The presumed purpose of animal evaluation programmes is to provide accurate and relevant information about the usefulness of alternative breeding stocks and breeding systems for defined breed roles and production-marketing environments. The two levels of animal germplasm evaluation programmes to be considered here are 1) Direct comparisons of genetic stocks and 2) Crossbreeding experiments. Programmes for genetic evaluation of individuals within breeds are considered only as they may contribute to comparisons among breeds.
Direct comparisons of genetic stocks include a wide range of usefulness, from only growth efficiency of young males to total input/output efficiency of herds or flocks. Usefulness depends on both accuracy and completeness of information obtained. Accuracy is affected by method of sampling the stocks and by the design and scale of performance recording. Inaccurate or incomplete descriptive data can be not only inadequate but even misleading to those using the information to guide choice of breeding stock for a commercial animal production system. Crossbreeding experiments are planned matings to estimate not only mean performance differences among straightbred animals of different breeds or strains of livestock, but also to measure the parameters of breed, heterosis and gene-recombination effects in crosses that will allow prediction of relative efficiency for alternative breeds and systems of breeding. Examples of these two levels of germplasm evaluation programmes will be discussed for major species of livestock.
The "on farm" improvement and record of performance plans are
useful primarily for selection within a breeder's own flocks. Both the possible
selection of birds enrolled and environmental differences among
breeders flocks limit the usefulness of published records for comparisons among
genetic stocks. The early publicly operated central tests compared performance for samples of adult birds from
different stocks under a common environment (Warren, 1958). However, the
weaknesses of small and selected samples from each stock soon led to development of so-called Random
Sample Performance Tests (Dickerson 1965). The avowed purpose of Random
Sample Tests has been "to provide a reliable guide for commercial
producers, hatcheries and breeders concerning the potential performance of
commercial chicks or poults offered for sale by hatchery outlets".
For egg production stocks, random samples are obtained from commercial
hatchery sources, preferably as hatching eggs rather than chicks.
Performance is compared under a commercial egg production environment. Pens of
each entry are replicated within a test location. Preferably, each stock is
entered in a large number of different test locations, thereby increasing the
reliability of the stock rankings based on data from all locations. Records of mortality, age at onset of lay, body
size, egg production, egg size, shell strength, blood and meat spots and
albumen quality are used to calculate egg and meat output value from each entry. Records of feed consumption and
chick cost constitute input. Economic comparisons are made in terms of
income over feed and chick cost per pullet housed (net income).
Feed per
unit weight of eggs produced provides a measure of biological efficiency, but
one which ignores output of salvage chicken meat at the end of the laying
period. A measure of economic efficiency influenced less by stock differences
in body size would be total input cost per unit of output value.
Two-year
combined summaries of results from all Random Sample Egg Production Tests in
U.S.A. and Canada provided overall estimates of breeding values, with 5%
confidence limits, for each performance trait of each stock (e.g., see ARS,
1966). Information also was included for each trait concerning the average
within-test correlation of the same stocks among replicates and between years,
as well as the repeatability of the same
stock between test locations in the same and different years. Because measures
of some traits are much more accurate than others as indicators of their
breeding values (i.e., more highly heritable), an index weighting of component
traits was shown to be a better predictor of net income in future tests and
years than net income itself (Kinney et al., 1969). This analysis also showed that future net income could
be predicted nearly as accurately without including the measures of feed intake, using only age at first egg, eggs
per hen housed, egg weights, laying mortality and hen body weight. Use
of the same unselected control stock at all
test locations over years permitted estimation of genetic changes over time for
the genetic stocks entered in the tests.
Such
analysis of test results helps users to realize the limited accuracy of results
from a single year and test location (e.g., repeatability of .4 to .5 for net
income) and the value of entering each stock at many test locations each year.
However, even estimates of net income based on entries at many locations are
far from perfectly accurate (e.g., $1.93 ± .10 per pullet housed for a stock
with 80 pens at 32 locations). Even this accuracy is low enough that small
changes in overall ranking of a stock can occur unpredictably from year to
year, and is a reason for breeders to avoid relying too heavily on results of
Random Sample Tests for their sales promotion.
In Random
Sample Tests of chicken and turkey meat production stocks, maternal effects of
parent flock health, age and egg size on chick size, mortality and later
broiler weights make unbiased sampling of
eggs for each entry difficult (Goodwin, 1961). Ideally samples for each
stock entered should come from several parent-stock flocks of the same standard age. Primary emphasis has been placed on
feed conversion or feed intake per unit weight of market birds. This
measure ignores any differences in dressing percentage, or carcass composition,
and in cost of chicks or poults, as affected by parent flock body size, egg production, fertility and hatchability of
eggs. Factors affecting cost of broiler chicks or turkey poults
presumably should be reflected in their price to growers. Evaluation of
carcasses is increasingly important but more difficult, requiring direct or
indirect measures of body fat and scores for conformation.
Random
sample testing of poultry stocks can be useful both to the industry and to
breeders, especially when there are many stocks to be compared and many
independent growers. Properly conducted,
such public evaluations of the available stocks direct industry attention
to the real merit of the alternative breeding stocks. However, continued
usefulness of such public testing to the
poultry industry depends upon the accuracy and relevance of test information
not obtainable more directly by individual producers and breeders.
At an
earlier period, there were many breeds of both egg, meat and dual purpose
chickens. Since then, there has been much Random Sample Testing and
experimental evaluation of breeds and their
crosses, leading to development of the present specialized egg and meat
production stocks (Warren, 1942, 1958). In the meat stocks, specialized
maternal and terminal sire lines have been
developed to maximize efficiency in production of market meat birds.
Parallel developments have occurred in meat turkeys. The extremely high
reproductive rate of poultry, the intensive mass-production management systems
and the intense competition in the poultry,
egg and meat industry, have now led to sharply reduced numbers of
surviving breeders in much of the world. Strain-crosses of Leghorn or part
Leghorn composites have become the dominant egg producing stocks.
Strain-crosses, with white feathers and
skin for market acceptance, now dominate the chicken and turkey meat industry.
Diallel or
partial diallel designs (Table 1) are generally used in crossbreeding
evaluation of poultry breeds or strains (see review by Jacobec et al., 1987)
because of their high reproductive rate and intensive management. Generally,
first cross-heterosis is important for sexual maturity, rate of lay, viability
and total egg mass per pullet housed. However, experiments extended to include
F2 or later generation progeny from inter se mating within the Fl cross have shown more loss of Fl
heterosis than the 50% expected from the
reduced heterozygosity in the F2 generation (review by Sheridan,
1981). This experience, and the
very small proportion of pure line populations necessary to produce the parents
of commercial chicks, logically have led to use of specific crosses
by commercial producers. If composite lines are developed, it is done to obtain
a desired blend of characteristics in a new
line intended for later use as the male or female parent in some specific F1 commercial cross.
Organized central performance testing of swine stocks began in Denmark
in the early 1900's (Clausen and Gerwig, 1958). Typically, four slaughter pigs
from each litter entered were fed together under standard conditions from about
20 to 90 kg to measure rate of gain, feed conversion and carcass traits. The
purpose was to improve accuracy in comparing the genetic merit of breeder stock
by testing all samples under a single uniform environment. This approach largely removed herd environmental
differences from the comparisons and facilitated
the uniform recording of feed consumption and of carcass traits. It also
permitted valuable analyses of genetic variation in performance traits
(Lush, 1936). However, the limited total capacity of the central testing
facilities allowed only small and potentially selected (unrepresentative)
samples of each breeder's stock for use as a sib or progeny test.
Similar
central testing of samples from breeders spread to the U.S.A. and other
countries in the 1920's (Craft, 1958). Then in the 1950's, testing samples of
full or half-sib sets of boars alone, or of boars and sib-sets of barrows and
gilts for slaughter was initiated in many states of U.S.A. and other countries
(King, 1955). Boars with the better records are offered for sale to breeders.
Such programmes allow comparison between boars from different breeders or even
breeds, but the limited proportion of all boars that can be accommodated limits
their accuracy in estimating differences among breeds or breeders. It also
means that most selection by breeders still must be based on their own records.
A more complete central testing procedure was developed in Britain (MLC,
1977) and the Netherlands (Schoonoord, 1981) to compare commercial cross
combinations offered by large breeders or
breeding companies, using entries of both boars and gilts to measure whole litter
growing performance and carcass traits (MLC, 1977). In Denmark (Jonsson, 1975),
facilities have been expanded to allow growth and carcass testing of a set of
four litter mates at one of twenty testing stations from each of one-half of
all approved breeding sows. Participating
breeders also record measures of age, weight and ultra-sonic sidefat,
eye-muscle and fat areas. Thus breeders and producers have rather
adequate summaries of both central test and
on farm records to use in selecting replacement breeders and for choices among
the breeds and breed crosses evaluated.
