Evolution I: Evolutionary Thought, Selection, Life History, and Population and Quantitative Genetics

Lamarck (Philosophie zoologique, 1809) proposed that organisms change through use and disuse of organs and that characters acquired in a lifetime are inherited; the giraffe’s neck lengthened by stretching. Darwin (On the Origin of Species, 1859), with Wallace, proposed natural selection: individuals vary, much of the variation is heritable, more offspring are produced than survive, and those whose heritable traits suit the environment leave more descendants. Darwin lacked a mechanism of inheritance; the Modern Synthesis of the 1930s–40s (Fisher, Haldane, Wright, Dobzhansky, Mayr, Simpson, Huxley) united Mendelian genetics with selection, showing that continuous variation arises from many Mendelian loci and that evolution is a change in allele frequencies driven by mutation, selection, drift and gene flow.

Fitness is reproductive success: absolute fitness is the expected number of offspring, relative fitness w is fitness divided by that of the best genotype (or by the mean), and the selection coefficient s = 1 − w measures the disadvantage. An adaptation is a trait shaped by selection for its current function; a trait that arose for one function and was later co-opted for another (feathers, perhaps first for insulation) is an exaptation. Not every trait is an adaptation — some are by-products of other traits or of history.

An organism has a finite budget of energy and time to divide among growth, maintenance, survival and reproduction, so allocation to one function comes at the expense of another. These trade-offs shape life histories: current reproduction against future survival (the cost of reproduction), offspring number against offspring size (the Smith–Fretwell trade-off), and early against late reproduction. Lack’s clutch size hypothesis says birds lay the number of eggs that maximises the number of fledglings raised, which is often fewer than the maximum they could lay, because very large broods are underfed.

Semelparous organisms reproduce once and die — Pacific salmon, many annual plants, bamboos that flower gregariously after decades, Agave; iteroparous organisms reproduce repeatedly — most vertebrates and perennial plants. Semelparity is favoured when adult survival is low or uncertain relative to juvenile survival, or when a large one-off investment gives disproportionately high returns (an Agave’s tall inflorescence attracting pollinators). Cole’s paradox noted that a semelparous annual needs only one more offspring per year than an iteroparous perennial to match it; the resolution is that juvenile survival is usually far lower than adult survival. The r/K scheme has largely given way to demographic life-history theory, in which age-specific mortality drives the evolution of these traits.

Genetic variation originates in mutation — point mutations, insertions and deletions, duplications, chromosomal rearrangements — and is reshuffled each generation by recombination, independent assortment and random fertilisation. Mendel’s laws of segregation (a monohybrid F₂ of 3:1, genotypes 1:2:1) and independent assortment (a dihybrid F₂ of 9:3:3:1) hold for unlinked loci. Epistasis, an interaction between loci in which one gene masks or modifies another, changes the dihybrid ratio: recessive epistasis gives 9:3:4 (coat colour in Labrador retrievers), duplicate recessive (complementary) epistasis 9:7, dominant epistasis 12:3:1 and duplicate dominant 15:1.

The Hardy–Weinberg principle: in a large, randomly mating population with no mutation, selection or migration, allele frequencies p and q (p + q = 1) stay constant and genotype frequencies reach p², 2pq and q² after one generation of random mating. Two uses dominate questions. Counting alleles from genotypes: with 360 AA, 480 Aa and 160 aa, p = (2 × 360 + 480)/2000 = 0.6, and the expected heterozygotes 2 × 0.6 × 0.4 × 1000 = 480 match the observed. Inferring carriers from a recessive condition: if 1 in 400 people shows a recessive trait, q² = 0.0025, q = 0.05, p = 0.95 and the carrier frequency is 2pq = 0.095, nearly 1 in 10 — far more carriers than affected individuals.

Population genetic structure: a panmictic population mates at random throughout; real populations are subdivided. Wright’s F_ST = (H_T − H_S)/H_T measures the fraction of total heterozygosity due to differences between subpopulations, where H_S is the mean expected heterozygosity within subpopulations and H_T the expected heterozygosity of the pooled population. With H_S = 0.30 and H_T = 0.40, F_ST = 0.25 — great differentiation by Wright’s rough guide (0–0.05 little, 0.05–0.15 moderate, 0.15–0.25 great, above 0.25 very great). Gene flow homogenises: in the island model at equilibrium, F_ST ≈ 1/(4Nm + 1), where Nm is the number of migrants per generation, so about one migrant per generation (F_ST ≈ 0.2) is enough to prevent subpopulations drifting to fixation for different alleles. Migration changes allele frequency by Δp = m(p_m − p).

