Directory UMM :Data Elmu:jurnal:A:Animal Feed Science and Technology:Vol86.Issue1-2.Jul2000:
Animal Feed Science and Technology
86 (2000) 53±69
Characterisation of feedstuffs for ruminants
using some physical parameters
Sylvie Giger-Reverdin*
Laboratoire de Nutrition et Alimentation (INRA), Institut National Agronomique Paris±Grignon,
16 rue Claude Bernard, 75231 Paris Cedex 05, France
Received 3 August 1999; received in revised form 14 April 2000; accepted 4 May 2000
Abstract
Various ingredients and roughages of known chemical composition have been analysed by
several physical methods: particle density, particle size, water holding capacity, feed solubilisation
and osmotic pressure. Some of these methods already exist and have been adapted and others have
been newly developed. All of the methods are quite easy to perform and the results are very
repeatable. The data obtained by the physical and chemical methods have been correlated together.
The physical methods gave new information about the nutritive value of feedstuffs for ruminants
and they can be used to differentiate between feedstuffs. They might explain part of the role played
by rumen ¯ora on feedstuffs which is not taken into account by the chemical approach.
# 2000 Elsevier Science B.V. All rights reserved.
Keywords: Feedstuffs; Ruminants; Physical properties; Particle size; Osmotic pressure; Water holding capacity
1. Introduction
The physical characteristics of feedstuffs for ruminants are rarely measured,
particularly in relation to their nutritional properties that could be taken into account
in feed formulation. It is known that some of these characteristics might partly explain the
interaction between rumen ¯ora and feedstuffs degradation. Particle density in¯uences
their rate of passage from the rumen (Ehle, 1984; Martz and Belyea, 1986) and thus
ruminal turnover rate of feeds and possibly their level of intake (Singh and Narang, 1991).
Particle size of feedstuffs in¯uences the surface area available for micro-organisms attack
(Owens and Goetsch, 1988) and thus their multiplication (Dehority and Orpin, 1988). It
*
Tel.: 33-1-44081768; fax: 33-1-44081853.
E-mail address: [email protected] (S. Giger-Reverdin)
0377-8401/00/$ ± see front matter # 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 7 - 8 4 0 1 ( 0 0 ) 0 0 1 5 9 - 0
54
S. Giger-Reverdin / Animal Feed Science and Technology 86 (2000) 53±69
also plays a role in the rate of passage of feedstuffs through the digestive tract (Balch,
1950). Water holding capacity (WHC) has an impact on microbial colonisation and
osmotic pressure is a factor that should be considered in the overall ecology of the rumen
(Bennink et al., 1978). Most methods concerning physical characteristics had been
applied either to roughages or to rations (Montgomery and Baumgardt, 1965; Allen et al.,
1984; Hooper and Welch, 1985; Singh and Narang, 1991), or to foods for human nutrition
(McConnell et al., 1974; Robertson and Eastwood, 1981).
The purpose of this paper is to describe easy-to-perform methods and to test them on a
set of feedstuffs representative of the range used in ruminant diets thereby allowing
discussion of their potential usefulness compared to more classical chemical methods.
2. Materials and methods
2.1. Feedstuffs
Twenty-four feedstuffs were tested. They can be grouped into six main classes:
cereals (oats, wheat, barley, maize, sorghum),
cereal by-products (wheat bran, brewers grains, corn gluten feed, corn gluten meal),
legumes (faba beans, peas, lupine seed),
oilmeals (soyabean meal, coconut meal, rapeseed meal, sunflower meal, palm kernel
meal),
agro-industrial by-products (soyabean hulls, citrus pulp, sugar beet pulp),
roughages (corn silage, corn stover, alfalfa hay, dehydrated alfalfa).
These ingredients were chosen to represent the wide range of dietary sources used in
rations or in compound feedstuffs for ruminants. Moreover, a part of the choice was
performed so that the ingredients differed as much as possible from a chemical point of
view. All the samples were ground through a 1 mm screen. Wet samples (corn silage,
citrus pulp) were oven-dried at 608C for 48 h.
