- Variance and covariance are important statistical concepts in animal breeding and quantitative genetics. They help researchers and breeders understand the amount of variation present in animal populations and how different traits vary together. Animals within the same breed often differ in body weight, milk production, growth rate, fertility, disease resistance, feed efficiency, and other economically important characteristics. These differences arise from genetic factors, environmental conditions, and interactions between genes and the environment. Statistical methods such as variance and covariance help describe these differences and provide a foundation for genetic evaluation and selection.
- Variance measures how much individual observations differ from the mean of a population or sample. When animals have similar values for a particular trait, the variance is low. When their values are widely spread around the mean, the variance is high. For example, if a group of cattle has similar birth weights, the variance in birth weight will be relatively low. If the animals have widely different birth weights, the variance will be higher. Understanding this variation is important because animal breeders need to know how much differences exist among animals before deciding which individuals should be selected for breeding.
- Variance is calculated by determining the difference between each observation and the mean, squaring these differences, and calculating their average or an appropriate sample estimate. Squaring the differences prevents positive and negative deviations from cancelling each other out and gives greater weight to larger deviations. For a population, variance is commonly represented by the symbol 2, while sample variance is represented by s2. Population variance is calculated by dividing the sum of squared deviations by the total number of observations, whereas sample variance is generally calculated by dividing by one less than the sample size. This adjustment helps provide an unbiased estimate of population variance when data are collected from a sample.
- Variance is closely related to standard deviation, which is the square root of variance. While variance is expressed in squared units, standard deviation is expressed in the original units of the trait and is often easier to interpret. For example, if the variance of body weight is expressed in kilograms squared, the standard deviation is expressed in kilograms. Both measures are useful for describing the distribution of animal traits, identifying variation within herds, and comparing the consistency of performance among groups of animals.
- μ-σ+σValue of XRelative densityVar(X)=E(X2)-[E(X)]2
- In quantitative genetics, variance has a particularly important role because observed differences among animals can be separated into genetic and environmental components. Phenotypic variance is the total variance observed in a measurable trait, such as milk yield, body weight, or egg production. It can be influenced by genetic differences, environmental conditions, and interactions between genotypes and environments. Under a simple model, phenotypic variance can be represented as the sum of genetic variance and environmental variance, although more detailed models may include additional components and covariance terms.
- Genetic variance refers to the portion of variation associated with genetic differences among animals. It is commonly divided into additive genetic variance, dominance variance, and epistatic variance. Additive genetic variance is especially important in animal breeding because it reflects the variation associated with the average effects of genes passed from parents to offspring. This component helps determine how effectively selection can improve a trait across generations. Dominance variance arises from interactions between alleles at the same gene location, while epistatic variance arises from interactions between genes at different locations. Environmental variance results from factors such as nutrition, housing, temperature, disease exposure, management practices, and access to resources.
- The amount of additive genetic variance present in a population influences the potential response to selection. If a trait has sufficient additive genetic variance, animals can differ genetically in ways that allow breeders to make progress by selecting superior individuals. If additive genetic variance is very low, selection may produce slower improvement, although genetic variation can sometimes be maintained or introduced through appropriate breeding strategies. For this reason, estimating genetic variance is an important step in designing long-term breeding programmes and monitoring genetic diversity.
- Heritability is another important concept related to variance. It describes the proportion of phenotypic variance attributable to genetic differences, with narrow-sense heritability specifically referring to the proportion attributable to additive genetic variance. Traits with higher heritability generally allow more accurate selection based on an animal’s own performance, provided the records and breeding population are suitable. Traits with lower heritability may require information from relatives, repeated measurements, progeny records, or genomic evaluations. However, heritability is specific to a population and environment, and it should not be interpreted as the percentage of an individual animal’s trait that is genetically determined.
- Covariance, in contrast, measures how two variables vary together. While variance describes variation in one trait, covariance describes the joint variation between two traits. A positive covariance indicates that individuals with values above the mean for one trait tend, on average, to have values above the mean for the other trait. A negative covariance indicates that individuals with above-average values for one trait tend to have below-average values for the other. A covariance close to zero indicates little or no linear joint variation, although a nonlinear relationship may still exist.
- For example, body weight and height may show positive covariance in a population of growing cattle because heavier animals tend to be taller. Milk yield and milk protein yield may also show positive covariance because cows producing more milk often produce more total protein. In contrast, some production and fitness traits may show an unfavorable relationship. For example, under certain management and breeding conditions, selecting strongly for high production without accounting for fertility or health may be associated with poorer reproductive performance. The actual direction and strength of these relationships depend on the population, the traits measured, and the environmental conditions.
