close

Вход

Забыли?

вход по аккаунту

?

Components of racial variation in finger ridge-counts.

код для вставкиСкачать
AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 52139-144 (1980)
Components of Racial Variation in Finger Ridge-Counts
RICHARD L. JANTZ AND CLEONE H. HAWKINSON
Department of Anthropology, University of Tennessee, Knoxville, Tennessee
KEY WORDS Dermatoglyphics, Finger ridgecounts, Race
ABSTRACT
Principal components analysis was used to evaluate finger
ridge-count variability as a n indicator of genetic relationships between populations. The analysis was carried out on American White, American Black and
African Black samples, each including both sexes. Each individual is represented
as a vector of 20 counts, a radial and a n ulnar count for each digit. No assumptions
were made prior to analysis concerning the number of meaningful components, and
all were examined sequentially. The first five eigenvectors extracted from the
within-groups correlation matrix have loadings very similar to those previously
described by Roberts and Coope ('75). However, it is the component scores derived
from the sixth eigenvector which show the most marked variation, accounting for
45% or more of the D' in all Black-White comparisons. A number of other components also show significant intergroup heterogeneity, but they often do not accord
with what is known of the genetic relationships between the populations. Apparently a large amount of ridge-count variation is not genetically meaningful, a t
least as far as these populations are concerned.
Dermatoglyphics have been used in population studies in one way or another since it was
recognized over 70 years ago that they show
racial and population variation. No consensus
has developed as to what analytical approaches
might prove most useful in population studies,
and the problem has become progressively
worse as anthropologists ask more sophisticated questions of their data. An early approach, which continues to have adherents, is
simply to tabulate frequencies of arches, loops
and whorls. As a result of Holt's ('56 ; '57) genetic studies of quantitative traits, total ridgecount seemed useful with its high heritibility
estimates and statistical qualities. As is now
well known, TRC is the sum of the larger counts
for each digit, a procedure which greatly simplifies matters, reducing twenty variables to
one. It may also, however, obscure a great deal
of information relevant to genetical studies of
human populations.
Holt's parent-child ('56) and sib-sib ('57) correlations are in almost perfect agreement with
theoretical expectations under the assumption
of additivity, suggesting a heritibility of nearly
lOWo. Since heritibility estimates are population specific, we might expect them to vary
somewhat. Loesch's ('71) Polish data and Froelich's ('76) southwest Pacific data both yield
0002-948318015201-0139 $01.40 0 1980 ALAN R. LISS, INC
heritibility estimates for TRC of around 70%.
These are substantially below Holt's estimates,
but even so indicate that the majority of variation is genetic. Unfortunately little is known
about heritibility of components of ridge-count
variability apart from TRC. There may be some
aspects of ridge-count variability which are
primarily environmental.
It is apparent that dermatoglyphics, when
incorporated into population studies along with
other types of data, often produce results difficult to interpret. The most extensive analysis
of this type is that of Friedlaender ('75) on
Bougainville. He calculated intervillage distances based on anthropometric, serological,
odontometric, kinship, geographical, and
dermatoglyphic data. Good agreement was
found between anthropometric and serological
data, which in turn agreed well with geography, anthropometry being the better of the
two. Dermatoglyphics had low agreement with
all three, producing a unique population arrangement which was not easily interpretable
in genetic terms. Moreover, this poor agreement between dermatoglyphics and other
biological data is the rule rather than the exception. Nee1 et al. ('74) found poor agreement
between dermatoglyphics and serology or
anthropometry, Chai ('72) between derma-
139
140
RICHARD L. JANTZ AND CLEONE H. HAWKINSON
toglyphics and anthropometry, and Jantz and
Chopra ('76) between dermatoglyphics and
anthropometry or serology. Jantz and Chopra
('76) found further that population relationships based on dermatoglyphics alone change,
depending upon the method used. Relationships based on ridge-counts vary with those
based on pattern classification. The latter further varies with the type of system used.
There is clearly a need for research directed
toward the problem of how, or even whether,
dermatoglyphics provides useful information
in anthropological contexts. Steps in this direction were taken by Roberts and Coope ('751, who
carried out a principal components analysis on
the correlation matrix of the twenty finger
ridge-count variables. They identified five
components and advanced field theory as a n
explanatory model. Jantz and Owsley ('77) carried out a similar analysis, rotating five extracted principal components to attempt to
simplify the interpretation. Both studies obtained results unattainable by previous investigations. It turns out that radial and ulnar
sides of the digits are relatively independent,
indicating that ignoring orientation in choosing the larger count is not appropriate. Moreover, certain digits covary to the extent that
they may be considered meaningful entities.
