H

Hardy–Weinberg Equilibrium and Random Mating

J Lachance, Georgia Institute of Technology, Atlanta, GA, USA
r 2016 Elsevier Inc. All rights reserved.

Glossary Genotype The pair of alleles present at a single locus.

Allele One of a number of alternative forms of a gene. Inbreeding Preferential mating between relatives.
Assortative mating Non-random mating on the basis of Natural selection Differential survival or reproduction.
phenotype. Phenotype The set of an organism’s observable traits.
Fitness The expected contribution of an individual or Ploidy The number of sets of chromosomes in a cell.
genotype to next generation’s gene pool. Single nucleotide polymorphism (SNP) DNA sequence
Genetic drift Changes in allele or genotype frequencies variation at a single nucleotide position that differs between
due to random sampling. individuals.

History equilibrium principle for a single locus with two segregating

alleles (Weinberg, 1908).
Darwin’s theory of evolution transformed biology in the mid-
nineteenth century. However, Darwin had an inaccurate
understanding of heredity and his belief in blending inherit- The Hardy–Weinberg Principle
ance was problematic. Because blending inheritance ultimately
results in homogenized populations full of intermediate The Hardy–Weinberg principle relates allele frequencies to
genotypes, it is unable to explain how genetic variation can genotype frequencies in a randomly mating population. Im-
persist over evolutionary time (Charlesworth and Charles- agine that you have a population with two alleles (A and B)
worth, 2009). It wasn’t until the rediscovery of Mendelian that segregate at a single locus. The frequency of allele A is
genetics in the early twentieth century that a potential solution denoted by p and the frequency of allele B is denoted by q. The
could be found. Importantly, Mendel treated genes as discrete Hardy–Weinberg principle states that after one generation of
units, which enabled a mathematical formulation of popu- random mating genotype frequencies will be p2, 2pq, and q2. In
lation genetics. Despite the rediscovery of Mendelian genetics the absence of other evolutionary forces (such as natural se-
and the pioneering work of biologists and statisticians (Castle, lection), genotype frequencies are expected to remain constant
1903; Pearson, 1904) there was still some confusion, and and the population is said to be at Hardy–Weinberg equilibrium.
during a 1908 meeting of the Royal Society of Medicine it was The Hardy–Weinberg principle relies on a number of as-
claimed that dominant mutations would increase in frequency sumptions: (1) random mating (i.e, population structure is
until phenotypes reached 3:1 Mendelian ratios (Provine, absent and matings occur in proportion to genotype fre-
2001). Something about this claim seemed wrong to R.C. quencies), (2) the absence of natural selection, (3) a very large
Punnett, who passed on the question to his cricket-playing population size (i.e., genetic drift is negligible), (4) no gene
friend G.H. Hardy. Hardy was one of the leading mathemat- flow or migration, (5) no mutation, and (6) the locus is
icians in Britain and he quickly deduced a solution. In a 1908 autosomal. When these assumptions are violated, departures
letter to the journal Science, Hardy began by writing “I should from Hardy–Weinberg proportions can result.
have expected the very simple point that I wish to make to One useful way to think about the Hardy–Weinberg prin-
have been familiar to biologists” before using the binomial ciple is to use the metaphor of a gene pool (Crow, 2001).
expansion to demonstrate that genotype frequencies will reach Here, individuals contribute alleles to an infinitely large pool
a stable equilibrium after one generation of random mating of gametes. In a randomly mating population without natural
(Hardy, 1908). Unbeknownst to Hardy, a Germany physician selection, offspring genotypes are found by randomly sam-
named Wilhelm Weinberg had published a similar paper on pling two alleles from this gene pool (one from their mother
the same topic six months earlier. Weinberg derived a general and one from their father). Because the allele that an

