D.J. Hunsaker et al. Agricultural and Forest Meteorology 104 2000 85–105 93
apply the FAO-56 calculations to the conditions of the present study. These included daily values for the crop
height, the fraction of the soil surface wetted by irri- gation or precipitation f
w
, and the fraction of the soil surface exposed to sunlight and air denoted as 1−f
c
, as well as two drying parameters for the specific soil.
For both the 1995–1996 and 1996–1997 experiments, daily values for crop height were estimated by lin-
ear interpolation of the weekly plant height measure- ments. Also, for both years, a value of 1.0 was used
for f
w
following precipitation events i.e. the entire soil surface was assumed to have been wetted. Follow-
ing subsurface drip irrigations, the values used for f
w
were calculated using the FAO-56 recommendations for subsurface drip systems as follows:
f
w
= 0.401 − 0.67f
c
4 where f
c
represents the fraction of the soil surface covered by canopy.
Values for the fraction of the exposed soil surface, 1−f
c
, were approximated from the fraction of PAR transmitted through the canopy, as determined from
light bar measurements. For the 1995–1996 treat- ments, daily values of fT
PAR
≈1−f
c
were estimated by linear interpolation between the days that the
fT
PAR
were measured. For the 1996–1997 treatments, estimates of daily fT
PAR
were obtained from the daily NDVI data for that season using a calibration model
developed from the 1995–1996 fT
PAR
and NDVI data. The soil parameters needed in the procedures are the
readily evaporable water REW, which is the maxi- mum depth of evaporation from the soil surface that
can occur during the energy limiting stage, and the to- tal evaporable water TEW, which is the maximum
depth of water that can be completely evaporated from an initially wetted soil surface. For the clay loam soil
at the FACE site, a value of 10 mm was assumed for REW, an average value suggested by FAO-56 for clay
soils. Assuming a 0.10 m soil evaporation layer, TEW was calculated as 20 mm using the equation provided
for estimating TEW in FAO-56, which is based on the field capacity and wilting point values for the surface
layer.
Values of the water use efficiency were calculated for each treatment plot using the total cumulative ET
estimates and measured grain yields, where WUE equals final grain yield per unit of ET, expressed as
kgm
3
. 2.8. Data analysis
Statistical analyses to evaluate significant treat- ment effects on ET and WUE were made using a
strip-split-plot model, as described by Little and Hills 1978. The analyses were performed using the Gen-
eral Linear Models GLM procedure of SAS SAS Institute Inc., 1988. The strip-split-plot model had
three parts. Part 1 included the replication effect ρ and the CO
2
main effect α; the error term used for evaluating the effect of CO
2
was ρα. Part 2 included the soil nitrogen effect β, which was eval-
uated with the error term, ρβ. Part 3 included the interaction term αβ, which was evaluated by the
residual mean-square error.
3. Results
3.1. Basal crop coefficients, K
cb
Fig. 3 1995–1996 and Fig. 4 1996–1997 show the seasonal variation of the average K
cb
values, de- rived from the soil water depletion measurements, for
each of the four, CO
2
by soil nitrogen treatment com- binations. In both years, the K
cb
for the C–H treatment increased to a maximum value of about 1.20 during
mid-April early grain-fill, and then decreased to a fi- nal value between 0.45 and 0.55 in early-May. As seen
in the figures, the K
cb
for the F–H treatment were gen- erally slightly lower than those for the C–H treatment
during the two seasons. The K
cb
for Low N treatments were similar to the K
cb
for High N treatments through about early-March in both seasons. However, starting
in mid-March, the K
cb
values for Low N treatments were lower relative to the High N treatments, indicat-
ing less water use among Low N plots during the latter half of the season.
