Genetic Drift

Historical Context & Motivation

For much of the early twentieth century, evolutionary biology was dominated by the idea that natural selection was the primary—if not sole—driver of changes in allele frequencies over time. Charles Darwin's theory, bolstered by the mathematical models of the Modern Synthesis, suggested that traits spread through populations because they improved an organism's fitness. But a growing body of observations hinted at a complementary force: the role of pure chance in shaping populations, especially small ones.

The central question that drift addresses is deceptively simple: Can allele frequencies change across generations even when no allele has a survival or reproductive advantage? The answer, as Sewall Wright first formalized, is a resounding yes—and the consequences are profound for how we understand biodiversity, conservation biology, and the molecular clock.

Core Principles & Definitions

Genetic drift is the change in allele frequency in a population due to random sampling of organisms. Unlike natural selection, which is deterministic and driven by differences in fitness, drift is inherently stochastic. It arises because the alleles present in a new generation are a random sample of the alleles from the parental generation—and any random sample is unlikely to perfectly represent the original.

Visual Explanation — Drift Over Generations

The following diagram illustrates how the frequency of a single allele (allele A) changes across 10 generations in three populations of different sizes. Each line represents an independent replicate population, all starting at an allele frequency of 0.5. Notice how the smaller population (N = 20) shows wild fluctuations and rapid fixation or loss, while the larger population (N = 500) remains relatively stable near the starting frequency.

In the diagram above, the pink lines represent two replicate populations of N = 20. One rapidly drifts to fixation (allele frequency reaches 1.0) and the other to loss (frequency reaches 0.0)—both within about 10 generations. The violet lines (N = 100) show moderate fluctuation, while the cyan lines (N = 500) hover near the starting frequency of 0.5 with minimal deviation. This visually captures the core relationship: smaller populations are more susceptible to drift.

Mathematical Framework

The Wright–Fisher model is the classical mathematical framework for genetic drift. It imagines an idealized diploid population of constant size N in which each generation is formed by randomly sampling 2N alleles (with replacement) from the parental gene pool. From this simple model, several powerful results emerge.

This equation reveals two key insights. First, the variance is greatest when p = 0.5 (the allele is at intermediate frequency), meaning drift is most "noisy" for alleles that aren't yet rare or common. Second, the variance is inversely proportional to 2N, so doubling the population size halves the randomness. In a population of N = 10, the standard deviation of Δp when p = 0.5 is √(0.25/20) ≈ 0.112—a massive 11% fluctuation per generation.

This elegantly simple result means that a new neutral mutation appearing as a single copy in a population of 500 diploid individuals has a fixation probability of just 1/1000 = 0.001. The vast majority of neutral mutations are lost by drift. However, because many new mutations arise every generation, the rate of neutral substitution (fixation of new mutations over evolutionary time) equals the mutation rate μ, independent of population size—a cornerstone of the molecular clock.

Together, these equations form the quantitative backbone of genetic drift theory. They allow us to predict how quickly populations lose genetic variation, how likely an allele is to become fixed, and how long fixation takes—all as functions of the effective population size N_e, which is often much smaller than the census population size due to factors such as unequal sex ratios, variance in reproductive success, and population fluctuations.

Types of Genetic Drift — Bottlenecks & Founder Effects

While drift operates continuously in all finite populations, two special scenarios dramatically amplify its effects by temporarily or permanently reducing population size. These are the population bottleneck and the founder effect, and they produce some of the most striking real-world examples of drift.

A population bottleneck occurs when a population's size is drastically reduced by a catastrophic event—a disease, natural disaster, or habitat destruction—and only a small subset of individuals survives to reproduce. The survivors carry only a fraction of the original genetic diversity. A classic example is the northern elephant seal, hunted to as few as ~20 individuals in the 1890s; despite recovering to over 100,000 today, the species has almost no genetic variation at many loci.

The founder effect occurs when a small group of individuals colonizes a new habitat, carrying only a subset of the genetic variation of the source population. The Amish populations of Pennsylvania, founded by a few hundred German immigrants in the 18th century, exhibit unusually high frequencies of certain rare genetic disorders like Ellis–van Creveld syndrome—conditions that happened to be present in the founding group and were amplified through drift and limited gene flow.

Worked Example

Let us work through a quantitative problem that illustrates the interplay between population size, drift, and heterozygosity loss.

Drift vs. Natural Selection — Strengths & Limitations

Genetic drift and natural selection are the two primary forces that change allele frequencies in populations, but they operate through fundamentally different mechanisms. Understanding when each dominates is essential to modern evolutionary biology.

The critical threshold for determining whether drift or selection dominates the fate of an allele is expressed by the product Nₑ × s, where s is the selection coefficient. When |Nₑs| ≪ 1 (roughly less than 1), the allele behaves as if it were effectively neutral, and its fate is governed primarily by drift. When |Nₑs| ≫ 1, selection is the dominant force. This explains why even a mildly beneficial mutation (say, s = 0.001) can be effectively invisible to selection in a population of Nₑ = 100, but strongly selected in a population of Nₑ = 100,000.

Connection to Advanced Theory

The Wright–Fisher model of genetic drift, while powerful, is only the beginning. Several advanced frameworks build upon its foundations to address more realistic biological scenarios.

Motoo Kimura's Neutral Theory (1968) extended drift to the molecular level, proposing that most observed sequence differences between species are due to the fixation of selectively neutral mutations via drift. This theory provides the theoretical basis for the molecular clock—the observation that neutral substitutions accumulate at a roughly constant rate over time, allowing biologists to estimate divergence times between lineages.

Tomoko Ohta's Nearly Neutral Theory refined this by recognizing that many mutations are not strictly neutral but have selection coefficients so small (|s| ≈ 1/N_e) that drift overwhelms selection. This predicts that organisms with smaller effective population sizes will accumulate more slightly deleterious mutations—a prediction confirmed by genomic comparisons across species.

In modern population genomics, coalescent theory provides a retrospective framework for understanding drift. Rather than simulating allele frequencies forward in time, coalescent models trace alleles backward to their most recent common ancestor. The expected coalescence time for two randomly chosen alleles is 2Nₑ generations in a diploid population, directly linking observed patterns of genetic variation to population history. This approach underpins methods like PSMC (pairwise sequentially Markovian coalescent) that reconstruct historical population sizes from single genome sequences—revealing ancient bottlenecks and expansions encoded in our DNA.

Practice Problems

Lesson Summary