.*5&12*38/'"2+"*.$/,..*5&12*38/'"2+"*.$/,.
*(*3",/--/.2.*5&12*38/'"2+"*.$/,.*(*3",/--/.2.*5&12*38/'"2+"*.$/,.
"$4,38"0&12".%4#,*$"3*/.2*..*-",
$*&.$&
.*-",$*&.$&&0"13-&.3
4,8
"1"-&3&1&23*-"3&2'/1(1&"28:&&$&6&*()3/'"-#/4*,,&3"1"-&3&1&23*-"3&2'/1(1&"28:&&$&6&*()3/'"-#/4*,,&3
2)&&0"3%*''&1&.3"(&22)&&0"3%*''&1&.3"(&2
!&&
.*5&12*38/'"2+"*.$/,.
!",%1/.
&7"2(1*$4,341",70&1*-&.33"3*/.&7"2.*5&12*38"..(&,/
",& ". ,&$+
.*5&12*38/'"2+"*.$/,.
%5".5,&$+4.,&%4
/,,/63)*2".%"%%*3*/.",6/1+2"3)3302%*(*3",$/--/.24.,&%4".*-",2$*'"$04#
"13/'3)&.*-",$*&.$&2/--/.2
&&!!",%1/.".% ". ,&$+",&"1"-&3&1&23*-"3&2'/1(1&"28:&&$&6&*()3/'
"-#/4*,,&32)&&0"3%*''&1&.3"(&2
"$4,38"0&12".%4#,*$"3*/.2*..*-",$*&.$&
)3302%*(*3",$/--/.24.,&%4".*-",2$*'"$04#
)*213*$,&*2#1/4()33/8/4'/1'1&&".%/0&."$$&22#83)&.*-",$*&.$&&0"13-&.3"3
*(*3",/--/.2.*5&12*38/'"2+"*.$/,.3)"2#&&."$$&03&%'/1*.$,42*/.*."$4,38"0&12".%
4#,*$"3*/.2*..*-",$*&.$"."43)/1*9&%"%-*.*231"3/1/'*(*3",/--/.2.*5&12*38/'"2+"*.$/,.
Parameter estimates for greasy fleece weight of Rambouillet
sheep at different ages
1,2
J. W. Lee*, D. F. Waldron†
,3
, and L. D. Van Vleck‡
*Department of Animal Science, University of Nebraska, Lincoln 68583-0908; †Texas Agricultural Experiment
Station, Texas A&M University, San Angelo 76901-9714; and ‡USDA, ARS,
Roman L. Hruska U.S. Meat Animal Research Center, Lincoln, NE 68583-0908
ABSTRACT: Variance components for greasy fleece
weight in Rambouillet sheep were estimated. Greasy
fleece weight was modeled either as repeated measure-
ments on the same trait or as different traits at different
ages. The original data were separated according to the
age of the ewe at shearing into three classes; 1 yr, 2
and 3 yr, and older than 3 yr. An animal model was
used to obtain estimates of genetic parameters with a
REML algorithm. Total numbers of animals in pedi-
grees for the different age classes were 696, 729, and
573, respectively, and 822 for the repeated measures
model across ages. The animal model included direct
genetic, permanent environmental, and residual envi-
ronmental random effects and fixed effects for age of
Key Words: Genetic Correlations, Genotype Environment Interaction, Heritability
2000 American Society of Animal Science. All rights reserved. J. Anim. Sci. 2000. 78:2108–2112
Introduction
Previous studies have reported estimates of heritabil-
ities and genetic correlations for several traits in sheep
(Shelton and Menzies, 1968; Vesely et al., 1970; Coelli
et al., 1998). Fogarty (1995) summarized estimates of
genetic parameters for live weight, wool, and reproduc-
tion traits. Ercanbrack and Knight (1998) estimated
genetic trends in lamb and wool production as part of
a study to investigate the effectiveness of four selection
protocols for improving production. Their study com-
pared the economic results of using these protocols in
Rambouillet, Targhee, Columbia, and Polypay sheep.
