656 Chapter 19 Variation and Selection in Populations

Figure 19.1 Tuberculosis in human populations. (a) Photo- One way to answer these questions is to examine
graph of M. tuberculosis bacteria colonizing the lungs and bones. (b) Death genetic variation and its expression as phenotypes
rate from tuberculosis in the United States from 1900–2000. within populations of organisms. The scientific disci-
(a) pline that studies what happens in whole populations at
the genetic level is known as population genetics. It
encompasses the evolutionary ideas of Darwin, the laws
of Mendel, and the insights of molecular biology.
In this chapter, we explore the nature of genotypic
and phenotypic variation within populations and the
role of this variation in evolution. We know from
Chapter 2 that variation exists in all populations. To
begin, we analyze the incidence of genetic diseases,
such as cystic fibrosis and retinoblastoma, that are
determined by a single gene. Through our analysis, we
develop a framework for understanding how the fre-
quency of a disease-causing allele determines the fre-
quency of diseased individuals in a population. We next
examine the effects of population size and chance on
changes in allele frequency.
Finally, we consider variation in multifactorial
traits, that is, in traits determined by two or more genes
(b) Death Rate from Tuberculosis in the United States
and their interaction with the environment (see Chap-
200 ter 3). In fact, most aspects of disease susceptibility
are multifactorial.
Deaths per 100,000 people

175
One general theme emerges from our discussion:
150
Population geneticists rely on mathematical models in
125
predicting a population’s potential for stasis or change
100 because most of the scientifically useful questions they
75 ask are statistical. Simple mathematical models not only
clarify the questions about frequency of genetic diseases
50
or rate of spread of pathogens, but, they also serve as
25
10 tools for analyzing data and making predictions about
0 future populations.
1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000
Year
Prior to 1933 data are for only areas with death-registration; after 1933 data are for entire U.S.

