Genetic drift

• 1. Introduction

  ›

  Genetic drift is a force of change that acts on the living world through pure chance, altering the genetic makeup of populations without any push from survival or reproduction. Picture a jar holding twenty marbles, half red and half blue. Each marble stands for an organism, and each color stands for a different version of a single gene. When the next generation is built by randomly drawing marbles one at a time, the new jar rarely ends up with exactly ten of each color. Sometimes the red ones dwindle; sometimes the blue ones take over. No selection pressure drove that outcome. Random chance did.

  That simple image captures something profound about how life changes over time. For much of the twentieth century, biologists debated whether this kind of undirected change was a significant player in evolution at all, or merely a footnote to the grand machinery of natural selection. The answer turns out to depend heavily on how many individuals a population contains, how genes connect to one another across a chromosome, and what happens when a catastrophe reduces a thriving species to a handful of survivors. Those questions will carry us from the mathematics of coin-flips to an atoll in Micronesia, from the prairie chickens of Illinois to the founding families of an Amish community in Pennsylvania, and into a scientific argument that has never fully quieted.

• 2. The Mathematics of Chance

  ›

  Consider a large colony of bacteria isolated in a drop of solution, genetically identical except for a single gene carrying two neutral alleles, labeled A and B. Suppose half the bacteria carry A and the other half carry B, giving each allele a frequency of exactly one-half. When the drop shrinks until only four bacteria can be fed, every other individual dies without reproducing. Among those four survivors, sixteen possible allele combinations exist, and each is equally likely to appear with a probability of one in sixteen.

  Counting those combinations by how many carry A versus B reveals something counterintuitive. Only six of the sixteen combinations preserve equal numbers of A and B. The other ten produce unequal counts. The probability of unequal representation is ten in sixteen, meaning unequal outcome is actually more likely than equal outcome. That shift in frequency due to random survival is genetic drift in action, and the sudden collapse to four survivors is a population bottleneck.

  The Wright-Fisher model, named after Sewall Wright and Ronald Fisher, provides one formal framework for tracking these shifts across generations. It assumes that generations do not overlap and that each copy of a gene in the new generation is drawn independently at random from all copies in the old one. An alternative, the Moran model, assumes overlapping generations: at each step, one individual is chosen to reproduce and one is chosen to die. In practice the two models produce qualitatively similar results, but drift runs twice as fast in the Moran model. The Wright-Fisher model is more convenient for computer simulation because fewer time steps need to be calculated, while the Moran model's tridiagonal transition matrix makes some analytical problems easier to solve.

• 3. Fixation and Loss

  ›

  When an allele reaches a frequency of 100 percent it is said to be fixed in the population; when it falls to zero percent it is lost. The Hardy-Weinberg principle establishes that without any disturbance from evolutionary forces, allele frequencies in a diploid population approximate a binomial distribution and hold steady across generations. Genetic drift is one mechanism that breaks that stability, alongside migration, mutation, and natural selection.

  Smaller populations achieve fixation far faster than large ones. In the limit of an infinitely large population, fixation never occurs through drift alone. At any given moment, the probability that a particular allele will eventually become fixed is simply its current frequency: if allele A sits at 75 percent and allele B at 25 percent, there is a 75 percent chance A ultimately takes over and a 25 percent chance that B does.

  The expected number of generations before fixation occurs is proportional to the effective population size, a figure that can differ substantially from the raw count of individuals. The effective population size accounts for inbreeding levels, whichever stage of the life cycle has the fewest individuals, and the fact that some neutral genes are linked to others that are under selection. Once an allele is fixed, drift stops operating on it entirely. A new allele can only re-enter the population through mutation or gene flow, meaning that even though drift is a random, directionless process, it consistently pushes populations toward genetic uniformity over time.

  When mutation is also present, a more complex picture emerges. If an allele is lost by mutation more often than it is gained, both drift and mutation shape the time to its disappearance. When such an allele begins fully fixed and is lost by mutation at rate m per replication, the expected time until loss depends on Euler's constant in the approximating formula, and the dominant term is mutation, not the effective population size.

• 4. Drift Versus Selection

  ›

  Natural selection has a direction: it drives populations toward heritable adaptations suited to their current environment. Genetic drift has no direction at all. Selection favors alleles whose effects increase survival or reproduction, suppresses alleles tied to unfavorable traits, and ignores neutral ones. Drift operates indifferently on all three categories, shifting allele frequencies purely through the mathematics of chance.

  The law of large numbers predicts that in small populations the magnitude of drift per generation is large. Drift can overwhelm selection whenever the selection coefficient is less than one divided by the effective population size. Non-adaptive evolution from the combined action of mutation and drift is therefore considered a consequential mechanism primarily within small, isolated populations.

