Partitioning_Nov-30

advertisement
1
Partitioning ecosystem respiration between plant and
microbial sources using natural abundance stable carbon
isotopes: A study of four California ecosystems
Kevin P. Tu* and Todd E. Dawson
Center for Stable Isotope Biogeochemistry,
Department of Integrative Biology,
University of California at Berkeley, CA 94720, USA
* corresponding author: (510) 642-1054
email: kevinptu@gmail.com
2
ABSTRACT
Partitioning plant and microbial respiration is important for understanding the sources
and therefore the mechanistic basis of ecosystem respiration, as each can respond to
changes in environmental conditions in different ways and at different timescales. Natural
abundance carbon isotope ratios (13C; ‰) of plant and microbial respiration can be used
to partition their respective contributions to ecosystem respiration if their isotopic
differences can be resolved and isotopic mass balance among ecosystem respiration
components is conserved. While resolvable differences in the 13C of plant and microbial
respiration fluxes have been observed, partitioning efforts have been confounded by the
lack of isotopic mass balance at the ecosystem scale. We examined the influence of
spatial and temporal variability of respiration 13C signals on isotopic mass balance and
respiration partitioning by characterizing the 13C of CO2 respired from major ecosystem
components including ecosystem, belowground, leaf, rhizosphere, litter, and soil organic
matter (SOM) sources at different spatial scales and time periods in four contrasting
ecosystem types in California; coastal redwood forest, annual grassland, oak savanna and
montane pine forest. We found consistent differences in the 13C of plant, microbial and
ecosystem respiration across the different ecosystems, with SOM decomposition enriched
in 13C relative to leaf respiration by 2-5‰. Leaf respiration differed between night and
day by as much as 4-6‰ and with height in a tall redwood crown by more than 4‰. As a
result, isotopic mass balance was only observed when 13C signals were measured at the
same time of the day and when the spatial heterogeneity within the crown was accounted
for, indicating that careful consideration of the timing and location of sample collection is
critical to partitioning efforts using natural abundance stable carbon isotopes. Our
isotope-based partitioning indicated that belowground respiration accounted for 84% of
ecosystem respiration in the pine forest, 69% in the oak savanna, 57% in the redwood
forest and 37% in the grassland. Microbial respiration provided the majority of the
belowground respiratory flux: 84% in the pine forest, 83% in the grassland and 72% in
the redwood forest. The majority of this microbial respiration originated from litter
decomposition: 84% in the grassland, 75% in the pine forest and 55% in the redwood
forest. Below the litter layer, autotrophic and heterotrophic sources were generally in
3
balance, with rhizosphere respiration and SOM decomposition each averaging about 20%
of belowground respiration across these three sites. Our results indicate that partitioning
ecosystem respiration using natural abundance stable carbon isotopes is possible due to
the relatively large isotopic differences between leaf respiration and SOM decomposition,
but extra care must be taken to adequately characterize the spatial heterogeneity and
temporal variability of source signatures to ensure that isotopic mass balance is
conserved. The presence of isotopic disequilibrium between plant and microbial sources
and variation in plant respiration throughout the course of a day suggests significant postphotosynthetic fractionation effects due to biochemical, physiological, or ecological
processes from the time C is initially fixed in the leaf to the time it is respired by
microorganisms in the soil. Understanding the magnitude and causes of such isotope
effects are essential for robust interpretation of Keeling plot data across time and space
and for studies aimed at partitioning autotrophic and heterotrophic contributions to
ecosystem respiration.
Keywords: stable carbon isotopes, Keeling plots, respiration partitioning,
4
Introduction
Respiration plays a central role in the global carbon (C) cycle, as nearly half of all
C fixed through terrestrial photosynthesis is returned to the atmosphere as CO2 during
plant and soil microbial metabolism (REFERENCE). Autotrophic and heterotrophic
respiration can respond to changes in environmental conditions in different ways and at
different timescales, thus methods for quantifying plant and microbial respiration are
essential for developing a mechanistic understanding of the processes regulating
ecosystem C metabolism and its potential response to environmental change (Goulden et
al. 1998, Reichstein et al. 2002). Current micrometeorological methods allow direct
measurement of soil respiration (the soil surface CO2 efflux) and ecosystem respiration
(the CO2 flux of a whole ecosystem at night). However, both soil and ecosystem
respiration include autotrophic and heterotrophic sources and partitioning between the
two remains problematic (Hanson et al. 2000, Trumbore 2006, Carbone et al. in review).
Plant and microbial respiratory fluxes are typically characterized by measuring
root, stem, leaf, and soil respiration rates independently with chambers and solving for
microbial respiration as the difference between soil and root respiration (Law et al. 2001,
Xu et al. 2001, Unger et al. 2009). Except for leaf and stem respiration, these
measurements cannot be done in situ and involve considerable disturbance to the root and
soil system, including excising the roots from the soil. Tree girdling has been used in
forest ecosystems to partition plant and microbial respiration and this method alleviates
some but not all of the disturbance issues common to component chamber measurements
(e.g. Högberg et al. 2001, Subke et al. 2004). Radiocarbon 14C signals have also been
used to partition autotrophic and heterotrophic sources to belowground respiration
(Cisneros-Dozal et al. 2005, Schuur and Trumbore 2005, Carbone et al. 2007), but the
method is typically limited by high costs, low sample numbers as well as the need for
different 14C signals of autotrophic and heterotrophic respiration. Similar to the
radiocarbon approaches, methods based on natural abundance stable C isotopes provide
an alternative to chamber or girdling techniques that holds the promise of in situ
partitioning without disturbance effects (Tu and Dawson 2005). However, realization of
this promise has been limited to ecosystems that have experienced either a change in the
photosynthetic pathway of the dominant vegetation from C3 to C4 or visa versa (Robinson
5
& Scrimgeour 1995, Rochette and Flanagan 1997, Rochette et al. 1999) or that have been
exposed to labeled CO2 which is either highly enriched or depleted in 13C (Hungate et al.
