🧬 Stable qpAdm Models of Ashkenazi Jewish Ancestry – Ignoring PCA No More

[Lard]

🧬 Realistic qpAdm Models of Ashkenazi Jewish Ancestry — Grounded in PCA, Not Cherry-Picking

🧠 Why This Matters
Ashkenazi Jews are the only major population where:


Most genetic studies ignore their PCA position, which clearly places them on top of or directly adjacent to Southern Italians
Instead, they use Northern Italians or even French as the “European” reference
This artificially inflates Levantine ancestry to 50–60%, despite PCA showing otherwise


These modeling choices reflect prior assumptions or narrative pressures more than empirical reality. PCA plots, which are unsupervised and data-driven, consistently place Ashkenazi Jews overlapping with Southern Italians — not halfway between Europe and the Levant.

The qpAdm models below are built on that PCA reality, and produce stable results with both modern and ancient Levantine sources.

---

📌 Summary of Results

These are all 2-way qpAdm fits, showing strong internal consistency:


~75–78% Southern Italian–like (Roman-era Jews and converts)
~22–25% Levantine (Israelite/Canaanite core)
Feasible p-values and strong standard errors for a highly admixed population


They match both:

✅ PCA structure: Ashkenazi Jews cluster with Southern Italians, not French or North Italians
✅ Historical records: Jews lived and proselytized in Southern Italy from the Roman through Byzantine periods


⚠️ Note: These models do not yet include Eastern European (~10–15%) or Iranian/Caucasus (~5–10%) ancestry — those will be included in upcoming 3-way models.

---

📊 Model 1: Modern Levantine Proxy
Target: Jew_Ashkenazi.HO
Sources:
• Italian_South.HO — 75.12% ± 5.29%
• Lebanese_Muslim.HO — 24.88% ± 5.29%
p-value: 0.1252 ✅

https://i.ibb.co/M0ZfVqC/1.png

---

🏺 Model 2: Bronze Age Coastal Levantine (Philistine-admixed)
Target: Jew_Ashkenazi.HO
Sources:
• Italian_South.HO — 77.39% ± 6.38%
• Israel_Ashkelon_LBA.AG — 22.61% ± 6.38%
p-value: 0.0545 ✅

https://i.ibb.co/S4FnRm4s/2.png

---

⛰️ Model 3: Inland Bronze Age Levantine (likely Israelite/Canaanite)
Target: Jew_Ashkenazi.HO
Sources:
• Italian_South.HO — 78.25% ± 10.81%
• Israel_C_o2.AG — 21.75% ± 10.81%
p-value: 0.2843 ✅

https://i.ibb.co/G43Kn5wk/3.png

---

More to come soon: 3-way models incorporating Slavic and Caucasus-related ancestry.

[celticdragongod]

Interesting. Did you run these models yourself?

[Lard]

Interesting. Did you run these models yourself?

Yes, I ran all of these models myself using AdmixLab on the Illustrative DNA platform.

I did it because what's being passed off as science in many published qpAdm studies on Ashkenazi Jews is, frankly, systematic misrepresentation. Here's the problem:

Every major population is modeled where it plots on PCA — except Ashkenazi Jews.
Southern Italians? Modeled with Southern Italian ancestry.
Greeks? Modeled with Greek or Balkan ancestry.
Georgians? Modeled with Caucasus ancestry.
Levantines? Modeled with Levantine ancestry.
Ashkenazi Jews? Suddenly modeled with Northern Italians or French — even though they plot directly on top of Southern Italians in every PCA from Reich Lab to Lazaridis to Behar.

That isn’t an oversight. That’s by design.

Using French or North Italians as the “European” source artificially pushes Ashkenazi Jews toward the Levant, inflating Levantine ancestry to 50–60% — even though their PCA position shows otherwise. PCA is unsupervised. It reflects real genetic distances. And it consistently places Ashkenazi Jews in or overlapping with Southern Italy, not halfway between Europe and the Levant.

Choosing reference populations that ignore this violates the basic logic of qpAdm, which assumes the sources bracket the target in PCA space. When that assumption is broken, the model is invalid — and yet it keeps getting published.

That’s not just flawed methodology — it’s scientific fraud. It reflects narrative pressure, not empirical rigor.

My models correct for this by using Italian_South.HO as the core European source. The resulting ancestry estimates — ~75–78% Southern Italian–like, ~22–25% Levantine — are PCA-consistent, historically grounded, and qpAdm-feasible. They reflect Roman-era Jewish ancestry plus local admixture in Southern Italy, not a fantasy of uninterrupted Israelite descent.

3-way models including Slavic and Caucasus-related ancestry are coming soon.

[celticdragongod]

So what results did you get when you ran Northern Italians and French in place of Southern Italians?

[Lard]

So what results did you get when you ran Northern Italians and French in place of Southern Italians?

I don’t run models using Northern Italians, Tuscans, Sardinians, or French to represent Ashkenazi Jewish ancestry — and there's a reason:
It’s not empirically supportable.

You can check any of the older studies — Behar, Hammer, Ostrer, etc. — and you’ll see the same flawed setup repeated:
They use Northern Italians or French as the “European” reference, which always produces a ~50/50 split with Levantines.
That’s because these models ignore PCA placement — which clearly shows Ashkenazi Jews cluster directly with Southern Italians, not in between Europe and the Levant.

Even Andrei DNA, who’s also a member here on Apricity, repeats this recycled framework in his video:


https://www.youtube.com/watch?v=5gxr79HfpYQ

In it, he classifies Ashkenazi Jews as a “Middle Eastern” population — which is completely detached from PCA reality.
Ashkenazim are no more Middle Eastern than Calabrians or Maltese, both of whom plot right next to them. Only by choosing intentionally distant references do you get that artificial outcome.

My models reflect the actual data structure:

✅ ~75–78% Southern Italian–like ancestry
✅ ~22–25% Levantine (Israelite/Canaanite core)
✅ Feasible p-values and low standard errors
✅ Consistent with known Roman/Byzantine diaspora history

These are stable 2-way models. I’ll be posting 3-way models soon that include Slavic and Caucasus input (~10–15% and ~5–10%, respectively).

Bottom line: I don’t entertain models that ignore both PCA and history just because prior studies recycled the same lazy reference sets. Ashkenazi Jews clearly reflect a Southern Italian + Levantine profile, not a fantasy mix of Northern Italians and Bedouins.

[Dušan]

They have minor Mongoloid admixture, too.

[Lard]

They have minor Mongoloid admixture, too.

Absolutely — as do the Slavic populations Ashkenazim admixed with in Northeastern Europe. That’s the most likely source of the minor Mongoloid or East Eurasian signal sometimes detected in Ashkenazi Jews — not the Khazars.

My qpAdm models here are 2-way, focusing on the dominant ancestry components reflected in PCA. Low-level admixture like that — often under 5% — is real, but difficult to model directly. qpAdm tends to absorb minor or divergent signals into the nearest major population in the model. So unless you specifically build a model to isolate those traces, they usually get folded into broader categories like Slavic or Steppe-related ancestry.

As for the Khazar theory, it can’t be tested in any rigorous way as of 2025. We still have virtually no ancient Khazar genetic data, so there's no scientific way to confirm or rule out Khazar input in Ashkenazi ancestry.

And even if we did have samples, it wouldn’t be straightforward — because Khazaria was a vast, multi-ethnic confederation, not a single people. It included:


Turkic-speaking elites (probably Oghuric)
Iranian-speaking Alans and Sarmatians
Slavs
Finno-Ugric tribes
North Caucasian peoples (Avars, Lezgins, Circassians, etc.)
Jews, Armenians, Greeks, and others through trade and resettlement


So even if “Khazar ancestry” existed, we’d have to ask: which part of Khazaria? Genetically, the concept is too vague to be useful without a clearer source population.

What can be modeled — and what’s supported by qpAdm, PCA, ADMIXTURE, and historical records — is the following stable ancestry profile for Ashkenazi Jews:


~65–75% Southern Italian–like ancestry (Roman-era Jews and converts)
~15–25% Levantine ancestry (speculative Israelite/Canaanite core)
~10–15% Slavic + minor Central/East Asian input
~5 –10% Central European ancestry
~5–10% Iranian-related ancestry

This profile is internally consistent, historically grounded, and reflected across multiple modeling strategies.

[Dušan]

Absolutely — as do the Slavic populations Ashkenazim admixed with in Northeastern Europe. That’s the most likely source of the minor Mongoloid or East Eurasian signal sometimes detected in Ashkenazi Jews — not the Khazars.

My qpAdm models here are 2-way, focusing on the dominant ancestry components reflected in PCA. Low-level admixture like that — often under 5% — is real, but difficult to model directly. qpAdm tends to absorb minor or divergent signals into the nearest major population in the model. So unless you specifically build a model to isolate those traces, they usually get folded into broader categories like Slavic or Steppe-related ancestry.

As for the Khazar theory, it can’t be tested in any rigorous way as of 2025. We still have virtually no ancient Khazar genetic data, so there's no scientific way to confirm or rule out Khazar input in Ashkenazi ancestry.

And even if we did have samples, it wouldn’t be straightforward — because Khazaria was a vast, multi-ethnic confederation, not a single people. It included:


Turkic-speaking elites (probably Oghuric)
Iranian-speaking Alans and Sarmatians
Slavs
Finno-Ugric tribes
North Caucasian peoples (Avars, Lezgins, Circassians, etc.)
Jews, Armenians, Greeks, and others through trade and resettlement


So even if “Khazar ancestry” existed, we’d have to ask: which part of Khazaria? Genetically, the concept is too vague to be useful without a clearer source population.

What can be modeled — and what’s supported by qpAdm, PCA, ADMIXTURE, and historical records — is the following stable ancestry profile for Ashkenazi Jews:


~65–75% Southern Italian–like ancestry (Roman-era Jews and converts)
~15–25% Levantine ancestry (speculative Israelite/Canaanite core)
~10–15% Slavic + minor Central/East Asian input
~5 –10% Central European ancestry
~5–10% Iranian-related ancestry

This profile is internally consistent, historically grounded, and reflected across multiple modeling strategies.

They didnt get East Eurasian admixture from Slavs.
They lived among Poles and Belarussians who dont have EE admixture.
Russians do have some, but vast majority of Ashkenazi Jews lived in area of former Polish Lithuanian Commonwealth.

Ashkenazi Jews got EE admixture either by assimilating Khazarian elite, either by interactions on trade Silk road.

https://www.reddit.com/r/illustrativeDNA/comments/1m3214x/is_this_rare_for_half_ashkenazi_half_sephardi_jew/

[Lard]

They didnt get East Eurasian admixture from Slavs.
They lived among Poles and Belarussians who dont have EE admixture.
Russians do have some, but vast majority of Ashkenazi Jews lived in area of former Polish Lithuanian Commonwealth.

Ashkenazi Jews got EE admixture either by assimilating Khazarian elite, either by interactions on trade Silk road.

https://www.reddit.com/r/illustrativeDNA/comments/1m3214x/is_this_rare_for_half_ashkenazi_half_sephardi_jew/

A few quick clarifications:


I’m not chasing Reddit threads or Wikipedia footnotes. I build my models from the actual genotype datasets (HO, AG, SG, DG) and the published qpAdm / ADMIXTURE pipelines.
You skipped the enormous Ukrainian Jewish population of the Pale of Settlement. By 1900 roughly 40 % of all Ashkenazim lived in today’s Ukraine — far more than in ethnic Poland or Belarus.
“Poles and Belarusians have zero East‑Eurasian” isn’t accurate. Modern genome panels consistently show ~2–4 % Nganasan‑like / Siberian ancestry in Poles, Belarusians, Lithuanians, and even some Czechs. It’s subtle, but it’s there and it does get sucked up by qpAdm when you use those groups as sources.

Historical perspective:


The Polish‑Lithuanian Commonwealth absorbed Tatar service nobles (Lipka Tatars) and Cossack hosts who carried Golden Horde ancestry.
The Grand Duchy of Moscow — later the Russian Empire — inherited far more of that steppe‑Tatar genetic layer through centuries of tribute, intermarriage, and migration.
Even the “Khazar elite” you mention were themselves a multi‑ethnic Turkic power created after the westward surge of the Göktürk and later Mongol polities.
If you invoke Silk‑Road contacts or Genghis‑era events, remember: those waves swept the entire steppe from the Dnieper to the Volga. Ashkenazi communities trading through Kyiv, Lviv, Berdychiv, and Odessa were not isolated from that gene flow.

Bottom line:
Minor East‑Eurasian (~2–5 %) signals in Ashkenazim can come via Slavic intermediaries, Cossack‑Tatar mixing, and yes, maybe a Khazar fragment — but the data don’t let us disentangle those threads yet. Until we get a high‑coverage Khazar genome, the safest empirical path is still:


Southern‑Italian–like ≈ 65–75 %
Levantine ≈ 15–25 %
Slavic / NE‑European (with that 2–4 % EE signal folded in) ≈ 10–15 %
Central European ≈ 5–10 %
Iranian‑related ≈ 5–10 %

[rothaer]

Yes, I ran all of these models myself using AdmixLab on the Illustrative DNA platform.

I did it because what's being passed off as science in many published qpAdm studies on Ashkenazi Jews is, frankly, systematic misrepresentation. Here's the problem:

Every major population is modeled where it plots on PCA — except Ashkenazi Jews.
Southern Italians? Modeled with Southern Italian ancestry.
Greeks? Modeled with Greek or Balkan ancestry.
Georgians? Modeled with Caucasus ancestry.
Levantines? Modeled with Levantine ancestry.
Ashkenazi Jews? Suddenly modeled with Northern Italians or French — even though they plot directly on top of Southern Italians in every PCA from Reich Lab to Lazaridis to Behar.

That isn’t an oversight. That’s by design.

Using French or North Italians as the “European” source artificially pushes Ashkenazi Jews toward the Levant, inflating Levantine ancestry to 50–60% — even though their PCA position shows otherwise. PCA is unsupervised. It reflects real genetic distances. And it consistently places Ashkenazi Jews in or overlapping with Southern Italy, not halfway between Europe and the Levant.

Choosing reference populations that ignore this violates the basic logic of qpAdm, which assumes the sources bracket the target in PCA space. When that assumption is broken, the model is invalid — and yet it keeps getting published.

That’s not just flawed methodology — it’s scientific fraud. It reflects narrative pressure, not empirical rigor.

My models correct for this by using Italian_South.HO as the core European source. The resulting ancestry estimates — ~75–78% Southern Italian–like, ~22–25% Levantine — are PCA-consistent, historically grounded, and qpAdm-feasible. They reflect Roman-era Jewish ancestry plus local admixture in Southern Italy, not a fantasy of uninterrupted Israelite descent.

3-way models including Slavic and Caucasus-related ancestry are coming soon.

Exactly. This is what I've been claiming for years on this topic.

(Btw. historically you have loads of Roman era Jews in Southern Italy and Rome but not in Northern Italy.)

Now, if you take into account that you also had converts in the Eastern Mediterranean outside of the Levant, say Anatolia and Greece, the 1/4 Levant part in modelling will likely shrink to abt. 10%.

Needless to say that also the Levant like Libanon and Syria contained non-Israelits that had been converted to Judaism, i. e. not even the Levant DNA proportion does equal the ancestry proportion of original Judaea Jews.

As a rough conclusion I can imagine that Ashkenazi Jews do have as much connection to Judaea as had early Christians: Both hail culturally from there, the initial core personnel was derived from there and most of the followers were gained by conversions within the Roman Empire, predominantly in the Eastern Mediterranean including Central and Southern Italy.

[Opie]

This is essentially what I've been claiming for years on this topic.

Now, if you take into account that you also had converts in the Eastern Mediterranean outside of the Levant, say Anatolia and Greece, the 1/4 Levant part in modelling will likely shrink to abt. 10%. (Needless to say that also the Levant like Libanon and Syria contained non-Israelits that had been converted to Judaism.)

As a rough conclusion I can imagine that Ashkenazi Jews do have as much connection to Judaea as had early Christians: Both hail culturally from there, the initial core personnel was derived from there and most of the followers were gained by conversions within the Roman Empire, predominantly in the Eastern Mediterranean including Central and Southern Italy.

You can’t be a Jew unless your mother is a Jew and even that is seen not enough by Jews. Jews are an ethnoreligious group that do not accept outsiders. This conversion story must be researched.

[Peterski]

Then how do you explain Near Eastern origins of Ashkenazi Jewish haplogroups?:

https://genoplot.com/discussions/topic/15102/major-jewish-y-lineages?page=1

https://www.sciencedirect.com/science/article/abs/pii/S2773044125000713#f0035

[Lard]

Then how do you explain Near Eastern origins of Ashkenazi Jewish haplogroups?:

https://genoplot.com/discussions/topic/15102/major-jewish-y-lineages?page=1

https://www.sciencedirect.com/science/article/abs/pii/S2773044125000713#f0035


Let’s Examine These So-Called “Near Eastern” Haplogroups in Ashkenazim — and See If They Align with Their PCA Plotting in Southern Italy

Ashkenazi Jewish Y-DNA haplogroups like J1, J2, and E-M123/M34 are often cited as proof of “Near Eastern” ancestry. But this framing fails basic population genetics. Haplogroups alone don’t determine ancestry — autosomal DNA does. And when we examine where these haplogroups are found and how they behave in other populations, the “Near Eastern” label begins to unravel.

1. Southern Italy Is a Hotspot for J1 — But Nobody Calls South Italians Levantine

Multiple studies — including Boattini et al. (2013) and Di Giacomo et al. (2002) — show substantial J1 frequencies across Southern Italy:


Agrigento, Sicily – 12%
Pescara, Abruzzo – 15%
North Gargano, Apulia – 17.2%
Reggio Calabria – 9.1%
Paola, Calabria – 11.1%
Benevento, Campania – 8.5%
Pistoia, Tuscany – 7.5%
Bologna, Emilia-Romagna – 7%


These numbers match or exceed the J1 frequency in Ashkenazi Jews (~5–8%). Yet no one refers to Southern Italians or Greeks — populations that Ashkenazim cluster near on PCA — as "Levantine."

Despite these high J1 frequencies, virtually no one investigates or questions J1 in Southern Europe. But when Ashkenazim carry the same lineages, it's treated as proof of Near Eastern roots. That’s selective and unscientific.

2. Ashkenazi J2 Is Even More Common — and Strongly European

J2 is more frequent than J1 among Ashkenazim. Dominant subclades like J-M67 and J-M92 are widespread in:


Southern Italy
Greece
Western Anatolia


These are precisely the populations that Ashkenazim cluster with in autosomal PCA. If Greeks and South Italians with the same J2 lineages aren’t labeled “Near Eastern,” why are Ashkenazim?

3. Founder Effects Skew Haplogroup Frequencies

A perfect example: the Arbëreshë, a Tosk Albanian minority in Southern Italy. Tosk Albanians today have less than 3% E-M123. Yet in some Arbëreshë villages, it exceeds 20%.

Why? Founder effect and drift — not Levantine ancestry.

This is the same subclade (E-M123) often labeled “Near Eastern” in Ashkenazim. Yet the Arbëreshë remain Southeast European in autosomal profile — just like Ashkenazim.

4. Napoleon Had E-M34 — Nobody Says He Was Levantine

Napoleon Bonaparte, of Corsican-Italian descent, carried haplogroup E-M34 — a cousin clade of E-M123. It’s also found in ~10% of Ashkenazi men.

Yet no one argues Napoleon was Levantine. Again: haplogroups do not equal total ancestry.

5. Bottom Line: Autosomal DNA Tells the Real Story


J1 and E-M123 are common in Southern Italy and Greece
Ashkenazim cluster autosomally with those same populations
J2 in Ashkenazim is clearly aligned with Greco-Roman Europe
Founder effects explain inflated frequencies in small populations
Haplogroups must be interpreted in genetic, historical, and geographic context


Calling Ashkenazi Y-DNA “Near Eastern” while ignoring identical haplogroups in Greeks and South Italians — populations with whom Ashkenazim share autosomal ancestry — is not science. It’s cherry-picking.

Key Sources:


Boattini et al. (2013) – "Uniparental Markers in Italy"
Di Giacomo et al. (2002)
Capelli et al. (2007)
Brisighelli et al. (2012)
Battaglia et al. (2008)

[Lard]

You can’t be a Jew unless your mother is a Jew and even that is seen not enough by Jews. Jews are an ethnoreligious group that do not accept outsiders. This conversion story must be researched.

That statement about Jewish identity is oversimplified and historically inaccurate.

1. Modern Reality:
Today in the United States, around 70% of non-Orthodox Jews are married to non-Jews, according to data from Pew Research. Intermarriage is extremely common in Reform and secular Jewish communities, and many children of mixed marriages are raised as Jews — especially if one parent identifies Jewishly, regardless of gender.

2. Israeli Law of Return:
Under Israel’s Law of Return, Jewish citizenship is granted to anyone with at least one Jewish grandparent, on either the maternal or paternal side. Religious conversion is not required. The law intentionally mirrors how Nazi Germany defined Jews under the Nuremberg Laws — to ensure broad protection.

3. Ancient Jewish Identity Was Not Matrilineal:
The idea that Jewishness is exclusively passed through the mother is a rabbinic rule that developed after the fall of the Second Temple (70 CE). In earlier periods — including during the time of the Greco-Roman world — descent could be patrilineal or mixed. Biblical tribes were typically traced through the father, and Jewish men often had children with non-Israelite women.

So yes, Judaism is an ethnoreligious group with internal boundaries — but those boundaries have shifted dramatically over time, and are not biologically rigid. The idea that Jews "do not accept outsiders" is not true historically or in modern legal or social terms.

[Lard]

Exactly. This is what I've been claiming for years on this topic.

(Btw. historically you have loads of Roman era Jews in Southern Italy and Rome but not in Northern Italy.)

Now, if you take into account that you also had converts in the Eastern Mediterranean outside of the Levant, say Anatolia and Greece, the 1/4 Levant part in modelling will likely shrink to abt. 10%.

