THE NEOLITHIC INVASION OF EUROPE 1
Sunday, 25. November 2007, 00:06:38
Martin Richards
Department of Chemical and Biological Sciences, University of Huddersfield,
Queensgate, Huddersfield HD1 3DH, United Kingdom; email: m.b.richards@hud.ac.uk
Annu. Rev. Anthropol. 2003. 32:135–62
doi: 10.1146/annurev.anthro.32.061002.093207
Copyright c° 2003 by Annual Reviews. All rights reserved
Key Words Neolithic farmers, Mesolithic foragers, mitochondrial DNA,
Y chromosome, phylogeography
Abstract Who are Europeans?
Both prehistoric archaeology and, subsequently,
classical population genetics have attempted to trace the ancestry of modern Europeans
back to the first appearance of agriculture in the continent; however, the question has
remained controversial. Classical population geneticists attributed the major pattern
in the European gene pool to the demographic impact of Neolithic farmers dispersing
from the Near East, but archaeological research has failed to uncover substantial
evidence for the population growth that is supposed to have driven this process. Recently,
molecular approaches, using non-recombining genetic marker systems, have
introduced a chronological dimension by both allowing the tracing of lineages back
through time and dating using the molecular clock. Both mitochondrial DNA and Ychromosome
analyses have indicated a contribution of Neolithic Near Eastern lineages
to the gene pool of modern Europeans of around a quarter or less. This suggests that
dispersals bringing the Neolithic to Europe may have been demographically minor and
that contact and assimilation had an important role.
INTRODUCTION: FARMERS OUR ANCESTORS?
Prehistoric archaeology grew up under the shadow of nationalism, providing the
means by which the newly established European nation-states could create a unitary
past for their peoples (Trigger 1989, Kohl & Fawcett 1995). There was traditionally
a tendency, therefore, for the narratives of European prehistorians to
divide their actors into “us” and “them.” At the deepest level, as Zvelebil (1995a)
argues, this amounted to a founding myth for European culture and civilization
that placed extraordinary emphasis on the Neolithic—a myth that idolizes farmers
at the expense of hunting and foraging ways of life.
There are, Zvelebil believes, three particular reasons why this may have happened.
The first arose from the prejudice against hunter gatherers that emerged to
justify the colonial persecution of hunter-gatherer communities in the NewWorld
(Gamble 1992). The second is the rise of urbanism and the resulting idealization
of the rural way of life. The third is the need of many of the new nation-states,
especially in central and northeast Europe, to forge a national identity among their
predominantly peasant populace. The effects were felt not only in archaeology but
also in literature from Wordsworth and Hardy to Tolstoy, and much more widely
throughout the popular culture.
In archaeology, of course, the triumph of these views came with V.G. Childe’s
proposal that the appearance of the first farming communities in Europe—the
future Marxist Childe’s “Neolithic revolution”—was the result of immigration
of populations from the Near East (Childe 1925). He proposed that following the
onset of the Neolithic in the Near East at the end of the last glaciation, about 10,000
years ago, farmers dispersed north and west into Europe, replacing the indigenous,
Mesolithic hunting and foraging “savages” by virtue of their superior technology
and culture. Zvelebil argues, along with a number of Palaeolithic and Mesolithic
archaeologists, that the contribution of Europe’s earlier, indigenous inhabitants to
European society has been underestimated ever since.
This presumption in favor of what Zvelebil calls “farmers our ancestors” was
perpetuated by both Neolithic archaeologists and population geneticists, as the
latter became involved in prehistoric reconstruction from the 1970s onward. As
recently as the 1980s, Bradley (1984) was able to make his famous observation
that “in the literature as a whole, successful farmers have social relations with
one another, while hunter-gatherers have ecological relations with hazelnuts.” Yet
the period since his remark has begun to witness a sea change in our view of
the Neolithic and, finally, a rehabilitation of Europe’s hunter-gatherer past. This
has happened for two reasons: first, a sustained critique of the mass-migration
perspective from the archaeological community, stressing in particular the importance
of the social context of the Neolithic transition, and, second, by the arrival
of molecular genetics on the scene.