1997, Lin et al. 1997, Andrews et al. 1999, Pendall et al. 2003, Bahn et al. 2009, Högberg
et al. 2010). Because these conditions are present only under special cases, methods for
flux partitioning based on natural abundance stable isotope signatures that could be
applied to a wide range of circumstances are highly desirable. Further, natural abundance
isotope-based partitioning methods used in conjunction with whole-ecosystem flux
measurements such as those routinely made within the FLUXNET eddy covariance
network could greatly expand our ability to quantify rates and controls on autotrophic and
heterotrophic respiration in different ecosystems and across a wide range of
environmental conditions.
In theory, natural abundance stable C isotopes can allow source partitioning when
the isotopic difference between the sources in question can be resolved (Tu and Dawson
2005). While the longstanding notion is that such isotopic differences among respiration
sources do not exist (Cerling 1991, Cheng 1996, Amundson et al. 1998, Lin et al. 1999),
a growing number of field studies have shown significant differences among natural
abundance carbon isotope ratios (13C) of respiration from different ecosystem
components that suggests partitioning may be possible (Bowling et al. 2003, Formánek
and Ambus 2004, McDowell et al. 2004, Mortazavi et al. 2005, Kodama et al. 2008,
Marron et al. 2009, Unger et al. 2010). However, partitioning efforts have been
confounded by the lack of isotopic mass balance at the ecosystem scale. Conservation of
mass requires that the isotopic composition of ecosystem respiration lies between that of
its contributing sources such that the 13C of ecosystem respiration equals the sum of the
flux-weighted 13C of all contributing sources. Inversion of the isotope mass balance
equation is the basis for partitioning among respiration sources, therefore mass balance is
required for successful partitioning.
Ecosystem respiration has typically been observed to be more enriched in 13C than
both soil respiration and leaf respiration (e.g., Bowling 2003, McDowell et al. 2004),
violating the necessary condition of isotopic mass balance. Ecosystem respiration has
also been observed to be more depleted than both soil and leaf respiration (Unger et al.
2010), similarly violating the necessary condition of isotopic mass balance. Since
6
isotopic mass balance must exist, these results suggest that the methods used for
characterizing the source signatures (e.g., belowground or aboveground respiration) or
the mixture itself (e.g., ecosystem respiration) are not yet reliable and require further
development. In addition to methodological uncertainties (e.g. Bowling et al. 2003,
Bowling et al. 2008), spatial and temporal variability in the 13C of respiration signals
from different ecosystem components may be confounding efforts to achieve isotopic
mass balance.
Recent field studies have shown large diel variations in the 13C of leaf (Hymus et
al. 2005, Prater et al. 2006) and ecosystem respiration (Bowling et al. 2003, Knohl et al.
2005, Werner et al. 2006). This variation is consistent with diel variations in the 13C of
soluble organic matter in leaves, stems and phloem sap (Gessler et al. 2008, Saveyn et al.
2010), the signal of which may transfer to the 13C of respiration either directly (Lin and
Ehleringer 1999) or indirectly (Kodama et al. 2008, Priault et al. 2009, Werner et al.
2009). As a result of this temporal variability, isotopic mass balance among ecosystem
respiration sources can only be ensured when all respiration signatures are determined at
the same time, and is not likely to be observed when respiration signals are compiled
from different times of the day. Similarly, isotopic mass balance is not likely to be
observed when the spatial heterogeneity of respiration, for example with leaf position
within a canopy or with soil conditions across the landscape, is not adequately
characterized.
We examined the influence of spatial and temporal variability in the 13C of
respiration on isotopic mass balance and respiration partitioning by characterizing the
13C of ecosystem respiration and its main components including belowground, leaf,
rhizosphere, litter, and soil organic matter (SOM) sources at different spatial and
temporal scales in four contrasting ecosystem types in California; coastal redwood forest,
annual grassland, oak savanna and montane pine forest. These field sites represented
different ecosystem types native to California and provided the opportunity to examine
similarities or differences based on species, life-form and prevailing climate and soils.
Materials and methods
Study Sites
7
Plant, soil and air samples were collected at four field sites located across a
precipitation gradient from the coast to the Sierra Nevada mountains of California that
spanned a range of plant functional types (Table 1). The regional climate is characterized
as Mediterranean with cool wet winters and hot dry summers. The coastal redwood forest
receives summer fog water inputs equal to roughly 35% of the mean annual precipitation
(Dawson 1998) whereas the inland Central Valley grassland and savanna sites receive
negligible fog inputs (Corbin et al. 2005). The inland grassland and oak savanna
experience large water deficits during the hot and dry summer months when evaporative
demand exceeds available water (PET>PPT in Table 1). The Sierra Nevada pine forest
receives similar rainfall as the redwood forest but lacking the fog water inputs of the
coast and with the hot summer temperatures of the Central Valley, its water deficit is
between that of the coastal and inland sites (Table 1).
Partitioning Approach
Ecosystem respiration was partitioned between plant and microbial sources using
a hierarchical approach as shown diagrammatically in Figure 1. Ecosystem respiration
(Reco) was expressed as the sum of aboveground (Rabove) and belowground (Rbelow) fluxes:
Reco  Rabove  Rbelow
and multiplied by their respective ecosystem ( δeco ), aboveground ( δabove) and
belowground ( δbelow ) isotope ratios following the conservation of mass (e.g. Bowling et
al. 2001) to give:
δeco Reco  δaboveRabove  δbelowRbelow
By rearranging, we solved for the aboveground fraction of ecosystem respiration:
f above 
( eco   below)
( above   below)
(1)
Microbial respiration was assumed to represent the decomposition of soil organic matter
(SOM). We did not attempt to separate root and rhizosphere microbial respiration, which
we assumed to have indistinguishable isotope ratios (Högberg et al. 2010). We therefore
treated the two together as rhizosphere respiration (Rrhiz). Belowground respiration
8
(Rbelow) was therefore expressed as the sum of three potential fluxes, rhizosphere (Rrhiz),
litter (Rlitter) and SOM (RSOM):
Rbelow  Rrhiz  Rlitter  RSOC
Expanding based on isotopic mass balance gives
 belowRbelow   rhiz Rrhiz   litter Rlitter   SOM RSOM
 below   rhiz f rhiz   litter f litter   SOM f SOM
where f rhiz , f litter and f SOM are the fractions of belowground respiration originating from
rhizosphere respiration, litter decomposition and SOM decomposition, respectively.
