Post by zarahan on Apr 9, 2010 11:55:25 GMT -5
Below is a detailed critique of the European academy's handling of
African genetic data, particularly the work of Cavalli Sforza.
The author questions whether the European academy can be
trusted to render a balanced picture of African genetic diversity.
Among the criticisms is that data on African peoples is being
analyzed in ways different from that accorded European samples,
for example with a lesser level of detail in some analyses that
masks the picture on the ground.
quote:
African ethnic units are not bounded, homogeneous
monoliths either frozen in place since before a.d. 1492
or caromming around the continent like culture-bearing
billiard balls. They are dynamic entities, manipulated in
response to changes in the natural and social environment,
and their composition can change very quickly
indeed..
In part because of this lack of chronological control, the
procedures used by Cavalli-Sforza et al. to associate genetic
patterning with cultural events in the past are quite
unsystematic, involving attempts to fit secondhand
knowledge of archaeological, ethnographic, and linguistic
reconstructions into a Procrustean bed of genetic patterning.
Clark’s (1998) description of the “ransacking”
of European genetic data for meaning applies equally
well in the African case.
Some argue that activities like the Human Geonome
Project simply replicate existing ideological
models of current elites only this time couched
in the language of biotechnology and soothing
tones of "diversity." A typical example is
"Genomics, divination, racecraft," by STEPHAN PALMIÉ 2007.
A critique of the critique by Keita is also shown below.
Others raise the spectre of a clamed "bio colonialism"..
In seeking to patent the material they “mine,” the bio-prospectors
will have opened a sort of final colonialism, contained here in a
chiasmus: from control over the raw material of nature to control
over the nature of raw materials."
excerpt of critique from:
Scott MacEachern
Genes, Tribes, and African History. Scott MacEachern.
Current Anthropology Volume 41, Number 3, June 2000
SOME BLURBS:
quote:
- African ethnic units are not bounded, homogeneous
monoliths either frozen in place since before a.d. 1492
or caromming around the continent like culture-bearing
billiard balls.
- A number of the groups used in the analyses
in The History and Geography of Human Genes did not
exist as such five centuries ago, and the modes of human
interaction in Africa are not in any way reducible to a
list of technological innovations and mass migrations.
In part because of this lack of chronological control, the
procedures used by Cavalli-Sforza et al. to associate genetic
patterning with cultural events in the past are quite
unsystematic, involving attempts to fit secondhand
knowledge of archaeological, ethnographic, and linguistic
reconstructions into a Procrustean bed of genetic patterning.
Clark’s (1998) description of the “ransacking”
of European genetic data for meaning applies equally
well in the African case.
-Presumably, this patterning is explained by the admixture
of “Caucasoid” and “Negroid” populations in
northeastern Africa..
--These “Caucasoid” genetic contributions are thus invoked
as the primary explanation for the variation exhibited
in the first, second, and third principal components
but seem to refer to different sets of genetic
characteristics and thus presumably to different populations
of “Caucasoids.” These interpretations would
seem to lack much real explanatory value.
--The extent of this confusion of different data sources is
made evident by a map (1994: fig. 3.9.3) that purports
to show a pattern of Bantu expansion
based upon archaeological data in which most
of the linkages (between Nilotic-speaking peoples and
Central-Western Bantu or between Bane populations and
Northeastern Bantu, for example) are sustained by no
archaeological data at all. No other mechanisms for genetic
or linguistic dispersal are considered, and the con-
nection of these different types of data seems quite
unsystematic.
--Cavalli-Sforza et al. (1994:189–92) do not attempt to give
historical explanations for African genetic frequency variation
to the sixth principal component as they do for Europe.
-- The data given in this work indicate that such processes
may be quite common on the African continent. In general,
the correspondences between genetic relationships and those
based upon linguistic and “tribal”/ethnic affiliation do
not appear to be particularly close. Cavalli-Sforza et al.’s
(1994:182–85) discussion of these correspondences
abounds with perplexing relationships between different
groups of people—Nubians with Moroccans, Ubangianspeakers
with southwestern Bantu populations, Sara
speakers of a Nilo-Saharan language with Hausa on the
one hand and Biaka Pygmies on the other, and so on.
The “Caucasoid” genetic affiliations of the San may
prove to be another example.
--The Hausa, we should note, become Nilo–Saharans rather than
speakers of Afroasiatic languages in the principal-component maps
of Cavalli-Sforza et al.
-- Cavalli-Sforza et al. recognize
the peculiarity of a situation in which one ethnonym
encompasses two distinct genetic populations
and respond by recalling Murdock’s (1959:414–15) distinction
between pastoral Fulbe, who are supposed to be
“Caucasoid-like,” and settled Fulbe, who are “Negroid.”
They intimate that the Peul sample, which also contains
genetic material from people identified as Tukolor, thus
contains more admixture from “Caucasoid” (specifically
Berber) populations. However, this well-known distinction
between “cattle Fulani” and “town Fulani” is very
frequently overdrawn..
-- the Wolof and Serer populations that cluster
with the Peul are not particularly “Caucasoid,” and
these westernmostWest African groups are certainly distinguished
from Saharan populations according to Cavalli-
Sforza et al.’s principal-component analysis (1994:
170).
--Indeed, historical data might lead
us to expect such genetic results. This possibility seems
to be comprehensively edited out of The History and
Geography of Human Genes, presumably because European
settlement in southern Africa took place after
a.d. 1492. This claimed association of Khoisan people
with “Caucasoids” is quoted in more recent works on
genetics and history in Africa, among them the work by
Passarino et al. (1998) on Ethiopian populations. This
latter paper exhibits the same reliance upon inappropriate
sources as does Cavalli-Sforza et al., with for example
use of a 1964 Encyclopedia Britannica article as a primary
reference on the prehistory of Ethiopia.
--------------------------------
MORE EXCERPTS FROM THE ARTICLE
It is thus extraordinarily
ironic that Cavalli-Sforza, Menozzi, and Piazza in
their attempts to elucidate the ancient genetic history
of the whole of humanity find it necessary to posit that
Africans, as well as members of indigenous groups in
other parts of the world, really have been people “without
history” (Trevor-Roper 1965:9) since a.d. 1492. If anything,
the continent of Africa, at least, has for the past
500 years suffered from a surfeit of history, not its lack.
The relatively recent and closely connected developments
of nationalism and colonialism have certainly
generated a hierarchy of states, ethnies, and “tribes” that
can command the loyalties of people living in different
parts of the world today, but the extension of those units
into the past is a proposition to be tested, not a given.
The recent creation of “tribal” identities is not the
only process of ethnic-group definition in Africa that fails
to accord with the assumptions made in The History and
Geography of Human Genes. The mosaic of peoples described
by Roderick McIntosh (1998; see also McIntosh
1993) in The Peoples of the Middle Niger: The Island of
Gold appears to be of very long standing indeed, with
roots ultimately to be located in the mid-Holocene, but
the complex and evolving relations between different
specialist communities convincingly depicted in that
work bear very little resemblance to ahistorical, static
models of African “tribes.” This mosaic of groups incorporated
people of different language affiliations; it involved
extremely sophisticated ritual, economic, and
technological accommodations to a diverse and changing
natural and cultural milieu; it allowed for processes of
immigration, emigration, and population interchange.
Such constellations of interacting groups can be found
elsewhere in Africa as well, although probably none have
been as intensively investigated and described as the
Middle Niger populations. Interaction between peoples,
in Africa and elsewhere, is a continuing and complex
process and only rarely involves the spasms of mass migration
that Cavalli-Sforza et al. invoke.
Modern concepts of ethnicity and the social groupings
that result from those concepts seem if anything more
closely comparable to the demes studied by population
geneticists than to the “tribes” created by colonialists
and ethnographers and adopted as units of study by Cavalli-
Sforza et al. Demes are not necessarily held to have
sharp boundaries, and intergrades between demes at particular
levels are both clinal and common. Similarly, the
exact characteristics of particular demes may be defined
in reference to particular research problems. This somewhat
resembles instrumentalist formulations that locate
ethnic identities in situationally contingent bundles of
social and cultural identifications that can assume differential
importance in different social situations while
simultaneously existing at multiple social and spatial
levels (Barth 1969, Cohen 1978, Handelman 1977, Mc-
Kay 1982). Ethnoarchaeological research by a number of
investigators working in Africa (David 1992, Hodder
1982, Larrick 1986, MacEachern 1998, Wiessner 1984)
has detected similar patterning in material culture. It is
unlikely that such patterning would derive from the
ahistorical monoliths identified as “tribes” by Cavalli-
Sforza et al. (1994).
African Examples
As in each other continental area investigated, Cavalli-
Sforza, Menozzi, and Piazza treat their data on African
genetic variation in different ways and for somewhat different
purposes. Trees of genetic distance and similarity
are generated through average linkage analysis in order
to elucidate phylogenetic relationships between modern
populations and the histories of ancient fissioning that
have resulted in those modern populations. Two-dimensional
principal-component analyses of variation in genetic
characteristics allow an alternative, graphic presentation
of modern population affinities for comparison
with the phylogenetic trees. Studies of the distributions
of single alleles of particular genes yield data on (and
allow speculation about) particular environmental adaptations
and interactions between different human
groups; these data are somewhat beyond the bounds of
this paper.
Finally, synthetic maps of each region are
derived from the results of principal-component analysis
of multiple gene frequencies—for 79 genes in the African
case, for example. As in the European instance noted
above, Cavalli-Sforza et al. usually assert that the spatial
patterning of values for each principal component is a
reflection of a specific population/language movement
(again, the two are often not differentiated) in the past.
I will first examine some features of the African genetic
tree and principal-component map given in Cavalli-
Sforza et al. (1994:169, 170) and then examine the synthetic
maps of the continent. I emphasize that this inquiry
is very far from being comprehensive, but I hope
that it will be illustrative.
Fig. 1. Genetic tree of 49 African populations (Cavalli-
Sforza, Menozzi, and Piazza 1994:fig. 3.5.1;
q1994, Princeton University Press, reprinted by permission
of Princeton University Press).
The continental genetic tree of
African populations generated by Cavalli-Sforza, Menozzi,
and Piazza (reproduced here as fig. 1) uses genetic
data from 49 human groupings in Africa. The most fundamental
differentiation shown is that between sub-Saharan
Africans, on the one hand, and Saharan and
Northern African populations, on the other. It is among
the latter group (16 of the 49) that we see the only national
identifications of populations (Algerians, Sudanese,
Moroccans, Tunisians, Libyans, Egyptians),
whereas, as noted above, such identifications are the rule
across the Mediterranean in Europe. The other Saharan/
Northern African and all the sub-Saharan population
samples identified are denoted by (1) “traditional” ethnonyms
(Amhara, Hausa, Sandawe, and so on), (2) geographical
designations (Bantu, N.E., Bantu, S.W., and so
on), (3) designations based on language affiliation (Nilotic,
Ubangian, Volta, San), or (4) designations based on
particular physical characteristics (Pygmy/“Pygmoid”).
As noted above, appendix 3 makes the human diversity
of many of these samples obvious.
