Natural Selection in Contemporary Human Society

Examples, Opportunity for Selection, Quantitative Traits and Behavior,

HbS, HbA, CCR5, The Genetics of Human Populations
Natural selection is the process by which gene frequencies in populations change by virtue of
the effects those genes have on the fitness of their bearers. Researchers often wish to know the
magnitude and direction of selective forces in populations in order to understand and predict
evolutionary change. It is difficult to measure selective forces with sufficient accuracy for that
purpose, because even very weak selection can lead to appreciable change when it
accumulates over many generations. Consider, for example, pedestrian deaths in traffic. These
may be random and thus not related to natural selection; however, if individuals with alleles
(alternative copies of the same gene) that lead them to be less fearful are disproportionately
victims, then these deaths are selective, and the alleles that suppress fear will decrease in
frequency through time.

To what extent is natural selection still operating in modern human society, now that many of
the historically important sources of mortality have been reduced? We begin by discussing
three common ways in which natural selection operates. We then identify ways of measuring
the potential for it in human populations. This article concludes with a discussion of several
traits on which natural selection has operated in the recent past and may still be operating in
human society.

Examples
Three important categories of natural selection are balanced polymorphism in which selection
maintains genetic diversity, selection in response to environmental change, and selection that
removes deleterious mutations. We will discuss an example of each category.

Falciparum malaria is one of the most important causes of premature death in much of the
tropical and subtropical Old World. Several genes are known to affect the fitness of
individuals through their response to falciparum malaria. A particularly well understood case
is that of the sickle-cell gene, HbS, which causes production of a modified form of
hemoglobin. Individuals with a single copy of the sickle-cell gene and a copy of the normal
allele (heterozygotes) enjoy better survival and reproduction in environments where infection
by falciparum malaria is common than do individuals with two copies of the normal allele
(HbA homozygotes). Individuals with two copies of the sickle-cell gene (HbS homozygotes)
ordinarily die before the age of reproduction unless aided by modern medical technology.

In malaria-ridden environments, the fitness advantage of heterozygotes is 10 to 20 percent.

This means that a cohort of heterozygotes leaves, one generation later, 10 to 20 percent more
offspring than a cohort of normal homozygotes. Using the lower figure of 10 percent and
assuming that sickle-cell homozygotes never survive to reproduce, natural selection will lead
to population frequencies of about 90 percent for the wild-type (HbA) allele, and ten percent
for the sickle-cell (HbS) allele. At this polymorphic evolutionary equilibrium, the average
fitness of the population is only 1 percent greater than the average fitness of a population
without the polymorphism.
The sickle-cell gene is the best-understood example of balanced polymorphism in our species:
heterozygote advantage leads to the persistence of both alleles in the population. The sickle-
cell allele does not spread through the whole population because a higher frequency of this
gene results in higher numbers of HbS homozygotes, who have very low fitness. It is not
known how many such self-destructive pathogen defense polymorphisms are present in our
species, but almost all known human polymorphisms causing illness in homozygotes are
pathogen defenses. The sickle-cell polymorphism is particular costly in personal terms
because homozygotes sicken and die soon after birth as fetal hemoglobin is replaced by
normal or adult hemoglobin. If the pathology in affected homozygotes were manifested early
in intrauterine life, there would be few or no personal consequences, and we would likely be
unaware of the system (Melanesian ovalocytosis, another malaria defense polymorphism, is
an example of this: no surviving homozygote has ever been observed). The personal cost of the
polymorphism also remains high in emigrant descendants of African and Mediterranean
populations with the polymorphism. Where malaria has been eliminated rapidly through
technology, illness and premature death from the polymorphism persist for many
generations. Its victims are just as much victims of malaria as are people suffering directly
from the disease.

The gene CCR5 codes for a cell surface receptor. In European populations there is a faulty
allele that, in homozygotes, causes the absence of this receptor (Schliekelman et al., 2001).
Such individuals infected with HIV do not progress to AIDS, while infected heterozygotes
enjoy delayed progression to AIDS. Other variant alleles of the gene are known in Africa; some
delay and some accelerate the progression to AIDS. In environments with high levels of
exposure to HIV, as in much of sub-Saharan Africa, natural selection favors alleles that delay
AIDS and disfavors those that accelerate the onset of the disease. Recent projections suggest
that the population frequency of resistant genes will increase from 40 percent to 53 percent in
the next century, while the frequency of susceptible alleles will decrease from 20 to 10
percent. The frequency of the faulty, AIDS-resistant allele in Europe is so high that it is
thought that it increased in response to some other major pathogen, perhaps plague, within
the last few thousand years.

