The Neandertals of Mount Doom

12 minute read

Well, I already snarked on the science headlines that have been claiming volcanoes “wiped out” the Neandertals. Some variation of this story, swapping in a different Neanderkiller, has been circulating since around 1890. But is there any truth to the headlines?

(see UPDATE below)

The source of the story is a paper in the October issue of Current Anthropology, by Golovanova and colleagues (2010). The paper reviews the chronology of Mezmaiskaya Cave, a site occupied by Neandertals and successive Upper Paleolithic peoples, in the Russian Caucasus. This site produced the skeleton of an infant, from which DNA evidence has been recovered. As Golovanova and colleagues describe, the deposit additionally contains volcanic ash from two eruptions that happened around 40,000 years ago.

The latter of the two eruptions appears to coincide with a long abandonment of the site:

Hominin occupation of Mezmaiskaya Cave changed dramatically after the later volcanic eruption represented in layer 1D. This eruption was probably more powerful than that in layer 2B-1. Layer 1D has a thickness up to 0.7 m and in some areas is composed of a relatively clean sediment lacking any inclusions (fig. A10). Limestone fragments, bones, and lithic artifacts are absent, and even pollen grains are rare. Pollen data show that extreme deterioration to a very cold and dry climate occurred in this time period....A chemical analysis of layer 1D indicates that the volcanic ash apparently derives from an eruption in the Kazbek volcanic province that occurred around 40,000 years ago. Because no Neanderthal specimens or MP lithic industries postdate layer 1D at Mezmaiskaya, this eruption seems to have significantly disrupted the ecological niche of local Neanderthals, possibly resulting in their rapid disappearance in this region.

The Kasbek volcanic province is in the Caucasus, so we’re talking about a large eruption relatively local to the site. This is the sort of event you might well expect to have a strong impact on a dispersed hunter-gatherer population. The Middle Paleolithic people (presumably Neandertals) might have locally declined in numbers, or they might have moved on. The region need not have been abandoned entirely; a new population might have entered the area without using the same site. In this case, when new people began to use the site much later, the newbies were using an Upper Paleolithic industry.

A relatively local effect of volcanism in the Caucasus is one thing, but the extinction of Neandertals across western Eurasia is quite a bit more. How does the paper go from local event to a regional extinction?

The local eruption was the second event to leave ash in the Mezmaiskaya sequence. The first was a different eruption from Mt. Elbrus, which had a smaller impact than the second, as discussed below.

At issue in the paper is the possible coincidence of the second eruption and consequent abandonment of the site with a much larger volcanic event in Italy:

The CI [Campanian Ignimbrite] eruption from the Phlegrean Fields, southern Italy---the largest eruption documented in the Mediterranean region during the past 200,000 years (Wohletz, Civetta, and Orsi 1999)drastically impacted European ecosystems. The most recent numerical (40Ar/39Ar) age determinations for CI eruption vary from to BP and cluster around 40,000 BP (Fedele, Giaccio, and Hajdas 2008:839).
This eruption produced the CI in Italy and Y5 tephra in Central and Eastern Europe and Eastern Mediterranean (Fedele et al. 2003; Fedele, Giaccio, and Hajdas 2008). High-altitude clouds of volcanic ash from this eruption had a significant effect on global climate. The resulting ash fall covered km2 of land and sea (fig. 1), and the Y5 tephra layer accumulated in the Eastern Mediterranean as far as Cyprusmore than 1,500 km from its source (Mussi 2001:191). The Y5 tephra is also identified in the EUP sequence at Kostenki in the Middle Don River in Russia (Holliday et al. 2007). In Eastern Europe, the ash layer varies from 12 cm in the eastern limit (between Penza and Rostov) to 58 cm in the west and southwest (southern Ukraine and Moldova) and averages about 34 cm (Laverov et al. 2005:51). Obviously, the area affected by this ash fall was much larger than the documented Y5 tephra sites.

Golovanova and colleagues propose the hypothesis that the climate effects of this CI event caused the demise of the Neandertals:

Our new data provide support for the hypothesis that the MUP transition in western Eurasia coincides with one of the most globally significant volcanogenic catastrophic events in the recent history of the earth. The large and coeval volcanic eruptions (from an unusually large CI eruption in the Apennines to a smaller eruption in the Caucasus) had a sudden and devastating effect on the ecology and forced the fast and extreme climate deterioration (so-called volcanic winter, perhaps comparable to the effect of nuclear winter) of the Northern Hemisphere in the beginning of Heinrich Event 4. We guess that this catastrophe likely may have both drastically destroyed the ecological niches of Neanderthals, possibly resulting in the mass death of hominins and prey animals and the severe alteration of foraging zones, and caused Neanderthal depopulation from Central Europe to the Caucasus.

