Influence of cranial deformation on facial morphology among prehistoric South Central Andean populations.код для вставкиСкачать
AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 130:462–470 (2006) Inﬂuence of Cranial Deformation on Facial Morphology Among Prehistoric South Central Andean Populations Matthew P. Rhode1* and Bernardo T. Arriaza2 1 Department of Anthropology, University of Missouri-Columbia, Columbia, Missouri 65201 Centro de Investigaciones del Hombre en el Desierto, Departamento de Arqueologı́a y Museologı́a, Universidad de Tarapacá, Arica, Chile 2 KEY WORDS artiﬁcial; intentional; South America; face; craniofacial; biodistance ABSTRACT Calculating biodistances among South American populations using cranial measurements is often hindered, as many available skeletal collections exhibit deformation. Acknowledging vault modiﬁcations, researchers have sought measurements in other regions which are unaffected by deformation. In the 1970s, a set of 10 ‘‘relatively’’ unaffected facial measurements was identiﬁed in Argentinean crania that later became the basis of numerous South American biodistance studies. These measurements include: minimum frontal breadth, bizygomatic breadth, orbit height, orbit breadth, palate breath, palate length, upper facial height, basion-prosthion length, nasal height, and nasal breadth. Palate length was excluded from the present analysis due to considerable measurement error. The suitability of these measurements in populations other than Argentineans has not been rigorously tested. Using a sample of 350 prehistoric crania from the Museo Arqueológico San Miguel de Azapa (MASMA, Arica, Chile), this project tested the hypothesis that these measurements are unaffected by either annular or tabular deformation. Results obtained from MANOVA analysis indicate this hypothesis cannot be fully supported. Among males, only 3 of the 9 measurements are unaffected by either form of deformation (palate breadth, basion-prosthion length, and nasal breadth), while analysis of females indicates that 4 of the 9 measurements remain unaltered (minimum frontal breadth, orbit breadth, basion-prosthion length, and nasal breadth). Additionally, analogous to the vault, facial measurements display patterns consistent with the deformation applied. Two implications can be drawn from this research: 1) previous studies using these measurements must be interpreted cautiously, and 2) researchers using these measurements must explicitly test their suitability in each population. Am J Phys Anthropol 130:462–470, 2006. V 2006 Wiley-Liss, Inc. Deciphering biocultural relationships among past populations continues to be a topic of interest to physical anthropologists, and studies have spanned the gamut using genetic (blood groups and DNA), linguistic, archaeological, and osteometric data. Of these sources, cranial measurements, both conventional linear (two-dimensional; 2D) and coordinate (three-dimensional; 3D) landmark or morphometric data, comprise a majority of these studies. Such research has as its goal the measurement and interpretation of the phenotypic/genetic relatedness or divergence among populations or subgroups within a population, using sets of quantiﬁable polygenic morphometric traits (Buikstra et al., 1990; Larsen, 1997). However, inﬂuences from confounding factors limit the application of these methods in some regions. One such factor is cranial deformation, given the likelihood that deformation skews or alters cranial measurements. Interpretations derived from such data may inaccurately reﬂect underlying genetic relationships, being spurious associations created by deformation, which may mirror patterns created by ethnic variation and/or reproductive isolation. This is especially true of South American materials, where more than 50% of available collections often exhibit deformation (Allison et al., 1981; Dingwall, 1931; Gerszten, 1993; Gerszten and Gerszten, 1995). Two types of deformation are recognized: intentional and unintentional. Intentional deformation is the culturally prescribed practice of artiﬁcially modifying ‘‘natural’’ head shape to a more desired form (Anton, 1989a,b; Dingwall, 1931; Flowers, 1881; Gerszten, 1993; Gerszten and Gerszten, 1995; Rogers, 1975). Desired head form is informed by cultural aesthetics, parental desires, and parental willingness to submit their children to procedures involving the application of boards, pads, bands, stones, or a combination of elements to the skull during the ﬁrst year of life or longer (Dingwall, 1931; Blackwood and Danby, 1955; Rogers, 1975). Unintentional cranial deformation results from a variety of factors, including genetics, hormones, individual nutrition and health, and preferred sleeping posture (Anton, 1989a,b; Dingwall, 1931; Flowers, 1881; Rogers, 1975). Two general types of deformed crania are recognized in South America: 1) annular, circumferential, circular, or Aymara; and 2) tabular, fronto-occipital, or antero-posterior (Allison et al., 1981; Dembo and Imbelloni, 1938; Dingwall, 1931; Gerszten, 1993; Gerszten and Gerszten, 1995; Imbelloni, 1924–1925, 1933; Munizaga, 1976; Neumann, 1942; Rogers, 1975; Romano, 1965; Weiss, 1962). These two forms can be further subdivided into oblique and erect variants, C 2006 V WILEY-LISS, INC. C Grant sponsor: University of Nevada, Las Vegas Study Abroad Grants; Grant sponsor: National Geographic; Grant number: 571296; Grant sponsor: Fondecyt; Grant number: 1950035/1970525. *Correspondence to: Matthew P. Rhode, Department of Anthropology, University of Missouri-Columbia, 107 Swallow Hall, Columbia, MO 65201. E-mail: email@example.com Received 24 September 2004; accepted 9 May 2005. DOI 10.1002/ajpa.20333 Published online 27 January 2006 in Wiley InterScience (www.interscience.wiley.com). 