If t h e components emerging from such
analyses have genetic meaning, they should
prove useful in population studies, and may
reveal unsuspected aspects of population variation. To date the components of variation as
elucidated by principal components analysis
have not been analyzed from the perspective of
interpopulation variation. That such a n approach may yield more meaningful and unsuspected information has been suggested by
Jantz and Hawkinson's ('79)analysis of African
ridge-counts. The object of this paper is to carry
out an analysis using component scores of three
samples to evaluate the genetic significance of
components extracted from a correlation matrix.
MATERIALS AND METHODS
The three samples employed in the analysis
are an American White sample (185males, 184
females),a n American Black sample (96 males,
119 females), and an African Black Yoruba
sample (120 males, 55 females). These are the
three samples previously used i n a factor
analytic study (Jantz and Owsley, '77) and the
Yoruba sample has recently been described
(Jantz and Brehme, '78). Standard procedures
were followed for ridge-counting (Holt, '68).
Each individual is represented as a vector of
twenty counts, a radial and an ulnar count for
each digit.
The 20 x 20 covariance matrix was formed
for each group. The matrices were pooled over
groups and sexes and converted to the withingroups correlation matrix, yielding a correlation matrix based on 741 degrees of freedom.
All 20 eigenvalues and eigenvectors were extracted using the SPSS subroutine FACTOR,
which also calculates a factor score coefficient
matrix. If A is the matrix of factor loadings, the
factor score matrix is F = (A'A)-'A. The principal component scores are then calculated as
Y = F Z where Z is the data matrix expressed
as standard scores. Principal components
scores obtained in this way are then standard
scores, having a grand mean of zero and a standard deviation of one. These scores have the
further advantage that they may be used to
'72) as:
obtain Mahalanobis'
- D'
- (Goodman,
- D' = (Y,-Y,) (Yl-YJ)'
where Y ,and YJare the mean vectors of component scores for groups i and j.
The pooled correlation matrix was used to
assure that all distances would have the same
meaning to facilitate comparison. There is
some evidence that the correlation matrices differ between Blacks and Whites (Jantz, '77;
Jantz and Owsley, '77) in that Blacks have
higher average correlations. Factor analysis
failed to reveal any systematic differences in
structure so we are probably justified in pooling
the matrices over groups.
We have computed component scores using
the unrotated factor loadings. It has been
shown that rotation often renders factors easier
to interpret, but it is also the case that the
loadings change with the number of factors included in the rotational solution. Since we wish
to m a k e no assumptions concerning t h e
number of meaningful components, we have
examined all of them sequentially.
RESULTS
It is not our purpose to attempt any thorough
examination of the component loadings from
the standpoint of structure. It may be useful to
look briefly a t the most important eigenvalues
and associated eigenvectors, which a r e
presented in Table 1. The eigenvalues behave
in a manner similar to that found previously
(Robertsand Coope, '75). The first eigenvalue is
large, accounting for almost half of the variance. Subsequent eigenvalues decrease i n
value very gradually, so that 12 are required to
account for 91% of the variation. The first five
141
RACIAL VARIATION IN RIDGE-COUNTS
TABLE I , First sir eigenuectors and eigenvalues of the finger ridge-count correlation matrix
Eigenvectors
Digits
1
2
3
4
5
6
0.74
0.44
0.80
0.71
0.81
0.68
0.69
0.64
0.63
0.56
0.72
0.46
0.76
0.75
0.76
0.69
0.63
0.67
0.61
0.58
-0.25
0.55
-0.22
0.39
-0.21
0.40
-0.14
0.11
-0.36
0.10
-0.25
0.55
-0.24
0.28
-0.24
0.34
-0.10
0.09
-0.42
0.07
-0.08
0.38
-0.07
0.14
0.13
-0.16
0.32
-0.43
0.13
-0.32
-0.13
0.40
-0.06
0.06
0.25
-0.25
0.44
-0.45
0.15
-0.33
-0.04
0.20
-0.19
-0.08
-0.18
-0.21
-0.07
-0.10
0.33
0.58
-0.04
0.21
-0.15
-0.10
-0.12
-0.23
-0.05
-0.16
0.31
0.58
-0.43
-0.25
-0.25
-0.01
0.07
0.22
0.20
0.15
0.31
-0.01
-0.45
-0.20
-0.28
0.07
0.11
0.21
0.17
0.15
0.25
-0.03
0.12
0.24
-0.03
-0.08
-0.10
0.07
-0.27
0.17
0.35
-0.33
0.12
0.17
-0.10
-0.10
-0.10
0.05
-0.29
0.15
0.37
-0.26
9.03
1.81
1.47
1.23
1.01
0.82
45.10
9.10
7.30
6.20
5.10
4.10
-~
L5
R
U
L4
R
L3
U
R
U
L!!
