J.A. Laws et al. Agriculture, Ecosystems and Environment 80 2000 243–254 247
device attached to the rear axle of the trailer in which the crop was transported ‘Weyload’, Chipping Sod-
bury, Avon. This instrument was calibrated by weigh- ing harvested grass in trailers on a public weighbridge
after recording the weight measured by the ‘Weyload’. A linear regression described by x=123.39y−1138
where x=weight of grass and y=the ‘Weyload’ read- ing accounted for 98.5 of the variation between
19 samples. Samples of fresh herbage were taken at the time of harvesting, and, after drying to constant
weight, dry matter DM content was determined and DM yield calculated.
Herbage mass on offer was assessed on four occa- sions during the year to coincide with the start and
end of grazing and the first and second silage cuts. On each occasion, samples 0.1 m
2
were selected at random and cut to ground level with hand shears. At
the start of grazing, one sample was taken on each of the 10 sectors of each farmlet. On subsequent occa-
sions, 10 samples were taken on each farmlet being divided equally over all sectors under grazing. The
herbage was dried at 80
◦
C in a forced-draught oven and the DM content was determined. A representative
sub-sample of the dried herbage was milled through a 0.8 mm screen and total N was determined by Kjel-
dahl digestion.
The same herbage samples were used to assess clover proportion under grazing and 10 additional
samples were taken on the sectors closed to grazing to assess clover proportion under cutting for treat-
ment GC. Prior to taking the samples, the clover fraction as a proportion of the total herbage present
within each sample was assessed visually. Four ad- ditional samples were assessed for clover proportion
in the same sector but were not cut. The herbage taken was separated into grass and clover fractions
before drying so that the proportion of clover in the biomass could be calculated. A linear regression
equation was derived for the relationship between the actual and assessed values for clover proportion.
The mean values of the four additional assessments made on each sector were used in the equation to
calculate the mean clover percentage. Separate re- gression equations for each site were established
for swards under cutting or grazing management at the time of sampling. Estimates of clover-N fixa-
tion were made using the models Ledgard, personal communication:
• N fixed for grazed swards kg ha
− 1
=clover yield kg DM ha
− 1
×0.06; •
N fixed for cut swards kg ha
− 1
=clover yield kg DM ha
− 1
×0.068. Measured amounts of mineral fertiliser were applied
and slurry was spread with tankers of pre-determined capacity. The total N content of the slurry was deter-
mined by Kjeldahl digestion.
3. Results
3.1. Weather conditions Table 3 shows rainfall and mean air temperature
for each month from April to October and total rain- fall during the preceding November–March period,
with 30-year 1966–1995 means, for 1992–1996. Considerable contrasts in rainfall occurred during the
experiment. Of particular note is the extreme wetness of the April–October 1993 period, with rainfall during
this period being 46 higher than the 30-year mean. The November–March period in 1993–1994 and
1994–1995 were 29 and 40 wetter than the 30-year mean, respectively. Dry periods were experienced in
all years, with June–July 1994, June–August 1995 and June–July and September 1996 being exceptional. The
July–September period in 1993 was cooler than aver- age and the summer of 1995 was particularly warm.
3.2. Slurry application Slurry application was impeded by wet ground
conditions at various times during the experiment, particularly in spring 1993 and 1994 after high win-
ter rainfall and in autumn 1992–1995 after excessive rainfall in August 1992 and in September 1993–1995.
Shortfalls in the amounts of slurry spread in any year were carried over to the following year. More slurry
was generated over winter when the turnout of cat- tle was delayed because of wet ground conditions in
spring, particularly at Site 2, and during the extended housed period for treatment TN. Difficulties in meet-
ing the required targets for slurry-N returns to the farmlets were further exacerbated by the generally
low DM content of the slurries used and high stocking rates in 1992 and 1993. Over all years, the calculated
amounts of slurry to be returned to the farmlets were
248 J.A. Laws et al. Agriculture, Ecosystems and Environment 80 2000 243–254
Table 3 Rainfall distribution mm and mean air temperature
◦
C November–March
April May
June July
August September
October Rainfall
1991–1992 309
73 57
23 74
127 48
97 1992–1993
494 88
143 84
118 19
148 112
1993–1994 730
78 108
24 24
84 129
99 1994–1995
794 52
47 15
70 15
115 74
1995–1996 505
81 81
23 14
84 26
106 30-year mean
565 62
65 59
54 62
76 105
Temperature 1992
– 8.1
17.7 14.9
16.1 14.9
13.1 7.9
1993 –
9.1 10.6
14.5 14.7
14.3 12.1
8.0 1994
– 7.7
10.1 13.7
16.5 15.5
12.7 10.9
1995 –
8.5 11.1
13.7 17.6
18.9 13.5
13.2 1996
– 8.3
8.9 13.9
15.7 15.5
13.3 11.7
30-year mean 7.7
10.5 13.5
15.3 15.2
13.3 10.8
49, 59 and 49 m
3
for treatments CN, TN and GC, respectively, at Site 1 and 46, 59 and 45 m
3
for treat- ments CN, TN and GC, respectively, at Site 2. Corre-
sponding values for the amounts of slurry applied as a proportion of the amounts required were 90, 93 and
92 at Site 1 and 87, 105 and 84 at Site 2.