2.2. Physical methods
2.2.1. Bulk density
Bulk density was determined by a modi®cation of the method of Montgomery and
Baumgardt (1965), where manual swirling of the container between the palms was
replaced by an automatic method, and the fourth step of feedstuff addition was not
performed. The following procedure was used to estimate the bulk density of the dried
ground samples:
A 100 ml glass graduated cylinder (2.7 cm internal diameter) was filled with sample to
the 50 ml mark and swirled for 15 s. Weight of the sample and volume occupied were
recorded.
Additional sample was added up to the 100 ml mark and the cylinder was swirled for
10 s.
S. Giger-Reverdin / Animal Feed Science and Technology 86 (2000) 53±69
55
The cylinder was refilled to 100 ml mark and swirled for 5 s. Total weight of sample
and final volume were recorded.
As density is de®ned as the mass of the substrate over the mass of an equivalent volume
of water, bulk density was equal to the weight of sample (expressed in mg) over the
volume occupied (expressed in ml).
2.2.2. Median particle size
Median particle size of feedstuffs was measured by a dry-sieving method according to
Melcion and de Monredon (1991). About 100 g of each feedstuff was weighed into the
upper sieve and shaken from side to side for 15 min. For wheat bran, alfalfa hay, corn
silage and corn stover, the largest particles clogged the sieve meshes. Therefore, 50 g of
these feedstuffs was used. The sizes of the six sieves were 0.400, 0.315, 0.250, 0.200,
0.125 and 0.080 mm. Refusals from each sieve were weighed, and the percentage of each
class was calculated. The cumulated percentage was calculated for each sieve from the
largest sieve to the smallest. Results were plotted against the logarithm of sieve size
(AFNOR, 1985). Median particle size (D50) was read directly as corresponding to a
virtual screen which would retain 50% of the particles, and D16 and D84 were,
respectively, those corresponding to 16 and 84% probability (Allen et al., 1984). In order
to estimate, the distribution of the particles around the median particle size, a relative
deviation was estimated as the difference between D16 and D84 divided by the median
particle size and called RD50.
2.2.3. Water holding capacity
WHC of feedstuffs was measured using an adaptation of one of the methods proposed
by Robertson and Eastwood (1981). The whole sample was used instead of its ®brous
part. A mass of 2.5 g of sample was left to soak for 16±24 h in 250 ml of distilled water,
so that water was in excess. The sample was then ®ltered on a fritted glass crucible
(porosity 2) and the walls of the beaker were carefully rinsed. The wet sample was
weighed after letting water decant for 10 min. WHC was the quantity of water retained by
the sample and expressed as l kgÿ1 sample dry matter.
2.2.4. Feed solubilisation
The ®ltrate sample collected after passage through the ®lter was oven-dried for 72 h at
1038C, weighed, and subsequently ashed at 5508C overnight and weighed. Dry matter
and ash solubilised were expressed as g lÿ1 or as percentage of initial weight of the
component considered.
2.2.5. Intrinsic osmotic pressure
2.5 g of sample was soaked in 50 ml of water for 24 h. Filtration was performed
according to the WHC method already described, but without the addition of water.
Osmotic pressure was measured on the ®ltrate using the freezing point depression
technique with a Mark 3 Osmometer, manufactured by Fiske (USA).
56
S. Giger-Reverdin / Animal Feed Science and Technology 86 (2000) 53±69
2.3. Chemical methods
The feedstuffs were analysed by standard methods: dry matter was estimated from
water content (AFNOR, 1982), ash (AFNOR, 1977), crude protein (ISO, 1997),
enzymatic starch (AFNOR, 1997a) and fat (AFNOR, 1997b). Cell wall content was
estimated by the neutral detergent ®bre (NDF) method of Van Soest and Wine (1967) as
modi®ed by AFNOR (1997c). Lignocellulose (ADF) and lignin (ADL) were obtained
using a sequential approach (AFNOR, 1997c) on the NDF residue as proposed by Giger
et al. (1987).
2.4. Statistical analysis
The SAS package (SAS, 1987) was used for statistical calculations.
When analyses were in duplicate for a sample, a variance analysis including a feed
effect was performed and the so-obtained residual standard deviation (RSD) was divided
by the mean value in order to estimate the precision of the method involved. This unitless
value which could be called coef®cient of variation would allow to estimate the
respective abilities of the methods to discriminate the feedstuffs.
Regressions were performed to study the relationships between physical and chemical
parameters. The link between the dependent variable and the one or more independent
variables was analysed using the RSD of the dependent variable.
