close

Вход

Забыли?

вход по аккаунту

?

Environment and morphology in Australian Aborigines A re-analysis of the Birdsell database.

код для вставкиСкачать
AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 134:75–91 (2007)
Environment and Morphology in Australian Aborigines:
A Re-analysis of the Birdsell Database
Ian Gilligan* and David Bulbeck
School of Archaeology and Anthropology, Australian National University, Canberra ACT 0200, Australia
KEY WORDS
climatic adaptation; Bergmann’s rule; human variation
ABSTRACT
Pursuant to his major research interest
in the cultural ecology of hunter-gatherers, Birdsell collected an unparalleled body of phenotypic data on Aboriginal Australians during the mid twentieth century.
Birdsell did not explicitly relate the geographic patterning in his data to Australia’s climatic variation, instead
arguing that the observable differences between groups
reflect multiple origins of Australian Aborigines. In this
article, bivariate correlation and multivariate analyses
demonstrate statistically significant associations between
climatic variables and the body build of Australians that
are consistent with the theoretical expectations of
Bergmann’s and Allen’s rules. While Australian Aborigines in comparison to Eurasian and New World populations can be generally described as long-headed, linear
in build, and characterized by elongated distal limbs, the
variation in this morphological pattern across the continent evidently reflects biological adaptation to local
Holocene climates. These results add to a growing body
of evidence for the role of environmental selection in the
development of modern human variation. Am J Phys
Anthropol 134:75–91, 2007. V 2007 Wiley-Liss, Inc.
The aim of this article is to utilize to best effect the
wealth of data on hunter-gatherer biological adaptation
to climatic variation in Australia. The latter is precisely
documented at numerous weather stations across the
continent, which record a standard set of meteorological
variables. Hunter-gatherer biological adaptation can be
investigated thanks to the monumental work by Joseph
Birdsell (1993) in recording 114 morphological variables
on 2,131 Australian Aboriginal subjects during the midtwentieth century. Notwithstanding the unique value of
Birdsell’s data set for an investigation like ours, its use
does entail some methodological issues, which our study
manages through a logical sequence of statistical analyses. The issues include the uneven distribution of Aboriginal settlements which Birdsell was able to survey,
the complexity of factors—different degrees of seasonality as well as temperature and rainfall differences—
which determines the critical aspects of local Australian
climates, and the possibility that apparent correlations
with climate may actually be the result of shared inheritance (‘‘Galton’s problem’’). Indeed our first analysis
will show that the frequency of the A1 blood group in
Australian populations is a very useful measure of relatedness, and its apparent climatic correlations are an
instantiation of Galton’s problem. We will then explore
the critical climate factors affecting morphological adaptation through a combination of significance testing on
correlation coefficient results and canonical correlation
analysis (CCA). The uneven geographical distribution of
Birdsell’s records is handled by subsidiary analyses,
which a) consider only populations of the temperate zone
and b) examine the climatic correlations by weather
station rather than Aboriginal population. Additional
complications, such as the extent of change from a traditional subsistence strategy, will be reviewed in our
Discussion. In particular, we stress that while we have
utilized Birdsell’s data for purposes quite different from
his, in our view this research continues and extends
Birdsell’s work on Australian hunter-gatherers in relation to their ecological setting.
C 2007
V
WILEY-LISS, INC.
C
BACKGROUND
Morphological adaptation
Studies in physical anthropology have demonstrated
consistent associations between environmental parameters and major trends in human morphology. Body
mass (or weight), body shape, head shape, and relative
limb proportions for instance correlate with thermal
conditions, particularly mean annual temperature. These
trends exist on a global and on a continental or regional
scale (e.g., Roberts, 1978; Houghton, 1990), and also
among fossil hominins (Trinkaus, 1981; Ruff, 1994;
Holliday and Falsetti, 1995).
The associations between thermal and morphological
variation are widely viewed as illustrating the operation
of certain ecogeographical ‘‘rules.’’ One is Bergmann’s
rule, first stated when Carl Bergmann (1847) noted the
importance of the surface area to volume ratio in affecting heat balance. A related rule is that of Allen (1877),
referring to the size of body appendages, with exposed
limbs becoming shorter in cooler climates. These trends
may arise through direct effects of environment on phenotypic development (Allen, 1877) and also niche choice,
whereby organisms tend to move into climatic zones
within which they are most thermally comfortable.
*Correspondence to: I.J. Gilligan, School of Archaeology and
Anthropology, Australian National University, Canberra ACT 0200,
Australia. E-mail: ian.g@bigpond.net.au
Received 18 July 2006; accepted 26 March 2007
DOI 10.1002/ajpa.20640
Published online 13 June 2007 in Wiley InterScience
(www.interscience.wiley.com).
76
I. GILLIGAN AND D. BULBECK
The physiological principles are summarized in Ruff
(2002). Briefly, body form affects thermal requirements
by virtue of the ratio between skin surface area and
body volume, or mass. The latter relates to heat production, whereas skin surface area plays a role in dissipating heat. Even minor variations in body shape or size
can be significant because volume and surface area vary
in unequal proportions to each other: volume varies as
the cube, whereas surface area varies as the square, of
any linear change in size. These principles predict a
lighter, more linear body build in warmer environments,
and a heavier and stockier build in cooler regions.
Like all rules, that of Bergmann has exceptions, as
emphasized by critics (e.g., Geist, 1987). With increasing
latitude, for instance, body mass first increases but then
decreases in some species. This reflects the fact that
greater body mass attracts a higher caloric cost, so its
net adaptive benefits reach limits, at which point alternative strategies such as increased fur cover can be
favored. Phenotype is ‘‘the result of a compromise between many conflicting selection pressures’’ (Mayr, 1956,
p 106), and exceptions serve only to illustrate the complexities involved (e.g., Schreider, 1975; Katzmarzyk and
Leonard, 1998). With this caveat, Bergmann’s rule has
been shown to apply among numerous bird and mammal
species (Ashton et al., 2000; Meiri and Dayan, 2003).
Based on Bergmann’s and Allen’s rules, we hypothesize that among Australian Aborigines, morphometric
variables relating to body size and shape should correlate with local climatic—primarily thermal—indices, independent of other factors. For example, we expect limb
segment lengths to correlate positively with temperature, more so for the lower limb and for distal segments
(radial and tibial lengths). Theoretically, body mass
(weight) should correlate negatively with temperature,
while variables denoting a more stocky body shape—
such as relative shoulder breadth—should likewise correlate negatively with temperature indices. Head size
and shape should also vary consistently, and we expect
the cranium to be larger and rounder in cooler regions
(e.g., the cranial module, reflecting overall size, and the
cephalic index, reflecting roundness, should both correlate negatively with temperature). Equally, we predict
that other features of phenotypic variation not considered subject to environmental selection effects—especially
those known to be under strong genetic control, including many dental and particularly serological variables—
will fail to show consistent climatic associations. To test
these hypotheses, we match each of Birdsell’s 217 variables to meteorological data from local weather stations
and compare correlation results for body form and for
those variables not expected to show environmental
patterning (Appendix A).
Australian Aborigines
In comparative studies of modern human groups, the
Australian Aborigines emerge with a distinctly tropical
pattern, i.e., a linear body build with relatively long,
slender limbs (Roberts, 1978). Nonetheless there are
reasons for anticipating thermal trends in Aboriginal
morphology that should include a less linear build, less
dolichocephalic head shape, and reduced limb proportions in the cooler southern latitudes. One reason is the
time depth of their occupation of the continent—at least
45,000 and perhaps 60,000 years (O’Connell and Allen,
2004). The long duration of the Aboriginal presence on
the continent implies an adequate time depth for any
morphological responses to become apparent across a
latitudinal range of 338, which today spans a 158C range
of mean annual temperatures. Exposure to cooler conditions during the late Pleistocene—more so in southern
parts of the continent—would have amplified regional
variation in selection for thermal adaptations. However,
we are concerned here only with the Holocene, a 10,000year period of comparatively stable conditions, to which
we expect the present population would have become
adapted.
The early literature on Australian morphological variability was polarized between Abbie (e.g., 1975, 1976)
who emphasized homogeneity, and Birdsell (Tindale and
Birdsell, 1941; Birdsell, 1967) who argued that a trihybrid constitution of the Aboriginal population was discernible from the extant phenotypic variation. Currently
the debate has moved onto whether there had been one
or two major founding populations during the Pleistocene, along with general agreement on an effective single population in continental Australia throughout the
Holocene (e.g., Thorne, 1971; Pietrusewsky, 1984; Habgood, 1986; Brown, 1987; Pardoe, 1994). Birdsell’s continued insistence until 1993 on the Holocene immigration of a third wave, which he has labeled Carpentarian,
appears anomalous in the current clime. We suspect that
Birdsell’s apparent reluctance to investigate environmental patterning in his Australian data reflects the fact
that many of the features he has cited as distinctive of
the Carpentarians, whose distribution he has placed
across the desert and most of the tropics, are precisely
those that would be expected of populations adapted to
hot environments.
From Abbie’s perspective, a climatic explanation of
anthropometric differences between Aboriginal populations would appear logical but, despite gathering much
data, he did not subject these to environmental analysis
(e.g., Abbie, 1975). A reanalysis of his data found significant regional variation in head and body measurements
and indices (Macho and Freedman, 1987). Results
showed that climatic factors made a typically small but
statistically significant contribution to the variation. The
authors urged caution in interpreting their findings
given the limited sampling, although they suggested
that the most important climatic variable was the temperature range of the coldest month.
Phylogenetic Issues
A potential confounding issue is that regional geographical patterning in phenotypic variation may be a
consequence not of adaptive effects but rather shared
group ancestry and environments, resulting in pseudoenvironmental correlations with morphology. This illustrates Galton’s problem, a long-standing and potentially
insoluble difficulty in population studies, whereby it
cannot be assumed that any observed relationship (for
instance between morphology and environment) is independent of historical associations between groups
(Strauss and Orans, 1975). Various methods can be used
to estimate the likely contribution of shared ancestry,
including linguistic analyses and genetic markers. In
Australia there is a major split between non-PamaNyungan (in the northwest) and the more widespread
Pama-Nyungan linguistic groups, but lack of reliable
precolonial data in some areas and uncertainties as to
the origins and time-depth of this pattern (e.g., Clendon,
American Journal of Physical Anthropology—DOI 10.1002/ajpa
ENVIRONMENT AND MORPHOLOGY IN ABORIGINES
2006) reduce its utility for our purposes. Extensive serological data in the Birdsell database, on the other hand,
provide suitable genetic markers to assess and test for
the confounding effect of phylogeny. Indeed, sampling for
the ABO system is the most extensive of all variables in
the database, covering virtually the entire continent.
Since A2 is absent and the distribution of gene O the
inverse of A, and gene B a relatively recent intrusion
limited to parts of the northern coastline, gene A1 provides the best (and simplest) such marker.
The frequency of the A1 allele among Australian
Aborigines shows great geographic variation with the
highest expression in the southern arid zone and gradually declining frequencies north, east, south and west
(Birdsell, 1993; see also Jurmain and Nelson, 1994). The
considerable variation in this genetically inherited trait
suggests that it may be a sensitive indicator of genetic
relatedness between populations. If so, then if body
shape and size patterning reflects common ancestry, it
should correlate at a statistically significant level with
variation in A1 frequencies. There are two null hypotheses to test: first, frequency of the A1 allele does not correlate with the frequencies of other traits under genetic
control; second, the frequency of the A1 allele does not
correlate with variation in body shape and size. Rejection of the first null hypothesis but nonrejection of the
second null hypothesis would allow us to reject an explanation of Australians’ body shape and size patterning in
terms of common ancestry.
MATERIALS AND METHODS
Subject selection and characteristics
77
Fig. 1. Main tribal distributions (showing the 56 tribal/
pooled groups with data on femoral length). Key: Dark gray—30
basic groups, medium gray—regional pooled series (1952–1954),
light gray—regional pooled series (1938–1939).
as discussed later. With respect to likely nutritional and
dietary variation, few led what may be described as a
fully traditional lifestyle at the time of data collection.
Among the main groups examined in the northwest, for
example, fieldwork was carried out at 51 locations, comprising 17 camps, 9 government settlements, 18 pastoral
stations, and 7 missions (Birdsell, 1993).
Morphological data
Birdsell’s fieldwork was directed at gathering as much
anthropological data as possible on Australian Aborigines as the world’s largest remaining hunter-gatherer
population, whose indigenous lifestyle was disintegrating
rapidly following white colonization. While Birdsell
began preliminary analyses of ecological patterning in
Aboriginal society (Birdsell, 1953, 1979), the phenotypic
data analyzed here were gathered to seek evidence for
divergent phylogenetic backgrounds (Birdsell, 1967,
1993). Given his stated aims, Birdsell’s database provides an opportunity to explore both environmental and
phylogenetic influences on morphological variation, but
is less suited to social and demographic variation. Given
his intent of examining hunter-gatherers, a greater proportion of the data derive from the remaining remote
groups in the interior and northwest of the continent.
