A trait is defined as an inherited characteristic, like eye color and height. A phenotype is defined as the specific physical manifestation of a trait, like blue eyes and 6 feet, respectively.

Now, genes code for traits. And because alleles are different versions of the SAME gene, they must code for the same trait. Let's say gene A codes for hair color (trait). Gene A(1) (allele) codes for brown hair (phenotype) while gene A(2) codes for blonde hair. Therefore, alleles code for the same TRAIT, but differing PHENOTYPES.

Here is the problem. When I google: "heterozygous", I get a bunch of websites defining it as "referring to a gene pair in which the two alleles do not code for the same trait"

This is not correct because a gene-pair refers to genes located at the same locus on a chromosome. In other words, they are alleles of the SAME gene and code for the SAME trait. They only code for different phenotypes associated with that trait.

So. Either my definition of traits is wrong. My definition of alleles is wrong. My understanding of alleles being different versions of the same gene is wrong. OR. The internet loves to be imprecise with their language and interchangeably use terms that mean different things.

Which is it?

  • $\begingroup$ Depends on what you call a trait. One gene (allele) can contribute to multiple traits and vice versa. IMHO, terms like "trait" and "phenotype" have little relevance in modern genetics. $\endgroup$ – WYSIWYG Oct 20 '16 at 6:00
  • 1
    $\begingroup$ I disagree that (1) the issue behind the question is related to the definition of trait and (2) that the concept of phenotype has no room in modern genetics. I don't think many geneticist, evolutionary biologist, medical geneticist or anyone would really agree that anything in modern genetics remove relevance of the concept of phenotypic traits. $\endgroup$ – Remi.b Oct 27 '16 at 6:54

I fear you won't get a definitive answer on this. A survey indicates both usages of trait all over the Internet. As a teacher, I generally used trait to apply to the gene. This made teaching the "central dogma" clear: DNA -> RNA -> protein -> trait. Of course this simplification gets added to later as the course progresses.

One interesting question this poses is in the case of skin color. Is that a trait? Last I read, at least four different genes interact to produce skin color.

I'd side with you on this but acknowledge the fact trait doesn't seem to have the consistent, well-defined meaning that allele does.

On the other hand, a visit to UC Berkeley's evolution site shows them applying trait to the variations due to allele combinations.

Hope that helps! :)

  • $\begingroup$ I appreciate your answer! I think you're probably right. NOTE: I would argue that skin color is indeed a trait. And you can have multiple genes code for one trait. In fact, that is the definition of a polygenic trait! $\endgroup$ – Nova Oct 19 '16 at 22:51

What is a trait?

Many people have talked about the issue of defining a trait. Of course a trait can be anything and there is no objective way of delimitating or counting the number of traits of an organism. However, the issue in this question is completely unrelated to the definition of traits, it is an issue of semantic of causality.

Semantic of causality

The whole issue boils don to what you mean by code for the same trait. In reality, this expression makes little sense. To explain the semantic behind this concept I'll use the existing semantic from probability theory (which applies not only as a vague analogy).

Semantic issue explained in probability theory

If you have some basis in probability theory, you know that you can consider a covariance between two random variables but it doe snot make any sense to consider covariance between the outcome of a random variable and another variable. For example, it makes sense to ask whether footedness is related to handedness, but it makes no sense to ask whether right handed is related to footedness.

Above semantic applied to genetics

Following the above logic, consider a locus (=position on the genome, may correspond to a gene) as being a random variable and the alleles at this locus being the possible outcomes of a random variable. Trait is another random variable which outcomes are the different trait value. It makes no sense to ask whether a given outcome (a given allele) affects a trait (random variable), but only if a locus (random variable) affect a trait (another random variable).

In other word, one can only consider whether variation at a given locus affect variation for a given trait, and not whether a given allele codes for a trait. A locus can indeed affect several traits. Typically, an allele will cause investing into building trait A, while the other allele at the same locus will cause investing into building trait B. As such this locus explains variation for both trait A and trait B but one could not say that an allele code for A, while the other codes for B.

The above case where one locus affect several traits is called pleiotropy. When several loci affect a single trait, it is called epistasis.


The way I see it, the semantics in this area of genetics are somewhat loose and you are correct about all your definitions.

I think the word 'trait' is not currently used as much as it used to e.g. in early 20th-century genetics, or earlier. The meaning of 'trait' can be compared to that of 'feature' (i.e. does it refer to 'having eyes' or 'having blue eyes'?).

In short, I would use 'phenotype' rather than trait when you refer to the different alleles, and if necessary use 'trait' as height, eye colour, etc.

  • $\begingroup$ Really? Every modern introductory genetics book I've encountered definitely use the term "trait" to describe a characteristic... $\endgroup$ – Nova Oct 19 '16 at 22:51
  • 1
    $\begingroup$ in other words, it is mostly used in Mendelian genetics... so it will not be a problem you will encounter in more advanced genetics, in my opinion $\endgroup$ – M.G. Oct 20 '16 at 10:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.