According to Hebert et al. 2003 mean genetic divergence in COI between congeneric species of insects range from 6.6 to 11.5. However, they found species pairs with divergences as low as 2%.

I would recomend you to use Arlequin to calculate genetic divergence and using the uncorrected pairwise difference model, as it's the more widely used.

As COI genetic divergence thresolds are ambiguous, I recommend you to calculate intraspecific COI divergence for both species then compare these values with the genetic divergence between the species. If the interspecific divergence is 3 times higher than the intraspecific divergence, or more, it's very likely you are working with two different species.

Finally, a small  COI divergence between species doesn't exclude two species from being different. It can mean their reproductive isolation was very recent.

For reference: Hebert, P. D., Ratnasingham, S., & de Waard, J. R. (2003). Barcoding animal life: cytochrome c oxidase subunit 1 divergences among closely related species. Proceedings of the Royal Society of London. Series B: Biological Sciences, 270(Suppl 1), S96-S99.

Dear Prof.Artur Burzynski,

Thanks for the comments . I took 3 sequences of each specimen and analyzed carefuly. This is first report in this insect group and both the specimens have few characters differences . However, since the specimens are paratypes, can't be destroyed again for molecular work. I have few remaining DNA in each sample can be used for ITS region amplification and as Daniel Fernandez Marchan mentioned I would like to go for Arlequin method . I believe to would get it. Thanks for all once again.

If you are sure that morphological traits support well the specific status, then your mtDNA data only tells you that COI may not be useful for species delimitation. You need more than one sequence though in order to test for the degree of clustering. The most likely explanation for the low divergence would be a recent origin for the new species. That's also a useful information that you can add to the publication, and call for further analyses, such as those suited for multilocus data to test the species status. You should also consider the distribution of the new species as is important for the interpretation of your data (e.g. sympatric, parapatric, or allopatric?).

I think I have been repeating myself in similar questions, but do it again. Mitochondrial genetic difference does not necessarily tell anything at species level. I know species complexes that for as surely as is possible contain separate species, with identical COI sequences, and on the other hand, over 7 % divergences within one species, and cases of introgression where populations or single individuals of populations possess MtDNA of another related species. Barcoding is a good screening tool in search of taxonomically intriguing, perhaps overlooked cases, or getting preliminary organization of largely unknown biotas, say, like small wasps in canopy of Amazonian rainforests. But, the evidence for the species status of a taxon must essentially be searched for from elsewhere (morphology, ecology, perhaps stridulation pattern in leafhoppers, and so on). There are no, I repeat, there are no thresholds existing, although practical experience tells that in particular groups certain divergences often suggest status of species and members of the same species often nicely cluster together and are separate from others - but again - they do not define it. It is of course nice if you can add that  your species, discovered by other means, have, on top, also some difference in the MtDNA, but that is not itself hard evidence to one or another direction. In my focal group  in microlepidoptera, difference as low as1 % is often indicative that there may be something behind (http://www.nature.com/srep/2013/131009/srep02901/full/srep02901.html); in other Lepidoptera groups, as is the case in Geometridae,  7 % may be very little. So, never generalize. Playing with algorithms does not help, either, in specific cases as yours.

Dear ProfsLauri Kaila  and Artur Burzynski

Thank you for precise answer for my query.

Unfortunately, I don't have many specimens for molecular work since al are  paratypes..I did PCR for only few samples and concluded to assist the new finding. In all the three times we got similar sequences and analysed with software carefully.

However, We would like to go for ITS regions and the software here mentioned.

Thank you