Introduction
Soil is the largest pool of terrestrial organic carbon in the biosphere,
storing around 2344 PgC in the top 3 m (Jobbágy and
Jackson, 2000). Soil respiration (SR) is the main source of carbon efflux from
ecosystems to the atmosphere, accounting for 60–90 % of the total
ecosystem respiration (Schimel et al., 2001; Raich et al., 2002). Thus, SR
plays an important role in the global carbon balance (Schimel et al., 2001;
Raich et al., 2002), and even small changes of SR may induce positive
feedbacks to climate change (Schlesinger and Andrews, 2000). Therefore,
information of how SR interacts with environmental conditions, such as the
response of specific components of soil respiration to temperature and
moisture changes, will be a key part of the improvement of process-based
models.
On large scales, such as in ecosystems and biomes, net primary production
(NPP) may be the most important factor controlling SR (Wardle, 2002). NPP
provides the inputs to the soil from aboveground litter and also belowground
organic detritus (Raich and Potter, 1995). Moreover, root respiration is
strongly dependent on the translocation of photosynthates from the
aboveground part of the plant (Curiel-Yuste et al., 2004). At the smaller
scale, SR has been found to be very sensitive to soil temperature and soil
moisture (Fang and Moncrieff, 2001). Soil temperature has been recognized as
the most important environmental factor controlling SR because it affects the
respiratory enzymes of both roots and soil microbial biomass (Xu et al.,
2011). In general, SR increases exponentially with increases of soil
temperature (Epron et al., 1999; Lloyd and Taylor, 1994; Mielnick and Dugas,
2000). In contrast to the positive relationship between SR and soil
temperature, both very high and very low soil moisture have been shown
to diminish the temperature response of SR (Londo et al., 1999; Welsch and
Hornberger, 2004) due to the potential oxygen limitations under high soil
moisture (Skopp et al., 1990) and due to metabolic drought stress under very
low soil moisture (Orchard and Cook, 1983). Soil moisture also affects plant
composition and productivity (Häring et al., 2013) and thus controls the
quantity and quality of both soil organic matter (SOM) and root exudate
supply (Rustad et al., 2000).
Numerous studies have reported the effects of temperature and moisture on SR.
However, studies about the combined effects of both factors are relatively
few, and the information of how soil moisture affects the relationship between
soil temperature and SR is scarce (Bowden et al., 1998; Davidson et al.,
2006; Curiel-Yuste et al., 2007). In Mediterranean and semiarid ecosystems,
SR is highly sensitive to soil moisture, and the temperature-driven increases
in SR are likely dampened by low soil moisture (Conant et al., 2004; Raich
and Potter, 1995; Rey et al., 2002). It is still unclear under which
circumstances or environmental conditions would the primary control factor of SR
switch from temperature to soil moisture.
SR can be divided into autotrophic and heterotrophic respiration by
different biological sources (Hanson et al., 2000). Autotrophic
respiration, also known as root respiration, is mainly dependent on NPP and
tree physiology such as photosynthesis substrate supply (Heinemeyer
et al., 2007; Hogberg et al., 2001). Heterotrophic respiration is the
sum of microbial decomposition of SOM (Fang et al., 2005; Knorr
et al., 2005). In theory, due to the different origins of autotrophic
and heterotrophic respiration, they may have different sensitivities
toward environmental factors and respond differently to seasonality
(Epron et al., 2001; Kuzyakov and Larionova, 2006; Yan et al., 2010).
Riparian areas are characterized with high soil moisture and sustained water table. (McGlynn and Seibert, 2003). In these ecosystems, tree species
composition and tree growth are strongly influenced by the topographic
position concomitant with the changes in the soil water content.
Thus, this may indirectly affect SR through litter input and nutrient
availability. Because of the retardation of microbial decomposition
with the frequent saturation of soil water, riparian areas tend to
accumulate more SOM than hillslope areas do (Sjögersten et al.,
2006).
