Directory UMM :Data Elmu:jurnal:S:Small Ruminant Research:Vol39.Issue1.Jan2001:
Small Ruminant Research 39 (2001) 47±57
Comparison of hay and silage for pregnant and
lactating Finnish Landrace ewes
R. Sormunen-Cristiana,*, L. Jauhiainenb
a
b
Animal Production Research, Agricultural Research Centre of Finland, FIN-31600 Jokioinen, Finland
Data and Information Services, Agricultural Research Centre of Finland, FIN-31600 Jokioinen, Finland
Accepted 16 June 2000
Abstract
A comparison of conserving timothy/meadow fescue/meadow grass as silage or hay was conducted with Finnish Landrace
ewes. Intake of hay (H), silage (S) and hay/silage (HS), and the performance of 30 Finnish Landrace ewes carrying twins,
triplets and quadruplets and suckling either twin or triplet lambs was measured during the last 8 weeks of pregnancy and
during the ®rst 6 weeks of lactation. The dry matter (DM) intake of HS was lower ( p < 0:03) than H during lactation, whereas
there was no difference in DM intake between pregnant and lactating ewes fed either H or S. The intake of metabolisable
energy was on an average 12.8, 14.3 and 13.4 MJ per day in pregnancy, and 17.7, 20.2 and 18.1 MJ in lactation for H, S and
HS, respectively. Daily intakes of protein assessed in terms of amino acids absorbed in the small intestine (AAT) were on an
average 92, 110 and 102 g for H, S and HS during pregnancy and 157, 184 and 165 g during lactation, respectively. Ewes in all
treatments were, relative to calculated requirements, de®cient in energy and protein during the last month of pregnancy and
early lactation. Ewes performed consistently better when fed silage than hay. Lamb mortality was low for all treatments. The
number of arti®cially reared lambs tended to be highest when ewes were fed both hay and silage ad libitum. Lamb growth was
higher on S than H based diet ( p < 0:005). With respect to winter feeding of ewes, grass silage compared favourably with hay
and indicated that hay could be replaced by well preserved grass silage. During late pregnancy and early lactation, there was
no advantage of supplementing silage with hay, relative to entirely silage based diets. # 2001 Elsevier Science B.V. All rights
reserved.
Keywords: Hay; Lactation; Pregnancy; Silage; Sheep
1. Introduction
In northern Europe, pregnancy and lactation of
most sheep ¯ocks occurs during winter. For 6±8
months ¯ocks are housed indoors and fed conserved
*
Corresponding author. Fax: 358-3-41883661.
E-mail address: [email protected]
(R. Sormunen-Cristian).
forages. Climate conditions in Finland offer greater
opportunities to make good-quality grass silage
than hay. However, hay is widely used for winter
feeding on Finnish sheep farms. Hence, silage
feeding is not typical, and therefore information
concerning the suitability for pregnant and lactating
Finnish Landrace ewes under Finnish conditions is
limited. Dry matter (DM) intake of ensiled crop
has often been reported to be lower in sheep
than that of the same crop conserved as hay (Thomas
et al., 1969; Orr and Treacher, 1989). However, in
0921-4488/01/$ ± see front matter # 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 1 - 4 4 8 8 ( 0 0 ) 0 0 1 6 7 - X
48
R. Sormunen-Cristian, L. Jauhiainen / Small Ruminant Research 39 (2001) 47±57
spite of higher intake, animal performance on hay
based diets has not always been improved (Thomas
et al., 1969). The reasons for inconsistencies between
forages remain unclear but may be explained by
differences between breed, ewe body weight, number
of foetuses and suckling lambs in addition to the often
great variations in quality of silage and differences in
forage digestibility.