The image of an invasion force marching across Europe at the beginning of the
Neolithic has always been a caricature, but the question of the relative contribution
of newcomers and natives to the European Neolithic has been vigorously fought
over during the last 15 years. This question in turn has implications for explaining
the social changes that took place with the shift to the Neolithic (e.g., Thomas 1996,
1998; Zvelebil 2000), and perhaps also for the arrival of new languages (Renfrew
1987, Zvelebil 1995b).Onthe genetic side, however, it appears that some consensus
may finally be emerging. This article aims to review that part of the story, in order
to assess the state of the argument over “farmers our ancestors” today.
THEWAVE OF ADVANCE
The subject of the genetic history of Europe was more or less created by Luca
Cavalli-Sforza and his colleagues in the 1970s. Their work has cast a long shadow,
to the extent that today’s genetic researchers into the ancestry of Europeans, and
even many archaeologists, continue to discuss their subject matter in the terms
that they laid down. Cavalli-Sforza’s work was pioneering in two ways. It was the
first sustained attempt to apply genetic data to a question of major archaeological
interest. In addition, however, it helped to bridge the gap between genetics and
archaeology by being carried out in collaboration with an archaeologist, Albert
Ammerman. The partnership began in 1970 and culminated in 1984 with The
Neolithic Transition and the Genetics of Populations in Europe, which remains a
seminal work in archaeogenetics.
Ammerman & Cavalli-Sforza (1984) drew upon the new processual archaeology
of Lewis Binford (1968) for a scientific model explaining the origins and
spread of farming in western Eurasia. Binford’s density equilibrium model suggested
that farming had arisen in the Near East in areas of optimal wild resources,
leading to sedentism, local population growth, and expansion into more marginal
environments. Ammerman and Cavalli-Sforza accepted the central role of sedentism,
population growth, and the resulting resource pressure in the early farming
communities.
Cavalli-Sforza had studied under the founder of orthodox statistics, R.A. Fisher,
and was keen to apply quantitative methods to the study of prehistoric Europe. His
work with Ammerman began by measuring the rate of the spread of farming
into Europe, drawing on the newly available radiocarbon maps compiled by Clark
(1965), showing an east-west trend in first Neolithic dates. They observed that there
was a complex of elements at Neolithic sites in Europe—the so-called Neolithic
package. This included cereal crops—especially emmer wheat, einkorn wheat, and
barley, whose wild progenitors occurred only in the Fertile Crescent of the Near
East—domestic animals, pottery, ground and polished stone tools, and houses. To
maximize their data set, Ammerman and Cavalli-Sforza sometimes chose to rely
upon single items, focusing on cereals where possible, as markers for the spread of
the Neolithic. They developed the use of isochron maps, plotting similarly dated
sites on a map of Europe as a series of isochrons or contour lines. The result was
a remarkably uniform rate of about one kilometer per year, or 25 kilometers per
generation, with minor variations in different parts of the continent. The entire
process, from Greece to the British Isles, had taken place in about 2500 years.
This is in fact rather rapid: less time, for example, than it took farming to spread
throughout the Near East, where it had originated (Price 2000a).
This discovery that the rate of spread across Europe appeared to be roughly
constant suggested a single overarching mechanism, a diffusionary process rather
than an old-fashioned model of directed colonization. Ammerman and Cavalli-
Sforza introduced the expression demic diffusion to contrast the immigration of
farmers themselves with the spread of farming as an idea through the indigenous
hunter-gatherer populations—cultural diffusion. Demic diffusion could imply a
rather traditional model of migration and colonization, butAmmermanand Cavalli-
Sforza argued that the observed rate of spread suggested something different: a
“wave of advance.”
They took the wave of advance model from Cavalli-Sforza’s mentor, Fisher,
who had used it to describe the spread of an advantageous gene though a population.
The wave of advance combines two features: logistic population growth
and random local migratory diffusion or range expansion. The population growth
was explained as the result of agriculturalist surpluses and storage in Neolithic
societies, which allowed the carrying capacity of the land to rise. The outcome of
growth combined with range expansion is a radial expanding population wave, in
which the culture spreads with the expansion of people. Ammerman & Cavalli-
Sforza (1984) described the wave of advance as “colonization without
colonists.”