Within the Saharan/Northern African group there are
two subclusters. One is comprised of a number of “national”
North African populations (Moroccan, Tunisian,
Libyan, Egyptian), as well as Nubian, Bedouin, Berber,
and Canarian. Cavalli-Sforza et al. (1994:172) note that
there are few data available on the genetic makeup of
the Canarian group, and it is not clear whether they are
data on modern peoples or on the extinct Guanches. The
other cluster consists of a number of Northeast African
populations, including the national Sudanese and a generalized
Cushitic-speaking group. Within this Northeast
African group, one subcluster does not conform to historical
or linguistic expectations, as it associates the Algerian
national population fairly closely with the Beja of
eastern Sudan and somewhat more distantly with the
Berber-speaking Tuareg; the latter group thus appears to
be rather distinct from other Berber populations. Cavalli-
Sforza et al. (pp. 172–73) posit an ancient relationship
between Tuareg and Beja, largely, it appears, on the basis
of their shared status as pastoralists.
There are no other
data that I know of that indicate such a link, and in any
case it does not explain the putative close relationship
of modern Algerians to both groups. The researchers note
that they have data on relatively few genes from this
Algerian group, but it should be pointed out that such
small data sets are no obstacle to the acceptance of particular
genetic associations when these fit their expectations—
the Canarian case noted above is a good example.
This pattern of acceptance and rejection of
generally equivalent data on the basis of their concordance
with expectations is common throughout The History
and Geography of Human Genes and is an example
of the strategy of post-hoc accommodative argument
noted by Clark (1998).
The distinction between Saharan/Northern African
populations and peoples living in sub-Saharan Africa is
explained by the varying contribution of genes from
“Caucasoid” populations in Europe and Southwest Asia
to the former. This is very likely a contributing factor,
given the archaeological and historical evidence of such
population interactions around the Mediterranean. It is
also quite likely that clines in gene frequencies across
the Sahara are in part the result of natural selection operating
upon characteristics that are not adaptively neutral
in the very different environments through this region.
There is a significant amount of evidence for both
climatic and latitudinal effects upon different gene frequencies
(Cavalli-Sforza et al. 1994:143; Mastana, Constans,
and Papiha 1996; O’Rourke, Suarez, and Crouse
1985; Spitsyn et al. 1998). The greater instability of Saharan
environments through time probably offered less
scope for such in situ adaptation than is the case among,
for example, the Nile Valley populations examined by
Brace et al. (1996).
Saharan and Sahelian groups (various Berber- and Arabic-
speaking populations, including Tuareg and groups
subordinated to them, such as the Bella and the Haratine
and Saharan-speakers such as the Chaamba, Reguibat,
Teda, and Kanembu) are not covered in detail in the work
(Cavalli-Sforza et al. 1994:173), although investigations
of biological variation among those populations have indicated
that their anthropometric and genetic affiliations
are very diverse and complex (Froment 1999). This lack
of data on intermediate groups may make human physical
and genetic distinctions across the Sahara appear
more clear-cut than they are. The status of these populations
is particularly important given that climatic
change rendered significant parts of the Sahara passable
(and in some cases habitable) through periods in the Holocene
at least, with the result that there is abundant
evidence of more extensive human contacts across the
desert than have existed in historic times. Sutton (1974)
and Ehret (1993) have suggested that the Saharo-Sudanese
Neolithic tradition was largely the province of Nilo-
Saharan-speakers. Populations speaking those languages
do not, however, occupy an intermediate position between
North African and sub-Saharan African populations,
suggesting that either the correlations between archaeology
and linguistics or those between genetics and
linguistics—or both—are erroneous.
While Cavalli-Sforza et al. emphasize the contribution
of immigrant genes to the modern genetic makeup of
Saharan/Northern African populations, they do not really
consider the possibility of an African genetic contribution
to either Europe or the Near East. It thus appears
that Africa accepts genetic contributions from
other areas but does not reciprocate them. A principa lcomponent
map of 42 world populations (Cavalli-Sforza
et al. 1994:82) indicates a somewhat more complex picture,
with a succession of Basques, Sardinians, Near Eastern
populations, and Berbers occupying a space intermediate
between African and European populations,
although certainly arrayed closer to European groups.
This assumption is also at variance with the known history
of the region, where we see evidence for two-way
relations throughout the Holocene, especially via Southwest
Asia and the Iberian and Italian peninsulas. People
from North, Saharan, and sub-Saharan Africa have
crossed the Mediterranean as settlers, conquerors, and
slaves through recorded history just as have Europeans.
In recent times such population flows may have tended
to be from north to south, but it should not be assumed
that this has always been the case.
Pygmies, Khoisan, and Caucasian connections. The
high-level cluster of sub-Saharan African populations
contains 33 of the 49 populations of the phylogenetic
tree. It is a considerably more diverse grouping than the
Saharan/Northern African, with multiple subclusters at
different fissioning points. The most famous “outlier”
populations of traditional African ethnography are of
course Pygmy and Khoisan-speaking groups, which are
to varying degrees physically distinct from their African
neighbors and also to varying degrees participate in foraging
economies. These latter are frequently seen by
Westerners as archaic, and Pygmy and Khoisan populations
have often been identified as unchanged relics of
earlier ages (e.g., Thomas 1959:6–8;Turnbull 1983:11–13,
157–58). Pygmy (Mbuti and Biaka) and “Pygmoid” populations
are found at various points on Cavalli-Sforza et
al.’s phylogenetic tree as outliers and with other groups.
As Froment (1998) points out, this separation of Pygmy
and other African populations is extremely imprecise; it
depends to a great extent upon linguistic criteria, ignores
the numerous transitional populations (not only those
denominated as “Pygmoid”), and systematically discounts
the fact that we know very little about the historical
and physical relations between these groups over
any significant period of time.
Similarly, Khoi and San populations cluster with a Somali
sample (which itself is held to be out of place, given
that Somali groups geographically sit within the
Northern African range), while Sandawe clusters with
populations from Senegambia and Hadza is an outlier
between the two. Cavalli-Sforza et al. (1994:169–70,
174–77, 189–93) posit that especially San populations are
the result of admixture between “Caucasoid” groups
originating in Southwest Asia and African “Negroid”
groups. This is supposed to be a different process of interaction
across the Red Sea from the one that yielded
the distinctive genetic and physical characteristics of
Ethiopian populations; indeed, the San and Ethiopian
peoples are held to be “similar to Caucasoids but . . .
otherwise very different [from one another]” ( p. 191).
The historical mechanisms—and even the demographic
meaning—of such multiple similarities are left unspecified.
This is unfortunate, given that hypotheses of immigration
into Africa by (often “Hamitic”) “Caucasoids”
have bedeviled African history and archaeology for much
of the past century, often being advanced to explain away
African cultural innovations and based on very unsatisfactory
evidence. One would have hoped that consciousness
of this situation would have led the authors
of The History and Geography of Human Genes to substantiate
this hypothesis in detail.
The nongenetic evidence marshaled in support of the
hypothesis of relations between San groups and populations
in the Near East is extremely weak. A putative
“Asian” genetic contribution to forager groups in Ethiopia
(Nijenhuis and Hendrikse 1986) is discussed only
with reference to “Pygmoid” populations, although Cavalli-
Sforza et al. (1994:174) imply that these groups are
related to the San. They claim ( pp. 160, 176) that skeletal
material “credibly identified as San” has been found in
various parts of North and East Africa, including northern
Egypt, but note only parenthetically that this assertion
in Nurse, Weiner, and Jenkins (1984) is based upon
a 30-year-old paper by Philip Tobias (1968 [1964]).
The Tobias paper does not in fact seem to make that claim,
and it is in any case disputed by more recent researchers
on the basis of the characteristics of the material involved,
the very fragmentary state of the collections, and
known problems with the accumulation of Khoi and San
skeletal reference collections (Froment 1998; Morris
1986, 1987; Rightmire 1975; 1984:193–98; Schepartz
Fig. 2. Principal-component analysis of 49 African populations. BNT,
Bantu and Bane; C, Central; ETH, Ethiopia;
KH, Khoisan; N, North Africa; NS, Nilo-Saharan; PG, Pygmies; W, West Africa (Cavalli-Sforza, Menozzi,
and Piazza 1994:fig. 3.5.2; q1994, Princeton University Press, reprinted by permission of Princeton University
Press 1988).
In fact, the identification of this skeletal material
from northeastern Africa as related to San skeletal material
from southern Africa is very doubtful; the material
indicates that ancient populations in the area were most
closely affiliated with the present-day inhabitants.
The only widely accepted evidence of ancient Khoisan
populations in East Africa is the ascription of the Sandawe
and Hadza languages to the Khoisan phylum (with
even less well-attested traces of Khoisan contacts in Dahalo
and Yaaku [Ehret 1974:11, 88]). However, the Khoisan
affiliations of Sandawe and/or Hadza are still disputed
by some linguists, and in any case the available
genetic data do not indicate a close relationship between
Sandawe and Hadza people, on the one hand, and San
and Somali people, on the other. The paradox is obvious:
Sandawe and Hadza provide the only firm link between
San populations and northeastern Africa (a linguistic
one), but according to the genetic data that provide the
basis for The History and Geography of Human Genes
they are more closely related toWest and Central African
groups (fig. 2).
There seems to be no a priori reason to
associate Khoisan-speaking populations with Southwest
Asia on the basis of San genetic data and not to associate
Khoisan-speaking populations with Senegambia on the
basis of Sandawe genetic data, but this is just what Cavalli-
Sforza et al. do. It is also, of course, possible that
either or both associations are spurious, especially given
the small size of some of these forager groups and the
attendant possibility of genetic drift.
We must remember, as well, that the southern tip of
Africa is not in fact occupied only by Khoisan and other
African peoples, as the discussion in Cavalli-Sforza et al.
(1994:158–68) would sometimes seem to imply. It is also
the only part of the continent to have seen extensive
settlement by Europeans (and Asian people arriving in
the context of colonization) over the past four centuries.
This has certainly involved a considerable amount of
genetic interchange between indigenous and immigrant
populations, along with the formation of a number of
communities of mixed descent, with genetic, cultural,
and linguistic affiliations to Khoisan, other African, European,
and Asian groups.
These communities and their
origins are discussed at some length in Nurse, Weiner,
and Jenkins (1985:221–40); they have at various times
and in various areas included !Kora, Baster, Orlam, Gyzikoa,
Griqua, and “Cape Coloured” populations, among
others, and have occupied territories in South Africa,
Namibia, and Botswana. As Morris (1992, 1997) points
out, “Caucasoid” genetic and morphological features
seem to be less important among Griqua populations (the
only cases that I know of for which these affinities for
these groups have been studied) than might have been
expected on historical grounds, but these groups are quite
diverse, and rather little is known of genetic exchange
between European and African populations on the frontiers
of European settlement in southern Africa.
