Extensions of Hardy-Weinberg: multiple alleles, X-linked loci, and inbreeding

Intuition Beginner

Hardy-Weinberg equilibrium generalises beyond two alleles. The ABO blood group has three alleles — $I^A$, $I^B$, $i$ — and the same random-mating logic gives genotype frequencies from allele frequencies: with frequencies $p$, $q$, $r$ summing to 1, the six genotype frequencies are $p^2$, $q^2$, $r^2$, $2pq$, $2pr$, $2qr$. For genes on the X chromosome, males carry one copy and females two. Male allele frequencies equal the maternal allele frequencies from the previous generation, so the sexes oscillate toward a shared equilibrium over several generations rather than one. Inbreeding — mating between relatives — increases the fraction of homozygotes above the Hardy-Weinberg prediction, raising the chance that harmful recessive alleles are expressed. The inbreeding coefficient $F$ quantifies this excess.

Visual Beginner

Two panels. The left panel shows the ABO system as a ternary (De Finetti) diagram: allele frequencies $p$, $q$, $r$ at the vertices of an equilateral triangle, with genotype-frequency points lying on a surface inside. Each point represents a population's allele-frequency composition, and the Hardy-Weinberg surface is the curved manifold of genotype-frequency triples satisfying the multinomial-square relation. The right panel plots male and female allele frequencies at an X-linked locus across generations, starting from different initial values, converging with damped oscillations toward a common equilibrium.

Worked example Beginner

The ABO blood group in humans is the textbook case of multiple alleles at a single locus. Three alleles — $I^A$, $I^B$, and $i$ — produce four phenotypes: type A ($I^A{I}^A$ or $I^{A}i$), type B ($I^B{I}^B$ or $I^{B}i$), type AB ($I^A{I}^B$), and type O ($ii$). $I^A$ and $I^B$ are codominant with each other and both are dominant over $i$.

In a population sample the allele frequencies are $p=0.3$ for $I^A$, $q=0.1$ for $I^B$, $r=0.6$ for $i$. The six genotype frequencies under Hardy-Weinberg are

$$P(I^A{I}^A)=p^2=0.09,P(I^A{I}^B)=2pq=0.06,P(I^{A}i)=2pr=0.36,$$ $$P(I^B{I}^B)=q^2=0.01,P(I^{B}i)=2qr=0.12,P(ii)=r^2=\mathrm{0.36.}$$

The phenotype frequencies are: type A $=0.09+0.36=0.45$, type B $=0.01+0.12=0.13$, type AB $=0.06$, type O $=0.36$. These sum to 1, and each phenotype frequency is computable from the allele frequencies alone — the same logic as two-allele Hardy-Weinberg, extended to three alleles.

Check your understanding Beginner

Formal definition Intermediate+

Multi-allele Hardy-Weinberg

Fix a single autosomal locus with $k$ alleles $A_1,\dots ,A_k$ with frequencies $p_1,\dots ,p_k$ summing to 1 in a diploid population. Under the five Hardy-Weinberg assumptions (no selection, no mutation, no migration, random mating, infinite population), the genotype frequencies after one generation of random mating are

$$P(A_i{A}_i)=p_i^2(i=1,\dots ,k),P(A_i{A}_j)=2p_i{p}_j(i\ne j).$$

There are $k$ homozygote classes and $\left(\genfrac{}{}{0ex}{}{k}{2}\right)$ heterozygote classes, for $k+\left(\genfrac{}{}{0ex}{}{k}{2}\right)=k(k+1)/2$ genotypes total. The sum of all genotype frequencies is

$$\sum _{i=1}^k{p}_i^2+\sum _{i<j}2p_i{p}_j={\left(\sum _{i=1}^k{p}_i\right)}^2=1.$$

The expected homozygosity is ${\sum }_i{p}_i^2$, the probability that two randomly drawn alleles are the same. The expected heterozygosity (gene diversity) is $H_e=1-{\sum }_i{p}_i^2$, the standard summary of allelic diversity at a locus. For $k$ equally frequent alleles, $H_e=1-1/k$, increasing with allele count.

