THE B BLOOD GROUP SYSTEM OF CHICKENS.
111. THE EFFECTS OF W O HETEROZYGOUS GENOTYPES ON
THE SURVIVAL AND EGG PRODUCTION OF MULTIPLE CROSSES
COURTNEY P. ALLEN
AND
DOUGLAS G. GILMOUR
Hy-LinePoultry Farms, Des Moines, Iowa,and S c h d
of Agriculture,
Cambridge University, England
Received July 23, 1962
N a previous paper of this series (BRILES,ALLENand MILLEN1957) data were
'presented showing that the B blood group system of chickens constituted a case
of genetic polymorphism. Of 73 closed populations tested, 71 had two or more
alleles. Moreover a detailed serological study disclosed the existence of 21 alleles
in 12 lines derived from nine different original stocks. GILMOUR
(1959) studied
four very highly inbred lines of widely differing origins. (inbreeding coefficients
98.0-99.4) and found two or more B alleles in each, with a total of nine different
alleles. Several other blood group systems are also polymorphic, although there is
less evidence available for them (BRILES1956; GILMOUR
1959, 1960).
It was concluded by BRILESet al. ( 1957) that the polymorphism of B was of the
balanced rather than the transient type in that segregation was continuing in
almost all the populations, even those highly inbred. GILMOUR'S
(1959) evidence
supported this conclusion for the B system, and extended it to other systems still
segregating at very high levels of inbreeding. The most obvious mechanism for
such continued segregation would be a selective advantage of heterozygotes over
homozygotes (although other mechanisms are possible, e.g., as suggested by
HALDANE1962). Direct evidence of heterozygote superiority during inbreeding
( 1960), and more fully for the B
was presented by BRILES(1956) and GILMOUR
system in the second paper of this series (BRILESand ALLEN1961). This latter
paper presented data based on 7732 fertile eggs of seven inbred lines, and provided evidence of association in some lines between B blood type and juvenile
viability, adult viability and egg production. The overall effect was one of
heterozygote superiority in viability and reproduction.
In the present paper experiments are described in which the comparisons of
viability and egg production were made, not between heterozygote and homozygote within inbred lines, but between two different heterozygotes in crossbred
individuals. The findings may thus have somewhat more relevance to the question
of polymorphism in outbreeding populations.
MATERIALS AND METHODS
The serological methods used were described by BRILESet al. (1957). Details
of the five inbred lines used in the present study, and of the genetics and serology
of the B locus antigens concerned, will be found in the same paper.
Genetics 47: 1711-1718 December 1962.
1712
C . P. ALLEN A N D D. G . GILMOUR
The male parents of the progeny whose performance was studied were inbred
males of genotype BZ1/Bz1selectd by blood typing from White Leghorn line H45.
The female parents were of a different origin in each test, as shown in Table 1,
being two-way crosses, three-way crosses, or backcrosses of White Leghorn inbred
lines; but in every case the alleles B13 or B14 were derived from males or line H36
as grandparents. Thus the primary comparisons made were between pullet genotypes BB1/B1$
and BZ1/B14,
while the effects of other B genotypes were assumed to
be equalized between the comparison groups. As an illustration of the mating
scheme and manner of blood typing, details will be given for the pullets hatched
in 1955 and tested over the season 1955-56. Their female parents were two-way
crosses of line H36 males and line H8 females. Preparation for the production of
birds of the appropriate B locus genotypes had begun in 1953. In each line two
alleles were segregating, BI3 and B'I4 in H36, and B" and B12 in H8. The progeny
of various matings in each line were bled and blood typed at about three months
of age, and the heterozygous birds selected and wing banded. In the next breeding season ( 1954) heterozygous B13/B14males of H36 were mated to heterozygous
B'I/B'* females of H8. The female progeny of this cross were similarly bled and
TABLE 1
Derivation and
Test
Year of hatch
of tested
progeny
B
locus genotypes of the comparison groups
Expected composition of
progeny companson groups
Female'
parents
Male grandparents (from
line H36)
Female
grandparents
Line i s ) of origin
of female
grandparents
* All genotypic cayories of female parents in this table were identified by blood typing the individuals concerned;
except those marked ,which were identified by their parents' blood genotypes.