Needless to say that also the Levant like Libanon and Syria contained non-Israelits that had been converted to Judaism, i. e. not even the Levant DNA proportion does equal the ancestry proportion of original Judaea Jews.

As a rough conclusion I can imagine that Ashkenazi Jews do have as much connection to Judaea as had early Christians: Both hail culturally from there, the initial core personnel was derived from there and most of the followers were gained by conversions within the Roman Empire, predominantly in the Eastern Mediterranean including Central and Southern Italy.

Exactly — and this is the key point too many people are still avoiding: even the small Levantine signal in Ashkenazi Jews doesn’t equal “Israelite ancestry.” It’s a generic Eastern Mediterranean component — one that could just as easily reflect converts from Syria, Lebanon, Anatolia, or Hellenized coastal cities as it does Judea itself.

Once you strip away the romanticism and look at the genetics honestly, it becomes clear that the core of Ashkenazi ancestry isn’t Levantine. It’s overwhelmingly Southern Italian-like — and the reason is simple: the bulk of the ethnogenesis happened in the Roman Empire, outside of Judea.

The myth of direct, uninterrupted Israelite descent falls apart under even mild scrutiny. Just like early Christianity, Judaism spread through urban centers, trade routes, and cultural networks — mostly by conversion. That’s why Roman-era Jewish communities existed all across Southern Italy and Greece, and why genetic data places Ashkenazi Jews directly on top of those populations in PCA.

So when models return ~20–25% Levantine ancestry (at most), that’s already a generous upper bound. A significant chunk of that could be Eastern Mediterranean converts — not Israelites. Realistically, we’re probably looking at 10–15% Judean ancestry at best, with the rest being local Southern Italian, Anatolian, and Balkan inputs.

The cultural origin was Judea — the genetic reality was Roman.

[majevica]

Yes, they're like an West Asian shifted version of the already heavily West Asian shifted South Italians

[Lard]

Yes, they're like an West Asian shifted version of the already heavily West Asian shifted South Italians

I see what you're getting at, but I’d refine that description — it oversimplifies both Southern Italians and Ashkenazi Jews.

Southern Italians do carry ancient West Asian-related ancestry, but it’s the product of long-term Mediterranean interaction — Neolithic farmers, Bronze Age Greeks, Phoenicians, Roman cosmopolitanism — not recent or direct West Asian gene flow. So rather than calling them “heavily West Asian–shifted,” it’s more accurate to say they sit on the eastern edge of the European Mediterranean cline.

As for Ashkenazi Jews, especially Eastern European ones, a clearer description would be:


Iranian-shifted Southern Italians, with a subtle Central European (Rhineland German) component, and substantial Slavic admixture visible in autosomal DNA.

In PCA, they plot directly over Southern Italians, but with a slight northeast drift. This reflects minor Iranian/Caucasus-related ancestry, a modest Central European input, and a significant autosomal pull from Slavic populations, especially among Eastern Ashkenazim. That Slavic ancestry, while not strong enough to shift them away from Southern Italy on PCA, contributes to their distinct northeast-leaning position within the Southern European genetic zone.

And here’s the broader context:
Italy forms a clear north-south genetic cline, and it's part of a larger European continuum:


Southern Italians transition gradually into Central Italians (Lazio, Tuscany)
Central Italians flow into Northern Italians (Lombardy, Veneto)
Northern Italians overlap with Central Europeans (Swiss, Austrians, South Germans)
Central Europe connects into Eastern Europe, including West and East Slavic populations


Ashkenazi Jews fit firmly on the southern end of this cline. Their autosomal profile centers over Southern Italy, with moderate shifts toward the Balkans, Central Europe, and Eastern Europe — reflecting their formation in the Roman world followed by centuries of contact and admixture in the north and east.

So to sum it up more precisely:


Ashkenazi Jews are Southern Italian–centered in autosomal DNA, with minor Iranian/Caucasus-related shift, a light Central European pull (especially Rhineland German), and substantial Slavic autosomal ancestry layered through medieval-era admixture.

This interpretation is consistent with PCA, ADMIXTURE, and qpAdm results — when modeled using geographically and historically realistic source populations.

[Alenka]

So to sum it up more precisely:


Ashkenazi Jews are Southern Italian–centered in autosomal DNA, with minor Iranian/Caucasus-related shift, a light Central European pull (especially Rhineland German), and substantial Slavic autosomal ancestry layered through medieval-era admixture.

This interpretation is consistent with PCA, ADMIXTURE, and qpAdm results — when modeled using geographically and historically realistic source populations.
So why not include Slavic and Rhineland German into your models...
Until you do, your models cannot be a realistic breakdown of their autosomal ancesry.



🧬 Realistic qpAdm Models of Ashkenazi Jewish Ancestry — Grounded in PCA, Not Cherry-Picking

🧠 Why This Matters
Ashkenazi Jews are the only major population where:


Most genetic studies ignore their PCA position, which clearly places them on top of or directly adjacent to Southern Italians
Instead, they use Northern Italians or even French as the “European” reference
This artificially inflates Levantine ancestry to 50–60%, despite PCA showing otherwise


These modeling choices reflect prior assumptions or narrative pressures more than empirical reality. PCA plots, which are unsupervised and data-driven, consistently place Ashkenazi Jews overlapping with Southern Italians — not halfway between Europe and the Levant.

The qpAdm models below are built on that PCA reality, and produce stable results with both modern and ancient Levantine sources.

---

📌 Summary of Results

These are all 2-way qpAdm fits, showing strong internal consistency:


~75–78% Southern Italian–like (Roman-era Jews and converts)
~22–25% Levantine (Israelite/Canaanite core)
Feasible p-values and strong standard errors for a highly admixed population


They match both:

✅ PCA structure: Ashkenazi Jews cluster with Southern Italians, not French or North Italians
✅ Historical records: Jews lived and proselytized in Southern Italy from the Roman through Byzantine periods



---

📊 Model 1: Modern Levantine Proxy
Target: Jew_Ashkenazi.HO
Sources:
• Italian_South.HO — 75.12% ± 5.29%
• Lebanese_Muslim.HO — 24.88% ± 5.29%
p-value: 0.1252 ✅

https://i.ibb.co/M0ZfVqC/1.png

---

🏺 Model 2: Bronze Age Coastal Levantine (Philistine-admixed)
Target: Jew_Ashkenazi.HO
Sources:
• Italian_South.HO — 77.39% ± 6.38%
• Israel_Ashkelon_LBA.AG — 22.61% ± 6.38%
p-value: 0.0545 ✅

https://i.ibb.co/S4FnRm4s/2.png

---

⛰️ Model 3: Inland Bronze Age Levantine (likely Israelite/Canaanite)
Target: Jew_Ashkenazi.HO
Sources:
• Italian_South.HO — 78.25% ± 10.81%
• Israel_C_o2.AG — 21.75% ± 10.81%
p-value: 0.2843 ✅

https://i.ibb.co/G43Kn5wk/3.png

---

More to come soon: 3-way models incorporating Slavic and Caucasus-related ancestry.

[Lard]

So why not include Slavic and Rhineland German into your models...
Until you do, your models cannot be a realistic breakdown of their autosomal ancesry.




As I wrote in the original post:

“More to come soon: 3-way models incorporating Slavic and Caucasus-related ancestry.”

This thread is focused specifically on 2-way modeling, which is the format most commonly cited — both in public discussion and in academic studies — to describe the core ancestry structure of Ashkenazi Jews.

And with all due respect, the assertion that these models “cannot be a realistic breakdown of their autosomal ancestry” is not accurate.

In fact, these 2-way models already match the autosomal PCA position of Ashkenazi Jews, which is repeatedly shown to overlap directly with Southern Italians and Sicilians — not Northern Europeans or Levantines. That’s not speculation — it’s a well-documented pattern across PCA from Reich Lab, Behar et al., Lazaridis, and others.

My newest model — using Sicilian.HO instead of Italian_South.HO — fits even better and shows a slightly lower Levantine component:

https://i.ibb.co/h1WSX7cx/jj.png (https://ibb.co/dJ2RbWts)

This result supports the PCA-visible overlap between Ashkenazi Jews and Southern Italians, and improves statistical fit over the previous model. It also suggests that even Italian_South.HO may slightly overestimate Levantine ancestry.

Also, for reference:
Waldman et al. 2022 used a 3-way qpAdm and found:


~70% Southern Italian
~15% Lebanese
~15% Russian


So yes — Slavic and Central European input is real and acknowledged. It’s being incorporated next in carefully constructed 3- and 4-way models. But these current 2-way models already represent a realistic and PCA-consistent breakdown of Ashkenazi Jewish autosomal ancestry.

More coming soon.

[Alenka]

In fact, these 2-way models already match the autosomal PCA position of Ashkenazi Jews, which is repeatedly shown to overlap directly with Southern Italians and Sicilians — not Northern Europeans or Levantines. That’s not speculation — it’s a well-documented pattern across PCA from Reich Lab, Behar et al., Lazaridis, and others.
Years go, this forum had a member 7eleven or something like that.
His mother was Ashkenazi, and his father WASP.
7eleven autosomal PCA position was in the Romanian cluster.

According to your logic, he was in fact of Romanian ancestry. :rolleyes:



My newest model — using Sicilian.HO instead of Italian_South.HO — fits even better and shows a slightly lower Levantine component:

https://i.ibb.co/h1WSX7cx/jj.png (https://ibb.co/dJ2RbWts)

This result supports the PCA-visible overlap between Ashkenazi Jews and Southern Italians, and improves statistical fit over the previous model. It also suggests that even Italian_South.HO may slightly overestimate Levantine ancestry.

Also, for reference:
Waldman et al. 2022 used a 3-way qpAdm and found:


~70% Southern Italian
~15% Lebanese
~15% Russian


So according to you, early Jews heavily proselytized in Sicily, but didnt even touch Germans with a 10-foot pole.
Makes sense. :picard1:



So yes — Slavic and Central European input is real and acknowledged. It’s being incorporated next in carefully constructed 3- and 4-way models. But these current 2-way models already represent a realistic and PCA-consistent breakdown of Ashkenazi Jewish autosomal ancestry.

More coming soon.
How is Slavic and Central European input acknowledged, when none of your models account for it yet? :confused:

[Lard]

Years go, this forum had a member 7eleven or something like that.
His mother was Ashkenazi, and his father WASP.
7eleven autosomal PCA position was in the Romanian cluster.

According to your logic, he was in fact of Romanian ancestry. :rolleyes:



So according to you, early Jews heavily proselytized in Sicily, but didnt even touch Germans with a 10-foot pole.
Makes sense. :picard1:


How is Slavic and Central European input acknowledged, when none of your models account for it yet? :confused:


Years ago, this forum had a member “7eleven.”
His mother was Ashkenazi, and his father WASP.
7eleven autosomal PCA position was in the Romanian cluster.

According to your logic, he was in fact of Romanian ancestry.

This is garbage-tier logic.

You're comparing a single mixed individual to the PCA centroid of a stable population of over 12 million people. PCA doesn’t randomly label people “Romanian” — it places them based on real genetic distances, and Ashkenazi Jews plot directly over Southern Italians/Sicilians, consistently, in every major dataset: Reich Lab, Behar, Lazaridis, Lazaridis et al. 2022, etc.

Try again.


So according to you, early Jews heavily proselytized in Sicily, but didn’t even touch Germans with a 10-foot pole.
Makes sense.

No — what makes no sense is your reply. And frankly, I’m done entertaining your low-effort goalpost-shifting.

You demanded a model that includes Slavic or Central European input. So I cited one:
Waldman et al. 2022 — a peer-reviewed qpAdm model with:

~70% Southern Italian

~15% Lebanese

~15% Russian

That’s a Slavic population. It directly answers your complaint. And what do you do next? Instantly move the goalpost and whine about Germans instead — as if you weren’t just asking about Slavic or Central European input two posts ago.

You’re not debating — you’re flailing.

For the record:

Jews lived in Southern Italy and Sicily since the Roman Republic, with real communities, proselytism, and integration.

The Rhineland presence came far later, and yes, Ashkenazi Jews later picked up Slavic and Central European ancestry — which I already stated, repeatedly.

I'm currently building 3- and 4-way qpAdm models to include that layer properly, because I actually care about statistical fit — not dropping random populations in for narrative reasons like you seem to want.

Your entire reply is a bad-faith misreading. You’re quote-mining posts, ignoring everything that contradicts your script, and now pretending I claimed Jews “never touched Germans” because I referenced Waldman’s Slavic model.

You wanted a model with Slavic input? You got it.
You didn't like that it wrecked your snark, so now you're pretending the issue was Germans all along.

You're done. If your next post isn’t backed by qpAdm, PCA, or ancient samples, I’m not replying. Go troll someone else.


How is Slavic and Central European input acknowledged, when none of your models account for it yet?

Because I said, word for word:


“It’s being incorporated next in carefully constructed 3- and 4-way models.”

You literally quoted it — and then pretended it wasn’t there.

You demanded evidence of Slavic or Central European ancestry in Ashkenazi Jews. I gave you Waldman et al. 2022, a peer-reviewed qpAdm study that modeled:


~70% Southern Italian
~15% Lebanese
~15% Russian (Slavic proxy)


And instead of engaging with that evidence — the exact kind you claimed was “missing” — you ignored it and moved the goalposts.

That’s not debate. That’s trolling.

You’re not here to discuss the data. You’re here to twist quotes, ignore studies, and act smug while offering nothing of substance. I’ve backed up every claim with modeling, PCA positioning, and published research. You’ve brought snark and zero sources.

This was your final spoon-fed answer. I won’t waste another reply on someone who doesn’t read, doesn’t listen, and contributes absolutely nothing.

[Gannicus]

Let’s Examine These So-Called “Near Eastern” Haplogroups in Ashkenazim — and See If They Align with Their PCA Plotting in Southern Italy

Ashkenazi Jewish Y-DNA haplogroups like J1, J2, and E-M123/M34 are often cited as proof of “Near Eastern” ancestry. But this framing fails basic population genetics. Haplogroups alone don’t determine ancestry — autosomal DNA does. And when we examine where these haplogroups are found and how they behave in other populations, the “Near Eastern” label begins to unravel.

1. Southern Italy Is a Hotspot for J1 — But Nobody Calls South Italians Levantine

Multiple studies — including Boattini et al. (2013) and Di Giacomo et al. (2002) — show substantial J1 frequencies across Southern Italy:


Agrigento, Sicily – 12%
Pescara, Abruzzo – 15%
North Gargano, Apulia – 17.2%
Reggio Calabria – 9.1%
Paola, Calabria – 11.1%
Benevento, Campania – 8.5%
Pistoia, Tuscany – 7.5%
Bologna, Emilia-Romagna – 7%


These numbers match or exceed the J1 frequency in Ashkenazi Jews (~5–8%). Yet no one refers to Southern Italians or Greeks — populations that Ashkenazim cluster near on PCA — as "Levantine."

Despite these high J1 frequencies, virtually no one investigates or questions J1 in Southern Europe. But when Ashkenazim carry the same lineages, it's treated as proof of Near Eastern roots. That’s selective and unscientific.

2. Ashkenazi J2 Is Even More Common — and Strongly European

J2 is more frequent than J1 among Ashkenazim. Dominant subclades like J-M67 and J-M92 are widespread in:


Southern Italy
Greece
Western Anatolia


These are precisely the populations that Ashkenazim cluster with in autosomal PCA. If Greeks and South Italians with the same J2 lineages aren’t labeled “Near Eastern,” why are Ashkenazim?

3. Founder Effects Skew Haplogroup Frequencies

A perfect example: the Arbëreshë, a Tosk Albanian minority in Southern Italy. Tosk Albanians today have less than 3% E-M123. Yet in some Arbëreshë villages, it exceeds 20%.

Why? Founder effect and drift — not Levantine ancestry.

This is the same subclade (E-M123) often labeled “Near Eastern” in Ashkenazim. Yet the Arbëreshë remain Southeast European in autosomal profile — just like Ashkenazim.

4. Napoleon Had E-M34 — Nobody Says He Was Levantine

Napoleon Bonaparte, of Corsican-Italian descent, carried haplogroup E-M34 — a cousin clade of E-M123. It’s also found in ~10% of Ashkenazi men.

Yet no one argues Napoleon was Levantine. Again: haplogroups do not equal total ancestry.

5. Bottom Line: Autosomal DNA Tells the Real Story


J1 and E-M123 are common in Southern Italy and Greece
Ashkenazim cluster autosomally with those same populations
J2 in Ashkenazim is clearly aligned with Greco-Roman Europe
Founder effects explain inflated frequencies in small populations
Haplogroups must be interpreted in genetic, historical, and geographic context


Calling Ashkenazi Y-DNA “Near Eastern” while ignoring identical haplogroups in Greeks and South Italians — populations with whom Ashkenazim share autosomal ancestry — is not science. It’s cherry-picking.

Key Sources:


Boattini et al. (2013) – "Uniparental Markers in Italy"
Di Giacomo et al. (2002)
Capelli et al. (2007)
Brisighelli et al. (2012)
Battaglia et al. (2008)


Don't forget J2 is common in the Levant as well.


141891
141892

[Lard]

Don't forget J2 is common in the Levant as well.

141891
141892
Sure — and it’s just as common, if not more so, in Southern Italy, Crete, mainland Greece, and across the Caucasus. So what?

You can find J2 all over the Eastern Mediterranean and West Asia. It’s a broadly distributed macro-haplogroup, not a Levantine fingerprint.

The relevant point is this:
Ashkenazi Jews don’t cluster with Levantines on PCA. They cluster with Southern Italians and Cretans — two regions that also carry J2, J1, and E-M123 in abundance. So if those haplogroups automatically meant “Levantine ancestry,” you’d have to call half of Southern Italy and Crete Levantine too — which no one does.

But that’s not how population genetics works. Haplogroups ≠ autosomal ancestry.

Trying to tie Ashkenazi J2 to the Levant without acknowledging that PCA and autosomal models consistently place them in Southern Europe — and that those same haplogroups are common there too — is cherry-picking trivia, not analysis.

If you’re going to bring up J2, then bring the full context — or don’t bring it at all.

[Alenka]

You demanded a model that includes Slavic or Central European input. So I cited one:
Waldman et al. 2022 — a peer-reviewed qpAdm model with:

~70% Southern Italian

~15% Lebanese

~15% Russian
Nice try, but no. I didnt ask for a model that includes Slavic or Central European input.
I asked for a model that includes Slavic AND Central European input.


I'm currently building 3- and 4-way qpAdm models to include that layer properly, because I actually care about statistical fit — not dropping random populations in for narrative reasons like you seem to want.
Great, curious to see how that turns out.

Regarding the second half of your sentence - are you sure youre not doing exactly what you accuse me of doing?
Dropping random populations in for narrative reasons.

If youre really not doing that, then please explain why you modeled them with Sicilians.

You correctly stated that Jews lived in Southern Italy since the Roman Republic, with real communities, proselytism, and integration.
But can you share any sources stating this was especially prevalent in Sicily as opposed to other regions of Southern Italy?


Your entire reply is a bad-faith misreading. You’re quote-mining posts, ignoring everything that contradicts your script, and now pretending I claimed Jews “never touched Germans” because I referenced Waldman’s Slavic model.

You’re not here to discuss the data. You’re here to twist quotes, ignore studies, and act smug while offering nothing of substance. I’ve backed up every claim with modeling, PCA positioning, and published research. You’ve brought snark and zero sources.

This was your final spoon-fed answer. I won’t waste another reply on someone who doesn’t read, doesn’t listen, and contributes absolutely nothing.
No, no, no.

Hear me out, I actually agree with some of the points you made in your initial post.
For example, studies using North Italians or French as proxy for their Euro input doesnt even make any historical sense.

South Italian indeed seems way more logical.
But I think adding rhineland German and Slavic is also needed to capture the mozaic of their autosomal ancestry as realistically as possible.
Curious to see what comes out if you model them with both of the above.

Regarding the Slavic, not sure if Russian is the best choice, considering the fact that most Ashkenazi populations never moved so far east. Perhaps using Belarusian or North Ukrainian would be better. Or simply early Slavic perhaps.

And last but not least, sorry if my previous replies seemed too rude.
Didnt mean to offend you.

Lets put the forum to good use and have a respectful discussion. :o

[Gannicus]

Sure — and it’s just as common, if not more so, in Southern Italy, Crete, mainland Greece, and across the Caucasus. So what?

You can find J2 all over the Eastern Mediterranean and West Asia. It’s a broadly distributed macro-haplogroup, not a Levantine fingerprint.

The relevant point is this:
Ashkenazi Jews don’t cluster with Levantines on PCA. They cluster with Southern Italians and Cretans — two regions full of J2, J1, and E-M123. If J2 proves Levantine ancestry, then so are half the Greeks and most Sicilians.

But that’s not how population genetics works. Haplogroups ≠ autosomal ancestry.

Trying to tie Ashkenazi J2 to the Levant without acknowledging that PCA and autosomal models consistently place them in Southern Europe — and that those same haplogroups are common there too — is cherry-picking trivia, not analysis.

If you’re going to bring up J2, then bring the full context — or don’t bring it at all.

The point is, J2 should not be called strongly European. West Eurasian is more appropriate. That's the issue I had. And tell ChatGPT to be more polite next time and drop the adversarial tone.

[Lard]

Nice try, but no. I didnt ask for a model that includes Slavic or Central European input.
I asked for a model that includes Slavic AND Central European input.


You either didn’t read the literature I posted or you're pretending it doesn’t exist.
Here’s a single peer-reviewed study that covers **all three** European sources — Slavic (East-EU), Central/Western European (West-EU), and Southern European (South-EU):

Xue et al. 2017 – “The Time and Place of European Admixture in Ashkenazi Jewish History,” PLoS Genetics
https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1006644

The paper uses IBD segment sharing to quantify autosomal input from each region into Ashkenazi Jews:


They analyzed all three in one coherent framework — exactly what you claim hasn't been done.

https://i.ibb.co/HpDVRF9j/555.png

Panel A: Normalized by genome length (relative IBD sharing)
Panel B: Raw segment count (total IBD events)

This figure is from Xue et al. 2017, "The time and place of European admixture in Ashkenazi Jewish history" (PLOS Genetics (https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1006644)).

