Ann. Hum.Genet.

(1991), 55, 51-67 51

Printed in Great Britain

A genetic reconstruction of the history of the population

of the Iberian Peninsula
J. BERTRANPETIT' AND L. L. CAVALLI-SFORZA2
Laboratori d 'Antropologia, Facultat de Biologia, Universitat de Barcelona,
Diagonal 645, 08028 Barcelona, Spain
Department of Genetics, Stanford University, Stanford, C A 94305, U S A

SUMMARY

The genetic patterns detectable in human populations of the Iberian Peninsula are shown by
means of 'synthetic genetic maps ', i.e. geographic maps of the highest principal components
(PC) of gene frequencies. This method of analysis separates independent patterns of the genetic
landscape, which hopefully represents different, major evolutionary events of the past. Among
these are clines established by ancient important migrations, and local differentiations of
populations due to barriers responsible for relative isolation. Only events of some magnitude
from a demographic point of view, involving populations having initially definite genetic
differences are detectable by the method. For this to be true, the genetic consequences of these
events must not have been entirely smoothed out by later, prolonged genetic exchange between
neighbours ; but simulations have shown that long clines produced by major migrations can be
rather stable in time.
The first synthetic map, corresponding to the first PC, shows that the major difference in the
Iberian Peninsula is that between people originally of Basque and non-Basque descent. The
recession in time of the boundaries of the Basque-speaking area seems correlated with the
progressive genetic dilution of the Basque genotype in modern populations, as we move away
from the centre of the Basque area. Clearly there must have been a close relationship in the
progressive loss of the Basque language and increasing genetic admixture with neighbours.
Most probably, Basques represent descendants of Paleolithic and/or Mesolithic populations
and non-Basques later arrivals, beginning with the Neolithic.
The second synthetic map is correlated with early Neolithic infiltrations from the eastern
edge of the Pyrenees. It has been shown by archaeologists that, in some areas, early Neolithics
lived side by side a t overlapping dates with well developed Mesolithics. The demographic
impact of Neolithic farmers versus Mesolithics, and therefore their genetic influence was thus
less important in the Iberian Peninsula than in Central Europe.
The third synthetic map shows a correlation with the linguistic and historical duality
between the Atlantic and the Mediterranean fringes, which developed in the first millennium
B.C.and was probably determined, to some extent, by infiltrations through the Pyrenees of the
Urnfield cultural elements as well as by several other later events.

INTRODUCTION

The reconstruction of the history of human populations has recently received valuable
contributions from a wide range of disciplines each of which, taken in isolation, may be unable
Author for correspondence: <J. Bertranpetit.
4- 2
52 J. BERTRANPETIT
AND L. L. CAVALLI-SFORZA

to supply conclusive results. In this very desirable multidisciplinary approach, the relative
value of genetic information is being increasingly recognized. For instance, Renfrew (1988; see
also Bradley, 1989) accepts the usefulness of genetic data for testing the wave-of-advance model
for the spread of the Neolithic in Europe (Ammerman & Cavalli-Sforza, 1971, 1973, 1984),
which he used in his discussion of the connections of archaeology and linguistic aspects of the
spread of the Neolithic and of Indo-European languages (Renfrew, 1987).
In a study of the spread of anatomically modern man, it was suggested that genetic,
archaeological and linguistic data are but reflections of the common, underlying history of
human populations (Cavalli-Sforza et al. 1988).
I n recent years, several methodologies have been developed to dig deeper into the history of
population from a genetic point of view. Genes of modern populations still echo a remote past
which may be uncovered to reconstruct their history through many different approaches,
ranging from the frequency of rare alleles to the analysis of phylogenetic trees and to the
detection of clines and isolates. Clines due to ancient migrations can be detected by the
geographic display of principal components of gene frequencies as shown by Menozzi et al.
(1978),who applied the method to the analysis of genetic patterns in Europe. By this technique
(' synthetic genetic maps ') they detected gradients believed to be due to the spread of Neolithic
farmers from the Near East, as confirmed also by independent techniques of analysis (Sokal &
Menozzi, 1982), and to other putative ancient migrations. Simulations have shown that ancient
clines are more stable over time than is ordinarily believed (Rendine et al. 1986; see also Jones,
1989).
The method of synthetic maps has the advantages of visualizing genetic landscapes based on
as large a number of genes as are available, from which those responding to known local effects
of natural selection are preferably eliminated. The total multidimensional genetic landscape is
partitioned into a variety of uncorrelated maps representing independent evolutionary events,
of which only those that have had a deeper demographic impact are likely to emerge. Here we
apply this method to data from Spain and Portugal.
The Iberian Peninsula represents the south-western edge of Europe, and has been exposed to
several distinct cultural worlds : European, African, Mediterranean and Atlantic, whose relative
importance has been for a long time and continues to be a matter of controversy. African
influence, for example, is nowadays seen to have been very slight from Mesolithic to historic
times, a view quite different from that held by prehistorians of the mid-century.
The Pyrenees are a natural barrier with northern Europe, but in some cases they have been
a true pathway for north-south interactions. The Basque population, well known for its
distinctiveness from a genetic and linguistic point of view, occupies the western edge and gives
us a first insight into the existence of a marked regionalization, detected in the Iberian
Peninsula as early as Upper Paleolithic and still present as cultural, genetic and linguistic
diversity.

