Mutation as a Source of Genetic Variation

A mutation is any heritable change in the nucleotide sequence of DNA (or, in RNA viruses, RNA). Mutations range from a single base substitution to the duplication of an entire genome. They occur in somatic cells and in germline cells, but only germline mutations — those in the cells that give rise to gametes — are transmitted to offspring and are therefore relevant to evolution. A mutation that is not passed on, however dramatic its effect on the individual carrying it, is an evolutionary dead end. The central conceptual claim of this chapter is that mutation is random with respect to fitness. This phrase is precise and easy to misread, so it is worth unpacking. It does not mean mutations occur at random positions in the genome with equal probability — some regions mutate far more than others due to local chemistry, chromatin structure, and repair efficiency. It does not mean all types of mutation are equally likely — transitions (purine-to-purine or pyrimidine-to-pyrimidine substitutions) occur more often than transversions. What it means is this: the probability that a particular mutation occurs is not influenced by whether that mutation would be useful, harmful, or neutral to the organism in its current environment. A mutation conferring antibiotic resistance is no more likely to appear in the presence of the antibiotic than in its absence. The environment can select among mutations that already exist; it cannot, in ordinary Darwinian biology, cause an adaptive mutation to preferentially arise because it is needed. Mutation is therefore correctly described as the ultimate source of genetic variation, in contrast to recombination and gene flow, which are proximate or redistributive sources — they shuffle existing alleles into new combinations but create no new allelic information themselves. Every gene, every allele, every base pair that natural selection or genetic drift has ever acted upon first appeared as a mutation in some ancestor. This does not mean mutation is the dominant force shaping allele frequencies generation-to-generation — that role usually belongs to Natural Selection and the Concept of Fitness and drift acting on standing variation — but without mutation, the well of variation would eventually run dry.

1. Point mutations arise from replication errors and DNA damage. DNA polymerases are remarkably accurate, but not perfect. During replication, a polymerase occasionally inserts a mismatched base — for example, pairing a tautomeric form of cytosine with adenine instead of guanine. Proofreading (the polymerase's 3′→5′ exonuclease activity) corrects most such errors immediately, and mismatch repair systems catch most of the rest after replication. What survives — roughly one error per 10⁹–10¹⁰ base pairs replicated in humans, after all repair — becomes a fixed substitution. Independent of replication, DNA is also damaged continuously: spontaneous deamination of cytosine to uracil, depurination (loss of a purine base leaving an abasic site), oxidative damage from reactive oxygen species, and UV-induced pyrimidine dimers. Left unrepaired or misrepaired, this damage becomes a permanent mutation when the cell next replicates its DNA.

2. Insertions, deletions, and slippage. Small insertions and deletions (indels) often arise from replication slippage, especially in repetitive sequences such as microsatellites (e.g., CA repeats). The polymerase and template strand transiently dissociate and mis-anneal, causing the newly synthesized strand to loop out or the template to loop out, adding or removing repeat units. Indels within a coding sequence that are not multiples of three cause a frameshift, scrambling every downstream codon — usually catastrophic for protein function.

3. Chromosomal-scale mutation. Larger-scale changes include inversions (a segment flips orientation), translocations (a segment moves to a different chromosome), duplications (a segment is copied, often via unequal crossing-over between misaligned repetitive sequences during meiosis), and whole genome duplication (polyploidy), common in plant speciation. Gene duplication is especially important evolutionarily: it creates a redundant copy that is initially free from selective constraint, one paralog can retain the ancestral function while the other accumulates mutations and potentially acquires a new function (neofunctionalization) — the mechanism behind gene families such as globins and opsins.

4. Transposable elements. Roughly half the human genome derives from transposable elements — sequences capable of copying or moving themselves within the genome. Their insertion can disrupt genes, alter regulatory regions, or occasionally be co-opted for new regulatory functions, making them a significant and ongoing mutational force.

5. The fitness effect of a mutation is independent of the mutational process that created it. This is the crux of "randomness with respect to fitness." The chemical or replication error that produces a given base change has no information about, and is not influenced by, whether that base change will help or hurt the organism in its current environment. A given mutation might be:

• Deleterious — most mutations that affect protein function fall here, because organisms are already reasonably well adapted and most novel changes disrupt some existing function.
• Neutral — many mutations, especially synonymous substitutions or changes in nonfunctional DNA, have no measurable fitness effect.
• Beneficial — a small minority increase fitness in the current environment; these are the raw material selection can act on.

