3.1 Conditions and fitness #
3.1.1 Three conditions, and the response #
Selection follows when variation, heritability, and differential reproductive success co-occur. The differential can sit in survival to reproductive age, mating success, fecundity, or offspring survival. A trait that helps only after reproduction is over is invisible to selection unless kin who share the allele are helped (Section 3.5). A trait that shortens life and raises mating success can spread. Selection does not prepare a population for next century, and it does not act for the good of the species. “The heart is for pumping” is acceptable only as shorthand for past selection, not as a claim about intention.
The quantitative statement for a measured trait is the breeder’s equation, R = h²S. S is the difference between the mean of those who breed and the population mean. R is the difference between the offspring mean and that original mean. Narrow-sense heritability, h², is the fraction of phenotypic variance that is additive genetic. If h² is zero, selection produces no response. If S is zero, there is nothing to respond to. Estimating heritability from relatives is Genetics. The evolutionary use is the prediction: the response has the sign of S and is smaller than S unless heritability is one.
Mutation is not aimed at utility. Most mutations are deleterious or neutral (Chapter 4). Selection sorts variation that already exists or that mutation supplies at random with respect to need.
3.1.2 Absolute and relative fitness #
Absolute fitness, W, is the expected contribution of a genotype to the next generation. Relative fitness, w, scales absolute fitness to a reference genotype, usually the fittest, so that the reference has w = 1. The selection coefficient is s = 1 − w. Allele-frequency change depends on relative fitness. Population growth depends on absolute fitness.
An illustration, not a census: absolute fitnesses 2, 3, and 1 for AA, Aa, and aa give relative fitnesses 2/3, 1, and 1/3. The heterozygote is favoured. All three absolute fitnesses can still lie below replacement, so the population shrinks while the heterozygote increases its share. A genotype can also have a high absolute fitness in a growing population and decline in frequency if its rivals do better. Fitness is not a label a genotype carries. It is a number relative to other genotypes in a stated environment. The melanic moth was favoured in soot and disfavoured after the air cleared.
Mean fitness is the frequency-weighted average of genotypic fitnesses. Selection with constant genotypic fitnesses often increases it. Frequency dependence (Chapter 6) is a standard exception, and drift can decrease it.
3.2 Modes of selection #
The study of Modes of selection provides fundamental principles governing cellular architecture, molecular interactions, and physiological regulation.
Speciation Modes-He
3.2.1 Directional, stabilising, and disruptive #
Directional selection favours one tail. The mean moves. In the 1977 drought on Daphne Major, medium ground finches (Geospiza fortis) that survived had larger beaks, on average, than the pre-drought population, and the shift was inherited. Large beaks handled the remaining hard seeds. In later, wetter years selection swung back. The mean is not on a permanent march.
Stabilising selection favours intermediates. The mean stays, and the variance falls if the extremes fail to breed. Human birth weight is the classical case: extremes survived worse than newborns near the mean, in mid-twentieth-century data. Medical care has since changed the coefficients. Stabilising selection is not an absence of selection. It removes deviations, and the optimum can move.
Disruptive selection favours both tails. Variance rises, and the distribution can become bimodal. In the black-bellied seedcracker, intermediate beaks do worse on the two seed types that the small and large beaks each handle. Disruptive selection splits a population only if mating is also non-random with respect to the trait (Chapter 7). Random mating refills the centre every generation.
| Mode | Favoured | Mean | Variance |
|---|---|---|---|
| Directional | One extreme | Shifts | Often reduced |
| Stabilising | Intermediate | Stays | Reduced |
| Disruptive | Both extremes | Stays if the tails balance | Increased |
| Balancing | Whatever protects rare alleles | Held inside (0, 1) | Polymorphism kept |
3.2.2 Balancing selection and sickle cell #
Balancing selection keeps more than one allele. Heterozygote advantage, negative frequency dependence (Chapter 6), and selection that changes in space or time are different mechanisms.
Sickle-cell haemoglobin is the heterozygote case, restated from Genetics rather than re-taught as a pedigree. Call the common β-globin allele A and the sickling allele S. Where falciparum malaria is intense, AS has higher fitness than AA, because one S allele alters the red cell in a way that resists the parasite, and higher fitness than SS, because SS causes sickle-cell disease. Where malaria is absent, S is only deleterious.
Write relative fitnesses 1 − s for AA, 1 for AS, and 1 − t for SS. The equilibrium frequency of S is about s / (s + t). Severe disease means large t, so the equilibrium lies well below one half. In parts of Africa with intense transmission it reaches about 0.1 and sometimes higher. That is a balance, not a march to fixation. Thalassaemia alleles and G6PD deficiency also track malaria, and not all of them are heterozygote advantage. At the sickle-cell equilibrium, mean fitness is below the fitness of AS. Both homozygotes are produced every generation. Balancing selection maintains variation. It does not give everyone the best genotype, because that genotype does not breed true.
3.3 Sexual selection #
3.3.1 Competition and mate choice #
Sexual selection is differential reproductive success from competition for mates. Intrasexual selection is competition within one sex: fighting, threat, sperm competition. Weapons and large male body size usually sit here. Intersexual selection is mate choice, and ornaments and displays sit here. The two co-occur.
The preference needs an account of its own. In a Fisherian process, preference and ornament become genetically correlated, and the trait can run past the viability optimum. In an indicator, or handicap, process, cost makes the ornament honest evidence of condition. In a sensory-bias process, the preference is older than the ornament and is a side effect of how senses already work. Different species support different mixtures. The peacock’s train is an ornament maintained against a viability cost. Which cue carries the choice is a narrower measurement, not a closed story.
