- Selection index is a quantitative method used in animal breeding to combine information from several traits or sources of information into a single numerical value for ranking animals for selection. Instead of evaluating animals on one trait at a time, a selection index combines measurements such as production, fertility, growth, health, feed efficiency, survival, welfare, and other economically or biologically important traits according to their genetic relationships and relative importance.
- The central purpose of a selection index is to identify animals that are expected to make the greatest contribution toward the overall breeding objective. Modern animal breeding rarely aims to maximize only one trait. Improving milk yield, meat production, egg production, wool production, or growth without considering fertility, health, longevity, welfare, feed efficiency, and adaptation can create undesirable genetic responses. A selection index provides a structured way to balance these traits and make selection decisions based on overall genetic merit.
- The genetic basis of selection index methodology comes from quantitative genetics. An animal’s observed phenotype can be represented conceptually as:
- P = G + E
- where P is the observed phenotype, G is the genetic component, and E is the environmental component. The genetic component may contain additive genetic effects, dominance effects, and epistatic effects. For most breeding-value-based selection decisions, the additive component is particularly important because it is the component that is predictably transmitted from parents to offspring.
- The main target of a selection index is therefore not simply the animal’s phenotype but its expected breeding value for the traits included in the breeding objective. A breeding value represents the additive genetic merit that an animal is expected to transmit to its descendants. Selection index methodology combines available information to predict this underlying genetic merit as accurately as possible.
- A breeding programme normally begins by defining a breeding objective. The breeding objective describes the long-term genetic changes that the programme wants to achieve. It may include traits such as milk yield, protein yield, fertility, somatic cell score, mastitis resistance, longevity, feed efficiency, body condition, heat tolerance, disease resistance, or welfare. The breeding objective can be represented as an aggregate breeding objective:
- H = a₁A₁ + a₂A₂ + … + aₙAₙ
- where H is the aggregate breeding objective, A represents the true additive genetic value for each objective trait, and a represents the relative importance or economic weight assigned to each trait.
- The economic weight does not necessarily mean that every trait must have a direct market price. Some traits, such as animal welfare, disease resistance, survival, resilience, and environmental adaptation, may have important biological, ethical, regulatory, or sustainability values even when their contribution is not easily expressed as a simple market price. Modern breeding objectives can therefore include both economic and non-market considerations.
- The selection index uses measurable information to predict the breeding objective. A general selection index can be expressed as:
- I = b₁x₁ + b₂x₂ + … + bₙxₙ
- where I is the selection index, x represents the available selection information, and b represents the index coefficients. The index coefficient determines how strongly each information source contributes to the final index value.
- The information included in an index can come from many sources. It may include an animal’s own performance, the performance of relatives, progeny records, pedigree information, genomic information, repeated records, indicator traits, or estimated breeding values. The index therefore provides a framework for combining different types of information into a single ranking criterion.
- For example, a dairy breeding programme may want to improve milk production while maintaining fertility, udder health, longevity, feed efficiency, and resistance to disease. An index could combine estimated breeding values for these traits, with coefficients reflecting their relative importance and genetic relationships. An animal with exceptionally high milk production but poor fertility and health may therefore rank below an animal with slightly lower production but substantially better overall genetic merit.
- This illustrates an important principle: selection index methodology is designed to optimize the overall breeding objective rather than maximize an individual trait.
- The relationship between the breeding objective and the selection index is fundamental. The breeding objective represents what the breeder ultimately wants to improve, whereas the selection index represents the statistical tool used to rank animals using available information. The selection criteria are the individual measurements or estimated values that provide information for constructing the index.
- For example, a breeding objective might include fertility, longevity, disease resistance, and production. The selection criteria might include age at first calving, reproductive records, survival records, disease observations, milk yield, genomic information, and progeny performance. The selection index combines these sources of information into a single ranking value.
- The quality of a selection index depends strongly on the heritability of the traits, the amount and quality of available information, the genetic relationships among traits, and the accuracy with which breeding values can be estimated. Heritability is commonly expressed as:
- h² = σ²_A / σ²_P
- where σ²_A is additive genetic variance and σ²_P is phenotypic variance.
- A trait with high heritability generally provides more information about an individual’s genetic merit from its own phenotype. However, selection index methodology is not restricted to highly heritable traits. Family, progeny, pedigree, repeated-record, and genomic information can improve prediction for traits with low heritability.
