taken into account Chartier et al., 1983. Thus, to account for the variation in ground cover until
the stem elongation phase, an effective index of ground cover was introduced to estimate the in-
tercepted radiation Jones, 1992, and Eq. 1 was applied within the row. The LAI, measured over
the whole plot, was thus related to the fraction occupied by the row within the plot, according to
a formula derived from Eq. 1:
PARa
L
= GCR × PAR
L
× 1 − r
L
× {1 − e
[ − K
L
× LAIGCR]
} 2
where GCR is the ground cover ratio unitless. The ground cover ratio was estimated for the
eight first sampling dates, from numerical treat- ment of digitised picture taken at a 1.5 m height
above the ground. Between measurements, the GCR was interpolated using a regression of LAI
on GCR obtained from the sampling dates.
2
.
5
. Calculation of RUEa In principle, RUEa may be calculated between
two successive sampling dates as the ratio of crop dry matter production to total absorbed PAR
over the corresponding time period. Here, dry matter
production was
calculated from
the changes in generated dry matter gTDM, as de-
scribed in the ‘plant measurements’ section. Al- though the measurements of gTDM were quite
precise CVs B 15, the RUEa values thus ob- tained were affected by a significant random vari-
ation, with a saw-tooth pattern. This may be explained by errors propagating from one time
interval to the next. For instance, if gTDM was under-estimated on one sampling date, then
RUEa was accordingly under-estimated on the time interval preceding this sampling, and over-es-
timated on the following time interval. In order to smooth this variability, one chose to estimate
RUEa over longer time periods spanning two to three sampling dates, and corresponding to differ-
ent development stages. Here, RUEa was taken as the slope of the linear regression unconstrained
between cumulative gTDM production and ab- sorbed PAR, over these time intervals.
2
.
6
. Calculation of NNI The N nutritional status for each treatment was
quantified using the NNI Lemaire and Gastal, 1997 which is calculated as follows:
NNI = NmNc 3
where Nm is the measured total N concentration for all the aerial parts and Nc is the critical total
N concentration calculated for the value of ADM measured in situ; Nm and Nc are expressed as
of ADM. Nc is the minimum N concentration needed to obtain the maximum dry matter pro-
duction by the crop. Nc is regarded as constant for low biomasses, and subsequently drops as
ADM increases, according to a relationship known as ‘N dilution’. It can be used for a large
number of herbaceous species: wheat, maize, for- age grasses, lucerne, peas, barley, durum wheat,
sorghum etc. Although N dilution is a general phenomenon, parameters of the critical N dilution
curve must be determined for each crop species Lemaire and Gastal, 1997. Nc has been estab-
lished for oilseed rape by Colnenne et al. 1998 from juvenile stages up to flowering, correspond-
ing to values of ADM from 0.2 to 6.3 t ha
− 1
: Nc = 4.63
if ADM 5 0.88 t ha
− 1
4a Nc = 4.48 × ADM
− 0.25
otherwise 4b
NNI was calculated for each sampling date and was then averaged for the period of two to three
measurement dates corresponding to the calcula- tion of RUEa as a function of development stage.
A value of NNI ] 1 indicates a crop with ample N supply N non-limiting; NNI = 1 represents
optimal N nutrition. The more NNI falls below 1, the more deficient is the crop in N.
3. Results
3
.
1
. Changes in NNI o6er time Table 3 shows the changes in NNI in relation
to developmental stage for the three treatments. The NNI was far greater than unity at the begin-
ning of growth for all the treatments, showing
Table 3 Mean air temperature °C, RUEa g MJ
− 1
PARa and nitrogen nutrition index NNI of the three treatments for each developmental stage
Radiation use efficiency Mean air temperature
NNI Oilseed rape stages
N0 N135
N270 N0
N135 N270
1.13 1.11
B1 to B5 early Autumn growth 1.14
11.3 1.16
0.03 0.05
0.06 0.07
2.45 2.93
0.92 B6 to B11 Autumn growth
9.9 1.13
0.09 0.19
0.09 0.08
0.74 0.98
B12 to B16 Winter growth 0.73
4.9
a
1.01 0.06
0.10 0.01
Frost period 0.04
5.8 C1 to C2 Spring stem elongation
1.75 1.61
2.17 0.72
0.80 1.06
0.04 0.15
0.16 0.01
0.09 0.06
6.8 D1 to D2 inflorescence formation
1.80 2.78
2.69 0.64
1.10 1.29
0.22 0.07
0.18 0.07
0.08 0.08
2.52 3.38
3.90 10.8
0.50 F1 to G2 flowering
0.83 1.20
0.06 0.08
0.12 0.02
0.09 0.10
1.95 2.27
3.40 12.4
0.37
b
G3 to G4 pod formation 0.56
b
0.92
b
0.07 0.15
0.26 0.04
0.08 0.11
0.42 0.48
0.52 16.9
G5 to maturity ripening 0.09
0.07 0.08
a
Mean air temperature for the period using mean daily values including negative values.
