Íslenskar landbúnaðarrannsóknir - 01.03.1970, Qupperneq 57

COLOUR INHERITANCE IN ICELANDIC SHEEP 55
tion in table 9, while the comparisons
made in the test were kept to a minimum
in order to avoid ambiguity in the inter-
pretation of the result o£ the test.
A. ESTIMATION OF ALLELE
PARAMETERS
Deviations from expected segregation ratios
can be caused by four different mechan-
isms, i.e.
1) differential representation of the two
alleles from either one of the parents,
2) differential union of a gamete from
one parent with the two gametes from
the other parent,
3) differential mortality of zygotes from
one genotypic class to another, and
4) phenotypic overlapping of different
genotypic classes.
The last mentioned cause of disturbed
segratation can be excluded in the present
study, because the observed colours in
table 9 agree well with those expected.
The only colours where misclassifcation
due to overlapping phenotypes could be
expected are grey vs. ldack and greybrown
vs. brown. The fact that neither black nor
brown lambs were observed in the matings
A2 A2 X A5A5 and A5Að X A2A2 shows
that misclassification w.r.t. these colours
must have been very rare. These colours
will be discussed further later, p. 61 and
onwards.
The first mentioned of the 4 mechanisms
above can be regarded as the average effect
of one allele as compared to the average
effect of the other allele within the same
genotype. The second and third mechan-
isms on the other hand can be regarded
as interaction between alleles from the two
parents.
In case of differential representation of
the two alleles from either one of the par-
ents the probability of each one of the
alleles being represented among the pro-
geny is different from 0.5. For this type of
deviation the relation between the geno-
types of parents and progeny can be ex-
]iressed in the general terrns shown in
table 14.
In case of eqnal representation of both
alleles from each of the parents, i.e. p,j =
Pji = qlk = qkl = 0.5, the expected value of
all the g’s is 0.25, while unequal representa-
tion of the two alleles from either one o£
the parents leads to g-values which differ
from the expected 0.25.
When all genotypic classes can be dist-
inguished, p(j ancl qkl can be estimated
from the observed frequencies and their
agreement with expectation tested by a
X2-test.
I lie number of phenotypic classes of
progeny from each mating clepends on the
dominance relationship among the A-allel-
es of the parents in cpiestion.
Matings where only one phenotypic class
is expected give no information about
segregation ratios and cannot tlierefore be
included in any segregation test. Matings
which result in 2 or more phenotypic
classes can be grouped into 7 dilferent
main situations, as shown in table 15.
Situation 4a can be treated as situation
4 by changing the order of the alleles of
the sire, and situation 5a can be treated
as situation 5 by changing the order of the
alleles of the dam. By going this each mat-
ing where segregation is expected in table
9 will belong to one of the 7 main situa-
tions in table 15.
Eour phenotypic classes are only expect-
ed in situation 7 of table 15. Direct estima-
tion and testing of p1} and qkl within in-
dividual matings can therefore only be
applied to a limited proportation of the
data in table 9.
The direct estimation of pi:i and qkl gives
estimates of the representation of allele At