Whether a specific mutation lands in one category or another is a matter of the current environment and genetic background, not of when or why the mutation occurred.

The problem. By the early 1940s, biologists knew that bacterial populations exposed to a bacteriophage (a virus that kills bacteria) sometimes contained a few surviving, resistant colonies. Two explanations were on the table. One, favored by some microbiologists at the time and resonant with Lamarckian ideas, held that exposure to the phage itself somehow induced resistance in the bacteria that encountered it — an adaptive, directed response. The other, consistent with Darwinian random variation, held that resistant mutants arose spontaneously and randomly at some point before the phage was ever introduced, and the phage merely revealed and selected them.

The Luria–Delbrück fluctuation test (1943). Salvador Luria and Max Delbrück devised an elegant statistical test to distinguish these hypotheses. They grew many small, independent cultures of E. coli, each starting from a small inoculum, allowed them to grow for the same number of generations, and then plated all the cells from each culture onto phage-laced agar, counting resistant colonies per culture. The logic of the test hinges on when mutations occur relative to plating:

• If resistance were induced by the phage at the moment of exposure, every culture should experience the same probability of induction and therefore show roughly the same, small number of resistant colonies — a Poisson distribution where the variance approximately equals the mean.
• If resistance arose spontaneously and randomly at various times during growth, before phage exposure, then a mutation occurring early in one culture's history would be copied into many descendant cells by the time of plating, while a mutation occurring late (or not at all) in another culture would yield very few or zero resistant cells. This produces occasional "jackpot" cultures with huge numbers of resistant colonies, and a variance far exceeding the mean.

Result. Luria and Delbrück found exactly the jackpot pattern: enormous culture-to-culture variance, far greater than the mean number of resistant colonies, with some cultures showing zero resistant cells and others showing hundreds. This variance was inconsistent with induced resistance and consistent only with mutations arising randomly at various points during growth, independent of the phage. They developed a mathematical distribution (the Luria–Delbrück distribution) to formalize this and used it to estimate mutation rates in bacteria for the first time — on the order of 10⁻⁸ per cell per generation for phage resistance in their system. The paper, "Mutations of Bacteria from Virus Sensitivity to Virus Resistance," won Luria and Delbrück (with Alfred Hershey) the 1969 Nobel Prize in Physiology or Medicine.

Replica plating: the Lederbergs' direct confirmation (1952). Joshua and Esther Lederberg devised a technique that made the same conclusion visually undeniable. They pressed a velvet pad onto a master plate of bacterial colonies grown without any phage or antibiotic present, then stamped that same pad onto several replica plates containing the selective agent, transferring cells in the identical spatial pattern to each replica. Resistant colonies appeared in the same physical position on multiple independent replica plates — meaning the resistant cells existed in the master plate, at that location, before they had ever been exposed to the phage or antibiotic. This directly demonstrated that resistance mutations pre-existed selection rather than being induced by it, closing the case opened by Luria and Delbrück nine years earlier.

What this proved, precisely. These experiments established that mutation is random with respect to the future selective environment — not that mutation is uniform across the genome, nor that all mutations are equally likely, nor that mutation rates cannot be influenced by cellular stress responses (a subtlety explored under adaptive mutation debates in bacteria, which remain narrow and contested exceptions rather than a refutation of the general principle). The core finding — that useful mutations do not arise preferentially because they are useful — has been repeatedly confirmed since and stands as one of the best-supported empirical results in evolutionary biology.

Example 1 — Estimating mutation rate from a fluctuation test (quantitative reasoning). Suppose a researcher repeats a simplified fluctuation-test-style experiment: 20 independent cultures are each grown from a single sensitive cell to a final population of N = 2 × 10⁸ cells, then plated on phage-containing medium. Eighteen of the 20 cultures show 0 resistant colonies; two cultures show 1,400 and 3,100 resistant colonies respectively (jackpots).