X-linked loci

For a locus on the X chromosome with two alleles $A$ and $a$, males (XY) are hemizygous and carry a single allele, while females (XX) carry two. Let $p_f^{(n)}$ and $p_m^{(n)}$ denote the frequency of allele $A$ in females and males at generation $n$. The recursion under random mating is

$$p_m^{(n+1)}=p_f^{(n)},p_f^{(n+1)}=\frac{p_f^{(n)}+p_m^{(n)}}{2}.$$

A male inherits his single X from his mother; a female inherits one X from each parent. In matrix form

$$\left(\begin{array}{c}{p}_f^{(n+1)}\\ p_m^{(n+1)}\end{array}\right)=\left(\begin{array}{cc}1/2& 1/2\\ 1& 0\end{array}\right)\left(\begin{array}{c}{p}_f^{(n)}\\ p_m^{(n)}\end{array}\right).$$

The transition matrix has eigenvalues ${\lambda }_1=1$ and ${\lambda }_2=-1/2$. The ${\lambda }_1$ eigenvector is $(1,1{)}^T$: both sexes converge to the same frequency. The mean allele frequency $\bar{p}=(2p_f+p_m)/3$ is invariant across generations. The sex difference $p_f^{(n)}-p_m^{(n)}$ decays by a factor $-1/2$ each generation, oscillating in sign and shrinking to zero. After $\sim 5$ generations the sex difference is negligible.

At equilibrium, both sexes carry allele $A$ at frequency $\bar{p}$. Female genotype frequencies are $P(AA)={\bar{p}}^2$, $P(Aa)=2\bar{p}\bar{q}$, $P(aa)={\bar{q}}^2$; males carry $A$ at frequency $\bar{p}$ and $a$ at $\bar{q}$. The equilibrium is the same Hardy-Weinberg proportions, but in females only — males remain hemizygous.

Inbreeding coefficient $F$

The inbreeding coefficient $F$ of an individual at a locus is the probability that the two gene copies it carries are identical by descent (IBD) — copies of the same ancestral allele, traced through the pedigree without mutation. Two alleles that are functionally identical but derived from different ancestral copies are identical by state (IBS) but not IBD.

In a population with allele frequency $p$ for allele $A$ and inbreeding coefficient $F$, the genotype frequencies are

$$P(AA)=p^2+Fpq,P(Aa)=2pq(1-F),P(aa)=q^2+Fpq.$$

The parameter $F$ interpolates between Hardy-Weinberg ($F=0$) and complete inbreeding ($F=1$, no heterozygotes). The excess homozygosity $Fpq$ in each homozygote class comes at the expense of the heterozygote class, which is reduced by $2Fpq$. The inbreeding coefficient can be estimated from observed genotype frequencies:

$$\hat{F}=1-\frac{H_{\text{obs}}}{2\hat{p}\hat{q}}.$$

Self-fertilisation. Under strict selfing (each individual mates with itself), $F$ increases from $F_n$ to $F_{n+1}=\frac{1}{2}(1+F_n)$, starting from $F_0=0$. Solving the recurrence gives $F_n=1-(1/2{)}^n$, approaching 1. After 10 generations of selfing, $F\approx 0.999$. Heterozygosity halves each generation.

Counterexamples to common slips

• Multiple alleles do not break Hardy-Weinberg. The equilibrium is the multinomial-square expansion of the allele-frequency vector for any $k$. The same one-generation reset holds.
• X-linked equilibrium is not one-step. The two-allele autosomal case reaches Hardy-Weinberg in a single generation; the X-linked case requires several generations because the male and female allele-frequency reservoirs are initially distinct and converge at rate $1/2$ per generation.
• Inbreeding does not change allele frequencies. The inbreeding coefficient $F$ redistributes alleles from heterozygotes to homozygotes without changing the allele frequency $p$. Inbreeding is not an evolutionary force on allele frequencies; it is a departure from Hardy-Weinberg genotype proportions.
• Identity by descent is not identity by state. Two alleles can be identical in molecular state (same nucleotide sequence) but descend from different ancestral copies. IBD requires a genealogical connection; IBS does not. The inbreeding coefficient $F$ measures IBD, not IBS.

Key theorem with proof Intermediate+

Theorem (Multi-allele Hardy-Weinberg equilibrium). Let a diploid population have $k$ alleles $A_1,\dots ,A_k$ at a single autosomal locus with frequencies $p_1,\dots ,p_k$ summing to 1. Under random mating, no selection, no mutation, no migration, and infinite population size, the genotype frequencies after one generation are

$$\boxed{P(A_i{A}_j)=\{\begin{array}{ll}{p}_i^2& i=j,\\ 2p_i{p}_j& i\ne j,\end{array}}$$

and the allele frequencies are unchanged.