19.1 The Hardy-Weinberg Law: Population geneticists describe

Predicting Genetic Variation in populations with well-defined terms
Populations To population geneticists, a population is a group of inter-
breeding individuals of the same species that inhabit the
Modern genetics began with Mendel’s elucidation of for- same space at the same time. An example would be all the
mal rules of probability that describe the likelihood of white-tailed deer on Angel Island in San Francisco Bay in
transmission of genes and traits from parents to offspring 1990 or all the rock cod at the mouth of the bay. The sum
in controlled breeding experiments. In this section, we total of all alleles carried in all members of a population is
describe an extension of Mendel’s work that provides that population’s gene pool. In nature, the genetic makeup
researchers with genetic tools for predicting transmission of a population changes over time as new alleles arise by
frequencies of traits and alleles in natural populations mutation or are introduced by immigration and as rare, pre-
having an unlimited size. existing alleles disappear when all individuals carrying
 19.1 The Hardy-Weinberg Law: Predicting Genetic Variation in Populations
them leave the population or die. Changes in the frequency
of alleles within a population are the basis of microevolu-
tion: alterations of a population’s gene pool.
Suppose you were to look at a human population of 20
in which 4 people have blue-colored eyes because they are
homozygous for the B allele at a particular “blue eyes” locus,
where the alternative allele is A. To predict how the number
of blue-eyed individuals in the population will change over
time, you need to determine the frequencies of each geno-
type (homozygous AA, heterozygous AB, and homozygous
BB), each phenotype (dark eyes and blue eyes), and each
allele (A and B). Population geneticists define phenotype
frequency as the proportion of individuals in a population
that express a particular phenotype. For our hypothetical
population, the phenotype frequencies are 4/20 = 20% blue
eyes (the number of homozygous BB individuals expressing
the recessive trait) and 16/20 = 80% dark-eyed (the remain-
ing fraction with either AA or AB genotypes).
Genotype frequencies
Genotype frequency is the proportion of total individuals
in a population that carry a particular genotype. To deter-
mine genotype frequencies, you simply count the number
of individuals of each genotype and divide by the total
number of individuals in the population (Fig. 19.2a and b).
For recessive traits such as blue eyes, it is not possible to
distinguish between homozygous dark eyes and heterozy-
gous genotypes: Both give rise to individuals with dark
eyes. Thus, the only way to determine genotype frequen-
cies directly is to use a molecular assay that distinguishes
between the different alleles. For our hypothetical popula-
tion, molecular analyses showed that 12 individuals (of 20)
are of genotype AA, 4 are of genotype AB, and 4 are BB.
This means that the AA genotype frequency is 12/20 5 0.6;
the AB genotype frequency is 4/20 5 0.2; and the BB geno-
type frequency is also 0.2. Note that these three frequencies
(0.6 1 0.2 1 0.2) sum to 1, the totality of genotypes in
the whole population.
Allele frequency
The definition of allele frequency is the proportion of
gene copies in a whole population that are of a given
allele type. (Initially, population geneticists used the term
“gene frequency” to describe what is now more accu-
rately called “allele frequency.” Because each individual
in a population has two copies of each chromosome, the
total number of gene copies is two times the number of
individuals in the population. Thus, for our hypothetical
population of 20 people, there would be 40 gene copies
or chromosomes. Of course, both homozygotes and
heterozygotes contribute to the frequency of an allele.
But homozygotes contribute to the frequency of a par-
ticular allele twice, while heterozygotes contribute only
once (Fig. 19.2b). To find the frequencies of A and B,
657
you first use the number of people with each genotype to
compute the number of A and B alleles.
12 AA n 24 copies of A
4 AB n 4 copies of A
4 BB n 0 copies A
Together, 24 1 4 1 0 5 28 copies of the A allele.
Similarly,
12 AA n 0 copies of B
4 AB n 4 copies of B
4 BB n 8 copies of B
Together, 0 1 4 1 8 5 12 copies of the B allele.
Next, you add the 28 A alleles to the 12 B alleles to find
the total number of chromosome copies.
28 1 12 5 40 copies of the gene, which is twice the
number of people in the population
Finally, you divide the number of each allele by the total
number of gene copies to find the proportion, or fre-
quency, for each allele.
For the A allele, it is 28/40 5 0.7
For the B allele, it is 12/40 5 0.3
Note that here again, the frequencies sum to a 1, repre-
senting all the alleles in the gene pool.
Gene pool
A gene pool is not a real material object. It is, rather, a
conceptual term used by population geneticists. A gene
pool represents all of the alleles present on the chromo-
somes of all members of a population and the relative
prominence or rareness of each allele. Although an indi-
vidual diploid organism can carry, at most, two alleles at
a locus, a whole population of N individuals could, in
theory, have a locus-specific gene pool of up to 2N alleles
(Fig. 19.2a). In reality, the allele number is much smaller
than 2N because nearly all children inherit unchanged
alleles from their parents. A population is defined by its
allele frequencies and genotype frequencies, which together
make up a gene pool. The precise frequency measurements
of a gene pool are fleeting, constantly changing over time
as population components (that is, individual organisms)
come into, or pass out of, existence.
The human cystic fibrosis transmembrane receptor
(CFTR) locus provides an interesting example of a gene
pool that is broad—with many disease-causing alleles—
but shallow in the sense that none of these alleles are present
at a very high frequency. Over 1600 disease-causing muta-
tions have been identified in the genomes of individuals
who express cystic fibrosis (Fig. 19.2c). The combined
frequency of all disease-causing alleles is 0.02, which is
quite small compared to the 0.98 frequency of the functional
class of CFTR alleles.
658 Chapter 19 Variation and Selection in Populations

Figure 19.2 From genotype frequencies to allele frequencies to genotype frequencies. (a) A first-generation population of
20 individuals who are each homozygous or heterozygous for alleles A and/or B at a locus. (b) Whole population numbers for genotypes and
alleles at single locus of interest. (c) Gene pool of alleles at the CFTR locus in European-American populations. The structure of the CFTR gene
is not drawn to scale; introns are much larger proportionally than shown. At the level of DNA sequence, thousands of nonfunctional biallelic SNP
and InDel variants have been identified at this locus like all others (locations shown in the Seq. Var. row). Nonfunctional alleles are generally
ignored in studies focused on disease phenotypes. The rows labeled missense, nonsense, frameshift, in-frame in/del, splicing, and promoter
show the locations of mutations that affect each of these aspects of gene expression and CFTR protein structure and function.
(a) (b)

(c)

