Selective breeding over generations

At the scale of one organism, breeding does not change that organism's genes. The change appears across a population over generations because humans decide which organisms reproduce.

The word desired describes a human aim, not a characteristic that is automatically good for the organism. A farmer might want wheat that resists a fungal disease; a dairy breeder might want cows that produce more milk; a dog breeder might select a particular body shape or behaviour.

The process needs inherited variation to begin with:

1. Decide the desired characteristic.
2. Find organisms that vary in that characteristic and choose parents that show it strongly.
3. Breed the selected parents. Their offspring inherit a mixture of alleles from both parents.
4. Measure or observe the offspring and select those that best show the desired inherited characteristic.
5. Use those offspring as parents and repeat the cycle for many generations.

Repeated selection makes alleles associated with the chosen characteristic more common. It does not guarantee that every offspring will have the characteristic: sexual reproduction produces genetic variation, many characteristics involve several genes, and the environment can also affect the phenotype.

Impacts on food plants and domesticated animals

Selective breeding can produce food plants with higher yield, improved food quality, or resistance to pests, disease and drought. It can produce domesticated animals with greater milk, egg or meat production, useful behaviour, disease resistance or improved health.

There can also be harmful impacts. Repeatedly breeding from a small group narrows the gene pool, so the population has less genetic variation with which to respond to a new disease or environmental change. If close relatives breed, both may carry the same harmful recessive allele, increasing the chance that offspring inherit two copies and develop an inherited disorder. Selecting only for production or an exaggerated body feature can also damage animal health and welfare.

These outcomes are not inevitable. Breeders can use unrelated parents, screen for harmful alleles and select for health, function and genetic diversity as well as production.

Applying genetic-cross probabilities to breeding

The Punnett-square method in Biology 3.14 and 3.16 can help a breeder predict one inherited characteristic. Consider a fictional crop with a single gene for resistance: allele $R$ is dominant for resistance and $r$ gives susceptibility. Two selected resistant parents are both heterozygous, $Rr$.

Each parent makes gametes carrying $R$ or $r$. Combining one gamete from each parent gives:

The genotype ratio is $1\,\mathrm{RR}:2\,\mathrm{Rr}:1\,\mathrm{rr}$. Both $RR$ and $Rr$ plants are resistant, so the phenotype ratio resistant:susceptible is $3:1$, and the probability that an offspring is resistant is $\frac34=75\%$. Resistance in both parents therefore does not guarantee resistance in every offspring.

For an expected count, multiply the total offspring by the probability. From 80 seedlings, predict $80\times\frac34=60$ resistant seedlings. If the cross and conditions stay the same, doubling the offspring total to 160 doubles the expected resistant count to $160\times\frac34=120$: expected count is directly proportional to the total. Individual outcomes are random, so observed counts need not exactly match these predictions. Assess the actual offspring before choosing the next parents. This one-gene example does not make complex features such as yield single-gene traits.