463 CRANIAL DEFORMATION AND FACIAL MORPHOLOGY TABLE 1. Nine measurements under analysis Basion-prosthion length Maxillo-alveolar length (dropped) Maxillo-alveolar breadth Upper facial height Minimum frontal breadth Nasal height Nasal breadth Orbital breadth Orbital height Bizygomatic breadth Fig. 1. Common MASMA cranial types. which can be further subdivided into a myriad of subtypes (Allison et al., 1981; Espoueys, 2004; Gerszten, 1993). For this paper, a simple classiﬁcation system delimiting three cranial types was used (Fig. 1): annular (ANN), tabular (T), and normal/nondeformed (N), following Anton (1989a,b), Cheverud et al. (1992), Cheverud and Midkiff (1992), Kohn et al. (1993), and Rogers (1975). Common data used in biodistance analysis include linear and 3D cranial measurements (Howells, 1973, 1989, 1995; Larsen, 1997; McKeown and Jantz, 2002). Observed variation in such data is assumed to reﬂect primary genetic relationships between sample populations, ‘‘undiluted’’ by external factors. Yet, as noted, populations like those from South America often possess a greater percentage of deformed than nondeformed crania. When confronted with collections possessing deformed crania, researchers have two alternatives: 1) not to perform biodistance analysis using deformed crania, proceeding with a reduced but potentially biased sample of nondeformed crania; or 2) to discover a way to use the deformed crania, which will permit the biodistance analysis to proceed. Given that the ﬁrst alternative is often not appealing, researchers have attempted to ﬁnd ways to make deformed crania usable. One method used to make deformed crania usable is to identify regions which are only ‘‘slightly’’ (nonsigniﬁcantly) affected by deformation. A review of the literature reveals numerous attempts to deﬁne the impacts of deformation on cranial measurements (Anton, 1989a,b; Blackwood and Danby, 1955; Bjork and Bjork, 1964; Cheverud et al., 1992; Cheverud and Midkiff, 1992; Friess and Baylac, 2003; Kohn et al., 1993; McGibbon, 1965; McNeill and Newton, 1965; Mizoguchi, 1991; Oetteking, 1924; Schendel et al., 1980; Suzuki et al., 1993). Overall results suggest that most common cranial measurements are altered by deformation to some degree. The degree and severity of modiﬁcation vary by population, deformation type, and measurements under study. BAPR PRALV ECMECM NPR FTFT NNS ALAL DEC ORBHGHT ZYZY Annular deformation is characterized by lengthening of the cranial vault and narrowing of medial-lateral dimensions, leading to increased height and decreased width measurements (Anton, 1989a,b; Kohn et al., 1993). Tabular deformation is characterized by vault compression along the sagittal plane, with expansion of mediallateral dimensions. This causes width measurements to increase, while height measurements are variably affected (Anton, 1989a,b; Cheverud et al., 1992; Cheverud and Midkiff, 1992). As the vault is the most signiﬁcantly altered portion of the cranium, other regions have received increased attention in the search for measurements unaffected by deformation. Of the remaining regions, the facial skeleton has generated the most research. This attention is tied to the assumption that facial features are more directly under genetic (population-speciﬁc) and cultural (health, diet, and subsistence) control (Devor, 1987; Kohn, 1991; Vandenberg, 1962), and thus are useful in establishing biological relatedness among populations. In South America, a series of inﬂuential publications by Cocilovo (1973, 1975, 1978) profoundly impacted the direction of future biodistance studies in the region. Cocilovo (1975) identiﬁed a set of 10 facial measurements among Argentinean crania that were shown to be unaffected or ‘‘slightly’’ affected by deformation (Table 1). This set of 10 facial measurements was subsequently adopted by researchers, and has been the basis for numerous South American biodistance studies over the past 20 years (Cocilovo et al., 1982, 1990, 1995; Dittmar, 1996; Guillén, 1992; Rivera, 1984, 1991; Rivera and Rothhammer, 1986, 1991; Rothhammer, 1994; Rothhammer and Silva, 1989, 1990, 1992; Rothhammer et al., 1981, 1982, 1983, 1984a,b, 1986; Rothhammer and Santoro, 2001). Upon review of this research, concerns arise over the use of measurement sets, which differ from study to study. This may by an artifact of data limitations, but there is an expectation that researchers would maintain consistency across publications that often used the same populations. This discrepancy makes evaluating published results complicated (Sutter, 1997, 2000; Rhode, 2001, 2002). More signiﬁcantly, there is doubt as to whether the reliability of these measurements has been fully tested. Speciﬁcally, no published study shows whether the measurements remain stable in deformed crania other than the Argentineans originally examined by Cocilovo (1973, 1975, 1978; see Rhode, 2001, 2002). The present research seeks to address this concern by testing the hypothesis that 9 of the 10 measurements identiﬁed by Cocilovo (1973, 1975, 1978) are unaffected or are only slightly affected (nonsigniﬁcantly) by deformation (annular or tabular) in a prehistoric Chilean population, and therefore can be used by researchers in biodistance studies of these populations. 