Foetal requirements increase rapidly 6 weeks before
lambing (SjoÈdin, 1983), whereas voluntary intake
declines particularly in multiparous ewes (Orr and
Treacher, 1984). On the other hand, voluntary feed
intake, at a minimum just before lambing, increases
progressively during lactation reaching a maximum
5±8 weeks post-lambing. However, requirements for
energy and protein are highest between the second and
third weeks of lactation (Sormunen-Cristian et al.,
1997). To meet these greater requirements during
the last weeks of pregnancy and early lactation, it is
not clear whether multiparous ewes should be fed hay
or silage. Silage feeding increases the proli®cacy of
ewes (Nedkvitne, 1969). However, adult Finnish
Landrace ewes lambing once a year, are very proli®c
producing on an average 2.8 lambs per ewe (Savolainen, 1999) and therefore an increase of proli®cacy is
no longer desirable.
The present study was undertaken to compare hay
and grass silage harvested in the typical farm conditions as the basal forage for pregnant and lactating
multiparous Finnish Landrace ewes supplemented
with the same amount of concentrates. Furthermore,
the in¯uence of conserved forages on the performance
of ewes, on lamb growth and development was
assessed.
2. Material and methods
2.1. Conserved forages
Hay was prepared from a timothy (Phleum pratense
L.)/meadow fescue (Festuca pratensis Huds.)/meadow grass (Poa pratensis L.) sward at the beginning
of July and second cut silage was prepared from the
same sward on 23 August. Silage was ensiled using an
acid based additive (80% formic acid, 2% orthophosphoric acid) at a rate of 4.0 l tÿ1 into a
tower silo. The study was carried out at Agricultural
Research Centre of Finland (618400 N, 278130 E, 107 m
above sea level).
2.2. Animals, feeding and management
One month before a 4-week mating period (26
October±22 November), 30 mature pure bred Finnish
Landrace ewes were randomly selected from the 120
ewe experimental ¯ock. Ewes weighed on an average
59.4 kg (S.D. 7.8) and were between 18 and 30 months
of age. Ewes were allocated according to weight and
age into 10 blocks of three animals. Ewes within each
block were fed one of three experimental rations,
which consisted of hay (H), or silage (S), or hay
and silage (HS), each offered ad libitum. Forages
were offered 15% in excess of the previous day's
consumption. For ¯ushing (12 October±22 November), ewes were daily supplemented with 0.3 kg of
oats. For mating, ewes were exposed to two Finnish
Landrace rams equipped with a crayon harness.
Marked ewes in heat were recorded daily.
During the last 2 months of pregnancy a mixture of
rolled barley and oats (50:50 on an air-dry basis) was
offered 0.3±0.4 kg per ewe and 0.4 kg per one suckling
lamb during lactation. For 3 weeks before, and 4
weeks after parturition 15% of molassed sugar
beet pulp was added into the concentrate mixture.
A mineral supplement, salt and water were freely
available. In addition to milk, the lambs were offered
the same forage as to their respective dam. Lambs
also had ad libitum access to a commercial protein
concentrate.
Approximately 10 weeks before parturition all ewes
were housed in individual wooden pens (each 2.4 m2).
Individual intake of ewes was recorded daily for 8
weeks before parturition and during the ®rst 6 weeks
of lactation. In the case that ewes were unable to rear
all her offspring, lambs were transferred to a nursery
facility.
Live weight of ewes were recorded at the beginning
of the mating period, 56, 28 and 7 days before
parturition, 2 days post-partum and every week during
lactation. Lambs were weighed at birth and at 2-week
intervals up to 6 weeks. Initial and ®nal weights for
ewes and lambs were measured over two consecutive
days. Dates of mating and lambing, the number of
lambs born and lamb mortality were recorded for all
experimental animals.
