All Iberian men were wiped out by Yamna men 4,500 years ago

[Olympus Mons]

To explain why Iberian (and indeed other) men might have been suddenly wiped out by a population that previously appears to have been very limited indeed, it makes sense to question what led to the abrupt change. The answer might differ, depending on whether the population concerned was a self-contained elite or struggling to survive. If it was an elite, what stopped it from expanding and branching earlier? If it was a struggler, what changed to to allow it to take off so spectacularly?

Here is an option - During Chalcolithic (3300bc - 1900bc) the region of south Portugal was thriving with a Oak tree forests and huge settlements held by agriculture, high pastoral life and game hunting. By bronze age all had disappear (as did places like Porto Torrão and Perdigoes). by end Iron age, when strabo passed there he stated that the Celts had entered portugal through that region (alentejo) and that it was a barren region where agriculture was extremely difficult. So climate change must have been a factor.

 

[Sile]

Hey!!! I go Crazy with anything.
Actually I go Crazy if they try to link Shulaveri-Shomu to anything Kura-Araxes. People have as big problem with time as they seem to have with space. there is a Gap of 1000 years between both cultures. And they share no traits at all.
I think its not at all possible that Haak or whomever is linking Kura-araxes to PIE or R1b... Nobody is that stupid. Or perhaps the madness of "pretend not to see" in Academia is even greater that what I could phantom. "PnS" is a particular lie close to politically very left leaning people.
To be fair, what they at Max Planck must be saying is that the Hittite is related to Kura-Araxes. Which would be wrong, but not really crazy.
Now... the Shulaveri, oh boy, oh boy, its coming!

Hittite texts state their origins are north of where they where ...............some modern historians place them in modern western georgia

 

[Sile]

@ Moesan.

"Linguistics can be very uncertain sometimes, when speaking of evolution speed. Nothing of what we say is very solidly based."

From Wiki : "Proto-Celtic is mostly dated to roughly 800 BC (Hallstatt C). In the first decade of the 21st century a number of scholars addressed this question using computational methods, with differing results. Gray and Atkinson estimated a date of 6100 BP (4100 BC) while Forster and Toth suggest a date of 3200 BC ±1500 years for the arrival of Celtic in Britain, but such early dates are not generally accepted."

--> Margin of uncertainty between the most extreme estimates : 3300 years !!!

Glauberg near frankfurt was the celtic "capital" as central and south germany are celtic lands.........halstatt celts came there via modern bavaria which was celtic lands.......
they are still digging in glauberg 30 years later

 

[Sile]

Well, apart from Venetic and Liburnian, which do not fit well into either Celtic or Italic, and the controversial cases of Tartessian and North Picene (which some linguists have tried to establish as extremely divergent IE languages, and not non-IE languages), there is Germanic as a pretty diverged IE branch (especially in terms of syntax, with a lot of innovations) in comparison with Celtic and Italic and certainly dating to much before the Urnfield expansion. I find it extremely unlikely that, if the earliest Celtic and Italic are both dated to around 1300-1200 BC (that is, early Urnfield times), their common links with the much more divergent Pre-Proto-Germanic would've taken place just a few centuries earlier. I'd say therefore that the Germanic-Italic-Celtic-Venetic/Liburnian common ancestor might date to the late BB times. The heavy non-IE substrate in Germanic may be very early, from the times where Central European BB was spreading in Europe or even when CWC was absorbing other cultures in Northern Europe.

One should not expect a divergence among Northwestern IE languages as large as that between Italic and Indo-Iranian, for instance, because of areal features when the early proto-languages were still closer to each other and also because they certainly diverged much later than the "eastern" IE branches from Northwestern IE (Italo-Celtic and Germanic are usually assumed to have been among the first IE branches to split off from the rest).

Unfortunately I don't think we have much if any reliable and well received data about the linguistic landscape of most of Europe until the Iron Age, and all the IE languages we have written evidence of before the Latin and Germanic expansions were spoken in Southern Europe, mainly around Italy. The supposed non-IE presence in Insular Celtic is pretty speculative and as far as I have researched not well received at all by most linguists. The non-IE languages we know of were all found in Southern Europe where Central European BB apparently left a smaller genetic imprint in many areas and met more populous societies. What is certain is that much like EBA steppe-like ancestry was spread to Northern Europe a bit earlier, that same ancestral admixture was spread to Central and Western Europe later. I doubt that genetic change had nothing to do with the spread of Indo-European languages, even if part of them may have adopted a local non-IE (EEF-derived) language as theirs and thus helped spread it in Europe, too, more or less like Turkified Scythians may have helped bring more steppe ancestry to some parts of Asia in the Middle Ages.

in the recent picene paper IIRC, they state north-picene was basically a colony of liburnians and south -picene where umbrians
venetic belonged to ancient indigenous Euganei language who belong to the danubian culture

 

[Saetrus]

Kura-Araxes works perfectly fine for me, very much in line with what I've said Indo-Europeans were both in terms of autosomal and Y-DNA.

K-A Y-DNA available so far:

Kura-Araxes ARM002 Y-HG G2b
Odd haplogroup, but then we see subclades connecting Italy to Pakistan at different ends of the Indo-European world:

Lots of Indo-European subclades behave like that.

Kura-Araxes I1635 Y-HG R1b-M415
R1b older than M269 showing the deep roots of M269 in that region.

Kura-Araxes VEK007 Y-HG J1
We have a Bactrian aristocratic sample who is J1 as we also have a Thracian aristocratic sample who is J2.

 

[ToBeOrNotToBe]

Kura-Araxes works perfectly fine for me, very much in line with what I've said Indo-Europeans were both in terms of autosomal and Y-DNA.

K-A Y-DNA available so far:

Kura-Araxes ARM002 Y-HG G2b
Odd haplogroup, but then we see subclades connecting Italy to Pakistan at different ends of the Indo-European world:

Lots of Indo-European subclades behave like that.

Kura-Araxes I1635 Y-HG R1b-M415
R1b older than M269 showing the deep roots of M269 in that region.

Kura-Araxes VEK007 Y-HG J1
We have a Bactrian aristocratic sample who is J1 as we also have a Thracian aristocratic sample who is J2.

It makes no sense in terms of the spread of IE though. IE has to be from the Steppe at the very least after Anatolian split off. KA can only exist as the homeland in a modified Kurgan theory.

I do see L23 as having originated from that region though. And that KA R1b basically means nothing.

 

[markod]

Well, apart from Venetic and Liburnian, which do not fit well into either Celtic or Italic, and the controversial cases of Tartessian and North Picene (which some linguists have tried to establish as extremely divergent IE languages, and not non-IE languages), there is Germanic as a pretty diverged IE branch (especially in terms of syntax, with a lot of innovations) in comparison with Celtic and Italic and certainly dating to much before the Urnfield expansion. I find it extremely unlikely that, if the earliest Celtic and Italic are both dated to around 1300-1200 BC (that is, early Urnfield times), their common links with the much more divergent Pre-Proto-Germanic would've taken place just a few centuries earlier. I'd say therefore that the Germanic-Italic-Celtic-Venetic/Liburnian common ancestor might date to the late BB times. The heavy non-IE substrate in Germanic may be very early, from the times where Central European BB was spreading in Europe or even when CWC was absorbing other cultures in Northern Europe.

One should not expect a divergence among Northwestern IE languages as large as that between Italic and Indo-Iranian, for instance, because of areal features when the early proto-languages were still closer to each other and also because they certainly diverged much later than the "eastern" IE branches from Northwestern IE (Italo-Celtic and Germanic are usually assumed to have been among the first IE branches to split off from the rest).

Unfortunately I don't think we have much if any reliable and well received data about the linguistic landscape of most of Europe until the Iron Age, and all the IE languages we have written evidence of before the Latin and Germanic expansions were spoken in Southern Europe, mainly around Italy. The supposed non-IE presence in Insular Celtic is pretty speculative and as far as I have researched not well received at all by most linguists. The non-IE languages we know of were all found in Southern Europe where Central European BB apparently left a smaller genetic imprint in many areas and met more populous societies. What is certain is that much like EBA steppe-like ancestry was spread to Northern Europe a bit earlier, that same ancestral admixture was spread to Central and Western Europe later. I doubt that genetic change had nothing to do with the spread of Indo-European languages, even if part of them may have adopted a local non-IE (EEF-derived) language as theirs and thus helped spread it in Europe, too, more or less like Turkified Scythians may have helped bring more steppe ancestry to some parts of Asia in the Middle Ages.

Neither of those examples are deeply diverged, they all fit into the North-Western Indo-European cluster. Steppe Bell Beakers appear in Iberia, Britain & Ireland around 2500-2300 B.C. . In the Chang et al. Bronze Age model Indo-Iranian seperates from NWIE-BS around 2600 B.C., Balto-Slavic separates from NWIE around 2200 B.C. . There's no IE language with such time depth in Western Europe. The BB = NWIE model would likely require a Neolithic timeframe for the breakup of PIE, which means both steppe and Chalcolithic Armenia are out.

