How do you explain the complete absence of T-Y16897 in Africans, except Somalis then? While I agree with you that we should be skeptical until further information comes out, aren't you making the same mistake by assuming that if we test the Rendille and Southern Cushites that they test positive for T-Y44591?I don’t think T-Y44591 is Arabian at all.
It so far has a demographic expansion exactly like R-Y8452 - a non-Afroasiatic branch (R-Y7771) and a Semitic branch, with respective Jewish specific (R-FGC20980) and Arab subclades.
The Somali Y44591 ancestor is already bound to head to a period o civilizations. It better matches matches the Ethiosemitic entrance into Northeast Africa. It also post dates the camel in the Horn of Africa. But if Habeshas don’t carry it - and it’s near non-existence especially outside Arabia, I doubt it has anything to do with Ethiosemites. I am very certain if we have a Rendille T-M70 year, they will increase the mrca by a couple hundred years, and up to several thousand.
It is very useless to use Arab matches as evidence of a non-Cushitic origin - every Somali subclade will have Arabs, like the Hawiye V32 guys are sharing with an Arab tribe from Yemen (Hubaysh). Clearly however the subclade is Somali.
If the ydna is
1) too localized
2) lacks a wide ethnic distribution and regional character
It can be ruled out as having a Eurasian origin.
If it is indeed Semitic, with its age we would expected for the largest demographic testers on ftdna (Jews and peninsular Arabs) to have shown up. Arabs are enriched for all ydna in Africa, even super rare African ydna like E-V16 and some unknown Central African HG ydna. If it is truly not from Africa - you need a wide ethnic distribution otherwise. And Y44591 does not have that - Z19971 does.
Statistically it is very unlikely Y44591 to be from Eurasia as it stands. Especially without the T-L208 Southeast Africans with Cushitic ancestry like the Karamoja Nilote and Iraqws and Bantus untested. The Arsi Oromo and other Ethiopian T-M70 heavily belong to young Semitic and West Asian subclades - not shocking because Arsi Oromos have a lot of J-P58 and even J2-M172, and very little T-M70. The Saho subclade is hard to make sense of, but does nothing to really dismiss the case at hand. Sahos and Afars have a lot of Ethiosemitic ancestry and have many clans with various P56 and P58 subclades.
Somali J guys also belong to a major Khawlan branch that they share with the other sibling clade housing a Ma’rib Khawlan tribe. The Somali-Khawlan mrca is way too young to be from classical pre-Arab South Arabia, it almost certainly comes from some Khawlan Arab that came to the Horn recently.
On Sahra and Afars - Afars from within the Afar region less than 100 miles from Djibouti are around 25% J-M304, and they are very similar to Amharas and Habeshas ydna-wise, the major difference is really just founder-effected for specific E-V6 and E-V22 and because of this have less E-M34 and E-V42. These Afars lack much T (around 5%) and have much more A3b2 (more than 10%) - so Afars in Djibouti must have gotten a lot of their T from Somalis, which is why they are so close.
Arab Y17859 also come from a diversity of clans - some are Hubaysh, others have a surname Badawi (Bedouin), an unidentified guy from Tripolitania, and Yemenis. So some few Arab guys from Bani Malik doesn’t mean anything - otherwise we have an easier case for E-Y17859 being Arabian.
The Garhajis T-FGC92488 subclade and the Dir T-BY181210* are past 1600 and now have an mrca of 1900 years. I know there is a Habr Gidir T-L208 aside from the Djibouti Hawiye T-L208, because of this I think T-BY18210 Somali ancestor will go further back and soon E-Y17859 and T-BY181210 will have similar mrca - since they already have identical similar formation dates.
If T-Y16897 has a TMRCA of 6900 bp (4000-5000BCE), then that would still mean some Eurasian HGs or pastoralists came fairly recently into the Somali Genepool.