One particular mutant allele, called DF508 because it
deletes amino acid codon number 508, a phenylalanine
(F) in normal CFTR polypeptides, accounts for approxi-
mately 70% of the total disease-causing chromosomes.
This value corresponds to an individual allele frequency
of 0.014; no other cystic fibrosis disease allele has a fre-
quency of greater than 0.001. In contrast, other loci, like
the genetic determinant of eye color that we will discuss
shortly, have allelic forms that predominate in some
human populations but are absent from others.
The scientific endeavor of genetics can be divided
into subfields based on the unit object that is the focus of
a geneticist’s attention. To molecular geneticists, the unit
entity is “the gene.” To formal geneticists, the unit entity
is the individual organism, which is defined by genotypes.
To population geneticists, the unit entity is the population
consisting of a group of interbreeding organisms.
Although an actual gene pool can only be determined
empirically by counting all of its constituent alleles (which
is nearly impossible to do for a large wild population),
population geneticists have developed analytical and
computational models for estimating the genetic and phe-
notypic variations of a population and how they may
change over time. The foundation for all of the more
sophisticated models lies within the Hardy-Weinberg law
that we discuss next.
The Hardy-Weinberg law is a binomial
equation that correlates allele and
genotype frequencies
Many rare diseases result from recessive disease-causing
alleles. For scientists seeking to predict the potential inci-
dence of such recessive conditions, an important question
is, How common are the heterozygous carriers of the
disease-causing allele in a population? At the start, the
scientists know only the phenotypic frequencies of healthy
 19.1 The Hardy-Weinberg Law: Predicting Genetic Variation in Populations
and diseased individuals. Can they use this information to
predict the frequency of heterozygous carriers?
The key to answering this question lies in establishing
a quantitative relationship between phenotype, genotype,
and allele frequencies within a population. The Hardy-
Weinberg law, named for the two men—G. H. Hardy and
W. Weinberg—who independently developed it in 1908,
clarifies the relationships between genotype and allele fre-
quencies within a generation and from one generation to
the next. The derivation of this general law requires cer-
tain simplifying assumptions:
1. The population is composed of a very large number
of individuals that, for all intents and purposes, is
infinite.
2. An individual’s genotype at the locus of interest has
no influence on his or her choice of a mate—that is,
mating is random.
3. No new mutations appear in the gene pool.
4. No migration takes place into or out of the
population.
5. Different genotypes at the locus of interest have no
impact on the ability to survive to reproductive age
and transmit genes to the next generation.
The assumptions behind the Hardy-Weinberg equilib-
rium enable the mathematical derivation of an equation
for predicting genotype (and thence phenotype) frequen-
cies for a population of diploid individuals. Of course, no
actual population is in perfect Hardy-Weinberg equilib-
rium. All populations are finite; alternative genotypes can
make a difference in mating, mutations occur constantly;
migration into and out of a population is common; and
many genotypes of interest, such as those that cause diseases,
affect the ability to survive or reproduce. Nevertheless, even
when many of the assumptions of the Hardy-Weinberg
equilibrium do not apply, the equation derived on the basis
of these assumptions is remarkably robust at providing
estimates of genotype and phenotype frequencies in real
populations over a limited number of breeding genera-
tions. Furthermore, the reverse situation, where allele fre-
quencies are found to be inconsistent with a Hardy-Weinberg
equilibrium, can sometimes provide scientists with insight
into special biological properties of the locus in question
or the population as a whole. Indeed, the equation has
always been the most powerful mathematical tool avail-
able to population geneticists.
Predicting frequencies from one
generation to the next
For a population of sexually reproducing diploid organ-
isms, two steps are needed in translating the genotype
frequencies of one generation into the genotype frequen-
cies of the next generation (Fig. 19.2).
First, if the likelihood that an individual will grow
into an adult does not depend on the genotype (that is, if
659
there is no difference in fitness among individuals), then
the allele frequencies in the adults should be the same as
in their gametes. For example, if p is the frequency of
allele A, and q is the frequency of allele B in the adults,
p and q will also be the frequencies of the two alleles in
the combination of gametes produced by the whole popu-
lation of those adults.
Second, the allele frequencies in the gametes can be
used to calculate genotype frequencies in the zygotes of
the next generation. An enhanced version of the Punnett
square, which provides a systematic means of considering
all possible combinations of uniting gametes, is the tool of
choice (Fig. 19.2b). For example, if fertilization is random
among individuals with any genotype and if the popula-
tion of gametes is very large, then the following pattern
emerges. Recall that AA zygotes result from fertilization
of A-carrying eggs by A-carrying sperm. If p is the fre-
quency of A gametes (eggs and sperm), then, applying the
product rule, the frequency of AA zygotes is p (frequency
of A eggs) 3 p (frequency of A sperm) 5 p2. Similarly,
