At the level of whole organisms, natural selection has traditionally been regarded as the principal force shaping observable traits, and numerous well-adapted structures the hooves of horses, the grasping hands of humans, the asymmetric body of flatfish that lie on one side, or the camouflage patterns seen across many animal groups — are readily explained as products of selection acting on simpler or more generalized ancestral forms. This type of selection, which favours new, better-adapted variants, is termed positive selection. When molecular sequence data first became available, it was widely assumed that positive selection would prove equally important in shaping variation at the level of DNA and protein sequences.
This assumption gave rise to two contrasting schools of thought regarding molecular evolution. Adaptationists maintain that positive selection is the principal driving force shaping molecular sequences. Neutralists, in contrast, maintain that most molecular-level change results from neutral processes chiefly mutation combined with random genetic drift with positive selection playing only a comparatively minor role, and mutation itself being regarded as the dominant driving force. It is useful, before proceeding, to distinguish the broad categories of selection referred to throughout this topic: positive (directional) selection favours new advantageous variants and drives their fixation; stabilizing (purifying) selection removes deleterious variants and preserves the existing functional sequence; and balancing selection actively maintains more than one allele within a population.
1. Techniques for Detecting Molecular Variation
1.1 Allozyme Electrophoresis
One of the earliest experimental approaches to measuring molecular variation, developed during the 1960s and 1970s, was the study of allozymes — different allelic forms of the same protein that can be distinguished by gel electrophoresis on the basis of differences in electrical charge. Surveys using this technique showed that a surprisingly large number of protein-coding loci were polymorphic, carrying more than one allele within a population. This finding sat awkwardly with the adaptationist expectation of the time, since natural selection was thought to eliminate less-fit alleles from a polymorphic locus fairly quickly, ultimately leaving a single fixed allele behind. It is also worth noting that allozyme electrophoresis captures only part of the true genetic variation present at a locus: substitutions that are synonymous leave the protein unchanged and go undetected, and even non-synonymous substitutions frequently escape detection if the resulting amino acid change does not alter the overall charge of the protein.
1.2 Restriction Fragment Length Polymorphisms (RFLPs)
A second widely used approach, emerging during the 1970s and 1980s, made use of restriction fragment length polymorphisms, commonly abbreviated as RFLPs. This method exploits restriction enzymes, a class of nucleases that recognize specific short DNA sequences and cleave DNA precisely at those recognition sites. Digesting a length of DNA with a restriction enzyme therefore produces a characteristic set of fragments, which can be resolved by size using gel electrophoresis, a technique that has been applied extensively to mitochondrial DNA. If individuals differ in the pattern of fragments produced, this indicates that a base-pair difference exists somewhere within a restriction site, since even a single-nucleotide change can be sufficient to create or destroy a cutting site.
Both allozyme electrophoresis and RFLP analysis share an important limitation: each is sensitive to only a restricted subset of the sequence variation that actually exists within a genome.
1.3 Large-Scale DNA Sequencing
More recently, large-scale DNA sequencing has become feasible, generating extensive datasets of complete gene sequences sampled from many individuals within a population. Unlike allozyme or RFLP surveys, full sequencing captures the entire pattern of single nucleotide polymorphism present in the genes under study, without the detection biases of the earlier techniques. With this complete picture available, the central question that arises, at both the DNA and protein level, is why particular sites remain polymorphic, and what this reveals about the underlying mechanism of molecular evolution.
2. Explanations for the Maintenance of Polymorphism
2.1 Adaptationist (Selective) Explanations
Positive selection, in its most straightforward action, tends to eliminate deleterious alleles and thereby reduces, rather than maintains, the number of polymorphic loci in a genome. To account for the observed abundance of polymorphism, adaptationists have therefore had to invoke more elaborate selective scenarios capable of actively preserving variation. Two such scenarios are commonly proposed:
- Selection favouring different alleles in different environments: an allele may be advantageous under one set of ecological conditions but disadvantageous under another. A species occupying two regions with contrasting climates, for example, might have distinct alleles that are each optimal in their respective region, so that selection maintains both alleles simultaneously within the overall population.