The main objectives of this study were (1) to investigate the
seasonal variations and relationships between SR and both soil
temperature and moisture in a Mediterranean riparian forest along
a groundwater level gradient; (2) to determine soil moisture
thresholds at which SR is controlled by soil moisture rather than by
temperature, even in such non-water-stressed environments; (3) to
compare SR responses under different tree species present in
a Mediterranean riparian forest (Alnus glutinosa, Populus
nigra and Fraxinus excelsior). With these aims, we
carried out measurements of SR under different tree species along
a groundwater level gradient in a riparian forest in NE Spain. The
results of our study may help to better the understanding of the interactions
between different components of SR with soil temperature and moisture
as well as the role of different tree species. It also provides
relevant information for SR model parameterization.
Material and methods
Site description
The experiment was conducted in a riparian forest growing along the Font
de Regàs stream, a headwater tributary of the Tordera River, in
Montseny Natural Park (north of Barcelona;
41∘50′ N, 2∘30′ E, altitudinal
range 300–1200 ma.s.l.). The forest community of our study
site consists of black alder (Alnus glutinosa L.), black
locust (Robinia pseudoacacia L.), common ash
(Fraxinus excelsior L.) and black poplar (Populus
nigra L.). As result of water and nutrient availability,
A. glutinosa and P. nigra trees are mostly distributed
near the river, whereas F. excelsior trees are located further
away on the upper site, near the hill. R. pseudoacacia
trees are scattered over the study area and were not monitored. Mean
annual temperature is 12 ∘C with maximum and minimum average
temperatures of 14 and 10 ∘C, respectively. The mean annual
precipitation is 872 mm (1951–2010). The riparian soil is
sandy loam with low rock content (<13 %), weakly acidic (pH of
6.7) and has an average bulk density of 1.09 gcm-3.
Experimental design
We divided the groundwater gradient (riparian–hillslope transect) into four
levels according to the distance from the riverside and by tree species
composition (Fig. 1). The distances of level 1 to level 4 (L1 to L4) from the
river centre were 2.7, 4.4, 6.8 and 11.8 m, respectively. The three
target tree species, A. glutinosa,
P. nigra and F. excelsior were located at levels L1, L2 and
L3, respectively. To examine the interaction effects on SR of tree species,
soil moisture and temperature, we set three
riparian–hillslope transects to measure the variation of total SR (sum of soil autotrophic
and heterotrophic respiration, hereafter referred to collectively as total
SR, SRtot) from different tree species. Soil chambers were
placed 1.5 m from the stem of the target tree species. Moreover, we
also set two transects to measure the topographic effects on soil
heterotrophic respiration (SRH). Due to the difficulty of
trenching next to the riverbank, chambers for SRH were
set only at levels L2, L3 and L4. To separate root respiration from
SRH, we inserted a PVC tube (diameter: 65 cm;
height: 40 cm) into the soil 5 months before starting the
measurements. To avoid constraints on groundwater table fluctuations by
the PVC tube, we cut two opposite windows into the PVC tube and covered by
65 µm mesh to prevent root growth through the windows.
Stainless steel rings were inserted permanently into the soil, down to
3 cm depth as the base of the soil chambers, and kept free
from seedlings throughout the experiment duration. The distances of
each soil chamber from the riverside varied slightly due to the tree
distribution.
Field measurement
SR and soil temperatures were measured seasonally from summer 2011 to
autumn 2012. These measurements were conducted continuously for 1
week within each season. A heavy rainfall event took place in winter
2012, resulting in elevated water levels of the river that washed away
most of the litter layer within 3 m from the river
bank.
Sketch of levels along a gradient of soil water availability
with tree species distribution and SRH chamber
positions.