Proof. Under random mating, each offspring gene copy is an independent draw from the gene pool with allele probabilities $(p_1,\dots ,p_k)$. For genotype $A_i{A}_i$, both draws yield $A_i$: probability $p_i\cdot p_i=p_i^2$. For genotype $A_i{A}_j$ with $i\ne j$, there are two ordered realisations ($A_i$ first, $A_j$ second, or vice versa): probability $2p_i{p}_j$. Allele frequency preservation:

$$p_i^{\prime }=p_i^2+\sum _{j\ne i}{p}_i{p}_j=p_i\left(p_i+\sum _{j\ne i}{p}_j\right)=p_i\cdot 1=p_i.$$

Stationarity follows as in the two-allele case. $\square$

Theorem (X-linked convergence). At an X-linked locus with two alleles, the sex-specific allele frequencies $(p_f^{(n)},p_m^{(n)})$ converge to $(\bar{p},\bar{p})$ where $\bar{p}=(2p_f^{(0)}+p_m^{(0)})/3$, with the deviation decaying as $(-1/2{)}^n$.

Proof. Diagonalise the transition matrix $M=\left(\begin{array}{cc}1/2& 1/2\\ 1& 0\end{array}\right)$. The eigenvalues are ${\lambda }_1=1$ and ${\lambda }_2=-1/2$. The ${\lambda }_1=1$ eigenvector is $(1,1{)}^T$ with eigenvalue $1$, so the component of the state vector along $(1,1{)}^T$ is preserved. The projection of $(p_f^{(0)},p_m^{(0)})$ onto this eigenvector gives the common limit: both coordinates converge to $\bar{p}=(2p_f^{(0)}+p_m^{(0)})/3$ (weighted by the female-to-male X-chromosome ratio $2:1$ in the population). The deviation from equilibrium along the ${\lambda }_2=-1/2$ eigenvector $(1,-2{)}^T$ decays as $(-1/2{)}^n$, oscillating in sign and halving in magnitude each generation. After $n$ generations:

$$p_f^{(n)}-p_m^{(n)}={\left(-\frac{1}{2}\right)}^n(p_f^{(0)}-p_m^{(0)}).$$

After 10 generations, the sex difference is $<0.1\%$ of its initial value. $\square$

Theorem (Inbreeding genotype frequencies). In a population with allele frequencies $p$ and $q=1-p$ at a biallelic autosomal locus and inbreeding coefficient $F$, the genotype frequencies are

$$\boxed{P(AA)=p^2+Fpq,P(Aa)=2pq(1-F),P(aa)=q^2+Fpq.}$$

Proof. With probability $F$ the two alleles are IBD, meaning they are copies of the same ancestral allele. Given IBD, both are $A$ with probability $p$ (giving genotype $AA$) and both are $a$ with probability $q$ (giving genotype $aa$). With probability $1-F$ the alleles are independent draws from the gene pool, giving Hardy-Weinberg frequencies. Therefore

$$P(AA)=F\cdot p+(1-F)\cdot p^2=p(F+(1-F)p)=p(p+F(1-p))=p^2+Fpq.$$ $$P(aa)=F\cdot q+(1-F)\cdot q^2=q^2+Fpq.$$ $$P(Aa)=(1-F)\cdot 2pq=2pq-2Fpq.$$

Sum: $p^2+Fpq+2pq-2Fpq+q^2+Fpq=p^2+2pq+q^2=1$. $\square$

The Wahlund effect

If a sample is pooled from two subpopulations with allele frequencies $p_1$ and $p_2$, each internally at Hardy-Weinberg, the pooled sample shows an apparent heterozygote deficit. With equal sample sizes the pooled allele frequency is $\bar{p}=(p_1+p_2)/2$, and the actual pooled heterozygote frequency is $\bar{H}=(2p_1{q}_1+2p_2{q}_2)/2$. The deficit relative to naive Hardy-Weinberg is

$$2\bar{p}\bar{q}-\bar{H}=\frac{1}{2}(p_1-p_2{)}^2\ge 0.$$

This is the Wahlund effect (Wahlund 1928): pooling differentiated subpopulations always produces an apparent heterozygote deficit, numerically identical to the signature of inbreeding. Distinguishing substructure from inbreeding requires sampling design or multi-locus data.

F-statistics

Wright's $F$-statistics partition the total heterozygote deficit into hierarchical components:

• $F_{IT}=(H_T-H_I)/H_T$: the overall inbreeding coefficient, measuring the deviation of individual genotypes from the total population.
• $F_{IS}=(H_S-H_I)/H_S$: the within-subpopulation inbreeding coefficient, measuring deviation of individual genotypes from their subpopulation.
• $F_{ST}=(H_T-H_S)/H_T$: the fixation index, measuring allele-frequency differentiation among subpopulations.