3.2. Evapotranspiration Figs. 5a and 6a present the average daily ET calcu-
lated for the four treatments and Figs. 5b and 6b de- pict the occurrence and magnitudes of the individual
wetting events in the 1995–1996 and 1996–1997 sea- sons, respectively. Note that the irrigations shown in
May of both years reflect only the amounts applied to the High N treatments. The ET rates for all treatments
94 D.J. Hunsaker et al. Agricultural and Forest Meteorology 104 2000 85–105
Fig. 3. Average basal crop coefficients measured for soil water depletion periods during the 1995–1996 season for the Control–High N C–H, Control–Low N C–L, FACE–High N F–H, and FACE–Low N F–H treatments.
were generally increased for several days following ir- rigation and precipitation events due to wet soil evap-
oration. In both years, the soil evaporation component of ET was most pronounced following irrigation and
precipitation during January and February, prior to full canopy development. The amount of total soil evapo-
ration calculated for the High N treatments averaged 25 and 22 mm in 1995–1996 and 32 and 30 mm in
1996–1997 for the C–H and F–H treatments, respec-
Fig. 4. Average basal crop coefficients measured for soil water depletion periods during the 1996–1997 season for the Control–High N C–H, Control–Low N C–L, FACE–High N F–H, and FACE–Low N F–H treatments.
tively. For the Low N treatments, the total soil evapora- tion averaged 39 and 38 mm in 1995–1996 and 44 and
50 mm in 1996–1997 for the C–L and F–L treatments, respectively. The total soil evaporation was greater in
Low N than High N treatments since the portion of the soil surface exposed to sunlight during wetting
periods was somewhat larger in Low N plots due to smaller leaf area in those plots Pinter et al., 1996b,
1997.
D.J. Hunsaker et al. Agricultural and Forest Meteorology 104 2000 85–105 95
Fig. 5. Average daily ET with day of year during the 1995–1996 season for the Control–High N C–H, Control–Low N C–L, FACE–High N F–H, and FACE–Low N F–H treatments a and the irrigation and precipitation dates and amounts applied during the season b.
Fig. 6. Average daily ET with day of year during the 1996–1997 season for the Control–High N C–H, Control–Low N C–L, FACE–High N F–H, and FACE–Low N F–H treatments a and the irrigation and precipitation dates and amounts applied during the season b.
96 D.J. Hunsaker et al. Agricultural and Forest Meteorology 104 2000 85–105
In Figs. 5a and 6a, the daily ET for Low N treat- ments was decreased relative to the High N treat-
ments after mid-March as expected, since K
cb
values for Low N treatments were also decreased after that
time. The GLM analyses, applied to ET accumula- tions over 10-day periods during 1995–1996 Table 2
and 1996–1997 Table 3, indicated that the soil N treatment effect on ET was significant p0.05 for
each 10-day period between late-March and the re- mainder of the season in both years. Although signif-
icant N treatment effects were indicated for the small amount of ET accumulated during the first two 10-day
periods in January in 1996 Table 2, these were as- sumed to be random effects and not related to soil ni-
trogen treatment, since the periods occurred prior to the treatment fertilizer applications that were begun
on 30 January. The period in late-March, when the effects of the nitrogen treatment on ET were first sig-
nificant, corresponded approximately to the beginning of anthesis. At that time the Low N treatment had re-
ceived a total of 45 and 10 kg N ha
− 1
for 1995–1996 and 1996–1997, respectively, compared to a total of
175 kg N ha
− 1
for the High N treatment Table 1. The limited N fertilizer applied to the Low N treatments
resulted in a significant decline in the leaf area and the rate of aboveground biomass production relative to
the High N treatments during the latter half of the two growing seasons. Thus the reductions in water use ob-
served for the Low N treatments during the latter half of the seasons were consistent with decreasing leaf
area and growth rate for the Low N treatments dur- ing that period. The total cumulative ET shown at the
bottom of Tables 2 and 3 for Low N treatments was 25 and 22 less than for High N treatments in Con-
trol and 23 and 20 less in FACE in the1995–1996 and 1996–1997 seasons, respectively. In both years,
the difference in total cumulative ET between soil N treatments was highly significant p0.01. Each treat-
ment in 1996–1997 attained a cumulative ET that was within 6 to 16 mm of the treatment’s cumulative ET in
1995–1996 suggesting little variation in climatic con- ditions between the two seasons. Table 4 indicates that
differences in average climatic data between the two seasons were small.