208 Encyclopedia of Evolutionary Biology, Volume 2 doi:10.1016/B978-0-12-800049-6.00022-6

 Hardy–Weinberg Equilibrium and Random Mating 209

individual receives from their mother is independent of the This occurs because q2 is much smaller than 2pq when q is
allele they receive from their father, the probability of ob- close to zero.
serving a particular genotype is found by multiplying maternal The Hardy–Weinberg principle can be generalized to in-
and paternal allele frequencies. Mathematically this involves clude polyploid organisms and genes that have more than two
the binomial expansion: (p þ q)2 ¼ p2 þ 2pq þ q2 (see the segregating alleles. Equilibrium genotype frequencies are
modified Punnett Square in Figure 1 for a graphical repre- found by expanding the multinomial (p1 þ … þ pk)n, where
sentation). Note that there are two ways that an individual can n is the number of sets of chromosomes in a cell and k is the
be an AB heterozygote: they can either inherit an A allele from number of segregating alleles. For example, tetraploid organ-
their mother and a B allele from their father or they can inherit isms (n ¼ 4) with two segregating alleles (k ¼ 2) are expected
a B allele from their mother and an A allele from their father. to have genotype frequencies of: p14 (AAAA), 4p13p2 (AAAB),
Additional insight can be found by considering an empir- 6p12p22 (AABB), 4p1p23 (ABBB), and p4 (BBBB). Similarly,
ical example (Figure 2). Consider a population that initially diploid organisms (n ¼ 2) with three segregating alleles
contains 18 AA homozygotes, 4 AB heterozygotes, and 3 BB (k ¼ 3) are expected to have genotype frequencies of: p12 (AA),
homozygotes. The alleles in the gene pool, 80% are A and 20% p22 (BB), p32 (CC), 2p1p2 (AB), 2p1p3 (AC), and 2p2p3 (BC).
are B. After a single generation of random mating we observe Genotype frequencies sum to one for each of the above
Hardy–Weinberg proportions: 16 AA homozygotes, 8 AB scenarios. Although the Hardy–Weinberg principle can also be
heterozygotes, and 1 BB homozygote. Note that allele fre- generalized to include genes located on sex chromosomes
quencies remain unchanged. (e.g., X chromosomes in humans), it is important to note that
There are a number of evolutionary implications of the it can take multiple generations for genotype frequencies at
Hardy–Weinberg principle. Most importantly, genetic vari- sex-linked loci to reach equilibrium values.
ation is conserved in large, randomly mating populations. A
second implication is that the Hardy–Weinberg principle
allows one to determine the proportion of individuals that are
carriers for a recessive allele. Third, it is important to note that Testing for Hardy–Weinberg Proportions
dominant alleles are not always the most common alleles in a
population. Another implication of the Hardy–Weinberg A Chi-square (χ2) test with one degree of freedom can be used
principle is that rare alleles are more likely to be found in to determine whether a population is at Hardy–Weinberg
heterozygous individuals than in homozygous individuals. equilibrium (Weir, 1996) (see Table 1 for an example). First,
the observed numbers of individuals with each genotype are
counted. These genotype frequencies are then used to obtain
allele frequencies (p and q). The expected numbers of indi-
viduals with each genotype are calculated by multiplying the
total number of individuals sampled by p2, 2pq, and q2. Once
observed and expected genotype counts are known, χ2 statistics
p p2 pq
Maternal
allele Table 1 Chi-square test of Hardy–Weinberg proportions. Allele
frequency frequencies are p ¼ 0.7 and q ¼ 0.3. The expected number of
individuals with each genotype are calculated by multiplying the total
number of individuals by p2 (AA), 2pq (AB), and q2 (BB). A χ2 of
q pq q2
8.724 with one degree of freedom yields a p-value of 0.00172,
indicating that there is a statistically significant departure from Hardy–
p q Weinberg proportions

Paternal allele frequency Genotype Observed Expected χ2

Figure 1 Graphical representation of the Hardy–Weinberg principle. AA 117 107.8 0.785
The frequency of A alleles is denoted by p and the proportion of B AB 74 92.4 3.664
alleles by q. AA homozygotes are represented by white, AB BB 29 19.8 4.275
heterozygotes by gray, and BB homozygotes by gold. Shaded areas Total 220 220 8.724
are proportional to the probability of observing each genotype.

Present generation Gene pool Next generation

72% 64%

16% 32%

12% p = 0.8 q = 0.2 4%

Figure 2 Hardy–Weinberg example. AA homozygotes (black circles), AB heterozygotes (black and gold circles), and BB homozygotes (gold
circles) contribute to the gene pool. A alleles are shown as black half-circles and B alleles are shown as gold half-circles. After a single generation
of random mating Hardy–Weinberg proportions are obtained.
210 Hardy–Weinberg Equilibrium and Random Mating

