Natural selection requires just a handful of conditions, but every one of them must be in place simultaneously or the process stalls. A population needs variation among individuals, that variation must be at least partly heritable, and some variants must survive or reproduce better than others in a given environment. These conditions sound simple, and in a sense they are, yet the interplay among them generates the staggering diversity of life on Earth. Where the story gets more interesting is in the details of each condition, and in how easily any one of them can be weakened or disrupted.
Variation Is the Raw Material
Without differences among individuals, there is nothing for selection to act on. A population of perfectly identical organisms facing a new disease or a shift in temperature would live or die as a group, with no mechanism for adaptation. In practice, no population is truly uniform. Mutation, recombination during sexual reproduction, and gene flow from neighboring populations constantly inject new variants. The variation that matters most for natural selection is phenotypic: differences in body size, behavior, disease resistance, metabolic efficiency, coloring, or any other trait that could affect how well an organism does in its environment.
A key insight from decades of research on selection in the wild is that selection acts on phenotypes regardless of their underlying genetic basis. In other words, natural selection does not “see” DNA directly. It sees the organism’s traits and how those traits perform in a particular setting. Whether a bird’s beak shape is controlled by one gene or fifty, selection simply favors the beak shape that cracks available seeds most efficiently. This distinction between studying selection on traits and studying the genetic architecture behind those traits has been a productive separation in evolutionary biology for over forty years.1Evolution. Phenotypic selection in natural populations: what have we learned in 40 years?
Heritability Makes Selection Stick
Variation alone is not enough. If a trait that helps an organism survive cannot be passed to offspring, the advantage dies with the individual. Heritability is the condition that allows selection to accumulate across generations. When parents who are faster, more resistant to parasites, or better at finding food pass some of that edge to their young, the population shifts over time toward those favorable traits.
In quantitative genetics, the expected change in a trait from one generation to the next depends on both the strength of selection and the trait’s heritability. If heritability is high, even modest selection pressure can shift a population substantially. If heritability is low, even intense selection may produce only a sluggish response. This relationship is central to predicting how quickly populations can adapt.2PubMed Central. The Missing Response to Selection in the Wild – Section: The Statistical Explanations
That said, heritability is not a fixed number stamped on a trait. It shifts depending on the population and the environment. A trait might be highly heritable in one setting and much less so in another, because environmental variation can swamp or reveal genetic differences. This is one reason why wild populations sometimes respond to selection more slowly than laboratory models predict.
Differential Fitness Drives the Process
The third essential condition is that variation in traits must translate into variation in fitness. “Fitness” in an evolutionary sense boils down to how many surviving offspring an individual contributes to the next generation. If all individuals in a population reproduce equally well regardless of their traits, natural selection has no lever to pull.
A useful way to think about this comes from a result known as the Secondary Theorem of Natural Selection, developed independently by Alan Robertson and George Price. The idea is that the amount a trait changes from one generation to the next depends on how strongly that trait covaries with fitness. Traits closely tied to survival and reproduction shift quickly; traits with no connection to fitness drift without direction.3PubMed Central. Fitness and its role in evolutionary genetics
Fitness is not a fixed property of an organism, though. It depends heavily on context. The same traits that confer high fitness in one environment may be neutral or harmful in another. Both the physical environment (temperature, humidity, available resources) and the biological environment (predators, competitors, parasites) shape which phenotypes do well.4PubMed Central. Defining fitness in evolutionary ecology A thick coat is an advantage in the Arctic and a liability in the tropics. Antibiotic resistance helps a bacterium in a hospital but may come at a metabolic cost when antibiotics are absent.
Why Environment Is Sometimes Called a Fourth Condition
Some treatments of natural selection list three conditions (variation, heritability, differential fitness) and leave it at that. Others emphasize the environment as a distinct condition, because it is the environment that determines which traits confer an advantage. Both framings describe the same process, but calling out the environment explicitly makes a practical point: change the environment, and you change which direction selection pushes.
Population density alone can flip the direction of selection. Organisms that thrive when a population is sparse, reproducing rapidly and colonizing open habitat, may be outcompeted by slower-reproducing organisms that are better at tolerating crowding when density is high. These density-dependent shifts produce different life-history strategies at different population levels.5PubMed. Density-dependent natural selection and trade-offs in life history traits The conditions for natural selection are all still met, but the selective outcome reverses because the environmental context changed.
Parasites and pathogens add another layer. One of the more vivid demonstrations of environment-driven selection comes from experiments on the Red Queen hypothesis, the idea that species must keep evolving just to stay in place against coevolving enemies. In laboratory populations of a small roundworm, researchers found that coevolution with a bacterial pathogen strongly favored sexual reproduction over self-fertilization. Populations that could not outcross went extinct, while sexually reproducing populations persisted through ongoing adaptation.6PubMed Central. Running with the Red Queen: host-parasite coevolution selects for biparental sex The pathogen created the selective pressure, the worm’s reproductive variation provided the raw material, and heritability let favorable gene combinations accumulate. All the conditions were in play, but the parasite was the environmental factor that made them matter.