3.3. Nitrogen inputs, outputs, efficiencies and surpluses
Nitrogen inputs, outputs, efficiencies and surpluses using a ‘surface balance’ for the soil Jarvis, 1999
for each treatment are shown in Table 4. Average fertiliser N inputs on treatment TN were 95 and
98 kg ha
− 1
less than for treatment CN for Site 1 and Site 2, respectively, with a wider range at Site 2 than
at Site 1. Estimates of biological fixation for treatment GC reflected a reduction in clover proportion in the
sward over the years at Site 1 and an increase at Site 2. Slurry-N addition was highest for treatment TN at
both sites with overall means being 24, 34 and 37 of total N inputs for CN, TN and GC, respectively, at Site
1, and 18, 37 and 57 for CN, TN and GC, respec- tively, at Site 2. Nitrogen outputs in LWG were small
in comparison to N inputs; values for treatment TN were lowest at both sites with treatment differences
at Site 2 being significant p0.05. Significantly less N was removed in cut herbage on treatment GC
compared with the other treatments at both sites Site 1, p0.01; Site 2, p0.001. Live weight gain kg
− 1
N applied was ranked in order from highest to lowest GCTNCN at Site 1 p0.001 and GCCNTN
at Site 2 p0.001. Nitrogen surpluses N inputs as fertiliserfixation and manure minus N outputs in
LWG and silage were calculated on a per unit area ha
− 1
and per unit product 100 kg
− 1
LWG basis. Mean values for N surplus per unit area were ranked
in order from highest to lowest CNTNGC at both Sites 1 and 2, with the differences being significantly
p0.001 greater on treatment CN than the other treatments at Site 1, and significantly p0.001
lower on treatment GC than the other treatments at Site 2. Nitrogen surpluses ha
− 1
for treatment TN and GC, showed reductions in comparison with CN of
34 and 56, respectively, at Site 1, and 25 and 75, respectively, at Site 2. For N surplus per unit product,
treatments were ranked in order CNTNGC at Site 1 and TNCNGC at Site 2. Mean values for treat-
ment GC were significantly less than the other treat- ments at Site 1 p0.01 and treatment TN at Site 2
p0.05.
3.4. Grazing periods and herbage measurements On average, the grazing period was 10 days longer
at Site 1 than at Site 2. Mean grazing periods over all years were 170, 123 and 168 days for CN, TN and
GC, respectively, at Site 1 with corresponding values
J.A. Laws et al. Agriculture, Ecosystems and Environment 80 2000 243–254 249
Table 4 N inputs as fertiliser CN and TN or biological fixation GC and slurry, outputs in live weight gain and cut herbage, efficiency and
surpluses
a
per unit area and per unit product Site 1
b
S.E.D. Significance
Site 2
b
S.E.D. Significance
CN TN
GC CN
TN GC
N inputs kg ha
− 1
Range 273–284
172–208 34–308
277–288 142–222
34–62 Mean
281 186
121 282
184 48
Slurry N inputs kg ha
− 1
Range 48–131
48–160 44–136
40–103 74–166
37–93 Mean
88 96
71 64
109 63
N removed in LWG kg ha
− 1
Range 15–25
10–23 14–26
11–19 5–11
7–18 Mean
18 15
18 2.08
NS 15 a
8 b 14 a
1.8 N removed in herbage kg ha
− 1
Range 87–145
80–156 52–91
44–121 43–150
28–46 Mean
112 a 113 a
70 b 12.6
99 a 110 a
38 b 16.8
LWG kg ha
− 1
per kg N applied Range
1.3–2.8 1.1–3.3
2.5–6.5 1.0–2.1
0.6–1.4 2.9–7.3
Mean 1.8 a
2.0 b 4.1 c
0.55 1.6 a
1.1 a 4.9 b
0.66 Surplus N kg ha
− 1
Range 192–319
100–226 36–251
205–347 116–305
52–107 Mean
239 a 153 b
104 b 29.4
232 a 175 a
58 b 30.0
Surplus N kg per 100 kg
− 1
LWG Range
21–55 17–64
6–27 31–92
37–177 10–27
Mean 41 a
32 a 15 b
6.5 47 ab
73 a 13 b
20.6
a
N inputs as fertiliserfixation and slurry minus N outputs as live-weight gain LWG and silage.