R. Sormunen-Cristian, L. Jauhiainen / Small Ruminant Research 39 (2001) 47±57
2.3. Analytical methods
Feed samples were collected at each feeding and
pooled over a 2-week period. DM content of feeds was
determined by oven drying at 1058C for 24 h. Silage
DM was corrected for loss of volatiles according to
Huida et al. (1986). Samples of forages and concentrates were analysed according to standard procedures
(AOAC, 1980). In addition, silage total and watersoluble nitrogen (Kjeldahl method), ammonium nitrogen (McCullough, 1967), pH and lactic acid (Barker
and Summerson, 1941), volatile fatty acid (Huida,
1973) and water-soluble carbohydrate (Somogyi,
1945) content was determined. Digestibility coef®cients were taken from in vivo experiments. The net
energy value of experimental feeds was calculated as
feed units (FFU) according to Salo et al. (1990) and
metabolisable energy (ME) according to MAFF
(1975). Protein intake was calculated in terms of
amino acids absorbed in the small intestine (AAT)
and protein balance in the rumen (PBV) (Tuori et al.,
1996). In the calculations of PBV and AAT a value of
75, 85, 77 and 79% was assumed for the rumen
degradability of hay, silage, barley±oats mixture and
commercial concentrate protein, respectively (Tuori
et al., 1996).
49
If rearing type treatment interaction was not statistically signi®cant ( p > 0:10), the covariate was
removed from the model (Littell et al., 1996), whereas
rearing type was retained in the model despite being
non-signi®cant.
When the response variable was lamb birth weight,
the statistical model contained birth type and birth
type treatment interaction as a covariate. Birth type
was classi®ed into two categories for analysis: less
than or at least three lambs. If birth type treatment
interaction was not statistically signi®cant ( p > 0:10),
the covariate was removed from the model.
Birth type was included in the model despite being
non-signi®cant.
Measurements of ewe intake and live weight of ewe
and lamb were repeated several times for each animal,
and were found to be correlated. Consequently this
correlation was taken into account in the selected
statistical model (2). Covariance structure of the
repeated measurements was chosen by comparing
potential structures using Akaike's and Schwarz's
Bayesian information criterion (Wol®nger, 1996).
Compound symmetry (CS) covariance structure
proved useful for all data, except lamb live weight
for which an unstructured (UN) approach was used.
Analysis of variance for repeated measurements was
performed according to the following model:
2.4. Statistical methods
Yij m Ai Bj eij Tk A Tjk
Experimental data were statistically analysed as a
randomised complete block design using the restricted
maximum likelihood (REML) estimation method
within the `mixed procedure' of SAS (SAS, 1992)
according to the following model:
Yij m Ai Bj eij
(1)
where m is the intercept, Ai the ®xed effect from the ith
treatment, Bj the random effect from the jth block, eij is
a random variable that represents the error associated
with the ijth cell. Model (1) assumes that there is no
interaction between treatment and block and that the
eij's are independent normal random variables with
zero means and the same unknown variance.
When the response variable was ewe daily intake
and lamb weight gain, the statistical model included
the number of lambs reared by ewe (rearing type) and
the rearing type treatment interaction as covariates.
B Tik dijk
(2)
where m, Ai, Bj and eij are equivalent terms to that in
model (1). Tk and (A T)jk are the ®xed effect of time
and treatment time interactions, respectively,
(B T)ik the random effect of block time interactions and dijk represents experimental error.
When the response variable was lamb live weight,
the statistical model (2) contained rearing type, rearing type treatment interaction, rearing type time
interaction and rearing type treatment time interaction as a covariate. Rearing type was included in the
model despite being non-signi®cant. Interactions
between the other terms were removed from model,
if they were not statistically signi®cant ( p > 0:10).
Assumption used for models (1) and (2) was
checked using graphical methods by a box-plot to
test normality of errors and plots of residuals to test the
constancy of error variance (Neter et al., 1996).
50
R. Sormunen-Cristian, L. Jauhiainen / Small Ruminant Research 39 (2001) 47±57
Logistic regression was employed for the analysis of
data from birth and rearing type. Data were collected
from 10 blocks, and therefore a different constant term
was required for each block, while the common
logistic regression had only one constant term.
Regression analyses were carried out using LogXact-software (Mehta and Patel, 1996).
acetic acid, respectively, with a pH of 4.11 and soluble
and ammonium N content of 53.7 and 5.7% of total N,
respectively. Silage offered during pregnancy and
lactation contained 2.4, 9.1, 2.8% water-soluble carbohydrate, lactic acid and acetic acid, respectively,
with a pH of 4.07 and soluble and ammonium N
content of 52.5 and 5.8% of total N, respectively.
