In assessing for genetic differentiation between two populations using 8 microsatellite loci, my traditional Fst/ Gst values suggest little population structure (i.e., < 0.15), while my standardized measurements of differentiation (Jost's Dest, G'st, G"st) are very high (0.35-0.39). Would relatively high allelic diversity at microsatellite loci greatly bias the latter measurements? What else might cause these values to be contra-indicative?
This is quite a complex issue and has been discussed many times in the literature (e.g Whitlock 2011; Hedrick 2005; Verity and Nichols 2014). Your result make perfect sense when using highly polymorphic markers (G'st > Gst) since you are correcting for the maximum possible value of Gst. The simple answer is: Yes, high allelic diversity will bias Gst and this is the exact reason why you calculated G'st. But, what makes you think these values are 'contra-indicative'? Are you using Wright's original description of Fst ~0.05-0.15 = moderate differentiation and Fst >0.15 = high differentiation? If so, then both your Gst and G'st suggest moderate to high differentiation. But remember, Wright's recommendations are based upon biallelic loci. The affect of allelic diversity will depend upon frequency: lots of very rare alleles do not contribute much to heterozygosity, so wont affect the maximum Gst very much. But, many, equally frequent alleles will result in high heterozygosity and a much lower maximum Gst.
Reading the articles given below will help more. Remember, these values are not really 'counter-indicative' since the same Gst-max will be used to correct ALL Gst calculations between populations. To better assess your G'st value it may be useful to simulate a microsatellite dataset under varying histories and compare G'st values from these simulated datasets to your observed values (~approximate Bayesian computation approach). Finally, what is the question you hope to answer? If you want to estimate migration rates, don't use Jost's D...
For your reading...
Whitlock, M. 2011. Mol Ecol. DOI: 10.1111/j.1365-294X.2010.04996.x
Hedrick, P. 2005. Evolution. 59(8):1633-8.
Verity, R and Nichols, R. 2014. Mol Ecol. DOI: 10.1111/mec.12856
F statistics are affected by gene diversity witihin populations, D-statistics are independent of within population diversity. I suggest show both measurements and do your conclusion considering your biological expectations.
Anthony
The answer by Robert is very accurate. Just to follow-up a little bit, imagine one SSR in two populations with very high gene diversity (close to 1) but no alleles in common. Because Gst=1-(Hs/Ht), traditional measures (Gst, Fst) will tell you differentiation is close to 0. But the populations have no alleles in common, so differentiation should be 1!!
This is one of the reasons for Jost's claim that "Gst and relatives do not measure genetic differentiation"
Hope this helps
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation