.4n1mal Science l&, 66: 349-355 ‘0 1998 ...
.4n1mal Science l&, 66: 349-355
‘0 1998 British Sacietv of Animal Sciewc
Estimates of genetic parameters for growth traits of Gobra cattle
M. Diop’ and L. D. Van VleckY
‘Department ofAnima Science, Uniuersity ofNebraska, Lizzcoln, NE 68583-0908, USA
WS Department ofAgriculture-Agricultural
Research Station, Clay Cenfer, NE 68933-0166, USA
Abstract
Estimates of (co)variance componenfs and genefic paramefers wcre obtained for birth (no. = 3909), weaning (no. =
3425), yearling (no. = 2763), and final weight (no. = 2142) for Gobra cattle at the Centre de Recherches
Zootechniques de Dahra (Senegal), using single trait animal models. Data were analysed by restricted maximum
likelihood. Four different animal models were fïtted for each trait. Mode1 1 corzsidered the animal as the only
random efiect. Mode1 2 included in addition to fhe additiae direct effect of fhe animal, the environmental eflect due
to fhe dam. Mode1 3 added the materna1 additive genetic effects and allowed a covariance befween fhe direct and
materna1 genefic eflecfs. Model 4 fitted both materna1 genefic and permanent environmental eflectç. Inclusion of
both types qf materna1 effects kenetic and environmental) provided a betterjït for birth and weaning weights than
models with one materna1 effect only. For yearling and final weights, the improvement was not significant.
Important materna1 effecfs were found for a11 traits. Estimates of direct heritabilities zoere substantially higher when
materna1 efiects were ignored. Estimates of direct ami materna1 heritabilities with mode1 4 were 0.07 (s.e. 0.03) and
0.04 (se. @02), 0.20 (se. 0.05) and 0.21 (se. 0.05), 0.24 (s.e. 0.07) and 0.21 (se. 0.06), and 0.14 (s.e. O$I6) and
0.26 (se. 0.06) for birfh, weaning, yearling and final weights, respectively. Correlations between direct and
materna1 genetic effects were negative for a11 traits, and large for weaning and yearling weights witlz estimates of
-0.61 (se. 0~33) and -0.50 (se. 0.31), respectively. There was a signifcant positive linear phenotypic trend for
weaning and yearling weights. Linear trends for additive direct and materna1 breeding values were net signifcant
for any trait except maternal breeding value for yearling weight.
Keywords: cattle, genefic parameters, Gobra, growth.
Introduction
Evaluation of breeding programmes in Senegal has
Senegal has two main breeds of cattle: the taurine
concluded that their impact on local herds has been
Ndama located in the sub-humid areas of the south
limited and that new breeding strategies need to be
and east and the zebu Gobra in the north and centre.
designed. The design of new breeding plans
The Gobra is a lyre-horned zebu raised primarily by
requires, first of all, knowledge of genetic parameters
Fulani herders. Attempts genetically to improve the
for the Gobra. Abassa (1984) and Sow et al. (1988)
local Gobra have considered within-breed selection
reported heritability estimates for body weight at
and upgrading with Indo-Pakistani
breeds (Sahiwal
different ages; however, only additive direct genetic
and Guzerat). Early evaluation of the upgrading
variantes were estimated in the analyses. The
programme resulted in the recommendation
that
importance of materna1 effects in beef production has
crossbreeding be put on hold in favour of selection
been widely reported (Koch, 1972; Baker, 1980;
within a local breed (Centre de Recherches
Robison, 1981; Meyer, 1992). Materna1 effects should
Zootechniques de Dahra, 1973). Selection of the
be taken into account in selection for beef cattle,
Gobra began in 1963 with a foundation herd
especially if there is an antagonistic relationship
established in 1952 with animais bought from local
between direct and materna1 genetic effects.
herds. Selection was described by Denis and
Estimates of genetic parameters for beef production
Thiongane (1974) and was basically mass selection
traits for different cattle breeds have been reviewed
on growth performance.
by Mohiuddin (1993) and Koots et al. (1994).
This paper presents estimates of genetic parameters
t To whom correspondence should be addressed.
for growth of Gobra cattle obtained from fitting four
349
350
Diop and Van Vleck
animal models, attempting to separate direct genetic,
Table 1 Characturistics of the data structure
materna1 genetic
a n d
materna1 permanent
environmental effects. Weights at birth, 6 months
Birth Weaning Yearling Final
(weaning), 12 months (yearling) and 18 months
No. of:
weight weight wcight weight
(final) were considered. Genetic estimates are needed
to design breeding programmes and genetic
Records
3909
3425
2736
2142
Animalst
4480
3881
3171
2506
evaluation systems for Gobra cattle.
Sires$
6 4
6 4
6 3
6 2
Material and methods
Dams$.
1341
1210
10hl
635
Progeny / sire
61.1
53.5
43.9
34.5
Records of weight at birth, 6, 12 and 18 months were
Progenyl dam
2.9
2-3
2.6
2.3
obtained from the Gobra herd at the Centre de
Means (kg)
24.9
108.4
158.1
202.4
Recherches Zootechniques de Dahra (Senegal).
s.d. (kg)
4.0
20.0
25.4
29.6
The production environment was described by Sow
t Animals in pedigrees.
et al. (1988). The climate is tropical dry with two
$ Numbers of sires or dams with progeny with records.
distinct seasons: a long dry season from October to
June and a short rainy season from July to
than five progeny were deleted from the analysis.
September. The mean annual rainfall from 1964 to
PROC MIXED of Statistical Analysis Systems
1988 was 360 mm. The mean annual temperature
Institute (SAS, 1992) was used to test the significance
was 28°C. The vegetation is described as savanna
of the fixed effects of month of birth, year of birth,
type dominated by Acacia sp. and annual grasses.
sex and parity in a mode1 with sire considered to be a
Annual biomass production is closely related to the
random effect. Table 1 presents the characteristics of
amount of rainfall the area receives each year.
the data for the different ages.
Natural pasture is the main source of food. The
Variante and covariance components and genetic
quant@ and quality of the pasture vary considerably
parameters were estimated using the MTDFREML
during the year. During the rainy season, pastures
program (Boldman ef al., 1993) by fitting four animal
are of good quality and abundant. With the dry
models.
season, the nutritive value of the forage decreases
and supplemental feeding with ground nut cake or
Mode1 1 was a simple animal mode1 with animal’s
cotton seed has to be provided especially to some
additive direct genetic effect as the only random
categories of animals (suckling cows and weaned
effect. Mode1 2 fitted in addition, the effect of the dam
calves).
as an uncorrelated random effect. Mode1 3 considered
materna1 genetic effects as the second random effect
Management of the herd has been described by Sow
and allowed for covariance between the direct and
et al. (1988). Breeding females were randomly
materna1 genetic effects. Mode1 4 included both
assigned to sires (30 to 50 cows per sire) for a
materna1 genetic and permanent environmental
breeding season from December to March. The cows
effects and allowed for a genetic covariance between
that were open 3 months after the breeding season
direct and materna1 genetic effects.
started were reassigned to a different bull. Over the
years the size of the herd has fluctuated around 300
The preliminary analysis of variante showed that
CO~S. Selection for both males and females was on
fixed effects of month and year of birth, sex and
weight at 6 months (weaning). Males were then
p a r i t y w e r e significant f o r a11 f o u r t r a i t s .
selected again on their 18-month weight and the best
Consequently, these effects were included in a11 four
10 males were submitted to a growth performance
models.
test before final selection was made for the two or
three bulls to be used. The data for the growth
The full general mode1 (mode1 4) was:
performance test were not available. Females were
selected as replacements
at 24 months. About 5% of
y=Xj3+Za+Mm+Wc+e
the males and 80% of the females selected after
weaning were used as replacements.
