Selection, drift, and what actually drives evolution.

Selection is differential survival and reproduction based on heritable phenotypic variation. Mutation generates variation, recombination shuffles it, drift randomly eliminates it, but only selection consistently builds adaptation. For a single locus under additive selection, Δp ≈ sp(1−p) / 2, where s is the selection coefficient. For a quantitative trait, the breeder's equation R = h²S ties the response to the product of heritability and the selection differential. Selection acts on phenotypes, not genotypes directly, which is where pleiotropy, epistasis, and GxE all live.

Directional selection shifts the mean. Stabilizing selection narrows the variance. Disruptive selection pulls it apart and can drive speciation. Frequency-dependent and density-dependent regimes can flip direction as conditions change.

Fisher (The Genetical Theory of Natural Selection, 1930) formalized selection in mathematical terms and gave us the fundamental theorem: the rate of increase in fitness equals the additive genetic variance in fitness. Price (1970, Nature, "Selection and Covariance") showed that evolutionary change is the covariance between trait and fitness, the most general description of selection we have. Lande (1979, Evolution 33: 402–416) extended the theory to multivariate traits so that correlated characters could be tracked simultaneously. Empirical estimates of the dominance coefficient h for new mutations cluster around 0.2–0.3 (Manna, Martin & Lenormand 2011), which is why weakly recessive deleterious alleles dominate the load in most outbred populations.

The lab's drift papers make a stubborn point: whether a chromosomal rearrangement fixes depends less on how selected it is than on how small the population is that carries it. In Coleoptera, Blackmon et al. (2024, Journal of Heredity 115(2): 173–182, "Drift drives the evolution of chromosome number I") show that effective population size tracks karyotype change across beetles better than any measure of selective pressure. The companion, Jonika et al. (2024, Journal of Heredity 115(5): 524–531, "Drift drives the evolution of chromosome number II"), does the same thing in Carnivora using range size as a proxy for N_e. Both point to nonadaptive forces as the main driver of chromosome evolution in these groups.

Kimura (1968, Nature 217: 624–626, "Evolutionary Rate at the Molecular Level") proposed that most molecular evolution is driven by drift. Ohta (1973, Nature 246: 96–98, "Slightly Deleterious Mutant Substitutions in Evolution") sharpened this to the nearly neutral case: most new mutations are slightly deleterious and their fate depends on N_e. The mature framework in Ohta (1992, Annu. Rev. Ecol. Syst. 23: 263–286) ties it to the molecular clock. Lynch (2007, The Origins of Genome Architecture, Sinauer) argued that much of genome structure (intron proliferation, genome size, mobile elements) reflects the inability of selection to prevent bloat when N_e is small. The lab's Blackmon et al. (2019, Evolution, "Meiotic drive shapes rates of karyotype evolution in mammals", doi:10.1111/evo.13682) extends this to chromosome rearrangements across mammals.

Darwin (1871, The Descent of Man, and Selection in Relation to Sex) named sexual selection as a distinct force, arising from differential mating success rather than differential survival. Lande (1981, PNAS 78: 3721–3725, "Models of speciation by sexual selection on polygenic traits") showed that Fisherian runaway can drive speciation without ecological divergence. Rice (1984, Evolution 38: 735–742, "Sex Chromosomes and the Evolution of Sexual Dimorphism") laid out how sex chromosomes resolve sexual antagonism by differential time spent in each sex.

"Sexual selection depends, not on a struggle for existence in relation to other organic beings or to external conditions, but on a struggle between the individuals of one sex, generally the males, for the possession of the other sex." (Darwin, 1871)

When an allele benefits one sex and harms the other, resolution mechanisms include sex-limited expression, gene duplication and divergence, movement onto sex chromosomes, genomic imprinting (especially in placental mammals), and condition-dependent expression (Bonduriansky & Chenoweth 2009, TREE). Resolution is rarely complete. Burch et al. (2024, Evolution 78(4): 624–634, "Wright was right", doi:10.1093/evolut/qpae003) argue that part of why resolution stalls is that the relevant architecture is non-additive, so a single sex-limited switch cannot fully decouple male and female fitness.

In an XY system, an X-linked allele spends two-thirds of its time in females, making the X a natural home for female-benefit antagonistic alleles. A Y-linked allele is male-exclusive. This creates selection to expand sex-linked regions and drives a lot of sex chromosome evolution. Blackmon & Brandvain (2017, Genetics, "Long-term fragility of Y chromosomes is dominated by short-term resolution of sexual antagonism", doi:10.1534/genetics.117.300382) show how sexually antagonistic selection drives Y fragility and turnover. Blackmon & Demuth (2015, Bioessays, "The fragile Y hypothesis: implications of autosome-sex chromosome fusions", doi:10.1002/bies.201500040) connects the conflict to chromosome evolution more broadly.

Charlesworth, Morgan & Charlesworth (1993, Genetics 134: 1289–1303, "The Effect of Deleterious Mutations on Neutral Molecular Variation") formalized background selection: purifying selection against deleterious mutations at linked loci reduces the effective population size at nearby neutral sites. Stronger purifying selection and lower recombination mean larger reductions. Regions with little recombination (near centromeres, on sex chromosomes, within inversions) show reduced diversity, reduced efficacy of selection, and faster accumulation of deleterious mutations.

Maynard Smith & Haigh (1974, Genetical Research 23: 23–35, "The Hitch-Hiking Effect of a Favourable Gene") showed that a beneficial mutation sweeping to fixation drags linked neutral (and even slightly deleterious) variants along with it, creating a selective sweep: a region of reduced diversity around the selected site. The signature is characteristic: reduced heterozygosity, excess rare alleles, elevated linkage disequilibrium. This is how we detect recent positive selection in genome scans.

Hill & Robertson (1966, Genetical Research 8: 269–294, "The Effect of Linkage on Limits to Artificial Selection") showed that selection at multiple linked loci interferes with itself: a beneficial allele at one locus can be linked to a deleterious allele at another, blocking both from their optimal frequencies. This mutual interference reduces the efficacy of selection and is one of the strongest theoretical arguments for why recombination itself should evolve. The same population-genetic core explains the well-documented selection limits in long-term breeding experiments.

A chromosomal inversion captures a block of genes into a non-recombining unit. If that block contains locally adapted or sexually antagonistic alleles, the inversion is favored indirectly because it preserves beneficial gene combinations. Kirkpatrick & Barton (2006, Genetics 173: 419–434, "Chromosome Inversions, Local Adaptation and Speciation") is the classic model. Joron et al. (2011, Nature 477: 203–206, "Chromosomal rearrangements maintain a polymorphic supergene controlling butterfly mimicry") is the Heliconius case. Thompson & Jiggins (2014, Heredity 113: 1–8, "Supergenes and their role in evolution") reviews the phenomenon across taxa: white-throated sparrow morphs, fire ant social chromosomes, and more.

Fusions change which genes are sex-linked and how they recombine. Anderson, Hjelmen & Blackmon (2020, Biology Letters, "The probability of fusions joining sex chromosomes and autosomes", doi:10.1098/rsbl.2020.0648) shows that fusions joining autosomes to sex chromosomes occur at non-random rates. Wilhoit, Alexander & Blackmon (2024, PeerJ, "Worse than nothing at all: the inequality of fusions joining autosomes to the PAR and non-PAR portions of sex chromosomes", doi:10.7717/peerj.17740) shows that where the fusion attaches matters profoundly: PAR and non-PAR fusions have different evolutionary consequences.

Modifiers that change recombination can be favored in either direction. On a rugged epistatic landscape, reduced recombination preserves good gene combinations. In a changing environment, increased recombination generates the novel combinations needed to adapt. Burch et al. (2024) on pervasive epistasis across more than 1000 line-cross datasets (doi:10.1093/evolut/qpae003) bears on this: the landscape really is rugged in most systems, which has implications for how often recombination modifiers should be favored for or against.

Artificial selection applies the same machinery (heritable variation plus differential reproduction) with human preference replacing ecological fitness. Darwin (1868, The Variation of Animals and Plants under Domestication, John Murray) began the Origin argument with artificial selection because it made the mechanism undeniable. Falconer & Mackay (1996, Introduction to Quantitative Genetics, 4th ed., Longman) is the standard reference for selection response theory and the breeder's equation.

Response can be rapid: wolves to Chihuahuas, teosinte to maize, red junglefowl to broilers, all in a few thousand generations. Correlated responses (the "domestication syndrome") reveal pleiotropy. Domestication bottlenecks reduce diversity. Linkage drag imports unwanted neighbors along with any introgressed trait.

Reduced effective population size during domestication pushes the boundary of effective neutrality outward (Ohta 1992), so slightly deleterious alleles drift to fixation. Dogs, rice, soy, and chickens all show elevated d_N/d_S in conserved regions and excess derived deleterious alleles relative to their wild progenitors (Marsden et al. 2016; Liu & Zhang 2017). This is the same nearly-neutral process at work in Y degeneration and small-island endemics, with humans as the agent of N_e reduction. Relaxed selection on wild-type traits (anti-predator behavior, immune function) lets them degrade, which is evidence that they were being actively maintained in the wild.

Betta fish (Betta splendens) have been bred for centuries for color, fin morphology, and aggression: a clean case of extreme phenotypic diversity from a single wild ancestor. Chickens, domesticated from red junglefowl, are now the world's most abundant bird and a model for rapid adaptation under intense artificial selection. We also work with tomatoes (fruit size, flavor, yield) and beetles (comparative genomics), using domesticates as tractable systems where the selection history is at least partially known.

Divergent natural selection in different environments creates reproductive isolation as a byproduct of adaptation: populations adapting to different niches accumulate differences that reduce hybrid fitness. Lande (1981, PNAS) showed that divergent mate preferences can drive rapid prezygotic isolation even without ecological divergence. When hybrids are less fit, selection favors assortative mating, a process called reinforcement (the "Wallace effect"). Dobzhansky (1937, Genetics and the Origin of Species, Columbia), Mayr (1942, Systematics and the Origin of Species, Columbia), Coyne & Orr (2004, Speciation, Sinauer), and Nosil (2012, Ecological Speciation, Oxford) are the canonical stack.

Independently derived mutations that work in their home background but interact badly in hybrids create Dobzhansky-Muller incompatibilities. Under the symmetric assumptions of Orr (1995), incompatibilities accumulate as roughly the square of divergence time (the "snowball" effect). The empirical scaling in Drosophila and yeast is accelerating but not cleanly quadratic: the exact curve depends on which loci are involved and on demographic history.

Haldane's rule: in crosses between species, the heterogametic sex (XY males, ZW females) is more often inviable or infertile. Contributing mechanisms include the dominance theory (recessive incompatibilities on the X exposed in hemizygous males), faster-X evolution, and meiotic drive. Blackmon & Demuth (2015, Current Opinion in Insect Science, "Genomic origins of insect sex chromosomes", doi:10.1016/j.cois.2014.12.003) ties this to sex chromosome evolution and speciation in insects.

With gene flow, selection creates "genomic islands of speciation": regions that resist introgression because they hold locally adapted or incompatible alleles, while the rest of the genome mixes freely. These islands track inversions and other low-recombination regions, which brings the speciation story back to the same genome-architecture questions as linked and indirect selection. It is also why our work on chromosome fusions and PAR vs. non-PAR attachment matters for speciation as well as for chromosome evolution.