A principal component analysis (PCA) was used to examine the relationships among
several quantitative variables, as described by Lebart and Fenelon (1971).
3. Results
3.1. General considerations about chemical and physical compositions
Chemical parameters for each feedstuff are summarised in Table 1. For some
feedstuffs, either starch or fat were not determined, because those feeds were assumed to
contain only traces of these elements. The statistical parameters were determined without
them (Table 1), but in the PCA, they were assume to be equal to 0 g kgÿ1 DM. As
expected, feedstuffs showed a wide variability in chemical composition.
All physical determinations have been performed in duplicate on all feedstuffs, what
allowed calculating a coef®cient of variation as de®ned before.
3.2. Bulk density
In fact, two types of bulk density were measured. The ®rst one was obtained after the
®rst swirling and was called density50, and the second one after the third swirling, called
density100. They were directly proportional and their mean values (0.561 vs 0.541) did
not statistically differ
Density100 0:966 density50
r 2 0:995; n 24; RSD 0:0110
57
S. Giger-Reverdin / Animal Feed Science and Technology 86 (2000) 53±69
Table 1
Chemical parameters of feedstuffs (g kgÿ1 DM)
Crude
protein
Fat
Starch
NDF
ADF
ADL
Ash
Oats
Wheat
Barley
Maize
Sorghum
Wheat bran
Brewers grains
Corn gluten feed
Corn gluten meal
Faba beans
Peas
Lupine seed
Soyabean meal
Coconut meal
Rapeseed meal
Sun¯ower meal
Palm kernel meal
Soyabean hulls
Citrus pulp
Sugar beet pulp
Corn silage
Corn stover
Alfalfa hay
Dehydrated alfalfa
92
103
141
104
114
165
300
188
650
294
233
382
472
146
415
381
164
124
81
85
73
44
172
157
47
17
20
40
36
30
57
31
24
15
13
110
19
29
11
20
97
22
25
ND
ND
ND
ND
ND
432
727
571
721
707
149
48
234
239
426
495
9
NDa
ND
ND
ND
ND
ND
ND
ND
250
ND
ND
ND
336
122
210
108
103
450
505
407
5
120
110
206
102
578
230
430
612
622
215
427
479
776
553
525
162
35
75
31
65
135
201
109
4
90
85
145
59
429
160
286
381
446
140
222
237
437
365
370
40
13
11
8
25
37
53
10
1
12
6
7
6
43
50
88
103
19
15
16
35
36
85
90
34
19
26
16
19
71
36
54
14
40
35
39
76
57
79
77
46
51
78
93
57
75
108
111
Statistical parameters
Mean
No. of feedstuffs
Standard deviation
212
24
152
35
19
27
385
13
251
343
24
214
195
24
142
34
24
30
55
24
28
a
ND: not determined.
Moreover, the coef®cient of variation of density100 (0.99%) was less than half that of
density50 (2.05%). This means that the density100 was a more repeatable method than
density50. This method with three steps of feedstuff addition and three swirlings was thus
used, and called simply bulk density. Bulk density varied greatly between feedstuffs
(Table 2). Corn stover had the lowest density (0.138 kg lÿ1) and peas the highest
(0.788 kg lÿ1).
Bulk density was negatively correlated with the cell wall content parameters of
feedstuffs The best relationship was with NDF content (Fig. 1):
Bulk density 0:743 ÿ 0:589 NDF kg kgÿ1 DM
r 2 0:644; n 24; RSD 0:0955
Therefore, for the same bulk density of around 0.5, NDF content varied from 206 g kgÿ1
DM (lupine seed) to 612 g kgÿ1 DM palm kernel meal, ADF explained 44.9% of the
variation of bulk density, and ADL, only 19.3%.