3.2. Feed intake
3. Results and discussion
3.1. Forage chemical composition
Chemical composition and feeding value of hay and
silage were constant throughout the experiment
(Table 1). Hay contained more ®brous constituents,
less protein and had a lower digestibility than silage.
Due to a late harvest, silage was also found to be too
coarse for feeding during lactation. It is recommended
that crude ®bre content of silage for lactating ewes
should not exceed 20% in the DM (SjoÈdin, 1983).
Fermentation quality of silage was good as assessed
by pH, ammonium nitrogen and butyric acid. In May,
when the preservation conditions deteriorated as a
result of higher ambient temperatures, silage contained NH3 just over 8% of total nitrogen, but no
butyric acid. Silage fed during mating contained 7.1,
8.3, 2.9% water-soluble carbohydrate, lactic acid and
3.2.1. Pre-lambing intake
Intake of concentrates during mating and pregnancy
was good. Refusals of molassed sugar beet pulp were
observed only on a few occasions. The hypothesis of
Andrieu et al. (Dulphy and VanOs, 1996) that ewes
consume more DM from hay than silage was not
supported by the current study (Table 2). Forage
DM intakes were similar between H and S diets
(1.03 versus 1.02 kg per ewe per day) during pregnancy, but lower for HS diets (0.95 kg per ewe
per day). Expressed on a metabolic body weight basis
DM intakes of H, S and HS during pregnancy were on
an average 45, 39 and 37 g (per kg W0.75). Differences
in DM intakes between forages were not as large as
that reported by Thomas et al. (1969). In this study
DM intake of hay was considered to be decreased by a
high ®bre and a low protein content due to a late
cutting date, which is consistent with the study of
Table 1
Chemical composition, quality and feed values of experimental feeds offered during mating (I), pregnancy (II) and lactation (II) for ewes and
during lamb rearing (III)
Hay
I (n2)
a
Dm (%)
In DM (%)
Organic matter
Crude protein
Ether extract
Crude fibre
Nitrogen-free extract
Digestible organic matter
Feed values (per kg DM)
NEa (FFU)
Mea (MJ)
AATa (g)
PBVa (g)
Silage
II±III (n4)
Concentrate
I (n3)
II±III (n8)
I (n1)
II (n4)
III (n1)
84.7
78.8
21.8
23.2
87.0
85.3
87.2
93.2
9.5
1.9
32.2
49.6
58.4
93.4
10.0
1.8
31.3
50.3
58.6
91.7
16.2
5.9
26.0
43.6
64.9
91.8
16.7
6.7
25.1
43.3
65.0
97.3
12.2
7.1
8.0
70.0
75.8
96.9
12.2
3.7
7.6
73.4
72.1
89.9
24.2
4.8
7.7
53.2
75.9
0.55
8.8
62
ÿ34
0.56
8.8
63
ÿ31
0.74
10.4
80
25
0.74
10.4
80
29
1.12
13.1
94
ÿ31
1.01
12.0
93
ÿ29
1.09
12.6
110
66
a
DM: dry matter, NE: net energy, FFU: fattening feed unit (0.7 kg starch equivalent), ME: metabolisable energy, AAT: amino acids
absorbed in the small intestine (Tuori et al., 1996) and PBV: protein balance in the rumen (Tuori et al., 1996).
Weeks 8±5 pre-partum
Forage (kg Dma)
Total DM intake (kg)
Total Oma intake (kg)
NE (FFU)a
ME (MJ)a
Crude protein (g)
AAT (g)a
PBV (g)a
a
S.E.M.
