Íslenskar landbúnaðarrannsóknir - 01.03.1970, Side 61

COLOUR INHERITANCE IN ICELANDIC SHEEP 59
TABLE 18
Maximum likelihood estimates (p) and standard errors of tlie allele pararneters
at for sires, bt for dams and for sexes combined
Parameters Estimate (P) S.E. | 1-0 —p | S.E.
ai 1.021 0.090 0.232
a2 0.876 0.087 1.429
33 1.094 0.201 0.469
a4 0.929 0.142 0.503
bi 0.945 0.087 0.629
b2 0.839 0.094 1.722
b3 0.882 0.191 0.616
b4 1.196 0.181 1.080
Cl 0.979 0.056 0.383
C2 0.855 0.062 2.345
C3 0.999 0.137 0.005
c4 1.044 0.111 0.393
colours. The resulting estimates, their
standard deviations and the deviations of
the parameters from the expected value,
1.0, divided by the standard error, are giv-
en in the upper part oí table 18.
In table 18 are also shown the maxim-
um likelihood estimates of new parameters,
Cj, which are obtained under the assump-
tion that aj = b; = c^ for each i. The Cj
parameters are therefore combined estimat-
es of a4 and bi; and for their estimation
the reciprocal matings are of identical im-
portance and have been added togetlier.
As table 18 sliows, none of the a^ and bs
parameters are significantly different from
1.0, and none of the difíerences between
corresponding a/s and ly’s are significant.
Two of the within sex parameters, how-
ever, deviate appreciably from 1.0, i.e. a2
by 1.429 and b2 by 1.722 standard devia-
tions, respectively, when estimated from all
matings.
Of the c-parameters only c2 is signífic-
antly different from 1.0 in all matings (P <
0.05).
The above significance tests relate only
to individual parameters. The total effect
of all parameters on the variation in segre-
gation was found by calculating expected
frequencies in individual matings in two
ways. First the expectations were calculat-
ed on basis of the value 1.0 for all para-
meters and then the maximum likelihood
estimates of the parameters were used for
a second calculation of the expected fre-
quencies. The ^-’s obtained frorn these
calculations are shown in table 19.
The results in table 19 confirm the
trends found in table 18, i.e. that the 8
parameters when all are estimated together
do not show a significant reduction in the
variation in the segregation ratios. When
a2 and b2 are the only parameters estimat-
ed, they account for a considerable, and