Transcript Slide 1

Topic 15. Lecture 21. Microevolutionary mechanisms of
Macroevolution.
We already considered genetic variation of natural populations and factors that affect this
variation (separately and together). Finally, we are prepared to address the key question:
what happens in natural populations? We will concentrate on changes driven by positive
selection, as they are behind adaptive Macroevolution. However, most of genetic variation
present in natural populations is irrelevant for adaptive Macroevolution, and we need to
briefly consider such variation, too. Specifically, we will treat the following subjects:
i) How strong could natural selection be in natural populations?
ii) Inbreeding depression and outbreeding depressions - two pervasive phenomena.
iii) How common are selectively neutral variation and allele replacements?
iv) How common are unconditionally deleterious alleles, which will never be fixed?
v) Microevolution of extinction.
vi) How common are positive selection-driven allele replacements?
vii) How Microevolution constraints the rate and the course of Macroevolution?
i) How strong could natural selection be in natural populations?
1) Is natural selection consistent with survival of the population? All kinds of natural
selection are involved with some genetic load. Thus, if the maximal contribution of an
individual to the next generation is one offspring (two for sexual diploids), there could be no
selection - any selection will drive the population to extinction.
However, in every population, there is an opportunity for selection, because maximal
fecundity is always higher than 2 - apparently, it is never below 10 (great apes, some birds of
prey). Also, contributions of males usually vary greatly, creating more opportunity for
selection. Apparently, any natural population can take a genetic load of at least 50%.
2) Is natural selection operating within natural populations? It is hard to measure genetic
load experimentally, but it can be done for variance of relative fitness Var[w i/W]. This
quantity can be referred to as "opportunity for selection": if Var[wi/W] = 0, there could be no
selection. However, data consistently show that fitness and its components are variable
within natural populations.
Opportunity for selection on some fitness-related traits in Drosophila melanogaster
Fecundity
0.15
Longevity
0.07
Developmental time
0.002
All natural populations, except abnormally small ones, possess a lot of non-neutral genetic
variation. Of course, not all of this variation may be used for adaptive evolution. Traits that
are closely related to fitness (viability, fecundity, mating success, longevity) have highest
evolvabilities (Genetics 130, 195-204, 1992).
Sexual selection creates extra opportunity for selection among males, because often a small
fraction of males obtain most of matings.
Number of mating partners of an individual in a natural population of green swordtails
(Molecular Ecology 17, 4522-4534, 2008).
ii) Inbreeding depression and outbreeding depressions - two pervasive
phenomena.
A) Inbreeding depression. Offspring from matings between closely related parents
consistently have reduced fitnesses. This inbreeding depression increases with the
coefficient of inbreeding F, the fraction of allele, at the same locus, that are "identical-bydescent", in the sense so they can be traced to the common ancestor.
Inbreeding depression has been detected in almost all experiments that studied the impact
of inbreeding on fitness. Traits that are closely related to fitness (viability, fecundity, mating
success, longevity) display the largest values of inbreeding depression. Very roughly, the
increase of F by 1% cause the decline of a fitness-related trait that is also ~1%. Thus,
complete homozygosity leads to fitness that is close to 0. This is not surprising, as long as
an average individual carries ~1-2 heterozygous recessive lethals. In fact, inbreeding
depression is caused not only by recessive lethals, but also by deleterious alleles with
individually mild effects.
Inbreeding depression of fertility in a tree-hole-breeding mosquito Aedes geniculatus (Proc.
Royal Society B, 267, 1939 - 1945, 2000).
Inbreeding depression may depend on the environment under which fitness is assayed. Not
surprisingly, it is often higher under tough, competitive environments.
The reproductive success of inbred (solid bars) and outbred (open bars) mice. (A) Males in
enclosures. (B) Males in laboratory (PNAS 97, 3324-3329, 2000). Male mice are really mean!
Naturally, inbreeding increases the fraction of homozygous loci in offspring, and this
probably is the cause of inbreeding depression. Still, there are two (not mutually exclusive)
feasible explanations of inbreeding depression - dominance and overdominance.
Overdominance (red): Aa has the highest fitness, and, thus, homozygosity is always bad.
Dominance of beneficial alleles (blue): one of homozygotes (aa) has reduced fitness, and,
thus, homozygosity is bad on average.
Overdominance is rare, so that dominance (incompatibility epistasis between deleterious
alleles) is probably the leading cause of inbreeding depression.
B. Outbreeding depression. Matings between close relatives lead to low fitness of offspring,
but matings between parents from different populations (from "the same species") may lead
to the same effect - and this is called outbreeding depression.
Mass of individual seeds versus pollination distance in the marine angiosperm Zostera
marina (Marine Biology 152, 793-801, 2007).
Thus, often there is an intermediate optimal genetic distance between parents. As
inbreeding depression, outbreeding depression is due to incompatibility epistasis, but
between alleles of different loci. We will return to this subject when we consider speciation.
iii) How common are selectively neutral variation and allele replacements?