Inadequate data on the age distribution within each
group is a deficiency: each is an adult sample but Birdsell provides only mean ages for the 57 goups with data
on major morphometric variables (minimum mean age
27.0 years, maximum mean age 61.8, average mean
41.9). Another limitation is the virtually complete absence of data for females, who for various (presumably
cultural) reasons were less accessible. Also absent are
data on other information of potential relevance here,
particularly details on lifestyle, diet, disease, and social
context. Furthermore, by present day standards the lack
of any formal arrangements for obtaining informed
consent (as opposed merely to cooperation) from the
participants is a serious omission, one which cannot be
rectified in retrospect. Fortunately, while the last problem must remain, additional sources can partially compensate for the paucity of information in some domains,
Included are the quantified data on Australian Aborigines presented by Birdsell (1993), transcribed onto a
digital database (SPSS 11.0, SPSS, 2001). Birdsell’s 185
variables include 114 morphological variables—measures
and indices relating to body form, the cranium, limbs,
and features such as skin color and hair form, together
with dental data (49 variables), serological data (10 variables), and composite scores (12 variables). The data
comprise mean values for 217 tribal and regional groups.
The number of individuals examined in each group
ranges from as few as five up to over 100, with an average of 67 for the 30 ‘‘basic’’ tribes in the northwest.
These latter groups alone provide data on all of the variables. While the whole continent is covered, data on
most of the body and limb measures are less comprehensive (data on limb segment lengths and ratios, for
instance, derive from 57 goups), with a bias towards the
northwestern region (Fig. 1).
Environmental data
Meteorological records in Australia are extensive and
provide an indication of climatic conditions, to which the
indigenous population may be expected to have become
adapted over the past 10,000 years. Nine environmental
variables are included on the database (Table 1). These
comprise seven derived from meteorological records at
weather stations (available on the website of Bureau of
Meteorology, 2006, Commonwealth of Australia), a wind
chill variable calculated from air temperature and wind
velocity records using the Steadman (1984) formula, and
a summer apparent temperature (AT) variable extrapolated from published maps of AT (Steadman, 1994).
Meteorological data from 103 weather stations were
American Journal of Physical Anthropology—DOI 10.1002/ajpa
78
I. GILLIGAN AND D. BULBECK
TABLE 1. Environmental variables
MT
ST
WT
H
R
C
W
WC
AT
Mean temperature
Summer temperature
Winter temperature
Relative humidity
Rainfall
Cloud cover
Wind speed (winter)
Wind chill
AT (summer)
Mean daily 3 p.m. air temperature (8C)
Mean daily January 3 p.m. temperature (8C)
Mean daily July 3 p.m. temperature (8C)
Mean daily 9 a.m. relative humidity (%)
Mean annual rainfall (mm)
Mean number of cloudy days (annual)
Mean daily July 3 p.m. wind speed (km/h)
Mean annual 3 p.m. temperature/wind (Steadman formula)
January average apparent temperature at solar noon
TABLE 2. Correlation matrix—environmental variables (bold type: significant at 0.0014 level, 2-tailed test)
MT
ST
WT
H
R
MT
ST
WT
H
R
C
W
WC
AT
—
þ0.7283
P ¼ 0.000
—
þ0.8561
P ¼ 0.000
0.8668
P ¼ 0.000
—
0.3768
P ¼ 0.000
þ0.0641
P ¼ 0.355
0.3433
P ¼ 0.000
—
0.1805
P ¼ 0.009
þ0.0162
P ¼ 0.815
þ0.2382
P ¼ 0.000
þ0.2503
P ¼ 0.000
—
þ0.0439
P ¼ 0.527
0.1536
P ¼ 0.026
0.1049
P ¼ 0.130
þ0.2057
P ¼ 0.003
þ0.1579
P ¼ 0.022
—
þ0.2253
P ¼ 0.001
þ0.0642
P ¼ 0.355
0.0078
P ¼ 0.910
0.0864
P ¼ 0.213
þ0.1262
P ¼ 0.068
þ0.2148
P ¼ 0.002
—
þ0.6030
P ¼ 0.000
0.0246
P ¼ 0.723
0.1535
P ¼ 0.026
þ0.1848
P ¼ 0.007
þ0.0851
P ¼ 0.219
þ0.0731
P ¼ 0.291
0.5950
P ¼ 0.000
—
0.0585
P ¼ 0.399
þ0.2034
P ¼ 0.003
þ0.1855
P ¼ 0.007
þ0.0672
P ¼ 0.332
þ0.2105
P ¼ 0.002
0.0791
P ¼ 0.254
þ0.3454
P ¼ 0.000
þ0.1260
P ¼ 0.068
—
C
W
WC
AT
matched to the locations of Birdsell’s 217 tribal groups.
Some weather stations were used more than once, as a
number of the tribal areas overlap. Each lies within the
corresponding tribal area, with distances from the
centers of the tribal areas ranging from 8 to 304 km; the
average distance is 79.8 km.
For the temperature variables, afternoon (3 p.m.)
rather then morning (9 a.m.) data are more suitable for
the analyses, as the former show greater regional variation. The converse applies with relative humidity, for
which the morning (9 a.m.) data show more variation
than the 3 p.m. data. Wind velocity shows marked diurnal variation, being generally lowest (and varying less
between locations) in the morning, hence 3 p.m. wind
data were utilized; winter (July) wind intensity is most
relevant in terms of wind chill.
Wind chill. The term ‘‘wind chill’’ used in the literature
is perhaps unfortunate because, for positive temperature
values, the calculated variable increases with decreased
exposure to cold winds. Given their greater variation
than 9 a.m. data, the 3 p.m. temperature and 3 p.m.
wind speed data are selected. Annual rather than seasonal averages, rather unexpectedly, show the greatest
regional variation. This presumably is due to a seasonal
contrast between tropical climates in the north (where
winds are strongest in winter months) and temperate
climates in the south (where winds are stronger in
summer months). The result is a cancelling effect
between tropical and temperate zones for both the
summer and winter wind chill estimates. For this reason, the correlation results for wind chill may be lower
than might otherwise be expected.
Apparent temperature. The combined effect of major
climatic variables such as air temperature, humidity,
wind velocity, and solar radiation can be estimated as
the ‘‘apparent temperature,’’ or AT (Steadman, 1984,
1994). AT provides a quantitative approximation for the
combined impact of these variables on human physiological requirements and the development of thermal
responses, including morphological adaptations. With
AT, primarily a measure of heat rather than cold stress,
summer rather than winter figures are preferred.
Colinearity. The potential problem of multicolinearity
between the environmental variables is examined by
means of partial correlation coefficients. These reflect
the correlations between each pair of variables, while
controlling for the effects of the other seven, as shown in
the correlation matrix (Table 2). For the 36 partial correlations between the nine environmental variables, a
Bonferroni correction gives a significance level of 0.05 7
36 ¼ 0.0014.
Summer and winter temperatures correlate negatively,
reflecting a contrast in the database between desert and
tropical climates (as discussed later). Winter (July) temperature correlates negatively with humidity (reflecting
a dominance of southeastern coastal and highland zones
over the northern tropical climatic pattern for humidity)
and positively with rainfall (a desert vs. tropical pattern), although the two moisture variables (humidity
and rainfall) nonetheless correlate positively with one
another. The wind chill correlations reflect the fact that
wind chill is a product of temperature and wind velocity.
AT emerges as relatively independent, as does cloud
cover. Notwithstanding the correlations between certain
variables, correlated variables need not exert similar
American Journal of Physical Anthropology—DOI 10.1002/ajpa
ENVIRONMENT AND MORPHOLOGY IN ABORIGINES
79
Canonical correlation analysis. CCA is one of the
most general methods for examining the relationships
between two sets of variables. It has been employed in a
range of disciplines, including physical anthropology
(e.g., Lahr and Wright, 1996). CCA may be conceptualized as the multivariate analogue of multiple regression
analysis (Lattin et al., 2003). Unlike multiple regression,
which combines the variables in one set to maximize the
correlation of this set with another (single) variable,
CCA allows for this procedure to be applied to two sets
of variables. It generates pairs of correlated variable
sets, and these are interpreted by examining the variable loadings. Unlike factor analysis, CCA also generates
output for testing the statistical significance of the
results. Analyses were performed using the canonical
correlation macro (‘‘cancorr’’) available through syntax in
the SPSS 11.0 (SPSS, 2001) program.
Additional analyses
Fig. 2. Tibial length and annual temperature: r ¼ þ0.664.
selection pressures on human morphology, and so all of
the environmental variables will be examined with
respect to each morphological trait.
The database
Each of Birdsell’s 217 Aboriginal groups is matched to
climatic data comprising the nine environmental variables from the local weather station. Tribal distributions
correspond to those given in Birdsell (1993), with reference to Tindale (1974) where clarification was required.
For each tribal (or pooled tribal) group, a weather station with data on all nine variables was identified as
close as practicable to the centre of the tribal or pooled
tribal territory.
Statistical analyses
Pearson correlations. A bivariate Pearson correlation
coefficient (r value) was calculated for each Birdsell variable on each of the nine environmental variables, using
SPSS 11.0 (SPSS, 2001) software. Histograms show that
in most cases, distribution frequencies for dependent
(morphological) variables approximate normal distributions. The environmental variables however are skewed,
being tied to the tribal distributions, which reduces the
likelihood of detecting environmental associations. Tests
for significance were mainly two-tailed; one-tailed tests
were used where there exist clear expectations of environmental trends occurring in a particular direction.
Bonferroni corrections were made to adjust for simultaneous testing on nine climatic variables, reducing the
significance level to no greater than 0.0056 (0.05 7 9),
with lower levels for related groups of tests [e.g., the significance level for each of the four limb segment
lengths—radial, humeral, femoral, tibial—is 0.05 7 (9 3
4) ¼ 0.0014]. Linear regression analysis was performed
for each morphological (dependent) variable on each
environmental (independent) variable. Scatterplots were
prepared for each analysis and a regression line fitted.
Figure 2 is an example, showing the results for tibial
length and annual temperature.
Phylogeny. For the phylogenetic analysis examining
correlations with the A1 allele, the traits recorded by
Birdsell were divided into six groups: those under simple
genetic control; those that reflect sexual dimorphism;
those under multiple genetic control; those that reflect
complex developmental expression; those that reflect
environmental effects or adaptation to the environment;
and those that measure body size and shape (Appendix
B). The last traits were assigned to a separate group
because they specifically relate to our second null
hypothesis. The first five groups of traits are ranked in
descending order of hereditability attributable to common ancestry. Under the first null hypothesis, the propensity of the A1 frequency to correlate with these traits’
expressions should not decrease with the descending
order of these five groups. Statistically significant correlation was tested with the Pearson product–moment
correlation coefficient, as in the main analysis. The Bonferroni correction is not appropriate here because we are
testing for statistical significance in the pattern of statistically significant correlations, rather than focusing on
individual results.
To avoid duplication in the correlation tests, Birdsell
traits which merely summarize individual traits (e.g.,
average frequency of 4-cusped lower molars, compared
with 4-cusped first, second, and third lower molars) are
excluded. In addition, with sets of traits whose expression is statistically interdependent (e.g., the frequency of
the different Rhesus alleles, or the frequency of the
different expressions of Carabelli’s anomaly), only one
expression is used. To increase the sensitivity of the test,
the expression focused on is the one that yields a statistically significant correlation or, failing that, a correlation as close as possible to statistical significance. Where
the sample size of Australian groups to be correlated
falls below 30, the test is excluded on the grounds of low
sample size (three cases, none statistically significant).
Simple genetic traits comprise blood groups and tawny
hair (cf. Birdsell, 1993). Sexually dimorphic traits, for
which the males’ Y chromosome is ultimately responsible, comprise hairiness, glabellar protrusion, and nasion
depression, the traits which Birdsell (1993) identifies as
particularly dimorphic sexually. Multiple genetic traits
include tooth size (Hillson, 1996) and dental morphology
(Scott and Turner, 1997), except with regard to the third
molars whose heightened variability reflects their eruption at the end of a complex developmental sequence.
American Journal of Physical Anthropology—DOI 10.1002/ajpa
80
I. GILLIGAN AND D. BULBECK
TABLE 3. Fisher exact test results for the proportion of traits that correlate with the frequency of the A1 allele
at a statistically significant level
Birdsell variable
na
sb
MT
ST
WT
H
R
C
W
WC
AT
Skin color
Body hair
Double hair whorl
Frontal hair tract 1D
Alveolar prognathism
Parietal sagittal elev.