 

[Ygorcs]

Neither of those examples are deeply diverged, they all fit into the North-Western Indo-European cluster. Steppe Bell Beakers appear in Iberia, Britain & Ireland around 2500-2300 B.C. . In the Chang et al. Bronze Age model Indo-Iranian seperates from NWIE-BS around 2600 B.C., Balto-Slavic separates from NWIE around 2200 B.C. . There's no IE language with such time depth in Western Europe. The BB = NWIE model would likely require a Neolithic timeframe for the breakup of PIE, which means both steppe and Chalcolithic Armenia are out.

Can you elaborate a bit on why you think so? The model presented by Chang et al. has estimates a bit on the lower end of the time spans I have seen for IE splits compared to other sources I've had access to. IIRC some of them assumed Germanic and Italo-Celtic for instance to have diverged from other branches even before 3000 BC, so in Early Yamna times. Probably a middle ground is closer to the truth. But in any case I did not get why you think a BB = NWIE would require the PIE breakup to be as early as the Neolithic (mind you, I do not believe undivided PIE was spoken in Yamnaya at all, I think that, if Anatolian and even Tocharian are included, the split started even before the Yamnaya expansion, and by the time of CWC it was already split into very close but distinct sister languages, with Proto-Anatolian probably already well developed).

As for NWIE, if Proto-Celtic and Proto-Italic started to break up roughly in 1200-1100 BC, we can assume that the languages had started to be spoken, breaking from their common ancestor, at least ~600-800 years before, so roughly by 2000-1800 BC. And if you add Germanic - if Germanic does indeed derive from a common ancestor with Italo-Celtic, which is not totally certain AFAIK - that date must go back several centuries further, so roughly to 3000-2500 BC (on the lower end coinciding with the begining of BBC in Central Europe). And of course we do not need to presume that NWIE started to be spoken only when the Central European BB phenomenon got formed. The language was brought with the newcomers from elsewhere and it didn't start from scratch straight from Undivided PIE. The timing in my opinion is totally compatible with a NWIE dialect (not necessarily the only one that existed, but the one that was successful in the long term) being spread by Rhine BB people to many areas west of Germany (though not necessarily all, as we all know the BBC had a very sparse occupation of territory, and they may have possibly adopted the local languages in some of the places they settled in, as in the hypothetical case of the Basques).

As for no IE language with such time depth existing in Western Europe, the vast majority of it had no written language virtually until the Common Era. The areas that had written inscriptions from earlier times had writing only during the later Iron Age (well after Celtic, Italic and Germanic expansions) and were some of those most impacted by Celtic and Italic conquests centuries earlier (Italy, France, Iberia). That situation is not very conducive to allow us to see remants of earlier NWIE languages (though Germanic, I insist, is definitely one such example, it's definitely divergent and innovative in comparison to Italo-Celtic). It'd be like judging the linguistic landscape of IE in continental Western Europe in the Middle Ages after daughter languages of Latin and Proto-Germanic dominated the region almost entirely.

I also have to doubt that had the all-important BBC been associated with another language family, we wouldn't see the non-IE remnants of that coherent and homogeneous language family spread to many parts of Western/Central Europe. Instead, we see Basque and Iberian, possibly but not certainly related, Tartessian, Etruscan/Rhaetic (Tyrrhenian), different language stocks, and not a formerly powerful and expansionist language family competing with IE. Compare that for instance with the expansion of Turkic over the former dominant language family of Central Asia and Anatolia/Azerbaijan, which left small pockets of languages of the same IE stock scattered across the region.

 

[ToBeOrNotToBe]

Neither of those examples are deeply diverged, they all fit into the North-Western Indo-European cluster. Steppe Bell Beakers appear in Iberia, Britain & Ireland around 2500-2300 B.C. . In the Chang et al. Bronze Age model Indo-Iranian seperates from NWIE-BS around 2600 B.C., Balto-Slavic separates from NWIE around 2200 B.C. . There's no IE language with such time depth in Western Europe. The BB = NWIE model would likely require a Neolithic timeframe for the breakup of PIE, which means both steppe and Chalcolithic Armenia are out.

What about Corded Ware = Germanic, Hungarian Yamnaya = Italo-Celtic?