GENOTYPES AND SURVIVAL
1713
blood typed, and sorted into the four possible genotypic classes. At this stage the
birds were wing banded with their B locus genotype. Meanwhile B”’/BB1males
were selected out of the progeny of various matings in line H45, and mated in
the 1955 breeding season to the four types of single-cross females. The female
progeny from these matings were then sorted into groups at hatching and wing
banded according to their dams’ B locus genotype. The comparisons between
versus B”/B14.
these 1955 progeny were treated as two separate tests of BS1’/B1$
The set of data with B” constant (female parents B”/B18 versus B11/B14) is
called test 1 in the present paper, and that with BIBconstant, test 2. It should be
realized that the progeny genotypes being compared were, in test 1, B2’/B” f
Bel/B” versus BB1/B” Bp1/B14and thus any effect of B’$ versus B’4 might be
expected to be “diluted” 50 percent.
The makeup of the other tests can be followed from Table 1. In test 3 the
female parents were derived from backcrosses to H36, and thus the comparison
1/4 Bzl’/BIB
classes were expected to be of approximate composition 3/4 BS1/BlS:
etc., with a corresponding reduction in “dilution” effect to 25 percent. In tests
4 and 5, the proportion of alleles other than Bl3 or B’4 was again one half, as in
tests 1 and 2.
The five tests were made over the four-year period 1955-58. Each took a
period of about 17 months, consisting of the incubation period, five months of
rearing, and 11 months of recording egg production and adult mortality. Tests
1 and 2 were “trap-tests,” while tests 3, 4 and 5 were “multiple variety tests.”
The former type of test was intended as a screening procedure for relatively
large numbers of possible varieties or crosses: accordingly each was tested on a
fairly small scale and on few locations. In tests 1 and 2, sire effects were
minimized as far as possible by rotating a number of males over all the females
twice weekly. Multiple variety tests were used for making more meticulous
comparisons of a few of the more promising varieties revealed by trap-testing.
Large numbers of birds were involved and tested a t many locations. In these
cases each dam category was separately mated to ten or more sires.
Detailed procedure was as follows. Chicks were hatched and sexed at the HyLine Hatchery at Des Moines. The pullets were wing banded with a variety
number corresponding to dams’ blood type and shipped to cooperating farms
(“locations”) all over the United States. Each location received all of one week‘s
hatch. No attempt was made to standardize conditions of rearing and management, which varied widely from farm to farm, although accuracy of recording
was checked regularly by traveling supervisors of the Hy-Line staff. At about five
months of age, the pullets were housed in laying quarters. In the trap-tests, birds
of the various dam types were intermingled and trapnested one day per week
(NORDSKOG
1948). The records were kept not individually but as totals for the
classes under comparison. Housing in the multiple variety tests was arranged on
a different basis, all the birds of each dam-type class being penned together and
total pen records of egg production kept. Alternatively, at some locations the
birds were housed in cages, but again no attempt was made to keep individual
records and only group totals were recorded. Despite the consequent confounding
+
1714
C. P. A L L E N A N D D. G. GILMOUR
of varietal and pen effects, this type of design of testing gives more reliable
results than the small-scale trap-test, partly because of the larger number of
birds tested in each group, but primarily because of the greater number of
locations used.
RES U LTS
W e were interested in testing the hypothesis of a relationship between B blood
type and overall fitness. I n this latter term we should include the individual
chicken’s s m i v a l from fertilization through its own reproductive period, together with its own contribution to the next generation. Separate measurable
variables contributing to this overall value for the genotypic classes would include
their o w n hatchability, survival through juvenile period (i.e. until sexual maturity), survival through the reproductive period, and their contribution to the
next generation as number of eggs laid and parental contribution to hatchability. We did not measure the last of these. Data on the other four suggested
components of survival value are given in Table 2. The data for hatchability
require separate consideration from the other results, since hatchability is probably subject to considerable parental influence. Thus in the present experiments
the effect of blood type on hatchability combines the effect of the various
maternal blood type comparisons, as laid out in Table 1, together with the
effects of the genotypic comparison we are primarily interested in, namely
BL1/Bl3versus Bz1/B14in the embryos. Paternal effects are expected to be randomized and in any case do not involve B blood type. The analysis of variance in
Table 3 indicates that the compounded maternal and progeny blood type difference is not a significant source of the total variation in hatchability.