Culling of cows
where y is the N X 1 vector of records, p denotes the
was based on poor reproductive performance or
vector of fixed effects (levels of month of birth, year
poor growth performance of offspring.
of birth, sex and parity of the dam), X is the matrix
that associates p with y; a is the vector of breeding
Data for these analyses consisted of records of
values for direct genetic effects, Z is the matrix that
animals born from 1963 to 1989. Consistency checks
associates a with y; m is the vector of breeding values
were performed on identification of animals and
for materna1 genetic effects, M is the matrix that
their pedigrees. Records of progeny of sires with less
associates m with y; c is the vector of environmental
Genetic parameters for growth of Gobra cattle
351
effects contributed by dams to records of their
variante (oc) to the total phenotypic variante (oy).
progeny; W is the matrix that associates c with y; and
Total (k:) heritability, is defined as (Willham, 1972):
e is the vector of residual effects.
kf = (o; + 0.54 + 1.50,) /o;.
For mode1 4,
Estimation of (co)variance components was carried
E(y) = Xp and
out using the MTDFREML program. A simplex
algorithm is used to search for variante components
rai
to minimize the function, -2log likelihood (L).
Convergence was assumed when the variante of the
function values (-210g L) of the simplex was less than
10m9. For a11 models a restart was performed after a
first convergence to verify that convergence was not
where d is the number of dams and N is the number
at a local minimum.
of records, A is the numerator relationship matrix
among animals including parents without records, oa
Approximate standard errors of the estimates of the
is the additive direct genetic variante, on, is the
parameters were derived from the inverse of the
materna1 genetic variante, o,,~ is the additive direct
average information matrix (Johnson and Thompson,
and materna1 genetic covariance, a: is the materna1
1994) which is the asymptotic variante-covariance
permanent environmental variante, 0: is the residual
matrix of the estimates.
variante;
and 1 , 1, are identity matrices of
appropriate order, the number of dams and the
Deviations from the overall mean of the least-squares
number of animals with records respectively.
estimates of effects of year of birth were used to
estimate the phenotypic trends. The solutions for
Estimates of additive direct (hz) and materna1 (hi)
direct and materna1 genetic effects under mode1 4 for
heritabilities were calculated as ratios of estimates of
animals born within each year were averaged to
additive direct (0:) and materna1 genetic (oz*)
examine the genetic trends for the four traits over the
variantes, respectively to the phenotypic variante
period covered by the study.
(oc). The direct-materna] correlation (Tarn) was
computed as the ratio of the estimates of direct-
Results
materna1 covariance (o,,,) to the product of the
Birtk weigkt
square roots of estimates of 0; and 0;; and c2 is the
Estimates of variante components and genetic
ratio of the estimates of materna1 environmental
parameters for birth weight are presented in Table 2.
Table 2 Estimates of Ccohariance componenfs (kg*) and genetic parameferst (alzd standard errors) for birtk aild uleaning zueigkfs
Birth weight
Weaning weight
Mode1 1
Mode1 2
Mode1 3
Mgdel4
Mode1 1
Mode1 2
Mode1 3
Mode1 4
2 2
2.04
1.34
1.16
1.18
137.0
62.9
85.3
83.1
0,
0.97
0.63
173.2
83.3
y
-0.14
-0.15
45.7
-50.7
DC
1.01
0.59
104.0
63.4
14.05
13.64
13.99
13.71
276.9
240.9
134.5
228.0
kr
$
16.09 0.127
15.99
0.084
15.97
0.073
15.95
0.074
413.9
0.331
407.8
0.154
427.3 0.200
409.6
A.
0.029
0.026
0.029
0.029
0.037
0.035
0.050
0.203 0.052
k2,
0.061
0.039
0.405
0.205
s.e.
0.022
0.022
0.041
0.053
r,,,
-0.132
-0.174
AI.541
-0.608
s.e.
0.337
0.396
0.233
0.328
C2
0.063
0.037
0.255
0.155
se.
0.016
0.020
0.022
0.034
12:
0.127
0.084
0.090
0.079
0.331
0.155
0.172
0.119
IogL
-13.316
-8.790
-2.042
0.000
-99.223
-32.254
-11.066
0.000
t o,$ direct additive genetic variante; o$, materna1 additive genetic variante; oam, direct-materna1 additive genetic
covariance; o:, materna1 permanent environmental variante; or temporary environmental variante; $, phenotypic variante;
ha, direct heritability; k$, maternai heritability; r,,,, direct-materna] genetic correlation; c2 = ~:/OS; k?, total heritability; log L,
expressed as deviation from mode1 with highest value.
/
-.
,
352
Diop and Van Vleck
Mode1 1 which ignored materna1 effects resulted in a
year of age. Under mode1 4, the estimate of km was of
higher estimate for 11: (0.13 (se. 0.03) compared with
similar magnitude to the estimate of ka: 0.21 (s.e.
O.OS*(s.e. 0.03), 0.07 (se. 0.03) and 0.07 (s.e. 0.03) for
0.06) and 0.24 (s.e. 0.07), respectively. The estimate of
models 2, 3 and 4, respectively). Judged by the log L,
c* for yearling weight was less than the c2 for
fitting materna1 effects (models 2 or 3) resulted in
weaning weight, using the same model: 0.05 (s.e.
significantly better fits compared with mode1 1, with
0.03) u. 0.16 (s.e. 0.03), respectively. Inclusion of
estimates of c* and hi of 0.06 (s.e. 0.02) and 0.06 (s.e.
materna1 permanent environmental effects in the
0.02), respectively. With both a materna1 permanent
mode1 did not improve mode1 4 compared with
environmental and a materna1 genetic effect in the
mode1 3 as shown by the values of Iog L. The
mode1 (mode1 4), the estimates of both L? and II~
estimate o f r,,, was negative and large: -0.50 (s.e.
were reduced. The estimate of r,,,, was -0.17 (s.e.
0.31).
0.40).
Final weigkt
Weaning weigkt
The estimates of the genetic parameters for final
Estimates of the genetic parameters for weaning
weight are reported in Table 3. Even at this age,
weight given in Table 2, show that inclusion of a
materna1 genetic effects accounted for a significant
materna1 effect in the mode1 resulted in a large
proportion of the total variante for final weight. The
increase in log L (models 2 or 3 D. mode1 1) and a
materna1 genetic and permanent environmental
reduction in the estimate of ha from 0.33 to 0.15 (s.e.
variantes represented proportionately 0.16 (s.e. 0.06)
0.04) and 0.20 (s.e. 0.05), respectively for models 2
and 0.04 (s.e. 0.04) respectively of total variante. As
and 3. The estimate of c* was 0.26 (s.e. 0.02) and that
with yearling weight, inclusion of materna1
of k2, was 0.41 (s.e. 0.04). With mode1 3, the estimate
permanent environmental effect in addition to the
of r,, was -0.54 (s.e. 0.23). With mode1 4, the estimate
materna1 genetic effect did not improve the mode1 as
of hm was reduced by one half with an estimate of r,,
shown by the values of log L. The direct-materna]
of Xl.61 (s.e. 0.33) and an estimate of c* of 0.16 (s.e.
genetic correlation was -0.29 (s.e. 0.36). Estimates of
0.03). In mode1 4, the genetic and environmental
direct and materna1 heritabilities were 0.14 (s.e. 0.06)
components of the effect of the dam are pulled apart
and O-16 (s.e. 0.06), respectively.
and this resulted in a significantly better fit
compared with models 2 and 3 (P < 0.05). Under
Phenotypic trends for the four traits are presented in
mode1 4, estimates of direct and materna1 genetic
Figure 1. Linear trends for weaning and yearling
variantes were similar in magnitude. As for birth
weights were significant (P < 0.05) with 0.954 and
weight, the standard errors of the r,,, were large.
