THE EVOLUTION OF BIOLOGICAL DIVERSITY
So far we have covered only anagenesis,
or evolution within populations and species. In this and subsequent chapters,
we will be applying the same principles to evolutionary diversification. The first case to consider is spatial
evolution across the geographic ranges of single species. Subsequent chapters cover the evolution
of new species, or cladogenesis.
Finally, we will apply these ideas also to higher forms of evolution, macroevolution, or evolution above the species level.
Spatial
evolution of populations
Genetic divergence can be
classified into two major geographic modes:
1. Local disruptive selection - sympatric
divergence
Although sympatric evolution
(within a single local population, while populations remain in contact) is the
normal kind of evolution that we have been treating so far, sympatric models of
divergence are somewhat controversial, especially with respect to speciation,
as we shall see in a subsequent chapter (SPECIATION).
2. Geographically varying
selection (or drift)
a) Parapatric divergence
There is plenty of evidence for
parapatric divergence (evolutionary divergence between populations found in
different areas, but in contact at their boundaries). Again, what is more controversial is whether speciation can
result from parapatric evolution.
b) Allopatric divergence
Allopatric divergence (divergence
between geographically isolated populations) has been considered almost the
gold standard for evolutionary divergence of populations, and there is plenty
of evidence that it occurs. For example, island populations are often
genetically differentiated from mainlands. In the past, this led to the idea that almost all speciation
was due to divergence in allopatry, but these ideas are now being challenged.
Geographic distributions and
geographic divergence:
obviously, allopatric divergence may result in a parapatric geographic distribution via secondary contact after a period of geographic
isolation. This has seemed so
obvious and likely that virtually all parapatric distributions and hybrid zones
between them were assumed to be due to secondary contact (e.g. Mayr 1970).
However, the reverse is also possible, and indeed likely; allopatric
distributions could result from extinction of populations in a contact zone between parapatrically
distributed forms. Thus, it is not easy to infer the geographical mode of origin of a distribution just from its current status.
Genetic divergence and
speciation
Necessarily, speciation must
involve genetic divergence. Many newly formed pairs of species have parapatric
or allopatric distributions. Parapatric distributions and zones of contact or
hybridisation are particularly interesting to evolutionists because they
represent a first evolutionary stage leading to local diversification of
species coexisting in sympatry. Evolution in parapatry is the subject of this
chapter.
Genetic
variation across a geographic area
Any consistent change in gene
frequency, or in heritable phenotype, across the geographical range is known as
a cline. Clines occur because dispersal across a region is limited,
because individuals from the whole geographical area do not form a single panmictic
population.
[Added for completeness here; in book will
be dealt with in Chapter 8] Population
geneticists and evolutionary biologists often use the term migration
rather than dispersal, though they do not mean the same thing at all as
ecologists and behaviourists. For
instance, they are not referring to migratory movements where birds leave for
the winter, and then return to near their parents’ nest. Only the movement from birthplace to
place of breeding is considered migration in the genetic sense. Evolutionists
also use the term gene flow –
movement of genes between populations -- though they usually mean genotype
flow, the movement of whole
individuals, or genotypes.
Dispersal is never universal across
the whole geographic range. Other
processes, especially genetic drift or selection, can therefore outweigh the
homogenizing effect of the movement of genes. This can lead to a cline
maintained as a balance between local genetic drift and gene flow, or between
local selection and gene flow. Genetic diversity may thus be maintained between
populations via the spatial uncoupling of selection and drift in different
populations, as well as by processes we have already encountered within single
populations (see Chapter 4: SELECTION FOR AND AGAINST DIVERSITY):
a) Clines
produced by drift/migration balance
Random drift on its own will not,
of course, produce consistent directional changes in gene frequency. Locally,
however, drift can result in temporary monotonic trends with distance, or
clines (see Fig.). Because the effect is not usually very powerful unless local
population sizes are very small, drift will usually produce broad, random
clines, which are prone to reversal over time.
