1. Gene350 Animal Genetics
Lecture 19
14 September 2009

2. Last Time Estimation of breeding values and accuracy Own performance
Progeny testing
Sibs
Ancestors

3. Today Multiple trait selection Methods Selection intensity
Correlations between traits

4. Multiple trait selection
Genetic, environmental & Phenotypic correlations are required for efficient multiple trait selection
Phenotypic correlation-2 traits such as height and weight can be correlated
Genetic correlation- the genetic values for 2 traits are correlated
Environmental correlation- the environmental effects on the 2 traits are correlated

5. Multiple trait selection
The definition of a correlation is the covariance between 2 traits divided by the square root of the product of the variances of the 2 traits
Phenotypic correlation =
Genetic correlation =
Environmental correlation =
Additive genetic correlation

6. Multiple trait selection
Response in correlated trait
CR =
Formula used to monitor what happens in other traits when we select for a single trait
Sign of genetic correlation determines the direction of correlated change
Selection intensity, phenotypic standard deviation and the heritabilities also determine the magnitude of the expected change

7. Multiple trait selection
The relative selection progress for trait 2 by selection for trait 1 as compared to progress from direct selection for trait 2 is

8. Multiple trait selection
economic efficiency in populations
reproductive efficiency
maternal ability
growth
feed efficiency
body composition
product quality
selection objective is very complex

9. Selection objective
all traits that contribute to
economic efficiency
Selection criterion
all traits that are evaluated for
improvement of the
selection objective
not necessarily the same traits

10. How might the selection objective and selection criterion have
different traits?
Some traits in objective might be
difficult to measure
expensive to measure
strongly associated with
other traits

11. How to decide what traits to include?
economic importance
heritability
standard deviation
genetic correlations among traits
phenotypic correlations among traits
ease or expense of measurement

12. Methods of multiple trait selection
tandem
independent culling levels
selection index .

13. Tandem selection select for one trait at a time
improve first trait until satisfied
improve second trait until satisfied
improve third trait until satisfied
etc .

14. Tandem selection not very effective
especially bad if adverse correlations
among traits
example
ADG and BF in pigs
select for  ADG   BF
select for  BF   ADG

15. Tandem selection Traits should be considered in order
of their economic value
Difficulties include:
Genetic correlations must be known
Correlations and heritabilities must
remain the same over several
generations of intense selection for one trait
3. Economic values are linear
4. Economic value does not change over time

16. Independent Culling Levels
establish culling criterion for each trait
cull any individual that fails any
criterion
advantage
can cull sequentially
can follow biological development of
animal. Selection will occur in
stages corresponding to level of maturity

17. Independent Culling Levels
disadvantage
strength in one trait cannot
make up for another weakness
If traits are correlated, calculating
expected response is difficult .

18. Selection index equation that combines all traits of importance
H = v1X1 + v2X2 + ….. VnXn
factors in calculation
economic importance
heritability
standard deviation
genetic correlations
phenotypic correlations

19. Selection index advantages most improvement in net merit
strength in one trait can make
up for weakness in
another
disadvantage
should measure all traits on
all selection candidates

20. Selection index Properties of selection index
Maximizes the correlation between the true
additive genetic merit and its predicted value
(accuracy of evaluation)
2. Minimizes the average squared prediction error
(minimizes prediction error)
3. Genetic gain is faster with this method than
with any other
4. Probability of correctly ranking pairs of
animals is maximized
5. Procedure is unbiased (average of prediction
error for all animals is zero)

21. National Swine Improvement Federation
L=number born alive minus the average number born alive
W=21-day litter weight minus adjusted 21-day litter weight
D=days to 250 pounds minus the average days to 250 pounds
B=backfat minus the average backfat
RECOMMENDED INDEXES
SPI = (L) + W
MI = (L) + .4(W) - 1.6(D) - 81(B)
TI = (D) - 168(B)

22. A practical approach set up loose culling levels
eliminate obviously inferior
cull sequentially
culling levels for soundness
eliminate problems
after all traits measured
selection index on remaining
selection candidates