It would seem reasonable to seek evidence for “Caucasoid”
genetic affiliations among southern African populations
in this recent and relatively well-attested contact
situation rather than in putative Khoisan contacts
with Northeast Africa and Southwest Asia for which no
firm evidence exists. Indeed, historical data might lead
us to expect such genetic results. This possibility seems
to be comprehensively edited out of The History and
Geography of Human Genes, presumably because European
settlement in southern Africa took place after
a.d. 1492. This claimed association of Khoisan people
with “Caucasoids” is quoted in more recent works on
genetics and history in Africa, among them the work by
Passarino et al. (1998) on Ethiopian populations. This
latter paper exhibits the same reliance upon inappropriate
sources as does Cavalli-Sforza et al., with for example
use of a 1964 Encyclopedia Britannica article as a primary
reference on the prehistory of Ethiopia.
West African cases. As noted above, the Sandawe sample
clusters with samples from westernmostWest Africa,
specifically from Serer, Wolof, and Peul populations. It
was this latter group that first caught my attention when
I looked at Cavalli-Sforza et al.’s phylogenetic tree because
of the distinction between the Peul and Fulani
samples that I have mentioned. These two ethnonyms
both denote Fulbe people, but the Fulani sample is
closely associated with a set of samples from Nigerian
populations, including Ibo, Yoruba, and Hausa—groups
a long way from Senegambia. Cavalli-Sforza et al. recognize
the peculiarity of a situation in which one ethnonym
encompasses two distinct genetic populations
and respond by recalling Murdock’s (1959:414–15) distinction
between pastoral Fulbe, who are supposed to be
“Caucasoid-like,” and settled Fulbe, who are “Negroid.”
They intimate that the Peul sample, which also contains
genetic material from people identified as Tukolor, thus
contains more admixture from “Caucasoid” (specifically
Berber) populations. However, this well-known distinction
between “cattle Fulani” and “town Fulani” is very
frequently overdrawn (for other considerations of the issue,
see e.g., Curtin 1975:19–22; Irwin 1981:46–55;
Schultz 1981), the Wolof and Serer populations that cluster
with the Peul are not particularly “Caucasoid,” and
these westernmos tWest African groups are certainly distinguished
from Saharan populations according to Cavalli-
Sforza et al.’s principal-component analysis (1994:
170).
It is more likely that this distinction between the populations
recorded as Fulani and Peul can in great part be
traced to social/political processes over the past 200
years in West/Central Africa, where among other events
the jihad of Uthman dan Fodio led to the establishment
of the Sokoto Caliphate in modern Nigeria and Cameroon
and to an attendant favorable change in the status
of a Fulbe ethnic identity. Under these circumstances,
we would expect that many indigenous people, especially
those of Hausa background but also including
members of other Nigerian and Cameroonian societies,
would attempt to assimilate to that Fulbe identity. Indeed,
such processes are often discussed and debated in
the Nigerian historical literature (Salamone 1985, Smith
1997) and can probably be seen in play in the mixed
ethnic affinities of the “Fulani” sample in appendix 3 of
The History and Geography of Human Genes (Cavalli-
Sforza et al. 1994:470).3 Difference between “Peul” and
“Fulani” thus appears to be not the result of a longstanding
division of two populations subsumed under
one ethnonym but quite the opposite—the relatively recent
creative manipulation of identity for political and
social advantage.
The significance of this conclusion lies not so much
in the ethnographic concepts used by Cavalli-Sforza et
al. as in the possibility of such a drastic change in lin-
3. The Hausa, we should note, become Nilo–Saharans rather than
speakers of Afroasiatic languages in the principal-component maps
of Cavalli-Sforza et al. (1994:170, 181; see fig. 2) and “elongated”
pastoral nomads a` la Hiernaux in the conclusions to the chapter
on Africa (p. 194). These incorrect ascriptions indicate the difficulty,
especially for nonspecialists, of working with such very large ethnographic and linguistic data sets.
This difficulty probably also accounts
for citations of, for example, works dealing with Chukotskii
and Brazilian populations in references to African groups (p. 471).
linguistic and ethnic affiliation by a significant group of
people in less than two centuries. If such a change happened
in northern Nigeria, how common is it elsewhere,
and how many such changes affect the analyses in The
History and Geography of Human Genes? The data given
in this work indicate that such processes may be quite
common on the African continent. In general, the correspondences
between genetic relationships and those
based upon linguistic and “tribal”/ethnic affiliation do
not appear to be particularly close. Cavalli-Sforza et al.’s
(1994:182–85) discussion of these correspondences
abounds with perplexing relationships between different
groups of people—Nubians with Moroccans, Ubangian speakers
with southwestern Bantu populations, Sara
speakers of a Nilo-Saharan language with Hausa on the
one hand and Biaka Pygmies on the other, and so on.
The “Caucasoid” genetic affiliations of the San may
prove to be another example.
Technological innovations, population expansions,
and especially migrations are invoked to explain as many
of those relationships as is possible, but these explanations
often seem ad hoc. For example, archaeological and
linguistic data involving the Bantu expansion are invoked
to explain the apparently very close genetic relations
between northwestern and southeastern Bantu
populations. Surely, however, any such explanation
should also account for the relative lack of genetic similarity
of the intervening Bantu populations (table 1). In
any case, the close relations between archaeological and
linguistic reconstructions (the former very dependent
upon the latter) in this case vitiate their separate use as
independent tests. Indeed, Vansina (1995) has cogently
critiqued the assumptions that there are close parallels
in archaeological and linguistic data bearing upon the
expansion of Bantu languages and that that expansion
was the result of a single, massive migration process.
The archaeological and linguistic data simply do not
yield the neat, uncomplicated picture of the past that
Cavalli-Sforza et al. present. The extent of this confusion
of different data sources is made evident by a map (1994:
fig. 3.9.3) that purports to show a pattern of Bantu expansion
based upon archaeological data in which most
of the linkages (between Nilotic-speaking peoples and
Central-Western Bantu or between Bane populations and
Northeastern Bantu, for example) are sustained by no
archaeological data at all. No other mechanisms for genetic
or linguistic dispersal are considered, and the con-
nection of these different types of data seems quite
unsystematic.
Synthetic maps of african genetic variation
Cavalli-Sforza et al. (1994:189–92) do not attempt to give
historical explanations for African genetic frequency variation
to the sixth principal component as they do for
Europe. (Perhaps they would have done so if there were
as many different regional historical reconstructions for
Africa as there are for Europe.) Instead, they limit their
explanations to the first four principal components,
which encompass about the same percentage of total variance
as in the European case. The synthetic map derived
from the first principal component (fig. 3) has a strong
north-south gradient, with opposing extremes along the
Mediterranean coast and from the Equator southward.
The researchers relate this to the ancient presence of
“Caucasoid” populations along the northern coast of Africa
and the gradual admixture of “Caucasoid” and “Negroid”
populations across the Sahara. They also comment
upon the paucity of data on Saharan populations.
As noted above, we should remember that this gradient
is also an environmental one and that it is quite likely
that environmental variability is affecting the genetic
makeup of African populations to some degree.
The discussion of patterning on the synthetic maps for
the second and third principal components is rather confusing.
Cavalli-Sforza et al. (1994:189) note that both
maps show maximum values in northeastern Africa, although
the specific association with Ethiopia is not as
clear as they claim. They state that the map of the second
principal component indicates a similar set of maximum
values in the area of southwestern Africa occupied by
Khoisan-speakers in precolonial times and that this indicates
that there existed a population ancestral to modern
Khoisan peoples somewhere near Ethiopia. Low values
on the second principal component in the areas
intervening between Ethiopia and the Kalahari are ascribed
to the later settlement of those areas by Bantu
peoples.
The maps (figs. 4 and 5) actually indicate that
this pattern is based upon variation in the third principal
component. The difference is significant, given that the
second principal component explains 18.6% of the total
variance and the third only 10.3%, but the text seems
to indicate that this pattern is in fact associated with
the second.
Presumably, this patterning is explained by the admixture
of “Caucasoid” and “Negroid” populations in
northeastern Africa that Cavalli-Sforza et al. believe gave
rise to the Khoisan peoples. However, the maps for the
second and third principal components show almost opposing
values along the Mediterranean coast, where the
presence of “Caucasoid” groups is held to be the explanation
for the patterning of the first. Similarly, northeastern
and southwestern Africa show maximum divergence
on the third principal component, which is
accounted for by their differences from one another in a
context of general relationship to “Caucasoid” groups.
These “Caucasoid” genetic contributions are thus invoked
as the primary explanation for the variation exhibited
in the first, second, and third principal components
but seem to refer to different sets of genetic
characteristics and thus presumably to different populations
of “Caucasoids.” These interpretations would
seem to lack much real explanatory value.
The fourth principal component accounts for only
7.0% of the total variance (fig. 6). Maximum values are
associated with the northwestern part of the Central African
tropical forest and minimum values with the border
regions of Mali, Niger, and Burkina Faso. Parsimoniously,
the researchers associate these maximum and
minimum values with two different agricultural expansions.
They associate the area of maximum values with
an expansion of Bantu populations, although one might
associate the patterning with Adamawa-Ubangian
groups as well (David 1982:88–91; Saxon 1982) and the
influence of this massive Bantu migration upon genetic
variation would seem to be extraordinarily low compared
with that of the “Caucasoid” gene flows that have determined
the nature of the first three principal components.
They admit that there is no equivalent evidence
for an agricultural expansion from the area of minimum
values in the West African Sudanic zone but counsel
future comparison with archaeological data. These syn-
thetic maps sometimes appear to function as historical
Rorschach tests, with genetic data being associated with
ancient cultural processes on the basis of general resemblances
in spatial patterning.
Conclusions
The History and Geography of Human Genes is an extraordinary
compendium of information on human genetic
variation. It links a huge amount of data on that
variation with archaeologically, historically, and linguistically
known cultural processes that took place at
various times and in various areas. Its hypotheses have
been provocative and in many cases useful, and researchers
in other fields have responded to them in ways that
will eventually increase our knowledge of our shared
human past. In the long run, Cavalli-Sforza et al. have
probably established the examination of patterns in genetic
characteristics as a valuable complement to older,
more established avenues for the examination of prehistory.
This genetic research was limited by the lack of
samples from many areas, by the necessity of using samples
collected under very different conditions, by the
great variation in the amount of information available
on the composition of the groups being sampled, and by
the lack of availability of statistical techniques that
could account for interactions between groups during the
process of constructing phylogenetic trees. To their
credit, Cavalli-Sforza, Menozzi, and Piazza recognize and
highlight a number of these problems and try to compensate
for them. Their solutions are often sophisticated
and ingenious, and the potential of these techniques is
obvious.
The danger for African archaeology is that the very
impressive amount of genetic data amassed in The History
and Geography of Human Genes will convince nonspecialists
of the validity of the hypothesized associations
between human groups. That would be unfortunate,
because this pioneering synthesis is hobbled by
a conceptual model of human behavior that takes little
account of ethnographic, historical, linguistic, and archaeological
research on the constitution of human societies
and of the various ways in which cultural meaning
is constructed in the continuing negotiation of
individual and group identities. Similarly, the discussion
of relations between language and ethnicity pays virtually
no attention to linguistic analyses of such relationships.
