Stabilizing selection is a form of natural selection in which individuals with intermediate values of a trait have higher biological fitness than individuals with more extreme values. It tends to maintain a population near a local optimum and can reduce variation in the selected trait. Unlike selection that consistently favors larger, smaller, earlier, or later values, stabilizing selection acts against departures in both directions from a favored range. Its defining feature is the relationship between a trait and reproductive success, not simply the absence of visible evolutionary change.
Mechanism
A trait’s contribution to fitness often involves competing benefits and costs. A structure that is too small may perform poorly, while an excessively large structure may require too much energy, interfere with movement, or increase exposure to predators. If intermediate values provide the most favorable balance, individuals near that range contribute disproportionately to subsequent generations.
The relevant trait is part of an organism’s phenotype, which reflects both inherited factors and environmental influences. Selection can sort phenotypic differences within a generation, but a sustained evolutionary response depends on how those differences relate to inherited variation. A trait may experience strong selection yet show little change across generations if the selected differences are largely environmental.
The favored value is called an optimum, but it is not necessarily the population’s current average. If the population mean lies away from the optimum, selection can initially move the mean toward it while also acting against extreme deviations. Stabilizing and directional effects can therefore occur together.
The optimum is also conditional: it depends on the environment, the organism’s other traits, and interactions with other organisms. Stabilizing selection does not imply an ideal form that remains best under all circumstances.
Comparison with other forms of selection
Three commonly distinguished patterns describe how fitness varies with a continuous trait:
- Directional selection favors movement toward one end of the observed trait range.
- Stabilizing selection favors an intermediate region over both extremes.
- Disruptive selection favors separated trait values over intermediate ones.
These are descriptions of fitness relationships, not rigid categories into which every population must fit. The pattern can change across locations, seasons, or life stages. A trait may experience stabilizing selection over a broad range but appear to experience directional selection within a population occupying only one side of that range.
Stabilizing selection is also distinct from purifying selection. Purifying selection generally refers to the removal of harmful genetic variants, especially in molecular evolution. Stabilizing selection describes selection around a phenotypic optimum. Purifying selection against mutations that move a trait away from its optimum can help produce stabilizing selection, but the terms are not interchangeable.
A mathematical model
A common model represents fitness as a bell-shaped function of a quantitative trait:
where:
- is an individual’s trait value;
- is the optimal trait value;
- is fitness at the optimum;
- determines the width of the fitness function.
A smaller means that fitness falls more sharply as the trait departs from the optimum, indicating stronger stabilizing selection.
If the initial trait distribution is normal, with mean and variance , weighting that distribution by this fitness function produces a selected distribution with mean
and variance
Under these assumptions, the selected mean lies closer to the optimum and the selected variance is smaller than the initial variance. If the initial mean already equals the optimum, selection leaves the mean unchanged while reducing variance.
These equations describe the distribution after the specified selection episode. They do not, by themselves, predict the distribution among offspring. Reproduction, inheritance, environmental variation, mutation, and migration can alter the next generation’s distribution.
Examples and ecological interpretation
Gall size in goldenrod gall flies
A well-studied example concerns the goldenrod gall fly, Eurosta solidaginis. Its larvae develop inside swellings, or galls, on goldenrod stems. Different natural enemies can impose opposing pressures on gall size: some parasitoid wasps more readily attack larvae in smaller galls, whereas bird predators may preferentially attack larger galls.
Where these pressures operate together, intermediate gall sizes can have a survival advantage. Their relative strengths vary among sites and years, so the example illustrates an ecological mechanism rather than an invariant rule. Gall size also reflects interactions between the insect and its host plant.
Resource-allocation traits
Traits involving allocation of limited resources can also produce intermediate optima. In reproductive investment, for example, investing too little may yield few surviving offspring, while investing too much may reduce parental survival or the resources available to each offspring.
Such trade-offs can generate stabilizing selection, but a plausible trade-off is not sufficient evidence. Demonstrating the pattern requires measurements showing that reproductive success declines on both sides of a favored trait range.
Genetic variation and long-term evolution
Stabilizing selection can help explain why some traits remain relatively constant over many generations. This does not mean that evolution has stopped. Allele frequencies, genetic interactions, and other traits may change while a particular phenotypic average remains near its optimum.
The persistence of genetic variation under stabilizing selection is an important problem in quantitative genetics. If departures from the optimum are disadvantageous, selection might be expected to remove the variants producing them. Several processes complicate that expectation:
- Mutation continually introduces new variants, including variants that shift traits away from the optimum.
- Gene flow introduces variants from populations experiencing different conditions.
- Genetic drift changes allele frequencies through chance, particularly in small populations.
- Environmental heterogeneity can produce different optima across places or times.
- Genetic interactions and effects on multiple traits mean that a variant’s contribution to fitness cannot always be inferred from its effect on one measured trait.
A stable phenotype can therefore coexist with substantial genetic variation. Conversely, low phenotypic variation does not by itself demonstrate stabilizing selection; developmental constraints or limited environmental variation may also contribute.
Detecting stabilizing selection
Researchers typically measure a trait and a component of fitness, such as survival, mating success, or offspring production. A common statistical approach fits a model containing both linear and quadratic terms:
where is relative fitness, describes a linear component, and describes curvature.
A negative quadratic coefficient is consistent with a fitness relationship that bends downward. However, it is not sufficient on its own to demonstrate that intermediate individuals are favored over both extremes within the observed population. The estimated fitness maximum may lie outside the sampled range, or the data may poorly describe the extremes.
Important complications include:
- Correlated traits: an apparent fitness relationship may arise because the measured trait is associated with another trait under selection.
- Shared environmental effects: favorable conditions may influence both trait expression and fitness, creating an association that is not caused by the trait itself.
- Incomplete fitness measurements: survival during one life stage may not represent lifetime reproductive success.
- Limited sampling: rare extreme individuals can be difficult to measure accurately.
- Changing conditions: combining observations across environments may obscure distinct local patterns.
Manipulative experiments, multivariate analyses, and repeated observations can help distinguish these possibilities.
Relationship to stasis and developmental stability
Stabilizing selection is one possible explanation for long-term phenotypic stasis, but stasis alone is not evidence for it. Little observed change may also reflect weak selection, limited heritable variation, developmental constraints, or opposing episodes of directional selection.
It is likewise distinct from canalization, the tendency of development to produce similar phenotypes despite genetic or environmental disturbances. Stabilizing selection can, under some conditions, favor mechanisms that buffer departures from an optimum. Canalization is a property of developmental robustness; stabilizing selection is a pattern of differential reproductive success.