For the other three components of survival value given in Table 2 we make the
assumption that the residual parental effect is less important, and thus that the
individual’s own genotype provides the main genetic effect. In the last column of
TABLE 2
The effects of two heterozygous genotypes on the performance of multiple crosses
Number of
replicates
Number of
Number of per
pullets housed
Test Incations location
BIs
B‘L
1
2
3
4
5
3
3
7
17
4
3
3
1
1
1
Meanss
103
101
650
1654
951
109
115
816
1630
918
Percent hatch
of all eggs
incubated
B’>
B’4
0-160 days
B’3
B’4
91.5
93.6
83.2
76.9
90.7
82.6
73.2
77.3
81.6
88.3
91.0
82.4
90.5
88.1
82.9
81.1
93.0
84.3
Percent viability*
161496 days
A13
BU
81.1
81.8
83.3
90.1
88.8
85.6
67.5 76.1
72.3 80.6
89.9 91.5
91.1 95.8
95.8 94.4
83.0 88.2
Percent:
hen-day eggs
RI3
B16
57.7
54.8
60.3
66.8
63.9
61.4
57.2
59.3
65.1
65.6
65.8
62.7
Mean of$
components
of fitness
~
1
3
66.1
68.1
77.3
82.1
83.5
75.6
gic
71.5
73.9
79.9
83.8
83.0
78.8
3459 3588
Totals
7047
for females progeny only.
+*$ Calculated
Computed from total eggs recorded divided by the summation of the number of hens alive on each recording day.
Simple arithmetic mean of percent viability 0-160 days, percent viability 161496 days and percent hen-day eggs.
$ Total of percentages of individual replicates and locations divided by 46.
1715
GENOTYPES A N D SURVIVAL
TABLE 3
Analysis of variance of hatchability of eggs set
Sources of variation
df
Sum of squares
Mean squares
B”’/B13 versus B”/B14 IB)
1
4
53.47
1,897.53
767.02
157.29
990.88
1,148.17
53.47
474.38
26.45
39.32
34.17
34.79
Tests (years and/or female sources) (T)
29
Hatches w/n sources (L)
4
BXT
29
BxL
B x T + B x L (&or)
33
F
<
1.54
12.06*
1.0
Significant at the .05 level of probability.
Table 2, these three components have been combined into a single percentage
figure which is a partial estimate of overall fitness. The analysis given in
Table 4 then allows a test of the dfect of blood type on this estimate of overall
fitness, while also making possible, by a study of the interaction B X C, a test of
its effect on the three components. The only estimate of error variance available
to us for testing the second-order interactions is derived from the replicates in
the small-scale tests 1 and 2. Each of these replicates averaged about 12 birds,
whereas the rest of the data was based on samples of a hundred or more birds.
Thus this error variance is likely to be biased in the upward direction, making
om test a conservative one. Since neither of the second order interactions was
significant when tested against this error term, the sums of squares of replicates
and the second-order interactions were pooled to give the error term used to test
significance for all other sources of variance except “components” and “tests.”
These were tested against their interaction, which was significant.
TABLE 4
Analysis of variance of three post-hatching traitsSurviua1 from hatching to 160 days,
from 161-496 days, and percent hen-day eggs
Sources of variation
B”’/Blj versus B*’/B14 (B)
df
1
Componentsof survival (C)
2
Tests (years and/or female sources) (T) 4
29
Locations w/n tests ( L w/n T)
2
BxC
4
BXT
B X Lw/nT
29
8
CXT
58
C x Lw/nT
8
BxCXT
58
B X C x Lw/nT
72
Replicates
Pooled error$
138
Sum of squares
704.00
31,930.12
10,218.62
5,131.77
177.87
291.82
1,303.08
1,811.91
5,918.28
278.34
1,133.50
6,751.25
8,163.09
704.00$
15,965.061&
2,554.6a
176.96$
88.%$.
72.96$
44.93$
226.48$
102.04.$
34.74119.54t
93.77
59.15
* * Significant at the .01 level of probability.
t Replicates used as error term.
2 Replicates plus both second order interactions totalled and used as error term.
S C X T used as error term.
F
Mean squares
11.902**
70.492**
11.280**
2.992**
1.504.