b) Clines
produced by selection/migration balance - EXTRINSIC selection
Extrinsic or environmental selection is imposed
directly by the environment. If two types of environment favour different genes
or phenotypes, if these two environments are sufficiently widely spaced, and if
migration rates are not too high, it is easy to see that this divergent
selection will set up a cline in gene or phenotype frequency. We have already
met an example of a cline maintained by extrinsic selection (peppered moths in
rural North Wales versus urban Manchester and Liverpool). Another example is
sickle-cell haemoglobin in malaria-infested vs. malaria-free areas of the
world, clines in heavy metal tolerance in plants, and even clines in the
frequency of insecticide resistance (see SELECTION AND THE SINGLE GENE).
[Added for completeness here; in book will
be dealt with in Chapter 8]
To
understand a little about the theory of clines maintained by a balance between
selection and gene flow, we will need to use a statistical description of
movement. The simplest method is
as follows:
Measuring dispersal
If the
dispersal of an individual between place of birth and breeding site is
essentially random, it resembles a "drunkards walk". It is called a
drunkard’s walk because the approximation is correct if the direction and
distance taken during each step is random. You may have encountered this sort of movement in
high-school physics; it has the same distribution as passive diffusion, a
two-dimensional normal distribution.
If this is
true, dispersal distance can simply measured as the standard deviation, s, of the distribution.
(Strictly, s is only a valid measure of dispersal if
dispersal is exactly normally distributed in two dimensions. Many field studies
have shown that dispersal is leptokurtic, i.e. non-normal where most
offspring breed very close to their parents, but some breed an enormous
distance away. But in practice, we can use the normal approximation even if
dispersal is non-normal, provided that leptokurtosis is not too extreme.)
Theory of
clines under extrinsic selection
If selection favours different
alleles in different areas, and if dispersal is not completely universal, it is
fairly clear that gene frequencies
may diverge and equilibrate in different parts of the range. The detailed
theory of clines requires solving some quite horrendous-looking differential
equations (a useful summary is
given by Roughgarden 1998), but we can appreciate the results without delving
into the mathematics too deeply.
At equilibrium between gene flow
and selection, the width of a cline (w = 1/[maximum
gradient in the centre], see Fig.) is proportional to dispersal divided by the
square root of selection. In fact:
w 1.7s/Ös
It is more important to understand
this result than to know how we obtain it.
First, it should, in retrospect,
obvious that the width of a cline scales directly to dispersal distance; the
cline will get wider as the dispersal s increases. Second, it is obvious that
stronger selection across an environmental gradient should result in a narrower
cline, i.e. w should be inversely proportional to some function of
selection. So the equation seems more or less sensible, though the square root
and the constant 1.7 comes out of the analytical maths, and assumes weak selection. You can test the details of this result
and its robustness to strong selection by running the CLINE SIMULATION model included
for different values of selection s
and migration s.
Also more important than
understanding the maths, we need to know why we want such an equation! It
provides us with a way to understand the evolutionary phenomenon of clines and
use it in analysing real data.
[BOX: Adaptation to patches
The
evolution of resistance to insecticides often causes insect control measures to
fail (see EVOLUTION AND THE SINGLE GENE). We might be able to use cline theory
to help with the problem. For example, if we were treating houses in an African
village with insecticides to prevent malaria transmission by Anopheles
mosquitoes, and we knew the selection pressures and dispersal distance, we
might be able to ensure that we avoid insecticide resistance evolving in a
village that is less than about 2w wide. This should work because two
back-to-back clines cannot form over a village unless it is the village is very
wide compared to the dispersal distance of the mosquitoes. Insecticide
resistance will not evolve in small villages because genetic swamping from
outside. Unfortunately for this
policy, many insecticides used in malaria control are also used in crop
protection outside the villages, so that resistance will evolve there as
well. The technique has frequently
been suggested, most recently by Lenormand & Raymond 1998, but has never
been tested carefully. A trial of
these ideas is currently under way in the control of Culex mosquitoes
near Montpellier in the south of France].