464 M.P. RHODE AND B.T. ARRIAZA TABLE 2. Cranial types studied Cranial type Normal Annular Tabular Total Males Females Total 66 45 40 151 54 62 83 199 120 107 123 350 0.46 mm. Further information on the procedures used to test measurement replicability can be found in Rhode (2001). Combined, these results indicate random, nondirectional error, reﬂecting a consistent measurement technique. Interobserver error could not be examined due to time constraints in the ﬁeld. Missing data MATERIALS AND METHODS Sample The sample used to test the hypothesis of measurement stability under deformation included several collections from the Museo Arqueológico San Miguel de Azapa (Arica, Chile). This is the same collection used by Rothhammer and Santoro (2001) in a biodistance study of Azapa Valley populations. The sample consisted of 350 crania representing 18 archaeological sites (8 coastal and 10 inland) and several cultural groups spanning a period of roughly 7,000–8,000 years. Additional background information on these sites can be obtained in Rhode (2001). Cases were selected for analysis based on the following criteria. 1) Only complete or nearly complete crania were used. 2) Crania were from individuals over 17 years old. 3) Crania with obvious pathologies or trauma were excluded. 4) Approximately equal numbers of males and females (151 males and 199 females) were used to permit analysis of sexual dimorphism. Each individual was coded for sex and deformation type prior to collecting measurements. Sex determinations were aided by the fact that many skeletons were formerly mummiﬁed and had been sexed using preserved external genitalia. Cranial deformation was scored visually, following the system of Dembo and Imbelloni (1938). Following Rogers (1975) and Anton (1989a,b), only three cranial types were recognized: normal (N), annular (ANN), and tabular (T) deformation (Fig. 1). Of the 350 crania studied, 230 (65.6%) exhibited deformation. One hundred and twenty crania were classiﬁed as normal/none, 107 displayed annular deformation, and 123 were classiﬁed as exhibiting tabular deformation. Table 2 provides a detailed breakdown of cranial types used. Measurements Table 1 lists the 10 linear measurements collected from each cranium, which were identiﬁed as unaffected by Cocilovo (1973, 1975, 1978). Initially, all 10 of Cocilovo’s measurements were to be examined, but palate length (pr-alv) was excluded from analysis early during data collection because it was found to be an inherently difﬁcult measurement to replicate, as noted by Heathcote (1981). Speciﬁcally, determining the location of alveolon (alv) is problematic, as it lacks a deﬁned anatomical landmark. After each of the 350 crania had been measured by M.P.R., a subset of 51 (25 males and 26 females), or 14.6%, of the sample was randomly selected and remeasured to conduct intraobserver error analysis, as recommended by Buikstra and Ubelaker (1994). The time lapse between initial and secondary measurement periods varied from 1 week to 1 month. Comparisons between the ﬁrst and second measurement sessions across all measurements produced Pearson correlations averaging r ¼ 0.96, an average mean difference of measurement of 6 0.02 mm, and an average method error of Missing data accounted for approximately 5% of the sample, indicating that data replacement was possible (Allison, 2002; Little and Rubin, 1987). Factor analysis of the missing values did not show patterning with age, sex, or health indicators, suggesting the pattern of missing data could be considered missing at random (MAR) (Allison, 2002; Little and Rubin, 1987, 1989; Little and Schenker, 1995; Schafer, 1997, 1999a,b; Schafer and Olsen, 1998). Missing data patterns by deformation type were not investigated, and therefore it was not possible to establish whether the data were MAR with respect to the variables of interest (Holt and Benfer, 2000). Several approaches are available for analyzing and imputing missing data (e.g., Allison, 2002; Little and Rubin, 1987). One common method is complete case analysis. If used, this method would have reduced the data set by 50%, possibly creating a substantially biased sample. Other common methods include mean replacement or linear regression, which may have created an overly homogenous data set, obscuring integral relationships. The method selected to address missing data in this project was the multiple imputation procedure of Schafer (1997), performed using the NORM 2.3 program (Schafer, 1999a,b). The NORM program is designed to analyze normally distributed continuous data. For this data set, the only non-normal data were sex and deformation type, which were determined for all crania. The only missing data were measurements from individuals who exhibited damaged crania. Given these conditions, it is