- Artificial selection is the deliberate choice of individuals for reproduction based on desirable characteristics. Unlike natural selection, which results from differences in survival and reproductive success under environmental conditions, artificial selection is directed by humans according to defined breeding objectives. In animal breeding, artificial selection is one of the fundamental processes used to change the genetic composition of populations and improve traits such as growth, milk production, meat quality, egg production, wool production, fertility, disease resistance, survival, adaptation, and other economically or biologically important characteristics.
- The basic principle of artificial selection is simple: animals with desirable genetic characteristics are preferentially used as parents of the next generation, while animals with less desirable genetic characteristics contribute fewer offspring or are excluded from breeding. When the selected differences have a genetic basis, the frequency of favorable alleles and genetic combinations can change over generations, resulting in a measurable change in the population’s average genetic merit.
- Artificial selection should be distinguished from simply choosing the best-looking or highest-performing animals. An animal’s observed phenotype is influenced by both genetic and environmental factors. The basic quantitative genetics model can be expressed as:
- P = G + E
- where P is the observed phenotype, G is the genetic component, and E represents environmental effects. An animal may perform exceptionally well because of superior genetics, favorable nutrition, excellent management, low disease exposure, or a combination of these factors. Effective artificial selection therefore attempts to identify animals with superior genetic merit, rather than selecting solely on observed performance.
- The genetic component of a phenotype includes additive genetic effects, dominance effects, and epistatic effects. Additive genetic variation is particularly important for artificial selection because additive effects are transmitted from parents to offspring in a predictable manner. The breeding value of an animal represents the additive genetic contribution that it is expected to transmit to its offspring.
- An animal’s estimated breeding value (EBV) is therefore more useful for selection than phenotype alone when reliable genetic evaluation is available. Breeding values can be estimated using information from the animal itself, its relatives, progeny, pedigree, and increasingly its genome.
- Artificial selection operates through differences in reproductive contribution. If selected animals produce more offspring than non-selected animals, the genetic composition of the next generation differs from that of the previous generation. Repeated over generations, this creates a selection response and can lead to sustained genetic gain.
- The difference between the mean phenotype or breeding value of selected animals and the mean of the population before selection is known as the selection differential. A larger selection differential generally creates greater opportunity for genetic change, provided that the trait has sufficient genetic variation and the selection criterion accurately reflects genetic merit.
- Selection intensity describes how strongly animals are selected. If only a small proportion of animals are chosen as parents, selection intensity is high. If a large proportion is retained, selection intensity is lower. Strong selection can increase the rate of genetic improvement, but excessive selection pressure can also reduce genetic diversity and increase the risk of genetic concentration.
- The relationship between selection differential, additive genetic variance, and response is central to quantitative genetics. The expected response to selection can be expressed as:
- R = h² × S
- where R is the expected response to selection, h² is narrow-sense heritability, and S is the selection differential.
- This relationship demonstrates why heritability is important in artificial selection. A trait with higher narrow-sense heritability generally provides greater opportunity for response to selection when the same selection differential is applied. However, heritability is a population-specific parameter and does not mean that a fixed percentage of an individual animal’s phenotype is genetic.
- Low heritability does not mean that a trait cannot respond to artificial selection. Even traits with relatively low heritability can show genetic improvement when selection is based on accurate information from relatives, repeated records, genomic information, or appropriate statistical models. Fertility, survival, disease resistance, behaviour, and other complex traits may have relatively low heritability but can still be important targets for genetic improvement.
- Artificial selection can be based on different sources of information. Individual selection uses the animal’s own phenotype. This can be effective when the trait is measurable in the individual and has adequate heritability. However, individual performance may be strongly influenced by environmental conditions, particularly for traits with substantial environmental variation.
- Family selection uses information from relatives. Family information can improve selection accuracy when an individual’s own phenotype is unavailable, difficult to measure, sex-limited, expressed late in life, or strongly influenced by environmental effects.
- Progeny testing evaluates an animal using the performance of its offspring. This approach has historically been important for traits such as milk production and other traits where the parent cannot directly express the relevant phenotype. Progeny testing can provide high accuracy but may require many offspring and can increase the generation interval.
- Modern breeding programs increasingly use BLUP (Best Linear Unbiased Prediction) to estimate breeding values. BLUP combines phenotypic records, pedigree relationships, fixed effects, and information from relatives to produce statistically adjusted predictions of genetic merit.
