Dear Chen,

different explanations are possible for the dicrepancy you found. I cannot give precise answers because I do not know the system you have studied. However, I suggest you to make the following further analyses:

1) to calculate (and test) Ramos-Onsins & Rozas statistics (R2). It is the most powerful test when sample size is small (otherwise Fs is ok)

2) to calculate (and test) the sum of squared deviations of mismatch distributions.

3) to calculate (and test) the raggedness index.

Tajima's D is considered the less powerful statiustics.

Multiple peaks may appear in mismatch distribution, despite population expansion, whether genetic substructuring is present in your population (this effect could be responsible of the discrepancies you found).

Also, remember that the significant threshold for Fs is not 0.05, but 0.02.

I hope you find this information somehow useful.

Best wishes

Ferruccio

Dear Chen

You can't really assume that multimodal mismatch distribution is a clear evidence for the absence of population expansion unless you analyze it statistically, and for this you have to test the SSD and rg demographic indexes as Ferruccio have already mentioned. If you find non significant SSD values then your data would really support the hypothesis of demographic expansion. I would probably expect this tendency based on the Fu's Fs results. Fu's Fs is considered to be more sensitive in detecting population expansion and suggesting possible past population growth. However if you don't find such result then either population reduction, population subdivision, a recent bottleneck, or migration which resulted in secondary contact among previously differentiated lineages could have caused the discrepancy in detecting population expansion.

Also, positive Tajima D values associated with multimodal mismatch distributions are often possible indicators of population bottleneck. This could be strongly accepted for population B, when you do you really have non significant and positive Tajima D value. The higly significant Fu's Fs for this population can be just a sign that it might have experienced historically rapid population growth from an ancestral population with a small effective population size. I advise you to make BSP (Bayesian Skyline Plot) analysis to test this hypothesis. Patterns of historical demography can also be inferred from estimates of the effective population size over time using the BSP method. 

Good luck!

I'm interested in Chen question since I'm doing similar analysis on my data, and reading your answers Ferruccio and Temim, I was wondering: what is the expected value of R2 in a scenario of no population expansion but a selective force acting on the population? Or, can you tell me how is the R2 value expected to varied under different scenarios?

Thanks!

Hola Carolina, you can find all details of R2 statistics in the paper by Ramos-Onsins and Rozas (2002). Remember that R2 assumes a neutral model.In practice, you first have to calculate R2 for your data with DnaSp, then you have to test its significance by coalescent simulations implemented in DnaSp.

Dear Ferruccio,

Thanks a lot! I did it indeed, and always obtaining positive values of R2 but just a few cases in which I found a significant one under coalescent simulations in DNAsp.

I will go in deep into my data and see.

Ti ringrazio un'altra volta! Saluti

Carolina

De nada Carolina. Si quieres, puedes escribirme en castellano la proxima vez. Yo entiendo y escribo castellano de vez en cuando. Tengo tios y primos en Rosario. Suerte con tu analisis. Un saludo. Ferruccio

Dear Deli

you mentioned "However if you don't find such result" what is the results you mean,  positive or negative results such as Fus Fs -2.035 and 3.524

Hi dear

I have one question about the rg and SSD indices which are calculated using the Arlequin. when we say there were not significant, what non significant exactly means here. Non significantly different from 0??? such as significantly different from 0 in Tajima’s D.

Ive search a lot but I cant find the answer of this.

Thanks in advances

Niloo

Dear Niloofar, SSD is a sort of "geometric" test. It evauates the deviation of the observed curve from the expected one. So, contrarilily to neutrality tests, its significance does not mean "demographic expansion", because the observed curve deviates significantly from the curve expected under the demograpic expansion model.

Dear All,

very good discussion. I am trying also to understand the results I have, all R2 are positive non-significant and the majority of Tajima's D and Fu's Fs values are negative non-significant. The mismatch distribution has three different situations: uni-modal (the majority), bi-modal and multi. My question here is that a demographic expansion?

Thanks,

Ahmed

In an analysis of population genetics, I found D= -1.51 (P=0.05); Fs= -2.88 (0.04); SSD= 0.169 (P=0.00) and Hri= 0.07 (P= 1.00). So, IIt seems the D and Fs value does not match SSD and Hri values. How I can interpret the result ? Your suggestion will be highly appreciated.

Can anybody help me how can I interpret my opposite results between neutrality test-mismatch curve and SSD-raggedness index (Hri) tests?

In my results, insignificant Tajima’s D= -0.76 (P= 0.24); Fu’s Fs= -1.59 (P= 0.27) and multi modal observed MMD curve denote populations are at equilibrium. However, SSD = 0.016 (P = 0.48) and Hri = 0.009 (P = 0.15) are insignificant which denote population expansion.

Please see the attached picture. I used Arlequin program for analysis. In my d-loop DNA sequence, there are long indel zones.

Can anybody help me, please?