DATA AND METHOD

I n this work the methodology used to study the European (Menozzi et al. 1978) and Italian
(Piazza et al. 1988a) genetic structure is applied to the Iberian Peninsula.
The analysis is based on all available gene frequencies for blood groups, enzymes and proteins
in the Iberian Peninsula, that is, Spain and Portugal. No information exists for Andorra. A
 Population history of the Iberian Peninsula 53

Table 1. Genetic systems used to generate the synthetic maps

Symbol Genetic polymorphism Alleles Samples
Blood groups
ABO ABO 171
A, A, ABO 34
RH Rhesus 67
CDE CDE (Rhesus) 37
MN MN 58
K Kell 23
FY Duffy 15
P P 16
LE Lewis 15
Proteins
HP Haptoglobin 42
PI a-1-antitrypsin 18
GC Group specific component I7
TF Transferrin 14
C3 Complement component 3 14
Enzymes
ACPl Acid phosphatase 1 3 21
ADA Adenosine deaminase 2 14
PGD Phosphogluconate dehydrogenase 2 I2
GLO 1 Glyoxalase I 2 20
AK1 Adenylate kinase 1 3 I1
PGMl Phosphoglucomutase 1 3 16
Total 54 635

significant part of the work published to date has appeared in journals and proceedings of local
diffusion. So, although the compilations by Mourant et al. (1976) and Tills et al. (1983) have
scarce information on Spain (including Basques) and Portugal, the actual available data are
much more abundant.
For this global study the islands (Balearic, Canaries, Azores and Madeira) are not considered,
and the genetic systems are restricted to those tested in a minimum of 11 different locations
(Table 1) reasonably scattered throughout the Peninsula. It is worth noting that the south-
east is less well represented than other regions, a fact to be considered in the interpretation.
Moreover, G-6-PD deficiency has been excluded because of its selective meaning, as has been
PTC tasting because its genetic basis is not completely understood.
In all, the analysis is based on 635 population studies, involving 20 genetic systems with 54
alleles, 34 of them independent (Table 1). In the polymorphisms with varying number of alleles
in different studies, these have been restricted to those more widely available and therefore
information on subtypes of various proteins is not used. For the two most widely studied
systems (ABO and Rh) much information is available, both for three and four alleles in the first
case, and for two and eight in the second. These have been treated separately, as redundant
information is automatically eliminated in the analysis. Alleles with 0 frequency in all cases
(such as CdE) are deleted. Unfortunately, not enough information is available for HLA to be
used.
Geographical coordinates have been assigned to each sampled location. Samples described
simply as ‘Spanish’ or ‘Portuguese’ without indication of their more specific geographic origin
have not been included.
54 J. BERTRANPETIT
AND L. L. CAVALLI-SFORZA

Table 2. Importance of the alleles in determining the three main principal components
illustrated by their correlation coegicients (positive or negative) with each of them
(Inside each range of values, alleles are listed in decreasing order of their correlation coefficient. x, rare
allele)
1st principal component
r > = 0.8
+: FY*a ADA*2 ACP*c
- : FY*b ABO*O ADA*1
08>r> =07
+: ABO*A LE(a-b-) RH*CDe
- : RH*d RH'*cde TF*B
07>r> =06
+: AK1*1 ABO*B
- : ACPl*b AK1*2
2nd principal component
r> =07
+ : PGM1*3 PI *x PI*Z
- : none
07>r> =06
+: P*2
- : PI*M P*l ACPl*a
0.6>r> = 0 5
+: GLO1*1 RH*cDE
- : GLOl*2 PGM1*1 RH*CDE PGMl*2
3rd principal component
r > = 0.5
+: PGM1*1 K*k
-: c3*3 K*K
05>r> =04
+: AK1*2 HP*2 C3*1 GLO1*1 RH*cDE RH*cDE
- : AK1 *X HP*1 AK1*1 PGM1*2 GL01*2 PI*S

Each gene frequency generates a surface on the geography (a genetic landscape), but the
geographical locations of samples vary from one system to another. In order to have a complete
set of variables to use for the multivariate analysis, a standard genetic landscape for each allele
has been constructed by estimating the gene frequency a t the intersection points of each
meridian and parallel (going from 36" t o 42" of latitude north and from -10" to +4O of
longitude). For 63 of these points falling on land, a geographical map of the 54 gene frequencies
has been calculated.
Several numerical methods have been described to handle the problem of mapping
observations to a surface (Piazza et al. 1981). In this case a smoothing technique using
Shepard's (1968) formula has been used to generate the complete data matrix.
Principal component analysis has been applied, allowing us to replace, for each location, the
original gene frequency vector by a new vector of factors, which are linear functions of the
original frequencies and are uncorrelated. A known and decreasing percentage of variation is
explained by the successive factors, the geographic variation of each of which can then be
examined separately.
To better visualize the results, a geographical contour map has been constructed for the first
three factors using six intensities of shading. Each of them may be called a synthetic genetic
 Population history of the Iberian Peninsula 55

Fig. 1 . First principal component of gene frequencies in the Iberian Peninsula from 54 alleles (34of
them independent) of 20 human loci. The percentage of variation explained by this factor is 27.1 Yo.

Fig. 2. Second principal component of gene frequencies in the Iberian Peninsula. The percentage
of variation explained by this factor is 1 4 5 % (and 41.6% accumulated).

landscape as it gives information on several genes a t once and forms a real landscape of hills and
valleys. The gradients shown are the most informative genetic gradients ; they include
information from all genes, from some more than from others, as shown by the correlations of
the original gene frequencies with the factors (Table 2). The direction of the gradient is
arbitrary, so the shading gradation could be inverted.
56 J. BERTRANPETIT
AND L. L. CAVALLI-SFORZA

Fig. 3. Third principal component, of gene frequencies in the Iberian Peninsula. The percentage of
variation explained is 12.3 % (and 53.9 % accumulated).

Figures 1-3 show the results for the three main factors. The first explains 27-1YOof the total
variation, the second 145% (and 41.6% accumulated), and the third 12.3% (and 53.9%
accumulated). Although it is difficult to exclude completely the influence of selection on the
observed patterns, the number of genetic systems involved gives some guarantee that the
gradients must mainly reflect the genetic history of the population, in terms of migrations and
local population diversification and relative growth.