acceptable to use the NORM program for missing data analysis. Multiple imputation uses an expected maximization procedure as the ﬁrst approximation of the missing data, and then performs data augmentation using a Monte Carlo Markov chain procedure, which creates multiple complete data sets that progressively (iteratively) converge towards a point of unity, where the variation between imputed data sets becomes negligible. Diverging from the procedures of Schafer (1999a,b), the ﬁnal imputed data matrix, exhibiting the least variation, was chosen as the data set to be studied, instead of using 4 or 5 selected during imputation. Males and females were not separated during this procedure, meaning that the imputed data set may exhibit slight homogenization, but less than could occur using mean replacement or linear regression. Statistical analyses Exploratory data analysis performed using SPSS 8.0 and 12.0 indicated that the nine measurements did not violate normality assumptions individually, by sex, or by deformation based on analysis of box plots, spread vs. level plots, and Q-Q plots (Norušis, 1988). Though several values for individual measurements could be classiﬁed as outliers, none were extreme outliers, and all 350 individuals were used in subsequent analyses. Table 3 presents summary statistics created during this analysis for the nine measurements by sex and deformation type. 465 CRANIAL DEFORMATION AND FACIAL MORPHOLOGY TABLE 3. Summary statistics of nine measurements under analysis Male Measurement Deformation type Mean Normal Annular Tabular Normal Annular Tabular Normal Annular Tabular Normal Annular Tabular Normal Annular Tabular Normal Annular Tabular Normal Annular Tabular Normal Annular Tabular Normal Annular Tabular 98.6 100.3 99.1 61.6 62 62.6 69.4 71.4 70.8 91 88.3 88.7 50.4 51.7 50.2 24.4 24.7 24.7 38.6 39.5 38.8 34.7 36.3 35.9 136.9 135.4 138.4 BAPR ECMECM NPR FTFT NNS ALAL DEC ORBHGHT ZYZY Female SD Deformation type Mean SD 5.2 5.1 4.2 3.3 3.4 3.7 4.4 4.2 4.2 4.8 5.3 5.6 3.4 2.5 2.9 1.6 1.9 1.6 1.7 1.7 1.7 1.9 1.9 2.1 5.6 5.7 4.9 Normal Annular Tabular Normal Annular Tabular Normal Annular Tabular Normal Annular Tabular Normal Annular Tabular Normal Annular Tabular Normal Annular Tabular Normal Annular Tabular Normal Annular Tabular 95.4 96.9 95.9 59.4 59.8 60.9 66.2 68.3 68.5 87.3 86 87.5 47.1 49.4 48.6 23.8 24.3 24.2 38.1 38 38.2 34.3 36.1 35.8 127.9 127.6 131.5 4.7 3.9 4.4 3.3 2.8 2.9 4.3 3.3 4.4 3.8 5.3 4.3 2.6 2.5 2.8 1.8 1.6 1.6 1.5 1.6 1.7 1.8 1.8 2 4.5 4.7 4.7 TABLE 4. Male and female multivariate MANOVA1 Male Deformation type Female Deformation type 1 Test Value F-value Hypothesized df Error df Signiﬁcance Pillai’s trace Wilks’ lambda 0.41 0.63 4.00 4.05 18 18 282 280 0.00 0.00 Pillai’s trace Wilks’ lambda 0.43 0.61 5.81 5.80 18 18 378 376 0.00 0.00 Signiﬁcant results in bold. The statistical method chosen to test the effect of deformation on the nine facial measurements was multivariate analysis of variance (MANOVA) under the general linear model (GLM) multivariate procedures module of SPSS 8.0 and 12.0. This procedure is an extension of the general ANOVA, for questions with several dependent variables measured in two or more samples (Norman and Striener, 2000; Norušis, 1998; Sokal and Rohlf, 1995). This method avoids spurious results caused by repeated testing, which would occur with multiple ttests, taking the correlation among dependent variables into consideration (Norman and Striener, 2000). In this study, the nine facial measurements were the dependent variables, and the independent variables were deformation types (normal, annular, and tabular). During preliminary analyses (not presented), sexual dimorphism was found to be a signiﬁcant inﬂuencing factor on all variables, and male and female data sets were subsequently separated during the MANOVA analysis. RESULTS Before running the MANOVA, a Box’s M test of covariance matrix equality was performed. Results did not reach signiﬁcance, with F ¼ 1.13 and P ¼ 0.09, indicating that the assumption of covariance matrix equality can be supported. These results also indicate that the MANOVA model was not suspect, meaning the analysis could reasonably proceed. The ﬁrst MANOVA analysis examined the overall relationship between measurements (dependent variables) and cranial types (independent variables). The results of this test, presented in Table 4, indicate that variation in cranial type signiﬁcantly altered measurements, with a Pillai’s trace and Wilk’s lambda values of P ¼ 0.00 for males and females. The low Pillai’s trace and high Wilk’s lambda values from Table 4 signify the presence of strong interactions among dependent and independent variables. These results by themselves imply that the hypothesis of measurement stability across deformation types cannot be supported in this population. Yet additional tests were needed to identify whether all measurements were equally affected by deformation, or whether effects varied by deformation type. Following the basic multivariate tests, a more detailed analysis of between-subjects effects was carried out for each measurement for each of the three cranial types. The results of this test, presented in Table 5, indicate 466 M.P. RHODE AND B.T. ARRIAZA TABLE 5. Male and female between-subjects