0.742 kg gain per year respectively.
Yearling weigh t
The trends for the additive direct and materna1
Results for yearling weight are summarized in Table
breeding values for the four traits analysed under
3. Materna1 genetic effects seem important even at 1
mode1 4 are illustrated in Figures 2 and 3. Except for
Table 3 Estirnafes of Ccobariance components (kg?) and genetic parameterst (and standard errors)for yarling undfinal wei@ts
Yearling weight
Final weight
Mode1 1
Mode1 2
Mode1 3
Mode1 4
Mode1 1
Mode1 2
Mode1 3
Mode1 4
02
209.9
133.5
159.5
158.6
242.1
135.3
122.6
124.7
0:
181.5
140.7
180.9
139.5
.<.
-83.4
-74.7
40.9
-37.6
OY
99.3
32.5
113.3
35.6
04
456.0
423.3
410.2
404.6
666.2
642.3
645.0
638.4
656.1
667.8
661.8
908.3
890.9
907.5
900.6
$
666.8
0.315
0.203
0.239
0.240
0.267
0.152
0.135
0.138
SP,.
0.042
0.044
0.065
0.065
0.049
0.047
0.054
0.055
hi
0.272
0.213
0.199
0.155
s.e.
0.045
0.057
0.048
0.060
Tarn
-0.490
-0.500
-0.275
-0.285
s.e.
0.278
0.314
0.326
0.362
C2
0.151
0.049
0.127
0.040
s.e.
0.024
0.033
0.029
0.040
h:
Il.335
0.203
0.188
0.177
0.267
0.152
0.167
0.153
log L
-37.621
-26.946
--la92
0.000
-14.971
-8.916
-0.390
0.000
t See Table 2 footnote.
Genetic parameters for growth of Gobra cattle
353
Discussion
Except for birth weight, estimates of h$, were as large
as or larger than the estimates of hz. This suggests
that matemal effects need to be considered in
selecting for growth in Gobra cattle.
For weaning weight, the estimates are consistent with
the values found with Herefords (Meyer, 1992), with
Nelore cattle (Eler et al., 1995), with zebu crosses
(Mackinnon et al., 1991), with Senepol (Wright et al.,
1991) and with Wakwa and Gudali cattle (Tawah et
6 3 6 5 6 7 6 9 7 1 7 3 7 5 7 7 7 9 8 1 8 3 8 5 8 7 8 9
al., 1993). However, for Mashona cattle (Khombe et
Years
RI., 1995), hz at weaning was estimated to be larger
than h:,.
Figure 1 Phenotypic trend for birth (+), weaning (Lt),
yearling (A) and final (0) weights.
For yearling and final weights, the estimates of ha
were consistent with most published estimates, but
estimates of hi were not. The high estimates of hm
were somehow surprising because at those ages,
materna1 effects are expected to fade out because the
animals no longer depend on their mother. Their
weights should reflect the direct genetic effects to
that age with only carry-over materna1 effects from
before weaning. Relatively high estimates of kn, were
also found by Eler et al. (1995) at yearling and
Mackinnon ef al. (1991) at later ages. This may be
explained by the fact that for animals raised on
pasture with little or no supplementary feeding, the
-4J
I
length of time between weaning and yearling may
6 3 6 5 6 7 6 9 71
7 3 7 5 7 7 7 9 81 8 3 8 5 8 7 8 9
not be enough to buffer materna1 effects existing at
Years
weaning (Eler et al., 1995). This explanation is
probably true for the situation here where calves are
Figure 2 Direct additive genetic trend for birth (+),
weaned in the dry season and often lose weight
weaning (Cl), yearling (A) and final (‘3) weights.
before the next rainy season
The correlations between the direct and materna1
genetic effects were negative for a11 traits and high for
weaning and yearling weights although not
significantly different from zero. Similar high
negative estimates of ram were reported for weaning
and yearling weights by Tawah et al. (1993), Meyer
(1992) with zebu crosses, and Wright et al. (1991). For
final weight, the estimate of r,, is comparable to
estimates found in Herefords (Meyer, 1992); but
different from the near-zero estimates reported by
I
Mackinnon et al. (1991) for zebu crosses.
6 3 6 5 6 7 6 9 7 1
7 3 7.5 7 7 7 9 8 1 8 3 8 5 8 7 8 9
These negative correlations between direct and
Years
materna1 genetic effects suggest that many of the
genes which favour the milking and mothering
Figure 3 Materna1 genetic trend for birth (+), weaning (Lt),
ability of the cow are partly detrimental for growth of
yearling (A) and final ((3) weights.
the Young calf (Mohiuddin, 1993). Koch (1972)
suggested that the negative correlation could be due
materna1 breeding values for yearling weight, no
to a negative direct influence of the dams on the
linear trends were significant (P > 0.05). However
materna1 ability of their female offspring through
overall breeding values (a + m) for final weight
overfeeding. These negative correlations may also be
showed a significant (P < 0.05), though small,
the result of an adaptation of the animals to the dry
positive linear trend of 0.097 kg/year.
tropical environment where food resources are scarce
354
Diop and Van Vleck
(Tawah et al., 1993). Small-size cows tend to meet
phenotypic values, improvement would be difficult
nutritional requirements for their maintenance and
to achieve (Willham, 1972).
growth of their calves more easily than larger-size
CO~S. The latter Will therefore produce smaller calves
It is also possible that evaluation of the bulls was not
especially at weaning than those of small-size cows
properly conducted. In an environment such as the
of similar ages. It might also be argued that under
one under study here, the weight an animal has at a
favourable conditions, the direct and materna1
particular age depends greatly on when during the
genetic correlation should be more nearly zero.
year that age is reached. Due to seasonal variation in
the amount and quality of pasture, animals that
Estimates of the ratio of permanent environmental
reach a particular age at the end of the dry season,
variante to the phenotypic variante, c*, for the four
are more likely to weigh less than animals that reach
traits were in agreement with the estimates by Eler et
the same age at the end of the rainy season. Even
al. (1995) and Meyer (1992) with Herefords and zebu
though supplemental feeding was provided during
crosses. The c2 was larger for weaning weight than
the dry season, the quantities were usually limited to
for the other traits. The permanent environmental
those needed to maintain and not to lose weight,
effects are due to incidents that affect a11 records of
because an attempt was made to keep the
progeny of the same cow. Before weaning, the effects
environment at the station similar to that of the local
may be due to sequels of diseases or accidents to the
farms. The possibility of genotype X environment
udder, which Will affect the milk production of the
interactions had to be minimized since bulls from the
cow. After weaning, the calf is no longer dependent
station were to be made available to local farmers.
on the milk of the dam and the estimate of c* for
Among the bulls used and born from 1963 to 1985,24
yearling and final weights is probably due to the
out of 46 or 52% were born in the months of June and
carry-over effects on weaning weight.