INTERPRETATION O F THE SYNTHETIC MAPS

First principal component : a pre-Neolithic differentiation

The first principal component shows the difference between the Basque population and the
rest of the Iberian Peninsula, with a clear gradation in all directions. The highest differences are
found both with the western and eastern edges.
The first isopleth (isogenic curve) coincides fairly well with what is now called the Basque
Country and the Basque-speaking region. Although the area of distribution of the Basque
language has been narrowing in historic times (Allihres, 1986 ; Echenique, 1987 ; see discussion
thereafter), the area remaining today clearly shows the coincidence of cultural and genetic
preservation of very ancient characteristics.
The genetic difference of the Basques has been well known since the classical studies on the
Rh blood groups in the 40s (Etcheverry, 1945 ; Mourant, 1947 ; Chalmers et al. 1949) showed the
highest frequency of Rh - in the world, with marked differences from surrounding populations.
Mourant (1947) proposed a first interpretation in terms of the history of the population, in the
 Population history of the Iberian Peninsula 57
sense that, by isolation, modern Basques would maintain the genetic characteristics of a ‘proto-
European’ population, mostly or entirely Rh-, as opposed to later R h + migration waves.
When more genetic information was gathered, differences were also detected in other
polymorphisms (ABO, HLA, Bf) but they did not show parallel clines to the Rh gene.
Nevertheless, the genetic differentiation of the Basques was accepted and they are considered
as a genetic isolate in gene frequency compilations (Mourant et al. 1976 ; Tills et al. 1983). Genetic
distances have shown Basques as clearly differentiated from Spanish or French samples (Nei &
Roychoudury, 1982; Piazza et al. 1988b).
With the results of the synthetic maps for gene frequencies in Europe (Menozzi et al. 1978 ;
Ammerman & Cavalli-Sforza, 1984), the importance of a Neolithic wave of advance from the
middle-east into Europe was strongly suggested and has received further support (Rendine et
al. 1986; Sokal & Menozzi, 1982).Within this context a pre-Neolithic (Paleolithic or Mesolithic)
hypothesis of the origin of the Basque population, which would have survived the impact of
genetic admixture with later comers to a greater extent than other old European populations,
seemed adequate (Cavalli-Sforza, 1988).
In the present study, the Basque identity has been ascertained in a geographical perspective
and with a large number of genes, allowing us to undertake an interpretation of its meaning on
a detailed spatial basis. The Basque phenomenon can now be studied in the light of other
evidence related to the history of populations : archaeology, historical linguistics, population
history and physical anthropology.
The model proposed for this first component is that it shows the remains of a genetic
differentiation which was already present in Neolithic times. When and in what conditions
could a high genetic differentiation take place ? Three conditions are important : small size of
population, isolation (lack of immigration) and length of time. Although it is not possible to
demonstrate directly that these conditions held for the prehistoric Basque population, there are
many signs of an identity in the Basque area since Upper Paleolithic times, of a delay in
absorbing most cultural innovations, and of the maintenance of special cultural aspects, mainly
the language, of unquestionable antiquity.
The Basque-Cantabrian area (the north Atlantic coast of Spain, between the mountains and
the sea, including the western Pyrenees) is the richest region of Iberia in known sites from
Solutrean to Magdalenian times in Upper Paleolithic and mainly in Mesolithic. In the latter, the
regionalization of the Iberian Peninsula is already clear : there is a contrast between
Mediterranean and Atlantic zones, with two lasting and isolated parts in the latter, the Basque-
Cantabric and the Central Atlantic, around the Tagus river mouth. I n the Mesolithic the
Azilian culture, with a microlaminar industry in silex, is a continuous development of the last
Magdalenian period and spans, in the Iberian Peninsula, only the Basque-Cantabric region
(Tarradell, 1980) (Fig. 4).
Today the Mesolithic is no longer seen as a starvation period of hunter-gatherers opposed to
the prosperity of the farmers of the Neolithic (Zvelebil, 1986). In fact, as already proposed by
Gordon Childe, the very success of the western European Mesolithic appears to have
significantly delayed the adoption of crop cultivation and husbandry (Lewthwaite, 1986).
Population densities calculated in hunter-gatherer societies range from 0.01 to 9 5 person/km2
(Hassan, 1981), but the upper limit refers to situations with extraordinarily high availability
of resources. When considering temperate climates, most values are between 0.05 and 1, the
58 J. BERTRANPETIT
AND L. L. CAVALLI-SFORZA

Fig. 4. Distribution of the main Mesolithic areas. 1 , Basque-Cantabrian;2, Tagus river mouth.
The dots in the Mediterranean region are a rough indicator of density of sites.

