Kym,

There is no "official consensus" among population geneticists because Fst is a parameter that only makes sense in comparative analyses among several populations.

The values of the fixation index Fst range from 0 to 1. A zero value indicates no population structuring or subdivision, or in other words, complete panmixis. A value of one implies that all genetic variation is explained by the population structure, which is mainly conditioned by the existence of barriers to gene flow (geographical, linguistical, sociocultural, even economical ...), It can, therefore, be concluded that the two populations examined share no genetic diversity.

In human populations, for instance, Fst values under 0.15 (approx.) are usually interpreted as a lack of significant genetic structuring or population subdivision. At the same time, Fst values above 0.25 are seen as moderate population structuring. Fst values above 0.4 are normally considered a strong population subdivision.

That's what I can tell you about the fixation index.

Good luck, Kym !!!

You can also look the significance of FST (using a correction for multiple tests, eg bonferroni or other correction method).

Josselin

kindly see the attachment. May be it may help you

I do not know the answer to your question... but here are some personal considerations.

If we have a continuous scale for FST, from 0 to 1, why should we want to break the values down into discrete classes (weak, moderate and strong)? I think that we lose information if we do this.

More importanly, we should not forget that FST is affected by marker heterozygosity and that different demographic processes can lead to the same value of FST (e.g. complete separation with accumulation of difference in time vs. an equilibrium situation with a balance between drift and gene flow...). So I would rather consider FST as a first indication of the processes that shaped genetic variation.

Bottom line. I would try to compare the value of FST against values found in other species/populations with the same markers... and I would not over-emphasize the biological meaning of this index.

Hi Kym,

Hartl and Clark (1997) had some classes for FST. 

0.25 = very great genetic diff.

Frankham et al. (2002; 2010) had FST>0.15 = significant differentiation

As per Marco's response above, I don't really agree the class structure (I think the descriptors run the risk of being somewhat subjective if used without the FST values themselves and an understanding of species life-history).  That said, the descriptors are often more meaningful for people not directly working in population genetics (e.g. species managers).

Kym,

In general, I have no problem with your rough classification (or the Hartl & Clark summary provided by Susan).  Although the scale ranges from 0-1, values greater than about 0.5 are quite rare in plants (for comparisons of demes that are clearly within the same species), but of course depends on sampling scale and natural history of the species.  And in my experience, measures of plant population differentiation often exceed that for most animal taxa, so what's "important" would clearly differ.

Kym, the answer has to be no, because Fst values depends not only on the amount of differentiation among populations but also on diversity within. This depends on your choice of markers. For example, if for the same populations you have data from both low substitution rate markers, such as non-synonymous changes in coding genes, and hyperdiverse markers, such as microsatellites, then, for a given level of structure Fst will be greater for the low rate markers. This comes from the fundamental (Ht-Hs)/Ht definition of Fst (the Weir and Cockerham analogues work similarly). As you increase the rate at which new alleles are produced, you increase average diversity within (Hs) relative to that across the pooled population (Ht), reducing Fst. Also, because the equilibrium diversity within a given population is determined by a combination of effective population size and mutation rate (theta), then species with larger populations will have lower Fst for a given migration rate than comparisons among smaller populations. This leads to Wright's result that it is the number of migrants (Nm) rather than the migration rate (m) that determines Fst, and the same number of migrants will lead to the same Fst value, despite vastly differing population sizes (equilibrium Fst = 1/[1+4Nm] for the infinite island model with diploids, but the same holds for other models of population structure)

Michael, regarding your comments on marker polymorphism influencing the magnitude of Fst, could microsatellites with relatively few alleles produce high Fst?  I'm doing a study on parasitic nematodes with microsats.  Each locus has only 3 or 4 alleles (not a lot for microsatellites).  And I'm getting high Fst (0.18).

Thanks

Doug, the FST value you report is quite large.  Mutation rate is not the only force, drift can be quite strong ( I am ignoring selection for now, microsats, assumed to be neutral marker).  I have compared results from allozymes and microsats (and from different mutational classes of microsats) in the same populations of pine and found little difference in FST when dispersal is large.  But, if you are interested in calculating a relative mutation rate, by taking Michaels lead and using effective population size (N), the relationship of effective number of alleles (ne) to the effective population size (N) can be used, as shown by Ewens (1979): Nu = (ne-1)/4 .  The effective population size (N) should be the same for all loci, and comparing Nu between loci is an indication of relative mutation rate.

I had a heck of a time figuring out the answer to this question for my masters thesis. The is no magic number- but from the literature I found, in population genetics I found Fst=0.15 to be a fairly good value for drawing the differentiation line.  doi:10.1007/s10592-014-0671-3

I would recommend to calculate z and set threshold of significant (e.g. z>5) depending on how the Manhattan plot looks like. This is a better estimate and it is used in may natural and directional selection papers.

z can be calculated as follow:

z = (FST - mFST) / s

where mFST is the average of FST values and s is the standard deviation of FST values.

You will have an extra column contains zFST values, whereas you need to keep FST values.

Cheers,

For SSR I totally agree with the rule Kim's rule of thumb. The most recent review for this in plants is Nybom et al (2004), which I use all the time. Maybe I would star do a third class = 0.25-0.40 and then a forth class > 0.40.

Rosalía Piñeiro, which Nybom et al 2004 paper are you referring to? Thanks!