effects by deformation1 Dependent variable Type III SS df Mean square F-value Signiﬁcance Males FTFT ORBHGHT ZYZY DEC ECMECM NPR BAPR NNS ALAL 236.79 76.62 191.96 270.03 25.35 118.20 83.93 60.43 4.38 2 2 2 2 2 2 2 2 2 118.39 38.31 95.98 13.51 12.67 59.10 41.97 30.21 2.19 4.48 9.80 3.20 4.71 10.09 3.22 1.71 3.31 0.74 0.01 0.00 0.04 0.01 0.34 0.04 0.18 0.04 0.48 Females FTFT ORBHGHT ZYZY DEC ECMECM NPR BAPR NNS ALAL 84.27 114.20 693.69 0.68 83.25 182.80 68.58 1530.09 8.50 2 2 2 2 2 2 2 2 2 42.13 57.10 346.85 0.34 41.63 91.40 34.29 76.55 4.25 20.07 16.22 160.03 0.13 4.68 5.56 1.80 10.89 1.54 0.13 0.00 0.00 0.88 0.01 0.00 0.17 0.00 0.22 1 Fig. 3. Male measurements affected by deformation. Signiﬁcant results in bold. Fig. 4. Female measurements not affected by deformation. Fig. 2. Male measurements not affected by deformation. that only 3 of 9 male measurements (palate breadth, basion-prosthion length, and nasal breadth) could be considered to remain stable under deformation. Figure 2 graphically displays the unaffected male measurements, and Figure 3 the affected male measurements. For the female data, only 4 of the 9 measurements (minimum frontal breadth, orbit breadth, basion-prosthion length, and nasal breadth) remain stable under deformation. Figure 4 graphically presents the unaffected female measurements, and Figure 5 the affected female measurements. Examining these tables and ﬁgures, it is clear that fewer than half of the measurements of Cocilovo (1973, 1975, 1978) may be considered unaffected by deformation in this population. Among the female signiﬁcance values (Table 5), all measurements were either strongly signiﬁcant or nonsigniﬁcant. Yet among male values this was not the case, as several measurements (bizygomatic breath, nasal height, and upper facial height) can be considered marginally signiﬁcant (P ¼ 0.04). These measurements may possibly be more affected by one type of deformation than the other, differences unable to be highlighted though this analysis. Between-subjects effects reveal no information concerning how a given measurement is altered by deformation. Post hoc testing using multiple comparisons can clarify these relationships. Post hoc tests (similar to repeated t-tests) were performed on each cranial type combination (normal-annular, normal-tabular, and annular-tabular) across all nine measurements for males and females, to determine the direction and magnitude in which measurements were altered by deformation, and to clarify male measurement relationships. Repeated testing can be problematic, because increasing the number of comparisons reduces the likelihood of at least one comparison being wrongly declared signiﬁcant, thus committing a type II error. Post hoc tests account for this possibility by applying a correction factor to reduce the possibility of inﬂated probability values (Legendre and Legendre, 1998; Sokal and Rohlf, 1995). Of several correction factors, Tukey’s honestly sig- CRANIAL DEFORMATION AND FACIAL MORPHOLOGY Fig. 5. Female measurements affected by deformation. niﬁcant difference (HSD) was selected for this study. This is a more liberal test than the more commonly used but conservative Bonferroni test (Norman and Striener, 2000). Table 6 presents the results of the multiple comparisons analysis. Reviewing the results in Table 6, the most obvious aspect is that none of the nine measurements can be considered signiﬁcant across all cranial types for either males or females. Focusing on the three male measurements considered marginally signiﬁcant, this patterning explains the observed marginal P-values. Only the annular-tabular comparison was signiﬁcant for bizygomatic breadth, while the annular-normal and tabular-normal comparisons were nonsigniﬁcant. In other words, bizygomatic breadth can be used to differentiate annular from tabular crania, but not normal from deformed crania. For upper facial height, only the normal-annular comparison was signiﬁcant, indicating that an increase in upper facial height can be used to distinguish normal and annular crania. Further, this suggests that in these populations, tabular and annular deformation cause upper facial height to be increased among males. The third questionable male measurement, nasal height, displayed a more interesting result, which illustrates that none of the three comparisons performed was significant with adjusted P-values. The normal-annular (P ¼ 0.07) and annular-tabular (P ¼ 0.06) comparisons nearly reach signiﬁcance, yet given the combined unlikelihood of these three differences, it is acceptable to include this measurement among those considered nonsigniﬁcant. However, for the purposes of this paper, it is included among the measurements considered signiﬁcantly affected. DISCUSSION The patterns obtained from the MANOVA analysis can more easily be assessed by examining Figures 2–5. Overall, it appears that cranial deformation has a variable but deﬁnitive impact on the facial skeleton. This impact corresponds both to the type of deformation applied and culturally prescribed factors (severity and/or apparatus) related to the deformation procedure. Speciﬁcally, the patterns noted in Table 6 and Figures 2–5 correspond 467 well with those noted by previous researchers regarding the general consequences of