Field experiments where the
supply of photosynthates to the roots has been
suppressed by forest girdling (Hogberg et al., 2001;
Subke et al., 2004b) have shown that under these
conditions, roots continue to respire starch reserves for
several weeks. Our samples only included roots of
diameters smaller than 2 cm, resulting in a smaller
starch store compared with field conditions, and it is
also likely that root growth and phloem transport had
(a) (b)
1
1.5
2
2.5
3
3.5
Q10
1
1.5
2
2.5
3
3.5
FieldBeg End Beg End Beg EndRewet Beg End Rewet
Organic Mineral
Fig. 9 (a) Average estimates of Q10 in the organic layer (left) and the mineral layer (right), at the beginning (beg), end (end) of the
drying period and after the rewetting (Rewet). Filled bars are from constantly wet treatment, open bars from drying–rewetting
treatment. Hatched or striped bars represent results from the dual-source model, solid bars from the single-source model. (b) Q10
estimate from parallel field measurements using the single-source model. Error bars denote standard errors. Only the effect of the dry
treatment is significant.
T E M P E R AT U R E S E N S I T I V I T Y O F D E C O M P O S I T I O N 1763
r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 1754–1767
still not reached high values in this montane forest at
the time of sampling.
The effect of the soil drying is expected and confirms
the results of earlier studies (Orchard & Cook, 1983;
Howard & Howard, 1993; Lomander et al., 1998). The
drying was not extreme (e.g. not up to the permanent
wilting point), but nevertheless respiration was clearly
reduced by approximately 50%. Such reduction is
expected by common models that relate soil microbial
activity to the logarithm of soil water potential (pF-
value) with a maximum at field capacity (�250 hPa; pF
2.3) and zero respiration at the permanent wilting point
(�15 000 hPa; pF 4.2) (e.g., Andren et al., 1992; Rodrigo
et al., 1997) where in our study the soil water potential
dropped to between �1000 and 2000 hPa (pF 3–3.3). It is
noteworthy that the difference in soil respiration
between wet and dry treatments hardly increased after
30 days of incubation. Although speculative, this may
indicate that after a certain time substrate limitation
becomes more and more co-limiting to respiration, so
that the sensitivity to water availability declines.
Rewetting has been reported to cause soil CO2 efflux
rates that exceed those rates before drying (Redeker
et al., 2004; Xu & Baldocchi, 2004). This so-called Birch
effect (Birch, 1958) is explained by remineralization of
dead microbes and release of easily decomposable
substrate through rewetting. In our experiment, this
effect was confirmed, as CO2 efflux rates after rewetting
exceeded those of the wet samples at the same time. In
addition to the pool of labile C supplied from dead
microbial biomass, a portion of the extra respiration
may be caused by decomposition of SOM which had
been ‘saved’ under drought conditions, when microbial
activity was limited by water rather than substrate.
That a labile carbon fraction had mobilized is sup-
ported by the fact, that respiration rate dropped within
10 days of the rewetting, even though samples were
kept constantly wet thereafter.
According to our analysis with the dual-source
model, the surface layer showed the largest variation
with incubation time and the most pronounced
responses to drying–rewetting. Thus, our study calls
attention to the importance of fluxes from the soil
surface. In particular, superficial rewetting events that
trigger surface respiration may not be accounted for by
models that are driven by observations at 5 or 10 cm
depth or that treat the soil as monolayer ‘bucket’.
Reichstein et al. (2003) and Xu & Baldocchi (2004)
experienced these problems with rain events after
drought, when fluxes were higher than expected
according to soil moisture at 10 cm. Changes in relative
contributions from different soil compartments may
well change the apparent Q10 of soil respiration, when
CO2 efflux data are correlated with a temperature at a
fixed position (discussed in Subke et al., 2003). Such a
correlation may result in a wide range of observed
apparent Q10 values (e.g. Janssens & Pilegaard, 2003),
which have to be treated with caution before conclu-
sions about soil physiological responses to temperature
are drawn. Also with respect to results from Irvine &
Law (2002), we propose that ecosystem models operat-
ing at subdaily to daily time steps should comprise a
surface, a main root zone and – at least in forests – a
deep soil layer to account for different dynamics in
different compartments.
Our results from the dual-source model are some-
what exploratory and we were not able to provide a
final validation of the surface flux estimate. But we
show that with such a still simple approach, the
descriptive power is clearly enhanced, and the results
we obtained concerning relative contributions are quite
plausible: (1) The surface layer consisting of fresh grass
and needle litter, very likely contains the highest
proportion of labile SOM, and thus our result that
surface flux declines more strongly than the flux from
the bulk soil is reasonable. (2) Similarly, our observation
(derived from the dual-source model) that the surface
layer responds most sensitively to the drying–rewetting
treatment is conceivable, because the surface layer dries
up most severely. Two field studies conducted at the
same site showed conflicting results concerning flux
contributions from the organic and mineral layers.