58
S. Giger-Reverdin / Animal Feed Science and Technology 86 (2000) 53±69
Table 2
Physical parameters of feedstuffs (bulk density, median particle size and WHC)a
D50c
Cereals
Oats
Wheat
Barley
Maize
Sorghum
0.469k,l
0.665e
0.566i
0.651f
0.686d
0.460d,e
0.315h,i,j,k
0.380f,g
0.530c
0.400f
2.860b,c
3.017b,c
2.048b,c,d,e
1.387d,e
1.788c,d,e,j
2.18i,j
1.98j
2.28i,j
1.62k
1.50k
Cereal by-products
Wheat bran
Brewers grains
Corn gluten feed
Corn gluten meal
0.340n
0.465k,l
0.604h
0.672e
0.740b
0.235l
0.275j,k,l
0.255k,l
3.162b
1.489d,e
1.127e
0.978E
3.07e,f
4.07c,d
2.53h,i
2.01j
Legumes
Faba beans
Peas
Lupine seed
0.701c
0.788a
0.460l
0.325h,i,j
0.343g,h
0.335g,h,i
2.190b,c,d,e
4.080a
1.108e
2.46h,i
2.44h,i
3.79d
Oilmeals
Soyabean meal
Coconut meal
Rapeseed meal
Sun¯ower meal
Palm kernel meal
0.701c
0.476k
0.651f
0.473k,l
0.519j
0.535c
0.480d
0.305h,i,j,k
0.420e,f
0.450d,e
2.972b,c
2.154b,c,d,e
1.739c,d,e
2.107b,c,d,e
1.241d,e
3.22e,f
4.08c,d
3.35e
3.24e,f
2.70g,h
Agro-industrial by-products
Soyabean hulls
Citrus pulp
Sugar beet pulp
0.448m
0.715b
0.625g
0.480d
0.270j,k,l
0.390f
1.750c,d,e
1.853c,d,e
2.455b,c,d
5.23b
4.33c
5.37b
Roughages
Corn silage
Corn stover
Alfalfa hay
Dehydrated alfalfa
0.318o
0.138p
0.333n
0.530j
0.280i,j,k,l
0.875a
0.260k,l
0.280i,j,k,l
1.873c,d,e
4.127a
1.808c,d,e
2.004b,c,d,e
4.09c,d
8.87a
3.80d
2.87f,g
Statistical parameters
Mean (24 feedstuffs)
Standard deviation
Coefficient of variation (%)
0.541
0.157
0.99
0.401
0.155
4.63
a
RD50d
WHC (l kgÿ1
DM of sample)
Bulk densityb
2.139
0.853
15.90
3.382
1.567
3.97
Values within columns with different superscript letters indicate a signi®cant feed effect at p
86 (2000) 53±69
Characterisation of feedstuffs for ruminants
using some physical parameters
Sylvie Giger-Reverdin*
Laboratoire de Nutrition et Alimentation (INRA), Institut National Agronomique Paris±Grignon,
16 rue Claude Bernard, 75231 Paris Cedex 05, France
Received 3 August 1999; received in revised form 14 April 2000; accepted 4 May 2000
Abstract
Various ingredients and roughages of known chemical composition have been analysed by
several physical methods: particle density, particle size, water holding capacity, feed solubilisation
and osmotic pressure. Some of these methods already exist and have been adapted and others have
been newly developed. All of the methods are quite easy to perform and the results are very
repeatable. The data obtained by the physical and chemical methods have been correlated together.
The physical methods gave new information about the nutritive value of feedstuffs for ruminants
and they can be used to differentiate between feedstuffs. They might explain part of the role played
by rumen ¯ora on feedstuffs which is not taken into account by the chemical approach.
# 2000 Elsevier Science B.V. All rights reserved.
Keywords: Feedstuffs; Ruminants; Physical properties; Particle size; Osmotic pressure; Water holding capacity
1. Introduction
The physical characteristics of feedstuffs for ruminants are rarely measured,
particularly in relation to their nutritional properties that could be taken into account
in feed formulation. It is known that some of these characteristics might partly explain the
interaction between rumen ¯ora and feedstuffs degradation. Particle density in¯uences
their rate of passage from the rumen (Ehle, 1984; Martz and Belyea, 1986) and thus
ruminal turnover rate of feeds and possibly their level of intake (Singh and Narang, 1991).
Particle size of feedstuffs in¯uences the surface area available for micro-organisms attack
(Owens and Goetsch, 1988) and thus their multiplication (Dehority and Orpin, 1988). It
*
Tel.: 33-1-44081768; fax: 33-1-44081853.
