Professional Documents
Culture Documents
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
TEAGASC-Agriculture and Food Development Authority is collaborating with JSTOR to digitize, preserve and
extend access to Irish Journal of Agricultural and Food Research.
http://www.jstor.org
The objective of this study was to compare the accuracy of four non-destructive
(PPCM) were evaluated. Estimates were obtained on five occasions (July, September
and November, 1997 and April and May, 1998). Mean dry matter (DM) yields
(kg/ha) of available herbage cut to 40 mm above ground level (AH) and total
herbage cut to ground level (TH) were 2161 (s.d. 850.2) and 2762 (s.d. 890.9) kg,
respectively. Relationships were obtained for VE, RPM, SS and PPCM with AH, and
for RPM, SS and PPCM with TH. For AH, the proportions of variance (R2)
accounted for by the model were 0.95, 0.94, 0.92 and 0.72 for VE, RPM, SS and
PPCM, respectively. The corresponding residual s.d. values were 193, 222, 249 and
458 kg DM/ha, and the coefficients of variation (CVs) were 9%, 10%, 12% and 21%,
respectively. For TH, the R2 values were 0.88, 0.87 and 0.76 for RPM, SS and PPCM,
respectively. The corresponding residual s.d. values were 318, 331, and 442 kg
DM/ha, and the CVs were 12%, 12%, and 16%, respectively. It is concluded that VE
was the most accurate method of HM prediction and that the PPCM method was an
inaccurate predictor of HM.
Keywords: Herbage mass; pasture probe; rising plate meter; sward stick; visual assessment.
17
(Lolium perenne L.) with some Poa gle probe type Pasture Wand (Ag
pratensis and Agrostis stolonifera. Nitro Products, Ohaupo, New Zealand). In the
gen fertiliser was applied at the rate of PPCM the capacitor is part of a circuit
375 kg N/ha over the grazing season. which generates a
signal of a certain fre
(Hutchings, 1991) supplied by the Scot cuts. The identified quadrats covered a
tish Agricultural College (Bush Estate, range inHM from 530 to 4454 kg DM/ha
Penicuik, Scotland). With this method, a over the five estimation periods. A metal
small plastic window is lowered until it quadrat (2 m X 2 m) was placed at
hits a part of any plant. The height of the random within the paddock. The herbage
first contact is then recorded. HM predic outside the quadrat was cut with a re
ciprocating motor scythe (Agria-Werke, (Y) and the various estimates of yield (X)
Moeckmuehl, Germany) and removed. using a model with effects for season and
This continued until all quadrat sites were the
regression of Y on X for each season.
identified. In order to minimise distur The differences, among seasons, in the
bance of the sward canopy, the measure coefficients were tested. Sea
regression
ments were taken at each site in son as referred to in the statistical
quadrat analy
the following order: VE, PPCM, SS and sis is a synonym for measurement period.
RPM. Observers did not consult with each The ratio of actual HM to estimated
other while estimating HM. For each HM was analysed according to a split-plot
quadrat, five PPCM, SS and RPM read design (SAS, 1991). The model had ef
were taken across each fects for season, within
ings diagonal. quadrat season,
(The duration of each observation period observer and season-by-observer interac
was 1 day). The herbage within the tion. The effects associated with quadrat
was harvested with the within season were considered random
quadrat recipro
cating motor scythe to a stubble height of and the associated means square was used
40 mm. It was then weighed and sub-sam to test differences due to season. The
pled (300 g per sample) in duplicate. DM other effects were tested against the
concentration was measured error =
by drying residual (d.f. 820).
Table 1. Mean, range and standard deviation (s.d.) of available herbage mass (AH) above 40-mm and
herbage mass measured from ground level (TH) (kg dry matter/ha) for the different harvest seasons
Harvest season
AH
3298 Mean 1968 2076 1905 1580
Range 1867-4454 530-3140 831-3465 851-3222 651-2368
666.2s.d. 638.3 735.9 538.5
515.6
TH
3847 Mean 2849 2746 2479 1891
Range 2375-5181 1238-4348 1263-3897 1540-3833 799-2675
766.5s.d. 671.7 654.1 553.5
543.0
herbage mass was recorded in April and assist in making decisions on grazing
the lowest was recorded in November. management (Stakelum 1996; O'Dono
The mean HM yields in the 0 to 40 mm van, 2000). This study compared four
horizon were 500 (s.e. 34.6), 880 (s.e. 35.6), methods of indirect estimation of HM
670 (s.e. 30.9), 575 (s.e. 34.6) and 311 (s.e. against DM yield measured by cutting.
