Hardy-Weinburg Equilibrium and You


There seems to a good bit of disagreement among dog breeders about Hardy-Weinburg Equilibrium and how, or whether, this mathematical model of population genetics applies to purebred dogs.  George Padgett says it's not really applicable, but it's the best we have and we should use it anyway, even though it doesn't really work.  This is a rather iffy argument, it seems to me.  If it doesn't work, then it really doesn't, and trying to use it is likely to seriously mislead breeders.

 Jackie Isabell says it is applicable and can be used to give us reliable information. 

One can easily figure out where these authors are coming from.  Here's the thing:  in order to use HWE math, the following assumptions must be met:

    1.  Mating within the population must be random.

    2.  Natural / artificial selection must not be occurring.

    3.  Mutations must not be occurring.

    4.  There must be no immigration / emigration of individuals into or out of the population.

    5.  Genetic drift must not be occurring.

Now, do populations of purebred dogs meet these criteria?

No, they obviously don't.  But neither do natural populations.  But natural populations come close enough to allow the use of HWE math.  Do purebred dogs come "close enough"?

Isabell, for example, says yes -- because although mating within purebred dogs is not random, it may be random with respect to the trait in question, and that's what counts.  Similarly, although selection is occurring, she points out that it is usually not selection with regards to the trait in question.

Although she is right as far as that goes, Isabell is missing the fact that HWE also implicitly assumes that each individual in the population has an equal chance to reproduce -- that's what that random-mating requirement is really all about -- and in addition, she does not appear to notice that popular-sire syndrome produces very powerful genetic drift (changes in allele frequency due to chance).  In small populations, as for any of the rarer breeds, drift is usually powerful anyway, even disregarding popular-sire syndrome.

Willis, I think, gets closer when he suggests that you can always use HW math to give you a "snapshot" picture of the incidence of heterozygotes for particular traits in a breed in a specific year, but unless the breed is actually at equilibrium, the incidence of carriers will not remain constant over generations.  However, even this suggestion is put at question by Jerold Bell.

According to Dr. Jerold Bell (the article is "Epidemiological studies of inherited disorders" -- a copy can be found at http://www.papillonclub.org/PapillonHealth/Article-Epidemiological-Studies.html), this question has been answered by generational studies of genetic disorders in domestic animals and the answer to whether HWE calculations apply to domestic dogs is No.  The actual quote is "While it is recognized that the frequency of carriers of recessive defective genes will far exceed that of affected individuals, there is no mathematical relationship between the two in domestic animal breeding."

Bell is the single geneticist I respect most and though I'd be interested in his references, I take him at his word.

HWE does not apply to dogs and cannot be used as though it did.  Padgett's estimates of carrier frequency -- anybody's estimates if they are using HWE -- are almost certainly wrong.  Isabell is wrong.  Even Willis, by this statement, seems to be wrong.

Bell recommends using "population-wide genetic testing" and pedigree analysis to obtain accurate estimates of carrier frequencies in show dog populations.

If you're interested in the topic despite Bell's opinion, you may be asking yourself:

But what is Hardy-Weinburg Equilibrium, exactly?

Here's what all these authors are talking about.  Notice that HWE is a population genetics model, which may explain why genetic counselors and so forth may not be quite up on the details of how this model is put together and meant to be applied.

The Hardy-Weinburg principle is a mathematic model used to estimate allele and genotype frequencies in natural populations.  It states that allele and genotype frequencies remain constant over time in populations that meet the criteria listed above.  It's meant to be used to estimate genotype frequencies for simple traits in which genes have only two possible alleles (but can be expanded for use even if there are more than two alleles), and what dog breeders try to do, evidently in vain if you believe Bell, is use this model to estimate the number of heterozygotes (carriers) of a recessive trait that exist in a population.  It does not have anything to do with identifying which specific individuals are carriers, just with estimating the overall number of carriers for the whole population.  The math involved in this particular use of the principle is simple.

It works like this:

Let p be the frequency of the normal allele (A)

Let q be the frequency of the recessive abnormal allele (a)

Let P be the frequency of the homozygous dominant genotype (AA)

Let H be the frequency of the heterozygous (carrier) genotype (Aa)

Let Q be the frequency of the homozygous recessive genotype (aa)

Then, if HWE assumptions hold,

P = p2

H = 2pq

Q = q2

p + q = 1

Given the above, it's pretty easy to estimate everything if you can start out knowing the frequency of affected individuals in the population.

Suppose that an extensive survey shows that 2% of all the wolves in Alaska are pure white and that this color is believed to be a simple recessive in a two-allele one-gene system.  If 2% of wolves are white, what proportion of the wolves are carriers for this trait?

Q = q2 = 0.02    so    √q2 = √0.02    so q = √0.02     so    q = 0.14

Now we know the frequency of the recessive allele (a).  That doesn't yet give us the frequency of carriers in the population, but it will, because:

p + q = 1    so     p = 1 - q    so     p = 1 - 0.14    so    p = 0.86


H = 2pq    so     H = 2 (0.86) (0.14)    so    H = 0.24

and that is the carrier frequency for the population.  24% of the wolves in this population are carriers for the white color.  Notice that a small percentage of affected animals (2%) implies a quite large percentage of carriers (24%).  Provided that the population is in Hardy-Weinburg Equilibrium, which most natural populations are, that percentage will be the same for the next generation, and the next -- unless natural selection selects against this color trait, at which point the breed will no longer be at equilibrium and the frequency of the recessive allele q will start to decline.  Or selection favors the trait, in which case, of course, the frequency of q will start to increase.  Or the population of wolves deviates substantially from one or more of the other HWE criteria.

I used wolves for this example, of course, because they would count as a natural population and allow me to explain HWE without regards to any questions about whether any of this applies to domestic dogs.  Let me repeat:

"While it is recognized that the frequency of carriers of recessive defective genes will far exceed that of affected individuals, there is no mathematical relationship between the two in domestic animal breeding."