- Estimated Breeding Values (EBVs) are statistical predictions of the genetic merit of animals for specific traits and are among the most important tools used in modern animal breeding. An EBV estimates the additive genetic value that an animal is expected to transmit to its offspring. Because the true breeding value of an animal cannot normally be observed directly, breeders use information from phenotypic records, relatives, pedigree, progeny, repeated measurements, genetically correlated traits, and genomic data to estimate it. EBVs therefore provide a practical basis for ranking animals and making genetic selection decisions.
- The concept of an EBV is closely related to the distinction between phenotypic value, genotypic value, true breeding value, and estimated breeding value. An animal’s phenotype is the observed expression of a trait and is influenced by genetics and environmental conditions. Genotypic value represents the total genetic contribution, including additive, dominance, and epistatic effects. The true breeding value represents the additive genetic component that is expected to be transmitted predictably to offspring, while the EBV is the statistical estimate of that true breeding value based on available information.
- The relationship between phenotype, genetics, and environment can be represented as P = G + E, where P is phenotypic value, G is genetic value, and E represents environmental effects. Genetic value can be separated into additive, dominance, and epistatic components. EBVs focus primarily on the additive component because additive genetic effects are transmitted from parents to offspring in a predictable statistical manner. This makes EBVs especially useful for selection decisions aimed at changing the genetic composition of future generations.
- The distinction between true breeding value and EBV is fundamental. The true breeding value is a theoretical quantity describing an animal’s actual additive genetic merit, whereas the EBV is an estimate of that quantity. The true breeding value is not directly observable, but an EBV can be calculated using statistical genetic evaluation methods. As more and better information becomes available, the EBV can be updated and generally becomes more reliable.
- An EBV is therefore not simply the observed performance of an animal. An animal may have a high phenotypic value because it received excellent nutrition, favourable management, low disease exposure, or other environmental advantages. Another animal may have a lower phenotype because it experienced poorer environmental conditions but may actually possess a higher genetic merit. Genetic evaluation attempts to separate these environmental effects from genetic differences so that animals can be ranked according to their predicted breeding value.
- Additive genetic variance is the genetic foundation underlying EBVs. It represents the variation among animals in additive genetic effects and therefore the variation in breeding values. When substantial additive genetic variance exists, animals differ in their genetic potential and selection can change the population mean. If additive genetic variance becomes very small, the potential response to selection is reduced because there are fewer genetically different animals from which to select.
- Narrow-sense heritability is closely related to the usefulness of individual phenotypic records for predicting EBVs. It is expressed as h² = σ²_A / σ²_P, where σ²_A is additive genetic variance and σ²_P is phenotypic variance. High heritability generally means that an animal’s own phenotype contains more information about its breeding value, whereas low heritability means that environmental effects contribute more strongly to observed differences. However, low heritability does not mean that a trait cannot be improved genetically because EBVs can incorporate information from relatives, progeny, repeated records, correlated traits, and genomic information.
- One of the main advantages of EBVs is that they can combine information from many sources. An evaluation may include the animal’s own performance, its parents, full and half siblings, offspring, more distant relatives, repeated records, genetically correlated traits, pedigree relationships, and genomic markers. Combining these sources provides a more complete picture of genetic merit than relying on any single measurement.
- Contemporary groups are particularly important when estimating EBVs from performance records. Animals should be compared with appropriate contemporaries that have been exposed to similar environmental conditions, management, feeding, housing, climate, health conditions, and other systematic influences. Proper adjustment for contemporary groups helps prevent environmental differences from being incorrectly interpreted as genetic differences.
- For example, suppose two animals have different growth rates but were raised under different feeding conditions. Simply ranking the animal with the greater weight could result in an incorrect selection decision. A genetic evaluation model attempts to account for the systematic environmental differences and determine how much of the observed difference is consistent with differences in additive genetic merit. The resulting EBV can therefore provide a more useful basis for selection than the raw phenotype.
- Pedigree information is another major source of information for EBVs. Animals that share common ancestors are expected to share some of their genes, so the performance of relatives provides information about an individual’s likely breeding value. Pedigree-based evaluations commonly use a relationship matrix, often represented as the A matrix, to describe expected additive genetic relationships among animals. The accuracy of pedigree-based EBVs depends on the completeness and correctness of the pedigree as well as the amount and quality of phenotypic information available.
