ABSTRACT
Data from 4,336 pullets, progeny of 144 sires and 779 dams for strain A and
4,843 pullets, progeny of 158 sires and 1108 dams for strain B belonging to six
generations under selection for part- period egg production to 280 days of age were used
for this study. The data were used to compare heritability (h2) estimates from daughterdam
regression method, with those estimated by variance component estimate method
from five procedures (Harvey method, TYPE 1, MIVQUE, ML and REML of SAS).
Response to selection, genetic and non-genetic correlations among egg production traits
in the two strains were investigated. Effective number of parents and co-efficient of
inbreeding were also calculated for both lines. The chickens were grouped into selected
and control populations within the male and female lines.
The estimates of genetic parameters over the years were obtained after correcting
the data for hatch and year effect. The traits considered in the computation of response to
selection and genetic parameters were egg number (EGG280D), age at sexual maturity
(ASM), average egg weight (EWTAV) and body weight at 40 weeks of age (BWT40).
The heritability estimates from different variance component methods were close
to one another in magnitude and in agreement with those found in literature. The
heritability estimates obtained from variance components (half-sib) were 0.18, 0.15, 0.24
and 0.16 for age at sexual maturity (ASM), egg number (EGG280D), egg weight average
(EWTAV) and body weight at 40weeks of age (BWT40), respectively, for the male line.
The corresponding values for the female line were 0.20, 0.16, 0.29 and 0.21. The
estimates obtained from daughter – dam regression for ASM, EGG 280D, EWTAV and
BWT40 were 0.19, 0.05, 0.28, and 0.27, respectively, for the male line and 0.19, 0.25,
0.27 and 0.20, respectively, for the female line. The standard errors associated with the
parameter estimates were very low which is an indication of their reliability.
Direct genetic response to selection was higher in the female than male line (3.4
vs 0.42 eggs per generation). Selection is therefore much more effective in improving
part year egg production in the female line as compared to the male line. The genetic
correlation estimates between the different economic traits over five generations ranged
from – 0.70 + 0.38 to 0.82 + 0.42 vs – 0.71 + 0.47 to 0.76 + 0.29 for the male and female
lines respectively. The correlation between egg number and egg weight was small and not
– 7 –
significantly different. Age at sexual maturity was highly and negatively correlated with
egg production to 280days in both lines, being higher than – 0.60 in most cases. However
the genetic correlation between egg number and matured body weight (BWT40) showed
no definite trend in the male and female lines.
In the female line, the correlated response in age at sexual maturity as a result of
direct selection for egg production to 280days had negative value. This is also true for
body weight due to selection for increased egg number to 280days. There was a reduction
of 0.89g per year in egg weight due to selection for increased number. In the male line
however except for body weight at 40 weeks, which showed a positive correlated
response of 3.4g per year, all other traits considered showed negative correlated
responses to selection for egg number to 280days of age.
The average inbreeding co-efficient due to finite population for both male and
female populations were equal with a value of 0.005 while values for the control
population were 0.008 vs 0.007 for the male and female line, respectively. The effective
number of parents in each generation averaged 174 vs 187 for male and female lines,
respectively.
It was observed that there was an increasing trend in the co-efficient of inbreeding
per generation over the period of study. Although inbreeding could be adjudged mild in
the study population, there is need to widen the genetic base of the population to avoid
intense inbreeding that could result in selection plateau in due course.
The use of the various estimation procedures or options in this study revealed that
generally the maximum likelihood estimators of SAS are more appropriate in dealing
with Animal breeding data as they are capable of dealing with both random and fixed
effects in a mixed model and are able to handle unbalanced data characteristics of Animal
breeding data. However, if Harvey’s method is to be used for analysis, data set with
missing cells should be edited out before analysis to obtain meaningful estimates.
Daughter – dam regression is recommended in the estimation of heritability due to the
fact that random sampling errors associated with estimates are minimized. The offspring
– parent method excludes the effects of environment more efficiently than those based on
half or full – sibs.
