Sickle cell trait kills. Individuals carrying two copies of the sickle cell allele (homozygotes) suffer severe pain, organ damage, and often early death. Sickle cell should be completely eliminated from populations by natural selection. Yet in Sub-Saharan African populations exposed to malaria, sickle cell frequency reaches 25–40% of the population—extraordinarily high for a trait that's lethal in homozygotes. This paradox reveals how disease operates as selection pressure. In regions without malaria, sickle cell is purely harmful—homozygotes die, heterozygotes suffer minor symptoms. Selection eliminates the allele. But in regions with endemic malaria, sickle cell heterozygotes gain an advantage: they're protected from malaria death without suffering sickle cell death (the one sickle allele is enough for malaria protection; the one normal allele prevents severe sickle cell). This means in malaria regions, heterozygotes outreproduce both homozygotes (sickle-sickle, who suffer sickle cell disease) and normals (normal-normal, who die from malaria). Over generations, heterozygotes become common. The allele frequency rises not because it's beneficial in absolute terms, but because it's beneficial relative to the two alternatives in a malaria-endemic environment.1
This is natural selection in real time. Malaria kills people without sickle cell protection. Sickle cell disease kills people with two sickle alleles. In the middle, heterozygotes survive better than either extreme. So heterozygotes become more common. The allele frequency—the proportion of individuals carrying sickle cell—drifts toward the point where heterozygote advantage is maximized. This is called balancing selection: the population settles at a frequency where the harmful allele is maintained because heterozygotes survive best. The mechanism is universal: any disease that kills non-resistant individuals will select for disease-resistance alleles, regardless of what those alleles cost in other contexts.
Alleles as Inherited Variants
An allele is one version of a gene. For blood type, you might have allele A, B, or O. For disease resistance, you might have a normal CCR5 receptor allele or a deletion allele (CCR5-Δ32). Allele frequency is the proportion of that allele in the population—if 10% of people carry the deletion allele, the allele frequency is 10%. Allele frequencies change over generations when certain alleles increase reproduction (or decrease death) relative to others. This change is evolution.
Selection Pressure and Fitness Differences
Selection pressure is any environmental factor that affects reproduction or survival differently for different genotypes. Malaria is a selection pressure: it kills people without sickle cell protection more often than it kills people with protection, creating a difference in survival probability. This survival difference translates to reproduction difference: people who survive longer can reproduce more. People who die young can't reproduce. This is the mechanism: environmental pressure (malaria) creates survival difference (some genotypes survive more often), which creates reproduction difference (survivors reproduce more), which changes allele frequency (their alleles become more common).1
Balancing Selection vs. Directional Selection
Directional selection is when one allele is simply better than others (produces more offspring consistently). This drives the allele to fixation (100% frequency) as the superior allele eliminates inferior ones. Balancing selection is when the allele's fitness depends on context: sickle cell is harmful in non-malaria regions (directional selection toward eliminating it) but beneficial in malaria regions (balancing selection maintaining both alleles at intermediate frequency). This is why sickle cell is common in malaria regions but nearly absent elsewhere—the environment literally inverts which allele is selected for.
Sickle cell emerges in Sub-Saharan African populations exposed to Plasmodium falciparum malaria. In these regions, individuals with genotype heterozygous (AS: one sickle, one normal allele) survive malaria better than AA (normal-normal) and SS (sickle-sickle). The frequency rises over generations:
In non-malaria regions (Europe, North America, areas without malaria exposure), sickle cell frequency is near 0% because the allele is purely harmful with no malaria advantage.1
The mathematical pattern: at equilibrium, allele frequency = s₂/(s₁ + s₂), where s₁ is the fitness cost of being homozygous sickle (severe disease risk) and s₂ is the fitness cost of being normal-homozygous without sickle protection (malaria death risk). When malaria death risk equals sickle cell disease risk, allele frequency stabilizes at ~50%. When one cost dominates, frequency shifts accordingly. The pattern is predictable from fitness differences.
CCR5-Δ32 is a 32-base-pair deletion in the CCR5 gene, which encodes a chemokine receptor used by plague bacteria (Yersinia pestis) to invade cells. Individuals homozygous for this deletion (Δ32/Δ32) are highly resistant to plague. Heterozygotes (CCR5/Δ32) have moderate resistance. Normals (CCR5/CCR5) are susceptible.
Plague swept through Europe repeatedly, killing 25–50% of population in each major epidemic (1347–1353, 1361–1362, recurring locally for centuries). Survivors included disproportionate numbers of Δ32 carriers because they were more likely to survive infection. Their offspring inherited the deletion allele. Over 400 years of repeated plague exposure (1347–1700s), the allele frequency increased from near-zero to ~10% in some Northern European populations.1
Geographic pattern: Δ32 frequency is highest in Northern Europe (10% in some populations), intermediate in Southern Europe (2–5%), nearly zero in Africa and Asia (where plague wasn't endemic or wasn't as lethal). The geographic pattern tracks plague history: populations repeatedly exposed to plague show high Δ32 frequency; populations without plague exposure have low frequency. This geographic correlation is strong evidence for plague as selection pressure.
HLA (human leukocyte antigen) genes control immune response to pathogens. Different HLA variants are more effective against different diseases. A population exposed to endemic tuberculosis will show selection for HLA variants that respond well to TB. A population exposed to endemic malaria will show selection for HLA variants that respond well to malaria parasites. The result: HLA allele frequencies vary by endemic disease environment.1
Comparative evidence:
The differences are not random genetic drift—they're predictable from disease history. Populations that faced similar diseases show similar HLA patterns. Populations that faced different diseases show different patterns. This is evidence that disease selects for immune alleles.