Falconer (1952) suggested that measurements of a ge-
notype in different environments might be considered
1
Published as paper no. 12745, Journal Ser., Nebraska Agric. Res.
Div., Univ. of Nebraska, Lincoln 68583-0908.
2
The authors gratefully acknowledge the contributions of Maurice
Shelton and Don Spiller in flock management and data recording.
3
Correspondence: 7887 US Highway 87 N. (phone: 915/653-4576;
fax: 915/658-4364; E-mail: d-waldron@tamu.edu).
Received August 23, 1999.
Accepted March 3, 2000.
2108
ewe, shearing date as contemporary group, and number
of lambs born. Days between shearings was used as a
covariate. Single-trait analyses were initially done to
obtain starting values for multiple-trait analyses. A
repeated measures model across ages was also used.
Estimates of heritability by age group were .42, .50,
and .58 from three-trait (age class) analyses and for
the repeated measures model the estimate was .57.
Estimates of genetic correlations between fleece yields
for 1 yr and 2 and 3 yr, 1 yr and >3 yr, and 2 and 3 yr
and >3 yr classes were .88, .89, and .97, respectively.
These estimates of genetic correlations suggest that a
repeated measures model for greasy fleece weight is
adequate for making selection decisions.
separate traits. If the genetic correlations are less than
.80, selection might be more effective if the trait is
defined by the environment where it is expressed (Fal-
coner, 1952; Robertson, 1959). Okut et al. (1999) esti-
mated the genetic correlations between expressions of
genotypes at different ages of animal for wool traits for
Columbia, Polypay, Rambouillet, and Targhee sheep.
Coelli et al. (1998) reported estimates of heritabilities
and genetic correlations for greasy fleece weight at dif-
ferent ages for Australian Merino sheep. The objective
of this study was to determine whether greasy fleece
weights at different ages in Rambouillet ewes should
be considered as repeated measurements on a single
trait or as different traits.
Materials and Methods
A flock of Rambouillet sheep was maintained by the
Texas Agricultural Experiment Station at a ranch in
Edwards County, Texas. The ewes were representative
of the Rambouillet population in Texas. Flock manage-
ment was typical for commercial conditions in the area
except that single-sire breeding pastures were used and
all lambing occurred in a barn to facilitate accurate
pedigree records. Ewes were typically exposed for
Fleece weight 2109
Table 1. Summary of observations, unadjusted means,
and standard deviations (SD) for greasy fleece
weight (kg) by age of ewe class
Age of ewe class (yr)
Item 1 2 and 3 >3All
Mean 3.62 4.37 4.26 4.16
SD .84 .87 .88 .91
Records 542 885 898 2,325
Ewes 542 577 413 691
Sires 42 40 47 60
Dams 306 322 253 370
breeding from August to October, with subsequent
lambing from January to March. All lambs were identi-
fied with their dam and ear-tagged within 18 h of birth.
All lambs were weaned on the same day within a year
when lambs averaged approximately 100 d of age.
Greasy fleece weights from the Rambouillet flock
were collected from 1978 through 1998. Actual fleece
weights were adjusted to a 365-d growth period. Fleece
weight was divided by number of days from the previous
shearing and multiplied by 365. For the first shearing,
the number of days since birth was used. Most shear-
ing dates were in April or May. All ewes were shorn
for the first time at about 14 mo of age. At the 14-mo
shearing, all ewes were shorn on the same day, within
each year.
Three data sets were generated from the original data
file to analyze three different age classes separately.
Records were divided by the age of ewe into three
classes: ages of 1 yr, 2 and 3 yr, and older than 3 yr.
The ages of ewe classes were the same categories as
used by Okut et al. (1999). Number of days between
shearings (or to first shearing) was fitted as a covariate
to account for non-linearity of the adjustment to 365 d.
