ZOOL 304

Neutral Evolution
and its effects on molecular evolution

"Population genetics remains the central intellectual connection between genetics and evolution."

"[T]he common thread throughout [population genetics] is what Gillespie calls 'the main obsession of our field,' the recurring question, 'Why is there so much genetic variation in natural populations?'"

"In a species with a million individuals, ...it takes roughly a million generations for genetic drift to change allele frequencies appreciably.  There is no conceivable way of verifying that genetic drift changes allele frequencies in most natural populations. Our understanding that it does in entirely theoretical."

 (The above points are quoted from an announcement for Population Genetics: A Concise Guide, by John H. Gillespie, Johns Hopkins Univ. Press)

The following notes (and the quoted headings) are adapted from Chapter 7 of Mark Ridley's textbook, EVOLUTION, 2nd ed. (1996), Blackwell Science, Inc., Cambridge MA. ISBN 0-86542-495-0.

• Introductory comments
• Overview
• CHECKLIST of important concepts and terms
• Highlights of a controversy
  • Two competing explanations.
  • Rates of molecular evolution.
  • Argument based on genetic load and cost of selection
    • Genetic load.
    • The 'cost' of natural selection.
    • Load appears too high molecular evolution based on selection.
    • Load appears too high for polymorphism based on selection.
    • Hard vs. soft selection; selection on many loci.
    • Argument is indecisive.
  • Argument based on consistency of rate for molecular evolution.
    • The molecular clock.
    • Ohta's "Nearly Neutral Theory".
  • Argument based on generation time effect.
  • Argument based on correlation of evolutionary rate with heterozygosity.

Introductory comments.

Protein and DNA sequences generally differ when compared in related species, with the number of differences approximately correlated with phylogenetic distance. To what extent is this molecular evolution driven by natural selection, and to what extent is it driven by neutral mutation and genetic drift?  This question remains controversial in current evolutionary theory.  

We shall be considering molecular evolution in terms of the processes that act to change (or stabilize) gene frequencies.  

Let's be perfectly clear.  No one disputes that natural selection is responsible for adaptation.  And no one disputes that drift can cause molecular evolution.  But it is not adequate simply to say that both selection and drift are important, since many other significant questions hinge on which one predominates.

In R.A. Fisher's widely accepted model of adaptive evolution, changes which have small effect are more likely to be successful than changes with large effect, as a general rule.  This is because small steps must move either toward or away from improvement, while large steps can miss the mark in any direction.  (This is most readily visualized in terms of the adaptive landscape.)  Therefore, most evolutionary changes will (must) be small.

In neutral theory, most evolutionary changes have no functional effect whatsoever (i.e.,they are "neutral").

The controversy concerns both the relative importance of these two real mechanisms (selection and drift) as the principal explanation for most molecular evolution and the relative influence of these two mechanisms in any particular example.

Since this issue remains unsettled, we will be considering not only the two competing hypotheses, but also the observations and areguments which attempt to distinguish between the two alternatives.

The following presentation roughly tracks the history of the controversy, with argument and counterargument as they were introduced over the past few decades.

  304 index page

CHECK LIST of important CONCEPTS and TERMS

• Neutral Theory controversy
• Mutation
  • Neutral mutation
  • Deleterious mutation (negative selection coefficient)
  • Favorable mutation (positive selection coefficient)
  • Frequency distribution for mutation rate vs. selection coefficients
• Protein sequence evolution
  • Amino acid substitution
• DNA sequence evolution
  • Base-pair substitution
  • Silent substitution
  • Replacement substitution (meaningful substitution)
• Rate of molecular evolution (rate of substitution)
  • Mutation rate
  • Substitution rate
• Genetic load
  • Optimal fitness
  • Mean fitness
• Cost of selection
  • Substitutional load
  • Segregational load
  • Hard and soft selection
  • Multiplicative (independent) fitnesses
• Molecular Clock
  • Relative rate test
  • Generation time effect
  • Functional constraint effect
  • Pseudogene
• Slightly deleterious mutations (the "Nearly Neutral Theory")
• Fisher's model of adaptive evolution
• Codon usage bias
• Correlation of heterozygosity with evolutionary rate

What is the controversy?