deformation. Brieﬂy, annularly deformed crania exhibit increased length measurements seen in the nasal and orbital heights, while reducing width measurements is evidenced in bizygomatic breadth and minimum frontal breadth. Palate breadth can be included in this list for females, but not males. This result may reﬂect variation across the sexes in performing deformation within these Chilean groups, or it may correspond to variation in deformation severity (data collected but not included in this analysis). The facial skeletons of tabularly deformed crania similarly follow patterns described in the literature, with increased width measurements evidenced by bizygomatic breadth. Minimum frontal breadth varies with tabular deformation in males but not females, further suggesting cultural differences in deformation practices. Height measurements in tabular crania are increased, paralleling the change noted with annular deformation. Two of the measurements unaffected by deformation were common to both sexes: basion-prosthion length and nasal breadth. The centralized placement of these measurements suggests the presence of an underlying structure to the human skull oriented along the inferior sagittal plane. Many of the unaffected measurements, with the exception of minimum frontal breadth (among females), are located centrally in the cranium, suggesting that both types of deformation have a greater effect on peripheral cranial structures, a ﬁnding which supports Anton (1989a,b). The differences observed between males and females can be explained through either of two hypotheses. The variation noted in measurements between males and females may reﬂect some underlying cultural aesthetics in deformation across the sexes. If this were the case, it would explain why minimum frontal breadth was unaffected in women but not men. Further, if males preferentially had greater force applied to the frontal portion of their skulls during tabular deformation, the pattern observed would be expected. Alternatively, given the inﬁnite range of possible head forms, these signiﬁcance differences may simply reﬂect the broad grouping methodology used, in which several deformation types and severities were combined, obscuring sex-limited expression in deformation subtypes and severity not addressed in the present research. CONCLUSIONS For this sample of several prehistoric Chilean populations, out of the original nine measurements of Cocilovo (1973, 1975, 1978) under consideration, only three male measurements (palate breadth, basion-prosthion length, and nasal breadth) and four female measurements (minimum frontal breadth, orbit breadth, basion-prosthion length, and nasal breadth) can be considered stable or unaffected by either annular or tabular cranial deformation. This means that even if these measurements exhibit some inﬂuence caused by the biomechanical forces used to modify the shape of the cranium, effects noted in these measurements are not signiﬁcant with the sample size used. The remaining six male and ﬁve female measurements show signiﬁcant effects by deformation, varying with cranial form. The obtained results suggest that caution is warranted in using these nine cranial measurements in biodistance studies in any population that displays intentional deformation. Clearly the original hypothesis that all nine cra- 468 M.P. RHODE AND B.T. ARRIAZA TABLE 6. Male and female Tukey’s HSD multiple comparisons Male1 FTFT ORBHGHT ZYZY DEC ECMECM NPR BAPR NNS ALAL 1 2 3 Deformation type2 Mean difference Standard error Signiﬁcance3 N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T 62.71 62.28 60.43 61.59 61.20 60.39 61.47 61.54 63.01 60.98 60.21 60.78 60.47 61.00 60.54 62.00 61.42 60.59 61.76 60.52 61.24 61.28 60.23 61.51 60.31 60.37 60.06 0.99 10.03 0.99 0.38 0.40 0.43 10.06 1.10 1.19 0.33 0.34 0.37 0.66 0.68 0.74 0.83 0.86 0.93 0.96 0.99 10.07 0.58 0.61 0.66 0.33 0.34 0.37 0.02 0.07 0.02 0.00 0.01 0.63 0.35 0.34 0.03 0.01 0.81 0.09 0.76 0.31 0.75 0.04 0.23 0.80 0.16 0.86 0.48 0.07 0.92 0.06 0.62 0.52 0.98 Female1 FTFT ORBHGHT ZYZY DEC ECMECM NPR BAPR NNS ALAL Deformation type2 Mean difference Standard error Signiﬁcance3 N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T N vs. Ann N vs. T Ann vs. T 61.33 60.12 61.45 61.88 61.51 60.37 60.34 63.60 63.94 60.028 60.10 60.13 60.38 61.48 61.10 62.05 62.22 60.17 61.50 60.52 60.97 62.27 61.51 60.77 60.50 60.42 60.08 0.84 0.79 0.76 0.35 0.33 0.31 0.87 0.81 0.78 0.30 0.28 0.27 0.56 0.52 0.50 0.76 0.71 0.68 0.81 0.76 0.73 0.49 0.46 0.45 0.31 0.29 0.28 0.25 0.99 0.14 0.00 0.00 0.47 0.92 0.00 0.00 1.00 0.93 0.88 0.78 0.01 0.07 0.02 0.00 0.96 0.16 0.77 0.38 0.00 0.00 0.20 0.23 0.31 0.95 Unaffected measurements in bold. N, normal; Ann, Anular; T, Tabular. Signiﬁcant comparisons in italic. nial measurements could be used in Chilean populations has been refuted. Moreover, these results cast serious doubts on the conclusions presented in previous biodistance studies of Chilean populations. Ultimately, those researchers wanting to use this set of measurements, would be well advised to explicitly test their suitability and stability in each population before proceeding with data analysis. These cranial measurements and others commonly collected are still of value, and can be used to obtain insights into individual or