Mutation creates variants; recombination and transfer alter how variants are distributed.
By the end: Explain why variation need not appear because an organism needs it, while allowing mutation rates and effects to depend on conditions.
Mutation changes sequence
Copying errors, DNA damage, and other molecular processes can change DNA sequence. Changes include substitutions, insertions, deletions, and duplications. In multicellular organisms, a mutation must enter the reproductive lineage to pass through ordinary reproduction; most mutations confined to body cells do not reach offspring.
Mutation is not uniform over sites or conditions. DNA repair, sequence context, and stress can change rates and patterns. The important contrast with foresight is that mutations do not generally arise because their particular future effects would solve an organism’s problem. Their consequences depend on the environment and genetic background.
Follow a sequence change into descendants
DNA stores sequence information in the order of its chemical bases. Before a cell divides, it copies that sequence. A copying error that remains after repair can leave one descendant with a different base at a position. Damage and imperfect repair can also change sequence without a copying error at that moment. Calling the result a mutation describes the change, not its usefulness. A substitution replaces a base, an insertion adds sequence, and a deletion removes sequence. These descriptions do not specify whether reproduction will increase or decrease.
A sequence change can alter a protein because the sequence helps specify the order of amino acids. It can also change when or where a gene produces its product. Other changes have little detectable effect in the conditions examined. Even a change outside a protein-coding region is not automatically irrelevant, because some such regions regulate gene activity. Conversely, a changed protein is not automatically beneficial or harmful. Connecting sequence to reproductive consequences requires several steps, including effects on cells, traits, and performance in a particular environment.
Inheritance imposes another condition. In a simple unicellular lineage, a changed sequence can pass to daughter cells during subsequent divisions. In a multicellular organism, a change confined to a patch of body cells ordinarily does not enter offspring through sexual reproduction. A change in a cell lineage that produces gametes can do so. The distinction concerns the route into descendants, not the size of the visible effect. A conspicuous body-cell change can leave no inherited population signal, while a small sequence change in a reproductive lineage can persist.
Which changed cells can reach offspring?
Schematic multicellular lineage, not a complete developmental map or a timeline to scale. A change in a body-cell branch need not reach gametes. A changed reproductive-lineage copy can reach an offspring; transmission is possible, not guaranteed.
Count the first appearance before asking about spread. Suppose a constructed diploid population has 100 individuals and 200 copies of an autosomal locus. Initially every copy is a. A mutation changes one copy to A in a reproductive lineage that contributes to the next census, leaving one Aa individual and 99 aa individuals there. A frequency is then 1/200, or 0.5%. The carrier fraction is 1/100, or 1%. If A is recessive for a defined visible trait, the fraction showing that trait can still be zero.
Those small numbers describe different observations. Sequencing copies could reveal A, counting carriers could reveal its distribution among individuals, and looking only for a recessive phenotype could miss it. None of the three measurements establishes an advantage. If the Aa individual contributes no descendants, A can disappear. If descendants inherit A, its later frequency depends on reproduction and sampling. The event that supplied a new sequence and the events that determine its persistence therefore need separate explanations, even when they occur close together in time.
Mutation rates also need a denominator. A probability per copied base per division is different from a probability per gene, genome, or generation. A larger genome offers more possible positions, and more divisions offer more copying opportunities. Without those units, saying one lineage has more mutations cannot distinguish a higher rate from more opportunities. Treat mutation counts as evidence about molecular events only after checking what was observed, what could have been detected, and how much copying or elapsed reproductive history the count represents.
One new copy has several denominators
Measurement
Constructed census
Fraction
A copies
1 of 200 locus copies
0.5%
A carriers
1 Aa of 100 individuals
1%
Recessive A phenotype
0 AA of 100 individuals
0%
Constructed diploid census after one inherited copy appears. Complete recessivity is assumed only for the named phenotype. No fitness effect follows from these counts.
Distinguish origin from spread
A resistance variant can exist before exposure to a selecting agent. Exposure can then change its reproductive advantage and reveal descendants that were rare before. Failing to detect a variant in an earlier sample is not proof that it was absent from the whole population.
The classic Luria and Delbruck experiment concerned bacterial resistance to viruses, not antibiotic treatment. Independently grown cultures produced highly variable resistant counts, including large groups consistent with an early mutation leaving many descendants. This supported mutations arising before selection rather than a simple model in which exposure induces resistance independently at a constant rate in each cell.
