Genetic drift is a mechanism of evolution that acts in concert with natural selection to change species characteristics over time. Like selection, it acts on populations, altering which traits and which alleles predominate among members and changing the diversity of the group. Unlike selection, however, drift occurs only in small populations and results in changes that need not be adaptive. A statistical effect, it arises from the role of chance in the production of offspring.

Reproduction is affected by chance because no trait dictates exactly to what age an individual survives or how many offspring he, she or it produces. Even when an organism inherits "perfect genes"--that is, all the alleles most associated with success or fitness--still an individual may be buried in an avalanche, frozen by a frost or caught in an awkward moment by a foe. By the time their moment comes, some individuals produce more offspring and some produce fewer. As a result, at best a trait confers a superior average reproductivity to the group of individuals that carry it.

Because individuals vary, however, group averages may vary too, especially groups with few members. When reproductivity varies, this affects the transmission of traits from one generation to the next. The prevalence of traits then rises and falls as one generation of carriers reproduces unusually well and the next reproduces an average number or worse. Over successive generations, chance trends in the deviations from the average are called "drift." Persistent drift causes an allele to either disappear from the gene pool or to supplant all other copies of a gene. This is one of the risks that population bottlenecks pose to genetic diversity.

Table of contents
1 Adaptive, neutral, and deleterious traits
2 Law of large numbers
3 Sex cells and chance
4 Population genetics perspective
5 When drift comes into play
6 Other sources of chance and variability
7 Examples
8 See also

Adaptive, neutral, and deleterious traits

Drift and natural selection are capable of acting in parallel, even on the same traits. At the same time as a disease, a predator or another selective pressure is predisposing individuals with adaptive traits to a higher rate of reproduction than their peers, runs of luck may enhance or oppose this tendency. Even the fittest will vary in reproductivity from one generation to the next, and whenever an advantageous allele diminishes in prevalence, it's a neutral or even deleterious allele that takes up the slack.

The ability to affect the prevalence of even neutral traits distinguishes drift from selection. As a result, theorists often appeal to drift to account for evolutionary changes that might have paved a way for adaptations, and yet offered no obvious immediate advantage. Yet while aiding in explanation, drift also creates a challenging ambiguity. How does one know whether a trait shared by every member of a species--yellow spots, for example--represents an adaptation to selective conditions, or instead represents just an accident? Not only can it be difficult to say, the answer may be "a little bit of both."

Teasing apart the relative influence that drift and selection have had over the course of evolution, both with respect to individual species as well as with regard to the history of life in general, is a primary aim of evolutionary biology.

Law of large numbers

Small populations' greater susceptibility to drift is a manifestation of the law of large numbers, which is especially easy to see in tossing coins. On average, of course, coins turn up heads or tails equally. Yet just a few tosses in a row are unlikely to produce heads and tails in equal number. The numbers are no more likely to be exactly equal for a large number of tosses in a row, but the inequality can be very small in percentage terms. As an example, ten tosses turn up 70% heads about once in every six tries. The chance of the same imbalance from a hundred tosses in a row is only about one in 25,000. Big deviations from average reproductivity, similarly, happen more frequently in small populations than in large ones. As a result, the smaller the population, the faster they drift.

Sex cells and chance

Beyond its influence on sheer numbers of offspring, chance affects the reproduction of sexual species in yet another way. Sperm and eggs each contain only half the amount of genetic material an offspring inherits at conception. The material in each half is selected somewhat haphazardly from the complete set of material that each parent carries. This random selection occurs during crossing over of the chromosomes when sex cells are produced.

In a narrow sense the random aspect is inconsequential, because every individual produces millions of sperm or eggs. If a diploid individual carries two versions of a gene, then each allele will be present in very nearly exactly half of their sex cells--because the number of cells is so large. But individuals produce few offspring, and the number produced by a group of genetically alike individuals may not be close to a million either. As a result, sexual reproduction can lead to a disproportionate transmission of alleles from one generation to the next, especially when the offspring population is small.

Population genetics perspective

From the statistical perspective of population genetics, drift is a "sampling effect". A chance over-production or under-production of offspring compared to the average represents what statisticians call a sampling error. According to this perspective, the frequency distribution of alleles among a population of offspring (how many carriers there are of each allele) reflects a sampling of the alleles of the preceding generation. When the alleles of a gene do not differ with regard to fitness, on average the number of carrieres in one generation is proportional to the number of carriers in the last. But the average is never tallied, because each generation parents the next one only once. Therefore the frequency of an allele among the offspring often differs from its frequency in the preceding generation.

Many sources of mortality, such as infectious diseases for which no immunity exists, may be regarded as randomly sampling a population. In other words, they randomly select some proportion of individuals for death. Because the sample is random, on average alleles are picked in proportion to how common they are. But because the sample size, the population size and the number of carriers of an allele are finite, deviations from the average or mean often occur. To the extent that the upward and downward deviations over successive generations do not exactly balance out, an allele drifts.

Drifting alleles are liable to disappear all together from the gene pool. When the number of individuals who carry an allele drifts to zero, so that no individuals are left to reproduce it, it disappears forever. Similarly, if all but one of the alleles for a given gene disappears, the proportion of individuals who carry it will never stray from 100%. That is, until in at least one individual a spontaneous mutation or other genetic change affects that carrier's allele. It is also possible in principle for such a change to reintroduce an allele that has disappeared from the gene pool.

When drift comes into play

Population bottlenecks, Founder's effect etc.

Other sources of chance and variability

The principle of independent assortment may also be involved in drift. According to this principle, during gamete formation many traits combine randomly. Thus, an individual may inherit alleles that increase fitness along with alleles that are neutral (that neither increase nor decrease fitness). Natural selection favors the alleles that increase fitness, but the associated neutral alleles will also increase in frequency, as an accidental byproduct.

Examples

Chance acts on allele frequency in a variety of ways.  Perhaps the most obvious input is lifespan.  For example, imagine a collision between a car and a bus.  If the collision was caused by the fact that the driver of the car had poor vision, that driver's death might be an example of 
natural selection. But for the driver and passengers of the bus, death was random. If these people died before reproducing, their death would alter the frequency of their genes in the subsequent generation. In other words, even when individuals are equally fit, they will differ in their success. Simply by being in the wrong place at the wrong time, the death of some and the survival of others can change the distribution of alleles in a population, and thus be a force in its evolution. "Differential morbidity" is the most important cause of drift in populations of asexual (or "clonal") organisms, and it is an important cause of drift in populations of sexual organisms as well.

See also