Coalescent units (branches) of length zero?

[jbernst]

Hi,

I am using ASTRAL-III to make a species tree for a set of taxa. I ran two versions of this tree, one in which three species are considered distinct between two populations (they are currently considered species), and another where I lump them into a single grouping (because concatenated analyses revealed that the two groups are not always reciprocally monophyletic). In the attached image, these are A2, C2, and E2 on the left tree. The right tree shows what happens when I lump these together, presuming that they are actually one species each and not two.

My question is why in the left tree the coalescent units are zero? I thought that the branches would be short and perhaps half of what it is in the right tree, but not that there would be a zero length for A2, C2, and E2. I just am looking for an explanation so I can understand how ASTRAL works better, and if I should proceed with just using the tree on the right if I think A1+A2, C1+C2, and E1+E2 are distinct species (each). Thank you!
ASTRAL_comparison.pdf

[smirarab]

Are you referring to the terminal branch above A1, C2, and E3? If so, how many individuals do you have from each of these population/species?

…

On Tue, Feb 6, 2024 at 5:24 PM Justin Bernstein ***@***.***> wrote: Hi, I am using ASTRAL-III to make a species tree for a set of taxa. I ran two versions of this tree, one in which three species are considered distinct between two populations (they are currently considered species), and another where I lump them into a single grouping (because concatenated analyses revealed that the two groups are not always reciprocally monophyletic). In the attached image, these are A2, C2, and E2 on the left tree. The right tree shows what happens when I lump these together, presuming that they are actually one species each and not two. My question is why in the left tree the coalescent units are zero? I thought that the branches would be short and perhaps half of what it is in the right tree, but not that there would be a zero length for A2, C2, and E2. I just am looking for an explanation so I can understand how ASTRAL works better, and if I should proceed with just using the tree on the right if I think A1+A2, C1+C2, and E1+E2 are distinct species (each). Thank you! ASTRAL_comparison.pdf <https://github.com/smirarab/ASTRAL/files/14186681/ASTRAL_comparison.pdf> — Reply to this email directly, view it on GitHub <#92>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAGJXOHPWO7YPPKMGCRW3N3YSKUTNAVCNFSM6AAAAABC4Y73U2VHI2DSMVQWIX3LMV43ASLTON2WKOZSGEZDCNZWHE4DAMA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- Siavash Mirarab

[jbernst]

Ah, yes, my apologies. I am talking about A1, C2, and E3. There are 5 of A1, 2 of C2, and 2 of E2.

If it helps, here are the numbers of individuals for each population

A1 = 5
A2 = 2
B1 = 2
B2 = 4
C1 = 2
C2 = 2
D = 4
E1 = 2
E2 = 2
F = 2
G = 2
H = 2