lower limit being most frequent. Estimates of the prehistoric population of England are from
0 0 3 to 0.07 person/km2 (Clark, 1972) and for the north-African Capsian around 0-3 person/km2.
Under these conditions considerable local genetic drift is expected.
The Neolithic period, directly following Mesolithic, appears later in the Cantabrian region
(3500 B . c . ) . The foraging economies persist longer; the decline of hunted species is only
apparent after the Eneolithic-Bronze transition, c. 1900 B.c., and there is no primary or indirect
evidence for cereal cultivation until the Bronze Age. These facts fit into the transition to the
farming model of Atlantic Europe, where a delay in the shift to farming occurred among the
more sedentary and socio-economically more complex hunter-gatherer societies (Zvelebil &
Rowley-Conwy, 1986). Sometimes the Basques have been regarded as a Mesolithic population
with no contact with new technologies until the Bronze Age (Tarradell, 1980) but more recent
evidence describes a slow and autochthonous use of new resources. The Mesolithic population
of the Cantabrian area thus probably grew slowly but for a long time (Zvelebil & Rowley-
Conwy, 1986) with little or no immigration from the outside.
Linguistics also supports the hypothesis of a very ancient autochthonous population in the
Basque region. The Basque language is a true isolate in Europe, as it is not related to any other
extant language and does not belong to the Indo-European family (Ruhlen, 1987). A
relationship suggested is with south Caucasian languages, of which Georgian is the best known
(Allikres, 1986). If this relationship was demonstrated to be genetic (in its linguistic meaning,
i.e. phylogenetic) this would lead to the conclusion that both groups are peripheral persistences
of a linguistic family that once spread over a wide area in pre-Neolithic times. This would be
true especially (but not only) if we accept the hypothesis that the first spread of Indo-European
languages took place with the Neolithic expansion (Renfrew, 1987, 1988).
 Population history of the Iberian Peninsula 59
Unquestionably all historical data attest to the ancient presence of the Basque language in
the region where it is spoken today. The Latin writer Mela, in the first century AD. (Garcia-
Bellido, 1947) wrote about certain people and rivers whose names could not be expressed in his
language, referring to non-Iberic and non-Celtic populations. And we know from different
sources (written documents, toponymy) that the linguistic area was, a t that time, much broader
in all directions than it is today (AlliBres, 1986; Badia, 1984; Echenique, 1987). At the
beginning of our era it corresponded approximately to the third isopleth of Fig. 1 and a similar
situation applies to the Basque area of southern France. Around the Ebro river, abundant
Basque toponymy is found in places where Basque was never spoken in historical times
(Beltran, 1978); in the eastern Pyrenees, a Basque dialect was spoken until the Middle Ages
(Badia, 1984, among others).
The Basque language is thus a clear relict from very ancient times, which is further proof of
isolation and may also have played an important role in maintaining the genet,ic (breeding)
isolation (Collins, 1985 ; Cavalli-Sforza, 1988).
Physical anthropology provides a further test of the hypothesis that Basques are a
differentiated group inside the Iberian Peninsula thus extending the genetic conclusions, and
may help to trace it back in time and space.
The morphology of the Basques has attracted the attention of several physical anthro-
pologists. A t the beginning of this century, a clearly different typology was described among
them and called western-pyrenaic (Atlanto-Mediterranean for the French anthropologists).
This typology was recognized both in living individuals and skeletal remains, allowing the
study of its spatial and temporal spread. It has recently been confirmed and refined through
multivariate analysis (de la Rua, 1985) allowing, for the first time, a populational interpretation
with typological implications.
The Basque characteristics have repeatedly been interpreted as paleomorphic, that is to say,
related to Upper Paleolithic morphology, which, represented mainly by the Cro-Magnon
typology, was predominant until the Mesolithic, after which the modern Mediterranean forms
began to predominate. It is worth remembering that Cro-Magnon itself is located in the Basque
area of France as defined by toponymy. So it is generally accepted that, a t the morphological
level, the Basques represent the persistence of very ancient characteristics (Marquer, 1963 ; de
la Rua, 1985, 1988). Ancient data, however, are very limited until the much later Megalithic
times.
The Mesolithic population of the Tagus river mouth in Portugal is the best known in Iberia
thanks to the remains of the Muge site (Roche, 1972) and is characterized by the high frequency
of paleomorphs, being, no doubt, the most differentiated group among the prehistoric and
historic series gathered until now (Garralda & Mesa, 1984). These ancient characteristics do not
survive among local modern human populations except those ih the Basque Country, showing
the incidence of later replacement, which also seems general in the rest of the Iberian Peninsula.
But the discontinuity cannot be as clearly established in other places as around Muge because
of the complete lack of Mesolithic or earlier series.
In all, physical anthropology supports the hypothesis of ancient persistence of populational
characteristics in the Basque Country and a replacement in other areas of the Iberian
Peninsula, where, as will be seen later, little or no change is observed from Neolithic to modern
times (Garralda, 1986).
60 J. BERTRANPETIT
AND L. L. CAVALLI-SFORZA

Fig. 5. Area of spread of the first Neolithic, with cardial pottery. The sites referred to in the text are:
1, Or; 2, Sarsa; 3, Cocina.

In conclusion, the first factor extracted from the genetic data of modern populations, in
conjunction with linguistic, archaeological and morphological observations, suggests that, in
Iberia, the Basques were less changed by subsequent admixture with later arrivals, while in
other areas of Mesolithic development, this dilution has largely altered the initial genetic
pattern. The genetic uniqueness of Basques is a consequence of genetic drift due to a long period
a t relatively low population densities in late Paleolithic and Mesolithic. The initial genetic
divergence thus arisen must have been maintained by the paucity of exchanges with later
arrivals, helped by endogamy tied to language and customs. Only a relatively short cline of the
first factor is visible in Iberia, probably due to later interchange with neighbours. The
uniqueness of Basques was also visible (detailed data not given) by calculating first factor
values for the gene frequencies of several external populations (central and northern Europe,
Sardinia, North Africa) which all show non-Basque values in an even more pronounced way.

The second component : the Neolithic wave of expansion

The second factor (Fig. 2) points to the genetic divergence of the north-east of the Iberian
Peninsula (Catalonia) from the central and south central parts. The first isopleths follow a
north-south direction, and the fourth turns towards the west and south to encircle the centre
(Meseta) and south-centre (Andalusia). The centre shows the greatest difference from the
Catalan area.
Evidence from archaeology (Tarradell, 1980 ; del Rincbn, 1987),linguistics (Corominas, 1976)
and also physical anthropology (Turbbn, 1984, 1989) points to the darkest area of Pig. 2 as the
most important pathway through the Pyrenees of influences from South Central France.
In prehistoric times prior to the Roman occupation of Iberia two waves of influence through
the eastern Pyrenees are recognized and clearly accepted.
 Population history of the Iberian Peninsula 61
1. Neolithization (del Rincdn, 1987; Tarradell, 1980) (Fig. 5). No doubt there is a primary
expansion of the Neolithic in the Mediterranean fringe, probably expanding from the north-east
edge to the south, close to the coast. A detailed map of the spread of Neolithic has not yet been
traced by archaeologists because of the poverty of informative settlements and of absolute
datings. /