Use timing to distinguish competing explanations
A resistant group discovered after exposure has two relevant times in its history. One is when a heritable difference first arose or arrived. The other is when the environment favored its descendants relative to alternatives. Discovery usually occurs later than origin. Imagine a resistant cell among a million susceptible cells before an environmental change. If the change suppresses susceptible reproduction, resistant descendants can become conspicuous without the change having generated their sequence. An increase in visibility is not a direct observation of the molecular event that began the lineage.
Failure to detect a rare type is expected in many small samples. For a constructed haploid population with resistant frequency 0.001, an independent random sample of 100 cells misses resistance with probability 0.999 multiplied by itself 100 times. That is about 0.905, or 90.5%. This calculation assumes random sampling from a large well-mixed population and perfect classification. Clustering or imperfect detection would require a different sampling model, not a stronger conclusion from the same negative result.
A negative sample can miss an existing variant
Random sample size
Probability all sampled cells lack resistance
1 cell
0.999 = 99.9%
100 cells
0.999¹⁰⁰ ≈ 90.48%
1,000 cells
0.999¹⁰⁰⁰ ≈ 36.77%
Constructed large, well-mixed haploid population at resistance frequency 0.001. Independent draws and perfect classification are assumed. These are probabilities of a negative sample given presence, not probabilities of absence given a negative sample.
The secondary historical account of Luria and Delbruck describes a different source of information, variation among independently grown populations. If a mutation occurs early in one population's history, many endpoint cells can descend from that event. A later event leaves fewer descendants before the endpoint. Populations can therefore differ greatly in resistant counts despite similar growth conditions and no advantage during the earlier growth phase. The historical experiment concerned bacterial resistance to viruses.
To see the arithmetic, extend a binary lineage through five divisions, ending with 32 cells. If one daughter after the first division has a mutation, four further doublings produce 16 mutant descendants. A mutation in one daughter after the fifth division leaves one mutant endpoint cell. Both histories contain one mutation event. Every cell divides at the same rate, so neither history includes selection. The sixteenfold difference comes from inherited timing alone. A large clone is therefore not sufficient evidence that the mutant had a reproductive advantage while that clone formed.
Contrast this with a narrowly specified exposure-triggered model. Suppose each endpoint cell independently acquires resistance only on exposure, with the same small probability, and there is no earlier inherited resistant lineage. That model lacks the early shared event that gives many cells the same history. It predicts different between-population count patterns from a model with inherited mutations during prior growth. The comparison tests stated probability models. Large fluctuations do not prove that every possible effect of environment on mutation is absent, or that mutation rates are constant at every site.
The timing hypotheses predict different shared histories
Specified model
Why counts can vary
What the comparison tests
Inherited mutation before exposure
An early event leaves many descendants
Shared lineage timing
Independent change only on exposure
Each endpoint cell has the same small chance
Independent endpoint events
Environmental change in mutation rate
More or fewer sequence changes may arise
Rate change, not proof of a directed solution
Hypothesis comparison, not experimental data or a complete list of mutation models. Historical interpretation relies on the secondary account. Luria and Delbruck studied virus resistance; the original full paper was not reviewed.
Historical inference must retain those qualifications. Stress may change damage, repair, or mutation rates without producing precisely the sequence that would be useful later. To infer a directed solution, one would need evidence beyond a later frequency increase. The teaching lineage isolates timing; it does not reproduce experimental counts or estimate a real mutation rate. The linked original paper has verified bibliographic metadata, but its full text was not reviewed for this course. The historical explanation here rests on the selected secondary account and the explicitly constructed inheritance calculation.
Work through the mechanism
Time runs downward through three divisions. Each marker is one cell; filled markers carry the new variant, open markers do not. Every cell divides into two equally; there is no selection or reversion. The drawings are invented lineage histories, not a reproduction of experimental results.
Process
What changes
Mutation
DNA sequence
Recombination
Combinations of existing variants
Horizontal transfer
Which lineage carries DNA
Selection
Relative reproductive contribution
The lineage drawing is invented; it illustrates timing, not measured experimental counts.
Read the lineage without adding selection
The existing lineage comparison ends after three divisions, with eight cells on each side. Read each fork as one cell producing two descendants, not as one allele becoming two different alleles. The filled markers track inheritance of a particular change. In the early history, one of the two cells after division one carries the change. Its two daughters and their four daughters inherit it. The final share is four of eight. In the late history, only one final daughter changes, giving one of eight.