Cline theory was
used by Jim Bishop in 1972 to study the cline of melanism in the peppered moth
[See picture from Bishop’s paper]. Bishop obtained the cline theory by computer
simulation rather than by the above analytical theory, but the
principle is the same. He used a mark-release-recapture experiment to estimate
selection and dispersal along a transect between North Wales and Liverpool (see
SELECTION AND THE SINGLE GENE). He then compared the actual cline in melanism
with the predicted cline based on his computer model. Bishop found that the
cline width was much as expected from the model, but that the melanics reached
further into relatively pristine woodland areas of North Wales than
predicted. Bishop explained this
was probably due to higher fitness of melanics during the caterpillar stage;
this was not accounted for in the mark-recapture experiments on adult moths.
c) Clines
produced by selection/migration balance - INTRINSIC selection
i) Heterozygous disadvantage
Not all natural selection is
dependent on the environment; selection may instead be completely intrinsic.
We have come across example of this in heterozygous advantage and disadvantage
acting on a single locus or a chromosomal rearrangement (see Fig. above, and
CHROMOSOMAL EVOLUTION, MAINTENANCE OF GENETIC VARIATION). Heterozygous
disadvantage creates a particularly interesting kind of spatial disruptive
selection: as we have seen, equilibrium gene frequency, t/(s+t)
is unstable, and selection tends to prevent polymorphism. There are two peaks
in mean fitness, known as adaptive peaks; fixation for A, and
fixation for a.
Perhaps surprisingly, intrinsic selection
will produce clines with very similar shape to those found when extrinsic
selection is operating. When A is rare, it is selected against, where it
is common, it is the favoured allele, and so areas with different starting
frequencies of A will produce patches, separated by clines. At
equilibrium between selection and gene flow, the constant of proportionality is
found to be different for different models of selection, but the equations
describing equilibrium shape and width of clines will be very similar. Under
heterozygous disadvantage (Barton 1979),
w 2.8s/Ös', ... where
s' is an average of s and t.
Again, it is fairly easy to
understand the general feel of this equation: the stronger the selection, s,
the narrower will be the cline. The greater the dispersal distance, the
more blurred and broader will be the cline. Once again, the constant 2.8
emerges from the maths, and we can ignore it here.
But there is a big
difference. Intrinsic selection does not depend on the outside environment, it
depends only on the "internal environment" of each population, or
local gene frequency. So this means that there will be no tendency, except for
inertia, for a cline to remain rooted to the spot. If one homozygote is more
strongly selected than another (s t), the
cline will trundle gently around the landscape.
Many examples of
patchy distributions of chromosomal variants are known with broad, nearly fixed
populations separated by narrow contact zones in which the chromosomes are
polymorphic (Fig. from MJD White on grasshoppers). In these parapatric chromosomal races, intrinsic selection
maintains near purity of each race, and enforces a narrow cline connecting the
two.
ii) Frequency-dependent selection
Another example of intrinsic
selection is frequency-dependent selection. Here, the strength of selection
depends on the frequency of alleles in the population. An example is warning
colour, where the commonest form is the fittest because it will presumably have
already taught more predators to avoid the colour pattern. During learning, rarer
colour patterns will be attacked more heavily. This is true even if the same number of individuals of each
colour pattern die while teaching predators, because a higher proportion
of the rarer form will die. If colour patterns start off with different abundances
in different regions, the geographic range overall will stay patchy for the
colour patterns, and clines under intrinsic selection will form between them.
This actually happens in warningly coloured butterflies such as Heliconius
(see mimicry chapter).
It turns out that
frequency-dependent selection will result in almost exactly the same kinds of
clines as for heterozygous disadvantage. Heterozygote advantage or disadvantage
is, in fact, frequency-dependent at the genic level, when you think about it:
the direction and strength of selection depends on gene frequency.
iii) Epistatic and disruptive selection
If the environment is constant, but
selection is disruptive, there is a special case where intrinsic
selection may be caused by the environment. Here selection may favour a
bimodal phenotypic distribution, that is two adaptive peaks
simultaneously within the same area. For example, the Darwin’s finches have
available large, tough seeds, as well as small soft seeds, which are hard to
get out of their pods or off grass stems. One type of seed selects for stout,
deep beaks; the other for narrow pincer-like beaks.