2. Migrations from central Europe at the end of the second millennium B.C. and the first half
of the first, which will be described in more detail in the next section.
Apart from these two clear contacts with northern Europe, other migrationist and/or
diffusionist movements have been postulated, but they are no longer accepted as major
migrations. Arguments about these movements include a whole range of observations made a t
the cultural level and historical facts ranging from those as ancient as Megalithism
(characterized by megalithic monuments) to as recent as the migrations of Visigoths or Francs.
For none of them is there a consensus, but it is unlikely that they involved important
population movements with serious demographic impact.
Evidence is largely in favour of Neolithization being more important in determining the
second factor. The genetic landscape can be influenced only by factors with a strong
demographic effect, and there is no doubt that the change to an economy of food production
characteristic of the Neolithic has a potentially major impact (Hassan, 1981 ; Champion et al.
1984). The Neolithic, as shown in the map, may have spread through the Pyrenees or along the
coast, from Liguria through Southern France (the coastal genetic picture could have been
changed by later genetic events), as shown by the distribution of the first ceramics (Fortea &
Marti, 1985). So even though not proved by absolute dating, it is likely that the first
introduction of Neolithic was through the north-eastern edge.
The spread of Neolithic in the Iberian Peninsula, however, has some peculiarities to be
discussed. Genetic data suggest a relatively smaller demographic impact of Neolithic people, as
can be deduced from the fact that the Neolithic does not markedly affect the first factor
(whereas it does in Central Europe) and that the second genetic factor, probably due to the
Neolithic, extends less widely to the south than indicated by archaeology (Fig, 5 ) .
An explanation can be found in the archaeological findings. Several sites confirm a long
coexistence of farmers and hunter-gatherers, as has been shown by the excellent excavations
around Valencia (centre of the Mediterranean coast), where in close sites and for overlapping
dates, full Neolithic is found (sites of Or and Sarsa among others, Fig. 5 ) close to the persisting
Mesolithic elements (site of Cocina, for example) (Fortea & Marti, 1985). The importance of the
Mesolithic substratum is, thus, well established.
These data suggest a model of neolithization in which a t the arrival of Neolithic there were
already strong concentrations of hunter-gatherers and they coexisted for a long time. The high
densities and high acculturation of Mesolithics under these conditions decreased the genetic
impact of Neolithics as compared with that observed in other areas. These facts could have been
even more important in the Atlantic fringe for climatic and ecological reasons. A similar model
to that of Zvelebil (1986) might be proposed, and would explain the relatively low effect of
Neolithization at the genetic level. Thus, while some archaeologists try to explain the Neolithic
entirely as an acculturation process (e.g. del RincBn, 1987), it seems more correct to view the
Neolithic in the Iberian Peninsula as having the mixed features of a demic (genetic) and a
cultural diffusion, with a preponderance of the latter further away from the point of origin.
62 J. BERTRANPETIT
AND L. L. CAVALLI-SFORZA

Fig. 6. Linguistic map of the Iberian Peninsula in the first millenium B.C. 1, Basque; 2, Iberian; 3,
Tartessian ; 4, Celtic languages (Indo-European). Tartessic culture is more ancient than Iberian and
in the eastern half of the area, the Turdetan culture (Iberian) developed, while the western became
Celtic.

Some authors discuss the potential importance of later Neolithic, that is to say, the
importance of a second generation of innovations. European societies were fundamentally
transformed by the arrival of these features, long after the advent of the Neolithic (discussed
in Sherratt, 1988). But he admits that the demographic increase was with the ‘first ’ Neolithic
(Sherratt, 1988). I n fact, the archaeological record does not offer evidence for significant
migrations into central, western and southern Europe a t the end of the Neolithic (Renfrew,
1988).
There is a delay of neolithization of the Meseta (the centre; del RincBn, 1987). This explains
why the pole of the second factor opposite to that of the eastern pole is found a t the centre. The
first occupation of the Meseta may have happened in the Calcolithic (early Bronze Age), by
peoples associated with the practice of collective burials. Recent evidence, however, seems to
indicate that Neolithic could have begun earlier.
I n contrast, there is no evidence of a strong demographic impact of post-Neolithic
immigrations. Two independent waves have classically been recognized (Maluquer, 1974, and
others), one around the change from second to first millennium B.c., with the introduction of
the Urnfield cultures, and the later Iron Age contacts, with the introduction of the Celtic
languages. Only the first may have had a direct external influence through migration (Ruiz-
Zapatero, 1983).
In fact the Urnfield expansion took place through the oriental edge of the Pyrenees and
intrusive elements are found in the Catalan Urnfield cultures. With a detailed analysis of the
sites, Ruiz-Zapatero (1983) corroborates the previous hypothesis by Almagro-Orbea of a
cultural influence from the north for the ancient Urnfield cultures ( l l W 9 0 0 B.c.), but by no
 Population history of the Iberian Peninsula 63
means a mass-migration : it is no longer possible to support the traditional idea of tumultuous
arrival of people, but only of a small ethnic contribution or the penetration of reduced supra-
family groups. And these immigrants met the well-established Bronze Age population of
Catalonia, which incorporated the incoming cultural elements. The demographic and therefore
genetic impact may, thus, have been small.
Physical anthropology has also centred its attention on the dark area of factor 2, but not
from a populational point of view. All studies that consider the temporal evolution of skull
morphology in the Iberian Peninsula point out, except for the already mentioned difference of
the Basques, the morphological continuity from Neolithic to modern times (Garralda & Mesa,
1984, 1986; Garralda, 1986; T u r b h , 1989), in spite of cultural changes and external influences.
Without direct evidence of the language spoken by Neolithics, linguistic evidence is of little
help. Further discussion on this point will be given in the next section.

