The neutral theory

Concept

The neutral theory

T-086Home BU-304Threads evolution

Statement

Most molecular evolution is selectively neutral.

Why it matters

natural-selection (from the introductory unit) establishes that differential reproductive success shapes trait frequencies, and it is tempting to assume every observed change in gene sequence must therefore reflect selection acting on it; the neutral theory is the deliberately contrarian, and now well-supported, claim that most molecular-level evolutionary change is not driven by selection at all, but by genetic-drift acting on mutations that are, at the level of fitness, essentially invisible to selection. molecular-clock depends directly on this idea: rate constancy is far more plausible for neutral sites, whose substitution rate the neutral theory shows is set by the mutation rate alone, than for sites under variable selective pressure.

It matters because it supplies the correct null model for detecting selection: without a clear expectation for what molecular evolution looks like in the complete absence of selection, there would be no baseline against which unusually fast or unusually slow substitution rates could be identified as evidence that selection is, in fact, acting.

Hypotheses

A substantial fraction of new mutations at the molecular level are selectively neutral or nearly neutral — they neither improve nor substantially reduce fitness.This is the theory's foundational and most contested claim: without a large neutral (or nearly neutral) class of mutations to draw on, genetic drift would have little raw material to act on, and molecular evolution would instead be dominated by selection as the classical, pre-1968 view assumed. Population size is finite, so genetic drift (random sampling of alleles between generations) has a real, nonzero effect on which neutral variants are ultimately fixed or lost.In an infinitely large population, drift would have no effect at all and every neutral variant's frequency would remain exactly constant indefinitely; it is specifically the finiteness of real populations that allows neutral alleles to drift to fixation or loss over time, which is what makes the whole theory's predictions about substitution rate (Step 3) apply to any real, finite population regardless of size.

Proof

\text{A new neutral mutation arises in a diploid population of effective size } N_e\text{, at rate } \mu\text{ per gene copy per generation.}

Total new neutral mutations arising per generation across the whole population is \(2N_e\mu\) (diploid, two gene copies per individual); each individual copy is, by definition of neutrality, equally likely as any other to eventually be fixed by drift, since none carries any fitness advantage over another. B

P(\text{fixation of any one particular new neutral copy}) = \frac{1}{2N_e}

Because a neutral allele's ultimate fate under drift depends only on random sampling, not on any fitness difference, its probability of eventually reaching fixation equals simply its initial frequency in the population — one copy out of \(2N_e\) total gene copies. B

k = (2N_e\mu)\times\frac{1}{2N_e} = \mu

The rate of neutral substitution \(k\) (fixations per generation) equals the number of new neutral mutations arising per generation (Step 1) multiplied by each one's fixation probability (Step 2); the population size \(N_e\) cancels exactly, leaving substitution rate equal simply to the mutation rate \(\mu\), entirely independent of population size. A

\text{Because } k=\mu\ \text{regardless of } N_e\text{, neutral substitution rate is predicted to be approximately constant across lineages with very different population sizes.}

This population-size independence is exactly the property molecular-clock needs to justify treating substitution rate as roughly constant across very different organisms, which otherwise differ enormously in population size, generation time, and every other demographic parameter that would ordinarily be expected to affect an evolutionary rate. A

\text{Nearly neutral extension: mutations with small fitness effect (}|s|\lesssim1/N_e\text{) behave effectively neutrally in small populations but not in large ones.}

Whether a mutation with a genuinely small (nonzero) fitness effect behaves as if neutral depends on the population size relative to that effect: in a small population, drift can overwhelm weak selection (\(|s|\lesssim1/N_e\)), while the identical mutation in a much larger population is efficiently selected against or for, since drift there is comparatively weaker. B

Result

k = \mu\ \ (\text{neutral substitution rate equals mutation rate, independent of population size})

Reading. For truly neutral sites, the long-run rate at which new variants become permanently fixed in a population depends only on the underlying mutation rate, not on how large the population is — a striking and testable cancellation, since fixation probability and mutation supply both individually depend on population size but in exactly offsetting ways.

Scope. Applies strictly to neutral sites (Hypotheses); sites under substantial positive or negative selection follow different dynamics not captured by \(k=\mu\) alone, and the nearly-neutral extension (Step 5) blurs the boundary for weakly selected sites in small populations.

Corollaries & converses

molecular-clock's assumption of approximately constant substitution rate across lineages (its own Hypotheses) is directly justified by Step 4, at least for the largely neutral, mostly synonymous sites typically used for clock calibration.
mutation-types' distinction between silent (synonymous) and missense (non-synonymous) substitutions maps closely onto the neutral/non-neutral distinction here: synonymous substitutions, having no effect on protein sequence, are the strongest practical candidates for the theory's neutral class, and are empirically observed to accumulate faster than non-synonymous substitutions in most genes, consistent with the latter being more often selected against.
Converse: if a gene's non-synonymous substitution rate is found to be unusually elevated relative to its synonymous rate (rather than suppressed, as purifying selection on protein sequence would predict), this deviation from the neutral expectation of Step 3 is itself standard evidence for positive selection having acted on that gene.

