Ludovic Slimak and colleagues  this week argue that Byzovaya, a site in the Russian far north, was produced by Neandertals.
If true, this is very newsworthy. It would be the highest-latitude Neandertal site, one that would clearly have required an effective adaptation to continental cold. The reported dates would place Byzovaya among the latest Neandertal sites -- showing that these people persisted longest not only the extreme south of their range but also in the far north. Such a finding would pretty much overturn two decades of literature on how and why the Neandertals disappeared.
I really don't see many reasons to doubt the results, except to note that the conclusions must be limited to the quality of the data. Most important, there are no skeletal remains, so we have to depend on the assumption that Mousterian assemblages in this late context were the product of Neandertals. So far that assumption is consistent with the record in Western Europe, but we should probably be cautious nonetheless.
Here's the abstract, which summarizes the paper admirably:
Palaeolithic sites in Russian high latitudes have been considered as Upper Palaeolithic and thus representing an Arctic expansion of modern humans. Here we show that at Byzovaya, in the western foothills of the Polar Urals, the technological structure of the lithic assemblage makes it directly comparable with Mousterian Middle Palaeolithic industries that so far have been exclusively attributed to the Neandertal populations in Europe. Radiocarbon and optical-stimulated luminescence dates on bones and sand grains indicate that the site was occupied during a short period around 28,500 carbon-14 years before the present (about 31,000 to 34,000 calendar years ago), at the time when only Upper Palaeolithic cultures occupied lower latitudes of Eurasia. Byzovaya may thus represent a late northern refuge for Neandertals, about 1000 km north of earlier known Mousterian sites.
I've wrote briefly about Byzovaya in 2005, as part of a discussion of Mamontovaya Kurya, a site slightly north of there ("Who colonized the European Arctic?"). As you can guess from the title of that post, the question of Neandertal occupation of extreme northern Russia was already at play. I quoted Pavlov and colleagues at the time :
The stone-working technology reflected in the Byzovaya material is similar to that of Sungir and other early Upper Palaeolithic sites of the eastern Szeletien tradition, indicating that these artefacts were manufactured by modern humans. However, whether the person who inflicted the marks on the tusk from Mamontovaya Kurya, as much as 8,000-9,000 years earlier, belonged to the same human lineage as the residents at Byzovaya and other Palaeolithic sites further to the south is more uncertain (Pavlov et al. 2001:66-67, citations omitted).
In that 2001 paper, Pavlov and colleagues accepted Szeletian as the product of early modern humans, but I pointed out that this association depended on unjustified assumptions about the technical relation of Sungir and Szeletian sites in Central Europe. Sungir is important because it has skeletal remains, which are not Neandertal. If other sites of equivalent age had similar archaeology, we would assume they were not made by Neandertals. How much does the archaeology at Byzovaya resemble Sungir?
In the current paper, Slimak and colleagues emphasize the differences between the Byzovaya and Sungir assemblages. The work reflects renewed excavations at Byzovaya started in 2007, now totalling more than 300 artifacts, of which 80 are typologically identifiable, the rest cores or unmodified flakes:
None of the 313 artefacts reflects a tool production technology typical of UP cultures. Furthermore, diagnostic tools that are common in any UP industry of Eurasia such as burins, backed tools, pointed blades, or bladelets are not represented. There are 11 end-scrapers, but none of these were prepared from UP blades. Varieties of end-scrapers, prepared from flakes, are common elements in any European MP industry, known since the first Mousterian typological analysis (16). Typological tools are mainly members of the Mousterian group (16), dominated by distinctive side-scrapers made out of flakes (fig. S5, nos. 1 and 2) that are typical for MP industries (17) (fig. S6 and table S4). Six of these tools have been retouched to form a bifacial tool. Most of the bifacial tools are thick, with a plano-convex section: one face shaped by large flakes and the opposite face formed by a semi-abrupt retouch. This way of shaping has been used for producing so-called Keilmesser tools (plano-convex and backed bifacial tools, Fig. 3, no. 1), which are considered to be specific artefacts of some archetypical MP industries of Central and Eastern Europe (18–20). Two of the bifacial tools from Byzovaya present a thin regular transformation of their faces that illustrates the technological similarities between this industry and the Eastern European MP (18, 19), where the so-called Blattspitzen (short foliate) tools occur frequently.
The apparent use of a Keilmesser-like approach is interesting, this is otherwise known from late Micoquian contexts in Germany and other parts of central Europe. It does seem to hang together as a technical package, involving a distinctive pattern of retouch on bifaces by flake removal from flat surfaces. The authors argue that the technology at Byzovaya is "technically homogeneous" with diagnostic features of central and eastern European Middle Paleolithic; this is supported by their data but it is worth noting that only 5 of the tools show these links -- the larger signature is the lack of anything that could belong only to an Upper Paleolithic context. The authors deal in a paragraph with the alternative hypotheses that raw material or the indended use (expedient butchery) may have limited the toolkit, concluding not. So it's Mousterian, likely similar to the kind found in Eastern Europe but quite a bit later in time.
