Hi Trojet,
I agree, looking at modern distribution it's easy to theorize "Celtic dispersal" of J-Z631. This view has been held many times before, including me
However, keep in mind that if we based things on modern distribution, R1b-M269>L23 would be from Iberia (as was theorized years ago on these forums), J-M241 in Pakistan/India would be as a result of Alexander the Great troops, etc, etc.
I do agree that modern distribution only is unsufficient (many thing can happen, including uni-directionnal diffusion, population replacement, ...), what was important in my post was the timescale.
From 900 to 500BC there is ~45 surviving branches of Z631 who started. This is a big indication that J-Z631 population experienced a big "boom" at that time.
Timeframe and distribution are fitting. This is the main point.
Also note that excluding very particular context, diversity centers (not density centers) are good proxy for haplogroup origin.
Placing L23 in Iberia sounds weird to me ... diversity of modern descendent haplogroups are not pointing there at all ... like not at all ! A superficial analysis should normaly be enough to exclude Iberia as a place of origin.
That's what is very important with modern distribution, what matter is not the number of people with a given haplogroup today, it is the number and distribution of branch at the moment of interest (which allows to remove most of the founder effects).
Same for J-M241 in India/Pakistan, what helps finding out how it arrives there, is to look a the time frame (and the diversity center), that's why J-Z2432 have likely been spread with neolithic peoples toward Pakistan/India and later integrated among Indus-valley peoples.
It's also true that Western Europeans are overrepresented at FTDNA. I know this situation quite well as I'm an admin there that deals precisely with this haplogroup. Also, it's not just J-Z631 showing distribution in "Celtic" or Northwestern European regions. Prominent examples are some other J-L283 subclades, as can be seen: Y23094, PH1602, CTS11100.
Truth to be said Y23094, PH1602, and CTS11100 are likely related to Urnfielders (what I was briefly mentionning in one of my first post here), their diffusion hapenned earlier than Z631 which is more likely Hallstatt/La Tène related.
After, it depends where you place the origin of the Kelts ... with Urnfielders or with Hallstatt-peoples, which is more an historical debate than a DNA-question.
I'm not discounting the possibility of J-Z631 = Celtic (or any other J-L283 subclade for that matter).
I don't like the "=" symbol for haplogroup and populations ... it is an oversimplification, caldes can partially enter as they can partially exit cultures. I'm picky, but I rather prefer it this way "J-Z631
⊂ Hallstatt". And even, I'm not fully satisfied with this notation.
However, in today's age, we could use evidence from ancient DNA to support theories or jump into conclusions, which currently doesn't exist in this case, when in fact those regions are some of the better tested.
Ancient DNA is precious and usefull, but it says things, and there is things it cannot say ! It is important to separate what it can do and what it can't.
The depth of coverage is way too faint (even in well covered area) to be able to exclude things. Ancient DNA is usefull for dominant haplogroups, not for "rare" ones (here you just rely on luck with ancient DNA, especially in cultures that were praticing cremation).
Ancient DNA is usefull to confirm the presence of an haplogroup, in no way it can infirms the presence of an haplogroup ... at best it can indicates it is highly unlikely there by finding it somewhere else in a totally unrelated context.
I also linked you the upcoming paper a few days ago which is supposed to be centered around the Celts and there is zero J-Z631.
This list perfectly illustrates my point if you think of it in a statistical way.
For instance, you have haplogroups in this list that only appear once in the full list (quick example : G-U1*) ... which means that a lot of haplogroup with a similar frequency have been missed by this sampling.
*I didn't checked for subclades if they appeared, but anyway ... there is a lot of one-time-found haplogroup in this list.
It is like hopping to find a sample of the line from M241 to L283 in the 9500-5500BP timeframe, not impossible ... but it would require a lot of luck, because L283 as the only surviving branche imply that the size of M241 population that will give L283 remained fairly small.
Also note that the list you quote contains obvious misasignation of haplogroup (ex: Z2432) ... thus I would not use that too much, it is a noisy information. I would consider it carefully.