A polygenic trait is a characteristic whose variation is influenced by multiple genes or genetic loci rather than predominantly by a single locus. Each contributing allele may have a small effect, although effects need not be equal and some loci may contribute substantially more than others. Polygenic inheritance helps explain variation in many phenotypes, including human height, flowering time in plants, and milk yield in cattle. Environmental conditions commonly influence these traits as well, so a polygenic trait is not necessarily determined by its genetic components alone.
Definition and scope
“Polygenic” describes the genetic basis of variation in a trait. A locus is a position or region in the genome; it may contain a protein-coding gene, a regulatory sequence, or another functional element. Consequently, the number of associated loci is not necessarily the number of causal genes. Multiple associated variants can point to the same biological mechanism, and a variant can affect more than one gene.
Polygenic inheritance differs from a simple single-gene model, in which variation at one locus has a predominant influence on the characteristic under study. This distinction is not absolute. A trait can have a broadly polygenic basis while particular individuals carry a rare variant with a large effect. Conversely, the expression of a predominantly single-gene condition can be modified by variants elsewhere in the genome.
Several related terms describe different aspects of inheritance:
- Polygenic means that multiple loci contribute to a trait.
- Quantitative means that a trait is measured numerically, such as height, mass, or concentration.
- Multifactorial means that genetic and nongenetic factors contribute.
- Complex trait is a broader term commonly used for characteristics that do not follow a simple single-gene inheritance pattern.
These categories overlap but are not interchangeable. A polygenic trait need not be continuously distributed, and a quantitative measurement is not by itself evidence of polygenic inheritance.
Genetic mechanisms
Additive effects
In a simple additive model, each allele contributes an increment to the expected value of a trait. An individual's phenotype can be represented as
where is the measured phenotype, is a baseline value, is the number of copies of a specified allele at locus , is its estimated additive effect, and represents environmental influences and other effects not captured by the model.
This is a statistical approximation, not a claim that development consists of independent biological increments. An allele's estimated effect can depend on the population, environmental conditions, and other variants present.
When many contributions combine, the resulting phenotype may show an approximately bell-shaped distribution. Such a distribution is common for polygenic quantitative traits but is neither necessary nor sufficient evidence of polygenicity.
Dominance and interactions
Genetic effects need not be purely additive. Dominance occurs when the phenotype of a heterozygote differs from the midpoint between the corresponding homozygotes. Epistasis occurs when the effect of a variant depends on variants at other loci.
Genes can influence a trait through shared developmental pathways, proteins, or regulatory networks. Variation affecting gene expression, for example, may change when and where other genes act. Biological interactions can be extensive even when an additive statistical model predicts much of the observed variation.
A further distinction concerns gene–environment interaction: a genetic variant may have different effects under different environmental conditions. A plant allele associated with growth under adequate water supply may have a different effect during drought.
Continuous traits and threshold traits
Many familiar polygenic traits vary continuously. Human height, for example, reflects contributions from numerous genetic variants as well as nutrition, health, and developmental conditions. Agricultural traits such as crop yield and animal growth rate also commonly have polygenic components.
Polygenic influences can also underlie categorical outcomes. In a threshold model, many genetic and environmental contributions combine into an unobserved liability. A categorical outcome occurs when liability exceeds a threshold. This model is used to study some binary traits and disease susceptibilities.
The threshold is a modeling device rather than necessarily a single identifiable biological switch. Likewise, a higher genetic liability does not establish that an individual will develop a condition. The relationship between inherited susceptibility and an observed outcome depends on the condition, exposures, age, and other factors.
Variance and heritability
Quantitative genetics studies the inheritance of variation in measured traits. Its central concern is often not whether genes matter in the abstract, but how differences among individuals relate to genetic and environmental differences in a specified population.
Under simplifying assumptions, phenotypic variance can be partitioned as
where is total phenotypic variance, is genetic variance, and is environmental variance. More detailed models may include gene–environment covariance and interaction components. Genetic variance itself can be partitioned into additive, dominance, and interaction components under specified statistical assumptions.
