Íslenskar landbúnaðarrannsóknir - 01.03.1979, Qupperneq 100

98 ÍSLENZKAR LANDBÚNAÐARRANNSÓKNIR
toelt, trot and gallop. This is to be expec-
ted since most additive genetic variation is
observed for these traits. The trait char-
acter (trait 9) has also got a fairly large
amount of additive genetic variation. But
since this particular trait, as it is judged, is
genetically negatively correlated with
other traits of more evaluated importance,
it gets negative value of b9 in the selection
index and the expected response is very
slight indeed. One could at íirst sight
wonder why trait number 10 (elegance) is
so much negatively evaluated in the selec-
tion index. The explanation is that this
particular trait has got a very low herit-
ability and a small variation, but is fairly
strongly correlated (both genetically and
phenotypically) with other traits having
more additive genetic variation. A greater
response in the trait ,,elegance“ is there-
fore obtained by simultaneous selection
on the index than by a sole selection on
,,elegance“ itself (0.069i against 0.048i).
Its function in the index then becomes
mainly that of indicating the environment
for other traits with more additive genetic
variation and of more importance.
Cunningham (1969) has described a
simple method of evaluating the relative
contribution, of inclusion of one or more
variate in the index, to the rate of response
in the aggregate genotype.
The variance of a reduced index where
ith to jth variates are ignored becomes
ali.j = al - b’i. jW_1i..jbi..j
wh’ere: bi. j = a vector of the ith to jth
veighing factors in the original index
Wi..j = the corresponding diagonal sub-
matrix of P~'
From the variance ratio of the reduced
index and original one, it may be shown
that, e. g. dropping the variate „pace“
from the index would reduce expected
genetic progress for the aggregate geno-
type by 25.42% Similarly ignoring „eleg-
ance“ would result in loss of 1.52% in
terms of genetic gain and ignorance of the
two lowest evaluated characters in the sel-
ection index, i.e. ,,character“ and „prop-
ortions“ would only reduce genetic prog-
ress by 0.11%. Discussion as to whether
retaining a variate in the index is worth-
while, or not, should be weighted by the
cost or labour of recording compared with
its relative contribution to the aggregate
genotype.
The scoring point index, which at pres-
ent is supposed to form a basis for the
selection is exactly the „base“ index as
described by Williams (1962). The exp-
ected response from a truncative selection
on a „base“ index can be calculated.
A i - 0.680
B Va’Pa
aggregate units per unit selection
intensity
A selection on the base index is theref-
ore expected to be 20.37% less eífective
than selection on the ,,optimal“ selection
index.
It should be stressed in this context that
the weighting factors in the selection index
are only optimal and the predicted genetic
gain from a selection on the index is only
accurate if the genetic and the phenotypic
parameters are estimated without an
error. This is only the case when the
sample used for the estimation is large