- This is particularly important for fertility, disease resistance, survival, longevity, behaviour, and many welfare-related traits, where individual phenotypic records may provide limited information about breeding value. In such situations, combining information from relatives, progeny, indicator traits, and genomic data can substantially improve selection decisions.
- The relationship between traits is another major component of selection index methodology. Traits are often genetically correlated because some genes or biological pathways influence more than one characteristic. Genetic correlation can be expressed as:
- r_A = Cov_A(X,Y) / (σ_A,X × σ_A,Y)
- where Cov_A(X,Y) is the additive genetic covariance between traits X and Y, and σ_A,X and σ_A,Y are their additive genetic standard deviations.
- A positive genetic correlation means that genetic improvement in one trait tends to be associated with improvement in another. A negative genetic correlation means that selection for one trait may produce an unfavorable response in another. Selection indexes allow these relationships to be incorporated into the overall selection decision.
- For example, selecting strongly for rapid growth without considering fertility or structural soundness could create undesirable correlated responses if the traits are genetically antagonistic. Similarly, selection for high production may need to be balanced against health, fertility, survival, or welfare. A multi-trait selection index can explicitly account for these relationships.
- The selection index coefficients are not normally chosen simply by intuition. They are derived using information about the covariance between the selection criteria and the breeding objective and the covariance among the information sources. In matrix notation, the index coefficients can be represented conceptually as:
- b = P⁻¹G a
- where P represents the covariance matrix among the selection criteria, G represents the covariance between selection criteria and objective traits, a represents the vector of economic weights, and b represents the vector of index coefficients.
- This statistical structure allows the index to account for both the relative importance of traits and the amount of overlapping information among them. If two measurements provide very similar information, simply adding both without accounting for their covariance could effectively give that information excessive weight.
- The selection index therefore differs from a simple weighted average. Its coefficients reflect the genetic and phenotypic relationships among the information sources and the breeding objective. This makes the method particularly powerful for complex breeding programmes involving many traits.
- The accuracy of an index is another important consideration. Selection accuracy describes the correlation between the predicted genetic merit and the true genetic merit of an animal. Greater accuracy generally means that selected animals are more likely to have the desired genetic merit.
- Accuracy can be improved by increasing the quality and quantity of information. Useful information may include individual performance, repeated measurements, family records, progeny records, pedigree relationships, genomic markers, and genetically correlated indicator traits.
- Traditional selection indexes often used phenotypic and pedigree information. Modern breeding programmes frequently integrate estimated breeding values (EBVs) generated through statistical genetic evaluation. EBVs can summarize information from an animal’s own performance and the performance of relatives, progeny, and other connected animals.
- Best linear unbiased prediction (BLUP) has become a major foundation of modern genetic evaluation. BLUP-based animal models can simultaneously account for fixed environmental effects and random genetic effects while using relationships among animals. Pedigree-based relationship matrices, often represented by the A matrix, allow information from relatives to contribute to breeding-value prediction.
- Modern selection indexes can therefore use EBVs from BLUP evaluations as selection criteria. Instead of directly combining raw phenotypic observations, the index can combine estimated genetic merits for different traits.
- Genomic information has further expanded selection index methodology. Genomic selection uses genome-wide genetic markers to improve prediction of genetic merit. Genomic evaluation can produce genomic estimated breeding values (GEBVs) that combine genomic relationships with phenotypic and pedigree information.
- The genomic relationship matrix, commonly called the G matrix, provides information about genetic similarity based on genomic markers. This can improve the ability to distinguish animals that have similar pedigree relationships but differ substantially in their actual inherited DNA.
- A modern multi-trait selection programme may therefore combine EBVs, GEBVs, genomic information, family records, progeny performance, repeated records, and indicator traits. The resulting index can provide a comprehensive ranking of candidates for selection.
- Multi-trait selection is one of the most important applications of selection index methodology. When several traits contribute to the breeding objective, selecting independently for each trait can produce conflicting decisions. A selection index provides a single framework for evaluating overall merit.