July when only 28% of the calvings occur. The month
of birth may have had a significant influence on
Except for weaning weight, estimates of total
selection at 18 months of age.
heritability, h:, were of similar magnitude to the
estimates of ha.
implications
There is no definitive agreement in the literature on
Standard errors for the estimates of direct and
the age at which materna1 effects become no longer
materna1 heritabilities were small; however large
important in beef cattle. This study of Gobra cattle
standard errors were found for estimates of r,,,.
showed that materna1 effects remained important at
Large negative estimates of r,, may result in
1 year and at 18 months of age. Selection at ages
inaccuracy of estimates of genetic parameters
when materna1 effects are not important would not
especially when materna1 genetic variante is the
be efficient because of the reduction in the rates of
same or smaller than direct genetic variante
response due to the lengthening of the generation
(Gerstmayr, 1992). Under mode1 4, the estimate of oz,
interval. Selection at weaning and yearling ages
was of the same size (weaning weight) or
based only on direct breeding values may not yield
smaller than the estimate of 0: for the different
optimal response because of the negative genetic
traits.
correlation between direct and materna1 effects (Van
Vleck et al., 1977). Future selection plans need to
Linear genetic trends in additive direct breeding
consider materna1 effects in order to optimize
values were not significant for a11 traits. The estimate
expected total response over the long run.
of hz under mode1 4 was low for birth and final
weights but moderate for weaning and yearling
Appropriate adjustments for factors such as month
weights. The expected genetic gain calculated using
or season of birth are also needed for proper
11: under mode1 4, the average selection intensity of
identification of the best animals to be selected. The
1.205 and the average generation interval of 7.5 years
use of mixed mode1 procedures
to obtain best linear
were 0.050, 0.382, 0.723 and 0.727 kg/ year for birth,
unbiased prediction breeding values should provide
weaning, yearling and final weight, respectively. The
a better way for evaluating and selecting animals
actual overall genetic gains (a + m) were 0.003,
than has been done previously.
-0.013, 0.097 and 0.097 kg/year. The latter two are
significantly different from those expected. The
Acknowledgements
length of the selection programme (26 years) which
The Senegaleese Institute of Agricultural Research (ISRA) is
corresponds to three to four generations may not be
thanked for providing the data and the US Agency for
enough to detect a significant change in genetic trend
International &Develop&ent
for funding the fell&wship of
(Khombe et al., 1995). It may also be argued that with
the senior author. Published as paper no. 11575, Journal
the negative correlation between direct and materna1
Series, Nebraska Agricultural Reséaich Division, Univers@
genetic effects and with selection solely based on
of Nebraska.
c
ti
Genetic parameters for growth of Gobra cattle
355
Ref erences
beef production traits. 1. Hcritability. A~~irrinl Rrerdiucy
Abassa, P. K. 1984. Systems approach to Gobra Zebu
Absfracts 62: 309-338.
production in Dahra, Senegal. PhD. dissertation, Univcrsity
Mackinnon, M. J., Meyer, K. and Hetzel, J. D. S. 1991.
ofFIorida, Gainesville, FL.
Genetic variation and covariation for growth, parasite
Baker, R. L. 1980. The role of materna1 effects on the
resistance and heat tolerance in tropical cattle. Lir~esfcçk
efficiency of selection in beef cattle. A review. Proceedings of
Production Science 27: 105-122.
fhe New Zealand Sociefy of Animal Production 40: 285-303.
Meyer, K. 1992. Variante components due to direct and
Boldman, K. G., Kriese, L. D., Van Vleck, L. D. and
materna1 effects for growth traits of Australian beef cattle.
Kachman, S. D. 1993. A manualfir use ofMTDFREML. A sef
Livestock Production Science 31: 179-204.
of programs to obtain estima& of oariances and couariances.
Mohiuddin, G. 1993. Estimates of genetic and phenotypic
ARS, USDA, Washington, DC.
parameters of some performance traits in beef cattle. Animal
Centre de Recherches Zootechniques de Dahra. 1973.
Breeding Absfracts 61: 495-522.
Rapport annuall973.
Robison, 0. W. 1981. The influence of materna1 effects on
Denis, J. P. and Thiongane, A. 1. 1974. L’aptitude a la
the efficiency of selection; a review. Livesfock Prodrrcfion
production de viande chez le zebu Cobra du Senegal.
Science 5: 121-137.
Proceedings of fhe jïrsf world congress on genefics applied to
Sow, R. S., Denis, J. P., Trail, J. M. C., Thiongane, P. I.,
animal production, Madrid, vol. 3, pp. 889-897.
Mbaye, M. and Diallo, 1.1988. Productivite du Zebu Gobra
Eler, J. P., Van Vleck, L. D., Ferraz, J. B. S. and Lôbo, R. B.
au Centre de Recherches Zootechniques de Dahra
1995. Estimation of variantes due to direct and materna1
(Senegal). ISRA Etudes et Documents, vol. 2, no. 2.
effects: for growth traits of Nelor cattle. /ournal oj Animal
Statistical Analysis Systems Institute. 1992. SAS/STAT
Science 73: 3253-3258.
release 6.08. SAS Institute lnc., Cary, NC.
Gerstmayr, S. 1992. Impact of the data structure on the
Tawah, C. L., Mbah, D. A., Rege, J. E. 0. and Oumate, H.
reliability of the estimated genetic parameters in an animal
1993. Genetic evaluation of birth and weaning weight of
mode1 with materna1 effects. journal oj Animal Breeding and
Gudali and two-breed synthetic Wakwa bref cattle
Genetics 109: 321-336.
populations under selection in Cameroon: genetic and
Johnson, D. L. and Thompson, R. 1994. Restricted
phenotypic parameters. Animal Production 57: 73-79.
maximum likelihood estimation of variante components for
Van Vleck, L. D., David, St L. and Miller, J. 1. 1977.
univariate animal models using sparse matrix ‘techniques
Expected phenotypic response in weaning weight of beef
and a auasi-Newton vrocedure. Proceedinps of the fifth world
calves from selection for direct and materna1 genetic effects.
congres; on genetics applied fo livesfock prodkfib, Guelph, vol.
Journal of Animal Science 44: 360-367.
18, p. 410-413.
Willham, R. L. 1972. The role of materna1 effects in animal
Khombe, C. T., Hayes, J. F., Cue, R. 1. and Wade, K. M.
breeding. III. Biometrical aspects of materna1 effects in
1995. Estimation of direct additive and materna1 additive
animals. lournat of Animal Science 35: 1288-1293.
genetic effects for weaning weight in Mashona cattle of
Zimbabwe using an individual animal model. Animal
Wright, D. W., Johnson, 2. B., Brown, C. J. and Wildeus, S.
Production 60: 41-48.
1991. Variante and covariance estimates for weaning
weight of Senepol cattle. Journal of Animaf Science 69:
Koch, R. M. 1972. The role of materna1 effects in animal
3945-3951.
breeding. VI. Materna1 effects in beef cattle. Journal of
Animal Science 35: 1316-1323.
Koots, K. R, Gibson, J. P, Smith, C. and Wilton, J. W.