- Information from parents can provide an initial prediction of an animal’s genetic merit. If both parents have reliable EBVs, their genetic information can contribute substantially to the prediction for their offspring. However, parental information does not completely determine the offspring’s breeding value because of Mendelian sampling. Each offspring receives a different random sample of parental alleles, meaning that full siblings can have different breeding values even though they have the same parents.
- Family information can improve EBVs when an animal has limited individual performance data. Records from full siblings, half siblings, cousins, and other relatives provide additional evidence about the genetic merit of the selection candidate. Family information is particularly useful for traits that are difficult to measure directly, expressed late in life, expressed in only one sex, expensive to record, or unsuitable for direct measurement on elite breeding animals.
- Progeny testing can provide especially strong information for EBVs. If a sire produces many offspring and those offspring perform consistently well for a particular trait, their performance provides evidence about the sire’s genetic merit. Progeny records are valuable because they represent the genetic contribution of the evaluated parent expressed in multiple descendants. Traditional progeny testing has been particularly important for traits such as milk production and other sex-limited traits, although genomic selection can now provide useful information much earlier in the animal’s life.
- Repeated measurements can also contribute to EBV estimation. For traits that are expressed multiple times, such as milk yield, egg production, body weight, or repeated reproductive records, several observations can provide more information than a single measurement. Repeatability describes the consistency of repeated records within an individual due to permanent differences among animals. Genetic evaluation models can account for both additive genetic effects and permanent environmental effects when using repeated records.
- Some traits are difficult to measure directly but can be predicted using indicator traits. If two traits have a sufficiently strong genetic correlation, information about the easier-to-measure trait can contribute to the prediction of breeding value for the more difficult trait. This approach is particularly useful for traits that require expensive phenotyping, slaughter, specialized equipment, long-term observation, or challenging field measurements.
- The genetic correlation between two traits can be expressed as r_A = Cov_A(X,Y) / (σ_A,X × σ_A,Y), where Cov_A(X,Y) is additive genetic covariance and σ_A,X and σ_A,Y are the additive genetic standard deviations of the two traits. Genetic correlations are important because selection for one trait can produce correlated genetic changes in another. Modern multi-trait EBV systems can use these relationships to improve prediction accuracy and manage favourable or unfavourable genetic relationships among traits.
- BLUP, or Best Linear Unbiased Prediction, is one of the most widely used approaches for estimating breeding values. BLUP combines information from multiple animals and multiple sources while accounting for fixed environmental effects and genetic relationships. In a general animal model, the statistical structure can be written as y = Xb + Za + e, where y represents observations, b represents fixed effects, a represents random additive genetic effects or breeding values, and e represents residual effects. The model uses the available data to predict the genetic merit of individual animals.
- The animal model is particularly powerful because every animal in the connected evaluation population can receive an EBV based on its own records and the records of relatives. The model can account for relationships across generations and can simultaneously adjust for systematic environmental effects. This allows EBVs to be estimated for young animals with limited information as well as older animals with extensive performance and progeny records.
- Modern genetic evaluations increasingly incorporate genomic information. Genomic markers provide information about the genetic similarity between animals and can improve the prediction of additive genetic merit. A genomic relationship matrix, commonly called the G matrix, describes genomic relationships among animals based on marker information. When genomic and pedigree information are combined, genetic evaluations can produce genomic estimated breeding values (GEBVs).
- Genomic selection is particularly useful for young animals because genomic information can provide evidence about genetic merit before extensive performance or progeny records are available. This can increase selection accuracy at an early age and allow breeders to identify promising animals earlier. Early selection can also reduce the generation interval, potentially increasing the rate of genetic improvement.
- The accuracy of an EBV describes how closely the estimated breeding value is expected to correspond to the animal’s true breeding value. A high-accuracy EBV gives greater confidence that the animal’s genetic ranking is close to its actual genetic merit, whereas a low-accuracy EBV contains greater uncertainty. Accuracy can increase as more records become available, as relatives and progeny are added to the evaluation, and as genomic information contributes additional evidence.
- The distinction between EBV and reliability is also important. Accuracy is commonly expressed as a correlation between the EBV and the true breeding value, while reliability is often related to the squared accuracy. In simplified terms, reliability indicates the proportion of variation in the true breeding value that is accounted for by the prediction. Breeders should consider both the EBV itself and its reliability when making selection decisions.