R
U
L1
R
IJ
R5
R4
R3
R
U
R
U
R
lJ
R2
R
U
Rl
R
U
Eigenvalues
Percent of
variation
components are very similar to those described
by Roberts and Coope ('75).The first component
is clearly size, all loadings being positive, reflecting the positive correlation between all
variables. The second component contrasts radial and ulnar sides of the digits, except for
digit 11. This suggests some sort of negative
interaction after removing the size effect. Component 3 contrasts the ulnar count on digit V
with the ulnar counts on digits I1 and I. In
addition, the radial and ulnar sides of digit I1
are contrasted. Component 4 is a thumb component, both radial and ulnar counts loading
positively but emphasizing the ulnar count
more than the radial. Component 5 takes the
form of an ulno-radial gradient, starting with
strong negative loadings on digit V and ending
in strong positive loadings on the radial count
of the thumb. The ulnar thumb count does not
participate in the gradient.
Roberts and Coope attempt to interpret only
five components, which generally correspond to
those with eigenvalues of one or greater. However, our component 6 is easily interpreted and
turns out to be very important. It contrasts
radial and ulnar counts on digits I and 11,and in
effect, the digits themselves since the direction
of the contrasts on the two digits is reversed.
Even later components are interpretable to
some extent. All components after 11 reflect
asymmetry, homologous variables having opposite loadings. While these collectively account for a very small fraction of the total variation, it is likely that even these contain some
biologically meaningful information. We will
return to some of these later.
The component scores for each sample for all
twenty components are presented in Table 2,
along with the F ratios derived from a one way
analysis of variance between groups. Several
points are noteworthy from these results. The
general finding is that there are several components which show significant heterogeneity,
the majority of which are in males. In males,
components 2,3,6,7,8,12, and 15show significant heterogeneity, while in females only 6,12,
and 17 are significantly heterogeneous. Only
components 6 and 12 show sex concordance for
statistical significance.
Component 12 is principally asymmetry of
the ulnar counts of digit V. Component 15 is a
rather complex reflection of the asymmetry of
radial ulnar contrasts on digit 111, and also involves the ulnar count of digit IV. Component
17 is a rather more straightforward whole hand
asymmetry centered on the radial count of digit
111. It is perhaps somewhat surprising that
asymmetry components show significant
interpopulation heterogeneity. However, the
concordance of sexes on component 12, and the
RICHARD L. JANTZ AND CLEONE H. HAWKINSON
142
TABLE 2. Principal component scores ofAmerican Whites (AWI, American Blacks (ABj und
Yoruba (Y), and F ratios for between-group heterogeneity
Males
Component
number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
F ratio
Females
AW
AB
Y
AW
AB
Y
d
0.058
-0.203
0.051
-0.003
-0.061
0.553
-0.191
-0.205
0.048
-0.080
-0.069
0.222
0.110
0.012
-0.171
0.067
0.025
0.006
-0.069
-0.169
0.033
0.146
0.201
0.066
0.111
-0.293
0.179
0.164
0.128
0.094
-0.003
0.003
-0.045
0.050
0.328
-0.012
0.017
0.014
-0.021
0.064
0.110
0.103
-0.148
0.008
-0.119
-0.319
-0.203
0.138
0.015
0.028
-0.006
-0.088
-0.036
-0.001
-0.002
0.012
-0.073
-0.081
0.138
0.045
-0.095
-0.062
-0.040
0.024
0.035
0.338
0.123
-0.048
-0.052
-0.108
0.006
0.081
0.014
0.051
0.140
-0.612
-0.188
0.234
-0.097
-0.309
-0.146
-0.457
0.354
0.076
0.048
0.142
0.100
-0.185
-0.144
-0,122
0.069
0.019
0.064
-0.164
0.169
0.064
0.17
10.5W
3.34*
0.17
1.51
20.9P*
11.51""
6.27"*
0.35
1.05
0.21
3.84*
1.13
0.08
7.87*'
0.23
0.38
0.34
1.61
2.47
-0.032
0.040
0.026
-0.001
0.024
-0.014
0.118
0.033
-0.036
0.019
0.003
0.085
-0.136
0.120
0.118
-0.236
-0.073
0.000
0.008
-0.095
-0.192
0.085
-0.036
0. I07
0
0.78
2.08
0.25
2.77
1.55
36.73*"
2.33
0.76
0.66
2.48
0.93
3.07"
0.75
0.34
0.19
0.33
3.584
1.20
0.98
0.28
**P- 005
-*P. 0 0 1
fact that both sexes show variation on a second
asymmetry factor, suggests that there may be
a n overlooked biological reality here.
