Biology › Evolution and speciation › Natural selection: what varies, what is selected, and what actually changes
Natural selection: what varies, what is selected, and what actually changes
Selection cannot invent anything. It works on variation that already exists, it acts on phenotypes it can see, and what it leaves behind is a population with different allele frequencies from the one it started with. Get those three sentences in the right order and most of this topic follows.
Before this Meiosis and the origin of variation · Population genetics and the Hardy-Weinberg principle · The genetic code and mutation
Before you start
Evolution is survival of the fittest, and the fittest are the strongest, fastest and biggest. Almost every part of that is wrong. Fitness in biology is not athletic — it is the number of offspring an individual leaves that themselves survive to breed, measured relative to everyone else in the population. A small, slow, unimpressive organism that raises eleven surviving young is fitter than a magnificent one that raises none. And survival on its own settles nothing: an individual that lives for thirty years and never reproduces has contributed exactly as much to the next generation's gene pool as one that died as an embryo.
What you should be able to do
- Name the sources of genetic variation and say which one produces an allele that did not exist before.
- Distinguish variation that can be selected on from variation that cannot be inherited.
- Define fitness and selection pressure in terms an examiner can credit.
- Recognise stabilising, directional and disruptive selection from a distribution and from a description.
- Explain why selection acts on phenotype while evolution is measured as a change in allele frequency.
- Write about selection without using language that implies organisms change on purpose.
Selection cannot act on variation that is not there
Start with the raw material, because most wrong answers on this topic come from getting the order of events backwards. A population contains individuals that differ, and only the genetic differences matter for evolution.
There are three ways a population acquires genetic variation, and they are not equivalent. Mutation is a change in the base sequence of DNA, and it is the only one that creates an allele which did not exist in the population before; everything else rearranges what mutation has already supplied. Crossing over in prophase I puts maternal and paternal alleles onto the same chromatid, and independent assortment at metaphase I gives a human 2²³ chromosome combinations before crossing over is counted. Random fertilisation then pairs any one of those gametes with any one of the other parent's.
Two things follow. A population with no genetic variation cannot respond to a new selection pressure at all, however severe it is, which is why a monoculture of genetically identical crop plants can be destroyed by one new strain of fungus.
The second point is about timing. Mutations happen whether or not they would be useful, and the environment does not induce the one it needs. It selects among mutations that had already occurred, most of which were neutral or harmful when they arose. Mutation is random with respect to fitness — a narrower claim than 'random', since mutation rates vary along a chromosome, but it is the claim that matters here.
- Mutation
- A change in the base sequence of DNA. The only source of new alleles.
- Genetic variation
- Differences between individuals in a population that are caused by differences in their alleles, and can therefore be inherited.
- Environmental variation
- Differences between individuals caused by the conditions they have experienced, which are not passed to offspring.
- Gene pool
- All the alleles of all the genes in a population at a given time.
Environmental variation deserves its own paragraph, because it is where the teleology usually starts. A plant grown in shade is taller and paler than the same clone grown in sun; a weightlifter's children are not born stronger. Selection can act on an environmentally produced phenotype — a small plant is shaded out whatever made it small — but if the difference has no genetic basis, nothing is passed on and no evolution occurs. Most characteristics fall between the two: height, yield and birth mass are polygenic and also responsive to conditions, which is why they show continuous variation, a smooth range with a peak in the middle rather than a few distinct classes.
Selection pressure, and a definition of fitness that survives contact with a mark scheme
A selection pressure is any environmental factor that causes individuals with different phenotypes to survive and reproduce at different rates: predators, parasites, drought, cold, competition for food or for mates, an antibiotic, a pesticide. The pressure does nothing to any individual's genotype. It changes only how many copies of that genotype reach the next generation.
Fitness is the term students think they already know, and the everyday meaning is close enough to the biological one to be dangerous. Properly written, an individual's fitness is its contribution to the gene pool of the next generation relative to other individuals in the same population. It is counted as surviving fertile offspring, and it is always relative: a fitness of 0.8 means only that this genotype leaves 80% as many surviving offspring as the best one, in that environment, at that time.
- Selection pressure
- An environmental factor that causes individuals with certain phenotypes to survive and reproduce more successfully than others.
- Fitness
- The contribution an individual makes to the gene pool of the next generation, relative to other individuals in the population — usually counted as offspring that themselves survive to reproduce.
- Adaptation
- A characteristic that increases an organism's chance of surviving and reproducing in its environment. Adaptations may be anatomical, physiological or behavioural.
- Natural selection
- The process by which individuals with phenotypes better suited to the environment reproduce more successfully, so the alleles responsible become more frequent.