Their compilation of the data that make up the
African sample should have indicated to the researchers
that these relations are extremely complex, while the
varying criteria—linguistic, geographic, national—used
for identifying populations make it seem unlikely that
the groups under study are comparable entities.
African ethnic units are not bounded, homogeneous
monoliths either frozen in place since before a.d. 1492
or caromming around the continent like culture-bearing
billiard balls. They are dynamic entities, manipulated in
response to changes in the natural and social environment,
and their composition can change very quickly
indeed. Differences in the temporal resolutions of genetic,
linguistic, archaeological, and historical analyses
are one of the main difficulties in these multidisciplinary
studies, and the ability to control chronologies is more
restricted in the former two disciplines than in the latter.
It is often difficult, therefore, to establish whether the
genetic patterning exhibited by modern populations can
be traced back to events that happened centuries or millennia
ago. A number of the groups used in the analyses
in The History and Geography of Human Genes did not
exist as such five centuries ago, and the modes of human
interaction in Africa are not in any way reducible to a
list of technological innovations and mass migrations.
In part because of this lack of chronological control, the
procedures used by Cavalli-Sforza et al. to associate genetic
patterning with cultural events in the past are quite
unsystematic, involving attempts to fit secondhand
knowledge of archaeological, ethnographic, and linguistic
reconstructions into a Procrustean bed of genetic patterning.
Clark’s (1998) description of the “ransacking”
of European genetic data for meaning applies equally
well in the African case.
The deme concept, often used in population genetic
research in other species, would seem to be a better and
more flexible fit for the study of human genetic variation
mac eachern Genes, Tribes, and African History F 371
than “tribal” typologies borrowed wholesale from ethnographic
research done 50 or more years ago, especially
since it seems to find resonances in some more recent
concepts of ethnicity. It remains to be seen whether it
will be applied to human genetic research in Africa and
elsewhere. It would be interesting to see, for example,
whether human demes on some level correspond to the
material distributions that we call archaeological
traditions, without any attempt to fit either genetics or
archaeology to the straitjacket of the ethnographic
“tribe.” Archaeological distributions of artifacts frequently
occur over much larger territories than do the
ethnic (“tribal”) units that ethnographers study in the
present, and it is possible that in some cases this reflects
movements by artifact producers, especially women,
that would also be reflected in patterns of genetic variation
(MacEachern 1998). In general, research on the African
past would at this point benefit far more from
small-scale studies of genetic patterning within and between
particular communities than from the continental
and subcontinental approaches that make up The History
and Geography of Human Genes.
The genetic data available for Africa have serious limitations,
but the patterns of variation demonstrated by
these data are interesting and probably do indicate patterns
of cultural interaction in the past. There are, however,
good reasons to be suspicious of efforts to correlate
those patterns with cultural processes at the level of
specificity attempted by Cavalli-Sforza et al. The project
of correlating the data from these different kinds of investigation
still lies well before us, and it would be a
great mistake to assume that such correlations will be
self-evident. Research that has as its goal the study of
the history and geography of human genes in Africa and
elsewhere must be truly interdisciplinary, working with
the best knowledge of other disciplines that can be acquired.
A study that uses the terminology of other disciplines
but takes insufficient account of the concepts
behind that terminology risks generating misleading reconstructions
of human history and prehistory.
--------------------------------------------------
Keita's critique of the critique:
MacEachern has given us an effective critique of the African
chapter in Cavalli-Sforza et al.’s book, while also
illustrating general theoretical problems in syntheses of
archaeology, ethnicity, linguistics, and human biology.
The insights offered are especially useful for those interested
in Africa, for which scholarship has long been
hampered by myths and stereotypes. Notable among
these is a paradigm which sets limits on authentic biocultural
indigenous Africanness; this surprisingly persists
in some scientific literature. The work of even some
Africanists is still influenced by this legacy, which usually
manifests itself around the problematic construct of
race. Although MacEachern has given a cogent presentation,
there are a few concerns.
One wishes that emphasis had been placed on the African
origin and differentiation of the Afro-Asiatic language
family and its subsequent spread to Asia (Greenberg
1966; Bender 1975; Diakonoff 1965, 1981; Ehret
1984, 1995; Ruhlen 1991). The dubious Nostratic construct
has been used to postulate the spread of food production
into Africa (from Asia) by Afro-Asiatic-speakers.
Significantly, reconstructed common Afro-Asiatic has no
terms for food production (Ehret 1984, 1995) and is accorded
the same time depth as Nostratic (cf. Ehret 1984
and Barbujani and Pilastro 1993). Furthermore, most
Nostraticists now conceptualize Afro-Asiatic to be a sister
to Nostratic, not a daughter (Ruhlen 1991). Finally,
archaeological data support migration from Africa into
the Near East during the time frame suggested by some
Nostratic models for immigration into Africa (cf. Bar-
Yosef 1987 and Barbujani and Pilastro 1993).
On a technical note, MacEachern has made a common
error, that of easily equating phenetic similarity with a
necessary genealogical relationship. The results of mathematical
techniques employed by human biologists must
be interpreted with sound biological principles and other
information (Rhoads 1984, Harrison 1984). The fact that
small samples of Beja and Algerians and then of Nubians
and Moroccans cluster clearly cannot be taken as necessarily
indicative of a close genealogical relationship.
The similarity may have many explanations including
chance, especially since, as MacEachern notes, the nature
of any links between earlier and later peoples having
these designations remains an open question.
This is even more so when selection, migration, gene flow, and
language shift have to be integrated into explanatory
models. Harrison (1984:61), considering the results of
simulations which evaluate interpopulation gene flow
with and without selection, points to
another important issue concerning migration and
population affinities: the causes of inter-population
patterns of gene frequency variation, and the meaning
of affinities based not on ancestry but on genetic
similarity. Clearly with varying levels of selection,
populations which are genetically similar may have
a much more remote common ancestry than populations
which are genetically more different.
The lack of criticism on this point does not, however,
seriously weaken the overall critique.
In fact, there is little to quibble with in MacEachern’s
effort. However, he is perhaps somewhat overconfident
in the belief that genetic research has led to the “dismemberment”
of racial taxonomies and racio-typological
thinking. While few write today in terms of Nordic
or Jewish races, and this is a result of research and the
World War II experience, there is still plenty of work on
Africa which constructs its interpretations of diversity
explicitly or implicitly in terms of conceptually idealized
“Caucasoid” and “Negroid” “taxa,” implying a causal or
foundational linkage between morphological complexes
and a variety of genes.
Cavalli-Sforza et al.’s book is one example of this kind
of work at some level. For some reason MacEachern does
not directly engage its underlying nonevolutionary raciotypological
model of interpretation or its use of racial
terms, which are ontologically misleading in that they
imply that supra-Saharan Africans were originally European-
derived. These people supposedly interacted with
“Negroids” and others, receiving some genes but contributing
a lot more and thus becoming the primary explanation
of African diversity. MacEachern’s critique
would be even more useful if he had pointed out that
(1)fossils indicate the presence of anatomically modern people
in supra-Saharan and Nile Valley Africa at a time
when hominids in Europe had Neanderthal morphology;
(2) coalescence (and therefore differentiation) times for
numerous genes sampled globally from living humans
date to a period when modern morphology had not yet
emerged or was to be found only in Africa; and
(3) global genetic diversity seems largely to be a subset of that
found in Africa. (Whether this dates back to the time of
Homo erectus or modern H. sapiens matters little; the
diversity denoting the mythical “racial divergence” exists
in Africa, as should be expected, and antedates the
existence of the morphologies used to define races.)
MacEachern fails to make these points, which together
indicate that the African genetic profiles and morphologies
being interpreted as solely resulting from European
(or Southwest Asian) colonization and/or admixture
(with “Negroids”) are largely the product of various authentic
intra-African biohistorical processes which perhaps
date to a time before there were any anatomically
modern Europeans.
However, the Holocene climatic
fluctuations of the Sahara (Hassan 1988), probable Late
Pleistocene (and Holocene) population increase in response
to changing subsistence regimes (Wetterstrom
1993), and the spread of early Afro-Asiatic-speakers from
the Horn surely also had some effects. Supra-Saharan
Africa is not primarily a product of extra-African colonization.
There are many ways to be biologically African—
many morphological combinations and genes
(Hiernaux 1974, Keita and Kittles 1997).
The frequency of neutral private alleles of the right date and
demonstrably originating outside of continental Africa will
have to be determined before the non-African contribution
to Africa’s genetic picture can be fully assessed.
MacEachern seems to accept the racio-typological model
of explanation at the larger level. Nevertheless, he contributes
to the paradigm shift in understanding African
biological variation, which will reject racio-typological
thinking and use the reality of indigenous African diversity
as an analytic tool.
----------------------------------------
OTHER COMMENTATORS WEIGHED IN:
QUOTE:
alan g. morris
Department of Anatomy and Cell Biology, University
of Cape Town Medical School, Observatory 7925,
South Africa (morris@anat.uct.ac.za). 17 xi 99
"MacEachern’s other major complaint about the compendium
by Cavalli-Sforza et al. is the use of principal
components to generate synthetic maps which are then
interpreted as signs of specific past events. The principal
components are not events. They are mathematical relationships
and as such as far too complex to be given
simple historical interpretations. MacEachern, in line
with Clark’s (1998) assessment, accuses Cavalli-Sforza
et al. of having a “pattern of acceptance or rejection of
generally equivalent data on the basis of concordance
with expectations.” Simplified, this means that they saw
what they wanted to see. I entirely agree with Mac-
Eachern’s concern that many students of anthropology,
genetics, and history are going to take Cavalli-Sforza et
al.’s work at face value. By doing that, they may be uncritically
reproducing interpretations ofnature of ethnic identity into account.
This is not the first time that I have looked at the
difficulties in interpreting genetic data. The equivalent
genetic history of southern African people was published
by Nurse, Weiner, and Jenkins in 1985. Again, the work
was a wonderful production of data on a wide range of
peoples, but the authors’ attempts at linking what they
saw in the genes with history were sometimes very wide
of the mark. Social theory and social events are not easily
mac eachern Genes, Tribes, and African History F 377
explained by biology. As does MacEachern here, I ended
my comments by warning of my great fear “that students
of archaeology, genetics and physical anthropology will
simply assume that this is the academic ‘bible’ on the
subject and will absorb many of the speculative passages
without criticism”
-----------------
mark pluciennik
Department of Archaeology, University of Wales,
Lampeter, Ceredigion SA48 7ED, Wales, U.K.
(m.pluciennik@lamp.ac.uk). 11 x 99
"MacEachern notes that the various “academic and political”
uses for human genetic research raise the most
troubling questions when driven by a regime of private
property. I share his concern that Western bio-prospectors
promote an “identification of indigenous peoples
with their surroundings” in part to performa metonymy
whereby people become resources. If thus identifying indigenous
people reduces their capacity to organize politically
to protect their rights of disposition, then serious
problems will arise. In seeking to patent the material
they “mine,” the bio-prospectors will have opened a sort
of final colonialism, contained here in a chiasmus: from
control over the raw material of nature to control over
the nature of raw materials."
African genetic data, particularly the work of Cavalli Sforza.
The author questions whether the European academy can be
trusted to render a balanced picture of African genetic diversity.
Among the criticisms is that data on African peoples is being
analyzed in ways different from that accorded European samples,
for example with a lesser level of detail in some analyses that
masks the picture on the ground.
quote:
African ethnic units are not bounded, homogeneous
monoliths either frozen in place since before a.d. 1492
or caromming around the continent like culture-bearing
billiard balls. They are dynamic entities, manipulated in
response to changes in the natural and social environment,
and their composition can change very quickly
indeed..
In part because of this lack of chronological control, the
procedures used by Cavalli-Sforza et al. to associate genetic
patterning with cultural events in the past are quite
unsystematic, involving attempts to fit secondhand
knowledge of archaeological, ethnographic, and linguistic
reconstructions into a Procrustean bed of genetic patterning.
Clark’s (1998) description of the “ransacking”
of European genetic data for meaning applies equally
well in the African case.
Some argue that activities like the Human Geonome
Project simply replicate existing ideological
models of current elites only this time couched
in the language of biotechnology and soothing
tones of "diversity." A typical example is
"Genomics, divination, racecraft," by STEPHAN PALMIÉ 2007.
A critique of the critique by Keita is also shown below.
Others raise the spectre of a clamed "bio colonialism"..
In seeking to patent the material they “mine,” the bio-prospectors
will have opened a sort of final colonialism, contained here in a
chiasmus: from control over the raw material of nature to control
over the nature of raw materials."
excerpt of critique from:
Scott MacEachern
Genes, Tribes, and African History. Scott MacEachern.
Current Anthropology Volume 41, Number 3, June 2000
SOME BLURBS:
quote:
- African ethnic units are not bounded, homogeneous
monoliths either frozen in place since before a.d. 1492
or caromming around the continent like culture-bearing
billiard balls.
- A number of the groups used in the analyses
in The History and Geography of Human Genes did not
exist as such five centuries ago, and the modes of human
interaction in Africa are not in any way reducible to a
list of technological innovations and mass migrations.
In part because of this lack of chronological control, the
procedures used by Cavalli-Sforza et al. to associate genetic
patterning with cultural events in the past are quite
unsystematic, involving attempts to fit secondhand
knowledge of archaeological, ethnographic, and linguistic
reconstructions into a Procrustean bed of genetic patterning.
Clark’s (1998) description of the “ransacking”
of European genetic data for meaning applies equally
well in the African case.
-Presumably, this patterning is explained by the admixture
of “Caucasoid” and “Negroid” populations in
northeastern Africa..
--These “Caucasoid” genetic contributions are thus invoked
as the primary explanation for the variation exhibited
in the first, second, and third principal components
but seem to refer to different sets of genetic
characteristics and thus presumably to different populations
of “Caucasoids.” These interpretations would
seem to lack much real explanatory value.
--The extent of this confusion of different data sources is
made evident by a map (1994: fig. 3.9.3) that purports
to show a pattern of Bantu expansion
based upon archaeological data in which most
of the linkages (between Nilotic-speaking peoples and
Central-Western Bantu or between Bane populations and
Northeastern Bantu, for example) are sustained by no
archaeological data at all. No other mechanisms for genetic
or linguistic dispersal are considered, and the con-
nection of these different types of data seems quite
unsystematic.
--Cavalli-Sforza et al. (1994:189–92) do not attempt to give
historical explanations for African genetic frequency variation
to the sixth principal component as they do for Europe.
-- The data given in this work indicate that such processes
may be quite common on the African continent. In general,
the correspondences between genetic relationships and those
based upon linguistic and “tribal”/ethnic affiliation do
not appear to be particularly close. Cavalli-Sforza et al.’s
(1994:182–85) discussion of these correspondences
abounds with perplexing relationships between different
groups of people—Nubians with Moroccans, Ubangianspeakers
with southwestern Bantu populations, Sara
speakers of a Nilo-Saharan language with Hausa on the
one hand and Biaka Pygmies on the other, and so on.
The “Caucasoid” genetic affiliations of the San may
prove to be another example.
--The Hausa, we should note, become Nilo–Saharans rather than
speakers of Afroasiatic languages in the principal-component maps
of Cavalli-Sforza et al.
-- Cavalli-Sforza et al. recognize
the peculiarity of a situation in which one ethnonym
encompasses two distinct genetic populations
and respond by recalling Murdock’s (1959:414–15) distinction
between pastoral Fulbe, who are supposed to be
“Caucasoid-like,” and settled Fulbe, who are “Negroid.”
They intimate that the Peul sample, which also contains
genetic material from people identified as Tukolor, thus
contains more admixture from “Caucasoid” (specifically
Berber) populations. However, this well-known distinction
between “cattle Fulani” and “town Fulani” is very
frequently overdrawn..
-- the Wolof and Serer populations that cluster
with the Peul are not particularly “Caucasoid,” and
these westernmostWest African groups are certainly distinguished
from Saharan populations according to Cavalli-
Sforza et al.’s principal-component analysis (1994:
170).
--Indeed, historical data might lead
us to expect such genetic results. This possibility seems
to be comprehensively edited out of The History and
Geography of Human Genes, presumably because European
settlement in southern Africa took place after
a.d. 1492. This claimed association of Khoisan people
with “Caucasoids” is quoted in more recent works on
genetics and history in Africa, among them the work by
Passarino et al. (1998) on Ethiopian populations. This
latter paper exhibits the same reliance upon inappropriate
sources as does Cavalli-Sforza et al., with for example
use of a 1964 Encyclopedia Britannica article as a primary
reference on the prehistory of Ethiopia.
--------------------------------
MORE EXCERPTS FROM THE ARTICLE
It is thus extraordinarily
ironic that Cavalli-Sforza, Menozzi, and Piazza in
their attempts to elucidate the ancient genetic history
of the whole of humanity find it necessary to posit that
Africans, as well as members of indigenous groups in
other parts of the world, really have been people “without
history” (Trevor-Roper 1965:9) since a.d. 1492. If anything,
the continent of Africa, at least, has for the past
500 years suffered from a surfeit of history, not its lack.
The relatively recent and closely connected developments
of nationalism and colonialism have certainly
generated a hierarchy of states, ethnies, and “tribes” that
can command the loyalties of people living in different
parts of the world today, but the extension of those units
into the past is a proposition to be tested, not a given.
The recent creation of “tribal” identities is not the
only process of ethnic-group definition in Africa that fails
to accord with the assumptions made in The History and
Geography of Human Genes. The mosaic of peoples described
by Roderick McIntosh (1998; see also McIntosh
1993) in The Peoples of the Middle Niger: The Island of
Gold appears to be of very long standing indeed, with
roots ultimately to be located in the mid-Holocene, but
the complex and evolving relations between different
specialist communities convincingly depicted in that
work bear very little resemblance to ahistorical, static
models of African “tribes.” This mosaic of groups incorporated
people of different language affiliations; it involved
extremely sophisticated ritual, economic, and
technological accommodations to a diverse and changing
natural and cultural milieu; it allowed for processes of
immigration, emigration, and population interchange.
Such constellations of interacting groups can be found
elsewhere in Africa as well, although probably none have
been as intensively investigated and described as the
Middle Niger populations. Interaction between peoples,
in Africa and elsewhere, is a continuing and complex
process and only rarely involves the spasms of mass migration
that Cavalli-Sforza et al. invoke.
Modern concepts of ethnicity and the social groupings
that result from those concepts seem if anything more
closely comparable to the demes studied by population
geneticists than to the “tribes” created by colonialists
and ethnographers and adopted as units of study by Cavalli-
Sforza et al. Demes are not necessarily held to have
sharp boundaries, and intergrades between demes at particular
levels are both clinal and common. Similarly, the
exact characteristics of particular demes may be defined
in reference to particular research problems. This somewhat
resembles instrumentalist formulations that locate
ethnic identities in situationally contingent bundles of
social and cultural identifications that can assume differential
importance in different social situations while
simultaneously existing at multiple social and spatial
levels (Barth 1969, Cohen 1978, Handelman 1977, Mc-
Kay 1982). Ethnoarchaeological research by a number of
investigators working in Africa (David 1992, Hodder
1982, Larrick 1986, MacEachern 1998, Wiessner 1984)
has detected similar patterning in material culture. It is
unlikely that such patterning would derive from the
ahistorical monoliths identified as “tribes” by Cavalli-
Sforza et al. (1994).
African Examples
As in each other continental area investigated, Cavalli-
Sforza, Menozzi, and Piazza treat their data on African
genetic variation in different ways and for somewhat different
purposes. Trees of genetic distance and similarity
are generated through average linkage analysis in order
to elucidate phylogenetic relationships between modern
populations and the histories of ancient fissioning that
have resulted in those modern populations. Two-dimensional
principal-component analyses of variation in genetic
characteristics allow an alternative, graphic presentation
of modern population affinities for comparison
with the phylogenetic trees. Studies of the distributions
of single alleles of particular genes yield data on (and
allow speculation about) particular environmental adaptations
and interactions between different human
groups; these data are somewhat beyond the bounds of
this paper.
Finally, synthetic maps of each region are
derived from the results of principal-component analysis
of multiple gene frequencies—for 79 genes in the African
case, for example. As in the European instance noted
above, Cavalli-Sforza et al. usually assert that the spatial
patterning of values for each principal component is a
reflection of a specific population/language movement
(again, the two are often not differentiated) in the past.
I will first examine some features of the African genetic
tree and principal-component map given in Cavalli-
Sforza et al. (1994:169, 170) and then examine the synthetic
maps of the continent. I emphasize that this inquiry
is very far from being comprehensive, but I hope
that it will be illustrative.
Fig. 1. Genetic tree of 49 African populations (Cavalli-
Sforza, Menozzi, and Piazza 1994:fig. 3.5.1;
q1994, Princeton University Press, reprinted by permission
of Princeton University Press).
The continental genetic tree of
African populations generated by Cavalli-Sforza, Menozzi,
and Piazza (reproduced here as fig. 1) uses genetic
data from 49 human groupings in Africa. The most fundamental
differentiation shown is that between sub-Saharan
Africans, on the one hand, and Saharan and
Northern African populations, on the other. It is among
the latter group (16 of the 49) that we see the only national
identifications of populations (Algerians, Sudanese,
Moroccans, Tunisians, Libyans, Egyptians),
whereas, as noted above, such identifications are the rule
across the Mediterranean in Europe. The other Saharan/
Northern African and all the sub-Saharan population
samples identified are denoted by (1) “traditional” ethnonyms
(Amhara, Hausa, Sandawe, and so on), (2) geographical
designations (Bantu, N.E., Bantu, S.W., and so
on), (3) designations based on language affiliation (Nilotic,
Ubangian, Volta, San), or (4) designations based on
particular physical characteristics (Pygmy/“Pygmoid”).
As noted above, appendix 3 makes the human diversity
of many of these samples obvious.