Given the three potential sources in the above equation and only one isotope, it was not
possible to partition rhizosphere, litter and SOM sources using a traditional two-source
mixing model. We therefore examined the range of possible partitioning outcomes given
the constraint of isotopic mass balance following the approach of Phillips and Gregg
(2003). We first eliminated one of the three unknowns by expressing it as a residual of
the other two:
f rhiz  1  flitter  f SOM
(2)
Substitution and rearranging for fSOM gives
f SOM 
 below  f litter litter   rhiz  f litter rhiz
 SOM   rhiz
(3)
Since flitter is not known, we considered the range of values of flitter that satisfied the
condition of isotopic mass balance. We then used these values of flitter to solve for fSOM
using Equation (3) then solved for frhiz using Equation (2).
In summary, ecosystem respiration was partitioned among six respiration sources
using measurements of the 13C of CO2 respired from six sources (Figure 1); ecosystem
(  eco ), belowground (  below ), aboveground (  above), rhizosphere (  rhiz ), litter (  litter ) and
soil (  SOM ). The methods for determining the isotope ratio of each component is
described in the following sections.
The Carbon Isotope Ratio of Ecosystem Respiration
The isotopic composition of whole ecosystem respiration was determined using
the ‘Keeling plot’ approach (Keeling 1958), as the intercept of a linear regression relating
9
the isotope ratios (13C) of air samples to the inverse of their CO2 concentrations (Pataki
et al. 2003). In the redwood forest, pine forest and oak savanna, air samples were
collected in 12 mL glass Exetainer vials by pulling air from various heights within and
above the canopy through Bev-A-Line IV tubing at 300 mL min-1 through the vials using
a double-holed needle inserted through the butyl-rubber septum (Tu et al. 2001). After
sufficient time to flush the tubing plus vial volume with the sample air, the flow was
stopped downstream of the vial and the pressure was allowed to equilibrate to ambient for
three seconds with the upstream sample air. Intercepts of all Keeling plots were
calculated using both ordinary least squares (OLS) and geometric mean regression (GM)
(see Zobitz et al. 2006). Outliers were removed by modifying the approach of Bowling et
al. (2002), by first calculating the regression line, calculating the standard deviation (SD)
of the residuals, removing the point farthest from the regression line that also exceed
three SDs of the residuals, then recalculating the regression and repeating this process
until all points fell within three SDs or until the standard error of the intercept (SE), as
calculated from the OLS regression (Pataki et al. 2003, Zobitz et al. 2006), was equal to
or less than our measurement precision of 0.05‰. Using this procedure, an average of
14% of the data points from each Keeling plot were excluded with a resulting mean r2 =
0.98.
In the redwood forest, air samples were collected from various heights near predawn at heights ranging from the tree top to above the soil surface through tubing
attached to a pulley affixed near the top of the tree. The exact sample height was adjusted
at the time of sampling to maximize the CO2 concentration gradients. In the oak savanna,
air samples were collected about every two hours at three heights from a tower
approximately located above the soil, mid-canopy, and above the canopy, with the height
adjusted to maximize the CO2 gradient. In the pine forest, eight air samples were
collected near pre-dawn at one height (~2m), as access to other heights was not available.
In the grassland, Keeling plot air samples were collected from the headspace of a 1L dark
airtight plastic chamber containing an intact ‘ecosystem’ core of about 7.5 cm diameter
and 10 cm depth. These cores were kept intact and therefore included all aboveground
and belowground components of the ecosystem. Two cores were collected at midday and
kept intact in the dark for at least 4 hours prior to sampling in the laboratory. After
10
sealing in the dark chamber, five air samples were sequentially sampled from the
headspace at intervals of 5-10 minutes by withdrawing 60ml of sample air and
simultaneously introducing 60ml of air from syringes previously filled with the same
background air used to initially flush the headspace of the chamber. The air sample was
then introduced into a 12-mL septum-capped vial by injecting through the septum while a
second needle was used as a vent. The 60mL of sample air flushed the vial with sample
air about six times over and effectively purged any air that was initially in the vial. The
typical range of CO2 concentrations achieved in the chamber over the course of collection
was ~350 ppmv. Leaks were negligible as there was no detectable change in the 13CCO2 in the headspace of an empty chamber (at ambient 13C-CO2) during a typical 10
minute sampling period. Self-closing quick-connect couplers with O-ring seals (ColeParmer) were used for all connections on the syringes and injection needles. Blanks were
not tested because the minimum detection limit of the mass spectrometer (~50 ppmv
CO2) was too low to resolve leaks of significant magnitude. However, using air of known
isotopic composition, no leaks or isotopic effects associated with this vial-filling method
or with the syringes and quick-connect couplers were detected. Leaks associated with the
septum-capped vials were assumed negligible based on previous tests (Tu et al. 2001) as
all samples were analyzed within 48 hours and air pressure differences between the field
sites and the laboratory were minimal.
The Carbon Isotope Ratio of Ecosystem Components
The isotopic composition of CO2 respired from belowground, SOM and litter
decomposition, leaves, and rhrizosphere was determined using two methods; Keeling
plots in chambers with ambient air and incubations in syringes with CO2-free air. To
ensure comparability between the Keeling plot and syringe incubation methods, we
compared respiration signatures using both techniques on two leaf and two stem samples
(Figure 2). The two methods appear to provide similar results as the slope and intercept
of the regression line was not significantly different from 1 and 0, respectively (P=0.05,
r2=0.991). Keeling plots were used for all samples collected from the grassland using the
dark 1L plastic chamber described above with the exception that soil, leaves, roots, litter
and root-free soil were placed in the chamber rather than intact cores. Keeling plots were
11
also used for belowground respiration in the savanna, redwood forest and pine forest,
with air samples collected from the headspace of a dark chamber placed on the soil
surface. Samples were withdrawn from this chamber with 60 mL syringes and then
transferred to Exetainer vials by flushing using two syringes, one for injecting, one for
venting. The 60mL of sample air flushed the vial with sample air about six times over
and effectively purged any background air that was previously in the vial (Tu et al. 2001).