BB zygotes result from fertilization of B-carrying eggs by
B-carrying sperm. If q is the frequency of B gametes (eggs
and sperm), the frequency of BB zygotes will be q (the
frequency of B eggs) 3 q (the frequency of B sperm) 5 q2.
Finally, AB zygotes result either from fertilization of A
eggs by B sperm, with a frequency of p 3 q 5 pq, or from
the fertilization of B eggs by A sperm, occurring at a
frequency of q 3 p 5 pq. The total frequency of AB
zygotes is thus pq 1 pq 5 2pq.
The resemblance of the Hardy-Weinberg square shown
in Fig. 19.3 to the Punnett square that we encountered in
the visual representation of formal genetics is not a coinci-
dence. The top and left sides of the Punnett square were
divided into sectors representing the frequency of each
genetically distinct class of sperm or egg produced by two
individual parents. But to population geneticists, individual
organisms are not significant. Instead, its the gametes pro-
duced by the population as a whole that matters.
In parallel to the Punnett square, the Hardy-Weinberg
square represents a metaphorical mixture of sperm
Figure 19.3 Gametes and offspring of first-generation
individuals.
660 Chapter 19 Variation and Selection in Populations
produced by all breeding males along one side, and a mix-
ture of eggs produced by all breeding females along the
second side.
To summarize: The genotype frequencies of zygotes
arising in a large population of sexually reproducing dip-
loid organisms are p2 for AA, 2pq for AB, and q2 for BB
(see Fig. 19.2b). These genotype frequencies are known
as the Hardy-Weinberg proportions; they exist in popula-
tions that satisfy the Hardy-Weinberg assumptions of a
large number of individuals, mating at random, with no
new mutations, no migration, and no genotype-dependent
differences in fitness. Since these genotype frequencies
represent the totality of genotypes in the population, they
must sum to 1. Thus the binomial equation representing
the Hardy-Weinberg proportions is
p2 1 2pq 1 q2 5 1 (19.1)
Because we have assumed no differences in fitness, the
genotype frequencies of the zygotes will be the genotype
frequencies of the adult generation that develops from
those zygotes.
Predicting the frequency of albinism:
A case study
This equation thus enables us to use information on geno-
type and allele frequencies to predict the genotype fre-
quencies of the next generation. Suppose, for example, that
in a population of 100,000 people carrying the recessive
allele a for albinism, there are 100 aa albinos and 1800
Aa heterozygous carriers. To find what the frequency of
heterozygous carriers will be in the next generation, you
compute the allele frequencies in the parent population.
98,100 AA individuals; 1800 Aa individuals,
and 100 aa individuals n 196,200 A alleles
1 1800 A alleles; 1800 a alleles 1 200 a alleles
Out of 200,000 total alleles the frequency of the A allele is
198,000/200,000 5 99/100 5 0.99; thus p 5 0.99
and the frequency of the a allele is
2000/200,000 5 1/100 5 0.01; thus, q 5 0.01
The Hardy-Weinberg equation for the albino gene in this
population is
p2 1 2pq 1 q2 5 1
(0.99)2 1 2(0.99 3 0.01) 1 (0.01)2 5 1
0.9801 1 0.0198 1 0.0001 5 1
It thus predicts that in the next generation of 100,000
individuals, there will be
100,000 3 0.9801 5 98,010 AA individuals
100,000 3 0.0198 5 1980 Aa individuals
100,000 3 0.0001 5 10 aa individuals
This example shows that in one generation, the geno-
type frequencies have changed somewhat. A natural ques-
tion is, Have the allele frequencies also changed? Recall that
the initial frequencies of the A and B alleles are p and q,
respectively, and that p 1 q 5 1. You can use the rules for
computing allele frequencies from genotype frequencies
(see Fig. 19.2) to compute the allele frequencies of the next
generation. From the Hardy-Weinberg equation, you know
that p2 of the individuals are AA, whose alleles are all A,
and 2pq of the individuals are AB, 1/2 of whose alleles are
A. Similarly, q2 of the individuals are BB, whose alleles are
all B, and 2pq of the individuals are AB, 1/2 of whose alleles
are B. If p 1 q 5 1, then q 5 1 2 p, and the frequency of
the A allele in the next-generation population is
p2 1 1/2[2p(1 2 p)] 5 p2 1 p(1 2 p)
5 p2 1 p 2 p2 5 p (19.2)
Similarly, p 5 1 2 q, and the frequency of the B allele in
the next-generation population is
q2 1 1/2[2q(1 2 q)] 5 q2 1 q(1 2 q)
5 q2 1 q 2 q2 5 q (19.3)
Using these equations to calculate the allele frequencies
of A and a in the second generation of 100,000 individuals,
some of whom are albinos, we find
for p 0.9801 1 0.99 2 0.9801 5 0.99
5 the frequency of the
A allele
for q 0.0001 1 0.01 2 0.0001 5 0.01
5 the frequency of the
a allele
These frequencies are the same as those in the previ-
ous generation. Thus, even though the genotype frequen-
cies have changed from the first generation to the next,
the allele frequencies have not. Note that this is true of
both the dominant and the recessive alleles.
Analysis of eye color shows
the power and limitations of
Hardy-Weinberg
As an example of dominant and recessive phenotypes that
result from different genotypes and alleles at a single
locus, let’s consider the actual genetics of eye color in
human populations. Until 10,000 years ago, eye color was
not polymorphic—all of our ancestors from this period
had brown irises.
Genetics of blue eyes
Blue eyes first appeared between 6000 and 10,000 years
ago, probably in a population living near the north shore
of the Black Sea, according to anthropological genetic
data. Today, the trait predominates in northern European
populations, appearing in more than 80% of the people
 19.1 The Hardy-Weinberg Law: Predicting Genetic Variation in Populations 661