Within the Saharan/Northern African group there are
two subclusters. One is comprised of a number of “national”
North African populations (Moroccan, Tunisian,
Libyan, Egyptian), as well as Nubian, Bedouin, Berber,
and Canarian. Cavalli-Sforza et al. (1994:172) note that
there are few data available on the genetic makeup of
the Canarian group, and it is not clear whether they are
data on modern peoples or on the extinct Guanches. The
other cluster consists of a number of Northeast African
populations, including the national Sudanese and a generalized
Cushitic-speaking group. Within this Northeast
African group, one subcluster does not conform to historical
or linguistic expectations, as it associates the Algerian
national population fairly closely with the Beja of
eastern Sudan and somewhat more distantly with the
Berber-speaking Tuareg; the latter group thus appears to
be rather distinct from other Berber populations. Cavalli-
Sforza et al. (pp. 172–73) posit an ancient relationship
between Tuareg and Beja, largely, it appears, on the basis
of their shared status as pastoralists.
There are no other
data that I know of that indicate such a link, and in any
case it does not explain the putative close relationship
of modern Algerians to both groups. The researchers note
that they have data on relatively few genes from this
Algerian group, but it should be pointed out that such
small data sets are no obstacle to the acceptance of particular
genetic associations when these fit their expectations—
the Canarian case noted above is a good example.
This pattern of acceptance and rejection of
generally equivalent data on the basis of their concordance
with expectations is common throughout The History
and Geography of Human Genes and is an example
of the strategy of post-hoc accommodative argument
noted by Clark (1998).
The distinction between Saharan/Northern African
populations and peoples living in sub-Saharan Africa is
explained by the varying contribution of genes from
“Caucasoid” populations in Europe and Southwest Asia
to the former. This is very likely a contributing factor,
given the archaeological and historical evidence of such
population interactions around the Mediterranean. It is
also quite likely that clines in gene frequencies across
the Sahara are in part the result of natural selection operating
upon characteristics that are not adaptively neutral
in the very different environments through this region.
There is a significant amount of evidence for both
climatic and latitudinal effects upon different gene frequencies
(Cavalli-Sforza et al. 1994:143; Mastana, Constans,
and Papiha 1996; O’Rourke, Suarez, and Crouse
1985; Spitsyn et al. 1998). The greater instability of Saharan
environments through time probably offered less
scope for such in situ adaptation than is the case among,
for example, the Nile Valley populations examined by
Brace et al. (1996).
Saharan and Sahelian groups (various Berber- and Arabic-
speaking populations, including Tuareg and groups
subordinated to them, such as the Bella and the Haratine
and Saharan-speakers such as the Chaamba, Reguibat,
Teda, and Kanembu) are not covered in detail in the work
(Cavalli-Sforza et al. 1994:173), although investigations
of biological variation among those populations have indicated
that their anthropometric and genetic affiliations
are very diverse and complex (Froment 1999). This lack
of data on intermediate groups may make human physical
and genetic distinctions across the Sahara appear
more clear-cut than they are. The status of these populations
is particularly important given that climatic
change rendered significant parts of the Sahara passable
(and in some cases habitable) through periods in the Holocene
at least, with the result that there is abundant
evidence of more extensive human contacts across the
desert than have existed in historic times. Sutton (1974)
and Ehret (1993) have suggested that the Saharo-Sudanese
Neolithic tradition was largely the province of Nilo-
Saharan-speakers. Populations speaking those languages
do not, however, occupy an intermediate position between
North African and sub-Saharan African populations,
suggesting that either the correlations between archaeology
and linguistics or those between genetics and
linguistics—or both—are erroneous.
While Cavalli-Sforza et al. emphasize the contribution
of immigrant genes to the modern genetic makeup of
Saharan/Northern African populations, they do not really
consider the possibility of an African genetic contribution
to either Europe or the Near East. It thus appears
that Africa accepts genetic contributions from
other areas but does not reciprocate them. A principa lcomponent
map of 42 world populations (Cavalli-Sforza
et al. 1994:82) indicates a somewhat more complex picture,
with a succession of Basques, Sardinians, Near Eastern
populations, and Berbers occupying a space intermediate
between African and European populations,
although certainly arrayed closer to European groups.
This assumption is also at variance with the known history
of the region, where we see evidence for two-way
relations throughout the Holocene, especially via Southwest
Asia and the Iberian and Italian peninsulas. People
from North, Saharan, and sub-Saharan Africa have
crossed the Mediterranean as settlers, conquerors, and
slaves through recorded history just as have Europeans.
In recent times such population flows may have tended
to be from north to south, but it should not be assumed
that this has always been the case.
Pygmies, Khoisan, and Caucasian connections. The
high-level cluster of sub-Saharan African populations
contains 33 of the 49 populations of the phylogenetic
tree. It is a considerably more diverse grouping than the
Saharan/Northern African, with multiple subclusters at
different fissioning points. The most famous “outlier”
populations of traditional African ethnography are of
course Pygmy and Khoisan-speaking groups, which are
to varying degrees physically distinct from their African
neighbors and also to varying degrees participate in foraging
economies. These latter are frequently seen by
Westerners as archaic, and Pygmy and Khoisan populations
have often been identified as unchanged relics of
earlier ages (e.g., Thomas 1959:6–8;Turnbull 1983:11–13,
157–58). Pygmy (Mbuti and Biaka) and “Pygmoid” populations
are found at various points on Cavalli-Sforza et
al.’s phylogenetic tree as outliers and with other groups.
As Froment (1998) points out, this separation of Pygmy
and other African populations is extremely imprecise; it
depends to a great extent upon linguistic criteria, ignores
the numerous transitional populations (not only those
denominated as “Pygmoid”), and systematically discounts
the fact that we know very little about the historical
and physical relations between these groups over
any significant period of time.
Similarly, Khoi and San populations cluster with a Somali
sample (which itself is held to be out of place, given
that Somali groups geographically sit within the
Northern African range), while Sandawe clusters with
populations from Senegambia and Hadza is an outlier
between the two. Cavalli-Sforza et al. (1994:169–70,
174–77, 189–93) posit that especially San populations are
the result of admixture between “Caucasoid” groups
originating in Southwest Asia and African “Negroid”
groups. This is supposed to be a different process of interaction
across the Red Sea from the one that yielded
the distinctive genetic and physical characteristics of
Ethiopian populations; indeed, the San and Ethiopian
peoples are held to be “similar to Caucasoids but . . .
otherwise very different [from one another]” ( p. 191).
The historical mechanisms—and even the demographic
meaning—of such multiple similarities are left unspecified.
This is unfortunate, given that hypotheses of immigration
into Africa by (often “Hamitic”) “Caucasoids”
have bedeviled African history and archaeology for much
of the past century, often being advanced to explain away
African cultural innovations and based on very unsatisfactory
evidence. One would have hoped that consciousness
of this situation would have led the authors
of The History and Geography of Human Genes to substantiate
this hypothesis in detail.
The nongenetic evidence marshaled in support of the
hypothesis of relations between San groups and populations
in the Near East is extremely weak. A putative
“Asian” genetic contribution to forager groups in Ethiopia
(Nijenhuis and Hendrikse 1986) is discussed only
with reference to “Pygmoid” populations, although Cavalli-
Sforza et al. (1994:174) imply that these groups are
related to the San. They claim ( pp. 160, 176) that skeletal
material “credibly identified as San” has been found in
various parts of North and East Africa, including northern
Egypt, but note only parenthetically that this assertion
in Nurse, Weiner, and Jenkins (1984) is based upon
a 30-year-old paper by Philip Tobias (1968 [1964]).
The Tobias paper does not in fact seem to make that claim,
and it is in any case disputed by more recent researchers
on the basis of the characteristics of the material involved,
the very fragmentary state of the collections, and
known problems with the accumulation of Khoi and San
skeletal reference collections (Froment 1998; Morris
1986, 1987; Rightmire 1975; 1984:193–98; Schepartz
Fig. 2. Principal-component analysis of 49 African populations. BNT,
Bantu and Bane; C, Central; ETH, Ethiopia;
KH, Khoisan; N, North Africa; NS, Nilo-Saharan; PG, Pygmies; W, West Africa (Cavalli-Sforza, Menozzi,
and Piazza 1994:fig. 3.5.2; q1994, Princeton University Press, reprinted by permission of Princeton University
Press 1988).
In fact, the identification of this skeletal material
from northeastern Africa as related to San skeletal material
from southern Africa is very doubtful; the material
indicates that ancient populations in the area were most
closely affiliated with the present-day inhabitants.
The only widely accepted evidence of ancient Khoisan
populations in East Africa is the ascription of the Sandawe
and Hadza languages to the Khoisan phylum (with
even less well-attested traces of Khoisan contacts in Dahalo
and Yaaku [Ehret 1974:11, 88]). However, the Khoisan
affiliations of Sandawe and/or Hadza are still disputed
by some linguists, and in any case the available
genetic data do not indicate a close relationship between
Sandawe and Hadza people, on the one hand, and San
and Somali people, on the other. The paradox is obvious:
Sandawe and Hadza provide the only firm link between
San populations and northeastern Africa (a linguistic
one), but according to the genetic data that provide the
basis for The History and Geography of Human Genes
they are more closely related toWest and Central African
groups (fig. 2).
There seems to be no a priori reason to
associate Khoisan-speaking populations with Southwest
Asia on the basis of San genetic data and not to associate
Khoisan-speaking populations with Senegambia on the
basis of Sandawe genetic data, but this is just what Cavalli-
Sforza et al. do. It is also, of course, possible that
either or both associations are spurious, especially given
the small size of some of these forager groups and the
attendant possibility of genetic drift.
We must remember, as well, that the southern tip of
Africa is not in fact occupied only by Khoisan and other
African peoples, as the discussion in Cavalli-Sforza et al.
(1994:158–68) would sometimes seem to imply. It is also
the only part of the continent to have seen extensive
settlement by Europeans (and Asian people arriving in
the context of colonization) over the past four centuries.
This has certainly involved a considerable amount of
genetic interchange between indigenous and immigrant
populations, along with the formation of a number of
communities of mixed descent, with genetic, cultural,
and linguistic affiliations to Khoisan, other African, European,
and Asian groups.
These communities and their
origins are discussed at some length in Nurse, Weiner,
and Jenkins (1985:221–40); they have at various times
and in various areas included !Kora, Baster, Orlam, Gyzikoa,
Griqua, and “Cape Coloured” populations, among
others, and have occupied territories in South Africa,
Namibia, and Botswana. As Morris (1992, 1997) points
out, “Caucasoid” genetic and morphological features
seem to be less important among Griqua populations (the
only cases that I know of for which these affinities for
these groups have been studied) than might have been
expected on historical grounds, but these groups are quite
diverse, and rather little is known of genetic exchange
between European and African populations on the frontiers
of European settlement in southern Africa.
