Molecular Evolution

Introductory article
Article Contents

Jan Klein, Max Planck Institute for Biology, Tubingen, Germany

. Introduction

Molecular evolution is the process by which molecules of living forms diverge from one
another over time. The study of molecular evolution embraces the evolution of the
molecules themselves, their origins, how they change from taxon to taxon, the effects of
these changes on function, and the integration of the molecules in the biochemical
pathways of an organism. It also infers phylogenetic relationships and contributes to the
formulation of general principles of evolution at the molecular level.

Introduction
For evolution to take place, changes must occur and they
must then be handed on further. In the case of biological
evolution, the changes (mutations) occur in the molecules
of the hereditary (genetic) material, the nucleic acids, and
their transmission from ancestor to descendant is guaranteed by the replication of these molecules. The changes may
remain conned to the nucleic acids or they may be
translated into an alteration of proteins encoded in the
genetic material; the modied proteins may then eect the
alteration of other molecules, sugars for example, and so
on, until, at the end of the causal chain, changes aecting
an observable character may take place. Ultimately, a
molecular change in the genetic constitution (genotype)
may thus be transliterated into altered appearance,
physiology or behaviour the organisms phenotype. To
become an evolutionary change, however, the mutation
must spread through a group of individuals, the population. There are thus two principal levels at which evolution
can be studied the molecular and the population levels.
Corresponding to a population at the molecular level is the
gene pool the total genetic material of an interbreeding
population at a given time (in a more restricted sense, a
gene pool is an assemblage of all alleles at a given locus in a
population in a given generation; here a gene is a unit of
inheritance, a sequence of nucleotides to which a specic
function can be assigned; alternative forms of a gene are
alleles and the position a gene or allele occupies on a
chromosome is a locus). Evolutionary changes at the
molecular level constitute molecular evolution.
The study of molecular evolution embraces three broad
areas. First, it is concerned with the evolution of the
molecules themselves, primarily sequences of nucleic acids
and proteins their origin, the manner in which they
change from taxon to taxon, the eects of these changes on
function, and the integration of the molecules in the
biochemical pathways of an organism. Although almost
any molecule or molecular segment can be selected for
evolutionary analysis, certain molecules have become
biologists favourites. These include haemoglobins, lysozymes, colour-sensitive pigment proteins, eye-lens crystal-

. History
. Principle
. Methods
. Molecular Clock
. The Neutral Theory of Evolution
. Evolution of the Genome

lins, homeobox genes, and genes of the immune system

such as those in the major histocompatibility complex
(Mhc) and immunoglobulins. In a broader context, the
study of molecular evolution also encompasses questions
concerning lifes origin, the inception of life-sustaining
molecules, and the emergence of fundamental life processes. This area of molecular evolution links up with
molecular and cellular biology and with biochemical
genetics.
In the second area, the information on the molecules is
used to infer phylogenetic relationships among organisms
and to classify taxa. The relationships studied range from
those among populations within a species to those among
lifes domains and among the main branches of the tree of
life. The selection of molecules (genes) to be studied
depends on the degree of the expected relationship. For
closely related taxa, microsatellites (short arrays of
tandemly repeated basic units of DNA) and mitochondrial
DNA are frequently used; for less closely related taxa, a
wide range of molecules, including some of those listed
above, is available; and for the most distant relationships,
the choice falls on slowly evolving molecules such as
ribosomal RNAs, transfer RNAs and heat-shock proteins,
shared by all organisms. In this second area, the study of
molecular evolution abuts on taxonomy and systematics.
The third area of molecular evolution is conceptual,
aimed at the formulation of general principles of evolution
at the molecular level. Here, researchers probe the nature
of evolutionary change at the molecular level and its
relationship to changes at the phenotypic level; they strive
to identify the forces responsible for the changes, and seek
to discover patterns in the evolutionary process. In this
area, the study of molecular evolution stands on common
ground with population genetics.