More
recently "on-farm" recording of reproduction, growth and backfat
performance in purebred herds of major US breeds (Stewart et al., 1991; Harris et al., 1989) has been
organized to provide estimates of breeding values intended primarily for use in
selection within breeds. However, such
complete herd records are also highly useful for comparisons between breeds within a common regional production
system. The main limitation of such comparisons
is the lack of direct feed conversion records and incomplete carcass
evaluation. These limitations can be overcome by joint use of the more limited
central-test comparisons along with predictions from the
"on-farm" body weights and backfat measures on live animals. The New Zealand
Voluntary Improvement Plan (VIP, 1979) is an example of central boar testing
combined with comprehensive on-farm performance recording. Clearly, the
information necessary to characterize differences among pure breeds of swine
can be obtained from both central test and on farm recording of performance or
a combination of these approaches. Central testing can provide more complete
information, but usually is handicapped by
problems of cost and of small, potentially selected, samples. Error in breed comparisons
from environmental variation among herds in on-farm performance recording can
be largely overcome by averaging unselected records from many herds and very
large numbers for each breed within a
region. However, the primary focus of "on-farm" recording must
be its use for within-herd selection, to avoid potential errors from
environmental competition among herds.
The
important improvements in viability and growth of market pigs and in
productivity of sows from crossing pure breeds have been demonstrated in
extensive crossbreeding experiments beginning in the early 1900's (Winters et al., 1937; Lush et al., 1939; reviews by
Jonsson, 1975; Sellier, 1976; Johnson. 1981). Complete or partial diallel mating designs have been used, including three-way
crosses to measure breed and heterosis effects on reproductive
performance of the F, crossbred sows (as in Table 2). Some of these experiments
also have compared F1 with purebred boars of the same breeds and
found significant heterosis in F1 male reproductive performance, but
negligible effects on performance of progeny.
Generally, crossbreeding results have indicated that maximum industry
efficiency in pork production can be realized by mating females of an F1,
cross chosen for superiority in reproductive, growth and carcass
traits with boars of the breed or F1, cross with best transmitted
viability, growth, carcass traits and superior male reproductive performance
(Bennett et al., 1983). If deviations of heterosis from degree of
heterozygosity are not importantly negative, composites of 3 or 4 maternal
breeds or irregular "periodic" rotations of sire breeds would retain
2/3 to 3/4 of the average F1 heterosis without requiring continued F1
replacements from purebred populations (Dickerson, 1973; Bennett, 1987).
A design
that would be useful in evaluating optimum fraction of an introduced breed in a
composite (e.g., Young, 1991) is comparison of reciprocal backcrosses (1/4 vs
3/4 or 1/8 vs 7/8) relative to a common control and the F2.
Here backcross comparisons can be made within the same level of retained
heterozygosity (Table 4).
A recent
experiment (Young et al., 1989) has compared parental, F1, F2
and F3 generations of crosses
among two sets of four breeds each, chosen either for market pig traits or for sow performance and pig traits, to see how
well F3 composite performance agrees with prediction from
parental and F1 performance. The F3 of the maternal breed
cross was above prediction for number weaned, but later in puberty and lower in
loin eye area. The F3 of the paternal breed crosses was slightly
above prediction for pig weight at weaning, earlier in puberty but lower in
loin eye area. These minor deviations of performance from predictions based on
only additive and dominant gene effects, along with the very small proportion
of purebred matings required to produce replacements, have encouraged swine industry use of specific crossbreeding systems
based upon estimates of the breed-average and F1 heterosis
effects for pig and sow performance traits.
Evaluation
of sheep and goat germplasm covers a broad spectrum, from summaries of on-farm or field performance records to
central testing of breed samples, to designed crossbreeding experiments
measuring average breed heterosis, and non-allelic gene interaction effects.
Performance traits studied vary with the major objective (meat, wool and/or
milk) and with the production environment (temperate, tropical, intensive,
extensive).