In the redwood forest, air samples were collected during pre-dawn hours from the
headspace of a 22L opaque PVC chamber that was sealed around the edges with small
sandbags. In the pine forest, air samples were collected seven different times during the
night using from a modified LI-COR 6400-9 soil respiration chamber as described by
Torn et al. (2003). In the oak savanna, air samples were withdrawn from the headspace of
a 170 L darkened chambers that were clamped onto collars that were inserted into the soil
several months prior to sampling. For all other components in the savanna, redwood
forest and pine forest, incubations in CO2-free air within syringes were used. The syringe
incubation method is the same as that described by Tu and Dawson (2005, see also
Werner et al. 2007). Briefly, samples were first placed in a 60mL plastic syringe and CO2
was then scrubbed from the headspace by pumping air repeatedly through a soda lime
column (~5 times) with the syringe plunger. Next, after a sufficient amount of time for
the CO2 concentration to reach near-ambient levels (generally after 5-15 minutes
depending on the respiration rate of the sample), a subsample was injected into a 12-mL
septum-capped vial by flushing the entire contents of the 60mL syringe through the vial,
effectively purging any air that was initially in the vial. Previous studies have shown that
the use of CO2-free air does not appear to affect leaf dark respiration signatures
(Ghashghaie et al. 2003, Xu et al. 2004). We attempted to minimize fractionation effects
for soil samples when CO2 diffuses out of the soil air spaces into the headspace of the
syringe by collecting all the air within the syringe by compressing the soil with the
plunger to removing any residual air within the soil. For both Keeling and syringe
incubation methods, leaves, roots and stems were detached from the plant and sampled
after ~10 minutes to avoid wound responses. Leaves that were collected during the day
were placed in the dark for at least 15 minutes before sampling to avoid isotopic effects
related to light enhanced dark respiration, LEDR (Barbour et al. 2007). Detaching the
12
leaves does not appear to affect dark respiration signatures (Prater et al. 2005) and this
was assumed to be true for roots as well (Wegener et al. 2010). Rhizosphere respiration
was collected from samples that were detached from the plant just prior to placing them
into the chamber or syringe. Rhizosphere respiration here refers to CO2 respired by the
plant root plus that respired by any microbes attached to the root. Respiration from SOM
decomposition was determined on root-free soil samples collected from below the litter
layer after removing all visible live roots.
Sun leaves near the ground (~2m) were used to estimate the whole-canopy
respiration signatures at the Sierra Nevada pine forest because of limited access to
different heights in the canopy. At the oak savanna site, sun leaves were collected and
assumed representative of the canopy because of their short stature and low leaf area
indices. Given the large variation in leaf respiration with height within the redwood
canopy, canopy respiration signatures were estimated as the LAI-weighted mean of leaf

δ 13Cleaf,z  LAI z 

0
Ccanopy 
where z is the height (m) and
z
LAI
0 z
z
respiration isotope ratios: δ
13
LAIz and 13Cleaf,z are the leaf area index LAI (m2 leaf/m2 ground), and 13Cleaf at height z,
respectively. Measurements of the 13C of leaf respiration were made at four heights;
upper (46m), middle (35m), lower (14m) and understory (2m). A logistic function was
found to fit these data well (Figure 3; r2=0.98, RMSE=0.16‰):
δ 13Cleaf,z  1.24 ln z   30.67 . LAIz was estimated as the difference in cumulative LAI
between two consecutive heights: LAI z   LAI z   LAI z 1 . Cumulative LAI was
estimated from the transmittance of photosynthetically active radiation (PAR) measured
with a quantum sensor at 22 heights within the canopy (  
PARz
) using Beer’s Law
PARtop
1
as  LAI z   ln  z . PAR was measured at 22 heights within the canopy (data not
k
shown; Steve Burgess, personal communication) and the following function was used to
estimate its variation with height z in the canopy: z=max (0.04518z-1.80415, 0.00458z
+0.02949 (r2=0.99, RMSE=0.019). For this purpose, the exact value of k, the light
extinction coefficient, is irrelevant because we only required the relative rather than
13
absolute distribution of leaf area. We did not have information on the difference between
leaf and plant area index but assumed that their relative distributions were similar.
Stable Isotope Analyses
Carbon isotope ratios of CO2 in the vials (13CPDB) were determined using a gasphase continuous-flow isotope ratio mass spectrometer (Finnigan MAT Delta+ XL;
Thermo Instruments, Breman Germany) interfaced to a PAL80 autosampler coupled to a
GasBench II, as described by Tu et al. (2001). Per vial measurement precision was
±0.05‰, as measured by the standard error (SE = SD
n ) of six replicate analyses from
sample vials containing a known standard. Water vapor was not removed while filling the
vials but was subsequently removed from the sample air stream (in helium carrier gas)
using an in-line NafionTM diffusion trap in the GasBench.
Carbon isotope ratios of organic matter were measured with a model 20-20
isotope ratio mass spectrometer (PDZ Europa Scientific. Manchester UK). All isotope
analyses were done at the Center for Stable Isotope Biogeochemistry, U.C. Berkeley
(http://ib.berkeley.edu/groups/biogeochemistry).
Uncertainty of Partitioning Estimates
Uncertainties of all partitioning estimates are presented as the SE using the
method of Phillips and Gregg (2001). These values include the uncertainties related to
both the isotopic analyses (SE = ~0.05‰) and the use of two-source mixing models, of
which the latter depend on the number of samples and isotopic difference between end
members. When two partitioning estimates were averaged, the propagation of the
uncertainties was determined by adding the uncertainties in quadrature,
U
U
2
a
 Ub
2
 , where U is the uncertainty.
RESULTS
Carbon Isotope Ratios of Plant, Microbial and Ecosystem Respiration
The 13C of CO2 respired from various ecosystem sources at each site for predawn hours is shown in Figure 4. The greatest within-site differences among sources was
found in the pine forest (5.9‰) followed by the grassland (4.9‰), oak savanna (4.7‰),
14
and redwood forest (2.4‰). The decomposition flux from SOM and rhizosphere
respiration tended to be the most 13C enriched sources at all of the sites and leaf
respiration the most 13C depleted. Thus, the within-site differences tended to be between
leaves and heterotrophic sources such as SOM decomposition, or between leaves and
heterotrophic root tissues of the plant. Belowground respiration was always more 13C
enriched than aboveground sources with differences between aboveground and
belowground respiration ranging from 3.9‰ in the pine forest, 1.9‰ in the redwood
forest to 1.6‰ in the oak savanna. Among microbial sources, respiration from
decomposition of SOM was always more 13C enriched than litter, by 4.4‰ in the oak
savanna, 2.8‰ in the pine forest to 1.2‰ in the redwood forest.