b
Abbreviations: CN, conventional nitrogen; TN, tactical nitrogen; GC, grasswhite clover. Values with different letters within rows for each site are significantly different NS, not significant; p0.05; p0.01; p0.001.
of 163, 104 and 166 days at Site 2. Generally, graz- ing starting earlier in spring at Site 1 and continued
longer in autumn at Site 2. More land was required to comply with grazing guidelines on treatment GC
than the other treatments and least land was grazed on treatment CN at both sites, with treatment differ-
ences over all years being significant p0.05 at Site 1. Means over all years were 0.57, 0.62 and 0.64 ha
S.E.D.=0.029, p0.05 at Site 1 and 0.53, 0.54 and 0.64 ha S.E.D.=0.059, p0.05 at Site 2, for treat-
ments CN, TN and GC, respectively.
Treatment comparisons for SSH, herbage mass on offer and the grassland area required for grazing were
made for the period from the start of grazing in spring to the end of grazing on treatment TN. Over all years,
SSH was highest on treatment CN at Site 1 and on treatment TN at Site 2 89, 86 and 86 mm, S.E.D.=2.8,
p 0.05 for Site 1 and 91, 96 and 88 mm, S.E.D.=3.4,
p 0.05 for Site 2 for CN, TN and GC, respectively.
Differences between treatment means were not signif- icant. The overall mean value for Site 2 was signif-
icantly greater than the corresponding value for Site 1 87 and 92 mm for Site 1 and Site 2, respectively,
S.E.D.=1.8, p0.05. Similarly, herbage mass on of- fer was highest for CN at Site 1 and for TN at Site
2, and was consistently lower in all years for GC at both sites 3.05, 2.85 and 2.46 t ha
− 1
, S.E.D.=0.136, p0.05 for Site 1, and 3.66, 4.21 and 3.15 t ha
− 1
, S.E.D.=0.288, p0.05 for Site 2, for CN, TN and
GC, respectively. The mean proportion of clover in the above-ground biomass for treatment GC declined
on both grazed and cut swards at Site 1, with values of 39 for grazed swards and 50 for cut swards
in 1992 and corresponding values of 12 and 7 in
250 J.A. Laws et al. Agriculture, Ecosystems and Environment 80 2000 243–254
Table 5 Herbage DM yield t and the total area cut for silage, for each treatment, in each year, with overall means
a
Site 1
b
S.E.D. Significance Site 2
b
S.E.D. Significance CN
TN GC
CN TN
GC 1992
DM yield 4.4 36 4.2 54
3.6 50 1.77
NS –
– –
– –
Area 1.2
2.1 1.2
1993 DM yield 4.3 53
3.9 48 3.2 38
0.68 NS
2.5 26 2.9 27
2.5 24 0.53
NS Area
1.1 1.8
1.0 0.6
1.4 0.7
1994 DM yield 6.6 208
5.0 116 4.8 156
0.66 NS
5.6 a 188 6.4 a 142 2.2 b 67 0.39 Area
1.3 1.9
1.1 1.3
2.2 1.0
1995 DM yield 4.5 145
4.6 127 2.7 86
0.76 NS
5.0 a 196 6.4 a 160 2.1 b 67 0.63 Area
1.4 2.0
1.1 1.4
2.1 0.9
1996 DM yield 4.7 168
5.2 169 2.6 104
0.73 NS
3.9 152 4.2 105
1.8 81 0.78
NS Area
1.4 2.2
1.2 1.4
2.2 0.8
Mean DM yield 4.9 a 122 4.6 a 103 3.4 b 87 0.48
4.2 a 141 5.0 a 108 2.1 b 60 0.64 Area
1.3 2.0
1.1 1.2
2.0 0.8
a
The values in parentheses show the yield of herbage DM as a proportion of the estimated DM required for winter feeding. Assuming a live weight at slaughter of 475 kg, rolled barley fed as a supplement 3 kg per head per day for the first 3 months of the
housed period and 4.5 kg per head per day thereafter and allowing for 25 total losses of conserved herbage during ensiling and feeding.
b
Values with different letters within rows for each site are significantly different NS, not significant; p0.05; p0.01; p0.001. Abbreviations as in Table 4.
