вход по аккаунту


Craniometric variation and the settlement of the Americas Testing hypotheses by means of R-matrix and matrix correlation analyses.

код для вставкиСкачать
Craniometric Variation and the Settlement of the
Americas: Testing Hypotheses by Means of R-Matrix
and Matrix Correlation Analyses
Rolando González-José,1* Silvia L. Dahinten,2,3 Marı́a A. Luis,4 Miquel Hernández,1
and Hector M. Pucciarelli3,4
Unitat d’Antropologia, Facultat de Biologia, Universitat de Barcelona, 08027 Barcelona, Spain
Centro Nacional Patagónico-Consejo Nacional de Investigaciones Cientı́ficas y Técnicas,
9120 Puerto Madryn, Argentina
Consejo Nacional de Investigaciones Cientı́ficas y Técnicas, Buenos Aires, Argentina
Departamento Cientı́fico de Antropologı́a del Museo de La Plata, Facultad de Ciencias Naturales y Museo,
1900 La Plata, Argentina
Amerindians; craniometrics; R-matrix methods; matrix permutation
New archaeological findings and the incorporation of new South American skull samples have
raised fundamental questions for the classical theories of
the Americas’ settlement. The aim of this study was to
estimate craniometric variability among several Asian
and Native American populations in order to test goodness
of fit of the data to different models of ancient population
entries and dispersions into the New World. Our data set
includes Howells’ variables recorded on East Asian, North
American, and South American natives (except for NaDene speakers). Five Fuego-Patagonian samples and one
Paleoamerican sample were also included. A multivariate
extension of the R-matrix method for quantitative traits
was used to obtain Fst values, which were considered
estimations of intergroup variation. Three main models
for the peopling of the New World were represented in
hypothetical design matrices. Matrix permutation tests
An inspection of the last 15 years’ literature in
biological anthropology evidences a great amount of
scientific activity referring to the original settlement
of the Americas (Greenberg et al., 1986; Guidon and
Delibrias, 1986; Neves and Pucciarelli, 1990, 1991,
1998; Szathmáry, 1993a,b; Torroni et al., 1993; Merriwether et al., 1995; Neves et al., 1999a,b; González-José et al., 2001). In fact, this subject has been
the topic of numerous disciplines such as archaeology, genetics, linguistics, physical anthropology, and
quaternary research. As expected, several hypotheses about the time and characteristics of human
entry into the New World arose and at the moment
compete in the explanation of that process.
Archaeologists and quaternary researchers tend
to focus the debate on the dates of first peopling
(Borrero and McEwan, 1997; McCulloch et al.,
1997). Radiocarbon dates and stratigraphy are of
major importance because they allow us to calibrate
hypotheses about the processes involved in the settlement. Despite the presence of archaeological sites
were performed to quantify the fit of the observed data
with 1) geographical separation of the samples and 2)
three ways of settlement, which were the Three Migration Model (TMM), the Single Wave Migration model
(SWM), and the Two Components Settlement Model
(TCS). R-matrix results showed high levels of heterogeneity among Native Americans. Matrix permutation
analyses suggested that the model involving high Amerindian heterogeneity and two different morphological
patterns or components (derived “Mongoloid” vs. generalized “non-Mongoloid”) explains better the variation
observed, even when the effects of geographical separation are removed. Whether these patterns arose as a
result of two separate migration events or by local evolution from Paleoamericans to Amerindians remains
unresolved. Am J Phys Anthropol 116:154 –165, 2001.
2001 Wiley-Liss, Inc.
in Tierra del Fuego and other regions by at least
11,000 –10,500 years ago, and despite the discovery
of several sites of pre-Clovis dates in South America,
the “Clovis First” model is still accepted by some
archaeologists. Nevertheless, solid evidence at the
Monte Verde site in southern Chile and other localities now indicates that dates of entering the New
World proposed by the Clovis model are incongruent
with any possible human migration from Beringia to
the southern tip of South America (Dillehay and
Collins, 1988; Roosevelt et al., 1996; Borrero and Mc
Ewan, 1997; Meltzer, 1997).
Geneticists vary in their opinions about the meaning of molecular variability in Amerindians. In a
*Correspondence to: Rolando González-José, Unitat d’Antropologia,
Facultat de Biologia, Universitat de Barcelona, Avinguda Diagonal
645, 08028 Barcelona, Spain. E-mail:
Received 24 April 2000; accepted 26 June 2001.
classic work, Greenberg et al. (1986) reviewed gene
and/or genotype frequency distributions, and haplotype frequencies. Most of the data were serological
in nature, involving blood-group antigens, serum
proteins, erythrocyte enzymes, immunoglobulins,
and leukocyte antigens. The interpretation of this
evidence led the authors to support the inference
based on linguistic and dental data: the Three-Migration Hypothesis. They, however, stated that from
a genetic perspective, the tripartite division of modern Native Americans is still without strong confirmation.
Later studies examined variation of blood polymorphism (Salzano and Callegari-Jacques, 1988),
Y-chromosome DNA (Bianchi et al., 1997, 1998),
mtDNA (Schurr et al., 1990; Torroni et al., 1993;
Bailliet et al., 1994; Merriwether et al., 1995; Lalueza et al., 1997), and nuclear DNA (Novick et al.,
1998; Da Silva et al., 1999) in Native Americans. In
general terms, these authors disagree about the existence of a severe bottleneck, about the coalescence
times of purported ancestors, and about the amount
of variation to have entered the continent (for reviews, see Szathmáry, 1993a,b; Lahr, 1995; Crawford, 1998). Overall, the genetic evidence for the
origins of Amerindian variation does not yet provide
a clear picture, although it indicates genetic heterogeneity within the “Amerindian” group of Greenberg et al. (1986; see also Szathmáry, 1993b; Lahr,
Linguistic studies were a major focus of interest in
the debate about the settlement of the Americas,
and an important support for the Three-Migration
Model. In fact, Greenberg et al. (1986) argued for a
clear separation of American natives into three different linguistic families: the Na-Dene speakers, the
Amerind speakers, and the Eskimo-Aleut speakers.
Nevertheless, this point of view received considerable criticism on linguistic grounds (Morell, 1990).
Morphological studies are not the exception in
terms of controversies, since different models seem
to be supported by the evidence. The dental and
craniometric evidence strongly supports an Asian
origin of Amerindians and Eskimos. The shovel
shape of the incisors of Northeast Asia and New
World populations occurs in 50 –100% of subjects, in
contrast to European and other Asian groups where
the frequency of this trait is under 50% (Harris,
1980; Turner, 1987). After a multivariate analysis
performed on measurements of skull samples from
different geographic sites, Howells (1989) noted that
Native American crania are placed metrically
among the East Asiatic groups. Simply put, it seems
like a morphological continuity from North-Asian to
Amerindian populations, with Eskimos and Aleuts
in an intermediate position (Crawford, 1998). However, several authors have challenged such a
scheme. Certainly, they state that Paleoamerican
cranial morphology is very distinct from that of modern native Indians and from that of Northern Asians
(Neves and Pucciarelli, 1990, 1991, 1998; Lahr,
1995; Neves et al., 1999a,b). Moreover, other authors were unable to attribute a derived East Asian/
Amerindian morphological pattern, to the FuegoPatagonian morphological series (Lahr, 1995;
González-José et al., 2001).