CLASSICAL MARKERS AND THE NEOLITHIC TRANSITION
The wave of advance model appeared to be compatible with the rate of spread
of the Neolithic measured from radiocarbon dates. More important, however, was
the introduction of genetic data into the equation. At the time, these necessarily
comprised only the “classical,” non-DNA markers: allele frequencies for blood
groups, the tissue antigen HLA system, and some enzymes. Assuming that the
Near East and Europe had been relatively isolated during the Upper Palaeolithic
and Mesolithic, and had therefore had the opportunity to differentiate genetically,
certain predictions were possible that might allow different hypotheses about the
spread of agriculture to be distinguished.
A demic diffusion of farmers from the Near East into Europe, involving
complete replacement, was thought to predict complete homogeneity of the
gene-frequency composition of European and Near Eastern populations (until subsequent
differentiation). On the other hand, an entirely cultural spread would leave
the two regions with different genetic compositions (assuming that was how they
started out). However, a mixed model of demic diffusion involving intermarriage
with the Mesolithic population would lead to a gene-frequency cline, or gradient,
along the main axis of expansion, with one pole in the Near East and the other in
northwest Europe.
It had already been shown by Mourant (1954) that the Rhesus-negative (Rh¡)
blood-group gene was virtually restricted to Europe, North Africa, and the Near
East, with its highest frequency amongst the non-Indo-European-speaking Basques
of southwest Europe. Mourant became one of the first biologists to use genefrequency
data to write prehistory when he suggested that this might be because
the Basques were a relict of an ancient proto-European population who had mixed
with newcomers later on. To Ammerman and Cavalli-Sforza, this idea suggested
that the newcomers might have been expanding Near Eastern farmers, spreading
agriculture into the continent. They were delighted to discover that the Rh¡ gene
was also found at high frequencies in northern and northwest Europe—precisely
at the supposed peripheries of the Neolithic expansion.
However, other genes often showed different patterns. Furthermore, the Near
East and Europe were not in fact highly differentiated from each other, weakening
the picture still further. It was therefore necessary to take a multivariate approach,
drawing on the results of many genetic systems, and to find an analytical method
that could dissect different patterns. Cavalli-Sforza, with colleagues Menozzi and
Piazza, chose principal-component (PC) analysis (Menozzi et al. 1978). They used
this to summarize the gene frequencies at a particular location and to represent as
much as possible of the information in just a few dimensions. Because the published
data sets they used were taken from a variety of sources, there were many
gaps, which had to be filled in by interpolation. The results could be graphed as a
two-dimensional plot (portraying, say, the first and second or first and third components),
or they could be presented as contour maps, component by component,
showing the changes in frequency with geography.
The map of the first PC, accounting for about 27% of the total variation in
classical marker frequencies across Europe and the Near East (initially using 39
genetic loci; later 95), has become something of an icon in the archaeogenetics of
Europe (Figure 1). It showed a gradient from the southeast to the northwest, with the
Near East at one pole and Europe at the other. The resemblance to the radiocarbon
map for the spread of the Neolithic was immediately obvious. This was, Cavalli-
Sforza and his colleagues believed, strong evidence for the mixed demic diffusion
hypothesis. The second and third components (accounting for about 22% and 11%
of the variation, respectively) showed clines that were oriented roughly southwestnortheast
(Figure 2) and east-west. Because their impact on the genetic variation
was lower, the processes that generated these were assumed to have taken place
more recently than the Neolithic.
The relative proportions of incoming farmers and indigenous hunter-gatherers
were difficult to assess, although the process of diffusion and interaction could be
simulated using the wave of advance model (Ammerman & Cavalli-Sforza 1984).
The simulations showed that the extent of acculturation—modeled as the marrying
of hunter-gatherers into the farming community—was critical.With high levels of
intermarriage, the survival of Near Eastern genes in the European population could
in fact be very low, even with the wave of advance in operation, and a rapid fall-off
of Near Eastern genes toward the northwest would be expected. Thus, whereas
genetic data could perhaps be used to assess the extent of demic diffusion, testing
the wave of advance model was going to be more difficult.