These traits are usually controlled
at multiple loci, and bimodal adaptive landscapes usually imply epistasis,
or interactions in selection between genes (see SELECTION ON MORE THAN ONE
GENE). Suppose A and B both cause larger beaks, whereas
alternative alleles a and b both result in smaller beaks. AA
BB will have very large beaks, while aa bb will have very small
beaks, both of which may be near the bimodal optimal beaks. Intermediates with
genotypes Aa bb or Aa Bb, for example, will have beaks that
"fall between two stools". Loci A and B depend on each
other (which is the definition of epistasis) to produce optimal phenotypes.
Another and very important example
is sexual selection and mating behaviour, in which genes for mate choice are
strongly epistatic on genes for the traits being chosen. Clearly, epistatic
genes in mating behaviour might cause speciation directly by altering mating patterns
(see SPECIES AND SPECIES DIFFERENCES).
It may be very difficult, or even
virtually impossible for a randomly-mating population to achieve a bimodal
phenotypic distribution, even if this is favoured, because Mendelian
inheritance and recombination ensures that phenotypes determined by multiple
loci will be approximately normally distributed with a single mode - see
QUANTITATIVE GENETICS). Specialization on a second type of seed may ruin
adaptation to the first, and vice versa. In this case, selection sets up
stresses which multiple loci cannot easily resolve. There are three possible
outcomes:
Polymorphism. A single locus or "supergene"
polymorphism could evolve, such that each morph occupies a different habitat.
This is unlikely in the case of a multilocus trait like size However, it is not
unknown for size in fish to be dimorphic due to a polymorphism at single
growth-hormone receptor genes.
Batesian mimics like Papilio
memnon achieve this kind of polymorphism also.
Speciation. If there are actually two different
species with little gene flow between them, bimodality is easy to achieve
overall by evolution in opposite directions within each. Competition, in fact, will usually
ensure that the populations diverge to form a bimodal distribution overall. Thus, a possible resolution of the
disruptive selection might be to evolve into two separate species (see
SPECIATION).
Loss
of one adaptive peak.
Perhaps the most probable outcome is that populations will evolve towards one
adaptive peak or the other. Otherwise the majority of individuals, near the
phenotypic mean, would suffer the consequences of poor adaptation if a
polymorphism existed covering exploitation of both habitats.
If there is loss of different
adaptive peaks in different areas, this will lead to a patchwork of different
forms over the geographic range of a species. Whether the differences are initially determined by the
environment, or perhaps by different kinds of sexual selection in different
populations (see SEXUAL EVOLUTION), the end result is the same: parapatric
contact between divergent populations will result in clines at the boundary
that are maintained by a balance between intrinsic selection against
intermediates and gene flow.
Clines at epistatic loci are
stabilized in a similar manner to clines with heterozygous disadvantage or
frequency-dependent selection. The
only difference is that, with more than one locus, the analytical mathematics
becomes more intractable, and either new approximations or numerical
simulations must be used to study these systems. Perhaps for this reason, few detailed studies have been done
on epistatic clines. However, such
clines are undoubtedly common, and epistasis is an extremely important topic in
speciation, as we shall see.
Evolution of
clines
Any one of the various extrinsic or
intrinsic modes of
selection can stabilize clines of gene frequency (in the case of epistasis, at
more than one gene). Under Ernst Mayr’s "biological species concept",
species are reproductively isolated from one another (see SPECIES AND SPECIES
DIFFERENCES). Intrinsic and even extrinsic selection across clines can
therefore give a degree of "reproductive isolation", in that crosses
between the types produce poorly adapted heterozygotes or other kinds of intermediate
phenotypes. Knowing how spatial evolution can produce clines and how to use
cline theory is therefore important for understanding speciation.