Here $H_I$ is the observed individual heterozygosity, $H_S$ is the expected heterozygosity within subpopulations (averaged across subpopulations), and $H_T$ is the expected heterozygosity under naive pooling. The three are related by

$$(1-F_{IT})=(1-F_{IS})(1-F_{ST}).$$

$F_{ST}$ ranges from 0 (no differentiation) to 1 (fixed differences). Empirical benchmarks: $F_{ST}<0.05$ indicates little differentiation, $0.05$–$0.15$ moderate, $0.15$–$0.25$ great, and $>0.25$ very great differentiation.

Exercises Intermediate+

Multiple alleles and X-linked equilibrium Master

Identity by descent and identity by state

Two alleles at a locus are identical by descent (IBD) if they are copies of the same ancestral allele, traced through the pedigree without intervening mutation. They are identical by state (IBS) if they carry the same molecular sequence, regardless of ancestry. IBD implies IBS, but not conversely: two alleles can independently mutate to the same nucleotide (homoplasy) or simply share the same ancestral state by chance.

The inbreeding coefficient $F$ is a probability of IBD. The relatedness coefficient $r$ between two individuals is twice their coefficient of coancestry — the probability that a random allele from one individual at a locus is IBD with a random allele from the other at the same locus. For parent-offspring: $r=1/2$. For full siblings: $r=1/2$. For half siblings: $r=1/4$. For first cousins: $r=1/8$.

Path analysis for $F$

Wright's (1922) path method computes the inbreeding coefficient of an individual from a pedigree. For individual $X$ with parents $A$ and $B$:

$$F_X=\sum _{\text{paths}}{\left(\frac{1}{2}\right)}^n(1+F_{\text{ancestor}})$$

where the sum is over all distinct paths from $A$ to $B$ through a common ancestor, $n$ is the number of individuals in the path (counting $A$, $B$, and the common ancestor), and $F_{\text{ancestor}}$ is the inbreeding coefficient of the common ancestor.

Example: offspring of first cousins. The parents share two common grandparents. Each path has $n=5$ individuals (parent, grandparent, other parent) through one grandparent. If the grandparents are not inbred:

$$F=2\times {\left(\frac{1}{2}\right)}^5=\frac{1}{16}=\mathrm{0.0625.}$$

This is the coefficient of inbreeding for the child of a first-cousin marriage.

Effective population size and inbreeding

In a finite population, random mating inevitably pairs related individuals over time. The rate of inbreeding per generation under the Wright-Fisher model is

$$\Delta F=\frac{1}{2N_e},$$

where $N_e$ is the effective population size — the size of an idealised Wright-Fisher population with the same rate of inbreeding as the real population. The inbreeding coefficient after $t$ generations is

$$F_t=1-{\left(1-\frac{1}{2N_e}\right)}^t.$$

Effective size is reduced below census size $N$ by unequal sex ratios ($N_e\approx 4N_m{N}_f/(N_m+N_f)$), variance in offspring number ($N_e\approx (4N-2)/({\sigma }_k^2+2)$ where ${\sigma }_k^2$ is the variance in offspring number), and fluctuating population size ($N_e\approx$ harmonic mean of $N_i$ across generations). A single bottleneck reduces $N_e$ dramatically and permanently elevates $F$.

Purging of deleterious alleles

Inbreeding increases homozygosity, which exposes recessive deleterious alleles to selection. In a population that inbreeds consistently over many generations, selection against homozygous deleterious genotypes can purge the deleterious alleles, reducing the genetic load. The efficacy of purging depends on the dominance coefficient $h$: fully recessive deleterious alleles ($h\approx 0$) are efficiently purged because they are almost entirely hidden in heterozygotes under random mating but fully exposed under inbreeding; partially dominant deleterious alleles ($h>0$) are already partially exposed to selection in outbred populations, so inbreeding provides less additional purging.

Purging is a double-edged process in conservation biology. Small populations inevitably accumulate inbreeding, which both exposes deleterious alleles (potentially purging them) and reduces fitness through inbreeding depression. Whether purging can rescue a small population depends on the initial genetic load, the dominance coefficients of the deleterious alleles, and the rate of inbreeding. Rapid inbreeding (e.g., a severe bottleneck) tends to reduce fitness faster than purging can compensate; slow inbreeding gives selection more time to act.