That’s a very clearly stated hypothesis. A volcanic eruption initiated climate effects that the regional population of Neandertals could not survive.

However, Golovanova and colleagues include in their paper several critical facts that run against this hypothesis:

1. The Mezmaiskaya sequence itself shows Middle Paleolithic people returning and proliferating after a large relatively local eruption. The Elbrus eruption apparently left ash in layer 2B-1, with a low density of bones and a very low frequency of bison compared to caprids. The excavators interpret the layer as a very low-intensity use of the cave. The pollen evidence suggests a “cold, dry climate”. In other words, the paleoclimate and faunal evidence are both consistent with the hypothesis that the eruption had effects on Neandertal populations in the Caucasus. But then the Neandertals apparently returned in force:

The intensity of site use increased, however, during the accumulation of the upper MP layers 2A and 2 when the climate become cool and wet. Although the lithic industry changed slightly after the environmental crisis of layer 2B-1, it still remained typically MP Eastern Micoquian. Skeletal and mtDNA evidence indicates that Neanderthals produced both the earlier and the later MP industries at Mezmaiskaya (Briggs et al. 2009; Golovanova et al. 1999; Green et al. 2010; Ponce de Leon et al. 2008). Thus, the late MP environmental crisis at the cave had repercussions for local Neanderthals but did not cause a break in the continuity of occupation or technology.

That makes it seem pretty unequivocal. Neandertals survived and effectively adapted to at least one volcanic event in this area. That eruption did not kill them off, and it did not leave the area devoid of Neandertals in a way that facilitated a “modern human invasion.”

It was only after the second volcanic event that Middle Paleolithic people declined at the site.

At issue is whether this second event was coincident with the CI eruption. The ash in the Mezmaiskaya sequence is not from the Y5 tephra, it is attributable to a much nearer source. I do not fully understand why the authors attribute this second event to the same time as the CI event; the time between layers 2 (terminal Mousterian) and 1C (early Upper Paleolithic) appears to have occupied a few hundred years, between 32,000 and 34,000 radiocarbon years BP. Calibration will move those dates older by a few thousand years (I discussed radiocarbon calibration a few years ago). But I think the CI eruption, around 40,000 years ago, doesn’t fit well with this later event. It might fit with the earlier eruption, in my view, as Elbrus lava flows include 40,000 BP.

In any event, I think the associations of either local volcanic event with the larger CI event is at best uncertain. The record at the site makes it pretty clear that Neandertals were effectively adapting to the changing local climates and faunal abundance that coincided with the first eruption.

2. The initial Upper Paleolithic of Kostenki had appeared before the Y5 tephra was deposited. I wrote about the identification of this Y5 tephra at Kostenki a couple of years ago (“An earlier initial Upper Paleolithic at Kostenki”). As my post indicated, the identification of the ash layer with the Campanian Ignimbrite event suggested an earlier date for the initial Upper Paleolithic on the Russian Plain.

From the standpoint of the Neandertal volcanic winter hypothesis, this sequence of events is a problem, which Golovanova and colleagues discuss:

In any case, with or without the Kostenki addition, the few CI-bearing sites show that this eruption could have also extinguished the first wave (Proto-Aurignacian) of EMH expansion into Europe (Fedele, Giaccio, and Hajdas 2008): At all key sites, where sedimentary resolution is good, the CI tephra directly seals archaeological layers that contain assemblages of the MUP mosaic, often variants of Aurignacian-like or so-called Early Upper Paleolithic Industries. The layers above the CI tephra, where they are not culturally sterile, contain later and often much later properly defined Upper Paleolithic industries (841). Thus, the CI-bearing sites demonstrate clear evidence of the break in habitation and culture changea whole gamut of archaeological attributes for population replacement.

The volcano is supposed to explain the MUP transition, but occurs earlier than the MUP transition in some areas, but later in others. Golovanova and colleagues propose an ad hoc hypothesis to account for this mismatch: some early Upper Paleolithic modern humans were also wiped out.

