264 S
. Hermesch et al. Livestock Production Science 65 2000 261 –270
to converge, standard errors for genetic correlations weights in the second and third parity with estimates
were obtained using the formula of Robertson of 0.22 and 0.20. A lower genetic variance rather
1959. than an increase in environmental variance is the
Significant random effects were tested with a log cause for this reduced heritability for litter birth
likelihood ratio test. For reproductive traits of the weight in the first parity. Gilts were farrowing at 337
sow only maternal effects were analysed as an days on average and their uterine capacity is smaller
additional random effect. Only 19 of all sows with in comparison to multiparous sows. This might have
records had at least one full sister available in the been a restriction on expression of their genetic
data set and the data structure did not allow to fit potential in litter birth weight. Heritabilities were
litter effect as an additional random effect. General- 0.15, 0.16, 0.15 for average piglet birth weight in the
ly, maternal effects were not significant for reproduc- first three parities.
tive traits of the sow and were therefore not fitted in Literature estimates of heritabilities for litter birth
the model. weight and average piglet weight at birth varied
between research data and field data. Heritabilities for litter birth weight were 0.13 and 0.21 in the study
3. Results and discussion of Crump et al. 1997, which was based on field
data. In contrast, high estimates of heritabilities with 3.1. Heritability estimates for reproductive traits of
a range of 0.42–0.65 were obtained by Irvin and the sow
Swiger 1984 and by Ferguson et al. 1985 for litter birth weight and average piglet weight at birth,
Heritabilities and variance components for re- using data from research herds. Rydhmer et al.
production traits of the sow are presented in Table 2 1992 pointed out that litter birth weight is in-
along with the number of records and means for each fluenced by the milk intake of piglets after birth
trait. Litter size was lowly heritable with estimates of which depends on the milk performance of the sow.
0.08, 0.09 and 0.08 for the first three parities pooled An early weighing of the litter after birth is therefore
over both breeds. These estimates are in agreement important, and delays in recording litter birth weight
with literature estimates Southwood and Kennedy, in field data might be the reason for lower
1990; Alfonso et al., 1994; Irgang et al., 1994; heritabilities. The recording policy for litter birth
¨ Rydhmer et al., 1994; Rohe and Kennedy, 1995;
weight changed in the third quarter of 1993 in this Ducos and Bidanel, 1996; Tholen et al., 1996.
herd, when litter birth weight was recorded 3 days Litter birth weight in the first parity had a lower
after farrowing and includes cross-fostered piglets. heritability 0.08 in comparison to litter birth
This inconsistency in data recording and weighing after cross-fostering leads to an increase in environ-
mental variation and therefore reduced heritability
Table 2
2
Tholen et al., 1996.
Number of records n, means, heritabilities h with standard
2
errors S.E. and phenotypic variances s for reproduction traits
Litter weight at 21-days recorded in the first parity
p a
of the sow
LW21 was lowly heritable Table 2, which is in
1
2 2
2
Trait n
Mean h
S.E. of h s
agreement with estimates presented by Kaplon et al.
p
1991, Siewerdt et al. 1995 and Tholen et al.
NBA 5986
9.61 0.08
0.02 5.90
1
1996. Litter weight at 21-days was strongly in-
NBA 4113
10.06 0.09
0.02 5.94
2
NBA 2965
10.79 0.08
0.03 6.11
fluenced by number of piglets weighed and the age
3
LBW Kg 4306
12.55 0.08
0.02 7.26
1
of the litter at weighing. Tholen et al. 1996
LBW Kg 2084
13.41 0.22
0.05 8.44
2
compared regression coefficients for number of
LBW Kg 1234
14.65 0.20
0.07 10.49
3
piglets after weighing. Linear and quadratic regres-
ABW g 4206
1342 0.15
0.03 124 129
1
sion coefficients varied considerably across parities
ABW g 2032
1390 0.16
0.04 98 286
2
ABW g 1216
1419 0.15
0.06 79 504
and herds. This shows possible limitations in adjust-
3
LW21 Kg 1111
41.73 0.07
0.06 86.52
1
ing environmental variation caused by different
a
For abbreviations of traits see Table 1.
policies in cross-fostering and weaning.