Isotopic mass balance was observed at all the sites. That is, the 13C value of the
CO2 respired from the entire ecosystem was between that respired by its potential
sources, namely aboveground and belowground respiration. Further, belowground
respiration was bounded by its potential sources that included litter, SOM, and
rhizosphere respiration (Figure 4).
The relationship between the 13C of bulk carbon and respired CO2 from different
ecosystem components is shown in Figure 5. Bulk carbon 13C explained 39% of the
variation in the 13C of respired CO2. Rhizosphere, soil and litter respiration were all
more enriched in 13C relative to bulk C, whereas leaf respiration was sometimes more
enriched by 3.9‰ and sometimes more depleted by 2.4‰.
Temporal Variation in the Carbon Isotope Ratios of Plant, Microbial and Ecosystem
Respiration
Diel measurements in the pine forest (Figure 6a) and oak savanna (Figure 6b)
indicated large temporal variation in the 13C of plant respiration. In the pine forest,
differences between night and day were as large as 4.3‰ for leaf respiration and 2.2‰
for rhizosphere respiration. Leaf and rhizosphere respiration tended to be similar during
the day but diverged at night, when leaf respiration became more 13C depleted and
rhizosphere respiration become more enriched (Figure 6a). The greatest leaf-rhizosphere
difference occurred at pre-dawn, when leaf respiration was depleted by 5.9‰ relative to
rhizosphere respiration. Belowground respiration varied by only 0.5‰ from day to night,
15
and this variation was strongly correlated with rhizosphere respiration (r2= 0.94, inset
graph in Figure 7). The 13C of ecosystem respiration was determined only during predawn hours at which time isotopic mass balance was observed, as its isotope ratio was
between that of aboveground and belowground respiration, and belowground respiration
was between that of rhizosphere and microbial sources in the soil.
In the oak savanna, there was large temporal variation in the 13C of leaf and
belowground respiration (Figure 6b). Differences between night and day were as large as
4.9‰ for leaf respiration and excluding one value (hr 1330 on JD 228), belowground
respiration tended to be exhibit less diel variation with a range of about 2‰ (5.5‰ with
this value included). Soil respiration also tended to be enriched in 13C relative to leaf
respiration. The greatest difference between the two occurred pre-dawn, when leaf
respiration was depleted by 6.7‰ relative to soil respiration. Isotopic mass balance was
observed, indicated by the fact that the 13C of ecosystem respiration was between that of
aboveground and belowground respiration, although the uncertainty of the ecosystem
respiration values was relatively large, as indicated by standard errors ranging from 2-6‰
over the course of the day. This uncertainty was consistent with that found by Pataki et al.
(2005) for the observed CO2 gradients of 10-20 ppmv. Based on this uncertainty it would
be difficult to statistically distinguish ecosystem respiration from aboveground and
belowground sources, although the means and the trends are consistent with isotopic
mass balance. Due to this diel variation, isotopic mass balance would not always be
conserved if component 13C values were compiled from different times of the day.
Due to the diel variation of the isotope ratios of the different ecosystem sources
shown in Figure 5, isotopic mass balance would not be found if isotope signatures were
combined from different times of the day. For example, at the pine forest site, the predawn ecosystem respiration value of -25.7‰ would be more negative than any other
possible respiration source during the day except for litter decomposition (Figure 6a).
Similarly, at the oak savanna site, the pre-dawn value of ecosystem respiration of around
-26‰ would be much more negative than leaf or belowground soil respiration during the
day, both of which range from -22 to -24‰ during the day. In contrast, isotopic mass
balance was found when isotope signatures were compared from the same time of the
day, as evidenced by the fact that the 13 C of ecosystem respiration was between that of
16
its above and belowground sources at any given time in both the pine forest and oak
savanna sites. Further, in the pine forest, belowground respiration was between that of its
respective, root and microbial sources ( δ leaf < δ eco < δ below < δ mic ).
Spatial Variation in the Carbon Isotope Ratios of Leaf Respiration
We found significant spatial variation of 4.1‰ in the 13C of leaf respiration with
height in the redwood tree crown, with the most 13C-depleted values at the bottom and
most enriched at the top (Figure 3). Due to this variation, isotopic mass balance cannot
be ensured if canopy respiration was characterized using measurements at only one
height.
Partitioning Plant and Microbial Respiration Using Stable Carbon Isotopes
Partitioning estimates during pre-dawn hours are shown in Figure 8 for the
redwood forest, grassland and pine forest. These estimates are based on the 13C values
of the component respiration sources shown in Figures 3 and Equations 1-11. Ecosystem
respiration was partitioned between aboveground and belowground sources with
belowground accounting for 5725% (meanrange) of ecosystem respiration in the
redwood forest, 3747% in the oak savanna and 84 12% in the pine forest.
Belowground respiration was further partitioned between rhizosphere and nonrhizosphere microbial (saprotrophic) respiration. Microbial respiration accounted for the
majority of the soil CO2 efflux at all three sites, 7226% in the redwood forest, 8316%
in the oak savanna, and 8415% in the pine forest. Microbial respiration was further
partitioned between litter and SOC decomposition, with litter decomposition exceeding
SOC decomposition at all three sites. Litter decomposition comprised 5531% of
microbial respiration in the redwood forest, 8412% in the oak savanna, and 7521%= in
the pine forest.
Plant respiration accounted for 5915% (meanrange) of ecosystem respiration in
the redwood forest, 696% in the oak savanna, and 2913% in the pine forest. Further,
the root/shoot ratio of plant respiration, calculated as the ratio of rhizosphere and
aboveground contributions to ecosystem respiration, was 0.4 in the redwood forest, 0.1 in
17
the oak savanna and 0.9 in the pine forest. The shoot contributions were therefore nearly
twice that of roots in the redwood forest, 10 times greater in the oak savanna and nearly
equal in the pine forest.