In short, very common. The highest values of H are present at sites that are under no or only
very weak selection. In compact genomes of prokaryotes and of some eukaryotes, such as
Drosophila melanogaster (Nature 437, 1149-1152, 2005), such sites are rare, but they
represent at least 80% of mammalian genomes. An average human carries ~5,000,000
(effectively) neutral variants (with s < 1/Ne < 10-5), most of them rare (H = 10-3, diploid
genome size = 6x109).
Because neutral alleles at a locus become fixed at a rate that equals to mutation rate, the
total number of selectively neutral allele replacements per generation was ~50 in the human
lineage. That is how humans and chimpanzees accumulated ~50,000,000 genetic differences
(mostly, single-nucleotide replacements) in the course of only ~1,000,000 generations of
independent evolution.
In nonmammalian species, these numbers may be lower, but still impressive. Thus bulk of
genetic differences between species is due to (effectively) neutral allele replacements.
Indeed, it is hard to imagine positive selection that drives 50 adaptive allele replacements
every generation (although theoretically this might be feasible).
iv) How common are unconditionally deleterious alleles, which will never be fixed?
Also quite common, because a lot of such alleles are only mildly deleterious - this is enough
to prevent them from being fixed, but not enough to keep them very rare.
An average Drosophila, fish (and probably human) carries 1-2 heterozygous recessive
lethals, which reduce fitness by ~1%.
An average normal human carries at least ~30 deleterious alleles that completely inactivate
the affected proteins (nonsense substitutions, frameshift insertions and deletions, etc.), but
still are not recessive lethals, because these proteins are not essential. Coefficients of
selection against such alleles are ~10-2-3.
An average "normal" human carries at least ~1000 nonsynonymous mildly deleterious
variants (out of ~10,000 nonsynonymous variants totally) with coefficients of selection
against them ~10-3-4 (Hum. Mol. Genet. 10, 591-597 2001).
Perhaps, a number of mildly deleterious non-coding alleles is even higher. However, almost
all drastic deleterious alleles affect protein sequences.
Data for other eukaryotes are probably similar. Humans are just better studied.
So, only ~10% of human non-synonymous variants are substantially deleterious, having
selection coefficients above 10-5, at the very least. However, we know that ~90% of human
non-synonymous mutations are substantially deleterious. Obviously, deleterious allele are
underrepresented in within-population variation, due to negative selection.
Probabilities that an amino acid
replacement is associated with the
coefficient of selection s > 10–2, 10–2–
10–4, 10–4–10–5 or < 10–5. Hum. Mol.
Genet. 14, 3191-3201, 2005). Grey:
probabilities among fresh mutations.
Red: probabilities among segregating
alleles. Neutral and mildly deleterious
variants are overrepresented.
Thus, the bulk of genetic variation
within every population is either
irrelevant for adaptive evolution
(effectively neutral variation) or has
not chance to become fixed within an
evolving lineage (unconditionally
deleterious variation). Nature is
wasteful, as usual.
v) Microevolution of extinction.
In fact, there is one aspect of Microevolution that is directly affected by deleterious
mutations and negative selection - extinction. Extinction is a complex phenomenon - unless
it is caused by a complete habitat destruction or by direct killing of every individual ("hard
extinction").
A soft extinction involves an ecological component:
1) critical low density (Allee effect) - it could be that if too few individuals are left,
they cannot reproduce
2) demographical stochasticity - in a small population there is a substantial chance
that the number of individuals will decline further "by chance"
and a genetical component:
1) inbreeding depression due to homozygosity of pre-existing deleterious alleles
2) fixation of new deleterious mutations due to inefficient selection under low Ne
3) loss of genetic variation, which prevents response to positive selection
Genetic deterioration of ecologically declining populations plays an important role in soft
extinctions. Low population size leads to genetic deterioration, and vice versa, leading to a
self-reinforcing process known as mutational meltdown. Two kinds of data demonstrate the
importance of genetic deterioration in extinction.
1) Very small populations on the verge of extinction often contain genotypes of low quality,
presumably due to fixation of mildly deleterious alleles. This can be seen by the results of
crosses of such genotypes with members of a related, "healthy" population, if such a
population is available.
South Florida houses a small, isolated,
inbred and distinct subspecies of
panthers (Puma concolor coryi). Once
part of a continuous, widespread
population, panthers became isolated
in south Florida more than a century
ago. Numbers declined and the
occurrence of genetic defects
increased. Hoping to reverse the
genetic damage, managers introduced
eight female panthers from Texas into
south Florida in the mid-1990s. Hybrid
kittens have about a three times higher
chance of becoming adults as do
purebred ones. Hybrid adult females
survive better than purebred females;
there is no obvious difference between
the males. Hybrids are expanding the
known range of habitats panthers
occupy and use (Animal Conservation
9, 115-122, 2006).