The GLM results also revealed that the effects of CO
2
enrichment on ET over 10-day periods were not often statistically significant. However, the GLM did
indicate that ET for FACE was significantly lower than for Control during two 10-day periods in March 1996
and during three 10-day periods in 1997. A CO
2
by nitrogen interaction effect on the ET was also present
for two periods during 1997. Although the ET dif- ferences between CO
2
treatments over the short time periods were usually not statistically significant, there
appeared to be a consistent tendency during both sea- sons for the ET values to be lower for F–H than C–H
during the 10-day periods. Under Low N, a consis- tently lower ET for F–L than C–L was not apparent. As
shown in Tables 2 and 3, the total cumulative ET was 21 and 23 mm lower for F–H than C–H in 1995–1996
and 1996–1997, respectively, whereas total cumula- tive ET for the F–L was only 3 and 6 mm smaller than
that for C–L in the first and second years, respectively. The GLM analyses indicated that the cumulative ET
differences between CO
2
treatments were not highly significant.
The effects of CO
2
enrichment on ET were fur- ther examined by evaluating the differences in cu-
mulative ET between FACE and Control on a daily basis as shown in Fig. 7a 1995–1996 and Fig. 7b
1996–1997 for High N treatments and Fig. 8a 1995–1996 and Fig. 8b 1996–1997 for Low N
treatments. For the High N treatment in 1995–1996, Fig. 7a shows that the daily cumulative ET for F–H
was initially lower than that of C–H, and then about equal to C–H for a short period coinciding with
mid-tillering i.e. late-January through mid-February 1996. However, from late-February stem extension
through about the end of March anthesis, the daily ET for F–H increased at a slower rate than that for
C–H, such that by DOY 83 23 March F–H had accumulated 6.3 less ET than C–H. After anthesis,
the cumulative ET for F–H continued to lag about 5–6 behind C–H until late-April. However, F–H
then achieved higher ET than C–H during early-May late grain-fill, which might be indicative of differ-
ences in the senescence between the two treatments i.e. F–H may have senesced slightly later than C–H.
As a result of the late season ET increase for F–H, the total cumulative ET for F–H for the 1995–1996
season was only 3.7 less than the total ET for C–H.
For the High N treatment in 1996–1997, Fig. 7b shows that the daily cumulative ET for F–H lagged
behind C–H considerably during January. The ET for F–H then increased at a more rapid rate than C–H, such
that by late-February DOY 54, the cumulative ET for
D.J. Hunsaker et al. Agricultural and Forest Meteorology 104 2000 85–105 97
Table 2 Treatment means and GLM results for wheat evapotranspiration ET determined for 10-day periods during the 1995–1996 season
Period 1996 Treatments mm
a
Effects PF
b
Julian days Dates
C–H C–L
F–H F–L
CO
2
N CO
2
× N
004–011
c
04–11 January 3
2 3
1 0.82
0.04 0.96
012–021 12–21 January
9 5
9 6
0.98 0.05
0.97 022–031
22–31 January 14
11 15
12 0.97
0.08 0.63
032–041 01–10 February
30 30
30 29
0.92 0.78
0.49 042–051
11–20 February 29
24 26
25 0.70
0.38 0.10
052–061 21 February–01 March
34 30
28 30
0.03 0.35
0.08 062–071
02–11 March 51
44 48
42 0.10
0.19 0.99
072–081 12–21 March
45 44
42 38
0.04 0.15
0.60 082–091
22–31 March 63
50 60
51 0.80
0.01 0.72
092–101 01–10 April
79 62
76 64
0.78 0.01
0.29 102–111
11–20 April 88
73 86
72 0.38
0.02 0.64
112–121 21–30 April
88 74
88 76
0.94 0.03
0.65 122–130
d
01–09 May 68
na
e
69 na
e
0.79 na
e
na
e
Cumulative total 601
449 580
446 0.30
0.01 0.06
a
Treatments are Control–High nitrogen C–H, Control–Low nitrogen C–L, FACE–High nitrogen F–H, and FACE–Low nitrogen F–L.
b
The probability of a larger F statistic PF for the CO
2
, nitrogen N, and the CO
2
× N interaction effects from the GLM.
c
The first period was 8 days.
d
The final period was 9 days.
e
na: Not applicable.