- A major advantage of BLUP is that it attempts to separate genetic effects from systematic environmental effects. For example, animals raised in different herds, years, seasons, management groups, or production systems can be compared more fairly when these environmental effects are appropriately modeled.
- Animal models are widely used for genetic evaluation. An animal model can incorporate records from the individual and its relatives while accounting for the relationship structure of the population. This makes it possible to estimate breeding values even when individual animals have limited phenotypic information.
- Artificial selection can also be based on family breeding values, maternal breeding values, or other specialized genetic evaluations. This is particularly important for traits affected by maternal performance, common environmental effects, or other sources of variation.
- For example, growth in young animals may be influenced by the individual’s own genes as well as the genetic and environmental effects of its mother. Maternal effects can therefore be included in genetic evaluations to avoid attributing maternal influences incorrectly to the offspring’s direct genetic merit.
- Similarly, permanent environmental effects can influence repeated records from the same animal. A breeding program must distinguish these effects from genetic differences to avoid selecting animals for environmental advantages that are not transmitted to offspring.
- Artificial selection can be applied to many different trait categories. Production traits include milk yield, growth rate, meat production, carcass composition, egg production, wool production, and fiber characteristics. Reproductive traits include fertility, age at sexual maturity, conception rate, litter size, calving interval, semen quality, and reproductive longevity.
- Health traits include disease resistance, disease susceptibility, immune function, survival, and resilience. Functional traits may include locomotion, structural soundness, longevity, udder health, claw health, and other characteristics affecting productive life.
- Behavioural and welfare-related traits can also be targets of artificial selection. Temperament, stress response, social behaviour, maternal behaviour, and other behavioural characteristics may have genetic components. Selecting for these traits can contribute to animal welfare and improve compatibility between animals and their production environments.
- Artificial selection can also target adaptation traits. Heat tolerance, disease resilience, feed efficiency under challenging conditions, water-use efficiency, and environmental adaptation may become increasingly important as production systems and climates change.
- Selection for a single trait can produce rapid improvement in that trait but may create undesirable correlated changes in other traits. This occurs because traits can share genetic pathways or because alleles affecting one trait also influence another. Genetic correlation describes the extent to which genetic effects affecting two traits are associated.
- For example, selection for increased production may be genetically correlated with fertility, health, survival, or longevity. If the correlation is unfavorable, intense single-trait selection may lead to deterioration in the correlated trait.
- This is why modern artificial selection often uses multi-trait selection. Rather than maximizing one trait, breeders define a breeding objective containing several traits and their relative economic, biological, welfare, or conservation importance.
- A selection index combines information from multiple traits into a single selection criterion. It can incorporate production, fertility, health, survival, welfare, adaptation, and other breeding objectives. This allows breeders to pursue balanced genetic improvement.
- Artificial selection can also produce correlated responses in traits that are not directly included in the selection criterion. These correlated responses can be favorable or unfavorable depending on the genetic correlations between traits.
- The long-term rate of genetic improvement depends on several factors. A commonly used expression is:
- ΔG/year = i × r × σ_A / L
- where i is selection intensity, r is accuracy of selection, σ_A is additive genetic standard deviation, and L is generation interval.
- This equation shows that genetic improvement can be increased through greater selection intensity, higher selection accuracy, greater additive genetic variation, or a shorter generation interval. However, changing one component can affect the others and can create trade-offs with genetic diversity, welfare, reproductive performance, or long-term sustainability.
- Accuracy of selection is particularly important because selecting the wrong animals reduces genetic progress. Accuracy can be improved by using better phenotypic records, larger reference populations, pedigree information, progeny information, repeated records, and genomic data.
- Genomic selection has transformed artificial selection in many modern breeding programs. Genomic selection uses large numbers of DNA markers, usually SNPs, to predict genomic breeding values. It can increase selection accuracy for young animals before they have extensive phenotypic or progeny records.
- One of the major advantages of genomic selection is the ability to shorten the generation interval. Young animals can be selected earlier, reducing the time between generations. This can increase the rate of genetic gain.
- However, accelerated selection also requires careful population management. If the highest genomic breeding values are concentrated in a small number of related families, rapid genomic selection can increase genetic concentration, reduce effective population size, and increase inbreeding.
- Artificial selection must therefore be balanced with genetic diversity. Genetic diversity provides the raw material for future selection and adaptation. Excessive selection pressure can reduce the frequency of some alleles and increase the representation of others.