can be calculated for each genotype using the equation χ2 ¼ selection present. Selection results in departures from Hardy–
(observed  expected)2/expected. These χ2 values are summed, Weinberg proportions whenever genotypic fitnesses are non-
and if the overall χ2 test statistic is greater than 3.84 the null multiplicative (i.e., wAB2 a wAA  wBB) (Lachance, 2008). Not
hypothesis that the population is in Hardy–Weinberg equi- surprisingly, overdominance (heterozygote advantage) results
librium can be rejected (p-value o0.05). The power of this in an excess of heterozygotes compared to Hardy–Weinberg
statistical test is directly related to sample size: departures from expectations, and underdominance (heterozygote disadvan-
Hardy–Weinberg proportions are unlikely to be statistically tage) results in an excess of homozygotes. Strong directional
significant if you have genotype frequency data from a small selection, such as when one allele is a recessive lethal, leads to
number of individuals. marked departures from Hardy–Weinberg expectations.
However, weak directional selection has only modest effects
on genotype frequencies, and detecting these effects can re-
Departures from Hardy–Weinberg Proportions quire sample sizes that are larger than 10 000 individuals
(Lachance, 2009). Note that the effects of natural selection
There are a number of reasons why one might fail to observe tend to be locus-specific rather than genome-wide.
Hardy–Weinberg proportions, and these departures from
Hardy–Weinberg equilibrium can involve either an excess of Other Causes
homozygotes or an excess of heterozygotes.
Additional population genetic phenomena that can result in
departures from Hardy–Weinberg proportions include genetic
Genotyping Error drift and mutation. In small populations the proportion of
individuals with each genotype often differs from p2, 2pq, and
While it may not be the most interesting scenario from an
q2. The idea here is that sampling effects due to genetic drift
evolutionary perspective, departures from Hardy–Weinberg
yield a slight excess of homozygotes or heterozygotes. These
proportions sometimes occur because of genotyping error.
effects are more pronounced in very small populations. This is
These errors can be due to low quality DNA samples, bio-
akin to what happens when you flip a coin a small number of
chemical artifacts, or human error (Pompanon et al., 2005). For
times and end up with an unequal number of heads and tails.
example, low-coverage DNA sequencing can cause genotypes to
Although mutation can lead to slight departures from Hardy–
be incorrectly inferred. To improve data quality, a common
Weinberg proportions, these departures are difficult to detect.
practice in genome-wide association studies is to filter out any
This is because mutation rates tend to be small (on the order
single nucleotide polymorphisms (SNPs) that show significant
of 10–8 mutations per base pair per generation in humans).
departures from Hardy–Weinberg proportions (Turner et al.,
Ultimately, departures from Hardy–Weinberg proportions
2011). However, this runs the risk of eliminating SNPs that
reveal the biological complexity of natural populations. By
have departures from Hardy–Weinberg proportions that are due
building upon the foundation of the Hardy–Weinberg prin-
to some other reason than genotyping error.
ciple, generations of population geneticists have been able to
construct more complex models of evolution. These math-
Non-Random Mating and Population Structure ematical models have led to an increased understanding of
how evolutionary forces shape natural patterns of genetic
Non-random mating leads to departures from Hardy–Wein- diversity.
berg proportions. For example, inbreeding and positive as-
sortative mating (where individuals prefer to mate with
phenotypically similar individuals) yield an excess of homo-
See also: Genetic Variation in Populations. Linkage Disequilibrium:
zygotes. By contrast, negative assortative mating (where op-
Population Genetics of Multiple Loci
posites attract and individuals prefer to mate with
phenotypically different individuals) results in excess of het-
erozygotes. Population structure also causes departures from
Hardy–Weinberg proportions. For example, consider what References
happens when samples are drawn from multiple populations
instead of a single randomly mating population. If these Castle, W.E., 1903. The laws of heredity of Galton and Mendel: And some laws
samples are pooled together and there are allele frequency governing race improvement by selection. Proceedings of the American Academy
of Arts and Sciences 39, 223–242.
differences between source populations the resulting mixture
Charlesworth, B., Charlesworth, D., 2009. Darwin and genetics. Genetics 183,
will have an excess of homozygotes. This reduction in het- 757–766.
erozygosity is known as the Wahlund effect (Wahlund, 1928). Crow, J.F., 2001. The beanbag lives on. Nature 409, 771.
In practice, departures from Hardy–Weinberg proportions due Hardy, G.H., 1908. Mendelian proportions in a mixed population. Science 28,
to non-random mating and population structure tend to be 49–50.
Lachance, J., 2008. A fundamental relationship between genotype frequencies and
genome-wide (i.e., their effects can be seen at multiple loci). fitnesses. Genetics 180, 1087–1093.
Lachance, J., 2009. Detecting selection-induced departures from Hardy−Weinberg
proportions. Genetics Selection and Evolution 41, 15.
Natural Selection Pearson, K., 1904. Mathematical contributions to the theory of evolution.
XII. On a generalised theory of alternative inheritance, with special
Natural selection modifies allele and genotype frequencies and reference to Mendel's laws. Philosophical Transactions of the Royal Society A.
these effects depend on both the magnitude and type of 53–86.
 Hardy–Weinberg Equilibrium and Random Mating 211

Pompanon, F., Bonin, A., Bellemain, E., Taberlet, P., 2005. Genotyping errors: Wahlund, S., 1928. Zusammensetzung von Populationen und
Causes, consequences and solutions. Nature Reviews Genetics 6, 847–859. Korrelationserscheinungen vom Standpunkt der Vererbungslehre aus betrachtet.
Provine, W.B., 2001. The Origins of Theoretical Population Genetics. Chicago: Hereditas 11, 65–106.
University of Chicago Press. Weinberg, W., 1908. Über vererbungsgesetze beim menschen. Molecular and
Turner, S., Armstrong, L.L., Bradford, Y., et al., 2011 Quality control procedures for General Genetics MGG 1, 440–460.
genome-wide association studies. Current Protocols in Human Genetics Weir, B.S., 1996. Genetic Data Analysis II: Methods for Discrete Population Genetic
Chapter 1: Unit1 19. Data. Sunderland, MA: Sinauer Associates.