DISCUSSION
Early research using natural abundance carbon isotope ratios of respiration from
plants and microbial sources suggests that their differences cannot be reliably resolved or
useful for partitioning between them (Cerling 1991, Cheng 1996, Amundson et al. 1998,
Lin et al. 1999). However, actual field measurements made to support this notion have
only been collected in a few studies and for the most part the data were inconclusive
(Bowling et al. 2003, McDowell et al. 2004, Mortazavi et al. 2005, Werner et al. 2006,
Unger et al. 2010). In these cases, differences in the carbon isotope ratios of respiration
from different ecosystem components appear resolvable, however the values do not
consistently satisfy the necessary condition of isotope mass balance. Specifically,
belowground and aboveground respiration together comprise ecosystem respiration and
mass balance dictates that the isotopic signal of ecosystem respiration must lie between
that of its belowground and aboveground respiration sources. Yet, ecosystem respiration
is typically found to be more enriched in 13C than both belowground soil respiration and
aboveground leaf respiration. Possible explanations have been suggested, ranging from
‘scaling effects’ (e.g., measurements made at one scale cannot be directly applied at
another scale; Xu et al. 2004) to sampling artifacts (Bowling et al. 2003). Regardless, it is
clear that the methods for characterizing the source signatures (i.e. belowground or
aboveground respiration) or the mixture itself (i.e. ecosystem respiration) are not yet
reliable and need further development. Our results indicate that reliable characterization
of source signatures in the context of isotopic mass balance also requires careful
consideration of the temporal (diel) and spatial (within crown) variability of the isotope
signals. Due to the fact that the isotopic signals of different respiration sources can
change throughout the day (Figure 6) or with height within tall crowns (Figure 3), mass
balance can only be ensured when isotopic signals from all ecosystem components are
measured at the same time and/or when their spatial variation is accounted for.
Measurements in the pine forest and oak savanna (Figure 6) clearly indicate that isotopic
18
mass balance would not be observed if isotope values were compiled from different times
of the day. Further, measurements along the height gradient in ~65m-tall redwood crown
indicate that isotopic mass balance would not be conserved if the isotope ratios of canopy
respiration were represented by measurements on only the uppermost sun leaves.
Although simultaneous measurements of ecosystem components are important, it
does not guarantee isotopic mass balance. For example, Unger et al. (2010) did not find
isotopic mass balance despite measuring all major ecosystem components throughout the
day. As they noted (Unger et al. 2010), difficulties in characterizing the isotopic
signatures of the ecosystem and its respiration sources must still be overcome, such as
those related to the spatial heterogeneity in leaf and soil respiration and different source
footprints when collecting ecosystem Keeling plot samples at different heights or at
different times.
Temporal Variability in Ecosystem Respiration Sources
Temporal variability in ecosystem respiration appears to be attributed primarily to
changes in plant respiration rather than microbial respiration (Figure 6). Leaf respiration
in both the pine forest (Figure 6a) and oak savanna (Figure 6b) exhibited significant 13C
depletion of 4-5‰ during the night with subsequent enrichment during the day. Recent
studies have also found enrichment of leaf respiration of a similar magnitude during the
day (Hymus et al. 2005, Prater et al. 2006, Werner et al. 2009) or following periods of
darkness when leaves are exposed to light (Tcherkez et al. 2003). In contrast, 13C
depletion of leaf respiration has also been measured when leaves are placed in the dark
(Park and Epstein 1961, Tu and Dawson 2005, Barbour et al. 2007), although not all
species exhibit this daytime enrichment (Werner et al. 2007). Further, these studies are
consistent with observations of diel variations in the carbon isotope composition of
soluble organic matter in leaves, stems and phloem sap (Gessler et al. 2008), the isotopic
signal of which should transfer to the respired CO2 (Lin and Ehleringer 1999). A
mechanistic explanation for this daytime enrichment of leaf respiration may lie in
fractionation that occurs during secondary metabolism (Tcherkez et al. 2003). Secondary
compounds such as fatty acids and amino acids and derivatives such as lipids, proteins,
isoprenoids, and lignin are preferentially incorporate the lighter 12C of carbohydrate
19
substrates (e.g. photosynthates), while the heavier 13C is preferentially respired and not
incorporated into metabolites. As a result, daytime 13C enrichment of leaf respiration is
expected to vary with the rate at which 12C is retained in secondary compounds, which
appears to vary with light availability and photosynthesis (Tcherkez et al. 2003, Hymus et
al. 2005, Prater et al. 2006) as well as plant functional type related to growth form and
growth rate (Priault et al. 2008).
Although the temporal variation in microbial respiration was not measured,
results from the pine forest indicate its temporal variation was minimal, as rhizosphere
respiration explained most (94%) of the variation in belowground soil respiration (Figure
7). The opposing trends in rhizosphere and leaf respiration signals (Figure 5a) could be
due to the time-lag caused by the transit time of phloem moving to the roots. For typical
phloem flow rates of 0.5-1.0 m hr-1 (e.g. Peuke et al. 2001) and a tree height of ~20m at
this site, the time-lag would be ~10-20 hrs and could explain the opposing trends in
rhizosphere and leaf respiration, with the rhizosphere signal temporally offset from the
leaf signal. However, this does not explain the different magnitudes of the isotope ratios
of rhizosphere and leaf respiration, as rhizosphere respiration varied by ~2‰ whereas
leaf respiration varied by ~4‰. This isotopic difference could be due to fractionation
(Helle and Schleser 2004) or isotope effects related to metabolic branch points (Schmidt
and Gleixner 1998) before, during or after transport of the phloem sugars to the site of
respiration in the roots (Badeck et al. 2005, Brandes et al. 2006), consistent with the
typical enrichment of root relative to leaf bulk carbon isotopes (Hobbie and Werner 2004,
Badeck et al. 2005, Gottlincher et al. 2006, Gessler et al. 2008, Wegener et al. 2010). The
results shown here are consistent with the enrichment found in soluble root sugars seen
by Gleixner et al. (1993), in root starch and soluble sugars by Gottlincher et al. (2006)
and with the general pattern of root enrichment in 13C relative to leaf bulk carbon
(Badeck et al. 2005, Cernusak et al. 2009). Hobbie and Werner (2004) suggest that the
allocation of carbon to lignin and lipid pools in leaves could leave the remaining
carbohydrates that are transported to the roots relatively enriched in 13C thereby resulting
in enriched root biomass. Our findings are also consistent with the typical enrichment of
phloem relative to leaf sugars (Damesin and Lelarge 2003, Scartazza et al. 2004, Gessler
et al. 2008). However, during the synthesis of 13C-depleted compounds like lignin or
20
lipids the enriched carbon is likely to be respired rather than retained (Tcherkez et al.