- For example, a pig breeding programme may simultaneously consider growth rate, feed efficiency, carcass quality, reproductive performance, survival, and robustness. A poultry programme may consider egg number, egg quality, body weight, fertility, hatchability, disease resistance, and persistency. A sheep programme may combine growth, carcass traits, wool quality, reproductive performance, parasite resistance, and survival.
- In cattle, selection indexes may combine milk yield, fat yield, protein yield, fertility, mastitis resistance, udder conformation, longevity, feed efficiency, hoof health, and other traits. The exact structure depends on the production system, economic environment, climate, management conditions, and breeding goals.
- Selection indexes are particularly useful when traits have different units of measurement. Milk yield may be measured in kilograms, fertility may be measured as a reproductive rate or days, disease resistance may be expressed as a score or probability, and longevity may be measured in years. The index converts these different sources of information into a common ranking system.
- The method can also incorporate indicator traits. An indicator trait is a measurable characteristic that provides information about another trait that may be difficult, expensive, or slow to measure. For example, a correlated physiological or production measurement may provide information about disease resistance, feed efficiency, fertility, or resilience.
- Indicator traits are especially valuable when the target trait has low heritability or is difficult to measure directly. Their usefulness depends on the strength and stability of the genetic relationship between the indicator and target traits.
- Selection indexes can also incorporate repeated records. Some traits are measured multiple times during an animal’s life, such as milk production across lactations, body weight over time, behavioural measurements, or disease events. Repeated records can provide more information about permanent genetic merit when appropriately modeled.
- The concept of repeatability is important in such situations because repeated observations on the same animal are not necessarily independent. A repeated-record model can separate genetic effects from permanent environmental effects and temporary environmental variation.
- Maternal effects and common environmental effects can also influence selection decisions. For example, litter mates may share the same dam, uterine environment, early nutrition, housing, and management. Their phenotypic similarity may therefore reflect environmental factors rather than shared genetic merit. Appropriate statistical models are required to avoid assigning environmental similarity incorrectly to genetic effects.
- Selection index methodology is also useful for traits that are recorded as categories or binary outcomes. Examples include survival, disease status, pregnancy success, calving difficulty, reproductive success, and certain welfare outcomes. These are often treated as threshold traits, where an underlying continuous liability produces an observed categorical outcome.
- Threshold models and other appropriate statistical methods can be incorporated into genetic evaluation before the resulting breeding-value information is included in a selection index.
- An important advantage of selection index methodology is that it can integrate traits with very different biological characteristics. Production traits may be continuous, fertility may involve reproductive events, disease resistance may involve binary outcomes, and welfare traits may involve behavioural scores. A carefully designed index can combine information from all of these areas.
- The economic component of an index is particularly important. Economic weights describe the relative value of genetic improvement in different traits within the production system. These weights should reflect the consequences of genetic change rather than simply the current price of a product.
- For example, an improvement in fertility may increase the number of productive animals, reduce replacement costs, decrease veterinary expenses, and improve labour efficiency. Its total economic value may therefore be greater than what would be apparent from a single market price.
- Similarly, improvement in disease resistance may reduce treatment costs, mortality, production losses, and welfare problems. Improvement in feed efficiency may reduce feed costs and environmental impact. Improvement in longevity may reduce replacement requirements and increase lifetime productivity.
- For this reason, selection indexes should be periodically reviewed. Economic conditions, production systems, regulations, climate, disease pressures, consumer preferences, and breeding goals can change over time.
- A selection index can also be designed to include non-economic traits. Modern breeding programmes increasingly consider animal welfare, disease resistance, resilience, environmental adaptation, methane emissions, heat tolerance, robustness, and survival. These traits may have social, ethical, environmental, or regulatory importance even when their economic value is difficult to quantify precisely.
- Genotype–environment interaction (G×E) is another consideration. The genetic ranking of animals may differ between environments. An animal that performs exceptionally well under intensive management may not have the same advantage under extensive, hot, cold, disease-challenging, or low-input conditions.
- Selection indexes intended for broad populations should therefore consider whether the breeding objective applies across multiple environments. In some situations, environment-specific breeding objectives or multi-environment genetic evaluations may be appropriate.
- Selection index methodology is also connected to selection response. For a simple single-trait situation, expected response can be represented as:
- R = h² × S
- where R is response to selection, h² is heritability, and S is the selection differential.
