|
Heritability |
|
by Fred Lanting and edited by John Cole |
Perhaps no term is
more misunderstood by dog fanciers when talking about inherited characteristics
and environmental effects than the concept of heritability. It is necessary for
you to take a little time and effort to get this idea fixed in your mind before
applying it to practical breeding programs. I will try to give the major ways
of looking at it as expressed by some of the leading authorities, in both
veterinary medicine and genetics. Try to differentiate between the similar
terms of heritability and inheritance. Think of HD as being 100% transmitted by
genes, while its expression (obvious or radiographic signs) is not as
predictable as that 100%. The first is how
the trait is obtained, while the second is how much of the trait is seen (obvious
or radiographic signs). Pin this on your wall:
INHERITANCE = HOW
HERITABILITY = HOW MUCH
Dr. Corley of the OFA describes heritability as a
percentage measurement of "how much of the phenotypic variation in a
studied population is of genetic origin." This means the differences
caused by the genes, not the disease itself.
In the late 1950s and early '60s it was thought that heritability of HD
was around 0.6 (60%), but experiments in
When
Olmstead says, "70% of the development of hip dysplasia
is influenced by genetics" (personal correspondence, 1992), the key word
is development. Environment, then, would
affect 30% of the development, including the severity, age at which HD became
apparent, and the specific nature of the lesions (osteophytes,
laxity, etc.). Environment would not
have anything to do with the actual existence (genotype) of the disorder. Therefore, to measure the total phenotypic
variance of HD, you must evaluate the pelvic radiographs of a large population
of dogs maintained in the same environment (identical nutritional intake per
pound of body weight, for example). Dr.
Lust of
Malcolm Willis,
professor of genetics at the University of Newcastle-Upon-Tyne, and one of
You can
also positively direct heritability by fine-tuning diagnostic and rating systems. With the old BVA scheme, in which hips were
classified as normal, near-normal, and dysplastic, HD
heritability in the German Shepherd Dog was said to be around 0.25, but when
their much more detailed 0-to-106-point scoring system was applied to the same
dogs, heritability increased to nearly 0.4, which supposedly represents greater
success in eliminating (or, at least, reducing) the HD genotype from your line.
If this is an accurate estimate, then it means that the presence of good and
bad genes have been more closely identified by the newer British scheme than
they had been by using the old one, and the effects of environment are of
lesser importance now that diagnosis was supposedly more accurate. The OFA,
using the AVMA protocol, is reported to use a seven-feature approach, also, but
in practice two things are of utmost importance in a diagnosis: laxity and any
sign of remodeling or DJD. this
is not something they publicize to the breeder.
Continuing with the idea of directing heritability and changing
the indexes, remember that HD is polygenic. When many genes are involved, more
accuracy might be had in “looking” at them by more methods or aspects. Using an
evaluation scale with many levels for example, would allow the geneticist to
determine how much of the trait is under the control of genetics, and how much
is under the control of the environment. When you put all individuals into two
or three groups in order to evaluate them in terms of heritability, you do a
poor job of describing “reality” either statistically or verbally. Consider an
equation for heritability: h2
= s2A / (s2A +s2E). The better job
we do of estimating the environmental and genetic variances in the equation,
the better estimate we get of heritability, h2. Usually when a scale
with few categories is used to measure a trait that exhibits continuous
variation, genetic variance is underestimated and environmental variance is
overestimated (this has been going on in the field of HD for a long time), and
h2 index appears to be smaller than it should be.
Good advice has been given that breeders perhaps should
not use the reported and variable heritability figures either as a tool in
breeding or as an excuse to avoid using good genetic sense. We should not interpret heritability to be
anything more than an aid in predicting results from a breeding program. Stick to the basics: HD (and other joint
disease) is a genetic problem and must be attacked by better selection of
genes. You can raise the heritability of
your own lines by using the right mix of outcrossing
and linebreeding on dogs with the best phenotypes.
LINEBREEDING,
OUTCROSSING, AND INBREEDING
Linebreeding (or its extreme application,
inbreeding), while perpetuating desirable characteristics also limits the gene
pool. Another
way of saying this is that when you linebreed/inbreed,
you end up with a smaller pedigree (contains fewer unique individuals). This
results in a reduction of additive genetic variance because you have fewer
genes to put together, and you have many individuals with similar genotypes.
That is why puppies from linebreeding practices look
so much more uniform than do pups resulting from outcrossing. Since there is less variability and lower
heritability in linebred dogs, one can expect much
less progress through phenotypic selection when the hips of all the near
ancestors have not been carefully "chosen." The typical German Shepherd
Dog is an example of extremely close linebreeding. In
recent decades this has been as true of the German dog as it has long been of
the phenotypically quite-different AKC version. So, it is not surprising that breeders who
have not "culled the ancestors" and selected strongly for good hips
far back in the pedigree, will often find a "percentage plateau" (a
point at which no further progress can be made because of no more variability
in the line).
It is
imperative that when you buy an outcrossed pup, you
make sure that the parents' hips are as good as possible. By obtaining all the information you can on
the parents' and their littermates and litters, especially the dam's, the less
likely you will be to encounter the "outcross surprise."
The
October and December 1992 issues of Shiba Journal
carried an excellent piece on inbreeding by Susan Houser, an attorney with another
degree in zoology and direct work experience in genetics. She points out that inbreeding (breeding of
very close relatives) "brings about a decline in characters concerned with
fitness (viability, fecundity, and growth)”, quotes Darwin in saying that
prolonged inbreeding brings about "loss of size, constitutional vigor, and fertility", and quotes Ehrlich and
colleagues in saying, "a loss of fitness referred to as inbreeding
depression occurs . . . [when] . . . inbreeding is imposed on populations that
are usually outbreeding". Willis also warns
against unwise inbreeding. Inbreeding limits small populations to small numbers
of genes, and too often, many of these are the "wrong" genes being
paired because there is no dominant "good" allele there. While inbreeding concentrates certain desired
characteristics such as milk production, rapid weight gain in pigs, more ears
per cornstalk, grain yield per acre, rear leg angulation,
short legs, chiseled head features, etc., it also
causes the breed of animal to lose certain other genes through attrition or
prevention of being reintroduced to the pool. At least, the result is a loss of
availability of genes that are not easily recovered. The idea that genes are
“lost” with increasing homozygosity is referred to by
population geneticists as “fixation”.
Inbreeding
depression is the result of achieving homozygosity in
recessive genes. This could partly
explain the reported rise in HD (9.7%) and other defects among American German
Shepherd Dogs while the rest of the world population in that breed is steadily
(but slowly) improving in hips, without losing other historic qualities of
type. Other breeds were reported by the
OFA to have increased in frequency of HD in the 1980s compared to the
1970s: Great Dane 3.8%, Golden Retriever
1.3%, and
In the light of this knowledge (or perhaps it is in
spite of this knowledge being available!) why do breeders use so much
inbreeding, since there is voluminous evidence that it badly affects health? Because they are focusing on one or a few specific traits and,
seeing some short-range success with perhaps a litter of champions, continue to
inbreed or heavily linebreed beyond the invisible
margin of safety. Then they wonder why their lines are notorious for intussusception, pancreatic deficiency, and other organ
problems as well as shortened lifespans. The puppy
buyers, rather than the breeders, are too often forced to pay the price of
inbreeding.
Fortunately,
breeders can improve general health by selecting ancestors for good joints
without inbreeding. However, those who
inbreed or linebreed without regard to x-ray knowledge
of their dogs' joints will undoubtedly add dysplasias
to the list of complaints of their doggy descendants.
PHENOTYPE SELECTION AND VARIABILITY IN REPORTING h2
INDEX
Heritabilities vary, both by
the method of evaluating the trait and by which particular traits are
considered (e.g., HD's average heritability may be different from heritability
of shoulder angulation or temperament). Heritability of HD varies also between
breeds, so that the Samoyed, for example, may have an estimated index
variability of 0.8 and the German Shepherd Dog may have one of 0.25 (we shall
see later that this might be an inaccurate and too-low figure) even if
calculated by the same people. In
HD heritability can also be a function or property of a
colony or population of dogs selected over a period of time so that they have a
different percentage than other groups not so selectively bred. A conscientious breeder who has been applying
the best principles of HD prevention to his program for generations may have a
different index in his dogs than exists in the breed at large. Heritability indicesfor
HD in purebred dogs have been estimated at levels as low as 0.2 and as high as
0.6 or even as high as 0.8 in Samoyeds, with most of the data supporting these
numbers having been gathered before much breeding progress in specific
colonies. Reasons for why heritability
estimates vary have been more recently analyzed and they are as much dependent
on the particular population (selected subgroup) as on the environment, if not
more.