S . Hermesch et al. Livestock Production Science 65 2000 261 –270
265
3.2. Estimation of genetic correlations first to third parity rg 5 2 0.15, 2 0.12, 2 0.20.
However, litter birth weight in later parities was 3.2.1. Reproductive traits of the sow
mainly positively correlated with number born alive Genetic and environmental correlations between
rg 5 2 0.04 to 0.43. In this study litter birth weight reproduction traits of the sow are presented in Table
was not corrected for number born alive. Using a 3. Estimates of genetic correlations between litter
similar model, Bereskin 1984, Irvin and Swiger size in the first to third parity NBA
were 1984 and Siewerdt et al. 1995 found positive
1,2,3
positive. Number born alive in the first parity was genetic correlations ranging from 0.42 to 0.92. The
genetically a different trait than litter size in the negative genetic correlations found in this study
second and third parity genetic correlation rg 5 between litter birth weight and number born alive
0.62 for NBA , rg 5 0.61 for NBA supporting the could be caused by the recording procedure of litter
2 3
analysis of number born alive in the first parity as a birth weight. Piglet mortality is higher in larger
separate trait to litter size in later parities. Although litters and in litters with a lower average birth weight
Haley et al. 1988 suggested analysing litter size as Fraser, 1990. Therefore, gilts with large litters are
repeated records, results from more recent studies more likely to have a higher mortality rate, especial-
¨ i.e., Alfonso et al., 1994; Rohe and Kennedy, 1995;
ly within the first few days after farrowing. With the Tholen et al., 1996 confirm that number born alive
delay of recording litter birth weight this will reduce in the first parity should be regarded as a different
the total litter weight. In addition, the cross-fostering trait than litter size in later parities. The genetic
practice of putting smaller piglets on to gilts and correlation between number born alive in the second
taking their own bigger piglets away to older sows and third parity was high 0.95 and these traits can
could also lead to this negative genetic correlation. It therefore be treated as repeated records in selection
was not possible to incorporate these cross-fostering programs. Similar, genetic correlations between litter
practices in the model. birth weight and average piglet weight at birth in the
Genetic correlations between litter size and aver- first parity and performance in these two traits in the
age piglet weight at birth varied from 2 0.86 to second and third parity were significantly different
2 0.27 Table 3. Rydhmer et al. 1992 also found
from one ranging from 0.52 to 0.79. Litter birth an unfavourable relationship between number born
weight and average piglet weight at birth in the first alive and average piglet weight at birth rg 5
litter should therefore be analysed as a separate trait 2
0.34, while Irvin and Swiger 1984 found no while performance in later parities should be re-
relationship rg 5 0.05 between these two traits. garded as repeated records.
Litter birth weight and average piglet weight at Litter birth weight in the first parity had low
birth were moderately correlated rg 5 0.29–0.87. negative genetic correlations with litter size in the
Irvin and Swiger 1984 found a positive genetic
Table 3 Genetic correlations above diagonal with standard errors in brackets and environmental correlations below diagonal between
reproduction traits
NBA NBA
NBA LBW
LBW LBW
ABW ABW
ABW LW21
1 2
3 1
2 3
1 2
3 1
a
NBA 0.620.19
0.610.30 2
0.150.17 0.130.22
2 0.040.20
2 0.740.10
2 0.340.25
2 0.270.32
2 0.140.32
1 a
NBA 0.11
0.950.04 2
0.120.17 0.430.18
0.280.29 2
0.410.21 2
0.690.16 2
0.400.35 2
0.150.30
2 a
NBA 0.09
0.12 2
0.200.30 0.220.27
0.250.29 2
0.510.24 2
0.860.36 2
0.560.25 2
0.750.17
3 a
LBW 0.42
0.06 0.10
0.780.16 0.520.28
0.780.13 0.870.23
0.550.36 0.140.32
1 a
LBW 0.08
0.45 0.05
0.10 0.980.34
0.470.17 0.290.21
0.410.32 2
0.100.31
2 a
LBW 0.06
0.04 0.62
0.11 0.00
0.290.28 0.480.29
0.730.27 0.430.31
3 a
ABW 2
0.66 2
0.09 0.00
0.29 2
0.02 2
0.01 0.790.19
0.580.30 2
0.110.29
1 a
ABW 2
0.05 2
0.67 2
0.07 0.07
0.27 0.03
0.07 0.780.33
0.420.27
2 a
ABW 2
0.03 2
0.09 2
0.57 0.09
0.05 0.21
0.04 0.18
0.480.32
3
LW21 2
0.07 0.05
0.10 0.14
0.07 0.00
0.28 2
0.04 0.00
1 a
Estimates of standard errors obtained from approximation of Robertson 1959.