2003). An alternative explanation is that the enrichment in phloem sugars arises from
remobilization of starch (Gessler et al. 2008), which are typically enriched relative to
sugars formed during the Calvin-Benson cycle of photosynthesis (Gleixner et al. 1998).
Badeck et al. (2005) further suggest that root-leaf isotopic differences could be explained
by dark CO2 fixation in the roots via PEPc (Raven and Farquhar 1990). Thus, the trends
we found in rhizosphere and leaf respiration shown in Figure 6a could be due to both a
~12 hour time-lag as well as post-photosynthetic fractionations.
Can Natural Abundance Carbon Isotopes be used to Partition Ecosystem Respiration?
The observed isotopic differences among ecosystem respiration components were
generally adequate for partitioning ecosystem respiration between plant and microbial
sources. However, at times, these differences were small, resulting in relatively large
partitioning uncertainties (Figure 8). For example, small differences between
aboveground and belowground respiration in the oak savanna of between 0.6-1.0‰
translated to relatively large partitioning uncertainty of 47%. Thus, natural abundance
carbon isotopes may not always provide reliable estimates of plant and microbial
partitioning.
The partitioning estimates shown in Figure 8 are consistent with partitioning
estimates from other studies based on eddy covariance measurements and chamber
fluxes. Our isotope-based estimate of 57% (± 25%) of ecosystem respiration originating
from belowground root and microbial sources in the redwood forest is similar to the
range of the 60-80% typically found for belowground respiration in other forest
ecosystems (Wofsy et al. 1993, Amthor et al. 1994, Goulden et al. 1996, Law et al. 2001,
Xu et al. 2001, Janssens et al. 2001). Similarly, our oak savanna estimate of 37 ± 47%
lies within the range of the 20-80% reported in the literature for oak savanna ecosystems
(Cernusca et al. 1978, Mathes and Schriefer 1985, Suyker and Verma 2001,
Franzluebbers et al. 2002). Further, our estimate of the average microbial contribution to
total ecosystem respiration across the sites (47 ± 21%, mean ± SD) is consistent with
chamber-based estimates of 30-35% (e.g. Law et al. 2001, Xu et al. 2001) and modeling
estimates of 20-50% (e.g. Frolking et al. 1997, Reichstein et al. 2002). On a site by site
21
basis, the redwood forest (41 ± 35%) and the oak savanna (31 ± 50%) are consistent with
the results of the aforementioned studies, while the estimates for the pine forest (71 ±
19%) are much higher. Finally, our estimate of the mean microbial contribution to
belowground respiration of 80 ± 7% (mean ± SD) is higher than the value of 52%
reported by Hanson et al. (2000) based on mid-season measurements in both forests and
non-forests using a variety of techniques and under a wide range of conditions. Working
within the same ponderosa pine site but in a different stand, Xu et al. (2001) report a
chamber-based estimate of 53%, lower than our value of 84 ± 15% (mean ± possible
range). Law et al. (2001) report a value similar to Xu et al. (2001) of 47% for a ponderosa
pine forest in Oregon. It should be noted that our estimates are only potential ranges
because we only had one isotope and three potential belowground sources (rhizosphere,
litter, SOC), thus the true estimate based on the isotope ratios can be anywhere between
61-100% microbial (the mean range across the sites), and we have no way to determine
which value within this range is the true value. Nevertheless, the isotope-based estimates
are higher than previous chamber-based estimates.
The average contributions of root and microbial respiration were similar across
the four sites (Figure 8). This similarity suggests an interdependence of root and
microbial metabolism, and is consistent with a direct dependence of microbial activity on
plant activity (Hogberg et al. 2001). As a result, variation in the microbial contribution to
total ecosystem respiration should parallel and will likely be driven by variation in root
activity. Given that plants exhibit seasonally variable root activity and growth, we expect
these values to change throughout the growing season, particularly as these California
ecosystem experience dynamic seasonal changes in water availability.
Can δ 13 C of Organic Matter be used as a Surrogate for δ 13 C of Respired CO2?
Many studies have found an apparent isotopic mass imbalances when comparing
13C signatures of respiratory fluxes to their respective source signatures inferred from C
stocks in plant or soil organic matter that preclude respiration partitioning (Bowling et al.
2003, Pataki et al. 2003, Scartazza et al. 2004, Barbour et al. 2005). Our results suggest
that such mass imbalances may result from the fact that the 13C of bulk tissue C is a poor
indicator of the 13C of the respiratory flux. This notion is consistent with previous
22
studies (Duranceau et al. 1999, Duranceau et al. 2001, Ghashghaie 2001, Ghashghaie
2003, Bowling et al. 2003, Tcherkez et al. 2003, Xu et al. 2004, Formánek and Ambus
2004, Tu and Dawson 2005) and with a recent review of carbon isotope variation in
plants and plant compounds (Cernusak et al. 2009). Our results shown in Figure 5, and
the previous discussion on the possible causes of temporal variability in leaf respiration,
support this conclusion as there is variability in the isotope ratios of respired CO2 which
is not present in bulk C isotope ratios.
The δ13 C of bulk SOC, belowground, and ecosystem organic C is not expected to
provide a reliable indicator of the δ13 C of respired CO2 for partitioning respiration
because the microbial contribution to the respiration from these components is rarely in
proportion to the relative contribution of microbial biomass to total soil C. That is, it is
not surprising that the δ13 C of microbial respiration differs from the δ13 C of SOC because
microbial biomass comprises only a small percentage of total SOC (<10%) while it may
dominate the respiration flux. In addition, microbes may selectively decompose the most
labile fraction of SOC that is also the most 13C-enriched, so it is unlikely that the δ13 C of
microbial respiration equals that of SOC (Ehleringer et al. 2000, Formánek and Ambus
2004). There is also evidence that microbes may fractionate during C uptake (Henn and
Chapela 2000, Henn et al. 2002) or during respiration (Santrukova et al. 2000b), further
contributing to differences between the δ13 C of SOC and microbial respiration.