Analyses were also done without this covariate. Age
of ewe, calendar day at shearing (as an indicator of
contemporary group), and number of lambs born be-
tween shearings were fixed factors for all models. Num-
ber of lambs weaned would have been preferred to num-
ber of lambs born but was not available for most of the
years. Total numbers of animals in pedigrees for the
age 1, age 2 and 3, and age > 3 yr classes were 696,
729, and 573, respectively, and 822 for the repeated
records model across ages. Numbers of records and un-
adjusted means and standard deviations by age class
of ewes are in Table 1.
A single-trait animal model was initially used to es-
tablish starting values for multiple-trait models as well
as for the repeated measures model for all observations.
The full single-trait model was
y = Xβ + Za + Wp + e
where
y isaN× 1 vector of observations
β is vector of fixed effects (age of ewe, contemporary
group, number of lambs born, and covariate of days
between shearings)
a is vector of additive genetic effects
p is vector of uncorrelated permanent environmental
effects
e is vector of residual effects, with X, Z, and W known
matrices relating observations in y to vectors of
fixed and random effects
For the model,
E[y] = Xβ and
V
a
p
e
=
Aσ
2
a
00
0I
N
a
σ
2
p
0
00I
N
σ
2
e
where
N
a
= number of ewes
N = number of records
A = numerator relationship matrix among an-
imals in the pedigree file
I = identity matrix of appropriate order
The environmental covariance due to permanent en-
vironmental effects of ewes in the different age classes
can be forced into the structure of covariance of perma-
nent environmental effects across age classes even
though ewes in the first age class can have only one
measurement (Okut et al., 1999). The correlation be-
tween total (including residual and permanent) envi-
ronmental effects can be calculated from the sum of
the residual and permanent environmental variance
components. The total environmental covariance be-
tween measures in two different age classes was com-
puted as
r
e
ij
=
r
p
ij
√
p
2
i
× p
2
j
√
(p
2
i
+ e
2
i
)(p
2
j
+ e
2
j
)
where
r
e
ij
= the correlation betweentotal (residual and
permanent) environmental effects for
measurements of a ewe in age classes i
and j,
r
p
ij
= the correlation between permanent envi-
ronmental effects for age classes i and j
= fraction of variance due to permanent en-
p
2
i
vironmental effects for age class i
= fraction of variance due to permanent en-
p
2
j
vironmental effects for age class j
Lee et al.2110
Table 2. Estimates of variance components and genetic
parameters (standard errors) for greasy fleece weight
from single-trait analyses by age of ewe class and
for a repeated measures model across ages (all)
Age of ewe class (yr)
Parameter
a
1 2 and 3 >3All
σ
2
a
1.00 1.61 1.79 1.58
σ
2
p
.11 .41 .18
σ
2
e
1.08 1.03 .69 1.01
σ
2
y
2.08 2.75 2.90 2.77
h
2
a
.48 .59 .62 .57
(.097) (.072) (.083) (.056)
p
2
.04 .14 .07
(.061) (.072) (.042)
e
2
.52 .37 .24 .36
(.097) (.037) (.024) (.023)
a
σ
2
a
= direct genetic variance, σ
2
p
= variance due to permanent envi-
ronmental effects, σ
2
e
= variance of residual effects, σ
2
y
= phenotypic
variance, h
2
a
= direct heritability, p
2
= fraction of variance due to
permanent environmental effects, and e
2
= fraction of variance due
to residual effects.
= fraction of variance due to residual effects
e
2
i
for age class i
= fraction of variance due to residual effects
e
2
j
for age class j
Estimates of genetic parameters were obtained with
a derivative-free algorithm for REML (Boldman et al.,
1995). The program was restarted with estimates at
previous apparent convergence as initial values until
a global minimum was found (i.e., minus twice the loga-
rithm of the likelihood did not change to the third deci-
mal after consecutive restarts).
For multiple-trait analyses, this model was expanded
to include covariances between additive genetic and
permanent environmental effects in different age
classes but with residual covariances assumed to be
zero.