History
Molecules are not visible to the naked eye; although the
large molecules of nucleic acids and proteins can be
visualized with the aid of an electron microscope, the

ENCYCLOPEDIA OF LIFE SCIENCES / & 2001 Nature Publishing Group / www.els.net

Molecular Evolution

discernible detail is insucient to spot changes in microstructure. The study of molecular evolution therefore only
became possible after methods for detecting structural
changes in macromolecules were developed. The rst
methods were indirect: they revealed that structural change
occurred but provided no information about the nature of
the change, nor about its location in the molecule. One
indirect method took advantage of the immune systems
ability to distinguish foreign substances from those of the
organisms own. For example, the inoculation of human
serum albumin into a rabbit leads to the production of
antibodies, which react strongly with human, less strongly
with chimpanzee and gorilla, and do not react with rhesus
macaque serum albumin at all. This immunological
reaction, which, if positive, results in the formation of a
visible precipitate, indicates that human serum albumin is
structurally somewhat dierent from the serum albumin of
a chimpanzee or a gorilla, and quite distinct from the
monkey protein. The ability of antibodies to dierentiate
between structurally distinct proteins and carbohydrates
was already recognized at the turn of the century and was
exploited in attempts to classify plants and animals. Later
the method fell into disuse, only to be resurrected in the
1960s and 1970s.
Three other methods protein electrophoresis, DNA
DNA hybridization and restriction enzyme analysis were
also developed to detect dierences between biological
macromolecules without being able to ascertain the
chemical nature of the changes. Electrophoresis distinguishes proteins according to their overall charge. DNA
DNA hybridization takes advantage of the observation
that the two complementary strands of a DNA molecule
dissociate (melt) on heating and then reanneal again on
cooling. If radioactively labelled, melted DNA of one
species is mixed with melted DNA of another species and
the mixture is cooled, the strands derived from the two
species form hybrid duplexes. The extent of this DNA
DNA hybridization depends on the similarity of the DNA
of the two species and it can be determined by heating the
hybrid molecules: the greater the similarity, the higher the
melting temperature of the hybrids. The third method puts
to use restriction enzymes, proteins that have evolved in
various bacteria as a means of providing protection from
an invasion of foreign DNA. The enzymes attack the
foreign DNA and hack it into pieces, each enzyme
recognizing specic regions of the molecule. Because of
the dierences in structure, each DNA molecule yields a
distinct set of fragments when digested with one or more
restriction enzymes. The fragments can be separated and
visualized, and their length can be determined by electrophoresis.
Ultimately, advances in biochemistry led to the development of direct methods by which the nature of
mutational changes could be established. Biochemical
investigations revealed proteins and nucleic acids to be
strings of amino acid and nucleotide residues, respectively,
2

arranged in a distinctive order. The methods enabled

researchers to determine this order, or sequence, and to
establish that it was dierent not only in distinct proteins or
nucleic acids but also in the same proteins or nucleic acids
of dierent species. Protein sequencing methods became
available in the 1960s and, although they were later
improved, they have remained laborious. Nucleic acid
sequencing methods were developed some two decades
later and quickly came into widespread use. Because a
protein sequence can be deduced from a nucleic acid
sequence and because nucleic acid sequencing is much
simpler and faster than protein sequencing, the nucleic acid
sequencing has now largely superseded protein sequencing
in the study of molecular evolution.
Three other methodological developments contributed
to the widespread use of nucleic acid sequences in the study
of molecular evolution. First, the establishment of DNA
cloning procedures solved both the problem of sorting out
mixtures of molecules and that of obtaining pure populations of molecules. Second, the introduction of the
polymerase chain reaction (PCR) technique provided a
means of nding and amplifying a particular DNA
fragment, even if it is present in the sample in a minute
amount (in the most extreme case, in only one copy), and of
nding homologous DNA segments in dierent, sometimes even extinct taxa. Third, the automation of the
sequencing methods signicantly shortened the time and
eort required to determine the order of nucleotides in a
nucleic acid.
The mass of sequence data would have become
unmanageable, however, had statistical, computerized
methods of sequence analysis not been developed. These
methods enable evolutionary biologists to align sequences
obtained from dierent taxa, to identify the similarities and
dierences, to infer the most probable relationships among
the various sequences, and to assess the statistical
signicance of the relationships.