Organized
on-farm field recording of information on unselected animals for reproductive rate, mortality, body weights and
wool yields can provide initial characterization of differences among breeds maintained under similar regional
management conditions. Of course, large total numbers are required to
reduce the errors from environmental variation among the flocks sampled from each breed. More precise breed
comparisons of growth rate, feed conversion, carcass composition, wool
yield and quality, as well as milk production can be obtained by comparing
samples from each pure breed under a uniform central test environment, provided
that adequate numbers of representative samples are obtained from each breed (Turner, 1969). Central tests have been
widely used to compare only the growth potential of rams from terminal
sire breeds (Waldron et al., 1989), as a means of identifying the better
sources of replacement rams. Addition of feed consumption records would
increase the value of such ram testing. Central tests also could be used to compare
sire progenies from several meat breeds for growth, feed conversion and carcass
characters of ewe and wether lambs. Usefulness of
such central test comparisons of breeds and breeders is heavily dependent upon
adequate and representative sampling.
Experimental comparison of breeds for maternal (ewe) performance in
market lamb production can be done most efficiently by mating representative
ewes of each candidate breed to the same rams of one or more meat-type breeds.
This experimental design minimizes sampling
error from random sire effects on progeny in direct comparison of ewe breeds
(e.g., Fimland et al., 1969).
When
adequate numbers of ewes from the candidate breed are not available,
representative rams of each maternal breed can be mated to ewes of one or more
"native" breeds to produce the F1 females, which, in turn
are evaluated in subsequent matings with sires of the meat-type breeds (e.g.,
Jacobec and Drizik, in EEAP, 1988). This indirect design is approximately
one-fourth as efficient because it measures only one-half of the maternal breed differences (Figure 1). However,
there is no difference in the efficiency of comparisons among the
meat-type sire breeds. Records needed for a comprehensive evaluation include
not only those of ewe lifetime reproductive performance but also the viability,
growth and carcass traits of the market lambs produced (Dickerson, 1977).
When the
objective is to determine the optimum proportion of an exotic breed in composite populations derived from crossing with
adapted native breeds, a mating design comparing
ewes of the F2 (i.e., from F1 x F1)
generation with those of the 1/4 and 3/4 exotic backcrosses (as in Oltenacu et al., 1981) is efficient, because the proportion of
maximum F1, heterosis retained is expected to be equal (50%)
for these three levels of exotic breed contribution (Table 4). A prime example
is the worldwide experimental evaluation of Finnsheep
and other prolific breeds to increase net productivity under environments
ranging from temperate to subtropical and from intensive to extensive
range management (EEAP, 1988). Some of these
experiments compared F1, with the less heterozygous backcrosses, but
the estimated fractional breed effects for litter size born were relatively
unbiased because heterosis was slight. A Canadian experiment compared levels
from 1/8 to 7/8 and purebred Finnsheep ewes
(Fahmy, 1990). The reviews by Baker and others in an EEAP symposium (EEAP,
1988) provide a comprehensive summary of experiments evaluating potential usefulness of Finnsheep crossbreeding under
diverse managements. Experiments comparing Finnsheep with Romanov,
Booroola and other prolific breeds also are discussed.
Choices
between systematic crossbreeding and the optimum composite require estimates of
the average heterosis in overall performance realized for the two systems,
including the dilution of average heterosis from the proportion of purebreds
required to sustain each system. Such a comparison requires prediction of
performance and over-all efficiency for the complete crossbreeding system, and
for the F2 or later generation of the inter se mated composite,
using deviations from weighted means for the pure breeds involved, as in Young et al. (1986). For meat
production in sheep, comparisons likely would include maternal composite or maternal crossbred ewes, when both are
mated to meat-breed rams, and a straightbred general-purpose composite.
Such comparisons should include all important traits and the relative values of
wool and lamb that influence lifetime ewe productivity
under the intended production-marketing system (Ercanbrack and Knight, 1989). Breed
differences for each breed role in crossbreeding could be based on input costs
per unit of output value in the experiment
and compared with expected breed differences in unit costs based on
prior estimates of the economic weightings for each trait (e.g., Wang et. al.,
1991).
Because of the large differences among sheep breeds in prolificacy vs
growth-carcass merit, industry breeding systems for market lamb production generally
use superior large growth-carcass breeds to sire lambs from ewes of smaller, more
prolific breeds, breed-crosses or composites. Choice of the ewe-breeds depends
partly upon the feasability of production
environments in which nutrition, matings and care at lambing can be controlled.