Third factor :Mediterranean versus Atlantic

Factor 3 shows the basic feature of the well-established regionalization of the Iberian
Peninsula according to archaeology (see Fig. 3) : the Mediterranean as opposed to the Atlantic.
This duality, described as clear differentiation of specific cultures and languages, takes its full
sense in the dichotomous division of the entire Peninsula (except for small parts, like the
Basque region) into Iberian peoples in the Mediterranean (speaking a non-Indo-European
language) and Celtic or Hispano-Celtic (sometimes referred to as Indo-Europeans) in the centre
and Atlantic fringes. This duality was fully developed in the last half of the first millennium B.C.
and was noted a t the arrival of the Romans (beginning in 218 B.C. in Emporion, Catalonia)
whose geographers made a detailed picture of it (Garcia-Bellido, 1947).
The linguistic picture is shown in Fig. 6, and is based on the distribution of toponymics. Many
other linguistic features show essentially the same map (Tovar, 1949). Iberian is not highly
documented, but it is probably a non-Indo-European language of obscure origin, possibly very
ancient. Relationship to Basque has been suggested but is not very likely (Echenique, 1987).
Indo-European languages spoken in the Atlantic moiety belong to the Celtic group and were
introduced from France. Tartessian probably is a non-Indo-European language related to
Iberian but more ancient, of which very little is known. There are also claims for the spread of
a pre-Celtic language (called Sorotaptic, with affinities with several Indo-European languages ;
Corominas, 1976), some traces of which are recognizable as a substratum in extant Catalan. It
is believed to have spread downstream via the Duero river to northern Portugal.
Archaeology of this period shows a distribution of artifacts tied especially to cemeteries
known as Urnfields. Several contacts through the Pyrenees between 1100 and 500 B.C. have
been suggested (Ruiz-Zapatero, 1983). Ancient Urnfields ( 1 100-900 B.c.) are located near the
eastern edge of the Pyrenees. Their location indicates a path of descent mostly along the river
Segre (a northern tributary of the Ebro) and also along the coast, towards the south.
For more recent Urnfields (90&700 B.c.) there is an autochthonous development of the
groups that migrated in the first phase, with some cultural influence through both eastern and
western Pyrenees. Finally, at the beginning of the Iron Age (700-500 B.c.) one can also
recognize influences from the Mediterranean coast towards the inside, in correspondence with
the settlements of Phoenician colonists (along most of the coast) and the less numerous Greek
colonists (Ruiz-Zapatero, 1983).
64 J. BERTRANPETIT
AND L. L. CAVALLI-SFORZA

The introduction of incineration rites typical of the Urnfield culture destroyed the
information that could have been made available through physical anthropology. Cremation
was general in the Iberian Peninsula at the time of Roman occupation.
Today archaeologists view this diffusion as entirely cultural, not as a mass migration. Some
immigrants must be assumed, however, to have introduced Indo-European (Celtic and possibly
other) languages from Central Europe, but the demographic contribution of these immigrants
may have been relatively small; the impact of the new cultures, however, may well have
determined local population growth and increased exogamy a t least within areas of comparable
culture. Archaeological evidence for demographic growth is available (Ruiz-Zapatero, 1983) ;
McEvedy & Jones (1978) estimate an important population increase between 400 and 200 B . c . ,
the highest for prehistory and until 1000 A.D.
An unsolved archaeological and linguistic problem is that of the passage of Indo-European
speakers from the eastern edge of the Pyrenees (documented archaeologically) to the Indo-
European speaking Atlantic area. I n the southern part of the Basque area (Alava) there are
indications of influence by Indo-European speaking intruders. But the postulated ‘ corridor ’
from east to west, south of the Pyrenees, is very unclear. The scarcity or absence of traces of
the passage may perhaps be explained if the passage was rapid.
The development of Celtics on the western side and the more vigorous increase of Iberians in
the East, with relatively little exchange between them, may be responsible for the duality
between East and West observed in the genetic map, even without assuming important genetic
contributions from external people. Some exchange a t the boundaries may of course have
occurred and the traditional name of Celtiberians, given to a population located in the northern
central part, suggests that this happened, a t least to some extent, and is linguistically well
documented, so that no really sharp boundary is expected.
In spite of differences between the migratory events associated with Neolithic and post-
Neolithic times, the assignment of the Neolithic to the second and of more recent events to the
third may have to be revised, but on the basis of present knowledge our choice seems easier to
defend.
The fourth and fifth factors also cover a reasonable portion of the variance (9.6YOand 9.0 YO);
their geographic maps, however, present several peaks and troughs which are more difficult to
interpret and inevitably of weaker significance than those corresponding to the first three. No
factor examined seems to involve the South-East of the Iberian Peninsula, the region most
heavily occupied by Arabs (Moorish), but also less studied genetically. The high population
density of the Iberian Peninsula at the time of Arab occupation (end of the seventh
century A . D . ) makes it unlikely that the Moorish genetic contribution (asserted by some other
authors, like Reyment, 1983) would be easily detectable by this method.

