That model involving Morocco LN in modern Sicilians is really curious. Morocco LN seems a mix of Barcin N-like (which they would have used as Anatolian Neo reference) and Taforalt-like, with some Natufian-like and traces of WHG. Apparently the problem in this case was the lack of another Natufian-related reference in the model, which forced the algorithm to pick up Morocco LN, which in turn could have been "drained" by this another reference if it were used.
Well, that would be a bit like saying that we can't use North African sources to model their contribution to the gene pool of some place because they won't be 100% made by indigenous components. By that token, even Natufian would be "misleading", because it was also ~70% directly related to the Anatolian HGs further back. PPNB is the earliest Neolithic culture in the region. Nonetheless, the picture won't change if you use Natufian instead. Since Natufians were long gone (or rather changed into something else) by the time of the Neolithic and Chalcolithic expansion, if you use Natufian admixture you will get lower percentages, but no realistic interpretation of such a model would consider that there was actually direct Natufian contribution to Neolithic or Chalcolithic populations, that'd be totally anachronistic, so you would have to consider at least 2x the amount of Natufian to have a realistic estimate of the Levantine-related gene flow into some population compared to some earlier period.
This ratio seems possible imo. Something like 1.7x. Unless there were flows in both directions between AAF/Boncuklu and ACF/Barcin periods. The former (AAF) seems to lack Natufian-related ancestry. So, the first flows North, from Levant, could perhaps have been slightly richer in Natufian-related ancestry (?), and they could have occurred at the same time there were flows South from somewhere in Anatolia. PPNB already had some CHG/Iran-related ancestry (as AAF/Bonculklu had). However, flows in both ways at the same time don't seem likely, and actually the proportion of CHG/Iran-like ancestry in AAF seem to fit well the proportion of this component in PPNB when non-Natufian ancestry is considered. So, indeed, it looks like there was this first AAF->pre-PPNB flow followed by a PPNB(-like)->Anatolia flow, in turn followed by another increase in both Natufian-related and CHG/Iran-related ancestries, resulting in Tepecik.
We could also consider that clines where/are common, at the same time there're important unsampled areas (such in part of Fertile Crescent). It'll be interesting to know how people from these other areas looked like, and if they could somehow illuminate the scenario.
By the way, according to Feldman et al. and also G25 PCA, Barcin N (ACF) already had Levant-PPNB related ancestry, which Boncuklu (AAF) lacked, and also CHG/Iran-related, which Boncuklu had. Tepecik, in turn, had an excess of both ancestries compared to Barcin N, which evidences a continuous ~East-West flow (bringing both Natufian-related and CHG/Iran-related ancestries) from somewhere else into parts of Anatolia, between 10200-8300 years ago. What I think is curious is this "coincidence" in both Natufian and CHG/Iran increase into Tepecik (did they arrive independently?), that's why I speculate a possible (already mixed) source somewhere in Fertile Crescent, comparatively heavier in CHG/Iran-like and Natufian-like, and poorer in Anatolian-like). But again, we cannot rule out independent sources, of course, also because AAF had only CHG/Iran, while ACF had both at similar proportion (which means at some point Natufian ancestry has grown more than CHG/Iran, suggestive of independent movements, at least till some time before ACF period).
Anyway, apparently this ~E-W flow continued, which could perhaps explain why Isparta prefers Tepecik over Barcin largely, while modern Sicilians seems to be a mix of a Barcin-like (supposedly arrived earlier) and a Tepecik-like (supposedly arrived in post-Neo and associated to ABA?). We'd be talking of an important influence of both SBA and ABA ancestries over South Italy in general.
Uniparental markers may provide clues on this, as well as some TMRCAs (when decently interpreted). An example that comes to my mind is G-M406. Despite being so old (slightly older than the "cousin" clade G-P303, more common in Europe), it was not identified in Neolithic Europe thus far, but it shows up above traces in South Italy and Greece (it's also present in other parts of Europe, of course; in lower frequencies though). I wonder if it could be related to this possible more eastern/Tepecik-like source.
I think once Lazaridis' pre-print on Dzudzurna finally comes out, there are going to be some rethinking of these models. Considering that Dzudzurna is very similar to Anatolian_N. While Natufian is Dzudzurna, plus Ancestral North African. Perhaps Anatolian_N is an isolated Dzudzurna-like population, and not necessarily a mixture of WHG and Natufian.
Precisely. According to Lazaridis et. al., Anatolian and Dzudzuana were virtually the same, the former descending from the latter, with very little input from other sources. In this sense, Anatolian would be kind of a "Dzudzuana relic".
However, Natufian would be mostly Dzudzuana plus some Taforalt, which in turn also had Dzudzuana (at ~55%). Both Dzudzuana and Taforalt had "deep ancestry" (BE-like or something), so at the end both would be a mix of WHG-like and deep ancestry, if I got it correctly.
I never bought this idea that AHG were basicall WHG+Natufian. It makes very little sense, and especially WHG is very divergent from AHG. I think Lazaridis et al.'s remark on the Dzudzuana pre-print that the Neolithic Anatolians (I assume he means the average typically Anatolian ancestry) resulted from a Dzuduana-like source added to something more WHG-like (but not WHG) and more Basal Eurasian is more credible.
IIRC, Lazaridis et al. suggested Anatolians and Dzudzuana formed a clade. Dzdzuana itself would be a mix of this WHG-like ancestry (~72%) with "deep ancestry", and that's why Anatolians also had WHG-like ancestry (taked directly from Dzudzuana ancestors).
By the way, Ygorcs, it would be really great if anyone of us could learn how to use qpAdm and other tools. This seems a good start:
https://bodkan.net/admixr/articles/tutorial.html (check also D statistic and f4 statistic)