Heterosis retained in prolific part-Finnsheep composites (Young et al.,
1986) has been encouraging for their use in crossing with terminal sire breeds,
thus simplifying matings required for production of replacement ewes. However,
there has been considerable variation among
sheep crossbreeding experiments from linear association of heterosis retained with level of increased heterozygosity
expected, possibly related to interaction with production environments.
Thus experimental evaluation of promising composite breed combinations seems
justified before recommending their adaption for industry use.
The various
types of on-farm "record of performance" programmes are intended
primarily for use in selection within a pure breed (Gregory et al., 1961). However, when averaged across many herds of each breed, they can
provide much useful information about differences
among breeds that exist within the same geographical and livestock management region.
Their value depends on accurate measurement and reporting of the important
performance traits for unselected animals. It is not usually feasible to
include on-farm records of feed consumption or of carcass composition. Central
Testing Stations can be used to compare samples from different breeds under a
common environment for some of the important performance traits, such as
growth, feed conversion, conformation and live fat measures of young bulls
(Olson, 1989) or these traits plus carcass traits of steers. However, the small number of
potentially selected animals and traits measured tend to limit both the
accuracy and completeness of breed comparisons based on information from
Central Tests.
The major
potential advantages of planned beef cattle crossbreeding experiments are 1) the measurement of both breed average and
crossbreeding heterosis effects, 2) minimizing environmental sources of
error and 3) more complete measurement of traits affecting production
efficiency. The production objective may be meat production only or a
combination of meat and milk production. In some cases, it may be desirable to
include more than one environment or
management system in the experiment, although this multiplies the
necessary scale of the experiment (e.g., Olson et al., 1991).
When the
objective is to evaluate several introduced or exotic breeds, the more feasible crossbreeding design is one comparing a
representative sample of sires from each introduced breed in matings
with one or more "native" breeds to improve performance potential or
adaptation to a difficult environment (Figure 1, Table 4 as in Gregory et al.,
1985; Trail et al., 1985, or Paschal et al.,
1991). The first generation allows comparison among
introduced breeds for the combination of transmitted (gI) and
heterosis (hI) effects of each breed of sire. Adding
information from the F2 and the two backcross generation matings would allow separate estimation of breed
differences in transmitted individual (gI), heterosis (hI), and recombination
effects (rI). It also would allow evaluation of the optimum fraction of each introduced breed from 1/4 to 3/4
at the same proportion of F1 increase in heterozygosity
(Table 4).
Estimation of maternal breed (gM), heterosis (hM)
and recombination (rM) effects requires third
generation matings of generation-two females with sires of an unrelated breed
(Table 4), and combining these results with information from generations one
and two. See Cundiff et al., (1986) and references cited for partial examples
of this approach. A summary of information
from these breed and crossbreeding evaluation experiments, as applied to the choices
among alternative crossbreeding systems, is given by Gregory and Cundiff
(1980).
When
adequate samples of females as well as males are available from each breed to
be evaluated, the complete diallel design (Table 1, as in Gregory et al., 1980;
Baker et al., 1989; or Comerford et al.,
1991) is more efficient for estimating breed individual (gI) and maternal (gM) and individual heterosis
(hI) effects. It can be extended to measure maternal heterosis
(hM) by including the three-way crosses (Table 2). However,
measurement of individual recombination effect (rI) deviations from
linear association with average changes in heterozygosity (i.e., from the additive plus dominance expectations)
would require comparison of F2 with mean of reciprocal backcrosses
(Table 4).
Possible non-allelic gene interaction deviations from linear
association with expected heterozygosity (rI,
rM, rp) could be measured most completely by comparing
deviations from purebred means for each four-way cross with those for the mean of
the corresponding four F3 generation two-way crosses (Table 3). If
paternal effects (gp, hp, rp) are negligible,
similar comparisons for each three-way cross with those for the mean F3
generation of the two corresponding two-way
crosses (Table 3) would provide similar estimates for rI and rM
deviations, e.g.,
Several large
scale studies of heterosis retention in beef cattle (Gregory and Cundiff, 1980;
Koch et al., 1985; Gregory et al., 1991a,b) under favorable temperate
environments have not detected important deviations from additive-dominance
expectations in advanced generations of composite populations. If these results
are representative of cattle in general, most breed and crossbreeding
evaluation studies need not extend beyond the three-way crosses needed to
evaluate heterosis in maternal performance. However, crossbreeding results with
dairy cattle (Madalena, 1989; Madalena et al., 1990a) have indicated important
recombination losses in composites under difficult tropical environments.
