Allele and Genotype Frequencies

A population is a group of interbreeding individuals of the same species sharing a geographic area; its gene pool is the sum total of all alleles at all loci carried by all its members. At any single locus with two alleles, say $A_1$ and $A_2$, every diploid individual carries one of three genotypes: $A_1A_1$, $A_1A_2$, or $A_2A_2$. The genotype frequency is simply the proportion of individuals in the population carrying a given genotype — for example, if 320 of 1,000 individuals are $A_1A_2$ heterozygotes, the genotype frequency of $A_1A_2$ is 0.32. The allele frequency is a different and more fundamental quantity: the proportion of all gene copies at that locus, across the whole population, that are a particular allele. Because each diploid individual carries two copies, a population of $N$ individuals has $2N$ total allele copies at an autosomal locus. If $n_{11}$, $n_{12}$, and $n_{22}$ are the counts of the three genotypes ($n_{11}+n_{12}+n_{22}=N$), then the frequency of allele $A_1$, conventionally denoted $p$, is:

and the frequency of $A_2$, denoted $q$, is $q = 1-p$. This formula literally counts every $A_1$ allele: homozygotes contribute two copies each, heterozygotes contribute one. The same logic extends to loci with more than two alleles (e.g., the ABO blood group has three: $I^A$, $I^B$, $i$) and to sex-linked loci, where allele counting differs between the homogametic and heterogametic sex (in humans, males carry only one X-linked allele copy, so a population's X-linked allele frequency must be computed by weighting males and females appropriately). It is essential to keep genotype frequency and allele frequency conceptually distinct, because they answer different questions. Genotype frequency describes the makeup of individuals; allele frequency describes the makeup of the gene pool — the pool from which the next generation's gametes are drawn. Evolution, formally defined, is change in allele frequency across generations. A population can show shifting genotype frequencies (e.g., due to non-random mating) with allele frequencies remaining perfectly constant — this is why The Hardy-Weinberg Equilibrium as a Null Model carefully separates the two. Conversely, any true evolutionary change — by Genetic Drift, Natural Selection and the Concept of Fitness, mutation, or migration — must manifest, sooner or later, as a change in $p$ and $q$.

Step 1: Define the locus and enumerate the alleles. Every calculation starts by specifying a genetic locus (a gene or a marker such as a microsatellite or SNP) and listing every allele observed at it in the population. This is only possible because of prior Genetic Variation as the Raw Material of Evolution — without polymorphism, there is nothing to have a "frequency" of; a monomorphic locus has one allele at frequency 1.0 and is evolutionarily inert until new variation arises.

Step 2: Sample the population and genotype individuals. In practice, biologists cannot genotype every individual in a wild population, so they take a representative random sample. Historically this was done using visible phenotypic markers with known, simple genetics — human ABO blood types, industrial-melanism moth color morphs, or gel-electrophoresis banding patterns of enzyme variants (allozymes). Today it is done by direct DNA sequencing or PCR-based genotyping (e.g., restriction fragment length polymorphisms, microsatellite length variants, or single-nucleotide polymorphism arrays), which reveal genotype directly rather than inferring it from phenotype.

Step 3: Tabulate genotype counts, then convert to genotype frequencies. Divide each genotype count by total sample size $N$. These three genotype frequencies must sum to 1.

Step 4: Convert genotype frequencies to allele frequencies using the gene-counting formula above — each homozygote counts twice, each heterozygote once, all divided by $2N$. This conversion is unambiguous and always possible; the reverse direction (predicting genotype frequencies from allele frequencies alone) is not automatic — it requires an assumption about how gametes combine, which is exactly the assumption tested by Hardy–Weinberg.

Step 5: Repeat across generations (or across space) to detect change. A single frequency estimate is a snapshot. Evolutionary inference requires comparing frequencies across time (longitudinal sampling of the same population in successive generations) or reconstructing historical frequencies from preserved specimens, herbarium sheets, or ancient DNA. A statistically significant shift in $p$ from one generation to the next — beyond what sampling error alone would produce — is empirical evidence that evolution has occurred at that locus.