Principle
From the point of view of evolution, the simplest structural
change (mutation) in a DNA molecule is the substitution of
a nucleotide of one kind by another nucleotide during
replication. The substitution may or may not eect a
replacement of an amino acid residue in a protein,
depending on the region and the site of its occurrence. A
mutation generates a new allele that falls into one of three
broad categories. An allele with the same chance of
contributing its own copy to the next generation is said
to be neutral. An allele that eects its own preferential
transmission, increasing the chance of survival of its bearer
or of producing a larger number of ospring (i.e. increases
its tness), is said to be selectively advantageous. Finally,
an allele that decreases the chance of survival and

ENCYCLOPEDIA OF LIFE SCIENCES / & 2001 Nature Publishing Group / www.els.net

Molecular Evolution

reproduction of its bearer is considered to be selectively

disadvantageous.
When a neutral mutation takes place in a population of
Ne randomly breeding individuals, that is in a pool of 2Ne
genes in a diploid organism, it constitutes one of 2Ne
alleles. If every allele in the pool contributed one copy of
itself to the next generation, the proportion of the mutant
allele relative to all the other alleles (its frequency) would
remain the same during the passage of generations. In
reality, however, this never happens. By chance, some
genes do indeed contribute their own copy, but others
contribute more than one and others contribute none at all.
The gene pool of each generation is then said to be
produced by random sampling of the pool in the preceding
generation. Because of the eects of random sampling, the
frequency of a mutant allele may, again by chance, increase
from one generation to the next, decrease, or stay the same.
Chance uctuation in allele frequency from generation to
generation as a result of random sampling is referred to as
random genetic drift. The most likely fate of a mutant allele
is that it is lost from the pool; it becomes extinct within a
few generations after its emergence. But from time to time a
particularly fortunate allele may increase in frequency
over many generations until it replaces all other alleles in
the pool it becomes xed. If each of the 2Ne alleles in the
pool has the same probability of becoming xed, the
probability of xation of the mutant gene is 1/(2Ne). The
xation rate, the probability of xation of one of 2Nem
mutations that occur every generation, is then equal to
(2Nem)/(2Ne) 5 m, the mutation rate. The average interval
between two successive xations is 1/m. The average
xation time (i.e. the time required for a mutation destined
for xation to achieve a frequency of 1) is, for diploid
organisms, 4Ne generations.
Mutations occur constantly in all organisms. The great
majority of them become extinct, regardless whether they
are neutral or selectively advantageous (although the latter
have a better chance of escaping elimination than the
former), but from time to time one of them rises in
frequency until it becomes xed. (Selectively disadvantageous alleles are normally all eliminated.) Since there are
many sites at which a mutation can occur in the DNA of
each organism, the probability that two species will x the
same mutation is low. The xation of distinct mutations
leads to molecular divergence of species their molecular
evolution. The essence of evolution at the molecular level
is, therefore, the dierential xation of mutations in
distinct groups of organisms.

Methods
Molecular evolution begins with allele-frequency changes
in a population. To study these initial changes, researchers
sample a population and test all individuals in the sample

for the presence of variants of a particular molecule. The

tests reveal dierent frequencies of alleles at a given locus in
the population polymorphism. In the case of neutral
alleles, polymorphism amounts to a snapshot of the
ascendancydescendancy process. For certain types of
advantageous alleles, polymorphism may be the result of a
balance between the eects of a selective advantage of one
allele relative to that of another. The balance may maintain
alleles in a population at relatively stable frequencies over
long time periods. Both neutral and balanced polymorphisms may be passed on to new species when these arise by
the splitting of an ancestral species (trans-species polymorphism). The choice of methods for the study of
polymorphism is inuenced by the necessity of testing
large sample sizes. The methods commonly used include
special modications of the electrophoresis technique, a
combination of PCR with restriction enzyme analysis, and
sophisticated immunological techniques.
Information about the xed dierences between taxa is
obtained by comparing their nucleotide sequences. It can
be expected that nucleic acid or protein sequences of more
closely related species will reveal a greater similarity to one
another than sequences of more distantly related organisms. Sequences can therefore be used to infer evolutionary
or phylogenetic relationships among taxa. To make such
inferences, a researcher rst aligns the sequences by
arranging them into rows in such a way that at each site
(position), the residues of the dierent sequences are in a
single column. Often alignments are only made possible
when gaps are articially introduced into the sequences;
these presumably correspond to insertions or deletions
(indels) of residues undergone during the evolution of the
molecule.
There are two principal methods for inferring phylogenetic relationships from aligned sequences. First, the
relationship can be assessed on the basis of the genetic
(evolutionary) distance between the sequences: the larger
the distance, the less related are the sequences and the
organisms. Genetic distance is obtained by dividing the
number of observed dierences between two sequences by
the total number of sites (positions) compared. In these
comparisons, however, the possibility that, at some sites,
identity may have arisen although there was a dierence in
an intermediate, or that the observed dierence has arisen
in more than one step, must be taken into account. In both
instances, the number of changes that actually occurred
during evolution would be greater than that indicated by
the observed dierences. However, the probability of these
hidden changes occurring can be assessed and the actual
distances can be estimated. The sequences can then be
clustered by placing those separated by smaller distances
closer together than those separated by larger ones. The
result is a phylogenetic tree that is presumed to reect the
evolutionary relationships among the sequences.
The second method strives to reconstruct the succession
of changes that led to the observed sequence dierences.

