- Selection criteria are the measurable sources of information used to identify and rank animals for breeding. They provide the practical basis for deciding which animals should contribute genes to the next generation. While a breeding objective defines the long-term genetic goals of a breeding programme, selection criteria describe the information used to predict which animals are most likely to contribute to those goals. Selection criteria may include an animal’s own phenotype, relatives’ performance, progeny records, pedigree information, genomic information, repeated measurements, indicator traits, or combinations of these sources.
- The distinction between a breeding objective, selection objective, and selection criterion is fundamental to modern animal breeding. A breeding objective describes the desired long-term genetic improvement of the population. A selection objective translates these goals into the traits and genetic changes that should receive emphasis. A selection criterion is the actual information used to rank selection candidates. For example, a breeding programme may have improved fertility as part of its breeding objective, while fertility records, reproductive indicators, pedigree information, and genomic predictions may serve as selection criteria.
- The usefulness of a selection criterion depends on how strongly it is associated with the animal’s true breeding value for the objective trait. The phenotype of an animal is influenced by both genetic and environmental effects and can be represented as:
- P = G + E
- where P is phenotype, G represents genetic effects, and E represents environmental effects. Because environmental influences can obscure genetic differences, an animal with superior observed performance is not necessarily genetically superior.
- The genetic component of performance includes additive genetic effects, dominance, and epistasis. Additive genetic effects are particularly important for selection because they represent the component of genetic variation that is predictably transmitted from parents to offspring. Selection criteria should therefore provide information that helps estimate additive genetic merit as accurately as possible.
- Individual performance is one of the simplest selection criteria. An animal can be ranked according to its own measured performance for traits such as body weight, growth rate, milk yield, egg production, wool production, feed intake, carcass traits, or other measurable characteristics. This approach is often referred to as individual selection or phenotypic selection.
- Individual performance can be particularly useful when the trait has moderate or high heritability and can be measured accurately before the animal is selected for breeding. Narrow-sense heritability is expressed as:
- h² = σ²_A / σ²_P
- where σ²_A is additive genetic variance and σ²_P is phenotypic variance. Higher heritability generally increases the extent to which an individual’s phenotype provides information about its breeding value, although heritability alone does not determine selection accuracy.
- Environmental effects must be considered when using individual performance as a selection criterion. Differences in nutrition, housing, disease exposure, season, management, age, sex, and other environmental conditions can cause animals to differ phenotypically without corresponding differences in genetic merit. Contemporary groups are therefore important because animals should generally be compared with other animals raised under similar environmental and management conditions.
- An animal that performs exceptionally well because it received superior nutrition should not automatically be considered genetically superior to an animal raised under poorer conditions. Genetic evaluation methods attempt to separate genetic effects from environmental effects so that selection decisions are based more accurately on inherited merit.
- Family information provides another important selection criterion. Performance records from parents, siblings, half-siblings, offspring, and other relatives contain information about an animal’s genetic merit because relatives share genes. Family information can be especially valuable for traits with low heritability or traits that cannot be measured reliably on the selection candidate itself.
- Family selection uses the performance of relatives or family groups to identify animals with desirable genetic potential. Full-sibling and half-sibling records can provide information about genetic merit, although the strength of the information depends on the degree of relatedness and the number and quality of records available.
- Within-family selection compares relatives raised in similar environments. This can reduce the influence of common environmental effects and help identify individuals with superior genetic performance within a family. It can be particularly useful when family members share environmental conditions that would otherwise make between-family comparisons misleading.
- Family information can also help separate common environmental effects from genetic effects. When several offspring share the same dam, housing environment, management, or other conditions, their performance may be similar because of environmental factors rather than shared additive genetic merit. Appropriate statistical models are therefore important when family information is used as a selection criterion.
- Progeny performance provides another powerful source of selection information. In progeny testing, the performance of an animal’s offspring is used to evaluate the parent’s breeding value. This can be particularly useful for sex-limited traits, late-expressed traits, expensive traits, or traits that cannot be measured directly on the selection candidate.
- For example, a male animal cannot directly express milk production, egg production, or many maternal traits. Information from daughters or other offspring can therefore provide valuable evidence about his genetic merit for these traits. Progeny records also average environmental variation across multiple offspring, potentially increasing the accuracy of parental evaluation when enough progeny are available.