The trait of
primary importance in dairy cattle obviously is milk production, qualified by
fat, protein and total solids content. However, efficiency of milk production
can be also greatly affected by fertility, mortality and culling as they affect
herd life and replacement costs, as well as by fixed and maintenance costs
related to health care and cow size (Blake et al., 1986a; Schmidt and
Pritchard, 1988, Holman et al., 1990). Resistance to disease and parasites and
the tolerance of heat and of marginal feed intake are especially important
under some tropical, low-input production systems. Adjustments should be
avoided for such gene-influenced components of milk production as age at first
calving or lactation length under stressful environments (Madalena et al., 1989).
Within-herd
recording of milk, fat and now protein production is the longest, most
systematic and best utilized system of performance recording for domestic
animals. Although the DHIA system of performance recording is intended for use
in within-breed selection, breed averages across herds under similar management
clearly are good measures of breed differences in performance, and can be used
for comparing genetic evaluations between countries as well (Philipsson, 1987).
Within-herd records of milk, fat and protein production, supplemented by body
weights and reproductive performance of pedigreed cattle could even be used to
estimate breed differences in the economic efficiency of milk production, under
ranges of relative prices for milk components, feedstuffs and other inputs
(Blake et al., 1986a; Schmidt and Pritchard, 1988). Properly controlled, such
field comparisons could even be used to compare breeds of dairy cattle with
crosses between breeds (Fimland, 1975; Ericson, 1987; Ahlborn-Bruer and
Hohenboken, 1991). Thus, designed breed evaluation experiments are needed
mainly for the comprehensive evaluation of breeds in crossbreeding, including
crossing of native with exotic breeds to improve performance in difficult
environments (Simpson and Wilcox, 1982; Blake et al., 1986b; Cunningham and
Syrstad, 1987; Syrstad, 1988; Cunningham, 1989; Tewolde et al., 1990).
The most
informative crossing design is a complete diallel extended to include the
three-way crosses of females from each F1 and contemporary purebreds
by the same sires (Tables 1 and 2). The USDA
diallel crossing of Holstein, Ayrshire and Swiss breeds (McDowell and McDaniel, 1968) included these matings plus the first
generation of a three-breed rotation. Heterosis for fat corrected milk
production was 8 to 10% in F1 crosses with Ayrshire and Brown
Swiss but negligible for Ayrshire x Swiss F1. Heterosis in milk
yield was even greater for the three-way crosses. However, only the Ayrshire x
Holstein, Swiss x Holstein and Holstein x
(Ayrshire - Swiss) exceeded purebred Holstein in first lactation net return,
after adjusting for differences in health, mortality and calving interval.
The Holstein
x Guernsey crossbreeding experiment at the University of Illinois (Touchberry, 1970, 1992) attempted to include all
outputs and inputs affecting efficiency of dairy production (e.g., milk
solids yield, viability, reproduction, body weights, veterinary service,
mastitis, etc.). Breeding groups compared were purebred Holstein and Guernsey,
and reciprocal F1 crosses, from the same sire within a year and the
same dams in different years. The second generation compared the two pure
breeds with 1/4 and 3/4 Holstein backcrosses, using the same Holstein or
Guernsey bulls to sire each pair of purebred and backcross progenies and
allowing each F1 female to be mated for both backcrosses in
different years. The third generation included the two pure breeds plus the 3/8
and 5/8 backcrosses, from matings of the same sires to produce either pure and
5/8 Holstein or pure and 5/8 Guernsey progeny. The next two generations were
from crisscross matings of Holstein sires with pure and 3/8 Holstein and
Guernsey sires with pure and 3/8 Guernsey females followed by the reverse
backcross, to produce 5/16, 11/16, 11/32 and 21/36 and 100% Holstein and
Guernsey progeny, approaching the equilibrium 1/3 to 2/3 of a two-breed
rotation. Effects of breed additive and crossbreeding heterozygosity were
estimated from partial regressions on fractional breed of sire or dam and
heterozygosity. The F1 heterosis
was about 7% for total milk solids, but nearer 22% in terms of net return after
adjustment for reproductive, health and
other traits. However, the pure Holsteins used still exceeded the F1 crosses by about 10%,
because the pure Holsteins exceeded Guernsey's used by over 100% in estimated income over input costs. Thus, the potential
advantage from crossbreeding would be
much greater between breeds of more nearly equivalent performance, as
for Jerseys and Holsteins under the seasonal-pasture, milk-solids production
system of New Zealand (Ahlborn-Breur and Hohenboken, 1991). There,
Jersey-Holstein rotation crossbreeding apparently would slightly exceed pure
Holstein fat production before taking into account crossbred advantages in
reproduction, viability, and other performance.