Table 3 Treatment means and GLM results for wheat evapotranspiration ET determined for 10-day periods during the 1996–1997 season
Period 1997 Treatments mm
a
Effects PF
b
Julian days Dates
C–H C–L
F–H F–L
CO
2
N CO
2
× N
002–012
c
02–12 January 3
3 3
2 0.66
0.65 0.99
013–022 13–22 January
12 12
10 11
0.40 0.90
0.95 023–032
23 January–01 February 17
18 15
13 0.02
0.91 0.65
033–042 02–11 February
21 23
23 21
0.49 0.47
0.07 043–052
12–21 February 38
37 39
36 0.91
0.17 0.37
053–062 22 February–03 March
28 26
27 26
0.70 0.33
0.88 063–072
04–13 March 52
52 47
44 0.05
0.80 0.67
073–082 14–23 March
64 55
63 58
0.74 0.08
0.03 083–092
24 March–02 April 72
62 67
61 0.03
0.01 0.51
093–102 03–12 April
68 53
64 58
0.69 0.02
0.41 103–112
13–22 April 83
68 81
70 0.91
0.02 0.05
113–122 23 April–02 May
83 56
79 59
0.57 0.03
0.23 123–131
d
03–11 May 54
na
e
54 na
e
0.84 na
e
na
e
Cumulative total 595
465 572
459 0.20
0.01 0.25
a
Treatments are Control–High nitrogen C–H, Control–Low nitrogen C–L, FACE–High nitrogen F–H, and FACE–Low nitrogen F–L.
b
The probability of a larger F statistic PF for the CO
2
, nitrogen N, and the CO
2
× N interaction effects from the GLM.
c
The first period was 11 days.
d
The final period was 9 days.
e
na Not applicable.
98 D.J. Hunsaker et al. Agricultural and Forest Meteorology 104 2000 85–105
Table 4 Seasonal average climatic data at the experiment site
Time period Solar radiation
Maximum Minimum
Grass-reference evapotranspiration, MJm
2
day temperature
◦
C temperature
◦
C ET
o
mm per day 01 January–15 May, 1996
20.5 26.3
7.1 5.1
01 January–15 May, 1997 20.2
25.6 7.4
5.0
Fig. 7. Average daily cumulative ET and average daily difference in cumulative ET between FACE F–H and Control C–H for the High N treatment in the 1995–1996 season a and for the High N treatment in the 1996–1997 season b.
D.J. Hunsaker et al. Agricultural and Forest Meteorology 104 2000 85–105 99
Fig. 8. Average daily cumulative ET and average daily difference in cumulative ET between FACE F–L and Control C–L for the Low N treatment in the 1995–1996 season a and for the Low N treatment in the 1996–1997 season b.
F–H was nearly equal to that for the C–H treatment. However, this trend was reversed during the remainder
of the season and the cumulative ET for F–H gradually decreased relative to the C–H treatment, reaching a
reduction of 4.7 on DOY 106 ≈early grain-fill. The total cumulative ET for F–H at the end of the
1996–1997 season was 4.0 less than the total for C–H.
For the Low N treatment in 1995–1996 Fig. 8a, the cumulative ET for F–L was higher than that
for C–L through about early-March. However, from mid-March through late-March, there was a change in
the ET response between the two Low N treatments, such that the cumulative ET for F–L fell behind that
for C–L by as much as 3.8 on DOY 082 22 March. This trend then reversed starting in late-March when
F–L attained more ET than C–L. Consequently, the total cumulative ET for F–L for the 1995–1996 season
was only 0.7 less than that of the C–L treatment.