Extensive variation is observed when homologous proteins are compared.  Do these differences reflect:

• Adaptive optimization of each variant by natural selection?
         OR
• Random shuffling by genetic drift among adaptively equivalent variants?

Prior to the advent of neutral theory, most evolutionary biologists were accustomed to thinking of evolutionary change as adaptive.  Therefore, there was an early presumption that molecular variation between species reflected adaptive changes as the species diverged.  But molecular variation proved to be so extensive that this presumption became difficult to defend.

The presumed distribution of selection coefficients among mutations is one part of the controversy.  Even if we ignore substantially deleterious mutations (which everyone agrees do occur and are quickly eliminated by selection), does the normal supply of mutation include:

• A multitude of possibilities for adaptive improvement?
         OR
• A variety of adaptively equivalent mutants?

The hypothesis that neutral mutation and drift together can account for most molecular evolution is called the Neutral Theory.  The formulation of this influential concept in the 1960s and 70s is most closely associated with the name Motoo Kimura.  T. Ohta later introduced the significant modification known as the "nearly neutral theory".

"Pan-neutralism", the extreme suggestion that ALL molecular variation is neutral, is not a serious contender.  Nevertheless, this idea should be noted just to clarify the contrast between the two major competing theories.  

Addressing the controversy calls for data, most specifically measurements of rates of molecular evolution and amounts of resulting genetic variation.

The rate of molecular evolution can be measured for both protein sequences and DNA sequences, by counting sequence differences between two related species and dividing by twice the time since they diverged from a common ancestor.  (Why twice the time?  Because differences could accumulate independently in each species since divergence.)

Genetic variation is measured as proportion of polymorphic loci within species.

What follows is an extended argument, intended to explore the consequences if most sequence differences resulted from the action of selection.

The concept of "genetic load" offers a way to estimate the effect of deleterious mutations.

Genetic load is defined as optimal fitness minus mean fitness.  Optimal fitness is to be understood not as some imagined perfect adaptation, but as the fitness of the actual genotype that has the highest fitness of all available genotypes. Since mean fitness is the proportion of survivors, genetic load is the fraction of the population removed by selection.  

The genetic load that exists as the result of deleterious mutation is called mutational load.  

A standing genetic load can also result from heterozygote advantage ( see below), where it is called segregational load.

Genetic load is also incurred, temporarily, whenever a better allele (which allows for a higher optimal fitness) is introduced into a population.  The load decreases as the advantageous allele increases in frequency, until the new allele is fixed.

Because of genetic load, the selection of a favored allele carries a cost, called by Haldane the cost of selection.

(J.B.S. Haldane was one of the founders of the Modern Synthesis.)

Whenever a "new, improved" allele is introduced into a population, there will be a genetic load associated with selection against the "old, inferior" allele.

The cost of selection is the accumulated genetic load over all the generations that pass until the "new, improved" allele is fixed.

If one knows how many alleles have been substituted over time (as one can simply by counting sequence differences between species, above), one can calculate the cost that must have been incurred if selection were responsible for each substitution.

We will see later (below) that the very idea of the cost of selection assumes that most excess reproductive capacity is just waste rather than loss to selection (with loss implying that any selection actually reduces the population below its capacity at optimal fitness).

304 index page

It can be argued that, because the genetic load and the cost of selection would be too high, that measured rates of molecular evolution are too high to be explained by natural selection.

Since we can measure the rate of gene substitution (above), we can calculate the overall cost of selection if all of these substitutions resulted from selection.

M. Kimura (the founder of neutral theory), calculating from Haldane's model, showed that the cost of selection, if all observed substitutions in molecular sequences resulted from selection, would far exceed the available number of organisms.  However, there is no genetic load when neutral alleles drift to fixation.

Two conclusions might be reached from this result.