population-level research questions, but only when used in combination with other sources such as geographic, chronological, and archaeological data, or health, subsistence, activity, and status indicators. Biodistance studies seeking to assess genetic differences between and among populations cannot use craniometric variables when the populations in question performed cultural modiﬁcations of the cranium. However, those measurements that are found to vary signiﬁcantly with a given type of deformation may prove useful in measuring the combined results of reproductive and cultural isolation, which probably accounts for their overall interpretability in characterizing biological populations, which are also ethnic groups. ACKNOWLEDGMENTS The authors thank the following individuals and organizations for their assistance during this project: the Museo Arqueológico San Miguel de Azapa researchers Leticia Latorre and Vivien Standen, University of Nevada at Las Vegas Study Abroad Grants, and University of Nevada at Las Vegas Summer Scholarships. The authors also thank the anonymous reviewers who provided constructive criticism and valuable suggestions. LITERATURE CITED Allison MJ, Gerszten E, Munizaga J, Santoro C, Focacci G. 1981. La práctica de la deformación craneana entre los pueblos Andinos Precolombinos. Chungara 7:238–260. Allison PA. 2002. Missing data. London: Sage Publications, Inc. p 136. Anton SA. 1989a. Facial structure: the inﬂuence of intentional vault deformation. Am J Phys Anthropol 78:184. Anton SA. 1989b. Intentional cranial vault deformation and induced changes of the cranial base and face. Am J Phys Anthropol 79:253–267. Bjork A, Bjork L. 1964. Artiﬁcial deformation and cranio-facial asymmetry in ancient Peruvians. J Dent Res 43:353–362. Blackwood B, Danby PM. 1955. A study of artiﬁcial cranial deformation in New Britain. J R Anthropol Inst Great Britain Ireland 85:173–193. Buikstra JE, Ubelaker DH, editors. 1994. Standards for data collection from human skeletal remains. Arkansas Archaeological Survey research series no. 44. Fayetteville: Arkansas Archaeological Survey. Buikstra JE, Frankenberg SR, Konigsberg LW. 1990. Skeletal biological distance studies in American physical anthropology: recent trends. Am J Phys Anthropol 82:1–7. Cheverud JA, Midkiff JE. 1992. Effects of fronto-occipital cranial reshaping on mandibular form. Am J Phys Anthropol 87: 167–181. Cheverud JA, Kohn LAP, Konigsberg LW. 1992. Effects of fronto-occipital artiﬁcial cranial vault modiﬁcation on the cranial base and face. Am J Phys Anthropol 88:323–345. Cocilovo JA. 1973. Dimorﬁsmo sexual y deformación craneana en Patagones de Chubut. Act II Simp Int Ciencias Morf 633–642. CRANIAL DEFORMATION AND FACIAL MORPHOLOGY Cocilovo JA. 1975. Estudio de los factores que inﬂuencian la morfologı́a craneana en una colección Andina: el sexo y la deformación. Rev Inst Antropol 2:197–212. Cocilovo JA. 1978. Estudio de dos factores que inﬂuyen en la morfologı́a craneana en una colección patagónica: el sexo y la deformación artiﬁcal. Volume 3. Rio de Janerio: O Instituto. p 113–141. Cocilovo JA, Rothhammer F, Quevedo S, Llop E. 1982. Microevolución en poblaciones prehistóricas del area Andina. III. La población del Morro de Arica. Craneometrı́a. Rev UNRC 2:91–111. Cocilovo JA, Cuarto R, Rothhammer F. 1990. Paleopopulation biology of southern Andes: craniofacial chronological and geographical differentiation. Homo 41:16–31. Cocilovo JA, Varela HH, Quevedo S. 1995. La deformación artiﬁcial del cráneo en la población prehistórica de San Pedro de Atacama, Chile. Chungara 27:117–124. Dembo A, Imbelloni J. 1938. Deformaciones intencionales del cuerpo humano de carácter étnico. Biblioteca Humanior Sección A3, Buenos Aires: Imprenta Luis L. Gotelli. Devor EJ. 1987. Transmission of human craniofacial dimensions. J Craniofac Genet Dev Biol 7:95–96. Dingwall EJ. 1931. Artiﬁcial cranial deformation: a contribution to the study of ethnic mutilations. London: John Bale, Sons and Danielsson, Ltd. Dittmar M. 1996. A study of the face structure of highland and lowland populations in prehispanic South America. Coll Antropol 20:95–110. Espoueys O. 2004. (ms) Proyecto Fondecyt no. 1020375, contribución de la morfometrı́a geométrica al estudio antropológico de las poblaciones prehispánicas de Chile: microevolución y variación de la forma craneana. MASMA, Arica, Chile. Flowers WH. 1881. Fashion in deformity. London: Macmillan and Co. Friess M, Baylac M. 2003. Exploring artiﬁcial cranial deformation using elliptic Fourier analysis of Procrustes aligned outlines. Am J Phys Anthropol 122:11–22. Gerszten PC. 1993. An investigation into the practice of cranial deformation among the pre-Columbian peoples of Northern Chile. Int J Osteol 3:87–98. Gerszten PC, Gerszten E. 1995. Intentional cranial deformation: a disappearing form of self-mutilation. Neurosurgery 37:374–381. Guillén SE. 1992. The Chinchorro culture: mummies and crania in the reconstruction of preceramic coastal adaptations in the south central Andes. Unpublished Ph.D. dissertation, University of Michigan. Heathcote GM. 1981. The magnitude and consequences of measurement error in human craniometry. Can Rev Phys Anthropol 3:18–40. Holt B, Benfer RA. 2000. Estimating missing data: an interative regression approach. J Hum