Repeatability for the repeated measures model was
estimated as (additive genetic variance + permanent
environmental variance)/phenotypic variance.
Results and Discussion
Single-Trait Analyses
Parameter estimates from single-trait analyses by
age classes for greasy fleece weight are shown in Table
2. Estimates of heritabilities by age classes for greasy
fleece weight gradually increased with age of ewes (.48,
.59, and .62) with an intermediate estimate for the re-
peated measures model (.57). Except for age of ewe class
1, the estimates were the same whether or not days in
shearing period was included in the model as a covari-
ate. Including the covariate of days to first shearing
reduced the heritability estimate for age class 1 from
.50 to .48. These estimates agreed well with those of
Okut et al. (1999), who reported estimates of heritabilit-
ies for the same age classes for greasy fleece weight for
Rambouillet sheep to be .50, .68, and .68, respectively.
The estimates were greater than those of Bassett et
al. (1967), who reported an estimate of heritability for
greasy fleece weight to be .11 for Rambouillet sheep.
Fogarty (1995) reported mean heritabilities of .35 and
.36 for greasy and clean fleece weight, respectively, and
weighted average genetic and phenotypic correlations
between greasy and clean fleece weights of .84 and .88,
respectively, in an extensive review of published pa-
rameter estimates. Therefore, literature estimates of
both greasy and clean fleece weights are discussed to-
gether. Saboulard et al. (1995) estimated direct herita-
bilities for clean fleece weight to be .60 also for Ram-
bouillet sheep. Vesely et al. (1970) estimated heritabil-
ity to be .31 for greasy fleece weight and to be .23 for
clean fleece weight for Rambouillet sheep. Shelton and
Menzies (1968) obtained estimates of heritabilities to
be .58 by the paternal half-sib method and .52 by the
dam-offspring regression method for mean greasy fleece
weight for Rambouillet sheep.
Estimates of relative variance due to total (including
permanent and residual) environmental effects were
.52, .41, and .38, for the last three age classes and .43
for the analysis over all ages. These estimates agreed
with those of Okut et al. (1999) who reported the frac-
tion of variance due to both environmental effects for
the same age classes for greasy fleece weight in Ram-
bouillet sheep to be .50, .33, and .32, respectively.
Estimate of heritability for greasy fleece weight when
observations were considered as repeated measure-
ments of the same trait was .57. Repeatability was .64.
The heritability estimate reflects the average of esti-
mates for the three age classes. Comparison of the heri-
tability and repeatability estimates shows that genetic
value of the ewe is much more important than perma-
nent environmental effect of ewe for greasy fleece
weight.
Multiple-Trait Analyses
Parameter estimates for two-trait and three-trait
analyses for fleece weight are shown in Tables 3 and
4. Estimates of correlations among total environmental
effects from the two-trait analyses were .21 for 1 yr
with 2 and 3 yr, .07 for 1 yr with >3 yr, and .36 for 2
and 3 yr with >3 yr. Estimates of direct heritability by
age class for the three-trait analysis were .42, .50, and
.58, respectively. These estimates were somewhat less
than with two-trait analyses, which were somewhat
less than with single-trait analyses. These results agree
with those of Coelli et al. (1998), who reported heritabil-
ities for fleece weight to be .37 for 1 yr, .40 for 2 yr, .49
for 3 yr, .48 for 4 yr, and .47 for 5 yr age classes for
Australian Merino sheep. For two-trait analyses, esti-
mates of genetic correlations between 1 yr and 2 and
3 yr, 1 yr and >3 yr, and 2 and 3 yr and >3 yr classes
Fleece weight 2111
Table 3. Parameter estimates from two-trait (age of ewe classes)
analyses for greasy fleece weight
Trait (age)
12σ
2
y
1
σ
2
y
2
h
2
a
1
h
2
a
2
r
a
1
a
2
p
2
1
p
2
2
e
2
1
e
2
2
r
e
1
e
2
1 2 and 3 2.05 2.84 .45 .61 .90 .41 .03 .14 .36 .21
1 >3 2.08 3.08 .47 .68 .93 .43 .10 .09 .22 .07
2 and 3 >3 2.69 2.81 .50 .57 .97 .15 .19 .35 .25 .36
a
σ
2
y
i
= phenotypic variance for trait i,h
2
a
i
= heritability for trait i,r
a
i
a
j
= genetic correlation between traits
i and j,p
2
i
= fraction of variance due to permanent environmental effects for traits i,e
2
i
= fraction of variance
due to residual effects for trait i, and r
e
i
e
j
= correlation between total (permanent and residual) environmental
effects for traits i and j.