Fails without

Drop the existence of a substantial neutral mutation class (Hypotheses): if essentially every new mutation had a significant fitness effect, drift alone (Step 2) would not be the main determinant of which mutations get fixed; selection would dominate at nearly every site, and the population-size-independent substitution rate of Step 3 would not generally hold, undermining the molecular clock's justification for treating substitution rate as roughly constant across very different organisms.
Treat the theory as claiming no molecular evolution is ever adaptive: this overstates the claim beyond what Step 3 actually establishes; the neutral theory specifically concerns the majority of molecular substitutions, not all of them, and does not deny that a genuine minority of substitutions are adaptive and selection-driven — conflating "most" with "all" produces predictions the theory does not actually make and that are readily falsified by known cases of adaptive molecular evolution.

Common errors

Interpreting the neutral theory as claiming natural selection is unimportant in evolution generally; it specifically concerns molecular-level substitution rate, not phenotypic evolution or the action of selection at the level of visible traits, where natural-selection remains central.
Assuming \(k=\mu\) (Step 3) applies to every site in the genome uniformly; it applies specifically to sites that are actually neutral, and different genomic regions differ substantially in what fraction of their mutations meet that condition.
Treating "neutral" and "nearly neutral" as identical; the nearly-neutral extension (Step 5) specifically predicts population-size-dependent behaviour that the strict-neutral model does not, and the two make different quantitative predictions for small versus large populations.
Assuming a fast substitution rate always indicates positive selection; under the neutral theory, a fast rate more commonly indicates weak functional constraint (a larger fraction of mutations there being neutral), not adaptive selection specifically favouring change.

Discussion

Motoo Kimura proposed the neutral theory of molecular evolution in 1968, directly motivated by protein sequence data showing substitution rates that appeared far too high, and too uniform across very different genes and lineages, to be plausibly explained by selection driving essentially every substitution — a genuinely controversial claim at the time, since it challenged the then-prevailing assumption that visible, phenotypic natural selection (natural-selection) should extend straightforwardly to explain molecular-level change as well.

Tomoko Ohta's nearly-neutral extension (Step 5), developed from the 1970s onward, resolved several observations the strict neutral theory struggled with, particularly the tendency for substitution rate to correlate somewhat with generation time and population size in ways \(k=\mu\) alone does not predict, by explicitly allowing the neutral/non-neutral boundary itself to depend on effective population size rather than being fixed.

Common misconception: that the neutral theory and natural selection are competing, mutually exclusive explanations for evolution as a whole. They operate largely at different levels: the neutral theory principally concerns the fate of molecular-level variation with little or no fitness effect, while natural-selection concerns traits and mutations that do affect fitness: the two frameworks are complementary components of a complete evolutionary picture, not rival theories about the same phenomena.

Worked examples

\mu = 5\times10^{-9}\ \text{neutral substitutions/site/generation (a typical order-of-magnitude estimate)}

Applying Step 3 directly, the predicted long-run neutral substitution rate \(k\) equals \(\mu\) itself, \(5\times10^{-9}\) per site per generation, regardless of whether the population in question is small or very large. A

k = \mu = 5\times10^{-9}\ \text{per site per generation, independent of }N_e

Reading. Two species with wildly different effective population sizes, but a similar underlying neutral mutation rate for a given gene, are predicted to show a similar long-run neutral substitution rate for that gene — the population-size terms exactly cancel (Step 3).

Scope. This prediction applies specifically to neutral sites; comparing non-synonymous substitution rates between the same two species would not show the same population-size independence if selection is acting differently on protein sequence in each.

Problems

Species A has effective population size \(N_e=10^4\) and species B has \(N_e=10^8\), but both share an identical per-generation neutral mutation rate \(\mu=2\times10^{-8}\) for a given gene. Using Step 3, compare their predicted neutral substitution rates.
Solution
By Step 3, \(k=\mu\) for both species regardless of \(N_e\): \(k_A=k_B=2\times10^{-8}\) substitutions/site/generation. Despite species B having a population 10,000 times larger than species A, their predicted neutral substitution rates are identical, since the population-size dependence exactly cancels between mutation supply and fixation probability.
A researcher observes that the synonymous (silent) substitution rate in a gene is much higher than its non-synonymous (missense) substitution rate. Using the Corollaries, explain what this pattern is generally taken to indicate.
Solution
Synonymous substitutions do not change the encoded protein and are the strongest practical candidates for neutrality (Corollaries), so their rate is expected to closely track the raw mutation rate (Step 3). A markedly lower non-synonymous rate indicates that most amino-acid-changing mutations are being removed by purifying (negative) selection before they can reach fixation, rather than accumulating at the neutral rate — a standard signature of functional constraint on the protein.
A gene shows a non-synonymous substitution rate that is higher than its own synonymous substitution rate in a particular lineage. Using the Converse in Corollaries, what does this unusual pattern suggest?
Solution
Since synonymous substitutions are expected to accumulate at roughly the neutral rate \(\mu\) (Step 3), and purifying selection ordinarily suppresses non-synonymous substitution below that neutral baseline, finding non-synonymous substitution rate exceeding the synonymous rate is the reverse of the expected pattern. This is standard evidence for positive (diversifying) selection actively favouring amino-acid-changing mutations in that gene and lineage (Corollaries' Converse), rather than the gene evolving neutrally.