- The number of progeny recorded is important. A small number of offspring may provide limited information, while larger numbers can increase the reliability of breeding value estimates. However, traditional progeny testing can require substantial time and resources and can increase the generation interval because selection decisions may need to wait until offspring mature and produce records.
- Pedigree information is another selection criterion or source of information. Pedigrees describe expected genetic relationships among animals and allow genetic evaluation models to account for information from ancestors and relatives. The numerator relationship matrix, commonly represented by the A matrix, can summarize expected additive genetic relationships among animals in a pedigree.
- Pedigree information does not directly measure which alleles an animal carries. Instead, it describes expected relationships based on inheritance probabilities. Therefore, genomic information can provide additional information about the actual genetic similarity between animals.
- Genomic information has become one of the most important modern selection criteria. Genotyping allows breeders to evaluate genetic markers distributed throughout the genome and use them to predict an animal’s genetic merit. This can generate genomic estimated breeding values (GEBV) that may be available early in life.
- Genomic selection can be particularly valuable when selection decisions need to be made before animals have sufficient own performance or progeny records. Young animals can potentially be evaluated for traits such as fertility, disease resistance, longevity, production, feed efficiency, and other traits using genomic information combined with phenotypic and pedigree data from a reference population.
- A genomic relationship matrix, often represented by the G matrix, provides information about realized genetic relationships among animals based on their genotypes. This can improve the estimation of breeding values compared with relying only on expected pedigree relationships.
- The accuracy of genomic selection depends heavily on the quality and relevance of the reference population. The reference population must contain animals with reliable phenotypes and genotypes that are sufficiently representative of the population in which genomic predictions will be applied. Prediction accuracy can decline when genetic relationships between the reference and target populations are weak or when the trait is poorly recorded.
- Indicator traits can also serve as selection criteria. An indicator trait is a measurable characteristic that provides information about the genetic merit of a more difficult-to-measure objective trait. This approach is especially useful when the objective trait is expensive, late in life, sex-limited, difficult to measure, or requires destructive testing.
- For example, an easily measured physiological, behavioural, production, or genomic trait may provide information about disease resistance, fertility, feed efficiency, carcass quality, or resilience. The usefulness of an indicator trait depends largely on its genetic correlation with the objective trait and the accuracy with which the indicator can be measured.
- Genetic correlation between traits can be represented as:
- r_A = Cov_A(X,Y) / (σ_A,X × σ_A,Y)
- A strong favourable genetic correlation can make an indicator trait particularly useful for indirect selection. However, a phenotypic correlation alone does not necessarily indicate that selection on one trait will produce a useful genetic response in another. The genetic relationship is what matters for predicting correlated genetic change.
- Repeated records can also be used as selection criteria. Some traits, such as milk production, egg production, growth, behaviour, or health indicators, can be measured multiple times on the same animal. Repeated observations may provide more information about persistent genetic differences when temporary environmental variation is appropriately modelled.
- Repeatability describes the correlation between repeated records on the same animal due to genetic and permanent environmental effects. When repeated measurements are informative, they can improve the reliability of evaluation and help distinguish temporary environmental deviations from more persistent differences.
- Maternal effects can be particularly important for early-life traits. Birth weight, weaning weight, survival, and early growth may be influenced by the dam’s genetics and environment as well as the offspring’s own genes. A selection criterion based solely on offspring phenotype can therefore be misleading if maternal and common environmental effects are not properly accounted for.
- Similarly, permanent environmental effects may influence repeated performance records. Statistical genetic models can separate these effects from additive genetic effects so that selection criteria more accurately reflect breeding value.
- Selection criteria can involve threshold traits and categorical records. Disease status, pregnancy success, calving difficulty, survival, and some fertility outcomes may be recorded as categories rather than continuous measurements. Appropriate threshold or generalized models can use such data to estimate genetic merit.
- The choice of selection criterion should therefore consider the biological nature of the trait. Continuous traits, repeated traits, binary traits, categorical traits, time-to-event traits, and longitudinal traits may require different statistical approaches.
- A critical property of any selection criterion is selection accuracy. Accuracy describes how closely the predicted genetic merit of an animal corresponds to its true breeding value. More informative selection criteria generally increase accuracy.
- Selection accuracy can be improved by combining multiple sources of information. For example, an animal may have its own phenotype, information from parents and siblings, progeny records, genomic data, and repeated measurements. Modern genetic evaluation systems can integrate these sources into a single breeding value estimate.