266 S
. Hermesch et al. Livestock Production Science 65 2000 261 –270
correlation of 0.76 between these two traits. There- in the numerator relationship matrix. However, with
fore, a high litter birth weight is associated with a such data it is not possible to estimate environmental
high average piglet weight at birth. correlations between these traits and therefore it is
Genetic correlations between litter weight at 21- also not possible to obtain phenotypic correlations.
days and other reproduction traits of the sow were in Genetic correlations between reproductive traits of
general not significantly different from zero. In the sow and performance traits recorded on boars are
contrast, Siewerdt et al. 1995 found positive ge- presented in Table 4. Litter size was negatively
netic correlations ranging from 0.37 to 0.89 between correlated with growth rate traits. This relationship
21-day litter weight and litter size and litter birth was
stronger in
the first
parity rg 5 2 0.30,
weight. Differing from the study of Siewerdt et al. 2
0.42, than in later parities rg 5 2 0.30 to 0.00. 1995, litter weight at 21-days was adjusted for
In comparison, genetic correlations between litter number of piglets after cross-fostering in this study
size and growth rate ranged from 2 0.15 to 0.23 in which explains differences in estimates. Tholen et al.
the studies of Rydhmer et al. 1992, Short et al. 1996 also adjusted 21-day litter weight for number
1994, Merks and Molendijk 1995, Tholen et al. of piglets after cross-fostering, and genetic correla-
1996 and Crump et al. 1997. In this study genetic tions between number born alive and 21-day litter
correlations between reproduction traits of sows and weight averaged 2 0.43 across parities, while aver-
performance traits of their sons were analysed. age piglet weight at birth was positively correlated
Therefore, the size of the litter might have directly with 21-day litter weight average rg across parities,
influenced performance traits and growth rate in 0.32.
particular. For example, Dwyer et al. 1994 showed In order to optimise genetic improvement for
that a higher level of nutrition increased the sec- number of piglets weaned per sow litter size should
ondary to primary muscle fibre number ratio and be analysed in a multitrait analysis together with
subsequently increased growth rate. In further analy- litter birth weight or average piglet weight at birth
ses, genetic correlations between these two traits and 21-day litter weight. However, Tholen et al.
were obtained dividing the total reproductive data set 1996 found that 21-day litter weight is mainly
in half. In the first analysis reproductive records of influenced by the number of piglets after cross-
dams of boars with performance data were excluded fostering and the length of the time period between
from the reproductive data set. In contrast, the farrowing and weighing. For example, heavier litters
second analysis was only based on reproductive tend to be weaned and therefore weighed earlier in
information from dams of boars with growth rate order to make room for other sows that are due to
records. Given the hypothesis that the negative farrow. These management factors cannot always be
genetic correlation is caused by the direct influence explained well by the model which might lead to a
of litter size on growth performance of the offspring decrease in heritabilities and might influence genetic
an increase in genetic correlations is expected when correlations with other reproduction traits. In this
dams of boars are excluded from the reproductive study litter birth weight and average piglet weight at
data set. In addition, genetic correlations are ex- birth were also influenced by cross-fostering prac-
pected to decrease when litter size is only available tices for part of the data which might have influenced
from dams of boars with performance data. Among genetic parameters. In summary, these management
all trait combinations between growth rate and litter practices reduce the value of litter weight traits for