(as an aside, I will point out that I have written quite a bit about the early Upper Paleolithic of the Russian Plain, many of the sites are discussed in this paper. For example, my post on Kostenki: "The initial Upper Paleolithic at Kostenki", which links to others.)
The news aspect of the story -- the reason it's in Science -- is the date. Several radiocarbon dates on fauna, including cutmarked bone and ivory, cluster around 28,500 years BP, which calibrates to between 31,300 and 34,500 years ago. This range of dates is also confirmed by OSL on sand grains. If Neandertals made this site (and if we admit some doubt about later dates for Mousterian sites in Spain) Byzovaya could have been made by the latest Neandertals anywhere in the world.
A site near the Arctic Circle is totally the opposite of where we've been finding other late Mousterian occupations. Up to now, the latest Neandertals apparently had lived in Iberia, sites not quite as late are found in France, Italy, Croatia, and the Caucasus. Those places are all in the southern tier of Europe, leading many archaeologists to conclude that the Neandertals couldn't cope with the deteriorating climate of the Heinrich IV event. With better-tailored clothing and a more complex logistical strategy, Upper Paleolithic people seemed to have had a better cultural strategy to handle the truly cold steppes of periglacial Europe. Neandertals were increasingly limited to areas with the mixed patches of forest that they favored, an ecology that was shrinking after 45,000 years ago.
Toss that hypothesis out the window. And close it, it's cold out there!
Oh, well I guess I don't really think this paper alone disposes so neatly of the lingering Neandertal sauna hypothesis. But it should inspire us to think of an alternative. I like that, no neat tidy package.
But this paper has a glaring problem, as I see it: Somebody was at Mamontovaya Kurya, more than 100 km north of Byzovaya, more than 5,000 years earlier. But the paper doesn't discuss Mamontovaya Kurya at all! The paper discusses earlier sites far to the south, but not the one that's closest. If there is a Neandertal persistence in the Russian Arctic, surely these two nearby sites must both represent that population. Are the toolkits similar? Are these the same people? Why is there no discussion of it? Do Science papers no longer have to cite Nature papers? Isn't this the obvious comparison?
Frustrating it is.
Pavlov and colleagues wrote that the Mamontovaya Kurya and Byzovaya assemblages were quite different. But at that time (2001), they also wrote that Byzovaya was similar to Sungir. Here, Sungir and Byzovaya are depicted as very different. There are only 313 artifacts here. This is my frustration with archaeologists: too much depends on a typological assessment, the details of which are underreported in many publications.
How sure can we be that the apparent technical connections with Eastern and Central European Micoquian are real, sustained cultural traditions? In light of the rapid cultural shifts further to the south, probably we should doubt such a persistence or at least provide some mechanism for it.
What does this signify about the radiocarbon story?
Of course, my other frustration of late has been the problems of radiocarbon chronology. Just this week, I wrote about a paper that questioned the persistence of Neandertals after 40,000 years ago anywhere in Europe. Now, here's a paper that posits Neandertals in an entirely unexpected part of Europe less than 35,000 years ago. What gives?
As I noted on Tuesday, one of the sticking points is that some archaeologists insist on dating human bone because of the doubt that always accompanies mere associations by level. Only a few sites have Neandertal or non-Neandertal skeletal material, but many, many sites have been dated and have archaeology that is typologically diagnostic -- Mousterian, Châtelperronian, Aurignacian or whatever. Many archaeologists are happy to assume that a Mousterian site was made by Neandertals, an Aurignacian site by modern humans. Transitional (Châtelperronian, Uluzzian, Szeletian) sites have always raised more objections, as does early Aurignacian for many because of the lack of skeletal associations.
From the current paper, you can see the assumption and its effects:
Most researchers agree that classical Mousterian industries in Europe were exclusively produced by Neandertals (30, 31). However, whether Byzovaya represents a Neandertal site or not cannot be demonstrated beyond doubt until human bones or DNA are found. If the Byzovaya artefacts were struck by modern humans, this would have major implications for understanding the MP-UP transition, as it would imply that these Arctic H. sapiens groups preserved older, traditional MP cultures far after the full expansion of UP modern societies in the rest of Eurasia.
Oh, yes. That would be interesting, wouldn't it? I don't want to reduce the dichotomy but to multiply it. There weren't only two populations, a single group of Neandertals and a single group of early Upper Paleolithic non-Neandertals -- there were many successive populations of both. The Russian Plain was probably covered by different modern human populations at different times, possibly none of whom were very closely related to today's Europeans.
If the population history of Europe during the Middle-Upper Paleolithic transition is demographically complex, I think we should be more skeptical about the association of stone assemblages. We should probably insist even more strongly on dates from human skeletal material. But we should be less certain of the affinities of the skeletal materials themselves -- which are rarely complete. As we know from Les Rois, a few Neandertal traits will not allow a satisfactory diagnosis of partial remains.
At the moment, the dispute about radiocarbon dates of Neandertals is quite simple. It is not about Neandertals, really; it's about the quality of evidence associating Neandertals with dates, which must (at present) go through the two indirect steps: Associating fragments with populations, and associating populations with tool assemblages. Some researchers leap through these two steps, others take them more cautiously, a few won't take them at all. And that's not going to change soon.