Despite the complexity and variety of interpretations, we can identify three clear and documented
ideas or hypotheses about the way in which the
continents were peopled. In the first case, we should
consider the Three-Migration Model, proposed in a
series of well-known papers by Turner (1987) and
Greenberg et al. (1986). This model is based mainly
on linguistic, dental, and genetic data studied across
Asian and Amerindian populations. In this model,
ancestral Amerind-speaking people would have undertaken the first (and oldest) migration into the
continent, followed by a second migration that involved the Na-Dene speakers, who currently inhabit
the interior of Alaska, the Yukon and Northwest
Territories of Canada, and the North Pacific coast
(with some large pockets of population further
south). The last independent group of people to enter
the continent would have been the Eskimo, who at
the moment inhabit the Arctic. However, it is unclear whether the authors regard the Na-Dene
speakers as the second or third migratory wave
(Greenberg et al., 1986; Turner, 1987).
Secondly, we may consider theories based on a
single migration into the New World. To be accurate,
this concept can be traced back to the sixteenth
century, when the Spanish naturalist José de Acosta
(Acosta, 1962 [1589]) first enunciated the idea of a
single origin for all Native Americans. Almost three
centuries later, Hrdlička (1925) picked up this theory and developed the notion of a “Mongoloid” ancestry for the Americans. Evidence to support a
“one-migration” model, including the possibility of a
migration before the coalescence of the great North
American ice shields, has been published by Szathmáry (1981, 1984, 1993a,b). More recently, and after
the analysis of the distribution of four founding mitochondrial DNA lineages, Merriwether et al. (1995)
concluded that a single migration and/or single
source is the most parsimonious explanation for this
distribution. Later studies on mtDNA (Bonatto and
Salzano, 1997a,b) and Y-chromosome polymorphisms (Bianchi et al., 1997, 1998) also defended a
one-migration and/or population source for Native
Finally, we consider the model proposed by Neves
and one of us (Neves and Pucciarelli, 1990, 1991,
1998; Neves et al., 1999a,b). This model was developed after several cranial comparative analyses of
the first known South and North American crania,
and rests on the fact that the morphology of the
Paleoamerican crania is more consistent with a
“non-Mongoloid” definition, and less “Sinodont-Mongoloid” than are recent Amerindians. According to
the authors, the best way of interpreting the observed cranial variation involves an earlier generalized “non-Mongoloid” population occupying the
TABLE 1. Populations considered in this analysis, codes, sample sizes, geographic origin, and sample description’s references
Sample size
Geographic origin
Description in
Hainan Island
North Japan
South Japan
Inugsuk Eskimo
Early Arikara
Santa Cruz
Yauyos, Peru
Chilean Araucano
Fueguian marine
Fueguian terrestrial
Fueguian marine
Patagonian terrestrial
South China
Honan Province, East China
Siberia, Baikal Region
Hokkaido, Japan
North Kyushu, Japan
West and southeast Greenland (as far
as Scoresby Sound)
South Dakota, USA
Santa Cruz Island, California, USA
50–100 km southeast of Lima, Peru
Lagoa Santa sites, Minas Gerais, Brasil
Central Chile
Chilean Pacific Coast, Chile
This work
Isla Grande, Tierra del Fuego, Chile,
and Argentina
Beagle Channel, Chile, and Argentina
Northeast Patagonia (Chubut and Rı́o
Negro), Argentina
This work
Howells, 1973
Howells, 1989
Howells, 1989
This work
This work
This work
This work
Americas. Native Americans can be, in consequence,
divided into two “components:” one early Paleoamerican component (not necessarily Paleoindian),
showing a generalized morphological pattern, and a
later component of derived characteristics, related
to modern Asiatic groups and including past and
modern Amerindian populations.
Even when several authors have discussed these
and other hypotheses, the attempts made to test the
models in numerical or statistical terms are almost
nonexistent. In this context, R-matrix methods
(Harpending and Jenkins, 1973) and matrix permutation test (Sokal et al., 1992; Waddle, 1994) could
be of great importance as an approach to test the
explanatory power of the several theories and mechanisms proposed.
An R-matrix is the normalized covariance matrix
of allele frequencies across populations. R-matrix
analysis has several advantages over other methods
of estimating genetic similarities and distances
(Relethford and Harpending, 1994). Moreover, it can
be applied to a wide range of data, including allele
frequencies (Harpending and Jenkins, 1973), metric
traits (Relethford and Blangero, 1990), and migration matrices (Rogers and Harpending, 1986). Further details of these methods are provided by Williams-Blangero and Blangero (1989), Relethford and
Blangero (1990), Relethford and Harpending (1994),
Relethford (1996), and Relethford et al. (1997).
The matrix permutation techniques were suggested and preliminary developed to test human
settlement and dispersion models for Europe by
Sokal et al. (1992). Waddle (1994) published an outstanding work in which three main models for the
origin of modern Homo were numerically represented and analyzed by means of matrix permutation tests. These techniques were then applied in
other studies and/or populations (see Sokal et al.,
1997; Waddle et al., 1998; González-José et al.,
2001), and were extensively criticized, improved,
and commented on by Konigsberg et al. (1994), Cole
(1996), and Konigsberg (1997).
Our purpose here is to test three of the current
models of the Americas’ settlement using craniometric information either published or collected by us.
Simply put, we will compare observed cranial dissimilarities among several samples of Asian and
Native American populations, with hypothetical differences expected under the three competing models
proposed for the settlement of the Americas, and
with several representations of geographical separation. Analyses will focus on the validity of including
all Amerindian morphological variability (with special emphasis on South American and Paleoamerican series) in a single group, rather than focusing on
North American group relationships.
Cranial dissimilarities (distances) will be obtained following a “model bound” approach in order
to estimate specific parameters as average withingroup phenotypic variance, Wright’s Fst, minimum
genetic distances, etc. (Relethford and Blangero,
1990; Relethford, 1994; Relethford and Harpending,
1994). Matrix permutation techniques will be used
to test the fit of observed data with hypothetical
models peopling of the Americas.
We also attempt to improve and update the comparative framework by including samples from Fuego-Patagonia and Paleoamerican sites. Since these
samples were not included in some earlier studies,
we will show their importance as a source of variation when included in the Amerindian group.
The sample
Table 1 lists the populations considered in this
study, their sample size, and the geographical origin
of the samples. To avoid sex differences, only male
TABLE 2. Craniometric variables used as markers
(for definitions see Howells, 1973)
Glabello-occipital length
Basion-nasion length
Basion-bregma height
Maximum cranial breadth
Maximum frontal breadth
Biauricular breadth
Nasion-prosthion height
Nasal height
Nasal breadth
Palate breadth, external
Orbit height, left
Bimaxillary breadth
Bifrontal breadth
Biorbital breadth
Nasion-bregma chord (frontal chord)
Bregma-lambda chord (parietal chord)
Lambda-opisthion chord (occipital chord)
to the south by the Magellan Straits. The Selk’nam
were confined to the steppe regions from Isla Grande
(Tierra del Fuego) and were also terrestrial huntergatherers. Both Tehuelches and Selk’nam were
adapted to the hunting of the guanaco (Lama guanicoe), one of the four South American camelids.
Finally, we included two marine hunter-gatherer
“canoeros:” the Kaweskar, or Alakaluf (KAWE), and
the Yahgan, or Yámana (YAHG) groups. The Yahgan inhabited the coastal areas around the Beagle
Channel, south to the Brecnock Peninsula, including
Cape Horn. They are the only human group ever to
have inhabited a region below 55°S (Hernández et
al., 1997). The Kaweskar lived on islands and channels of the Chilean Pacific coast, from the Brecknock
Peninsula to the Gulf of Penas.