Figure 19.4 Blue-eye population genetics. (a) Geographical differences in the proportions of European populations expressing the “blue
eyes” phenotype. (b) The SNP rs12913832 that controls blue eye color is located upstream from OCA2 in an intron of HERC2. (c) Representation
of haplotype structure and coinheritance of SNPs across extended DNA blocks in the OCA2-HERC2 gene region of chromosome 15. The SNP
allele rs12913832G is part of conserved 50 kb chromosomal block that spans the 39 half of HERC2. (d) Pie diagrams depict frequencies of
alleles G and A at the SNP locus rs12913832 in different Old World populations.
(a) (c)

(d)

(b)

living around the Baltic Sea (Fig. 19.4a). Blue-eyed indi-

viduals can be seen throughout most of Europe with
diminishing frequencies at lower latitudes. Outside of
Europe and the Mediterranean region, the blue eye phe-
notype is rare, appearing sparsely in some central and
northern Asian populations, but essentially absent from
sub-Saharan Africa and east Asia.
High resolution linkage mapping has demonstrated
an association of blue eyes with a single base substitu-
tion, from an A to a G, that defines the SNP locus
rs12913832, located in an intron of the HERC2 gene on
chromosome 15 (Fig. 19.4b). The rs12913832G allele
is always found as a part of a larger haplotype of con-
served SNP alleles across a 50 kb chromosomal region
(Fig. 19.4c).
The extent of the conserved DNA region and the lim-
ited distribution of the G allele primarily to European
populations strongly suggests that this base substitution
was a one-time event. All people with blue eyes today
can trace themselves back genetically to the single indi-
vidual in which the mutation occurred.
Scientists were puzzled at first by the association of
rs12913832 to eye color because the product of the
HERC2 a gene—within which the SNP lies—does not
have any connection to pigment production and is not
even expressed in the iris or its precursors. The puzzle was
solved with studies that demonstrated transcription factor
binding to the DNA sequence surrounding rs12913832A,
with a large reduction in binding affinity to the DNA
sequence containing the G substitution. Further analysis
showed that rs12913832 was inside a highly conserved
enhancer of the OCA2 gene, which plays a critical role in
the biosynthesis of the dark pigment melanin. When a
person is homozygous for the rs12913832G allele, which
damages the iris-specific OCA2 enhancer, the OCA2 gene
product is greatly reduced, melanin is not synthesized, and
the resulting eye color in the absence of brown pigment
is blue.
Pie diagrams illustrating allele frequencies
Pie diagrams provide an intuitive tool for visualizing allele
frequency differences that distinguish one population from
another. The two alleles of an SNP locus are represented
in contrasting colors that occupy radial portions of a circle,
or pie, corresponding to allele frequencies.
Pie diagrams are placed on an appropriate geographic
map at the locations of the screened populations (Fig. 19.4d).
To obtain useful data, researchers typically assign people to
662 Chapter 19 Variation and Selection in Populations
populations according to the geographic locations where
their recent ancestors were born and lived.
By viewing Fig. 19.4d, you can see immediately that
allele frequencies at the SNP locus rs12913832 are dra-
matically different in geographically separated popula-
tions. The highest frequency of the rs12913832G allele in
populations screened for this locus is 0.84, associated
with a population from northern Finland, whereas all of
the Chinese and sub-Saharan African populations do not
carry the G allele at all.
Use of the Hardy-Weinberg equation
with mixed populations
To understand the implications of the Hardy-Weinberg law
in the analysis of a population formed from a mixture of
previously differentiated populations, imagine that 100 adults
from northern Finland and 100 from the Yakut people of
eastern Sibera (both men and women from both populations)
decide to move to a newly built offshore oil rig in the Arctic
Sea. Imagine further that the 200 men and women on the oil
rig marry each other without regard to their ancestry.
Now let’s ask two questions and see how the data we