The isotopic signatures were collected either pre-dawn or after an extended period
of time in the dark for the oak savanna samples (>4 hours) to avoid potentially
confounding effects related to enriched daytime leaf respiration, as found recently by
several studies (Hymus et al. 2005, Prater et al. 2005). Thus, pre-dawn may be the most
appropriate time to sample component respiratory sources (leaves, roots, soil, etc.) when
CO2 gradients between the atmosphere and inside the canopy are greatest and therefore
most suitable for Keeling plot analyses. If leaf respiration signatures become more
enriched during the day as reviewed and discussed by Cernusak et al. (2009) and
Bowling et al. (2009), then isotopic mass balace requires that ecosystem respiration
signatures must also become more enriched during the day, unless microbial respiration
signatures perfectly offset (i.e. vary inversely) with leaf respiration signatures, which is
23
unlikely. If true, this could have significant implications to studies that assume daytime
ecosystem respiration signatures can be estimated from nigttime Keeling plots.
CONCLUSIONS
We provide the first evidence that natural abundance stable carbon isotopes in
respired CO2 can be used to partition ecosystem respiration between plant and microbial
sources for a range of different California ecosystems. The largest isotopic differences
appear to be between leaf and microbial respiration and suggest significant isotope effects
due to biochemical, physiological, or ecological processes from the time C is initially
fixed in the leaf to the time it is ultimately respired by microorganisms in the soil.
Exploiting these natural 13C differences, we partitioned whole ecosystem respiration
between leaf, rhizosphere, litter, SOM decomposition sources. Our results not only
indicate that there are resolvable and adequate carbon isotope differences among plant
and microbial respiration sources for partitioning respiration, but that due to temporal and
spatial variability in these isotope signals it is critical for partitioning purposes that all
respiratory signatures be determined at the same time of day and that spatial
heterogeneity in component signatures are accounted for. Such variation in respiration
signals might explain the apparent isotopic mass imbalances observed in previous studies
when nighttime Keeling plots are combined with daytime measurements of respiration
from various ecosystem components such as leaves, soil, and stems.
Due to the small differences that may occur between belowground and whole
ecosystem 13C values of respiration, isotope-based partitioning of aboveground and
belowground respiration may not always be robust. Fortunately, alternative methods for
partitioning above and belowground respiration using chamber measurements of soil
respiration are well established. However, separating belowground respiration between
root and microbial sources using chamber measurements can be problematic. In this case,
due to the large differences between plant and microbial signatures which we found to
range between 2-5‰, partitioning belowground sources of autotrophic and heterotrophic
respiration should be more robust. The principle limitation to this approach is the
characterization of the root and microbial signatures.
24
While our results demonstrate the potential for using stable C isotopes in respired
CO2 for partitioning respiratory fluxes, future applications should include better
characterization of the spatial heterogeneity in plant and microbial sources and greater
sampling frequencies to reduce the uncertainies in partitioning estimates. Spatial
heterogeneity of soil conditions can be enormous and therefore improved methods need
to be developed to ensure representative root and microbial respiration signatures.
While the observed differences between plant and microbial sources are beneficial
for the purpose of partitioning respiration between autotrophic and heterotrophic
respiration, they challenge the assumptions of isotopic equilibrium inherent to many
terrestrial ecosystem based models. Further research is needed to develop a predictive
understanding of the causes of the observed plant-microbial isotopic differences under
different physiological and environmental conditions.
Acknowledgements
We thank Paul Brooks for his assistance with the isotope analyses, Francesca Ponti, Jia
Hu and Stefania Mambelli for field assistance, The A.W. Mellon Foundation and U.C.
Berkeley for financial support, numerous participants from the NSF-supported BASIN
working group for valuable feedback during the course of this work, and Alexander
Knohl and Christiane Werner for comments that improved the manuscript.
25
REFERENCES
Amthor et al. 1994. Testing a mechanistic model of forest-canopy mass and
energy Australian Journal of Plant Physiology 21:623-651.
Cernusak, L.A., G. Tcherkez, C. Keitel, W.K. Cornwell, L.S. Santiago, A. Knohl, M.M.
Barbour, D.G. Williams, P.B. Reich, D.S. Ellsworth, T.E. Dawson, H.G. Griffiths,
G.D. Farquhar and I.J. Wright. 2009. Why are non-photosynthetic tissues
generally 13C enriched compared to leaves in C3 plants? Review and synthesis of
current hypotheses. Functional Plant Biology 36: 199-213.
Dawson, T.E, S. Mambelli, A.H. Plamboeck, P.H. Templer and K.P. Tu. 2002. Stable
Isotopes in Plant Ecology. Annual Review of Ecology and Systematics 33: 507559.
Gleixner, G., H.-J. Danier, R.A. Werner, and H.-L. Schmidt. 1993. Correlations between
the 13C content of primary and secondary plant products in different cell
compartments and that in decomposing basidiomycetes. Plant Physiol., 102:
1287–1290.
Goulden et al. 1996 Measurements of carbon sequestration by long-term eddy
covariance: methods and a critical evaluation of accuracy. Global Change
Biology, 2
Running SW, Baldocchi DD,Turner DP, Gower ST, Bakwin PS,Hibbard KA.
1999. A global terrestrial monitoring network integrating tower fluxes,
flask sampling, ecosystem modeling and EOS satellite data. Remote
Sensing of Environment 70: 108–127.
Saveyn, A., Steppe, K., N. Ubierna and T.E. Dawson. 2010. Woody tissue
photosynthesis and its contribution to trunk growth and bud development in
young plants. Plant, Cell & Environment 33: 1949-1958.
Tu, K.P. and T.E. Dawson. 2005. Partitioning ecosystem respiration using stable carbon
isotope analyses of CO2. Pp 125-153 In: L.B. Flanagan, J.R, Ehleringer & D.E.
Pataki (eds) Stable Isotopes and Biosphere-Atmosphere Interactions: Processes
and Biological Controls. Elsevier Science, San Diego.
Tu, K, Brooks PD, Dawson TE. 2001. Using septum capped vials with continuous flow
isotope ratio mass spectrometric analysis of atmospheric CO2 for Keeling plot
applications. Rapid Communications in Mass Spectrometry 15: 952-956.
Wofsy et al. 1993 Net exchange of…Science 260:1314-1317.
Download