CONCLUSION

Analysis of synthetic maps by principal components of gene frequencies (or factors, as they
are called .here) can detect gradients and areas of genetic differentiation and separate
independent genetic events of the past from each other. One way in which a particular
‘landscape’ may arise is by migration of one expanding population into an area already
inhabited by people of different genetic background. The smoothing of these gradients by later
 Population history of the Iberian Peninsula
exchanges between neighbours is slow and it has been shown by earlier simulations that
gradients have a high persistence.
Another way in which marked genetic heterogeneity may arise is when local populations,
differentiated genetically from their neighbours because of early genetic drift or for other
reasons, undergo substantial demographic growth, usually due t,o some technological
development (autochthonous or imported). Numerical increase of the population tends to
freeze, a t least for some time, the genetic divergence of that population from its neighbours.
The analysis of genetic data from the Iberian Peninsula shows that the major component of
genetic variation reflects the difference between early inhabitants through Paleolithic and
Mesolithic times, whose descendants are represented today in the Basque region, and later
arrivals in Neolithic or Post-Neolithic times. The isogenic lines follow quite faithfully modern
(or, for lower isogenic curves, earlier) boundaries between the Basque area and the rest of the
Peninsula.
The northern part of the Iberian Peninsula is among the areas most remote from the origins
of Neolithic farmers in the Middle East (Ammerman & Cavalli-Sforza, 1971, 1973, 1984;
Mbnozzi et al. 1978; Renfrew, 1987, 1988). It is also relatively protected from incomers by the
Pyrenees. Prolonged cultural preservation of successful food economy, of ancient language and
customs have helped preserve the Basque identity by a relative closure to later comers, which
was both cultural and genetic. Thus, in a way, Basques represent the opposite pole to that of
incoming Neolithic farmers.
All archaeological and historical linguistic evidence strongly reinforces the notion that
Basque genetic differentiation could not have taken place later than assumed here, and was
largely maintained in spite of subsequent events, which affected other parts of the Peninsula
more deeply. This is thus a clear case of long term preservation of a genetic cline, which was
already described in the whole of Europe (Menozzi et al. 1978). Here we see an enlargement of
this picture regarding the Iberian Peninsula.
The geographic distribution of the second factor shows a correlation with the entry of the
Neolithic, whose influence is here less marked than in the rest of Europe, for the reasons just
mentioned, and also because in the area most influenced by early Neolit,hic, the Mediterranean
coast, Mesolithic people had reached a relatively greater development than in other parts of
Europe.
The third synthetic map is in correlation with well known archaeological and especially
linguistic patterns of divergence between the Atlantic and the Mediterranean parts, summarized
by the dualism between the Celtic (Atlantic) and Iberian (Mediterranean) cultures. This duality
may have been established primarily by events of the first millennium B . C . Thus the observed
order of the first three factors reflects in this case the chronological order of events, which is not
surprising if we consider the temporal growth of the population size (McEvedy & Jones, 1978)
which makes the genetic impact of demographic events more difficult.
We feel that these conclusions reinforce the view that correlating results from the genetic
analysis of human populations with information from archaeology, physical anthropology,
paleodemography , history and linguistics can lead to an improved understanding of the history
of human populations.
Most of this work has been carried out during the stay of J . Bertranpetit a t the University of Stanford thanks
to a Grant from the ‘Direccidn General de Investigacidn Cientifica y Tdcnica’.
5 H G E 55
66 J. BERTRANPETIT
AND L. L. CAVALLI-SFORZA

We are particularly grateful to P. Moral (University of Barcelona) and A. Abade (University of Coimbra) for
help in collecting the data and to P. Menozzi and 8 . Piazza for supplying already computerized gene data.
M. A. del Rincbn (University of Barcelona) offered endless help, valuable comments and discussion, which
also came from B. Marti (Museu Arqueolbgic de Valkncia), A. Badia (University of Barcelona), Elisenda Vives
(Patrimoni Artistic Nacional, Andorra) and other colleagues.
Robin Rycroft helped with the English manuscript.

REFERENCES"

ALLIERES,J. (1986).Les Basques, (3rd edn). Paris: Presses Universitaires de France.