CO2 concentration was measured in situ with an automatic
changeover open system. The system consists of an infrared gas
analyser (IRGA, LiCor 6262, LiCor, Inc., Lincoln, NE, USA),
a data logger (CR10, Campbell Scientific Inc., UT, USA), 12 pairs of
channels, 12 soil chambers, 12 pairs of rotameters, 6 pumps and 2
flowmeters. Each pair of channels consists of two tubes connected to
a soil chamber, one attached on the top of chamber (reference
CO2 concentration) and another attached at the base for
calculating the increment of CO2 concentration provided by
SR. Soil chambers were placed at the beginning of each field
campaign, and CO2 concentrations were analysed and recorded
sequentially over 1 min intervals at each chamber. Air was
continuously forced through all chambers by pumps. Only one chamber was connected at
a time to the IRGA to analyse the CO2
concentration of the respective chamber, while air from the others was
exhausted to the atmosphere until their own turn. The sequence was
programmed every 4 cycles of differential IRGA measurements from
12 chambers, and an additional cycle of absolute IRGA measurement,
which was then used to calculate the actual absolute ambient air
concentration of CO2 in ppm. The CO2 concentration of
the ambient air was determined as the difference between the scrubbed
sample, which flows through soda lime and Mg(ClO4)2, and
the ambient air sample.
Soil chambers were protected by placing a 50cm×50cm green fine mesh on top to avoid possible heating by
direct sunlight during the measurements. Soil temperature of
5 cm depth was continuously measured with Pt100 temperature
sensors and recorded in parallel with the CO2 concentration
analysis. Thirty centimetre integral soil moisture (cm3cm-3,
SM30) in each level were determined and recorded
half-hourly with a moisture reflectometer (CS616, Campbell Scientific).
Additionally, we also measured 5 cm integral soil moisture
(SM5) next to each soil chamber once per day during each
measuring field campaign with impedance probes (ThetaProbe
soil moisture sensor, MI2x, Delta-T Devices, Cambridge, UK). A grid of 28 wells (PVC tubes of 35 mm in diameter)
was installed to monitor groundwater table oscillation. Wells were
distributed along the study site and at different distances from the
stream: 2.7, 4.4, 6.8, 11.8 m (n=7). Groundwater levels were
monitored manually every 2 weeks using a sounding device with
acoustic and light signal (Eijkelkamp Agrisearch Equipment). In autumn
of 2012, after concluding the measurements, litter layer and soil
samples (15 cm depth) inside each chamber were
collected. Litter layer samples were weighted after oven drying at
65–70 ∘C for 24 h. Soil samples were first
oven dried at 105 ∘C and then analysed to determine their
organic carbon and nitrogen content by using the Walkley–Black and
Kjeldahl methods, respectively.
Statistical analysis
Statistical analyses were performed with PASW statistics 18 (SPSS
Inc., 2009, Chicago, IL, USA). The missing data of soil temperatures were
estimated from air temperature values based on a regression analyses
between air and soil temperatures. SR, soil temperature and soil
moisture data were analysed using ANOVA to examine whether seasonal SR
rates were different between levels and tree species. Data used to test
the significance in ANOVA were based on daily means. Least significant
difference (LSD) was used to detect differences between levels and tree
species for each season. We used regression analysis to examine the
relationship between SR and soil temperature. An univariate
exponential equation was fitted (van 't Hoff, 1898):
SR=aebT,
where SR is soil respiration rate
(µmolCm-2s-1), T is soil temperature
(∘C), a and b are fitted parameters.
The apparent Q10 was calculated
as
Q10=e10b
A Q10 value for the whole measurement period
was computed for each topographic position and tree species on the
basis of daily average SR rate and soil temperature. In addition, we
estimated specific Q10 values for summer of 2011 and 2012. Data
collected were fitted to the exponential equation.
In order to understand the interaction between soil temperature and
soil moisture and the effect of soil moisture on regulating SR, we
applied recursive partitioning analysis to search for the threshold of
soil moisture. As models based on partitioning can only handle linear
models, the Eq (1) was transformed by linearizing with
logarithms:
lnSR=lna+bT
Logarithmic transformed SR values were used as the dependent
variable. Once the soil moisture thresholds were obtained, linear and
nonlinear regression analyses were used to determine the relationship
between SR, soil temperature and soil moisture at each soil moisture
interval. The recursive partitioning analysis was conducted in the R
statistical environmental using the party package (Zeileis
et al., 2008).