- Young animals often have lower reliability because they have limited individual and progeny information. An older sire with hundreds or thousands of informative offspring may have a much more reliable EBV than a young animal with only genomic and pedigree information. However, genomic selection can substantially improve the early evaluation of young animals and reduce the need to wait for extensive progeny testing.
- EBVs are also influenced by the genetic base used in an evaluation system. Breeding values are generally expressed relative to a defined population or genetic base, so the numerical value of an EBV should be interpreted within the appropriate evaluation system. Changes in the genetic base can change the numerical scale or average of EBVs without necessarily representing an actual change in the animal’s underlying genetic merit. This is important when comparing EBVs across different populations, breeds, or evaluation systems.
- EBVs are commonly expressed in the units of the trait or on a scale related to the expected genetic contribution to offspring. For example, an EBV for body weight may be expressed in kilograms, while an EBV for milk production may be expressed in kilograms of milk. The interpretation of the scale depends on the specific evaluation system. Breeders should therefore understand what one unit of EBV means and whether the value represents the animal’s breeding value or the expected difference in progeny performance relative to a reference.
- Because an offspring receives approximately half of its additive genetic material from each parent, the expected genetic contribution of a parent to its offspring is related to one-half of its breeding value. Thus, if an animal has a breeding value of +20 units for a trait, its expected average contribution to the breeding value of its offspring is approximately +10 units relative to the relevant population reference, assuming the scale is interpreted appropriately. This does not mean every offspring will show exactly half of the parent’s EBV because Mendelian sampling and the other parent’s genetic contribution create differences among offspring.
- EBVs are particularly useful for selection objectives involving several traits. A breeding programme may aim to improve production while maintaining fertility, health, survival, feed efficiency, welfare, and adaptation. Selecting animals according to a single EBV can create undesirable correlated responses if traits are genetically related. Therefore, breeders often combine several EBVs through a selection index or another multi-trait decision-making system.
- A general selection index can be expressed as I = b₁x₁ + b₂x₂ + … + bₙxₙ, where x values represent available selection information and b values are coefficients that determine how the information contributes to the index. The index can combine EBVs or other information according to the overall breeding objective. This allows breeders to select animals based on balanced genetic merit rather than simply choosing the animal with the highest value for one trait.
- A breeding objective can be represented as H = a₁A₁ + a₂A₂ + … + aₙAₙ, where A values represent breeding values for the traits included in the objective and a values represent their relative importance. In practice, the true breeding values are unknown, so EBVs are used as predictions of the genetic merit underlying the breeding objective. Economic weights, biological importance, welfare considerations, sustainability goals, and production-system requirements can all influence the weighting of traits.
- EBVs can be used for production traits such as growth rate, body weight, milk production, milk composition, meat production, carcass traits, egg production, egg quality, wool, fibre, feed intake, and feed efficiency. They can also be estimated for reproductive traits, including fertility, age at sexual maturity, litter size, calving, lambing, farrowing, semen quality, and maternal performance. The usefulness of an EBV depends on the quality of the phenotypic data and genetic evaluation system supporting the trait.
- Increasingly, EBVs are also used for health traits, disease resistance, disease susceptibility, immune function, survival, longevity, temperament, welfare-related traits, adaptation, heat tolerance, and stress resistance. These traits may be more difficult to measure and may be strongly affected by environmental conditions, but they can still have meaningful additive genetic variation. Appropriate data collection and statistical modelling are essential for obtaining useful EBVs for such complex traits.
- For binary and categorical traits, such as disease status, survival, pregnancy outcome, or certain reproductive outcomes, genetic evaluation may use threshold models or other statistical approaches designed for non-continuous observations. The observed phenotype may be recorded as categories, but the underlying genetic liability can be treated as a continuous variable. EBVs can then be used to rank animals according to predicted genetic liability or genetic merit for the trait.
- Maternal effects, common environmental effects, and permanent environmental effects must also be considered when appropriate. For example, early growth can be influenced by the offspring’s own genes as well as the dam’s milk production, maternal behaviour, uterine environment, and other maternal influences. If these effects are ignored, the EBV for the offspring’s own genetic merit may be biased. Modern genetic evaluation models can separate these different sources of variation when sufficient data are available.
- Genotype–environment interaction (G×E) can also influence the interpretation and usefulness of EBVs. Animals may rank differently under different climates, feeding systems, disease pressures, housing conditions, or management systems. For traits such as heat tolerance, disease resistance, adaptation, and resilience, genetic evaluations may benefit from data collected across the environments in which animals are expected to perform.