1994. Analyses of published genetic parameter estimates for
(Received 29 April 1997-Acc@cd 31 Orfober 1997)
--
^_
.-. -_
‘0 1998 British Sacietv of Animal Sciewc
Estimates of genetic parameters for growth traits of Gobra cattle
M. Diop’ and L. D. Van VleckY
‘Department ofAnima Science, Uniuersity ofNebraska, Lizzcoln, NE 68583-0908, USA
WS Department ofAgriculture-Agricultural
Research Station, Clay Cenfer, NE 68933-0166, USA
Abstract
Estimates of (co)variance componenfs and genefic paramefers wcre obtained for birth (no. = 3909), weaning (no. =
3425), yearling (no. = 2763), and final weight (no. = 2142) for Gobra cattle at the Centre de Recherches
Zootechniques de Dahra (Senegal), using single trait animal models. Data were analysed by restricted maximum
likelihood. Four different animal models were fïtted for each trait. Mode1 1 corzsidered the animal as the only
random efiect. Mode1 2 included in addition to fhe additiae direct effect of fhe animal, the environmental eflect due
to fhe dam. Mode1 3 added the materna1 additive genetic effects and allowed a covariance befween fhe direct and
materna1 genefic eflecfs. Model 4 fitted both materna1 genefic and permanent environmental eflectç. Inclusion of
both types qf materna1 effects kenetic and environmental) provided a betterjït for birth and weaning weights than
models with one materna1 effect only. For yearling and final weights, the improvement was not significant.
Important materna1 effecfs were found for a11 traits. Estimates of direct heritabilities zoere substantially higher when
materna1 efiects were ignored. Estimates of direct ami materna1 heritabilities with mode1 4 were 0.07 (s.e. 0.03) and
0.04 (se. @02), 0.20 (se. 0.05) and 0.21 (se. 0.05), 0.24 (s.e. 0.07) and 0.21 (se. 0.06), and 0.14 (s.e. O$I6) and
0.26 (se. 0.06) for birfh, weaning, yearling and final weights, respectively. Correlations between direct and
materna1 genetic effects were negative for a11 traits, and large for weaning and yearling weights witlz estimates of
-0.61 (se. 0~33) and -0.50 (se. 0.31), respectively. There was a signifcant positive linear phenotypic trend for
weaning and yearling weights. Linear trends for additive direct and materna1 breeding values were net signifcant
for any trait except maternal breeding value for yearling weight.
Keywords: cattle, genefic parameters, Gobra, growth.
Introduction
Evaluation of breeding programmes in Senegal has
Senegal has two main breeds of cattle: the taurine
concluded that their impact on local herds has been
Ndama located in the sub-humid areas of the south
limited and that new breeding strategies need to be
and east and the zebu Gobra in the north and centre.
designed. The design of new breeding plans
The Gobra is a lyre-horned zebu raised primarily by
requires, first of all, knowledge of genetic parameters
Fulani herders. Attempts genetically to improve the
for the Gobra. Abassa (1984) and Sow et al. (1988)
local Gobra have considered within-breed selection
reported heritability estimates for body weight at
and upgrading with Indo-Pakistani
breeds (Sahiwal
different ages; however, only additive direct genetic
and Guzerat). Early evaluation of the upgrading
variantes were estimated in the analyses. The
programme resulted in the recommendation
that
importance of materna1 effects in beef production has
crossbreeding be put on hold in favour of selection
been widely reported (Koch, 1972; Baker, 1980;
within a local breed (Centre de Recherches
Robison, 1981; Meyer, 1992). Materna1 effects should
Zootechniques de Dahra, 1973). Selection of the
be taken into account in selection for beef cattle,
Gobra began in 1963 with a foundation herd
especially if there is an antagonistic relationship
established in 1952 with animais bought from local
between direct and materna1 genetic effects.
herds. Selection was described by Denis and
Estimates of genetic parameters for beef production
Thiongane (1974) and was basically mass selection
traits for different cattle breeds have been reviewed
on growth performance.
by Mohiuddin (1993) and Koots et al. (1994).
This paper presents estimates of genetic parameters
t To whom correspondence should be addressed.
for growth of Gobra cattle obtained from fitting four
349
350
Diop and Van Vleck
animal models, attempting to separate direct genetic,
Table 1 Characturistics of the data structure
materna1 genetic
a n d
materna1 permanent
environmental effects. Weights at birth, 6 months
Birth Weaning Yearling Final
(weaning), 12 months (yearling) and 18 months
No. of:
weight weight wcight weight
(final) were considered. Genetic estimates are needed
to design breeding programmes and genetic
Records
3909
3425
2736
2142
Animalst
4480
3881
3171
2506
evaluation systems for Gobra cattle.
Sires$
6 4
6 4
6 3
6 2
Material and methods
Dams$.
1341
1210
10hl
635
Progeny / sire
61.1
53.5
43.9
34.5
Records of weight at birth, 6, 12 and 18 months were
Progenyl dam
2.9
2-3
2.6
2.3
obtained from the Gobra herd at the Centre de
Means (kg)
24.9
108.4
158.1
202.4
Recherches Zootechniques de Dahra (Senegal).
s.d. (kg)
4.0
20.0
25.4
29.6
The production environment was described by Sow
t Animals in pedigrees.
et al. (1988). The climate is tropical dry with two
$ Numbers of sires or dams with progeny with records.
distinct seasons: a long dry season from October to
June and a short rainy season from July to
than five progeny were deleted from the analysis.
September. The mean annual rainfall from 1964 to
PROC MIXED of Statistical Analysis Systems
1988 was 360 mm. The mean annual temperature
Institute (SAS, 1992) was used to test the significance
was 28°C. The vegetation is described as savanna
of the fixed effects of month of birth, year of birth,
type dominated by Acacia sp. and annual grasses.
sex and parity in a mode1 with sire considered to be a
Annual biomass production is closely related to the
random effect. Table 1 presents the characteristics of
amount of rainfall the area receives each year.
the data for the different ages.
Natural pasture is the main source of food. The
Variante and covariance components and genetic
quant@ and quality of the pasture vary considerably
parameters were estimated using the MTDFREML
during the year. During the rainy season, pastures
program (Boldman ef al., 1993) by fitting four animal
are of good quality and abundant. With the dry
models.
season, the nutritive value of the forage decreases
and supplemental feeding with ground nut cake or
Mode1 1 was a simple animal mode1 with animal’s
cotton seed has to be provided especially to some
additive direct genetic effect as the only random
categories of animals (suckling cows and weaned
effect. Mode1 2 fitted in addition, the effect of the dam
calves).
as an uncorrelated random effect. Mode1 3 considered
materna1 genetic effects as the second random effect
Management of the herd has been described by Sow
and allowed for covariance between the direct and
et al. (1988). Breeding females were randomly
materna1 genetic effects. Mode1 4 included both
assigned to sires (30 to 50 cows per sire) for a
materna1 genetic and permanent environmental
breeding season from December to March. The cows
effects and allowed for a genetic covariance between
that were open 3 months after the breeding season
direct and materna1 genetic effects.
started were reassigned to a different bull. Over the
years the size of the herd has fluctuated around 300
The preliminary analysis of variante showed that
CO~S. Selection for both males and females was on
fixed effects of month and year of birth, sex and
weight at 6 months (weaning). Males were then
p a r i t y w e r e significant f o r a11 f o u r t r a i t s .
selected again on their 18-month weight and the best
Consequently, these effects were included in a11 four
10 males were submitted to a growth performance
models.
test before final selection was made for the two or
three bulls to be used. The data for the growth
The full general mode1 (mode1 4) was:
performance test were not available. Females were
selected as replacements
at 24 months. About 5% of
y=Xj3+Za+Mm+Wc+e
the males and 80% of the females selected after
weaning were used as replacements.