- Covariance is calculated by examining how observations for two traits deviate from their respective means. For each animal, the deviation in the first trait is multiplied by the deviation in the second trait. These products are then averaged for a population or used to estimate sample covariance. When both traits tend to be above their means in the same animals, or below their means in the same animals, the products tend to be positive. When one trait tends to be above its mean while the other is below its mean, the products tend to be negative. Unlike correlation, covariance is measured in the product of the units of the two traits, so its numerical size depends on the scales used to measure them.
- Correlation is closely related to covariance and is often used alongside it. The correlation coefficient is obtained by standardizing covariance using the standard deviations of the two traits. It ranges from −1 to +1 and describes the direction and strength of a linear relationship. A positive correlation indicates that the traits tend to increase together, while a negative correlation indicates that an increase in one trait is associated with a decrease in the other. A correlation near zero indicates a weak linear relationship. Correlation is easier to compare across traits measured in different units because it is dimensionless, whereas covariance retains information about the original measurement scales.
- In animal breeding, genetic covariance is especially important because it describes how genetic effects influencing one trait are associated with genetic effects influencing another. Genetic covariance between two traits may be positive, negative, or close to zero. A positive genetic covariance between growth rate and mature body weight, for example, would suggest that genetic effects associated with faster growth tend to be associated with greater mature weight in the population being studied. A negative genetic covariance between two traits may indicate a potential trade-off, such as one trait improving while another tends to decline under selection. These relationships must be estimated from appropriate data rather than assumed to be the same across breeds or production systems.
- Genetic covariance is central to understanding genetic correlations. The genetic correlation between two traits describes the relationship between their additive genetic effects and ranges from −1 to +1. A strong positive genetic correlation means that genes associated with higher values of one trait tend to be associated with higher values of the other. A strong negative genetic correlation means that genes associated with higher values of one trait tend to be associated with lower values of the other. When the genetic correlation is close to zero, the additive genetic effects of the traits have little linear association. Genetic correlations are useful for predicting correlated responses to selection, especially when breeders select for one trait but also want to understand how another trait may change.
- For instance, a breeding programme designed to increase milk production should consider how milk yield is genetically related to fertility, udder health, longevity, and feed efficiency. If the traits have unfavorable genetic correlations, selection based solely on milk production could lead to undesirable changes in other characteristics. Breeders can reduce this risk by using selection indices that combine estimated breeding values for several traits and assign appropriate economic or management-based weights to them. This approach supports balanced improvement rather than maximizing a single trait at the expense of overall animal performance and welfare.
- Variance and covariance are also essential for estimating breeding values. Modern genetic evaluations use statistical models that account for relationships among animals, repeated performance records, environmental effects, and the genetic architecture of traits. Variance components help estimate the amount of genetic and environmental variation, while covariance components describe relationships between traits and sometimes between different records of the same trait. These estimates are used in best linear unbiased prediction and other evaluation methods to estimate the genetic merit of animals. Genomic evaluations can also use genetic relationships and information from DNA markers, but reliable estimates of variance and covariance remain important for building and validating these systems.
- In practical livestock production, variance and covariance can help breeders evaluate the consistency of performance, identify sources of variation, and design more effective selection strategies. A high variance in a trait may indicate substantial differences among animals, but it does not automatically mean that the trait can be improved genetically. The variation may be mostly environmental rather than genetic. Similarly, a positive covariance between two traits does not establish that one trait causes the other to change. Breeders must distinguish statistical relationships from causal mechanisms and use suitable genetic and environmental models to interpret their data.
- The estimation of variance and covariance requires accurate and representative records. Data quality can be affected by measurement errors, missing observations, unequal management conditions, differences in animal age, and inappropriate grouping of animals. For example, comparing milk yields without accounting for lactation stage, parity, herd management, and other relevant factors may lead to misleading conclusions. Appropriate statistical models help separate genetic differences from systematic environmental effects and improve the reliability of estimates. Large and well-designed datasets, accurate pedigrees, genomic information when available, and consistent recording systems can strengthen genetic evaluations.
- Variance and covariance also contribute to the conservation of genetic diversity. Genetic variation provides the raw material for future selection and adaptation, while relationships among traits can help identify possible consequences of intensive selection. Excessive emphasis on a narrow set of traits or a small number of highly used breeding animals can reduce genetic diversity and increase inbreeding. Breeding programmes should therefore balance genetic improvement with the maintenance of diversity, fertility, health, resilience, and adaptation to changing environments.
- In conclusion, variance and covariance are fundamental statistical tools in animal breeding and quantitative genetics. Variance measures the degree of variation within a single trait, while covariance describes how two traits vary together. Their genetic components help breeders understand the inheritance of economically important characteristics, estimate breeding values, predict correlated responses to selection, and develop balanced breeding objectives. When combined with reliable data, appropriate statistical models, and sound breeding strategies, these concepts support sustainable genetic improvement and help livestock populations remain productive, healthy, and adaptable over generations.