The conclusions of Ammerman and Cavalli-Sforza and their colleagues were
supported by work using a different analytical approach: spatial autocorrelation
analysis (Sokal et al. 1989, 1991). Again, about a third of the genetic markers analyzed
appeared to be arranged in a southeast-northwest cline. Despite Ammerman
and Cavalli-Sforza’s early caveats about the acculturation coefficient, and despite
the fact that only about a quarter to a third of the variation could be explained
by these gradients, the assumed model of surplus-driven population growth and
expansion led both groups to tend to play up the role of the Neolithic newcomers at
the expense of the indigenous Mesolithic peoples. After all, it was the newcomers
who had won in the end. No matter how careful and qualified the argument put
forward in 1984 was, by the time of the publication of Cavalli-Sforza’s magnum
opus, The History and Geography of Human Genes (Cavalli-Sforza et al. 1994),
the view that the genetic data supported an overwhelmingly Neolithic ancestry
for modern Europeans had firmly taken root. It was further reflected in the view
that much of the remaining variation was the result of subsequent migrations into
Europe and that the principal components could provide relative dating, like an
archaeological stratigraphy (Cavalli-Sforza 1996).
Further consequences ensued. Although he was suitably cautious about the
genetic evidence for demic diffusion, Renfrew (1987) enthusiastically embraced
the wave of advance model. To Renfrew, it was a good processual model that
could be rigorously formulated, anti-migrationist (in the traditional sense) and yet
allowing for the expansion of a group of people over, potentially, an enormously
wide area. It seemed perfectly suited to Renfrew’s radical hypothesis that the spread
of Indo-European languages through Europe had been mediated by agriculturalist
expansion. He suggested that the Basques were indeed a Palaeolithic relict and
that the Indo-European languages had been dispersed through Europe by Neolithic
populations originating in Anatolia. The choice of Anatoliawas necessitated by the
fact that this region was home to a number of extinct Bronze-Age Indo-European
languages, which were extremely archaic and believed by some linguists to form a
sister branch to the surviving members of the family. Many linguists believed them
to be intrusive to Anatolia; Renfrew’s suggestion was not popular among Indo-
Europeanists, who mostly preferred something akin to the hypothesis of Marija
Gimbutas that the Proto-Indo-Europeans had spread west with the Bronze-Age
“Kurgan culture” of the eastern European steppe (Mallory 1989, Hines 1991). It
was, however, received with considerable interest among archaeologists (Zvelebil
& Zvelebil 1988, Sherratt & Sherratt 1988).
CRITIQUES OF THEWAVE OF ADVANCE
Despite some rather vigorous methodological disagreements between the two main
proponents of large-scale Neolithic demic diffusion, the Cavalli-Sforza and Sokal
groups, the southeast-northwest gradient itself was robust to a number of tests.
It appeared there was a genuine pattern that needed explaining, rather than some
artefact of the analysis. Nevertheless, interpreting the gradients was not as unproblematic
as first thought. This was originally argued not by geneticists, however,
but by an archaeologist sympathetic to Renfrew’s perspective, Marek Zvelebil.
Firstly, Zvelebil (1989, 1998a) pointed out that there was no strong reason
for identifying the first PC solely with a Neolithic expansion. Europe is a small
peninsula of the Eurasian landmass and, as such, is likely to have been the sink
for many such dispersals throughout prehistory.Movements into Europe may well
have taken place many times via Anatolia because this represents one of the main
possible points of entry into the continent. The term palimpsest came into play
(Renfrew 1998, Zvelebil 1998a): The gradients identified by PC analysis might
be the result of many dispersals, each one overwriting the last. The Neolithic may
have been one of these, or it may not; if itwas, it may or may not have been the most
significant. Whereas Zvelebil’s argument for “incremental palimpsest” (Zvelebil
2000) focused on post-Neolithic migrations, Richards et al. (1997) compiled a
radiocarbon map for the first spread of the Early Upper Palaeolithic into Europe
(analogous to the early maps for the Neolithic produced by J.G.D. Clark). They
pointed out that the spread of the first modern humans into Europe, from about
45,000 years ago, followed very similar routes to the later spread of the Neolithic—
a rapid dispersal around the Mediterranean and a more gradual expansion along the
southeast-northwest axis, following the river systems of central Europe. Whereas
any patterns set up by the arrival of early modern humans may well have been
erased by subsequent demographic events, this reinforced the point that the same
routes into Europe may have been used time and time again.