- BLUP, or Best Linear Unbiased Prediction, is one of the major statistical frameworks used to combine information from different sources. Animal models can simultaneously account for pedigree relationships, contemporary groups, environmental effects, repeated records, maternal effects, and other relevant factors.
- Rather than treating each selection criterion independently, BLUP-based genetic evaluation can combine the available information to estimate an animal’s estimated breeding value (EBV). The EBV represents the predicted additive genetic merit of the animal for a particular trait.
- The difference between phenotype and EBV is important. A phenotype describes what an animal has actually measured, whereas an EBV attempts to predict the inherited genetic component underlying that performance. An animal may have a high phenotype but a moderate EBV if much of its superior performance is attributed to environmental conditions. Conversely, an animal with average phenotype may have a high EBV if environmental conditions suppressed its observed performance.
- Genomic selection extends this framework by incorporating DNA-marker information. A GEBV can combine genomic information with pedigree and phenotypic information to predict genetic merit. This can improve selection decisions, particularly for young animals and difficult-to-measure traits.
- Selection criteria can also be combined through a selection index. If several pieces of information are available, the index can combine them into a single score:
- I = b₁x₁ + b₂x₂ + … + bₙxₙ
- where I is the selection index, x represents selection information, and b represents coefficients determined from the genetic and statistical relationships among the information sources and the breeding objective.
- A selection index therefore provides a practical bridge between breeding objectives and selection criteria. The breeding objective defines what genetic improvement is desired, while the selection index determines how available information can be combined to identify animals that are expected to contribute most effectively to that objective.
- Selection criteria can be applied to single-trait selection or multiple-trait selection. In multiple-trait selection, information from several traits may be combined because the breeding programme aims to improve several characteristics simultaneously. This is particularly important when traits are genetically correlated or when improvement in one trait can produce undesirable changes in another.
- For example, a breeding programme may seek increased production while maintaining fertility, health, longevity, and welfare. Selection criteria might therefore include production records, reproductive records, health data, longevity information, genomic predictions, and correlated indicator traits.
- The relative contribution of each criterion depends on its accuracy, cost, availability, genetic relationship with the breeding objective, and potential contribution to genetic gain. A highly accurate but very expensive measurement may not always be practical for every candidate. Conversely, a low-cost but less accurate measurement may be useful when applied to large numbers of animals.
- Cost of measurement is therefore an important practical consideration. Breeding programmes must decide which traits can realistically be recorded across the population. Automated phenotyping, sensors, imaging, electronic identification, precision livestock technologies, and genomic testing are increasingly expanding the range of traits that can be measured economically.
- Selection criteria should also be standardized. Differences in measurement protocols, trait definitions, equipment, laboratories, scoring systems, or recording practices can reduce comparability among animals. Consistent data collection improves the quality of genetic evaluation.
- Contemporary group design is particularly important for traits influenced strongly by management. Animals should be compared with appropriate contemporaries so that environmental differences do not become incorrectly attributed to genetic differences. This is essential for accurate selection criteria in commercial breeding systems.
- Genotype–environment interaction (G×E) can also affect the usefulness of selection criteria. A selection criterion recorded in one environment may not predict genetic performance equally well in another. For example, heat tolerance measured in a temperate environment may provide limited information about performance under severe heat stress. Similarly, disease-resistance records collected under one pathogen exposure level may not fully predict resistance under another challenge.
- Selection criteria should therefore be relevant to the environments in which selected animals and their descendants will be used. In some cases, environment-specific records or reaction norms may be needed.
- The choice of selection criteria also has implications for generation interval. Traditional progeny testing may provide highly informative records but requires waiting for offspring to mature and produce performance data. Genomic selection can provide useful information at a much younger age, potentially shortening the generation interval and increasing the rate of genetic improvement.
- The annual rate of genetic improvement can be approximated as:
- ΔG/year = i × r × σ_A / L
- Improving selection criteria can increase r, the accuracy of selection, while earlier evaluation through genomic information can reduce L, the generation interval. Both changes can increase the rate of genetic improvement.
- However, maximizing selection accuracy and genetic gain should not be the only goal. Genetic diversity must also be considered. If the same small number of animals consistently receive the highest rankings and are used extensively, genetic concentration may increase.
