AP Biology · Unit 7 · Topic 7.3
Artificial Selection
A grower wants flowers to open earlier in the season. Giving plants warmer conditions might make this year's flowers open sooner. Choosing early-flowering plants as parents might change future generations. Those are different interventions: one changes growing conditions, while the other changes which inherited variants contribute to the next generation.
The choice matters only if it changes reproduction
Artificial selection occurs when human preferences influence which organisms reproduce and how much they contribute to later generations. Breeders might favor fruit quality, flowering time, or an animal's behavior. A sustained inherited response requires variation that can be transmitted; wishing for a trait does not generate the necessary alleles.
For the flower grower, measuring plants and labeling the earliest ones is not yet enough. If all plants contribute equally to breeding, those labels have not imposed selection. If the chosen plants contribute disproportionately, inherited variants associated with early flowering can become better represented in their descendants.
In a sexually reproducing plant, both the seed parent and the pollen parent contribute nuclear alleles. Saving seed from a chosen plant does not automatically establish the identity or traits of its pollen parent. This is why a breeding explanation must specify the sources of both gametes rather than treating the seed parent as a genetic copy machine.
What changes compared with natural selection?
| Question | Artificial selection | Natural selection |
|---|---|---|
| What favors a variant? | Human decisions about breeding or propagation. | Consequences of expressed traits in the prevailing environment. |
| What must be passed on for an evolutionary response? | Heritable differences associated with the selected trait. | Heritable differences associated with reproductive success. |
| Does the process require a useful new mutation on demand? | No; selecting parents does not direct the mutations they need. | No; selection does not anticipate a future need. |
Human preference can favor a feature that would be costly without human care. Yet artificial and natural selection can operate together: a selected parent must still survive and reproduce, and a crop must still function in its growing environment.
Not every human-caused environmental change is artificial selection. If a pollutant changes which wild organisms survive, but people are not selecting breeding stock, that is an environmental selective pressure. The distinction concerns the mechanism, not simply whether humans were involved.
Choosing parents does not guarantee identical offspring
Many breeding targets are quantitative traits influenced by several loci and environmental conditions. An early-flowering plant may carry inherited tendencies toward early flowering, may have occupied a warmer spot, or both. Its offspring inherit alleles, not the parent's exact surroundings or a numerical copy of its flowering date.
Meiosis and fertilization also make new combinations. Even when both parents have the desired phenotype, their offspring can vary. Conversely, breeders can accumulate favorable combinations over repeated generations without introducing a new allele in every generation. Mutation creates new alleles; crossing and selection change how inherited variation is combined and represented.
Crop domestication provides real examples of cumulative breeding change. OpenStax describes maize breeding from teosinte ancestry. The relevant lesson is that repeated differences in propagation can yield substantial changes in a population, not that a single breeder transforms an individual into a different species.
Calculate the choice before interpreting its outcome
Consider an invented example of a compatible, sexually reproducing flowering plant. Ten candidate parents reach first flower at 30, 32, 34, 36, 38, 42, 44, 46, 48, and 50 days after germination under standardized conditions. The grower chooses the three earliest plants. Assume their contributions are balanced in a planned breeding design, with both pollen and seed parents drawn from the selected group.
The candidate mean is the sum of all ten measurements divided by ten. The selected-parent mean uses only the three chosen values:
The selection differential, \(S\), measures how far the selected parents differ from the candidates:
The negative sign records selection toward earlier flowering. It is not an error or evidence that selection failed. But −8 days describes the parents chosen, not an observed change in their offspring. The next generation must be grown and measured before making that claim.
Follow the descendants alongside a control
Suppose the following hypothetical results summarize a breeding study. Generation 0 is the starting candidate population. In each later cycle, the selected line is bred from early-flowering parents; control parents are chosen without regard to flowering time. Both lines use comparable numbers of parents and are grown side by side under matched conditions.
| Generation | Selected line | Unselected control |
|---|---|---|
| 0 | 40.0 | 40.0 |
| 1 | 37.0 | 40.0 |
| 2 | 35.0 | 39.5 |
| 3 | 34.0 | 40.0 |
From generation 0 to 1, the selected-line mean changes by \(37-40=-3\) days, not by the parental differential of −8 days. The controls remain at 40 days, so the generation-1 line difference is also −3 days. The selected-parent mean of 32 days is deliberately not plotted as a generation mean: it describes a subset of generation 0.