HWE testing: chi-square and exact tests

Testing conformity to Hardy-Weinberg proportions is a standard quality-control step in genetic studies. The chi-square goodness-of-fit test compares observed genotype counts $n_{AA}$, $n_{Aa}$, $n_{aa}$ (summing to $N$) with expected counts $E_{AA}=N{\hat{p}}^2$, $E_{Aa}=2N\hat{p}\hat{q}$, $E_{aa}=N{\hat{q}}^2$ where $\hat{p}=(2n_{AA}+n_{Aa})/(2N)$:

$${\chi }^2=\sum _g\frac{(n_g-E_g{)}^2}{E_g},\text{df}=1.$$

The test has one degree of freedom (three categories minus one constraint from the total minus one estimated parameter). Reject at $\alpha =0.05$ if ${\chi }^2>3.84$.

The chi-square test is unreliable when expected counts are small (any $E_g<5$). The exact test (Guo-Thompson 1992) computes the probability of the observed or more extreme genotype configurations under Hardy-Weinberg by enumerating all possible genotype tables with the same allele counts. The exact test is the standard for multi-allele loci and small samples.

Significant deviation from Hardy-Weinberg in a dataset can indicate genotyping error, population substructure (Wahlund effect), assortative mating, selection, or inbreeding. In genome-wide association studies (GWAS), departures from Hardy-Weinberg in controls are used as a quality-control filter to remove SNPs with probable genotyping artifacts.

Runs of homozygosity and autozygosity mapping

Runs of homozygosity (ROH) are long contiguous stretches of the genome where both chromosomes are identical. ROH arise when an individual inherits two copies of a chromosomal segment from a single recent ancestor — the segments are IBD, hence autozygous. The length of an ROH segment is inversely proportional to the number of generations to the common ancestor: segments from a recent ancestor are long (tens of megabases for first-cousin inheritance), while segments from a distant ancestor are short (kilobases). The distribution of ROH lengths across the genome provides a forensic record of an individual's demographic history.

Autozygosity mapping uses ROH to map recessive disease genes: if affected individuals in an inbred population share long ROH segments at a particular genomic region, the causative recessive variant is likely located within the shared segment. This approach has identified numerous disease genes in consanguineous families.

The genomic inbreeding coefficient $F_{\text{ROH}}$ is estimated as the total fraction of the autosomal genome contained in ROH above a specified length threshold. $F_{\text{ROH}}$ correlates with pedigree-based $F$ but captures realised autozygosity rather than expected autozygosity, and can detect ancient inbreeding that is not apparent from recent pedigrees.

Inbreeding, identity, and F-statistics Master

Genetic load

The genetic load $L$ of a population is the reduction in mean fitness relative to the theoretical maximum, caused by deleterious mutations segregating in the population:

$$L=\frac{w_{\max }-\bar{w}}{w_{\max }}.$$

Under mutation-selection balance, a population carries a load from recurrent deleterious mutations. Inbreeding increases the expression of recessive deleterious alleles, converting the masked load (hidden in heterozygotes) into the realised load (expressed as reduced fitness in homozygotes). The total load does not change with inbreeding — it is merely redistributed from the masked to the realised component.

Inbreeding depression is the reduction in mean fitness of inbred individuals relative to outbred individuals, measured as

$$\delta =1-\frac{{\bar{w}}_{\text{inbred}}}{{\bar{w}}_{\text{outbred}}}.$$

Inbreeding depression is observed in virtually every studied outbreeding species and is typically linear in $F$: $\delta \approx B\cdot F$, where $B$ is a constant that depends on the number and effect sizes of deleterious recessive alleles. Empirical estimates of $B$ in human populations are in the range 2–6 lethal equivalents per diploid genome — equivalent to each individual carrying 2–6 recessive alleles that would be lethal in homozygous combination.

Inbreeding in conservation biology

Small endangered populations face an inbreeding dilemma. As $N_e$ decreases, $F$ increases at rate $1/(2N_e)$ per generation, and inbreeding depression reduces fitness components (survival, fecundity, disease resistance). The minimum viable population size — the smallest $N_e$ at which a population can persist with acceptable extinction risk — is typically set at $N_e\ge 50$ in the short term (to limit inbreeding depression) and $N_e\ge 500$ in the long term (to maintain evolutionary potential). These thresholds (Franklin 1980) are guidelines; actual requirements depend on the species' life history, generation time, and genetic architecture.