- Heterozygote advantage (overdominance): a heterozygous individual may possess higher fitness than either corresponding homozygote. Selection under this scenario actively drives allele frequencies towards an intermediate equilibrium at which both alleles are maintained, typically approaching equal frequency.
2.2 Neutralist Explanation
While such selective scenarios undoubtedly apply to particular genes, it is doubtful that they can account for the sheer scale of polymorphism observed across genomes as a whole. The neutralist account is considerably simpler: new alleles are continually generated by mutation, while existing alleles are continually lost through random genetic drift. Because these two opposing processes are always in operation, at any given moment some fraction of loci will inevitably be found in a polymorphic state, without requiring any special selective explanation. Variability within populations, on this view, is simply a reflection of an appreciable underlying mutation rate.
3. Heterozygosity as a Measure of Genetic Variability
A convenient quantitative measure of genetic variability at a locus is heterozygosity. In a diploid population, heterozygosity is defined as the proportion of individuals that are heterozygous at that locus, that is, individuals carrying two different alleles, one inherited from each parent. Provided mating occurs at random with respect to the alleles present at the locus, the two alleles carried by any given individual can be treated as two independently, randomly drawn alleles from the parental generation’s gene pool. This allows heterozygosity to be defined equivalently as the probability that two alleles, chosen at random from the population, differ from each other.
At a neutral locus, heterozygosity reflects a dynamic balance between the continual input of new variation by mutation and its continual removal by random drift. The average heterozygosity maintained at this equilibrium can be shown to depend on the compound parameter Nu, the product of population size N and mutation rate u. When Nu is greater than 1, mutation supplies new variation faster than drift can remove it, so heterozygosity is high and most loci remain polymorphic. Conversely, when Nu is much less than 1, drift removes variation faster than mutation can replace it, heterozygosity is correspondingly low, and few loci remain polymorphic.
4. The Neutral Theory as a Null Hypothesis
One of the principal strengths of the neutral theory is its mathematical tractability: a large number of quantities of interest can be derived exactly under its assumptions. This makes the neutral model extremely useful as a null hypothesis against which observed molecular data can be compared. If real data are found to depart significantly from the predictions of the neutral model, this is taken as evidence that selection has been acting.
An important complication, however, is that predictions of the neutral theory are subject to very large stochastic variation around their expected values. The mean time for two gene lineages to coalesce to a common ancestor, for example, is N generations, yet the actual distribution of coalescence times is exponential in shape, broad and skewed rather than tightly clustered around this mean. A comparable pattern is seen for heterozygosity: although the mean heterozygosity expected under neutrality is straightforward to calculate, the distribution of heterozygosity values observed across different loci is itself complex and broadly spread, rather than sharply concentrated near the mean. Because of this substantial underlying variability, statistical tests designed to detect departures from neutrality, such as those developed by Watterson and by Tajima, which compare different estimators of genetic diversity derived from the same dataset, tend to be mathematically involved and generally require large sample sizes before they can reliably distinguish genuine selective effects from the scatter expected under neutrality alone.
5. Sequence Divergence Between Species
5.1 The Adaptationist View
The adaptationist–neutralist debate extends beyond within-population variation to the question of why protein sequences differ between species. According to the adaptationist position, most amino acid differences observed between species arise because positively selected changes have occurred independently in one lineage or another. A new sequence variant, on this view, may function more effectively within the particular cellular and organismal context of one species than of another, a context shaped by the other molecules with which the protein interacts, as well as by lifestyle and environmental factors specific to that species. Selection would therefore continually fine-tune each species’ proteins towards optimal function within its own context, making selection the principal engine of sequence divergence between species.
5.2 Evidence for Stabilizing Selection
In practice, however, examination of actual sequence alignments (such as comparisons of the BRCA1 gene across species) tells a rather different story. The dominant signature of selection detectable in most alignments is not positive selection favouring new advantageous variants, but rather stabilizing (purifying) selection, which acts to eliminate deleterious mutations and preserve the existing, functional version of a gene. Several lines of evidence point to the pervasiveness of stabilizing selection:
- Synonymous substitutions consistently outnumber non-synonymous substitutions, indicating that changes altering the protein sequence are selectively disfavoured relative to those that do not.