What Happens When a Condition Weakens
Because all the conditions must hold for natural selection to work effectively, weakening any single one can slow or stall adaptation. The most commonly underappreciated case involves population size.
In very small populations, random chance (genetic drift) overpowers selection. A beneficial mutation might simply be lost because its carrier happened to be eaten by a predator before reproducing, or a harmful mutation might spread because its carrier got lucky. The technical way to frame this is that the efficiency of natural selection depends on effective population size. Studies comparing fruit fly species have shown that species with smaller effective populations carry more slightly harmful mutations, have less finely tuned gene usage patterns, and accumulate more repetitive DNA in their genomes, all signatures of weakened selection.7PubMed. Selection efficiency and effective population size in Drosophila species The conditions for natural selection were formally present (there was variation, heritability, and fitness differences), but the population was too small for selection to reliably sort winners from losers.
This matters beyond fruit flies. Endangered species squeezed into tiny remnant populations face exactly this problem. Even if favorable genetic variation exists, random drift may prevent selection from fixing it. Conservation biologists worry about this dynamic constantly.
Trade-Offs Limit What Selection Can Achieve
Even when all conditions are firmly in place, natural selection does not produce perfection. One reason is trade-offs: improving one trait often comes at the cost of another. A genetic variant that boosts one component of fitness may simultaneously reduce another, a phenomenon called antagonistic pleiotropy.8PubMed. Trade-Offs (and Constraints) in Organismal Biology
A classic example is the trade-off between offspring size and offspring number. Investing more resources per egg or baby tends to improve each offspring’s survival, but it also means producing fewer of them. This tension has been documented in organisms from fish to lizards to birds. In sand lizards, for instance, researchers found that this trade-off is shaped by genetic, physiological, and structural constraints, and that differences among individual mothers can make the underlying trade-off hard to detect without careful analysis.9PubMed. Selection and constraints on offspring size-number trade-offs in sand lizards (Lacerta agilis)
Trade-offs mean that natural selection is not climbing toward a single peak of perfection. It is navigating a landscape of compromises, finding solutions that are good enough given the constraints. An organism that could produce many large offspring with no cost would outcompete everything, but biology does not hand out free lunches.
Selection Can Operate at More Than One Level
When people think of natural selection, they usually picture individual organisms competing for survival and reproduction. That picture is accurate but incomplete. Selection can also act at the group level, creating situations where what is good for the individual may conflict with what is good for the group, and vice versa.
A recent study of wild animal populations found that selection on group-level social structure was as strong as, or stronger than, selection on individual social behavior. The mean selection gradient for group social phenotype was actually slightly higher than for individual social phenotype, contrary to the longstanding criticism that group-level selection is too weak to matter. The study also found antagonistic selection, meaning selection sometimes pushed in opposite directions at the individual and group levels.10PubMed Central. Multilevel selection on individual and group social behaviour in the wild This finding complicates the standard picture. Natural selection’s conditions (variation, heritability, differential fitness) can be met at multiple levels simultaneously, and the outcomes at different levels can conflict.
Selection That Maintains Variation Instead of Removing It
A common misconception is that natural selection always reduces variation by driving the “best” variant to fixation. In reality, several forms of selection actively maintain variation within populations. One well-studied mechanism is negative frequency-dependent selection, where rare types have an advantage precisely because they are rare. As a variant becomes more common, its advantage fades, preventing any single type from taking over.
Experimental work has demonstrated this principle at the level of mitochondrial DNA. By manipulating the starting frequencies of genetic variants and tracking their dynamics over generations, researchers confirmed that negative frequency-dependent selection acts to preserve genetic diversity in this region of the genome.11PubMed Central. Negative frequency dependent selection contributes to the maintenance of a global polymorphism in mitochondrial DNA All the usual conditions for selection are met, but the outcome is the opposite of what people expect: instead of reducing variation, selection actively sustains it. The conditions for natural selection do not predetermine its direction or outcome. They just set the stage.
Watching Natural Selection Happen in the Lab
One of the most powerful confirmations that these conditions are both necessary and sufficient comes from long-term evolution experiments with bacteria. Since 1988, Richard Lenski’s lab has propagated twelve populations of E. coli from a single ancestor, now past 50,000 generations. The bacteria have all the ingredients: variation arises through mutation, that variation is heritable because bacteria copy their DNA when they divide, and some mutants grow faster than others in the glucose-limited environment they inhabit.