 

[markod]

Can you elaborate a bit on why you think so? The model presented by Chang et al. has estimates a bit on the lower end of the time spans I have seen for IE splits compared to other sources I've had access to. IIRC some of them assumed Germanic and Italo-Celtic for instance to have diverged from other branches even before 3000 BC, so in Early Yamna times. Probably a middle ground is closer to the truth. But in any case I did not get why you think a BB = NWIE would require the PIE breakup to be as early as the Neolithic (mind you, I do not believe undivided PIE was spoken in Yamnaya at all, I think that, if Anatolian and even Tocharian are included, the split started even before the Yamnaya expansion, and by the time of CWC it was already split into very close but distinct sister languages, with Proto-Anatolian probably already well developed).

As for NWIE, if Proto-Celtic and Proto-Italic started to break up roughly in 1200-1100 BC, we can assume that the languages had started to be spoken, breaking from their common ancestor, at least ~600-800 years before, so roughly by 2000-1800 BC. And if you add Germanic - if Germanic does indeed derive from a common ancestor with Italo-Celtic, which is not totally certain AFAIK - that date must go back several centuries further, so roughly to 3000-2500 BC (on the lower end coinciding with the begining of BBC in Central Europe). And of course we do not need to presume that NWIE started to be spoken only when the Central European BB phenomenon got formed. The language was brought with the newcomers from elsewhere and it didn't start from scratch straight from Undivided PIE. The timing in my opinion is totally compatible with a NWIE dialect (not necessarily the only one that existed, but the one that was successful in the long term) being spread by Rhine BB people to many areas west of Germany (though not necessarily all, as we all know the BBC had a very sparse occupation of territory, and they may have possibly adopted the local languages in some of the places they settled in, as in the hypothetical case of the Basques).

As for no IE language with such time depth existing in Western Europe, the vast majority of it had no written language virtually until the Common Era. The areas that had written inscriptions from earlier on had writing only during the later Iron Age and were basically those most impacted by Celtic and Italic conquests. That situation is not very conducive to allow us to see remants of earlier NWIE languages (though Germanic, I insist, is definitely one such example, it's definitely divergent and innovative in comparison to Italo-Celtic). I also have to doubt that had the all-important BBC been associated with another language family, we wouldn't see the non-IE remnants of that coherent and homogeneous language family spread to many parts of Western/Central Europe. Instead, we see Basque and Iberian, possibly but not certainly related, Tartessian, Etruscan/Rhaetic (Tyrrhenian), different language stocks, and not a formerly powerful and expansionist language family competing with IE.

Chang & previously Bouckaert are the only unbiased models we have. All other estimates are mere guesswork by historical linguists, usually made to fit their hypotheses of choice. It's difficult to discuss those because they aren't based on data. To the contrary, I suspect that Chang's estimate might even be a bit too old due to the strong non-IE substrates in Western IE and Anatolian.

Italic and Celtic are only marginally closer to each other than either is to Germanic in Chang's model. Germanic / Italo-Celtic separate 1900 B.C. and Italic and Celtic separate 1700 B.C. . Germanic is innovative in the sense that so many of its words aren't Indo-European. To reiterate my previous point: BB spread IE then there should be traces of extant languages that are not NWIE, which existed in a unified population until 1900 B.C. . We should find something akin to Slavic or Indo-Iranian.

 

[Pip]

Well if ATP3 is worth anything, it was about 1/3 CHG, 1/3 EEF and 1/3 European Hunter-Gatherer of some form (I'm not sure how much WHG and how much EHG), according to Genetiker at least. Maciamo looking at this basically described it (excluding its EEF ancestry) as Steppe-derived but with enriched CHG - though I think this is just an illusion, and instead it was just originally WHG+EHG that moved into West Asia.

To clarify, according to Genetiker's most recent (K=14) analysis - ATP3 is quite a bit less than 1/3 CHG and its European HG is entirely EHG (no WHG). Its greatest similarity is to Balkan Z2103, and looks to me like a basal Z2109. I think its paternal ancestor was most likely sitting with formative L51 in the Balkans only a few hundred years earlier.

 

[markod]

To clarify, according to Genetiker's most recent (K=14) analysis - ATP3 is quite a bit less than 1/3 CHG and its European HG is entirely EHG (no WHG). Its greatest similarity is to Balkan Z2103, and looks to me like a basal Z2109. I think its paternal ancestor was most likely sitting with formative L51 in the Balkans only a few hundred years earlier.