Genetic drift is random change in allele frequency from generation to generation because only a sample of gametes forms the next generation. Its effects are stronger in small populations: allele frequencies wander, alleles are lost or fixed, and heterozygosity declines. A new neutral mutation in a diploid population of N has a probability 1/(2N) of eventually being fixed, and every neutral allele’s probability of fixation equals its current frequency. The founder effect (a new population started by a few individuals) and bottlenecks (a drastic temporary reduction) are episodes of intense drift; the northern elephant seal, hunted to a few tens of animals in the 1890s, still shows very low genetic variation.

The effective population size Nₑ is the size of an idealised population that would drift at the same rate; it is usually much smaller than the census size. Three standard corrections: an unequal sex ratio, Nₑ = 4N_mN_f/(N_m + N_f) — 10 males and 40 females give Nₑ = 1600/50 = 32, not 50; fluctuating size, where Nₑ is the harmonic mean of the generation sizes — 1000, 10 and 1000 give 3/(0.001 + 0.1 + 0.001) = 29.4, dominated by the bottleneck; and variance in family size, which lowers Nₑ further. Heterozygosity is lost at a rate of 1/(2Nₑ) per generation: H_t = H₀(1 − 1/(2Nₑ))^t. With Nₑ = 25 the loss is 2 % per generation and after 10 generations H₁₀/H₀ = 0.98¹⁰ = 0.817.

Give the genotypes AA, Aa and aa relative fitnesses w₁₁, w₁₂ and w₂₂. After selection the frequency of a is q′ = (pq w₁₂ + q² w₂₂)/w̄, where the mean fitness is w̄ = p²w₁₁ + 2pq w₁₂ + q²w₂₂. For selection against a recessive (w₁₁ = w₁₂ = 1, w₂₂ = 1 − s) this simplifies to q′ = q(1 − sq)/(1 − sq²) and Δq = −sq²(1 − q)/(1 − sq²). Worked example, q = 0.4 and s = 0.5: w̄ = 1 − 0.5 × 0.16 = 0.92 and q′ = 0.4 × 0.8/0.92 = 0.32/0.92 = 0.348. Selection against a rare recessive is very slow, because rare recessive alleles hide in heterozygotes where selection cannot see them.

The sickle-cell allele Hb^S in malarial regions is the classic case of overdominance: the heterozygote is protected against Plasmodium falciparum malaria, the Hb^A homozygote is susceptible and the Hb^S homozygote suffers sickle-cell disease. If w(AA) = 0.9 (s₁ = 0.1) and w(SS) = 0.2 (s₂ = 0.8), the equilibrium frequency of S is 0.1/0.9 = 0.11, and it persists at that frequency despite being nearly lethal in homozygotes. Frequency-dependent selection can also maintain polymorphism: when rare types are favoured (negative frequency dependence), as in self-incompatibility alleles or scale-eating cichlids with left- and right-bent mouths, each allele rises when rare.

Polygenic traits — height, body mass, yield — are controlled by many loci of small effect plus the environment, and vary continuously. The phenotypic variance partitions as V_P = V_G + V_E (plus a genotype-by-environment interaction V_G×E and covariance where they exist), and the genetic variance as V_G = V_A + V_D + V_I: additive, dominance and interaction (epistatic) variance. Broad-sense heritability H² = V_G/V_P; narrow-sense heritability h² = V_A/V_P, the part that makes offspring resemble their parents and the part selection acts on. It is estimated from the slope of offspring on mid-parent values (the slope equals h²), from twins or from the response to selection.

The breeder’s equation R = h²S predicts the response to selection R (the change in the mean between generations) from the selection differential S (the mean of the selected parents minus the population mean). If the population mean is 50, the selected parents average 60 and their offspring average 54, then S = 10, R = 4 and the realised heritability h² = R/S = 0.4. Heritability is a property of a population in an environment, not of a trait: it can differ between populations, and a high heritability within groups says nothing about the cause of a difference between groups.

Phenotypic plasticity is the ability of one genotype to produce different phenotypes in different environments, described by its norm of reaction (phenotype plotted against environment). When the norms of different genotypes are not parallel — they cross or diverge — there is a genotype-by-environment interaction: the best genotype in one environment is not the best in another. Examples are the helmet induced in Daphnia by predator chemicals, temperature-dependent sex determination in many turtles and crocodilians, and the leaf forms of amphibious plants. Epigenetic inheritance passes on states of gene expression without a change in DNA sequence — DNA methylation, histone modification and small RNAs. Genomic imprinting in mammals and paramutation in maize are established examples; the extent and stability of transgenerational epigenetic effects in animals remain actively studied.