duce complexity and to adapt the simulation to an intended use in decision support.
2. Materials and methods
2
.
1
. Experiments The effects of interspecific radiation competi-
tion between common lambsquarters aka. fat hen, Chenopodium album L., Conrad Appel,
Darmstadt, Germany and cauliflower Brassica oleracea L. convar. botrytis var. botrytis L. cv.
Fremont, Royal Sluis, Neustadt, Germany were examined in field trials in 1994 and 1995. The trial
in the second year was carried out as a spring ‘experiment 1’ and a summer planting ‘experi-
ment 2’ with identical layouts Table 1. The experiments were made on a silty loam at the
experimental station in Ruthe near Hanover, Ger- many. To achieve different degrees of competi-
tion, planting density of C. album was varied at two levels. In addition to mixed stands, plots with
C. album in monoculture were planted on both densities. The experiments were set up in com-
pletely randomised blocks with four replications. Within each plot, separate subplots were used for
four successive samplings throughout the growing period.
2
.
1
.
1
. Culti6ation Cauliflower seedlings were cultivated in a green-
house until the transplants had an average fresh weight of c. 5 g and had four to five leaves. At
least 1 week prior to planting the transplants were hardened off in a cold frame. In the field, the
plants were arranged in a rectangular pattern in 1994 and in an isometric pattern in 1995. Nitro-
Table 1 Cultivation and sampling data of field experiments
Cauliflower 1994
1995 experiment 1 1995 experiment 2
18 February Sowing date
24 February 8 June
18 April 24 April
18 July Date of
transplanting Harvest dates
24 May 16 May
9 August 31 May
9 June 23 August
13 September 28 June
20 June 10 July
9 October 1 July
a
Planting pattern m 0.45×0.52
0.5×0.6 0.45×0.52
3.33 Plants m
− 2
4.27 4.27
Plot area m
2
7.28×2.0 8.4×4.0
7.28×2.0 Plants per harvest
4 4
10 Chenopodium album
Sowing date 19 April
3 April 27 June
20 July –
Date of 3 May
transplanting Plants m
− 2
Target: 25, emerged: 24 Low density
Mixed: 12.8, pure: 17.1 Mixed: 12.8, pure: 17.1
Target: 50, emerged: 41 High density
Mixed: 38.5, pure: 42.7 Mixed: 38.5, pure: 42.7
Five or three early or late Plants per harvest
Five or three early or late Five or three early or late
growth stages, respectively growth stages, respectively
growth stages, respectively
a
No dry matter data available.
gen fertilisation was given shortly before and 4 weeks after transplanting: after soil sampling, the
mineral N supply was replenished to 130 kg N ha
− 1
at the first and to 270 kg N ha
− 1
at the second application Lorenz et al., 1989. In 1994,
C. album was sown by hand directly into the field. After a germination test, a quantity of seeds ap-
propriate to obtain target densities was mixed with quartz sand and spread out uniformly over
the plot. In low density plots, the target density was achieved, but it fell short in high density plots
Table 1. Still, a considerable variation between the treatments could be expected. Due to a long
juvenile phase, the onset of competition with cauliflower was notably delayed. To achieve early
competition, C. album plants were raised and transplanted to the field for the 1995 experiments.
Four to five seeds were laid into peat-filled cone- trays and singled after emergence. After raising
the seedlings in a greenhouse, they were trans- planted when having about three nodes and an
average height of 3 cm. The weed densities were achieved by placing one or three C. album be-
tween two cauliflower plants for low and high density plots, respectively. In pure weed stands,
the same pattern was used, except that the posi- tion of the cauliflower plant was filled by another
weed. This resulted in slightly higher planting densities in pure than in mixed stands. All plots
were well irrigated to avoid water limitations and weeds emerging spontaneously were removed by
hand. Pests and insects were controlled chemically for both crops and weeds.
2
.
1
.
2
. Data collection At four successive sampling dates, a number of
cauliflower and C. album plants were taken from the designated sub-plots. Plants were dug out to a
depth of c. 20 cm; the roots were washed and separated from the shoot just above the first
lateral root branching off. Measurements of shoot length from base to top and its greatest width
gave the plant height and diameter, respectively, both with an accuracy of c. 5 mm. The shoots
were divided into stems including petioles, green leaves, senescent leaves and reproductive organs
only cauliflower. All plant parts were dried for at least 3 days at 100°C and weighted. The area of
green leaves was measured with a LI-COR 3100 Area Meter LI-COR Inc., Lincoln, NE, USA.
