Directory UMM :Data Elmu:jurnal:S:Soil Biology And Chemistry:Vol33.Issue1.Jan2001:
Soil Biology & Biochemistry 33 (2001) 83±92
www.elsevier.com/locate/soilbio
Modeling the incorporation of corn (Zea mays L.) carbon from roots
and rhizodeposition into soil organic matter
J.A.E. Molina a,*, C.E. Clapp b, D.R. Linden b, R.R. Allmaras b, M.F. Layese b,
R.H. Dowdy b, H.H. Cheng a
b
a
Department of Soil, Water, and Climate, University of Minnesota, St. Paul, MN 55108, USA
USDA-ARS and Department of Soil, Water, and Climate, University of Minnesota, St. Paul, MN 55108, USA
Received 5 October 1999; received in revised form 17 April 2000; accepted 6 May 2000
Abstract
Experimental data reported in the literature over the last decennium indicate that roots and rhizodeposition are important sources of carbon
for the synthesis of soil organic carbon. Our objective was to verify the capability of the simulation model NCSWAP to reproduce the general
conclusions from the experimental literature, and to gain some insight about the processes that control the incorporation of corn belowground production into the soil organic matter. The model was calibrated against the experimental data gathered from a long-term ®eld
experiment located near St. Paul, Minnesota. The simulation model updated daily the soil conditions to reproduce over a 13 year period the
measured kinetics of seven variables: above-ground corn production, and the total soil organic matter, soil d value, and the soil organic
matter derived from corn in the 0±15 and 15±30 cm depth. The simulation gave a root-plus-rhizodeposition 1.8 times larger than stalks plus
leaves. The translocation ef®ciency of corn-C into soil organic C at the 0±15 cm depth gradually decreased to 0.19 of the below-ground
deposition. The sensitivity of below-ground photosynthate incorporation into the soil organic matter was analyzed relative to variations in the
parameters that control the formation and decay of roots and rhizodeposition. Roots had a greater effect than rhizodeposition on the soil
organic matter, though more photosynthates were translocated to rhizodeposition than to roots. q 2001 Elsevier Science Ltd. All rights
reserved.
Keywords: Root exudate; Rhizosphere; Natural C enrichment; Tracer C; Lignin; NCSOIL; NCSWAP
1. Introduction
Roots and rhizodeposition are important sources of
carbon for the synthesis of soil organic carbon (SOC). The
translocation of plant C into SOC is quantitatively documented with tracer C. For example, Johnen and Sauerbeck
(1977) working with 14C enriched mustard and wheat
showed that tracer C in soil at harvest was 20±50% higher
than that mechanically isolated from roots, and that only
20% of the 14C mineralized to CO2 originated from root
respiration. Experiments with 14C require elaborate methods
that limit measurements to plants maintained in growth
chambers for a short period. By contrast, the different
natural 13C enrichments in C3 and C4 plants facilitate the
estimation in situ of the long term transformation of corn
photosynthates into SOC (Cerri et al., 1985; Balesdent and
Balabane, 1996; Huggins et al., 1998; Clapp et al., 2000).
The amount of SOC originating from corn increases as
* Corresponding author. Tel.: 11-612-625-6259; fax: 11-612-625-2208.
E-mail address: jmolina@soils.umn.edu (J.A.E. Molina).
annual crops of corn are grown continuously. It nevertheless
represents only a fraction of the corn-C incorporated into the
soil as some is lost, mostly as CO2, during the decay
processes of the corn residues and the SOC of corn origin.
Experimental results summarized by Bolinder et al. (1999)
suggest that the percentage of the below ground corn-C
incorporated into SOC (range 16±30%) is higher than that
from the above ground corn biomass (range 7.7±20%). The
multiplicity and interdependence of C and N transformations make it dif®cult to appreciate the origin of the percentage range, as well as the predominance of below ground
over above ground contribution, without a process oriented
simulation model. Several computer models are available to
describe crop growth and nutrient cycling in soil, water and
plant (Molina and Smith, 1998). One such model,
NCSWAP, simulates below and above ground photosynthate input into soil layers. It distinguishes between
mechanically harvested roots and rhizodeposition. Total
and tracer C and N released from plant are incorporated
into the soil processes as computed with the model NCSOIL
(Molina et al., 1997). These two models have been tested
0038-0717/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0038-071 7(00)00117-6
84
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
(23 references in Current Contents 1993 to date), and are
registered with the Soil Organic Network of the Global
Change and Terrestrial Ecosystem of the International
Geosphere±Biosphere Programme.
A long-term ®eld experiment located near St. Paul was
initiated in 1980 to investigate the effect of N and tillage
management on the incorporation of corn residues into the
SOC. Information from this experiment documents the
origin of the SOC (corn or soil), and the partition of inorganic fertilizer N between the plant and the soil fractions
through the use of total and tracer elements (Clapp et al.,
2000).
Our objective was to gain some insight about the
processes that control the incorporation of corn belowground production into the soil organic matter. In this
paper, we focus on C transformations and a ®eld treatment (no above-ground residues returned, no fertilizer N
added) that limited inputs to the SOC from roots and
rhizodeposition. We ascertained the ability of NCSWAP
to reproduce the experimental data, and a sensitivity
analysis was performed to quantify the in¯uence of
roots, rhizodeposition, and soil C transformations on
the system over 13 years.
2. Materials and methods
2.1. Field experiment and measured data
Data were obtained from one of the 12 treatments Ð the
chisel-plough tillage with no inorganic N addition and
stover-residue plus grain removed Ð from a long-term
continuous corn cultivation (Zea mays L.) study, which
took place during the years 1980±1993 at the University
of Minnesota Research and Outreach Center, Rosemount,
MN (Clay et al., 1989; Clapp et al., 2000). The soil at the
site is a Waukegan silt loam (®ne-silty over sandy or sandyskeletal, mixed, mesic Typic Hapludolls). Soil depth to
gravel is about 100 cm, and the climate is sub-humid. Chisel
ploughing to a depth of about 17 cm was conducted
annually in the fall. Soil samples were taken from 0 to 15
and 15±30 cm layers, sieved (2 mm), the retained organic
debris ground and returned to the soil sample, and analyzed
for d 13C value. Thus, plant organic debris are included in
the SOC data. Above-ground corn production was measured
annually from 1980 to 1993 (Linden et al., 2000).
Speci®c ®eld management, analytical techniques, and
results were detailed in Clapp et al. (2000) and Linden
et al. (2000).
is parameterized for mass increase measured during a year
(the reference year, 1981) that gave maximum production
with minimal or no N and water stress. For other years on
the same site with the same crop, growth is adjusted daily to
account for the soil conditions (N and water availability)
and air temperature. The N and C transformations are simulated by the model NCSOIL included as a subroutine in
NCSWAP (Molina et al., 1983; Nicolardot et al., 1994;
Hadas et al., 1998). The soil organic matter is represented
in each 5 cm layer of the 100 cm deep soil pro®le by four
pools: labile and resistant Pool I, Pool II, and Pool III.
NCSWAP identi®es in each soil layer the following organic
debris: (1) roots and (2) rhizodeposition from the current
year; (3) roots and (4) rhizodeposition from the preceding
years; (5±7) three different above-ground crop residues; and
(8±11) four different organic amendments. Each debris is
described by two pools, each de®ned by its ®rst-order
decay-rate-constant and C to N ratio. NCSWAP requires
several input ®les: the initial soil conditions, and yearly
®les for the driving variables (crop, management, temperature, and precipitation). The soil conditions, roots and rhizodeposition included, are updated daily or every 0.2 d during
water in®ltration and redistribution. The source and executable code of NCSWAP and NCSOIL with the description
of the variables in the input ®les are available at the web site
(http://www.soils.umn.edu/research/ncswap-ncsoil/). Since
this work focuses on the dynamics of roots and rhizodeposition, some speci®cs are given in reference to the simulation
of the below-ground crop.
Root and above-ground daily mass increase are linked
by the shoot-to-root mass ratio (SR); SR R1 1 R2 £ t;
where R1 and R2 are constants and t is the time (d) from
emergence. Repartition of the root mass in the soil
layers is controlled by two parameters: the fraction
(R3) of the daily root mass increase that is added to
the top R4 layers of the pro®le. The parametric value
of R3 is de®ned for different physiological stages of the
crop. The left-over daily-root-mass increase is distributed below the R4th layer according to an algorithm that
considers the vertical depth of root penetration. Rhizodeposition is computed as R5 times the root mass
formed per day at each layer. Roots grow until they
start to decay at crop maturity or harvest, whichever
comes ®rst. Rhizodeposition decays as soon as formed.
