Hi,

The figure on the bottom gives you the number of species for each prior. With the initial approach, it is very stable: 6 species. In the recursive approach, 1 or 2 additional species are recognized. Then to decide if you have finally 6, 7 or 8 species, you need to integrate other data, criteria or methods (anything that could support one or another partition).

Please note that each point on this figure (in the ABGD website) is clickable. It opens another page with the list of the species and the sequences included in it. If your dataset is less than 700 sequences, ABGD also provides a NJ tree (newick format, you just have to copy it in a document that you open with your favorite tree editor). In the tree, the species number as defined by ABGD is added at the end of each sequence title, at the tip of each branch.

All the best

I'm not familiar with this program, but it looks to me as if there are 3 groups. The distances close to 0 are within-group. Then there are distances around 7%, around 11% and 14%. You can think of these as the sides of a triangle. I think theoretically there could be a fourth and even further groups whose distance to each of the first three by coincidence overlaps with the distances you already have.

Maybe I'm conservative but I'd still prefer a tree graph and a table that identifies the clades and shows the distances between them.

How you further interpret these results depends on the other evidence you have. I think 7% distance usually but not always implies different species, whereas this is certainly so for 11 and 14%.

Strange - I thought I had answered to this question, but it is not shown anymore... New attempt:

I'm not familiar with this program, but I think:

The histogram show 4 categories of distances: 0-1% which are within-group; then around 7%, 11% and 14%. These most likely represent 3 groups. You can imagine them ordered in a triangle with sides 7, 11 and 14 long. In theory there could be a 4th or further groups whose distances to the first three coincidentally also is 7, 11 or 14.

These graphs don't identify the groups by name. Maybe it's old-fashioned but as illustration I would prefer a tree graph (unrooted if you have no outgroup) with names of the groups as labels and a table indicating the pairwise distances between them as ranges.

As for what these distances mean: 11 and 14% are large, taking into account the saturation effect, and do not suggest that the groups are closely related. 7% would be much for intra-specific mt variation but not entirely impossible.

Maybe I should include my rooted  phylogenetic tree as well. Since this analysis (ABGD) is based on pairwise distance matrix, thus it may not congruent with phylogenetic tree (evolution based).

Please anyone help me...

Hi Haslawati,

Yes, it is always nice to see the divergences in your dataset in form of a graph. Please note that a NJ-tree like the one you present is based on a clustering algorithm on genetic distances without an optimality criterion. Thus, it represents a more poorly founded phylogenetic hypothesis compared to results from ML or Bayesian inference. To preserve the distinction, the groups are not referred to as clades but clusters. You should probably pay extra attention to sample NC006381. To me it looks like this must be a species different from cluster E.

Cheers and good luck!

Yes Dr Ekrem.  NC006381 is a different species definitely. I did not do ML or BA since majority of my reference articles on barcoding just use NJ tree (very few used ML). Only seen BA when used for phylogenetic studies.

By the way, please forgive my ignorance, what is the difference to use clusters instead of clades?

Thank you for your contribution.

Hi,

The figure on the bottom gives you the number of species for each prior. With the initial approach, it is very stable: 6 species. In the recursive approach, 1 or 2 additional species are recognized. Then to decide if you have finally 6, 7 or 8 species, you need to integrate other data, criteria or methods (anything that could support one or another partition).

Please note that each point on this figure (in the ABGD website) is clickable. It opens another page with the list of the species and the sequences included in it. If your dataset is less than 700 sequences, ABGD also provides a NJ tree (newick format, you just have to copy it in a document that you open with your favorite tree editor). In the tree, the species number as defined by ABGD is added at the end of each sequence title, at the tip of each branch.

All the best

Thanks Dr Nicolas for the clear definition on the useful help on ABGD method. It really helps!

Hi Haslawati,

One question for you. How can you obtain the d value in Figure2 (last column)? I am the newcomer of ABGD and I cannot find this value for my dataset.

Feng Xu I am really sorry I did not see your message earlier. It has been 3 years already. Do you still need the answer? I hope you already have it anyway.

I hope I can still contribute, even though now I am on another project.

Hi Haslawati Baharuddin

If you have finished your project on ABGD, So please share with us some discussion and conclusion about the ABGD analysis graphs (results).

I was wondering what can I find using these graphs obtained from ABGD...

Thanks for your Kind