Selection is not the right explanation for most molecular substitutions.
           OR
Some assumptions in Haldane's model are not appropriate.

Kimura chose the first conclusion, and gave us the neutral theory.  With a high enough rate for neutral mutation, genetic drift can explain molecular evolution.

304 index page

By a similar argument, if the amount of polymorphism in a typical population appears too high to explain by selection for for heterozygote advantage.

If genetic variation were maintained by heterozygote advantage, genetic load would result from selection against the homozygotes.  This is called segregational load.

Kimura calculated this load.  If all of the many numerous molecular polymorphisms that can be observed were independently maintained by heterozygote advantage, the total mortality resulting from this segretional load would be impossibly high.  But random genetic drift of neutral mutations can maintain polymorphism with no genetic load.

Again (see above), two conclusions might be reached.

Selection must not be the right explanation for most molecular polymorphism.
           OR
Some assumptions about segregational load involving multiple loci may not be appropriate.

Again, Kimura chose the first conclusion and gave us the neutral theory.

However, the observed frequency of molecular polymorphism is not nearly as high as naively expected from neutral theory.

304 index page

However, there are some plausible explanations for extensive polymorphism and high rates of molecular variation, that are based on selection.

A much higher genetic load may be sustained if selection is "soft" rather than "hard".

• Hard selection is defined as selection which reduces standing population size.
• Soft selection is defined as selection which does NOT reduce standing population size, but occurs entirely within the excess reproductive capacity of the population.
• Note that with hard selection, most "excess" reproductive capacity essentially goes to waste.  With soft selection, selection occurs within this excess capacity so population size is not reduced by selection.
• The argument regarding cost of selection (above) assumed hard selection.
• If selection is soft, then the cost of selection poses no problem for selection as an explanation for molecular evolution at observed rates.
• Note that the possibility of soft selection in no way refutes the neutral theory, it merely preserves the possibility that selection might still be a viable explanation.

"Natural selection can act jointly on many genetic loci."

• An argument regarding segregational load assumed that selection must operate independently on each locus.
• If loci are not independent (technically, if loci do not have multiplicative effects on fitness), then the segregational load of many suboptimal alleles can be greatly reduced.
• Again, note that the possibility of non-independent effects on fitness in no way refutes the neutral theory but merely preserves the possibility that selection might still be a viable explanation.
• One more point. Selection can maintain polymorphism not only by heterozygote advantage but also by frequency-dependent selection.  And, unlike heterozygote advantage, there is no necessary cost associated with frequency-dependent selection.

Considerations of genetic load and cost of selection are therefore indecisive.

"A neutralist could 'explain' almost any evolutionary rate or heterozygosity by positing suitable values for N or u" (effective population size and mutation rate).

"The selectionist can [also] explain almost any observation by inventing suitable values for unknown numbers."

304 index page

A second argument may be based on the apparent constancy of rates of molecular evolution.

Rates of molecular evolution are surprisingly steady across related lineages, at least when homologous molecules are being compared.

This result is called the molecular clock, although the term can be misleading.

Although this result does not contradict a selectionist model, it was NOT expected from a selectionist model.

This result IS expected from the Neutral Theory.

The comparison of alpha and beta hemoglobins among human, shark and carp is intended to highlight this surprising result.  However, the argument presented is incomplete.  To support the conclusion that differences between the alpha and beta forms have separately accumulated at the same rate in human, shark, and carp lineages (details are described in Ridley's text), we need not only data showing how many substitutions have happened but also data to show when they occurred.  (From the data given, it could be possible that the observed substitutions all happened immediately after the globin gene duplicated but before the human, shark and carp lineages had diverged.)  If we can assume a consistent molecular clock for this gene, then all three lineages show similar rates of change (since similar numbers of substitutions have occurred).  Tabulation of alpha-alpha differences and beta-beta differences between the various pairs of species do support this conclusion.

304 index page

There are some additional complications.

• The protein molecular clock appears to count elapsed time rather than number of generations.
   