Radial length
Rel. shoulder breadth
Relative sitting height
Head length
Cranial module
Upper molar cusps
Incisor winging
A1 allele
O allele
Rhesus R1
Rhesus R2
Lewis (aþ)
14
13
6
6
6
6
14
18
14
41
36
6
6
30
30
13
14
9
1
1
2
2
1
2
1
1
1
1
1
2
2
2
2
2
2
2
þ0.714
–0.432
þ0.149
–0.891
þ0.951
þ0.534
þ0.661
0.595
0.263
0.122
0.639
þ0.123
þ0.546
þ0.442
0.432
þ0.256
0.100
þ0.202
þ0.725
–0.418
þ0.154
–0.935
þ0.918
þ0.641
þ0.633
0.511
0.316
0.083
0.658
þ0.151
þ0.611
þ0.494
0.492
0.025
þ0.313
0.012
þ0.607
–0.429
þ0.437
–0.868
þ0.974
þ0.580
þ0.549
0.588
0.174
0.130
0.501
þ0.303
þ0.485
þ0.273
0.247
þ0.521
0.522
þ0.221
0.679
þ0.502
–0.005
þ0.749
0.973
–0.293
0.448
þ0.611
þ0.370
þ0.093
þ0.629
þ0.131
0.284
0.516
þ0.501
þ0.105
0.360
0.309
–.404
þ0.065
–0.825
þ0.218
–0.031
–0.628
0.260
þ0.251
þ0.249
0.176
þ0.369
0.599
0.142
0.499
þ0.502
þ0.474
0.758
þ0.573
0.667
þ0.339
–0.714
þ0.795
0.795
–0.701
0.694
þ0.594
þ0.178
þ0.095
þ0.500
0.671
0.627
0.544
þ0.532
0.051
0.234
þ0.335
0.118
þ0.249
þ0.420
þ0.645
–0.434
–0.472
0.594
þ0.144
þ0.130
þ0.099
þ0.100
þ0.269
0.479
0.192
þ0.153
0.100
þ0.232
þ0.162
þ0.698
–0.450
þ0.074
–0.909
þ0.916
þ0.575
þ0.717
0.577
0.303
124
0.583
þ0.079
þ0.579
þ0.447
0.434
þ0.234
0.124
þ0.109
þ0.743
–0.606
–0.007
–0.562
þ0.364
0.000
þ0.739
0.625
0.127
0.108
0.601
þ0.237
þ0.251
þ0.317
0.280
þ0.233
0.229
þ0.333
a
b
Number of tribal groups in analyses.
Significance test: 1 ¼ one-tailed, 2 ¼ two-tailed.
The division of non-dental traits between those that
have multiple genetic inheritance and those that are
complex developmental traits is based on these traits’
description in Birdsell (1993) and our understanding of
the general literature in human biological development;
both sets of traits are similar in how they correlate with
A1 frequencies, so any arbitrariness in our assignments
should not be of consequence. Finally, tooth displacement
and tooth wear appear to reflect the subjects’ use of their
teeth (Birdsell, 1993), while iris color and skin color are
textbook examples of adaptation to environmental factors (e.g., Jurmain and Nelson, 1994).
Weather station groups. There being more than twice
as many Aboriginal groups (217) as weather stations
(103), this could compromise the independence of the
meteorological indices in the analysis. To address this
problem, correlations were also performed using a modified database, in which group data were averaged for
each weather station, creating 103 dependent variables.
Results are compared with those obtained using Birdsell’s
217 goups.
Population densities. Population density has been
shown to affect body size in various hunter-gatherer
populations (Walker, Life history consequences of density
dependence and the evolution of human body sizes, Curr
Anthropol, submitted for publication), with higher densities linked to smaller body size (possibly a consequence
of greater competition for resources). In his study of ecological influences on the Aboriginal population, Birdsell
(1953) found rainfall to be the dominant factor in
Australia, with a strong correlation between tribal area
(hence population density, given relatively uniform tribal
size) and mean annual rainfall. He derived a formula
expressing density as a function of rainfall, which accords
well with density figures obtained independently from
ethnographic work across a wide range of Australian
climatic zones (Keen, 2004). Mean annual rainfall can
thus be used to calculate a population density variable,
yielding a tenth independent variable (in addition to the
nine environmental variables), which can be correlated
with the 103 goups based on meteorological stations
(and with the Bonferroni significance level adjusted
accordingly).
RESULTS
Relatedness analysis
The Pearson correlation tests summarized in Appendix
B strongly suggest rejection of the null hypothesis that
the frequency of the A1 blood group does not correlate
with other traits under genetic control. In the descending order from simple genetic traits to sexually dimorphic traits, to multiple genetic traits and complex developmental traits, and finally to environmental traits, the
proportion of statistically significant correlations with
the frequency of A1 drops dramatically from 80 to 25%,
then gradually to 24, 16, and finally 0%. This impression
is confirmed with the Fisher exact test where we compare the five groups of traits for their number of traits
which do, and do not, correlate with the frequency of A1
at a statistically significant level (Table 3). Compared
with every other group of traits, those under simple
genetic control have a tendency, which is statistically
significant, to correlate more frequently with the population’s A1 frequency.
The results of the Pearson correlation tests equally
strongly suggest that a second null hypothesis (that A1
does not correlate with variables expected to reflect environmental adaptation) cannot be rejected. Indeed there
are good grounds for positively accepting the second null
hypothesis, given that we would expect at least one of
the body size and shape traits to correlate (at P ¼ 0.05)
with the frequency of A1 by chance alone, and none does
(Appendix B). When we compare these body variables
with the other groups of traits, in their number of traits
that correlate with the frequency of A1 at a statistically
significant level, using the Fisher exact test (Table 3),
we find a statistically significant difference with simple
genetic traits, multiple genetic traits and complex developmental traits, and a nearly significant difference with
sexually dimorphic traits. With environmental traits
there is no difference at all. Based on the earlier results,
variation in the body size and shape variables would
appear to be entirely independent of genetic relatedness
American Journal of Physical Anthropology—DOI 10.1002/ajpa
ENVIRONMENT AND MORPHOLOGY IN ABORIGINES
between the populations, and may well reflect environmental adaptation.
Correlation coefficients
A total of 1,665 Pearson correlation coefficients were
calculated, for each of the 185 dependent variables on
each of the nine environmental variables. All results are
listed in Appendix A; also shown is the number of tribal
groups in each analysis (n), whether tests were one- or
two-tailed (s), and the result of the significance test.
Where the correlation is statistically significant, it is
highlighted in bold type.
Pigmentation. Skin color fails to correlate with cloud
cover but there are correlations with temperature. A
thermal association is reflected in the correlation with
wind chill (i.e., a darker skin color in hotter areas) and
also the correlation with AT. Oral pigmentation tends to
increase in cooler, windier and wetter environments.
Tawny or blonde hair, most common in the Western
Desert areas, correlates with hotter summers and lower
moisture levels.
Hair. Body and facial hair are more pronounced in
regions with colder winters and less rainfall. With the
exception of nasal tip hair, the minor hair variables (e.g.
nostril and finger hair) show no correlations. Scalp hair
presents a distinctive picture, with deep wave (curly)
hair correlating with moisture and cloud cover. Gray
head hair occurs at an earlier age where winters are
warmer, and male baldness increases at colder wind chill
levels.
Other nonmetric variation. This category covers a
disparate range of features (e.g., left-handedness, relative size of upper and lower lips), which are generally
poorly correlated with environmental variation. Alveolar
prognathism is more pronounced where conditions are
dry and winters are colder. The size of brow ridges does
not correlate with environmental variables, but brow
ridges tend to be more continuous in areas of higher
rainfall. Cryptose iris structure, a degenerative eye condition, is more common in areas with hot, dry summers.
Body and limb metrics. Body weight (or mass) correlates with dryness. Stature correlates with all of the
temperature variables, most strongly with wind chill.
Shoulder breadth fails to correlate with any environmental variables, while sitting height replicates the results
for stature. Limb lengths show consistent environmental
correlations (e.g., Fig. 2)—positive for temperature, negative for humidity and wind velocity. Wind chill correlates positively, meaning that limb lengths are reduced
with lower (colder) wind chill levels. Correlations are
higher for the lower compared with the upper limbs, and
for distal compared with proximal limb segments.
Body and limb indices. The ponderal index (stature 7
cube root of weight) measures overall body shape.
Results show a more linear build correlating with higher
mean temperatures, especially with milder winters.
Relative shoulder breadth also measures body shape,
and the results indicate a broader or stockier build with
lower mean temperature. Most of this association is
attributable to cold exposure, given the correlation with
winter but not with summer temperatures. The intermembral index (upper 7 lower limb lengths) shows that
the lower limbs become shorter than upper limbs in
relation to the wind chill effect.
81
Head and facial metrics. Head length and breadth
increase as winter temperatures fall; head length and
basal breadth increase with dryness. Total facial height
correlates negatively with winter temperature and positively with humidity. Nasal and mouth breadth correlate
with low humidity and with higher summer temperatures.
Head and facial indices. The cranial module, a measure of overall cranial size, increases as winter temperatures decline. The cephalic index, a measure of cranial
shape (cranial breadth 7 length), correlates negatively
with mean and summer temperatures, indicating a less
elongated (or dolichocephalic) head shape in cooler conditions. Both the cranial module and cephalic index
correlate negatively with wind chill. In other words, as
temperatures fall and the wind chill effect increases,
heads tend to become larger and rounder. Among facial
indices, the total facial index (breadth 7 height) correlates mainly with humidity. The nasal breadth index
(breadth 7 height) gives an overall indication of nasal
shape in relation to the frontal (or coronal) plane. A
broader shape is seen to correlate with higher summer
temperatures and with lower humidity. The nasal depth
index (depth 7 height) describes shape in the sagittal
plane, and shows the opposite trend: a more elongated
profile occurs with colder temperatures and higher
humidity.
Dental variables. Size of dentition correlates with
higher summer temperatures and dryness, and this
trend is also evident with tooth displacement, which correlates with mean annual temperature. Such a scenario
is consistent with the proposal that large tooth size in
Australian Aborigines reflects long-term environmental
selection, with drier conditions favoring larger grinding
surfaces (Wright, 1976). Although dental wear fails to
correlate with any environmental variables, it can be
noted that traditional hunter-gatherer practices had
largely disappeared by the time of Birdsell’s studies.
Full development of the fourth cusp on the maxillary
molars, and presence of Carabelli’s cusp (an accessory
cusp on maxillary molars), correlate with windier and
therefore cooler climates; in the case of fourth cusp
development there is a further correlation with dryness,
suggesting this may be a characteristic of desert populations. Incisor winging occurs, according to Birdsell, in
most human groups but most frequently among those in
eastern Asia (see also Scott and Turner 1997, p 1781);
his western Australian data suggest it is more frequent
along the northwestern coast and least frequent in
southern areas, and in this analysis the frequency correlates with milder winter temperatures.
Serological variables. For the ABO system, the A1
allele is more common in central and southern Australia,
while the B allele is restricted to Cape York and the Gulf
of Carpentaria. This pattern may contribute to the negative winter temperature and moisture correlations for
the A1 allele; the O allele shows an inverse pattern to
the A1 allele, as expected. For the MN groups, the N
gene is commonest in the western and southern desert
and southeast coastal areas, and also in the northeast
Queensland rainforest area, and it shows modest negative winter and wind chill temperature correlations. For
the Rhesus genes, R2 shows the most environmental
correlations; its frequency is maximal in the southern
desert areas, which may account for the correlations
American Journal of Physical Anthropology—DOI 10.1002/ajpa
82
I. GILLIGAN AND D. BULBECK
with hotter summers, colder winters and low rainfall. As
may be expected given the R2 pattern, the R1 gene shows
an environmental pattern that is the inverse of R2. The
Lewis aþ antigen has a similar pattern of associations,
though only two are statistically significant correlations.
The tendency for the serological variables to show similar patterns of correlations with climate variables is
consistent with the statistically significant correlations
between the frequency of A1, on one hand, and R1 and
Lewis aþ on the other hand (Appendix B).
Composite gradients. Composite gradients are of limited use in this study. They are calculated on the number
of clinal isophenes separating neighboring tribes, and
are used by Birdsell (1993, p 173) as an indirect measure
of ‘‘biological differentiation.’’ Their usefulness is compromised by their being derived from his ‘‘basic’’ tribal series, which encompasses barely 30% of the continent.
Also, each incorporates a restricted range of variables,
and neither the biological significance nor the statistical
validity of these composite indices is clear. The only
suggested correlations would relate Birdsell’s ‘‘body composite’’ to rainfall and wind velocity, and his ‘‘cranial
composite’’ to winter temperature and rainfall. Not
surprisingly, when all these composite gradients are
merged (i.e., the average of nonmetric morphological,
metric morphological, dental, and serological composites), no environmental correlations are found.
Fig. 3. CCA1: 15 metrical variables.
Each CCA uses six environmental variables for the
first (independent) variable set—mean annual temperature, wind chill and AT are excluded to reduce problems
of multicolinearity. Four CCA’s are needed to examine
separately the four different data types in the database:
metrical variables, indices, proportions (e.g., percentage
frequencies), and scaled variables (e.g. skin color and
body hair). The maximum number of variables included
is ideally related to the number of cases. A figure of at
least twenty times as many cases as variables is recommended, otherwise statistical power is reduced and even
strong canonical correlations between the sets may not
emerge as significant (Stevens, 2002). The small number
of cases (as few as 37) in the Birdsell database is barely
sufficient for CCA, so the number of variables included
in each analysis must be minimized, and is restricted
here to ~15.
especially facial height. While head length would not
seem to show much correlation with the degree of exposure to freshening winds, longer heads are clearly associated with drier conditions, especially desert conditions
(as shown by the loadings of hotter summer temperatures, colder winters and less cloud cover on this factor).
Aside from the strengths of the canonical correlations,
the utility of a CCA can be assessed by means of redundancy analysis. This measures the proportion of variance
of the variables in each set explained by its own canonical variate and also by the canonical variate of the other
set. In CCA1, the first two variates in Set 1 (environmental variables) account for 17.4 and 44.1%, respectively (a total of 61.5%) of the variance in the environmental set. For the morphological variables (Set 2), the
first two canonical variates account for 25.7 and 18.6%,
respectively, a total of 44.3%. Of particular interest is
the proportion of morphological variance (in Set 2)
accounted for by the first two environmental variates.
The figures are 22.8 and 14.4% for the wind exposure
and dryness factors, respectively, indicating that a total
of 37.2% of the morphological variance is explained by
these two environmental factors.
CCA1. The first CCA includes metrical variables (Fig. 3).
The environmental and morphological sets are correlated, with the first two canonical variates being significant (P < 0.000 and P < 0.047). Variable loadings for the
environmental set show the first factor is negative for
winter temperature and positive for wind velocity: it
represents the effects of exposure to cold winds (‘‘wind
exposure’’). The second factor loads negatively for
humidity, rainfall and cloud cover: it represents a dryness effect (and also correlates with warmer summers).
While there are no definitive criteria for the interpretation of variable loadings, a loading of at least 0.300 is
generally considered the minimum if a variable is to be
interpreted as part of a factor (Tabachnick and Fidell,
2001). Loadings of the morphological variables on the
first factor point to stature and its components, or more
precisely vertical variables, as those most related to
wind exposure. Loadings on the second (dryness) factor
essentially contrast head length with the other variables,
CCA2. Results of CCA2 using fifteen indices are shown
in Figure 4. Two canonical variates are significant (P <
0.000 and P < 0.015). Indices corresponding to linearity
of body build (the ponderal, tibial-sitting height and
tibial-femoral indices) load positively on the factor that
represents lack of wind exposure, whereas those relating
to more stocky build (relative shoulder breadth and
the calf-tibial index) load negatively. The intermembral
index also loads negatively: lower limbs become relatively
shorter as wind exposure increases. The second (dryness)
factor loads positively with the nasal breadth and limb
segment/sitting height indices, and negatively with the
facial (bizygomatic breadth 7 facial height), relative
sitting height, relative shoulder breadth, and cephalic
indices—the last corresponding to a more brachycephalic
head shape in moister conditions. Redundancy analysis
for CCA2 shows the two canonical factors accommodating 72.8% of the environmental variance and 37.6%
Canonical correlation analyses
American Journal of Physical Anthropology—DOI 10.1002/ajpa
ENVIRONMENT AND MORPHOLOGY IN ABORIGINES
83
Fig. 6. CCA4: 18 scaled variables.
Fig. 4. CCA2: 15 indicial variables.
tropical Australia (e.g., the Lewisþ antigen, Carabelli’s
fissures). Here we seem to be dealing with differences
between the two main regions sampled, rather than independent correlations between environment and phenotype. Only the A1 allele, generally more common in the
cooler climate of southern Australia, would suggest itself
as an environmentally correlated variable. As discussed
later, we interpret this result as a practical illustration
of Galton’s problem.
Fig. 5. CCA3: 14 proportional variables.
of the morphological variance, with the environmental
variates accounting for 25.8% of the latter.
CCA3. Results of CCA3 using fourteen proportional (or
percentage) variables in Set 2 are shown in Figure 5.
Two canonical variates are significant at the P < 0.000
level. Environmental loadings suggest the first factor
corresponds to moisture, and the second factor corresponds to an absence of cold winds. Redundancy analysis
shows 44.2 and 38.2% of variance in Sets 1 and 2,
respectively being accommodated by the two canonical
variates; 34.4% of Set 2 variance is accounted for by the
two environmental factors.
This CCA includes serological, dental and hair variables. Of the last, tawny hair loads negatively with both
environmental factors, consistent with its geographical
concentration in the Western Desert area. Serological
and dental variables tend to fall in a linear pattern from
those found among desert Aborigines (tawny hair, fourcusped maxillary molars) and those prevalent in moist
CCA4. This includes 18 scaled variables (Fig. 6). The
first two canonical variates are significant at the P <
0.000 and P < 0.003 levels. Canonical loadings show the
first environmental factor corresponds mainly to winter
wind exposure, and the second to moisture. Redundancy
analysis shows 50.6 and 22.9% of variance in Sets 1 and
2, respectively, accommodated by the two canonical variates; a modest 20.1% of Set 2 variance is accounted for
by the environmental factors, with most of this (15.3%)
being the first (wind exposure) factor.
The hair variables load positively with exposure to
colder winds, while darker skin loads negatively,
together with larger brow ridges, larger teeth, and a less
parabolic palate shape. Few variables load effectively on
the moisture factor, with only brow ridge and ear lobe
size loading positively and alveolar prognathism negatively. As with CCA3, the morphological variables follow
a linear plot contrasting colder winters and aridity
(e.g., alveolar prognathism, two-cusped lower premolars)
with warmer winters and humidity (e.g., darker skin
color, large brow ridges), suggesting a distinction
between desert and northern tropical populations.
Population density
In comparison to the climatic variables, the group
density variable correlates with few of Birdsell’s variables. In every instance where it reaches statistical significance, this is in concert with the rainfall variable from
which it derives. Of interest, however, is that it does
correlate significantly (and negatively) with body weight,
as may be predicted from research relating body size to
population density (Walker, submitted).
American Journal of Physical Anthropology—DOI 10.1002/ajpa
84
I. GILLIGAN AND D. BULBECK
TABLE 4. Tropical versus desert patterns
of environmental correlations
Tropical
Desert
MT
ST
WT
H
R
C
W
WC
AT
:
;
;
:
:
;
:
;
:
;
:
;
;
:
:
;
:
;
Environmental correlations—Temperate database
A major problem is evident from the number of significant environmental correlations seen with dental and
serological variables. These suggest there is regional
patterning in the Birdsell database, which is causally
unrelated to environmental variation, but which results
in what may be termed pseudoenvironmental correlations. The regional patterning in the distribution of the
ABO system genes and Rhesus variants shows a contrast
between the central and southern parts of the continent
on the one hand, and the coastal—especially the northern coastal—areas on the other. This patterning may
correspond to a contrast between desert and tropical
populations, with the serological results revealing the
bias in Birdsell’s data structure, viz., sampled groups
tend to be either desert groups or tropical (especially
northwest Australian) groups. The pattern of significant
correlations between the serological and environmental
variables indeed corresponds to an environmental contrast between tropical and desert regions (Table 4), quite
different from the pattern seen with morphological
variables.
One way to explore this issue is to remove from the
correlation analyses all those cases that lie within the
tropics. The simplest method is to include only those
cases deriving from south of the Tropic of Capricorn, i.e.,
those lying at latitudes greater than 238300 S. Since
most of Birdsell’s cases derive from north of the Tropic
(and especially from the northwest), the remaining number of cases is greatly reduced, to as low as six cases for
many variables, and only very strong linear associations
can reach statistical significance. The results for representative, illustrative variables using cases located south
of latitude 238300 S are shown in Table 5.
The tropical versus desert patterns either disappear or
are greatly diminished using only the subtropical groups.
In particular, the unexpected environmental correlations
for variables such as hair whorls, dental cusp patterns,
and serological genotypes essentially disappear when the
tropical cases are removed from the correlation analyses.
Accordingly, their appearance of environmental patterning can be attributed to genetic differentiation between
tropical and desert Aborigines. Of further interest, the
expected negative correlation between skin color and
cloud cover emerges. Evidently, the association between
higher temperatures and greater cloud cover in tropical
climates (the former having a positive correlation and
the latter a negative correlation with skin color) means
that the opposing correlations tended to cancel out.
When tropical cases are removed, strong and independent correlations with temperature (positive) and cloud
cover (negative) are unmasked.
The distorting effects of the tropical sampling bias in
the Birdsell database are clearly evident with the cranial
module. Being the average of cranial length, breadth
and height, this variable provides a simple measure of
overall cranial size, and is expected on thermal grounds
to correlate negatively with temperature. The cranial
module shows only modest correlations using the full
database but, with tropical cases excluded, stronger correlations emerge that accord well with thermal considerations, despite a substantial reduction in case numbers
(from 93 to 36).
While the desert versus tropical correlation pattern
largely disappears for the serological variables, the A1
and O alleles retain their (complementary) environmental correlations to some extent. The particular associations of A1 are with higher summer temperatures and
with lower humidity, rainfall and cloud cover, reflecting
the distinction between the more arid and better
watered parts of temperate Australia. Recall, however,
that the A1 allele loaded on the wind exposure and moisture factors in CCA3 (Fig. 5), corresponding instead to
cold, windy, and wet environments. If the A1 allele
reflects climatic adaptation, we could not say what that
climate was. The geographical distribution of A1, with its
remarkable pattern of clinal decrease away from its area
of concentration in the southern arid zone (Birdsell,
1993, Fig. B1), may well be important in terms of population relationships across the Australian continent, but
the resulting climatic correlations are explicable in
terms of Galton’s problem.
Environmental correlations by weather station
As previously discussed, an alternative strategy to
ameliorate the geographical bias in Birdsell’s coverage is
to use weather stations rather than Aboriginal groups.
To keep the length of this article manageable, we restrict
ourselves to describing the results of the weather station
analysis, which are available from the corresponding
author on request. Correlation results using the weather
station database (103 merged Aboriginal groups corresponding to the weather stations) are very similar to
those obtained using the original Birdsell database. This
removes the overlap between Aboriginal groups and
weather stations as a confounding factor in the main
analysis. Smaller sample sizes (number of groups with
data on each variable) reduce the number of significant
correlations only marginally. With the exception of humeral length (which shows the weakest correlation on the
original database, consistent with the expectations of
Allen’s rule), significant environmental correlations of
comparable magnitude are seen for body size and shape
variables, while the number of significant dental and
serological correlations is lower.
DISCUSSION
Body and head shape
The results show that considerable regional patterning
of morphological variation exists within the Australian
Aboriginal population. Most of the major regional trends
documented by Birdsell (1993) are found to vary in concert with environmental conditions. Significant correlations exist for the main measures relating to body shape,
limb lengths and proportions, and head size and shape.
As seen in both the bivariate (Pearson correlation) and
multivariate (CCA) analyses, thermal (including wind
chill) trends are prominent but the moisture variables
also show significant trends. That the trends among the
body shape variables largely reflect morphological
responses to environmental conditions is suggested both
by these trends occurring in directions predictable from
biological principles, and by a lower incidence of signifi-
American Journal of Physical Anthropology—DOI 10.1002/ajpa
85
ENVIRONMENT AND MORPHOLOGY IN ABORIGINES
0
a
TABLE 5. Subtropical (>23830 S latitude) environmental correlations
Birdsell variable
b
n
sc
MT
ST
WT
H
R
C
W
WC
AT
Skin color
Body hair
Double hair whorl
Frontal hair tract 1D
Ear axis
Alveolar prognathism
Parietal sagittal elev.
Thumb extension
Radial length
Rel. shoulder breadth
Relative sitting height
Head length
Cranial module
Upper molar cusps
Incisor winging
A1 group
O group
Rhesus R1
Rhesus R2
Lewis (aþ)
14
13
6
6
6
6
6
6
14
18
14
41
36
6
6
30
30
13
14
9
1
1
2
2
2
1
2
2
1
1
1
1
1
2
2
2
2
2
2
2
þ0.714
0.432
þ0.149
0.891
þ0.377
þ0.951
þ0.534
þ0.100
þ0.661
0.595
0.263
0.122
0.639
þ0.123
þ0.546
þ0.442
0.432
þ0.256
0.100
þ0.202
þ0.725
0.418
þ0.154
0.935
þ0.277
þ0.918
þ0.641
þ0.154
þ0.633
0.511
0.316
0.083
0.658
þ0.151
þ0.611
þ0.494
0.492
0.025
þ0.313
0.012
þ0.607
0.429
þ0.437
0.868
þ0.559
þ0.974
þ0.580
0.075
þ0.549
0.588
0.174
0.130
0.501
þ0.303
þ0.485
þ0.273
0.247
þ0.521
0.522
þ0.221
0.679
þ0.502
0.005
þ0.749
0.497
0.973
0.293
0.122
0.448
þ0.611
þ0.370
þ0.093
þ0.629
þ0.131
0.284
0.516
þ0.501
þ0.105
0.360
0.309
0.404
þ0.065
0.825
þ0.218
0.138
0.031
0.628
þ0.237
0.260
þ0.251
þ0.249
0.176
þ0.369
0.599
0.142
0.499
þ0.502
þ0.474
0.758
þ0.573
0.667
þ0.339
0.714
þ0.795
0.471
0.795
0.701
þ0.253
0.694
þ0.594
þ0.178
þ0.095
þ0.500
0.671
0.627
0.544
þ0.532
0.051
0.234
þ0.335
0.118
þ0.249
þ0.420
þ0.645
þ0.306
0.434
0.472
0.527
0.594
þ0.144
þ0.130
þ0.099
þ0.100
þ0.269
0.479
0.192
þ0.153
0.100
þ0.232
þ0.162
þ0.698
0.450
þ0.074
0.909
þ0.274
þ0.916
þ0.575
þ0.179
þ0.717
0.577
0.303
124
0.583
þ0.079
þ0.579
þ0.447
0.434
þ0.234
0.124
þ0.109
þ0.743
0.606
0.007
0.562
þ0.314
þ0.364
0.000
þ0.497
þ0.739
0.625
0.127
0.108
0.601
þ0.237
þ0.251
þ0.317
0.280
þ0.233
0.229
þ0.333
a
b
c
Correlations significant at a Bonferroni level of 0.0056 are shown in bold.
Number of tribal groups in analyses.
Significance test: 1 ¼ one-tailed, 2 ¼ two-tailed.
cant trends among those morphological variables that
would not be expected to show such correlations.
The major trends evident in these analyses involve
associations between morphological variation and the
temperature variables. This is seen for instance with
stature and other measures reflecting body build, such
as the ponderal index, which correlates most strongly
with winter temperature. Body weight however fails to
show any correlation with temperature variables. Birdsell (1993) observed that this measure is scarcely indicative of hunter-gatherer variation, as all the data were
collected among Aboriginal populations exposed for at
least some decades and often several generations to a
nonindigenous diet, and rapid weight gain was common.
Sitting height, which in studies of other groups worldwide correlates negatively with temperature (Roberts,
1978), shows a positive relationship here. This anomaly
could reflect a predominant total stature effect among a
population with an overall tropical linear morphology.
Relative sitting height shows the expected negative temperature trend, at least with summer temperature. Limb
measures show temperature trends in accordance with
Allen’s rule: they become shorter with lower temperatures, and this applies particularly to the distal limb
segments.
measures tend to show more pronounced correlations
with moisture than with the temperature variables. In
other studies, nasal shape correlates more strongly with
humidity than with temperature (Wolpoff, 1968; Carey
and Steegmann, 1981).
Moisture in the form of humidity and rainfall also
affects the relationship between temperature and body
shape. The cooling advantage of a higher surface/volume
ratio in hot environments depends largely on evaporation of perspiration from the skin surface. Evaporative
cooling is compromised in conditions of high atmospheric
moisture content, especially if coupled with reduced wind
chill as it is in relatively closed (e.g., heavily forested)
environments. This has been posited as an explanation
for the reduced body size of the Pygmy and ‘‘Negrito’’
populations residing in hot, humid rainforest habitats.
In these circumstances, where keeping cool is the main
priority, a large skin surface is of little use for evaporative cooling but a lower body mass means less metabolic
heat production, so the net thermal result is reduced
body size (Hiernaux et al., 1975; Cavalli-Sforza, 1986).
In Australia, the small stature of Aboriginal groups in
the rainforests of north Queensland probably reflects adaptation to local conditions, removing any explanatory
value from Birdsell’s claim that these groups are the
surviving relicts of a Pleistocene ‘‘Negrito’’ immigration.
Moisture, nasal shape, and stature
Besides temperature, the other key environmental variables identified in studies of human variation include
humidity, wind velocity, and solar radiation. Each tends
to be correlated with temperature but can also have an
independent association with certain morphological features. For instance, body hair correlates negatively with
both winter temperature and rainfall. The latter correlation may relate to the reduced thermal value of hair
where moisture levels are high. Variables relating to the
form of head hair also provide some interesting trends,
suggesting a relationship between more curled scalp hair
and moister conditions. Generally, the facial and nasal
Skin color
Skin color has a long history in biology as an example
of environmental variation, with Gloger’s Rule stating
that pigmentation increases in warmer environments.
Gloger was an ornithologist who documented how the
plumage of birds (particularly their coloration) varies
consistently with thermal conditions. He generalized his
observations to mammals, noting that the latter are less
mobile than birds and so are more subject to local conditions (Gloger, 1833). Birdsell (1993) himself noted the
general trend for Australian Aboriginal skin color to
become lighter moving from lower to higher latitudes.
American Journal of Physical Anthropology—DOI 10.1002/ajpa
86
I. GILLIGAN AND D. BULBECK
In humans, skin pigmentation correlates with levels of
sunlight, although other factors are involved (Jablonski
and Chaplin, 2000). In the present study, skin color
(intensity of pigmentation) among Australian Aborigines
is found to correlate with temperature and humidity
and, at least south of the Tropic of Capricorn, with cloud
cover. One reason for an association between temperature and skin color is that darker skin could be advantageous in hotter environments, as it results in a warmer
skin surface, which aids in cooling through direct radiation (Harrison, 1975). In terms of vitamin D synthesis, a
less pigmented skin surface that facilitates penetration
of UV light is especially advantageous beyond the latitude of around 408, corresponding to that of Tasmania,
where the Aboriginal population is often described as
having a lighter or coppery skin tone (Roth, 1899).
Nonclimatic factors
Given the lack of suitable detailed information in
Birdsell’s database, the possible role of influences such as
nutrition, health status, or social environment cannot be
explored in full. However, it may be noted that besides tribal
area and density (influenced strongly by rainfall, the main
determinant of enviromental productivity), Birdsell stressed
that the main elements of Aboriginal social organization
(including marriage and kinship systems) do not vary in
any consistent manner across the full range of ecological
conditions in Australia (Birdsell, 1979). This, in his view,
reflects the adaptive flexibility of their social arrangements
and economic strategies. While some 17 major culture areas
are delineated by the main drainage systems (Peterson,
1976), the basic social unit—the patrilineal band—is found
everywhere, as is polygyny (Peterson, 1986). Although
higher polygyny (i.e., large numbers of wives) may have
been more common in areas with higher rainfall and population densities, the organization of production was broadly
similar across differing ecological zones and the environment appears to be a ‘‘poor predictor’’ of kinship systems
(Keen, 2004, p 331, 396). In this analysis, the one attribute
of Aboriginal social structure that does vary consistently
with environment—population density—correlates (negatively) with body weight but not to any marked extent with
other measures of body size and shape.
While it may be anticipated that social factors would
have minimal effects on the present findings, nutrition
and diet could exert some confounding effect on the observed environmental patterning of body size and shape.
This can arise because, when compared with northern
(especially inland) groups, southern groups in cooler
environments had a longer contact history since white
settlement. Greater exposure to a Western diet could
have resulted in a greater propensity to excessive weight
(and diabetes) among the southern groups, hence contributing to a more stocky body shape. However, the preservation of significant correlations between climate and
body shape within subtropical groups (i.e., south of the
Tropic of Capricorn) suggests an independent environmental association, even allowing for a nutrition effect.
Also, the merged weather station groups in the second
database will have a more mixed dietary background
(comprising a greater blend of indigenous, station and
mission samples), yet these still show the predicted
climatic correlations.
Having established the existence of systematic relationships between major body and craniofacial shape variables
and climatic variables (seasonal and wind chill tempera-
tures and also relative humidity), it remains true that the
correlations, while statistically significant, have r2 values
indicating that only a modest proportion of total variance is
explained by these relationships. For example, the r value
for tibial length and summer temperature is þ0.664, with
the r2 showing only 44% of variance is explained. While statistically significant, a reasonable question is how much important biological variation is explained? Clearly, not all variation is explained by climate, and numerous other factors
may contribute a proportion to the remaining unexplained
variance. Nonetheless, 44% is a respectable proportion for
any single factor, and is consistent with Allen’s Rule.
Another question is whether our simple test for phylogenetic
dependence, using A1 as a genetic, marker is sufficient to
address this issue here. We suggest it is sufficient to show
that such a confounding influence is unlikely to be a major
problem in these analyses, although concede that additional
analyses such as multiple regressions of body metrics as
functions of phylogenetic markers and climate would enable
better separation (and quantification) of these independent
effects. However, this would introduce considerable further
complexities to an already complex study, and we reserve it
for a future work.
It remains possible that people who differed to some
degree biologically may have gravitated into regions
where they were most suited morphologically. In this
case, the environmental correlations could reflect ecological niche preference, a component of niche construction
(Odling-Smee et al., 2003). Since natural selection and
ecological niche preference would both tend to produce
the same observed correlations, these two explanations
would be difficult to distinguish. They are not mutually
exclusive, and may be expected to have operated together. Genetic drift will also account for a component of
present morphological variation, especially among inland
groups where population densities are lower (and intergroup contacts fewer), although environmental adaptation appears the more salient factor across the continent
as a whole (Pardoe, 2006).
CONCLUSION
These analyses indicate that a sizeable proportion of
morphological variation within the Australian Aboriginal
population can be explained in terms of adaptations to
the physical environment. Correlations (particularly
with temperature) are found for measures relating to
body shape, head size and shape, limb lengths, and limb
proportions—correlations predictable from biological
principles of long-term human adjustment to environmental conditions. All of these climate-related trends
can also be discerned in the clinal distribution maps
published by Birdsell (1993), even though clinal maps
are a crude technique for representing climatic adaptation owing to the very imperfect correlation between latitude and climatic variation. In many cases, Birdsell had
observed the geographical patterning, but tended to play
down its significance because of his particular interest in
extracting evidence in seeming support of his trihybrid
theory of Australian origins. While the results of our
analysis leave little room to support the view that the
deep ancestry of Australian Aborigines can somehow be
discerned in their geographical patterns of phenotypic
variation as recorded by Birdsell, we trust that our
analysis builds on his legacy of an extraordinary documentation of biological variation across the huntergatherer continent of precontact Australia.
American Journal of Physical Anthropology—DOI 10.1002/ajpa
87
ENVIRONMENT AND MORPHOLOGY IN ABORIGINES
ACKNOWLEDGMENTS
The authors wish to thank the following individuals:
Bob Steadman for his correspondence on wind chill and
apparent temperature, Peter White for his comments on
an earlier draft of the manuscript, Robert Walker for drawing our attention to Galton’s problem and pointing out the
need to consider phylogenetic and other issues, Richard
Wright for his advice on the CAA analyses and other statis-
tical matters, Ian Keen for advice on traditional Aboriginal
social organization, and an anonymous reviewer for useful
comments. Also, appreciation is expressed to the Bureau of
Meteorology, Australia, for its policy of allowing use of
copyright meteorological data for academic purposes.
Finally, the cooperation of the many Aboriginal people who
allowed their persons to be recorded by Birdsell, at a time
when informed consent was not required in anthropological fieldwork, should be acknowledged.
Appendix A
TABLE A1. Full Birdsell database—Environmental correlations (statistics and abbreviations as in Table 4)
Birdsell variable
Pigmentation
Skin color
Oral pigment–area
Oral pigment–intensity
Tawny (blonde) head hair
Eye color (iris)
Hair systems
Beard abundance
Body hair
Nasal tip hair
Nostril hair (narial)
Ear hair (tragus)
Finger hair
Head hair–deep wave
Head hair–crisp
Head hair–gray
Baldness
Hair whorl–clockwise
Hair whorl–anticlockwise
Hair whorl–double
Hair whorl–frontal
Frontal hair tract type I
Frontal hair tract type II
Frontal hair tract type III
Frontal hair tract type 1A
Frontal hair tract type 1B
Frontal hair tract type 1C
Frontal hair tract type 1D
Frontal hair tract type 1E
Supra-nasionhair streams
Widow’s peak
Circumcaruncular (medial
eyelid) hair
Other non-metric
Left-handedness
Finger formula (4th > 2nd)
Toe formula (2nd > 1st)
Hammertoes
Nasal profile (convex)
Nasal cartilage anomaly
Ear pili (preauricular pit)
Darwin’s points
Ear protrusion
Ear axis (rearward)
Ear helix roll
Ear lobe– hanging
Ear lobe size
Eye axis obliquity
Prognathism–total
Prognathism–mid-facial
Prognathism–alveolar
Lip proportions
(upper > lower)
Bilateral chin
n
s
MT
ST
WT
H
R
C
W
WC
AT
65
37
57
66
57
1
1
1
1
2
þ0.783
0.779
0.584
0.102
þ0.323
þ0.432
0.211
0.689
þ0.460
þ0.117
þ0.728
0.659
0.311
0.437
þ0.351
0.260
þ0.179
þ0.784
0.589
0.132
þ0.128
0.184
þ0.477
0.591
þ0.108
0.243
0.065
þ0.544
0.386
þ0.147
0.252
þ0.443
þ0.027
þ0.168
0.124
þ0.767
0.772
0.537
0.113
þ0.329
þ0.775
0.694
0.497
0.104
þ0.286
65
64
56
24
37
57
57
57
55
56
37
37
37
23
37
37
37
37
37
37
37
37
16
37
56
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
0.452
0.254
0.473
0.052
0.364
0.040
0.230
0.018
0.484
0.501
þ0.292
0.122
0.314
þ0.048
0.176
þ0.015
þ0.249
0.095
þ0.379
þ0.109
0.774
0.269
þ0.144
þ0.108
0.311
þ0.039
þ0.249
0.163
þ0.442
0.076
þ0.143
0.405
0.089
þ0.006
0.363
0.155
0.015
þ0.277
þ0.507
0.212
þ0.210
þ0.099
þ0.001
þ0.237
0.179
0.108
þ0.105
0.237
0.243
þ0.157
0.621
0.504
0.518
0.479
0.321
0.155
0.058
þ0.027
0.641
0.446
þ0.383
0.141
0.437
0.290
0.083
0.100
þ0.231
0.132
þ0.269
þ0.131
0.747
0.313
þ0.347
þ0.258
0.512
0.283
0.410
þ0.118
0.205
þ0.054
0.119
þ0.375
þ0.078
0.167
þ0.191
þ0.290
0.184
0.218
0.442
þ0.146
0.196
0.021
0.013
0.315
þ0.260
þ0.030
0.034
þ0.155
þ0.316
0.276
0.515
0.596
0.214
0.412
0.221
0.235
þ0.429
þ0.234
0.401
0.119
þ0.345
0.131
0.386
0.391
þ0.155
0.115
0.114
0.118
0.072
þ0.294
0.386
0.299
þ0.336
þ0.228
0.487
0.138
0.360
þ0.016
0.475
0.233
0.161
þ0.515
þ0.309
0.041
þ0.223
þ0.303
0.097
0.363
0.385
þ0.442
0.345
0.315
þ0.009
0.262
þ0.272
0.050
0.340
þ0.290
þ0.137
0.205
þ0.241
þ0.108
þ0.254
þ0.249
þ0.253
þ0.231
þ0.095
þ0.175
þ0.425
þ0.423
0.498
þ0.132
þ0.637
þ0.259
0.164
þ0.087
þ0.184
þ0.042
0.128
0.144
þ0.545
þ0.424
0.172
0.338
þ0.375
0.477
0.287
0.472
0.112
0.382
0.071
0.207
0.029
0.497
0.536
þ0.326
0.125
0.365
þ0.034
0.096
0.012
þ0.152
0.070
þ0.356
þ0.027
0.759
0.341
þ0.173
þ0.105
0.353
0.542
0.400
0.439
þ0.115
0.279
0.118
0.094
þ0.103
0.556
0.628
þ0.206
0.145
0.139
þ0.011
0.216
0.006
þ0.343
0.147
þ0.288
þ0.082
0.688
0.135
þ0.056
þ0.118
0.337
56
37
60
37
75
13
31
37
56
37
56
56
56
56
37
37
37
37
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
þ0.093
0.135
þ0.115
þ0.222
þ0.130
þ0.023
0.140
0.208
þ0.266
þ0.592
þ0.237
0.388
0.017
0.151
0.045
0.098
0.357
þ0.305
þ0.141
þ0.210
þ0.502
þ0.060
þ0.240
0.244
þ0.144
0.305
þ0.447
0.222
þ0.359
0.369
þ0.043
0.058
þ0.287
þ0.274
þ0.403
þ0.063
þ0.038
0.254
0.179
þ0.180
þ0.010
þ0.373
0.238
0.048
þ0.049
þ0.740
þ0.103
0.296
0.050
0.146
0.202
0.271
0.603
þ0.266
0.149
0.192
0.554
0.038
0.294
þ0.363
0.079
þ0.294
0.533
þ0.252
0.474
þ0.415
0.055
þ0.081
0.206
0.150
0.576
0.134
0.145
0.207
0.576
0.107
0.176
þ0.056
0.185
þ0.143
0.480
þ0.569
0.330
þ0.093
0.159
0.088
0.275
0.416
0.600
þ0.215
0.206
0.100
0.507
0.183
0.158
þ0.057
0.349
þ0.186
0.493
þ0.252
0.481
þ0.303
0.259
þ0.045
0.207
0.283
0.372
þ0.269
þ0.458
þ0.215
þ0.037
0.096
0.124
0.323
þ0.448
0.200
þ0.088
0.402
0.008
þ0.010
0.045
þ0.085
þ0.187
þ0.257
þ0.384
0.352
þ0.027
0.130
þ0.078
þ0.219
þ0.129
þ0.008
0.260
0.181
þ0.221
þ0.579
þ0.191
0.356
0.043
0.162
0.027
0.096
0.346
þ0.328
þ0.137
0.086
þ0.081
þ0.027
þ0.086
þ0.072
þ0.050
0.145
þ0.127
þ0.485
þ0.221
0.404
.000
0.255
0.081
0.186
0.379
þ0.192
56
2
þ0.232
þ0.362
þ0.049
0.514
0.362
0.201
þ0.078
þ0.212
þ0.167
American Journal of Physical Anthropology—DOI 10.1002/ajpa
88
I. GILLIGAN AND D. BULBECK
TABLE A1. (Continued)
Birdsell variable
n
s
MT
ST
WT
H
R
C
W
WC
AT
Chin slope (rearward)
Brow ridge size
Brow ridge form–continuous
Forehead slope (retreating)
Minimum frontal diameter
(height)
Sagittal elevation (parietal)
Sagittal elevation (frontal)
Parietal bosses
Sagittal fossa
Cryptose iris
Contraction iris furrows
Interrupted iris furrows
Body metrics
Weight
Stature
Thumb extension
Shoulder breadth
Sitting height
Limb metrics
Humeral length
Radial length
Femoral length
Tibial length
Calf girth
Body indices
Ponderal index
Relative shoulder breadth
Relative sitting height
Limb indices
Radial-humeral index
Tibial-femoral index
Intermembral index
Calf-tibial index
Limb-body indices
Humeral–sitting height
Radial–sitting height
Femoral–sitting height
Tibial–sitting height
Head metrics
Head length
Head breadth (maximum)
Head breadth (basal)
Head height
Frontal diameter (minimum)
Glabellar protrusion
Nasion depression
Bizygomatic diameter
Bigonial diameter
Facial metrics
Facial height (total)
Facial height (upper)
Nasal height
Nasal breadth
Nasal depth
Mandibulardepth
Lip thickness
Mouth breadth
Head indices
Cranial module
Cephalic index
Basal breadth index
Height-length index
Height-breadth index
Cephalo-facial index
Fronto-parietal index
Fronto-gonial index
Zygo-frontal index
Zygo-gonial index
37
59
37
60
37
2
2
2
2
2
0.327
þ0.247
þ0.035
0.138
0.023
þ0.078
þ0.075
0.271
0.051
0.318
0.388
þ0.277
þ0.206
0.147
þ0.153
0.111
0.152
þ0.266
þ0.081
þ0.374
0.195
0.029
þ0.508
0.114
þ0.425
þ0.140
0.086
þ0.307
0.094
þ0.542
0.040
0.174
0.225
0.139
0.095
0.251
þ0.226
þ0.053
0.137
þ0.022
0.303
þ0.25
þ0.009
0.147
þ0.144
37
37
40
37
56
37
37
2
2
2
2
2
2
2
0.388
þ0.105
0.289
þ0.410
þ0.395
0.343
þ0.342
þ0.469
þ0.103
0.658
0.003
þ0.408
0.454
0.310
0.633
þ0.053
þ0.007
þ0.435
þ0.262
0.085
þ0.493
0.437
0.206
þ0.614
0.012
0.430
þ0.489
þ0.291
0.554
0.069
þ0.630
þ0.434
0.195
þ0.219
þ0.362
0.527
0.029
þ0.692
þ0.059
0.386
þ0.301
þ0.153
þ0.440
0.051
þ0.060
0.295
0.302
þ0.284
0.458
0.398
þ0.123
0.261
þ0.400
þ0.379
0.347
þ0.377
0.332
þ0.186
0.278
þ0.242
þ0.349
0.086
þ0.243
87
93
36
74
56
1
1
2
1
1
þ0.116
þ0.542
þ0.589
þ0.062
þ0.585
þ0.262
þ0.360
0.096
þ0.145
þ0.335
0.010
þ0.467
þ0.658
0.020
þ0.548
0.278
0.361
þ0.166
0.162
0.319
0.328
0.070
þ0.369
0.225
0.041
0.342
0.258
þ0.149
0.189
0.318
0.254
0.366
0.534
0.140
0.389
þ0.107
þ0.546
þ0.584
þ0.050
þ0.564
0.005
þ0.444
þ0.498
0.072
þ0.504
57
57
57
57
57
1
1
1
1
1
þ0.429
þ0.577
þ0.571
þ0.664
þ0.047
þ0.331
þ0.506
þ0.405
þ0.513
0.129
þ0.331
þ0.423
þ0.471
þ0.534
þ0.151
0.378
0.490
0.447
0.520
þ0.149
0.208
0.238
0.169
0.213
þ0.153
0.323
0.471
0.354
0.440
0.032
0.321
0.387
0.445
0.424
0.227
þ0.412
þ0.559
þ0.570
þ0.657
þ0.037
þ0.317
þ0.488
þ0.402
þ0.561
þ0.007
86
75
56
1
1
1
þ0.464
0.568
0.272
þ0.202
0.229
0.377
þ0.514
0.529
0.103
0.149
þ0.150
þ0.456
þ0.204
0.127
þ0.354
0.031
þ0.151
þ0.310
0.126
þ0.284
þ0.139
þ0.470
0.569
0.287
þ0.493
0.540
0.133
56
56
56
56
1
1
1
1
þ0.251
þ0.521
0.559
0.505
þ0.341
þ0.506
0.333
0.559
þ0.131
þ0.373
0.515
0.286
0.250
0.417
þ0.355
þ0.590
0.101
0.236
þ0.037
þ0.379
0.303
0.432
þ0.218
þ0.350
0.032
0.183
þ0.418
þ0.148
þ0.248
þ0.506
0.585
0.509
þ0.291
þ0.558
0.405
0.426
56
56
56
56
2
2
2
2
0.072
þ0.247
þ0.229
þ0.473
þ0.132
þ0.441
þ0.262
þ0.530
0.206
þ0.034
þ0.105
þ0.269
0.237
0.494
0.369
0.565
0.359
0.338
0.217
0.355
0.136
0.452
0.177
0.421
0.038
0.136
0.251
0.284
0.069
þ0.239
þ0.256
þ0.484
0.174
þ0.195
þ0.052
þ0.361
99
102
59
85
77
63
62
98
72
1
1
1
1
2
2
2
2
2
0.328
0.390
þ0.052
0.160
þ0.390
þ0.403
þ0.368
þ0.227
þ0.236
þ0.036
0.368
þ0.281
0.248
þ0.410
þ0.100
þ0.523
þ0.258
þ0.273
0.494
0.322
0.131
0.081
þ0.256
þ0.477
þ0.194
þ0.142
þ0.116
0.297
þ0.245
0.449
þ0.150
0.372
0.135
0.461
0.338
þ0.290
0.565
0.068
0.522
þ0.130
0.241
þ0.084
0.257
0.302
0.199
0.087
þ0.199
0.432
þ0.185
0.448
0.130
0.508
0.308
0.280
þ0.075
þ0.008
0.199
0.073
0.044
0.314
0.127
0.215
0.073
0.324
0.382
þ0.043
0.148
þ0.358
þ0.382
þ0.361
þ0.229
þ0.197
0.428
0.417
0.083
0.106
þ0.350
þ0.368
þ0.351
þ0.187
þ0.249
93
83
59
75
56
56
56
56
2
2
2
2
2
2
2
2
0.226
0.009
þ0.126
þ0.468
0.236
þ0.213
0.119
þ0.356
0.315
0.080
0.063
þ0.403
0.152
þ0.057
0.347
þ0.554
0.121
þ0.015
þ0.178
þ0.311
0.222
þ0.264
þ0.055
þ0.099
þ0.316
0.003
þ0.017
0.413
þ0.131
0.124
þ0.346
0.637
þ0.059
0.030
þ0.011
0.219
0.095
þ0.136
þ0.325
0.425
þ0.204
þ0.056
þ0.015
0.381
0.074
0.325
þ0.257
0.394
0.168
0.238
0.240
0.135
0.047
þ0.023
þ0.010
0.088
0.216
0.003
þ0.124
þ0.426
0.246
þ0.170
0.117
þ0.333
0.277
0.094
