Variation and inheritance

In one study of people in Jordan, 26.1 percent of the ABO blood-group genes carried the A allele; a change in a number like that is what biologists mean by evolution.

Natural selection can only work on differences between individuals, and only on differences that are inherited. A child who grows tall because of good food will not have taller children for that reason. A muscle built in the gym does not pass to the next generation either. Darwin and Wallace knew that much variation was inherited, but neither knew how. The answer was already in print: Gregor Mendel published his pea experiments in 1866, a few years after the Origin, but almost nobody read them until 1900. By the 1940s biologists had joined Mendel's genetics to Darwin's selection, a combination known as the modern synthesis.

Where new variation comes from

There are two main sources. Mutation, a change in DNA, is the only source of brand new alleles. A mutation can lower an organism's chance of surviving or reproducing, raise it, or make no difference at all; the last kind is called neutral, and many mutations are neutral. Sexual reproduction is the second source. It does not make new alleles, but every egg and sperm carries a different mix of the parent's alleles, so each offspring gets a new combination. There is a catch for animals: a mutation in a skin cell or a liver cell stays in that body. Only changes in the cells that make eggs or sperm can reach the next generation.

The lab below starts with 100 made-up beetles that are all the middle shade, and birds that ignore colour (selection pressure 0). Press Next generation a few times. Mutation makes a few lighter and darker beetles, and with no selection their shares wander up and down by chance from one generation to the next. Then raise the selection pressure and watch the bars start to lean towards the bark colour.

Counting alleles in a population

Population genetics looks at a whole population rather than one family. The allele frequency is the share of all copies of a gene in the population that are one particular allele. OpenStax gives a real example: a study in Jordan found the ABO blood-group alleles at 26.1 percent for A, 13.4 percent for B and 60.5 percent for O, which add up to 100 percent. In population genetics, evolution means a change in allele frequencies over generations. Natural selection is one way that happens. Another is genetic drift, a change by pure chance, which is strongest in small populations. When a few individuals start a new population, for example on an island, their alleles may not match the original population, a special case called the founder effect.

A population that does not change

In the early 1900s the mathematician Godfrey Hardy and the physician Wilhelm Weinberg described what a population looks like when nothing is changing it. Call the frequency of one allele p and the other q, so p + q = 1. If mates pair up at random, the genotypes come out as p² for the homozygous dominant, 2pq for the heterozygotes and q² for the homozygous recessive, and p² + 2pq + q² = 1. The model assumes a huge population with no mutation, no migration and no selection. No real population meets that, which is the point: if the counts in the field differ from the prediction, something is acting on the population. In OpenStax's example, a population of 500 plants has p = 0.8 for violet V and q = 0.2 for white v. The model predicts 0.64 × 500 = 320 VV, 0.32 × 500 = 160 Vv and 0.04 × 500 = 20 vv plants.

Fast-changing populations make this practical. Flu viruses copy themselves quickly and mutate often, so the strains going around shift from year to year. Each year scientists study how the strains have been changing and pick the ones to put in the next vaccine, which is why a flu shot from last year may not protect well this year.