Artificial Selection: Study Guide (Biology 7.3)

Humans as the selecting force

In Topic 7.2 you saw that the environment “selects”: individuals whose heritable traits suit the conditions leave more offspring, so those traits become more common. Artificial selection uses the same logic, but a person makes the choice. A farmer keeps seed only from the highest-yielding plants. A breeder mates only the calmest animals. Over many generations the chosen traits become more common, and the population changes.

Artificial selection needs the same three conditions as natural selection:

1. Variation. Individuals in the population differ in the trait.
2. Heritability. At least part of that variation is caused by genes, so offspring tend to resemble their parents.
3. Differential reproduction. Some individuals have more offspring than others. Here, humans decide who breeds.

	Natural selection	Artificial selection
What decides who reproduces	the biotic and abiotic environment	human choice
Which traits increase	traits that raise survival and reproduction in that environment	traits humans want, even if they lower survival in the wild
Speed	often slow	often fast, because breeders can be very strict
Raw material	existing heritable variation	existing heritable variation

The last row is the key. Breeders cannot order a new trait. They can only choose among the variations that are already there. New alleles come from mutation, and breeders can also bring in alleles by crossing their population with another one.

What selection does to a population

Selection changes a population in two ways at once.

The mean shifts. If breeders keep only the largest individuals, the next generation’s mean size rises, as long as size is partly heritable. Repeated over generations, the mean keeps moving in the chosen direction.

Variation within the population usually falls. Alleles that produce unwanted phenotypes are removed, because their carriers do not breed. Breeders also tend to use few parents, so many alleles are lost simply because they were not among the parents chosen. The population becomes more uniform. A field of a modern crop variety can be almost genetically identical, plant to plant.

Variation between populations grows. Different breeders choose different traits. Dog breeds, all descended from wolves, now differ hugely in size, coat and behaviour, yet each breed is fairly uniform inside itself. Humans have therefore increased diversity between breeds while decreasing diversity within each breed.

One wild plant, many vegetables

Wild cabbage (Brassica oleracea) is a classic case. Over centuries, growers in different places selected different parts of the same species:

Part selected	Vegetable produced
large leaves	kale
tight terminal bud	cabbage
side (lateral) buds	Brussels sprouts
swollen stem	kohlrabi
immature flower clusters	broccoli and cauliflower

All of these are still one species. They look different because people kept different heritable variations in different lines. Maize is another example: it was domesticated from a wild grass called teosinte in Mexico about 9,000 years ago, and it now looks very different from its wild ancestor.

Selection without meaning to

Humans also select other species by accident. Spraying an insecticide kills susceptible insects, so resistant ones leave more offspring. Using antibiotics does the same to bacteria. Heavy fishing that removes the largest fish favours fish that mature young and small. Strictly, these are cases of natural selection in a human-made environment, because no one chose the winners on purpose. Either way, human activity changes the variation in another species.

Representing selection data

A selection experiment usually has three groups: a line selected up (only the largest, fastest or highest-yielding individuals breed), a line selected down, and an unselected control line (parents chosen at random). The control shows what happens without selection, for example because of changes in food, temperature or measurement over time.

The graph shows the mean, but selection also changes the spread. Figure 2 compares the starting population with the high line after five generations.

Reading Figure 2. In generation 0, only 6 of 100 beetles (6%) were 2.8 mg or heavier. In the high line at generation 5, 45 of 100 (45%) were. The beetles now occupy five mass classes instead of seven. The whole distribution has shifted right and narrowed. That narrowing is a sign of lost variation.

Worked example 1: describing a selection experiment

Question. Use Figure 1 (data below) to (a) describe the trend in each line, (b) calculate the percentage change in each selected line, and (c) explain why the control line is needed.

Generation	0	1	2	3	4	5
High line / mg	2.40	2.48	2.57	2.63	2.72	2.80
Control / mg	2.40	2.41	2.39	2.42	2.40	2.41
Low line / mg	2.40	2.33	2.27	2.20	2.15	2.08

(a) Describe the trends (use numbers).

1. High line: mean mass rises every generation, from 2.40 mg to 2.80 mg. The rise each generation is fairly steady, between 0.06 and 0.09 mg.
2. Low line: mean mass falls every generation, from 2.40 mg to 2.08 mg, by 0.05 to 0.07 mg each time.
3. Control: mean mass stays between 2.39 and 2.42 mg, with no clear trend.
4. Relationship: the lines diverge; by generation 5 the high line is 0.72 mg heavier than the low line.