Because of the general superiority of the Holstein-Friesian breed for
milk production, especially in temperate climates, interest has focussed on
differences among Friesian strains from different countries. The FAO
sponsored comparison in Poland of 10 strains of Friesian cattle (Stolzman et
al., 1988; Jasiorowski et al.,
1988a,b) compared the F1
and the 3/4 and 1/4 backcrosses of nine other strains with Polish Friesian.
Differences among the nine F1
crosses with the Polish Friesians would include 1/2 gI + hI
effects for each strain. Those among the nine breed of sire 3/4 backcrosses
would contain 3/4 gI + 1/2 hI + 1/8 rI + 1/2 gM
+ hM effects for each strain. Comparisons of 3/4 with 1/4
backcrosses would contain only the 1/2 gI
effect of each strain, and this would indirectly permit estimation of
differences in hI from
the combination of F1 and backcross information. Use of estimates
for gI and hI differences between strains would
then permit estimates of differences in 1/2 gM + hM + 1/8
rI. If contemporary purebred Polish Friesians had been included, the
experiment would have been much more
efficient for estimating both individual and maternal heterosis (hI
and hM). Inclusion of the F2 generation of each
cross would have allowed estimation of epistatic recombination effects as well (Table 4).
Another important question concerns the role of high producing dairy
breeds from temperate climates in the crossbreeding improvement of milk
production in more difficult tropical environments (Simpson and Wilcox, 1982;
Cunningham and Syrstad, 1987a; Cunningham, 1989; Tewolde et al., 1990) or role of Zebu cattle
crossbreeding in semitropical environments (Blake et al., 1986b). This question
can be approached from analysis of well
planned experiments on cooperator farms as in Madalena et al. (1989; 1990a),
being careful to avoid adjustment for gene-influenced components of milk
production, such as lactation length and age at first calving. Partial
regressions on fractions for breed composition
and on relative crossbreeding increase in heterozygosity (Robison et al., 1981) can be used. When inter se matings
of crossbreds are included along with levels of backcrossing to both exotic and adapted native breeds (as in Madalena et
al., 1989; 1990a), epistatic
recombination deviations from additive breed and dominance effects can be
detected (Table 3). For example, performance of the 5/8 inter se in this
analysis was markedly below the additive plus dominance expectations, relative
to those for the 1/4 to 31/32 Holstein crosses with Guzera. Madalena et al.
(1990b) also compared profit/day of herd life for the F1, 3/4
Holstein, 5/8 inter se, rotational cross and modified 2 Holstein: 1 Guzera
rotation, under high and low management
levels. Results emphasized the greater advantage of F1 over 3/4 Holstein for low than for high management.
Also results from the 2 Holstein: 1 Guzera (2H:1G) rotation relative to F1 were good (75%) under high
management, while the 1H:1G rotation was 60% of F1 and better
than the 2H:1G rotation under low management. However, Syrstad's (1990) summary
analysis of many studies comparing F1 and 3/4 backcross milk yields of first parity Holstein
and Jersey crosses indicated that the increased exotic breed effect at
least compensated for the reduced heterozygosity of the 3/4 backcross over a
wide range of herd production levels. Also, the ratio of Jersey to Holstein FI,
crosses was similar from low to high herd production levels. Perhaps, more
evidence for interaction of heterosis or
breed effects with herd production level would have been detected if viability,
lactation length and age at first calving could have been examined.
These
results suggest caution in assuming linear association of heterosis with
heterozygosity retained in breed composite populations until further
experimental evidence is obtained for the full array of important component
traits, especially under difficult production environments.