Biological distances
individuals were considered. In Table 2, we show the
craniometrical variables used in the present study.
In order to eliminate size influence on the variables,
a Q-mode standardization was performed, where
each original measurement is divided by the object
(individuals) arithmetic mean, calculated over all
variables (Corruccini, 1973). This cancels out size
differences by giving each individual the same average character state or magnitude over all measurements taken on it (Corruccini, 1973). Q-mode standardized measurements are analogous to the
C-scores used by Howells (1989).
Detailed information about the variables is available in Howells (1973, 1989), and original data were
downloaded from the web site
edu/howells.htm, and combined with data collected
by us, in order to cover East Asian and Native American populations. The materials used in this work
comprise 656 modern human crania divided into 16
cranial series: 6 from East Asia, and 10 from North
and South America. Since our interest is focused on
the validity of the “Amerind” or “Paleoindian” group
(sensu Greenberg et al., 1986) we included no samples of crania from Na-Dene speakers. Nevertheless,
we extend the data set of Howells (1973, 1989) to
include six new populations. One of the new American series (LAGO) corresponds to a prehistoric
sample of male individuals recovered from Paleoamerican horizons of Brazil (Lagoa Santa) with an
estimated age dating of 12,000 – 8,000 years BP
(Neves and Pucciarelli, 1991). The Araucanian
(ARAU) sample corresponds to a historical population which came from central Chile and settled at
the Argentinean Pampas from the 18th century.
This ethnic group was mainly influenced by the Andean culture and by the South American version of
the horse complex. Fuego-Patagonian samples include the Tehuelches (TEHU), Selk’nam (SELK),
Kaweskar (KAWE), and Yahgan (YAHG) groups.
The Tehuelches were terrestrial hunter-gatherers
who occupied continental Patagonia. Their territory
was limited to the north by the Colorado River and
Biological distances were assessed using Mahalanobis generalized distances (D2), after the modifications of Williams-Blangero and Blangero (1989).
These modifications assume an additive polygenic
model for the traits in which the expectation of environmental deviations is zero. The phenotypic variance, composed of genetic and environmental components (␴p2 ⫽ ␴g2 ⫹ ␴e2), must be greater than or
equal to the genetic variance (␴p2 ⱖ ␴g2). Those
authors demonstrated that “Dp2 represents a matrix
containing the minimum genetic distances derived
from the phenetic variation” (Williams-Blangero
and Blangero, 1989, p. 5).
The resulting equation can be written as:
d ij 2 ⫽ r ii ⫹ r jj ⫺ 2r ij
where rij are the elements of an R-matrix computed
for each trait in populations i and j (Relethford et al.,
The diagonal elements rii also give the genetic
distance of each population to the group centroid,
and the average diagonal element of the R-matrix
weighted by population size is equal to Wright’s Fst,
a measure of average genetic differentiation relative
to the contemporary gene pool (Relethford, 1996).
The level of craniometric differentiation was assessed here from both Fst values, and a biological
distance matrix. This matrix (BIO) represents the
distances or dissimilarities (dij2) based on 17 Howells craniometric variables observed in 13 populations and obtained following the methodology described above. Equal relative population sizes were
assumed for the hunter-gatherer groups (ESKI,
twice for the nonhunter-gatherers.
Analysis of data and computation of distances
were performed using the software RMet for Windows, version 4.0 (Relethford, 1998), provided by Dr.
J. Relethford at the World Wide Web (http://konig.
Spatial separation
In order to test geographical patterns of cranial
variation, we constructed two matrices of spatial
distance. When testing geographic variation for significant departures from randomness, investigators
will want to know the effects of coding their data in
various ways (Sokal, 1979). The choice of a given
geographic distance model depends on the nature of
the mechanisms involved in the origin of geographic
variation. Thus, we focused our analyses on two
different patterns of geographic separation. Firstly,
we assumed a simple isolation-by-distance model, in
which biological distance should increase with increased geographic distance (Wright, 1943). If one
consider a process working on a continuous surface
between all pairs of localities, an ordinary geographic distance matrix should be used (Sokal,
1979). In this matrix (GEO), the elements are equal
to the linear distance in kilometers between populations i and j. Obviously, distances between Asian
and American samples were computed considering a
route across the Bering Strait rather than direct
distances to avoid illusory (transoceanic) distances.
Secondly, one may state that phenetic differences
will be for short distances and that all the great
distances have the same effect and can be grouped
into a non-neighbor group. This is the case with our
asymptotic distance connectivity matrix (ADM), in
which distances between neighbor localities are
equal to their actual separation in thousand of kilometers, and the distance between not-connected localities get a constant value slightly superior to the
higher distance observed between any pair of connected localities (9,500 km in our sample). Following
Sokal and Oden (1978), we consider that A is the
nearest neighbor of B if no other locality lies on or
within the circle whose diameter is the line AB.
Thus, A and B are connected when d2AB ⬍ d2AC ⫹
d2BC, where d2AB is the squared distance between A
and B, and C is any third locality.
Models and design matrices
In a typical matrix permutation study, dissimilarities between the samples are estimated after any
character observable in the population, and are then
represented in a distance matrix. Then, hypothetical
dissimilarities expected under a particular model are
set and described in connection schemes and design
matrices (Waddle et al., 1998). A design matrix describes the relative distances among populations expected under a particular model (Waddle et al.,
1998). An element of a biological distance matrix, as
much as an element of a design one, describes the
strength of the link between two populations. Construction and handling of design matrices was welldescribed in several papers by Sokal et al. (1992,
1997), Waddle (1994), and Waddle et al. (1998).
Mantel tests (Mantel, 1967) are usually used to compare the degree of association between the observed
and the design matrices. An appropriate design ma-
Fig. 1. Connection scheme for one model of human settlement
of the Americas: the Three-Migration Model (Greenberg et al.,
1986). Numbers represent hypothesized distances.
trix is critical to successful discrimination of an observed distance matrix (Sokal et al., 1997).
We intend to evaluate here three main documented models of dispersion of humans into the
Americas. Models are quite simple and, of course,
can be improved in several ways. Nevertheless, simple models are preferable and easy to express upon
design matrices.
In Figure 1, we show the connectivity scheme
corresponding to the Three-Migration Model (TMM).
In this scheme, samples are regrouped in four boxes.
The first corresponds to Atayal (ATAY) and Hainan
(HAIN) samples, and represents the more generalized sundadont dental pattern, from which typical
“Mongoloids” are derived (Turner, 1992; Lahr,
1995). Turner (1987) also recognizes a derived expression of this dental complex, typical of Eastern
and Northeastern Asian populations and Amerindians, which he named sinodonty. Sinodont morphology is derived from the sundadont and is represented by the Anyang (ANYA), Buriats (BURI),
North Japan (NJAP), and South Japan (SJAP) samples grouped in a single box and playing a role of
“source” for the peopling of the Americas. In fact,
from this group we derived the hypothetical first
wave, represented in a third box containing all the
Amerindians and, separately in a fourth box, the
Eskimos (Greenberg et al., 1986). A distance of zero
was assumed for samples included in the same box
and a value of one for samples of different but connected boxes, except for the connection between the
sinodonts and Amerindians, which is assumed to be
equal to two in order to represent an older separation and divergence from the Asian stock in comparison with Eskimos. As a general rule for all models,
the distance between any pair of samples of nonconnected boxes was obtained by adding together the
values along the path (e.g., distance between HAIN
and SANT is equal to 3).