The best evidence for disease selection pressure comes from selection in real time. When humans migrate to new disease environments, genetic change happens fast. African populations migrating to malaria-free regions show sickle cell frequency declining over generations (no longer selected for). European populations migrating to malaria-endemic regions (if they interbreed with local populations) show sickle cell frequency increasing as the allele is selected for locally. This rapid change—occurring in decades to centuries, not thousands of years—proves that disease is a real selection pressure operating on current populations, not just a historical phenomenon.
Tension 1: Selection Pressure Takes Extreme Cost
Sickle cell selection required thousands of deaths per year from malaria (creating selection pressure for protection) but also killed thousands per year from sickle cell disease in homozygotes. The "solution" (widespread sickle cell) is itself a catastrophe. Did selection pressure select for or against overall population health? The paradox: the adaptation is tragic—it's "better" than malaria alone or sickle cell alone, but worse than neither. Tension: is this selection beneficial? Does benefit depend on perspective (if you have the heterozygote genotype, you benefit; if you're unlucky enough to inherit two sickle alleles, you suffer).1
Tension 2: Genetic vs. Behavioral Adaptation
Humans can develop behavioral immunity: use nets to avoid mosquitoes, drain swamps to reduce breeding habitat, use quinine to treat malaria. These don't require genetic selection—they work within a generation. Yet populations historically relied on genetic selection while behavioral solutions existed. Why did Sub-Saharan populations experience thousands of years of malaria selection pressure when mosquito nets (invented centuries ago) could prevent transmission? The tension: are genetic adaptations to disease actually adaptive, or are they mark of failure to develop behavioral alternatives?
Tension 3: Genetic Disease and Moral Judgment
Carriers of sickle cell trait, Tay-Sachs, cystic fibrosis face genetic disease as consequence of ancestral selection pressure. A person carrying two CF alleles suffers severe disease with no benefit. The genetic adaptation selected for heterozygote advantage, but some individuals inherit homozygous harmful state. This creates suffering for individuals who gain no benefit from the ancestral selection. Is the population-level adaptation worth the individual suffering?
Diamond emphasizes disease as selection pressure—showing how epidemic diseases selected for resistance alleles in Eurasian populations but not American populations, creating immune system divergence. But he largely sidesteps the question of whether selection-driven genetic adaptation is "good" or "bad"—he presents it as mechanical fact (populations exposed to disease show genetic resistance, naive populations don't), not as moral judgment. A fuller analysis would grapple with whether genetic resistance is preferable to behavioral prevention of disease. Diamond's framework captures the genetic mechanism without fully resolving whether genetic selection is adaptive or tragic.1
Antibiotic Resistance and Human-Driven Selection Pressure — Modern medicine creates selection pressure for antibiotic resistance in bacteria. When doctors prescribe antibiotics, they kill susceptible bacteria but allow resistant bacteria to survive and reproduce. Over generations, resistant bacteria become common. This is identical to disease selection pressure on humans: an environmental factor (antibiotics) kills non-resistant individuals (bacteria) and selects for resistant individuals. The insight that transfers: selection pressure operates the same way regardless of whether it's disease selecting on humans or antibiotics selecting on bacteria. In both cases, the mechanism is: survival difference creates reproduction difference, which changes population genetics. The difference is timescale: bacterial selection happens in months; human selection happens in generations. But the principle is identical. This means humans today are potentially repeating the evolutionary process we historically experienced: creating selection pressure (antibiotic overuse) that selects for resistant populations (antibiotic-resistant bacteria). We're not aware that we're selecting for resistance the way malaria selected for sickle cell, but we are.
Vaccination as Population Immunity Without Selection Pressure — Vaccination is a technological intervention that provides immunity without genetic selection pressure. Vaccinated individuals get protection without the genetic cost—no need for sickle cell trait, no need for CCR5-Δ32 deletion. This means vaccination breaks the historical pattern: instead of populations evolving genetic resistance through deadly selection pressure, populations can achieve herd immunity through technology. The structural parallel to sickle cell: sickle cell is genetic immunity (heterozygotes survive malaria through genetic mechanism); vaccination is technological immunity (everyone survives through injected antibodies, no genetic selection required). The insight that transfers: technological solutions to disease can prevent the genetic selection that would otherwise occur. This means vaccination has a hidden cost: it relieves selection pressure, allowing disease-susceptibility alleles to remain in populations. If vaccination were ever abandoned, modern populations would lack the genetic immunity that ancestral selection pressure created. This is a tension Diamond doesn't fully explore: modern medicine breaks the selection pressure that historically shaped human genetics, with unknown long-term consequences if medical technology fails.
The Sharpest Implication
If disease acts as selection pressure on human populations, then modern medicine is disrupting selection processes that operated for 10,000 years. Antibiotics, vaccines, and hygiene remove selection pressure for disease resistance—allowing disease-susceptibility alleles to persist in modern populations rather than being eliminated as they historically would be. This means we're becoming genetically more disease-susceptible as a species, not less, because we've removed the selection pressure that maintained resistance. The uncomfortable implication: modern medicine prevents the genetic evolution that would historically have occurred, potentially making us dependent on continued medical intervention. If antibiotics were unavailable tomorrow, modern populations would have no evolved resistance to bacterial infection. We'd be more vulnerable than pre-medicine populations because we lack the genetic hardening of ancestral selection pressure. This doesn't mean medicine is bad—it's better to have safe antibiotics than to rely on painful genetic selection—but it reveals an invisible cost: technological dependence replacing genetic adaptation.
Generative Questions