It would seem reasonable to seek evidence for “Caucasoid”
genetic affiliations among southern African populations
in this recent and relatively well-attested contact
situation rather than in putative Khoisan contacts
with Northeast Africa and Southwest Asia for which no
firm evidence exists. Indeed, historical data might lead
us to expect such genetic results. This possibility seems
to be comprehensively edited out of The History and
Geography of Human Genes, presumably because European
settlement in southern Africa took place after
a.d. 1492. This claimed association of Khoisan people
with “Caucasoids” is quoted in more recent works on
genetics and history in Africa, among them the work by
Passarino et al. (1998) on Ethiopian populations. This
latter paper exhibits the same reliance upon inappropriate
sources as does Cavalli-Sforza et al., with for example
use of a 1964 Encyclopedia Britannica article as a primary
reference on the prehistory of Ethiopia.
West African cases. As noted above, the Sandawe sample
clusters with samples from westernmostWest Africa,
specifically from Serer, Wolof, and Peul populations. It
was this latter group that first caught my attention when
I looked at Cavalli-Sforza et al.’s phylogenetic tree because
of the distinction between the Peul and Fulani
samples that I have mentioned. These two ethnonyms
both denote Fulbe people, but the Fulani sample is
closely associated with a set of samples from Nigerian
populations, including Ibo, Yoruba, and Hausa—groups
a long way from Senegambia. Cavalli-Sforza et al. recognize
the peculiarity of a situation in which one ethnonym
encompasses two distinct genetic populations
and respond by recalling Murdock’s (1959:414–15) distinction
between pastoral Fulbe, who are supposed to be
“Caucasoid-like,” and settled Fulbe, who are “Negroid.”
They intimate that the Peul sample, which also contains
genetic material from people identified as Tukolor, thus
contains more admixture from “Caucasoid” (specifically
Berber) populations. However, this well-known distinction
between “cattle Fulani” and “town Fulani” is very
frequently overdrawn (for other considerations of the issue,
see e.g., Curtin 1975:19–22; Irwin 1981:46–55;
Schultz 1981), the Wolof and Serer populations that cluster
with the Peul are not particularly “Caucasoid,” and
these westernmos tWest African groups are certainly distinguished
from Saharan populations according to Cavalli-
Sforza et al.’s principal-component analysis (1994:
170).
It is more likely that this distinction between the populations
recorded as Fulani and Peul can in great part be
traced to social/political processes over the past 200
years in West/Central Africa, where among other events
the jihad of Uthman dan Fodio led to the establishment
of the Sokoto Caliphate in modern Nigeria and Cameroon
and to an attendant favorable change in the status
of a Fulbe ethnic identity. Under these circumstances,
we would expect that many indigenous people, especially
those of Hausa background but also including
members of other Nigerian and Cameroonian societies,
would attempt to assimilate to that Fulbe identity. Indeed,
such processes are often discussed and debated in
the Nigerian historical literature (Salamone 1985, Smith
1997) and can probably be seen in play in the mixed
ethnic affinities of the “Fulani” sample in appendix 3 of
The History and Geography of Human Genes (Cavalli-
Sforza et al. 1994:470).3 Difference between “Peul” and
“Fulani” thus appears to be not the result of a longstanding
division of two populations subsumed under
one ethnonym but quite the opposite—the relatively recent
creative manipulation of identity for political and
social advantage.
The significance of this conclusion lies not so much
in the ethnographic concepts used by Cavalli-Sforza et
al. as in the possibility of such a drastic change in lin-
3. The Hausa, we should note, become Nilo–Saharans rather than
speakers of Afroasiatic languages in the principal-component maps
of Cavalli-Sforza et al. (1994:170, 181; see fig. 2) and “elongated”
pastoral nomads a` la Hiernaux in the conclusions to the chapter
on Africa (p. 194). These incorrect ascriptions indicate the difficulty,
especially for nonspecialists, of working with such very large ethnographic and linguistic data sets.
This difficulty probably also accounts
for citations of, for example, works dealing with Chukotskii
and Brazilian populations in references to African groups (p. 471).
linguistic and ethnic affiliation by a significant group of
people in less than two centuries. If such a change happened
in northern Nigeria, how common is it elsewhere,
and how many such changes affect the analyses in The
History and Geography of Human Genes? The data given
in this work indicate that such processes may be quite
common on the African continent. In general, the correspondences
between genetic relationships and those
based upon linguistic and “tribal”/ethnic affiliation do
not appear to be particularly close. Cavalli-Sforza et al.’s
(1994:182–85) discussion of these correspondences
abounds with perplexing relationships between different
groups of people—Nubians with Moroccans, Ubangian speakers
with southwestern Bantu populations, Sara
speakers of a Nilo-Saharan language with Hausa on the
one hand and Biaka Pygmies on the other, and so on.
The “Caucasoid” genetic affiliations of the San may
prove to be another example.
Technological innovations, population expansions,
and especially migrations are invoked to explain as many
of those relationships as is possible, but these explanations
often seem ad hoc. For example, archaeological and
linguistic data involving the Bantu expansion are invoked
to explain the apparently very close genetic relations
between northwestern and southeastern Bantu
populations. Surely, however, any such explanation
should also account for the relative lack of genetic similarity
of the intervening Bantu populations (table 1). In
any case, the close relations between archaeological and
linguistic reconstructions (the former very dependent
upon the latter) in this case vitiate their separate use as
independent tests. Indeed, Vansina (1995) has cogently
critiqued the assumptions that there are close parallels
in archaeological and linguistic data bearing upon the
expansion of Bantu languages and that that expansion
was the result of a single, massive migration process.
The archaeological and linguistic data simply do not
yield the neat, uncomplicated picture of the past that
Cavalli-Sforza et al. present. The extent of this confusion
of different data sources is made evident by a map (1994:
fig. 3.9.3) that purports to show a pattern of Bantu expansion
based upon archaeological data in which most
of the linkages (between Nilotic-speaking peoples and
Central-Western Bantu or between Bane populations and
Northeastern Bantu, for example) are sustained by no
archaeological data at all. No other mechanisms for genetic
or linguistic dispersal are considered, and the con-
nection of these different types of data seems quite
unsystematic.
Synthetic maps of african genetic variation
Cavalli-Sforza et al. (1994:189–92) do not attempt to give
historical explanations for African genetic frequency variation
to the sixth principal component as they do for
Europe. (Perhaps they would have done so if there were
as many different regional historical reconstructions for
Africa as there are for Europe.) Instead, they limit their
explanations to the first four principal components,
which encompass about the same percentage of total variance
as in the European case. The synthetic map derived
from the first principal component (fig. 3) has a strong
north-south gradient, with opposing extremes along the
Mediterranean coast and from the Equator southward.
The researchers relate this to the ancient presence of
“Caucasoid” populations along the northern coast of Africa
and the gradual admixture of “Caucasoid” and “Negroid”
populations across the Sahara. They also comment
upon the paucity of data on Saharan populations.
As noted above, we should remember that this gradient
is also an environmental one and that it is quite likely
that environmental variability is affecting the genetic
makeup of African populations to some degree.
The discussion of patterning on the synthetic maps for
the second and third principal components is rather confusing.
Cavalli-Sforza et al. (1994:189) note that both
maps show maximum values in northeastern Africa, although
the specific association with Ethiopia is not as
clear as they claim. They state that the map of the second
principal component indicates a similar set of maximum
values in the area of southwestern Africa occupied by
Khoisan-speakers in precolonial times and that this indicates
that there existed a population ancestral to modern
Khoisan peoples somewhere near Ethiopia. Low values
on the second principal component in the areas
intervening between Ethiopia and the Kalahari are ascribed
to the later settlement of those areas by Bantu
peoples.
The maps (figs. 4 and 5) actually indicate that
this pattern is based upon variation in the third principal
component. The difference is significant, given that the
second principal component explains 18.6% of the total
variance and the third only 10.3%, but the text seems
to indicate that this pattern is in fact associated with
the second.
Presumably, this patterning is explained by the admixture
of “Caucasoid” and “Negroid” populations in
northeastern Africa that Cavalli-Sforza et al. believe gave
rise to the Khoisan peoples. However, the maps for the
second and third principal components show almost opposing
values along the Mediterranean coast, where the
presence of “Caucasoid” groups is held to be the explanation
for the patterning of the first. Similarly, northeastern
and southwestern Africa show maximum divergence
on the third principal component, which is
accounted for by their differences from one another in a
context of general relationship to “Caucasoid” groups.
These “Caucasoid” genetic contributions are thus invoked
as the primary explanation for the variation exhibited
in the first, second, and third principal components
but seem to refer to different sets of genetic
characteristics and thus presumably to different populations
of “Caucasoids.” These interpretations would
seem to lack much real explanatory value.
The fourth principal component accounts for only
7.0% of the total variance (fig. 6). Maximum values are
associated with the northwestern part of the Central African
tropical forest and minimum values with the border
regions of Mali, Niger, and Burkina Faso. Parsimoniously,
the researchers associate these maximum and
minimum values with two different agricultural expansions.
They associate the area of maximum values with
an expansion of Bantu populations, although one might
associate the patterning with Adamawa-Ubangian
groups as well (David 1982:88–91; Saxon 1982) and the
influence of this massive Bantu migration upon genetic
variation would seem to be extraordinarily low compared
with that of the “Caucasoid” gene flows that have determined
the nature of the first three principal components.
They admit that there is no equivalent evidence
for an agricultural expansion from the area of minimum
values in the West African Sudanic zone but counsel
future comparison with archaeological data. These syn-
thetic maps sometimes appear to function as historical
Rorschach tests, with genetic data being associated with
ancient cultural processes on the basis of general resemblances
in spatial patterning.
Conclusions
The History and Geography of Human Genes is an extraordinary
compendium of information on human genetic
variation. It links a huge amount of data on that
variation with archaeologically, historically, and linguistically
known cultural processes that took place at
various times and in various areas. Its hypotheses have
been provocative and in many cases useful, and researchers
in other fields have responded to them in ways that
will eventually increase our knowledge of our shared
human past. In the long run, Cavalli-Sforza et al. have
probably established the examination of patterns in genetic
characteristics as a valuable complement to older,
more established avenues for the examination of prehistory.
This genetic research was limited by the lack of
samples from many areas, by the necessity of using samples
collected under very different conditions, by the
great variation in the amount of information available
on the composition of the groups being sampled, and by
the lack of availability of statistical techniques that
could account for interactions between groups during the
process of constructing phylogenetic trees. To their
credit, Cavalli-Sforza, Menozzi, and Piazza recognize and
highlight a number of these problems and try to compensate
for them. Their solutions are often sophisticated
and ingenious, and the potential of these techniques is
obvious.
The danger for African archaeology is that the very
impressive amount of genetic data amassed in The History
and Geography of Human Genes will convince nonspecialists
of the validity of the hypothesized associations
between human groups. That would be unfortunate,
because this pioneering synthesis is hobbled by
a conceptual model of human behavior that takes little
account of ethnographic, historical, linguistic, and archaeological
research on the constitution of human societies
and of the various ways in which cultural meaning
is constructed in the continuing negotiation of
individual and group identities. Similarly, the discussion
of relations between language and ethnicity pays virtually
no attention to linguistic analyses of such relationships.