- Amino acid substitutions that do occur are predominantly conservative, replacing an amino acid with one of similar chemical properties, rather than radical changes between very dissimilar amino acids.
- Frameshift mutations, which would disrupt the entire downstream reading frame, are strongly selected against and rarely observed to persist.
- In genes encoding structural RNAs, compensatory substitutions frequently occur in pairs, such that the underlying sequence changes while the secondary structure of the RNA molecule, and hence its function, is preserved.
A further striking observation to emerge from genome sequencing projects is the extent to which vertebrate proteins possess clearly recognizable homologs even in organisms as evolutionarily distant as yeast and bacteria. What stands out most, in other words, is the underlying similarity of sequences across widely divergent organisms, rather than the specific adaptations that distinguish them. Taken together, this evidence suggests that sequence divergence occurs in spite of, rather than because of, ongoing stabilizing selection that continually acts to resist change. Neutralists interpret this pattern as showing that mutation, rather than selection, is the dominant force driving divergence between sequences, with most of the differences observed between species representing the chance fixation of nearly neutral mutations in one lineage or another over evolutionary time.
6. Historical Development and the Precise Neutralist Position
The case for the importance of neutral evolution was first put forward influentially by King and Jukes in 1969, in a paper titled “Non-Darwinian Evolution”, a title that itself signalled how provocative the idea was considered at the time. The neutral theory was subsequently developed into a comprehensive framework and championed prominently by Motoo Kimura, whose 1983 monograph, “The Neutral Theory of Molecular Evolution”, remains the definitive statement of the theory. This position attracted, and continues to attract, considerable criticism from adaptationists, who are often uncomfortable with the prominent role assigned to chance in neutral evolution.
It is important, however, to be precise about what the neutral hypothesis actually asserts, since it is frequently mischaracterized. Neutralists do not claim that natural selection is absent or unimportant in an absolute sense. They fully accept that stabilizing selection continually removes deleterious mutations, and they accept that advantageous mutations do occasionally arise and spread. Their specific claim is a comparative one: that neutral, or nearly neutral, mutations arise far more frequently than advantageous mutations do, and that, as a consequence, the majority of mutations that ultimately achieve fixation in a population are neutral rather than advantageous.
7. Selective Sweeps, Genetic Hitchhiking, and the Search for Positive Selection
7.1 Selective Sweeps and Hitchhiking
One reason advantageous mutations are so difficult to detect directly is that they spread through a population rapidly once favoured by selection, making it hard to observe the process while it is actually under way. By the time the mutation has reached fixation, all of the sequence variation that previously existed at that location has been erased, a phenomenon known as a selective sweep. A further consequence of a selective sweep is genetic hitchhiking: as the advantageous mutation sweeps to fixation, other mutations located at closely linked sites nearby may be carried to fixation alongside it, even if those additional mutations are themselves neutral or mildly deleterious. This occurs whenever such linked sites lie close enough to the selected mutation that no recombination separates them during the relatively short timescale of the sweep, so that they remain physically associated with the advantageous variant and are fixed together with it.
7.2 Identifying Genes Under Positive Selection
In more recent years, the intensity of debate between adaptationist and neutralist positions has diminished considerably, even though the underlying question has never been fully settled. The current consensus tends to view neutral and selected mutations as occupying different points along a continuous spectrum, with the neutral theory retaining a firmly established role as the standard null model against which molecular data are evaluated. Even granting that the neutral hypothesis holds broadly on a statistical, genome-wide basis, it remains valuable to search for specific genes that show clear evidence of directional selection in particular lineages. One established approach is to identify genes with an unusually high ratio of non-synonymous to synonymous substitutions (often denoted dN/dS), since an excess of amino-acid-changing substitutions relative to silent ones is a hallmark of positive selection rather than neutral drift or purifying selection. Genes meeting this criterion are relatively uncommon, though the BRCA1 gene provides a notable example, with evidence pointing to adaptive evolution having occurred specifically within the human and chimpanzee lineage following their divergence from gorillas (Huntley et al., 2000).