Three consequences are examinable and routinely missed. Fitness is environment-specific: a pesticide resistance allele is usually mildly costly when the pesticide is absent, which is why resistance often declines after spraying stops. Fitness is population-specific, so a rare male colour form can do well precisely because it is rare. And fitness is about reproduction, so a peacock's tail is favoured if it buys enough matings to pay for the risk it carries.
Note what fitness is not. It is not a property of a species, so 'humans are the fittest species' is not a sentence with a meaning; it is not something an individual can improve during its life; and it is not a prediction, because the environment that defines it changes.
Three shapes of selection, and the shape of the distribution each leaves
For a characteristic showing continuous variation, plot the number of individuals against the value of the characteristic and you get a distribution, usually a rough bell. Selection can change it in only three ways, depending on which part of the range is favoured.
Stabilising selection favours the middle of the range and acts against both extremes. The mean stays where it is and the variance falls, so the curve becomes taller and narrower. This is the commonest mode in a population living in a stable environment, and it removes variation rather than creating change. Human birth mass is the standard example: in Karn and Penrose's 1951 data, infant mortality was lowest at around 3.6 kg and rose steeply either side of it. Use it carefully now, because modern obstetric care has weakened the effect considerably — the pressure was on birth mass, and medicine changed the pressure.
Directional selection favours one end of the range. Individuals at that end leave more offspring, the whole distribution moves that way, and the mean shifts. This is the mode that produces visible evolutionary change, and it is what happens when an environment alters. Peter and Rosemary Grant watched it on Daphne Major in the Galápagos in 1977, when a drought left mainly large hard seeds. Ground finches with deeper, stronger beaks could crack them and most of the rest starved. The survivors' beaks averaged around 5% deeper than the pre-drought population, and their offspring inherited most of that difference.
Disruptive selection is the rarest and the most interesting, because it acts against the middle and favours both extremes. Variance increases and the single peak splits into two. Thomas Bates Smith's work on the black-bellied seedcracker in Cameroon is the cleanest case: the finches feed on two sedges, one with hard seeds and one with soft, and birds have either large bills or small ones with almost no intermediates, even though the two forms interbreed freely. Disruptive selection matters later in this unit, because it is one of the few routes by which a population can begin to split while still living in one place.
| Stabilising | Directional | Disruptive | |
|---|---|---|---|
| Favoured | The mean phenotype | One extreme | Both extremes |
| Selected against | Both extremes | The other extreme | The mean |
| Effect on the mean | Unchanged | Moves towards the favoured extreme | May not move at all |
| Effect on variation | Reduced | Roughly unchanged | Increased |
| Curve after selection | Taller and narrower | Shifted sideways | Two peaks with a trough between |
| Typical situation | A stable environment | An environment that has changed | Two distinct resources, nothing in between |
| Example | Human birth mass | Beak depth in ground finches after drought | Bill size in the black-bellied seedcracker |
A board note before you revise the table. AQA's specification names stabilising and directional selection and does not list disruptive selection; OCR Biology A and Cambridge International expect all three. The third costs one row of a table and protects you against a data question with a two-peaked distribution, so learn it whichever board you sit.
Selection sees phenotypes; evolution is counted in alleles
Here is the join with the population genetics you have already met. A predator, a drought and an antibiotic all act on the phenotype — how fast the animal runs, how deep the beak is, whether the cell wall survives — and none of them detects a genotype directly. But what persists into the next generation is not the phenotype, which dies with its owner; it is the alleles that helped build it. Selection operates at one level and its effect is recorded at another, and the definition of evolution is written at the second: a change in the allele frequencies of a population over time.
That two-level structure explains several things that look odd at first. A recessive allele in a heterozygote has no effect on the phenotype, so selection cannot touch it — which is why a harmful recessive allele is never eliminated by selection alone, and why the rarer it becomes, the larger the fraction of it sheltered in carriers. A dominant harmful allele is exposed in everyone carrying it and disappears quickly.
evolution = a change in allele frequency in a gene poolHardy-Weinberg gives the null: with no selection, no mutation, no migration, random mating and a large population, the frequencies do not change at all. A measured change is therefore evidence that at least one of those conditions is being broken.
Following one generation of selection through the arithmetic
A dominant allele A gives resistance to a fungal disease, and in a large randomly mating population its frequency is 0.10. The disease arrives. The resistant phenotype has a relative fitness of 1.0 and homozygous recessive individuals 0.5. Calculate the frequency of A in the next generation. Idealised figures, chosen to keep the arithmetic clean.
Start from the genotype frequencies before selection, using Hardy-Weinberg with p = 0.10 and q = 0.90: p² = 0.01 for AA, 2pq = 0.18 for Aa, q² = 0.81 for aa.
Weight each genotype by its fitness. AA and Aa both show the resistant phenotype, so they are unchanged at 0.01 and 0.18; the aa class is halved to 0.405.