34.6) kg DM/ha for the seasons, April to
November, respectively.
Table 2 shows the parameters of the Visual assessment
pended on season. A common slope with (1981) reported CVs up to 38%. How
different intercepts could be used for SS. ever, the regression equations in Table 2
Table 3 shows the parameters of the for VE, while giving information on the
regression equations calculated for TH precision and accuracy of the method,
and the significance of the terms in the cannot be used in practice for predictive
model. The CVs were 12, 12 and 16% for purposes. The ratio of actual AH to visu
RPM, SS and PPCM, respectively. The ally estimated AH (in conjunction with its
analysis indicates that a pooled equation s.e.) rather than the difference between
for all seasons would not be appropriate actual and estimated HM was chosen to
for RPM, SS or PPCM because there show both bias and error together. It was
were differences between seasons in considered a more measure
appropriate
slope.
of bias because a constant difference (100
There were significant (P < 0.001) ef kg/ha for example) can be a large bias at
fects of season, observer, season by low HM but a much smaller bias at high
observer, actual HM and actual HM X HM.
season on the ratio of actual HM to the The significant interactions between
VE estimate. Table 4 shows the ratios for observer and season and between level of
the effect of season by observer. The AH and season raise questions about the
largest under-estimate was in November, usefulness of the VE method. In general,
while the smallest was in September. Fig there was considerable under-estimation
ure 1 shows the relationships for the ra of AH, especially inNovember, April and
tio and actual HM for the five harvest July. However, in those harvest seasons,
seasons. The bias in VE for April (-0.17) all observers under-estimated AH by
and July (-0.13) was stable across HM roughly similar amounts. In May and
levels. The bias for May (-0.05) and September when the ratio was 1.06 and
November (-0.25) declined as HM in 1.00, respectively, the observers, while
creased. The bias (+ 0.01) increased under- and over-estimating AH, were
****193.3
0.95
Q * ***
249.0
0.92
?
? 0.94
221.6
***
**
**
#**
***
07?
S
4.S7
O
of
regression
regression
Heterogeneity
Pooled
b
Season
Xa
2Significance
Season
(?
Regression
R"
factors
of
Residual
equation
s.e.)
s.d.
estimation
method
effects
and
(a)
the
(b)
with
for
using
Y
within-season
and
model
of
X
season
regression
on
a
The
2.
Table
linear
between
40-mm)
(above
available
relationships
(Y
^herbage
(X)
in
kg
dry
matter/ha)
and
herbage
available
of
estimates
for
each
mass
Visual
estimation
2 j*j
Sward
stick
Plate
meter
^ probe
Pasture
- 55 -
372
May
0.817
?0.0698
?118.3
C May
5?216.0
158.0
?19.35
?
?July
JulyApril 409
164.4 > ?111.4
1.016 266
0.965
0.0680
?0.0748?? April
?-1
237.6
250.6
?23.36
July
>May
-?61
195.3
192.8
?22.00
154
174.5
? 174.3
-226 ?20.39
2 316
April
256.5
181.0
>+
?19.95 +July
May
+ g106
-1864
664.0
? 0.1503
1067
0.563
562.3
April + 607.2
+151.0+18.10
0.9830.527
?0.1651
0.1761 ^ X
September2 0.885
162 ?81.3
0.0564
398
November
??0.0796
139.1
? 0.900 q
September
November November September
-461
November
256.1
+151.4
-183
?20.29
168.1
?<258.1
175.7
?16.28
225.8 r ? ? 25.25
18.76 140
^-C
147.3 September
-2213
+
777.6
0.896
0.1855
239.1 209.8
+ 448.1
+ ? ?