ENCYCLOPEDIA OF LIFE SCIENCES / & 2001 Nature Publishing Group / www.els.net

Molecular Evolution

The reconstruction is guided by the principle of parsimony,

which assumes that if multiple paths could theoretically
lead to the same state, the one taken by evolution is that
requiring the smallest number of evolutionary changes.
The justication for the use of this principle is that since the
changes are rare, the probability of multiple hits at the
same site is lower than that of a single change. The path
obtained by this maximum parsimony method, too, is
expressed in the form of a phylogenetic tree. To obtain the
most parsimonious tree, the researcher ideally considers all
theoretically possible pathways that could yield the
observed sequence dierences and then selects the one
requiring the fewest changes along the way.
Both distance and parsimony methods exist in many
versions. Various computer programs have been designed
to enable a tree to be drawn from the aligned sequences;
indeed, the alignment of sequences itself has also been
computerized. A single molecule-based tree, however, may
not accurately reect the evolutionary history of the
organisms from which the molecules have been isolated:
the gene tree may dier from the species tree. There are
numerous complicating factors that may be responsible for
the incongruence. One of them is trans-species polymorphism and the ultimate dierential xation of the polymorphic genes in the descendant species. Another is the
inaccuracy associated with attempts to estimate the true
number of changes at individual sites. Inferences that are
made regarding the relationships of species must therefore
be based on the examination of several genes or proteins.

the hypothesis holds true, it should be possible to use

proteins and nucleic acids for determining not only
relationships among organisms, but also their divergence
times from shared ancestors. The clock, of course, has to be
calibrated by nonmolecular (geological, palaeontological)
events, but once this is accomplished the clock can be
applied to groups for which no fossil record is available.
The molecular clock hypothesis, rst proposed in 1965,
has been controversial. Nevertheless, a certain regularity in
the accumulation of amino acid replacements can no
longer be doubted. However, proteins dier in the rate at
which they diverge. Some, such as brinopeptides, which
are fragments of proteins participating in blood clotting,
evolve rapidly, while others, such as histones, which form
the spools for winding DNA strands in a chromosome,
evolve very slowly. The reason for these dierences is that
proteins are dierentially constrained in their evolution by
their functions. Some, such as the brinopeptides, can
change considerably and still remain functional, while
most amino acid replacements in others, such as histones,
would endanger the proteins function. In the former,
therefore, many more mutations have a chance of being
xed than in the latter, in which virtually all mutations are
eliminated. Also controversial is the speed of the molecular
clock in dierent evolutionary lineages. Some researchers
have obtained evidence that in some lineages, for example
rodents among the mammals, the clock runs faster than in
others, such as primates. Other researchers contest these
ndings. At the nucleotide level, some regions of the
genome and some sites within a given region evolve faster
than others.

Molecular Clock
Phenotypic evolutionary changes are highly erratic. Some
lineages may change rapidly within a short time, while
others appear not to have changed their phenotypes for
many millions of years. Early protein sequence comparisons suggested that, at the molecular level, amino acid
replacements might accumulate at a much more regular
pace. Indeed, if a large proportion of nondeleterious
mutations were neutral in terms of their eect on tness,
their fate in the gene pool would be determined by random
genetic drift and xations could be expected to occur at
more or less regular intervals. It would, however, be a
stochastic regularity, rather like that leading to the
expectation of obtaining 50% heads in coin ipping: the
more times you ip a coin, the closer the match between the
observed and the expected result. The supposition that
molecules evolve at an approximately constant rate is
termed the molecular clock. According to the molecular
clock hypothesis, the number of amino acid replacements
(and by extension also the number of nucleotide substitutions) found between proteins (nucleic acid segments) of
two species is proportional to the time elapsed since the
divergence of these species from their common ancestor. If
4

