THE MECHANISMS OF EVOLUTION 465

23.5 Artificial Selection Reveals Genetic

Variation In artificial selection experiments
Abdominal with Drosophila melanogaster, changes in bristle
Population bristles number evolved rapidly. The graphs show the
selected for number of flies with different numbers of bristles
low number after 35 generations of artificial selection.
Number of individuals

of bristles. Low- and high-selected

populations do not overlap
the original population.
Population
selected for
high number
of bristles.

Original
population

0 10 20 30 40 50 60 70 80 90 100 110
Number of bristles

How do we measure genetic variation?
A locally interbreeding group within a geographic popula-
tion is called a Mendelian population. Mendelian popula-
tions are often the subjects of evolutionary studies. To meas-
ure genetic variation in a Mendelian population precisely, we
would need to count every allele at every locus in every in-
dividual in it. By doing so, we could determine the relative
proportions, or frequencies, of all alleles in the population.
Fortunately, we do not need to make such complete meas-
urements, because we can reliably estimate allele frequencies
for a given locus by counting alleles in a sample of individ-
uals from the population. The sum of all allele frequencies at
a locus is equal to 1, so measures of allele frequency range
from 0 to 1.
An allele’s frequency is calculated using the following
formula:
number of copies of the allele in the population
p=
sum of alleles in the population
If only two alleles (for example, A and a) for a given locus are
found among the members of a diploid population, they may
combine to form three different genotypes: AA, Aa, and aa.
Using the formula above, we can calculate the relative fre-
quencies of alleles A and a in a population of N individuals
as follows:

Let NAA be the number of individuals that are homozy-
gous for the A allele (AA).

Let NAa be the number that are heterozygous (Aa).

Let Naa be the number that are homozygous for the a
allele (aa).
Note that NAA + NAa + Naa = N, the total number of individ-
uals in the population, and that the total number of copies of
both alleles present in the population is 2N because each in-
dividual is diploid. Each AA individual has two copies of the
A allele, and each Aa individual has one copy of the A allele.
Therefore, the total number of A alleles in the population is
2NAA + NAa. Similarly, the total number of a alleles in the pop-
ulation is 2Naa + NAa.
If p represents the frequency of A, and q represents the fre-
quency of a, then
2 N AA + N Aa
p=
2N
and
2 N aa + N Aa
q=
2N
To show how this formula works, Figure 23.6 calculates al-
lele frequencies in two populations, each containing 200
diploid individuals. Population 1 has mostly homozygotes
(90 AA, 40 Aa, and 70 aa); population 2 has mostly heterozy-
gotes (45 AA, 130 Aa, and 25 aa).
The calculations in Figure 23.6 demonstrate two impor-
tant points. First, notice that for each population, p + q = 1. If
there is only one allele in a population, its frequency is 1. If
an allele is missing from a population, its frequency is 0, and
the locus in that population is represented by one or more
other alleles. Since p + q = 1, then q = 1 – p. So when there are
only two alleles at a given locus in a population, we can cal-
culate the frequency of one allele and then easily obtain the
second allele’s frequency by subtraction.
The second thing to notice is that both population 1 (con-
sisting mostly of homozygotes) and population 2 (consisting
mostly of heterozygotes) have the same allele frequencies for
A and a. Therefore, they have the same gene pool for this lo-
cus. However, because the alleles in the gene pool are dis-
tributed differently, the genotype frequencies of the two popu-
lations differ. Genotype frequencies are calculated as the
466 CHAPTER T WENT Y-THREE
In any population:

Frequency p = 2N
AA +N
Aa Frequency q = 2N + N
aa Aa
number of individuals that have the genotype =
2N
=
2N
of allele A of allele a
divided by the total number of individuals in
the population. In population 1 in Figure 23.6, where N is the total number of individuals in the population.
the genotype frequencies are 0.45 AA, 0.20 Aa,
and 0.35 aa. For population 1 (mostly homozygotes): For population 2 (mostly heterozygotes):
The frequencies of different alleles at each NAA = 90, NAa = 40, and Naa = 70 NAA = 45, NAa = 130, and Naa = 25
locus and the frequencies of different geno- so so
types in a Mendelian population describe its p=
180 + 40
= 0.55 p=
90 + 130
= 0.55
genetic structure. Allele frequencies measure 400 400
the amount of genetic variation in a popula- 140 + 40 50 + 130
q= = 0.45 q= = 0.45
tion; genotype frequencies show how a popu- 400 400
lation’s genetic variation is distributed among
its members. With these measurements, it be-
23.6 Calculating Allele Frequencies The gene pool and allele fre-
comes possible to consider how the genetic quencies are the same in two different populations, but the alleles
structure of a population changes or does not change over are distributed differently between heterozygous and homozygous
generations. genotypes. In all cases, p + q must equal 1.