b
NNI calculated with an extrapolation of the critical dilution curve beyond flowering Colnenne et al., 1998. Numbers in brackets corresponds to the standard error of RUEa and NNI.
that early growth was not limited by N, even when no fertiliser was applied at sowing N135
and N0 treatments. Thereafter, the course of NNI differed across treatments: i it remained
above 1 for N270, confirming that this treatment was non-limiting in N until the end of flowering;
ii it fell steadily throughout the whole growth period for N0, indicating a progressive increase in
N deficiency, which proved considerable by the time of flowering; and iii varied as a result of the
two spring applications of N for treatment N135. For the latter, N deficiency appeared and devel-
oped from the pod formation stage stage G2 onwards, implying that the N135 dose did not
sustain maximum growth after flowering.
As the N270 treatment was always non-limiting in N, it was assumed to have achieved the maxi-
mum RUEa as proposed by Be´langer et al. 1992 for tall fescue swards, also bearing in mind that
possible limitations set by other nutrients, or fungal diseases or insects were completely allevi-
ated.
3
.
2
. Green LAI, PAI and FAI changes o6er time The changes in green LAI, PAI and FAI ac-
cording to thermal-time basis 4.5°C or develop- ment stage are shown in Fig. 2. LAI increase was
approximately exponential and then linear in au- tumn. In winter time, a frost period at the begin-
ning of January caused green leaves to drop from the base of the canopy, and LAI decreased. From
the end of winter to flowering, green LAI in- creased again linearly, after which it declined very
rapidly to reach 0 at harvest.
The maximum PAI was reached at the end of pod formation G4 stage, when it amounted to
2.6 m
2
m
− 2
for N270. This high value illustrates the importance of including it when calculating
the crop PARa. The development of FAI, which depends on the
duration of flowering ca. 1 month, was very sporadic. However the maximum FAI was 0.55
m
2
m
− 2
for N270, which justifies including it in the radiation balance.
N had a very large effect on the morphogenesis of oilseed rape since the N deficiencies achieved
with N0 and to a lesser extent with N135 caused a very clear decrease in green LAI and also in PAI
and FAI comparing to N270.
3
.
3
. PAR assessment for each layer of the canopy As the values of green LAI and PAI were
significantly less for N0 and N135 than for N270, the sum of PARa PARa absorbed by the green
LAI and PAI was significantly different between the three treatments Fig. 3a. At harvest, the
PARa absorbed by the sole green LAI was 765, 683 and 521 MJ m
− 2
for N270, N135 and N0 respectively; when including the pods the PARa
became substantially larger, at 1129, 1001 and 766 MJ m
− 2
for N270, N135 and N0, respectively.
Fig. 3. Calculated cumulative absorbed photosynthetically ac- tive radiation PAR: by green LAI and PAI a, FAI and soil
b, and cumulative reflected PAR from the canopy and soil c vs. thermal-time from emergence and development stage
for the three N treatments.
Fig. 2. Measured green LAI a, PAI b, and FAI c vs. thermal-time from emergence and development stage for the
three N treatments. Bars represent S.E..
Conversely, the PAR absorbed by the soil and that reflected by the crop unused for photosyn-
thesis were higher for N0 and N135 than for N270 Fig. 3b,c. Neither the amount of PAR
absorbed by the FAI range: 40 – 60 MJ m
− 2
, nor the one reflected by the crop range: 112 – 163 MJ
m
− 2
were negligible Fig. 3c.
3
.