- The quality of an EBV depends heavily on the quality of the underlying data. Accurate phenotypic records, reliable pedigrees, appropriate contemporary groups, consistent trait definitions, sufficient sample sizes, and suitable statistical models are essential. For genomic evaluations, the quality and relevance of the reference population are particularly important because genomic prediction depends on relationships between genotypes and reliable phenotypic or breeding-value information.
- EBVs can change when new information becomes available. A young animal may initially receive an EBV based largely on pedigree and genomic information. As it grows and receives its own performance records, the EBV may change. When its offspring are recorded, additional evidence can further improve or alter the estimate. Therefore, an EBV should be viewed as an evolving statistical prediction rather than a permanent biological measurement.
- The use of EBVs has transformed animal breeding from selection based primarily on visible performance toward selection based on predicted genetic merit. Phenotypic selection remains useful, but EBVs provide a way to account for environmental effects and integrate information that cannot be observed directly in the selection candidate. This is especially important for traits with low heritability, traits expressed late in life, sex-limited traits, difficult-to-measure traits, and traits for which genomic information is available.
- EBVs also play an important role in managing genetic diversity. Selecting only animals with the highest EBVs can increase genetic concentration if a small number of elite animals are used excessively. This can increase relatedness and the rate of inbreeding. Therefore, modern breeding programmes increasingly combine EBV-based selection with genetic relationship information, mate allocation, and optimal contribution selection to balance genetic gain and long-term population health.
- The expected inbreeding of an offspring from a particular mating can be expressed as E(F_offspring) = φ(sire, dam), where φ is the coefficient of coancestry between the parents. The expected additive relationship between two animals is approximately r ≈ 2φ. These relationships can be considered when selecting breeding pairs so that animals with high EBVs are not repeatedly mated to closely related individuals.
- The effective population size is another important consideration. Under simplified assumptions, the expected increase in inbreeding per generation can be approximated as ΔF ≈ 1 / (2Ne), where Ne is the effective population size. A small effective population size can result in faster accumulation of inbreeding and loss of genetic diversity. Consequently, the goal of genetic evaluation should not be simply to maximize EBVs but to achieve useful genetic progress while maintaining a healthy and sufficiently diverse breeding population.
- Genomic information can also be used to monitor runs of homozygosity (ROH) and genomic relatedness. ROH are long homozygous segments that can provide information about autozygosity and common ancestry. A genomic measure of homozygosity can be expressed as F_ROH = Total length of ROH / Total autosomal genome length. Such information can complement pedigree-based relationships and help breeders manage inbreeding risk when making mating decisions.
- EBVs are also useful in managing deleterious genetic variants. Harmful recessive alleles can remain hidden in heterozygous carriers and may become expressed when carriers are mated. Genomic testing can identify known variants, while relationship information and mate allocation can reduce the probability of producing affected offspring. This allows breeders to manage genetic health while avoiding unnecessary loss of valuable genetic diversity.
- The relationship between EBVs and genetic gain can be described through the factors that influence the rate of selection response. A commonly used 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. Increasing EBV accuracy increases the likelihood that selected animals have genuinely superior genetic merit, while reducing generation interval can accelerate the transfer of favourable genes to future generations.
- The practical importance of EBVs therefore extends beyond ranking animals. EBVs help breeders make decisions about which animals should become parents of the next generation, which animals should be used more or less intensively, how different traits should be balanced, and which mating combinations can achieve genetic progress while controlling inbreeding. They form the quantitative foundation of many modern breeding programmes.
- Ultimately, Estimated Breeding Values (EBVs) provide a practical prediction of the additive genetic merit of animals when the true breeding value cannot be observed directly. By combining individual performance, relatives, pedigree, progeny, repeated records, correlated traits, and genomic information, genetic evaluation systems can produce increasingly accurate predictions of genetic merit. EBVs allow breeders to move beyond phenotype-based selection and make decisions based on the expected genetic contribution of animals to future generations.
- The effective use of EBVs requires more than simply choosing animals with the highest numbers. Breeders must consider selection accuracy, reliability, genetic correlations, breeding objectives, economic importance, environmental conditions, generation interval, genetic diversity, inbreeding, and long-term sustainability. When these factors are integrated appropriately, EBVs provide one of the most powerful tools available for improving production, fertility, health, welfare, adaptation, resilience, and overall genetic performance in animal populations.