The charts show that Ashkenazi Jews share the greatest amount of IBD segments with:


East-EU (Slavic) — highest overall segment sharing
West-EU (Central Europe/Rhineland) — substantial sharing
Iberia — also significant
South-EU (Southern Italy, Crete) — lower than above
Levant, South-ME, Druze — minimal sharing

This empirically confirms that Slavic and Central European input is real and measurable in Ashkenazi Jews — and not just hypothetical.
It also undermines the idea that Levantine ancestry dominates. These patterns of IBD sharing align with later diaspora contacts (e.g. Eastern and Central Europe), not just ancient Near Eastern origins.



At this point, you're just moving the goalposts for trolling.
You keep shifting the demand — I met it.

You’ve ignored multiple citations, introduced zero of your own, and continue to play bad-faith games.

You're now persona non grata in this thread.
I won’t reply to you again. If others want serious discussion, I’m here.

[AndreiDNA]

this is a funny approach.
On PCA, jews are very different from Southern Italians. Only on a 2D PCA (which greatly oversimplifies the ancestral components) do jews plot over south italians. By the way, on a 2D PCA, guess where a half european half MENA person would plot? That's right - exactly over Ashkenazim and Sicilians.
Here I've made my own tiny PCA system to estimate jewish ancestry on the basis of an Fst matrix.

Egyptian.HO,0,20,4,16,8,12,11,21,9,16,18,5,5,21,19 ,13
Ukrainian.HO,20,0,24,10,9,5,11,0,8,3,15,15,16,4,3, 5
Tunisian.HO,4,24,0,22,14,16,15,25,14,20,21,11,12,2 5,22,17
Lezgin.HO,16,10,22,0,5,9,12,11,8,10,21,9,11,12,11, 11
Turkish.HO,8,9,14,5,0,4,6,10,3,7,13,4,5,11,9,6
Italian_North.HO,12,5,16,9,4,0,6,7,2,2,7,8,10,5,3, 1
Russian.HO,21,0,25,11,10,7,13,0,9,4,18,16,18,4,4,6
Italian_South.HO,9,8,14,8,3,2,5,9,0,5,8,5,7,9,6,3
French.HO,16,3,20,10,7,2,8,4,5,0,10,12,14,2,1,1
Sardinian.HO,18,15,21,21,13,7,12,18,8,10,0,15,17,1 5,12,7
Lebanese.HO,5,15,11,9,4,8,8,16,5,12,15,0,4,17,13,1 0
Syrian.HO,5,16,12,11,5,10,9,18,7,14,17,4,0,18,15,1 2
Norwegian.HO,21,4,25,12,11,5,12,4,9,2,15,17,18,0,2 ,4
English.HO,19,3,22,11,9,3,9,4,6,1,12,13,15,2,0,2
Spanish.HO,13,5,17,11,6,1,7,6,3,1,7,10,12,4,2,0

Jew_Ashkenazi.HO,11,11,15,12,6,6,0,13,5,8,12,8,9,1 2,9,7

Target: Jew_Ashkenazi.HO
Distance: 829.2003% / 8.29200338
43.2 Italian_South.HO
19.8 Syrian.HO
14.2 English.HO
10.0 Sardinian.HO
7.4 Tunisian.HO
5.4 Turkish.HO

Here we see that indeed the biggest contributor to Ashkenazi DNA is South Italian. At least in this model.
Let's compare something though.
Lard proclaims that a South Italian + Lebanese model is ideal for Ashkenazi modeling. I disagree. I think the ideal model is South Italian + Lebanese + Russian.
Thankfully, I included all these populations in my Fst matrix. So let's compare them, using fit to gauge plausibility.
Target: Jew_Ashkenazi.HO
Distance: 1125.7990% / 11.25798987 | R3P
63.2 Italian_South.HO
36.8 Lebanese.HO

vs
Target: Jew_Ashkenazi.HO
Distance: 983.6796% / 9.83679643 | R3P
42.2 Lebanese.HO
40.2 Italian_South.HO
17.6 Russian.HO

Clearly the model with Russian gives a better fit.
Notice how the portion of Lebanese drastically increased too!

Honestly, the whole thread seems politically motivated. Clearly Lard has a problem with Israel, and is trying to spin the age old tale that Jews are not indigenous to this land. Very thinly disguised motives.
I don't support Israel and condone the brutality with which they suppress Palestinians, but I will never concede that jews are indeed indigenous to it - Ashkenazim, Sephardim, Mizharim - these people have suffered and paid with blood for ownership of the land they come from. And they absolutely deserve to have a country of their own - they have been persecuted everywhere they've been, and Israel must remain a refuge for them in the event of a second holocaust. The Arabs have many countries. The Jews only have one.

[Lard]

The point is, J2 should not be called strongly European. West Eurasian is more appropriate. That's the issue I had. And tell ChatGPT to be more polite next time and drop the adversarial tone.

What you’re doing is classic goalpost-shifting and semantic hair-splitting.

Nobody claimed J2 is "native" to Europe in the deep prehistoric sense. The point — which you either missed or chose to ignore — is that **J2 is widespread in the exact Southern European populations that Ashkenazi Jews cluster with on PCA**, including Sicily, Calabria, Crete, and parts of Greece.

We're talking **15–25% frequencies** across Southern Italy and Crete. That’s not fringe — it’s a **major component** of local Y-DNA. If J2 makes Ashkenazim Levantine, then by your logic, so are a huge chunk of Sicilians, Calabrians, and Cretans. You won’t say that, of course — because you’re not actually following the data. You’re cherry-picking.

This is how bad arguments unravel:
– You ignore autosomal DNA, which places Ashkenazim squarely in Southern Europe
– You ignore the fact that their haplogroups are common in the same populations they cluster with
– And then you try to police terminology like a Reddit mod, as if changing “European” to “West Eurasian” matters when the entire point was about **modern regional distribution**, not ancient origins

And that jab about ChatGPT? Please. I don’t need AI to expose weak arguments — your posts do that on their own. If you dish out condescension, don’t whine when you get firm replies in return.

This discussion was about evidence. Try engaging with it next time, instead of tone-policing your way around getting fact-checked.

That’s my last reply to you. If you want to keep spamming my thread with passive-aggressive nonsense, feel free — maybe other posters will waste time on it. I won’t.

[Alenka]

You either didn’t read the literature I posted or you're pretending it doesn’t exist.
Here’s a single peer-reviewed study that covers **all three** European sources — Slavic (East-EU), Central/Western European (West-EU), and Southern European (South-EU):

Xue et al. 2017 – “The Time and Place of European Admixture in Ashkenazi Jewish History,” PLoS Genetics
🔗 https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1006644

The paper uses IBD segment sharing to quantify autosomal input from each region into Ashkenazi Jews:


They analyzed all three in one coherent framework — exactly what you claim hasn't been done.
Please stop putting words into my mouth.
When I said both Slavic and Central European has not been accounted for yet, I was referring to your models.
I never claimed it generally hasn't been done or analyzed by anyone.



https://i.ibb.co/HpDVRF9j/555.png

Panel A: Normalized by genome length (relative IBD sharing)
Panel B: Raw segment count (total IBD events)

This figure is from Xue et al. 2017, "The time and place of European admixture in Ashkenazi Jewish history" (PLOS Genetics (https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1006644)).

The charts show that Ashkenazi Jews share the greatest amount of IBD segments with:


East-EU (Slavic) — highest overall segment sharing
West-EU (Central Europe/Rhineland) — substantial sharing
Iberia — also significant
South-EU (Southern Italy, Crete) — lower than above
Levant, South-ME, Druze — minimal sharing

This empirically confirms that Slavic and Central European input is real and measurable in Ashkenazi Jews — and not just hypothetical.
It also undermines the idea that Levantine ancestry dominates. These patterns of IBD sharing align with later diaspora contacts (e.g. Eastern and Central Europe), not just ancient Near Eastern origins.

At this point, you're just moving the goalposts for trolling.
You keep shifting the demand — I met it.

You’ve ignored multiple citations, introduced zero of your own, and continue to play bad-faith games.

You're now persona non grata in this thread.
I won’t reply to you again. If others want serious discussion, I’m here.
Do you realize the findings regarding shared IBD segments you just posted are proving your models from your initial post wrong?

Accoding to the above, Ashkenazi Jews share more IBD segments with East-EU (Slavic), West-EU (Central Europe/Rhineland), and even Iberia, moreso than South-EU (Southern Italy, Crete) and Levant, South-ME, Druze.

That goes to show your models from the initial post are faulty and absolutely unrealistic.

[Gannicus]

Sure — and it’s just as common, if not more so, in Southern Italy, Crete, mainland Greece, and across the Caucasus. So what?

You can find J2 all over the Eastern Mediterranean and West Asia. It’s a broadly distributed macro-haplogroup, not a Levantine fingerprint.

The relevant point is this:
Ashkenazi Jews don’t cluster with Levantines on PCA. They cluster with Southern Italians and Cretans — two regions that also carry J2, J1, and E-M123 in abundance. So if those haplogroups automatically meant “Levantine ancestry,” you’d have to call half of Southern Italy and Crete Levantine too — which no one does.

But that’s not how population genetics works. Haplogroups ≠ autosomal ancestry.

Trying to tie Ashkenazi J2 to the Levant without acknowledging that PCA and autosomal models consistently place them in Southern Europe — and that those same haplogroups are common there too — is cherry-picking trivia, not analysis.

If you’re going to bring up J2, then bring the full context — or don’t bring it at all.

Hit a nerve didn't I?

I have seen you post on here and I have grown sick of it. For once just type some of your own forum responses. I used it to argue with Afrocentrists in the Youtube comments section out of boredom and out of fatigue with their crap. So, I know what it looks like.

I can use ChatGPT too!

✅ Yes, J2 is broadly distributed — but distribution ≠ origin
It’s true that J2 is found all over the Eastern Mediterranean, Anatolia, the Caucasus, and Iran, so no single modern region owns it exclusively.

But that doesn’t mean it can’t function as a “Levantine fingerprint” in the right context.

�� J2 can absolutely indicate Levantine ancestry when:
It appears in historically Levantine populations (e.g., ancient Canaanites, Israelites, Phoenicians).

It shows up in diaspora groups with documented Levantine origin (e.g., Jews, Syriacs, Phoenician colonies).

The subclade is specific to Levantine populations, such as:

J2a-M67 (especially some downstream clades) in Levantine Jews, Druze, and Syrians

J2b-M205, occasionally linked to Levantine ancestry in some studies

�� Examples:
Ancient DNA from Sidon (Bronze Age Lebanon) showed J2a among Canaanites — a clear Levantine context.

Ashkenazi and Sephardic Jews carry J2a clades that match lineages seen in ancient and modern Levantines.

Phoenician colonization brought J2 to places like Carthage and parts of Southern Europe — again, a Levantine fingerprint spreading outward.



Here:

You keep repeating that “haplogroups ≠ autosomal ancestry” like you’re delivering a revelation. Trust me — I know. But haplogroups do reflect deep male-line migrations, and mine is a perfect example. I descend from J2a1a1a2b2 (J-FTC5373) — a lineage that runs through J-M67, which is frequent among Levantines, ancient Canaanites, Jews, and Eastern Mediterranean populations. It has also turned up in ancient samples from the Bronze Age Levant, Imperial Rome, and Phoenician-linked sites.

You dismissed J2 as a "non-Levantine fingerprint" while ignoring the fact that certain subclades are Levantine indicators, especially when paired with autosomal and historical context. Saying otherwise is like pretending R1b doesn’t mean anything in the context of Indo-European migrations — it’s disingenuous.

As for your PCA obsession: PCA shows proximity, not origin. Ashkenazim cluster near Southern Italians partly because Southern Italians have substantial ancient Near Eastern ancestry, especially from Anatolia and the Levant — something you conveniently downplay. Modeling Ashkenazim as ~50% Levantine isn’t “inflated” — it reflects real ancestry components that PCA alone can’t resolve. qpAdm, f-statistics, and ancient DNA show that clearly.

Finally, the tone and downvote were unnecessary. If you're here to debate, great — but don’t pretend you're correcting a novice when I'm bringing actual lineage data, models, and context. You don’t have to like what I said, but dismissing well-supported points as “cherry-picking” is pure projection. I brought the full picture — you just didn’t like the view.

[Gannicus]

What you’re doing is classic goalpost-shifting and semantic hair-splitting.

Nobody claimed J2 is "native" to Europe in the deep prehistoric sense. The point — which you either missed or chose to ignore — is that **J2 is widespread in the exact Southern European populations that Ashkenazi Jews cluster with on PCA**, including Sicily, Calabria, Crete, and parts of Greece.

We're talking **15–25% frequencies** across Southern Italy and Crete. That’s not fringe — it’s a **major component** of local Y-DNA. If J2 makes Ashkenazim Levantine, then by your logic, so are a huge chunk of Sicilians, Calabrians, and Cretans. You won’t say that, of course — because you’re not actually following the data. You’re cherry-picking.

This is how bad arguments unravel:
– You ignore autosomal DNA, which places Ashkenazim squarely in Southern Europe
– You ignore the fact that their haplogroups are common in the same populations they cluster with
– And then you try to police terminology like a Reddit mod, as if changing “European” to “West Eurasian” matters when the entire point was about **modern regional distribution**, not ancient origins

And that jab about ChatGPT? Please. I don’t need AI to expose weak arguments — your posts do that on their own. If you dish out condescension, don’t whine when you get firm replies in return.

This discussion was about evidence. Try engaging with it next time, instead of tone-policing your way around getting fact-checked.

That’s my last reply to you. If you want to keep spamming my thread with passive-aggressive nonsense, feel free — maybe other posters will waste time on it. I won’t.

Other posters have expressed their fatigue with your AI responses.

If you want, we can keep this going. The dueling of the AI.

ChatGPT:

You just spent 300 words accusing me of everything from cherry-picking to tone-policing, while ironically evading the actual point: J2 subclades like J-M67 do appear in ancient Levantine populations and are meaningful in context — especially when we’re discussing deep ancestry and migrations.

You keep saying “nobody claimed J2 was native to Europe,” but the original post literally called it “strongly European”. That’s what I responded to. Not “modern distribution,” but the implication of origin. You're now shifting your own goalposts by pretending this was only ever about PCA and modern frequency. That’s rhetorical sleight-of-hand, not “fact-checking.”

I never denied that J2 is common in Southern Europe. It is — because Southern Europe received multiple waves of gene flow from the Near East over thousands of years, starting with the Neolithic. You say Ashkenazim clustering with Sicilians proves they're Southern European — but Sicilians themselves have significant Near Eastern ancestry, which is exactly why they cluster together. You’re proving my point while thinking you’re dismantling it.

As for autosomal models: I’ve referenced qpAdm, f-statistics, ancient DNA, and Levantine Bronze Age sources — so pretending I “ignore” autosomal data is pure projection. What I don’t do is rely solely on PCA and pretend it tells the whole story. That’s your move.

You can dismiss the ChatGPT jab all you like, but if your posts weren’t so indistinguishable from automated synthesis, the thought wouldn’t have crossed my mind. The tone, the structure, the shifting — it’s textbook. If that bothers you more than the fact I challenged your oversimplified narrative, that’s on you.

Anyway, your “last reply” speech is noted — a bit dramatic, but I’ll let you have it. I’m here to engage with data, not with ego flares. Others can decide who actually brought sources, logic, and context to the table.

[Shrubby]

Please stop putting words into my mouth.
When I said both Slavic and Central European has not been accounted for yet, I was referring to your models.
I never claimed it generally hasn't been done or analyzed by anyone.


Do you realize the findings regarding shared IBD segments you just posted are proving your models from your initial post wrong?

Accoding to the above, Ashkenazi Jews share more IBD segments with East-EU (Slavic), West-EU (Central Europe/Rhineland), and even Iberia, moreso than South-EU (Southern Italy, Crete) and Levant, South-ME, Druze.

That goes to show your models from the initial post are faulty and absolutely unrealistic.

More shared IBD segments indicate more recent contact or interaction, not level of admixture. By the time the ancestors of Ashkenazim settled in Central and Eastern Europe they were isolated from East Med populations. The higher sharing with Iberian populations could potentially be explained by small numbers of assimilated Sephardim possessing minor Iberian admixture.

[Alenka]

this is a funny approach.
On PCA, jews are very different from Southern Italians. Only on a 2D PCA (which greatly oversimplifies the ancestral components) do jews plot over south italians. By the way, on a 2D PCA, guess where a half european half MENA person would plot? That's right - exactly over Ashkenazim and Sicilians.
Here I've made my own tiny PCA system to estimate jewish ancestry on the basis of an Fst matrix.

Egyptian.HO,0,20,4,16,8,12,11,21,9,16,18,5,5,21,19 ,13
Ukrainian.HO,20,0,24,10,9,5,11,0,8,3,15,15,16,4,3, 5
Tunisian.HO,4,24,0,22,14,16,15,25,14,20,21,11,12,2 5,22,17
Lezgin.HO,16,10,22,0,5,9,12,11,8,10,21,9,11,12,11, 11
Turkish.HO,8,9,14,5,0,4,6,10,3,7,13,4,5,11,9,6
Italian_North.HO,12,5,16,9,4,0,6,7,2,2,7,8,10,5,3, 1
Russian.HO,21,0,25,11,10,7,13,0,9,4,18,16,18,4,4,6
Italian_South.HO,9,8,14,8,3,2,5,9,0,5,8,5,7,9,6,3
French.HO,16,3,20,10,7,2,8,4,5,0,10,12,14,2,1,1
Sardinian.HO,18,15,21,21,13,7,12,18,8,10,0,15,17,1 5,12,7
Lebanese.HO,5,15,11,9,4,8,8,16,5,12,15,0,4,17,13,1 0
Syrian.HO,5,16,12,11,5,10,9,18,7,14,17,4,0,18,15,1 2
Norwegian.HO,21,4,25,12,11,5,12,4,9,2,15,17,18,0,2 ,4
English.HO,19,3,22,11,9,3,9,4,6,1,12,13,15,2,0,2
Spanish.HO,13,5,17,11,6,1,7,6,3,1,7,10,12,4,2,0

Jew_Ashkenazi.HO,11,11,15,12,6,6,0,13,5,8,12,8,9,1 2,9,7

Target: Jew_Ashkenazi.HO
Distance: 829.2003% / 8.29200338
43.2 Italian_South.HO
19.8 Syrian.HO
14.2 English.HO
10.0 Sardinian.HO
7.4 Tunisian.HO
5.4 Turkish.HO

Here we see that indeed the biggest contributor to Ashkenazi DNA is South Italian. At least in this model.
Let's compare something though.
Lard proclaims that a South Italian + Lebanese model is ideal for Ashkenazi modeling. I disagree. I think the ideal model is South Italian + Lebanese + Russian.
Thankfully, I included all these populations in my Fst matrix. So let's compare them, using fit to gauge plausibility.
Target: Jew_Ashkenazi.HO
Distance: 1125.7990% / 11.25798987 | R3P
63.2 Italian_South.HO
36.8 Lebanese.HO

vs
Target: Jew_Ashkenazi.HO
Distance: 983.6796% / 9.83679643 | R3P
42.2 Lebanese.HO
40.2 Italian_South.HO
17.6 Russian.HO

Clearly the model with Russian gives a better fit.
Notice how the portion of Lebanese drastically increased too!
I would be curious to see if you add rhineland German also besides Russian.
Wondering how much it would allocate to each.


Honestly, the whole thread seems politically motivated. Clearly Lard has a problem with Israel, and is trying to spin the age old tale that Jews are not indigenous to this land. Very thinly disguised motives.
I don't support Israel and condone the brutality with which they suppress Palestinians, but I will never concede that jews are indeed indigenous to it - Ashkenazim, Sephardim, Mizharim - these people have suffered and paid with blood for ownership of the land they come from. And they absolutely deserve to have a country of their own - they have been persecuted everywhere they've been, and Israel must remain a refuge for them in the event of a second holocaust. The Arabs have many countries. The Jews only have one.
Here I disagree a bit. I think Uncle Sam supports them way too much.
They are ready to drag the whole world into war for their own selfish interest.

If they are unable to keep the land by themselves, then they dont deserve it.
Israel is the one who shouldnt have nukes, they are way more impulsive than Iran.

And its not true that Jews have only one country.
Stalin gifted them the Jewish Autonomous Oblast (https://en.wikipedia.org/wiki/Jewish_Autonomous_Oblast) before Israel even existed.

I realize JAO is not very conveniently placed, and they have no attachment to that area, but oh well.
Beggars cant be choosers. And at least they would be safer, far away from populations they would eternally be feuding with.

[Alenka]

More shared IBD segments indicate more recent contact or interaction, not level of admixture. By the time the ancestors of Ashkenazim settled in Central and Eastern Europe they were isolated from East Med populations. The higher sharing with Iberian populations could potentially be explained by small numbers of assimilated Sephardim possessing minor Iberian admixture.
Thanks for your input, I am far from an expert.

But still it shows that Central and Eastern Europe input is real, thus modeling them with just Southern Europe and Levant is incomplete, isnt it?

[Gannicus]

One thing I have noticed with PCA is that if one has a real but small piece of admixture from elsewhere, it won't shift one out of the cluster. In my case I'm still trying to figure out what this Near East is.

141897

Despite this result I'm within the British Isles, French, German, and Dutch clusters

141899

141898

[Lard]

this is a funny approach.
On PCA, jews are very different from Southern Italians. Only on a 2D PCA (which greatly oversimplifies the ancestral components) do jews plot over south italians. By the way, on a 2D PCA, guess where a half european half MENA person would plot? That's right - exactly over Ashkenazim and Sicilians.
Here I've made my own tiny PCA system to estimate jewish ancestry on the basis of an Fst matrix.


Egyptian.HO,0,20,4,16,8,12,11,21,9,16,18,5,5,21,19 ,13
Ukrainian.HO,20,0,24,10,9,5,11,0,8,3,15,15,16,4,3, 5
Tunisian.HO,4,24,0,22,14,16,15,25,14,20,21,11,12,2 5,22,17
Lezgin.HO,16,10,22,0,5,9,12,11,8,10,21,9,11,12,11, 11
Turkish.HO,8,9,14,5,0,4,6,10,3,7,13,4,5,11,9,6
Italian_North.HO,12,5,16,9,4,0,6,7,2,2,7,8,10,5,3, 1
Russian.HO,21,0,25,11,10,7,13,0,9,4,18,16,18,4,4,6
Italian_South.HO,9,8,14,8,3,2,5,9,0,5,8,5,7,9,6,3
French.HO,16,3,20,10,7,2,8,4,5,0,10,12,14,2,1,1
Sardinian.HO,18,15,21,21,13,7,12,18,8,10,0,15,17,1 5,12,7
Lebanese.HO,5,15,11,9,4,8,8,16,5,12,15,0,4,17,13,1 0
Syrian.HO,5,16,12,11,5,10,9,18,7,14,17,4,0,18,15,1 2
Norwegian.HO,21,4,25,12,11,5,12,4,9,2,15,17,18,0,2 ,4
English.HO,19,3,22,11,9,3,9,4,6,1,12,13,15,2,0,2
Spanish.HO,13,5,17,11,6,1,7,6,3,1,7,10,12,4,2,0

Jew_Ashkenazi.HO,11,11,15,12,6,6,0,13,5,8,12,8,9,1 2,9,7

Target: Jew_Ashkenazi.HO
Distance: 829.2003% / 8.29200338
43.2 Italian_South.HO
19.8 Syrian.HO
14.2 English.HO
10.0 Sardinian.HO
7.4 Tunisian.HO
5.4 Turkish.HO

Here we see that indeed the biggest contributor to Ashkenazi DNA is South Italian. At least in this model.
Let's compare something though.
Lard proclaims that a South Italian + Lebanese model is ideal for Ashkenazi modeling. I disagree. I think the ideal model is South Italian + Lebanese + Russian.
Thankfully, I included all these populations in my Fst matrix. So let's compare them, using fit to gauge plausibility.
Target: Jew_Ashkenazi.HO
Distance: 1125.7990% / 11.25798987 | R3P
63.2 Italian_South.HO
36.8 Lebanese.HO

vs
Target: Jew_Ashkenazi.HO
Distance: 983.6796% / 9.83679643 | R3P
42.2 Lebanese.HO
40.2 Italian_South.HO
17.6 Russian.HO

Clearly the model with Russian gives a better fit.
Notice how the portion of Lebanese drastically increased too!

Honestly, the whole thread seems politically motivated. Clearly Lard has a problem with Israel, and is trying to spin the age old tale that Jews are not indigenous to this land. Very thinly disguised motives.
I don't support Israel and condone the brutality with which they suppress Palestinians, but I will never concede that jews are indeed indigenous to it - Ashkenazim, Sephardim, Mizharim - these people have suffered and paid with blood for ownership of the land they come from. And they absolutely deserve to have a country of their own - they have been persecuted everywhere they've been, and Israel must remain a refuge for them in the event of a second holocaust. The Arabs have many countries. The Jews only have one.



Let’s break this down. You literally just confirmed everything I said — and everything peer-reviewed research already proved.

In your own YouTube video The Ethnogenesis of Ashkenazi Jews (G25),

https://www.youtube.com/watch?v=l4-hpbMKLgg
you posted this model:
https://i.ibb.co/7x6CbwHg/vlcsnap-2025-07-20-19h29m55s115.png

G25 model shows:


63.2% Italian_Calabria
25.2% Lebanese_Druze
10.8% Belarusian
0.8% Kumyk


So… 2/3rds Southern Italian, 1/4 Levantine, and ~10% Slavic — exactly what I said this whole time.
That was my 2-way model on this thread: ~70–75% Southern Italian, ~25% Lebanese.
I told viewers I’d later build a 3-way qpAdm to flesh out the Slavic signal at 10–15%.

You tried to mock that — and accidentally confirmed all of it.

Even more ironic? Your numbers are nearly identical to the peer-reviewed qpAdm study from Waldman et al. 2022, published in Cell:
https://pubmed.ncbi.nlm.nih.gov/36455558/


Ancient Ashkenazi Jews modeled as:
▸ 70% Southern Italian
▸ 15% Lebanese
▸ 15% Russian

So what does that make you? A “self-hating Zionist” for replicating Waldman’s data?

You accused me of being anti-Israel — while you posted the same model in your own video. Hypocrisy level: critical.

---

Let’s talk about your methods, too:

✔ FST tables ≠ PCA ≠ qpAdm
✔ You presented a homemade FST matrix and called it PCA
✔ Then claimed it’s “scientific” while you literally charge people to run free tools
✔ qpAdm is public — illustratedDNA.com lets anyone run it themselves

If any forum user wants legit qpAdm modeling done, just PM me if we’re friends on here.
I’ll do it for free. No scam. No YouTube grifting.

---

Finally, this is what real science says:

• Drineas et al. 2019 (Ann Hum Genet)
https://pubmed.ncbi.nlm.nih.gov/31162867/


“.In the PCA of Crete vs Europe, the Cretans overlap with three populations: the Peloponneseans, the Sicilians and the Ashkenazi Jews (see Figures 4a, S17, and S18). Southern European and Mediterranean ancestry of the Ashkenazi Jews has also been demonstrated before (Atzmon et al., 2010; Behar et al., 2010; Bauchet et al., 2007; Price et al., 2008; Seldin et al., 2006; Tian et al., 2008). Furthermore, we find in both PCA and ADMIXTURE analysis, that the Ashkenazi are more similar to the Cretans than to the two Levantine Semitic populations.”

• Waldman et al. 2022 (Cell)
https://pubmed.ncbi.nlm.nih.gov/36455558/

70% Southern Italian
15% Lebanese
15% Russian

So you can call it “funny” all you want — but you just validated everything I posted, and then some.

I’d call that checkmate.

[AndreiDNA]

Let’s break this down. You literally just confirmed everything I said — and everything peer-reviewed research already proved.

In your own YouTube video The Ethnogenesis of Ashkenazi Jews (G25),

https://www.youtube.com/watch?v=l4-hpbMKLgg
you posted this model:
https://i.ibb.co/7x6CbwHg/vlcsnap-2025-07-20-19h29m55s115.png

G25 model shows:


63.2% Italian_Calabria
25.2% Lebanese_Druze
10.8% Belarusian
0.8% Kumyk


So… 2/3rds Southern Italian, 1/4 Levantine, and ~10% Slavic — exactly what I said this whole time.
That was my 2-way model on this thread: ~70–75% Southern Italian, ~25% Lebanese.
I told viewers I’d later build a 3-way qpAdm to flesh out the Slavic signal at 10–15%.

You tried to mock that — and accidentally confirmed all of it.

Even more ironic? Your numbers are nearly identical to the peer-reviewed qpAdm study from Waldman et al. 2022, published in Cell:
https://pubmed.ncbi.nlm.nih.gov/36455558/



So what does that make you? A “self-hating Zionist” for replicating Waldman’s data?

You accused me of being anti-Israel — while you posted the same model in your own video. Hypocrisy level: critical.

---

Let’s talk about your methods, too:

✔ FST tables ≠ PCA ≠ qpAdm
✔ You presented a homemade FST matrix and called it PCA
✔ Then claimed it’s “scientific” while you literally charge people to run free tools
✔ qpAdm is public — illustratedDNA.com lets anyone run it themselves

If any forum user wants legit qpAdm modeling done, just PM me if we’re friends on here.
I’ll do it for free. No scam. No YouTube grifting.

---

Finally, this is what real science says:

• Drineas et al. 2019 (Ann Hum Genet)
https://pubmed.ncbi.nlm.nih.gov/31162867/



• Waldman et al. 2022 (Cell)
https://pubmed.ncbi.nlm.nih.gov/36455558/


So you can call it “funny” all you want — but you just validated everything I posted, and then some.

I’d call that checkmate.

You have used qpadm for 2 days. I have used it for over half a year. What I charge people for, is services you can't get on illustrative. If someone wants to model themselves with a sample that isn't on illustrative's aadr + HO dataset, you wouldn't be able to do that for them - and I would.

Man you need to remove the fingers from your own butt. It seems you don't know what PCA is - as an analysis of a matrix of Fst values constitutes a PCA (Principal component analysis) because by definition it is an analysis of principal components. Each component in that matrix represents an affinity to one of the multiple anchor populations. It is no different in principle than G25 or T25 or any other component system.

" Hypocrisy level: critical."
Lol what a load of crap. That video is 3 years old.

Go ahead and call me a grifter all you want, you only show people that you're acting like a pissed off child. I have a solid reputation for doing what I do and your screeching won't damage it one bit.

[Gannicus]

🎯 You’re not just analyzing haplogroups or PCA — you’re clearly trying to reframe Ashkenazi Jews as “mostly European” by downplaying their Near Eastern ancestry.

🧬 J2 is Near Eastern in origin, not European
✅ Originated in the Fertile Crescent, not Europe
✅ Spread into Southern Europe via Neolithic and Bronze Age migrations
✅ Found in ancient samples from:

📍Levant (e.g., Megiddo, Ashkelon – Bronze Age)

📍Minoan Crete (Early Bronze Age)

📍Anatolia (Chalcolithic & Bronze Age)

📍Imperial Rome (eastern migrants)

🛑 Modern frequency ≠ point of origin — Southern Italians and Greeks carry J2 today because of past admixture from the Near East.

📌 PCA overlap ≠ same ancestry
You argue:

“Ashkenazi Jews cluster with Southern Italians on PCA, so they’re basically the same.”

But that’s a shallow take on PCA:

✅ PCA compresses thousands of years of genetic drift and admixture into 2–3 visual dimensions
✅ Populations with different ancestries can cluster together if they share certain mixture profiles
✅ PCA shows similarity, not identity — and certainly not origin

Ashkenazi Jews cluster between Southern Europeans and Levantines because they are admixed — not because they’re just Italians with a twist.

You can have two populations sit close on PCA and still have very different uniparental lineages, different ancient source populations, and different historical trajectories.

🔍 Let’s be honest — your standards shift when it comes to Jews
If any other population had:

~25–40% Near Eastern autosomal DNA, and

Near Eastern Y-DNA like J1, J2, E-M123,

You’d call them Near Eastern-admixed without hesitation.

But when it’s Ashkenazi Jews, you suddenly want to redefine J2 as “European” and reduce Levantine input to a footnote.

That’s not population genetics — that’s spin.

🧠 Bottom Line:
✅ J2 is Near Eastern, not European in origin
✅ PCA overlap with Southern Italians reflects shared admixture, not common origin
✅ Ashkenazi Jews are West Eurasian, but their ancestry is biogeographically intermediate, not “just another European population”

📣 Trying to “Europeanize” Ashkenazi Jews by redefining Near Eastern markers and cherry-picking PCA context isn’t scientific — it’s agenda-driven.
And it’s transparent.


ChatGPT off
Kinda looks like a Lard response doesn't it?

[Gannicus]

You have used qpadm for 2 days. I have used it for over half a year. What I charge people for, is services you can't get on illustrative. If someone wants to model themselves with a sample that isn't on illustrative's aadr + HO dataset, you wouldn't be able to do that for them - and I would.

Man you need to remove the fingers from your own butt. It seems you don't know what PCA is - as an analysis of a matrix of Fst values constitutes a PCA (Principal component analysis) because by definition it is an analysis of principal components. Each component in that matrix represents an affinity to one of the multiple anchor populations. It is no different in principle than G25 or T25 or any other component system.

" Hypocrisy level: critical."
Lol what a load of crap. That video is 3 years old.

Go ahead and call me a grifter all you want, you only show people that you're acting like a pissed off child. I have a solid reputation for doing what I do and your screeching won't damage it one bit.

That's the problem with the internet. Content is up basically forever. You'd have to constantly have to refer to older work and clarify that you have made new conclusions based on new data.

[Alenka]

Let’s Examine These So-Called “Near Eastern” Haplogroups in Ashkenazim — and See If They Align with Their PCA Plotting in Southern Italy

Ashkenazi Jewish Y-DNA haplogroups like J1, J2, and E-M123/M34 are often cited as proof of “Near Eastern” ancestry. But this framing fails basic population genetics. Haplogroups alone don’t determine ancestry — autosomal DNA does. And when we examine where these haplogroups are found and how they behave in other populations, the “Near Eastern” label begins to unravel.

1. Southern Italy Is a Hotspot for J1 — But Nobody Calls South Italians Levantine

Multiple studies — including Boattini et al. (2013) and Di Giacomo et al. (2002) — show substantial J1 frequencies across Southern Italy:


Agrigento, Sicily – 12%
Pescara, Abruzzo – 15%
North Gargano, Apulia – 17.2%
Reggio Calabria – 9.1%
Paola, Calabria – 11.1%
Benevento, Campania – 8.5%
Pistoia, Tuscany – 7.5%
Bologna, Emilia-Romagna – 7%



Where did you get these figures?
I see you named a couple studies, and I tried to look this up.
So please, for the sake of transparency, provide us a link to your source.

You named the study Boattini et al. (2013) as one of your sources and I have trouble finding those figures.
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0065441
Is this your source or is it some other link? Because I skimmed through and I found no such content.

The other one you named is Di Giacomo et al. (2002). Wasnt able to even find this one.
Maybe figures are there, so please share the links to both of your sources.

I ask because charts Ive seen thus far have Southern Italy at quite lower rates of J1 than your figures.

https://www.eupedia.com/europe/european_y-dna_haplogroups.shtml
At Eupedia, Southern Italy is listed at 4% and Sicily 3.5%, for example.
Much less than the Ashkenazi 19%.

If your figures are not just some BS that ChatGPT spewed to tell you what you want to hear, please share a link to your source.

[AndreiDNA]

That's the problem with the internet. Content is up basically forever. You'd have to constantly have to refer to older work and clarify that you have made new conclusions based on new data.

That + I don't remove old videos which I now disagree with out of principle.
I view and present my channel as a learning experience where I learn stuff and share it with my audience.
There are a lot of videos I've made which I'd probably delete if I wanted to present everything on my channel as some objective truth. But I leave them so people can see what I thought then and compare that to what I think now. So they'd go through the learning experience with me.

[Lard]

You have used qpadm for 2 days. I have used it for over half a year. What I charge people for, is services you can't get on illustrative. If someone wants to model themselves with a sample that isn't on illustrative's aadr + HO dataset, you wouldn't be able to do that for them - and I would.

Man you need to remove the fingers from your own butt. It seems you don't know what PCA is - as an analysis of a matrix of Fst values constitutes a PCA (Principal component analysis) because by definition it is an analysis of principal components. Each component in that matrix represents an affinity to one of the multiple anchor populations. It is no different in principle than G25 or T25 or any other component system.

" Hypocrisy level: critical."
Lol what a load of crap. That video is 3 years old.

Go ahead and call me a grifter all you want, you only show people that you're acting like a pissed off child. I have a solid reputation for doing what I do and your screeching won't damage it one bit.

You’re still missing the point — or deliberately twisting it.

My 2-way model was always:
• ~⅔ Southern Italian
• ~¼–⅓ Levantine (Lebanese)

Your own video shows:
• 63.2% Italian_Calabria
• 25.2% Lebanese_Druze
• 10.8% Belarusian

Which is effectively the same breakdown I posted — just with the Slavic signal added in.
So while you were busy trying to discredit me, you ended up confirming my model down to the decimal.
That’s not a rebuttal — it’s an endorsement.


My 2-way model matches peer-reviewed Waldman et al. 2022 (Cell) — minus the Slavic component,
which I explicitly said I’d incorporate in a later 3-way model.
YOU ended up adding that Slavic element yourself, reinforcing Waldman’s findings even more directly.


About your methods:
You’re confusing the output of an Fst matrix with the methodology of PCA.
PCA reduces covariance in multivariate data using eigen decomposition — you didn’t do that.
You made a visual affinity matrix. That’s not PCA.
That’s not opinion, it’s math.

Calling it “my PCA system” misleads people into thinking you performed an actual Principal Component Analysis,
when in reality you just manually plotted distances.
That’s fine for visualization — just don’t conflate it with formal PCA or qpAdm.

As for your “half a year” with qpAdm — congratulations.
That’s still less than the decade+ of cumulative work from academic teams publishing in Cell, Nature, and PNAS.
I cite them. You dismiss them. Big difference.

If you're charging for custom modeling with non-public samples, fine — just say that.
But don’t dress up a basic Fst matrix as PCA, and don’t pretend it’s interchangeable with formal stats like qpAdm or f3/f4.
It’s not. You know that. Every geneticist knows that.

Also, charging people to run free tools is your choice, but let’s not pretend it’s out of reach for others.
You just admitted you use the same underlying methods — AADR, HO, Fst, ADMIXTURE, and sometimes G25.
IllustratedDNA and other public interfaces let users run these tools themselves without needing to “pay a middleman.”
Anyone with sample IDs and the patience to learn can do it — and many are.

The difference is: I don’t gatekeep.

Bottom line:
You can deflect with tone and brag about your “reputation,” but that won’t change the fact that:

1. Your model replicated mine
2. My model matches peer-reviewed Waldman et al. 2022 (Cell) — you just filled in the Slavic part I already planned
3. The dominant ancestry in Ashkenazim is Southern Italian, not Levantine

You helped prove that.
So again — thank you.

[Lard]

That + I don't remove old videos which I now disagree with out of principle.
I view and present my channel as a learning experience where I learn stuff and share it with my audience.
There are a lot of videos I've made which I'd probably delete if I wanted to present everything on my channel as some objective truth. But I leave them so people can see what I thought then and compare that to what I think now. So they'd go through the learning experience with me.

Okay, fair enough — you say your channel is a learning journey.
So here’s the obvious follow-up:

What exactly has changed?
You now reject your own model that showed Ashkenazim as ~63% Southern Italian, ~25% Levantine, ~10% Slavic?

If so — tell us why using Southern Italian modeling is supposedly “wrong” for Ashkenazim,
when they literally plot there on PCA, have their deepest autosomal pull in that direction,
and formed as a population in the Roman world, not the medieval Levant.

Let me guess, champ — you now avoid Southern Italy because it “soaks up Levantine signal,” right?
You think that hides Levantine ancestry in Ashkenazim?

That’s backwards. Completely.

Using Southern Italians is accurate because they reflect actual historical admixture partners —
Romans, Greeks, Phoenicians, and yes, even Jews — that helped form the Ashkenazi population in Late Antiquity.

Using Tuscans or Northern Italians instead is both:
1. Historically inaccurate — Ashkenazim didn’t form in Tuscany or the Po Valley.
2. Genetically misleading — those populations have much less Near Eastern ancestry than Southern Italians,
so they artificially inflate the Levantine signal in models by contrast.

That’s not “scientific rigor,” that’s stacking the deck.

So again:
What exactly do you now believe, and why is your own past model — which matched Waldman et al. 2022 almost exactly — suddenly invalid?

Because it sure looks like you’re walking backward from data that didn’t fit your new narrative.

[Gannicus]

Okay, fair enough — you say your channel is a learning journey.
So here’s the obvious follow-up:

What exactly has changed?
You now reject your own model that showed Ashkenazim as ~63% Southern Italian, ~25% Levantine, ~10% Slavic?

If so — tell us why using Southern Italian modeling is supposedly “wrong” for Ashkenazim,
when they literally plot there on PCA, have their deepest autosomal pull in that direction,
and formed as a population in the Roman world, not the medieval Levant.

Let me guess, champ — you now avoid Southern Italy because it “soaks up Levantine signal,” right?
You think that hides Levantine ancestry in Ashkenazim?

That’s backwards. Completely.

Using Southern Italians is accurate because they reflect actual historical admixture partners —
Romans, Greeks, Phoenicians, and yes, even Jews — that helped form the Ashkenazi population in Late Antiquity.

Using Tuscans or Northern Italians instead is both:
1. Historically inaccurate — Ashkenazim didn’t form in Tuscany or the Po Valley.
2. Genetically misleading — those populations have much less Near Eastern ancestry than Southern Italians,
so they artificially inflate the Levantine signal in models by contrast.

That’s not “scientific rigor,” that’s stacking the deck.

So again:
What exactly do you now believe, and why is your own past model — which matched Waldman et al. 2022 almost exactly — suddenly invalid?

Because it sure looks like you’re walking backward from data that didn’t fit your new narrative.

Do you have a problem with Ashkenazi Jews being nearly half Near Eastern?

I don’t. There’s nothing wrong with that.

People forget that Middle Easterners and Europeans are part of the same broader West Eurasian family. There’s a long history of overlap, admixture, and shared ancestry — especially around the Mediterranean.

Even the so-called “European core” like the Yamnaya weren’t purely local — they carried significant CHG and Iran_N ancestry, which are themselves tied to the broader Near East. So if we’re being honest, most Europeans today already carry ancient West Asian ancestry to some degree.

If anything, acknowledging the Near Eastern side adds depth to the story of Ashkenazi origins — it doesn’t take away from it.

Trying to erase or minimize that just to make them “fit” a cleaner European label isn’t scientific — it’s narrative-driven. And frankly, it misses the point of how complex population history really is.

[Lard]

Do you have a problem with Ashkenazi Jews being nearly half Near Eastern?

I don’t. There’s nothing wrong with that.

People forget that Middle Easterners and Europeans are part of the same broader West Eurasian family. There’s a long history of overlap, admixture, and shared ancestry — especially around the Mediterranean.

Even the so-called “European core” like the Yamnaya weren’t purely local — they carried significant CHG and Iran_N ancestry, which are themselves tied to the broader Near East. So if we’re being honest, most Europeans today already carry ancient West Asian ancestry to some degree.

If anything, acknowledging the Near Eastern side adds depth to the story of Ashkenazi origins — it doesn’t take away from it.

Trying to erase or minimize that just to make them “fit” a cleaner European label isn’t scientific — it’s narrative-driven. And frankly, it misses the point of how complex population history really is.

Do you have a problem with Ashkenazi Jews being mostly Southern Italian?