-569-388
-C
? 318.3
0.88
*** ******
*441.5
0.76
^**
***
^n
&7
n7
J of
regression
Season
Heterogeneity
bOa
Pooled
(?
Regression
Season
Significance
Residual
factors
of
equation
qs.e.)
R
s.d.
3.
Table
The
linear
kg
in with
the
effects
(a)
model
for
within-season
using
and
Y
of
regression
X
on
season
a?.
different
(Y
between
herbage
total
relationships
three
dry
(X)
matter/ha)
and
total
herbage
of
mass
methods
each
for
estimation
^ Sward
stick Pasture
probe
m
^
Plate
meter
- ^-
May
891
+
276.9
?31.60
188.4
??
114
April
283.5
341.2
33.54
X+ gJuly
1319
?250.6
?29.29
129.5 May
July
+
316
380.7
26.49
206.2156.5
+
25.70
g?365.8
626
-325+436.2
April
+W 111.5
?24.04
> May
July
-1005
1250
April
0.656
?542.70.989
?0.1450
& 373
?585.9
?0.1700
g0.635
?640.8 +0.1593
>
3September
61
343.3
36.26
220.7
? November ? November
-96
?432.4
0.560
0.1203
+
-123
211.6
+264.2
+
26.95
g September
-378
?437.8
147.3
?26.94
S?
November
-618
?302.9
+181.9
21.63 September
-1540
750.4
+0.874
?0.1790
g
-^
This content downloaded from 129.68.65.223 on Fri, 16 Aug 2013 08:47:40 AM
All use subject to JSTOR Terms and Conditions
24 IRISH JOURNAL OF AGRICULTURAL AND FOOD RESEARCH, VOL. 41, NO. 1, 2002
20 r
Y = 1.21-IX 10'*X
April
R2 0.02
1.5
0.5 I->-'-'
500 1500 2500 3500 4600
2.0 r
May y ?u?-i x io^x
R2 0.28
1.5
0.5 '- - -
1500
500 2500 3500 4500
2.0 r
July Y= 1.20-3 X lO"^
R2 0.04
15
0.5 I-'-'-1-<
500 1500 2500 3600 4500
2.0 r
September y=oj?+5x io-*x
R2 0.05
1.5
0.5 I-1-1-i
500 1500 2600 3500 4500
2.0 r
November y = 1.61-2 x io^x
f 4 R2 0J4
Figure 1: The linear relationship of the ratio of available (above 40-mm) herbage yield to
visually estimated herbage yield (Y) and the herbage mass (X; kg dry matter / ha) for each
of five harvest seasons.
Table 4. The mean ratios of available (above 40-mm horizon) herbage yield to visually estimated yield for
season of harvest by individual observer
garding the interaction between season herbage. Differences in cutting height and
and the level of AH in April, July and method of continuous calibration of the
September, the degree of under-estima observers together with the cutting
tion did not change with increasing HM method used may explain the under
while it decreased for May and especially estimation of HM by the VE method in
for November. The November data (Fig this study compared to others.