Neurofibromatosis is a severely damaging dominant disorder: people who inherit the

mutation usually die early in life. Its incidence is determined entirely by the balance between
the rate of the mutation that causes new cases and selection against bearers of the mutation.
Approximately one or two births per ten thousand have this disorder, one of the highest
mutation rates known for a deleterious gene; it reflects the large size of the affected gene and
hence the length of DNA susceptible to mutational damage. As in the case of the sickle-cell
disorder, we know about this disorder because its effects occur well after birth. Similar
disorders that caused very early fetal wastage would have few or no personal consequences
and would not be known to us.

In all three of these cases, natural selection has essentially been maintaining the status quo,
except for the recent increase in CCR5 in response to the AIDS pandemic. In general, we
would like to know, first, how many individual differences in health and fertility are caused by
gene differences; and second, how many of these differences are balanced polymorphisms,
how many are responses to environmental change, and how many represent mutational
damage. We have almost no solid information about the first question, and very little about
the second. Extensive studies of fruit flies suggest that the removal of deleterious mutations
has a larger contribution to fitness than does the maintenance of balanced polymorphism. It is
not known whether this finding can be extrapolated to mammals.

Opportunity for Selection

With some simplifying assumptions, it is possible to estimate the opportunity for natural
selection in human populations. A statistic developed by James Crow (1958) describes the
variance in fitness among individuals. If the variance is low, then there is not much difference
in number of successful offspring between people and, hence, not much opportunity for
selection. If the variance is large, then there are large differences, and the opportunity for
selection is great. Crow's index I can be interpreted as follows: if all variation in reproductive
performance (fitness) is attributable to additive gene differences, then the fitness of the next
generation will change by 1 + I. However, there are three reasons why we cannot use this
index directly: first, the fraction of variation in fitness that is related to gene differences is
unknown; second, much of the fitness variation attributable to gene differences must be used
up in eliminating deleterious mutations, many of which are newly arisen; and third, this
demographic method cannot detect very early deaths. For example, an individual bearing a
new dominant lethal mutation that causes death within the first few weeks after conception
would lose zygotes bearing the mutation, but these early losses of half her conceptions would
mean only a slight, probably undetectable, fertility reduction.

Nevertheless, in humans Crow's index shows that there is enough fitness variation among
individuals to accommodate high levels of selection and evolution. In high-mortality
populations, the index reaches values of 2 to 3, so that under the simplest assumptions fitness
could triple or quadruple each generation. Much of the variation arises from pre-reproductive
mortality, but even in contemporary low-mortality populations the index hovers around 0.5, a
value that would more than double fitness in two generations under the simplest assumptions.
Among Utah Mormons early in the nineteenth century, the index for males was greater than
that of females, reflecting the effect of polygyny and acccompanying higher variance in male
fitness; after polygyny ended, the values in two sexes became similar.

Quantitative Traits and Behavior

Crow's work shows that there is ample opportunity for ongoing natural selection in modern
human populations. The examples of the HbS and CCR5 loci suggest that a substantial
fraction of ongoing natural selection in human populations occurs in response to changing
environments—movement away from malarial areas in the case of selection against HbS, and
the advent of widespread HIV infection in the case of CCR5. What other changes in human
populations are occurring or will occur in the future in response to environmental change?
There are many ongoing social changes in industrial societies that will, if continued, result in
significant changes in the genetics of these populations. Among the most interesting are those
affecting partially heritable quantitative traits, especially behavioral traits.

A quantitative trait is a metric trait that is variable in a population, with part of the variation
determined by gene differences. For example, height is known to be influenced both by genes
and by environmental factors like diet. Geneticists use heritability as a statistic describing the
fraction of variability in a trait that is genetically transmitted in a population. If height in
humans were entirely determined by diet and other environmental agents, then its heritability
would be zero. In contrast, if differences in height were determined entirely by additive gene
differences, the heritability would be unity. In reality, height is affected by both, and
similarities among relatives can be used to estimate heritability. It should be clear that there is
nothing fundamental about the heritability of a trait. In a sample of people from a population
with large differences in nutritional status, we would find that some individuals suffered
marginal nutrition and much chronic infectious disease during development, while others
were well fed and well supplied with antibiotics; then, much of the variation in height would
reflect early environmental conditions. In contrast, height in a sample of people who were all
raised in a wealthy community with plentiful food and medical treatment would be highly
heritable; all individuals were exposed to a similar highly favorable environment, so
phenotypic differences among them would reflect mostly genetic differences.