The Neutral Theory of Evolution

As mentioned earlier, according to their eect on the tness
of their bearers, mutations are divided into three categories
disadvantageous, advantageous, and neutral. Disadvantageous mutations have a negative eect because they
lower the bearers chances of survival and reproduction
relative to that of other individuals, or they reduce the
number of its progeny. These mutations are eliminated
from the gene pool by negative (purifying) selection soon
after they have arisen. Advantageous mutations have a
positive eect because they improve the bearers chances of
survival and reproduction and thus increase the probability of being passed on to the following generations.
Relative to other mutations, advantageous mutations have
a greater chance of becoming xed: they are under positive
selection. Neutral mutations have no eect on the tness of
their bearers; they are inuenced neither by negative nor by
positive selection, but some of them can become xed by
drift. Researchers agree that deleterious mutations,
although they constitute the majority in terms of frequency
of occurrence, do not contribute to evolution. They also

ENCYCLOPEDIA OF LIFE SCIENCES / & 2001 Nature Publishing Group / www.els.net

Molecular Evolution

agree that most evolutionary changes of the phenotype are

eected by the xation of advantageous mutations. They
disagree, however, on the relative contribution of advantageous and neutral mutations to the divergence of nucleic
acids and proteins at the molecular level. The neutralists,
the proponents of the neutral theory of evolution, maintain
that the majority of sequence dierences observed between
nucleic acids (proteins) of two species are the result of
xation of neutral or nearly neutral mutations by random
genetic drift. The more traditionally minded biologists, the
selectionists, believe that not only at the phenotypic but
also at the molecular level most evolutionary changes are
the result of natural selection. Although the controversy
continues, it has become clear that if natural selection
inuences the xation of mutations, most of the time its
eects are very dicult to demonstrate. The debate has
also led to the realization that drift is a factor that cannot
be ignored in the explanations of evolutionary changes at
the molecular level. Finally, the neutral theory, with its
elegant mathematical backing, provides the foundations
on which more complex explanations of molecular
evolution can be built.

Evolution of the Genome

The replacement of one nucleotide by another at a given
site is one of the simplest evolutionary changes. Another
relatively simple change is the insertion or deletion of a
nucleotide or a group of nucleotides during DNA
replication. Such changes presumably take place when
the copying mechanism skips one or more nucleotides of
the template by mistake or when it copies the same
nucleotide twice. Reiteration of this replication slippage
generates repeats of the same nucleotide or of a group of
two or more nucleotides. These simple sequence repeats or
microsatellites then assume a life of their own, expanding
and contracting, and dierentiating individuals, populations and species. They usually do not exceed a length of
100 nucleotides, but occasionally they expand way beyond
their normal limits and can then cause genetic disorders
such as Huntington disease or the fragile X syndrome in
humans.
Repetitiveness is a general characteristic of eukaryotic
genomes. It occurs at several levels, in addition to that of
the simple sequence repeats. In the minisatellites, a core
sequence ranging in length from 9 to 65 nucleotides is
repeated tandemly 12 to 2000 times. The sequences of the
individual repeats are similar but not identical. A genome
may contain several thousands of scattered minisatellite
loci, each characterized by a unique core sequence that
itself is not internally repetitive, at least not in a simple
manner. There are many alleles at each of the minisatellite
loci that dier in the number and sequences of the repeats.
This variable number tandem repeat (VNTR) polymorph-