4
. ADM, DDM and gTDM changes o6er time The evolution of the different biomass fractions
measured on the three treatments according to thermal-time or development stage is shown in
Fig. 4. The dynamics of the accumulation of ADM were similar to those obtained for green
LAI, namely: i at first, exponential growth; ii
next a fall in ADM resulting from leaf fall in winter; iii linear growth after the end of winter
until the end of flowering; and lastly iv very slow or zero growth at the end of the growing period
Fig. 4a. The amount of estimated dead DM DDM was very large at 480 g m
− 2
for N270 at the end of season Fig. 4b. Estimates of DDM
were consistent with the rest of the DM since positive variations were always obtained when
calculating the generated TDM gTDM, despite heavy leaf falls in winter Fig. 4c. The amount of
gTDM produced was very large for N270, being 2347 g m
− 2
at harvest. Also, as with green LAI, there was a strong effect of N on DM production
with a reduction in gTDM of 51 for N0 and 22 for N135 Fig. 4c. This resulted in signifi-
cant yield losses: 5.36 t ha
− 1
of grains were produced for N270 compared to only 4.54 and
3.07 t ha
− 1
for N135 and N0, respectively Gabrielle et al., 1998a.
3
.
5
. Variation in RUEa Table 3 lists the values and standard errors of
calculated RUEa,
for different
development stages, along with gTDM and PARa for LAI
and PAI. The RUEa significantly varied with the development stage and also with the N treatment,
from 0.74 to 3.90 g MJ
− 1
. It was also found that the RUEa on treatment N270 was always signifi-
cantly greater than on treatments N0 and N135 when their NNI was less than 1 N deficiency,
confirming the effect of N on RUEa in a situation of N deficiency. On the other hand the RUEa was
the same for all treatments when NNI was above 1 N non-limiting situation. However, as the
RUEa varied much more with the development stage than between the other two N treatments,
there was no strict single relationship between RUEa and NNI Table 3. Thus, to remove any
effect other than crop N status NNI value on the RUEa, the ratio of actual to maximum RUE
RUEaRUEa
max
was calculated for each devel- opment stage, as proposed by Be´langer et al.
1992 for tall fescue swards fertilised with differ- ent rates of N. Since the NNI of the N270 treat-
ment was always ] 1 N non limiting until flowering, it was assumed that the maximum
RUEa RUEa
max
was always obtained for the N270 treatment. Consequently, by definition,
RUEaRUEa
max
was always equal to 1 for N270. Fig. 5 shows the relationship between NNI and
RUEaRUEa
max
for the other two N treatments, when NNI B 1. A close relationship was obtained
between NNI and the RUEa ratio showing a sharp decrease of RUEa from a NNI value of ca.
1 – 0.4. A linear regression yielded a good fit of the response of RUEa to N deficiency:
RUEaRUEa
max
= 0.74 × NNI + 0.23
5 Statistical
indicators were
satisfactory: RMSE = 0.028 g MJ
− 1
PARa and r
2
= 0.919 8
d.f..
Fig. 4. Measured aerial dry matter ADM a, cumulative dead DM b, and generated total dry matter gTDM c vs.
thermal-time from emergence and developmental stage for the three N treatments. Bars represent S.E..
Fig. 5. Relationship between nitrogen nutrition index NNI and RUEaRUEa
max
. Bars represent S.E..
However, since the NNI value of N270 was not ]
1 after the pod formation stage, Eq. 6 was used to calculate an estimate of RUEa
max
at this later stage for the treatment N270. Fig. 6 shows
the relationship between periodic mean air tem- perature T and RUEa
max
for stages B6 6 leaves to G1 beginning of pod formation. A significant
relationship was obtained between temperature and RUEa
max
; the following exponential model gave a satisfactory fit to the data:
RUEa
max
= 3.5 × [1 − 13.1 × e
0.6 × T
] 6
Statistical indicators were acceptable: RMSE = 0.27 g MJ
− 1
PARa; r
2
= 0.907 4 d.f.. This may
be considered as an approximation of the temper- ature effect on RUEa for oilseed rape. However,
this clearly shows that below a given threshold 6 – 7°C the RUEa was dramatically reduced. On
the other hand, the temperature does not explain the low value of RUEa
max
at early stages after emergence B1 – B5 leaves or during the ripening
stage.
4. Discussion