We also assembled a connectivity scheme for the
Single Wave Migration Model (SWM) (Fig. 2). It
represent the conclusions obtained by Merriwether
Fig. 2. Connection scheme for one model of human settlement
of the Americas: the Single Wave of Migration Model (Merriwether et al., 1995). Numbers represent hypothesized distances.
et al. (1995) after examining distribution patterns of
the four founding mtDNA lineages. These authors
postulated that all Native Americans came from a
single migratory wave as did Bonatto and Salzano
(1997a,b). The model was symbolized here by simply
joining the Eskimo sample within the Amerindian
box into the anterior scheme. This scheme attempts
to equate a molecular common ancestor with a morphological one. The elements on the design matrix
were obtained from the SWM connection scheme in
the same manner as for the preceding model.
The Two-Components Settlement Model (TCM)
was developed by Neves and Pucciarelli (1990, 1991,
1998) and Neves et al. (1999a,b), and was supported
by other studies (Steele and Powell, 1992, 1993;
Powell and Neves, 1999). It was observed that traits
characterizing both typical East Asians and recent
Amerinds are absent from Paleoamericans. These
differences lie in the more generalized morphological pattern of the Paleoamericans, and indicate that
early people were not part of the typical “Mongoloids” we currently associate with Asia. In previous
papers, Neves et al. (1999a) and Powell and Neves
(1999) discussed this theory under the title FourMigration Model, a modification of the “Three-Migration Model” by Greenberg et al. (1986). In the
following, however, we will refer to this theory as
Two-Component Settlement Model (TCM), since this
label reflects better the validity of Amerindians as a
heterogeneous group, without reference to any cause
of heterogeneity (i.e., migration). In this context, we
must note that discussion about the grade of genetic,
biological, or linguistic closeness between Eskimos,
Aleuts, and Indians of the northwestern part of
North America is not of primary interest in this
model, since TCM is based mainly on relationships
between Paleoamerican remains and modern Amerindian variation. This may be more relevant to the
TMM model, though, since blood group analyses
showed that Eskimos and Indians of the Na-Dene
language phylum (Haida, Tlingit, and Athapaskan)
are more closely related to each other than to other
Fig. 3. Connection scheme for one model of human settlement
of Americas: the Two-Component Settlement Model (Neves and
Pucciarelli, 1990, 1991, 1998; Neves et al., 1999a,b; Powell and
Neves, 1999). Two modifications are included in this scheme: 1)
the distance between Paleoamerican and Fuego-Patagonia is increased (model TCM2 and TCM3, distance value in parentheses);
and 2) Eskimos are connected with Amerindians rather than with
Asians (model TCM3, dotted line). Numbers represent hypothesized distances.
Americans or Asians (Szathmáry, 1979). Furthermore, Ossenberg (1976) demonstrated the biological
closeness between Na-Dene speakers (specifically
Athapaskan speakers) and Aleuts (Eskimo-Aleut
family) after the analysis of a battery of 24 discrete
cranial traits (see also Szathmáry and Ossenberg,
1978; Szathmáry, 1993b; more recent molecular
data confirm the relative closeness of Eskimo and
Na-Dene speakers).
The connectivity scheme representing the TCM is
shown in Figure 3. Samples were divided into five
boxes by geographical location and according to the
divergences predicted by the TCM: Paleoamerican
and SJAP), Eskimo (ESKI), Amerindian (ARIK,
SANT, PERU, and ARAU), and Fuego-Patagonia
(TEHU, SELK, YAHG, and KAWE). In this scheme,
the Asian and Amerindian boxes are connected by a
relatively small distance value (1) to indicate the
predicted dispersion from a modern differentiated
stock. The Paleoamerican sample is connected to
Fuego-Patagonia by the same small distance value
(1) regarding their belonging to the hypothetical
early wave. In this context, Lahr (1995) concluded
that Fueguian-Patagonian groups also show this
mosaic of primitive, generalized pattern and stated
that this phenomenon could represent a retention of
traits from Paleoamerican populations. Next, the
Paleoamerican box is connected to the Amerindian
one by a distance of 2, representing the predicted
TABLE 3. Distance matrices considered in this study
Biological distance matrix obtained after an R-matrix
Geographical separation expressed in thousands of
Asymptotic distance matrix (spatial separation)
“Three-Migration Model”
“Single Wave of Migration Model”
“Two-Component Model”
TCM, but equal distance between Paleoamerican,
Fuego-Patagonia, and Asia
TCM2, but ESKI connected with Amerindian rather
than Asian groups
separation between a morphologically generalized
modern Homo sapiens (early wave) and a specialized
East Asian-Amerindian group. Finally, we derived
the Eskimos from the Asian stock, joining both with
the minimum distance (0.5) predicted in order to
represent a recent divergence.
Two additional design matrices departing from
this initial scheme were also constructed. Basically,
we changed the original scheme of TCM by deriving
the Eskimos from the Amerindian box rather than
the Asian one (TCM2), and finally we altered the
triangle formed by Paleoamerican, Fueguian-Patagonians, and Amerindians by increasing the distance value between Paleoamerican and Fuego-Patagonia from one to two (TCM3). Modifications are
shown in Figure 3, with dotted lines for the TCM3
model and in parentheses for TCM2. In Table 3, we
give a brief description of the eight matrices (one
biological, two geographic, and five design ones) involved in this study.
Correlation among distance matrices
Mantel statistic tests (Mantel, 1967) were used to
analyze correlations among different types of distance matrices. The Mantel statistic tests associations between distance matrices, and gives us a way
of testing the significance of these associations. Significance of the correlation was determined by a
permutation test: the rows and columns of one matrix are permuted and the Mantel statistic is calculated 9,999 times, creating a distribution that is
used to evaluate the significance of the observed
correlation (Mantel, 1967; Smouse et al., 1986; Waddle, 1994; Sokal and Rohlf, 1995).
Alternatively, the Smouse-Long-Sokal test (Smouse
et al., 1986) was used to yield partial matrix correlations. The Smouse-Long-Sokal method extends
Mantel’s statistic to three or more matrices and
tests whether an association between matrix A and
B is significant when one or more matrices C, D, . . .
are held constant. The Smouse-Long-Sokal test was
used to test partial correlations after removing the
effects of geography.
Figure 4 shows the plot of the first two principal
cooordinates obtained after the BIO matrix. The
first two eigenvalues collectively account for 54.8 %
of the variation. Clearly, LAGO appears to be distinct from the remaining samples. Except for BURI,
Asian groups tend to form a separate cluster. Conversely, Native Americans constitute a highly heterogeneous and polymorphic group. Cranial breadth
(XCB), frontal breadth (XFB), and auricular breadth
(AUB) highly contributed to the first coordinate,
whereas cranial length (GOL), basion-nasion length
(BNL), basion-bregma height (BBH), and frontal,
parietal, and occipital chords (FRC, PAC, and OCC)
negatively contributed to the first coordinate. This
pattern clearly separates between groups with short
and wide crania (e.g., BURI) from those with a more
long and narrow skull (e.g., LAGO). Variance along
the second axis was mainly generated by cranial
breadth (XCB) and frontal breadth (XFB) (positive
values), and bimaxillary breadth (ZMB), bifrontal
breadth (FMB), and biorbital breadth (EKB) (negative values).