These no longer add to 1, because part of the population has been removed. The total is 0.01 + 0.18 + 0.405 = 0.595, so divide each by it: AA = 0.0168, Aa = 0.3025, aa = 0.6807.
Allele frequency is counted over alleles, so every AA individual contributes two copies of A and every Aa one: p′ = 0.0168 + ½ × 0.3025 = 0.168.
A has risen from 0.100 to 0.168 in a single generation, because selection acted on the phenotype of the aa individuals. Notice what did not happen: no allele changed into another, and nothing new appeared.
Worth noticing for later: q has fallen only from 0.900 to 0.832, even though half the aa individuals were removed. Most copies of the recessive allele were sitting in heterozygotes, out of reach.
TRY IT — Reading a distribution rather than describing it
An annual plant is measured for stem height in 1990 and again in 2010. In 1990 the distribution is symmetrical, mean 42 cm, standard deviation 9 cm. In 2010 the mean is 41 cm and the standard deviation is 4 cm. Grazing animals were excluded throughout and the climate did not change.
Name the type of selection acting, and explain how the two statistics support your answer.
Check your answer
Stabilising selection.
The mean is essentially unchanged — 42 cm to 41 cm — so the population has not moved towards either extreme, which rules out directional selection.
The standard deviation has fallen from 9 cm to 4 cm, so variation around the mean has been more than halved. Plants much taller and much shorter than the mean contributed fewer offspring than plants near it, so the alleles producing extreme heights became less frequent.
The stable environment stated in the question is the expected setting: the phenotype already best suited is still the best suited.
An answer saying only 'the plants adapted to stay the same height' scores nothing. The marks are in the two statistics.
Five sentences that lose marks, and what to write instead
This is the topic where careless phrasing costs the most, because the convenient shorthand implies a mechanism that does not exist. Examiners are briefed on these, and they are worth reading slowly.
| Do not write | Why it is wrong | Write instead |
|---|---|---|
| The organism adapted to the cold | An individual cannot change its alleles during its lifetime. Populations become adapted; individuals do not adapt. | Individuals with alleles for thicker fur survived the cold better and left more offspring, so those alleles became more frequent |
| A mutation arose because the population needed it | Mutation is random with respect to fitness. Need does not cause mutation. | A mutation had already occurred by chance; when conditions changed it happened to be advantageous |
| The fittest survived | Circular unless fitness is defined, and survival alone changes nothing without reproduction. | Individuals with the advantageous phenotype survived and reproduced more successfully than the others |
| The species evolved a longer beak so it could reach the seeds | Puts the purpose before the variation. Nothing is produced in order to be useful. | Beak depth varied; birds with deeper beaks could crack the available seeds and left more offspring |
| The bacteria became resistant when the antibiotic was applied | Implies the antibiotic caused the change. It selected among variation already present. | A few bacteria already carried a resistance allele; the antibiotic killed the rest, so the resistant ones came to dominate the population |
A reliable check on your own writing is to find the word 'so' and ask which side of it the purpose sits on. 'Deeper beaks meant more seeds could be cracked, so those birds left more offspring' runs cause to consequence and is fine. 'The birds needed to crack more seeds, so they evolved deeper beaks' runs the other way and is not.
Four steps, in this order, answer almost any 'explain how natural selection produced X' question. Genetic variation exists in the population, caused by mutation. A selection pressure is named. Individuals with a specified phenotype survive and reproduce more successfully. The alleles responsible are passed on, so their frequency increases over generations. Miss the first step and the answer describes Lamarck; miss the last and it describes one generation of survival rather than evolution.
In the exam
- Say 'variation caused by mutation' at the start of every natural selection answer. The variation must exist before the pressure arrives, and the first mark is usually for saying so.
- Name the selection pressure the question has given you. 'The environment changed' is not a selection pressure; 'the antibiotic', 'the drought', 'the predator' are.
- Define fitness as reproductive contribution relative to others, not as strength or as survival. Answers that say 'the strongest survive' are marked wrong on sight.
- For a distribution question, quote the mean and the spread separately. Stabilising keeps the mean and cuts the spread; directional moves the mean; disruptive raises the spread and splits the peak.
- Finish on allele frequency. An answer that stops at 'the resistant ones survived' has described selection but not evolution.
- Never write that an organism adapts, wants, needs or tries. Populations become adapted; alleles become more frequent.
Check yourself
A species of moth rests on tree trunks. In an industrial area the trunks are darkened by soot, and over fifty years the proportion of dark moths in the population rises from 2% to over 90%. Explain this change, and then explain why the same alleles were present in the population before the trunks darkened.
Answer
Mutation had already produced an allele for dark colouring, so both phenotypes were present before the soot arrived. Mutation happens whether or not the allele is useful.