Heritability summarizes the genetic contribution to observed variation. Broad-sense heritability is
whereas narrow-sense heritability is
with denoting additive genetic variance.
These quantities are population- and environment-specific. They do not describe what fraction of an individual's trait was “caused by genes,” nor do they measure how difficult a trait is to change. High heritability can coexist with substantial environmental effects on the population mean. Heritability also does not reveal how many loci are involved: a highly heritable trait may have either a simple or a highly polygenic genetic architecture.
Historical development
Nineteenth-century studies of continuous variation initially appeared difficult to reconcile with the discrete inheritance patterns described by Gregor Mendel. Work on crop characteristics in the early twentieth century, including Herman Nilsson-Ehle's studies of wheat kernel color, showed how several segregating factors could jointly generate multiple phenotypic classes.
In 1918, Ronald Fisher demonstrated mathematically how Mendelian inheritance at many loci could explain correlations between relatives for continuously varying traits. This helped establish quantitative genetics and connect Mendelian inheritance with the statistical study of variation.
Later developments in DNA sequencing, genetic mapping, and large-scale genotyping made it possible to investigate polygenic inheritance at the molecular level. These methods revealed that many traits involve numerous contributing regions distributed across different chromosomes.
Identifying contributing loci
Family studies and experimental crosses
Resemblance among relatives provides evidence about inherited variation, although relatives may also share environments. Pedigree, twin, and adoption designs use different comparisons and assumptions to distinguish these contributions.
In experimental organisms, controlled crosses can identify quantitative trait loci—genomic regions associated with variation in a measured characteristic. A mapped region may contain several genes, and locating it does not by itself identify the causal variant or mechanism.
Genome-wide association studies
A genome-wide association study tests whether genetic variants are statistically associated with a phenotype across many individuals. Large studies can detect small effects that would be difficult to distinguish in smaller samples.
An association is not automatically evidence that the measured variant directly causes the trait. Nearby variants may be inherited together through linkage disequilibrium, allowing a tested marker to tag a causal variant elsewhere in the region. Differences in ancestry, environment, or sampling can also complicate interpretation.
Follow-up analyses may combine association results with functional experiments, gene-expression measurements, and biological information. Distinguishing statistical prediction from biological causation remains central to interpreting polygenic findings.
Polygenic scores
A polygenic score combines information from many genetic variants, commonly as a weighted sum:
where is an estimated weight and is the allele count at variant .
Such scores summarize the component of variation captured by a particular model. They are used in research on trait prediction, genetic architecture, and inherited susceptibility. They do not capture all genetic effects, directly measure environmental influences, or determine an individual's eventual phenotype.
Performance depends on the study population, sample size, phenotype definition, and analysis method. Scores can perform less well in populations that differ from those used to estimate their weights, partly because allele frequencies and linkage disequilibrium patterns differ. For health-related outcomes, a score is not itself a diagnosis, and its practical value requires separate evaluation.
Applications and interpretive limits
In plant and animal breeding, polygenic models support selection for characteristics such as yield, growth, and product quality. Genomic prediction uses markers distributed throughout the genome to estimate breeding values, including when individual causal variants remain unknown.
In evolutionary biology, polygenic models help explain how natural selection can change a trait through shifts in allele frequencies at many loci. Such changes may be subtle at each individual locus and difficult to distinguish from demographic history or sampling effects.
Several limits are especially important:
- Genetic architecture depends on context. The variants contributing to observed variation, and their estimated effects, can differ across populations and environments.
- Undetected effects remain important. Rare variants, structural variation, small effects, and imperfect measurement can contribute to gaps between different estimates of genetic influence.
- Shared genetic associations are not necessarily direct causal relationships. Correlations between traits can arise through several mechanisms, including variants affecting multiple traits.
- Within-population results do not establish causes of between-population differences. Heritability estimates and association results alone cannot determine why population averages differ.
- Polygenic inheritance is not genetic determinism. Development, environmental exposure, and chance can all contribute to the observed phenotype.