- For multiple traits, the expected response becomes more complex because selection on one trait can cause correlated changes in other traits. The covariance structure among traits therefore becomes a central component of the selection index.
- The ultimate goal is not simply to maximize the index value but to maximize expected improvement in the breeding objective. A well-designed index should therefore provide a high correlation between index value and aggregate breeding objective.
- The rate of genetic improvement also depends on selection intensity, accuracy, genetic variation, and generation interval. A commonly used conceptual expression is:
- ΔG/year = i × r × σ_A / L
- where i is selection intensity, r is selection accuracy, σ_A is additive genetic standard deviation, and L is generation interval.
- Selection index methodology primarily contributes to improving the accuracy and direction of selection, but it interacts with the other components of genetic gain. Increasing accuracy can increase genetic progress, while reducing generation interval can increase annual progress. However, excessive selection intensity or rapid turnover can create undesirable consequences for genetic diversity and population structure.
- This makes genetic diversity an essential consideration in selection programmes. A highly accurate index can produce substantial short-term genetic gain, but using only a small number of elite animals can increase relatedness and inbreeding.
- Genetic concentration may occur when a small number of influential sires or families contribute disproportionately to future generations. The popular sire effect is a well-known example. A highly successful sire may produce very large numbers of offspring, causing his genes to become widespread throughout the population.
- Such concentration can reduce effective population size and increase the probability that related animals will be mated in future generations.
- The approximate relationship between effective population size and the rate of inbreeding in an idealized population is:
- ΔF ≈ 1 / (2Ne)
- where ΔF is the change in inbreeding per generation and Ne is the effective population size.
- Selection index methodology should therefore be combined with appropriate population-management strategies. Optimal contribution selection (OCS) can be used to balance expected genetic gain against the contribution of animals to future generations. Instead of simply selecting the animals with the highest index values, OCS can restrict or optimize their genetic contributions to maintain diversity.
- Mate allocation can also be used after selection. The objective is to avoid undesirable matings while making efficient use of selected animals. The expected inbreeding of offspring from a particular sire-dam combination can be related to their coancestry:
- E(F_offspring) = φ(sire, dam)
- where φ represents the coefficient of coancestry between the prospective parents.
- The coefficient of relationship between two animals is approximately twice their coefficient of coancestry:
- r ≈ 2φ
- These relationships allow breeding programmes to combine selection decisions with mating decisions.
- Genomic information provides additional opportunities for managing diversity. Genomic relatedness can identify animals that are more closely related than their pedigree records suggest. Runs of homozygosity (ROH) can provide information about recent and historical autozygosity and can be used as one indicator of genomic inbreeding.
- A commonly used measure is:
- F_ROH = Total length of ROH / Total autosomal genome length
- Monitoring genomic diversity alongside index-based selection can help breeding programmes achieve genetic improvement while reducing unintended increases in homozygosity.
- Selection indexes can also be integrated with genetic management of deleterious recessive variants. Genetic testing and genomic information can identify carriers of known harmful variants. Breeding programmes can then use appropriate mating strategies to reduce the probability of affected offspring without unnecessarily eliminating valuable genetic diversity from the population.
- This is particularly important because eliminating every carrier immediately may reduce genetic diversity and increase genetic concentration. In some cases, controlled mating and gradual management of genetic load are more sustainable than rapid elimination.
- Selection index methodology also has an important role in low-heritability traits. When a trait has low heritability, individual phenotype may be a poor indicator of breeding value. However, information from relatives, progeny, repeated records, correlated traits, and genomic markers can increase accuracy.
- For example, fertility may have relatively low heritability but high economic and biological importance. A selection index can combine direct fertility records with correlated production, reproductive, health, and genomic information to improve selection decisions.
- The same principle applies to disease resistance, survival, longevity, welfare, stress resistance, heat tolerance, and adaptation traits. Genetic improvement in these characteristics can be slower when direct measurement is difficult, but structured information systems and multi-trait genetic evaluation can make them suitable for selection.
- Selection indexes are also useful when direct measurement of a trait is expensive. Modern breeding programmes increasingly use automated sensors, precision livestock technologies, imaging systems, electronic identification, activity monitors, milk sensors, and other high-throughput phenotyping systems. These technologies can generate large quantities of information that can be incorporated into genetic evaluation and selection indexes.