þ0.046
þ0.401
0.267
þ0.194
0.096
þ0.343
93
109
59
85
87
104
56
56
59
56
1
1
2
2
1
2
2
2
2
2
0.344
0.318
þ0.295
þ0.054
0.352
þ0.587
þ0.603
þ0.080
þ0.130
þ0.084
0.190
0.388
þ0.530
0.239
0.378
þ0.522
þ0.636
þ0.076
þ0.099
þ0.107
0.390
0.216
þ0.048
þ0.235
0.262
þ0.510
þ0.443
þ0.046
þ0.141
þ0.040
0.078
þ0.293
0.684
þ0.316
þ0.360
0.351
0.543
0.055
0.012
0.079
0.343
þ0.087
0.483
þ0.496
þ0.164
þ0.024
0.144
0.130
þ0.167
0.129
þ0.050
þ0.213
0.459
þ0.198
þ0.430
0.343
0.423
0.113
0.085
0.303
0.039
þ0.079
0.039
0.095
þ0.116
0.114
0.089
0.155
0.005
0.048
0.334
0.315
þ0.273
þ0.062
0.345
þ0.569
þ0.586
þ0.076
þ0.136
þ0.056
0.386
0.313
þ0.154
þ0.147
0.299
þ0.622
þ0.616
þ0.076
þ0.168
þ0.111
American Journal of Physical Anthropology—DOI 10.1002/ajpa
89
ENVIRONMENT AND MORPHOLOGY IN ABORIGINES
TABLE A1. (Continued)
Birdsell variable
Facial indices
Facial index (total)
Facial index (upper)
Nasal breadth index
Nasal depth index
Lip index
Dentition
Dental wear
Size of dentition
Palate breadth
Palate shape (square)
Bi-canine breadth
Bi-incisor breadth
M1 (lower) breadth
M2 (lower) breadth
M3 (lower) breadth
M2 > M1 (lower)
M2 > M1 (upper)
M1 (lower) – four cusps
M2 (lower) – four cusps
M3 (lower) – four cusps
Lower molarcusps (average)
M1 (upper) – four cusps
M2 (upper) – four cusps
M3 (upper) – four cusps
Upper molarcusps (average)
M1 (lower) fissures
M2 (lower) fissures
M3 (lower) fissures
Lower molar fissures
Crowned M3 (lower)
Irregular M3 (lower)
M3 (lower) impacted
Supernumerary molars
Sixth cusp (lower molars)
Seventh cusp (lower molars)
Bolk’s cusp
Campbell’s cusp
Carabelli’s features
Carabelli’s cusp
Carabelli’s fissures
Carabelli’s pits
Musgrave cusplet
Double Musgrave cusplet
Incisor gap (upper)
Incisor winging
Reduced lateral incisors (upper)
Shovel-shaped incisors
Basal tubercles
Premolar displacement
Premolar rotation
Premolar size disharmony
Multi-cusped lower premolars
Anterior open bite
Tooth displacement
D5 þ P6/D5
Serology
A1 allele
B group
O allele
N gene
Rhesus R1
Rhesus R2
Rhesus R0
Rhesus Rz
Lewis (aþ)
P1þ variant
Composites
Non-metric complex A
Non-metric complex B
n
s
MT
ST
WT
H
R
92
85
56
56
56
2
2
2
2
2
0.360
0.195
þ0.366
0.591
0.166
0.471
0.259
þ0.479
0.445
0.392
0.204
0.122
þ0.197
0.513
þ0.024
þ0.500
þ0.211
0.454
þ0.428
þ0.402
þ0.230
þ0.078
0.225
þ0.050
þ0.342
37
55
60
37
37
37
41
41
41
37
37
59
59
60
37
37
37
37
37
59
59
59
37
37
37
59
49
37
37
37
42
37
37
37
37
37
37
37
37
37
61
57
37
37
37
37
37
37
28
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
0.215
þ0.372
0.527
þ0.334
þ0.630
þ0.611
0.043
0.083
þ0.060
0.575
þ0.185
0.373
0.260
0.186
þ0.381
0.521
0.661
0.464
0.532
0.380
0.241
þ0.307
þ0.087
þ0.193
þ0.043
þ0.162
þ0.011
þ0.104
þ0.212
þ0.030
þ0.367
0.090
0.548
þ0.463
þ0.226
þ0.429
þ0.308
þ0.229
þ0.432
þ0.290
þ0.202
0.408
þ0.270
þ0.280
þ0.195
0.739
þ0.217
þ0.448
0.228
0.253
þ0.521
0.473
þ0.040
þ0.042
0.134
þ0.180
þ0.119
þ0.205
0.068
0.101
0.292
0.213
0.317
0.061
þ0.359
þ0.103
þ0.128
þ0.240
0.440
0.320
þ0.291
0.077
0.143
þ0.325
þ0.095
þ0.179
0.063
þ0.041
0.390
þ0.097
0.362
0.174
0.194
0.094
0.223
þ0.455
þ0.172
0.088
þ0.111
þ0.113
0.215
0.022
0.060
0.190
0.169
þ0.277
þ0.065
0.068
0.069
þ0.139
0.484
þ0.305
þ0.641
þ0.718
0.166
0.185
0.056
0.540
þ0.254
0.312
0.230
0.015
þ0.452
0.724
0.717
0.512
0.657
0.252
0.135
þ0.232
þ0.142
þ0.291
0.135
þ0.157
0.130
þ0.155
þ0.224
þ0.237
þ0.394
þ0.083
0.452
þ0.551
þ0.250
þ0.529
þ0.073
þ0.142
þ0.474
þ0.247
þ0.178
0.389
þ0.289
þ0.307
þ0.308
0.655
þ0.356
þ0.408
0.232
þ0.035
0.621
þ0.344
0.125
þ0.179
þ0.325
0.138
0.131
0.200
þ0.079
þ0.100
þ0.442
þ0.244
þ0.326
þ0.166
0.371
0.113
0.119
0.208
þ0.374
þ0.388
0.282
þ0.146
þ0.129
0.326
0.096
0.335
þ0.043
0.040
þ0.473
0.088
þ0.376
þ0.243
þ0.155
þ0.013
þ0.226
0.420
0.063
þ0.173
þ0.029
0.096
þ0.195
þ0.043
0.100
þ0.157
0.019
þ0.136
þ0.064
þ0.093
124
124
124
102
76
74
75
76
55
53
2
2
2
2
2
2
2
2
2
2
0.232
þ0.028
þ0.243
0.304
þ0.232
0.314
þ0.066
þ0.114
þ0.244
0.026
þ0.321
0.204
0.371
0.076
0.324
þ0.432
0.286
þ0.216
0.342
þ0.103
0.468
þ0.163
þ0.524
0.281
þ0.452
0.609
þ0.234
0.006
þ0.499
0.074
34
34
2
2
þ0.145
0.210
þ0.253
þ0.373
þ0.009
0.418
C
W
WC
AT
þ0.355
þ0.230
0.417
þ0.337
þ0.264
0.035
þ0.097
þ0.105
þ0.036
0.036
0.345
0.209
þ0.325
0.559
0.153
0.349
0.206
þ0.432
0.650
0.200
þ0.139
0.505
0.092
þ0.238
þ0.484
þ0.596
0.266
0.321
0.202
0.283
0.028
þ0.223
þ0.087
þ0.262
þ0.436
0.681
0.425
0.359
0.518
þ0.077
þ0.179
0.143
þ0.125
þ0.140
0.297
0.018
0.360
þ0.207
þ0.147
þ0.428
þ0.164
þ0.295
0.056
þ0.343
þ0.061
þ0.284
0.338
0.037
þ0.384
þ0.021
0.002
0.107
þ0.163
þ0.176
þ0.300
0.234
þ0.310
þ0.288
0.239
þ0.065
445
þ0.280
þ0.129
þ0.194
þ0.216
0.109
0.174
0.139
0.137
0.078
þ0.277
þ0.324
þ0.297
0.083
0.579
0.371
0.473
0.518
þ0.269
þ0.285
0.265
þ0.176
0.068
0.269
0.227
0.180
0.140
0.182
þ0.490
0.115
þ0.434
þ0.108
þ0.348
þ0.118
þ0.393
0.425
0.102
þ0.157
0.221
0.088
þ0.253
0.057
þ0.151
þ0.176
0.247
þ0.257
þ0.108
0.033
þ0.137
0.063
þ0.118
0.313
0.405
0.451
0.185
0.202
0.031
þ0.553
þ0.144
þ0.091
þ0.138
þ0.153
0.188
þ0.573
þ0.459
þ0.500
þ0.567
0.074
0.149
0.292
0.280
0.167
þ0.294
0.157
þ0.108
0.203
þ0.017
0.245
0.237
0.253
þ0.404
0.598
0.543
0.509
þ0.086
0.243
0.379
0.211
0.160
þ0.337
0.192
0.153
0.208
þ0.399
0.294
0.399
þ0.046
0.202
þ0.331
0.512
þ0.356
þ0.617
þ0.591
0.020
0.079
þ0.040
0.632
þ0.126
0.330
0.231
0.162
þ0.341
0.562
0.659
0.482
0.567
0.363
0.200
þ0.302
þ0.127
þ0.195
0.006
þ0.161
þ0.002
þ0.134
þ0.187
þ0.083
þ0.364
0.017
0.560
þ0.527
þ0.287
þ0.464
þ0.259
þ0.218
þ0.440
þ0.272
þ0.193
0.412
þ0.256
þ0.278
þ0.214
0.729
þ0.207
þ0.444
0.206
0.102
þ0.212
0.504
þ0.135
þ0.761
þ0.642
þ0.078
þ0.059
þ0.027
0.385
þ0.224
0.325
0.273
0.083
þ0.390
0.427
0.592
0.427
0.449
0.366
0.280
þ0.343
þ0.189
þ0.199
þ0.045
þ0.138
0.262
þ0.049
þ0.269
þ0.150
þ0.239
0.013
0.165
þ0.184
þ0.124
þ0.341
þ0.195
þ0.235
þ0.308
þ0.125
þ0.187
0.324
þ0.335
þ0.204
þ0.127
0.673
þ0.051
þ0.502
0.313
0.398
þ0.253
þ0.434
þ0.027
þ0.316
0.395
þ0.237
0.194
þ0.336
0.136
0.455
þ0.205
þ0.528
þ0.018
þ0.467
0.589
þ0.285
0.182
þ0.543
0.067
0.357
þ0.096
þ0.399
þ0.086
þ0.258
0.322
þ0.009
0.034
þ0.318
0.070
0.008
0.020
þ0.036
þ0.241
þ0.029
þ0.082
þ0.003
0.162
0.116
0.175
0.228
þ0.051
þ0.233
0.322
þ0.220
0.320
þ0.077
þ0.132
þ0.242
0.000
0.321
þ0.057
þ0.370
0.267
0.181
0.298
þ0.141
þ0.110
þ0.252
0.177
0.138
0.181
0.315
0.368
0.343
0.289
þ0.138
þ0.127
þ0.100
0.210
þ0.338
0.097
American Journal of Physical Anthropology—DOI 10.1002/ajpa
90
I. GILLIGAN AND D. BULBECK
TABLE A1. (Continued)
Birdsell variable
n
s
MT
ST
WT
H
R
C
W
WC
AT
Non-metric A þ B
Metric – body
Metric – cranial
Metric – facial
Metric – total
Dental complex A
Dental complex B
Dental A þ B
Serology composite
Non-metric, metric,
dental þ serological
34
34
34
34
34
34
34
34
34
34
2
2
2
2
2
2
2
2
2
2
0.030
þ0.184
þ0.480
0.244
þ0.199
þ0.416
0.304
0.150
þ0.042
þ0.088
þ0.411
0.301
0.183
0.036
0.254
0.458
þ0.109
0.245
þ0.378
þ0.294
0.257
þ0.363
þ0.621
0.247
þ0.353
þ0.684
0.391
0.022
0.162
0.071
0.206
þ0.299
þ0.259
þ0.029
þ0.280
þ0.500
0.102
þ0.314
0.236
0.097
0.431
þ0.533
þ0.495
0.080
þ0.460
þ0.567
0.324
þ0.083
0.242
0.142
0.409
þ0.364
þ0.110
þ0.094
þ0.281
þ0.233
0.338
0.139
0.134
0.145
þ0.132
0.482
0.277
þ0.048
0.354
0.403
þ0.183
0.089
0.035
0.180
0.055
þ0.224
þ0.438
0.222
þ0.212
þ0.419
0.301
0.129
þ0.048
þ0.089
þ0.139
þ0.145
þ0.536
0.176
þ0.228
þ0.321
0.380
0.236
þ0.141
þ0.220
Appendix B
TABLE B1. Statistical significance of Pearson r correlations between the frequency of the A1
blood group and other variables in Birdsell (1993)
Trait category
Simple genetic (5)
Sexual dimorphism—
males only (8)
Multiple genetic (29)
Complex
developmental
(73)
Significant at P < 0.05
Significant at P < 0.01
N, R1,
Lewis aþ (3)
Body hair,
nasion
depression (2)
4-Cusped M1,
4-cusped M2,
M2 fissure,
multicusped
lower premolars
(4)
Tawny hair (1)
4-Cusped M3,
anterior open
bite, oral
pigmentation,
iris contraction
furrows, ear
protrusion, ear
axis rearward
orientation,
ear helix roll,
total prognathism,
thumb extension,
upper/lower
head index,
total facial
index (11)
Parietal
sagittal
elevation (1)
Carabelli’s
cusplets,
Musgrave
cusplets,
circumancular
hair (3)
Evironment/
environmental
adaptation (4)
Body size and shape
(20)
American Journal of Physical Anthropology—DOI 10.1002/ajpa
Nonsignificant
P1þ (1)
Beards, nasal tip terminal hair, ear tragus
terminal hair, baldness, finger hair,
glabellar protrusion (6)
Dental size, M1 breadth, M2 breadth, M2
breadth > M1 breadth, M2 breadth > M1
breadth, 4-cusped M1, 4-cusped M2, M1
fissure, lower molars’ cusp 6, lower
molars’ cusp 7, Bolk’s paramolar cusp,
Campbell’s paramolar cusp, double
Musgrave’s cusplets, I winging, I
shoveling, I tubercles, curly hair, double
occipital hair whorl, frontal hair
Type 1, widow’s peak, left-handedness,
ear lobe attachment (22)
Palate breadth, palate shape, bi-canine breadth,
bi-incisor breadth, M3 breadth, 4-cusped M3,
M3 fissure, crowned M3, irregular M3, impacted