Culling of cows
where y is the N X 1 vector of records, p denotes the
was based on poor reproductive performance or
vector of fixed effects (levels of month of birth, year
poor growth performance of offspring.
of birth, sex and parity of the dam), X is the matrix
that associates p with y; a is the vector of breeding
Data for these analyses consisted of records of
values for direct genetic effects, Z is the matrix that
animals born from 1963 to 1989. Consistency checks
associates a with y; m is the vector of breeding values
were performed on identification of animals and
for materna1 genetic effects, M is the matrix that
their pedigrees. Records of progeny of sires with less
associates m with y; c is the vector of environmental
Genetic parameters for growth of Gobra cattle
351
effects contributed by dams to records of their
variante (oc) to the total phenotypic variante (oy).
progeny; W is the matrix that associates c with y; and
Total (k:) heritability, is defined as (Willham, 1972):
e is the vector of residual effects.
kf = (o; + 0.54 + 1.50,) /o;.
For mode1 4,
Estimation of (co)variance components was carried
E(y) = Xp and
out using the MTDFREML program. A simplex
algorithm is used to search for variante components
rai
to minimize the function, -2log likelihood (L).
Convergence was assumed when the variante of the
function values (-210g L) of the simplex was less than
10m9. For a11 models a restart was performed after a
first convergence to verify that convergence was not
where d is the number of dams and N is the number
at a local minimum.
of records, A is the numerator relationship matrix
among animals including parents without records, oa
Approximate standard errors of the estimates of the
is the additive direct genetic variante, on, is the
parameters were derived from the inverse of the
materna1 genetic variante, o,,~ is the additive direct
average information matrix (Johnson and Thompson,
and materna1 genetic covariance, a: is the materna1
1994) which is the asymptotic variante-covariance
permanent environmental variante, 0: is the residual
matrix of the estimates.
variante;
and 1 , 1, are identity matrices of
appropriate order, the number of dams and the
Deviations from the overall mean of the least-squares
number of animals with records respectively.
estimates of effects of year of birth were used to
estimate the phenotypic trends. The solutions for
Estimates of additive direct (hz) and materna1 (hi)
direct and materna1 genetic effects under mode1 4 for
heritabilities were calculated as ratios of estimates of
animals born within each year were averaged to
additive direct (0:) and materna1 genetic (oz*)
examine the genetic trends for the four traits over the
variantes, respectively to the phenotypic variante
period covered by the study.
(oc). The direct-materna] correlation (Tarn) was
computed as the ratio of the estimates of direct-
Results
materna1 covariance (o,,,) to the product of the
Birtk weigkt
square roots of estimates of 0; and 0;; and c2 is the
Estimates of variante components and genetic
ratio of the estimates of materna1 environmental
parameters for birth weight are presented in Table 2.
Table 2 Estimates of Ccohariance componenfs (kg*) and genetic parameferst (alzd standard errors) for birtk aild uleaning zueigkfs
Birth weight
Weaning weight
Mode1 1
Mode1 2
Mode1 3
Mgdel4
Mode1 1
Mode1 2
Mode1 3
Mode1 4
2 2
2.04
1.34
1.16
1.18
137.0
62.9
85.3
83.1
0,
0.97
0.63
173.2
83.3
y
-0.14
-0.15
45.7
-50.7
DC
1.01
0.59
104.0
63.4
14.05
13.64
13.99
13.71
276.9
240.9
134.5
228.0
kr
$
16.09 0.127
15.99
0.084
15.97
0.073
15.95
0.074
413.9
0.331
407.8
0.154
427.3 0.200
409.6
A.
0.029
0.026
0.029
0.029
0.037
0.035
0.050
0.203 0.052
k2,
0.061
0.039
0.405
0.205
s.e.
0.022
0.022
0.041
0.053
r,,,
-0.132
-0.174
AI.541
-0.608
s.e.
0.337
0.396
0.233
0.328
C2
0.063
0.037
0.255
0.155
se.
0.016
0.020
0.022
0.034
12:
0.127
0.084
0.090
0.079
0.331
0.155
0.172
0.119
IogL
-13.316
-8.790
-2.042
0.000
-99.223
-32.254
-11.066
0.000
t o,$ direct additive genetic variante; o$, materna1 additive genetic variante; oam, direct-materna1 additive genetic
covariance; o:, materna1 permanent environmental variante; or temporary environmental variante; $, phenotypic variante;
ha, direct heritability; k$, maternai heritability; r,,,, direct-materna] genetic correlation; c2 = ~:/OS; k?, total heritability; log L,
expressed as deviation from mode1 with highest value.
/
-.
,
352
Diop and Van Vleck
Mode1 1 which ignored materna1 effects resulted in a
year of age. Under mode1 4, the estimate of km was of
higher estimate for 11: (0.13 (se. 0.03) compared with
similar magnitude to the estimate of ka: 0.21 (s.e.
O.OS*(s.e. 0.03), 0.07 (se. 0.03) and 0.07 (s.e. 0.03) for
0.06) and 0.24 (s.e. 0.07), respectively. The estimate of
models 2, 3 and 4, respectively). Judged by the log L,
c* for yearling weight was less than the c2 for
fitting materna1 effects (models 2 or 3) resulted in
weaning weight, using the same model: 0.05 (s.e.
significantly better fits compared with mode1 1, with
0.03) u. 0.16 (s.e. 0.03), respectively. Inclusion of
estimates of c* and hi of 0.06 (s.e. 0.02) and 0.06 (s.e.
materna1 permanent environmental effects in the
0.02), respectively. With both a materna1 permanent
mode1 did not improve mode1 4 compared with
environmental and a materna1 genetic effect in the
mode1 3 as shown by the values of Iog L. The
mode1 (mode1 4), the estimates of both L? and II~
estimate o f r,,, was negative and large: -0.50 (s.e.
were reduced. The estimate of r,,,, was -0.17 (s.e.
0.31).
0.40).
Final weigkt
Weaning weigkt
The estimates of the genetic parameters for final
Estimates of the genetic parameters for weaning
weight are reported in Table 3. Even at this age,
weight given in Table 2, show that inclusion of a
materna1 genetic effects accounted for a significant
materna1 effect in the mode1 resulted in a large
proportion of the total variante for final weight. The
increase in log L (models 2 or 3 D. mode1 1) and a
materna1 genetic and permanent environmental
reduction in the estimate of ha from 0.33 to 0.15 (s.e.
variantes represented proportionately 0.16 (s.e. 0.06)
0.04) and 0.20 (s.e. 0.05), respectively for models 2
and 0.04 (s.e. 0.04) respectively of total variante. As
and 3. The estimate of c* was 0.26 (s.e. 0.02) and that
with yearling weight, inclusion of materna1
of k2, was 0.41 (s.e. 0.04). With mode1 3, the estimate
permanent environmental effect in addition to the
of r,, was -0.54 (s.e. 0.23). With mode1 4, the estimate
materna1 genetic effect did not improve the mode1 as
of hm was reduced by one half with an estimate of r,,
shown by the values of log L. The direct-materna]
of Xl.61 (s.e. 0.33) and an estimate of c* of 0.16 (s.e.
genetic correlation was -0.29 (s.e. 0.36). Estimates of
0.03). In mode1 4, the genetic and environmental
direct and materna1 heritabilities were 0.14 (s.e. 0.06)
components of the effect of the dam are pulled apart
and O-16 (s.e. 0.06), respectively.
and this resulted in a significantly better fit
compared with models 2 and 3 (P < 0.05). Under
Phenotypic trends for the four traits are presented in
mode1 4, estimates of direct and materna1 genetic
Figure 1. Linear trends for weaning and yearling
variantes were similar in magnitude. As for birth
weights were significant (P < 0.05) with 0.954 and
weight, the standard errors of the r,,, were large.
