Animal Molecular Breeding, 2013, Vol.3, No.1, 1-8
http://amb.sophiapublisher.com
13
Rank correlations for the trait under all the methods
were highly significant, ranging from 0.724 (between
model 8 and multivariate animal model for GFY
1
)
to
0.997 (
between model 8 and univariate animal model
for BWT). Rank correlations were above 0.950 for all
the traits when sires were ranked by univariate and
multivariate animal model. So, the ranking by these
two animal models was almost same. Rank
correlations of model 8 with univariate and
multivariate animal models were lower than univariate
with multivariate animal models. The rank correlation
among the different methods were though high and
significant (P<0.01), yet not perfect, revealing that
ranking of sires by different methods not similar.
The rank correlations ‘among traits within method’
were lower than ‘among within trait’. In ‘among traits
within method’, the ranking of the sires changed
resulting in to decreased rank correlation coefficients.
The change in ranking of sires with increase in age or
weights of their daughter might be due to non-unity in
genetic correlations between different weights. Within
the method, the rank correlation ranged from 0.124
(
between 6WT and GFY
1
in multivariate animal model) to
0.934 (
between 6WT and GFY
1
in multivariate animal
model). All the rank correlation coefficients were significant,
except between BWT and GFY
1
in univariate animal
model and between 6WT and GFY
1
in multivariate
animal model. Model 8 had higher rank correlations
among traits as compared to other two. In general,
WWT had highest rank correlation with 6WT in all
the methods, ranging from 0.728 in model 8 to 0.934
in multivariate animal model. This high rank
correlation might be due to high genetic correlation
between WWT and 6WT. These findings are in close
agreement with the reports of Ahmad, (2002) in
Avikalin sheep.
2
Materials and Methods
2.1
Data
The present study includes data collected from 1974 to
1998
in the Chokla sheep flock at of Chokla sheep
was maintained at the Institute under “All India
Coordinated Research Project on Sheep Breeding
(
AICRP-SP)” for fine wool until 1992 and from April
1992
the flock was maintained under, Network Project
on Sheep Improvement in the research project
“
Evaluation and Improvement of Chokla Sheep for
Carpet Wool”.
The animals with known pedigree and complete
records on all traits viz. birth weight, weaning weight,
6
month weight and first greasy fleece yield were
considered for the present study. The animals were
given new identity on the basis of their date of birth to
avoid the pedigree check. While the identity no of the
animal must be higher, the point is the data checks are
to make sure parents are, n fact, older ten progeny.
The sires with less than 4 progeny were deleted from
the analysis. Similarly, years in which less than 20
observations were deleted from the analysis.
2.2
Statistical Methodology
For the estimation of parameters and (co) variance
components, least-squares analysis (LSA) and derivative
free restricted maximum likelihood (DFREML) methods
were employed. Data were subjected to LSMLMW and
MIXMDL package of Harvey (1990) under different
models. A total of two models were considered to examine
the effect of genetic and non-genetic factors on various
body weight traits and on first greasy fleece yield.
2.3
Model 2
The model 2 considered was from LSMLMW and
MIXMDL package of Harvey (1990) which consists
one set of cross classified non-interacting random
effect. All four traits were analyzed simultaneously,
the model is as follows.
Y
ijkl
= ų + s
i
+ c
j
+ p
k
+ e
ijkl
where, Y
ijkl
is observation on 1
th
progeny of j
th
sex in
k
th
year; ų is the over all mean; s
i
is the random effect
of i
th
sire (i = 1,2,……., 110); c
j
is the fixed effect of
the j
th
sex (j = 1, 2); p
k
is the fixed effect of k
th
year of
birth (k =1, 2, ………., 20), and e
ijkl
is the random
error which is normally and independently distributed
with mean 0 and variance σ
2
e
.
The analysis was computed with the mixed model
least squares program which utilizes the method 3 of
Henderson (1953).
2.4
Model 8
The model 8 considered was same as above which
also consists one set of cross classified non interacting