Genetic rescue — introducing migrants from a different population to increase $N_e$ and reduce $F$ — can reverse inbreeding depression in small populations. The Florida panther (Puma concolor coryi) is the canonical example: the population declined to $\sim 25$ individuals with severe inbreeding depression (cryptorchidism, kinked tails, heart defects), and the introduction of 8 Texas panthers in 1995 produced a rapid increase in fitness and population size. Genetic rescue works by introducing new alleles that break up autozygous deleterious combinations, restoring heterozygosity.

Consanguinity and health

Consanguineous marriages (between second cousins or closer) are practiced in many human populations, particularly in the Middle East, North Africa, and South Asia. The offspring of first-cousin marriages have $F=1/16=0.0625$, corresponding to an approximately 2–3% increase in risk of serious congenital disorders above the baseline population risk (typically 2–3%). The excess risk is modest in absolute terms but detectable at the population level. The risk increases with closer relatedness: offspring of uncle-niece marriages have $F=1/8=0.125$.

Structured populations and coancestry

In a spatially structured population, the coancestry coefficient ${\theta }_{ij}$ — the probability that a random allele from individual $i$ and a random allele from individual $j$ are IBD — increases with geographic proximity and decreases with distance. Wright's isolation by distance model (Wright 1943) predicts that $F_{ST}$ between populations increases with geographic distance at a rate determined by the dispersal distance $\sigma$ and the effective population density. The relationship $F_{ST}\approx 1/(1+4\pi D_e{\sigma }^2)$ at equilibrium for a two-dimensional population provides a method for estimating dispersal from genetic data.

The structured coalescent extends the Kingman coalescent to subdivided populations: lineages coalesce within demes at rate $1/(2N_d)$ (where $N_d$ is the deme size) and migrate between demes at rate $m$ per generation. The coalescent rate within demes is elevated relative to the total population because lineages in the same deme are more likely to share a recent ancestor. The ratio of within-deme to total coalescent rates maps directly onto $F_{ST}$.

Connections Master

• Hardy-Weinberg equilibrium 19.02.01. This unit extends the two-allele Hardy-Weinberg result of the prerequisite unit in three directions: multiple alleles, sex-linked inheritance, and non-random mating (inbreeding). The two-allele autosomal case is the base upon which all three extensions build; the multi-allele generalisation replaces the binomial square $(p+q{)}^2$ with the multinomial square $(\sum p_i{)}^2$, the X-linked case replaces the single-generation reset with a multi-generation convergence, and the inbreeding extension replaces the genotype-frequency proportions $p^2:2pq:q^2$ with $p^2+Fpq:2pq(1-F):q^2+Fpq$.

• Wright-Fisher model and diffusion approximation 19.02.05. The effective population size $N_e$ introduced here governs the rate of inbreeding accumulation $\Delta F=1/(2N_e)$ in finite populations. The Wright-Fisher chain is the stochastic substrate on which inbreeding builds; the inbreeding coefficient $F$ at time $t$ in a finite population equals $1-(1-1/(2N_e){)}^t$, which is the same expression derived from heterozygosity decay in the Wright-Fisher model. The $F$-statistics partition the Wright-Fisher variance into within- and between-population components.

• Natural selection 19.03.01. Inbreeding exposes recessive alleles to selection, changing the dynamics of allele-frequency change. The selection coefficient against a deleterious recessive is effectively multiplied by $1+F$ in an inbred population, accelerating the removal of deleterious alleles (purging) but also reducing fitness through inbreeding depression. Mutation-selection balance in inbred populations shifts the equilibrium frequency of deleterious alleles downward.

• Genetic drift 19.04.01. Drift and inbreeding are two faces of the same finite-population process. In a Wright-Fisher population of size $N_e$, drift reduces heterozygosity at rate $1/(2N_e)$ per generation, and the inbreeding coefficient $F$ increases at exactly the same rate. $F_{ST}$ measures the proportion of total genetic variance attributable to between-population differentiation, which is the population-structure signature of drift acting independently in isolated subpopulations.

• Quantitative genetics 19.05.01. Inbreeding reduces the additive genetic variance $V_A$ within a population by converting heterozygous loci to homozygous ones, changing the variance components that underlie the breeder's equation $R=h^{2}S$. The dominance variance $V_D$ is particularly affected by inbreeding because it depends on heterozygote frequencies, which are reduced by $F$. The response to selection in an inbred population differs from that in an outbred population at the same allele frequencies.

