Directional, Stabilising, and Disruptive Selection
~11 min read · Tier toggle changes depth (T1 plain · T2 standard · T3 rigour)
You should understand the basic logic of natural selection, specifically how heritable variation leads to differential reproduction, and be comfortable interpreting basic frequency distribution graphs.
Why this matters
Natural selection is not a single force but a set of distinct statistical patterns that shape how populations evolve over time.
While the previous lecture established the logical necessity of selection, this lecture categorises the specific ways in which selection acts on phenotypic variation. By distinguishing between directional, stabilising, and disruptive selection, we can predict how the shape of a population's trait distribution will change. This classification is fundamental to understanding evolutionary dynamics, as it explains why some populations shift their mean trait value, why others maintain a narrow range of optimal traits, and why others may split into distinct subgroups. Recognising these patterns allows biologists to interpret evolutionary pressures in response to environmental changes, stabilising conditions, or heterogeneous niches.
The Three Modes of Selection in Plain Language
Imagine a population of birds with beak sizes that vary from small to large. The way natural selection acts on this variation depends entirely on the relationship between beak size and survival or reproductive success. There are three primary patterns, or modes, of selection. First, consider directional selection. This occurs when individuals at one extreme of the trait distribution have higher fitness than those in the middle or at the other extreme. For example, if a population of insects is exposed to a new pesticide, only those with a specific genetic resistance might survive. Over time, the average resistance level in the population increases. The entire distribution curve shifts to the right. This is the most common form of selection observed during rapid environmental changes, such as climate change or the introduction of a new predator. It drives the population towards a new optimum.
Second, consider stabilising selection. This is the opposite of directional selection in terms of the extremes. Here, individuals with intermediate phenotypes have the highest fitness, while those at both extremes are selected against. A classic example is human birth weight. Babies that are too small may lack the energy reserves to survive, while babies that are too large may cause complications during birth, reducing survival rates for both mother and child. Consequently, the population maintains a narrow range of birth weights around the mean. The variance decreases, and the curve becomes taller and narrower. This mode of selection is common in stable environments where an existing trait is already well-adapted.
Third, consider disruptive selection (also known as diversifying selection). In this scenario, individuals at both extremes of the trait distribution have higher fitness than those in the middle. This can happen if the environment offers two distinct niches. For example, in a bird population where both small seeds and large seeds are abundant, but medium-sized seeds are scarce, birds with small beaks (efficient for small seeds) and large beaks (efficient for large seeds) will outcompete birds with medium beaks. Over time, the single peak in the distribution may split into two distinct peaks. This mode of selection can lead to polymorphism or even speciation if the two groups stop interbreeding. Understanding these three modes is crucial because they describe the different statistical outcomes of the same underlying process: differential survival and reproduction based on heritable traits.
Directional Selection: Shifting the Mean
Directional selection is the process by which the mean value of a quantitative trait shifts in one direction over generations. This occurs when the fitness function is monotonic; that is, fitness increases as the trait value increases (or decreases). Mathematically, if we let \( x \) represent the trait value and \( w(x) \) represent the relative fitness, directional selection implies that \( \frac{dw}{dx} \) is consistently positive or negative across the range of variation. The result is a change in the population mean, \( \bar{x} \), while the variance, \( \sigma^2 \), may remain relatively constant or change slightly depending on the strength of selection at the tails.
A concrete example is the evolution of antibiotic resistance in bacteria. When a population of bacteria is exposed to an antibiotic, the individuals with low resistance die, while those with high resistance survive and reproduce. The next generation has a higher average resistance level than the previous one. This is a direct application of the principle that heritable variation leads to evolutionary change. The selection pressure is unidirectional: more resistance is better, up to a certain physiological limit. In this context, the fitness landscape is a slope. The population moves "uphill" towards the peak of fitness. The rate of this shift depends on the strength of selection, \( s \), and the heritability of the trait, \( h^2 \). The response to selection, \( R \), can be estimated using the breeder's equation: \( R = h^2 S \), where \( S \) is the selection differential. In directional selection, \( S \) is the difference between the mean of the selected parents and the mean of the entire population.
It is important to note that directional selection does not necessarily mean that the trait will increase indefinitely. There are usually trade-offs. For instance, in the case of beak size, larger beaks might be better for cracking hard seeds, but they might require more energy to maintain or reduce the bird's agility. Eventually, the fitness curve will flatten out, and the mean will stabilise at a new optimum. However, during the period of active selection, the defining characteristic is the consistent shift in the mean. This is distinct from stabilising selection, where the mean remains constant, and disruptive selection, where the mean may remain constant but the variance increases significantly.
Stabilising Selection: Minimising Variance
Stabilising selection acts to reduce the variance of a trait around the mean. In this mode, the intermediate phenotype has the highest fitness, while both extremes are selected against. The fitness function is concave, with a maximum at the optimal trait value \( x_{opt} \). Individuals deviating from \( x_{opt} \) have lower fitness, regardless of the direction of deviation. The result is a narrowing of the phenotypic distribution. The mean \( \bar{x} \) remains approximately constant, but the standard deviation \( \sigma \) decreases over generations. This mode of selection is prevalent in stable environments where the current phenotype is already well-adapted to the prevailing conditions.
A classic example is the clutch size in birds. If a bird lays too few eggs, it may not produce enough offspring to replace itself. If it lays too many eggs, it may not be able to provide sufficient food for all the chicks, leading to high mortality. Therefore, there is an optimal clutch size that maximises the number of surviving offspring. Birds with clutch sizes far from this optimum have lower fitness. Over time, the population maintains a tight distribution of clutch sizes around this optimum. This is stabilising selection. Another example is the length of a mammal's gestation period. Too short a gestation results in underdeveloped offspring, while too long a gestation increases the risk of maternal mortality and reduces the number of reproductive cycles. Thus, the population stabilises around a specific gestation length.