Minimum Fst values were computed for different
subsets of data according to several comparison criteria. Results are shown in Table 4. Minimum Fsts
are obtained assuming that all heritabilities are
equal to 1, and that the additive genetic covariance
matrix is proportional to the phenotypic covariance
matrix (Williams-Blangero and Blangero, 1989).
This is a conservative statistic because it implies
that the minimum Fst should be less than an estimate of Fst computed from genetic markers (Relethford, 1994).
In order to compare our results with further studies, minimum Fsts were estimated considering the
following subsets: Asian, Native Americans, South
Americans, South Americans except Paleoamericans (LAGO), and North Americans. The minimum
Fst values range from 0.074 – 0.162. As expected, Fst
values differ much from either array of samples. The
smallest value is obtained when only Asiatic samples are considered (minimum Fst ⫽ 0.074). In contrast, South American groups give the highest Fst
value (0.176). Geographic distribution of the samples is an important factor that could be responsible
for this difference: the local samples in the Americas
are widely separated, and the Asian ones are close
together geographically. Because of isolation by distance, populations further apart geographically are
expected to be more dissimilar from each other and
from the regional centroid. Nevertheless, the results
do not always match that rule, since Native Americans in general and South Americans in particular
show higher Fst values (Fst ⫽ 0.161 and Fst ⫽
0.162, respectively) than the entire sample, which
covers the widest geographic area. Variability in
South Americans is clearly generated by Paleoamericans: Fst value drops from 0.162 to 0.116 when
this sample is removed.
Pairwise correlations between the biological distance matrix (BIO), spatial separation matrices
(GEO, ADM), and five design matrices (TMM, SWM,
TCM, TCM2, and TCM3) are given in Table 5. As
Fig. 4.
Principal coordinate plot obtained from minimum genetic distances (BIO) between samples.
TABLE 4. Minimum Fst values for craniometric traits
calculated for several arrays of the total set of populations
Samples included
All samples
Native Americans
South Americans
South Americans except Paleoamerican
North Americans
expected, the two arrays of spatial separation are
correlated (r ⫽ 0.440; P ⬍ 0.0001). The settlement
models as much as the biological distances matrix
were correlated with the separation in kilometers
between samples (GEO) and with the array representing long-distance isolation (AMD).
The Three-Migration Model and Single-Wave
Model failed to correlate with biological distances
expressed in BIO (Table 5). Conversely, biological
distances gave significant, positive correlations with
the three variants of the Two-Component Settlement Model. The correlation between BIO and
TCM2 was the strongest of these comparisons (r ⫽
0.386; P ⬍ 0.05).
Because both biological variation and the models
are closely related with geographic distances, results of correlations could be best interpreted as
partial matrix correlations holding geography constant (Sokal et al., 1992; Waddle et al., 1998). In
Table 6, we show the partial Mantel correlations
among the BIO and five design matrices (TMM,
SWM, TCM, TCM2, and TCM3) with spatial separation held constant. Neither the Three-Migration
Model nor the Single-Wave Model correlated well
with the biological distance matrix (rBIO.TMM ⫽
0.033, P ⬎ 0.05 and rBIO.SWM ⫽ 0,013, P ⬎ 0.05,
respectively). Conversely, the TCM models give positive, significant correlations when tested against
the BIO matrix. The strongest correlation was observed between BIO and design TCM2 (r ⫽ 0.364;
P ⬍ 0.02). This design describes equal distances
between Fuego-Patagonians, Paleoamericans, and
Amerindians. However, note that these results cannot clearly differentiate the effects of the three variant forms of TCM (TCM, TCM2, TCM3).
In order to evaluate the explanatory power of
TCM by alternative methods, we performed two additional tests. First, we applied the method developed by Dow and Cheverud (1985). This test assumes the null hypothesis of no difference between
the correlations of an observed distance matrix C
with two explanatory matrices A and B. Thus, the
Dow-Cheverud test determines whether two correlations r(AC) and r(BC) differ significantly from
each other. Table 7 shows the results of the DowCheverud test. Matrix correlations for TCM vs.
TMM and SWM are given as the difference between
design matrices (after standardization). Negative
correlation indicates a better fit between BIO and
TABLE 5. Complete matrix correlations among two spatial separation matrices (GEO and ADM), five design matrices
(TMM, SWM, TCM, TCM2, and TCM3), and one biological distance matrix (BIO)1
0.440 (0.000)
0.720 (0.000)
0.699 (0.000)
0.651 (0.002)
0.629 (0.002)
0.640 (0.001)
0.233 (0.009)
0.299 (0.008)
0.294 (0.016)
0.177 (0.043)
0.187 (0.037)
0.185 (0.043)
0.176 (0.006)
0.749 (0.052)
0.476 (0.009)
0.443 (0.008)
0.378 (0.010)
0.084 (0.275)
0.270 (0.022)
0.240 (0.024)
0.318 (0.019)
0.064 (0.286)
0.988 (0.000)
0.961 (0.001)
0.354 (0.009)
0.975 (0.001)
0.386 (0.010)
0.384 (0.016)
Exact probabilities are in parentheses, under each correlation value.
TABLE 6. Partial Mantel correlations among one measure of
biological distance (BIO) and five design matrices
(TMM, SWM, TCM, TCM2, and TCM3), with spatial
separation held constant1
TABLE 8. Matrix correlation test between BIO and TCM and
partial correlations involving these distances and geographic,
TMM, and SWM distances1
0.354 (0.009)
0.333 (0.012)
0.356 (0.043)
0.358 (0.036)
0.033 (0.373)
0.013 (0.390)
0.333 (0.012)
0.364 (0.016)
0.363 (0.023)
Column r gives matrix correlations. Exact probabilities are in
parentheses, under each correlation value.
TABLE 7. Matrix correlation tests for models of New World’s
peopling using Dow-Cheverud approach
(Dow and Cheverud, 1985)1
⫺0.264 (0.841)
⫺0.240 (0.864)
⫺0.286 (0.831)
⫺0.261 (0.860)
⫺0.268 (0.817)
⫺0.275 (0.849)
Exact probabilities are in parentheses, under each correlation
TCM, whereas a positive correlation indicates a better fit of the TWM or SWM models. Although this
test suggests that the hypothesis depicted in TCM
better fits the data (negative values imply a stronger
effect of TCM on BIO when compared against TMM
and SWM), P values were not significant. However,
since Oden and Sokal (1992) demonstrated that this
test is highly vulnerable to spatially autocorrelated
data, those results must be taken with caution. Another important issue regarding the application of
this method is that the Dow-Cheverud approach
detects only additive effects of a data matrix on a
pair of predictor matrices. The generalized regression approach advocated in the Smouse-Long-Sokal
test can easily incorporate both types of dependency,
additive and nonadditive (Smouse and Long, 1992),
and can be extended to multiple matrices.
Since we need to control the effects of more than
one matrix (e.g., GEO and SWM) to further corroborate the significance of the BIO-TCM correlation,
we applied an alternative method described in Sokal
et al. (1992). This approach consists of the computation of partial correlations of BIO on TCM holding
Partial correlations, which are all of BIO against TCM with
various other distances held constant, are indicated by a period
followed by the constant variables. Thus, .GEO, TMM stands for
rBIO,TCM.GEO,TMM. Exact probabilities are in parentheses, under
each correlation value.
the remaining models constant until all matrices are
considered. Thus, we tested whether any correlation
remained between BIO and TCM, once the correlation between these two variables due to one or more
regressor variables (spatial separation and competing models) was eliminated. In Table 8, we present
the results of the Smouse-Long-Sokal test performed
in this way. Note that partial correlation of BIO on
TCM with added distance matrices held constant
does not decrease further and continues to be significant. In general terms, results indicate that spatial
separation and a model considering two different
patterns of morphological variability in the Americas are the most dominant influence on the degree
and pattern of craniometric differentiation.