In the very same analysis Lithuanians are like 80% 'EHG'. Supervised admixture run have to be interpreted carefully.

 

[Pip]

I can't figure out why you guys want those L51 people to have been a small group. Soon after 2500 BC, they are all over Central and Western Europe, Britain included. They didn't need to just move, they had to confront new populations, fight and conquer. Those places were not empty. There were long-established cultures everywhere, and population density was probably high in some Beaker and Megalithic areas. It must have taken some military might to impose oneself on such structured cultures.

L51 was indeed a large group by 2,500 BC, but before 3,000 BC the bearers of its surviving subclades were a small group. And as there is no sign of L51 in 5th and 4th millennium BC samples, my guess is that the bearers of its non-surviving subclades were most likely smallish in number as well.

 

[Pip]

OK, we are unclear about exactly how Balkan-like DNA in the mid 4th millennium BC ended up developing in Western Europe in the early-mid 3rd millennium BC. Let's move on a little, and focus on the so-called Yamna paternal lineages that would have proliferated in Iberia around 2,500 BC.

Iberia's largest such lineage today is R-DF27. Available data indicates that (i) this lineage moved and expanded there fairly early on (pretty much in keeping with Reich's dating of 2,500 BC), so it is a prime candidate for one of the main developing Yamna groups in Iberia; and (ii) its main track was most likely North Western France down into North Central Castille, i.e. at least its immediate ancestors look to have been external to Iberia.

My tentative view is that:
1. Early surviving L51 was most likely a fringe collaborator with populations that developed from EEF.
2. It probably rose to prominence when it successfully mobilised these populations against hostile R1a Corded Ware expansionism. Despite the similarities, Corded Ware was most likely its main enemy. Its first point of triumph against Corded Ware was in Northern France, from where it led the over-running of CW populations in both directions along the Rhine. Unlike in some other subjugated populations, it looks like most of Corded Ware's female population as well as its male population were eliminated.
3. In doing so, it acquired acceptance and status across the region, leading to colonial-style dispersion and reproductive success. It infiltrated, policed, protected and supervised a multi-ethnic chain of related communities and cultures (rather than trying to destroy them). Both EEF and HG communities that cooperated with it (in Iberia and elsewhere) survived, and those that did not were eliminated.
4. L51 lines probably dispersed quickly and continued out-growing other haplogroups until well into the Bronze Age.

 

[hrvclv]

Italic and Celtic are only marginally closer to each other than either is to Germanic in Chang's model. Germanic / Italo-Celtic separate 1900 B.C. and Italic and Celtic separate 1700 B.C. . Germanic is innovative in the sense that so many of its words aren't Indo-European. To reiterate my previous point: BB spread IE then there should be traces of extant languages that are not NWIE, which existed in a unified population until 1900 B.C. . We should find something akin to Slavic or Indo-Iranian.

As you people must have realized by now, my "bible" for the time being is that Homeland Timeline, which I abundantly referred to upthread. For most samples, what is given is not a definite dating, but a date bracket. Most of the time : 2500 to 2000 BC. If we take into account that margin of uncertainty, then Markod's timings for the successive language splits fit in rather nicely.

We could posit some still-undifferentiated IE language in the Hungarian plains some time around 2500 BC for the L51 group. At some point in time during the next two to three centuries, those people would have moved north of the Carpathians. U106 tribes went their own way, steered north - northeast, conquered and/or mixed with CW people, and over another two to three centuries (ie by 2000 BC) developed some form of pre-Proto-Germanic ; while P312 veered west into BB territory and developed Proto-Italo-Celtic. Italic and Celtic gradually separating from 1700 BC onwards would then turn out to be a pretty coherent estimate.

So that the earliest people to arrive in, eg, the British Isles, would have spoken a language still close to NWIE but already on its way to becoming some form of Celtic, while the NWIE in Iberia would have been some sort of pre-Lusitanian (that riddle language standing somewhere in between Italic and Celtic).

