Thank you stephen, I tried to interpret it but what I found that Tajima's D and Fu and Li's should be same charge ( positive value for  Tajim's follow by   Fu and Li's also positive) but my data showed opposite

Hello Salama,

The numbers seem a bit odd since they are really high and especially since Tajima's D has the opposite sign as Fu's F. With that said since the p-values are all non-significant all you can say is that the do not deviate from neutral expectations. However, I would make sure you are running everything correctly since the values are so odd.

Best Regards

Hi, i have a similar problem. I am analyzing a set of 297 sequences of COI representing 6 geographically different populations. Tajima D: 0.244 with not significant P value; Fu & Li's D: -2,743 with significant P (P

Hello Dilruba,

Fu's F is one of the most sensitive to recent populations expansions, so there are cases where Fu's F is negative and significant and Tajima's D is near zero, which is what my best explanation would be.

My suggestion is to double check that everything is running correctly and maybe make a haplotype network to a visual looked at your sequences (i.e. should have low nucleotide diversity with high haplotype diversity or in other words a "star" network). The haplotype network might help explain your results (i.e. maybe the population maintained a few haplotypes that are far apart during the bottleneck and now there is a gain in haplotypes that are one base off of these distance haplotypes making pairwise comparison of nucleotide diversity high even though the population is clearly a population expansion). Also, the network would help you see if there is any population structure you are over looking, which would also give you odd results like this.

What ever you do I would not ignore Tajima's D but instead try to figure out why you have negative Fu's F and near-zero Tajima's D.

Hope this helps.

Does grouping the populations in DNAsp affects these results? When i grouped relatively close populations together i get both positive Tajima D and Fu's F. and one more question about the p values.. if the values are not significant is that mean it not happening? Max R. Bangs

Best regards and thank you

There has been some debate on this but if you have population structure you can get false significance and false values (not under every scenario, but there are some ways this can happen and in my option more often then not). In any case this statistic are meant to be ran on within a single population, and violating this is highly advised against.

My suggestion is to:

1) Make a haplotype network (or phylogeny) and see if there is any pattern in the association of geography or phenotype (i.e. is there population structure).

2) If there is a clear pattern of population structure then run Tajima's D and Fu's F (or other statistics you like) on each independent population. If there is no structure then run it as if it is one population.

I should also note that if you are using a low number of samples you can also get bad estimates of Tajima's D and Fu's F. For example, in a similar oddity with these statistics it was later discovered that the research was calculating using two samples, as you would expect this does not work well. I would suggest at least 20-30 samples per population (30 is rule of thumb if the are looking a single genes or small set of loci) and would not trust any values for anything calculated with

Thank you for your precious advices Max R. Bangs

Best regards

Hi, both the Tajima’s D and Fu ald Li’s D are parameters calculated from DNA sequences that principally indicate molecular evolution under different demographic models in a single population setting in loci that are putatively neutral. Biological interpretaions can also be made to infer natural selection at the DNA sequence fragment from bith the parameters. Overall, interpretaion drpends on the DNA marker fragment and the evolutionary question asked in a particular research setting.

Please share with me how you solved it since I had the same problem Salama Al-Hamidhi @

A negative value is not significant base on the articles described.