• Speciation 19.06.01. $F_{ST}$ measures reproductive isolation between populations at the genetic level. As populations diverge (allopatric speciation), $F_{ST}$ increases from near zero (panmixia) toward 1 (fixed differences). The speciation process can be framed as the accumulation of $F_{ST}$ across the genome, with islands of high $F_{ST}$ marking loci under divergent selection or reduced gene flow. The genomic landscape of $F_{ST}$ is a primary tool in speciation genomics.

• Conservation biology 19.14.01. Inbreeding depression, effective population size, minimum viable population, and genetic rescue — all central concepts in conservation biology — are grounded in the $F$ coefficient and $F_{ST}$ framework developed here. The $50/500$ rule, ROH-based inbreeding estimation, and the design of captive-breeding programmes to minimise $F$ all apply the machinery of this unit directly.

Historical & philosophical context Master

Sewall Wright introduced the inbreeding coefficient $F$ in his 1922 paper Coefficients of inbreeding and relationship (Am. Nat. 56, 330–338) ^[Wright1922], developing the path method for computing $F$ from pedigrees. Wright's broader programme — the shifting balance theory (Wright 1931 Genetics 16) — used $F$ as the central parameter connecting individual inbreeding to population subdivision to species-level evolution. The $F$-statistics ($F_{IS}$, $F_{ST}$, $F_{IT}$) were formalised in Wright 1951 (Ann. Eugen. 15) as a hierarchical decomposition of inbreeding at individual, subpopulation, and total-population levels. $F_{ST}$ has since become the most widely used summary of genetic differentiation in population genetics.

The multi-allele Hardy-Weinberg generalisation is implicit in the work of both Hardy (1908) and Weinberg (1908) — the multinomial-square expansion is the natural extension of the binomial square — but was developed explicitly by Weinberg in his 1909 and 1910 papers on the inheritance of human traits with multiple allelic forms. The formal treatment of multiple alleles in the context of the ABO blood group awaited the identification of the ABO system by Landsteiner (1900) and the genetic analysis by Bernstein (1924–1930), who demonstrated that three alleles at a single locus explain the four blood-type phenotypes.

The X-linked equilibrium was first analysed by Jennings (1916) and independently by Robbins (1918), who derived the convergence rate of $(-1/2{)}^n$ for the sex difference. The result was refined by Wright (1933) and embedded in the broader framework of sex-linked selection by Haldane (1926).

The Wahlund effect was described by Wahlund (1928) in Hereditas ^{[Wahlund1928]}, though the underlying algebra is a straightforward consequence of the law of total variance applied to genotype frequencies. The effect is routinely encountered in human genetics when samples are drawn from ethnically or geographically heterogeneous populations.

The distinction between identity by descent and identity by state was formalised by Malecot (1948) in Les mathematiques de l'heredite, which recast Wright's path coefficients in probability-theoretic language. Malecot's coefficient of coancestry — the probability that two randomly chosen alleles, one from each of two individuals, are IBD — is the foundational quantity for all of modern quantitative genetics and pedigree analysis.

Runs of homozygosity as a tool for estimating individual inbreeding from genomic data were developed by Lander and Green (1987) in the context of linkage mapping and applied systematically by McQuillan et al. (2008) in Am. J. Hum. Genet. The approach has become standard with the advent of dense SNP arrays and whole-genome sequencing.

Philosophically, the inbreeding coefficient $F$ is a rare quantity in biology that connects an individual-level genealogical property (two alleles tracing to a common ancestor) to a population-level statistical property (heterozygote deficit). The dual interpretation — $F$ as a probability of IBD within an individual and $F$ as a measure of population structure — reflects the duality between the genealogical and frequency-based views of population genetics that runs through the entire field. The extension of Hardy-Weinberg to inbred populations shows that equilibrium is not a binary property (present or absent) but a continuum indexed by $F$, with random mating at one endpoint ($F=0$) and complete inbreeding at the other ($F=1$).

