Transcript No Slide Title

Archived at http://orgprints.org/00001564
Genetic improvement on feather pecking behaviour by selection is promising
Guosheng Su, Jørgen B. Kjaer and Poul Sørensen
Danish Institute of Agricultural Sciences, P.O. Box 39, Foulum, DK - 8830 Tjele, Denmark
Objectives
To investigate genetic variation in feather pecking behaviour
and genetic improvement on this trait by selection, using an
appropriate data-transformation and an animal model.
Materials and Methods
Selection experiment:
Line LP was selected for low feather pecking, line HP for high
feather pecking, conducted for four generations.
Selection was based on number of feather pecking bouts.
Feather pecking for females was recorded at about 30 weeks
in floor pens for 3 hours.
A bout = a series of continuous peaking direct to the same
body part in the same chicken.
Table 1. Phenotypic mean and standard Deviation of bout
number in observed scale and in Box-Cox trans. scale with
=-0.2).
Line
GeneN
Bouts (obs.)
Bouts (trans.)
ration
Mean
Std
Mean
Std
LP
0
193
2.82
6.15
0.701
0.677
1-4
867
2.24
7.67
0.499
0.622
HP
0
200
2.42
5.31
0.661
0.656
1-4
850
7.20
23.9
0.878
0.815
No effects of observing pen and full-sib group can be found, but
a heritability of 0.18 and 0.15 for number of bouts in line LP and
HP, respectively (Table 2).
2
(p ),
Statistical analysis:
The data for each line were analysed separately using an
animal model:
Table 2. Phenotypic variance
ratio of variance for
2
2
pen effects (c ) and for full-sib group effects (f ) and
2
heritability (h ) (Box-Cox trans. scale with =-0.2).
2
2
2
2
Line
c
f
p
h SE
LP
0.3984
0.009
0.000
0.181  0.08
HP
0.6313
0.007
0.000
0.154  0.07
where yt is the number of bouts of feather pecking in Box-Cox
transformation scale, i.e.,
yt = (y - 1)/, for   0; yt = log(y), for  = 0; ( y>0)
where y is the number of bouts in original scale plus one (to
meet y>0). A set of  ranging from -1 to 1 with interval 0.1 were
used to do the transformation.
Results
The transformation giving the largest maximum likelihood
was that with  = -0.2 in both lines (Fig. 1).
Genetic changes by selection was significant (Figure 2). After 4
generations of selection the total response was 62% of base
population mean in line LP and 84% in line HP.
0.8
Mean breeding value
yt = year-hatch + pen + full-sib + additive genetic+ residual
0.6
0.4
0.2
LP
0
HP
-0.2
-0.4
-0.6
0
HP Bout
LP Bout
1
2
3
4
5
- 2 log L
- 2 log L
Generation
-1.5
-1
-0.5
0

0.5
1
1.5
Figure 2. Mean breeding value for bout number in each
Generation (Box-Cox trans. scale with =-0.2).
-1.5
-1
-0.5
0
0.5
1
1.5

Figure 1. Maximum likelihood for the model based on the
transformed data with different values of , expressed
as -2log likelihood (-2log L).
Bout number averaged over selection generations in line
HP was much higher than that in line LP (Table 1). The
coefficient of variation was about 100% in transformed scale
and 300% in observed scale.
Conclusions
•Large phenotypic
behaviour.
variation
for
feather
pecking
•No full-sib group effect can be found, indicating that
maternal effect is negligible.
•Moderate or low heritability.
•Genetic improvement on feather pecking by selection is
effective.