Because you seem oddly uncomfortable with that — despite the fact that:

• Ashkenazim plot directly over Southern Italians on PCA
• The oldest and largest Jewish communities in Europe were in Southern Italy and Sicily
• Peer-reviewed studies (like Waldman et al. 2022, Cell) confirm ~70% Southern Italian ancestry in ancient Ashkenazim

So why do you resist it?

You just told me there’s “nothing wrong” with acknowledging the Near Eastern side — great. I agree.
But by that logic, there should be nothing wrong with acknowledging the overwhelmingly Southern Italian side either.

Yet here you are, downplaying, avoiding, or rejecting that component — even though it’s consistent across PCA, qpAdm, and ancient DNA.

Ignoring Ashkenazi PCA placement right over regions like Calabria, Naples, and Sicily —
where Jews had a major presence in antiquity —
starts to look less like scientific disagreement and more like historical denial.

So I’ll turn your words back at you:

“Trying to erase or minimize that just to make them ‘fit’ a cleaner Near Eastern label isn’t scientific — it’s narrative-driven.”
What exactly is motivating that bias in you?

[Gannicus]

Do you have a problem with Ashkenazi Jews being mostly Southern Italian?

Because you seem oddly uncomfortable with that — despite the fact that:

• Ashkenazim plot directly over Southern Italians on PCA
• The oldest and largest Jewish communities in Europe were in Southern Italy and Sicily
• Peer-reviewed studies (like Waldman et al. 2022, Cell) confirm ~70% Southern Italian ancestry in ancient Ashkenazim

So why do you resist it?

You just told me there’s “nothing wrong” with acknowledging the Near Eastern side — great. I agree.
But by that logic, there should be nothing wrong with acknowledging the overwhelmingly Southern Italian side either.

Yet here you are, downplaying, avoiding, or rejecting that component — even though it’s consistent across PCA, qpAdm, and ancient DNA.

Ignoring Ashkenazi PCA placement right over regions like Calabria, Naples, and Sicily —
where Jews had a major presence in antiquity —
starts to look less like scientific disagreement and more like historical denial.

So I’ll turn your words back at you:

“Trying to erase or minimize that just to make them ‘fit’ a cleaner Near Eastern label isn’t scientific — it’s narrative-driven.”
What exactly is motivating that bias in you?

�� Do you have a problem with Ashkenazi Jews being partly Near Eastern?
Because you seem oddly uncomfortable with that — despite the fact that:

✅ Peer-reviewed qpAdm models (Waldman et al. 2022, Behar et al. 2010, etc.) show ~25–30% Levantine ancestry in ancient Ashkenazim
✅ They carry Near Eastern Y-DNA (J1, J2, E-M123) — not typically European lineages like R1b or I2
✅ On PCA, Ashkenazim plot between Southern Italians and Levantines — not inside the Italian cluster
✅ Ancient DNA from the Levant (Israel_C, Ashkelon_LBA, etc.) shows strong continuity with the Near Eastern component in Ashkenazim

You said:
“There should be nothing wrong with acknowledging the overwhelmingly Southern Italian side.”

I agree — and I don’t reject that component.

But by that logic, there should also be nothing wrong with acknowledging the Levantine/Near Eastern side, especially when it’s clear in both autosomal and uniparental data.

�� But here’s what you’re doing:
�� Rebranding Near Eastern haplogroups like J2 and E-M123 as “strongly European”
�� Acting like PCA overlap with Southern Italians means same origin — not shared admixture
�� Ignoring the fact that Southern Italians themselves carry substantial Near Eastern ancestry
�� Failing to consider that some of the “Italian” signal in qpAdm models is likely misallocated Near Eastern ancestry shared across the Eastern Mediterranean

�� Let’s break that last point down:
Modern Southern Italians are not a clean “European” source. They absorbed:

Neolithic Levantine ancestry (via Anatolia)

Bronze Age Aegean & Near Eastern input

Phoenician, Greek, Roman, and Byzantine Easterners

So when you model Ashkenazim as 70% “Southern Italian,” some of that is shared Near Eastern ancestry, not unique Italian ancestry.

You’re not identifying a purely “Southern European” origin — you’re pointing to a shared genetic blend across the Eastern Mediterranean. And qpAdm reflects that blend, not neat ethnic categories.

�� And your historical argument?
Yes, Jews lived in Southern Italy. That doesn’t mean Ashkenazim descend from Italians. It means there was interaction and admixture, but the Near Eastern component remains distinct and substantial, both genetically and historically.

�� And don’t forget the Yamnaya…
Even the “European” ancestral layer — Yamnaya — was ~50% CHG/Iran_N, which are Near Eastern-adjacent ancestries. West Eurasia is deeply interconnected.

⚖️ Final Point:
Ashkenazim are a West Eurasian population, with ancestry from both Southern Europe and the Levant.
Pretending one half tells the full story — whether European or Near Eastern — is what’s narrative-driven.

You quoted me:

“Trying to erase or minimize that just to make them ‘fit’ a cleaner Near Eastern label isn’t scientific.”

Here’s the flip side:

Trying to erase or minimize the Near Eastern side just to make them ‘fit’ a cleaner European label isn’t scientific either.

Let’s not play that game.

[rothaer]

(...) So according to you, early Jews heavily proselytized in Sicily, but didnt even touch Germans with a 10-foot pole.
Makes sense. :picard1:

This consideration is legit. But the answer is simple. In the 4th century (380 AD) Christianity became official state religion in the Roman Empire and at the same time the rights of Jews were restricted and repeatedly religious mission by Jews became forbidden. In the Middle Ages there was even death penalty on such mission in various European countries. This was a total game changer for Judaism. The Southern Italian converts were from before 380 AD and the Ashkenazis in the Rhineland formed from 917 AD onwards.

[Dušan]

This consideration is legit. But the answer is simple. In the 4th century (380 AD) Christianity became official state religion in the Roman Empire and at the same time the rights of Jews were restricted and repeatedly religious mission by Jews became forbidden. In the Middle Ages there was even death penalty on such mission in various European countries. This was a total game changer for Judaism. The Southern Italian converts were from before 380 AD and the Ashkenazis in the Rhineland formed from 917 AD onwards.

Are these Ashkenazis from Rhineland came from Italy or Spain?

[Alenka]

This consideration is legit. But the answer is simple. In the 4th century (380 AD) Christianity became official state religion in the Roman Empire and at the same time the rights of Jews were restricted and repeatedly religious mission by Jews became forbidden. In the Middle Ages there was even death penalty on such mission in various European countries. This was a total game changer for Judaism. The Southern Italian converts were from before 380 AD and the Ashkenazis in the Rhineland formed from 917 AD onwards.
I see.
So it makes sense German input is likely much lower than South Italian.

But i dont buy that there is absolutely none of it.
Even IBD segment sharing that Lard posted proves it is there.

Either way, it should be easy enough to prove me wrong (or right :cool:) with a model including:
South Italian proxy, and Levantine proxy, and Slavic proxy, and Rhineland German proxy.

If Rhineland German input is actually miniscule, it will allocate just noise level range percentage at most.
My guess is that its not so extremely low though.

[Lard]

Update: The 3-Way qpAdm Model Has Arrived
As promised, here's the first stable three-way qpAdm model of Ashkenazi Jewish ancestry, incorporating Southern Italian, Levantine, and Slavic/Eastern European input. This model is based on unsupervised PCA structure and produces a statistically valid fit with strong standard errors.

Model 4: Southern Italian + Levantine + Slavic
Target: Jew_Ashkenazi.HO
Sources:
• Italian_South.HO — 66.88% ± 5.86%
• Lebanese_Muslim.HO — 23.85% ± 4.10%
• Russian.DG — 9.28% ± 2.58%
p-value: 0.1033
Feasibility: TRUE

https://i.ibb.co/yLp1S7T/777.png (https://ibb.co/JXq4QgL)

This model shows a much smaller Levantine minority than what is commonly claimed in mainstream population genetics papers. Those inflated figures often rely on cherry-picked source populations such as Northern Italians, Tuscans, or French, which do not align with PCA clustering, do not reflect historically inhabited regions of early Jewish communities in Italy, and are inconsistent with Y-DNA lineages (J1, J2a, and E subclades) that Ashkenazi Jews share with Southern Italians and related Greek populations. This model corrects that by grounding itself in both unsupervised data structure and historical plausibility.

What’s Next:
4-way modeling with Central Europeans (e.g. Czech, Polish, German and Hungarian proxies) is currently in development and will be posted shortly. These models will test whether adding an additional Central/East European source improves resolution of the Slavic component.

Reminder:
My offer still stands — if you want a free qpAdm model done, just PM me.
The only condition:
If you've spammed my threads or lied about me on this site, don't bother. I have no issue with posters who disagree in good faith, but liars and hate-spammers are not welcome.

More coming soon.

[Lard]

Are these Ashkenazis from Rhineland came from Italy or Spain?

Yeah, they came from Italy — not Spain.

The core Ashkenazi population formed mainly from Jews already settled in Southern and Central Italy during the Roman period. After things got worse under Christian rule post-380 AD, a lot of those communities gradually moved northward over centuries — into the Rhineland by around 900–1000 AD. There's no solid evidence of a Spanish origin for early Ashkenazim. The Iberian Jews (Sephardim) were a totally separate group with a different migration path, and the major expulsion from Spain didn’t even happen until 1492 — long after the Ashkenazi population had already formed in the Rhineland and moved into Eastern Europe.

[Dušan]

Yeah, they came from Italy — not Spain.

The core Ashkenazi population formed mainly from Jews already settled in Southern and Central Italy during the Roman period. After things got worse under Christian rule post-380 AD, a lot of those communities gradually moved northward over centuries — into the Rhineland by around 900–1000 AD. There's no solid evidence of a Spanish origin for early Ashkenazim. The Iberian Jews (Sephardim) were a totally separate group with a different migration path, and the major expulsion from Spain didn’t even happen until 1492 — long after the Ashkenazi population had already formed in the Rhineland and moved into Eastern Europe.

Map from wikipedia :dunno:

What a mess

https://upload.wikimedia.org/wikipedia/commons/thumb/6/65/Expulsion_judios-en.svg/2560px-Expulsion_judios-en.svg.png

[Lard]

Map from wikipedia :dunno:

What a mess

https://upload.wikimedia.org/wikipedia/commons/thumb/6/65/Expulsion_judios-en.svg/2560px-Expulsion_judios-en.svg.png
Dušan, do yourself a favor and stay away from Wikipedia for anything population genetics–related. It’s run by a handful of power-drunk moderators who gatekeep the content and push a sanitized, often misleading narrative. They routinely suppress legitimate findings or alternative models if it doesn't fit their preferred framing. It’s not a reliable source — especially when you’re dealing with something as technical and politically sensitive as human ancestry.

And just to be clear — Reddit and Quora are no better. They’re both cesspools of recycled misinformation, driven by clout-chasing posters and upvote farming. Most of the top answers are just copy-pasted summaries from outdated papers or cherry-picked snippets from Wikipedia itself. There's no quality control, no critical analysis — just parroting.

If you actually want to understand this stuff, skip the crowdsourced junk. Go straight to the primary academic literature, actual model results (qpAdm, f-statistics, PCA plots, ADMIXTURE runs), and discussions where the raw data is being worked with — not filtered through five layers of agenda or ignorance.

[cass]

Are these Ashkenazis from Rhineland came from Italy or Spain?

The population was exiled to Sardinia, and that's where it consolidated, as far as I know.

[rothaer]

I see.
So it makes sense German input is likely much lower than South Italian.

But i dont buy that there is absolutely none of it.
Even IBD segment sharing that Lard posted proves it is there.

Either way, it should be easy enough to prove me wrong (or right :cool:) with a model including:
South Italian proxy, and Levantine proxy, and Slavic proxy, and Rhineland German proxy.

If Rhineland German input is actually miniscule, it will allocate just noise level range percentage at most.
My guess is that its not so extremely low though.

Cross checking can never be objected. :thumb001:

[Lard]

The population was exiled to Sardinia, and that's where it consolidated, as far as I know.

The core of Ashkenazi ancestry wasn’t formed in Sardinia — it was shaped much earlier, primarily in Magna Graecia (Southern Italy), where Greek-speaking Jews were actively converting locals during the Hellenistic period. This continued through Roman times, before Christianity took over.

There were also massive Jewish communities in places like Alexandria and Cyrene (Libya), and all along the shores of the Black Sea in Greek colonies — again, all in the pre-Christian era. A lot of integration and conversion happened among Greeks and Romans well before any major exile narrative. That’s the real ancestral core, not Sardinia.

[cass]

The core of Ashkenazi ancestry wasn’t formed in Sardinia — it was shaped much earlier, primarily in Magna Graecia (Southern Italy), where Greek-speaking Jews were actively converting locals during the Hellenistic period. This continued through Roman times, before Christianity took over.

There were also massive Jewish communities in places like Alexandria and Cyrene (Libya), and all along the shores of the Black Sea in Greek colonies — again, all in the pre-Christian era. A lot of integration and conversion happened among Greeks and Romans well before any major exile narrative. That’s the real ancestral core, not Sardinia.

That's possible, but the Ashkenazi population is very specific, and its genetic lineage has been tested multiple times and published in scientific literature. What's more, it went through, AFAIR, three population bottlenecks before expanding significantly.


https://systemsbiology.columbia.edu/sites/default/files/Ashkenazi_Bottleneck_Model.jpg

[Lard]

That's possible, but the Ashkenazi population is very specific, and its genetic lineage has been tested multiple times and published in scientific literature. What's more, it went through, AFAIR, three population bottlenecks before expanding significantly.


https://systemsbiology.columbia.edu/sites/default/files/Ashkenazi_Bottleneck_Model.jpg
You're misunderstanding what I said — this isn’t a theory, it’s historical record. Greek-speaking Jews were living in what is now Southern Italy (Magna Graecia) for over 2300 years. They were well-established in the Hellenistic world, including Greek colonies along the Black Sea, Alexandria (which was up to 1/3 Jewish at one point), and Cyrene. These weren’t fringe communities — they were large, active, and integrated with the broader Greco-Roman world well before Christianity and long before any supposed Sardinian bottleneck.

As for genetics: yes, Ashkenazi Jews have gone through multiple bottlenecks and founder effects — no disagreement there. But that’s exactly why it’s misleading to treat modern Ashkenazi haplogroups as a reliable window into deeper ancestry. Those bottlenecks severely distort lineage signals and make it easy to draw the wrong conclusions if you're just recycling the same limited modeling frameworks.

Most studies do just that — they keep using the same reference populations (like Northern Italians or French) that have no historical connection to early Jewish communities. Rinsing and repeating flawed assumptions doesn't make the results more accurate — it just reinforces the same blind spots.

If you want to understand Ashkenazi origins, you have to take both the archaeological/historical record and the unbiased PCA and qpAdm modeling seriously. And those point consistently to a Southern Italian–centered core formed in the Greco-Roman world — not Sardinia, and not Northern Europe.

[rothaer]

Yeah, they came from Italy — not Spain.

The core Ashkenazi population formed mainly from Jews already settled in Southern and Central Italy during the Roman period. After things got worse under Christian rule post-380 AD, a lot of those communities gradually moved northward over centuries — into the Rhineland by around 900–1000 AD. (...)

Mostly they came from Italy and a few from France afaik. The data is anecdotal just, but, however, one very important founder Ashkenazi family came from Lucca in Northern Italy, where there in earlier times had not been recorded any notable Jewish population, and from Rome. Tellingly it's name is Kalonymos, i. e. a Greek name. https://en.wikipedia.org/wiki/Kalonymos_family This fits well to Magna Graecia (Southern Italy).

[Gannicus]

Your Quote:

"And that jab about ChatGPT? Please. I don’t need AI to expose weak arguments — your posts do that on their own. If you dish out condescension, don’t whine when you get firm replies in return."

The only reason I ripped you is because of your rude condescending tone in response to my first comment in this thread.

Post #22 I said:

"Don't forget J2 is common in the Levant as well." < Nothing rude or condescending about that.

You replied to me:

Sure — and it’s just as common, if not more so, in Southern Italy, Crete, mainland Greece, and across the Caucasus. So what?

You can find J2 all over the Eastern Mediterranean and West Asia. It’s a broadly distributed macro-haplogroup, not a Levantine fingerprint.

The relevant point is this:
Ashkenazi Jews don’t cluster with Levantines on PCA. They cluster with Southern Italians and Cretans — two regions that also carry J2, J1, and E-M123 in abundance. So if those haplogroups automatically meant “Levantine ancestry,” you’d have to call half of Southern Italy and Crete Levantine too — which no one does.

But that’s not how population genetics works. Haplogroups ≠ autosomal ancestry.

Trying to tie Ashkenazi J2 to the Levant without acknowledging that PCA and autosomal models consistently place them in Southern Europe — and that those same haplogroups are common there too — is cherry-picking trivia, not analysis.

If you’re going to bring up J2, then bring the full context — or don’t bring it at all.


Pretty rude if you ask me. If you reply respectfully to me, I will do the same. Talk to me like that and you will receive it back, probably worse.

This was not produced by Dragon or any speech to text software and AI prompting for flow.

[Lard]

Mostly they came from Italy and a few from France afaik. The data is anecdotal just, but, however, one very important founder Ashkenazi family came from Lucca in Northern Italy, where there in earlier times had not been recorded any notable Jewish population, and from Rome. Tellingly it's name is Kalonymos, i. e. a Greek name. https://en.wikipedia.org/wiki/Kalonymos_family This fits well to Magna Graecia (Southern Italy).

Appreciate the reply — and you're actually pointing toward something deeper that often gets glossed over in mainstream discussions.

Yes, the Kalonymos family is a known founder lineage, and while it shows up in Northern Italy (Lucca), the name itself is Greek — not Latin or Hebrew — which already hints at the real origins. That name didn’t originate in Lucca; it came up from the Hellenized Jewish world. Most genetic narratives skip over that entire layer and jump straight from ancient Judea to medieval Rhineland, ignoring centuries of Greco-Roman Jewish presence across the Mediterranean.

I actually made a full thread breaking all of this down — genetics, Y-DNA, mtDNA, and the historical record — showing how Ashkenazi Jews formed as a Mediterranean population rooted in Magna Graecia and the broader Hellenistic world:

Thread: Buried in Plain Sight – Southern European Ethnogenesis of Ashkenazi Jews
https://www.theapricity.com/forum/showthread.php?396824-–-Buried-in-Plain-Sight-Southern-European-Ethnogenesis-of-Ashkenazi-Jews (https://www.theapricity.com/forum/showthread.php?396824-%26%23129516%3B-Buried-in-Plain-Sight-Southern-European-Ethnogenesis-of-Ashkenazi-Jews)

Not many replies, but the discussion was civil and on-point. Some key takeaways:


Romaniote and Italkim Jews show clear genetic continuity with Southern Italians, Greeks, and Aegean populations — not Northern Italians or French.
PCA and G25 plots consistently place Ashkenazi Jews overlapping Southern Italians and Sicilians — not halfway between Europe and the Levant.
The presence of J1 in Ashkenazim is often overstated as "Levantine," but J1 — including the P58 (Cohen modal) subclade — is well-documented in Southern Italy.
mtDNA studies (e.g., Costa et al. 2013) show over 80% of Ashkenazi maternal ancestry is from prehistoric European women.
Recent autosomal models (Brace 2022, Waldman 2022) place Ashkenazi Jews as around 67–70% Southern Italian/Sicilian, with minority inputs from the Levant and Eastern Europe.


In short: Ashkenazi Jews didn’t form in isolation in the Rhineland or from some direct Israelite-to-Germany migration. Their ancestry reflects centuries of Jewish life, conversion, and integration throughout the Greco-Roman Mediterranean — especially Magna Graecia. That’s the real ethnogenesis, and the data backs it up.

Let me know what you think if you check out the thread.

[Alenka]

I dont know a whole lot about this stuff, can anyone familiar with genetics please explain if this is possible?

In post #1, (https://www.theapricity.com/forum/showthread.php?397054-%26%23129516%3B-Stable-qpAdm-Models-of-Ashkenazi-Jewish-Ancestry-%96-Ignoring-PCA-No-More&p=8265155&viewfull=1#post8265155) Lard posted a model with South Italian and Lebanese Muslim:
https://i.ibb.co/M0ZfVqC/1.png

In post 64 (https://www.theapricity.com/forum/showthread.php?397054-%26%23129516%3B-Stable-qpAdm-Models-of-Ashkenazi-Jewish-Ancestry-%96-Ignoring-PCA-No-More&p=8265991&viewfull=1#post8265991) Lard posted a model with South Italian and Lebanese Muslim and Russian:
https://i.ibb.co/yLp1S7T/777.png

In his model, adding Russian resulted in decreased Levantine. Is this realistic?
Or does it imply he used different proxies for South Italian and/or Lebanese Muslim for each of these two models?

It would be very interesting if any of the genetic enthusiast members tried replicating something akin to these two models and posted what results they get.
I would do it myself, but I dont know how to run these tools.

[rothaer]

(...)


(...)

The modelling with Southern Italians instead of Northern Italians that dramatically reduces the (direct) Levant ancestry proportion is not just some fringe imagination that is unbacked in scientific papers. The first time I saw this modelling - long after I claimed in different forums that that for obvious reasons had to be done - was in the Erfurt Jew paper from 2022. Expectedly it shows a notable drop in the (direct) Levant ancestry. It states:

"We used qpAdm to test quantitative models for the ancestral sources of EAJ (STAR Methods). Based on the PCA above and previous modeling (Xue et al., 2017), we considered a model where EAJ is a mixture of the following sources: Southern European (South Italians or North Italians), Middle Eastern (Druze, Egyptians, Bedouins, Palestinians, Lebanese, Jordanians, Syrians, or Saudis), and Eastern European (Russians). We used modern populations as sources, as modeling with ancient sources was unsuccessful (Data S1, section 7). Multiple models with South-Italians were plausible (p>0.05; Table S3), which would be consistent with historical models pointing to the Italian peninsula as the source for the AJ population (Data S1, section 16; though see below for alternatives and caveats). The mean admixture proportions (over all of our plausible models; Table S3) were 65% South Italy, 19% ME, and 16% East-EU (Figure 3A). We validated that our results did not qualitatively change when using only transversions vs. all SNPs, a different outgroup population, or fewer SNPs (Table S3; Data S1, section 7). Estimates of the admixture time were unreliable in our setting (Data S1, section 8).

Models with other sources, in particular Mediterranean, could also fit the EAJ data (Data S1, section 7). A model with North Italians as a source (Table S3) had ancestry proportions 37% North Italy, 43% ME, and 20% East-EU (Figure 3A). Models with Greek as a source had average ancestry proportions 51% Greek, 32% ME, and 17% East-EU (Table S3). Models with Spanish or North African sources (in addition to ME and East-EU) were not plausible. A model with all Levant populations merged together as the ME source fit the EAJ data (p = 0.07), with ancestry proportions 65% South Italy, 19% Levant, and 16% East-EU. A model with all Mediterranean populations merged as a single source (Middle Eastern, Greek, and Italian, with East-EU as the other source) fit the data (p = 0.11) with ancestry proportions 89% Mediterranean and 11% East-EU. Models with a Western European source (Germans) instead of Russians were not plausible. There was also no support for a minor ancestry component from East Asians."

https://www.sciencedirect.com/science/article/pii/S0092867422013782

[rothaer]

Quote from the paper above: "There was also no support for a minor ancestry component from East Asians."

This statement has legitimately been much criticised by Davidski that asked why they used modern Russians as proxy for a Slavic input in Central Europe. And indeed there hides a small East Asian contribution that was not in the proto Slavs.

However, it has to be kept in mind that the Erfurt Jews with their partly notable Slavic contributions were not typical for Ashkenazi Jews. But, ofc, they contained the same stock of Jews that had immigrated to the Rhineland and had formed the AJ. The secondary admixture with Slavs is irrelevant for the matter of modelling AJ with Southern Italians instead of Northern Italians.

[Alenka]

Here I've made my own tiny PCA system to estimate jewish ancestry on the basis of an Fst matrix.

Egyptian.HO,0,20,4,16,8,12,11,21,9,16,18,5,5,21,19 ,13
Ukrainian.HO,20,0,24,10,9,5,11,0,8,3,15,15,16,4,3, 5
Tunisian.HO,4,24,0,22,14,16,15,25,14,20,21,11,12,2 5,22,17
Lezgin.HO,16,10,22,0,5,9,12,11,8,10,21,9,11,12,11, 11
Turkish.HO,8,9,14,5,0,4,6,10,3,7,13,4,5,11,9,6
Italian_North.HO,12,5,16,9,4,0,6,7,2,2,7,8,10,5,3, 1
Russian.HO,21,0,25,11,10,7,13,0,9,4,18,16,18,4,4,6
Italian_South.HO,9,8,14,8,3,2,5,9,0,5,8,5,7,9,6,3
French.HO,16,3,20,10,7,2,8,4,5,0,10,12,14,2,1,1
Sardinian.HO,18,15,21,21,13,7,12,18,8,10,0,15,17,1 5,12,7
Lebanese.HO,5,15,11,9,4,8,8,16,5,12,15,0,4,17,13,1 0
Syrian.HO,5,16,12,11,5,10,9,18,7,14,17,4,0,18,15,1 2
Norwegian.HO,21,4,25,12,11,5,12,4,9,2,15,17,18,0,2 ,4
English.HO,19,3,22,11,9,3,9,4,6,1,12,13,15,2,0,2
Spanish.HO,13,5,17,11,6,1,7,6,3,1,7,10,12,4,2,0

Jew_Ashkenazi.HO,11,11,15,12,6,6,0,13,5,8,12,8,9,1 2,9,7

Target: Jew_Ashkenazi.HO
Distance: 829.2003% / 8.29200338
43.2 Italian_South.HO
19.8 Syrian.HO
14.2 English.HO
10.0 Sardinian.HO
7.4 Tunisian.HO
5.4 Turkish.HO

Here we see that indeed the biggest contributor to Ashkenazi DNA is South Italian. At least in this model.
Let's compare something though.
Lard proclaims that a South Italian + Lebanese model is ideal for Ashkenazi modeling. I disagree. I think the ideal model is South Italian + Lebanese + Russian.
Thankfully, I included all these populations in my Fst matrix. So let's compare them, using fit to gauge plausibility.
Target: Jew_Ashkenazi.HO
Distance: 1125.7990% / 11.25798987 | R3P
63.2 Italian_South.HO
36.8 Lebanese.HO

vs
Target: Jew_Ashkenazi.HO
Distance: 983.6796% / 9.83679643 | R3P
42.2 Lebanese.HO
40.2 Italian_South.HO
17.6 Russian.HO

Clearly the model with Russian gives a better fit.
Notice how the portion of Lebanese drastically increased too!

For example, a few pages back member AndreiDNA posted two models very similar to Lards.

First model just Italian_South and Lebanese, in the second model he added Russian.

Yet he got the opposite result, the portion of Lebanese increased.

Makes sense to look into why they got seemingly opposite results.

[Peterski]

By the way, before the migration of Ashkenazi Jews to Poland, there were already some Slavic-speaking Jews in Poland:

https://en.wikipedia.org/wiki/Knaanic_language

It is possible that Slavic admixture in Ashkenazi Jews is not directly from Slavs but rather from those Knaanim Jews.

[Lard]

By the way, before the migration of Ashkenazi Jews to Poland, there were already some Slavic-speaking Jews in Poland:

https://en.wikipedia.org/wiki/Knaanic_language

It is possible that Slavic admixture in Ashkenazi Jews is not directly from Slavs but rather from those Knaanim Jews.

That’s true — the Knaanim did exist in early medieval Poland, but it doesn’t change much in terms of ancestry. They themselves were already mixed with local Slavs and spoke a West Slavic language. So even if some of the Slavic admixture in Ashkenazi Jews came via Knaanic Jews rather than directly from surrounding Slavic populations, the genetic input is still Slavic in origin either way. It’s just a different vector for the same component.

[Feiichy]

what about Sephardic Jews, got any plausible models for them?

[rothaer]

I dont know a whole lot about this stuff, can anyone familiar with genetics please explain if this is possible?

In post #1, (https://www.theapricity.com/forum/showthread.php?397054-%26%23129516%3B-Stable-qpAdm-Models-of-Ashkenazi-Jewish-Ancestry-%96-Ignoring-PCA-No-More&p=8265155&viewfull=1#post8265155) Lard posted a model with South Italian and Lebanese Muslim:
https://i.ibb.co/M0ZfVqC/1.png

In post 64 (https://www.theapricity.com/forum/showthread.php?397054-%26%23129516%3B-Stable-qpAdm-Models-of-Ashkenazi-Jewish-Ancestry-%96-Ignoring-PCA-No-More&p=8265991&viewfull=1#post8265991) Lard posted a model with South Italian and Lebanese Muslim and Russian:
https://i.ibb.co/yLp1S7T/777.png

In his model, adding Russian resulted in decreased Levantine. Is this realistic?
Or does it imply he used different proxies for South Italian and/or Lebanese Muslim for each of these two models?

It would be very interesting if any of the genetic enthusiast members tried replicating something akin to these two models and posted what results they get.
I would do it myself, but I dont know how to run these tools.

You see that the second modelling has a somewhat better fit. Adding Russian made the Levant part decrease by 1 percent point (only) and made the South Italian proportion decrease by 8 percent points. I see no inconsistency in that. When you model with two components just, you force everything into them, no matter if that is suitable or not. You can model me with Chinese and Nigerian just. I will get 100% of these two components then. Although the fit will not be the best. When you then add the component Moroccan, I will get a lot of that. This will make both Chinese and Nigerian notably shrink. Also, if I would happen to have some real Nigerian that would not make it unplausible that Nigerian shrinks when Moroccan is added.

[Lard]

what about Sephardic Jews, got any plausible models for them?

Good question — and honestly, if Ashkenazim are already complicated to model, Sephardic Jews are even more so. The main reason is their post-expulsion diaspora: after being expelled from Iberia in 1492, Sephardim mixed with a wide range of local Jewish communities across North Africa, the Ottoman Empire (including Turkish, Syrian, and Balkan Jews), and even Italy.

I’m using qpAdm via Illustrative DNA, and while the tool is great, the model datasets for Sephardim are still pretty limited — especially when it comes to differentiating “pure” Iberian Sephardim from later mixed populations.

That said, on unsupervised PCA plots, Sephardim from Iberia cluster very closely with Ashkenazim, which strongly suggests a shared ancestral base rooted in the Greco-Roman Mediterranean. The key difference is that Sephardim lack the Eastern European and Germanic ancestry components seen in Ashkenazim.

Also worth noting — Y-DNA studies of Iberian Sephardim show a substantial number of Spanish/Portuguese male lineages, which makes sense given centuries of intermarriage and conversion within Iberia prior to expulsion.

Bottom line: Sephardim and Ashkenazim likely share a Mediterranean genetic core — probably stemming from Roman-era and Hellenistic Jewish populations — but later histories sent them on very different admixture paths.

[rothaer]

For example, a few pages back member AndreiDNA posted two models very similar to Lards.

First model just Italian_South and Lebanese, in the second model he added Russian.

Yet he got the opposite result, the portion of Lebanese increased.

Makes sense to look into why they got seemingly opposite results.

Indeed. I see the difference in Labels between Lebanese and Lebanese_Muslim. I've no possibility to assess what data exactly are used in Lard's model. For enlightening Ashkenazi ancestry, I'd use Ashkenazi_Germany only. They are the purest and essentially without a secondary Slavic contribution. The latter may be assessed as well, ofc.

[Alenka]

You see that the second modelling has a somewhat better fit. Adding Russian made the Levant part decrease by 1 percent point (only) and made the South Italian proportion decrease by 8 percent points. I see no inconsistency in that. When you model with two components just, you force everything into them, no matter if that is suitable or not. You can model me with Chinese and Nigerian just. I will get 100% of these two components then. Although the fit will not be the best. When you then add the component Moroccan, I will get a lot of that. This will make both Chinese and Nigerian notably shrink. Also, if I would happen to have some real Nigerian that would not make it unplausible that Nigerian shrinks when Moroccan is added.
But Russian is way more northern compared to South Italian.

So, when you add Russian, it will eat up some of the northernmost stuff out of South Italian.
And therefore, some of the southernmost stuff out of South Italian will have to be eaten by Levant.

Isnt it so?

[rothaer]

(...) It’s just a different vector for the same component.

After my claim that Peterski's Y DNA most likely was mediated by Germanics (Restgermanen) tormented him, he has become very sensitive in the matter of vectors. :p

[Lard]

Indeed. I see the difference in Labels between Lebanese and Lebanese_Muslim. I've no possibility to assess what data exactly are used in Lard's model. For enlightening Ashkenazi ancestry, I'd use Ashkenazi_Germany only. They are the purest and essentially without a secondary Slavic contribution. The latter may be assessed as well, ofc.

Just to clarify — I built the model using qpAdm through AdmixLab on Illustrative DNA, which uses datasets based on curated samples from Allen Ancient DNA Resource and Human Origins panels. These are some of the most widely used and published reference populations in population genetics.

Here are the exact sources I used in that 3-way model of Ashkenazi Jews:


Italian_South.HO — modern individuals from Southern Italy, including regions like Calabria, Apulia, and Sicily. These are the closest proxies for Roman-era and Magna Graecia Jewish communities.
Lebanese_Muslim.HO — modern Lebanese Muslims, often used as a general Levantine proxy in qpAdm when ancient Bronze Age samples aren’t available or don’t pass model feasibility.
Russian.DG — from the DG (Diverse Genome) panel, representing modern ethnic Russians. Used here to represent the Slavic/Eastern European component.


As for the Ashkenazi sample, I used Jew_Ashkenazi.HO, which is a general composite sample found in the Human Origins panel. You're right that Ashkenazi_Germany might be slightly more "unmixed" in terms of Slavic input, but even Jew_Ashkenazi.HO typically overlaps with the same PCA position — clustered over Southern Italy — and the models are still very informative.

The goal was to reflect the PCA structure rather than exaggerate Levantine or Northern European ancestry by using out-of-place references like French or North Italians. This model does that well.

[rothaer]

But Russian is way more northern compared to South Italian.

So, when you add Russian, it will eat up some of the northernmost stuff out of South Italian.
And therefore, some of the southernmost stuff out of South Italian will have to be eaten by Levant.

Isnt it so?

IF it eats out some such stuff from Southern Italian what you say would be a logical consequence, yes.

Hard to say what will be the change for the Levant proportion at the bottom line if some Levant from the first 2-way-model would tend more to Russia than to the Levant (or Southern Italy), f. i. for containing much CHG.

[rothaer]

Just to clarify — I built the model using qpAdm through AdmixLab on Illustrative DNA, which uses datasets based on curated samples from Allen Ancient DNA Resource and Human Origins panels. These are some of the most widely used and published reference populations in population genetics.
(...)

The question is rather into what data (k10, k13, k25...) these samples are simplified for the purpose of calculations. But I guess that that is not visible using the tools at IllustrativeDNA. Compared to this Andrei posted the used data rows that maybe have some 13 positions (I didn't bother to count).

[Lard]

The question is rather into what data (k10, k13, k25...) these samples are simplified for the purpose of calculations. But I guess that that is not visible using the tools at IllustrativeDNA. Compared to this Andrei posted the used data rows that maybe have some 13 positions (I didn't bother to count).

You're right that with Illustrative DNA, especially when using AdmixLab’s qpAdm interface, you don’t directly see the reduced data formats (like K10, K13, K25, etc.) because it’s not based on admixture calculators like those used in GEDmatch or Dodecad. What you're working with there is full f-statistics-based modeling — not ADMIXTURE components or synthetic calculator percentages.

In other words, the modeling runs on the raw allele frequency differences via qpAdm using source and outgroup populations — not simplified component breakdowns. The numbers I posted (e.g., 66.88% Italian_South.HO, etc.) are direct qpAdm output proportions, not K-values or admixture calculator results.

So this isn’t a case of converting to a K13 or K25 structure. It's a supervised method based on real population references and allele frequency comparisons — closer to academic modeling than hobbyist calculators.

Andrei's dataset may be something else entirely (probably a custom ADMIXTURE setup), but in the case of qpAdm on Illustrative DNA, you're getting structure based on formal stats, not component bins.

[rothaer]

You're right that with Illustrative DNA, especially when using AdmixLab’s qpAdm interface, you don’t directly see the reduced data formats (like K10, K13, K25, etc.) because it’s not based on admixture calculators like those used in GEDmatch or Dodecad. What you're working with there is full f-statistics-based modeling — not ADMIXTURE components or synthetic calculator percentages.

In other words, the modeling runs on the raw allele frequency differences via qpAdm using source and outgroup populations — not simplified component breakdowns. The numbers I posted (e.g., 66.88% Italian_South.HO, etc.) are direct qpAdm output proportions, not K-values or admixture calculator results.

So this isn’t a case of converting to a K13 or K25 structure. It's a supervised method based on real population references and allele frequency comparisons — closer to academic modeling than hobbyist calculators.

Andrei's dataset may be something else entirely (probably a custom ADMIXTURE setup), but in the case of qpAdm on Illustrative DNA, you're getting structure based on formal stats, not component bins.

Wow, interesting. I don't fully grasp how qpAdm works but I understand the direction. That must consume a lot of calculatory power compared to the reduced data formats.

However, we now can point out a notable difference in the modelling calculations (which was Alenka's question).

[Lard]

Wow, interesting. I don't fully grasp how qpAdm works but I understand the direction. That must consume a lot of calculatory power compared to the reduced data formats.

However, we now can point out a notable difference in the modelling calculations (which was Alenka's question).

Exactly — that’s the key distinction. Tools like K13, K25, etc., are based on ADMIXTURE-style modeling, which simplifies ancestry into component percentages. It’s useful for broad strokes, but it tends to miss nuance — especially with complex populations like Ashkenazi Jews.

qpAdm, on the other hand, is a supervised method that models the target population as a mixture of real reference populations using allele frequency-based f-statistics. It doesn’t generate clusters; it tests whether a proposed model fits the data — using specific source and outgroup populations. And yes, it’s far more computationally demanding, but much more robust for detailed ancestry modeling.

That said — I haven’t been reading any of Alenka’s posts. She’s been uncivil with me in the past, and I’m not interested in fielding questions from people who just want to spam threads and stir drama. If she genuinely wants answers, she can go directly to the tool I used:
https://lab.illustrativedna.com/
It’s open to anyone and very straightforward once you learn the basics.

Now back to the modeling — qpAdm is extremely sensitive to outgroup selection. Choosing the right set of outgroups can make or break a model’s feasibility. It takes trial and error, lots of tweaking, and deep familiarity with the datasets to get something that holds up — especially when you’re modeling admixed populations with multiple ancestral layers like Ashkenazim.

Bottom line: Ashkenazi Jews are consistently modeled as about 2/3 to 3/4 Southern Italian–like, with a notable Slavic/Eastern European contribution. I’m currently working on a 4-way model to refine this even further, but it’s not simple — and anyone who hasn’t actually worked with qpAdm has no idea what a juggling act it can be.

So yes — I do get frustrated when people like Alenka come in acting like experts while being rude and dismissive. I’m here to discuss data, not deal with that nonsense.

[Gannicus]

Exactly — that’s the key distinction. Tools like K13, K25, etc., are based on ADMIXTURE-style modeling, which simplifies ancestry into component percentages. It’s useful for broad strokes, but it tends to miss nuance — especially with complex populations like Ashkenazi Jews.

qpAdm, on the other hand, is a supervised method that models the target population as a mixture of real reference populations using allele frequency-based f-statistics. It doesn’t generate clusters; it tests whether a proposed model fits the data — using specific source and outgroup populations. And yes, it’s far more computationally demanding, but much more robust for detailed ancestry modeling.

That said — I haven’t been reading any of Alenka’s posts. She’s been uncivil with me in the past, and I’m not interested in fielding questions from people who just want to spam threads and stir drama. If she genuinely wants answers, she can go directly to the tool I used:
https://lab.illustrativedna.com/
It’s open to anyone and very straightforward once you learn the basics.

Now back to the modeling — qpAdm is extremely sensitive to outgroup selection. Choosing the right set of outgroups can make or break a model’s feasibility. It takes trial and error, lots of tweaking, and deep familiarity with the datasets to get something that holds up — especially when you’re modeling admixed populations with multiple ancestral layers like Ashkenazim.

Bottom line: Ashkenazi Jews are consistently modeled as about 2/3 to 3/4 Southern Italian–like, with a notable Slavic/Eastern European contribution. I’m currently working on a 4-way model to refine this even further, but it’s not simple — and anyone who hasn’t actually worked with qpAdm has no idea what a juggling act it can be.

So yes — I do get frustrated when people like Alenka come in acting like experts while being rude and dismissive. I’m here to discuss data, not deal with that nonsense.

You responded to me earlier:

"And that jab about ChatGPT? Please. I don’t need AI to expose weak arguments — your posts do that on their own. If you dish out condescension, don’t whine when you get firm replies in return."

The only reason I ripped you is because of your rude condescending tone in response to my first comment in this thread.

Post #22 I said:

"Don't forget J2 is common in the Levant as well." < Nothing rude or condescending about that.

You replied to me:

Sure — and it’s just as common, if not more so, in Southern Italy, Crete, mainland Greece, and across the Caucasus. So what?

You can find J2 all over the Eastern Mediterranean and West Asia. It’s a broadly distributed macro-haplogroup, not a Levantine fingerprint.

The relevant point is this:
Ashkenazi Jews don’t cluster with Levantines on PCA. They cluster with Southern Italians and Cretans — two regions that also carry J2, J1, and E-M123 in abundance. So if those haplogroups automatically meant “Levantine ancestry,” you’d have to call half of Southern Italy and Crete Levantine too — which no one does.

But that’s not how population genetics works. Haplogroups ≠ autosomal ancestry.

Trying to tie Ashkenazi J2 to the Levant without acknowledging that PCA and autosomal models consistently place them in Southern Europe — and that those same haplogroups are common there too — is cherry-picking trivia, not analysis.

If you’re going to bring up J2, then bring the full context — or don’t bring it at all."


Pretty rude and condescending if you ask me. If you reply respectfully to me, I will do the same. Talk to me like that and you will receive it back, probably worse.

[rothaer]

(...)

Now back to the modeling — qpAdm is extremely sensitive to outgroup selection. Choosing the right set of outgroups can make or break a model’s feasibility. It takes trial and error, lots of tweaking, and deep familiarity with the datasets to get something that holds up — especially when you’re modeling admixed populations with multiple ancestral layers like Ashkenazim.


I'm in the AdmixLab of IllustrativeDNA. Is it there where you have the qpAdm runs? I understand what is "target" and "source" from Vahaduo, but what function does "references" have?

[rothaer]

By the way, before the migration of Ashkenazi Jews to Poland, there were already some Slavic-speaking Jews in Poland:

https://en.wikipedia.org/wiki/Knaanic_language


To me the existence of Knaanic and the somewhat unusual genetics of the explored Erfurt Jews are connected. Those Erfurt Jews - re-settlers after a pogrom - are said to have come predominantly from Silesia and Bohemia.

[rothaer]

(...)
Russian.DG — from the DG (Diverse Genome) panel, representing modern ethnic Russians. Used here to represent the Slavic/Eastern European component.
[/list]

You could consider using some Eastern Polish (or some of the known ancients like Hun_Av2) as a proto Slav proxy and some East Asian separately to meet Davidski's criticism on using Russian in this context.

[Lard]

I'm in the AdmixLab of IllustrativeDNA. Is it there where you have the qpAdm runs? I understand what is "target" and "source" from Vahaduo, but what function does "references" have?

Yes, I’m running the qpAdm models in the AdmixLab section of IllustrativeDNA. It’s part of their paid subscription — $16/month for 100 runs per day, though to be honest, they don’t strictly enforce that limit. I’ve been able to go over it without any issues.

To answer your question:

Target = the population you're modeling (e.g., Ashkenazi Jews)
Sources = candidate ancestral populations contributing to that target
References (aka "right pops") = outgroups used to help differentiate the sources. They don’t contribute ancestry themselves, but they’re essential for helping the model detect which sources are statistically valid.


If you're interested, I’d be happy to share some of the reference/outgroup sets I use to model Ashkenazi Jews — just PM me. I’m not posting them here because, honestly, there are people in this thread (like Andri DNA) who seem less interested in actual population genetics and more focused on turning qpAdm into some kind of sales pitch. It’s like they’re using The Apricity as their personal infomercial for selling paid runs.

I’m still learning too, but if you ever want help setting up models — or if you don’t have a paid plan — I’d be glad to run a qpAdm model for you for free. Just let me know.

[Lard]

You could consider using some Eastern Polish (or some of the known ancients like Hun_Av2 as a proto Slav proxy and some East Asian separately to meet Davidski's criticism on using Russian in this context.

Thanks again for the input—really appreciate it.

Just to clarify about Russian.DG: it's a diverse dataset from the Diverse Genome panel, and it includes individuals with varying levels of East and West Slavic ancestry. It’s not just Siberian-shifted or Finno-Ugric-heavy. A lot of these samples actually cluster very closely with Belarusians, Ukrainians, and even Eastern Poles on PCA, so while not perfect, it does capture a good chunk of the broader Slavic genetic space. That's why I’ve used it as a placeholder for the Slavic/Eastern European component in my qpAdm models so far.

That said, I’m still in the process of refining my Ashkenazi models—I’m not stopping at just Russian.DG. I’m now running a 4-way model, which is already pretty ambitious for qpAdm and takes some tweaking to balance sources and outgroups properly.

Also, I’ve come to think it’s pretty clear that Ashkenazim are better modeled with Sicilians rather than general Southern Italians. Sicilians show a stronger genetic affinity with Ashkenazim, and they help reduce what I think is an inflated Levantine signal that shows up when using Southern Italian samples (which often get overly Lebanese-like). Sicilian samples give a cleaner, more historically realistic fit and help separate genuine Near Eastern ancestry from what's actually deep Mediterranean structure.

Here’s a direct comparison to illustrate what I mean:

Model 1 – Using Sicilian.HO
Target: Jew_Ashkenazi.HO
p-value: 0.2283 | Feasibility: TRUE

Sicilian.HO — 83.25%
Lebanese_Muslim.HO — 16.75%

https://i.ibb.co/6cHr3zFG/650.png
Full-size link (https://ibb.co/1Yr7D1TB)

Model 2 – Using Italian_South.HO
Target: Jew_Ashkenazi.HO
p-value: 0.1252 | Feasibility: TRUE

Italian_South.HO — 75.12%
Lebanese_Muslim.HO — 24.88%

https://i.ibb.co/tPXgHQJy/South.png
Full-size link (https://ibb.co/Dgph1zR3)

The Sicilian model doesn't just fit better — it has a significantly stronger p-score of 0.2283 compared to 0.1252 in the Southern Italian model.
More importantly, it assigns only 16.75% to Lebanese_Muslim.HO, versus a much higher 24.88% when using Italian_South.HO.
That nearly 8-point drop in Levantine ancestry shows how Sicilian.HO more effectively absorbs Near Eastern-related signals, without inflating Levantine input unnaturally — which is crucial for accurate Ashkenazi modeling.

And just being honest—I try to stay away from the politics side of all this. I think Polako (Davidski) brings a strong ethnonationalist slant to the discussion. Based on how he interacts, I doubt any model I run would satisfy him regardless of the stats or historical context. There's a lot of negativity and gatekeeping in that space, and I’m not interested in that vibe.

I'm here to learn, run solid models, and share findings with people genuinely interested in population genetics. Always happy to compare notes or collaborate.

[Lard]

One poster in this thread was trolling so hard I had to banish her to the Phantom Zone. 👋

She claimed her 6th cousin’s friend’s ex-girlfriend was half Wakandian, half Narnian, and plotted Romanian on PCA — and from that, tried to argue that population structure is basically a lottery, so Ashkenazi genetics can't be valid either. 🤡

Then she tried to sound serious with this:


For example, a few pages back member AndreiDNA posted two models very similar to Lard’s.

First model just Italian_South and Lebanese, in the second model he added Russian.

Yet he got the opposite result, the portion of Lebanese increased.

Makes sense to look into why they got seemingly opposite results.

It does — if you actually understand how qpAdm works.

Adding a Slavic source like Russian.DG doesn’t automatically lower the Levantine signal. If your outgroups (right pops) aren’t designed to differentiate Slavic from East Med ancestry, then the model will blur those lines — and the Levantine percentage can actually go up, not down.

That’s why I don’t just copy-paste the same reference set blindly. I adjust the outgroups depending on the sources I use — Sicilian.HO vs. South_Italian.HO, Lebanese.HO vs. Lebanese_Muslim.HO, etc.

To illustrate that, here’s what happens when you actually model Ashkenazi Jews with those two Levantine references:

Model 1 – Using Lebanese.HO
Target: Jew_Ashkenazi.HO
p-value: 0.3887 | Feasibility: TRUE

Sicilian.HO — 89.13%
Lebanese.HO — 10.87%

https://i.ibb.co/Y4SzGJ1M/777.png
Full-size PNG (https://ibb.co/gb85CQfp)

Model 2 – Using Lebanese_Muslim.HO
Target: Jew_Ashkenazi.HO
p-value: 0.2283 | Feasibility: TRUE

Sicilian.HO — 83.25%
Lebanese_Muslim.HO — 16.75%

https://i.ibb.co/v4yhMB8X/hhh.png
Full-size PNG (https://ibb.co/chfwsvdY)

Conclusion: The model using Lebanese.HO gives a significantly higher p-score, lower standard errors, and a more realistic Levantine estimate. In other words: it’s a statistically stronger and cleaner model across the board.

The reason these models differ isn’t a “genetic lottery.” It’s because:

• qpAdm and PCA are sensitive to input — SNP count, reference sample composition, and even dataset merging all affect the outcome.
• Outgroups must be chosen carefully — they need to help the model distinguish between sources, not blur them.
• qpAdm doesn’t guess — it tests your model mathematically. If it says TRUE with a high p-value and tight errors, that’s the benchmark — not some fantasy PCA anecdote.

At the end of the day, what matters are p-value (which tells you how plausible your model is) and standard error (which tells you how stable your source estimates are).

So no — I’m not going to toss out valid models just because someone thinks ancestry components shift around like Pokémon evolutions or slot machines. If your framework relies on mythical relatives and cherry-picked exceptions, you’re not doing population genetics — you’re doing improv.

[Figaro]

Do sephardim have similar Caucasus to Ashkenazim or is that the Iranic confederati in AJ’s?

[Lard]

Do sephardim have similar Caucasus to Ashkenazim or is that the Iranic confederati in AJ’s?

Do Sephardim have similar Caucasus ancestry to Ashkenazim, or is that the “Iranic Confederation” component in AJ?

It's a tough question because Caucasus-related ancestry is hard to isolate cleanly in models. The southern Caucasus shows strong genetic affinity with the ancient southern Levant — and so do Anatolian and Iranian signals. These can overlap heavily in admixture tools, making it hard to untangle which is which.

As for Sephardim, the modeling used to define them is often pretty shaky: small sample sizes, inconsistent definitions, and lots of post-expulsion admixture. Many so-called “Sephardic” samples today have ancestry absorbed from Iberian Jews, Ottoman Turkish communities, Syrian Jews, and North African Jews. So whatever Caucasus-like component they show may not be directly comparable to what's in Ashkenazim — or may be coming in through totally different channels.

In contrast, Ashkenazim seem to carry a clearer and more persistent “Iranic” or Caucasus-adjacent signal — possibly reflecting deeper layers of ancestry that weren’t diluted by later diaspora admixture.

[Gannicus]

One poster in this thread was trolling so hard I had to banish her to the Phantom Zone. ��

She claimed her 6th cousin’s friend’s ex-girlfriend was half Wakandian, half Narnian, and plotted Romanian on PCA — and from that, tried to argue that population structure is basically a lottery, so Ashkenazi genetics can't be valid either. ��

Then she tried to sound serious with this:



It does — if you actually understand how qpAdm works.

Adding a Slavic source like Russian.DG doesn’t automatically lower the Levantine signal. If your outgroups (right pops) aren’t designed to differentiate Slavic from East Med ancestry, then the model will blur those lines — and the Levantine percentage can actually go up, not down.

That’s why I don’t just copy-paste the same reference set blindly. I adjust the outgroups depending on the sources I use — Sicilian.HO vs. South_Italian.HO, Lebanese.HO vs. Lebanese_Muslim.HO, etc.

To illustrate that, here’s what happens when you actually model Ashkenazi Jews with those two Levantine references:

Model 1 – Using Lebanese.HO
Target: Jew_Ashkenazi.HO
p-value: 0.3887 | Feasibility: TRUE

Sicilian.HO — 89.13%
Lebanese.HO — 10.87%

https://i.ibb.co/Y4SzGJ1M/777.png
Full-size PNG (https://ibb.co/gb85CQfp)

Model 2 – Using Lebanese_Muslim.HO
Target: Jew_Ashkenazi.HO
p-value: 0.2283 | Feasibility: TRUE

Sicilian.HO — 83.25%
Lebanese_Muslim.HO — 16.75%

https://i.ibb.co/v4yhMB8X/hhh.png
Full-size PNG (https://ibb.co/chfwsvdY)

Conclusion: The model using Lebanese.HO gives a significantly higher p-score, lower standard errors, and a more realistic Levantine estimate. In other words: it’s a statistically stronger and cleaner model across the board.

The reason these models differ isn’t a “genetic lottery.” It’s because:

• qpAdm and PCA are sensitive to input — SNP count, reference sample composition, and even dataset merging all affect the outcome.
• Outgroups must be chosen carefully — they need to help the model distinguish between sources, not blur them.
• qpAdm doesn’t guess — it tests your model mathematically. If it says TRUE with a high p-value and tight errors, that’s the benchmark — not some fantasy PCA anecdote.

At the end of the day, what matters are p-value (which tells you how plausible your model is) and standard error (which tells you how stable your source estimates are).

So no — I’m not going to toss out valid models just because someone thinks ancestry components shift around like Pokémon evolutions or slot machines. If your framework relies on mythical relatives and cherry-picked exceptions, you’re not doing population genetics — you’re doing improv.

You keep attacking this person and probably others that you have ignored. That's pretty low if you ask me.

[rothaer]

Thanks again for the input—really appreciate it.

Just to clarify about Russian.DG: it's a diverse dataset from the Diverse Genome panel, and it includes individuals with varying levels of East and West Slavic ancestry. It’s not just Siberian-shifted or Finno-Ugric-heavy. A lot of these samples actually cluster very closely with Belarusians, Ukrainians, and even Eastern Poles on PCA, so while not perfect, it does capture a good chunk of the broader Slavic genetic space. That's why I’ve used it as a placeholder for the Slavic/Eastern European component in my qpAdm models so far.

That said, I’m still in the process of refining my Ashkenazi models—I’m not stopping at just Russian.DG. I’m now running a 4-way model, which is already pretty ambitious for qpAdm and takes some tweaking to balance sources and outgroups properly.

Also, I’ve come to think it’s pretty clear that Ashkenazim are better modeled with Sicilians rather than general Southern Italians. Sicilians show a stronger genetic affinity with Ashkenazim, and they help reduce what I think is an inflated Levantine signal that shows up when using Southern Italian samples (which often get overly Lebanese-like). Sicilian samples give a cleaner, more historically realistic fit and help separate genuine Near Eastern ancestry from what's actually deep Mediterranean structure.

Here’s a direct comparison to illustrate what I mean:

Model 1 – Using Sicilian.HO
Target: Jew_Ashkenazi.HO
p-value: 0.2283 | Feasibility: TRUE

Sicilian.HO — 83.25%
Lebanese_Muslim.HO — 16.75%

Full-size link (https://ibb.co/1Yr7D1TB)

Model 2 – Using Italian_South.HO
Target: Jew_Ashkenazi.HO
p-value: 0.1252 | Feasibility: TRUE

Italian_South.HO — 75.12%
Lebanese_Muslim.HO — 24.88%

Full-size link (https://ibb.co/Dgph1zR3)

The Sicilian model doesn't just fit better — it has a significantly stronger p-score of 0.2283 compared to 0.1252 in the Southern Italian model.
More importantly, it assigns only 16.75% to Lebanese_Muslim.HO, versus a much higher 24.88% when using Italian_South.HO.
That nearly 8-point drop in Levantine ancestry shows how Sicilian.HO more effectively absorbs Near Eastern-related signals, without inflating Levantine input unnaturally — which is crucial for accurate Ashkenazi modeling.

And just being honest—I try to stay away from the politics side of all this. I think Polako (Davidski) brings a strong ethnonationalist slant to the discussion. Based on how he interacts, I doubt any model I run would satisfy him regardless of the stats or historical context. There's a lot of negativity and gatekeeping in that space, and I’m not interested in that vibe.

I'm here to learn, run solid models, and share findings with people genuinely interested in population genetics. Always happy to compare notes or collaborate.

Here's my first qpAdm modelling of Ashkenazis. I used 4 components. For Judaean-like Levant I used Druze as I think that they will be best preserved locals from the era in question and hence proxies. Following the Davidski criticism on Russians used in the Erfurt Jew paper I used Polish (for Slavs) and Turkmen (for Khazars). The used out-groups are visible. As you can see there is a small Turkmen signal.

Can you enlighten whether the figures for standard errors are percentages, i. e. relative to the result, or percentage-points? If the former is applicable then the Turkmen signal seems pretty safe.

https://i.imgur.com/5ZefTX1.jpeg

[Feiichy]

I saw plenty of Ashkenazi Jews scoring minor Siberian/mongoloid on gedmatch so small Turkic signal makes sense.

Also I think it's even better (maybe) to use Samaritians for ancient Levantines, afaik they are considered closest living thing to ancient Judeans. But if they are not available Druze is fine too, I assume they are both fairly similar genetically.

[rothaer]

I saw plenty of Ashkenazi Jews scoring minor Siberian/mongoloid on gedmatch so small Turkic signal makes sense.

Also I think it's even better (maybe) to use Samaritans for ancient Levantines, afaik they are considered closest living thing to ancient Judeans. But if they are not available Druze is fine too, I assume they are both fairly similar genetically.

Ofc! I had them in mind and must have had a black-out! ;)

Here we go!

https://i.imgur.com/1571qvf.jpeg

The Turkmen proportion went up half a percentage-point. Maybe the Druze were affected by the Turkish incursions while the Samaritans kept more isolated.

Now the standard errors seem pretty low. Also, the p value is very nice as well (0.9181!).

[Opie]

Ofc! I had them in mind and must have had a black-out! ;)

Here we go!

https://i.imgur.com/1571qvf.jpeg

The Turkmen proportion went up half a percentage-point. Maybe the Druze were affected by the Turkish incursions while the Samaritans kept more isolated.

Now the standard errors seem pretty low. Also, the p value is very nice as well (0.9181!).

Dont use Turkmen. If there is no Khazar sample, use Cuman-Kipchak samples. This is what historically makes sense.

[Opie]

Ofc! I had them in mind and must have had a black-out! ;)

Here we go!

https://i.imgur.com/1571qvf.jpeg

The Turkmen proportion went up half a percentage-point. Maybe the Druze were affected by the Turkish incursions while the Samaritans kept more isolated.

Now the standard errors seem pretty low. Also, the p value is very nice as well (0.9181!).

Btw, it’s better to use North Italian samples because they claimed to yield better statistical fits than Central or Southern Italians.

[Opie]

I saw plenty of Ashkenazi Jews scoring minor Siberian/mongoloid on gedmatch so small Turkic signal makes sense.

Also I think it's even better (maybe) to use Samaritians for ancient Levantines, afaik they are considered closest living thing to ancient Judeans. But if they are not available Druze is fine too, I assume they are both fairly similar genetically.

Probably Khazar connection.

[rothaer]

Btw, it’s better to use North Italian samples because they claimed to yield better statistical fits than Central or Southern Italians.

In what function? As you see I model with Sicilian as discussed in this thread. The suitable outgroup should not be less extreme, hence North Italian as an outgroup would not suit.

[Opie]

In what function? As you see I model with Sicilian as discussed in this thread. The suitable outgroup should not be less extreme, hence North Italian as an outgroup would not suit.

Here is a model with better p value

141983

[rothaer]

Here is a model with better p value

141983

0.2818 is better than 0.9181?!

Also, we have discussed a lot why it's better to model with something Southern Italian (former Magna Graecia, known history of where were most Jewish communities), not with IA Central Italian (Italy Republic).

[Opie]

0.2818 is better than 0.9181?!

Isn’t lower P value better?

[rothaer]

Isn’t lower P value better?

No.

And models are even regularly to be rejected as not plausible if the p value is below 0.05. I just checked.

[Opie]

No.

How so?

[rothaer]

How so?

Simply google qpadm p value.

[Opie]

No.

And models are even regularly to be rejected as not plausible if the p value is below 0.05. I just checked.

I don’t oppose that


In evaluating the qpAdm results, unless otherwise stated, we consider plausible models to have a P-value ≥ 0.05 and admixture weights between zero and one ([0:1]).

I mean a high p value may suggest that specific model is too broad and overfitting.

[Feiichy]

Ofc! I had them in mind and must have had a black-out! ;)

Here we go!

https://i.imgur.com/1571qvf.jpeg

The Turkmen proportion went up half a percentage-point. Maybe the Druze were affected by the Turkish incursions while the Samaritans kept more isolated.

Now the standard errors seem pretty low. Also, the p value is very nice as well (0.9181!).

Excellent model, great work.

[Dušan]

Dont use Turkmen. If there is no Khazar sample, use Cuman-Kipchak samples. This is what historically makes sense.

Or maybe Crimean Tatars.

[Dušan]

Distance below 1 on G25

https://i.imgur.com/nZKlGlv.png



Askenazi_Jew,0.09908,0.136553,-0.009933,-0.043669,0.01004,-0.015689,-0.001939,-0.00147,0.007359,0.015071,0.002384,-0.000192,0.002276,-0.000197,-0.004173,0.000478,-0.002117,-0.000502,-0.000567,-0.00397,-0.003064,-0.002667,0.001215,0.002792,0.000525

Samaritan,0.085367,0.148674,-0.057398,-0.092508,-0.009602,-0.035029,-0.002209,-0.008169,0.016321,0.008237,0.009159,-0.009921,0.020485,0.011863,-0.005537,0.002121,-0.011239,0.001014,0.002112,-0.003652,-0.000524,0.003116,0.000567,-0.00294,0.004646
Italian_Calabria,0.103437,0.14814,-0.010324,-0.046997,0.012618,-0.015025,-0.002438,-0.001586,0.004448,0.021231,0.002213,0.002997,-0.004887,-0.001737,-0.006549,-0.002784,0.005802,0.000728,0.002938,-0.005362,-0.002605,-0.000325,0.000339,1.5e-05,-0.002425
Tatar_Crimean_steppe,0.094473,-0.0556,0.038089,0.00751,-0.008309,0.002022,0.006874,0.010269,-0.006238,-0.004875,-0.011489,-0.002061,0.002825,0.006571,-0.004106,0.005834,0.01017,-0.003706,0.001446,0.004127,-0.00627,0.000433,-0.002064,0.002259,0.003443
Polish,0.13184,0.12927,0.069868,0.057738,0.040676, 0.021713,0.008678,0.010863,-0.000933,-0.018553,-0.004349,-0.006419,0.013093,0.018629,-0.007024,-0.00056,0.001546,-7.4e-05,0.002695,0.001287,-0.003129,-0.003119,0.005606,-0.003333,-8.8e-05

[Feiichy]

Distance below 1 on G25

https://i.imgur.com/nZKlGlv.png



Askenazi_Jew,0.09908,0.136553,-0.009933,-0.043669,0.01004,-0.015689,-0.001939,-0.00147,0.007359,0.015071,0.002384,-0.000192,0.002276,-0.000197,-0.004173,0.000478,-0.002117,-0.000502,-0.000567,-0.00397,-0.003064,-0.002667,0.001215,0.002792,0.000525

Samaritan,0.085367,0.148674,-0.057398,-0.092508,-0.009602,-0.035029,-0.002209,-0.008169,0.016321,0.008237,0.009159,-0.009921,0.020485,0.011863,-0.005537,0.002121,-0.011239,0.001014,0.002112,-0.003652,-0.000524,0.003116,0.000567,-0.00294,0.004646
Italian_Calabria,0.103437,0.14814,-0.010324,-0.046997,0.012618,-0.015025,-0.002438,-0.001586,0.004448,0.021231,0.002213,0.002997,-0.004887,-0.001737,-0.006549,-0.002784,0.005802,0.000728,0.002938,-0.005362,-0.002605,-0.000325,0.000339,1.5e-05,-0.002425
Tatar_Crimean_steppe,0.094473,-0.0556,0.038089,0.00751,-0.008309,0.002022,0.006874,0.010269,-0.006238,-0.004875,-0.011489,-0.002061,0.002825,0.006571,-0.004106,0.005834,0.01017,-0.003706,0.001446,0.004127,-0.00627,0.000433,-0.002064,0.002259,0.003443
Polish,0.13184,0.12927,0.069868,0.057738,0.040676, 0.021713,0.008678,0.010863,-0.000933,-0.018553,-0.004349,-0.006419,0.013093,0.018629,-0.007024,-0.00056,0.001546,-7.4e-05,0.002695,0.001287,-0.003129,-0.003119,0.005606,-0.003333,-8.8e-05


Great and pretty realistic. Crimean Tatars are part Anatolian tho, so real Turkic will be lower.

[majevica]

Great and pretty realistic. Crimean Tatars are part Anatolian tho, so real Turkic will be lower.