From these data the green leaf area index LAI was calculated. Hourly data of air temperature,
global incoming radiation and relative humidity were supplied by an on-site automatic weather
station Campbell Scientific, UK.
For further details see Ro¨hrig et al. 1999 and Ro¨hrig 1999.
2
.
2
. Model description The model presented here was derived from a
crop growth simulation for faba beans Stu¨tzel, 1995a,b, which calculates plant growth based on
the insolation. The profile of absorbed radiation in the canopy is computed assuming a leaf area
distribution in three dimensions. From the ab- sorbed radiation and the efficiency of its use, the
total growth rate of the competing species is estimated. The biomass produced is partitioned
into the various plant organs. From the leaf dry weight and the specific leaf area, the LAI is
calculated, which is distributed in the canopy according to plant positions and dimensions. De-
velopmental processes such as vernalisation and ontogeny are modelled as a function of ambient
temperature Kage and Stu¨tzel, 1999.
2
.
2
.
1
. Calculations Fifty percent of the incoming global radiation
400 – 3000 nm is assumed photosynthetically ac- tive radiation PAR, 400 – 700 nm. The thermal
time scale is defined in the model as the tempera- ture sum TS °C day:
TS = T
av
with T
av
= T
min
+ T
max
2 1
where T
av
, T
min
and T
max
are the daily average, minimum and maximum temperatures in °C, re-
spectively. A base temperature of 0°C for both cauliflower Wurr et al., 1990; Grevesen and
Olesen, 1994 and C. album Angus et al., 1981 was used.
2
.
2
.
2
. Growth processes The radiation absorption submodel is described
in detail by Ro¨hrig et al. 1999. In brief, the
complete canopy volume is divided into cubic subunits, which are either empty or filled with leaf
area. Leaf area of each plant of a species is distributed in a geometric foliage envelope, which
is defined in the model domain by plant height, diameter and position. Foliage envelopes may
overlap resulting in a multiple occupation of a cube by one or more species. Radiation transmis-
sion is then calculated by following the path of a solar ray through the canopy volume until ground
level is reached. The absorption of diffuse radia- tion is calculated by selection and integration of
solar rays from 15 directions according to the Gaussian
integration algorithm
Goudriaan, 1986. This routine is also used to integrate the
diurnal course, yielding a daily total of absorbed radiation, I
a
MJ m
− 2
, Eq. 19 in Ro¨hrig et al., 1999, for each species. The increase of total plant
biomass, W
T
g m
− 2
, is calculated as a function of I
a
and the radiation-use efficiency, RUE g MJ
− 1
: dW
T
dt =
I
a
RUE 2
In 1995 it was found that a RUE of cauliflower showed a seasonal variation between the experi-
ments and b crop dry matter decreased due to profound shading by C. album. The seasonal vari-
ation of RUE can be attributed to a differing insolation that was also found in potatoes Man-
rique et al., 1991. A log – normal function was used to describe both findings:
RUE = e
1
+ e
2
e
− 0.5lnI
0, c
e
3
e
4 2
3 where e
1
to e
4
are calibration constants and I
0, c
MJ m
− 2
is the daily radiation incident on a horizontal plane above the cauliflower plants. The
constant e
1
is the minimum of the function and e
3
denotes the maximum, i.e. the irradiance at which RUE is maximum. If I
MJ m
− 2
is the insola- tion and I
a,t
MJ m
− 2
denotes the radiation absorbed by the entire canopy, then the irradiance
above species S, I
0, S
, amounts to: I
0, S
= I
− I
a, t
− I
a, H
S
4 where I
a, H
S
MJ m
− 2
is the radiation absorbed by the canopy from the height of species S, H
S
, down to ground level. In the case of monospecific
stands I
0, S
equals I ; in mixtures the term in
parentheses represents the radiation absorbed by species with leaf area between the top of the
canopy and the top of the species under consider- ation. Newly produced biomass is allocated to
root, W
R
g m
− 2
, and shoot dry matter, W
Sh
g m
− 2
, using an allometric growth equation Per- sall, 1927. As shown by Kage and Stu¨tzel 1999,
Eq. 18, the growth rates of plant organs growing allometrically are obtained by:
dW
Sh
dt =
dW
T
dt 1
1 + e
a
bW
Sh b − 1
and dW
R
dt =
dW
T
dt −
dW
Sh
dt 5
Considering roots and shoots, the term e
a
is the ratio between the initial weights Persall, 1927
and the allometric growth coefficient b denotes the fixed ratio between the relative growth rates
Thornley and Johnson, 1990.