– 8 –
TABLE OF CONTENTS
TITLE PAGE:……………………..…………………………………………………..i
CERTIFICATION:…………………………………….……………………………..ii
DECLARATION:…………………………………………………………………….iii
DEDICATION: ……………………………………………………………………….iv
ACKNOWLEDGEMENT:……………………………………………………………v
ABSTRACT:……………………………………………………………………….… vi
TABLE OF CONTENT:……………………………………………………………viii
LIST OF TABLES:……………………………………………………………………xi
CHAPTER ONE:………………………………………………………………………1
INTRODUCTION:……………………………………………………………………1
CHAPTER TWO:………………………………………………………………………6
LITERATURE REVIEW:…………………………………………………………..…6
2.1 Variance components estimation:…..………………………………………….6
2.2 Tools for the estimation of variance components:…………………………….7
2.3 Some traditionally important methods of variance component estimation:…9
2.4 Analysis of variance based methods:………………………………………….. 9
2.5 Minimum Variance (or Norm) Quadratic unbiased Estimation:…………..11
2.6 Maximum likelihood:………………………………………………………….12
2.7 Restricted Maximum Likelihood:…………………………………………….13
2.8 Heritability from variance component:………………………………………15
2.9 Heritability from offspring- parent regression:………………………………17
2.10 Correlations:…………………………………………………………………..20
2.11 Primary and secondary traits of economic importance……….…………….24
2.12 Selection for traits of economic importance:…………………….…………..26
– 9 –
2.13 Factors affecting Genetic progress:………………………………………….27
2.14 Population size and short term response to selection:………………………29
2.15 Prediction of effective population size:………………………………………31
2.16 Inbreeding and reproductive performance of laying hens:…………………33
CHAPTER THREE:……………………………………………………………..……37
MATERIALS AND METHODS:…………………………………………………..37
3.1 The Location of Study:………………………………………………………37
3.2 Stock Composition:……………………………………………………………37
3.3 Stock Management:……………………………………………………………38
3.4 Data:……………………………………………………………………………40
3.5 Selection Procedure:…………………………………………………………..40
3.6 Data Collection………………………………………………………….….….42
3.7 Estimation of Genetic Parameters:……………………………………….…..42
3.7.1 Estimating Variance Component:………………………………………..…..44
3.7.2 Heritability:……………………………………………………………….…..46
3.8 Parent Offspring Regression (intrasire regression of offspring on dam)…. 46
3.9 Estimation of Correlation:………….…………………………………………47
3.10 Estimation of Expected Genetic progress:……………………………….…..49
3.11 Estimation of Expected and Realized response to selection
for primary traits under selection:………………………………………..…49
3.12 Estimation of Realised responses (Phenotypic and Genetic):.…………….…50
3.13 Selection Differential:…………………………………………………………51
3.14 Effective Population Size and Rate of Inbreeding: ………………………….52
CHAPTER FOUR:………………………………………………………….………..55
4.0 RESULTS:……………..………………………………………………..……..55
4.1 Effective population size and inbreeding:…………………………..……….55
– 10 –
4.2 Maternal effects:…………………………………………………..………….55
4.3 Heritability estimates:….……………………………………………….……56
4.4 Genetic Correlation:…………………………………………………………56
4.5 Genetic and non-genetic estimates using one way analysis of
variance model:………….….…………………………………………….….57
4.6 Selection differential:………….………………………………………..……57
4.7 Responses of egg number (primary trait) to selection:…………………….58
4.8 Correlated Responses:……………………………………………………….58
CHAPTER FIVE:…….….…………………………………………………………..78
5.0 DISCUSSION:……………………………………………………………….78
5.1 Effective Population and inbreeding:….……………………………………78
5.2 Selection response:……………………………………………………………79
5.3 Correlated responses:………………………………………………………..81
5.4 Genetic Correlation:…………………………………………………..……..82
5.5 Heritability estimates:………………………………………………….……83
CHAPTER SIX:………………………………………………………………….…87
6.0 CONCLUSION AND RECOMMENDATION…………………..………….87
6.1 CONCLUSION:………………………………………………………………87
6.2 RECOMMENDATION:………….…………………………………………..88
REFERENCES:……………………………………………………………………..89
CHAPTER ONE
INTRODUCTION
Breeding practices in poultry aim at genetic improvement of birds through
successive generations. This requires intimate knowledge of the various characteristics of
the breeds. Since the ultimate objective of poultry breeding is to improve those qualities,
which have a definite market value such as increased egg production, improved quality of
egg and improved quality and quantity of meat, the success of a breeder in evolving
valuable strains would depend on how best he can combine the desirable qualities. In
practical breeding, selection is used and it involves making decision based on available
information. This information becomes even more relevant especially in those flocks
under selection, in view of the fact that continued selection tends to bring about changes
in the heritability and genetic correlations among traits (Sharma and Krishna, 1998).
Reliable estimates of genetic variances, covariances and heritabilities are needed
to formulate breeding plans, predict response to selection and estimate genetic merit of
animals. If the objective is solely to estimate genetic parameters such as heritability,
simple methods of estimation, which involve only parent and offspring generations,
based on parent- offspring regression or collateral relatives (full or half-sibs) can be used
(Falconer and Mackey, 1998). With records on one or both parents and their offspring,
the regression of offspring on one parent or both parents gives an unbiased estimate of
heritability assuming no environmental covariance and no selection (other than on the
parents only for that trait). If both parental records are available, heritability can also be
estimated as the sum of the partial regressions of progeny on sire and dam in a multiple
– 14 –
regression analysis (Gimelfarb and Willis, 1994). In a short-term experiment, the
estimated heritability (h2) can be used in the classical equation of quantitative genetics,
G= ih2 p to predict the response in offspring to selection, where ΔG is genetic change,
i is the standardized selection differential and p is the phenotypic standard deviation of
the trait under selection (Falconer and Mackey, 1998). The prediction equation assumes
that the regression of offspring on parent is linear, which would be so if the genotypes
and the phenotypes have a multivariate normal distribution. The use of linear regression
for estimating heritability or for the prediction of change from selection is justified only
when such linear relationships can be reasonably explained on genetic grounds
(Robertson, 1977). Thus an average estimate based on linearity either underestimates or
overestimates the true heritability (Ibe and Hill, 1988). In this case the response by a
character to selection predicted by non linear offspring-parent regression fitted to family
data may be quite different from the response predicted by linear regression fitted to the
same family data (Kempthorne, 1960; Gimelfarb and Willis, 1994).