Heritability
differences between breeds can reflect the numbers of dogs and families used by
the amount of linebreeding and by differences in
selection practices. They may also
reflect the number of grades and, hence, the accuracy of the rating/scoring
system used. Heritability differences
within a breed can be influenced by the same things. The reasons heritability of HD in American
guide dogs for the blind is so high (0.54) probably include conscious
pre-selection for better hips and possibly the use of other breeds in addition
to the "historic" German Shepherd Dog blind
guide dog.
Traits
that are high or moderately high in heritability (perhaps over 0.35, as an
arbitrary level) respond comparatively rapidly to phenotype selection (i.e.,
selection of breeding partners based on no radiographic evidence of
abnormalities in the joints). High
heritability means that environmental differences will have less effect on the
expression of bad genes, because there are fewer bad genes to be affected. In such animals, phenotype is a better
predictor of genotype than is the case in those dogs with low heritability
indexes. The polygenic trait of HD,
after generations of phenotypic breeding (i.e., exclusion of dysplastic dogs from the program) begins to act more like a
"simple" trait created by one or two gene pairs.
For
example, no matter what you do with environment, be it nutrition, exercise, or
anything else, you cannot change the phenotype of a black Labrador Retriever to a yellow Labrador Retriever. Heritability of coat color,
then, can be said to be 1.0 (100% controlled by the genes, zero by environment). The yellow Lab doesn't have any genes for
black and cannot produce black puppies if bred to another yellow Lab. The goal of the breeder seriously fighting HD
is to develop a line of dogs with so few bad hip genes that in a somewhat
similar manner, it cannot produce differently than their own phenotypes. At least, we should go as far in this
direction as is feasible. If you are
dealing with a trait or a breed with low heritability (maybe 0.1 to 0.2 or so),
selecting parental stock by phenotype (normal hips, in this context) will yield
slower progress in better hips in the first few generations. However, if you keep at the selection process
for normal hips, you will see progress in your line. Stick with dogs that have been selected for
normal phenotype in your own and others' lines and you will raise the
heritability in your stock above that of the breed in general. Do you want to lower the influence of
environmental factors such as overnutrition on your
production of dogs? You can do so by selecting
dogs that have hips that are relatively unaffected by such factors, which is
another way of saying you should choose dogs with higher heritability.
On the
other hand, the example used might not be the best, since it really does not
make a lot of sense to talk about heritability for a qualitative trait such as
coat color. For one reason, you would have a hard
time trying to measure color on an empirical scale in
many breeds such as the Bloodhound, Airedale, and German Shepherd,
all of which are saddle-marked dogs but with differences in expression of the
pattern. Such dogs are hard, if not impossible, to refer to as “36% black” or
“4% saddle”. Since there is no good numerical value measured, you would not be
able to compute the variances that form the mathematical expression for
heritability.
If all breeds are considered together, HD is a
less-heritable (and in most cases more polygenic) trait than many other
characteristics. So, compared to those
others, we would expect to see a less-rapid change in frequency when using
strictly phenotypic breeding. This includes X-ray pictures as well as regular
visual signs and traits. When we speak of phenotype in the context of a
discussion on HD, we are referring to the evaluation of the “standard” pelvic
radiograph. If we say that a breed or
population within a breed has a heritability of 0.25, we are saying that the
difference in genes between two dogs is responsible for 25% of the x-ray
picture differences between them. That
means that differences in environments (feeding practices, road work, etc.)
account for the other 75% of the phenotypic differences as seen on the
radiographs.
One reason for the fact that heritabilities can vary from one breed to another is the
history of intense linebreeding/inbreeding that is
most often used to found a breed. Look at the books on your own favorite breeds, and you will find a very few individuals
in all of the early pedigrees. A natural consequence of such inbreeding is that
there will be differences between breeds in regard to genetic variation for the
same trait. In dairy cattle, heritabilities for
traits such as volume of milk or level of butterfat produced are quite similar,
but that is because the major dairy breeds, while heavily selected, have not
undergone the same kind of pedigree-linebreeding that
many dog breeds have. The Border Collie, as soon as it
was “accepted” by the AKC as a “showdog”, was doomed
to eventual loss of abilities, the same as happened to many a Spaniel, Setter,
Shepherd, and “Non-Sporting” breed. Once breeding is based on pedigrees and a
narrow slice of the phenotype spectrum, instead of production records or
verifiable working ability, the gene pool becomes more limited, perhaps very
much so.
The interplay of genetic and environmental components of
the expression of pelvic phenotype is complex and understandably misinterpreted
by vets and breeders alike. But the
bottom line is still: environment may have
a great effect on the expression of genes an individual dog has inherited, but
no effect at all on the genes it will pass along to its progeny.
