$F_{ST}$ is one of the most famous and most important statistics of all of evolutionary biology. Yet, many people misunderstand it or misuse the classical results from the literature on $F_{ST}$ (Whitlock and McCaughley,1999).

Fixation Index in infinite islands model

In a panmictic population, the probability of identity by descent $F(t)$ in generation $t$ is the probability of sampling the same allele twice plus the probability of not sampling the same allele multiplied by the probability of identity by descent in the previous generation

$$F(t) = \frac{1}{2N} + \left(1-\frac{1}{2N}\right) F(t-1)$$

, where $N$ is the population size. Here I am assuming diploid population and no mutation. In an infinite allele model, these probabilities must be weighted by the probability that none of the parent migrated in the previous generation.

$$F(t) = (1-m)^2\left(\frac{1}{2N} + \left(1-\frac{1}{2N}\right) F(t-1)\right)$$

, where $m$ is the migration rate between any two demes. Setting $F(t) = F(t-1) = \hat F = F_{ST}$, assuming that $m$ is low and solving for $F_{ST}$ yield to the classic result from Sewall Wright

$$F_{ST} = \frac{1}{1+4Nm}$$

Definitions of $F_{ST}$

$F_{ST}$ was defined by S. Wright as

$$F_{ST} = \frac{var(p)}{\bar p(1-\bar p)}$$

, where $var(p)$ is the variance of allele frequency among population and $\bar p$ is the overall average allele frequency.

From Nei (1973)

Wright showed that the variation in gene frequency among subpopulations may be analyzed by the fixation indices or F-statistics. He derived the formula $$1 - F_{IT}= (1 - F_{IS}) (l-F_{ST})$$ where FIT and F1s are the correlations between two uniting gametes to produce the individuals relative to the total population and relative to the subpopulations, respectively, while $F_{ST}$ is the correlation between two gametes drawn at random from each subpopulation. $F_{IT}$ and $F_{IS}$ may become negative, but $F_{ST}$ is nonnegative.


In the derivation of the fixation index in infinite island model, $F_{ST}$ is a probability of identity. In the quotation from Nei (1973) (and other sources), $F_{ST}$ is presented as a correlation coefficient. In Wright's and Nei's definition, I don't see the relation from the equations to either a probability or a correlation coefficient. Can you help to clarify this for me?

For example, I would expect that $E\left[\frac{var(p)}{\bar p (1-\bar p)}\right] ≈ \frac{1}{4Nm+1}$, where $E[X]$ is the expected value of the variable $X$. Can you demonstrate that this is true?


Your Answer

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

Browse other questions tagged or ask your own question.