The Hardy-Weinberg principle

Concept

The Hardy-Weinberg principle

T-046Home BU-204Threads information · evolution

Statement

Allele and genotype frequencies in an idealised population.

Why it matters

Hardy-Weinberg establishes the idealised null model against which every other force treated in this unit — genetic-drift, gene-flow, mutation-selection-balance, and non-random mating (inbreeding-heterozygosity) — is measured as a deviation. Without a precisely stated baseline for what "no evolutionary change" looks like at a single locus, none of these forces could be detected or quantified from real genotype data; the equilibrium is therefore less a description of any real population and more the essential measuring stick population genetics is built around.

It is also, notably, a purely mathematical consequence of random mating and Mendelian segregation, requiring no assumption about which alleles are advantageous — which is precisely what makes deviation from it such a clean diagnostic signal that something else (selection, drift, non-random mating, migration, or mutation) must be acting on the locus in question.

Hypotheses

The population is infinitely large.This eliminates genetic-drift by construction; a finite population would show random, sampling-driven frequency fluctuation from one generation to the next even with every other Hardy-Weinberg assumption satisfied exactly. Mating is random with respect to the locus in question.Random mating is what allows genotype frequencies to be computed as the simple product of allele frequencies (Step 2 below); non-random mating (assortative mating, or inbreeding-heterozygosity's mating among relatives) breaks this independence even while leaving allele frequencies themselves unaffected. No mutation, no migration, and no selection act on the locus.Each of these forces can change allele frequencies between generations on its own; genotype-frequency equilibrium (Step 4) follows from random mating alone in a single generation, but allele-frequency constancy generation after generation additionally requires all three of these forces, along with drift, to be absent throughout.

Proof

p+q=1

At a diploid, autosomal, two-allele locus, let \(p\) and \(q\) be the frequencies of alleles \(A\) and \(a\) respectively; since every allele at the locus is one or the other, the two frequencies must sum to exactly 1. A

(p+q)^2 = p^2 + 2pq + q^2 = 1

Under random mating (Hypotheses), gametes combine independently, so genotype frequencies are simply the product of allele frequencies: \(AA\) at \(p^2\), \(Aa\) at \(2pq\) (heterozygotes can arise two ways, \(A\) from one parent and \(a\) from the other, or vice versa), and \(aa\) at \(q^2\) — exactly the binomial expansion of \((p+q)^2\). A

p' = p, \quad q' = q

In the absence of selection, mutation, migration, and drift (Hypotheses), allele frequencies do not change between generations; applying Step 2 again to the following generation reproduces exactly the same genotype frequencies, an equilibrium that persists indefinitely under these idealised conditions. A

\text{A single generation of random mating brings genotype frequencies to Hardy-Weinberg proportions, even starting from arbitrary initial genotype frequencies.}

Because gamete formation under random mating depends only on the current allele frequency, not on the history of genotype frequencies that produced it, Step 2's derivation applies immediately to whatever allele frequency is currently present — genotype-frequency equilibrium is reached in one generation of random mating, even before allele frequencies themselves have had a chance to change at all. B

\text{Observed deviation of genotype frequencies from } p^2,\,2pq,\,q^2 \text{ (tested e.g. by a chi-squared goodness-of-fit test) is direct evidence that at least one Hardy-Weinberg assumption is violated at that locus.}

Because Steps 2–4 are a direct mathematical consequence of the Hypotheses alone, a statistically significant departure from the predicted proportions in real genotype data is diagnostic: it identifies that something (non-random mating, selection, drift, migration, or mutation) is acting at that specific locus in that specific population, without yet identifying which one. A

Result

p^2 + 2pq + q^2 = 1 \qquad (p+q=1)

Reading. Under random mating alone (in one generation) and in the further absence of selection, mutation, migration, and drift (across generations), allele and genotype frequencies at a locus remain constant indefinitely.

Scope. A single-locus, two-allele, diploid, autosomal, randomly mating idealisation; real populations violate at least one of the Hypotheses to some degree, which is exactly what makes deviation from the equilibrium a useful diagnostic tool rather than a limitation of the model.

Corollaries & converses

Deviation from Hardy-Weinberg genotype proportions is direct, quantitative evidence that genetic-drift, gene-flow, mutation-selection-balance, non-random mating (inbreeding-heterozygosity), or direct selection is acting at that specific locus.
Heterozygote frequency \(2pq\) is maximised at \(p=q=0.5\), a useful reference point when comparing observed heterozygosity across loci or populations.
The frequency of a rare recessive allele can be estimated directly from the observed frequency of the recessive homozygous phenotype, since \(q^2=\) observed frequency, without needing to directly identify or count heterozygous carriers in the population.

Fails without

Mating is non-random with respect to genotype (Hypotheses): if individuals preferentially mate with genetically similar (assortative mating) or related (inbreeding-heterozygosity's specific case) partners, gametes no longer combine independently, and genotype frequencies deviate systematically from \(p^2,\,2pq,\,q^2\) even while allele frequencies \(p\) and \(q\) themselves remain entirely unchanged.
Population size is finite rather than infinite (Hypotheses): genetic-drift alone, with no other force acting, would still cause allele frequencies to wander randomly generation to generation, violating the equilibrium's constancy (Step 3) even though random mating alone still gives Hardy-Weinberg genotype proportions each generation from whatever the population's current allele frequency happens to be (Step 4).