Unlike the previous season, cumulative ET for the F–L treatment lagged far behind the cumulative ET for
the C–L treatment through early-March in 1996–1997
100 D.J. Hunsaker et al. Agricultural and Forest Meteorology 104 2000 85–105
Table 5 Treatment means for water use efficiency kgm
3
in 1995–1996
a
Treatments High soil N
Low soil N Ratio of Low NHigh N
Control 1.233 a
1.289 a 1.05
FACE 1.466 b
1.449 b 0.99
Ratio of FACEControl 1.19
1.12
a
Data are means of four replicates for each treatment; means followed by different letters in a row or column are significantly different at the 0.05 probability.
Fig. 8b. On DOY 036 05 February, F–L had accu- mulated 25 less ET and on DOY 075 16 March,
10 less ET than that for C–L. However, similar to the first year, the F–L treatment attained more ET than
the C–L treatment between mid-March and the end of the season. On DOY 090 31 March, the reduction
in cumulative ET for the F–L treatment had declined to 5. By the end of the 1996–1997 season, the total
cumulative ET for F–L was only 1.2 less than the total cumulative ET for C–L.
3.3. Water use efficiency 3.3.1. 1995–1996
Final grain yield of spring wheat in 1995–1996 in- creased an average of 15 and 12 under CO
2
enrich- ment for the High and Low N treatments, respectively.
The Low N treatment reduced grain yield an average of 22 and 24 within Control and FACE treatments,
respectively. Note that the reduction in grain yield for the Low N treatments was nearly proportional to the
reduction in total cumulative ET for the treatments i.e. 25 and 23 less ET occurred in the C–L and F–L
treatments than the C–H and F–H treatments, respec- tively.
The mean values for WUE, expressed as final grain yield per unit of seasonal ET kgm
3
, are presented for the 1995–1996 treatments in Table 5. The FACE
treatment resulted in a 19 and 12 increase in WUE
Table 6 Treatment means for water use efficiency kgm
3
in 1996–1997
a
Treatments High soil N
Low soil N Ratio of Low NHigh N
Control 1.004 a
1.085 a 1.08
FACE 1.232 b
1.163 b 0.94
Ratio of FACEControl 1.23
1.07
a
Data are means of four replicates for all treatments except the Control, Low N treatment, which had three replicates; means followed by different letters in a row or column are significantly different at the 0.01 probability.
over Control for the High N and Low N treatments, respectively. The GLM results revealed that the effect
on WUE due to CO
2
was significant at p=0.05. For FACE, the Low N treatment achieved a WUE that
was 1 lower than that for the High N treatment, but for Control, the WUE for the Low N treatment was
5 higher than that for the High N treatment. The GLM also indicated that the effects on WUE due to
nitrogen level p=0.77 or CO
2
by nitrogen interaction p=0.30 were not significant.
3.3.2. 1996–1997 Final grain yield of spring wheat in 1996–1997 in-
creased an average of 17 and 5 under CO
2
enrich- ment for the High and Low N treatments, respectively.
The Low N treatment reduced grain yield an average of 13 and 24 in the Control and FACE treatments,
respectively. Thus, the reduction in grain yield for the Low N treatments was less than the reduction in to-
tal cumulative ET under the Control treatment 22, but under FACE, the yield reduction for the Low N
treatment was slightly greater than the ET reduction of 20.
For 1996–1997, the FACE treatment resulted in a 23 increase in WUE over the High N and a 7 in-
crease over the Low N treatments Table 6. The GLM results revealed that the effect on WUE due to CO
2
was significant at p=0.01. As in the previous year, the Control Low N treatment achieved a WUE that was
D.J. Hunsaker et al. Agricultural and Forest Meteorology 104 2000 85–105 101
higher 8 than that for the Control High N treatment, whereas the WUE for the FACE Low N treatment was
again lower 6 than that for the FACE High N treat- ment. As in the first year, the effects on WUE due
to nitrogen level p=0.90 or due to CO
2
by nitrogen interaction p=0.25 were not highly significant.
4. Discussion