While Buchmann (2000) concluded from direct mea-
surements following the removal of organic horizons
that the majority of respired CO2 originates from the
mineral layer, Subke et al. (2003) argued that the
correlation between temperature and surface efflux
suggest that the organic layer has the larger contribu-
tion to the total soil CO2 efflux. (The conflict between
the results of both field studies is discussed in Subke
et al., 2003). Thus, we suggest that such inverse estima-
tion of flux partitioning between different compartments
should be further investigated and could be validated
nondestructively (e.g. with isotope-labelling studies).
One of the main questions posed in the current study
was about the commonly assumed invariance of the
temperature sensitivity of soil respiration processes,
which has recently been criticized. The effect of organic
matter age on the temperature sensitivity of soil
respiration was inferred indirectly only, but from two
perspectives: (a) With incubation time the relative
contribution of young SOM pools declines and the
contribution of older pools increases, thus if older pools
had a different temperature sensitivity the Q10 should
change with incubation time. (b) The mineral-SOM is
older than the organic layer carbon, thus the mineral
soil should have had a different Q10 than the organic
layer, as (e.g. suggested by Liski et al. 1999). As virtually
1764 M . R E I C H S T E I N et al.
r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 1754–1767
no effects of incubation time or soil layer on the Q10 of
soil respiration were detected, our study suggests no or
only minor effects of SOM age on the Q10 of soil
respiration. A limitation to this conclusion is that we
did not explicitly determine the age of the carbon
respired and we cannot infer the temperature sensitiv-
ity of very old pools, if their contribution to the overall
flux is very small.
The effect of soil water availability on the tempera-
ture sensitivity of decomposition of SOM was statisti-
cally significant, but modest (o0.4 Q10 units on
average) at the levels of drying in this experiment and
much less drastic than suggested by recent field studies
(Xu & Qi, 2001; Reichstein et al., 2003). However, in this
case the results are somewhat equivocal, as (a) there
might be a threshold type of relationship between Q10
and soil water availability and (b) the drying might not
have been advanced enough to pass such a threshold.
The drop of Q10 at the end of the drying period is
indicative of this, but is too singular, as we did not
continue the drying past this point, and this possibility
should be investigated further.
Conclusions
Our study analysed the temperature sensitivity of soil
respiration in the laboratory under minimization of
confounding effects that often occur under field
conditions and separated the effects of soil moisture,
soil horizon and decomposition dynamics (incubation
time). Under our conditions we did not find evidence of
important changes of the direct sensitivity of soil
respiration to temperature in response to the mentioned
factors. This contrasts recent results from field studies
and shows that one must be careful with interpretation
of statistical results from ecosystem level studies, as
these results can be ‘emergent properties’ that do not
hold at the lower organizational level (e.g. Q10 of
soil respiration vs. Q10 of decomposition rate constants
in a process model). From our experiment we cannot
conclude that in process models the temperature
sensitivity of rate constants should either differ
between organic and mineral layer carbon, or should
change along ongoing decomposition of SOM. Our
results demonstrate that laboratory experiments, which
are commonly criticized for being unnatural, should be
realized as a successful complementary tool to field
measurements, as they allow a more elaborate elucida-
tion of confounded factors. The disturbance caused by
sampling of soil cores can be minimized if the structure
of the soil is retained, and is more than compensated by
the possibility of controlling abiotic conditions in the
laboratory.
Acknowledgements
The authors would like to thank A. Suske, D. Otieno and A.Bobeva for help with field and laboratory work during thisexperiment and Riccardo Valentini for valuable discussions. Weappreciate two anonymous reviewers’ and E. Davidson’s helpfulcomments on an earlier draft. The current study was performedwithin the CARBODATA and CARBOEUROFLUX projects.During data analysis and interpretation MR was supported bythe EU Marie-Curie fellowship INTERMODE (MEIF-CT-2003-500696).
References
Andren O, Steen E, Rajkai K (1992) Modelling the effects of
moisture on barley straw and root decomposition in the field.
Soil Biology & Biochemistry, 24, 727–736.
Baath E, Wallander H (2003) Soil and rhizosphere microorgan-
isms have the same Q10 for respiration in a model system.
Global Change Biology, 9, 1788–1791.
Birch H (1958) The effect of soil drying on humus decomposition
and nitrogen availability. Plant and Soil, 10, 9–31.
Boone RD, Nadelhoffer K, Canary JD et al. (1998) Roots exert a
strong influence on the temperature sensitivity of soil
respiration. Nature, 396, 570–572.
Bosatta E, Agren G (1999) Soil organic matter quality inter-
preted thermodynamically. Soil Biology & Biochemistry, 31,
1889–1891.
Bottner P, Austrui F, Cortez J et al. (1998) Decomposition of 14C-
and 15N-labelled plant material, under controlled conditions,
in coniferous forest soil from a north–south climatic sequence
in western Europe. Soil Biology & Biochemistry, 30, 597–610.
Bottner P, Couteaux M, Anderson JM et al. (2000) Decomposition
of 13C labelled plant material in a European 65–40 degree
latitudinal transect of coniferous forest soils: simulation of
climate change by translocation of soils. Soil Biology &
Biochemistry, 32, 527–543.
Buchmann N (2000) Biotic and abiotic factors controlling soil
respiration rates in Picea abies stands. Soil Biology & Biochem-
istry, 32, 1625–1635.
Carlyle JC, Ba Than U (1988) Abiotic controls of soil respiration
beneath an eighteen-year-old Pinus radiata stand in south-east
Australia. Journal of Ecology, 76, 654–662.
Cotrufo MF, Ineson P, Roberts JD (1995) Decomposition of birch
leaf litters with varying C-to-N ratios. Soil Biology &
Biochemistry, 27, 1219–1221.
Cox PM, Betts RA, Jones CD et al. (2000) Acceleration of global
warming due to carbon-cycle feedbacks in a coupled climate
model. Nature, 408, 184–187.
Cramer W, Bondeau A, Woodward FI et al. (2001) Global
response of terrestrial ecosystem structure and function to
CO2 and climate change: results from six dynamic global
vegetation models. Global Change Biology, 7, 357–373.
Davidson EA, Belk E, Boone RD (1998) Soil water content and
temperature as independent or confounded factors controlling
soil respiration in temperate mixed hardwood forest. Global
Change Biology, 4, 217–227.
Davidson EA, Verchot LV, Cattanio JH et al. (2000) Effects of soil
water content on soil respiration in forests and cattle pastures
of eastern Amazonia. Biogeochemistry, 48, 53–69.
T E M P E R AT U R E S E N S I T I V I T Y O F D E C O M P O S I T I O N 1765
r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 1754–1767
Draper N, Smith H (1981) Applied Regression Analysis. Wiley,
New York.
Fang C, Moncrieff JB (2001) The dependence of soil CO2
efflux on temperature. Soil Biology and Biochemistry, 35,
155–165.
Giardina CP, Ryan MG (2000) Evidence that decomposition rates
of organic carbon in mineral soil do not vary with tempera-
ture. Nature, 404, 858–861.
Hogberg P, Nordgren A, Buchmann N et al. (2001) Large-scale
forest girdling experiment shows that current photosynthesis
drives soil respiration. Nature, 411, 789–792.
Howard DM, Howard PJA (1993) Relationships between
CO2 evolution, moisture content and temperature for a
range of soil types. Soil Biology & Biochemistry, 25, 1537–
1546.
Irvine J, Law BE (2002) Contrasting soil respiration in young and
old-growth ponderosa pine forests. Global Change Biology, 8,
1–12.
Janssens IA, Pilegaard K (2003) Large seasonal changes of Q10
of soil respiration in a beech forest. Global Change Biology, 9,
911–918.
Kalbitz K (2004) Bodenkundliche Charakterisierung der Inten-
sivmessflache am Waldstein. In: Biogeochemistry of Forested
Catchments in a Changing Environment, Vol. 172 (ed. Matzner
E), Springer, Berlin.
Katterer T, Reichstein M, Andren O et al. (1998) Temperature
dependence of organic matter decomposition: a critical review
using literature data analysed with different models. Biology
and Fertility of Soils, 27, 258–262.
Kirschbaum MUF (2000) Will changes in soil organic carbon act
as a positive or negative feedback on global warming?
Biogeochemistry, 48, 21–51.
Koizumi H, Kontturi M, Mariko S et al. (1999) Soil respiration in
three soil types in agricultural ecosystems in Finland. Acta
Agriculturæ Scandinavica, Section B, Soil and Plant Science, 49,
65–74.
Kutsch W, Kappen L (1997) Aspects of carbon and nitrogen
cycling in soils in the Bornhoved Lake district: II. Modelling
the influence of temperature increase on soil respiration and
organic carbon content in arable soils under different manage-
ments. Biogeochemistry, 39, 207–224.
Liski J, Ilvesniemi H, Makela A et al. (1999) CO2 emissions from
soil in response to climatic warming are overestimated – the
decomposition of old soil organic matter is tolerant of
temperature. Ambio, 28, 171–174.
Lloyd J, Taylor JA (1994) On the temperature dependence of soil
respiration. Functional Ecology, 8, 315–323.
Lomander A, Katterer T, Andren O (1998) Carbon dioxide
evolution from top- and subsoil as affected by moisture and
constant and fluctuating temperature. Soil Biology & Biochem-
istry, 30, 2017–2022.
Moren A-S, Lindroth A (2000) CO2 exchange at the floor
of a boreal forest. Agricultural and Forest Meteorology, 101,
1–14.
O’Connell AM (1990) Microbial decomposition (respiration) of
litter in eucalypt forests of south-western Australia: an
empirical model based on laboratory incubations. Soil Biology
& Biochemistry, 22, 153–160.
Orchard VA, Cook FJ (1983) Relationship between soil respira-
tion and soil moisture. Soil Biology and Biochemistry, 15,
447–454.
Raich JW, Potter CS (1995) Global patterns of carbon dioxide
emissions from soils. Global Biogeochemical Cycles, 9, 23–36.
Raich JW, Potter CS, Bhagawati D (2002) Interannual variability
in global soil respiration, 1980–94. Global Change Biology, 8,
800–812.
Raich JW, Schlesinger WH (1992) The global carbon dioxide flux
in soil respiration and its relationship to vegetation and
climate. Tellus, 81–99.
Redeker KR, Treseder KK, Allen MF et al. (2004) Rapid and
transient response of soil respiration to rain. Global Change
Biology, 10, 1017–1026.
Reichstein M, Bednorz F, Broll G et al. (2000) Temperature
dependence of carbon mineralization: conclusions from a
long-term incubation of subalpine soil samples. Soil Biology &
Biochemistry, 32, 947–958.
Reichstein M, Rey A, Freibauer A et al. (2003) Modelling
temporal and large-scale spatial variability of soil respiration
from soil water availability, temperature and vegetation
productivity indices. Global Biogeochemical Cycles, 17, 15/11–
15/15, doi: 10.1029/2003GB002035.
Reichstein M, Tenhunen JD, Ourcival JM et al. (2002) Ecosystem
respiration in two Mediterranean evergreen holm oak forests:
drought effects and decomposition dynamics. Functional
Ecology, 16, 27–39.
Rodrigo A, Recous S, Neef C et al. (1997) Modelling temper-
ature and moisture effects on C–N transformations in
soils: comparison of nine models. Ecological Modelling, 102,
325–329.
Russell CA, Voroney RP (1998) Carbon dioxide efflux from the
floor of a boreal aspen forest. I. Relationship to environmental
variables and estimates of C respired. Canadian Journal of Soil
Science, 78, 301–310.
Savage KE, Davidson EA (2001) Interannual variation of soil
respiration in two New England forests. Global Biogeochemical
Cycles, 15, 337–350.
Schimel DS, House JI, Hibbard KA et al. (2001) Recent patterns
and mechanisms of carbon exchange by terrestrial ecosystems.
Nature, 414, 169–172.
Subke J-A, Buchmann N, Tenhunen JD (2004a) Soil CO2 fluxes
in spruce forests – temporal variation, and environ-
mental controls. In: Biogeochemistry of forested catchments in a
changing environment, Vol. 172 (ed. Matzner E), Springer,
Berlin.
Subke J-A, Hahn V, Battipaglia G et al. (2004b) Feedback
interactions between needle litter decomposition and rhizo-
sphere activity. Oecologia, 139, 551–559.
Subke J-A, Reichstein M, Tenhunen M (2003) Explaining
temporal variation in soil CO2 efflux in a mature spruce
forest in Southern Germany. Soil Biology & Biochemistry, 35,
1467–1483.
Van Genuchten MT (1980) A closed-form equation for predicting
the hydraulic conductivity of unsaturated soils. Soil Science
Society of America Journal, 44, 892–898.
Visual Numerics Inc. (2001) PV-Wave Advantage Reference. VNI
Press, Houston.
1766 M . R E I C H S T E I N et al.
r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 1754–1767
Winkler JP, Cherry RS, Schlesinger WH (1996) The Q10 relation-
ship of microbial respiration in a temperate forest soil. Soil
Biology & Biochemistry, 28, 1067–1072.
Witkamp M (1966) Decomposition of leaf litter in relation to
environment, microflora, and microbial respiration. Ecology,
47, 194–201.
Xu L, Baldocchi DD (2004) Seasonal variation in carbon dioxide
exchange over a Mediterranean annual grassland in Califor-
nia. Agricultural and Forest Meteorology, 123, 79–96.
Xu M, Qi Y (2001) Soil-surface CO2 efflux and its spatial and
temporal variations in a young ponderosa pine plantation in
northern California. Global Change Biology, 7, 667–677.
T E M P E R AT U R E S E N S I T I V I T Y O F D E C O M P O S I T I O N 1767
r 2005 Blackwell Publishing Ltd, Global Change Biology, 11, 1754–1767