By generation 3, the selected line flowers 6 days earlier than the control. Relative to the control mean, that is \(6/40=0.15\), or 15% fewer days to first flower. The early-flowering shift slows from 3 days to 2 days to 1 day per generation. That pattern does not establish its cause. Reduced selectable variation, environmental effects, or chance could contribute; it does not prove a permanent limit has been reached.
What would make the experiment convincing?
One selected line and one control can diverge partly by chance, especially when few parents are used. A stronger hypothetical design uses independent replicate lines, keeps records of parentage and contributions, randomizes growing positions, and measures the same developmental event in both groups. Report the number of plants and the spread of their measurements, not only their means.
If a warmer greenhouse advances both lines by three days, a comparison only with the original generation could misleadingly attribute that change to breeding. A contemporary control helps identify shared environmental shifts. However, common conditions do not automatically eliminate maternal effects, unequal seed provisioning, or all environmental carryover. Following additional generations and comparing families strengthens the inheritance argument.
The graph supports a directional difference between the supplied means. Without replicate variation, it cannot establish statistical significance or identify particular alleles. A larger numerical shift is not a substitute for those missing observations.
More distinct varieties can coexist with less diversity within a line
Humans can develop contrasting varieties by selecting different traits from related starting material. Yet repeatedly breeding from a small subset can lose alleles within each line. Diversity among named varieties and genetic variation within one breeding population are therefore different quantities.
Breeding close relatives increases homozygosity and can expose harmful recessive alleles already present. It does not create those alleles merely because the parents are related. Selection for one target can also carry linked variants along, while a gene affecting several traits can generate an unwanted correlated response.
These risks do not mean all artificial selection inevitably eliminates diversity. Maintaining multiple families, avoiding excessive reliance on a few parents, and preserving diverse seed collections can retain options for future breeding. Genetic diversity is relevant to resilience when pests or growing conditions change, although diversity alone cannot guarantee resistance to every threat.
Define “better” before celebrating the result
Earlier flowering might help a crop mature in a short season, but the grower may also care about yield, reproductive success, or performance during cold weather. A hypothetical early line that flowers six days sooner but produces far fewer viable seeds has improved one measurement while worsening another. The breeding objective determines how that trade-off is evaluated.
Artificial selection is also distinct from directly editing DNA. Choosing parents and crossing them uses inheritance and recombination; it does not require insertion or editing of a sequence. These approaches can be combined, but neither “bred by humans” nor “genetically changed over generations” specifies which molecular technique was used. The central question here remains who contributes inherited variation to the next generation.
Check the claim against the data
1. Does choosing parents averaging 32 days mean their offspring must average 32 days?
No. The selected-parent mean measures the choice, not the response. Genetic recombination and environmental contributions can make the offspring distribution different. The measured generation-1 mean in this example is 37 days.
2. What is the line difference at generation 2?
The selected mean minus the control mean is \(35-39.5=-4.5\) days. Thus the selected line flowers 4.5 days earlier than its contemporary control, not 5 days earlier. Five days is the comparison with its own generation-0 mean.
3. A grower improves fertilizer and every plant flowers earlier. Is this evidence of artificial selection?
Not by itself. The intervention altered the environment. Show that human breeding choices changed reproductive contributions and that the resulting difference is inherited before attributing a generational shift to artificial selection.
4. Why might a control line change even when parents are chosen randomly?
Random choice does not freeze the line. Environmental differences, genetic drift, or natural selection can still affect it. Comparable handling and multiple independent lines help distinguish these effects from the intended breeding treatment.
5. If two varieties look very different, must each contain high genetic diversity?
No. The contrast describes differences between varieties. Each could be genetically uniform at many loci. Examine variation within each population separately rather than inferring it from their contrasting appearances.
Concept references: OpenStax Biology 2e: Pollination and Fertilization; Iowa State University Digital Press, Crop Genetics: Inheritance of Quantitative Traits; OpenStax Biology 2e: Population Genetics; OpenStax Concepts of Biology: Importance of Biodiversity. Scope reference: College Board AP Biology Course and Exam Description, Topic 7.3. The explanation, flowering-time dataset, calculations, graph, and questions were independently composed. External textbook figures and classroom worksheets are not reproduced.
Official curriculum reference: College Board AP Biology course page. The explanation and worked example are independently written for this study site.