Hay
(n10)
Silage
(n10)
Hay/silage
(n10)
1.13
1.39
1.31
1.23
13.20
146
95
ÿ41
1.08
1.34
1.25
1.07
14.53
201
111
13
0.98
1.24
1.16
1.02
13.20
176
100
3
0.054
0.054
0.051
0.044
0.557
7.9
4.2
2.0
Significance
Weeks 4±1 pre-partum
S.E.M.
p1
p2
Hay
(n10)
Silage
(n10)
Hay/silage
(n10)
0.50
0.45
0.31
Comparison of hay and silage for pregnant and
lactating Finnish Landrace ewes
R. Sormunen-Cristiana,*, L. Jauhiainenb
a
b
Animal Production Research, Agricultural Research Centre of Finland, FIN-31600 Jokioinen, Finland
Data and Information Services, Agricultural Research Centre of Finland, FIN-31600 Jokioinen, Finland
Accepted 16 June 2000
Abstract
A comparison of conserving timothy/meadow fescue/meadow grass as silage or hay was conducted with Finnish Landrace
ewes. Intake of hay (H), silage (S) and hay/silage (HS), and the performance of 30 Finnish Landrace ewes carrying twins,
triplets and quadruplets and suckling either twin or triplet lambs was measured during the last 8 weeks of pregnancy and
during the ®rst 6 weeks of lactation. The dry matter (DM) intake of HS was lower ( p < 0:03) than H during lactation, whereas
there was no difference in DM intake between pregnant and lactating ewes fed either H or S. The intake of metabolisable
energy was on an average 12.8, 14.3 and 13.4 MJ per day in pregnancy, and 17.7, 20.2 and 18.1 MJ in lactation for H, S and
HS, respectively. Daily intakes of protein assessed in terms of amino acids absorbed in the small intestine (AAT) were on an
average 92, 110 and 102 g for H, S and HS during pregnancy and 157, 184 and 165 g during lactation, respectively. Ewes in all
treatments were, relative to calculated requirements, de®cient in energy and protein during the last month of pregnancy and
early lactation. Ewes performed consistently better when fed silage than hay. Lamb mortality was low for all treatments. The
number of arti®cially reared lambs tended to be highest when ewes were fed both hay and silage ad libitum. Lamb growth was
higher on S than H based diet ( p < 0:005). With respect to winter feeding of ewes, grass silage compared favourably with hay
and indicated that hay could be replaced by well preserved grass silage. During late pregnancy and early lactation, there was
no advantage of supplementing silage with hay, relative to entirely silage based diets. # 2001 Elsevier Science B.V. All rights
reserved.
Keywords: Hay; Lactation; Pregnancy; Silage; Sheep
1. Introduction
In northern Europe, pregnancy and lactation of
most sheep ¯ocks occurs during winter. For 6±8
months ¯ocks are housed indoors and fed conserved
*
Corresponding author. Fax: 358-3-41883661.
E-mail address: [email protected]
(R. Sormunen-Cristian).
forages. Climate conditions in Finland offer greater
opportunities to make good-quality grass silage
than hay. However, hay is widely used for winter
feeding on Finnish sheep farms. Hence, silage
feeding is not typical, and therefore information
concerning the suitability for pregnant and lactating
Finnish Landrace ewes under Finnish conditions is
limited. Dry matter (DM) intake of ensiled crop
has often been reported to be lower in sheep
than that of the same crop conserved as hay (Thomas
et al., 1969; Orr and Treacher, 1989). However, in
0921-4488/01/$ ± see front matter # 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 1 - 4 4 8 8 ( 0 0 ) 0 0 1 6 7 - X
48
R. Sormunen-Cristian, L. Jauhiainen / Small Ruminant Research 39 (2001) 47±57
spite of higher intake, animal performance on hay
based diets has not always been improved (Thomas
et al., 1969). The reasons for inconsistencies between
forages remain unclear but may be explained by
differences between breed, ewe body weight, number
of foetuses and suckling lambs in addition to the often
great variations in quality of silage and differences in
forage digestibility.
Foetal requirements increase rapidly 6 weeks before
lambing (SjoÈdin, 1983), whereas voluntary intake
declines particularly in multiparous ewes (Orr and
Treacher, 1984). On the other hand, voluntary feed
intake, at a minimum just before lambing, increases
progressively during lactation reaching a maximum
5±8 weeks post-lambing. However, requirements for
energy and protein are highest between the second and
third weeks of lactation (Sormunen-Cristian et al.,
1997). To meet these greater requirements during
the last weeks of pregnancy and early lactation, it is
not clear whether multiparous ewes should be fed hay
or silage. Silage feeding increases the proli®cacy of
ewes (Nedkvitne, 1969). However, adult Finnish
Landrace ewes lambing once a year, are very proli®c
producing on an average 2.8 lambs per ewe (Savolainen, 1999) and therefore an increase of proli®cacy is
no longer desirable.