And Calabrians are heavily Levanto-Anatolian, so the real South European (like Old Italic) will be lower too

https://www.reddit.com/r/illustrativeDNA/comments/1ictt7z/south_italy_calabria_updated_periodical/

[celticdragongod]

Distance below 1 on G25

https://i.imgur.com/nZKlGlv.png



Askenazi_Jew,0.09908,0.136553,-0.009933,-0.043669,0.01004,-0.015689,-0.001939,-0.00147,0.007359,0.015071,0.002384,-0.000192,0.002276,-0.000197,-0.004173,0.000478,-0.002117,-0.000502,-0.000567,-0.00397,-0.003064,-0.002667,0.001215,0.002792,0.000525

Samaritan,0.085367,0.148674,-0.057398,-0.092508,-0.009602,-0.035029,-0.002209,-0.008169,0.016321,0.008237,0.009159,-0.009921,0.020485,0.011863,-0.005537,0.002121,-0.011239,0.001014,0.002112,-0.003652,-0.000524,0.003116,0.000567,-0.00294,0.004646
Italian_Calabria,0.103437,0.14814,-0.010324,-0.046997,0.012618,-0.015025,-0.002438,-0.001586,0.004448,0.021231,0.002213,0.002997,-0.004887,-0.001737,-0.006549,-0.002784,0.005802,0.000728,0.002938,-0.005362,-0.002605,-0.000325,0.000339,1.5e-05,-0.002425
Tatar_Crimean_steppe,0.094473,-0.0556,0.038089,0.00751,-0.008309,0.002022,0.006874,0.010269,-0.006238,-0.004875,-0.011489,-0.002061,0.002825,0.006571,-0.004106,0.005834,0.01017,-0.003706,0.001446,0.004127,-0.00627,0.000433,-0.002064,0.002259,0.003443
Polish,0.13184,0.12927,0.069868,0.057738,0.040676, 0.021713,0.008678,0.010863,-0.000933,-0.018553,-0.004349,-0.006419,0.013093,0.018629,-0.007024,-0.00056,0.001546,-7.4e-05,0.002695,0.001287,-0.003129,-0.003119,0.005606,-0.003333,-8.8e-05


Distance to: CDG_scaled
0.06193090 Polish
0.13828453 Askenazi_Jew
0.14119774 Italian_Calabria
0.21270099 Tatar_Crimean_steppe
0.21486603 Samaritan

Target: CDG_scaled
Distance: 5.6942% / 0.05694188
84.2 Polish
15.8 Italian_Calabria

[Opie]

Great and pretty realistic. Crimean Tatars are part Anatolian tho, so real Turkic will be lower.

Yes and they have Slavic ancestry too, so it probably absorbed some of the Askhenazi Slavic too.

[Lard]

🧬 The Ashkenazi Redpill No One Wants to Swallow:

📍 PCA shows Ashkenazi Jews directly overlapping with Southern Italians
❌ Not French
❌ Not Northern Italians
❌ Not Lebanese
✅ Southern Italians — consistently

🔬 qpAdm models reflecting PCA reality =
✔️ ~65–75% Southern Italian–like
✔️ ~10% Eastern European (Slavic)
✔️ ~15–25% Levantine
✔️ Stable p-values
✔️ Matches Roman + Byzantine Jewish history

😂 But what does the mainstream give us instead?

→ “Let’s model Ashkenazim as 50% North Italian + 50% Lebanese”
→ “Southern Italian models are unrealistic”

🤡 So now Zionists are pushing 2-way models made entirely of gentiles to defend Jewish continuity?
Make it make sense.

🔥 There’s no other population on Earth where:

• They overlap a totally different ethnic group on PCA
• There’s historical evidence of admixture with that group
• But scientists insist it’s all a coincidence

This only happens with Ashkenazi Jews — because the real ancestry threatens both anti-Semitic and Zionist narratives.

[Opie]

Best modern and ancient models to date with highest P values and lowest standard errors. I believe this is an excellent qpAdm model to represent core Ashkenazi Jewish ancestry. References used:
Mbuti.DG
Han.DG
Papuan.DG
Chukchi.DG
Mixe.DG
Dai.DG
Brahmin_UP.DG
Yakut.DG
BedouinB.DG
French.DG
Sardinian.DG

142114
142115

[rothaer]

(...) This only happens with Ashkenazi Jews — because the real ancestry threatens both anti-Semitic and Zionist narratives.

In fact it even challenges the Judaistic narrative "We are one people ever since". It made me realise that Jewish scholars, even if they may not be religious anymore, will likely be biased when it comes to the interpretation of genetic results, hence it's not the very best idea to completely leave it up to them.

[Peterski]

Here are closest distances to Ashkenazi-Germany based on my Eurogenes K36 averages:

(maybe later I will check how they plot in a PCA based on my Eurogenes K36 populations)

Distance to: Ashkenazi-Germany
2.80601140 Ashkenazi-France
3.52404597 Ashkenazi-Czechoslovakia
3.71717635 Sephardic-Turkey
3.74060156 Ashkenazi-Mixed
4.01736232 Sephardic-Bulgaria
4.14967469 Ashkenazi-Hungary
4.43004515 Sephardic-Greece
4.55887047 Ashkenazi-General
4.64877403 ITA_Sicily-Caltanisetta
4.66842586 Ashkenazi-Eastern
4.72890051 Ashkenazi-Prussia
4.79046971 Ashkenazi-Poland
4.82082980 Italkim-Jews
5.01228491 Ashkenazi-Galitzianer
5.03147096 Ashkenazi-Ukraine
5.03501738 Ashkenazi-Romania
5.08785810 Ashkenazi-Litvaks
5.09873514 ITA_Sicily-Messina
5.33275726 ITA_Sicily-Agrigento
5.33504452 Ashkenazi-Austria
5.34861664 ITA_Sicily-Palermo
5.40533070 Ashkenazi-Latvia
5.48723063 Ashkenazi-Netherlands
5.48723063 Ashkenazi-West
5.49804511 MT_Malta:GEDmatch
5.59576626 Ashkenazi-Average:N471
5.60573813 Ashkenazi-Lithuania
5.77743888 Ashkenazi-Moldova
5.83788489 Ashkenazi-Belarus
6.05588144 MT_Malta:academic
6.17543521 Sephardic-Algeria
6.32391493 ITA_Sicily-Trapani
6.35358167 Ashkenazi-Russia
6.51179699 ITA_Calabria-Catanzaro
6.53372788 ITA_Sicily-Catania
6.54451679 ITA_Sicily-Ragusa
6.67119929 Romaniote-Jews
6.71707526 ITA_Basilicata
6.76530118 Sephardic-Morocco
6.81566578 ITA_Calabria-Cosenza
6.85229888 ITA_Campania-Salerno
7.64321268 GR_Crete
7.65800235 Sephardic-General
7.76007088 ITA_Campania-Naples
7.79676215 GR_Dodecanese
8.23085658 ITA_Campania-Caserta
8.32407953 GR_Kalymnos
8.58269189 ITA_Apulia-Foggia
8.59382336 Sephardic-Maghreb
8.87138095 GR_Ikaria
(...)

As can be seen, Sephardic Jews are close to Ashkenazi Jews in terms of admixture proportions.

The main difference is that Sephardic Jews did not experience a bottleneck like Ashkenazi Jews.

Italkim and Romaniote Jews are also similar in terms of Eurogenes K36 admixture proportions.

[Lard]

Here are closest distances to Ashkenazi-Germany based on my Eurogenes K36 averages:

(maybe later I will check how they plot in a PCA based on my Eurogenes K36 populations)

Distance to: Ashkenazi-Germany
2.80601140 Ashkenazi-France
3.52404597 Ashkenazi-Czechoslovakia
3.71717635 Sephardic-Turkey
3.74060156 Ashkenazi-Mixed
4.01736232 Sephardic-Bulgaria
4.14967469 Ashkenazi-Hungary
4.43004515 Sephardic-Greece
4.55887047 Ashkenazi-General
4.64877403 ITA_Sicily-Caltanisetta
4.66842586 Ashkenazi-Eastern
4.72890051 Ashkenazi-Prussia
4.79046971 Ashkenazi-Poland
4.82082980 Italkim-Jews
5.01228491 Ashkenazi-Galitzianer
5.03147096 Ashkenazi-Ukraine
5.03501738 Ashkenazi-Romania
5.08785810 Ashkenazi-Litvaks
5.09873514 ITA_Sicily-Messina
5.33275726 ITA_Sicily-Agrigento
5.33504452 Ashkenazi-Austria
5.34861664 ITA_Sicily-Palermo
5.40533070 Ashkenazi-Latvia
5.48723063 Ashkenazi-Netherlands
5.48723063 Ashkenazi-West
5.49804511 MT_Malta
(...)

As can be seen, Sephardic Jews are close to Ashkenazi Jews in terms of admixture proportions.

The main difference is that Sephardic Jews did not experience a bottleneck like Ashkenazi Jews.

Thanks for posting that, Peterski. I’ll be making an account at the forum link you shared (https://genarchivist.net/) soon. I already have MyBB installed locally since I’m planning to launch a forum focused on serious population genetics discussions.

As for the K36 distances, they’re pretty telling. Some of the Southern Italian samples like ITA_Sicily-Caltanisetta, ITA_Sicily-Messina, ITA_Sicily-Agrigento, and ITA_Sicily-Palermo are actually closer to Ashkenazi-Germany than certain Ashkenazi subgroups are to each other. Caltanisetta, for example, is closer to Ashkenazi-Germany than Ashkenazi-Poland or Ashkenazi-Litvaks are. That fits perfectly with what I’ve seen across PCA and formal modeling: Ashkenazim overwhelmingly plot with Southern Italians and Sicilians, not Northern Italians or modern-day Levantines.

The idea that Ashkenazim are “half Northern Italian and half Lebanese” is just narrative padding. The actual breakdown that fits both the PCA and history is something more like 65 to 75 percent Southern Italian, 10 to 15 percent Slavic, and 15 to 25 percent Levantine or Caucasus-related ancestry. That’s consistent with where they plot, and with what we know about Jewish communities in Southern Italy during Late Antiquity.

Sephardic Jews, as your distances show, are also close — but unlike Ashkenazim, they don’t carry that Eastern European admixture pulse. That ~10–15 percent Slavic-like layer is a distinct Ashkenazi feature, and it shows up clearly in PCA, ADMIXTURE, and qpAdm models.

And just to be upfront, I’ve already been inundated with people saying stuff like “I have a friend with random admixture who plots with Romanians on PCA” — and they use that to explain away Ashkenazi clustering. It’s ridiculous. No other population on Earth plots tightly with another group on PCA unless they have both major shared ancestry and historical context to explain it. Ashkenazim plotting directly with Southern Italians is not some fluke. I’m not interested in dealing with that kind of nonsense again.

I’m also not interested in the Y-DNA cope trying to "prove" Levantine origin. That entire line of reasoning misses the point. Ashkenazi Y-DNA is severely skewed because of bottlenecks and founder effects. The dominant J1, J2, and E clades don’t prove a Levantine origin — they prove drift. Many of the same subclades, like J1-P58, E-M34, and major J2a lines, are also found in Southern Europeans — the same populations Ashkenazim plot with on PCA. Y-DNA doesn't override autosomal reality, and the autosomal data is clear.

So yeah, I’ll be browsing GenArchivist to get a sense of the discussion quality. If it's solid and grounded in real data interpretation, I’m all in. But if it’s more of the same outdated narratives and circular reasoning, I’ll probably just stick to building my own space.

[Lard]

In fact it even challenges the Judaistic narrative "We are one people ever since". It made me realise that Jewish scholars, even if they may not be religious anymore, will likely be biased when it comes to the interpretation of genetic results, hence it's not the very best idea to completely leave it to them.

You're absolutely right, rothaer. The truth is hidden in plain sight. What we are dealing with is not just flawed interpretation but intentional narrative control. Population genetics, when it comes to Jews, has been co-opted by figures like Harry Ostrer and turned into pseudoscientific spin to support a fragile idea of a unified Jewish people "returning" to their ancient homeland.

Ostrer and his circle act more like ideological operators in lab coats than neutral scientists. Their job is to selectively curate reference populations that inflate Levantine ancestry in Ashkenazim while completely erasing Southern Italy, which consistently shows up as a key ancestral source in independent PCA and qpAdm analyses.

And the irony is that even with all their cherry-picking and model-rigging, the best they can come up with is a "half Northern Italian, half Lebanese" model. That still makes Ashkenazim fully gentile in terms of ancestry. So even their preferred Zionist framing collapses under its own logic.

When you run the data without narrative bias, Ashkenazim consistently show a strong Southern Italian core, some Eastern European admixture, and a smaller Levantine or Caucasus-related layer. That fits both the genetics and the historical record. It is not a mystery. It is just inconvenient for those who need the genetics to prop up a political ideology.

[rothaer]

(...) Y-DNA cope trying to "prove" Levantine origin. That entire line of reasoning misses the point. Ashkenazi Y-DNA is severely skewed because of bottlenecks and founder effects.

OMG I've been emphasising this to Y DNA enthusiasts sooo many times that in the context of bottlenecks, founder effects and genetic drifts the Y DNA proportions become close to irrelevant for enlightening the ancestry proportions.

[rothaer]

And the irony is that even with all their cherry-picking and model-rigging, the best they can come up with is a "half Northern Italian, half Lebanese" model. That still makes Ashkenazim fully gentile in terms of ancestry. So even their preferred Zionist framing collapses under its own logic.


Lol, and indeed Levantine or its proxy by far does not yet equal Judaean. There were loads of Levantine non-Jewish folks from Lebanon and Syria. Also, Jewish proselytes from Anatolia will in modelling rather evoke Levantine than Southern Italian. And there were plenty of them (in Ephesos etc.).

[Lard]

OMG I've been emphasising this to Y DNA enthusiasts sooo many times that in the context of bottlenecks, founder effects and genetic drifts the Y DNA proportions become close to irrelevant for enlightening the ancestry proportions.

Exactly, rothaer. This whole obsession with Ashkenazi Y-DNA is just misplaced, especially when it's being spun by political ideologues like Harry Ostrer. What they’re doing is less genetics and more Zionist fan fiction dressed up in lab coats. They treat Y-DNA as if it reflects overall ancestry, completely ignoring what genetic drift, founder effects, and bottlenecks actually do to small, isolated populations.

A perfect real-world example of this is the Arbëreshë of Southern Italy and Sicily. They arrived from Albania in the 15th century and settled across mainland Southern Italy and Sicily. Like Ashkenazim, they were a small founding population that remained endogamous and preserved their distinct language and culture for centuries while living among local Southern Italians. Yet despite descending from Tosk Albanians, who have around 2% or less E-M123, some Arbëreshë communities today show over 20% E-M123 — even higher than what is found in Ashkenazim.

This is a textbook case of genetic drift. A rare Y-DNA lineage in the source population can become massively overrepresented in a bottlenecked descendant group. That’s exactly what we see in Ashkenazim — and in Arbëreshë.

But here's the key: autosomal DNA is far less susceptible to this kind of distortion. Despite their skewed Y-DNA, the Arbëreshë still cluster with Tosks on PCA and other autosomal analyses. That’s because autosomal DNA reflects the full ancestry picture, not just a single male lineage. So when people point to E-M34, J1, or J2a in Ashkenazim as “proof” of Levantine ancestry, they’re either misunderstanding the science or deliberately misrepresenting it.

In short, Y-DNA tells you about lineages, not ancestry proportions. Autosomal data — PCA, ADMIXTURE, qpAdm — consistently shows Ashkenazim clustering tightly with Southern Italians and Sicilians, not with modern Levantines. That’s the ancestry signal that actually matters.

[Lard]

Lol, and indeed Levantine or its proxy by far does not yet equal Judaean. There were loads of Levantine non-Jewish folks from Lebanon and Syria. Also, Jewish proselytes from Anatolia will in modelling rather evoke Levantine than Southern Italian. And there were plenty of them (in Ephesos etc.).
When even the most committed Zionist propagandists like Dr. Harry Ostrer can only model Ashkenazim using two fully gentile populations — Northern Italians and Lebanese — while deliberately ignoring the one gentile population Ashkenazim actually overlap with on PCA and in historical records, namely Southern Italians, the agenda becomes obvious.

None of this is real science. It’s politicized propaganda, designed to generate a quick, misleading headline in the New York Times or on Zionist platforms like The Forward. Sure, that narrative rules Reddit, Quora, and YouTube, but that’s exactly why I avoid those cesspools of disinformation — unless I’m jumping in to point out lies. Like with Andrei’s DNA video, where he quietly used the same Ashkenazi model I did, after spending so much time posturing and pushing contradictions.

[Lard]

Everyone reading this — if you want to see a masterclass in disinformation and cherry-picking data to serve a Zionist narrative, watch this TED Talk by Nathan Pearson. His European comparison dataset for Ashkenazi Jews is, wait for it… French, Hungarians, and Russians. He deliberately excluded Southern Italians — the population Ashkenazim consistently cluster nearest to on PCA plots and in historical context. Instead, he turned Italy into a complete dead zone in his model to steer the narrative.

Even the charlatans like Harry Ostrer at least include Italian samples — albeit from every region except the one Ashkenazim actually show the strongest genetic affinity with. That omission isn’t science — it’s agenda-driven storytelling.


https://www.youtube.com/watch?v=-dEL2yhT7Uo

[Etelfrido]

The modelling with Southern Italians instead of Northern Italians that dramatically reduces the (direct) Levant ancestry proportion is not just some fringe imagination that is unbacked in scientific papers. The first time I saw this modelling - long after I claimed in different forums that that for obvious reasons had to be done - was in the Erfurt Jew paper from 2022. Expectedly it shows a notable drop in the (direct) Levant ancestry. It states:

"We used qpAdm to test quantitative models for the ancestral sources of EAJ (STAR Methods). Based on the PCA above and previous modeling (Xue et al., 2017), we considered a model where EAJ is a mixture of the following sources: Southern European (South Italians or North Italians), Middle Eastern (Druze, Egyptians, Bedouins, Palestinians, Lebanese, Jordanians, Syrians, or Saudis), and Eastern European (Russians). We used modern populations as sources, as modeling with ancient sources was unsuccessful (Data S1, section 7). Multiple models with South-Italians were plausible (p>0.05; Table S3), which would be consistent with historical models pointing to the Italian peninsula as the source for the AJ population (Data S1, section 16; though see below for alternatives and caveats). The mean admixture proportions (over all of our plausible models; Table S3) were 65% South Italy, 19% ME, and 16% East-EU (Figure 3A). We validated that our results did not qualitatively change when using only transversions vs. all SNPs, a different outgroup population, or fewer SNPs (Table S3; Data S1, section 7). Estimates of the admixture time were unreliable in our setting (Data S1, section 8).

Models with other sources, in particular Mediterranean, could also fit the EAJ data (Data S1, section 7). A model with North Italians as a source (Table S3) had ancestry proportions 37% North Italy, 43% ME, and 20% East-EU (Figure 3A). Models with Greek as a source had average ancestry proportions 51% Greek, 32% ME, and 17% East-EU (Table S3). Models with Spanish or North African sources (in addition to ME and East-EU) were not plausible. A model with all Levant populations merged together as the ME source fit the EAJ data (p = 0.07), with ancestry proportions 65% South Italy, 19% Levant, and 16% East-EU. A model with all Mediterranean populations merged as a single source (Middle Eastern, Greek, and Italian, with East-EU as the other source) fit the data (p = 0.11) with ancestry proportions 89% Mediterranean and 11% East-EU. Models with a Western European source (Germans) instead of Russians were not plausible. There was also no support for a minor ancestry component from East Asians."

https://www.sciencedirect.com/science/article/pii/S0092867422013782

"Models with a Western European source (Germans) instead of Russians were not plausible."

A more exact model preferably should include both, but from my understanding that implies Ashkenazim mixed more with Slavs than with Germans. Is there any historical explanation for that?

[Lard]

"Models with a Western European source (Germans) instead of Russians were not plausible."

A more exact model preferably should include both, but from my understanding that implies Ashkenazim mixed more with Slavs than with Germans. Is there any historical explanation for that?
At their height before WWII, Jews made up less than 1% of the total German population (about half a million out of about 67 million). By contrast, in Poland there were about 3.5 million Ashkenazim, making up around 10% of the entire Polish population, and in Ukraine there were roughly 2.7 to 3 million Jews, about 10% of the total population. The demographic core of Ashkenazi Jewry was clearly in the Slavic lands, not in Germany.

Ashkenazi ethnogenesis drew on several sources.
They absorbed Knaanic Jews who were already admixed with Western Slavs, and layered onto a Jewish base that picked up substantial Mediterranean input, especially southern Italian and other Mediterranean ancestry. Waldman et al. (2022, Erfurt sample) modeled roughly two thirds of their ancestry as southern Italian like, with a minor Near Eastern substratum and later Slavic admixture.

A Germanic signal is present, but because of strong founder effects and bottlenecks haplogroup analysis cannot cleanly resolve it.
In qpAdm modeling it is also difficult to separate Germanic from Slavic input, since many Germans, particularly in the east, share strong genetic affinity with Western Slavic groups such as Poles, Czechs, Slovaks, and Sorbs.

The expansion that defined Ashkenazi Jewry happened above all in the Slavic lands of central and eastern Europe, where their population grew into the millions. Modeling them with Germans alone misses this reality.

[Opie]

Missclicked

[Opie]

Here's a proposed model for Askhazi Jewish ancestry

142720

[Peterski]

Here is a different opinion than OP:


https://www.youtube.com/watch?v=vhJzWpdKi-Q

[Karol Klačansky]

the problem with your whole thesis is that southern italians themselves are about half eastern med levantine genetically and half northern like italian in ancestry. So Jews plotting with them being half levantine is no surprise. Y-dna analysis also doesn’t agree with your analysis that jews are just southern italians basically. Their language and history etc is there for a reason, it didn’t come from no where. Northern african jews and others Jewish populations also genetically show links to ashkenazi and on a pca plot run along the same sort of admixture line between europe and the levant. I don’t think ur thread in any way contradicts what we already knew about the Jews. Also at the time of the destruction of the tempe, Roman italy shifted genetically almost entirely east med towards the levant. When Jews left to italy if anything it was italians at the time who essentially became anotolian/levantine greek