Hardy-Weinberg Equilibrium

The single idea to hold onto is that Hardy-Weinberg describes a population that is not evolving. Evolution means a change in allele frequencies across generations; Hardy-Weinberg is the baseline where those frequencies stay constant forever. It is a null model — a “nothing is happening” reference point — and it exists so you have something to compare a real population against.

That stillness is not automatic. A population holds at equilibrium only while five conditions are all met: no mutation, random mating, no gene flow (migration), a population large enough that genetic drift is negligible, and no natural selection. Four of these — no mutation, no gene flow, no drift, no selection — shut off the only things that can shift allele frequencies. The fifth, random mating, is doing a different job: it is what licenses the $p^2/2pq/q^2$ shuffle. Nonrandom mating rearranges which genotypes get built without, by itself, changing $p$ or $q$ one bit. Shut all five off and both levels hold steady — the alleles because nothing is pushing them, the genotypes because random mating keeps rebuilding them in the same proportions. The conditions are not a random list to memorize; each one names something specific that would disturb the picture.

When the conditions hold, two equations pin down the numbers. $$p+q=1$$ says the two allele frequencies add to one, and $$p^2+2pq+q^2=1$$ says the three genotype frequencies add to one. Keep those two jobs separate — $p$ and $q$ count alleles; $p^2$, $2pq$, and $q^2$ count genotypes — and the whole topic clicks into place. The real payoff comes later: if a population’s observed frequencies do not match what these equations predict, that mismatch tells you at least one condition is violated — which is a lead, not a verdict.

Most Hardy-Weinberg questions give you one number — usually the fraction of recessive individuals — and ask you to rebuild the rest. Walk the steps in order and the reason for each move is clear. The trick is knowing which quantity is an allele frequency and which is a genotype frequency, and never mixing them up.

1. Start from the recessive genotype to get $q^2$. Only one genotype is visible with certainty: the homozygous recessive one ($aa$), because that is the only way the recessive phenotype shows. Its frequency is $q^2$. So if 16% of a population shows the recessive trait, then $q^2 = 0.16$.
2. Take the square root to find $q$. $q$ is the frequency of the recessive allele, and $q^2$ is the frequency of the recessive genotype. Undo the square: $q = \sqrt{q^2}$. From $q^2 = 0.16$, you get $q = 0.4$. This is the step students skip — do not report $q^2$ as if it were $q$.
3. Use $p + q = 1$ to find $p$. There are only two alleles, so their frequencies must sum to one. Rearrange to $p = 1 - q$. With $q = 0.4$, you get $p = 0.6$. Never square anything here — $p$ and $q$ are allele frequencies and add to 1.
4. Build the genotype frequencies with $p^2 + 2pq + q^2 = 1$. Now assemble the three genotypes: homozygous dominant is $p^2 = 0.36$, heterozygous is $2pq = 2(0.6)(0.4) = 0.48$, and homozygous recessive is $q^2 = 0.16$. They add to $1.00$, which is your check that the arithmetic is sound.
5. Answer what was actually asked. “Carriers” means the heterozygotes, $2pq$ — not $q$. “Dominant phenotype” means $p^2 + 2pq$, because both genotypes show it. Read the question for whether it wants an allele frequency ($p$ or $q$) or a genotype frequency ($p^2$, $2pq$, $q^2$), and give that one.

The through-line: recessive genotype gives $q^2$, its root gives the allele frequency $q$, subtraction gives $p$, and the squared terms rebuild the genotypes. All of it assumes the population is at equilibrium — that the five conditions hold — which is exactly why the equations are allowed to work.

Quick reference card. For each term, read what it is and where students most often trip — the recurring theme is keeping allele frequencies ($p$, $q$) separate from genotype frequencies ($p^2$, $2pq$, $q^2$), and remembering that the model describes a population that is not evolving.

Two things trip up almost everyone on Hardy-Weinberg. The first is treating the five conditions as a random list of trivia to memorize. The second is sliding between allele frequencies and genotype frequencies as if $p$ and $q$ could stand in for $p^2$, $2pq$, or $q^2$. Get these two straight and the rest is arithmetic.

Each condition blocks something specific — but they are not all blocking the same kind of thing. Evolution is a change in allele frequencies, and only four things can change them: mutation (creates new alleles), gene flow (adds or removes alleles by migration), genetic drift (random loss in small populations), and natural selection (non-random survival and reproduction). The fifth condition, random mating, sits in its own category. Non-random mating — inbreeding, assortative mating — changes which genotypes get built out of the same allele pool. An inbred population can have far fewer heterozygotes than $2pq$ predicts while $p$ and $q$ sit at exactly the values they had last generation. Genotypes move; alleles do not. So the conditions are not arbitrary — but four of them define “nothing is changing the alleles” and one defines “the genotypes are being shuffled fairly.”

$p$ and $q$ are allele frequencies; $p^2$, $2pq$, $q^2$ are genotype frequencies. $p$ and $q$ count how common each allele is and must add to one: $p + q = 1$. The squared expression, $p^2 + 2pq + q^2 = 1$, distributes those alleles into genotypes under random mating. Mixing the two levels is the single most common error — for example, reporting $q$ when the question asked for the recessive phenotype ($q^2$), or reporting $q$ for the fraction of carriers ($2pq$).

The recessive genotype is your only direct read. You cannot see $p$ or $q$ directly, and you cannot separate homozygous dominants ($p^2$) from heterozygotes ($2pq$) by eye — both show the dominant phenotype. What you can count is the recessive phenotype, which equals $q^2$. That is why every calculation starts there: from $q^2$ take the square root for $q$, then $p = 1 - q$, then rebuild $p^2$ and $2pq$.

Deviation from Hardy-Weinberg is the whole point. The model matters less as a description of real life — almost no population meets all five conditions — and more as a ruler. When observed genotype frequencies differ from the $p^2/2pq/q^2$ prediction, you know at least one condition is broken — and that is exactly as far as the deviation carries you. To say the population is evolving, you have to name the culprit: mutation, gene flow, drift, or selection moves the alleles, but non-random mating alone leaves $p$ and $q$ untouched while still wrecking the genotype prediction. A deviation is a lead to chase, not a conclusion. Keep four ideas straight — the conditions are specific, not arbitrary; alleles versus genotypes; $q^2$ is your entry point; and deviation flags a broken condition you still have to identify — and Hardy-Weinberg stops feeling like a formula to memorize.