Common errors

Assuming Hardy-Weinberg equilibrium requires allele frequencies to be exactly 0.5 and 0.5; any fixed pair of frequencies \(p,q\) satisfying \(p+q=1\) constitutes a valid equilibrium, not only an even split.
Interpreting a population "in Hardy-Weinberg equilibrium" at one locus as a claim that no evolution is occurring in that species generally, rather than specifically at the one locus or trait actually being tested.
Forgetting that genotype-frequency equilibrium (Step 4) is reached in a single generation of random mating alone, while allele-frequency constancy generation after generation additionally requires the much stronger, less commonly true condition that none of selection, mutation, migration, or drift are acting (t3 Hypothesis).
Computing \(q\) incorrectly by taking the square root of an observed heterozygote frequency, rather than the observed homozygous recessive phenotype frequency (\(q=\sqrt{q^2}\), Corollaries).

Discussion

The mathematician G.H. Hardy and physician Wilhelm Weinberg independently derived this result in 1908, prompted partly by a debate over whether a dominant allele should be expected to increase in frequency over successive generations simply by virtue of being dominant. The derivation shows it does not: dominance alone has no effect on allele frequency under random mating absent other forces, a point Hardy himself reportedly regarded as a fairly straightforward piece of algebra rather than a deep biological insight, even though it became a foundational result of population genetics.

Extensions of the basic two-allele, single-locus result exist for multiple alleles at one locus, for sex-linked loci (where genotype proportions differ between the two sexes), and for multi-locus systems; the core logic of Steps 2–4 — random combination of gametes under the stated assumptions — generalises to each of these, though the specific resulting formulas differ from the simple \(p^2+2pq+q^2\) form.

Common misconception: that observing Hardy-Weinberg genotype proportions in a real population is direct proof that no evolutionary forces are acting on it at all. Because several different forces (e.g. balanced migration and selection in opposite directions) can, in principle, combine to leave genotype proportions looking approximately Hardy-Weinberg despite each individually acting, matching the predicted proportions is consistent with, but does not strictly prove, the complete absence of every underlying force.

Worked examples

\text{A recessive genetic disorder occurs in 1 in 10{,}000 live births in a large, randomly mating population.}

The observed disorder frequency is \(q^2 = 1/10{,}000 = 0.0001\); taking the square root, \(q=\sqrt{0.0001}=0.01\), so the recessive allele's frequency in this population is estimated at 1%, using only the observed homozygous recessive phenotype frequency (Corollaries). A

p = 1-q = 0.99, \qquad 2pq = 2(0.99)(0.01) = 0.0198

The carrier (heterozygote) frequency is therefore estimated at roughly 1.98%, or about 1 in 50 individuals — substantially more common than the roughly 1-in-10,000 affected individuals, illustrating how many more unaffected carriers than affected individuals typically exist for a rare recessive condition. A

q^2=0.0001 \Rightarrow q=0.01,\ 2pq\approx0.0198\ (\text{about 1 in 50 carriers})

Reading. Even a rare recessive disease, affecting only 1 in 10,000 individuals, corresponds to a substantially more common carrier frequency in the population — a direct, practically important consequence of the \(q^2\) versus \(2pq\) relationship in the Result.

Scope. This carrier-frequency estimation technique applies to any recessive condition whose homozygous phenotype frequency is known and whose locus can reasonably be assumed close to Hardy-Weinberg proportions in the population being studied.

Problems

In a population, 16% of individuals show a recessive phenotype. Assuming Hardy-Weinberg proportions, find the allele frequencies and the percentage of heterozygous carriers.
Solution
\(q^2=0.16\), so \(q=0.4\), and \(p=1-0.4=0.6\). Carrier frequency: \(2pq=2(0.6)(0.4)=0.48\), so 48% of the population are heterozygous carriers.
A researcher samples a wild population and finds genotype counts of 640 \(AA\), 320 \(Aa\), and 40 \(aa\) individuals (total 1000). Compute the observed allele frequencies, the Hardy-Weinberg-expected genotype counts, and state what a substantial discrepancy between observed and expected counts would indicate.
Solution
Observed allele frequencies (counting alleles directly): \(p = (2\times640+320)/2000 = 1600/2000=0.80\), \(q=1-0.80=0.20\). Expected Hardy-Weinberg counts: \(AA=p^2\times1000=640\), \(Aa=2pq\times1000=320\), \(aa=q^2\times1000=40\) — these match the observed counts exactly in this case, consistent with the locus being at Hardy-Weinberg equilibrium. A substantial discrepancy, had one been found, would indicate (Step 5 of the Proof) that at least one Hardy-Weinberg assumption is violated at this locus — e.g. non-random mating, selection, drift, or migration — without, on its own, identifying which.
Explain why a single generation of strictly random mating is enough to restore Hardy-Weinberg genotype proportions at an autosomal locus, even in a population that starts with genotype frequencies very different from \(p^2,2pq,q^2\), provided the population's allele frequency itself is \(p,q\).
Solution
Genotype-frequency equilibrium depends only on how gametes combine under random mating, not on the genotype-frequency history that produced the current allele frequency (Step 4 of the Proof): each gamete produced carries allele \(A\) with probability \(p\) and \(a\) with probability \(q\), regardless of which specific genotype it came from, so random combination of gametes under Step 2's logic immediately yields \(p^2,2pq,q^2\) genotype proportions in the very next generation, independent of the parental generation's starting genotype distribution.