were .90, .93, and .97, respectively, which are highly
positive. For three-trait analyses, estimates of genetic
correlations between 1 yr and 2 and 3 yr, 1 yr and >3
yr, and 2 and 3 yr and >3 yr classes were slightly smaller
but still large; .88, .89, and .97, respectively. These
estimates of genetic correlations among age classes
were greater than those Okut et al. (1999) reported
between the same age classes for Rambouillet sheep at
the U.S. Sheep Experiment Station of .76, .74, and .94,
respectively. These results also agree with those of
Coelli et al. (1998), who reported genetic correlations
between 2 yr and older than 2 yr age classes for fleece
weight in Australian Merino sheep to be greater than
.88.
The guideline of Robertson (1959) is that, if the ge-
netic correlation between traits is greater than .80, then
the trait need not be divided into separate traits defined
by age class. The genetic correlations among greasy
fleece weight in the three age classes indicate that selec-
tion for increasing fleece weight at young ages would
result in increasing fleece weight at older ages. The
large estimates of genetic correlations indicate that
greasy fleece weights could be modeled as repeated
measurements rather than separately by age class.
These results agree with the recommendation of Okut
et al. (1999) that fleece weight measured at different
ages be considered as one trait. The estimates of herita-
bility and of total variance, however, in this study were
Table 4. Parameter estimates from three-trait (age of ewe classes)
analyses for greasy fleece weight
Correlation matrix
e
Trait (age) σ
2a
y
h
2b
p
2c
e
2d
12∼3 >3
1 2.01 .42 .45 .12 .88 .89
2∼3 2.75 .50 .16 .33 .29 .97
>3 2.88 .68 .19 .24 .24 .37
a
σ
2
y
= phenotypic variance.
b
h
2
= heritability.
c
p
2
= fraction of variance due to permanent residual effects.
d
e
2
= fraction of variance due to residual effects.
e
Genetic correlations above diagonal and total environmental correlations (permanent and residual) below
diagonal.
smaller for the 1 yr class than for the two older age
classes. Thus, some consideration might be given to age
classes of 1 yr and >1 yr or to standardizing phenotypic
variance across ages.
Fixed Effects of Number of Lambs Born
and Days Between Shearings
The estimated effect of producing twin lambs, com-
pared with producing a single lamb, on the weight of
fleece being grown at the time of parturition was −.26
± .10 kg (P < .05) for 2- and 3-yr-old ewes and −.26 ±
.08kg(P < .05) for ewes older than 3 yr. These estimates
are higher than those of Snowder and Shelton (1988),
who reported that Rambouillet ewes that weaned two
lambs had .14 kg less greasy fleece weight than did
ewes that raised a single lamb. Ray and Sidwell (1964)
reported that the effects of lactation were greater than
those of pregnancy in reducing wool production of Nav-
ajo and Targhee ewes. Because records of the number
of lambs weaned were not available for this flock, the
number of lambs born was the best available predictor
of number of lambs weaned. The estimated effects of
producing an additional lamb in the present study were
within the range of estimates in the review of Cor-
bett (1979).
The regression of greasy fleece weight on days be-
tween shearings was significant (.0203 ± .0053) for the