Results
Seasonal variation of groundwater level, soil moisture, soil nitrogen and carbon content
Seasonal variation of air temperature and precipitation was
remarkable. The precipitation in 2011 was significantly higher than in
2012, especially in summer. Summer precipitation in 2011 was 4
times higher (183 mm) than in 2012
(39 mm). SM30 was significantly higher at L1
(Fig. 2). In summer 2012, due to a remarkable drought,
SM30 at L1 only showed a small decrease with respect to
summer 2011; while at the other levels (L2, L3 and L4)
SM30 was markedly decreased. Groundwater levels showed no
seasonal variation but were significantly different between them.
Soil carbon and nitrogen content and dry weight of litter L and F organic horizons from soil respiration chambers.
Groundwater level
C/N
SOC %
Nitrogen%
Litter Layer
(kgm-2)
L2 – Near river
10.40
2.73
0.16
0.97
L3 – Intermediate
10.00
4.38
0.26
1.20
L4 – Uphill
9.15
3.36
0.23
1.67
L1 – A. glutinosa
12.13
2.29
0.11
0.69
L2 – P. nigra
10.27
3.52
0.20
1.18
L3 – F. excelsior
9.67
4.85
0.30
2.21
Seasonal changes of summer 2011 (Su 11), autumn 2011
(Au 11), winter 2012 (Wi 12), spring 2012 (Sp 12) summer 2012 (Su 12) and autumn 2012
(Au12) in
(a) mean seasonal air temperature and precipitation;
(b) 30 cm integral soil moisture (SM30);
(c) groundwater level, value represents the depth of
groundwater level from soil surface (L1, L2, L3 and L4).
Soil near the river contained less organic carbon and nitrogen, but
a higher C : N ratio, with a C : N ratio of 12.13 (Table 1). Soil
C : N ratio decreased from the riverside going uphill, whereas the dry
weight of litter layer increased from the riverside going uphill. The
largest amount of dry weight of litter layer was found under
F. excelsior, and coincided with the highest soil organic
carbon (SOC) and soil nitrogen concentrations between all levels.
Comparison of soil respiration rates (SR), soil moisture (SM) and
Q10 values in 2011 and 2012 summer campaigns. Heterotrophic SR
(SRH). Total SR (SRtot). Five centimetre integral soil
moisture (SM5). Thirty centimetre integral soil moisture (SM30).
SR (µmolCm-2s-1)
SM5 (%)
SM30 (%)
Q10
Chamber
2011
2012
Reduction
2011
2012
Reduction
2011
2012
Reduction
2011
2012
SRH
L2 – Near river
1.65
0.84
49 %
27.10
14.94
45 %
22.22
14.51
35 %
1.09
0.76
L3 – Intermediate
0.98
0.70
28 %
31.68
14.91
53 %
12.60
9.22
27 %
1.04
0.88
L4 – Uphill
0.74
0.50
32 %
38.02
14.19
63 %
10.87
8.13
25 %
0.97
0.84
SRtot
L1 – A. glutinosa
1.24
0.78
37 %
27.24
13.04
52 %
42.49
36.58
14 %
1.31
0.80
L2 – P. nigra
1.42
1.13
21 %
26.22
12.93
51 %
22.22
14.51
35 %
1.17
0.63
L3 – F. excelsior
1.26
0.76
40 %
26.45
12.87
51 %
12.60
9.22
27 %
1.40
1.14
All data of SR, SM5 and SM30 were
significantly different between 2011 and 2012. All P values <0.001.
Seasonal variation of SRH along hillslope transect
SRH rates ranged from
0.17 µmolCm-2s-1 (in winter, L4) to
1.69 µmolCm-2s-1 (in summer, L2,
Fig. 3a–d). SRH decreased significantly from
riparian zone (L2) to hill zone (L4), especially in
summer. SRH measured from different levels were
significantly different in all seasons (P<0.05). SRH at L2 had a higher variability during
the whole experiment. Minimum soil temperature coincided with maximum
SM5 in winter while maximum soil temperature was recorded
in summer when SM5 was lowest. SRH
varied markedly during the year following the change of soil
temperature from summer 2011 to spring 2012, and the changes of
SM5 for summer and autumn 2012. As expected,
SRH was lower during winter when soil
temperatures were the lowest of the year, and SRH
was higher during the growing season.