For cauliflower, the shoot dry matter is allo- cated to vegetative, W
V
g m
− 2
, and reproductive organs, W
G
g m
− 2
, but only after curd initiation Kage and Stu¨tzel, 1999. They proposed that
during the reproductive phase the ratio of curd growth rate to shoot growth rate increases logisti-
cally in time. On the transition from the exponen- tial to the linear phase of the logistic function, the
sink strength of the curd becomes substantial. At this point, it is assumed that a fraction of the
vegetative shoot dry weight is reallocated to the curd. The reproductive growth rate is then:
dW
G
dt =
Á Ã
Í Ã
Ä f
dW
Sh
dt if
f 5 0.1 f
dW
Sh
dt +
r
t
T
av
W
6
if f \ 0.1
6
where f is a logistic growth function of thermal time after curd initiation see Eq. 21 in Kage
and Stu¨tzel, 1999 and r
t
°C
− 1
day
− 1
is the translocation rate. For the distribution of vegeta-
tive shoot dry matter into stems, W
S
g m
− 2
, and leaves, W
L
g m
− 2
, the approach of Stu¨tzel et al. 1988 and Stu¨tzel and Aufhammer 1991 is used,
which mainly assumes allometric growth. Thus, in analogy to Eq. 5 the growth rates of leaves and
stem are obtained by:
dW
L
dt =
dW
V
dt 1
1 + e
h
gW
L g − 1
and dW
S
dt =
dW
V
dt −
dW
L
dt 7
where h and g are constants. In case of low irradiance Eq. 3 may become negative. Since
stems and roots are less physiologically active, these losses are solely attributed to leaf dry
weight and all other growth rates are set to zero. It is furthermore assumed that the leaf area in-
dex, L m
2
m
− 2
, is directly linked to leaf dry weight by:
dL dt
= Á
à Í
à Ä
dW
L
dt SLA
if dW
L
dt \
dW
L
dt 0.5
L W
L
if dW
L
dt 5
8
where SLA m
2
g
− 1
is the specific leaf area of newly formed leaves. Eq. 8 also implies that per
unit dry weight loss, leaf area is decreasing only by half the actual leaf area ratio. SLA is a con-
stant input parameter for C. album, but in cauliflower it is made a function of the radiation
environment quantified by Alt 1999 personal communication:
SLA = 0.059I
0, c, av −
0.85
9 where I
0, c, av
is I
0, c
MJ m
− 2
as a running aver- age of 5 days. This relationship takes into ac-
count leaves becoming thinner when growing under low radiation. Dry matter production and
distribution of C. album are quantified in Ro¨hrig 1999. The radiation-use efficiency and the co-
efficients of the allometric growth equation are estimated and the effects of competitive stress on
the growth processes are discussed. All data from the 1995 experiments were used for calibration.
2
.
2
.
3
. De6elopmental processes Vernalisation
and flower
initiation in
cauliflower are modelled by the temperature sum approach of Wiebe 1979 as modified by Kage
and Stu¨tzel 1999. The spatial expansion of cauliflower plants is calculated as a hyperbolic
function of stem dry weight for plant height, H m, and vegetative shoot dry weight for plant
diameter, D m. Plant dimensions, however, are not allowed to be less than the initial values H
and D at transplanting, thus:
H = max H
max
W
S
w
S,h
+ W
S
, H and
D = max D
max
W
V
w
V,d
+ W
V
, D 10
where H
max
and D
max
m denote the maximum height and diameter, respectively. The calibration
constants w
S, h
g m
− 2
and w
V, d
g m
− 2
are the stem and vegetative shoot dry weights at which
half of the maximum stem heights and plant diameters are achieved. Canopy development of
C. album is modelled explicitly in response to the competitive environment Ro¨hrig, 1999. In brief,
C. album as a shade avoiding weed species adapts its morphology to the degree of crowding. The
leaf area around the extending internodes of C. album ‘effective’ LAI is assumed mainly respon-
sible for plant height modulation. In the ap- proach, stem extension is related to the thermal
time and an estimate of the ‘effective’ LAI result- ing in a different plant height of C. album for
each treatment. The relationships were calibrated using data from all treatments in the 1995 trial.
All plants in the model were assumed to have an ellipsoidal foliage envelope and the cube edge
length was set to 0.01 m Ro¨hrig et al., 1999. The latitude of the experimental site 52°15 N
was input in the model as well as the north – south alignment of the field.
Crop canopies usually have a specific planting pattern and are thus defined by a repeating basic
unit. This basic unit is specific for each treatment and is modelled accordingly by a its boundaries
in a west – east x-axis and south – north y-axis direction and b the number and position of
each species under consideration Fig. 1. In the 1994 experiment, C. album was sown by hand. In
the model, the position of C. album plants was thus distributed at random and the number of
plants corresponds with the observed planting density Table 1. In 1995 C. album was trans-
planted in a defined pattern and this was re- peated in the simulation.
Fig. 1. Distribution of cauliflower and C. album plants in the field experiments as repeated in the model. The top
and the middle row represent the mixed stands of the cauliflower experiments in 1994 A – C and 1995 D – F, re-
spectively. The bottom row shows the pure weed plots in the 1995 experiments G – H. The columns from left to right
represent weed free, low and high weed planting densities.
calculated per layer. Within a layer, a homoge- nous leaf area distribution is assumed, whereas in
the vertical dimension a parabolic density func- tion is used. Summation over all layers yields the
total radiation absorbed by each species.
The modification used here applies to the ex- tinction coefficient for diffuse radiation. Instead
of using a constant value, the absorption of solar rays from selected elevation angles in a layer is
calculated to be integrated over the hemisphere according to the Gaussian integration procedure
Goudriaan, 1986. Five points, p, are selected at defined distances, d
5p
, from the central point of the integration range of the solar elevation angle
b . The absorption of radiation originating at
these solar heights b
p
is calculated and weighted, w
5p
. For the distances and weights for the Gaus- sian integration see Ro¨hrig et al. 1999, Table 1.
The solar elevation angles are determined by: b
p
= p
2 0.5 + d
5p
for 0 B b 5 p
2 with p = [1…5]
11 The radiation extinction coefficient to be used for
integrating the absorption of diffuse radiation is Goudriaan, 1977, Eq. 2.41:
K
a
= O
av
sin b
p
1−s 12
where O
av
is the average projection of leaves, which is 0.5 for a spherical leaf angle distribution
and s is the scattering coefficient, which amounts to 0.2 for visible radiation. According to the
Gaussian integration the relative absorption of diffuse radiation, F
a,f
, is then the weighted sum of all rays considered:
F
a, f
= p
2 5
p = 1
I
p
1 − e
− K
a
L
LY
sin b
p
cos b
p
w
5p
13 where I
p
is the relative irradiance of the selected solar ray, which amounts to 1p for each solid
angle above the canopy and L
LY
is the total leaf area in layer LY. Multiplication by sin b
p
consid- ers the fraction of radiation effective on a hori-
zontal plane and the factor cos b
p
is needed when integrating the solid angles over the hemisphere.
Since the radiation absorption of canopy with a
2
.
2
.
4
. Run control The model runs start at the date of cauliflower
transplanting. C. album is initiated at the three- node stage, i.e. at about 300°C days after sowing
in the 1994 experiment and at the date of trans- planting in the 1995 experiments. The simulation
is terminated when the crop has reached harvest time, which is determined by a specified curd
diameter in cauliflower. For those treatments where curds did not reach a marketable size, the
model is stopped at the day of the final harvest.
2
.
2
.
5
. Model simplification To reduce complexity, the detailed model was
compared to a modified version of the radiation interception module XASSNM of SUCROS87
Spitters, 1989, which stratifies the canopy into a number of horizontal leaf layers, amounting to
one layer per 0.01 m canopy height. For each competing species, the absorption of diffuse and
direct radiation by sunlit and shaded leaves is
horizontally homogeneous leaf area distribution is insensitive to changes of the solar azimuth angle,
the weighted sum has finally to be multiplied by the integration range of both solar angles 2pp
2 = p
2
. Calculating the absorption of diffuse radiation in a layer with the above procedure
yields a value that is dependent on the leaf area Goudriaan, 1977 and on the depth in the
canopy. Subsequently, the ‘effective’ diffuse radia- tion extinction coefficient, K
f
, is calculated, which is needed to realise the relative absorption of
diffuse radiation given by Eq. 13, thus: F
a,f
= 1 − e
− K
f
L
LY
14 which can be rewritten as
K
f
= −
ln1 − F
a, f
L
LY
15 Eq. 13 estimates the fraction of absorbed diffuse
radiation in relation to the irradiance above the canopy. If the absorption is calculated from the
irradiance above a layer I
0, LY
, Eq. 15 is modified to give
K
f, LY
= −
ln1 − F
a, f
I
0, LY
L
LY
16 yielding the ‘effective’ diffuse radiation extinction
coefficient in a layer K
f, LY
, which is used in the routines of XASSMN.
3. Results and discussion