Roots and rhizodeposition decay as a two-component
material (labile and resistant) with the rate constants
kl and kr (per day), modi®ed by reduction factors to
account for edaphic conditions.
2.3. d value and SOC from corn
2.2. Simulation model
The model NCSWAP simulates N and C Ð total and
tracer Ð transformations and the ¯ow of water in the soil
pro®le (Molina et al., 1997). The above-ground plant growth
is described by a generalized form of the logistic equation. It
Tracer and total C concentrations are simulated in parallel. At each time step the rate of change in tracer concentration in any pool (D tr) is given by the equation,
Dtr Dto £ E £ m; where D to is the rate of change in
total concentration, E 13 C=12 C 1 13 C is the tracer
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
85
Fig. 1. Root mass distribution in the soil pro®le at harvest. Simulated for the 1984 season and measured during the 1986 season.
concentration at the beginning of the time step, and m
the discrimination factor, set to 1.0 in this study. Tracer
and total concentrations are both updated, and the
process repeated. Simulations were performed with the
same initial d 13C assigned to the resistant and labile
corn-debris pools.
The initial SOC- 13C tracer concentrations were set to
their measured values against the Pee Dee Belemnite
standard Rpdb 13 Cpdb =12 Cpdb 0:0112372 : Es
13 Cs =12 C 1 13 Cs 0:0108926 corresponding to a
and
Es
d13 Cs Rs =Rpdp 2 1 1000 219:99½;
0:0109157 or d13 Cs 217:89½ in the 0±15 and 15±
30 cm layers, respectively at t 0: The root 13C concentration was taken as that of the measured corn stover
residues: Ec 0:0109804 or d13 Cc 212:00½:
The fraction ( f ) of SOC from corn in the SOC at time t is
usually computed as
13
13
13
13
f d Ct 2 d Cs = d Cc 2 d Cs ;
1
13
where d Ct Rt =Rpdb 2 11000 is the measured value at
time t. It is an approximation of equation
f Et 2 Es = Ec 2 Es ;
13
12
2
13
where Et Ct = C 1 Ct ; and linearity between f and
the 13C tracer concentrations is implied. It is de®ned for f
0 when Et Es ; the measured tracer concentration of the
SOC at the beginning of the experiment t 0; and f 1
for Et Ec when the tracer concentration is equal to that of
the corn (Ec), after the experiment has been maintained long
enough to obtain complete saturation of the SOC with cornC. The potential error induced by assuming linearity
between f and 13C tracer concentrations (or d 13C) was estimated by computing SOC from corn without the use of f.
This was achieved by setting 100% tracer enrichment in the
photosynthates Ec 1:0; and no tracer in the SOC at t
0 Es 0:0: With these initial conditions, the simulated
tracer concentration in the SOC for t . 0 is the concentration of corn derived C in the SOC. Simulated 13C and total C
concentrations in the SOC were reported with corn debris
included, in line with the experimental procedure that
returned ground organic debris to the soil sample after
sieving.
2.4. Calibration
Calibration was performed by adjusting some soil and
crop parameters to obtain, by trial and error, a good
visual ®t of measured and simulated data during the
13 year simulation (1980±1992). The parameters of
soil transformations used in previous simulations with
NCSOIL were retained (e.g. half-lives of labile and
resistant Pool I, and Pool II: 2, 17, and 115 d, respectively), with the exception of the initial content of Pool
I plus Pool II, and the decay rate constant of Pool III.
86
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
Fig. 2. Simulated (Ð) and measured (V) above-ground production.
The root parameters R1 through R5 were determined in
two steps: ®rst, the parameters R1 through R4 were calibrated against root mass production and distribution in
the soil, then R5 Ð the mass ratio of rhizodeposition-toroot daily production Ð was calibrated against SOC,
d 13C, and SOC derived from corn.
2.5. Sensitivity analysis
Deviation from calibrated values of the parameters was
tried one at a time. Parameters that control the formation
and distribution of roots and rhizodeposition in the soil
pro®le and their decay and assimilation into the SOC were
assayed for their in¯uence on total SOC, d 13C, corn-derived
SOC in the top 15 cm layer, and the above-ground
production.
Different d 13C values were assigned to the labile and
resistant pools of roots to assess the effect of 13C depletion
in lignin (Benner et al., 1987). Simulations were also
performed to estimate the contribution of non-decayed
root and rhizodeposition debris to the total SOC, d 13C,
and SOC from corn; it corresponds to the experimental
error incurred when these organic debris are removed
from the soil sample prior to analysis.
3. Results
3.1. Computation of d values and SOC from corn
It was veri®ed that errors caused by the use of Eq. (1)
instead of Eq. (2) were low (,1% on SOC from corn), as
expected since natural 13C concentrations are low enough to
maintain 13 C=12 C 1 13 C ù 13 C=12 C:
The potential error induced by assuming linearity
between f and 13C tracer concentrations was estimated by
computing SOC from corn with and without the use of f. The
discrepancy of results between the two methods was low: it
peaked for a few days to ,10% in 1980, ,4% from 1981 to
1984, and ,2% in subsequent years.
3.2. Calibration and simulated kinetics
The initial content of Pool I plus Pool II (385, 300, 150,
and 50 mg N g 21 in the top 0±15, 15±30, 30±50, and 50±
100 cm layers, respectively) and decay rate constant of Pool
III (27 year 21) re¯ected the high initial fertility of the soil.
The parameters controlling root mass production and
distribution in the soil pro®le were set to the following
values: initial shoot-to-root ratio, R1 0:5; and slope of
the linear relationship with time, R2 0:07; fraction (R3)
of the daily root mass increase distributed in the top 15 cm
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
87
Fig. 3. Simulated SOC (Ð). Measured SOC: 0±15 cm (V) and 15±30 cm (O).
R4 3 of the soil pro®le: 0.65, 0.60, and 0.55 during plant
growth from 0±10, 10±20, and 40 d after emergence,
respectively; mass ratio of rhizodeposition-to roots daily
production R5 6:0: The decay rate constants for
the roots were selected from Hunt (1977): root labile,
27%, kl 0:05 d21 ; root resistant, 73%, kr 0:0002 d21 :
The rate constants for the rhizodeposition were set to those
of Pool I, which decays as cellular debris (Molina et al.,
1983): labile pool, 56%, kl 0:332 d21 ; resistant pool,
44%, kr 0:04 d21 : The ratio of microbial C assimilation to
roots and rhizodeposition decay was set to 0.6, the same value
used in connection with the decay of soil organic pools.
Simulated and measured root mass distribution in the soil
pro®le, above ground production, total SOC, d 13C, and the
SOC from corn are shown in Figs. 1±5. The year-to-year
change of above-ground production was correctly simulated, though the simulated levels differed from the
measured ones for some years (Fig. 2). A gradual decline
in SOC was observed in the soil top 15 cm, and correctly
reproduced by the model (Fig. 3). In the 15±30 cm layer,
measured and simulated SOC were lower than in the top
15 cm, but the simulated gradual decrease in organic matter
content was not observed. Difference of 13C enrichment
observed between the SOC of the 0±15 and the 15±30 cm
layer were reproduced by the model (Fig. 4). The SOC from
corn was higher Ð equivalently, d 13C values were lower Ð
in the 0±15 cm layer where corn roots and rhizodeposition
were more abundant than in the 15±30 cm layer (Fig. 5).
3.3. Incorporation of corn-C from roots and rhizodeposition
into the SOC
The simulated ef®ciency of below-ground corn-C incorporation into the total SOC (ratio of SOC from corn-tobelow-ground corn-C cumulative production), and fraction
of below-ground corn-C in the SOC (ratio of SOC from
below-ground corn-C-to-total SOC) in the top 15 cm of
the soil are shown in Fig. 6. The ef®ciency of corn-C incorporation into SOC decreased as the fraction of corn-C in the
SOC increased.