size traits, growth rate from 3 to 18 weeks ADG1 genetic improvement of sow productivity and should
and litter size in the first parity NBA were the
1
be minimised in nucleus herds. only trait combination where genetic correlations
changed in this expected way. The genetic correla- 3.2.2. Reproduction and production traits
tion was 2 0.05 for the data set when dams of boars Reproduction traits are measurements of the sow
were excluded from the reproduction data set and while performance, carcase and meat quality traits
2 0.68 when the reproductive data set included only
were recorded on offspring of sows. Genetic correla- dams of boars with performance records. For the
tions were therefore estimated through genetic links other trait combinations, genetic correlations did not
S . Hermesch et al. Livestock Production Science 65 2000 261 –270
267 Table 4
a
Genetic correlations with standard errors in brackets between reproduction and production, carcase and meat quality traits
NBA NBA
NBA LBW
LBW LBW
ABW ABW
ABW
1 2
3 1
2 3
1 2
3 b
b b
b b
b b
b b
ADG1 2
0.30 0.14
2 0.01
0.14 2
0.26 0.17
0.39 0.13
0.38 0.12
0.35 0.18
0.33 0.12
0.27 0.14
0.33 0.17
b b
b b
b b
b b
b
ADG2 2
0.42 0.16
2 0.30
0.17 0.00
0.20 0.18
0.20 0.08
0.19 0.12
0.22 0.45
0.14 0.35
0.17 0.09
0.25 FDINT
2 0.19
0.22 2
0.24 0.25
2 0.05
0.29 2
0.11 0.22
2 0.20
0.22 2
0.22 0.31
0.16 0.20
0.03 0.26
0.10 0.35
b
FCR 0.09
0.30 0.00
0.30 0.08
0.22 2
0.53 0.27
2 0.60
0.28 2
0.57 0.45
2 0.38
0.25 2
0.43 0.33
2 0.13
0.46
b b
LFDP2 0.10
0.17 0.17
0.19 2
0.07 0.23
2 0.39
0.19 2
0.43 0.21
2 0.30
0.25 2
0.07 0.11
2 0.33
0.09 2
0.14 0.13
b b
LFD3 4 0.09
0.17 0.16
0.19 2
0.05 0.23
2 0.35
0.09 2
0.23 0.11
2 0.06
0.09
b
LMD3 4 2
0.13 0.23
0.25 0.25
0.00 0.30
2 0.16
0.24 0.31
0.29 0.15
0.33 2
0.18 0.16
2 0.09
0.15 0.18
0.34 FDP2
2 0.15
0.21 0.16
0.23 2
0.07 0.27
2 0.23
0.22 2
0.44 0.25
2 0.08
0.31 0.05
0.32 FD3 4
2 0.28
0.23 2
0.02 0.26
2 0.28
0.28 2
0.26 0.25
2 0.54
0.27 2
0.22 0.36
0.12 0.17
2 0.13
0.26 0.13
0.37
b b
b b
b b
b b
BLW 2
0.45 0.13
2 0.23
0.15 2
0.24 0.19
0.46 0.13
0.42 0.13
0.42 0.16
0.29 0.14
0.27 0.16
b b
b b
b b
b b
b
LMW 2
0.31 0.14
2 0.26
0.13 2
0.08 0.18
0.52 0.11
0.55 0.10
0.61 0.11
0.13 0.13
0.32 0.14
0.37 0.17
b b
b b
b b
b b
b
pH45 2
0.34 0.16
2 0.19
0.17 2
0.26 0.21
2 0.30
0.17 2
0.22 0.14
2 0.40
0.17 0.13
0.16 0.05
0.18 0.30
0.21
b b
b b
b b
b b
b
pH24 2
0.26 0.18
0.10 0.18
2 0.25
0.22 0.08
0.19 0.17
0.17 2
0.23 0.20
0.19 0.16
2 0.03
0.19 0.08
0.24
b b
b b
b b
b b
b
CLD 2
0.11 0.16
2 0.53
0.11 2
0.27 0.18
2 0.11
0.16 2
0.28 0.14
0.41 0.12
0.00 0.15
0.36 0.14
0.31 0.18
b b
b b
b b
b b
b
CMD 2
0.42 0.12
2 0.45
0.11 2
0.34 0.16
0.00 0.14
2 0.27
0.13 2
0.09 0.16
0.21 0.12
0.00 0.14
0.25 0.19
b b
b b
b b
b b
b
DLP 0.17
0.16 0.34
0.14 2
0.08 0.20
0.05 0.16
0.10 0.15
0.42 0.15
2 0.23
0.14 0.34
0.14 0.25
0.21
b b
b b
b b
b b
b
IMF 2
0.11 0.14
0.08 0.14
0.11 0.18
2 0.37
0.13 2
0.32 0.12
2 0.26
0.15 2
0.15 0.13
2 0.19
0.14 2
0.12 0.18
a
ADG1, average daily gain from 3 to 18 weeks; ADG2, average daily gain during station testing from 18 to 22 weeks; FDINT, feed intake recorded during station testing from 18 to 22 weeks; FCR, feed conversion ratio defined as feed intake over growth rate 18–22
weeks; LFDP2, backfat depth at P2 measured with real time ultrasound; LFD3 4, backfat depth between the third and fourth last ribs measured with real time ultrasound; LMD3 4, muscle depth of m
. longissimusdorsi between the third and fourth last ribs on the live animal; FDP2, backfat depth at P2 measured with Hennesy Chong grading machine; FD3 4, backfat depth between third and fourth last ribs
measured with Hennesy Chong grading machine; BLW, weight of whole left back leg; LMW, weight of slash boned left back leg; pH45, pH measured 45 min after slaughter; pH24, pH measured 24 h after slaughter; CLD, L value of Minolta chromameter of m
. longissimus dorsi; CMD, L value of Minolta chromameter of m
. multifidus dorsi; DLP, drip loss percentage; IMF, intramuscular fat content.
