Hi,
I am looking for evidence for hybridization among different species and each of them is represented by several sequences in my alignment. GARD detected break points in my alignment, but how do I know which sequence pairs are involved in recombination? I guess there is recombination within each species, and GARD may detect it, because I have several sequences for each species, but that is not what I am interested in. I would like to know whether sequences belonging to different species show evidence for recombination. Would it be reasonable to say that if a pair of sequences group together on any of the trees corresponding to different alignment fragments that means they have recombination? Do they need to be identical or just similar? Can it be that they group together just because this fragment is conserved and they share it because of common ancestry and not because of current recombination?

Thank you!

Dear Artpoon,
thanks a lot for your reply. Yes, I know that GARD is based on phylogenetic incongruence. And I think that rate variation is quite probable in my data set. But how can I be sure that what GARD detected is due to recombination or due to rate variation? GARD has detected two break points. I have examined the trees and the first two trees seem to be due to rate variation. The first tree makes no sense: all sequences grouped together except few from different species placed on long branches. The second tree groups all taxa together and places the most divergent species on a long branch. The last tree looks plausible: the two taxa are grouped together (although the sequences are not identical) that could possibly hybridize, but I do not know it for sure, that’s what I wanted to check with GARD... So, your advice to examine the output trees closely is surely useful as it helps to sort out the trees which make no sense. But if there is a tree that does make sense, then how do I decide whether it is due to rate variation or to recombination?

Another question: I have tested my data set with GENECONV before, and it detected no significant recombination fragments. Can it be that GARD is more sensitive than GENECONV and can find recombination that actually happened but was not found by other methods? Or should I interpret these results so that there is in fact no recombination and phylogenetic incongruence detected by GARD is due to rate variation?