The present study was undertaken to compare hay
and grass silage harvested in the typical farm conditions as the basal forage for pregnant and lactating
multiparous Finnish Landrace ewes supplemented
with the same amount of concentrates. Furthermore,
the in¯uence of conserved forages on the performance
of ewes, on lamb growth and development was
assessed.
2. Material and methods
2.1. Conserved forages
Hay was prepared from a timothy (Phleum pratense
L.)/meadow fescue (Festuca pratensis Huds.)/meadow grass (Poa pratensis L.) sward at the beginning
of July and second cut silage was prepared from the
same sward on 23 August. Silage was ensiled using an
acid based additive (80% formic acid, 2% orthophosphoric acid) at a rate of 4.0 l tÿ1 into a
tower silo. The study was carried out at Agricultural
Research Centre of Finland (618400 N, 278130 E, 107 m
above sea level).
2.2. Animals, feeding and management
One month before a 4-week mating period (26
October±22 November), 30 mature pure bred Finnish
Landrace ewes were randomly selected from the 120
ewe experimental ¯ock. Ewes weighed on an average
59.4 kg (S.D. 7.8) and were between 18 and 30 months
of age. Ewes were allocated according to weight and
age into 10 blocks of three animals. Ewes within each
block were fed one of three experimental rations,
which consisted of hay (H), or silage (S), or hay
and silage (HS), each offered ad libitum. Forages
were offered 15% in excess of the previous day's
consumption. For ¯ushing (12 October±22 November), ewes were daily supplemented with 0.3 kg of
oats. For mating, ewes were exposed to two Finnish
Landrace rams equipped with a crayon harness.
Marked ewes in heat were recorded daily.
During the last 2 months of pregnancy a mixture of
rolled barley and oats (50:50 on an air-dry basis) was
offered 0.3±0.4 kg per ewe and 0.4 kg per one suckling
lamb during lactation. For 3 weeks before, and 4
weeks after parturition 15% of molassed sugar
beet pulp was added into the concentrate mixture.
A mineral supplement, salt and water were freely
available. In addition to milk, the lambs were offered
the same forage as to their respective dam. Lambs
also had ad libitum access to a commercial protein
concentrate.
Approximately 10 weeks before parturition all ewes
were housed in individual wooden pens (each 2.4 m2).
Individual intake of ewes was recorded daily for 8
weeks before parturition and during the ®rst 6 weeks
of lactation. In the case that ewes were unable to rear
all her offspring, lambs were transferred to a nursery
facility.
Live weight of ewes were recorded at the beginning
of the mating period, 56, 28 and 7 days before
parturition, 2 days post-partum and every week during
lactation. Lambs were weighed at birth and at 2-week
intervals up to 6 weeks. Initial and ®nal weights for
ewes and lambs were measured over two consecutive
days. Dates of mating and lambing, the number of
lambs born and lamb mortality were recorded for all
experimental animals.
R. Sormunen-Cristian, L. Jauhiainen / Small Ruminant Research 39 (2001) 47±57
2.3. Analytical methods
Feed samples were collected at each feeding and
pooled over a 2-week period. DM content of feeds was
determined by oven drying at 1058C for 24 h. Silage
DM was corrected for loss of volatiles according to
Huida et al. (1986). Samples of forages and concentrates were analysed according to standard procedures
(AOAC, 1980). In addition, silage total and watersoluble nitrogen (Kjeldahl method), ammonium nitrogen (McCullough, 1967), pH and lactic acid (Barker
and Summerson, 1941), volatile fatty acid (Huida,
1973) and water-soluble carbohydrate (Somogyi,
1945) content was determined. Digestibility coef®cients were taken from in vivo experiments. The net
energy value of experimental feeds was calculated as
feed units (FFU) according to Salo et al. (1990) and
metabolisable energy (ME) according to MAFF
(1975). Protein intake was calculated in terms of
amino acids absorbed in the small intestine (AAT)
and protein balance in the rumen (PBV) (Tuori et al.,
1996). In the calculations of PBV and AAT a value of
75, 85, 77 and 79% was assumed for the rumen
degradability of hay, silage, barley±oats mixture and
commercial concentrate protein, respectively (Tuori
et al., 1996).
49
If rearing type treatment interaction was not statistically signi®cant ( p > 0:10), the covariate was
removed from the model (Littell et al., 1996), whereas
rearing type was retained in the model despite being
non-signi®cant.
When the response variable was lamb birth weight,
the statistical model contained birth type and birth
type treatment interaction as a covariate. Birth type
was classi®ed into two categories for analysis: less
than or at least three lambs. If birth type treatment
interaction was not statistically signi®cant ( p > 0:10),
the covariate was removed from the model.
Birth type was included in the model despite being
non-signi®cant.
Measurements of ewe intake and live weight of ewe
and lamb were repeated several times for each animal,
and were found to be correlated. Consequently this
correlation was taken into account in the selected
statistical model (2). Covariance structure of the
repeated measurements was chosen by comparing
potential structures using Akaike's and Schwarz's
Bayesian information criterion (Wol®nger, 1996).
Compound symmetry (CS) covariance structure
proved useful for all data, except lamb live weight
for which an unstructured (UN) approach was used.
Analysis of variance for repeated measurements was
performed according to the following model:
2.4. Statistical methods
Yij m Ai Bj eij Tk A Tjk
Experimental data were statistically analysed as a
randomised complete block design using the restricted
maximum likelihood (REML) estimation method
within the `mixed procedure' of SAS (SAS, 1992)
according to the following model:
Yij m Ai Bj eij
(1)
where m is the intercept, Ai the ®xed effect from the ith
treatment, Bj the random effect from the jth block, eij is
a random variable that represents the error associated
with the ijth cell. Model (1) assumes that there is no
interaction between treatment and block and that the
eij's are independent normal random variables with
zero means and the same unknown variance.
When the response variable was ewe daily intake
and lamb weight gain, the statistical model included
the number of lambs reared by ewe (rearing type) and
the rearing type treatment interaction as covariates.
B Tik dijk
(2)
where m, Ai, Bj and eij are equivalent terms to that in
model (1). Tk and (A T)jk are the ®xed effect of time
and treatment time interactions, respectively,
(B T)ik the random effect of block time interactions and dijk represents experimental error.
When the response variable was lamb live weight,
the statistical model (2) contained rearing type, rearing type treatment interaction, rearing type time
interaction and rearing type treatment time interaction as a covariate. Rearing type was included in the
model despite being non-signi®cant. Interactions
between the other terms were removed from model,
if they were not statistically signi®cant ( p > 0:10).
Assumption used for models (1) and (2) was
checked using graphical methods by a box-plot to
test normality of errors and plots of residuals to test the
constancy of error variance (Neter et al., 1996).
50
R. Sormunen-Cristian, L. Jauhiainen / Small Ruminant Research 39 (2001) 47±57
Logistic regression was employed for the analysis of
data from birth and rearing type. Data were collected
from 10 blocks, and therefore a different constant term
was required for each block, while the common
logistic regression had only one constant term.
Regression analyses were carried out using LogXact-software (Mehta and Patel, 1996).
acetic acid, respectively, with a pH of 4.11 and soluble
and ammonium N content of 53.7 and 5.7% of total N,
respectively. Silage offered during pregnancy and
lactation contained 2.4, 9.1, 2.8% water-soluble carbohydrate, lactic acid and acetic acid, respectively,
with a pH of 4.07 and soluble and ammonium N
content of 52.5 and 5.8% of total N, respectively.