Empirical studies of twins and other categories of relatives demonstrate overwhelmingly that
any trait we examine is partially heritable. Even such traits as political liberalism or
conservatism, or the tendency to divorce, are correlated with differences that are transmitted
in large part genetically.

The response of quantitative traits to selection is particularly simple. The change in one
generation is the product of the trait's heritability and the difference between the population
mean and the mean of those who reproduce (called the selection differential).

Intelligence quotient (IQ), the score obtained on any of various standard written tests of
cognitive ability, is a familiar example of a quantitative genetic trait (Herrnstein and Murray,
1994). We do not know what genes are involved, but we are sure from a variety of natural
experiments, such as identical twins reared apart, that the obtained score has a heritability
greater than 0.5. Consider a social system that favored IQ in a population such that those
individuals with IQ scores greater than the mean had 5 percent more offspring. If the initial
population mean were 100 with standard deviation of 15 points, then the mean IQ of parents
would be 100.6, so that the selection differential is 0.6 points. The mean IQ of the next
generation would be the product of the differential 0.6 and the heritability 0.5, or 0.3. In a
single generation, there would be little change in IQ—from 100 to 100.3. If the social system
were to persist for a thousand years, or 40 generations, the net change at the end of the
millennium would be 40 × 0.3, or 12 IQ points: the population would then have a mean IQ
score of 112. There has been speculation that the high IQ test scores of Ashkenazi Jews reflect
a social history like this.

Because different social systems reward different abilities and proclivities, and because almost
everything so far examined is more or less heritable, we expect social systems to cause genetic
changes in their members over time scales of centuries to millennia. A possible example is
that of the D4 dopamine receptor locus, at which long alleles are differentially distributed
among human populations (Chen et al., 1999). There is a correlation between long alleles at
this locus and novelty-seeking and perhaps activity levels in individuals. Populations that have
migrated long distances since the end of the Pleistocene have a higher frequency of long
alleles, perhaps because bearers of long alleles were more likely to join population movements
in the past. For example, the world's highest frequencies of long alleles, over 50 percent, are in
South American populations, furthest from the African home of our species.

We can only speculate about the consequences for genetic change of contemporary social
patterns in industrial societies, but there are several obvious candidates. First, the easy
availability of birth control technology should lead to selection in females for maternalism and
a positive desire to have children. In the past, women rarely had such precise control over
conception; now, there is often conscious decision-making involved. Second, human females
around the world obtain provisioning from males, but in varying degrees. In dense
agricultural societies such as those of Europe, male provisioning has been crucial for female
fitness, and this has favored social systems with durable marriages. In many subsistence
horticultural societies—notably those in Amazonia, Southeast Asia and Oceania, and central
Africa—females essentially feed themselves and their children; provisioning by fathers is less
essential to female fitness, and marriage bonds are fragile. As male provisioning becomes less
and less relevant to the health and survival of children in industrial societies, we expect
selection to favor females who do not seek a provisioning mate and males who are less willing
to provision offspring. Finally, new reproductive technologies, if they become widespread,
have the potential for drastically altering selection. Currently the wealthiest individuals have
the fewest children, but if reproduction could be carried out by surrogates, this negative
correlation between wealth and fitness might be reversed.

Bibliography and More Information about Natural Selection in Contemporary Human Society

• Cavalli-Sforza, L. L., and W. F. Bodmer. The Genetics of Human Populations. San Francisco, 1971. A classic text on human population
genetics, written before the explosion of molecular biology but unexcelled in its treatment of demography and quantitative traits.

• Chen, C., et al. “Population Migration and the Variation of Dopamine D4 Allele Frequencies around the Globe.” Evolution and
Human Behavior 20 (1999): 309–324. Describes worldwide variation in population frequency of long alleles at a dopamine receptor locus that
may be related to population differences in personality and behavior.

• Crow, J. “Some Possibilities for Measuring Selection Intensities in Man.” Human Biology 30 (1958): 1–13. Crow's classic paper is a
description of how to relate demographic patterns to the potential intensity of natural selection.

• Herrnstein, R. J., and C. Murray. The Bell Curve: Intelligence and Class Structure in American Life. New York, 1994. A
standard work on IQ, its transmission, and its social correlates in the United States.

• Schliekelman, P., et al. “Natural Selection and Resistance to HIV.” Nature 411 (2001): 545. Describes the very rapid evolutionary
change at the CCR5 locus that must be occurring in sub-Saharan Africa.
Henry Harpending and Gregory Cochran: http://science.jrank.org/pages/48650/Natural-Selection-in-Contemporary-Human-Society.html