Step 6: Attribute the change to a mechanism. Once a frequency change is documented, the harder scientific work begins: distinguishing whether the shift was driven by selection (fitness differences among genotypes), drift (random sampling in a finite population), migration (gene flow bringing in different allele frequencies), or mutation (new alleles arising) — the four forces that are elaborated in the natural-selection and drift articles that follow this one.

Classical marker studies: peppered moths and blood groups. Long before DNA sequencing, biologists needed traits whose genotype could be inferred unambiguously from phenotype. Human ABO and MN blood groups, controlled by simple codominant/multiple-allele Mendelian systems, let population geneticists in the 1920s–1950s calculate allele frequencies across ethnic groups and geographic regions directly from serotyping data, generating some of the first quantitative human population-genetics datasets. Similarly, Bernard Kettlewell's mid-1950s mark-release-recapture studies of the peppered moth (Biston betularia) in polluted and unpolluted English woodlands measured the phenotype frequencies of melanic versus light morphs (controlled largely by a single dominant locus) before and during the decline of industrial pollution, documenting genotype-frequency shifts consistent with differential bird predation — a direct empirical link between measured allele-frequency change and a proposed selective mechanism.

Lewontin and Hubby's allozyme electrophoresis survey (1966). The single most influential experiment for establishing that natural populations harbor extensive, measurable genetic variation — the prerequisite for any frequency to be interesting at all — was Richard Lewontin and John Hubby's 1966 study of Drosophila pseudoobscura. Using starch-gel electrophoresis, which separates protein variants (allozymes) by their charge-dependent migration rate in an electric field, they assayed 18 enzyme-coding loci across wild-caught flies. They found that roughly 30% of loci were polymorphic (having no single allele at frequency above 0.99) and that a typical individual was heterozygous at roughly 12% of its loci. This was a landmark quantitative demonstration that gene pools are not homogeneous — they are seething with allelic diversity — a finding that reframed the entire question of how such variation is maintained (balancing selection versus neutral drift), a debate that still motivates much of molecular evolution.

Direct longitudinal frequency tracking: the Grants and Darwin's finches. Peter and Rosemary Grant's multi-decade study of medium ground finches (Geospiza fortis) on Daphne Major in the Galápagos measured beak-depth-associated allele frequencies (later tied directly to variants in genes such as HMGA2 and ALX1 through genomic sequencing) before and after the 1977 drought. Seed availability shifted toward larger, harder seeds; birds with deeper beaks survived disproportionately; and the population's mean beak depth — and the underlying allele frequencies at beak-development loci — shifted measurably in a single generation. This is one of the cleanest field demonstrations that allele-frequency change can be observed in real time, not merely inferred from static snapshots.

Sickle-cell allele and malaria: frequency as a signature of selection. Anthony Allison's studies in 1950s East Africa compared frequencies of the sickle-cell allele ($Hb^S$) across regions with differing malaria endemicity. He found $Hb^S$ allele frequencies as high as 0.10–0.20 in malaria-endemic regions of tropical Africa versus near 0.00 in malaria-free regions, and demonstrated that heterozygotes ($Hb^AHb^S$) showed markedly reduced parasitemia and clinical malaria severity compared to $Hb^AHb^A$ homozygotes. This geographic correlation between allele frequency and an environmental selective agent (malaria prevalence), combined with the fitness data, is a textbook case of using measured frequencies as direct evidence for the maintenance of a polymorphism by heterozygote advantage — a preview of concepts fully developed in the article on natural selection and fitness. Together, these studies establish the empirical backbone of the field: frequencies can be measured precisely, they vary geographically and temporally, and — when tracked longitudinally or correlated with environmental variables — their changes can be attributed to specific evolutionary mechanisms.

Example 1 — from a genetic cross-like sample to allele frequencies (MN blood group). The human MN blood group locus has two codominant alleles, $L^M$ and $L^N$, producing three unambiguous phenotypes/genotypes: MM, MN, NN. Suppose a sample of 200 individuals from a population yields:

Total $N = 200$, so total allele copies $= 2N = 400$. Count $L^M$ alleles: each MM contributes 2, each MN contributes 1.

Count $L^N$ alleles: each NN contributes 2, each MN contributes 1.