(b) Percentage changes.

1. High line: (2.80 − 2.40) ÷ 2.40 × 100 = 0.40 ÷ 2.40 × 100 = +16.7%.
2. Low line: (2.08 − 2.40) ÷ 2.40 × 100 = −0.32 ÷ 2.40 × 100 = −13.3%.
3. Mean change per generation: high line 0.40 ÷ 5 = 0.08 mg; low line −0.32 ÷ 5 = −0.064 mg.

(c) Why a control? The control experiences the same food, temperature and handling but no selection. Because its mean hardly changes (+0.01 mg), the large changes in the selected lines are best explained by selection, not by a change in the conditions. Without the control, a better diet in later generations could also explain a rise in mass.

Interpretation. The response shows that pupal mass is partly heritable: choosing heavy parents produced heavier offspring. Note the limits. These are fictional means for one population, with no measure of spread for each generation, so you cannot say how large the overlap between lines is. The data support heritability of the trait; they do not show which genes are involved.

Worked example 2: selection and crop diversity

Question. Researchers counted the mean number of different alleles per gene at 20 genes in three groups of a fictional grain crop. They also planted two fields next to a source of a fungal disease.

Group	Mean number of alleles per gene
Wild relatives	8.2
Traditional (landrace) varieties	5.6
Modern commercial variety	2.1

Field	Planting	% plants infected (3 plots)	Mean
A	one modern variety	81, 76, 77	78%
B	mixture of six landraces	19, 24, 23	22%

(a) Describe the pattern in allele numbers. (b) Explain how artificial selection produced it. (c) Use the infection data to explain a risk of this pattern.

(a) Allele number falls from wild relatives (8.2) to landraces (5.6) to the modern variety (2.1). The modern variety has (8.2 − 2.1) ÷ 8.2 × 100 = 74.4% fewer alleles per gene than the wild relatives, about one quarter as many (the wild relatives have 8.2 ÷ 2.1 ≈ 3.9 times as many).

(b) Each round of selection kept only plants with the wanted traits, so alleles carried only by rejected plants were lost. Early farmers took a small sample of wild plants, and modern breeders often start from a few elite parents, so many alleles were never passed on at all. The more intense and longer the selection, the fewer alleles remain.

(c) Field A was infected about 3.5 times as much as field B (78% against 22%), and the plot ranges do not overlap (76–81% against 19–24%). In a uniform field, if one plant is susceptible, nearly all are, so the fungus spreads from plant to plant. In the mixed field, some plants carry alleles for resistance; they survive and also slow the spread. The data are fictional and from one season, so they support the idea rather than prove it.

A real parallel. Bananas sold for export are mostly one clone, propagated without sex, so the plants are genetically identical. In the 1950s a fungal wilt (Panama disease) wiped out much of the earlier export variety, Gros Michel, in Central America, and growers switched to the Cavendish type. A newer strain of the same fungus now threatens Cavendish plantations. Breeders respond by crossing crops with wild relatives and by storing seed from many varieties, to restore variation that selection removed.

Common misconceptions

• “Breeders create new traits.” Selection only sorts variation that already exists. New alleles arise by mutation, or are brought in by crossing.
• “Artificial selection is genetic engineering.” Selection chooses whole organisms as parents; genetic engineering changes DNA directly.
• “A trait improved by training or feeding will be passed on.” Only heritable variation responds to selection. A well-fed animal’s larger size is not inherited if its genes are the same.
• “Selection always increases diversity, because we get many breeds.” It increases differences between breeds but usually reduces variation within each one.
• “Selected traits help the organism survive.” Many selected traits (huge seed heads, very short legs) would lower fitness in the wild. Humans, not the environment, decide what is “fit”.
• “Selection can continue for ever at the same rate.” As useful alleles become fixed and others are lost, the response slows and can stop (a plateau) unless new variation enters.
• “If two lines differ after selection, the change must be due to selection.” Only a comparison with an unselected control rules out changes in the conditions.

Where this leads

Artificial selection shows natural selection at work on a timescale you can measure. Next, Topic 7.4, Population Genetics, adds the random processes (mutation, genetic drift and gene flow) that also change allele frequencies. Test yourself with the practice questions, then use the revision notes and the checklist to consolidate.