3.2. Feed intake
3. Results and discussion
3.1. Forage chemical composition
Chemical composition and feeding value of hay and
silage were constant throughout the experiment
(Table 1). Hay contained more ®brous constituents,
less protein and had a lower digestibility than silage.
Due to a late harvest, silage was also found to be too
coarse for feeding during lactation. It is recommended
that crude ®bre content of silage for lactating ewes
should not exceed 20% in the DM (SjoÈdin, 1983).
Fermentation quality of silage was good as assessed
by pH, ammonium nitrogen and butyric acid. In May,
when the preservation conditions deteriorated as a
result of higher ambient temperatures, silage contained NH3 just over 8% of total nitrogen, but no
butyric acid. Silage fed during mating contained 7.1,
8.3, 2.9% water-soluble carbohydrate, lactic acid and
3.2.1. Pre-lambing intake
Intake of concentrates during mating and pregnancy
was good. Refusals of molassed sugar beet pulp were
observed only on a few occasions. The hypothesis of
Andrieu et al. (Dulphy and VanOs, 1996) that ewes
consume more DM from hay than silage was not
supported by the current study (Table 2). Forage
DM intakes were similar between H and S diets
(1.03 versus 1.02 kg per ewe per day) during pregnancy, but lower for HS diets (0.95 kg per ewe
per day). Expressed on a metabolic body weight basis
DM intakes of H, S and HS during pregnancy were on
an average 45, 39 and 37 g (per kg W0.75). Differences
in DM intakes between forages were not as large as
that reported by Thomas et al. (1969). In this study
DM intake of hay was considered to be decreased by a
high ®bre and a low protein content due to a late
cutting date, which is consistent with the study of
Table 1
Chemical composition, quality and feed values of experimental feeds offered during mating (I), pregnancy (II) and lactation (II) for ewes and
during lamb rearing (III)
Hay
I (n2)
a
Dm (%)
In DM (%)
Organic matter
Crude protein
Ether extract
Crude fibre
Nitrogen-free extract
Digestible organic matter
Feed values (per kg DM)
NEa (FFU)
Mea (MJ)
AATa (g)
PBVa (g)
Silage
II±III (n4)
Concentrate
I (n3)
II±III (n8)
I (n1)
II (n4)
III (n1)
84.7
78.8
21.8
23.2
87.0
85.3
87.2
93.2
9.5
1.9
32.2
49.6
58.4
93.4
10.0
1.8
31.3
50.3
58.6
91.7
16.2
5.9
26.0
43.6
64.9
91.8
16.7
6.7
25.1
43.3
65.0
97.3
12.2
7.1
8.0
70.0
75.8
96.9
12.2
3.7
7.6
73.4
72.1
89.9
24.2
4.8
7.7
53.2
75.9
0.55
8.8
62
ÿ34
0.56
8.8
63
ÿ31
0.74
10.4
80
25
0.74
10.4
80
29
1.12
13.1
94
ÿ31
1.01
12.0
93
ÿ29
1.09
12.6
110
66
a
DM: dry matter, NE: net energy, FFU: fattening feed unit (0.7 kg starch equivalent), ME: metabolisable energy, AAT: amino acids
absorbed in the small intestine (Tuori et al., 1996) and PBV: protein balance in the rumen (Tuori et al., 1996).
Weeks 8±5 pre-partum
Forage (kg Dma)
Total DM intake (kg)
Total Oma intake (kg)
NE (FFU)a
ME (MJ)a
Crude protein (g)
AAT (g)a
PBV (g)a
a
S.E.M.
Hay
(n10)
Silage
(n10)
Hay/silage
(n10)
1.13
1.39
1.31
1.23
13.20
146
95
ÿ41
1.08
1.34
1.25
1.07
14.53
201
111
13
0.98
1.24
1.16
1.02
13.20
176
100
3
0.054
0.054
0.051
0.044
0.557
7.9
4.2
2.0
Significance
Weeks 4±1 pre-partum
S.E.M.
p1
p2
Hay
(n10)
Silage
(n10)
Hay/silage
(n10)
0.50
0.45
0.31