8. Codon Bias: Weak Selection Among Synonymous Codons
A final phenomenon worth considering in this context is codon bias. Synonymous substitutions, changes at the DNA level that do not alter the encoded amino acid, might reasonably be expected to represent the clearest possible case of strictly neutral evolution, since they have no effect whatsoever on the resulting protein. Contrary to this expectation, however, there is substantial evidence that weak selection can act even among synonymous codons, with the result that the different codons encoding the same amino acid are not used with equal frequency within gene sequences. This unequal usage is termed codon bias, and it reflects an apparent preference for certain codons over their synonymous alternatives.
Codon bias arises from at least two distinct causes:
- A purely mutational cause: if the mutation rates between the four nucleotide bases are unequal, the expected base composition at the third, or wobble, position of a codon will differ from one base to another even in the complete absence of selection.
- A selective cause related to translational efficiency: codon usage is often found to differ systematically between genes within the same genome, with highly expressed genes typically showing stronger codon bias, that is, a greater tendency to use certain preferred codons, than genes expressed at lower levels. This pattern is attributed to selection acting to improve the efficiency of translation. Because different transfer RNA (tRNA) species are present within the cell at different concentrations, and because the preferentially used codons tend to correspond to the anticodons of the more abundant tRNAs, using a preferred codon reduces the time a ribosome must spend waiting for the matching tRNA to become available during translation.
Because it results from weak selective effects whose strength can vary considerably between different genes and genomic contexts, codon bias remains a comparatively subtle phenomenon to characterize and interpret.
Conclusion
The neutralist–adaptationist debate over the relative importance of mutation-driven drift versus natural selection in molecular evolution has, over time, moved away from an either–or framing towards a more nuanced, spectrum-based understanding. Evidence from allozyme surveys, restriction fragment analysis, and, most comprehensively, large-scale DNA sequencing consistently reveals extensive polymorphism within populations, which the neutral theory explains parsimoniously through the ongoing balance of mutation and drift, without requiring elaborate selective scenarios in every case. At the same time, clear signatures of selection are detectable throughout the genome: stabilizing selection dominates the pattern of divergence between species, occasional episodes of positive selection can be identified through elevated non-synonymous substitution rates and through the hitchhiking signatures left by selective sweeps, and even ostensibly silent synonymous substitutions show evidence of weak selection in the form of codon bias. The position established by Kimura, and defended against sustained adaptationist criticism, holds that neutral and nearly neutral mutations account for the majority of fixation events, while selection, both purifying and, more rarely, positive, remains an essential and detectable component of molecular evolution.
Summary of Key Concepts
- Adaptationists hold that positive selection is the principal driver of molecular evolution; neutralists hold that mutation and random drift are the principal drivers, with selection playing a comparatively minor role.
- Allozyme electrophoresis and RFLP analysis were early techniques revealing molecular polymorphism, but each detects only part of the true sequence variation present; large-scale DNA sequencing later provided a complete picture.
- Adaptationist explanations for maintained polymorphism: spatially/environmentally varying selection and heterozygote advantage (overdominance).
- Neutralist explanation for polymorphism: continuous balance between mutation (creates new alleles) and random drift (removes alleles).
- Heterozygosity: fraction of heterozygous individuals in a diploid population, equivalently the probability two randomly chosen alleles differ; its equilibrium value depends on the compound parameter Nu.
- The neutral theory serves as a null hypothesis for detecting selection but shows large stochastic variance in its predictions (e.g., exponential coalescence-time distribution), making neutrality tests (Watterson, Tajima’s D) statistically demanding.
- Evidence for stabilizing selection dominating interspecies divergence: excess synonymous over non-synonymous substitutions, conservative amino acid changes, frameshift avoidance, compensatory RNA substitutions, and deep cross-species homology.
- King and Jukes (1969) and Kimura (1983) established the neutral theory; neutralists accept that selection occurs but argue neutral fixations are far more frequent.
- Selective sweep: rapid fixation of an advantageous mutation that erases linked variability; genetic hitchhiking:linked neutral or deleterious mutations fixed alongside it.
- The dN/dS ratio is used to detect positive selection (e.g., BRCA1 in the human–chimpanzee lineage).
- Codon bias: unequal use of synonymous codons, arising from mutational bias and/or selection for translational efficiency via matching to abundant tRNAs.