- The concept of effective population size (Ne) is useful for evaluating this risk. A simplified approximation for the increase in inbreeding per generation is:
- ΔF ≈ 1 / (2Ne)
- A smaller effective population size generally results in a faster increase in inbreeding. Unequal reproductive contribution, popular sires, population bottlenecks, restricted gene flow, and strong concentration of selection can all reduce effective population size.
- Artificial selection can therefore influence the genetic structure of a population beyond simply changing average breeding values. If a small number of animals produce a large proportion of offspring, their genetic contribution becomes disproportionately large.
- The popular sire effect is an important example. An exceptionally high-ranking male may be used extensively through artificial insemination or other reproductive technologies. This can spread favorable alleles rapidly but can also increase relatedness among future breeding animals.
- For this reason, selection should be combined with mate allocation and optimal contribution selection. Mate allocation considers which individuals should be paired, while optimal contribution selection determines how much each selected animal should contribute to the next generation.
- The goal is not necessarily to prevent the use of elite animals. Instead, breeders can use superior animals while controlling their reproductive contribution to avoid excessive genetic concentration.
- Mean kinship can also be incorporated into selection decisions. Animals with low mean kinship may represent underrepresented genetic lineages. If they have acceptable breeding values, including them in the breeding population can help maintain genetic diversity.
- Genomic relatedness provides another tool for managing artificial selection. Genomic information can identify animals that are more closely related than their pedigree suggests and can help prevent mating combinations that produce excessive expected inbreeding.
- Runs of homozygosity (ROH) can provide information about genomic homozygosity and recent common ancestry. The proportion of the autosomal genome contained within ROH can be expressed as:
- F_ROH = Total length of ROH / Total autosomal genome length
- Monitoring genomic inbreeding and ROH can therefore complement pedigree-based management of artificial selection.
- Artificial selection is also closely related to genetic load. If selection repeatedly favors animals carrying harmful recessive variants because they have superior performance for other traits, those variants may become widespread. Genetic testing and genomic information can help identify known deleterious variants and reduce the risk of their concentration.
- However, eliminating every carrier from a population is not always the best strategy. If carriers are rare and removing them would substantially reduce genetic diversity, controlled mating may be preferable. For example, carrier animals may be mated only with animals that are not carriers, preventing affected offspring while retaining valuable genetic variation.
- Artificial selection can also interact with genotype–environment interaction (G×E). Animals selected under one environment may not perform equally well under another. Genetic rankings can therefore change across production systems, climates, management conditions, or disease challenges.
- This is especially relevant to climate adaptation. Selection for performance under favorable conditions may produce animals that are highly productive but less resilient under heat stress, disease pressure, poor feed availability, or other environmental challenges. Breeding objectives increasingly need to consider both productivity and adaptation.
- Artificial selection is not limited to intensive commercial breeding. It has been used for centuries in livestock, poultry, aquaculture, companion animals, laboratory animals, and other domesticated populations. Historical selection has produced major differences among breeds in body size, morphology, production characteristics, behaviour, reproductive patterns, and environmental adaptation.
- Breed formation itself is often a result of long-term artificial selection combined with reproductive isolation and population management. Over time, repeated selection for specific characteristics can create distinct genetic populations.
- Artificial selection can be directional, stabilizing, or disruptive depending on the breeding objective. Directional selection favors increasing or decreasing a trait toward a desired value. Stabilizing selection favors an intermediate phenotype and may be useful when extremes are undesirable. Disruptive selection can favor different extremes and may contribute to population differentiation under particular breeding systems.
- Selection thresholds may also be used for categorical or binary traits. For example, animals may be classified as affected or unaffected by a disease. Underlying genetic liability can be modeled using a threshold model, where an unobserved continuous liability determines whether the observed category crosses a threshold.
- This is important because many economically and biologically important traits are not continuously measured. Fertility, survival, disease status, calving difficulty, reproductive success, and some behavioural traits may be recorded as binary or categorical outcomes.
- Artificial selection can be based on direct measurements or indirect indicators. An indicator trait may be useful when the target trait is difficult, expensive, late-expressed, sex-limited, or impossible to measure directly in selection candidates.
- For example, genetic markers may be used as indicators of disease resistance, while early-life measurements may provide information about later production. The usefulness of an indicator depends on its genetic correlation with the target trait and the accuracy with which it predicts genetic merit.