The Florida panther Puma concolor coryi .
2) If natural selection is artificially abolished, mean population fitness declines rapidly, by 12% per generation, due to unopposed accumulation of new deleterious mutations. This can
be seen from data on "Middle Class Neighborhood" (MCN) populations and on
reintroductions of individuals from captive populations into the wild.
An MCN population: if every pair contributes exactly two offspring to the next generation,
natural selection is impossible (although genetic drift still occurs slowly).
In an MCN population, fitness declines by 1-2% per generation.
Decline of fitness, measured under tough conditions, in populations of different sizes
maintained in captivity under relaxed natural selection (Proc. Natl. Acad. Sci. USA 94, 1303413039, 1997; Conservation Genetics 3, 277–288, 2002; Biological Conservation 126, 131-140,
2005; Ann. Rev. Ecol. Evol. Syst. 37, 433-458 2006).
vi) How common are positive selection-driven allele replacements?
Finally, we can consider the only aspect of Microevolution that is directly responsible for
adaptive Macroevolution - allele replacements driven by positive selection. There are two
main questions:
1) What is the per generation number of number of selection-driven allele replacements
accepted by an evolving population?
2) What are coefficients of selection that drives these replacements?
Due to difficulties in detecting positive selection, we still do not have definite answers to
these fundamental questions. However, there are some estimates, for a small number of
carefully studied species, first of all for several species of Drosophila and for humans.
Positive selection in Drosophila melanogaster lineage. D. melanogaster is originally an
African species that dispersed into other parts of the Old World ~15,000 years ago. Since the
divergence of African and non-African D. melanogaster, there were ~200 selection-driven
allele replacements in Africa and ~300 such replacements outside Africa. Assuming 20
generations per year, this implies 1 adaptive allele replacement per 1000 generations. Most
of these replacements were driven by selection with coefficients 0.05%-0.5% (PLoS Genet 2,
e166, 2005). I tend to believe this estimate, although I am not 100% sure. Perhaps, adaptive
allele replacements are more common, but hardly more than 1 per 100 generations (Trends
in Ecology & Evolution 21, 569-575, 2006).
Positive selection in the human lineage. As we know, it must be easier to detect recent or
even ongoing episodes of positive selection, through the pattern of genetic variation.
Extensive data on genetic variation within humans are available, but
so far they did not translate into any credible quantitative estimates,
although several cases, such as lactose tolerance, are well
established. What we have now is a mess.
This table demonstrates discrepancies in estimates of positive selection in humans. For
each pair of studies, is presents the number of genes that were identified as being under
positive selection in both of them (Nature Reviews Genetics 8, 857-868, 2007).
It seems very likely, however, that the average rate of selection-driven allele replacements in
human evolution was less than 1 replacement per 100 generations, but we do not know how
much less, and there are no credible estimates of the strengths of selection involved.
Putative conclusions about positive selection:
1) A vast majority of non-neutral genetic variation within every population is totally
irrelevant to its evolution - it is due to segregation of rare, derived, deleterious alleles that
will never be fixed.
2) A vast majority of allele replacements is irrelevant to adaptive evolution - these
replacements involve selectively neutral (or almost neutral) pairs of alleles.
3) Allele replacements driven by substantial positive selection are not very common (1 per
100 or 1000 generations) and, thus, probably overlap only very little.
Still, we have no idea, how many adaptive allele replacements are responsible for the
transition from humans to apes. How many chimpanzee alleles one need to replace to turn it
into a passable human? 100? 1000? 10000? 10000?
vii) How Microevolution constraints the rate and the course of Macroevolution?
Darwinian mechanism of Microevolution is very inefficient, in comparison with imaginary
Lamarckian mechanism. Thus, it makes sense to ask the above question. There is no
comprehensive answer, but we probably can say the following:
1) Microevolutionary mechanism prevent long-term survival of lineages represented by too
few individuals at a time. At least 1,000 or even 10,000 individuals are needed.
2) The course of Macroevolution is severely constrained by availability of mutations - an
insertion of a particular sequence of 10 amino acids into the protein may never occur.
3) The rate of Macroevolution is limited by availability of even relatively common mutations:
a population of 100,000 individuals will need to wait for 100,000 generations before a
nucleotide substitution will occur with probability 10-8 and survive the initial random drift
with probability 10-2 (assuming 1% selective advantage).
4) Fixations of individual mutations make Macroevolution unpredictable. This is a huge
magnification of quantum effects.
5) It does not seem that inherent inefficiency of natural selection imposes a strong
constraint on Macroevolution.
Quiz:
What features of a population and of its environment can be expected to cause a
high rate per generation number of adaptive allele replacements in it? In
particular, who should accumulate beneficial mutations faster, fruit flies or
humans?
Hint: think of possible effects of mutation rate, population structure, drift, and
changes of external conditions.