Comparison of soil respiration rates (SR) and soil moisture (SM)
after a rainfall event of 13.5 mm in summer 2012. Heterotrophic SR
(SRH). Total SR (SRtot). Five centimetre integral soil
moisture (SM5). Thirty centimetre integral soil moisture (SM30). Data were
averaged for the 2 days before and 2 days after the rainfall event.
SR (µmolCm-2s-1)
SM5 (%)
SM30 (%)
Chamber
before
after
increase
before
after
increase
before
after
increase
SRH
L2 – Near river
0.66
1.00
52 %
14.09
18.84
34 %
14.45
14.50
0 %
L3 – Intermediate
0.59
0.80
34 %
15.19
18.37
21 %
8.46
10.15
20 %
L4 – Uphill
0.41
0.59
45 %
12.06
17.51
45 %
6.97
9.64
38 %
SRtot
L1 – A. glutinosa
0.67
1.04
54 %
11.27
16.91
50 %
36.13
37.48
4 %
L2 – P. nigra
0.99
1.66
68 %
10.86
18.86
74 %
14.45
14.50
0 %
L3 – F. excelsior
0.68
0.98
44 %
11.10
17.20
55 %
8.46
10.15
20 %
Exponential relationships between soil respiration (SR) and soil
temperature (T), and Q10 for different SM5 intervals.
Heterotrophic SR (SRH). Total SR (SRtot).
(SM5) is 5 cm integral soil moisture.
SM5>23 %
23 % >SM5>20 %
SM5<20 %
Fn
R2
Q10
Fn
R2
Q10
Fn
R2
Q10
SRH
L2 – Near river
SRH=0.52e0.05T
0.77∗∗∗
1.58
SRH=0.68e0.02T
0.74∗
1.25
SRH=2.10e0.4T
0.58∗∗
0.02
L3 – Intermediate
SRH=0.51e0.04T
0.72∗∗∗
1.49
SRH=0.67e0.05T
0.70∗
1.65
SRH=2.11e-0.04T
0.57∗∗
0.66
L4 – Uphill
SRH=0.40e0.05T
0.84∗∗∗
1.58
SRH=0.64e0.02T
0.66∗
1.19
SRH=1.34e-0.03T
0.34∗
0.76
SM5>27 %
27 % >SM5>17 %
SM5<17 %
SRtot
L1 – A. glutinosa
SRtot=0.53e0.04T
0.77∗∗∗
1.54
SRtot=0.69e0.03T
0.83∗∗∗
1.30
SRtot=0.77e0.01T
0.01
1.06
L2 – P. nigra
SRtot=0.52e0.05T
0.78∗∗∗
1.60
SRtot=0.61e0.04T
0.80∗∗∗
1.46
SRtot=1.39e-0.02T
0.19∗∗
1.17
L3 – F. excelsior
SRtot=0.32e0.08T
0.68∗∗∗
2.14
SRtot=0.56e0.03T
0.62∗∗∗
1.40
SRtot=1.30e-0.02T
0.25∗∗
0.82
∗ P<0.05; ∗∗ P<0.01; ∗∗∗ P<0.001
Seasonal variation of soil respiration, soil temperature and soil
moisture. (a–d) Data of soil heterotrophic respiration:
(a) SRH along groundwater level gradient. (b) 5 cm soil
temperature. (c) 5 cm integral soil moisture (SM5). (d)
30 cm integral soil moisture (SM30). (e–h) Data of total soil
respiration rates (SRtot) of three tree species. (e)
SRtot under different tree species. (f) 5 cm soil
temperature. (g) 5 cm integral soil moisture (SM5). (h)
30 cm integral soil moisture (SM30) . All values are mean SD. Data
points marked with indicate significant differences between species at P
<0.05 (For details, please see to Appendices A and B).