3.3.2 Why dimorphism follows #
Sexual selection is stronger on the sex with the higher variance in reproductive success. That is typically the sex that invests less per offspring and can gain more by extra matings. Eggs are costly and sperm are cheap — anisogamy — and in mammals gestation and lactation add further female investment. Males then compete, and sexual dimorphism is the usual result.
The rule is about variance and investment, not about maleness. Male pipefish brood the eggs, females compete, and females are the ornamented sex. Sexual selection and viability selection are often opposed. The equilibrium ornament is larger than the viability optimum and smaller than the mating optimum. A display that also attracts predators is selection with two signs, not a refutation of selection.
3.4 Adaptation as a hypothesis #
3.4.1 Tests, not stories #
An adaptation is a trait shaped by selection for its current function. The slide into story-telling is easy: a trait exists, a benefit is imagined, and the benefit is called the cause. Gould and Lewontin called these just-so stories. The correction is to treat adaptation as a hypothesis that can fail.
An experiment manipulates the trait or the environment and predicts the fitness effect. A comparative test asks whether the trait evolves repeatedly where the proposed cause evolves. Species are not independent replicates. Close relatives share traits by inheritance, so the comparison has to count independent origins. A quantitative model that predicts a value, and not only a direction, is a third test. Present benefit is not proof of origin, and present uselessness is not proof that a trait was never an adaptation.
3.4.2 Spandrels and constraints #
Gould and Lewontin (1979) compared pan-adaptationism to reading a cathedral so that every architectural byproduct is praised as the point of the building. The triangular spaces where a dome meets its arches — the spandrels of San Marco — are decorated. The decoration is not why the spaces exist. Biological spandrels are traits that exist as consequences of other traits, of materials, or of history, and that may later be used.
The paper is not an attack on selection. It attacks the habit of assuming adaptation in advance and of ignoring alternatives. A trait may be present because of phyletic constraint (the lineage inherited a body that cannot be rebuilt in one step), developmental constraint (the embryos do not produce the variant the story needs), genetic correlation (selection on one trait drags another), or physics. It may also be an adaptation. Constraint means the variation selection would need is absent, correlated, or developmentally costly. That is h² = 0 for the imagined change, sometimes for a deep historical reason.
3.5 Kin selection #
3.5.1 Hamilton’s rule #
An allele can spread if it causes its bearer to help other bearers, even at a cost to the bearer’s own offspring. Hamilton’s rule is rB > C. C is the fitness cost to the actor, B the fitness benefit to the recipient, and r the relatedness of actor to recipient, properly a regression of breeding value. For outbred diploids the pedigree values are an approximation.
| Relationship | Approximate r |
|---|---|
| Parent–offspring, or full siblings | 0.5 |
| Half siblings, or uncle and nephew | 0.25 |
| First cousins | 0.125 |
| Haplodiploid full sisters, one father | 0.75 |
| Haplodiploid sister to her brother | 0.25 |
Inclusive fitness adds effects on an actor’s own offspring to effects on others’ offspring, each weighted by relatedness. It is an accounting method, not a second substance. Kin selection is selection through the indirect term. Helping a full sibling pays when the sibling gains more than twice what the actor loses.
“For the good of the species” is not the same idea. A behaviour that helps the species and hurts the actor’s inclusive fitness is removed inside the population. Group structure and kin selection can be two ways of partitioning the same differential. That equivalence does not rescue sacrifice for the species as such.
3.5.2 Eusociality, and the limit of relatedness #
Eusociality combines overlapping generations, cooperative brood care, and a reproductive division of labour in which some individuals reproduce little or not at all. Honeybees, many ants, termites, and naked mole-rats are the usual examples.
Hymenoptera are haplodiploid. Full sisters who share one father are related by 0.75, and a female is related to her own daughter by 0.5, so raising full sisters can transmit more of her genes than raising daughters. Hamilton suggested this predisposes ants, bees, and wasps to sterile female castes. It is not the whole explanation. Sisters are related by 0.75 only if the queen mated once; many eusocial queens mate more often, and relatedness falls. Most haplodiploid species are not eusocial. Termites are eusocial and diploid. Naked mole-rats are diploid, and inbreeding raises their relatedness by another route. Ecology has to make staying pay: a defensible nest, young that need prolonged care, and a benefit large enough that rB > C holds at the real relatedness. The rule is the inequality, not the number 0.75.
3.6 Key terms, summary, and self-check #
Key terms. Breeder’s equation; absolute fitness; relative fitness; directional selection; stabilising selection; disruptive selection; heterozygote advantage; sexual selection; anisogamy; adaptation; spandrel; constraint; Hamilton’s rule; inclusive fitness; eusociality.
Summary. Selection is heritable variation in reproductive success; the response equals heritability times the selection differential. Allele frequencies follow relative fitness, population growth absolute fitness. Directional selection shifts the mean, stabilising selection cuts variance, disruptive selection favours the tails, and sickle-cell heterozygotes are favoured only where malaria makes AA costly, at an equilibrium set by the two homozygote disadvantages. Sexual selection tracks variance in mating success and produces dimorphism in that sex. Adaptation is a hypothesis; spandrels and constraints are the alternatives that have to be ruled out. Kin selection spreads a costly act when rB > C. Haplodiploid relatedness of 0.75 encouraged the eusociality hypothesis and does not, alone, explain it.
Self-check.
1. A genotype has high absolute fitness but lower fitness than its rival. What happens to its frequency?
2. Why is stabilising selection not “no selection”?
3. With sickle-cell fitnesses 1 − s, 1, and 1 − t, why is S uncommon when disease is severe, and why does S decline where malaria is absent?
4. Why are female pipefish the ornamented sex, if ornaments are “male” traits?
5. Write Hamilton’s rule. Why does hymenopteran r = 0.75 fail, by itself, to explain eusociality?