AMMERMAN, A. J. & CAVALLI-SFORZA, L.L. (1971).Measuring the rate of spread of early farming in Europe.
M a n 6,674-688.
AMMERMAN, A. J. &, CAVALLI-SFORZA, L.L. (1973).A population model for the diffusion of early farming in
Europe. I n The explanation of culture change (ed. C. Renfrew), pp. 343-357. London : G. Duckworth &, Co.
AMMERMAN, A. J. & CAVALLI-SFORZA, L.L. (1984).The Neolithic Transition and the Genetics of Populations in
Europe. Princeton : Princeton University Press.
BADIA,A. M. (1984).L a formacid de la llengua catalana. Barcelona: Publicacions de I'Abadia de Montserrat.
BELTRAN, A. (1978).ProblemStica general de la iberizacibn en el Valle del Ebro. Ampurias 3840: 197-209.
BRADLEY, R. (1989).Archaeological cross-talk. Nature 338, 675-676.
CAVALLI-SFORZA, L. L. (1988).The Basque population and ancient migrations in Europe. Munibe (Antropologia
y Arqueologia) Suplemento 6,129-137.
CAVALLI-SFORZA, L. L., PIAZZA, A., MENOZZI,P. &, MOUNTAIN,J. (1988).Reconstruction of human evolution:
Bringing together genetic, archaeological, and linguistic data. Proc. Natl. Acad. Sci. U S A 85, 6002-6006.
CHALMERS,J . M. N., IKIN, E. W. & MOURANT,A. E. (1949).The ABO MN and Rh blood groups of the Basque
people. Amer. J . Phys. Anthrop. 7, 529-544.
CHAMPION,T., GAMBLE, C., SHENNAN, S. & WHITTLE,A. (1984).Prehistoric Europe. London : Academic Press.
CLARCK,J. G. D. (1972).Star Carr: a Case Study in Bioarchaeology. McCaleb module 10 in Anthropology.
Addison-Wesley Modular Publications.
COLLINS, R. (1985).The Basques. New York: Basil Blackwell.
COROMINAS,J. (1976).Entre dos llenguatges, vol 2. Barcelona: Curial.
ECHENIQUE, M. T. (1987).Historia linguistica vasco-romdnica. Madrid : Paraninfo.
ETCIIEVERRY, M. A. (1945).El factor rhesus; su genktica e importancia clinica. El Dia Midico 17,1237-1259.
FORTEA, J. & MARTI, B. (1985).Consideraciones sobre 10s inicios del Neolitico en el MediterrSneo espafiol.
Zephyrus 37-38, 167-199.
GARCIA-BELLIDO, A. (1947).L a EsparZa del siglo Z de nuestra era (segzin P . Mela y C. Plinio). Buenos Aires:
Espasa Calpe.
GARRALDA, M. D. (1986).Etnogenkse del Peuples Ibkriques (env. 1000 ~.c.-500A.D.). In Ethnogenese
europaischer Volker (ed. Bernhard & Kandler-PSlsson), pp. 187-207. Stuttgart : Gustav Fisher Verlag.
GARRALDA, M. D. & MESA, M.S. (1984).Variabilite morphologique dans la Phinsule Ib6rique: Epi-
pal6olithique-Age Ancien. Bull. M i m . Soc. Anthrop. Paris 1, skr. 14,207-220.
GARRALDA, M. D. & MESA, M. S. (1986).Morphological variability in the Iberian Peninsula: Middle-Modern
Ages. Int. J . Anthrop. 1, 97-106.
HASSAN, F. A. (1981).Demographic Archaeology. New York: Academic Press.
JONES, J . S. (1989).A tale of three cities. Nature 339, 17C-177.
LEWTKWAITE, J. (1986).The transition to food production: a Mediterranean perspective. In Hunters in
transition (ed. M. Zvelebil), pp. 53-66. Cambridge : Cambridge University Press.
MALUQUER,J. (1974).Prehistoria y Protohistoria. In Historia Antigua y Medieval, pp. 22-67. Barcelona :
Planeta.
MARQUER,P. (1963).Contribution a 1'6tude anthropologique du peuple basque et au problkme de ses origines
raciales. Bull. M i m . Soc. Anthrop. Paris 4,skr. 11, 1-240.
MCEVEDY, C. & JONES, R. (1978).Atlas of World Populatimz History. Harmondsworth: Penguin.
MENOZZI,P., PIAZZA, A. & CAVALLI-SFORZA, L. L. (1978).Synthetic maps of human gene frequencies in
Europeans. Science 201,786-792.
MOURANT,A. E. (1947).The blood groups of the Basques. Nature, 160,505.
MOURANT,A. E., K O P E ~A. , C. & DOMANIEWSKA-SOBCZAC, K. (1976).The Distribution of the Human Blood
Groups and other Polymorphisms. Oxford : Oxford University Press.
* The bibliography of this article does not list the references to the original papers containing gene frequency
information. It would be impossible to list them all here. However, a full list of references will be made available
by writing to the first author.
 Population history of the Iberian Peninsula
NEI, M . & ROYOHOIJDTJRY, A. K. (1982).Genetic relationship and evolution of human races. Evol. Biol. 14, 1-59.
PIAZZA, A., CAPELLO,N.. OLIVETTI,E. & RENDINE,S. ( 1 9 8 8 ~ )A. genetic history of Italy. Ann. Hum. Genet.
52, 203-213.
PIAZZA, A , , CAPELLO,N., OLIVETTI,E. & RENDINE,S. (1988b). The Basques in Europe: A genetic analysis.
Munibe (Antropologia) Suplemento 6, 169-177.
PIAZZA, A , , MENOZZI, P., CAVALLI-SFORZA, L. L. (1981). The Making and Testing of Geographic Gene-
Frequency Maps. Biomrtrics 37, 635-659.
RENDINE. S., PIAZZA: A. & CAVALLI-SFORZA, L. L. (1986). Simulation and separation by principa,l components
of multiple demic expansions in Europe. A m . Nat. 128, 681-806.
RENFREW, C. (1987).Archaeology and Language. The Puzzle of Indo-European Origins. London : Jonathan Cape.
RENFREW, C. (1988). Book Review, Archaeology and Language. Author's PrBcis. Current Anthropology 29,
437441,
REYMENT,R. A. (1983). Moors and Christians: an example of multivariate analysis applied to human
blood groups. A n n . Hum. Biol. 10, 505-521.
DEL RINCON, M. A. (1987).El neolitico y el calcolitico en la Peninsula IbBrica. In Laprotohistoria de Europa, pp.
354-411. Barcelona : Editorial Labor.
RocrrE, J . (1972). Le Gisrment Mekolithigue de iWoita do Sebastiao, Muge, Portugal. Second edition. Lisboa:
Instituto de Aka Cultura.
DE LA RUA,C. (1985).El craneo vasco: Morfologia y factores craneofaciales. Bilbao : Servicio de Publicaciones de
la Diputaci6n Foral de Vizcaya.
DE LA RUA.C. (1988). Revisi6n de 10s craneos prehist6ricos de Urtiaga (Guipuzcoa, Pais Vasco). Munibe
(Antropologia y Arquelogt'a) Suplemento 6, 26Y-280.
RUHLEN, M . (1987). A Guide to World's Languages. Volume 1 : ClassiJication. Stanford: Stanford IJniversity
Press.
RUIZ-ZAPATERO, G. (1983). Loa campos de urnas del LVE de la Peninsula Ibdrrica. Tesis Doctoral. Universidad
Complutense de Madrid.
SHEPARD, D. (1968).A two-dimensional interpolation function for irregularly spaced data. In Proceedings of the
1968 ACM National Conference, pp. 517-524. Princeton : Brandon Systems Press.
SHERRATT, A. (1988).Book Review. Archaeology and Language. Review. Current Anthropology 29, 458463.
SOKAL, R. R. & MENOZZI,P. (1982). Spatial autocorrelations of HLA frequencies in Europe support demic
diffusion of early farmers. A m . Nut. 119. 1-17.
TARRADELL, M. (1980). Primeras Culturas. In Historia de E'sparta, tom0 I , pp. 49-195. Barcelona: Editorial
Labor.
TILLS,I)., KoPE~!, A. C. &TILLS,R. E. (1983).Thedistributionof the human blood groupsandotherpolymorphisms.
Supplement 1. Oxford : Oxford University Press.
TOVAR, A . (1949).Estudio sobre las primitivas lenguas hispanicas. Buenos Aires : Facultad de Filosofia y Letras.
TIJRB~N D., (1984). Antropologia de las poblaciones prehistdricas e histdricas de Cataluiia. Trabajos de
Antropologia 19, 145-162.
TURBON,D. (1989).El passat biolbgic de l'home a Catalunya. Cota Zero 5 , 65-71.
ZVELEBIL, M. (1986). The Mesolithic context of the transition to farming. In Hunters in Transition (ed. M.
Zvelebil), pp. 5-15. Cambridge : Cambridge University Press.
ZVELEBIL, M. & ROWLEY-CONWY, P. R. (1986). Foragers and farmers in Atlantic Europe. In Hunters in
Transition (ed. M. Zvelebil), pp. 67-93. Cambridge : Cambridge University Press.