Bibliography Master

@article{Wright1922,
  author    = {Wright, S.},
  title     = {Coefficients of Inbreeding and Relationship},
  journal   = {American Naturalist},
  year      = {1922},
  volume    = {56},
  pages     = {330--338},
}

@article{Hardy1908,
  author    = {Hardy, G. H.},
  title     = {Mendelian Proportions in a Mixed Population},
  journal   = {Science},
  year      = {1908},
  volume    = {28},
  pages     = {49--50},
}

@article{Weinberg1908,
  author    = {Weinberg, W.},
  title     = {{\"U}ber den Nachweis der Vererbung beim Menschen},
  journal   = {Jahreshefte des Vereins f{\"u}r vaterl{\"a}ndische Naturkunde in W{\"u}rttemberg},
  year      = {1908},
  volume    = {64},
  pages     = {369--382},
}

@article{Wahlund1928,
  author    = {Wahlund, S.},
  title     = {Zusammensetzung von Populationen und Korrelationserscheinungen vom Standpunkt der Vererbungslehre aus betrachtet},
  journal   = {Hereditas},
  year      = {1928},
  volume    = {11},
  pages     = {65--106},
}

@article{Bernstein1924,
  author    = {Bernstein, F.},
  title     = {Ergebnisse einer biostatistischen zusammenfassenden Betrachtung {\"u}ber die erblichen Blutstrukturen des Menschen},
  journal   = {Klinische Wochenschrift},
  year      = {1924},
  volume    = {3},
  pages     = {1495--1497},
}

@article{Jennings1916,
  author    = {Jennings, H. S.},
  title     = {The Numerical Results of Diverse Systems of Breeding},
  journal   = {Genetics},
  year      = {1916},
  volume    = {1},
  pages     = {53--89},
}

@article{Robbins1918,
  author    = {Robbins, R. B.},
  title     = {Applications of Mathematics to Breeding Problems {II}},
  journal   = {Genetics},
  year      = {1918},
  volume    = {3},
  pages     = {73--92},
}

@book{Malecot1948,
  author    = {Mal{\'e}cot, G.},
  title     = {Les math{\'e}matiques de l'h{\'e}r{\'e}dit{\'e}},
  publisher = {Masson},
  year      = {1948},
  address   = {Paris},
}

@article{Wright1931,
  author    = {Wright, S.},
  title     = {Evolution in {M}endelian Populations},
  journal   = {Genetics},
  year      = {1931},
  volume    = {16},
  pages     = {97--159},
}

@article{Wright1951,
  author    = {Wright, S.},
  title     = {The Genetical Structure of Populations},
  journal   = {Annals of Eugenics},
  year      = {1951},
  volume    = {15},
  pages     = {323--354},
}

@article{GuoThompson1992,
  author    = {Guo, S. W. and Thompson, E. A.},
  title     = {Performing the Exact Test of {H}ardy-{W}einberg Proportion for Multiple Alleles},
  journal   = {Biometrics},
  year      = {1992},
  volume    = {48},
  pages     = {361--372},
}

@article{McQuillan2008,
  author    = {McQuillan, R. and Leutenegger, A.-L. and Abdel-Rahman, R. and Franklin, C. S. and Pericic, M. and Barac-Lauc, L. and Smolej-Narancic, N. and Janicijevic, B. and Polasek, O. and Tenesa, A. and Macleod, A. K. and Farrington, S. M. and Dunlop, M. G. and Wilson, J. F.},
  title     = {Runs of Homozygosity in European Populations},
  journal   = {American Journal of Human Genetics},
  year      = {2008},
  volume    = {83},
  pages     = {359--372},
}

@book{HartlClark2007,
  author    = {Hartl, D. L. and Clark, A. G.},
  title     = {Principles of Population Genetics},
  publisher = {Sinauer Associates},
  edition   = {4},
  year      = {2007},
}

@book{Futuyma2017,
  author    = {Futuyma, D. J. and Kirkpatrick, M.},
  title     = {Evolution},
  publisher = {Sinauer Associates},
  edition   = {4},
  year      = {2017},
}

@book{Ewens2004,
  author    = {Ewens, W. J.},
  title     = {Mathematical Population Genetics {I}: Theoretical Introduction},
  publisher = {Springer},
  edition   = {2},
  year      = {2004},
}

@article{Franklin1980,
  author    = {Franklin, I. R.},
  title     = {Evolutionary Change in Small Populations},
  journal   = {Conservation Biology: An Evolutionary-Ecological Perspective},
  year      = {1980},
  pages     = {135--149},
  publisher = {Sinauer Associates},
}

@article{Landsteiner1900,
  author    = {Landsteiner, K.},
  title     = {Zur Kenntnis der antifermentativen, lytischen und agglutinierenden Wirkungen des Blutserums und der Lymphe},
  journal   = {Zentralblatt f{\"u}r Bakteriologie, Parasitenkunde und Infektionskrankheiten},
  year      = {1900},
  volume    = {27},
  pages     = {357--362},
}