have obtained on allele frequencies can be combined with the
Hardy-Weinberg law to provide answers. First, we can esti-
mate how many adults on the oil rig have blue eyes. Second,
we can estimate both the number of children with blue eyes
and the allele frequency in this second generation.
Because the Finnish and Yakut populations have differ-
ent rs12913832G allele frequencies, it is easiest to use the
Hardy-Weinberg equation to determine the composition of
each separately. Let’s use p to represent the allele frequency
of the brown eyes associated rs12913832A allele, and q to
represent the allele frequency of rs12913832G. The data
sampled from the Finnish yield q 5 0.84. Blue eyes is reces-
sive, which means that we can estimate the number of Finnish
adults with blue eyes from the genotype frequency for GG:
q2 5 0.84 3 0.84 5 0.71
By multiplying 0.71 by the total number of Finns on the
oil rig, we get an estimate for the number of blue-eyed
Finns at
0.71 3 100 5 71 individuals
The frequency of brown-eyed Finnish carriers of the
rs12913832G allele is calculated as
2pq 5 2 3 .16 3 .84 5 .27
which yields 27 carrier individuals on the oil rig.
Similarly, from the Yakut rs12913832G allele fre-
quency of 0.10, we obtain
q2 5 (0.1)2 5 0.01
which translates into a single Yakut adult with blue eyes,
and
2pq 5 2 3 0.9 3 0.1 5 0.18
for 18 Yakut adults who are carriers. Together, among the
200 people on the oil rig, 72 (71 1 1) will have blue eyes,
and 45 (27 1 18) will be carriers. With this information,
we can calculate the combined G allele frequency from a
count of chromosomes—two from each blue-eyed person
and one from each carrier yields
2 3 72 1 45 5 189 4 400 (the number of chromosomes is
twice the number of people) for
an answer of q 5 0.47
Now let’s use the parental G allele frequency to pre-
dict the number of children expected with blue eyes. The
frequency of blue eyes in this second generation is q2
5 0.47 3 0.47 5 0.22. If we assume that the population
of children is equal in size to the population of adults, then
the number of blue-eyed children is 0.22 3 200 5 44, far
fewer than the 72 blue-eyed adults. Nevertheless, the G
allele frequency remains unchanged at 0.47. If this second
generation intermarries to produce a third generation of
200, the expected number with blue eyes will be the same
as the second generation. Indeed, all future generations will
have the same expected genotype and allele frequencies.
Properties of populations described
by Hardy-Weinberg
The application of Hardy-Weinberg analysis to the study of
eye color in human populations generates two important con-
clusions. First, even though the proportion of individuals
expressing the blue-eye phenotype changed dramatically
from the first generation of the mixed population to the sec-
ond generation, no change in allele frequency occurred. The
conservation of allele proportions principle holds from each
generation to the next, as long as the population is suffi-
ciently large, alleles are not lost by mutation or selection,
and alleles are not gained by mutation or immigration.
Populations with the same allele frequency don’t nec-
essarily have the same genotype or phenotype frequencies.
The reason is that a single allele can exist in homozygote
or heterozygote genotypes, but a recessive phenotype is
only expressed in homozygotes. In the most extreme hypo-
thetical example, a population could have a blue-eyed
allele frequency of 0.5 without actually having any people
with blue eyes. This situation would arise if everyone in
the population is a heterozygote.
Even in this extreme example, the Hardy-Weinberg
equilibrium tells us that the Hardy-Weinberg genotype
frequencies described by p2, 2pq, and q2, will appear in
the very next generation. This is the second significant
Hardy-Weinberg implication: a population that is initially
stratified because of its founding by individuals from two
or more distinct populations will become completely bal-
anced in a single generation.
Once a population is known in be in Hardy-Weinberg
equilibrium, it becomes a simple task to predict allele fre-
quencies from genotype frequencies, and genotype and
phenotype frequencies from allele frequencies.