  • A molecular clock is expected to count generation time because known mutational mechanisms seem to associate with DNA replication rather than with simple passage of time.  DNA is generally replicated less often in species with long generation times.
  • However, the molecular clock for protein evolution actually matches absolute time better than generation time.
  • This result was not only not expected from initial neutral theory, but could be seen as contradiction of neutral theory.  However, see below.
     

• The molecular clock for silent substitutions in DNA evolution does show a generation time effect.
  • So, in this case where mutations are expected to be neutral in fact, the neutral theory does seem to fit.

304 index page

"Slightly Deleterious Mutations in the Neutral Theory"

• Ohta introduced a variation called the "Nearly Neutral Theory".
• Basically, this theory offers a friendly modification of neutral theory to correct for the generation time effect.
• Creatures with longer generation times generally have relatively smaller population sizes.  With small population sizes, slightly deleterious mutations (which would be effectively selected against in large populations) will behave by drifting like effectively neutral mutations.  Thus as generation time increases, its effect on clock rate will be compensated by an increase in the rate of effectively neutral mutations

304 index page

A third argument is based on functional constraint.

The basic idea here is that different parts of proteins, and different parts of genes, differ in how much change can be tolerated without deleterious effect.  If little change can be tolerated, that region is said to be "constrained".    Constrained molecules are expected to evolve at lower rates.

• In neutral theory, functional constraint means that neutral mutation rate is reduced so evolutionary rate is also reduced.
• Neutral theory does not deny that selection operates to resist deleterious changes; it only presumes that most molecular changes that do evolve result from neutral drift rather than advantageous mutations.

Neither explanatory mode (selection or neutralism) has a distinct advantage.

• Under neutral theory, the rate of neutral mutations relative to the rate of deleterious mutations will be lower in the the more constrained regions, so neutral substitutions will accumulate more slowly.
• Under selectionist theory, favorable mutations are less likely in constrained sites (because in such sites any change is likely to have a large effect), so selectively advantageous small changes are more likely in less constrained locations.

However, the argument and counterargument can become more subtle than this.

• If there were such a thing as a totally unconstrained sequence, where any mutation would be neutral, then the substitution rate at that site would indicate the actual mutation rate.  (The average rate of substitution by drift is expected to equal the mutation rate, for truly neutral mutations.)
• To a first approximation, silent codon substitutions and substitutions in pseudogenes might be expected to be totally unconstrained.
• The observed more-rapid evolution in silent sites and pseudogenes IS expected by neutral theory.
• Significant differences between between different pseudogenes or between the silent sites of different genes are NOT expected by neutral theory.  These differences suggest that substitutions in such sites are constrained after all, and hence selection must be acting on such sites.
• If silent substitutions were truly unconstrained (i.e., if all synonymous codons were functionally equivalent), then only random variation should appear in the numbers of synonymous codons.  
• However, in fact there are well-established biases that favor some codons over others.
• Something, maybe selection in relation to DNA secondary structure, maybe selection in relation to available tRNAs, maybe directional mutation pressure, must be preventing silent substitutions from evolving simply by drift.

 

  304 index page

A fourth argument is based on the correlation of rates of evolutionary change of molecules with their heterozygosities.

In the neutral theory, the rate of evolutionary change (nucleotide substitution) should be equal to the neutral mutation rate, so the ratio of substitution rates for two sequences should be the same as the ratio of mutation rates.

If the the neutral mutation rates vary because of functional constraints, so also should the evolutionary rates.

In the neutral theory, the ratio of heterozygosities for two sequences should also be (approximately) proportional to the ratio of neutral mutation rates.

Therefore, rates of change due to neutral drift should correlate with heterozygosities due to neutral drift, and the ratios should be equal, regardless of functional constraint.

Selection offers no reason to expect such a correlation, except by chance.

Data from this test are ambiguous.

304 index page

Comments and questions: dgking@siu.edu
Department of Zoology e-mail: zoology@zoology.siu.edu
Comments and questions related to web server: webmaster@science.siu.edu