It should be pointed out at this
stage that pure intrinsic or extrinsic clines are unlikely. Intrinsically selected chromosomal
morphs are often differentiated at many loci due to the suppression of
recombination; thus they can carry loci which affect extrinsic adaptation as
well. Similarly, ecological or
extrinsic adaptations may have pleiotropic effects on mating behaviour or
survival of hybrids which have intrinsic effects.
Hybrid zones
Many species and/or races are
distributed parapatrically. Hybrid zones are the narrow zones of contact
between divergent forms or species in parapatric contact. Hybrid zones may include
few hybrids or many, and the hybrids themselves may consist only of F1
only, or of F1, F2 and every kind of
backcross.
We have already seen examples in
the case of chromosomal races of mammals, and grasshoppers, where heterozygote
or hybrid disadvantage presumably maintain the clines. But we also see them for
frequency-dependent traits like warningly coloured butterflies, sexually
selected birds [see Fig. of parapatric manakin distributions in Colombia and
Panama].
Hybrid zones are usually first noticed
because they separate morphologically different taxa or chromosomal races. However, these races often differ at
multiple other traits as well. Races separated by hybrid zones usually differ
at one or many molecular markers, such as enzyme loci (allozymes) and DNA
markers like mtDNA, microsatellites, or nuclear DNA sequences (Barton &
Hewitt 1983). Thus hybrid zones consist of clines at multiple genes or genetic
elements like chromosomes.
Perhaps the record for character
differences across a hybrid zone is held by the European fire-bellied toad, Bombina.
There are two forms, Bombina bombina and B. variegata (Fig., Szymura & Barton 1986).
Bombina bombina Bombina variegata
Habitat Lowland Hilly
Water bodies Large ponds,lakes Small ponds, puddles
Skin thickness Thin Thick
Eggs Small, many Large, many
Belly warning colour Yellow Red
Other differences Male mating call
Hybrids develop less successfully
Immunological differences
Multiple allozyme differences
[see Fig.]
mtDNA differences
Hybrid zones, then,
are just places where narrow clines at multiple loci overlap. In a few cases,
however, hybrid zones for morphology or chromosomes are not associated with
other genetic differences. For example, in hybrid zones between races of the
warningly coloured butterfly Heliconus [see MIMICRY], there are no
differences in allozymes, chromosomes, or DNA, and hybrids are apparently as
fertile and viable as the normal pure forms on either side of the zone. In Heliconius,
the strong selection on colour pattern differences acts at only a handful of
gene loci that affect warning colour.
Summary:
importance of clines and hybrid zones
Cline theory is important for
understanding the limits to environmental adaptation. In some cases, migration
will swamp adaptations to a particular area. But wherever the cline width w
is substantially smaller than the environmental patches, there will be no
problem for adaptation; adaptation can occur in parapatry.
Cline theory also shows how genes
for intrinsic selection will evolve spatially-- that is, very similarly to
genes for environmental adaptation. The patchy structure of chromosomal races,
mating types, and other kinds of genetic architectures under intrinsic
selection can be almost as fine-grained as genes for environmental adaptation.
Furthermore, clines that are
maintained by purely intrinsic selection, may be able to move around the
landscape, and can, in theory, spread novel adaptations, even if they are
disfavoured when rare. The
evolution of these adaptations would have been impossible if the populations
were not connected with some spatial structure. There is evidence that some clines do indeed move rapidly
(refs.), but the positions of many clines and hybrid zones seem to have
remained stable over long periods of evolutionary history (Szymura & Barton
1991). Moving clines are an
intriguing possibility, but we cannot yet tell whether they are simply a
theoretical curiosity or whether they represent a common means by which
coadapted groups of genes spread to new populations.
Because hybrid zones consist of
multiple clines, cline theory enables us to understand the widespread
phenomenon of hybrid zones.
Hybrid zones separate forms, that,
like species, usually differ at many genes. Therefore the study of hybrid zones
give us a glimpse of an intermediate stage of evolution between simple
intraspecific polymorphisms and "good" species. Unlike
"good" species, hybridizing forms can be crossed, and we can
therefore use clines and hybrid zones study the kinds of genes that lead to
incipient speciation.