Thus, the idea of “one PC—one migration,” suggested quite specifically by
Cavalli-Sforza, seemed implausible, as did his analogy between principal components
and an archaeological stratigraphy. He and his colleagues arranged principal
components in temporal succession, according to their magnitude. The first PC,
the supposed Neolithic southeast-northwest gradient, came first, accounting for
about 27% of the total variation. The third, running east-west, was attributed to
Gimbutas’s Proto-Indo-European Kurgan warriors and other invaders from the
steppes, and the fourth was ascribed to Greek colonization of the eastern Mediterranean.
The second, showing a southwest-northeast gradient and accounting for
about 22% of the variation, was thought to be problematic because no plausible
demographic explanation in the appropriate time frame (between the early Neolithic
and the Bronze Age) could be divined for it. It now seems surprising that
the suggestion that the second component might have been in part the result of the
Lateglacial re-expansions across Europe from the southwest was not mooted until
1998 (Torroni et al. 1998). Cavalli-Sforza and his colleagues were not ignorant of
Palaeolithic processes in Europe, but the fallacious idea that the magnitude of the
PCs reflected their age was a blinker to the full range of explanatory possibilities.
Further critiques of the wave of advance have been mounted on archaeological
grounds. In the first place, the radiocarbon map of Neolithic spread used by
Ammerman & Cavalli-Sforza (1984) was flawed in a number of ways, and the
Neolithic package they had used to map the expansion was gradually picked apart.
Whereas Ammerman and Cavalli-Sforza took the package as comprising domesticated
cereals and animals, ceramics, and so on, various authors showed that
these items rarely moved together, except in southeast and central Europe, and
that they might often be exchanged into Mesolithic communities (Thomas 1996,
Price 2000a). This work suggested that Ammerman and Cavalli-Sforza’s strategy
of using the presence of pottery, for example, to identify a settlement as “Neolithic”
(adopted to maximize the number of sites included) may have led them
to overestimate the Neolithic impact and the uniformity of its spread (Zvelebil
1986). Although Ammerman & Cavalli-Sforza (1984) did emphasize that their
rate was merely an average, the whole thrust of the wave of advance model was
to impose uniformity on the process. Archaeologists now widely agree that the
process was very heterogeneous and that there is no evidence in the archaeological
record for large-scale continent-wide immigration (e.g., Pluciennik 1998, Zvelebil
2000). Indeed, detailed studies of northern Europe suggest, in fact, that the Neolithic
developed in many areas over a very long period of time and did not arrive
in the Baltic region, for example, until well into the Bronze Age or even the Iron
Age (Zvelebil 1993, 1998b).
Ammerman and Cavalli-Sforza’s coupling of Neolithic populations with high
population density and Mesolithic ones with low density has been another casualty
of further research. Despite the enormous growth potential of agricultural
populations—the rationale for the wave of advance—the archaeological and palynological
records suggest that this was never remotely approached during the
Neolithic itself. Early Neolithic communities in Europe are likely to have been
quite small-scale and mobile (Willis & Bennett 1994; van Andel & Runnels 1995;
Roberts 1998, pp. 154–58). By contrast, lacustrine, riverine, and coastal Mesolithic
communities are likely to have been home to affluent, complex foraging communities,
with much greater population densities and a much higher degree of sedentism
than used to be assumed. Ethnographic comparisons with Australian aborigines,
or the southern African Khoisan-speaking hunter-gatherers, for example, seem
to have been inappropriate for these parts of Mesolithic Europe; comparisons
with Northwest Coast Native Americans, for example, now seem more suitable
(Zvelebil 1986). In this light, the Mesolithic should not be regarded as a mere prelude
to the Neolithic revolution but as a western alternative to it. The situation in
the forested interior of central Europe, where the Neolithic is known to have spread
extremely rapidly, is much less clear: Mesolithic remains are still scanty in these
areas. However, it is curious that in this particular case—in which colonization is
almost universally accepted to be the best explanation of the archaeological record
(the Bandkeramik or LBK horizon)—the spread was in fact far more rapid than
the average rate estimated by Ammerman & Cavalli-Sforza (1984) (Gronenborn
1999, Bogucki 2000, Price 2000b). The same is true for the appearance of Cardial
ware in the western Mediterranean, which is also thought likely to represent a case
of colonization, in this case by sea (Barnett 2000; Zilh˜ao 2000, 2001).
WIDGETIZE
How to use Quote function: