Natural selection comes in several distinct forms, each shaping populations in a different direction. The three classic types taught in biology courses are directional, stabilizing, and disruptive selection, which describe whether a trait’s average value shifts, stays put, or splits into extremes. But the full picture is richer than that. Frequency-dependent selection, sexual selection, balancing selection, kin selection, and coevolutionary selection all operate on populations in ways the classic trio doesn’t capture, and real organisms often experience several of these forces simultaneously.
Directional Selection
Directional selection favors individuals at one end of a trait’s range, pushing the population average in that direction over time. If larger bodies survive winters better, the population gradually gets bigger. If shorter beaks crack a newly abundant seed more efficiently, the population shifts toward shorter beaks. The key feature is that one extreme has a fitness advantage over the other, so the trait distribution slides.
One of the best-documented examples involves Darwin’s finches on the Galápagos Islands. During a drought on the island of Daphne Major, medium ground finches with large beaks faced strong selection against that trait when a competing large-beaked species arrived and monopolized the big seeds. Researchers estimated the selection coefficient against large-beaked genotypes at 0.59, which is remarkably intense by the standards of field biology. The result was rapid character displacement: the medium ground finch population shifted toward smaller beaks within just a few generations.1PubMed. A beak size locus in Darwin’s finches facilitated character displacement during a drought
A large review of selection studies in wild populations found that the median strength of directional selection across hundreds of measured traits was relatively modest, suggesting that the finch example is unusually dramatic. Strong directional selection exists in nature, but it’s the exception rather than the rule for most traits in most populations.2PubMed. The strength of phenotypic selection in natural populations
Stabilizing Selection
Where directional selection moves the average, stabilizing selection holds it in place. Individuals near the middle of a trait’s range do best, and those at either extreme are penalized. Think of birth weight in humans: babies that are too small or too large face higher health risks, so intermediate birth weight is favored. Over time, this narrows the spread of the trait without shifting where the center sits.
Research in contemporary human populations has found evidence that stabilizing selection is widespread across many traits but relatively weak compared to what’s been measured in other species.3PubMed Central. Evidence of directional and stabilizing selection in contemporary humans That weakness makes intuitive sense: modern medicine and technology buffer us from many of the extreme environmental pressures that would otherwise punish individuals at the tails of a distribution. In wild populations without those buffers, stabilizing selection can be considerably stronger.
Disruptive Selection
Disruptive selection is the mirror image of stabilizing selection. Instead of the middle being favored, the extremes win. Individuals with intermediate trait values are at a disadvantage, so the population tends to split into two distinct groups. This is considered a potential route to speciation, because once a single population develops two peaks in a trait distribution, those peaks can eventually diverge into separate species.
Darwin’s finches provide a clean example here too. The medium ground finch on Santa Cruz Island shows a bimodal distribution in beak size: there are small-beaked birds and large-beaked birds, but relatively few in between. Studies confirmed that disruptive selection was strong between the two beak-size groups.4PubMed Central. Disruptive selection in a bimodal population of Darwin’s finches Longer-term monitoring showed that this disruptive selection persists over many years but fluctuates in intensity depending on environmental conditions, becoming strongest when resources are scarce and competition for food is high.5PubMed Central. Temporally varying disruptive selection in the medium ground finch (Geospiza fortis)
This means disruptive selection isn’t simply “on” or “off.” The environment modulates it. In good years with abundant food, the pressure against intermediate beak sizes relaxes. In harsh years, it sharpens. This variability may explain why disruptive selection doesn’t always rapidly produce new species: the selective pressure isn’t constant enough to finish the job.
Frequency-Dependent Selection
Frequency-dependent selection is different from the three classic types because what matters isn’t where you fall on a trait distribution but how common your version of the trait is. In negative frequency-dependent selection, being rare is an advantage. In positive frequency-dependent selection, being common is an advantage. Negative frequency-dependent selection is particularly important in nature because it actively maintains diversity: as soon as one form becomes common, it loses its edge, and rarer forms gain ground.
A striking example comes from scale-eating cichlid fish in Lake Tanganyika. These fish have mouths that twist either left or right, which determines which flank of a prey fish they attack. If most scale-eaters in the population are left-mouthed, prey fish learn to guard their right side, giving right-mouthed scale-eaters a higher success rate on attacks. This keeps the two mouth forms at roughly equal frequencies in the population.6PubMed. Mouth dimorphism in scale-eating cichlid fish from Lake Tanganyika advances individual fitness It’s one of the most powerful forces for maintaining genetic variation, and it shows up across many systems, from color polymorphisms in damselflies to immune-system genes in vertebrates.7Evolution. Negative frequency-dependent selection in female color polymorphism of a damselfly
Balancing Selection and Heterozygote Advantage
Balancing selection is a broader category that includes frequency-dependent selection but also encompasses heterozygote advantage, where carrying two different versions of a gene provides higher fitness than carrying two copies of either version alone. The textbook case is sickle cell trait and malaria.
People who carry two copies of the sickle cell gene develop sickle cell disease, which is severely debilitating. People who carry zero copies have normal red blood cells but are fully vulnerable to malaria. Those who carry one copy, the heterozygotes, get the best of both worlds: their red blood cells are largely normal, but the parasite-infected cells sickle preferentially and get cleared by the immune system, providing substantial protection against malaria.8PubMed Central. Sickle cell anaemia and malaria Studies in East African children found that the sickle cell trait was roughly 40% protective against clinical malaria, though the degree of protection varied with age, peaking at around 60% during the first decade of life before declining.9PubMed Central. An Immune Basis for Malaria Protection by the Sickle Cell Trait
This is why the sickle cell gene persists at high frequencies in malaria-endemic regions despite the severe cost of sickle cell disease in homozygotes. Natural selection is maintaining both versions of the gene because the combination is more fit than either version alone. It’s a vivid reminder that “good” and “bad” genes don’t exist in a vacuum; fitness is always relative to the environment.
Sexual Selection
Sexual selection is a subset of natural selection driven by competition for mates rather than by survival in the environment. Darwin himself recognized it as distinct because it can produce traits that actively reduce survival, like the peacock’s enormous tail, as long as those traits increase mating success enough to compensate.
The peacock’s train is an interesting case because it appears to involve both types of sexual selection simultaneously. Male-male competition (intrasexual selection) plays a role, with peacocks using their elaborate tails to establish dominance hierarchies, somewhat like how deer use antlers. But unlike deer, female choice (intersexual selection) is also at work: peahens evaluate and select males based on the quality and display of their trains.10PubMed Central. A long tail of truth and beauty: A zigzag pattern of feather formation determines the symmetry, complexity, and beauty of the peacock’s tail Most species lean more heavily toward one form of sexual selection or the other, making the peacock something of a dual-purpose case study.
Sexual selection explains a lot of the most flamboyant traits in the animal kingdom: bright plumage, elaborate songs, oversized horns, and complex courtship dances. These traits are often costly to produce and maintain, and their persistence is only understandable once you factor in their role in securing reproductive opportunities.
Kin Selection and Social Behavior
Kin selection explains the evolution of traits that seem to hurt the individual performing them while helping relatives. The underlying logic, captured in Hamilton’s rule, is that a gene promoting altruistic behavior can spread if the benefit to related individuals, weighted by how closely related they are, exceeds the cost to the altruist. Worker bees that never reproduce, ground squirrels that give alarm calls at their own risk, and cooperative breeders that help raise siblings instead of having their own young all make sense under this framework.
Reviews of studies that have quantified the components of Hamilton’s rule confirm that altruism, defined as a net loss of direct reproductive output, does occur even when social behavior is optional. In most documented cases, the indirect fitness benefits gained through helping relatives exceed the direct fitness costs.11PubMed Central. Hamilton’s rule and the causes of social evolution Kin selection and its associated theoretical tools have been extraordinarily productive for understanding the evolution of sociality across insects, birds, mammals, and even microorganisms.12Nature. Kin selection and eusociality
Coevolutionary Selection and the Red Queen
When two species are locked in an ongoing evolutionary relationship, each one’s adaptations create new selective pressures for the other. This is coevolutionary selection, and the most famous version of it is the Red Queen hypothesis, named after the character in Lewis Carroll’s story who has to keep running just to stay in place. The idea is that hosts and parasites are constantly co-adapting: the host evolves defenses, the parasite evolves ways around them, and neither side gains a lasting advantage.
Field studies of freshwater snails and their trematode parasites have provided some of the strongest evidence. Researchers found that rare host genotypes were significantly less likely to be infected by the local parasites, and that parasites tracked common host genotypes in a time-lagged fashion, just as the Red Queen model predicts.13PubMed. Host-parasite coevolution: evidence for rare advantage and time-lagged selection in a natural population Mathematical modeling of these dynamics shows that while two genotypes may dominate the cycling, other genotypes can persist at low frequencies in synchronized but subordinate oscillations, explaining why many rare genotypes stay rare rather than cycling to high frequency as the simple model would predict.14PubMed Central. Host-parasite Red Queen dynamics with phase-locked rare genotypes
The Red Queen hypothesis also provides one of the leading explanations for why sexual reproduction is so common despite its substantial costs. Experimental work with the nematode C. elegans and a bacterial pathogen showed that coevolution with the pathogen drove populations toward more outcrossing (sexual reproduction), while self-fertilizing populations were rapidly driven to extinction. Outcrossing populations, by generating novel genetic combinations each generation, could keep pace with the evolving pathogen in a way that clonal or self-fertilizing populations could not.15PubMed Central. Running with the Red Queen: host-parasite coevolution selects for biparental sex
Fluctuating Selection
Some selective pressures aren’t constant. They shift with the seasons, or with multi-year cycles in climate, food availability, or predator abundance. Fluctuating selection occurs when the trait value favored by natural selection changes over time, sometimes favoring one extreme and sometimes the other. If the fluctuations are regular enough, this can maintain genetic diversity that would otherwise be eroded by consistent directional selection.
Studies of fruit fly populations have detected hundreds to thousands of genetic variants whose frequencies rise and fall with the seasons, suggesting that seasonally fluctuating selection may be far more common than previously appreciated.16PubMed. Fluctuating selection and the determinants of genetic variation Theoretical work shows that this kind of cycling can maintain variation at many genes simultaneously, through a mechanism where the currently favored version of a gene tends to be dominant, giving heterozygotes an edge in each season. This process works best when the fitness effects of different genes interact in certain ways and doesn’t run into the genetic load problems that plague some other models of variation maintenance.17PubMed Central. Seasonally fluctuating selection can maintain polymorphism at many loci via segregation lift
Fluctuating selection blurs the line between the classic types. A population experiencing directional selection toward larger size during cold winters and directional selection toward smaller size during hot summers is, over the full year, experiencing something that looks more like balancing selection from the gene’s perspective. The categories are useful, but nature doesn’t always sort cleanly into them.
Multilevel Selection
Most discussions of natural selection focus on the individual: which individuals survive and reproduce best? But selection can also act at the level of groups. In multilevel selection, traits that are costly to the individual but beneficial to the group can spread if groups containing cooperators outcompete groups of selfish individuals.
Microbial systems have become a productive testing ground for this idea. In bacteria that cooperate by swarming collectively, groups with many cooperators expand faster, produce more offspring groups, and are less likely to collapse from random cell death. Group-level selection can thus favor cooperation even when individual-level selection would favor cheaters who benefit from others’ cooperation without paying the cost. However, the analysis also shows this doesn’t always work: cooperation through group selection requires high relatedness among group members, and constitutive cooperation can only be favored when the costs aren’t too high and returns don’t diminish too steeply.18PubMed Central. Multilevel selection analysis of a microbial social trait One important finding is that bacteria with regulated cooperation genes, which turn cooperation on only when it’s needed, are a more robust strategy than constitutive cooperators because they capture the benefits without bearing the constant costs.19PubMed Central. Multilevel selection favors fragmentation modes that maintain cooperative interactions in multispecies communities
Selection and Climate Change
One of the pressing questions in contemporary biology is whether natural selection can keep pace with the rate of anthropogenic environmental change. When a population encounters a new thermal environment, selection on traits related to heat tolerance can be strong and rapid, but only if the population has sufficient genetic variation to work with.
Experiments transplanting tropical lizards to warmer, more thermally variable environments found strong directional selection on heat-performance traits. The same traits were not under detectable selection in the reference population living in less stressful conditions, confirming that the novel thermal environment was the driver.20PubMed Central. Natural selection on thermal performance in a novel thermal environment Other studies have found that selection can favor unexpected combinations of thermal traits. In one case, individuals that preferred warmer body temperatures but had lower maximum heat tolerance were favored, suggesting that selection acts on the interplay between preferred and critical temperatures rather than simply pushing all thermal traits in the same direction.21PubMed Central. Natural selection on thermal preference, critical thermal maxima and locomotor performance
These results suggest that rapid adaptation to warming is possible, at least for some species, but the outcome depends on whether the relevant traits are heritable and whether the pace of environmental change outstrips the population’s capacity to respond. For species with long generation times or low genetic diversity, selection may simply not work fast enough.
Selection and Antibiotic Resistance
Antibiotic resistance is natural selection observed in real time, on a timescale of days rather than generations of vertebrates. When bacteria are exposed to an antibiotic, those with mutations conferring resistance survive and proliferate. The process follows the same principles as directional selection in finches or lizards, just compressed.
Experimental evolution studies tracking bacterial populations exposed to increasing antibiotic concentrations have quantified how different forces shape resistance outcomes. Early on, historical differences between lineages accounted for most of the variation in resistance levels. But after about twelve days of exposure at high antibiotic concentrations, selection explained nearly half the variation, and the effects of lineage history had dropped substantially. Populations that started with lower resistance gained resistance faster, while populations already above the effective concentration showed slower gains, suggesting that resistance levels converge toward a predictable ceiling.22PubMed Central. The roles of history, chance, and natural selection in the evolution of antibiotic resistance This convergence is a hallmark of strong directional selection: it doesn’t matter much where you start if the selective pressure is intense enough.
Trade-Offs and Constraints on Selection
Natural selection doesn’t operate on traits in isolation. Organisms are integrated systems, and improving one trait often comes at the expense of another. These trade-offs place real limits on what selection can achieve. The conflict between survival and reproduction is the most universal example: energy invested in producing more offspring is energy not available for immune defense or tissue repair, and vice versa.23PubMed. Constraints, Trade-offs and the Currency of Fitness
At the genetic level, antagonistic pleiotropy describes situations where a single gene variant improves one aspect of fitness while simultaneously worsening another. Shared biochemical pathways, often mediated by hormones or other signaling molecules, can link traits in ways that prevent selection from optimizing each one independently.24PubMed. Trade-Offs (and Constraints) in Organismal Biology This is why organisms are never “perfectly adapted.” Selection pushes toward better compromises, not toward perfection in any single trait.
The practical implication is that understanding what selection favors requires looking at the whole organism, not just the trait in question. A mutation that makes a bacterium more resistant to one antibiotic might make it grow more slowly in the absence of that drug. A gene variant that improves heat tolerance might reduce cold tolerance. Selection produces solutions that work well enough across the full range of conditions an organism faces, which means they’re rarely optimal for any single condition.
How Artificial Selection Differs
Comparing natural and artificial selection sharpens the picture of what makes natural selection distinctive. When humans breed animals or plants for specific traits, the selection is usually intense, highly directional, and focused on a narrow set of characteristics. Natural selection, by contrast, acts on everything at once, is often weaker per trait, and fluctuates with environmental conditions.
A genomic comparison between Mexican turkeys, which descended from originally domesticated wild turkeys and have experienced primarily natural (adaptive) selection, and commercial hybrid turkeys, which have undergone about forty years of intense artificial selection, illustrates the difference. The commercial hybrids showed striking genomic homogeneity, a signature of strong directional selection narrowing the gene pool. The Mexican turkeys retained enough genomic diversity to separate into subgroups based on geographic origin, reflecting the more diffuse and variable nature of adaptive selection in the wild.25PubMed Central. Hybrid Versus Autochthonous Turkey Populations: Homozygous Genomic Regions Occurrences Due to Artificial and Natural Selection
This genetic homogeneity is a double-edged sword. It produces the traits breeders want, like rapid weight gain or uniform egg production, but it also strips away the variation that would allow the population to respond to new challenges. Wild populations maintain diversity precisely because natural selection is multidirectional and fluctuating. That diversity is the raw material for future adaptation, and it’s something artificial selection tends to erode.