3.4. Sensitivity analysis
Deviations from the calibrated values of the parameters
are shown in Table 1. The percentage change relative to the
calibrated data was most pronounced for the SOC from
corn; and to a lesser extent, the cumulative corn aboveground production. The simulated removal of non-decayed
root and rhizodeposition debris during sampling reduced the
SOC and d 13C by about 3% (Fig. 4), and the SOC from corn
by about 50% (Fig. 5). Changes in the initial 13C root
88
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
Fig. 4. Simulated d 13C of the SOC, with (Ð
Ð ) or without (Ð) roots and rhizodeposition. Measured d 13C: 0±15 cm (V) and 15±30 cm (O).
concentrations to account for lignin 13C depletion induced a
2.9% reduction in the SOC after 13 years of continuous
corn.
4. Discussion
The simulation model updated daily the soil conditions to
reproduce over a 13 year period the kinetics of seven
variables: above-ground corn production, and the total
SOC, soil d value, and the SOC from corn in the 0±15
and 15±30 cm depth. The parameters R1 to R4 were selected
to obtain a simulated total root mass and percentage root
mass distribution with depth that would follow the data
observed from the same experiment in Rosemount in 1986
(Fig. 1). The simulated data did not correctly partition the
percentage mass in the top two 15 cm depths. The measured
data included the crown, the portion of the above-ground
corn stalk that is not harvested and thus belongs to the roots.
Simulated root dry mass production ranged from 0.95
(1992) to 1.96 (1981) t ha 21, but the percentage mass distribution in soil remained almost the same during the 13 years
(e.g. 57±59% in the top 15 cm). Measured root mass at the
experimental site in 1986 ranged from 1.65 to 2.06 t ha 21, of
which 23±32% was found in the crown (R.H. Dowdy,
unpublished data). Comparison with other experimental
results is dubious when it is not possible to assert how
much of the crown is included (Allmaras et al., 1975;
Jones, 1985; Anderson, 1988; Huggins and Fuchs, 1997).
Calibration indicated the daily rate of rhizodeposition
to be 6 times that of root production. By comparison,
Johnen and Sauerbeck (1977) estimated with 14C that
roots released the equivalent of 2.5 times the C
mechanically recovered from roots. The simulated
amount of photosynthate translocated below-ground
was large. For example at harvest in 1984, 9.91 t dry
mass rhizodeposition ha 21 were released into the soil
with an above-ground and root dry mass, productions
of 12.74 and 1.65 t ha 21, respectively, to give a shoot
to root ratio (7.7) within the range of the experimental
values mean ^ std: dev: 5:31 ^ 3:95 computed from
the literature by Bolinder et al. (1999). Of this 1984
production, however, only 0.46 t dry mass rhizodeposition ha 21 were left in the soil on d 365 of the same
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
89
Fig. 5. Simulated SOC from corn with (Ð
Ð ) or without (Ð) roots and rhizodeposition. Measured SOC from corn: 0±15 cm (V) and 15±30 cm (O).
year. The root, which decayed more slowly, totaled
1.24 t dry mass ha 21 to give a shoot-to-root-mass ratio
of 10.3.
The simulated percentages of rhizodeposition in the total
net C ®xed photosynthetically (above- plus below-ground
production in 1984) were 40.8; or 24.4%, relative to the total
CO2-C ®xed. Experimental estimates with 14C methods have
ranged from 7 to 48% (Bottner et al., 1999; Biondini, 1988;
Whipps, 1984; Wood, 1987) and 3.5±18% (Barber and
Martin, 1976; Meharg and Killham, 1988) relative to the
net and total C ®xed phosynthetically, respectively. Balesdent and Balabane (1996) worked with naturally 13C
enriched corn to compute a below-ground average yearly
production 1.5 higher than that of the stalks and leaves;
ªmuch higher than any known estimation of in situ belowground production of C in maize or any other cereal ª and
consisting ªof ephemeral products such as root exudatesº.
By comparison, the present simulation gave a root-plusrhizodeposition production in 1984, equal to 1.8 that of
stalk plus leaves (assuming a harvest index of 0.50), thus
corroborating that corn may translocate more to the soil
from below- than above-ground.
The kinetics of SOC from corn presented peaks caused by
the abundant formation and rapid decay of rhizodeposition,
which did not accumulate in soil (Fig. 5). Rhizodeposition
was incorporated into the SOC through the process of
microbial assimilation, thus with a 40% loss to CO2 for
each unit decayed. Residual root debris, however, accumulated and accounted for about 50% of the SOC from
corn (Fig. 5). Roots had a greater effect on the SOC
from corn than rhizodeposition, though more photosynthates were translocated to rhizodeposition than
roots. This was also re¯ected in the higher sensitivity
of the SOC from corn per unit percentage change in R2
or R3 than in R5 (Table 1). These observations underline
the necessity to include root debris Ð crown included
Ð when the SOC is measured.
Information gathered from these ®eld experiments would
be richer if the rates of CO2 ± 13C production were measured.
It would allow to calibrate the parameters R2 through R5
with a higher weight given to the information on rhizodeposition than presently done, whereby R5 is obtained after
consideration of the overbearing root mass. For example,
when the simulated root mass percentage in the top 15 cm
was increased from 57±67% (parameter R3, Table 1) to
bring it up closer to the measured data (Fig. 1), the calibrated amount of SOC from corn (5.90 t C ha 21) could be
computed with R5 3:5 instead of 6.0.
The SOC from corn was the most sensitive variable to
variations of the parameters that controlled roots and rhizodeposition (Table 1). The cumulative above-ground production was also in¯uenced by those parameters that modi®ed
the root or rhizodeposition mass: microbial ef®ciency
R2, and R5. Corn with higher root and rhizodeposition
90
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
Fig. 6. Simulated ef®ciency of below-ground corn-C incorporation into the total SOC (ratio of SOC from corn-to-cumulative roots plus rhizodeposition
produced), and fraction of below-ground corn-C in the SOC (ratio of SOC from roots and rhizodeposition-to-SOC) in the top 15 cm of the soil pro®le.
mass required more N and thus induced more N stress
for growth under limited N supply, as was the case for
the treatment considered in this study (no fertilizer N
added).
The resistant root pool in soil accounted for about half the
SOC from corn. Symmetric 10-fold variations in the decay
rate constant kr (0.002 and 0.00002) around the calibrated
value (0.0002) induced an uneven response: the lower decay
rates affected less the SOC from corn than the higher decay
rates (Table 1). Davenport et al. (1988) also modeled root
decay with a double exponential equation: labile to recalcitrant ratio of 15/85 with kl 0:120 d21 ; and kr
0:0019 d21 ; values, which are closer to those used by Hunt
(1977). The double pool model does not parallel the fractions
that chemically characterize plant material (e.g. cellulose,
hemicellulose, and lignin). It creates a dif®culty when
the effects of differences in the 13C concentration
between the fractions are investigated by simulation.
Benner et al. (1987) measured a 4.7½ depletion in
the 13C lignin content relative to the whole-plant material. It corresponds to a 0.598½ depletion when it is
pro-rated from the lignin dry weight percentage (9.3%)
to that of the resistant pool (73%); and it induced a
2.9% decrease in the SOC from corn over 13 years,
which is small relative to the effects of the other
parameters.
The ef®ciency of below-ground corn-C incorporation
into the SOC, and fraction of below-ground corn-C in
the SOC are important indices used to estimate the
effect of agriculture on the global C dynamic. After
13 years of continuous corn, the simulated fraction and
ef®ciency indices in the top 15 cm depth were 0.12 and
0.19, respectively. The ef®ciency jumped to 1.00 at corn
emergence, then dropped as decay began and some
corn-C escaped the soil as CO2 (Fig. 6). The fall of
the index was rapid in the ®rst year, but as the amount
of SOC from corn built up, it decreased more slowly.
