I think this is accepted as I have seen 3% genetic distance for cyt b used to delimit species of the genus Bungarus. However, do the species in question morphological differences?

The short answer is, yes you can. The long answer is that you can, but the OTUs that you define are just arbitrary groupings that are based on a value judgment about a particular level of molecular divergence that you have made. In your case, this is based on prior information about molecular divergence in an existing taxon pair.

The problem with these types of barcoding approaches is that we no idea if the groupings are biologically or evolutionarily meaningful. That is not to say that these approaches aren't useful for defining conservation units and so on. It's just important to recognize the limitations of using an arbitrary cutoff. There is plenty of literature on this about the limits of DNA barcoding! Check it out!

Its also worth pointing out that this problem is not new. Taxa delimited based on geography or morphology almost always suffer from the same problem. People make value judgments about how different looking, geographically isolated or genetically different populations have to be before they are considered species. 

I would strongly recommend against doing so. From several species are deep mitochondrial divergences, of the same magnitudes as between species  known. If you  not also use other data it is no way of saying if the OTUs you find represent species in any traditional way or if they are only divergent mt-lineages of the same species 

Thanks Zeeshan, they are not morphologically different! However, geographically they are isolated and I guess there will be some hidden character among them to distinguish them, which normally happens in cryptic diversification. 

Sean Stankowski and Svante Martinsson,  thanks a lot for your explanations. I am delimiting them based on three genes and other empirical evidences. 

This has been a matter of some debate and I side with those who say that one cannot differentiate species on the basis of genetic distance.  The reason is that there does not seem to be any way to arrive at a consensus of what percentage to use and even if one was selected it would inevitably be an arbitrary decision.  I and many others prefer character based approaches.  If a character is fixed in one population and dose not appear in any other population it is evidence that the population with the fixed character is reproductively isolated.  Of course the size of the sample has an impact on the confidence one can have in these cases.  If there is a single individual it is impossible to say if the character is fixed.  On the other hand if you have sampled 100 individuals you could have greater confidence in your argument that the population is a species.  This issue is discussed within the context of molecular barcoding by DeSalle (2006).  I recommend this article as well as some of the other articles that he references for a more thorough discussion of this issue.  Here is the reference.

DeSalle R. (2006) Species Discovery versus Species identification in Molecular Barcoding Efforts: Response to Rubinoff. Conservation Biology Volume 20, No. 5, 1545–1547.

I suggest you reading the following articles that might help you to delimiting species based on genetic data by using the evolutionary genetic species concept (namely the kapa/theta method):

Birky CW Jr. 2013. Species detection and identification in sexual organisms using population genetic theory and DNA sequences. PLoS ONE 8:e52544. DOI 10.1371/journal.pone.0052544

Birky CW Jr, Adams J, Gemmel M, Perry J. 2010. Using population genetic theory and DNA sequences for species detection and identification in asexual organisms. PLoS ONE 5:e10609. doi:10.1371/journal.pone.0010609

http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0052544

http://journals.plos.org/plosone/article?id=10.1371%2Fjournal.pone.0010609

Hi James! In your answer you say that it might be better to use characters, and argue that if there is a differentially fixed character that it might indicate RI??

But, hold on, aren't nucleotide states characters too? Of course they are, and they are effected by gene flow in the same way as the variable nucleotides that encode the traits that you are talking about?? In fact, synonymous changes have an advantage of morph characters because they are less likely to be fixed by selection in the presence of gene flow (for a morph character, this would give the appearance of complete RI when there is ongoing gene flow).

The fact is, almost all taxonomic approaches are based on value judgements about how different looking, geographically isolated or genetically different populations have to be before they are considered species. I'n other words, taxonomic delimitation is never really about reproductive isolation at all. Its about levels of divergence, which may or may not be due to biologically based RI.