M3, super-numerary molars, I gap, reduced I2,
premolar lateral displacement, premolar lateral
rotation, premolar size disharmony, molar/
premolar eruption, cryptose iris structure, gray
hair median age, finger formula, toe formula,
nasal profile, ear pit, ear lobe size, mid-facial
prognathism, alveolar prognathism, bilateral
chins, rearward chin slope, brow ridge size, brow
ridge type, forehead slope, minimum frontal
diameter height, frontal sagittal elevation,
parietal boss size, inter-parietal sagittal fossa,
head length, head breadth, basal head breadth,
head height, minimum frontal diameter,
bizygomatic diameter, bigonial diameter, facial
height, upper facial height, nasal height, nasal
breadth, nasal depth, mandibular depth,
lip thickness, mouth breadth, cranial module,
cephalic index, head height/length index, head
height/breadth index, cephalo-facial index,
fronto-parietal index, fronto-gonial index,
zygo-frontal index, zygo-gonial index, nasal
breadth/height index, nasal depth/breadth
index (61)
Iris color, skin color, tooth wear, tooth displacement (4)
Weight, stature, shoulder breadth, sitting height,
humeral length, radial length, femoral length, tibia
length, maximum calf girth, ponderal index, relative
shoulder breadth, relative sitting height,
radial-humeral index, tibia-femoral index,
intermembral index, humeral-sitting height
index, radial-sitting height index, femoral-sitting
height index, tibia-sitting height index, calf-tibia
index (20)
ENVIRONMENT AND MORPHOLOGY IN ABORIGINES
LITERATURE CITED
Abbie AA. 1975. Studies in physical anthropology, Vol. II.
Canberra: Australian Institute of Aboriginal Studies.
Abbie AA. 1976. Morphological variation in the adult Australian
Aboriginal. In: Kirk RL, Thorne AG, editors. The origin of the
Australians. Canberra: Australian Institute of Aboriginal
Studies. p 211–214.
Allen JA. 1877. The influence of physical conditions in the genesis of species. Rad Rev 1:108–140.
Ashton KJ, Tracy MC, de Queiroz A. 2000. Is Bergmann’s rule
valid for mammals? Am Nat 156:390–415.
Bergmann C. 1847. Uber die verhältniesse der wärmeokonomie
der thiere zu ihrer grösse. Gottingen Stud 1:595–708.
Birdsell JB. 1953. Some environmental and cultural factors
influencing the structuring of Australian Aboriginal populations. Am Nat 87:171–207.
Birdsell JB. 1967. Preliminary data on the trihybrid origin of
the Australian Aborigines. Archaeol Phys Anthrop Oceania
2:100–155.
Birdsell JB. 1979. Ecological influences on Australian Aboriginal
social organization. In: Bernstein IS, Smith EO, editors.
Primate ecology: ecological influences on social organization.
New York: Garland STPM. p 117–151.
Birdsell JB. 1993. Microevolutionary patterns in Aboriginal
Australia: a gradient analysis of clines. New York: Oxford
University Press.
Brown P. 1987. Pleistocene homogeneity and Holocene size
reduction: the Australian human skeletal evidence. Archaeol
Oceania 22:41–67.
Bureau of Meteorology, Commonwealth of Australia. 2006. Climate averages map. http://www.bom.gov.au/climate/map/climate_avgs/clim_avg1.shtml.
Carey JW, Steegmann AT. 1981. Human nasal protrusion,
latitude, and climate. Am J Phys Anthropol 56:313–319.
Cavalli-Sforza LL. 1986. African Pygmies: an evaluation of the
state of research. In: Cavalli-Sforza LL, editor. African Pygmies.
Orlando: Academic Press. p 361–426.
Clendon M. 2006. Reassessing Australia’s linguistic prehistory.
Curr Anthropol 47:39–61.
Geist V. 1987. Bergmann’s rule is invalid. Can J Zool 65:1035–
1038.
Gloger CWL. 1833. Das abandern der vogel durch einfluss des
klimas. Breslau: August Schulz.
Habgood PJ. 1986. The origin of the Australians: a multivariate
approach. Archaeol Oceania 21:130–137.
Harrison GA. 1975. Pigmentation. In: Roberts DF, editor.
Human variation and natural selection. London: Taylor &
Francis. p 179–197.
Hiernaux J, Rudan P, Brambati A. 1975. Climate and the
weight/height relationship in sub-Saharan Africa. Ann Hum
Biol 2:3–12.
Hillson S. 1996. Dental anthropology. Cambridge: Cambridge
University Press.
Holliday TW, Falsetti AB. 1995. Lower limb length of European
early modern humans in relation to mobility and climate.
J Hum Evol 29:141–153.
Houghton P. 1990. The adaptive significance of Polynesian body
form. Ann Hum Biol 17:19–32.
Jablonski NG, Chaplin G. 2000. The evolution of human skin
coloration. J Hum Evol 39:57–106.
Jurmain R, Nelson H. 1994. Introduction to physical anthropology, 6th ed. Minneapolis: West.
Katzmarzyk PT, Leonard WR. 1998. Climatic influences on
human body size and proportions: ecological adaptations and
secular trends. Am J Phys Anthropol 106:483–503.
Keen I. 2004. Aboriginal economy and society: Australia at
the threshold of colonisation. Melbourne: Oxford University
Press.
Lahr MM, Wright RVS. 1996. The question of robusticity and
the relationship between cranial size and shape in Homo
sapiens. J Hum Evol 31:157–191.
Lattin JM, Carroll JD, Green PE. 2003. Analyzing multivariate
data. Pacific Grove: Brooks/Cole.
91
Macho G, Freedman L. 1987. A re-analysis of the Andrew A
Abbie morphometric data on Australian Aborigines. Canberra:
Australian Institute of Aboriginal Studies.
Mayr E. 1956. Geographical character gradients and climatic
adaptation. Evolution 10:105–108.
Meiri S, Dayan T. 2003. On the validity of Bergmann’s rule.
J Biogeogr 30:331–351.
O’Connell JF, Allen J. 2004. Dating the colonization of Sahul
(Pleistocene Australia-New Guinea): a review of recent
research. J Archaeol Sci 31:835–853.
Odling-Smee FJ, Laland KN, Feldman MW. 2003. Niche
construction: the neglected process in evolution. Princeton,
NJ: Princeton University Press.
Pardoe C. 1994. Isolation and evolution in Tasmania. Curr
Anthropol 35:1–21.
Pardoe C. 2006. Becoming Australian: evolutionary processes
that structure biological variation from origin to modern
times. Before Farming 1 (article 4).
Peterson N. 1976. The natural and cultural areas of Aboriginal
Australia: a preliminary analysis of population groupings
with adaptive significance. In: Peterson N, editor. Tribes and
boundaries in Australia. Canberra: Australian Institute of
Aboriginal Studies. p 50–71.
Peterson N. 1986. Australian territorial organization. Oceania
Monograph 30. Sydney: University of Sydney Press.
Pietrusewsky M. 1984. Metric and non-metric cranial variation
in Australian Aboriginal populations compared with populations from the Pacific and Asia. Canberra: Australian Institute of Aboriginal Studies.
Roberts DF. 1978. Climate and human variability. Chicago:
Cummings.
Roth HL. 1899. The Aborigines of Tasmania, 2nd ed. Hobart:
Fullers Bookshop.
Ruff CB. 1994. Morphological adaptation to climate in modern
and fossil hominids. Yearbk Phys Anthropol 37:65–107.
Ruff CB. 2002. Variation in human body size and shape. Annu
Rev Anthropol 31:211–232.
Schreider E. 1975. Morphological variations and climatic differences. J Hum Evol 4:529–539.
Scott GR, Turner CG II. 1997. The anthropology of modern
human teeth. Cambridge: Cambridge University Press.
SPSS. 2001. SPSS graduate pack 11.0 for Windows. Chicago:
SPSS.
Steadman RG. 1984. A universal scale of apparent temperature.
J Appl Meteorol 23:1674–1687.
Steadman RG. 1994. Norms of apparent temperature in
Australia. Aust Met Mag 43:1–15.
Stevens JP. 2002. Applied multivariate statistics for the social
sciences, 4th ed. Mahwah, NJ: Lawrence Erlbaum.
Strauss DJ, Orans M. 1975. Mighty sifts: a critical appraisal of
solutions to Galton’s problem and a partial solution. Curr
Anthropol 16:573–594.
Tabachnick BG, Fidell LS. 2001. Using multivariate statistics,
4th ed. Boston: Allyn and Bacon.
Thorne AG. 1971. The racial affinities and origins of the
Australian Aborigines. In: Mulvaney DJ, Golson J, editors.
Aboriginal man and environment in Australia. Canberra:
Australian National University Press. p 316–325.
Tindale NB. 1974. Aboriginal tribes of Australia. Berkeley:
University of California Press.
Tindale NB, Birdsell JB. 1941. Results of the Harvard-Adelaide
Universities Anthropological Expedition, 1938–1939: Tasmanoid tribes in north Queensland. Rec South Aust Mus 7:1–9.
Trinkaus E. 1981. Neanderthal limb proportions and cold adaptation. In: Stringer CB, editor. Aspects of human evolution.
London: Taylor & Francis. p 187–223.
Wolpoff MH. 1968. Climatic influence on the skeletal nasal aperture. Am J Phys Anthropol 29:405–424.
Wright RVS. 1976. Evolutionary process and semantics: Australian prehistoric tooth size as a local adjustment. In: Kirk RL,
Thorne AG, editors. The origin of the Australians. Canberra:
Australian Institute of Aboriginal Studies. p 265–274.
American Journal of Physical Anthropology—DOI 10.1002/ajpa
Документ
Категория
Без категории
Просмотров
0
Размер файла
368 Кб
Теги
environment, morphology, aboriginal, analysis, database, australia, birdsell
1/--страниц
Пожаловаться на содержимое документа