An ef®ciency index of 0.38 was observed after 1 year
with the input of the below-ground corn-C only (Balesdent and Balabane, 1996). The fraction index from
several reports of long-term studies ranged from 0.14
to 0.30 Ð values computed with various approximations for the amount of rhizodeposition in the SOC
(Bolinder et al., 1999). The simulated kinetics of the
ef®ciency and fraction indices leveled off at about
0.20 when the labile pools had established a steady
91
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
Table 1
Simulated d value, total SOC, SOC of corn origin, and above-ground production on d 365, 1992, associated with deviations from the calibrated parameters
Controlling parameters a
d value in
top 15 cm b
SOC in top 15 cm b
(t C ha 21)
SOC from corn in top Above-ground dry mass
15 cm b (t C ha 21)
(cumulative, t ha 21)
Microbial assimilation ef®ciency
8.0
0.6
0.4
218.96 (20.5)
219.05 (0.0)
219.15 (0.5)
50.6 (1.1)
50.0 (0.0)
49.3 (21.3)
6.49 (10.0)
5.90 (0.0)
5.17 (212.4)
136.3 (22.3)
139.6 (0.0)
142.7 (2.2)
Resistant root, decay rate
constant (per day)
0.00002
0.0002
0.002
219.01 (20.2)
219.05 (0.0)
219.24 (1.0)
50.3 (0.6)
50.0 (0.0)
48.6 (22.8)
6.19 (4.9)
5.90 (0.0)
4.53 (223.2)
139.8 (0.1)
139.6 (0.0)
138.9 (20.5)
R2: daily increase of the
shoot-to-root ratio
0.05 {5.7}
0.07 {7.7}
0.11 {11.7}
218.85 (21.0)
219.05 (0.0)
219.29 (1.3)
51.5 (3.0)
50.0 (0.0)
48.4 (23.1)
7.36 (24.8)
5.90 (0.0)
4.26 (227.8)
133.7 (24.2)
139.6 (0.0)
146.2 (4.7)
R3: fraction of daily root mass
increase set in top 15 cm depth c
0.75, 0.65, 0.60 [67] 218.92 (20.7)
0.65, 0.60, 0.55 [57] 219.05 (0.0)
0.55, 0.50, 0.45 [44] 219.23 (0.9)
51.0 (1.9)
50.0 (0.0)
48.8 (22.4)
6.85 (16.1)
5.90 (0.0)
4.64 (221.4)
139.8 (0.1)
139.6 (0.0)
139.0 (20.4)
51.3 (2.0)
50.0 (0.0)
49.0 (21.9)
7.19 (21.9)
5.90 (0.0)
4.49 (223.9)
126.2 (29.6)
139.6 (0.0)
148.4 (6.3)
R5: mass ratio of
rhizodeposition-to-root daily
production
12
6.0
3.0
218.87 (20.9)
219.05 (0.0)
219.19 (0.7)
a
Bold numbers correspond to the calibrated values. Numbers in braces ({) give the shoot-to-root mass ratio at harvest in 1984. Numbers in brackets ([) give
the ratio of root mass in top 15 cm at harvest in 1984.
b
Numbers in parenthesis give the percentage change relative to the calibrated value.
c
Parameter values assigned during plant growth from 0 to 10, 10 to 20, and 40 d after emergence.
state (Fig. 6). Further increase would be very slow and
controlled by the dual effect of the incorporation of the
C from corn into the stable SOC, and the decay of the
stabilized SOC of soil origin.
References
Allmaras, R.R., Nelson, W.W., Voorhees, W.B., 1975. Soybean and corn
rooting in southwestern Minnesota: II. Root distributions and related
water in¯ow. Soil Science Society of America Proceedings 39, 771±
777.
Anderson, E.L., 1988. Tillage and N fertilization effects on maize root
growth and root:shoot ratio. Plant and Soil 108, 245±251.
Balesdent, J., Balabane, M., 1996. Major contribution of roots to soil carbon
storage inferred from maize cultivated soils. Soil Biology & Biochemistry 28, 1261±1263.
Barber, D.A., Martin, J.K., 1976. The release of organic substances by
cereal roots into soil. New Phytologist 76, 69±80.
Benner, R., Fogel, M.L., Sprague, E.L.K., Hudson, E., 1987. Depletion of
13
C lignin and its implications for stable carbon isotopic studies. Nature
329, 708±710.
Biondini, M., 1988. Carbon and nitrogen losses through root exudition by
Agropyron cristatum, A. Smithiii and Bouteloua gracilis. Soil Biology
& Biochemistry 20, 477±482.
Bolinder, M.A., Angers, D.A., Giroux, M., LaverdieÁre, M.R., 1999. Estimating C inputs retained as soil organic matter from corn (Zea mays L.).
Plant and Soil 215, 85±91.
Bottner, P., Pansu, M., Sallih, Z., 1999. Modelling the effect of active roots
on soil organic matter turnover. Plant and Soil 216, 15±25.
Cerri, C., Feller, C., Balesdent, J., Victoria, R., Plenecassagne, A., 1985.
Application du tracage isotopique naturel en 13C aÁ l'eÂtude de la dynamique de la matieÁre organique dans les sols. Comptes Rendus de l'AcadeÂmie des Sciences de Paris T.300 II (9), 423±428.
Clapp, C.E., Allmaras, R.R., Layese, M.F., Linden, D.R., Dowdy, R.H.,
2000. Soil organic carbon and carbon-13 abundance as related to tillage,
crop residue, and nitrogen fertilization under continuous corn management in Minnesota. Soil and Tillage Research 55, 127±142.
Clay, D.E., Clapp, C.E., Linden, D.R., Molina, J.A.E., 1989. Nitrogentillage-residue management: 3. Observed and simulated interactions
among soil depth, nitrogen mineralization, and corn yield. Soil Science
147, 319±325.
Davenport, J.R., Thomas, R.L., Mott, S.C., 1988. Carbon mineralization of
corn (Zea mays L.) and bromegrass (Bromus inermis Leyss)
components with an emphasis on the below-ground carbon. Soil
Biology & Biochemistry 20, 471±476.
Hadas, A., Parkin, T.B., Stahl, P.D., 1998. Reduced CO2 release
from decomposing wheat straw under N-limiting conditions: simulation of carbon turnover. European Journal of Soil Science 49,
487±494.
Huggins, D.R., Fuchs, D.J., 1997. Long-term N management effects on corn
yield and soil C of an Haplustoll in Minnesota. In: Paul, E.A., Paustian,
K., Elliott, E.T., Cole, C.V. (Eds.). Soil Organic Matter in Temperate
Agroecosystems: Long-Term Experiments in North America, CRC
Press, Boca Raton, pp. 121±128.
Huggins, D.R., Clapp, C.E., Allmaras, R.R., Lamb, J.A., Layese, M.F.,
1998. Carbon dynamics in corn-soybean sequences as estimated from
natural carbon-13 abundance. Soil Science Society of America Journal
62, 195±203.
Hunt, H.W., 1977. A simulation model for decomposition in grasslands.
Ecology 58, 469±484.
Johnen, G.G., Sauerbeck, D.R., 1977. A tracer technique for measuring
growth, mass and microbial breakdown of plant roots during vegetation.
Ecological Bulletin (Stockholm) 25, 366±373.
Jones, C.A., 1985. C4 Grasses and Cereals: Growth, Development, and
Stress Response, Wiley, New York.
Linden, D.R., Clapp, C.E., Dowdy, R.H., 2000. Long-term corn grain and
stover yields as a function of tillage and residue removal in east central
Minnesota. Soil and Tillage Research (in press).
Meharg, A.A., Killham, K., 1988. A comparison of carbon ¯ow from prelabelled and pulse-labelled plants. Plant and Soil 112, 225±231.
92
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
Molina, J.A.E., Smith, P., 1998. Modeling carbon and nitrogen processes in
soils. Advances in Agronomy 62, 253±297.
Molina, J.A.E., Clapp, C.E., Shaffer, M.J., Chichester, F.W., Larson, W.E.,
1983. NCSOIL, a model of nitrogen transformations in soil: description,
calibration and behavior. Soil Science Society of America Journal 47,
85±91.
Molina, J.A.E., Crocker, G.J., Grace, P.R., Klir, J., Korschens, M., Poulton,
P.R., Richter, D.D., 1997. Simulating trends in soil organic carbon in
long-term experiments using the NCSOIL and NCSWAP models.
Geoderma 81, 91±107.
Nicolardot, B., Molina, J.A.E., Allard, M.R., 1994. C and N ¯uxes between
pools of soil organic matter. Model calibration with long-term incubation data. Soil Biology & Biochemistry 26, 235±243.
Whipps, J.M., 1984. Environmental factors affecting the loss of carbon
from the roots of wheat and barley seedlings. Journal of Experimental
Botany 35, 767±773.