E-mail address: [email protected] (S. Giger-Reverdin)
0377-8401/00/$ ± see front matter # 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 7 - 8 4 0 1 ( 0 0 ) 0 0 1 5 9 - 0
54
S. Giger-Reverdin / Animal Feed Science and Technology 86 (2000) 53±69
also plays a role in the rate of passage of feedstuffs through the digestive tract (Balch,
1950). Water holding capacity (WHC) has an impact on microbial colonisation and
osmotic pressure is a factor that should be considered in the overall ecology of the rumen
(Bennink et al., 1978). Most methods concerning physical characteristics had been
applied either to roughages or to rations (Montgomery and Baumgardt, 1965; Allen et al.,
1984; Hooper and Welch, 1985; Singh and Narang, 1991), or to foods for human nutrition
(McConnell et al., 1974; Robertson and Eastwood, 1981).
The purpose of this paper is to describe easy-to-perform methods and to test them on a
set of feedstuffs representative of the range used in ruminant diets thereby allowing
discussion of their potential usefulness compared to more classical chemical methods.
2. Materials and methods
2.1. Feedstuffs
Twenty-four feedstuffs were tested. They can be grouped into six main classes:
cereals (oats, wheat, barley, maize, sorghum),
cereal by-products (wheat bran, brewers grains, corn gluten feed, corn gluten meal),
legumes (faba beans, peas, lupine seed),
oilmeals (soyabean meal, coconut meal, rapeseed meal, sunflower meal, palm kernel
meal),
agro-industrial by-products (soyabean hulls, citrus pulp, sugar beet pulp),
roughages (corn silage, corn stover, alfalfa hay, dehydrated alfalfa).
These ingredients were chosen to represent the wide range of dietary sources used in
rations or in compound feedstuffs for ruminants. Moreover, a part of the choice was
performed so that the ingredients differed as much as possible from a chemical point of
view. All the samples were ground through a 1 mm screen. Wet samples (corn silage,
citrus pulp) were oven-dried at 608C for 48 h.
2.2. Physical methods
2.2.1. Bulk density
Bulk density was determined by a modi®cation of the method of Montgomery and
Baumgardt (1965), where manual swirling of the container between the palms was
replaced by an automatic method, and the fourth step of feedstuff addition was not
performed. The following procedure was used to estimate the bulk density of the dried
ground samples:
A 100 ml glass graduated cylinder (2.7 cm internal diameter) was filled with sample to
the 50 ml mark and swirled for 15 s. Weight of the sample and volume occupied were
recorded.
Additional sample was added up to the 100 ml mark and the cylinder was swirled for
10 s.
S. Giger-Reverdin / Animal Feed Science and Technology 86 (2000) 53±69
55
The cylinder was refilled to 100 ml mark and swirled for 5 s. Total weight of sample
and final volume were recorded.
As density is de®ned as the mass of the substrate over the mass of an equivalent volume
of water, bulk density was equal to the weight of sample (expressed in mg) over the
volume occupied (expressed in ml).
2.2.2. Median particle size
Median particle size of feedstuffs was measured by a dry-sieving method according to
Melcion and de Monredon (1991). About 100 g of each feedstuff was weighed into the
upper sieve and shaken from side to side for 15 min. For wheat bran, alfalfa hay, corn
silage and corn stover, the largest particles clogged the sieve meshes. Therefore, 50 g of
these feedstuffs was used. The sizes of the six sieves were 0.400, 0.315, 0.250, 0.200,
0.125 and 0.080 mm. Refusals from each sieve were weighed, and the percentage of each
class was calculated. The cumulated percentage was calculated for each sieve from the
largest sieve to the smallest. Results were plotted against the logarithm of sieve size
(AFNOR, 1985). Median particle size (D50) was read directly as corresponding to a
virtual screen which would retain 50% of the particles, and D16 and D84 were,
respectively, those corresponding to 16 and 84% probability (Allen et al., 1984). In order
to estimate, the distribution of the particles around the median particle size, a relative
deviation was estimated as the difference between D16 and D84 divided by the median
particle size and called RD50.
2.2.3. Water holding capacity
WHC of feedstuffs was measured using an adaptation of one of the methods proposed
by Robertson and Eastwood (1981). The whole sample was used instead of its ®brous
part. A mass of 2.5 g of sample was left to soak for 16±24 h in 250 ml of distilled water,
so that water was in excess. The sample was then ®ltered on a fritted glass crucible
(porosity 2) and the walls of the beaker were carefully rinsed. The wet sample was
weighed after letting water decant for 10 min. WHC was the quantity of water retained by
the sample and expressed as l kgÿ1 sample dry matter.