Low genetic correlation could result from data not corrected for hatch effect as
observed in the report of Oliver et al. (1957) who observed 0.22 genetic correlations
between short-term egg numbers and percent production. When estimates were however
made within hatches Bohren et al. (1966) observed a correlation of 0.79 while Morris
(1964) obtained 0.74 from data corrected for hatch effect. Using four strains of White
Leghorn, Srivastava et al. (1989) reported genetic response of 2.48 to 3.23 eggs per
generation or per year after four generations of selection. They observed close agreement
between predicted and realised responses for strain one while in the remaining three
strains the realised response was approximately one and a half times that of the predicted
– 15 –
response. Such disparity in response could be caused, among other factors, by genetic
drift, error of measurement, genotype – environmental interactions, time trend and natural
selection (Hill, 1972,a, b)
Srivastava (1985) reported that the only possible reason for variable responses
among the four strains could be either genotype – environment interaction or correlated
responses. Kinney and Shoffner, (1967) suggested that the nature of genetic variance
present in each strain might also contribute to the estimated response. Poggenpoel and
Erasmus, (1978), Ayyagari et al. (1980) and Barua, (1983) reported variable results
between realised and predicted response. Gowe et al., (1959a) reported significant
regression of 3.71 eggs per generation without correction for environmental effects. After
correction for environmental effects and using random bred control line, regression
reduced to 1.26 eggs per generation, but was not significant. Poggenpoel and Erasmus
(1978) reported control corrected regression coefficient of 3.04 eggs. Morris (1963), and
Mohapatra and Srivastava, (1971) reported gains of 3.0 and 2.01 eggs per generation.
Johari et al. (1989) reported positive and significant response to selection for part period
egg production ranging from 1.44 to 2.21. The results of Gowe and Fairful, (1986),
Poggenpoel (1987) and Lie (1988) supported these reports from their work on White
Leghorn. Brah et al. (1986) reported gains ranging from 2.60 to 3.40 eggs per generation
while Liljedahl et al. (1979) reported gains ranging from 4.4 to 6.2 eggs per generation
over four years of selection. However Nordskorg et al. (1974) found no appreciable
response in White Leghorn lines selected for part record rate of lay until the 8th of the 11
generations of selection for which the realised heritability was estimated to be 0.07.
– 16 –
Inbreeding, a system of mating where the mates are more closely related than the
average members of the population being mated has been used for the production of
genetically uniform strains for subsequent crossing to utilize heterosis. The higher the
effective population size, the lower the expected inbreeding depression. Rates of inbreeding
are largely inflated in selected populations because of the reduced effective
population size. Inbreeding reduces genetic variability, vigor and reproductive
performance and increases the probability of fixation of unfavorable genes. In recent
years, various methods have been proposed to reduce the rate of inbreeding in selection
programmes while keeping genetic gains at the same level (Nomura et al. 2002). These
methods assume various selection and mating strategies. For example, a reduction in the
weight of family mean in index selection (Toro and Perez-Enciso, 1990) for weighted
ancestral Mendelian sampling estimates (Grundy et al. 1998). The limited use of selected
parents (Toro and Nieto, 1984: Wei 1995) have been shown to be efficient methods.
Other methods include non-random mating of selected parents, such as factorial mating
designs (Woolliams 1989), minimum co – ancestry mating (Toro et al. 1988) and
compensatory mating (Santiago and Caballero, 1995). Among these, minimum co –
ancestry mating is a simple and intuitively appealing method, since it directly aims at
minimizing the average inbreeding of progeny. Also there is some evidence that the
depressive effects of inbreeding on certain traits can be dampened by rigid selection for
that trait.
In quantitative genetics, animal breeding includes among other things the
estimation of genetic parameters with which the genetic differences in the traits of
animals are evaluated since selection is made based on genetic content rather than
– 17 –
phenotypic manifestation. The accuracy of these estimates depend on the method used
and the data structure. Various methods have been used to estimate genetic parameters.
However the accuracy of these methods are yet to be fully evaluated from available
literature. This study attempts to provide such needed information.
The broad objective of this study is therefore to estimate genetic parameters using
different methods while the specific objectives are: –
1. To estimate and compare heritability values from variance components obtained
from various SAS subroutines (TYPE1, MIVQUE, ML, AND REML),
Harvey’s and daughter – dam regression methods
2. To estimate the genetic, phenotypic and environmental correlations between some
economic traits.
3. To evaluate the response to selection in the primary trait.
4. To compute inbreeding co- efficient in the selected population
– 18 –
IF YOU CAN'T FIND YOUR TOPIC, CLICK HERE TO HIRE A WRITER»