Use the drawing to make a prediction before counting the filled endpoints. Move the imagined mutation one division later than the early event, into one daughter after division two. That daughter has one remaining division and leaves two mutant cells out of eight. The sequence of possible clone sizes is four, two, then one as the event moves later. This paper extension is not a widget control. It assumes faithful inheritance, no reversion, no cell loss, and exactly synchronized binary division.
Moving one event changes its descendant count
Constructed three-division binary histories, schematic branch lengths. Each endpoint circle is one cell. Filled A markers inherit the single event; a markers lack it. Equal division, faithful inheritance, no death, and no reversion are assumed.
Now change a biological assumption rather than the drawing's timing. Suppose the early mutant daughter fails to divide during one interval when every other cell divides. Its descendants no longer follow the equal-growth tree. Alternatively, if one mutant descendant dies, the endpoint proportion depends on whether the total count also falls. The original picture is useful because it holds these complications absent. It separates the effect of an early inherited event from effects that would require different reproductive or survival rules.
Keep mutation events separate from mutant cells when comparing histories. Four endpoint cells carrying the same inherited change can represent one event followed by descent. Four separate late mutations can produce four carriers too. Sequence information at other positions, lineage observations, and independent population comparisons may help, but each supplies different evidence. Shared surrounding sequence can support common descent of copies, while recombination or transfer can complicate that inference. Do not count every carrier as an independent confirmation of a new origin.
A mutation event also need not explain an entire phenotype. Two sequence changes can have different effects in different genetic backgrounds, and several molecular routes can sometimes produce a similar measured trait. If two lineages both appear resistant, matching phenotype does not establish the same allele or a shared recent mutation. Conversely, descendants can inherit the same variant yet differ in measured performance because their environments differ. A useful account records the sequence observation, the phenotype observation, and the connection being proposed, instead of treating these as interchangeable labels for one event.
For a final guided comparison, imagine sampling one endpoint cell from each eight-cell history. The chance of sampling a carrier is one half in the early history and one eighth in the late history. Those are sampling probabilities conditional on the pictured histories, not probabilities that a mutation occurred early or late. Inferring event timing from an observed carrier requires information about how often the possible histories arise. This distinction prevents a common reversal of reasoning, where a large chance of detection given an early event becomes an unsupported claim that any detected variant arose early.
Equal carrier counts can have different origins
Constructed endpoint history
Events
Carriers
Chance one draw is a carrier
One early event, inherited
1
4 of 8
1/2
Four independent final events
4
4 of 8
1/2
One final event
1
1 of 8
1/8
One uniformly sampled endpoint cell per history, with no reversion. Carrier counts and detection probabilities alone do not identify the number of origins. The histories are constructed possibilities, not estimated event probabilities.
Changes can be rearranged or imported
Recombination reshuffles alleles within sexual lineages. Horizontal gene transfer moves genetic material between organisms outside ordinary parent-to-offspring transmission. In microbes, a resistance gene can arrive on mobile DNA. A lineage acquiring resistance therefore need not have invented the variant by a new mutation.
Mutation, selection, drift, and transfer can operate together. A resistance frequency rising during treatment does not alone reveal when the causal variant originated or whether it arrived from elsewhere. An explanation should distinguish the source of variation from the process that changed its abundance.
Separate new sequence from new local combinations
A variant can be new to a population without being new in evolutionary history. Consider a recipient population with no A copies. A migrant that passes through without reproducing does not add A to the local offspring gene pool merely by being observed there. A migrant that contributes A through reproduction can introduce it. The sequence may have existed in the source population for many generations. The recipient population gains variation through gene flow, while the original mutation occurred elsewhere.
Recombination raises a different question. Suppose an organism has two homologous chromosome copies carrying the two-locus combinations AB and ab. Without exchange between the loci, its gametes carry the parental combinations. An exchange can produce Ab and aB combinations. No new A, a, B, or b allele is required in this example. A new combination may affect performance even though each component allele already existed. A one-locus count would miss that change because it does not record which alleles travel together.