The mathematical consequence of stabilising selection is a reduction in genetic variance. If we consider the fitness of an individual with trait value \( x \) as \( w(x) = 1 - s(x - x_{opt})^2 \), where \( s \) is the strength of selection, then the change in variance per generation is negative. This is in contrast to directional selection, where the change in mean is the primary effect. Stabilising selection is often responsible for the maintenance of conserved traits across species. While new mutations can introduce variation, stabilising selection removes it, maintaining the status quo. This mode of selection is crucial for understanding why many traits do not change rapidly despite the presence of genetic variation. It highlights that evolution is not always about change; sometimes it is about maintaining an optimal state.
Disruptive Selection: Maximising Variance
Disruptive selection, or diversifying selection, occurs when individuals at both extremes of the phenotypic distribution have higher fitness than those in the middle. The fitness function is convex, with minima at the intermediate values and maxima at the extremes. This mode of selection increases the variance of the trait. The mean \( \bar{x} \) may remain constant, but the distribution becomes bimodal, with two peaks at the extremes. This can lead to the formation of distinct morphs or ecotypes within a population. If the two groups do not interbreed, this can lead to sympatric speciation.
A well-known example is the seed-eating finches in the Galápagos Islands. In years when both small and large seeds are abundant, but medium-sized seeds are scarce, finches with small beaks are efficient at eating small seeds, and finches with large beaks are efficient at eating large seeds. Finches with medium-sized beaks are inefficient at both. Consequently, the small-beaked and large-beaked finches have higher fitness. Over time, the population may split into two groups: one with small beaks and one with large beaks. This is disruptive selection. Another example is the colouration of peppered moths in environments with both light and dark trees. If the environment is heterogeneous, with patches of light and dark bark, then both light-coloured and dark-coloured moths may have an advantage over intermediate colours, depending on the specific microhabitat. However, in the classic case of industrial melanism, the environment was homogeneous (dark trees), leading to directional selection for dark colouration. Disruptive selection requires a heterogeneous environment or a trade-off that favours extremes.
The evolutionary consequence of disruptive selection is an increase in genetic diversity. For a polygenic trait, selection against intermediates keeps the extreme genotypes common. For a single locus, selection against heterozygotes alone does not keep both alleles stably: whichever allele is commoner tends to fix. Lasting polymorphism usually needs disruptive selection combined with frequency dependence, where each type does best when it is rare (the black-bellied seedcracker, with large- and small-billed morphs feeding on hard and soft seeds, is a well-studied case). This is in contrast to stabilising selection, which tends to eliminate variation. Disruptive selection is a key mechanism in the origin of species, as it can drive the divergence of populations within the same geographic area. It is also important in understanding the maintenance of genetic variation in natural populations. Where it is frequency-dependent, it prevents the fixation of a single type and so preserves genetic diversity. This mode of selection is less common than directional or stabilising selection, but it is crucial for understanding the diversity of life.
The concept
Natural selection
Differential survival and reproduction of heritable variation.
A common misconception is that stabilising selection eliminates all genetic variation. In reality, stabilising selection reduces phenotypic variance, but genetic variation can be maintained if there is a balance between selection and mutation, or if the fitness landscape changes over time. Additionally, stabilising selection does not necessarily mean that the mean trait value is optimal in an absolute sense; it is optimal relative to the current environment. If the environment changes, the optimum may shift, and stabilising selection may become directional. C
Worked example
Problem. Lizard tail length is normally distributed with mean 100 mm and standard deviation \(\sigma = 5\ \mathrm{mm}\). Survival follows a Gaussian fitness curve centred on the mean, \(w(x) = \exp\!\left[-(x-100)^2/(2\omega^2)\right]\) with \(\omega = 10\ \mathrm{mm}\) (a measure of how tolerant selection is). Find the standard deviation of tail length among the survivors.
- For a normal distribution multiplied by a Gaussian fitness curve with the same centre, the survivors are still normally distributed with the same mean.
- Their variance is \(\sigma^{*2} = \dfrac{\sigma^2 \omega^2}{\sigma^2 + \omega^2}\).
- Substitute: \(\sigma^2 = 25\ \mathrm{mm^2}\), \(\omega^2 = 100\ \mathrm{mm^2}\), so \(\sigma^{*2} = \dfrac{25 \times 100}{125} = 20\ \mathrm{mm^2}\).
- Standard deviation: \(\sigma^* = \sqrt{20} = 4.47\ \mathrm{mm}\), a 20% drop in variance in one bout of selection.
- The mean is unchanged at 100 mm because selection is symmetric about it. Only the heritable part of the variance reduction carries into the offspring, so the change seen in the next generation is smaller than this.
Units check. \(\mathrm{mm^2} \times \mathrm{mm^2} / \mathrm{mm^2} = \mathrm{mm^2}\) for the variance, and its square root is in millimetres.
Three quick tests
- Which mode of selection results in a shift of the mean trait value?
Answer
Directional selection. - What is the effect of stabilising selection on the variance of a trait?
Answer
It reduces the variance, narrowing the distribution around the mean. - Disruptive selection can lead to which evolutionary outcome?
Answer
Polymorphism or speciation, as it favours extremes and can split the population into distinct groups.
Back to BU-104 →
Every lecture and every concept in this unit, with prerequisites and threads.
Lecture notes checked line by line by the editor; every number in the worked example was recomputed.
Concepts cited: T-017 · Threads: evolution