Phenotypic data in the form of craniometrics were
analyzed in order to describe the genetic characteristics of the Asian and American populations considered
here. Inferences performed using this approach are
based on the premise that phenotypic variation adequately reflects genetic variation (Williams-Blangero
and Blangero, 1989). We used the Relethford-Blangero
method, because it provides an analytical framework
that explicitly delimits the genetic inferences that can
be obtained from purely phenetic data.
O’Rourke et al. (1992) give Fst values in a summary of statistics for Amerindian genetic data (see
Table 1 in O’Rourke et al., 1992). Fst values for
North America and South America were 0.090 and
0.091, respectively. Results presented in this report
tend to show a greater level of differentiation between populations (0.111 and 0.162, respectively).
Several explanations could account for those differences. First, the populations sampled in both studies
are not the same (e.g., Fuego-Patagonian groups
were not considered by O’Rourke et al. (1992), and
our study included no crania from Na-Dene-speaking populations). This raises once again the question
of sampling in these kinds of analyses. As stated by
Schanfield (1992), a broader approach requires that
the origin of all Amerindians be considered not the
limited picture that is formed by looking at the “usual” groups. Moreover, Lahr (1995) demonstrated
how much the variation between modern populations is affected by varying the populations that are
sampled. In this context, the lack of clustering between Fuego-Patagonian populations and the remaining Amerindians was also demonstrated in
other studies (Lahr, 1995; Lalueza et al., 1997), and
further research on those groups is necessary.
Secondly, natural selection and developmental acclimatization could affect cranial variation. This issue was discussed by several authors. Recent analyses gave little basis to consider selection as a cause
of such differences (see Relethford, 1994; Relethford
and Harpending, 1994; Neves et al., 1999b).
Finally, one may suspect that genetic drift generated high levels of variability within the Native
American subset. For instance, O’Rourke et al.
(1992) suggested that the levels of heterogeneity
observed in modern populations of South Amerindians could have arisen in part because of the great
density of population in Central America. This increase in population, according to Ward (O’Rourke
et al.,1992), could have blocked further migrants
who sought to enter South America via the north.
Relethford (1996) demonstrated that when population sizes are different, as in the present case, differential genetic drift can obscure the underlying
pattern of population history. The author developed
a method in which the R-matrix is modified (scaled)
after a migration matrix, and estimations population sizes control differential drift to some extent. A
preliminary application of this method to the data
set considered here showed that the general pattern
of between-group dissimilarities remained unaltered. Nevertheless, estimations of relative population sizes based on scaled R-matrices are just broad
approaches. Scaling methods can be improved by
better estimated demographic parameters.
Results presented here agree with early findings
made by several researchers that levels of variability observed in South America cannot be accommodated in a single group (Neves and Pucciarelli, 1990,
1991, 1998; Neves et al., 1999a,b). Moreover, high
variability is also found in several earlier studies
with different genetic markers. Classic protein
markers, as well as mitochondrial and nuclear DNA
polymorphisms, revealed that Amerindian populations present a higher within-group genetic heterogeneity than any other major human ethnic group
(Deka et al., 1995; Urbanek et al., 1996; Zago et al.,
1996; Bortolini et al., 1997; Novick et al., 1998; Da
Silva et al., 1999). As stated by Stone and Stoneking
(1998), it does not appear that European contact, on
the whole, significantly altered patterns of Amerind
mtDNA variation, despite the accompanying sudden
and drastic decrease in population size, although
some reduction in diversity may have occurred in
many populations (Wallace and Torroni, 1992).
To further explore underlying and/or hypothetical
reasons for the arrangement of the samples, we performed matrix permutation tests analyzing several
patterns of spatial separation and models for the
Americas’ settlement. Cole (1996) and Konigsberg
(1997) have extensively criticized the use of matrix
permutation test in paleoanthropology, and in particular, the analysis of modern human origins by
Waddle (1994). We intended to take into account the
points reviewed by them about matrix permutation
methods (e.g., sample sizes are stable and customary in such studies, Mahalanobis generalized distance was used, significance values were obtained by
means of randomization tests, and models fulfilled
the triangle inequality).
Tables 5– 8 show connections between hypothesized matrices and a particular way to estimate genetic distances between samples of skulls. Thus, the
“explanatory power” of the different models can be
evaluated. Despite the possibility of misunderstanding the true magnitude of correlations because of the
spatial separation effect, results show that TMM
and SWM are untenable, and TCM emerges as more
strongly associated with morphological variation.
This association cannot be reduced after geography
has been held constant, since the TCM model improved performance in comparison with the remaining design matrices. In fact, the association between
biological distances and TCM remains positive and
significant, even when the effects of spatial separation are removed. Keeping other models constant
does not reduce this correlation (Table 8).
The present study was unable to provide a clear
separation between the three variants based on the
Two-Components Settlement Model. This model implies that an early component corresponds to a generalized “non-Mongoloid” group, represented by Paleoamerican populations dated from 12,000 – 8,000
years BP. The remaining component corresponds to
differentiated groups, which arose in South America
around 9,000 – 8,000 years BP, giving rise to most of
the historic Amerindian groups. A central issue is
that this theory does not necessarily imply that
early inhabitants made any contributions to the genetic pool of later Native American populations.
Matrix permutation analyses were unable to discriminate clearly between models in which the relative position of Fuego-Patagonia was arbitrarily
altered (see Tables 5 and 6). Lahr (1995) suggested
that both Fuego-Patagonian and Paleoamerican
populations might reflect the morphology of a more
generalized ancestral group. In agreement with this
idea, Lalueza et al. (1997) suggested that the distinctive distributions in the continent of mtDNA
haplogroups A and B, which appear to have been
absent from Fueguian-Patagonians, traced back to a
population of more ancient ancestry, distinct from
those Amerindian populations harboring all four
primary lineages. According to Lalueza et al. (1997),
such a process is possible despite the potential influence of genetic drift, reasonably presumed to have
been rather small, under the demographic conditions depicted by missionaries’ census surveys and
conservative estimates. However, this assertion was
challenged by Moraga et al. (2000), and it is evident
that more research is needed in this field.
The results obtained here show that the Paleoamerican sample departs considerably from the
morphological pattern of the remaining Native
Americans, and also places Fueguians and Patagonians at one extreme of Amerindian morphological
variability (Fig. 4).
The present study was conclusive about two points.
First, models involving clustering of Native Americans
in a single migration, or deriving from a single ancestor, have little support from craniometric variability.
Second, from the three hypotheses examined, the one
that represented departures from a typical East Asian
morphology in Paleoamerican and Fuego-Patagonian
series gives the best fit to the data tested.
However, it remains unresolved whether the two
settlement components arose as a result of two separate migrations, by local evolution from Paleoamericans (whose remains are dated between 12,000 – 8,000
years ago) to typical Amerindians (present in America
since 9,000 – 8,000 years BP), or by shared ancestry
between Paleoamericans and the Amerindians.
Our thanks go to Dr. Erik Ozolins for allowing us
to use his Lagoa Santa data, and to W.W. Howells
for making his data available via the Internet. We
also thank Dr. John Relethford for providing the
Rmet statistical package on the Internet. Two anonymous reviewers and Dr. Emöke Szathmáry provided valuable advice, suggestions, and criticism
which improved the quality of the paper.