ure 1) show a major under-esti
clearly
mate at low HM values. Stockdale (1984) The rising plate meter
found that observers over-estimated low The RPM has been widely investigated as
herbage yields and under-estimated high a predictor of HM (Castle, 1976; Earle
herbage yields. and McGowan, 1979; Mitchell, 1982;
O'Donovan (2000) has described the Stockdale, 1984; Stockdale and Kelly,
of distribution of HM across
pattern 1984; Douglas and Crawford, 1994; Karl
and seasons under a and Nicholson, De
grazing paddocks 1987; O'Riordan,
of intensive summer milk and Stockdale
system produc vaney French, 1997). (1984)
tion. The optimum grass supply from
reported mean CVs of 14% and 28%
April to September is ca. 1000 kg/ha. A with 0.71 and 0.74 of the variation in
bias of +0.05 in the estimation of grass herbage yield accounted for, in pre- and
would be and over- or
supply acceptable post-grazing perennial ryegrass swards,
under-estimation of HM by this fraction respectively. Murphy et al. (1995) ob
would not restrict cow from tained correlation coefficients
performance relating
grazed pasture (O'Donovan, 2000). Other HM to RPM of 0.72 and 0.05 for pre- and
information collected routinely on dairy post-grazing HM measurements, respec
farms, such as daily milk yield, grazing tively. Earle and McGowan (1979) ob
severity and rotation length, will help to tained a CV of 18% when data were
improve the accuracy of farm grass sup pooled from a large number of calibra
ply estimation. tions, while Michell (1982) obtained a
The five observers in this study were correlation coefficient for the calibration
calibrated using the small quadrat (0.5 m relationship of 0.80 with a residual s.d. of
X 0.5 m) method (O'Donovan, 2000). In 500 kg DM/ha. The results of the pre
this method, the herbage within the sent study show that separate seasonal
nally grazed swards. The correlation co reported that the PPCM explained only a
efficients to the SS were small amount of variation (0.44) and had
relating HM
0.70 and 0.31 and re a high CV (31%). The present results are
pre- post-grazing,
consistent with those of the latter authors
spectively (Murphy et al, 1995). L'Huil
lier and Thomson (1988) reported a mean (Bryant et al, 1971; Virkajarvi, 1999). It
correlation coefficient of 0.81 with a is not clear why the PPCM performed so
residual s.d. of 415 kg DM/ha and a CV poorly in this experiment.
of 16%. The slopes of the regression
equations for the different seasons for Acknowledgemen ts
AH were similar but the intercepts were The authors wish to acknowledge the technical as
different. The similar can be con sistance of N, Byrne, M. Reidy, J. O'Dywer, M.
slopes
Kearney and P. Power. The assistance of the
sidered an of the method, as
advantage observers used in this study is also acknowledged.
each 10mm change in sward height would The financialassistance provided by the Irish Dairy
be equivalent to 150 to 180 kg DM/ha. Farmers through the Dairy Levy Fund and by Irish
Both the slopes and intercepts of the Fertilizer Industries Ltd. (IFI) and Farm Business
for the different Development (FBD) is gratefully acknowledged.
regression equations
This research was part-funded by European
seasons for TH were different. This re
Union Structural Funds (EAGGF).
duces the practical usefulness of SS. The
SS differs from RPM in that the SS read
ing is adjusted by 40 mm for AH com References
to TH, whereas the same is Baars, J.A. and Dyson, C.B. 1981. Visual estimates
pared reading
of herbage on hill country sheep pastures. New
used with the RPM for both AH and TH.
Zealand Journal of Experimental Agriculture 9:
157-160.
meter Barthram, G.T. 1986. Experimental techniques: the
Capacitance
HFRO sward stick. The Hill Farming Research
A number of studies have been carried
Organisation, Biennial Report 1984-85, pages
out examining the performance of the 29-30.
PPCM compared to both RPM and SS. Bryant, A.M., Parkes, O.F., Cook, M.A.S. and Tay
Stockdale and Kelly (1984) obtained a lor, M.J. 1971. An evaluation of the perfor
mance of the capacitance meter for estimating Murphy, W.M., Silman, J.P. and Mena Barreto,
the yield of dairy pastures. Proceedings of the A.D. 1995. A comparison of quadrat, capaci
New Zealand Grassland Association 33: 83-90. tance meter, HFRO sward stick, and rising plate
Campbell, N.A. and Arnold, G.W. 1973. The visual meter for estimating herbage mass in a smooth
assessment of pasture yield. Australian Journal stalked, meadowgrass-dominant white clover
of Experimental Agriculture and Animal Hus sward. Grass and Forage Science 50: 452-455.
bandry 13: 263-267. O'Donovan, M. 1997. Improved techniques for graz
Castle, M.E. 1976. A simple disc instrument for ing management. MAgrSc Thesis, National Uni
estimating herbage yield. Journal of the British versity of Ireland, 192 pages.
Grassland Society 31: 37-40. O'Donovan, M., Stakelum, G., Dillon, P., Connolly,
Clark, D.A. and Jans, F. 1995. High forage use in J. and Rath, M. 1997. The estimation of pasture
sustainable dairy systems. In: "Recent Develop mass for grazing management. Irish Journal of
ments in the Nutrition of Herbivores' (ed. M. Agricultural and Food Research 36: 130 (Ab
Journet, E. Grenat, M.H. Farce, M. Theriez and stract).
C. Demarquilly), INRA Edition, Paris, pages O'Donovan, M. 2000. The relationship between the
497-526. performance of dairy cows and grassland man
Dillon, P., Crosse, S., Stakelum, G. and Flynn, F. agement practice on intensive dairy farms in
1995. The effect of calving date and stocking Ireland. PhD Thesis, National University of Ire
rate on the performance of spring calving dairy land, 366 pages.
cows. Grass and Forage Science 50: 286-299. O'Riordan, E.G., Devaney, S. and French, P. 1997.
Douglas, J.T. and Crawford, C.E. 1994. An evalua Sward height as a measure of pasture herbage
tion of the drop-disc technique for measure supply. Irish Journal of Agricultural and Food
ments of herbage production in ryegrass for Research 36: 107 (Abstract).
silage. Grass and Forage Science 49: 252-255. O'Sullivan, M., O'Keeffe, W.F. and Flynn, MJ. 1987.
Earle D.F. and McGowan A.A. 1979. Evaluation The value of pasture height in the measurement
and calibration of an automated rising plate of dry matter yield. Irish Journal of Agricultural
meter for estimating DM yield of pasture. Aus Research 26: 63-68.
tralian Journalof Experimental Agriculture and Stakelum, G. 1996. Practical grazing management
Animal Husbandry 19: 337-343. for dairy cows. Proceedings of the Irish Grassland
Frame, J. 1993. Herbage mass. In: 'Sward Measure and Animal Production Association Journal 30:
ment Handbook' (ed. A. Davies, R.D. Baker, 33-45.
S.A. Grant and S. Laidlaw), 2nd edition, British Stockdale, C.R. 1984. Evaluation of techniques for
Grassland Society, UK, pages 39-69. estimating the yield of irrigated pastures inten
Hutchings, NJ. 1991. Spatial
heterogeneity and sively grazed by dairy cows. (1) Visual assess
other sources of variation
in sward height as ment. Australian Journal of Experimental Agricul
measured by sonic and the HFRO sward sticks. ture 24: 300-304.
Grass and Forage Science 46: 277-282. Stockdale, C.R. and Kelly, K.B. 1984. A comparison
Karl, M.G. and Nicholson R.A. 1987. Evaluation of of a rising plate meter and an electronic capaci
the forage disk method in mixed grass range tance meter for estimating the yield of pastures
lands of Kansas. Journal of Range Management grazed by dairy cows. Grass and Forage Science
40: 467-471. 39: 391-394.
L'Huillier, P.J. and Thomson, N.A. 1988. Estima SAS. 1991. SAS/STAT User's Guide: Statistics.
tion of herbage mass in ryegrass/white clover Statistical Analysis Systems, SAS Institute Inc.,
dairy pastures. Proceedings of the New Zealand Cary, North Carolina, USA.
Grassland Association 49: 117-122. Vickery, P.J., Bennett, LL. and Nicol, G.R. 1980.
Michell, P. 1982. Value of a rising plate meter for An improved electronic capacitance meter for
estimating herbage mass of grazed perennial estimating herbage mass. Grass and Forage Sci
ryegrass-white clover swards. Grass and Forage ence 35: 247-252.
Science 37: 81-87. Virkajarvi, P.
1999. Comparison of three indirect
Michell, P. and Large, R.V. 1983. The estimation of methods for prediction of herbage mass on
herbage mass of perennial ryegrass swards: a timothy-meadow fescue pastures. Acta Agricul
comparative evaluation of a rising plate meter turae Scandinavica, Sect B, Soil and Plant Science
and a single probe capacitance meter calibrated 49: 75-81.
at and above ground level. Grass and Forage
Science 38: 295-299. Received 16 January 2001