Íslenskar landbúnaðarrannsóknir - 01.09.1976, Síða 58

56 ÍSLENZKAR LANDBÚNABARRANNSÓKNIR
By setting Cov = o, one obtains
MS
From this expression one can then esti-
mate the heritability of maternal effect thus:
, 2 _ . , 2 12.
hM 4^öSD 4 V
2
Örji
Results and discussion.
The heritability estimates obtained are shown
in table 1. The progeny groups in the data
have been of reasonable size (k = 16.583
for males). The genetic relationship between
halfsibs, rg was found to be 0.253, by use
of Osborne’s (1957) formula, so that no
serious error is introduced by multiplying
the sire component by 4.
It should be pointed out that the analysis
is done on a within year and farm basis. In
order to obtain a combined estimate, the re-
spective sums of squares and degrees of free-
dom have been added. In the data used in
this smdy, several rams were used in succes-
sive years, giving altogether 391 sire groups.
As a result of this, one utilizes data on each
sire each year as on independent source of
information although the sire may already
have occurred in a previous year. It is there-
fore a matter of discussion whether the pro-
cedure of calculating sums of squares and
degrees of freedom as described here is valid.
A search trough available literature for a
description and discussion of this problem
has proved unsuccessful. It must therefore
remain an open question whether the com-
bined estimate will be affected and in which
way.
The heritability estimate for female lambs
was found to be higher than that for male
lambs. When correction factors were calcu-
lated (JÓnmundsson 1976) it was found
that male lambs were affected to a greater
extent by systematic environmental effects
than female lambs. Pattie (1965), found a
higher realized heritability estimate for
female than for male lambs. HallgrÍmsson
(1971), on the other hand, found higher
heritability for single males than for single
females.
The lowest heritability estimates were
found on the farm Reykhólar. One of the
reasons for this is the use of closely related
rams on this farm. This high degree of re-
lationship between sires results in reduced
variation between sires, which again may
lead to zero or negative estimates of the sire
component.
Assortative mating which may have oc-
curred on the farms under study will also
affect the estimates. Selection of sires will
have the same effect. Intensive selection for
other characters than autumn weight of
lambs might also lead to biased estimates.
Hallgrímsson (1971) obtained a herita-
bility of 0.18 for twin lambs but a higher
estimate for single lambs. No other estimates
of heritability of autumn weight of lambs in
Iceland are available. Bowman (1966) gives
an account of heritability estimates for live
weight of lambs, which show a wide range
of variation. Several of these estimates are
based on relatively limited amounts of data,
b. Matemal effect.
The same analysis of variance as described
earlier in this paper used for estimating the
heritability of the maternal effect, except that
the data were now grouped by sire of dam.
The results from this analysis are shown in
table 2. For all farms combined heritability