Acosta J. 1962. Historia natural y moral de las Indias [1589].
Bailliet G, Rothhammer F, Carnese FR, Bravi CM, Bianchi NO.
1994. Founder mitochondrial haplotypes in Amerindian populations. Am J Hum Genet 54:27–33.
Bianchi NO, Bailliet G, Bravi CM, Carnese FR, Rothammer F,
Martı́nez-Marignac VL, Pena SDJ. 1997. Origin of Amerindian
Y-chromosomes as inferred by the analysis of six polymorphic
markers. Am J Phys Anthropol 102:79 – 89.
Bianchi NO, Catanesi CI, Bailliet G, Martı́nez-Marignac VL,
Vidal Roja LB, Bravi CM, Herrera RJ, López Camelo JS. 1998.
Characterization of ancestral and derived Y-chromosome haplotypes of New World native populations. Am J Hum Genet
Bonatto SL, Salzano FM. 1997a. A single and early migration for
the peopling of the Americas supported by mitochondrial DNA
sequence data. Proc Natl Acad Sci uSA 94:1866 –1871.
Bonatto SL, Salzano FM. 1997b. Diversity and age of the four
major mtDNA haplogroups, and their implications for the peopling of the New World. Am J Hum Genet 61:1413–1423.
Borrero LA, Mc Ewan C. 1997. The peopling of Patagonia. In:
McEwan C, Borrero LA, Prieto A, editors. Patagonia. Natural
history, prehistory and ethnography at the uttermost end of the
earth. London: British Museum Press. p 32– 45.
Bortolini MC, Salzano FM, Zago MA, Silva WA Jr, Weimer TA.
1997. Genetic variability in two Brazilian ethnic groups: a
comparison of mitochondrial and protein data. Am J Phys Anthropol 103:147–156.
Cole TM. 1996. The use of matrix permutation tests for evaluating competing hypotheses of modern human origins. J Hum
Evol 31:477– 484.
Corruccini RS. 1973. Size and shape in similarity coefficients
based on metric characters. Am J Phys Anthropol 38:743–754.
Crawford MA. 1998. The origins of Native Americans. Evidence
from anthropological genetics. New York: Cambridge University Press.
Da Silva WA, Bortolini MC, Meyer D, Salzano FM, Elion J,
Krishnamoorty R, Schneider MPC, Castro de Guerra D, Layrisse Z, Mendez Castellan H, Azevedo Weimer T, Zago MA.
1999. Genetic diversity of two african and sixteen South American populations determined on the basis of six hypervariable
loci. Am J Phys Anthropol 109:425– 437.
Deka R, Shriver MD, Yu LM, Ferrel RE, Chakraborty R. 1995. Intra
and inter-populational diversity at short tandem repeat loci in
diverse populations of the world. Electrophoresis 16:1659 –1664.
Dillehay TD, Collins MB. 1988. Early cultural evicence from
Monte Verde in Chile. Nature 332:150 –152.
Dow MM, Cheverud JM. 1985. Comparison of genetic distance
matrices in studies of population structure and genetic microdifferentiation. Quadratic assignement. Am J Phys Anthropol
González-José R, Dahinten S, Hernández MH. 2001. The settlement of Patagonia: a matrix correlation study. Hum Biol. 73:
Greenberg JH, Turner CG, Zegura SL. 1986. The settlement of
the Americas: a comparison of the linguistic, dental and genetic
evidence. Curr Anthropol 27:477– 495.
Guidon N, Delibrias G. 1986. Carbon-14 dates point to man in the
Americas 32,000 years ago. Nature 321:769 –771.
Harpending HC, Jenkins T. 1973. Genetic distance among Southern African populations. In: Crawford MH, Workman PL, editors. Methods and theories of anthropological genetics. Albuquerque: University of New Mexico Press. p 177–199.
Harris EF. 1980. Sex differences in lingual marginal ridging on
the human maxillary central incisor. Am J Phys Anthropol
Hernández M, Lalueza C, Garcı́a-Moro C. 1997. Fueguian cranial
morphology: the adaptation to a cold, harsh environment. Am J
Phys Anthropol 103:103–117.
Howells WW. 1973. Cranial variation in man: a study by multivariate analysis of patterns of difference among recent human
populations. Papers of the Peabody Museum of Archaeology
and Ethnology, volume 67. Cambridge: Harvard University.
Howells WW. 1989. Skull shapes and the map. Craniometric
analyses in the dispersion of modern Homo. Papers of the
Peabody Museum of Archaeology and Ethnology, volume 79.
Cambridge: Harvard University.
Hrdlička A. 1925. The origin and antiquity of the American Indian. Annu Rep Smithson Inst Washington 1923:481– 494.
Konigsberg LW. 1997. Comments on matrix permutation tests in
the evaluation of competing models for modern human origins.
J Hum Evol 32: 479 – 488.
Konigsberg LW, Kramer A, Donelly SM, Relethford JH, Blangero
J. 1994. Modern human origins. Nature 372:229.
Lahr MM. 1995. Patterns of modern human diversification: implications for Amerindian origins. Am J Phys Anthropol 38:163–198.
Lalueza C, Pérez-Pérez A, Prats E, Cornudella L, Turbón D. 1997.
Lack of founding Amerindian mitochondrial DNA lineages in
extinct Aborigines from Tierra del Fuego-Patagonia. Hum Mol
Genet 6:41– 46.
Mantel N. 1967. The detection of disease clustering and a generalized regression approach. Cancer Res 27:209 –220.
McCulloch RD, Clapperton CM, Rabassa J, Currant A. 1997. The
natural setting. The glacial and post glacial environmental
history of Fuego-Patagonia. In: McEwan c, Borrero LA, Prieto
A, editors. Patagonia. Natural history, prehistory and ethnography at the uttermost end of the earth. London: British Museum Press. p 12–31.
Meltzer DJ. 1997. Monte Verde and the Pleistocene peopling of
the Americas. Science 276:754 –755.
Merriwether DA, Rothammer F, Ferrel RE. 1995. Distribution of
the four founding lineage haplotypes in native americans suggest a single wave of migration for the New World. Am J Phys
Anthropol 98:411– 443.
Moraga ML, Rocco P, Miquel JF, Nervi F, Llop E, Chakraborty R,
Rothhammer F, Carvallo P. 2000. Mitochondrial DNA polymorphisms in Chilean aboriginal populations: implications for the
peopling of the Southern cone of the continent. Am J Phys
Anthropol 113:19 –29.
Morell V. 1990. Confusion in earliest America. Science 248:439 –
Neves WA, Pucciarelli HM. 1990. The origin of the first Americans: an analysis based on the cranial morphology of early
South American human remains. Am J Phys Anthropol 81:274.
Neves WA, Pucciarelli HM. 1991. Morphological affinities of the
first Americans: an exploratory analysis based on early South
American human remains. J Hum Evol 21:261–273.
Neves WA, Pucciarelli HM. 1998. The Zhoukoudien Upper Cave
skull 101 as seen from the Americas. J Hum Evol 34:219 –222.
Neves WA, Powell JF, Ozolins EG. 1999a. Extra-continental morphological affinities of Lapa Vermelha IV, Hominid 1: a multivariate analysis with progressive numbers of variables. Homo
Neves WA, Munford D, Zanini MC, Pucciarelli HM. 1999b. Cranial morphological variation in South America and the colonization of the New World: towards a four migration model?