1 .W3
<I
3.829**
1.725
<I
<I
1716
C. P. A L L E N A N D D. G. G I L M O U R
The effect of the blood type difference on the partial overall fitness is significant at the .01 level of probability. The source of variation expressing interaction between blood type difference and components of fitness is nonsignificant
and relatively small. This implies that blood type differs little in its effect on the
three components. The superiority of Bz1/B14over B Z 1 / B fin
S each of the measures
thus adds up consistently to a highly significant superiority in the partial overall
fitness. Moreover the B x T source is also small and nonsignificant, implying
consistency of the blood type effect over the different tests. This contrasts with a
significant value for the C X T source (P < O.Ol), which implies considerable
variation in the average values of the three components as affected by the
different tests.
DISCUSSION
The selective advantages of the BZ1/B'4 genotype over B"/BIS, as shown in
Table 2, do not appear very large, although they are consistent. It should be
noted however, that the means in the last column, which express an arithmetical
combination of the components contributing to overall fitness, considerably
underestimate the magnitude of the total difference in fitness. A more expressive
combination is obtained by multiplying the three percentages. I n terms of the
means for B2'/BfS,82.4 percent of the pullets hatched survived to puberty, 83.0
percent of these survived through the laying (i.e. reproductive) period, and these
provided 61.4 percent of the potentially fertile eggs maximally possible, giving a
combined notional partial survival value for B"/B19 of 82.4 x 83.0 X 61.4 = 42.1
percent. For B2'/B1h the corresponding figure is 47.3 percent. From these figures
the selection coefficient against Bz1'/Bfsis 0.11. Even this is an underestimate
since in most of the comparison groups the effect of BP1/BlSversus BZf'/BI4was
"diluted" about 50 percent. Our results thus suggest a selection coefficient of the
order of 0.2.
The particular value of our findings is that they demonstrate for the first time
in chickens that differences can occur in survival value of different blood group
heterozygotes, and that these differences are stable over a fair variety of noninbred genetic backgrounds. Previous evidence on the selective value of blood
group genes in chickens has been derived so far almost entirely from the study of
inbred lines. BRILESet al. (1957) found that 71 out of 73 somewhat inbred
populations were still segregating at the B locus, while GILMOUR
(1959, 1960)
reported that at least 2 4 blood gro~iploci were still segregating in each of four
very highly inbred lines (computed inbreeding coefficients 98.6-99.4 percent).
I n several of the more highly inbred lines described by BRILESet aZ. (1957) one
of US (C.P.A.) has found continuing segregation at several loci additional to B.
All these findings are indicative of a powerful selective advantage of heterozygotes over homozygotes under the conditions of inbreeding. Direct evidence of
heterozygote superiority during inbreeding in factors related to survival, such as
reproductive performance and viability, was presented by BRILES(1956) and
GILMOUR
(1960). BRILESand ALLEN(1961 ) have recently presented fuller data,
based on 7732 fertile eggs in seven inbred lines, showing association of the B
GENOTYPES A N D SURVIVAL
1717
locus with viability during incubation, juvenile and adult life. Overall the effect
was one of an advantage of each heterozygote over the corresponding homozygotes.
This type of evidence from inbred lines is not entirely relevant to the question
of the polymorphism of blood groups under conditions of outbreeding. As GILMOUR (1962) pointed out, the process of inbreeding operating on multiple allelic
loci (such as B ) may tend to have a sieve-like effect, selecting and maintaining
in segregating form those alleles showing the more marked heterozygote advantage. It is noteworthy that BRILESet al. (1957) in their study of 73 populations found that the less inbred populations had as niany as eight alleles, but that
these were always reduced in number, usually to two, which then continued
segregating in the more highly inbred populations, because of a heterozygote
advantage (BRILESand ALLEN1961). MANDEL(1959) postulated on mathematical grounds that in a polymorphic system with several multiple alleles there
would be an overall margin of superiority of heterozygotes over homozygotes,
but that this margin could be quite small and still maintain polymorphism.
Some heterozygotes could even be inferior to homozygotes. Moreover, each
heterozygote would have a different fitness in the selective circumstances operating. It is evidence on this latter point which we have obtained. For in using
crossbred birds, we have avoided the special conditions of an inbred background,
and have been able to compare survival value of two heterozygous B locus genotypes in a variety 0-f noninbred gene backgrounds. Even so, we recognize that the
two B alleles compared, BIJ and BIJ, have a special relationship one to another,
in that they were the pair of continuing segregants in an inbred line (H36 of
BRILESet al. 1957, 1961). It is possible that a pair of alleles surviving the “sieve”
effect of inbreeding by showing heterozygote advantage might produce a marked
difference in survival value when combined in heterozygotes with a third allele
(here B2’).Certainly the magnitude of the selection coefficient operating between
the two heterozygous genotypes was quite large, and we should usually expect
much smaller ones.