Wood, M., 1987. Predicted microbial biomass in the rhizosphere of barley
in the ®eld. Plant and Soil 97, 303±314.
www.elsevier.com/locate/soilbio
Modeling the incorporation of corn (Zea mays L.) carbon from roots
and rhizodeposition into soil organic matter
J.A.E. Molina a,*, C.E. Clapp b, D.R. Linden b, R.R. Allmaras b, M.F. Layese b,
R.H. Dowdy b, H.H. Cheng a
b
a
Department of Soil, Water, and Climate, University of Minnesota, St. Paul, MN 55108, USA
USDA-ARS and Department of Soil, Water, and Climate, University of Minnesota, St. Paul, MN 55108, USA
Received 5 October 1999; received in revised form 17 April 2000; accepted 6 May 2000
Abstract
Experimental data reported in the literature over the last decennium indicate that roots and rhizodeposition are important sources of carbon
for the synthesis of soil organic carbon. Our objective was to verify the capability of the simulation model NCSWAP to reproduce the general
conclusions from the experimental literature, and to gain some insight about the processes that control the incorporation of corn belowground production into the soil organic matter. The model was calibrated against the experimental data gathered from a long-term ®eld
experiment located near St. Paul, Minnesota. The simulation model updated daily the soil conditions to reproduce over a 13 year period the
measured kinetics of seven variables: above-ground corn production, and the total soil organic matter, soil d value, and the soil organic
matter derived from corn in the 0±15 and 15±30 cm depth. The simulation gave a root-plus-rhizodeposition 1.8 times larger than stalks plus
leaves. The translocation ef®ciency of corn-C into soil organic C at the 0±15 cm depth gradually decreased to 0.19 of the below-ground
deposition. The sensitivity of below-ground photosynthate incorporation into the soil organic matter was analyzed relative to variations in the
parameters that control the formation and decay of roots and rhizodeposition. Roots had a greater effect than rhizodeposition on the soil
organic matter, though more photosynthates were translocated to rhizodeposition than to roots. q 2001 Elsevier Science Ltd. All rights
reserved.
Keywords: Root exudate; Rhizosphere; Natural C enrichment; Tracer C; Lignin; NCSOIL; NCSWAP
1. Introduction
Roots and rhizodeposition are important sources of
carbon for the synthesis of soil organic carbon (SOC). The
translocation of plant C into SOC is quantitatively documented with tracer C. For example, Johnen and Sauerbeck
(1977) working with 14C enriched mustard and wheat
showed that tracer C in soil at harvest was 20±50% higher
than that mechanically isolated from roots, and that only
20% of the 14C mineralized to CO2 originated from root
respiration. Experiments with 14C require elaborate methods
that limit measurements to plants maintained in growth
chambers for a short period. By contrast, the different
natural 13C enrichments in C3 and C4 plants facilitate the
estimation in situ of the long term transformation of corn
photosynthates into SOC (Cerri et al., 1985; Balesdent and
Balabane, 1996; Huggins et al., 1998; Clapp et al., 2000).
The amount of SOC originating from corn increases as
* Corresponding author. Tel.: 11-612-625-6259; fax: 11-612-625-2208.
E-mail address: jmolina@soils.umn.edu (J.A.E. Molina).
annual crops of corn are grown continuously. It nevertheless
represents only a fraction of the corn-C incorporated into the
soil as some is lost, mostly as CO2, during the decay
processes of the corn residues and the SOC of corn origin.
Experimental results summarized by Bolinder et al. (1999)
suggest that the percentage of the below ground corn-C
incorporated into SOC (range 16±30%) is higher than that
from the above ground corn biomass (range 7.7±20%). The
multiplicity and interdependence of C and N transformations make it dif®cult to appreciate the origin of the percentage range, as well as the predominance of below ground
over above ground contribution, without a process oriented
simulation model. Several computer models are available to
describe crop growth and nutrient cycling in soil, water and
plant (Molina and Smith, 1998). One such model,
NCSWAP, simulates below and above ground photosynthate input into soil layers. It distinguishes between
mechanically harvested roots and rhizodeposition. Total
and tracer C and N released from plant are incorporated
into the soil processes as computed with the model NCSOIL
(Molina et al., 1997). These two models have been tested
0038-0717/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0038-071 7(00)00117-6
84
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
(23 references in Current Contents 1993 to date), and are
registered with the Soil Organic Network of the Global
Change and Terrestrial Ecosystem of the International
Geosphere±Biosphere Programme.
A long-term ®eld experiment located near St. Paul was
initiated in 1980 to investigate the effect of N and tillage
management on the incorporation of corn residues into the
SOC. Information from this experiment documents the
origin of the SOC (corn or soil), and the partition of inorganic fertilizer N between the plant and the soil fractions
through the use of total and tracer elements (Clapp et al.,
2000).
Our objective was to gain some insight about the
processes that control the incorporation of corn belowground production into the soil organic matter. In this
paper, we focus on C transformations and a ®eld treatment (no above-ground residues returned, no fertilizer N
added) that limited inputs to the SOC from roots and
rhizodeposition. We ascertained the ability of NCSWAP
to reproduce the experimental data, and a sensitivity
analysis was performed to quantify the in¯uence of
roots, rhizodeposition, and soil C transformations on
the system over 13 years.
2. Materials and methods
2.1. Field experiment and measured data
Data were obtained from one of the 12 treatments Ð the
chisel-plough tillage with no inorganic N addition and
stover-residue plus grain removed Ð from a long-term
continuous corn cultivation (Zea mays L.) study, which
took place during the years 1980±1993 at the University
of Minnesota Research and Outreach Center, Rosemount,
MN (Clay et al., 1989; Clapp et al., 2000). The soil at the
site is a Waukegan silt loam (®ne-silty over sandy or sandyskeletal, mixed, mesic Typic Hapludolls). Soil depth to
gravel is about 100 cm, and the climate is sub-humid. Chisel
ploughing to a depth of about 17 cm was conducted
annually in the fall. Soil samples were taken from 0 to 15
and 15±30 cm layers, sieved (2 mm), the retained organic
debris ground and returned to the soil sample, and analyzed
for d 13C value. Thus, plant organic debris are included in
the SOC data. Above-ground corn production was measured
annually from 1980 to 1993 (Linden et al., 2000).
Speci®c ®eld management, analytical techniques, and
results were detailed in Clapp et al. (2000) and Linden
et al. (2000).
is parameterized for mass increase measured during a year
(the reference year, 1981) that gave maximum production
with minimal or no N and water stress. For other years on
the same site with the same crop, growth is adjusted daily to
account for the soil conditions (N and water availability)
and air temperature. The N and C transformations are simulated by the model NCSOIL included as a subroutine in
NCSWAP (Molina et al., 1983; Nicolardot et al., 1994;
Hadas et al., 1998). The soil organic matter is represented
in each 5 cm layer of the 100 cm deep soil pro®le by four
pools: labile and resistant Pool I, Pool II, and Pool III.
NCSWAP identi®es in each soil layer the following organic
debris: (1) roots and (2) rhizodeposition from the current
year; (3) roots and (4) rhizodeposition from the preceding
years; (5±7) three different above-ground crop residues; and
(8±11) four different organic amendments. Each debris is
described by two pools, each de®ned by its ®rst-order
decay-rate-constant and C to N ratio. NCSWAP requires
several input ®les: the initial soil conditions, and yearly
®les for the driving variables (crop, management, temperature, and precipitation). The soil conditions, roots and rhizodeposition included, are updated daily or every 0.2 d during
water in®ltration and redistribution. The source and executable code of NCSWAP and NCSOIL with the description
of the variables in the input ®les are available at the web site
(http://www.soils.umn.edu/research/ncswap-ncsoil/). Since
this work focuses on the dynamics of roots and rhizodeposition, some speci®cs are given in reference to the simulation
of the below-ground crop.
Root and above-ground daily mass increase are linked
by the shoot-to-root mass ratio (SR); SR R1 1 R2 £ t;
where R1 and R2 are constants and t is the time (d) from
emergence. Repartition of the root mass in the soil
layers is controlled by two parameters: the fraction
(R3) of the daily root mass increase that is added to
the top R4 layers of the pro®le. The parametric value
of R3 is de®ned for different physiological stages of the
crop. The left-over daily-root-mass increase is distributed below the R4th layer according to an algorithm that
considers the vertical depth of root penetration. Rhizodeposition is computed as R5 times the root mass
formed per day at each layer. Roots grow until they
start to decay at crop maturity or harvest, whichever
comes ®rst. Rhizodeposition decays as soon as formed.
Roots and rhizodeposition decay as a two-component
material (labile and resistant) with the rate constants
kl and kr (per day), modi®ed by reduction factors to
account for edaphic conditions.
2.3. d value and SOC from corn
2.2. Simulation model
The model NCSWAP simulates N and C Ð total and
tracer Ð transformations and the ¯ow of water in the soil
pro®le (Molina et al., 1997). The above-ground plant growth
is described by a generalized form of the logistic equation. It
Tracer and total C concentrations are simulated in parallel. At each time step the rate of change in tracer concentration in any pool (D tr) is given by the equation,
Dtr Dto £ E £ m; where D to is the rate of change in
total concentration, E 13 C=12 C 1 13 C is the tracer
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
85
Fig. 1. Root mass distribution in the soil pro®le at harvest. Simulated for the 1984 season and measured during the 1986 season.
concentration at the beginning of the time step, and m
the discrimination factor, set to 1.0 in this study. Tracer
and total concentrations are both updated, and the
process repeated. Simulations were performed with the
same initial d 13C assigned to the resistant and labile
corn-debris pools.
The initial SOC- 13C tracer concentrations were set to
their measured values against the Pee Dee Belemnite
standard Rpdb 13 Cpdb =12 Cpdb 0:0112372 : Es
13 Cs =12 C 1 13 Cs 0:0108926 corresponding to a
and
Es
d13 Cs Rs =Rpdp 2 1 1000 219:99½;
0:0109157 or d13 Cs 217:89½ in the 0±15 and 15±
30 cm layers, respectively at t 0: The root 13C concentration was taken as that of the measured corn stover
residues: Ec 0:0109804 or d13 Cc 212:00½:
The fraction ( f ) of SOC from corn in the SOC at time t is
usually computed as
13
13
13
13
f d Ct 2 d Cs = d Cc 2 d Cs ;
1
13
where d Ct Rt =Rpdb 2 11000 is the measured value at
time t. It is an approximation of equation
f Et 2 Es = Ec 2 Es ;
13
12
2
13
where Et Ct = C 1 Ct ; and linearity between f and
the 13C tracer concentrations is implied. It is de®ned for f
0 when Et Es ; the measured tracer concentration of the
SOC at the beginning of the experiment t 0; and f 1
for Et Ec when the tracer concentration is equal to that of
the corn (Ec), after the experiment has been maintained long
enough to obtain complete saturation of the SOC with cornC. The potential error induced by assuming linearity
between f and 13C tracer concentrations (or d 13C) was estimated by computing SOC from corn without the use of f.
This was achieved by setting 100% tracer enrichment in the
photosynthates Ec 1:0; and no tracer in the SOC at t
0 Es 0:0: With these initial conditions, the simulated
tracer concentration in the SOC for t . 0 is the concentration of corn derived C in the SOC. Simulated 13C and total C
concentrations in the SOC were reported with corn debris
included, in line with the experimental procedure that
returned ground organic debris to the soil sample after
sieving.
2.4. Calibration
Calibration was performed by adjusting some soil and
crop parameters to obtain, by trial and error, a good
visual ®t of measured and simulated data during the
13 year simulation (1980±1992). The parameters of
soil transformations used in previous simulations with
NCSOIL were retained (e.g. half-lives of labile and
resistant Pool I, and Pool II: 2, 17, and 115 d, respectively), with the exception of the initial content of Pool
I plus Pool II, and the decay rate constant of Pool III.
86
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
Fig. 2. Simulated (Ð) and measured (V) above-ground production.
The root parameters R1 through R5 were determined in
two steps: ®rst, the parameters R1 through R4 were calibrated against root mass production and distribution in
the soil, then R5 Ð the mass ratio of rhizodeposition-toroot daily production Ð was calibrated against SOC,
d 13C, and SOC derived from corn.
2.5. Sensitivity analysis
Deviation from calibrated values of the parameters was
tried one at a time. Parameters that control the formation
and distribution of roots and rhizodeposition in the soil
pro®le and their decay and assimilation into the SOC were
assayed for their in¯uence on total SOC, d 13C, corn-derived
SOC in the top 15 cm layer, and the above-ground
production.
Different d 13C values were assigned to the labile and
resistant pools of roots to assess the effect of 13C depletion
in lignin (Benner et al., 1987). Simulations were also
performed to estimate the contribution of non-decayed
root and rhizodeposition debris to the total SOC, d 13C,
and SOC from corn; it corresponds to the experimental
error incurred when these organic debris are removed
from the soil sample prior to analysis.
3. Results
3.1. Computation of d values and SOC from corn
It was veri®ed that errors caused by the use of Eq. (1)
instead of Eq. (2) were low (,1% on SOC from corn), as
expected since natural 13C concentrations are low enough to
maintain 13 C=12 C 1 13 C ù 13 C=12 C:
The potential error induced by assuming linearity
between f and 13C tracer concentrations was estimated by
computing SOC from corn with and without the use of f. The
discrepancy of results between the two methods was low: it
peaked for a few days to ,10% in 1980, ,4% from 1981 to
1984, and ,2% in subsequent years.
3.2. Calibration and simulated kinetics
The initial content of Pool I plus Pool II (385, 300, 150,
and 50 mg N g 21 in the top 0±15, 15±30, 30±50, and 50±
100 cm layers, respectively) and decay rate constant of Pool
III (27 year 21) re¯ected the high initial fertility of the soil.
The parameters controlling root mass production and
distribution in the soil pro®le were set to the following
values: initial shoot-to-root ratio, R1 0:5; and slope of
the linear relationship with time, R2 0:07; fraction (R3)
of the daily root mass increase distributed in the top 15 cm
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
87
Fig. 3. Simulated SOC (Ð). Measured SOC: 0±15 cm (V) and 15±30 cm (O).
R4 3 of the soil pro®le: 0.65, 0.60, and 0.55 during plant
growth from 0±10, 10±20, and 40 d after emergence,
respectively; mass ratio of rhizodeposition-to roots daily
production R5 6:0: The decay rate constants for
the roots were selected from Hunt (1977): root labile,
27%, kl 0:05 d21 ; root resistant, 73%, kr 0:0002 d21 :
The rate constants for the rhizodeposition were set to those
of Pool I, which decays as cellular debris (Molina et al.,
1983): labile pool, 56%, kl 0:332 d21 ; resistant pool,
44%, kr 0:04 d21 : The ratio of microbial C assimilation to
roots and rhizodeposition decay was set to 0.6, the same value
used in connection with the decay of soil organic pools.
Simulated and measured root mass distribution in the soil
pro®le, above ground production, total SOC, d 13C, and the
SOC from corn are shown in Figs. 1±5. The year-to-year
change of above-ground production was correctly simulated, though the simulated levels differed from the
measured ones for some years (Fig. 2). A gradual decline
in SOC was observed in the soil top 15 cm, and correctly
reproduced by the model (Fig. 3). In the 15±30 cm layer,
measured and simulated SOC were lower than in the top
15 cm, but the simulated gradual decrease in organic matter
content was not observed. Difference of 13C enrichment
observed between the SOC of the 0±15 and the 15±30 cm
layer were reproduced by the model (Fig. 4). The SOC from
corn was higher Ð equivalently, d 13C values were lower Ð
in the 0±15 cm layer where corn roots and rhizodeposition
were more abundant than in the 15±30 cm layer (Fig. 5).
3.3. Incorporation of corn-C from roots and rhizodeposition
into the SOC
The simulated ef®ciency of below-ground corn-C incorporation into the total SOC (ratio of SOC from corn-tobelow-ground corn-C cumulative production), and fraction
of below-ground corn-C in the SOC (ratio of SOC from
below-ground corn-C-to-total SOC) in the top 15 cm of
the soil are shown in Fig. 6. The ef®ciency of corn-C incorporation into SOC decreased as the fraction of corn-C in the
SOC increased.