Tree species and topographic effects on SRtot and SM30
The observed variation of SRtot for the three tree
species followed the change of soil temperature over the year
(Fig. 3e–h). SRtot of P. nigra was the
highest one, especially during summer, and SRtot of
A. glutinosa was the lowest one throughout the year. There
were no significant differences of soil temperatures between tree
species locations. SM5 did not differ between tree species
locations but there was a tendency towards a higher SM5
under F. excelsior. SM30 was significantly
different between levels for all seasons. The variation of
SM30 at L1 was lower and showed less seasonal variability,
maintaining most of the SM30 values around 40 %.
During both summers 2011 and 2012, SM30 at L3 dropped to
around 10 %, which is even lower than the SM5 at L4
where F. excelsior is found.
Drought and rain pulse effects on SR
The low precipitation of summer 2012 caused a significant reduction of around
50 % of SM5, 14–35 % of SM30 and at the same time a reduction of SR
between 21 and 49 %. The Q10 values ranged from 0.97–1.40 in summer
2011 and 0.63–1.14 in summer 2012 (Table 2).
Differentiation of soil temperature (ST) and soil moisture
(SM) as primary controlling factors for
SRH. At SM5≥23 %, there are positive correlations of SRH with
soil temperature in all levels (a). At 20 %
≤SM5<23 %, there is a transition with no clear relationship
of either SM or ST with SRH (b). At
SM5<20 %, there is no relationship between SR and ST as the
inset figure shows (c); it switches from ST to SM5 as
the
controlling factor with positive correlations between
SRH and SM5 for all
levels. Campaigns with SM5<20 % were all from
spring and summer 2012. SM5 (5 cm integral soil
moisture).
Differentiation of soil temperature (ST) and soil moisture
(SM) as primary controlling factors for
SRtot. At SM5≥27 %, there are positive correlations of SRtot with
soil temperature under all tree species (a). At 17 %
≤SM5<27 %, there are positive but slightly weaker
correlations of SRtot with soil temperature under
all tree species (b). At SM5<17 %, there is no
relationship between SR and ST as the inset figure shows (c); it
switches from ST to SM5 as the controlling factor with
positive correlations between SRtot and
SM5 for all tree species. Campaigns with SM5<17 % were all from spring and summer 2012. SM5
(5 cm integral soil moisture).
A rainfall event (13.5 mm) during the measurement period of
summer 2012 caused a significant increase of soil moisture and SR
rates at all levels (L1 to L4). The SM5 increased around
21–74 % after the rainfall event even though it only caused
a 0–38 % increase of the SM30 (Table 3). This
rainfall event caused a sharp increase of SR from
0.41–0.99 µmolCm-2s-1 to
0.59–1.66 µmolCm-2s-1, which corresponds to
an increase of SR around 34 to 68 %.
The switch of primary control factor of SR
We identified three SM5 intervals for each
SRH and SRtot (Table 4), which
suggests the existence of thresholds in soil moisture effects. SR was
positively related (P<0.001) to soil temperature when SM5 was
higher than 23 % for SRH or higher than 27 %
for SRtot. The lower thresholds for
SRH and SRtot were 20 and 17 % of
SM5 respectively. Under the lower bound value,
SRH showed a significantly positive relationship with
SM5 (Fig. 4, linear regression with r2 of 0.89, 0.92 and
0.91 for L2, L3 and L4) while SRtot showed a weak positive
relationship with SM5 (Fig. 5, linear regression with r2 of
0.56, 0.11 and 0.10 for L1, L2 and L3). The exponential model based on soil
temperature accounts for 68 to 84 % of the variation in both
SRH and SRtot rates at the higher
SM5 interval values. The fitted Q10 values in high
SM5 intervals ranged from 1.49 to 2.14. Generally the Q10
values of SRH were lower than the Q10 of
SRtot.