0.742 kg gain per year respectively.
Yearling weigh t
The trends for the additive direct and materna1
Results for yearling weight are summarized in Table
breeding values for the four traits analysed under
3. Materna1 genetic effects seem important even at 1
mode1 4 are illustrated in Figures 2 and 3. Except for
Table 3 Estirnafes of Ccobariance components (kg?) and genetic parameterst (and standard errors)for yarling undfinal wei@ts
Yearling weight
Final weight
Mode1 1
Mode1 2
Mode1 3
Mode1 4
Mode1 1
Mode1 2
Mode1 3
Mode1 4
02
209.9
133.5
159.5
158.6
242.1
135.3
122.6
124.7
0:
181.5
140.7
180.9
139.5
.<.
-83.4
-74.7
40.9
-37.6
OY
99.3
32.5
113.3
35.6
04
456.0
423.3
410.2
404.6
666.2
642.3
645.0
638.4
656.1
667.8
661.8
908.3
890.9
907.5
900.6
$
666.8
0.315
0.203
0.239
0.240
0.267
0.152
0.135
0.138
SP,.
0.042
0.044
0.065
0.065
0.049
0.047
0.054
0.055
hi
0.272
0.213
0.199
0.155
s.e.
0.045
0.057
0.048
0.060
Tarn
-0.490
-0.500
-0.275
-0.285
s.e.
0.278
0.314
0.326
0.362
C2
0.151
0.049
0.127
0.040
s.e.
0.024
0.033
0.029
0.040
h:
Il.335
0.203
0.188
0.177
0.267
0.152
0.167
0.153
log L
-37.621
-26.946
--la92
0.000
-14.971
-8.916
-0.390
0.000
t See Table 2 footnote.
Genetic parameters for growth of Gobra cattle
353
Discussion
Except for birth weight, estimates of h$, were as large
as or larger than the estimates of hz. This suggests
that matemal effects need to be considered in
selecting for growth in Gobra cattle.
For weaning weight, the estimates are consistent with
the values found with Herefords (Meyer, 1992), with
Nelore cattle (Eler et al., 1995), with zebu crosses
(Mackinnon et al., 1991), with Senepol (Wright et al.,
1991) and with Wakwa and Gudali cattle (Tawah et
6 3 6 5 6 7 6 9 7 1 7 3 7 5 7 7 7 9 8 1 8 3 8 5 8 7 8 9
al., 1993). However, for Mashona cattle (Khombe et
Years
RI., 1995), hz at weaning was estimated to be larger
than h:,.
Figure 1 Phenotypic trend for birth (+), weaning (Lt),
yearling (A) and final (0) weights.
For yearling and final weights, the estimates of ha
were consistent with most published estimates, but
estimates of hi were not. The high estimates of hm
were somehow surprising because at those ages,
materna1 effects are expected to fade out because the
animals no longer depend on their mother. Their
weights should reflect the direct genetic effects to
that age with only carry-over materna1 effects from
before weaning. Relatively high estimates of kn, were
also found by Eler et al. (1995) at yearling and
Mackinnon ef al. (1991) at later ages. This may be
explained by the fact that for animals raised on
pasture with little or no supplementary feeding, the
-4J
I
length of time between weaning and yearling may
6 3 6 5 6 7 6 9 71
7 3 7 5 7 7 7 9 81 8 3 8 5 8 7 8 9
not be enough to buffer materna1 effects existing at
Years
weaning (Eler et al., 1995). This explanation is
probably true for the situation here where calves are
Figure 2 Direct additive genetic trend for birth (+),
weaned in the dry season and often lose weight
weaning (Cl), yearling (A) and final (‘3) weights.
before the next rainy season
The correlations between the direct and materna1
genetic effects were negative for a11 traits and high for
weaning and yearling weights although not
significantly different from zero. Similar high
negative estimates of ram were reported for weaning
and yearling weights by Tawah et al. (1993), Meyer
(1992) with zebu crosses, and Wright et al. (1991). For
final weight, the estimate of r,, is comparable to
estimates found in Herefords (Meyer, 1992); but
different from the near-zero estimates reported by
I
Mackinnon et al. (1991) for zebu crosses.
6 3 6 5 6 7 6 9 7 1
7 3 7.5 7 7 7 9 8 1 8 3 8 5 8 7 8 9
These negative correlations between direct and
Years
materna1 genetic effects suggest that many of the
genes which favour the milking and mothering
Figure 3 Materna1 genetic trend for birth (+), weaning (Lt),
ability of the cow are partly detrimental for growth of
yearling (A) and final ((3) weights.
the Young calf (Mohiuddin, 1993). Koch (1972)
suggested that the negative correlation could be due
materna1 breeding values for yearling weight, no
to a negative direct influence of the dams on the
linear trends were significant (P > 0.05). However
materna1 ability of their female offspring through
overall breeding values (a + m) for final weight
overfeeding. These negative correlations may also be
showed a significant (P < 0.05), though small,
the result of an adaptation of the animals to the dry
positive linear trend of 0.097 kg/year.
tropical environment where food resources are scarce
354
Diop and Van Vleck
(Tawah et al., 1993). Small-size cows tend to meet
phenotypic values, improvement would be difficult
nutritional requirements for their maintenance and
to achieve (Willham, 1972).
growth of their calves more easily than larger-size
CO~S. The latter Will therefore produce smaller calves
It is also possible that evaluation of the bulls was not
especially at weaning than those of small-size cows
properly conducted. In an environment such as the
of similar ages. It might also be argued that under
one under study here, the weight an animal has at a
favourable conditions, the direct and materna1
particular age depends greatly on when during the
genetic correlation should be more nearly zero.
year that age is reached. Due to seasonal variation in
the amount and quality of pasture, animals that
Estimates of the ratio of permanent environmental
reach a particular age at the end of the dry season,
variante to the phenotypic variante, c*, for the four
are more likely to weigh less than animals that reach
traits were in agreement with the estimates by Eler et
the same age at the end of the rainy season. Even
al. (1995) and Meyer (1992) with Herefords and zebu
though supplemental feeding was provided during
crosses. The c2 was larger for weaning weight than
the dry season, the quantities were usually limited to
for the other traits. The permanent environmental
those needed to maintain and not to lose weight,
effects are due to incidents that affect a11 records of
because an attempt was made to keep the
progeny of the same cow. Before weaning, the effects
environment at the station similar to that of the local
may be due to sequels of diseases or accidents to the
farms. The possibility of genotype X environment
udder, which Will affect the milk production of the
interactions had to be minimized since bulls from the
cow. After weaning, the calf is no longer dependent
station were to be made available to local farmers.
on the milk of the dam and the estimate of c* for
Among the bulls used and born from 1963 to 1985,24
yearling and final weights is probably due to the
out of 46 or 52% were born in the months of June and
carry-over effects on weaning weight.