3.4. Sensitivity analysis
Deviations from the calibrated values of the parameters
are shown in Table 1. The percentage change relative to the
calibrated data was most pronounced for the SOC from
corn; and to a lesser extent, the cumulative corn aboveground production. The simulated removal of non-decayed
root and rhizodeposition debris during sampling reduced the
SOC and d 13C by about 3% (Fig. 4), and the SOC from corn
by about 50% (Fig. 5). Changes in the initial 13C root
88
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
Fig. 4. Simulated d 13C of the SOC, with (Ð
Ð ) or without (Ð) roots and rhizodeposition. Measured d 13C: 0±15 cm (V) and 15±30 cm (O).
concentrations to account for lignin 13C depletion induced a
2.9% reduction in the SOC after 13 years of continuous
corn.
4. Discussion
The simulation model updated daily the soil conditions to
reproduce over a 13 year period the kinetics of seven
variables: above-ground corn production, and the total
SOC, soil d value, and the SOC from corn in the 0±15
and 15±30 cm depth. The parameters R1 to R4 were selected
to obtain a simulated total root mass and percentage root
mass distribution with depth that would follow the data
observed from the same experiment in Rosemount in 1986
(Fig. 1). The simulated data did not correctly partition the
percentage mass in the top two 15 cm depths. The measured
data included the crown, the portion of the above-ground
corn stalk that is not harvested and thus belongs to the roots.
Simulated root dry mass production ranged from 0.95
(1992) to 1.96 (1981) t ha 21, but the percentage mass distribution in soil remained almost the same during the 13 years
(e.g. 57±59% in the top 15 cm). Measured root mass at the
experimental site in 1986 ranged from 1.65 to 2.06 t ha 21, of
which 23±32% was found in the crown (R.H. Dowdy,
unpublished data). Comparison with other experimental
results is dubious when it is not possible to assert how
much of the crown is included (Allmaras et al., 1975;
Jones, 1985; Anderson, 1988; Huggins and Fuchs, 1997).
Calibration indicated the daily rate of rhizodeposition
to be 6 times that of root production. By comparison,
Johnen and Sauerbeck (1977) estimated with 14C that
roots released the equivalent of 2.5 times the C
mechanically recovered from roots. The simulated
amount of photosynthate translocated below-ground
was large. For example at harvest in 1984, 9.91 t dry
mass rhizodeposition ha 21 were released into the soil
with an above-ground and root dry mass, productions
of 12.74 and 1.65 t ha 21, respectively, to give a shoot
to root ratio (7.7) within the range of the experimental
values mean ^ std: dev: 5:31 ^ 3:95 computed from
the literature by Bolinder et al. (1999). Of this 1984
production, however, only 0.46 t dry mass rhizodeposition ha 21 were left in the soil on d 365 of the same
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
89
Fig. 5. Simulated SOC from corn with (Ð
Ð ) or without (Ð) roots and rhizodeposition. Measured SOC from corn: 0±15 cm (V) and 15±30 cm (O).
year. The root, which decayed more slowly, totaled
1.24 t dry mass ha 21 to give a shoot-to-root-mass ratio
of 10.3.
The simulated percentages of rhizodeposition in the total
net C ®xed photosynthetically (above- plus below-ground
production in 1984) were 40.8; or 24.4%, relative to the total
CO2-C ®xed. Experimental estimates with 14C methods have
ranged from 7 to 48% (Bottner et al., 1999; Biondini, 1988;
Whipps, 1984; Wood, 1987) and 3.5±18% (Barber and
Martin, 1976; Meharg and Killham, 1988) relative to the
net and total C ®xed phosynthetically, respectively. Balesdent and Balabane (1996) worked with naturally 13C
enriched corn to compute a below-ground average yearly
production 1.5 higher than that of the stalks and leaves;
ªmuch higher than any known estimation of in situ belowground production of C in maize or any other cereal ª and
consisting ªof ephemeral products such as root exudatesº.
By comparison, the present simulation gave a root-plusrhizodeposition production in 1984, equal to 1.8 that of
stalk plus leaves (assuming a harvest index of 0.50), thus
corroborating that corn may translocate more to the soil
from below- than above-ground.
The kinetics of SOC from corn presented peaks caused by
the abundant formation and rapid decay of rhizodeposition,
which did not accumulate in soil (Fig. 5). Rhizodeposition
was incorporated into the SOC through the process of
microbial assimilation, thus with a 40% loss to CO2 for
each unit decayed. Residual root debris, however, accumulated and accounted for about 50% of the SOC from
corn (Fig. 5). Roots had a greater effect on the SOC
from corn than rhizodeposition, though more photosynthates were translocated to rhizodeposition than
roots. This was also re¯ected in the higher sensitivity
of the SOC from corn per unit percentage change in R2
or R3 than in R5 (Table 1). These observations underline
the necessity to include root debris Ð crown included
Ð when the SOC is measured.
Information gathered from these ®eld experiments would
be richer if the rates of CO2 ± 13C production were measured.
It would allow to calibrate the parameters R2 through R5
with a higher weight given to the information on rhizodeposition than presently done, whereby R5 is obtained after
consideration of the overbearing root mass. For example,
when the simulated root mass percentage in the top 15 cm
was increased from 57±67% (parameter R3, Table 1) to
bring it up closer to the measured data (Fig. 1), the calibrated amount of SOC from corn (5.90 t C ha 21) could be
computed with R5 3:5 instead of 6.0.
The SOC from corn was the most sensitive variable to
variations of the parameters that controlled roots and rhizodeposition (Table 1). The cumulative above-ground production was also in¯uenced by those parameters that modi®ed
the root or rhizodeposition mass: microbial ef®ciency
R2, and R5. Corn with higher root and rhizodeposition
90
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
Fig. 6. Simulated ef®ciency of below-ground corn-C incorporation into the total SOC (ratio of SOC from corn-to-cumulative roots plus rhizodeposition
produced), and fraction of below-ground corn-C in the SOC (ratio of SOC from roots and rhizodeposition-to-SOC) in the top 15 cm of the soil pro®le.
mass required more N and thus induced more N stress
for growth under limited N supply, as was the case for
the treatment considered in this study (no fertilizer N
added).
The resistant root pool in soil accounted for about half the
SOC from corn. Symmetric 10-fold variations in the decay
rate constant kr (0.002 and 0.00002) around the calibrated
value (0.0002) induced an uneven response: the lower decay
rates affected less the SOC from corn than the higher decay
rates (Table 1). Davenport et al. (1988) also modeled root
decay with a double exponential equation: labile to recalcitrant ratio of 15/85 with kl 0:120 d21 ; and kr
0:0019 d21 ; values, which are closer to those used by Hunt
(1977). The double pool model does not parallel the fractions
that chemically characterize plant material (e.g. cellulose,
hemicellulose, and lignin). It creates a dif®culty when
the effects of differences in the 13C concentration
between the fractions are investigated by simulation.
Benner et al. (1987) measured a 4.7½ depletion in
the 13C lignin content relative to the whole-plant material. It corresponds to a 0.598½ depletion when it is
pro-rated from the lignin dry weight percentage (9.3%)
to that of the resistant pool (73%); and it induced a
2.9% decrease in the SOC from corn over 13 years,
which is small relative to the effects of the other
parameters.
The ef®ciency of below-ground corn-C incorporation
into the SOC, and fraction of below-ground corn-C in
the SOC are important indices used to estimate the
effect of agriculture on the global C dynamic. After
13 years of continuous corn, the simulated fraction and
ef®ciency indices in the top 15 cm depth were 0.12 and
0.19, respectively. The ef®ciency jumped to 1.00 at corn
emergence, then dropped as decay began and some
corn-C escaped the soil as CO2 (Fig. 6). The fall of
the index was rapid in the ®rst year, but as the amount
of SOC from corn built up, it decreased more slowly.
An ef®ciency index of 0.38 was observed after 1 year
with the input of the below-ground corn-C only (Balesdent and Balabane, 1996). The fraction index from
several reports of long-term studies ranged from 0.14
to 0.30 Ð values computed with various approximations for the amount of rhizodeposition in the SOC
(Bolinder et al., 1999). The simulated kinetics of the
ef®ciency and fraction indices leveled off at about
0.20 when the labile pools had established a steady
91
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
Table 1
Simulated d value, total SOC, SOC of corn origin, and above-ground production on d 365, 1992, associated with deviations from the calibrated parameters
Controlling parameters a
d value in
top 15 cm b
SOC in top 15 cm b
(t C ha 21)
SOC from corn in top Above-ground dry mass
15 cm b (t C ha 21)
(cumulative, t ha 21)
Microbial assimilation ef®ciency
8.0
0.6
0.4
218.96 (20.5)
219.05 (0.0)
219.15 (0.5)
50.6 (1.1)
50.0 (0.0)
49.3 (21.3)
6.49 (10.0)
5.90 (0.0)
5.17 (212.4)
136.3 (22.3)
139.6 (0.0)
142.7 (2.2)
Resistant root, decay rate
constant (per day)
0.00002
0.0002
0.002
219.01 (20.2)
219.05 (0.0)
219.24 (1.0)
50.3 (0.6)
50.0 (0.0)
48.6 (22.8)
6.19 (4.9)
5.90 (0.0)
4.53 (223.2)
139.8 (0.1)
139.6 (0.0)
138.9 (20.5)
R2: daily increase of the
shoot-to-root ratio
0.05 {5.7}
0.07 {7.7}
0.11 {11.7}
218.85 (21.0)
219.05 (0.0)
219.29 (1.3)
51.5 (3.0)
50.0 (0.0)
48.4 (23.1)
7.36 (24.8)
5.90 (0.0)
4.26 (227.8)
133.7 (24.2)
139.6 (0.0)
146.2 (4.7)
R3: fraction of daily root mass
increase set in top 15 cm depth c
0.75, 0.65, 0.60 [67] 218.92 (20.7)
0.65, 0.60, 0.55 [57] 219.05 (0.0)
0.55, 0.50, 0.45 [44] 219.23 (0.9)
51.0 (1.9)
50.0 (0.0)
48.8 (22.4)
6.85 (16.1)
5.90 (0.0)
4.64 (221.4)
139.8 (0.1)
139.6 (0.0)
139.0 (20.4)
51.3 (2.0)
50.0 (0.0)
49.0 (21.9)
7.19 (21.9)
5.90 (0.0)
4.49 (223.9)
126.2 (29.6)
139.6 (0.0)
148.4 (6.3)
R5: mass ratio of
rhizodeposition-to-root daily
production
12
6.0
3.0
218.87 (20.9)
219.05 (0.0)
219.19 (0.7)
a
Bold numbers correspond to the calibrated values. Numbers in braces ({) give the shoot-to-root mass ratio at harvest in 1984. Numbers in brackets ([) give
the ratio of root mass in top 15 cm at harvest in 1984.
b
Numbers in parenthesis give the percentage change relative to the calibrated value.
c
Parameter values assigned during plant growth from 0 to 10, 10 to 20, and 40 d after emergence.
state (Fig. 6). Further increase would be very slow and
controlled by the dual effect of the incorporation of the
C from corn into the stable SOC, and the decay of the
stabilized SOC of soil origin.
References
Allmaras, R.R., Nelson, W.W., Voorhees, W.B., 1975. Soybean and corn
rooting in southwestern Minnesota: II. Root distributions and related
water in¯ow. Soil Science Society of America Proceedings 39, 771±
777.
Anderson, E.L., 1988. Tillage and N fertilization effects on maize root
growth and root:shoot ratio. Plant and Soil 108, 245±251.
Balesdent, J., Balabane, M., 1996. Major contribution of roots to soil carbon
storage inferred from maize cultivated soils. Soil Biology & Biochemistry 28, 1261±1263.
Barber, D.A., Martin, J.K., 1976. The release of organic substances by
cereal roots into soil. New Phytologist 76, 69±80.
Benner, R., Fogel, M.L., Sprague, E.L.K., Hudson, E., 1987. Depletion of
13
C lignin and its implications for stable carbon isotopic studies. Nature
329, 708±710.
Biondini, M., 1988. Carbon and nitrogen losses through root exudition by
Agropyron cristatum, A. Smithiii and Bouteloua gracilis. Soil Biology
& Biochemistry 20, 477±482.
Bolinder, M.A., Angers, D.A., Giroux, M., LaverdieÁre, M.R., 1999. Estimating C inputs retained as soil organic matter from corn (Zea mays L.).
Plant and Soil 215, 85±91.
Bottner, P., Pansu, M., Sallih, Z., 1999. Modelling the effect of active roots
on soil organic matter turnover. Plant and Soil 216, 15±25.
Cerri, C., Feller, C., Balesdent, J., Victoria, R., Plenecassagne, A., 1985.
Application du tracage isotopique naturel en 13C aÁ l'eÂtude de la dynamique de la matieÁre organique dans les sols. Comptes Rendus de l'AcadeÂmie des Sciences de Paris T.300 II (9), 423±428.
Clapp, C.E., Allmaras, R.R., Layese, M.F., Linden, D.R., Dowdy, R.H.,
2000. Soil organic carbon and carbon-13 abundance as related to tillage,
crop residue, and nitrogen fertilization under continuous corn management in Minnesota. Soil and Tillage Research 55, 127±142.
Clay, D.E., Clapp, C.E., Linden, D.R., Molina, J.A.E., 1989. Nitrogentillage-residue management: 3. Observed and simulated interactions
among soil depth, nitrogen mineralization, and corn yield. Soil Science
147, 319±325.
Davenport, J.R., Thomas, R.L., Mott, S.C., 1988. Carbon mineralization of
corn (Zea mays L.) and bromegrass (Bromus inermis Leyss)
components with an emphasis on the below-ground carbon. Soil
Biology & Biochemistry 20, 471±476.
Hadas, A., Parkin, T.B., Stahl, P.D., 1998. Reduced CO2 release
from decomposing wheat straw under N-limiting conditions: simulation of carbon turnover. European Journal of Soil Science 49,
487±494.
Huggins, D.R., Fuchs, D.J., 1997. Long-term N management effects on corn
yield and soil C of an Haplustoll in Minnesota. In: Paul, E.A., Paustian,
K., Elliott, E.T., Cole, C.V. (Eds.). Soil Organic Matter in Temperate
Agroecosystems: Long-Term Experiments in North America, CRC
Press, Boca Raton, pp. 121±128.
Huggins, D.R., Clapp, C.E., Allmaras, R.R., Lamb, J.A., Layese, M.F.,
1998. Carbon dynamics in corn-soybean sequences as estimated from
natural carbon-13 abundance. Soil Science Society of America Journal
62, 195±203.
Hunt, H.W., 1977. A simulation model for decomposition in grasslands.
Ecology 58, 469±484.
Johnen, G.G., Sauerbeck, D.R., 1977. A tracer technique for measuring
growth, mass and microbial breakdown of plant roots during vegetation.
Ecological Bulletin (Stockholm) 25, 366±373.
Jones, C.A., 1985. C4 Grasses and Cereals: Growth, Development, and
Stress Response, Wiley, New York.
Linden, D.R., Clapp, C.E., Dowdy, R.H., 2000. Long-term corn grain and
stover yields as a function of tillage and residue removal in east central
Minnesota. Soil and Tillage Research (in press).
Meharg, A.A., Killham, K., 1988. A comparison of carbon ¯ow from prelabelled and pulse-labelled plants. Plant and Soil 112, 225±231.
92
J.A.E. Molina et al. / Soil Biology & Biochemistry 33 (2001) 83±92
Molina, J.A.E., Smith, P., 1998. Modeling carbon and nitrogen processes in
soils. Advances in Agronomy 62, 253±297.
Molina, J.A.E., Clapp, C.E., Shaffer, M.J., Chichester, F.W., Larson, W.E.,
1983. NCSOIL, a model of nitrogen transformations in soil: description,
calibration and behavior. Soil Science Society of America Journal 47,
85±91.
Molina, J.A.E., Crocker, G.J., Grace, P.R., Klir, J., Korschens, M., Poulton,
P.R., Richter, D.D., 1997. Simulating trends in soil organic carbon in
long-term experiments using the NCSOIL and NCSWAP models.
Geoderma 81, 91±107.
Nicolardot, B., Molina, J.A.E., Allard, M.R., 1994. C and N ¯uxes between
pools of soil organic matter. Model calibration with long-term incubation data. Soil Biology & Biochemistry 26, 235±243.
Whipps, J.M., 1984. Environmental factors affecting the loss of carbon
from the roots of wheat and barley seedlings. Journal of Experimental
Botany 35, 767±773.
Wood, M., 1987. Predicted microbial biomass in the rhizosphere of barley
in the ®eld. Plant and Soil 97, 303±314.