1996. Conversely, values for treatment GC at Site 2 increased from 6 for both grazed and cut swards in
1993 to 10 for grazed and 9 for cut swards in 1996. Overall means, adjusted according to the proportion
of the total area under cutting or grazing management for each farmlet, were 20 at Site 1 and 9.5 at
Site 2.
Table 5 shows that, over all years, the greatest area of land was available for cutting on treatment TN than
on the other treatments, and least land was cut on treatment GC. Yields were generally lower on treat-
ment GC than on the other treatments at both sites, with the differences being significant at Site 2 in 1994
p0.01 and 1995 p0.05 and over all years at both sites Site 1, p0.01; Site 2, p0.001.
3.5. Self-sufficiency for winter fodder Estimated levels of silage DM required during the
housed period were calculated for each treatment, assuming an animal live weight at slaughter of 475 kg,
a mean total daily intake of 8.0 kg DM per head per day Wilkinson, 1984 during the housed period, a diet
of silage fed ad libitum with a rolled-barley supple- ment of 3 kg per head per day for the first 3 months,
rising to 4.5 kg per head per day for the remainder of the housed period Wilkinson and Tayler, 1973 and
allowing for 25 total losses of herbage DM during ensiling and feeding. Table 5 shows that the high
stocking rates in 1992–1993 effectively reduced the amount of silage made so that a deficit was realised
for all the treatments at both sites in these years. In 1994–1996 when the stocking rate was 4 steers ha
− 1
, surplus DM was harvested on treatments CN and TN
at both sites. For treatment GC, a deficit occurred in 1995 at Site 1 and in all years at Site 2. Average values
for the 1994–1996 period were 174, 137 and 115 for CN, TN and GC, respectively, for Site 1, and 179,
136 and 72 for CN, TN and GC, respectively, for Site 2.
J.A. Laws et al. Agriculture, Ecosystems and Environment 80 2000 243–254 251
Table 6 Daily live weight gain LWG per head kg up to the removal of cattle on TN and overall LWG grazed ha
− 1
kg for each treatment, in each year, with overall means
Site 1
a
S.E.D. Significance
Site 2
a
S.E.D. Significance
CN TN
GC CN
TN GC
1992 LWG head
0.97 0.91
0.96 0.071
NS –
– –
– LWG ha
1139 1209
1179 133.4
NS –
– –
– 1993
LWG head 0.56 a
0.56 a 0.73 b
0.046 0.30
0.25 0.32
0.111 NS
LWG ha 823
828 621
265.8 NS
688 324
592 137.5
NS 1994
LWG head 0.95
0.79 0.98
0.074 NS
1.03 0.73
0.89 0.1222
NS LWG ha
868 593
872 127.0
NS 1084
669 947
115.9 NS
1995 LWG head
0.88 0.97
0.90 0.127
NS 0.94
0.93 0.93
0.133 NS
LWG ha 908
714 738
89.7 NS
1018 a 719 b
801 b 51.3
1996 LWG head
1.11 1.22
1.22 0.043
NS 1.08 a
0.85 b 1.19 a
0.071 LWG ha
951 990
934 83.3
NS 929 a
643 b 946 a
66.6 Mean
LWG head 0.90
0.89 0.96
0.092 NS
0.84 0.69
0.83 0.166
NS LWG ha
938 867
869 98.2
NS 930 a
588 b 821 a
88.6
a
Values with different letters within rows for each site are significantly different NS, not significant; p0.05; p0.01. Abbreviations as in Table 4.
3.6. Animal production Daily LWG per head from turnout until the removal
of the cattle from treatment TN was calculated for all treatments. LWG grazed ha
− 1
, being the product of daily LWG per head, stocking rate of the grazed area,
and the number of days spent grazing, was calculated for the overall grazing period for each treatment.
Table 6 shows that, at Site 1, LWG per head was sim- ilar for all treatments, except in 1993 when the value
for treatment GC was significantly p0.05 greater than the other treatments. At Site 2, values were
generally lower on treatment TN than the other treat- ments with the differences being significant p0.05
in 1996. Values for CN and GC were similar in all years. LWG grazed ha
− 1
, for Site 1, was highest at the higher stocking rate in 1992. On average, values
were highest on treatment CN but treatment differ- ences were small and non-significant p0.05. For
Site 2, LWG grazed ha
− 1
was significantly p0.05 greater for treatment CN than the other treatments in
1995 and significantly p0.05 lower for treatment TN in 1996 and over all years.
4. Discussion
