Genetic drift

If everyone in a population lived a long life, mated, and reproduced absolutely equally (two offspring per person), then the population size would never change. There would always be approximately the same number of individuals, allowing for variations in when people are born or die. In this population, every gene has an equal chance of being passed into the next generation. Natural selection depends on differences in the chance that genes will survive and reproduce, so this population would not evolve by natural selection.

But the population would still evolve by random chance. A single chromosome can illustrate this potential for evolution. The Y chromosome determines whether humans will be male or female: males have one X chromosome and one Y, females have no Y and two X chromosomes. Mendelian genetics predicts that if a father has two offspring, each of these children has a 50 percent chance of inheriting his Y chromosome and thereby being a son. But these odds mean that the man has a substantial chance of having no sons at all — 25 percent of the time, both children will be daughters. If the man has no sons, then his Y chromosome is simply lost from the next generation. Genes disappear due to chance, even if everyone mates and reproduces equally.

Genetic drift is a random change in allele frequencies.

These random changes in allele frequency can accumulate over time. Across many generations, the frequency of an allele can gradually increase, gradually decrease, or fluctuate back and forth. In other words, the frequencies of different alleles seem to ``drift’’ up and down, without any direction. This is why the random change in allele frequencies is called \term{genetic drift}. Over time, genetic drift can make once rare alleles common, or eliminate alleles altogether.

Genetic drift is stronger in small populations.

\begin{figure} \centering \includegraphics[width=4in]{genetic_variation_drift.png} \label{fig:genetic_drift} \caption[Genetic variation under genetic drift]{Genetic variation under genetic drift as a function of population size. The expected amount of genetic variation increases as a linear function of the size of the population, when genetic drift and mutation are the only causes of evolution. Larger populations are more variable; smaller populations are less variable. } \end{figure}

The most obvious factor affecting the rate of genetic drift is the size of the population. If the population is small, then a small sample is taken of the gametic population in every generation. Small samples can vary more markedly from the larger sets from which they are selected than larger samples, so genetic drift is more powerful in smaller populations. For example, in a population of five individuals, an allele that exists in a single copy in one individual has a frequency of ten percent. Nevertheless, this allele is in constant jeopardy of being eliminated from the population, requiring only the chance of not being passed on once to never again be found. Likewise, it is very possible that in a very few generations this allele might increase from one copy to ten, eliminating all other alleles. In contrast, in a population of a thousand individuals, an allele with a frequency of ten percent exists in 200 copies. While random sampling of gametes will cause this number to fluctuate over time, it is extremely unlikely that chance alone would allow no copy of this allele to be passed on in any given generation. Indeed, it would likely take many hundreds of generations for random events to either eliminate this allele or all the others.