- Repeatability is a concept in quantitative genetics that describes the consistency of repeated measurements of the same trait in the same individuals. It is particularly useful when a trait can be measured more than once during an individual’s lifetime, such as body weight, milk production, egg production, litter size, growth rate, behavior, disease resistance, or reproductive performance. Repeatability helps determine how reliably an individual’s performance at one time predicts its performance at another time and provides information about the sources of variation underlying repeated observations.
- In quantitative genetics, repeatability is commonly defined as the proportion of total phenotypic variance that is attributable to permanent differences among individuals. A simplified expression is r=VI/VPr = V_I/V_P, where rr is repeatability, VIV_I represents permanent individual-specific variance, and VPV_P is total phenotypic variance. Permanent individual differences can include additive genetic variance, dominance variance, permanent environmental effects, and other persistent sources of individual variation. Consequently, repeatability is generally broader than narrow-sense heritability.
- A useful way to understand repeatability is to consider repeated measurements of a trait. Suppose an animal’s body weight is measured several times. Some differences between animals may be caused by their genes, while other differences may result from permanent environmental conditions experienced throughout their lives. Temporary environmental effects, measurement error, disease episodes, weather, nutrition, and other short-term influences can cause additional variation between measurements. Repeatability reflects the proportion of total variation that is due to differences that remain relatively consistent within individuals across repeated measurements.
- Repeatability is therefore different from heritability. Narrow-sense heritability is the proportion of phenotypic variance attributable specifically to additive genetic variance, h2=VA/VPh^2 = V_A/V_P. Repeatability includes additive genetic effects plus other permanent sources of individual differences. In a simplified model, repeatability can be expressed as r=(VA+VPERM)/VPr = (V_A + V_{PERM})/V_P, where VPERMV_{PERM} represents permanent non-additive genetic and permanent environmental effects not already included in the additive component. Because repeatability can include these additional sources of persistent variation, it is generally greater than or equal to narrow-sense heritability when estimated under comparable conditions.
- The distinction between permanent environmental effects and temporary environmental effects is particularly important. A permanent environmental effect can influence an individual’s performance over a long period. For example, an early-life nutritional limitation, chronic developmental condition, or permanent injury may affect an animal’s performance across multiple measurements. A temporary environmental effect, such as short-term illness or a brief change in weather, may affect only one measurement. Repeatability captures persistent individual differences but not purely temporary deviations.
- Repeatability can also be understood through covariance among repeated measurements. If two measurements of the same trait on the same individual are strongly correlated, the trait has high repeatability. If measurements fluctuate substantially from one occasion to another, repeatability is lower. In statistical terms, repeatability is closely related to the intraclass correlation coefficient (ICC), which measures the similarity of observations belonging to the same individual or grouping unit.
- For two repeated measurements, repeatability can be expressed conceptually as the correlation between measurements taken on the same individual. More generally, it is the ratio of between-individual variance to total variance. A high repeatability means that individuals tend to maintain their relative differences across repeated measurements, whereas a low repeatability means that temporary environmental effects and measurement noise account for a larger proportion of observed differences.
- Repeatability is especially useful in animal breeding because many economically important traits are measured repeatedly. Milk yield may be recorded across multiple lactations, body weight may be measured repeatedly during growth, egg production may be monitored across laying cycles, and reproductive traits may be recorded across successive breeding opportunities. Repeatability helps determine whether an individual’s early performance provides useful information about its future performance.
- When repeatability is high, a single observation may provide a relatively reliable indication of an individual’s long-term performance. When repeatability is low, repeated measurements can provide additional information and may be necessary before making selection decisions. This makes repeatability an important concept in determining the value of repeated records for selection and breeding value prediction.
- Repeatability also places an upper limit on the accuracy that can be achieved by selection based on repeated phenotypic records. If multiple records are collected from the same individual, the average of those records can provide a more reliable estimate of the individual’s persistent performance. As the number of repeated records increases, temporary environmental noise can be averaged out. However, the improvement in accuracy diminishes as additional records are added.
- The relationship between repeatability and the accuracy of the mean of repeated records can be expressed using the standard repeatability formula rn=nr1+(n−1)rr_n = \frac{nr}{1+(n-1)r}, where rnr_n is the reliability of the average of nn records and rr is repeatability. This relationship illustrates why repeated measurements are especially useful when repeatability is moderate or high. When repeatability is already very high, additional measurements provide relatively little improvement.
- Repeatability is also closely related to breeding value. An individual’s observed phenotype contains information about its genetic potential, but the phenotype may also contain environmental effects. When repeated records are available, averaging those records can reduce the influence of temporary environmental variation and provide better information about the individual’s persistent performance. However, repeatability itself should not be interpreted as equivalent to breeding value or additive genetic variance.
- The relationship between repeatability and heritability can be represented using a variance-component framework. Suppose phenotypic variance is partitioned into additive genetic variance VAV_A, other permanent individual variance VP∗V_P^{*}, temporary environmental variance VEV_E, and residual or measurement variance. Repeatability represents the proportion of total variance associated with persistent individual differences, whereas narrow-sense heritability represents only the additive genetic component. Therefore, repeatability can be used as an upper bound on narrow-sense heritability under appropriate assumptions.
- For example, if a trait has a repeatability of 0.70, individuals tend to maintain relatively consistent differences across repeated measurements. However, this does not mean that 70% of the trait is genetically determined. Some of the persistent differences may be caused by permanent environmental effects, dominance, or other sources of non-additive variation. Similarly, a repeatability of 0.20 does not mean that the trait is only 20% genetic. Repeatability and heritability answer different statistical questions.