Discussion
Effect of groundwater level and soil moisture on SR
In the studies of Martin and Bolstad (2005) and Pacific et al. (2008), it was
indicated that the amount and availability of soil water varies depending on
landscape position and topography. Both studies also show that small
differences in micro-topography appear to be important in driving soil
moisture conditions. This is in accordance with our results; the overall
seasonal trends of soil moisture were similar, but differences in the
relative magnitude of soil moisture still can be found between different levels.
In our study site, the SRH was significantly
higher at L2 and decreased with the distance from the river. At the
same time, SRtot of A. glutinosa at L1 was
significantly lower than the other two species found at L2 and
L3. This result could be explained by limitations of SR imposed by
groundwater level in two different ways. First, when the groundwater level
is low, the drought stresses soil microbial and root respiration activity; second, when groundwater level is high and close to topsoil
surface, it limits soil aeration and likely reduces the effective
respiring soil volume. Pacific et al. (2008) showed that the soil
CO2 concentrations were significantly higher in the riparian
zone as a result of higher soil moisture. In contrast, Zanchi
et al. (2011) found lower SR in plots after drainage, and suggested
that the low C and N content in the topsoil near the river, where
most of the soil CO2 respiration is produced, could partially
explain that low SR. The discrepancy of these two studies could be
associated with the different drainage regimes as the poorly drained
plots imply the anaerobic inhibition of SR. In our study, however,
SRH was measured at L2, L3 and L4 under
well-drained conditions, and SRH decreased
concomitantly with the decrease in the availability of soil water.
Nonetheless, SRtot of A. glutinosa was
measured at L1, where the soils sometimes experienced flooding or poor draining conditions, and the root respiration may be inhibited by
the high groundwater level.
Additionally, landscape position and topography not only altered the
availability of soil water but also affected the annual range of soil
moisture. This was shown in Zanchi et al. (2011), studying riparian SR in
Amazonia. They indicated that riparian soil is very sensitive to the changes
of water-flooding regimes. The high groundwater table in riparian zones
implies intermittent anaerobic conditions and the inhibition of diffusion
during water saturation. These differences in soil moisture caused by site
topography may result in differences in SR even though the soil temperatures
were similar among all sites. The different behaviours of
SRH and SRtot from L1 to L4 from our
results indicate a different contribution of SRH to
SRtot. As the root system of A. glutinosa may
constantly experience a saturated water regime, the relative contribution
from root respiration may be much lower than the one of the other two
species.
Rain pulse and drought effects on SR
The Mediterranean climate is characterized by summer droughts that
particularly affect the top soil layers; therefore, rainfall events during
these dry periods can trigger abrupt increases in SR that last for days
(Bowling et al., 2011; Cisneros-Dozal et al., 2007; Lee et al., 2004; Unger
et al., 2010). Lee et al. (2004) simulated precipitation and found that
hardwood forest floors were very sensitive to changes in moisture in the
upper soil layers. Moreover, Wang et al. (2012) noted that the response of
litterfall respiration is very sensitive to rainfall, and the increase in
soil moisture by rainfall primarily enhanced the litterfall respiration but
decreased mineral SR. Similar results were published by Casals et al. (2011),
who reported that SR after a precipitation pulse was mostly derived from
SRH with a contribution up to 70 % of
SRtot. Hence, our findings seem to be consistent with
these previous studies.
Confounding effects of temperature and moisture on SR
This study aimed at assessing the importance of soil moisture on soil
respiration and determining the threshold of soil moisture at which
soil moisture overrules temperature in controlling SR. The response of
SR to soil moisture has been widely studied and described by various
types of functions such as linear or logarithmic functions, depending
on the soil type, climate or vegetation type (Comstedt et al., 2010;
Epron et al., 1999; Orchard and Cook, 1983). In our study, the
seasonal courses of SRH and
SRtot generally followed the seasonal cycle of
temperature, but were moderated by soil moisture. Such a relationship is in
agreement with other previous studies (Davidson et al., 1998; Martin
and Bolstad, 2005; Wang et al., 2013).