Use a constructed gamete pool to check conservation. Start with 50 AB gametes and 50 ab gametes. Compare a pool containing 40 AB, 40 ab, 10 Ab, and 10 aB. Each pool has 100 copies at the A locus and 100 at the B locus. Each still contains 50 A and 50 B copies. Yet the share of recombinant combinations changes from zero to 20%. These are illustrative alternative pools, not measured recombination rates. The example shows why changing combinations need not change either single-locus allele frequency.
Allele counts stay fixed as associations change
Combination or count
Parental-only pool
Mixed-combination pool
AB gametes
50
40
ab gametes
50
40
Ab gametes
0
10
aB gametes
0
10
A copies of 100 at A locus
50
50
B copies of 100 at B locus
50
50
Recombinant gametes of 100
0
20
Constructed alternative pools of 100 gametes, each with one copy at each of two loci. The 20% recombinant share is illustrative, not a measured rate. It changes associations without introducing new allele labels.
Horizontal transfer can instead move a stretch of DNA into a lineage outside ordinary parent-to-offspring reproduction. If a bacterium receives DNA carrying an already existing variant and retains it, that variant can be new to the recipient lineage without a new sequence mutation there. Later descendants may inherit the acquired DNA. The history of that DNA segment then includes a transfer route that much of the recipient genome did not follow. Explaining its presence solely by the ancestry of whole organisms would omit part of its inherited history.
These distinctions help evaluate a spatial observation. If A appears at several neighboring sites, possible explanations include repeated mutations, descendants spreading from one origin, and transfer among resident lineages. Similar environmental conditions could favor A at all sites without causing identical origins at all sites. Mapping presence alone cannot show which route occurred. A useful comparison would examine sequences and surrounding ancestry across sites, while allowing that recombination can erase some associations and sampling can miss intermediate populations.
Organize a proposed explanation into a sequence of claims. Identify the source of the sequence, how a lineage acquired it, and what changed its subsequent representation. A variant may originate by mutation, arrive through gene flow or transfer, and then increase through selection while drift still affects its descendants. None of these labels excludes the others. Present usefulness also does not establish why the variant originally arose or what function past selection favored. That historical claim needs evidence about earlier environments and descent, rather than a story inferred only from what the variant does today.
A DNA segment can have a different route
Schematic constructed ancestry. Paths mark inheritance or transfer, not time or genetic distance. A transfers from the donor branch into one recipient branch, then passes to its descendant. The rest of the recipient genome need not share this transfer history.
Exercises
Timing creates different clone sizes
In an ideal binary lineage ending with eight cells, a mutation is inherited by one daughter after the first division. Compare its descendants with a mutation in one daughter of the final division. Assume no fitness difference.
Hint
Count descendant branches, not just mutation events.
Worked answer
Time runs downward through three divisions. Each marker is one cell; filled markers carry the new variant, open markers do not. Every cell divides into two equally; there is no selection or reversion. The drawings are invented lineage histories, not a reproduction of experimental results.
Mutation timing
Mutant endpoint cells
First division
4 of 8
Final division
1 of 8
All descendants inherit the change in this no-reversion toy lineage.
The early mutant daughter has four descendants at the endpoint. The late mutant daughter has one. A larger mutant group can therefore arise from earlier timing without any selective advantage. Actual cultures involve more divisions, losses, and sampling than this diagram.
Do not infer directed mutation
A resistant variant rises from undetected to common after an environmental change. What alternatives must you examine before claiming the environment created precisely the needed mutation?
Hint
Detection, pre-existing variation, and transfer are different possibilities.
Worked answer
Missing evidence
Why it matters
Earlier population sampling
Rare variants can be missed
Lineage or sequence history
Can distinguish origin and arrival
Reproductive differences
Can explain spread after appearance
An observation of spread is not by itself an observation of origin.
The variant might have been present below the sampling detection limit, arisen without regard to its eventual benefit, or arrived through gene transfer or migration. The environment may then favor it. Environmental effects on mutation rate are also possible, but a frequency increase alone cannot establish an adaptive direction for mutation.
Sources and further reading
The explanations and exercises are complete without these optional links. The references cover inheritance, population mechanisms, and evidence for ancestry. Source annotations distinguish textbook explanations, secondary history, and a primary paper whose bibliographic identity alone was checked. The examples are constructed teaching models, not empirical estimates or professional decision advice.
Classic primary paper. Bibliographic identity was checked through Europe PMC; the full paper was not independently reviewed. The historical explanation relies on the linked secondary account.