Our evidence thus leads the argument only a little way out of the special situation of inbreeding, and towards the goal of understanding blood group polymorphisms in random-breeding populations. For a variety of reasons we are
pessimistic about direct studies of blood groups in such populations. There are
technical difficulties in identifying particular heterozygotes in such populations,
1959), and the numbers
the selective differentials need be quite small (MANDEL
of animals studied must be large (e.g. we studied 7047 females for one comparison). It is also entirely possible that blood group polymorphisms are involved in,
and depend on, the complexity of coadaptation of an integrated gene pool, with
multiple interactions between genes at different loci. Evidence that such interactions exist was provided by ALLEN (1962), who showed that survival from
hatching to 160 days of age, survival from 160 to 500 days of age and egg production variation were associated with an interaction of B blood group alleles
from one line with the genomes of two other lines. Detection of a single withinlocus effect would then require stabilization of much of the rest of the gene com-
1718
C. P. A L L E N A N D D. G. GILMOUR
plex, such as occws during inbreeding, but is not often otherwise possible without
the most sophisticated experimental designs.
It is with these reservations that we approach the essentially negative findings
on association between blood groups and phenotypic characters in cattle, reported
by N I E M A N N - S 0 R E N S E N and ROBERTSON(1961 ) .
SUMMARY
Comparisons of crossbred Single Comb White Leghorn chicks of contrasting
and BB1/B14were produced in five different tests
blood group genotypes BZ1/B13
over a four year period. The chicks were hatched from matings of B z f / B Z 1males
of inbred line H45 to crossbred females having either a BI8 or a B14 allele as one
of the two B locus alleles. A total of 7047 pullets was produced from such matings.
Hatchability (number of chicks hatched as a percentage of the eggs incubated)
was one of the traits studied. Analysis of variance showed that the blood group
difference was not a significant source of the variation.
A partial estimate of overall fitness in the evolutionary sense was derived by
using the arithmetic mean of the three percentage traits: juvenile mortality,
adult mortality and egg production. Analysis of variance revealed that the B
blood group system was a highly significant source of variation in this estimate
of overall fitness. Examination of the relevant interaction terms indicated that
this effect of blood type was consistent for the three traits studied and for the five
tests.
It is estimated that the advantage of Bz1/B14over Bs1/B13represented a selection coefficient of about 0.2.
LITERATURE CITED
ALLEN,C. P., 1962 The effect of parental B locus genotypes on multiple cross performance in
chickens. Ann. N.Y. Acad. Sci. 97: 184-193.
BRILES,W. E., 1956 Superiority of birds heterozygous for blood group genes. 5th Poultry
Breeders' Roundtable, 78-100.
BRILES,W. E., and C. P. ALLEN,1961 The B blood group system of chickens. 11. The effects of
genotype on livability and egg production in seven commercial inbred lines. Genetics 46:
1273-1 293.
BRILES,W. E., C. P. ALLENand T. W. MILLEN,1957 The B blood group system of chickens.
I. Heterozygosity in closed papulations. Genetics 42 : 631-648.
GILMOUR,D. G., 1959 Segregation of genes determining red cell antigens at high levels of
inbreeding in chickens. Genetics 44: 14-33.
1960 Blood groups in chickens. Brit. Poultry Sci. 1 : 75-100.
1962 Current status of blood groups in chickens Ann. N.Y. Acad. Sci. 97: 166-172.
HALDANE,
J. B. S., I962 Conditions for stable polymorphism at a n autosomal locus. Nature
193: 1108.
MANDEL,S. P. H., 1959 The stability of a multiple allelic system. Heredity 13: 289-302.
NEIMANN-SBRENSEN,
A., and A. ROBERTSON,
1961 The association between blood groups and
several production characteristics in three Danish cattle breeds. Acta Agr. Scand. 11:
163-196.
NORDSKOG,
A. W., 194.8 Periodical trapnesting and family selection for egg production.
Poultry Sci. 27: 713-718.