Check: $p + q = 0.58 + 0.42 = 1.00$. ✓ Genotype frequencies (a separate calculation, needed for later Hardy–Weinberg comparison) are simply counts divided by $N$: $f(MM)=0.34$, $f(MN)=0.48$, $f(NN)=0.18$. Note that the heterozygote frequency (0.48) is not simply $2pq$ by definition — whether it matches $2pq = 2(0.58)(0.42) = 0.4872$ is precisely the empirical test that the Hardy–Weinberg model performs, foreshadowed here but not concluded.

Example 2 — an asymmetric, real-world clinical/population scenario: cystic fibrosis carrier frequency. Cystic fibrosis (CF) is an autosomal recessive disorder. In populations of Northern European descent, roughly 1 in 2,500 newborns is affected (genotype $aa$, homozygous recessive). Suppose you are asked: what proportion of the unaffected population are silent heterozygous carriers, and how does the carrier allele's frequency compare to the affected-genotype frequency? Assuming the population is approximately at Hardy–Weinberg proportions (a working assumption we can make here, formally justified only in the next article), the affected genotype frequency equals $q^2$:

So the disease allele frequency is $q = 0.02$ (2%) — already a striking result: a rare disease phenotype (1/2500) corresponds to a not-so-rare allele frequency (1/50), because recessive alleles hide in heterozygotes. Then $p = 1 - q = 0.98$, and the heterozygote (carrier) genotype frequency is:

This asymmetry — a disease affecting 1 in 2,500 people but with roughly 1 in 26 people silently carrying the allele — is the single most important quantitative lesson of recessive-genetics population screening, and it depends entirely on correctly distinguishing allele frequency ($q=0.02$) from genotype frequency ($q^2=0.0004$ for the affected class, $2pq\approx0.039$ for carriers). Genetic counselors use exactly this arithmetic to estimate a client's prior probability of being a carrier before confirmatory testing.

The mathematical formalization of allele frequency as evolution's fundamental variable is largely credited to three architects of the Modern Synthesis and population genetics: Ronald Fisher, Sewall Wright, and J. B. S. Haldane. Among them, Sewall Wright stands out for explicitly modeling gene pools as frequency distributions subject to sampling and drift, work that directly required the counting framework this article describes. Wright trained as an animal breeding geneticist at the USDA before turning to theoretical population genetics; unlike Fisher, who emphasized large populations and near-deterministic selection, Wright was fascinated by small, semi-isolated populations and the random component of allele-frequency change — what became his shifting balance theory. His mathematical treatment required, as a first step, an unambiguous, countable definition of allele frequency in finite populations — precisely the gene-counting apparatus formalized here. Wright also pioneered path-coefficient analysis and inbreeding coefficients, tools still used to correct allele-frequency estimates for non-random mating.

Clinical genetics and carrier screening. The allele-frequency-to-genotype-frequency arithmetic worked out above (cystic fibrosis) is used daily in prenatal and preconception carrier screening panels, which report population-specific carrier frequencies for dozens of recessive conditions (Tay-Sachs, sickle-cell disease, spinal muscular atrophy) so that couples can assess reproductive risk before conception.

Conservation genetics. Wildlife managers estimate allele frequencies and heterozygosity in endangered populations (cheetahs, Florida panthers, black-footed ferrets) from tissue or scat DNA to quantify loss of genetic diversity, detect inbreeding depression risk, and design supplementation or translocation programs that restore allele frequencies toward historically healthier baselines.

Agricultural and livestock breeding. Marker-assisted selection programs track allele frequencies at loci for disease resistance, milk yield, or herbicide tolerance across breeding generations, allowing breeders to quantify genetic gain per generation using exactly the gene-counting formalism introduced here.

Public health and pathogen surveillance. Tracking allele (or variant) frequencies of antibiotic-resistance genes in bacterial populations, or of spike-protein mutations across SARS-CoV-2 lineages, uses the identical mathematics — frequency of a variant in a sampled population over time — to detect the rise of resistant or immune-evasive strains in near-real time, informing treatment guidelines and vaccine updates.

Forensic and population-identity genetics. DNA profiling databases (CODIS in the United States) rely on measured allele frequencies at short tandem repeat loci across reference populations to calculate the statistical rarity of a DNA match, directly underpinning courtroom probability statements.

Ask the tutor: How does the Hardy–Weinberg equilibrium use these allele frequencies to predict expected genotype frequencies, and what does a deviation from that prediction actually tell us?