The Hardy–Weinberg Equilibrium

If certain conditions are met, the genetic structure of a
population may not change over time. The necessary condi-
tions for such an equilibrium were deduced independently
in 1908 by the British mathematician Godfrey Hardy and the
German physician Wilhelm Weinberg. Hardy wrote his equa-
tions in response to a question posed to him by the geneticist
Reginald C. Punnett (the inventor of the Punnett square) at
the Cambridge University faculty club. Punnett wondered at
the fact that even though the allele for short, stubby fingers
(a condition called brachydactyly) was dominant and the al-
lele for normal-length fingers was recessive, most people in
Britain have normal-length fingers. Hardy’s equations explain
why dominant alleles do not necessarily replace recessive al-
leles in populations, as well as other features of the genetic
structure of populations.
The Hardy–Weinberg equilibrium applies to sexually re-
producing organisms. The particular example we will illus-
trate here assumes that the organism in question is diploid,
its generations do not overlap, the gene under consideration
has two alleles, and allele frequencies are identical in males
and females. The Hardy–Weinberg equilibrium also applies if
the gene has more than two alleles and generations overlap,
but in those cases the mathematics is more complicated.
Several conditions must be met for a population to be at
Hardy–Weinberg equilibrium:

Mating is random

Population size is very large

There is no migration between populations

There is no mutation

Natural selection does not affect the alleles under
consideration
If these conditions hold, two major consequences follow.
First, the frequencies of alleles at a locus will remain constant
from generation to generation. And second, after one gener-
ation of random mating, the genotype frequencies will re-
main in the following proportions:
Genotype AA Aa aa
2
Frequency p 2pq q2
Stated another way, the equation for Hardy–Weinberg equi-
librium is 2 2
p + 2pq + q = 1
To see why, consider population 1 in Figure 23.6, in which the
frequency of A alleles (p) is 0.55. Because we assume that in-
dividuals select mates at random, without regard to their
genotype, gametes carrying A or a combine at random—that
is, as predicted by the frequencies p and q. The probability
that a particular sperm or egg in this example will bear an A
allele rather than an a allele is 0.55. In other words, 55 out of
100 randomly sampled sperm or eggs will bear an A allele.
Because q = 1 – p, the probability that a sperm or egg will bear
an a allele is 1 – 0.55 = 0.45.
To obtain the probability of two A-bearing gametes com-
ing together at fertilization, we multiply the two independ-
ent probabilities of their occurring separately (see the dis-
cussion of probability in Chapter 10):
p × p = p2 = (0.55)2 = 0.3025
Therefore, 0.3025, or 30.25 percent, of the offspring in the next
generation will have the AA genotype. Similarly, the proba-
bility of bringing together two a-bearing gametes is
q × q = q2 = (0.45)2 = 0.2025
 THE MECHANISMS OF EVOLUTION 467

Thus, 20.25 percent of the next generation will have the aa Generation I
genotype (Figure 23.7).
Figure 23.7 also shows that there are two ways of produc-
ing a heterozygote: An A sperm may combine with an a egg,
the probability of which is p × q; or an a sperm may combine
Genotypes AA Aa aa
with an A egg, the probability of which is q × p. Conse-
quently, the overall probability of obtaining a heterozygote Frequency of 0.45 0.20 0.35
genotypes in
is 2pq. population
It is now easy to show that the allele frequencies p and q
remain constant for each generation. If the frequency of A al- Frequency of 0.45 + 0.10 0.10 + 0.35
alleles in
leles in a randomly mating population is p2 + pq, this fre- population p = 0.55 q = 0.45
quency becomes p2 + p(1 – p) = p2 + p – p2 = p, the original al-
A Gametes a
lele frequencies are unchanged, and the population is at
Hardy–Weinberg equilibrium.
If some agent, such as emigration, were to alter the allele Generation II
frequencies, the genotype frequencies would automatically
settle into a predictable new set in the next generation. For
A A
instance, if only AA and Aa individuals left the population,
Eggs Sperm
p and q would change, but there would still be aa individu-
als in the population.
a a
AA (p2)
= 0.55 × 0.55
Why is the Hardy–Weinberg equilibrium important? = 0.3025

You may already have realized that populations in nature Aa (pq) Aa (pq)
rarely meet the stringent conditions necessary to maintain = 0.55 × 0.45 = 0.55 × 0.45
= 0.2475 = 0.2475
them at Hardy–Weinberg equilibrium. Why, then, is the
Hardy-Weinberg equilibrium considered so important for the p = 0.55 p = 0.55
study of evolution? The answer is that without it, we cannot aa (q2)
tell whether or not evolutionary agents are operating. The = 0.45 × 0.45
= 0.2025
most important message of the Hardy–Weinberg equilibrium
is that allele frequencies remain the same from generation to gen- q = 0.45 q = 0.45
eration unless some agent acts to change them.
In order to ascertain that evolutionary agents are in play,
we must estimate the actual allele or genotype frequencies
present in a population and then compare them with the fre- Adding the four genotype frequencies
gives the Hardy–Weinberg equation:
quencies that would be expected at Hardy–Weinberg equi- p2 + 2pq + q2 = 1
librium. The pattern of deviation from the Hardy–Weinberg
expectations tells us which assumptions are violated. Thus,
23.7 Calculating Hardy–Weinberg Genotype Frequencies The
we can identify the agents of evolutionary change on which areas within the squares are proportional to the expected frequen-
we should concentrate our attention. cies of possible matings if mating is random with respect to geno-
type. Because there are two ways of producing a heterozygote, the
probability of this event occurring is the sum of the two Aa squares.
Evolutionary Agents and Their Effects
Evolutionary agents are forces that change the genetic struc-
ture of a population. In other words, they cause deviations Mutations are changes in the genetic material
from the Hardy–Weinberg equilibrium. The known evolu- The origin of genetic variation is mutation. A mutation, as we
tionary agents are mutation, gene flow, genetic drift, non- saw in Chapter 12, is any change in an organism’s DNA. Mu-
random mating, and natural selection. Although only natu- tations appear to be random with respect to the adaptive
ral selection results in adaptation, to understand evolutionary needs of organisms. Most mutations are harmful to their
processes we need to discuss all of these evolutionary agents bearers or are neutral, but if environmental conditions
before considering natural selection in detail. change, previously harmful alleles may become advanta-