- However, more information does not automatically mean better selection. Data must be accurate, appropriately standardized, genetically informative, and correctly modeled. Contemporary groups are particularly important because animals should be compared with appropriate peers exposed to similar environmental conditions.
- Management, nutrition, housing, disease exposure, season, farm, age, sex, parity, and other environmental factors can influence phenotypic records. Genetic evaluation must separate these systematic environmental effects from genetic differences as effectively as possible.
- The design of a selection index should therefore begin with a clear breeding objective and a careful assessment of the available data. The breeder should identify the traits that matter, determine which traits are economically or biologically important, evaluate genetic correlations, identify reliable information sources, and select an appropriate statistical model.
- A practical selection index development process can be viewed as a sequence. First, define the long-term breeding objective. Second, identify the traits included in that objective. Third, assign appropriate economic or biological weights. Fourth, identify the available selection criteria and information sources. Fifth, estimate genetic and phenotypic variances and covariances. Sixth, calculate index coefficients. Seventh, evaluate selection accuracy and expected response. Eighth, assess the effects on genetic diversity, inbreeding, health, fertility, welfare, and other long-term outcomes. Finally, monitor realized genetic change and update the index when the breeding objective or production environment changes.
- Selection indexes should not be considered fixed forever. A breeding programme may begin with a strong emphasis on production but later need greater emphasis on fertility, longevity, disease resistance, welfare, climate adaptation, or feed efficiency. Changes in market conditions or production systems may also alter the relative importance of traits.
- The success of an index should therefore be evaluated using both genetic and practical outcomes. Important indicators may include genetic trends, realized response, selection accuracy, fertility, health, survival, inbreeding, effective population size, economic performance, and welfare outcomes.
- A major advantage of selection index methodology is that it makes trade-offs explicit. When two traits have antagonistic genetic relationships, the index forces the breeding programme to decide how much improvement in one trait is worth accepting in exchange for change in another. This makes selection decisions more transparent and consistent.
- Selection indexes also help prevent excessive emphasis on easily measured traits. A trait that is cheap and highly heritable may dominate selection if the breeding programme relies only on simple phenotypic ranking. A properly designed index can ensure that difficult but important traits such as fertility, health, welfare, longevity, and adaptation receive appropriate weight.
- The relationship between selection index, BLUP, and genomic selection is therefore complementary rather than competitive. BLUP and genomic evaluation provide accurate estimates of genetic merit from large amounts of information, while the selection index can combine these estimates according to the breeding objective.
- In advanced breeding programmes, a selection index may therefore operate at the final decision stage. Genetic evaluation produces EBVs or GEBVs for many traits, and the selection index combines those values into an overall ranking that reflects the breeding objective.
- The same framework can be extended to different species and production systems. In dairy cattle, it can balance production, fertility, udder health, longevity, feed efficiency, and welfare. In beef cattle, it can combine growth, carcass traits, maternal ability, fertility, survival, and feed efficiency. In pigs, it can balance growth, feed efficiency, carcass quality, reproduction, survival, and robustness. In poultry, it can combine egg production, egg quality, fertility, hatchability, health, and feed efficiency. In sheep and goats, it can include growth, carcass characteristics, reproduction, wool or fibre traits, disease resistance, survival, and adaptation.
- The exact coefficients and traits will differ among populations, but the underlying principle remains the same: select animals according to their expected contribution to the overall breeding objective rather than according to one trait alone.
- Selection index methodology is therefore one of the most important tools in modern animal breeding. It connects breeding objectives with measurable selection criteria and provides a quantitative framework for combining information from phenotypes, relatives, progeny, pedigrees, repeated records, indicator traits, EBVs, GEBVs, and genomic data.
- When properly designed, a selection index can increase the accuracy and efficiency of genetic selection while balancing production, reproduction, health, welfare, adaptation, and sustainability. Its greatest value is not simply producing animals with higher performance, but directing genetic change toward a balanced and clearly defined long-term breeding objective.
- The most effective breeding programmes therefore combine selection index methodology, accurate genetic evaluation, appropriate selection intensity, suitable generation intervals, genomic information, genetic diversity management, mate allocation, and continuous monitoring. The goal is sustainable genetic improvement that increases desirable performance without compromising fertility, health, survival, welfare, adaptability, or the long-term genetic health of the population.