- Repeatability is particularly important for traits expressed repeatedly during an individual’s lifetime. For a trait expressed only once, such as a single developmental event, conventional repeatability is generally not directly applicable because there are no repeated observations of the same trait on the same individual. In contrast, traits such as growth, production, behavior, physiological measurements, and reproductive performance may be measured repeatedly and therefore provide suitable data for repeatability analysis.
- The estimate of repeatability depends on the population and conditions under which it is measured. Differences in environmental variation, management, age structure, measurement methods, population genetics, and sampling design can influence the estimate. Repeatability should therefore be interpreted as a population- and context-specific parameter rather than an immutable characteristic of a trait.
- Genotype–environment interaction (G×E) can further complicate repeatability. If individuals respond differently to changing environments, their relative performance may change across conditions. For example, an animal that performs exceptionally well under one feeding regime may not maintain the same advantage under another. In such cases, repeatability across environments may be lower than repeatability measured under relatively stable conditions.
- Repeatability is also affected by age and developmental stage. Genetic and permanent environmental effects may change as individuals develop, and a trait measured during early life may not predict the same trait later in life with equal accuracy. Consequently, repeatability estimated during one developmental period may not necessarily apply to another.
- The estimation of repeatability commonly uses linear mixed models and variance-component methods. In these models, repeated observations from the same individual are linked through a random individual effect. The variance associated with the individual effect represents persistent differences among individuals, while residual variance represents temporary environmental effects and measurement error. Statistical approaches such as restricted maximum likelihood (REML) are frequently used to estimate these variance components.
- Pedigree and genomic information can also help separate the genetic contribution to repeatability. A pedigree relationship matrix can be incorporated into mixed models to estimate additive genetic variance, while a genomic relationship matrix can use genome-wide markers to estimate relationships among individuals. These approaches allow researchers to distinguish persistent genetic effects from permanent environmental effects more effectively than repeatability alone.
- Repeatability is closely connected to covariance among relatives, genetic covariance, and genetic correlation. Repeated measurements on the same individual provide information about consistency within individuals, whereas measurements from relatives provide information about genetic resemblance between individuals. Combining these sources of information can improve estimates of genetic variance and breeding values, particularly for traits that are difficult or expensive to measure.
- In animal and plant breeding, repeatability can help determine the optimal timing and number of records required for selection. If a trait has high repeatability, breeders may be able to make selection decisions using relatively early or limited phenotypic information. If repeatability is low, additional records, family information, pedigree information, or genomic information may be required to improve the accuracy of selection.
- Repeatability also has practical importance in genomic selection. Genome-wide markers can provide information about an individual’s genetic potential, while repeated phenotypic measurements provide information about persistent performance. Combining genomic information with repeated records can improve the accuracy of genomic estimated breeding values (GEBVs), particularly for traits with complex genetic architectures.
- Repeatability is relevant beyond agricultural breeding. In ecological and behavioral studies, researchers may repeatedly measure traits such as body size, activity, aggression, movement, reproductive behavior, or physiological responses. High repeatability can indicate consistent individual differences that may have biological significance. Such persistent differences can be important for studying personality-like traits, behavioral syndromes, life-history strategies, and individual variation in natural populations.
- In human genetics and biomedical research, repeated measurements are also common. Traits such as blood pressure, metabolic measurements, physiological characteristics, and behavioral or clinical measures may vary within individuals over time. Repeatability analysis can help distinguish stable individual differences from short-term fluctuations and measurement error. However, the interpretation depends strongly on the measurement design, population, and environmental context.
- Repeatability is also useful for understanding the quality of phenotypic measurements. A low repeatability estimate may indicate substantial temporary environmental variation or measurement error. Improving measurement precision, standardizing environmental conditions, increasing the number of observations, or collecting measurements at appropriate developmental stages can increase the reliability of phenotypic assessment.
- A key limitation is that high repeatability does not necessarily imply high evolutionary potential. Evolutionary response depends primarily on additive genetic variance, not simply on persistent phenotypic differences. A trait may have high repeatability because of a strong permanent environmental effect while having relatively little additive genetic variance. Conversely, a trait may have moderate repeatability but substantial additive genetic variation and therefore respond effectively to selection.
- The relationship between repeatability and realized heritability is also important. Repeatability describes the consistency of individual differences across repeated observations, whereas realized heritability is estimated from the observed response to selection relative to the selection differential. Repeatability can help evaluate the reliability of phenotypic selection, while realized heritability provides empirical evidence about the genetic response achieved across generations.
- Repeatability should also be distinguished from simple phenotypic correlation between measurements. A phenotypic correlation describes the association between two variables or measurements, whereas repeatability specifically evaluates the consistency of repeated measurements of the same trait within individuals. Appropriate variance-component models can separate between-individual variation from within-individual variation and therefore provide a more informative measure of consistency.
- Overall, repeatability is a fundamental concept in quantitative genetics for understanding persistent differences among individuals when traits are measured repeatedly. It incorporates additive genetic effects as well as other permanent sources of individual variation and therefore differs from narrow-sense heritability. Repeatability is commonly estimated as the proportion of phenotypic variance attributable to persistent individual differences and is closely related to the intraclass correlation coefficient.
- Repeatability is especially valuable for determining how reliably one measurement predicts future performance, how many repeated records are needed for accurate evaluation, and how effectively phenotypic information can be used for selection and breeding value estimation. When interpreted together with heritability, additive genetic variance, genetic variance, environmental variance, covariance among relatives, and genomic information, repeatability provides an important framework for understanding the consistency, inheritance, and genetic improvement of repeatedly measured traits.