Darkened trunks are the selection pressure. Against a dark background, pale moths are more conspicuous to birds hunting by sight and are eaten more often; dark moths are better camouflaged.
Dark moths therefore reproduce in greater numbers and pass on the allele. Over many generations its frequency in the gene pool rises, so the proportion of dark moths rises. That change in allele frequency is the evolution.
Before the trunks darkened the dark allele was present but rare, because against pale lichen-covered bark the dark phenotype was the conspicuous one: held at low frequency by selection, not absent. This is Biston betularia, and the detail is unusually well documented — the carbonaria form went from under 2% in Manchester in 1848 to about 98% by 1895, and the mutation has been identified as a transposable element in the cortex gene, dated to around 1819.
The move that scores is refusing to say the soot caused the mutation.
Questions
Question 14 marks
A field of crops is sprayed with a new insecticide every year, and after ten years most of the insects in that field survive the spray. Explain how this change came about.
Mark scheme
- B1 genetic variation was already present in the population, produced by random mutation, so a few insects carried an allele giving resistance before the insecticide was ever used
- B1 the insecticide is the selection pressure: insects without the allele are killed, while those carrying it survive
- B1 the survivors reproduce and pass the allele to their offspring, so a greater proportion of the next generation carries it
- B1 repeated over many generations the frequency of the resistance allele in the gene pool rises, and that change in allele frequency is the evolution — the spray selected among variation already present rather than causing the mutation
Question 24 marks
Compare stabilising selection with disruptive selection, referring to which phenotypes are favoured and to what happens to the mean and to the variation.
Mark scheme
- B1 stabilising selection favours the mean phenotype and acts against both extremes, whereas disruptive selection acts against the mean and favours both extremes
- B1 under stabilising selection the mean is unchanged, and under disruptive selection the mean may not move either, so the mean alone does not distinguish them
- B1 stabilising selection reduces the variation, so the curve becomes taller and narrower, whereas disruptive selection increases it and the single peak splits into two with a trough between
- B1 stabilising selection is the usual mode in a stable environment, whereas disruptive selection needs two distinct resources with nothing in between, as for a finch feeding on hard seeds and on soft ones
Question 34 marks
A recessive allele causes a metabolic condition, and 1 person in every 2500 is born with it. Assuming the Hardy-Weinberg conditions hold, calculate the frequency of the recessive allele and calculate the percentage of the population who are heterozygous carriers.
Mark scheme
- M1 the homozygous recessive genotype has frequency q2, so q2 = 1 ÷ 2500 = 0.0004; the square root is taken of that genotype frequency, not of 2500
- A1 q = the square root of 0.0004 = 0.02
- M1 p = 1 − q = 0.98, and the frequency of heterozygotes is 2pq = 2 × 0.98 × 0.02
- A1 0.0392, so 3.92 per cent of the population are carriers
Question 43 marks
An allele giving resistance to a pesticide reaches a frequency of 0.6 in a field where the pesticide is used every season. Spraying then stops, and over the following years the frequency of the allele falls. Suggest why.
Mark scheme
- B1 with no pesticide present the allele confers no advantage, so it is no longer favoured and the individuals lacking it are no longer killed
- B1 resistance usually carries a cost, such as a protein that does its normal job less well or energy spent running a mechanism the insect does not need, so carriers leave slightly fewer surviving offspring than the rest
- B1 the frequency of the allele in the gene pool therefore falls generation by generation, which shows that fitness belongs to a genotype in a particular environment rather than being a fixed property of it
Question 52 marks
Name the only source of alleles that did not exist in a population before, and name two processes that produce new combinations of alleles that already exist.
Mark scheme
- B1 mutation, a change in the base sequence of DNA
- B1 crossing over in prophase I and independent assortment at metaphase I of meiosis, with random fertilisation combining any gamete of one parent with any of the other's
Question 62 marks
State what is meant by the fitness of an individual, and state why survival on its own does not raise it.
Mark scheme
- B1 an individual's contribution to the gene pool of the next generation relative to other individuals in the same population, counted as offspring that themselves survive to reproduce
- B1 an individual that lives a long time and never reproduces contributes nothing to the next generation's gene pool, so survival counts only in so far as it leads to reproduction
Worth remembering
- Mutation is the only source of new alleles; meiosis and random fertilisation make new combinations of alleles that already exist.
- Mutation is random with respect to fitness. Environments select; they do not instruct.
- Fitness is relative reproductive contribution, counted in surviving fertile offspring.
- Stabilising narrows the distribution, directional moves it, disruptive splits it. AQA names the first two; OCR A and Cambridge International expect all three.
- Selection acts on the phenotype; evolution is measured as the change in allele frequency it leaves behind.
- Recessive alleles hide in heterozygotes, which is why selection removes them slowly and never completely.