2.2.4. Feed solubilisation
The ®ltrate sample collected after passage through the ®lter was oven-dried for 72 h at
1038C, weighed, and subsequently ashed at 5508C overnight and weighed. Dry matter
and ash solubilised were expressed as g lÿ1 or as percentage of initial weight of the
component considered.
2.2.5. Intrinsic osmotic pressure
2.5 g of sample was soaked in 50 ml of water for 24 h. Filtration was performed
according to the WHC method already described, but without the addition of water.
Osmotic pressure was measured on the ®ltrate using the freezing point depression
technique with a Mark 3 Osmometer, manufactured by Fiske (USA).
56
S. Giger-Reverdin / Animal Feed Science and Technology 86 (2000) 53±69
2.3. Chemical methods
The feedstuffs were analysed by standard methods: dry matter was estimated from
water content (AFNOR, 1982), ash (AFNOR, 1977), crude protein (ISO, 1997),
enzymatic starch (AFNOR, 1997a) and fat (AFNOR, 1997b). Cell wall content was
estimated by the neutral detergent ®bre (NDF) method of Van Soest and Wine (1967) as
modi®ed by AFNOR (1997c). Lignocellulose (ADF) and lignin (ADL) were obtained
using a sequential approach (AFNOR, 1997c) on the NDF residue as proposed by Giger
et al. (1987).
2.4. Statistical analysis
The SAS package (SAS, 1987) was used for statistical calculations.
When analyses were in duplicate for a sample, a variance analysis including a feed
effect was performed and the so-obtained residual standard deviation (RSD) was divided
by the mean value in order to estimate the precision of the method involved. This unitless
value which could be called coef®cient of variation would allow to estimate the
respective abilities of the methods to discriminate the feedstuffs.
Regressions were performed to study the relationships between physical and chemical
parameters. The link between the dependent variable and the one or more independent
variables was analysed using the RSD of the dependent variable.
A principal component analysis (PCA) was used to examine the relationships among
several quantitative variables, as described by Lebart and Fenelon (1971).
3. Results
3.1. General considerations about chemical and physical compositions
Chemical parameters for each feedstuff are summarised in Table 1. For some
feedstuffs, either starch or fat were not determined, because those feeds were assumed to
contain only traces of these elements. The statistical parameters were determined without
them (Table 1), but in the PCA, they were assume to be equal to 0 g kgÿ1 DM. As
expected, feedstuffs showed a wide variability in chemical composition.
All physical determinations have been performed in duplicate on all feedstuffs, what
allowed calculating a coef®cient of variation as de®ned before.
3.2. Bulk density
In fact, two types of bulk density were measured. The ®rst one was obtained after the
®rst swirling and was called density50, and the second one after the third swirling, called
density100. They were directly proportional and their mean values (0.561 vs 0.541) did
not statistically differ
Density100 0:966 density50
r 2 0:995; n 24; RSD 0:0110
57
S. Giger-Reverdin / Animal Feed Science and Technology 86 (2000) 53±69
Table 1
Chemical parameters of feedstuffs (g kgÿ1 DM)
Crude
protein
Fat
Starch
NDF
ADF
ADL
Ash
Oats
Wheat
Barley
Maize
Sorghum
Wheat bran
Brewers grains
Corn gluten feed
Corn gluten meal
Faba beans
Peas
Lupine seed
Soyabean meal
Coconut meal
Rapeseed meal
Sun¯ower meal
Palm kernel meal
Soyabean hulls
Citrus pulp
Sugar beet pulp
Corn silage
Corn stover
Alfalfa hay
Dehydrated alfalfa
92
103
141
104
114
165
300
188
650
294
233
382
472
146
415
381
164
124
81
85
73
44
172
157
47
17
20
40
36
30
57
31
24
15
13
110
19
29
11
20
97
22
25
ND
ND
ND
ND
ND
432
727
571
721
707
149
48
234
239
426
495
9
NDa
ND
ND
ND
ND
ND
ND
ND
250
ND
ND
ND
336
122
210
108
103
450
505
407
5
120
110
206
102
578
230
430
612
622
215
427
479
776
553
525
162
35
75
31
65
135
201
109
4
90
85
145
59
429
160
286
381
446
140
222
237
437
365
370
40
13
11
8
25
37
53
10
1
12
6
7
6
43
50
88
103
19
15
16
35
36
85
90
34
19
26
16
19
71
36
54
14
40
35
39
76
57
79
77
46
51
78
93
57
75
108
111
Statistical parameters
Mean
No. of feedstuffs
Standard deviation
212
24
152
35
19
27
385
13
251
343
24
214
195
24
142
34
24
30
55
24
28
a
ND: not determined.