• Mean resistant colonies per culture ≈ (1400 + 3100) / 20 = 225.
• Variance is enormous relative to this mean because two cultures dominate the total — a hallmark of the Luria–Delbrück jackpot distribution, not a Poisson process.
• Using the method of the median (a standard simplified estimator for these distributions), the mutation rate per cell per generation can be approximated as roughly

μ ≈ (median number of resistant cells) / (N × 1.24), giving a mutation rate on the order of 10⁻⁸ per cell per generation — consistent with Luria and Delbrück's original estimate for phage resistance in E. coli. The key qualitative takeaway, independent of the exact formula: a small number of jackpot cultures driving a huge variance-to-mean ratio is only possible if the resistant mutants arose stochastically at different, earlier times during growth — precisely the pattern incompatible with induction by the phage at the moment of plating.

Example 2 — A clinical scenario: antibiotic resistance in a hospital ward. A hospital treats a ward of patients colonized by Staphylococcus aureus with a new antibiotic. Consider a bacterial population of 10¹⁰ cells in a chronic wound, with a per-cell, per-generation mutation rate to resistance of μ = 10⁻⁹.

• Expected number of new resistant mutants arising per generation, before any antibiotic is given: 10¹⁰ × 10⁻⁹ = 10 cells.
• These 10 cells exist whether or not the antibiotic is ever applied — this is the direct clinical analogue of the Luria–Delbrück logic.
• When the antibiotic is applied, it does not create these mutants; it kills the roughly 10¹⁰ − 10 sensitive cells and leaves the ~10 pre-existing resistant cells to proliferate unopposed.
• This explains why using sub-lethal doses, or stopping treatment early, is so dangerous: it removes competitors from the resistant minority without eliminating it, accelerating its rise to dominance in the population — a process of selection, not induction, acting on standing mutational variation.

This reasoning underlies clinical guidance to complete full antibiotic courses and to use combination therapy (multiple drugs simultaneously), which requires a much rarer double mutation (probability ≈ μ² if the two resistance mutations are independent) to survive both drugs at once.

Salvador Luria was born in Turin, Italy, in 1912 and trained as a physician before turning to phage research, fleeing to the United States in 1940 after Mussolini's racial laws barred Jewish scientists from academic positions. He became a founding figure of the "phage group," a loose network of researchers (including Max Delbrück and Alfred Hershey) who used bacteriophages as simple, tractable systems to ask fundamental genetic questions before DNA's role as the hereditary material was even settled — that story is told through Genetic Variation as the Raw Material of Evolution and the Hershey–Chase experiment. Luria's fluctuation test, developed jointly with the physicist-turned-biologist Delbrück, brought quantitative, statistical rigor to a question that had previously been debated on largely qualitative and philosophical grounds.

Antibiotic and antiviral resistance management. The Luria–Delbrück logic directly informs modern infectious disease practice: because resistance mutations pre-exist drug exposure, clinicians use combination therapy (e.g., in HIV and tuberculosis treatment) so that a pathogen must acquire multiple independent resistance mutations simultaneously — a far rarer event — to survive treatment.

Cancer biology. Tumors evolve by the same logic. Chemotherapy-resistant cancer cell clones typically pre-exist treatment as rare mutants within a heterogeneous tumor; treatment selects for them rather than inducing resistance, explaining why relapse tumors are often already resistant at the moment therapy begins.

Agricultural pest and herbicide resistance. Pesticide and herbicide resistance in insects and weeds follows the identical dynamic — resistance alleles exist at low frequency before a chemical is ever sprayed, and repeated use simply selects for their increase, motivating rotation of chemical classes to slow resistance evolution.

Genome sequencing and human disease. Direct measurement of human germline mutation rates (via whole-genome sequencing of parent-offspring trios) gives ~1.0–1.5 × 10⁻⁸ mutations per base pair per generation, translating to roughly 60–70 new mutations per genome per generation — the same molecular process studied by Luria and Delbrück, now measured directly and used to calibrate molecular clock estimates of divergence times and to identify de novo mutations causing rare pediatric diseases.

Conservation genetics. Small, bottlenecked populations have reduced mutational and standing genetic variation, limiting their capacity to generate novel adaptive alleles in response to environmental change (e.g., emerging disease or climate shifts), a key concern in managing endangered species with critically low effective population sizes.

Ask the tutor: How does the mutation rate itself evolve, and can natural selection favor organisms with higher or lower mutation rates?