b
Estimates of standard errors obtained from approximation of Robertson 1959. Estimate did not converge.
differ in accordance with expectations. Therefore, Litter birth weight and average piglet weight at
these analyses indicate that the negative genetic birth were both positively correlated with growth rate
correlation between growth rate measured earlier in rg 5 0.08–0.45. Positive genetic correlations be-
life ADG1 and litter size in the first parity may tween growth rate and litter birth weight and average
have been influenced by the direct effect of litter size piglet weight at birth have also been reported by
on growth rate. However, genetic correlations be- Vangen 1980, Hutchens et al. 1981, Rydhmer et
tween other trait combinations seem not to be al. 1992, Tholen et al. 1996 and Crump et al.
influenced by the direct effect of litter size. 1997.
Considering that a high growth rate was associated Low negative genetic correlations were found
with a high feed intake Hermesch et al., 2000b, the between litter birth weight and feed intake. Litter
lowly negative genetic correlations rg 5 2 0.19, birth weight and average piglet weight at birth were
2 0.24, 2 0.05 between litter size and feed intake
both negatively correlated with feed conversion ratio. are in agreement with relationships between litter
This favourable relationship was stronger for litter size and growth rate. The average of genetic correla-
birth weight, with estimates of 2 0.53, 2 0.60 and tions between litter size and feed intake presented by
2 0.57 for the three parities, than for average piglet
Short et al. 1994 was 2 0.12, thus supporting these weight at birth, with genetic correlations with feed
results. In contrast, Crump et al. 1997 presented efficiency of 2 0.38, 2 0.43 and 2 0.13 for the first
slightly positive genetic correlations between litter three parities. The strong genetic correlations be-
size and feed intake and feed conversion ratio of tween litter birth weight and feed conversion ratio
0.20 and 0.23, respectively. result from positive genetic correlations between
268 S
. Hermesch et al. Livestock Production Science 65 2000 261 –270
litter birth weight and growth rate and lowly negative apparent between pH45 and litter birth weight rg
genetic correlations between litter birth weight and range, 2 0.40 to 2 0.22. A darker colour is related
feed intake. Furthermore, this favourable genetic to a larger litter size rg range, 2 0.53 to 2 0.11
relationship is supported by Kerr and Cameron and a higher litter birth weight rg range, 2 0.28 to
1995 who found a reduced litter weight in the 0.00. Genetic correlations ranged from 2 0.08 to
selection line selected for high lean feed conversion 0.34 for litter size in the first to third parity and drip
ratio. loss percentage. Overall, genetic correlations be-
tween reproduction traits and meat quality traits were inconsistent and mostly of low magnitude indicating
3.2.3. Reproduction and carcase traits no clear genetic relationships between these trait
Genetic correlations between number born alive groups. Estimates of genetic correlations between
and backfat measured with real time ultrasound reproduction traits and meat quality traits were not
ranged from 2 0.07 to 0.17 Table 4. Estimates of available in the literature.
genetic correlations between litter size and carcase Finally, intramuscular fat content had no genetic
backfat measurements varied from 2 0.28 to 0.16. relationship with litter size. However, a higher
This indicates no genetic relationship between lean- intramuscular fat content was associated with a lower
ness and litter size which is also apparent in genetic litter birth weight with genetic correlations ranging
correlations of muscle depth with litter size. Litera- from 2 0.37 to 2 0.26 and a lower average piglet
ture estimates between litter size and leanness were weight at birth. Genetic correlations were 2 0.15,
generally close to zero Johansson and Kennedy, 2
0.19 and 2 0.12 for the first three parities. In- 1983; Short et al., 1994; Ducos and Bidanel, 1996;
tramuscular fat content was reduced for leaner pigs Tholen et al., 1996.
Hermesch et al., 2000b. Leanness was not ge- Weight of the back leg and lean meat weight of
netically related to litter size Table 4, while litter the back leg were closely related to average daily
birth weight and average piglet weight at birth were gain Hermesch et al., 2000b. Negative genetic
favourably related to leanness. These genetic correla- correlations in the range of
2 0.45 to
2 0.08
tions between reproduction traits and intramuscular between these two weight measurements and litter
fat content are therefore in agreement with results size are therefore in agreement with genetic correla-
between leanness and reproduction traits. tions found between average daily gain traits and
litter size. In contrast to litter size, litter weight at birth had a
4. Conclusions