By tracking which mutations rose to prominence, researchers found that insertion sequences, a type of mobile genetic element, accounted for roughly 35% of the mutations that reached high frequency in populations that kept the ancestral mutation rate.12PubMed Central. Insertion-sequence-mediated mutations both promote and constrain evolvability during a long-term experiment with bacteria Some of these mutations were clearly beneficial. In one well-characterized population, mutations in genes controlling DNA supercoiling arose early and provided measurable fitness gains in competition experiments. Both mutations were independently beneficial, and their effects were additive.13PubMed Central. Long-term experimental evolution in Escherichia coli. XII. DNA topology as a key target of selection
These experiments are valuable because they strip natural selection down to its essentials in a controlled setting. There is no migration, no sexual reproduction, and a tightly regulated environment. Variation enters through mutation, heritability is automatic, and fitness differences in a constant environment are easy to measure. The result is textbook natural selection, observable in real time.
Does Inheritance Have to Be Genetic?
The traditional conditions for natural selection specify heritable variation, and for most of evolutionary history, “heritable” has been assumed to mean “encoded in DNA.” But growing evidence suggests that epigenetic changes, chemical modifications to DNA or its associated proteins that alter gene activity without changing the DNA sequence, can also be inherited across generations and subject to selection.
Epigenetic marks triggered by environmental stress can persist over surprisingly long timeframes and contribute to phenotypic changes in traits that matter for fitness.14PubMed. Epigenetic inheritance in adaptive evolution This does not replace genetic inheritance as the primary channel for evolution, but it adds a layer. If an environmental exposure causes epigenetic changes that alter an organism’s phenotype, and those changes are passed to offspring, then selection can act on them. The epigenetically-fixed phenotype does not need to become encoded in the genotype to affect evolution; the differential survival of individuals carrying advantageous or disadvantageous epigenetic marks is enough to shift the genetic composition of the population over time.15PubMed Central. Epigenetic Inheritance and Its Role in Evolutionary Biology: Re-Evaluation and New Perspectives – Section: 8.2.4. Indirect Epigenetic Effects on Evolution: Altered Natural Selection
Some researchers argue that epigenetic transgenerational inheritance should be fully integrated into evolutionary theory, expanding the inheritance condition beyond strict DNA sequence changes.16Environmental Epigenetics. Role of environmentally induced epigenetic transgenerational inheritance in evolutionary biology: Unified Evolution Theory This is still an active area of debate, but the core point stands: the “heritability” condition for natural selection may be broader than traditionally thought.
Antibiotic Resistance as a Real-World Case
If the conditions for natural selection feel abstract, antibiotic resistance brings them sharply into focus. In a hospital or a farm where antibiotics are used, the conditions align with textbook precision. Bacterial populations harbor variation, including rare mutants that happen to tolerate the drug. That resistance is heritable. And the antibiotic creates intense differential fitness: susceptible bacteria die, resistant ones survive and reproduce.
What makes this example especially striking is how little antibiotic it takes to set the process in motion. Research has shown that resistant bacteria can be selected at concentrations several hundred-fold below the levels that kill susceptible cells. The threshold is reached when the antibiotic’s effect on the susceptible strain is just enough to offset the metabolic cost of carrying resistance. For one well-studied mutation conferring resistance to ciprofloxacin, the minimum concentration needed to select for resistance was roughly 1/230th of the concentration needed to kill susceptible bacteria.17PubMed Central. Selection of antibiotic resistance at very low antibiotic concentrations All three conditions for selection are met at vanishingly low drug levels, which is part of why antibiotic resistance has become so difficult to contain.
Before Biology Itself
A fascinating edge case for the conditions of natural selection is the question of where those conditions first arose. Before cells existed, before DNA and proteins, could natural selection operate on the chemical systems that preceded life?
The short answer is that natural selection requires entities with at least three basic features: a way to store information, a metabolism to sustain themselves, and a boundary (like a membrane) to separate self from non-self.18PubMed. From prebiotic chemistry to cellular metabolism–the chemical evolution of metabolism before Darwinian natural selection Before those features emerged, chemical evolution proceeded by different rules, driven by thermodynamics and reaction kinetics rather than by heritable variation and differential reproduction. The transition from chemistry to biology is, in essence, the moment when the conditions for natural selection first came together. Understanding that boundary helps clarify why the conditions are not just a checklist for biology class but a description of when a fundamentally new kind of process becomes possible.
Digital evolution experiments have pushed this idea further. Computer scientists have built systems where digital organisms copy their genomes instruction by instruction through a noisy channel that introduces errors. Most errors are harmful, but occasionally a mutant has higher fitness than its parent. With all the necessary components for Darwinian evolution present, these digital populations adapt, speciate, and develop complex ecological interactions in ways that mirror biological evolution remarkably well.19PubMed Central. Evolving Digital Ecological Networks The medium does not matter. Carbon or silicon, chemistry or code, when variation is heritable and fitness differs, natural selection follows.