From Wiki : "Prósper, in her Lusitanian etymologies (2002; 2008), demonstrates that not only does Lusitanian not agree closely with the usual Celtic reflexes but that it is closer to Ligurian Italic. This suggests there may have been two well-differentiated branches of Indo-European in the Iberian Peninsula before the Romans, with Lusitanian belonging to the non-Celtic branch. Villar and Pedrero (2001) connect Lusitanian with ancient Ligurian. They base their finding on parallels in the names of deities and some lexical items (e.g., the similarity of Umbrian gomia and Lusitanian comaiam), and some grammatical elements.[2] This once again, raises more questions about the relation of the Lusitanian language with Celtic, because the ancient Ligurian language, in many ways like Lusitanian; is considered Celtic[7] by some and non-Celtic by others. Adding to lack of evidence and its geographical location, it has not been yet determined whether Lusitanian was part of the Ligurian language sub-group, Celtic or Celticised, or an even older Indo-European language. Prósper also sees Lusitanian as predating the introduction of Celtic and shows that it retains elements of Old European, making its origins possibly even older."

 

[brick]

Well, apart from Venetic and Liburnian, which do not fit well into either Celtic or Italic, and the controversial cases of Tartessian and North Picene (which some linguists have tried to establish as extremely divergent IE languages, and not non-IE languages), there is Germanic as a pretty diverged IE branch (especially in terms of syntax, with a lot of innovations) in comparison with Celtic and Italic and certainly dating to much before the Urnfield expansion.

Liburnian, North Picene and Tartessian are all non-IE languages (more specifically preindoeuropean).

 

[Pip]

In the very same analysis Lithuanians are like 80% 'EHG'. Supervised admixture run have to be interpreted carefully.

Yes, I agree we have to be cautious, but if we make a provisional assessment and ATP3's y-DNA calls are positive for R-M269 and its aDNA is closest to early Balkan R-Z2103, the most likely scenario is that it is either Z2103 or some other basal form of M269. In the absence of other evidence, I cannot see any alternative possibility that is more, or even equally, likely.

 

[brick]

in the recent picene paper IIRC, they state north-picene was basically a colony of liburnians and south -picene where umbrians

North Picene language is probably a Liburnian dialect (preindoeuropean) and South Picene is Oscan-Umbrian.

 

[Ygorcs]

Liburnian, North Picene and Tartessian are all non-IE languages (more specifically preindoeuropean).

Are you sure about Liburnian? Some linguists have made connection between it and Venetic.

 

[Ygorcs]

Chang & previously Bouckaert are the only unbiased models we have. All other estimates are mere guesswork by historical linguists, usually made to fit their hypotheses of choice. It's difficult to discuss those because they aren't based on data. To the contrary, I suspect that Chang's estimate might even be a bit too old due to the strong non-IE substrates in Western IE and Anatolian.

Italic and Celtic are only marginally closer to each other than either is to Germanic in Chang's model. Germanic / Italo-Celtic separate 1900 B.C. and Italic and Celtic separate 1700 B.C. . Germanic is innovative in the sense that so many of its words aren't Indo-European. To reiterate my previous point: BB spread IE then there should be traces of extant languages that are not NWIE, which existed in a unified population until 1900 B.C. . We should find something akin to Slavic or Indo-Iranian.

Starostin's (okay, I know they are almost always controversial) glottochronological model did find much earlier dates, IIRC around 2600-2500 BC, for the split of Italic and Germanic and especially between these two and Celtic (in his model Celtic split first, believe it or not).

So you basically think that CWC gave birth to Balto-Slavic, Indo-Iranian, Germanic, Italic and Celtic, even if these two latter are clearly much less linked to Balto-Slavic and Indo-Iranian in particular than the first 3 are among themselves (particularly, of course, Balto-Slavic and Indo-Iranian)? Or what is your personal hypothesis? I have a really hard time putting myself to believe that the earliest Celtic and Italic attested texts were removed a mere 1200 years from their undifferentiated common ancestor... lol. That'd be basically like Italian and Spanish nowadays (Old Spanish and Old Italian definitely having started to split from the common Romance at least by the 8th century AD).

Besides, I think we're forgetting to consider that languages remain a common speech, with little dialectal differentiation, for at least some time. Latin, for instance, had a recorded undivided history of some 1,000 years before it certainly split into recognizably distinct languages. A proto-language that split by 1900-1700 BC probably started to be spoken as the common speech of a language community by at least ~2500 BC (and from which earlier IE proto-language did it split? That's another mystery, but probably not directly from Common Late PIE in my opinion).

As for possible extant IE languages that were totally superseded by Central European NWIE languages (Celtic, Italic and Germanic), we have the possible examples of Venetic, Liburnian (controversial, but many linguists believe it's IE) and also Ligurian, who is still assumed by some linguists to have been at least a "Para-Celtic" language, not Celtic proper. As for languages much to the north of the Mediterranean, well, they simply did not write at all until the Late Antiquity times, so we'll never know.