- Selection decisions can also incorporate repeatability when animals have multiple records for the same trait. Repeated records provide additional information about permanent differences among animals, although repeated observations are not independent because they may share genetic and environmental components.
- Artificial selection should also account for maternal effects, common environmental effects, and other sources of non-independence among records. Ignoring these effects can result in inaccurate estimates of genetic merit.
- The effectiveness of artificial selection depends on the presence of genetic variation. If a trait has no usable additive genetic variation within the population, selection cannot produce a sustained additive genetic response regardless of selection intensity. Maintaining genetic diversity is therefore essential for long-term breeding progress.
- At the same time, genetic variation does not automatically guarantee improvement. Selection must be based on accurate information and aligned with the breeding objective. Poor measurement, environmental bias, incorrect pedigrees, inadequate statistical models, or inappropriate selection criteria can reduce realized genetic gain.
- The relationship between selection differential, heritability, and response provides a simple conceptual framework, but modern breeding programs generally use more sophisticated prediction methods. BLUP and genomic evaluation can account for multiple sources of information and complex population structures.
- Artificial selection can also be integrated with crossbreeding. In a crossbreeding program, selection within parental populations can improve additive genetic merit, while crossbreeding can exploit heterosis and breed complementarity. This combination can be particularly useful for fitness-related traits such as fertility, survival, and disease resistance.
- The choice between purebred selection and crossbreeding depends on the production system. Purebred selection is important for maintaining and improving specialized populations, while crossbreeding can combine complementary genetic strengths and increase heterozygosity.
- Artificial selection also has an important role in conservation breeding. Conservation programs may select animals differently from commercial systems. Instead of maximizing production, the objective may be to preserve genetic diversity, maintain rare alleles, minimize inbreeding, and retain adaptation to local environments.
- In small populations, unrestricted selection can quickly increase relatedness. Conservation breeding therefore often places greater emphasis on balanced reproductive contributions, mean kinship, and effective population size.
- The ethical dimension of artificial selection should also be considered. Genetic improvement can provide substantial benefits, but selection for extreme phenotypes can create welfare problems if correlated biological functions are compromised. Breeding objectives should therefore consider animal health, welfare, functional ability, fertility, longevity, and resilience.
- A successful breeding program does not simply ask which animal is the most productive. It asks which animals will best contribute to the long-term breeding objective while maintaining a healthy and genetically sustainable population.
- The distinction between phenotypic selection and genetic selection is therefore fundamental. Phenotypic selection chooses animals according to observed performance. Genetic selection uses information designed to predict inherited genetic merit. Modern breeding increasingly combines phenotypic, pedigree, genomic, and environmental information.
- Artificial selection can produce substantial cumulative change because genetic improvement is inherited across generations. Small improvements repeated over many generations can produce large differences between the original population and the modern breeding population.
- However, genetic change is not necessarily permanent or universally favorable. If selection objectives change, previously favored alleles may become less valuable. Environmental conditions may also change, altering the relative importance of traits.
- This is why breeding objectives should be periodically reviewed. Economic conditions, consumer preferences, disease threats, climate, welfare expectations, and production systems can change over time.
- A sustainable artificial selection program therefore needs both short-term and long-term objectives. Short-term objectives focus on measurable genetic improvement, while long-term objectives protect genetic diversity, health, fertility, adaptation, and population resilience.
- The most effective programs combine selection accuracy, appropriate selection intensity, adequate genetic variation, suitable generation intervals, balanced breeding objectives, and responsible reproductive contribution. Genetic gain should be pursued without creating unnecessary increases in inbreeding or loss of genetic diversity.
- In conclusion, artificial selection is the deliberate use of reproductive decisions to change the genetic composition of a population in a desired direction. It is one of the central foundations of animal breeding and has enabled major improvements in production, health, reproduction, adaptation, and other traits. The effectiveness of artificial selection depends on genetic variation, heritability, selection differential, selection intensity, accuracy of genetic evaluation, generation interval, and the breeding objective. Modern approaches such as BLUP, genomic selection, genetic testing, mate allocation, and optimal contribution selection have greatly increased the precision with which artificial selection can be practiced. The long-term goal is not simply to maximize genetic gain, but to achieve balanced improvement while maintaining genetic diversity, controlling inbreeding, protecting animal health and welfare, and preserving the capacity of breeding populations to adapt to future challenges.