July when only 28% of the calvings occur. The month
of birth may have had a significant influence on
Except for weaning weight, estimates of total
selection at 18 months of age.
heritability, h:, were of similar magnitude to the
estimates of ha.
implications
There is no definitive agreement in the literature on
Standard errors for the estimates of direct and
the age at which materna1 effects become no longer
materna1 heritabilities were small; however large
important in beef cattle. This study of Gobra cattle
standard errors were found for estimates of r,,,.
showed that materna1 effects remained important at
Large negative estimates of r,, may result in
1 year and at 18 months of age. Selection at ages
inaccuracy of estimates of genetic parameters
when materna1 effects are not important would not
especially when materna1 genetic variante is the
be efficient because of the reduction in the rates of
same or smaller than direct genetic variante
response due to the lengthening of the generation
(Gerstmayr, 1992). Under mode1 4, the estimate of oz,
interval. Selection at weaning and yearling ages
was of the same size (weaning weight) or
based only on direct breeding values may not yield
smaller than the estimate of 0: for the different
optimal response because of the negative genetic
traits.
correlation between direct and materna1 effects (Van
Vleck et al., 1977). Future selection plans need to
Linear genetic trends in additive direct breeding
consider materna1 effects in order to optimize
values were not significant for a11 traits. The estimate
expected total response over the long run.
of hz under mode1 4 was low for birth and final
weights but moderate for weaning and yearling
Appropriate adjustments for factors such as month
weights. The expected genetic gain calculated using
or season of birth are also needed for proper
11: under mode1 4, the average selection intensity of
identification of the best animals to be selected. The
1.205 and the average generation interval of 7.5 years
use of mixed mode1 procedures
to obtain best linear
were 0.050, 0.382, 0.723 and 0.727 kg/ year for birth,
unbiased prediction breeding values should provide
weaning, yearling and final weight, respectively. The
a better way for evaluating and selecting animals
actual overall genetic gains (a + m) were 0.003,
than has been done previously.
-0.013, 0.097 and 0.097 kg/year. The latter two are
significantly different from those expected. The
Acknowledgements
length of the selection programme (26 years) which
The Senegaleese Institute of Agricultural Research (ISRA) is
corresponds to three to four generations may not be
thanked for providing the data and the US Agency for
enough to detect a significant change in genetic trend
International &Develop&ent
for funding the fell&wship of
(Khombe et al., 1995). It may also be argued that with
the senior author. Published as paper no. 11575, Journal
the negative correlation between direct and materna1
Series, Nebraska Agricultural Reséaich Division, Univers@
genetic effects and with selection solely based on
of Nebraska.
c
ti
Genetic parameters for growth of Gobra cattle
355
Ref erences
beef production traits. 1. Hcritability. A~~irrinl Rrerdiucy
Abassa, P. K. 1984. Systems approach to Gobra Zebu
Absfracts 62: 309-338.
production in Dahra, Senegal. PhD. dissertation, Univcrsity
Mackinnon, M. J., Meyer, K. and Hetzel, J. D. S. 1991.
ofFIorida, Gainesville, FL.
Genetic variation and covariation for growth, parasite
Baker, R. L. 1980. The role of materna1 effects on the
resistance and heat tolerance in tropical cattle. Lir~esfcçk
efficiency of selection in beef cattle. A review. Proceedings of
Production Science 27: 105-122.
fhe New Zealand Sociefy of Animal Production 40: 285-303.
Meyer, K. 1992. Variante components due to direct and
Boldman, K. G., Kriese, L. D., Van Vleck, L. D. and
materna1 effects for growth traits of Australian beef cattle.
Kachman, S. D. 1993. A manualfir use ofMTDFREML. A sef
Livestock Production Science 31: 179-204.
of programs to obtain estima& of oariances and couariances.
Mohiuddin, G. 1993. Estimates of genetic and phenotypic
ARS, USDA, Washington, DC.
parameters of some performance traits in beef cattle. Animal
Centre de Recherches Zootechniques de Dahra. 1973.
Breeding Absfracts 61: 495-522.
Rapport annuall973.
Robison, 0. W. 1981. The influence of materna1 effects on
Denis, J. P. and Thiongane, A. 1. 1974. L’aptitude a la
the efficiency of selection; a review. Livesfock Prodrrcfion
production de viande chez le zebu Cobra du Senegal.
Science 5: 121-137.
Proceedings of fhe jïrsf world congress on genefics applied to
Sow, R. S., Denis, J. P., Trail, J. M. C., Thiongane, P. I.,
animal production, Madrid, vol. 3, pp. 889-897.
Mbaye, M. and Diallo, 1.1988. Productivite du Zebu Gobra
Eler, J. P., Van Vleck, L. D., Ferraz, J. B. S. and Lôbo, R. B.
au Centre de Recherches Zootechniques de Dahra
1995. Estimation of variantes due to direct and materna1
(Senegal). ISRA Etudes et Documents, vol. 2, no. 2.
effects: for growth traits of Nelor cattle. /ournal oj Animal
Statistical Analysis Systems Institute. 1992. SAS/STAT
Science 73: 3253-3258.
release 6.08. SAS Institute lnc., Cary, NC.
Gerstmayr, S. 1992. Impact of the data structure on the
Tawah, C. L., Mbah, D. A., Rege, J. E. 0. and Oumate, H.
reliability of the estimated genetic parameters in an animal
1993. Genetic evaluation of birth and weaning weight of
mode1 with materna1 effects. journal oj Animal Breeding and
Gudali and two-breed synthetic Wakwa bref cattle
Genetics 109: 321-336.
populations under selection in Cameroon: genetic and
Johnson, D. L. and Thompson, R. 1994. Restricted
phenotypic parameters. Animal Production 57: 73-79.
maximum likelihood estimation of variante components for
Van Vleck, L. D., David, St L. and Miller, J. 1. 1977.
univariate animal models using sparse matrix ‘techniques
Expected phenotypic response in weaning weight of beef
and a auasi-Newton vrocedure. Proceedinps of the fifth world
calves from selection for direct and materna1 genetic effects.
congres; on genetics applied fo livesfock prodkfib, Guelph, vol.
Journal of Animal Science 44: 360-367.
18, p. 410-413.
Willham, R. L. 1972. The role of materna1 effects in animal
Khombe, C. T., Hayes, J. F., Cue, R. 1. and Wade, K. M.
breeding. III. Biometrical aspects of materna1 effects in
1995. Estimation of direct additive and materna1 additive
animals. lournat of Animal Science 35: 1288-1293.
genetic effects for weaning weight in Mashona cattle of
Zimbabwe using an individual animal model. Animal
Wright, D. W., Johnson, 2. B., Brown, C. J. and Wildeus, S.
Production 60: 41-48.
1991. Variante and covariance estimates for weaning
weight of Senepol cattle. Journal of Animaf Science 69:
Koch, R. M. 1972. The role of materna1 effects in animal
3945-3951.
breeding. VI. Materna1 effects in beef cattle. Journal of
Animal Science 35: 1316-1323.
Koots, K. R, Gibson, J. P, Smith, C. and Wilton, J. W.
1994. Analyses of published genetic parameter estimates for
(Received 29 April 1997-Acc@cd 31 Orfober 1997)
--
^_
.-. -_