Many researchers might find this idea tempting. It might, for example, explain why the (few) skeletal remains of the earliest Aurignacian people have such a high proportion of Neandertal features. We could propose that the initial Upper Paleolithic represents a degree of population mixture that later populations do not; the discontinuity between them could have been caused by climate extremes.

But we don’t need climate or volcanism. Later Upper Paleolithic people retained similarities to Neandertals, which reduced in frequency over time. This is most readily explained by continued gene flow into a sparse European population from West Asia. A volcano-induced climate catastrophe is superfluous: It doesn’t add to the explanation of a sustained genetic transformation of Europe that continued through the later Upper Paleolithic and Neolithic.

3. “Catastrophes” are not rare. The record of climate change during the last glaciation shows frequent strong oscillations. Some of these occurred at the same time as known eruptions, and so might be associated with them, but most climate oscillations have no obvious cause. Up to 40,000 years ago or so, the Neandertals survived them all. They survived the Toba event, largest eruption by volume in the Pleistocene, with no evidence of ill effects.

The “intensity of occupation” of archaeological sites naturally fluctuated for many reasons. In Paleolithic contexts, sites were almost never inhabited continuously. We usually don’t know why a local population returned more often to a site, or why the later population may have used the site less often, but those changes in pattern will make a big difference to the occupation intensity. It’s not enough to show that a fluctuation in occupation intensity was coincident with an eruption or climate event – such coincidences are inevitable even when “occupation intensity” changed randomly.

What role volcanoes?

Bad things happened in the past. Many of those bad things – megadroughts, volcanoes, asteroid impacts, flesh-eating bacteria – probably killed a lot of people.

But our ability to find the effects of these death-dealing events is a lot more limited than you might assume. Less than a thousand years after the Black Death, how many signs of it are still evident today? To the exceedingly clever, who know where to look, there are a few. If we discount historical records, which do not exist for the Pleistocene, and limit ourselves to very small samples of bone and stone remains, it becomes very difficult to demonstrate this widespread epidemic, which reduced the population of some parts of Europe by up to half.

Most Paleolithic sites document exceedingly low-intensity use of an area by ancient people, and have gaps of thousands of years. The hope of finding a single event with a short duration is near zero, unless it affected many sites in the same way.

The extinction of a widespread group of hominids would be one kind of event we might test. In the current example, I think the data point to a clear conclusion: Not all Neandertals were killed, starved, or slowly declined due to the effects of any single volcanic eruption. Too many of them clearly survived the time of the large eruptions, and the available archaeological indicators suggest that their populations tended to recover after climate extremes had been reached. They were very resilient to climate change, more than many other mammals.

It’s not possible to rule out that one or more eruptions may not have had important effects, even ones that may have devastated some local populations. This is possibly the case at Mezmaiskaya. Nor is it possible to exclude the hypothesis that climate changes of greater and greater amplitude may have stressed their populations, contributing to the Neandertal demise.

That’s one of the returning frustrations of the archaeological record. An event might have been a major tragedy in human terms, but essentially invisible to us today. Meanwhile, the large-scale dynamics of human populations, including speciation and extinction, do not appear to fit the record of catastrophic eruptions. I don’t see that as the end of the story, but a more interesting prologue to our understanding of ancient human dynamics.

UPDATE (2010-10-16): I received a note from Golovanova and Doronichev, kindly pointing out a serious error in my post. I had misread their paper – I described it as supporting a coincidence of CI with the first ash evidence at Mezmaiskaya, but the paper clearly argues that the CI event was “coeval” with the second ash, in layer 1D of the site.

I have extensively updated the part of the post that refers to the CI eruption.

I’m skeptical that the CI and Kazbek ashfalls could have happened near the same time, because the latter seems by radiocarbon evidence to be 2000 or more years later than 40,000 years ago. But the ESR dates are arguably consistent with the idea that the two eruptions coincided. I wouldn’t push a chronology argument very far, not without a list of calibrated radiocarbon and ESR/TL dates from the relevant eruptions. But the multiplicity of events helps to reiterate the basic point that geological events happened, and fluctuations of site intensity happened, and it will take a coincidence across many sites to correlate the two.


Golovanova, L. V., Doronichev, V. B., Cleghorn, N. E., Koulkova, M. A., Sapelko, T. V., & Shackley, M. S. (2010). Significance of ecological factors in the Middle to Upper Paleolithic transition. Current Anthropology, 51(5), 655-691. doi:10.1086/656185