ism is widely used to distinguish and identify individuals of

a population.
In satellite sequences, the repeat unit may be as short as 5
and as long as 100 nucleotides, but it is repeated millions of
times. Satellite sequences are concentrated in specic
regions of chromosomes (the heterochromatin, centromere, telomere) or in specic chromosomes (e.g. the
mammalian Y chromosome). The sequences vary much
less than microsatellites and minisatellites, both in the
number and in the sequences of the repeats.
At the highest level, reiteration, in the simplest case
doubling or duplication, may aect part of a gene, the
entire gene, part of a chromosome, the entire chromosome,
or the whole genome (the entire hereditary material of a
cell). Duplication of part of a gene may result in the
addition of modules or domains to the encoded proteins
and the acquisition of new functions. Transfer of domainencoding segments between genes at dierent loci (exon
shuing) leads to the creation of mosaic genes.
Gene duplication is believed to be the most important
mechanism for the generation of new genes. Although the
extra copies of the gene commonly accumulate inactivating
mutations that turn them into nonfunctional pseudogenes,
occasionally some of them may acquire new functions and
evolve into new genes. Much of the complexity characterizing eukaryotic organisms may have evolved as a result of
gene duplications. Repeated duplications of a particular
gene or of a gene cluster result in the formation of gene
families, groups of loci engaged in similar or related
functions. The duplicated genes may remain in a tandem
arrangement or be scattered through the genome by
translocations and other mechanisms.
In plants and certain other eukaryotes, duplications of
individual chromosomes or whole chromosome sets often
lead to the generation of new species. Among animals,
large-scale duplications may have accompanied the
emergence of chordates and vertebrates.
A variety of mechanisms have been postulated to be
responsible for the observed changes of the genome. As
noted earlier, replication slippage seems to be primarily
responsible for the size variation at microsatellite loci.
Minisatellites may owe their variability primarily to
unequal sister-chromatid exchange, which involves recombination between two DNA molecules of the same
chromosome. Satellite sequences may be amplied by
extrachromosomal rolling circle replication, a copying
process in which the replication fork rotates around a
circular template for an indenite number of revolutions,
followed by reinsertion into the genome. Exon shuing
may either occur by nonhomologous recombination (i.e.
exchange between chromosomal segments containing
dissimilar genes) or be mediated by transposons (see
below; the excision of two copies of the same transposable
element anking an exon and their integration at another
chromosomal location may transpose the exon to another
gene). Most gene duplications probably arise by a

ENCYCLOPEDIA OF LIFE SCIENCES / & 2001 Nature Publishing Group / www.els.net

Molecular Evolution

misalignment of anking repeats in two homologous

chromosomes, followed by recombination between the
genes (unequal crossing-over). Once two or more copies of
a gene have arisen, the probability of further misalignments and unequal crossing-overs increases, leading to
more duplications of genes in some chromosomes and gene
deletions in others. As a consequence, the gene family
expands and contracts with time and the genes tend to have
similar sequences within a species as compared to another
species they are homogenized in the process. Another
process leading to sequence homogenization is gene
conversion, in which one gene interacts with another in
such a way that the sequences of the two genes begin to
resemble each other, often without the exchange of
anking markers. The process that leads to individual
members of a gene family in one species resembling one
another more than they resemble any members of the same
family in another species is called concerted evolution.
Duplications of whole chromosomes may result from a
failure of homologous chromosomes to separate during
meiosis (nondisjunction). In all these instances, the change
always occurs rst in a single individual and then spreads
through the population by genetic drift, selection, or both
processes.
Reiterations are responsible for changes in the size and
organization of the eukaryotic genome. Another factor
aecting the genome is the transposable elements. These
are DNA segments capable of inserting copies of
themselves into dierent regions of the genome. Some
elements transpose via RNA and others via DNA

intermediates. The former either possess their own reverse

transcriptase, the enzyme necessary for the conversion of
RNA back to DNA (retrovirus-like elements and retrotransposons), or rely on the use of reverse transcriptase
from other elements (retropseudogenes). The latter, the
transposons, possess a transposase, an enzyme facilitating
the excisioninsertion process. Transposable elements are
present in thousands of copies dispersed over most
eukaryotic genomes; they presumably maintain themselves
in the genomes by their ability to replicate (the selsh DNA
hypothesis). They often have deleterious eects on their
carriers in that their insertions disrupt the function of the
target genes. In some instances, however, they may
contribute to the emergence of evolutionary innovations
such as the generation of diversity in immunoglobulin
genes. Transposable elements may also be responsible for
horizontal transfers of genes from one species to another.

Further Reading
Gillespie JH (1991) The Causes of Molecular Evolution. Oxford: Oxford
University Press.
Kimura M (1983) The Neutral Theory of Molecular Evolution. Cambridge: Cambridge University Press.
Li W-H (1997) Molecular Evolution. Sunderland, MA: Sinauer
Associates.
Li W-H and Graur D (1991) Fundamentals of Molecular Evolution.
Sunderland, MA: Sinauer Associates.
Nei M (1987) Molecular Evolutionary Genetics. New York: Columbia
University Press.
Ridley M (1996) Evolution, 2nd edn. Cambridge: Blackwell Science.

ENCYCLOPEDIA OF LIFE SCIENCES / & 2001 Nature Publishing Group / www.els.net