  Genetic linkage complicates the picture further. When a neutral allele sits near a gene under active selection on the same chromosome, selection on that neighboring gene can reduce the effective population size experienced by the neutral one, amplifying drift's influence. Higher recombination rates weaken this linkage effect. The fingerprint of that dynamic is visible in molecular data as a correlation between local recombination rate and genetic diversity, alongside a negative correlation between gene density and diversity at noncoding DNA regions.

  Low allele frequency itself makes any allele more vulnerable to random elimination. Even a new advantageous mutation is nearly as likely to be lost by drift as any neutral mutation, until its frequency climbs to a threshold where drift's influence fades. This phenomenon has a named analogue called genetic draft, the effect on a locus caused by selection on linked loci, which is distinct from genetic drift because the direction of the random change it produces is autocorrelated across generations, unlike the sampling error at the heart of drift.

• 5. Bottlenecks and Founders

  ›

  A population bottleneck occurs when a group contracts sharply in size over a short period due to some random environmental event. In a true bottleneck, survival is purely random; no genetic advantage improves any individual's odds. The aftermath can radically reshape allele frequencies, independent of selection entirely.

  Pingelap atoll in Micronesia provides a striking example. The island's population experienced a bottleneck that produced a relatively high proportion of individuals with total rod-cell color blindness, a condition called achromatopsia. After any bottleneck, inbreeding rises, and with it the damage inflicted by recessive harmful mutations, a process called inbreeding depression. The worst of those mutations are selected against, pulling other alleles linked to them out of the population through background selection. A process called genetic purging can enhance that effect for recessive harmful mutations. Reduced genetic diversity then raises the odds of further allele fluctuations from drift in later generations.

  Over-hunting drove a severe bottleneck in the northern elephant seal during the nineteenth century. The resulting decline in their genetic variation can be estimated by comparing it to that of the southern elephant seal, which was not hunted as aggressively.

  In Illinois, greater prairie chicken numbers fell from roughly 100 million birds in 1900 to about 50 birds in the 1990s because of hunting and habitat destruction. DNA analysis comparing birds from the mid-century to birds in the 1990s documents a steep decline in genetic variation across just those few decades. The species currently experiences low reproductive success.

  The founder effect is a specific variety of bottleneck, produced when a small group splinters away from the original population and starts a new colony. The random allele sample carried by those founders can grossly misrepresent the source population. A well-documented case involves the Amish migration to Pennsylvania in 1744. Two members of that founding colony shared the recessive allele for Ellis-Van Creveld syndrome. Because members of the colony and their descendants have remained religious isolates with relatively little intermarriage with outsiders, Ellis-Van Creveld syndrome is now far more prevalent among the Amish than in the wider population.

  Sewall Wright was the first to argue that random drift in small, newly isolated populations is a significant driver of new species formation, a position he embedded in his shifting balance theory of speciation. Ernst Mayr later built on Wright's framework to argue that the genetic narrowing following a founder event was critically important for speciation. That view commands considerably less support today; repeated experimental tests have produced equivocal results at best.

• 6. The Debate That Would Not End

  ›

  Arend L. Hagedoorn and Anna Cornelia Hagedoorn-Vorstheuvel La Brand first outlined the role of random chance in evolution in 1921, highlighting how random survival drives the loss of variation from populations. Ronald Fisher responded the following year with the first mathematical treatment of what he called the Hagedoorn effect, though that treatment was later shown to be marginally incorrect. Fisher expected that many natural populations were too large, with an effective size on the order of 10,000, for drift to matter substantially.

  Sewall Wright provided the corrected mathematics and coined the term genetic drift itself. His first use of the word drift appeared in 1929, though at that point he used it to describe a directed process, closer in meaning to natural selection. The concept of random drift through sampling error eventually came to be called the Sewall-Wright effect, though Wright was never entirely comfortable with that attribution. He himself distinguished between what he called steady drift, meaning selection, and random drift, meaning sampling error, and he worried that restricting the word drift to sampling error alone would cause confusion.

  Ronald Fisher became the most vocal critic of the view that drift played any major role in evolution. Fisher conceded drift had some effect but considered it insignificant, and he has been accused of misreading Wright's position as a near-total rejection of selection. Fisher saw evolution as a long, steady, adaptive progression, the only framework he found adequate to explain the growing complexity of life from simpler ancestors. The argument between gradualists aligned with Fisher and those who followed Wright's model has never fully resolved.

  In 1968, population geneticist Motoo Kimura reignited the debate with his neutral theory of molecular evolution. Kimura argued that most instances in which a genetic change spreads across a population are caused by genetic drift acting on neutral mutations, rather than by selection. More recently, John H. Gillespie and William B. Provine have criticized even that framing, arguing that selection on linked sites, the genetic draft mechanism, is a more important stochastic force than sampling error. Emerging in the 1990s, a framework called constructive neutral evolution extended the conversation further, using genetic drift to theorize how complex biological systems can arise through neutral transitions without any selective pressure pushing them forward.