- The use of popular sires can be particularly important in this context. A highly successful sire can contribute genes to a large proportion of the population within a short period. Although this may accelerate genetic improvement, excessive use can increase relatedness and inbreeding and reduce the effective population size.
- A sustainable selection system therefore needs to balance selection accuracy and genetic gain with long-term diversity. Optimal contribution selection (OCS) can be used to determine appropriate contributions from selected animals while controlling coancestry and expected inbreeding.
- Expected offspring inbreeding can be related to parental kinship:
- E(F_offspring) = φ(sire, dam)
- where φ represents parental coancestry. The relationship between parents is approximately twice their kinship coefficient:
- r ≈ 2φ
- Genomic relatedness can provide more precise information about realized genetic similarity than pedigree alone. This can be particularly valuable when managing mating decisions and controlling inbreeding.
- Runs of homozygosity (ROH) can provide another source of information about genomic autozygosity. A genomic inbreeding measure based on ROH can be represented as:
- F_ROH = Total length of ROH / Total autosomal genome length
- This information can complement pedigree-based inbreeding estimates and help identify populations or animals with elevated levels of homozygosity.
- Selection criteria may also include genetic testing for deleterious variants. If a known recessive mutation is present in a population, carrier information can be incorporated into selection and mating decisions. The goal is often to avoid carrier-by-carrier matings rather than automatically eliminating every carrier, because excessive removal of carriers may unnecessarily reduce genetic diversity.
- Health, fertility, welfare, and adaptation traits deserve particular attention when developing selection criteria because they are often more difficult to measure than conventional production traits. They may have lower heritability, complex genetic architecture, strong environmental influences, or expensive phenotyping requirements. However, modern genetic evaluation, genomic selection, automated phenotyping, and large-scale recording systems are making these traits increasingly accessible.
- For disease resistance, selection criteria may include disease incidence, pathogen load, immune response, clinical scores, survival, or genomic indicators. For fertility, criteria may include age at first reproduction, conception rate, pregnancy success, calving interval, litter size, semen quality, or reproductive survival. For welfare, criteria may include structural soundness, injury incidence, temperament, disease frequency, behavioural indicators, and survival.
- For adaptation and resilience, selection criteria may include performance under heat stress, disease challenge, variable feed availability, climatic stress, or other environmental disturbances. The objective is not necessarily to select animals that perform maximally under ideal conditions, but animals that maintain appropriate performance and health under realistic production conditions.
- Selection criteria should ultimately be evaluated according to how well they support the overall breeding objective. A criterion is useful when it provides reliable information about genetic merit for one or more traits in the objective and can be measured with sufficient accuracy and practicality.
- No single selection criterion is universally superior. Individual performance may be highly informative for some traits, family information may be more valuable for others, progeny testing may be essential for sex-limited or late-expressed traits, and genomic information may provide the greatest advantage for young animals. The most effective systems often combine several sources.
- This is the basis of combined selection, where individual, family, pedigree, progeny, and genomic information are integrated into genetic evaluation. Modern breeding programmes increasingly use such combined approaches because different sources of information provide complementary advantages.
- The ultimate goal of selection criteria is therefore not simply to identify animals with the best observed performance. It is to identify animals with the greatest expected genetic merit for the breeding objective while accounting for environmental effects, relationships among traits, selection accuracy, genetic diversity, and long-term sustainability.
- Well-designed selection criteria make it possible to transform broad breeding goals into practical selection decisions. They connect breeding objectives with measurable evidence and provide the information required for EBV, GEBV, BLUP, selection indexes, genomic selection, and other genetic evaluation methods.
- In modern animal breeding, selection criteria are becoming increasingly diverse and precise. Phenotypic records, pedigree relationships, family information, progeny performance, genomic data, automated phenotyping, health records, behavioural measurements, and environmental information can all contribute to predicting genetic merit. The challenge is to combine these sources intelligently while maintaining reliable data quality and appropriate statistical models.
- A successful breeding programme therefore chooses selection criteria that are accurate, relevant, measurable, cost-effective, genetically informative, and aligned with the breeding objective. When these criteria are combined with appropriate genetic evaluation and responsible mating strategies, they can support sustained improvement in production, fertility, health, welfare, adaptation, resilience, and overall population performance while protecting genetic diversity for future generations.