Their compilation of the data that make up the
African sample should have indicated to the researchers
that these relations are extremely complex, while the
varying criteria—linguistic, geographic, national—used
for identifying populations make it seem unlikely that
the groups under study are comparable entities.
African ethnic units are not bounded, homogeneous
monoliths either frozen in place since before a.d. 1492
or caromming around the continent like culture-bearing
billiard balls. They are dynamic entities, manipulated in
response to changes in the natural and social environment,
and their composition can change very quickly
indeed. Differences in the temporal resolutions of genetic,
linguistic, archaeological, and historical analyses
are one of the main difficulties in these multidisciplinary
studies, and the ability to control chronologies is more
restricted in the former two disciplines than in the latter.
It is often difficult, therefore, to establish whether the
genetic patterning exhibited by modern populations can
be traced back to events that happened centuries or millennia
ago. A number of the groups used in the analyses
in The History and Geography of Human Genes did not
exist as such five centuries ago, and the modes of human
interaction in Africa are not in any way reducible to a
list of technological innovations and mass migrations.
In part because of this lack of chronological control, the
procedures used by Cavalli-Sforza et al. to associate genetic
patterning with cultural events in the past are quite
unsystematic, involving attempts to fit secondhand
knowledge of archaeological, ethnographic, and linguistic
reconstructions into a Procrustean bed of genetic patterning.
Clark’s (1998) description of the “ransacking”
of European genetic data for meaning applies equally
well in the African case.
The deme concept, often used in population genetic
research in other species, would seem to be a better and
more flexible fit for the study of human genetic variation
mac eachern Genes, Tribes, and African History F 371
than “tribal” typologies borrowed wholesale from ethnographic
research done 50 or more years ago, especially
since it seems to find resonances in some more recent
concepts of ethnicity. It remains to be seen whether it
will be applied to human genetic research in Africa and
elsewhere. It would be interesting to see, for example,
whether human demes on some level correspond to the
material distributions that we call archaeological
traditions, without any attempt to fit either genetics or
archaeology to the straitjacket of the ethnographic
“tribe.” Archaeological distributions of artifacts frequently
occur over much larger territories than do the
ethnic (“tribal”) units that ethnographers study in the
present, and it is possible that in some cases this reflects
movements by artifact producers, especially women,
that would also be reflected in patterns of genetic variation
(MacEachern 1998). In general, research on the African
past would at this point benefit far more from
small-scale studies of genetic patterning within and between
particular communities than from the continental
and subcontinental approaches that make up The History
and Geography of Human Genes.
The genetic data available for Africa have serious limitations,
but the patterns of variation demonstrated by
these data are interesting and probably do indicate patterns
of cultural interaction in the past. There are, however,
good reasons to be suspicious of efforts to correlate
those patterns with cultural processes at the level of
specificity attempted by Cavalli-Sforza et al. The project
of correlating the data from these different kinds of investigation
still lies well before us, and it would be a
great mistake to assume that such correlations will be
self-evident. Research that has as its goal the study of
the history and geography of human genes in Africa and
elsewhere must be truly interdisciplinary, working with
the best knowledge of other disciplines that can be acquired.
A study that uses the terminology of other disciplines
but takes insufficient account of the concepts
behind that terminology risks generating misleading reconstructions
of human history and prehistory.
--------------------------------------------------
Keita's critique of the critique:
MacEachern has given us an effective critique of the African
chapter in Cavalli-Sforza et al.’s book, while also
illustrating general theoretical problems in syntheses of
archaeology, ethnicity, linguistics, and human biology.
The insights offered are especially useful for those interested
in Africa, for which scholarship has long been
hampered by myths and stereotypes. Notable among
these is a paradigm which sets limits on authentic biocultural
indigenous Africanness; this surprisingly persists
in some scientific literature. The work of even some
Africanists is still influenced by this legacy, which usually
manifests itself around the problematic construct of
race. Although MacEachern has given a cogent presentation,
there are a few concerns.
One wishes that emphasis had been placed on the African
origin and differentiation of the Afro-Asiatic language
family and its subsequent spread to Asia (Greenberg
1966; Bender 1975; Diakonoff 1965, 1981; Ehret
1984, 1995; Ruhlen 1991). The dubious Nostratic construct
has been used to postulate the spread of food production
into Africa (from Asia) by Afro-Asiatic-speakers.
Significantly, reconstructed common Afro-Asiatic has no
terms for food production (Ehret 1984, 1995) and is accorded
the same time depth as Nostratic (cf. Ehret 1984
and Barbujani and Pilastro 1993). Furthermore, most
Nostraticists now conceptualize Afro-Asiatic to be a sister
to Nostratic, not a daughter (Ruhlen 1991). Finally,
archaeological data support migration from Africa into
the Near East during the time frame suggested by some
Nostratic models for immigration into Africa (cf. Bar-
Yosef 1987 and Barbujani and Pilastro 1993).
On a technical note, MacEachern has made a common
error, that of easily equating phenetic similarity with a
necessary genealogical relationship. The results of mathematical
techniques employed by human biologists must
be interpreted with sound biological principles and other
information (Rhoads 1984, Harrison 1984). The fact that
small samples of Beja and Algerians and then of Nubians
and Moroccans cluster clearly cannot be taken as necessarily
indicative of a close genealogical relationship.
The similarity may have many explanations including
chance, especially since, as MacEachern notes, the nature
of any links between earlier and later peoples having
these designations remains an open question.
This is even more so when selection, migration, gene flow, and
language shift have to be integrated into explanatory
models. Harrison (1984:61), considering the results of
simulations which evaluate interpopulation gene flow
with and without selection, points to
another important issue concerning migration and
population affinities: the causes of inter-population
patterns of gene frequency variation, and the meaning
of affinities based not on ancestry but on genetic
similarity. Clearly with varying levels of selection,
populations which are genetically similar may have
a much more remote common ancestry than populations
which are genetically more different.
The lack of criticism on this point does not, however,
seriously weaken the overall critique.
In fact, there is little to quibble with in MacEachern’s
effort. However, he is perhaps somewhat overconfident
in the belief that genetic research has led to the “dismemberment”
of racial taxonomies and racio-typological
thinking. While few write today in terms of Nordic
or Jewish races, and this is a result of research and the
World War II experience, there is still plenty of work on
Africa which constructs its interpretations of diversity
explicitly or implicitly in terms of conceptually idealized
“Caucasoid” and “Negroid” “taxa,” implying a causal or
foundational linkage between morphological complexes
and a variety of genes.
Cavalli-Sforza et al.’s book is one example of this kind
of work at some level. For some reason MacEachern does
not directly engage its underlying nonevolutionary raciotypological
model of interpretation or its use of racial
terms, which are ontologically misleading in that they
imply that supra-Saharan Africans were originally European-
derived. These people supposedly interacted with
“Negroids” and others, receiving some genes but contributing
a lot more and thus becoming the primary explanation
of African diversity. MacEachern’s critique
would be even more useful if he had pointed out that
(1)fossils indicate the presence of anatomically modern people
in supra-Saharan and Nile Valley Africa at a time
when hominids in Europe had Neanderthal morphology;
(2) coalescence (and therefore differentiation) times for
numerous genes sampled globally from living humans
date to a period when modern morphology had not yet
emerged or was to be found only in Africa; and
(3) global genetic diversity seems largely to be a subset of that
found in Africa. (Whether this dates back to the time of
Homo erectus or modern H. sapiens matters little; the
diversity denoting the mythical “racial divergence” exists
in Africa, as should be expected, and antedates the
existence of the morphologies used to define races.)
MacEachern fails to make these points, which together
indicate that the African genetic profiles and morphologies
being interpreted as solely resulting from European
(or Southwest Asian) colonization and/or admixture
(with “Negroids”) are largely the product of various authentic
intra-African biohistorical processes which perhaps
date to a time before there were any anatomically
modern Europeans.
However, the Holocene climatic
fluctuations of the Sahara (Hassan 1988), probable Late
Pleistocene (and Holocene) population increase in response
to changing subsistence regimes (Wetterstrom
1993), and the spread of early Afro-Asiatic-speakers from
the Horn surely also had some effects. Supra-Saharan
Africa is not primarily a product of extra-African colonization.
There are many ways to be biologically African—
many morphological combinations and genes
(Hiernaux 1974, Keita and Kittles 1997).
The frequency of neutral private alleles of the right date and
demonstrably originating outside of continental Africa will
have to be determined before the non-African contribution
to Africa’s genetic picture can be fully assessed.
MacEachern seems to accept the racio-typological model
of explanation at the larger level. Nevertheless, he contributes
to the paradigm shift in understanding African
biological variation, which will reject racio-typological
thinking and use the reality of indigenous African diversity
as an analytic tool.
----------------------------------------
OTHER COMMENTATORS WEIGHED IN:
QUOTE:
alan g. morris
Department of Anatomy and Cell Biology, University
of Cape Town Medical School, Observatory 7925,
South Africa (morris@anat.uct.ac.za). 17 xi 99
"MacEachern’s other major complaint about the compendium
by Cavalli-Sforza et al. is the use of principal
components to generate synthetic maps which are then
interpreted as signs of specific past events. The principal
components are not events. They are mathematical relationships
and as such as far too complex to be given
simple historical interpretations. MacEachern, in line
with Clark’s (1998) assessment, accuses Cavalli-Sforza
et al. of having a “pattern of acceptance or rejection of
generally equivalent data on the basis of concordance
with expectations.” Simplified, this means that they saw
what they wanted to see. I entirely agree with Mac-
Eachern’s concern that many students of anthropology,
genetics, and history are going to take Cavalli-Sforza et
al.’s work at face value. By doing that, they may be uncritically
reproducing interpretations ofnature of ethnic identity into account.
This is not the first time that I have looked at the
difficulties in interpreting genetic data. The equivalent
genetic history of southern African people was published
by Nurse, Weiner, and Jenkins in 1985. Again, the work
was a wonderful production of data on a wide range of
peoples, but the authors’ attempts at linking what they
saw in the genes with history were sometimes very wide
of the mark. Social theory and social events are not easily
mac eachern Genes, Tribes, and African History F 377
explained by biology. As does MacEachern here, I ended
my comments by warning of my great fear “that students
of archaeology, genetics and physical anthropology will
simply assume that this is the academic ‘bible’ on the
subject and will absorb many of the speculative passages
without criticism”
-----------------
mark pluciennik
Department of Archaeology, University of Wales,
Lampeter, Ceredigion SA48 7ED, Wales, U.K.
(m.pluciennik@lamp.ac.uk). 11 x 99
"MacEachern notes that the various “academic and political”
uses for human genetic research raise the most
troubling questions when driven by a regime of private
property. I share his concern that Western bio-prospectors
promote an “identification of indigenous peoples
with their surroundings” in part to performa metonymy
whereby people become resources. If thus identifying indigenous
people reduces their capacity to organize politically
to protect their rights of disposition, then serious
problems will arise. In seeking to patent the material
they “mine,” the bio-prospectors will have opened a sort
of final colonialism, contained here in a chiasmus: from
control over the raw material of nature to control over
the nature of raw materials."