The positive linear relationship between SR and soil moisture in low
soil moisture conditions found in our work agrees with many previous
studies where low soil moisture constrains SR (Almagro et al., 2009;
Davidson et al., 1998; Keith et al., 1997; Rey et al., 2002; Wang
et al., 2013; Xu and Qi, 2001). In our study, the low soil moisture
and warmer temperatures actually reduced SR rates, resulting in lower
Q10 values at the lower soil moisture. A similar decline of
Q10 with decreasing soil moisture was reported by Conant
et al. (2004), Curiel-Yuste et al. (2003) and Wen et al. (2006). Low
soil water content not only reduces the contact between the substrate,
enzymes and microbes, it also decreases the substrate supply due to
the increased drying-out of litter and topsoil layer (Davidson et al.,
2006). Another possible reason for the observed lower Q10 is that
the reduction of photosynthesis decreases the translocation of
photosynthates to the rhizosphere (Hogberg et al., 2001; Nordgren
et al., 2003).
In a Norway spruce stand, Gärdenäs (2000) found that litter moisture
explained most of the variation of SR, whereas mineral soil moisture, air and
litter temperature had no effects on SR. Our results showed that the seasonal
variations of SRH and SRtot were
mainly controlled by soil temperature, with a secondary influence of soil
moisture (SM5). Using the recursive partitioning method, we have
identified clear thresholds for SM5 effects on the temperature
sensitivity of SR. Soil moisture thresholds at which SR temperature
sensitivity is reduced have been found in several studies from different
ecosystems (Fang and Moncrieff, 2001; Gaumont-Guay et al., 2006; Jassal
et al., 2008; Lellei-Kovács et al., 2011; Palmroth et al., 2005; Wang
et al., 2013). However, the threshold values in soil moisture seem to be site
specific as the factors limiting water uptake by plants and microbes may
differ by ecosystem. Even in the same climate region, different soil moisture
thresholds have been found in previous studies. For example, Almagro
et al. (2009) investigated how soil moisture modulated the sensitivity of
soil respiration in different ecosystems in the Mediterranean region and
found that the threshold value of soil moisture was 10 %. Above this soil
moisture values, Q10 ranged from 1.86 to 2.20 and decreased to 0.44 to
0.63 when soil moisture was lower than 10 %. However, Rey et al. (2002)
found in a Mediterranean oak forest that soil temperature accounted for
85 % of the variation of SR when soil moisture was above 20 % with
a Q10 value of 2.34. Furthermore, Xu and Qi (2001) found that with soil
moisture higher than 14 %, the Q10 value was 1.8 and decreased to
1.4 when soil moisture was lower than 14 %.
Other factors affecting SR
In addition to soil moisture threshold values, we also found
variations of SRH and SRtot
between location and tree species in each soil moisture interval. For
example, when SM5 was lower than 20 %,
SRH measured at L4 was always lower than
SRH measured at L2 and L3. When SM5
was lower than 17 %, SRtot of P. nigra
was significantly higher than for the other two species, suggesting
that there are still other factors affecting SRH
and SRtot variations. Several explanations for this
result are plausible. First, spatial variability in vegetation can
affect SR due to differences in root respiration and the quantity and
quality of detritus (Raich and Tufekcioglu, 2000). These biophysical
gradients across landscape positions can lead to strong spatial
heterogeneity in SR. Tree species in our study site exhibit different
litterfall temporal patterns and may also contribute to the seasonal
variation of the availability of SOC and nutrients to the microbial
community and roots. Second, the vitality of tree species in
responses to soil water regime could generate different root
respiration rates. Additional data of daily sap flow of the studied
trees from our study site (data not shown) confirmed the differences in
tree transpiration and growth activity. For example, the water use
efficiency of P. nigra was highest, followed by
F. excelsior and A. glutinosa. Additionally, the mean
diameter at breast height (DBH) of P. nigra is larger than
mean DBH of the other tree species. P. nigra may be more
efficient in taking up water and nutrients compared to the other two
tree species.