Ciencia Cultura J Braz Assoc Adv Sci 51:151–165.
Novick GE, Novick CC, Yunis J, Yunis E, Antunez de Mayolo P,
Scheer DW, Deininger PL, Stoneking M, York DS, Batzer MA,
Herrera RJ. 1998. Polymorphic Alu insertions and the Asian
origin of Native American populations. Hum Biol 70:23–29.
Oden NL, Sokal RR. 1992. An investigation of three matrix permutation tests. J Classif 9:275–290.
O’Rourke DH, Mobarry A, Suarez BK. 1992. Patterns of genetic
variation in Native America. Hum Biol 64:417– 434.
Ossenberg NS. 1976. Within and between race distances in population studies based on discrete traits of the human skull.
Am J Phys Anthropol 45:701–716.
Powell JF, Neves WA. 1999. Craniofacial morphology of the first
Americans: pattern and process in the peopling of the New
World. Yrbk Phys Anthropol 42:153–158.
Relethford JH. 1994. Craniometric variation among modern human populations. Am J Phys Anthropol 95:53– 62.
Relethford JH. 1996. Genetic drift can obscure population history: problem and solution. Hum Biol 68:29 – 44.
Relethford JH. 1998. Rmet for Windows. Version 3.0. Oneonta,
NY: SUNY College at Oneonta.
Relethford JH, Blangero J. 1990. Detection of differential gene
flow from patterns of quantitative variation. Hum Biol 62:5–25.
Relethford JH, Harpending HC. 1994. Craniometric variation,
genetic theory and modern human origins. Am J Phys Anthropol 95:249 –257.
Relethford JH, Crawford MH, Blangero JH. 1997. Genetic drift and
gene-flow in post-famine Ireland. Hum Biol 69:443– 465.(1984)
Rogers AR, Harpending HC. 1986. Migration and genetic drift in
human populations. Evolution 40:1312–1327.
Roosevelt AC, Lima da Costa M, López Machado C, Michab M,
Mercier N, Valladas H, Feathers J, Barnett W, Imazzio da Silveira M, Henderson A, Sliva J, Chernoff B, Reese DS, Holman JA,
Toth N, Schick K. 1996. Paleoindian cave dwellers in the Amazon:
the peopling of the Americas. Science 272:373–384.
Salzano FM, Callegari-Jacques SM. 1988. South American Indians: a case study in evolution. Oxford: Clarendon Press.
Schanfield MS. 1992. Immunoglobulin allotypes (GM and KM)
indicate multiple founding populations of Native Americans:
evidence of at least four migrations to the New World. Hum
Biol 64:381– 402.
Schurr TG, Ballinger SW, Gan YY, Hodge JA, Merriwether DA,
Lawrence DN, Knowler WC, Weiss KM, Wallace DC. 1990.
Amerindian mitochondrial DNAs have a rare Asian mutations
at high frequencies, suggesting they derived from four primary
maternal lineages. Am J Hum Genet 46:613– 623.
Smouse PE, Long JC. 1992. Matrix correlation analysis in anthropology and genetics. Yrbk Phys Anhtropol 35:187–213.
Smouse PE, Long JC, Sokal RR. 1986. Multiple regression and
correlation extensions of the Mantel test matrix correspondence. Syst Zool 35:627– 632.
Sokal RR. 1979. Testing statistical significance of geographic
variation patterns. Syst Zool 28:227–233.
Sokal RR, Oden NL. 1978. Spatial autocorrelation in biology. 2:
some biological implications and four applications of evolutionary and ecological interest. Biol J Linn Soc 10:229 –234.
Sokal RR, Rohlf FJ. 1995. Biometry. 3rd ed. New York: W.H.
Sokal RR, Oden NL, Thomson BA. 1992. Origins of the Indo-Europeans: genetic evidence. Proc Natl Acad Sci USA 89:7669 –7673.
Sokal RR, Oden NL, Walker J, Waddle DM. 1997. Using distance
matrices to choose between competing theories and an application to the origin of modern humans. J Hum Evol 32:501–522.
Steele DG, Powell JF. 1992. Peopling of the Americas: paleobiological evidence. Hum Biol 64:303–336.
Steele DG, Powell JF. 1993. Paleobiology of the first Americans.
Evol Anthropol 2:138 –146.
Stone AC, Stoneking M. 1998. mtDNA analysis of a prehistoric
Oneonta population: implications for the peopling of the New
World. Am J Hum Genet 62:1153–1170.
Szathmáry EJE. 1979. Blood groups of Siberians, Eskimos, subarctic and Northwest Coast Indians: the problem of origin and
genetic relationships. In: Laughlin WS, Harper AB, editors.
The first Americans: origins, affinities and adaptations. New
York: Gustav Fischer. p 185–209.
Szathmáry EJE. 1981. Genetic markers in Siberian and northern
North American populations. Yrbk Phys Anthropol 24:37–73.
Szathmáry EJE. 1984. Peopling of northern North American:
clues from genetic studies. Acta Anthropogenet 8:79 –109.
Szathmáry EJE. 1993a. MtDNA and the peopling of the Americas. Am J Hum Genet 53:793–799.
Szathmáry EJE. 1993b. Genetics of aboriginal North Americans.
Evol Anthropol 1:202–220.
Szathmáry EJE, Ossenberg NS. 1978. Are the biological differences between North American Indians and Eskimos truly
profound? Curr Anthropol 19:673–701.
Torroni A, Schurr TG, Cabell MF, Brown MD, Neel JV, Larsen M,
Smith DG, Vullo CM, Wallace DC. 1993. Asian affinities and
continental radiation of the four founding Native American
mtDNAs. Am J Hum Genet 53:563–590.
Turner CG. 1987. Late Pleistocene and Holocene population history of East Asia based on dental variation. Am J Phys Anthropol 73:305–321.
Turner CG. 1992. Microevolution of East Asian and European
populations: a dental perspective. In: Akazawa T, Aoki K,
Kimura T, editors. The evolution and dispersal of modern humans in Asia. Tokyo: Hokusen-sha. p 414 – 438.
Urbanek M, Goldman D, Long JC. 1996. The apportionment of
dinucleotide repeat diversity in Native Americans and Europeans. a new approach to measuring gene identity reveals asymmetric patterns of divergence. Mol Biol Evol 13:943–953.
Waddle DM. 1994. Matrix correlation tests support a single origin
for modern humans. Nature 368:452– 454.
Waddle DM, Sokal RR, Rudan P. 1998. Factors affecting population variation in eastern Adriatic isolates (Croatia). Hum Biol
70:845– 864.
Wallace DC, Torroni A. 1992. American Indian prehistory as written
in the mitochondrial DNA: a review. Hum Biol 64:403– 416.
Williams-Blangero S, Blangero J. 1989. Anthropometric variation
and the genetic structure of the Jirels of Nepal. Hum Biol 61:1–12.
Wright S. 1943. Isolation by distance. Genetics 28:114 –138.
Zago MA, Silva WA Jr, Tavella MH, Santos SEB, Guerreiro JF,
Figueiredo MS. 1996. Interpopulational and intrapopulational
genetic diversity of Amerindians as revealed by six variable
number of tandem repeats. Hum Hered 46:274 –289.
Без категории
Размер файла
136 Кб
correlation, testing, craniometric, matrix, variation, mean, hypothesis, settlements, american, analyse
Пожаловаться на содержимое документа