The specific finding of greatest interest is
that component 6 shows remarkable variation
between groups. Its F ratio exceeds the next
highest ratio by a factor of two in males and ten
in females. Quite clearly, this factor is the
single most important source of interpopulation variation. It accounts for 45% or more of
the D2, shown in Table 3, in all Black-White
comparisons. In the case of the White-Yoruba
TABLE 3 . LY values for race and sex
American Whites
American Blacks
Yoruba
DISCUSSION
AW
AB
Y
0.36*
1.5P*
1.29%
1.26**
0.49
0.51
1.22*"
0.68
0.64
Sex differences within race are shown in the diagonal. Race differences
for males are In the lower left triangle, far females in the upper right
triangle.
D' values were tested for significance by the variance ratio
n , * n 2 ( n , + n,-m- 1 )
F A __ ___ D*
n , + n, ( n , + n,-2) m
where n , and n, are the sample sizes and m is the number of variables.
The F ratio has m and ( n , + n, - 1 - m) degrees of freedom.
*P<0.05.
* * P < 0.01.
.
female comparison, component 6 is responsible
for 7PkJof the D' between them!
The D2 values themselves seem to reflect a
general genetic reality. American Whites show
relatively large differences in relation to American Blacks and Yorubas. The latter two show
lower distances between each other. However,
in both sexes Yoruba are actually slightly
closer to American Whites than American
Blacks are, which would not accord with our
expectations based on what is known of White
gene flow into the American Black population.
The sex D' is low for all groups and statistically
significant in Whites only.
We now turn to the several implications of
these results for the anthropologic utilization of
finger ridge-count data. An interesting finding
is t h a t t h e most important component extracted from the correlation matrix shows no
statistically significant population variation.
The first component, being principally a size
factor, is nearly equivalent to the sum of the
twenty counts, a commonly used summary
character known as absolute ridge-count. This
same result has been found for total ridge-count
as well (Jantz, '74; Jantz and Hawkinson, '791,
so t h e conclusion t h a t summing counts
RACIAL VARIATION IN RIDGE-COUNTS
obscures a great deal of information now seems
well established.
A second finding, perhaps the most important of our findings, pertains to the extraordinary Black-White difference seen on component 6. This component has an eigenvalue of
only 0.82 and accounts for only 4.1% of the total
variation in within-group terms. Yet it is by far
the most important source of racial variation.
From a methodological point of view, this finding seems to offer strong support for the idea of
using twenty counts, rather than summing
counts to simplify matters. This component is
exclusively a radial-ulnar contrast so there is
no way the information could be obtained using
only the larger count from each digit. Some of
this has been alluded to in pattern type analysis, as in the lower frequency of radial loops on
digit I1 (Jantz, '74) and relatively higher frequency of arches on the thumb (Glanville and
Huizinga, '66) found in Black populations.
The developmental significance of component 6 also requires preliminary discussion a t
this time. Roberts and Coope ('75) applied field
theory as an explanatory model for their components, and Jantz and Owsley ('77) also felt
field theory to be appropriate. Thus component
6 may represent some sort of developmental
field affecting the radial digits only. This is
quite in line with the results from embryological data (Babler, '77; Okijima, ' 7 5 ) which
shows a radial-ulnar gradient in maturation of
the ridged skin system. But the most important
feature of this component would seem to be the
radial-ulnar contrasts drawn between sides of
digits I and 11. This type of contrast is quite
apparent elsewhere, as in components 2 and 3 ,
so it represents a common phenomenon of the
finger ridge system. One way to interpret it,
although at this stage it is purely speculative,
is that there is a racial difference in the nature
or strength of a growth gradient during
morphogenesis of the ridged skin system. The
relationship of pattern types to ridge maturation, as well as Black-White differences in ridge
maturation, have been documented (Babler,
'77). The form this maturational difference
might take is, of course, not clear. That it relates to differences in timing of pad regression
in relation to ridge development seems likely.
It is noteworthy here that the sum of the radial
and ulnar counts on each digit in Blacks and
Whites is very similar, so differences pertain
almost exclusively to radial-ulnar distribution.