Moreover, the coef®cient of variation of density100 (0.99%) was less than half that of
density50 (2.05%). This means that the density100 was a more repeatable method than
density50. This method with three steps of feedstuff addition and three swirlings was thus
used, and called simply bulk density. Bulk density varied greatly between feedstuffs
(Table 2). Corn stover had the lowest density (0.138 kg lÿ1) and peas the highest
(0.788 kg lÿ1).
Bulk density was negatively correlated with the cell wall content parameters of
feedstuffs The best relationship was with NDF content (Fig. 1):
Bulk density 0:743 ÿ 0:589 NDF kg kgÿ1 DM
r 2 0:644; n 24; RSD 0:0955
Therefore, for the same bulk density of around 0.5, NDF content varied from 206 g kgÿ1
DM (lupine seed) to 612 g kgÿ1 DM palm kernel meal, ADF explained 44.9% of the
variation of bulk density, and ADL, only 19.3%.
58
S. Giger-Reverdin / Animal Feed Science and Technology 86 (2000) 53±69
Table 2
Physical parameters of feedstuffs (bulk density, median particle size and WHC)a
D50c
Cereals
Oats
Wheat
Barley
Maize
Sorghum
0.469k,l
0.665e
0.566i
0.651f
0.686d
0.460d,e
0.315h,i,j,k
0.380f,g
0.530c
0.400f
2.860b,c
3.017b,c
2.048b,c,d,e
1.387d,e
1.788c,d,e,j
2.18i,j
1.98j
2.28i,j
1.62k
1.50k
Cereal by-products
Wheat bran
Brewers grains
Corn gluten feed
Corn gluten meal
0.340n
0.465k,l
0.604h
0.672e
0.740b
0.235l
0.275j,k,l
0.255k,l
3.162b
1.489d,e
1.127e
0.978E
3.07e,f
4.07c,d
2.53h,i
2.01j
Legumes
Faba beans
Peas
Lupine seed
0.701c
0.788a
0.460l
0.325h,i,j
0.343g,h
0.335g,h,i
2.190b,c,d,e
4.080a
1.108e
2.46h,i
2.44h,i
3.79d
Oilmeals
Soyabean meal
Coconut meal
Rapeseed meal
Sun¯ower meal
Palm kernel meal
0.701c
0.476k
0.651f
0.473k,l
0.519j
0.535c
0.480d
0.305h,i,j,k
0.420e,f
0.450d,e
2.972b,c
2.154b,c,d,e
1.739c,d,e
2.107b,c,d,e
1.241d,e
3.22e,f
4.08c,d
3.35e
3.24e,f
2.70g,h
Agro-industrial by-products
Soyabean hulls
Citrus pulp
Sugar beet pulp
0.448m
0.715b
0.625g
0.480d
0.270j,k,l
0.390f
1.750c,d,e
1.853c,d,e
2.455b,c,d
5.23b
4.33c
5.37b
Roughages
Corn silage
Corn stover
Alfalfa hay
Dehydrated alfalfa
0.318o
0.138p
0.333n
0.530j
0.280i,j,k,l
0.875a
0.260k,l
0.280i,j,k,l
1.873c,d,e
4.127a
1.808c,d,e
2.004b,c,d,e
4.09c,d
8.87a
3.80d
2.87f,g
Statistical parameters
Mean (24 feedstuffs)
Standard deviation
Coefficient of variation (%)
0.541
0.157
0.99
0.401
0.155
4.63
a
RD50d
WHC (l kgÿ1
DM of sample)
Bulk densityb
2.139
0.853
15.90
3.382
1.567
3.97
Values within columns with different superscript letters indicate a signi®cant feed effect at p