Additional inferences might arise from our
expectations from admixture studies. To the
extent that a component variable represents
143
genetic variation, we might expect that Whites
and Africans would be most dissimilar, with
American Blacks displaced towards Whites
somewhat, due t o admixture. This expectation
is borne out in the male scores for component 6,
but in females the American Blacks are more
extreme than the Yoruba. Given the small
sample size of the Yoruba females, this result
does not invalidate a genetic interpretation.
Even though component 6 produces somewhat ambiguous results in terms of our genetic
expectations, they are much better than results
of the other components. Component 7 is especially noteworthy in this regard. It shows
highly significant heterogeneity in males, but
it is not patterned in any intelligable way.
Yoruba and American Whites have similar
scores, American Blacks being the extreme
group. American Black males are also extreme
on component 15. Perhaps these represent
some sort of unidentified gene-environment interaction.
ACKNOWLEDGMENTS
The data for this research were analyzed on
the IBM 360165 at the University of Tennessee
Computer Center.
LITERATURE CITED
Babler, W.J. (1977) The prenatal origins of populational
differences in human dermatoglyphics. Diss. Abst. Int.,
38:(6)A:3591.
Chai, C.K. (1972) Biological distances between indigenous
populations of Taiwan. In: The Assessment of Population
Affinities in Man. J.S. Weiner and J. Huizinga, eds.,
Clarendon Press, Oxford, pp. 182-210.
Friedlaender, J.S. (1975) Patterns of Human Variation.
Harvard University Press, Cambridge.
Froehlich, J.W. (1976) The quantitative genetics of fingerprints. In: The Measures of Man. Methodologies in Biological Anthropology. E. Giles and J.S. Friedlaender, eds.,
Peabody Museum Press, Cambridge, pp. 26&320.
Glanville, E.V., and J. Huizinga (1966) Digital dermatoglyphics of the Dogon, Peul, and Kurumba of Mali and
Upper Volta. Proc. Koninkl. Nederl. Akad. Wetensch.,
Series C, 69t664-674.
Goodman, M.M. (1972) Distance analysis in biology. Syst.
Zool., 21:174-186.
Holt, S.B. (1956) Genetics of dermal ridges: parent-child
correlations for total finger ridge-count. Ann. Hum.
Genet., 20:270-281.
Holt, S.B. (1957) Genetics of dermal ridges: sib-pair correlations for total finger ridge-count. Ann. Hum. Genet.,
21t352-362.
Holt, S.B. (1968)The Genetics of Dermal Ridges. Charles C.
Thomas, Springfield, Ill.
Jantz, R.L. (1974) Finger ridge-counts and inter-finger variability in Negroes and Whites. Hum. Biol., 46:66%675.
Jantz. R.L. (1977) Sex and race differences in finger ridgecount correlations. Am. J. Phys. Anthropol., 461171-176,
Jantz, R.L.. and H. Brehme (1978)Finger and Dalmar dermatoglyphics of a Yoruba (Nigeria)sample. Ann. Hum. Biol.,
5t539-546.
144
RICHARD L. JANTZ AND CLEONE H. HAWKINSON
Jantz, R.L., and V.P. Chopra (1976) A comparison of dermatoglyphic methodologies i n population studies. Paper
presented to the 11th International Congress of Human
Genetics, Mexico City.
Jantz, R.L., and C.H. Hawkinson (1979) Finger ridge-count
variability in sub-Saharan Africa. Ann. Hum. Biol.,
6:41-53.
Jantz, R.L., and D.W. Owsley (1977)Factor analysis of finger
ridge-counts in Blacks and Whites. Ann. Hum. Biol.,
4.357-366.
Loesch, D. (1971) Genetics of dermatoglyphic patterns on
palms. Ann. Hum. Genet., 34:277-293.
Neel, J.V., F. Rothhammer, and J.C. Lingoes (1974) The
genetic structure of a tribal population, the Yanomama
Indians. X. Agreement between representation of village
distances based on different sets of characteristics. Am. J.
Hum. Genet., 26.281-303.
Okijima, M. (1975) Development of dermal ridges in the
fetus. J. Med. Genet., 12:24%250.
Roberts, D.F., and E. Coope (1975) Components of variation
in a multifactorial character: a dermatoglyphic analysis.
Hum. Biol., 47:16%188.
Документ
Категория
Без категории
Просмотров
2
Размер файла
470 Кб
Теги
finger, variation, components, count, racial, ridges
1/--страниц
Пожаловаться на содержимое документа