Evol 39:289–296. Howells WW. 1973. Cranial variation in man: a study by multivariate analysis of patterns of difference among recent human populations. Papers of the Peabody Museum of Archaeology and Ethnology. Harvard University. Volume 67. Cambridge, MA: Harvard University. Howells WW. 1989. Skull shapes and the map: craniometric analyses in the dispersion of modern Homo. Papers of the Peabody Museum of Archaeology and Ethnology. Harvard University. Volume 79. Cambridge, MA: Harvard University. Howells WW. 1995. Who’s who in skulls: ethnic identiﬁcation of crania from measurements. Papers of the Peabody Museum of Archaeology and Ethnology. Harvard University. Volume 82. Cambridge, MA: Harvard University. Imbelloni J. 1924–1925. Deformaciones intencionales del cráneo en Sud América. Rev Mus La Plata 28:329–407. Imbelloni J. 1933. Los pueblos deformadores de los Andes. Antropologı́a, etnologı́a y arqueologı́a. Publicación no. 75. Anales del Museo Argentino de Ciencias Naturales, Bernardino Rivadavia (editor). Buenes Aires, Argentina. p 209–254. Kohn LAP. 1991. The role of genetics in craniofacial morphology and growth. Annu Rev Anthropol 20:261–278. Kohn LAP, Leigh SR, Jacobs SC, Cheverud JM. 1993. Effects of annular cranial vault modiﬁcation on the cranial base and face. Am J Phys Anthropol 90:147–168. 469 Larsen CS. 1997. Bioarchaeology: interpreting behavior from the human skeleton. Cambridge: Cambridge University Press. Legendre P, Legendre L. 1998. Numerical ecology. Amsterdam: Elsevier Science B.V. Little RJA, Rubin DB. 1987. Statistical analysis with missing data. New York: John Wiley. Little RJA, Rubin DB. 1989. The analysis of social science data with missing values. Sociol Methods Res 18:292–326. Little RJA, Schenker N. 1995. Missing data. In: Arminger G, Clogg C, Sobel M, editors. Handbook of statistical modeling for the social and behavioral sciences. New York: Plenum. p 39–75. McGibbon H. 1965. Artiﬁcially deformed skull with special reference to the temporal and its typanic portion. Laryngoscope 22:1165–1184. McKeown AH, Jantz RL. 2002. Comparing traditional craniometric and nontraditional landmark based methods for investigating cranial variation [abstract]. Am J Phys Anthropol [Suppl] 117(S2):111. McNeill RW, Newton GN. 1965. Cranial base morphology in association with intentional cranial vault deformation. Am J Phys Anthropol 23:241–254. Mizoguchi Y. 1991. Covariation in craniofacial measurement caused by artiﬁcial deformation of the cranial vault. Bull Nat Sci Mus 17:31–50. Munizaga JR. 1976. Intentional cranial deformation in the preColumbian populations of Ecuador. Am J Phys Anthropol 45: 687–694. Neumann GK. 1942. Types of artiﬁcial cranial deformation in the eastern United States. Am Antiq 7:306–310. Norman GR, Striener DL 2000. Biostatistics: the bare essentials. 2nd ed. London: B.C. Decker. Norušis MJ. 1998. SPSS 8.0 guide to data analysis. Englewood Cliffs, New Jersey: Prentice Hall. Oetteking B. 1924. Declination of the pars basilaris in normal and in artiﬁcially deformed skull. Indian notes and monographs, miscellaneous. New York: Museum of the American Indian, Heye Foundation. Rhode MP. 2001. Craniofacial morphology and the consequences of intentional cranial deformation among prehistoric Andean populations. Unpublished Masters thesis, University of Nevada, Las Vegas. Rhode MP. 2002. Cranial deformation and measurement stability among prehistoric South Central Andean populations [abstract]. Am J Phys Anthropol [Suppl] 117(S32):30. Rivera MA. 1984. Altiplano and tropical lowland contacts in northern Chile prehistory: Chinchorro and Alto Ramirez revisited. In: Social and economic organization of the prehistoric Andes. International Congress of the Americanists. Browman DL, Burger RL, Rivera MA (editors). BAR International series 194. p 117–160. Rivera MA. 1991. The prehistory of northern Chile: a synthesis. J World Prehist 5:1–47. Rivera MA, Rothhammer F. 1986. Evaluación biológica y cultural de poblaciones Chinchorro: nuevos elementos para la hipótesis de contactos transaltiplánicos, cuenca AmazonasCosta Pacı́ﬁco. Chungara 16–17:295–306. Rivera MA, Rothhammer F. 1991. The Chinchorro people of northern Chile 5000 BC–500 BC: a review of their culture and relationships. Int J Anthropol 6:243–255. Rogers SL. 1975. Artiﬁcial deformation of the head: New World examples of ethnic mutilations and notes on its consequences. San Diego Museum papers no. 8. California: San Diego Museum of Man. Romano A. 1965. Estudio morfologio de la deformacion craneana en Tamuin, S. L. P. y en la Isa del Idolo, Ver. Mexico City: Instituto Nacional de Antropologia e Historia. Rothhammer F. 1994. Microevolution in prehistoric Andean populations: geographic craniometric differentiation. Homo 45:1–7. Rothhammer F, Santoro CM. 2001. El desarrollo cultural en el valle de Azapa, extremo norte de Chile y su vinculación con los deplazamientos poblacionales altiplánicos. Lat Am Antiq 12:59–66. 470 M.P. RHODE AND B.T. ARRIAZA Rothhammer F, Silva C. 1989. Peopling of Andean South America. Am J Phys Anthropol 78:403–410. Rothhammer F, Silva C. 1990. Craniometrical variation among South American prehistoric populations: climatic, altitudinal, chronological, and geographic contributions. Am J Phys Anthropol 82:9–17. Rothhammer F, Silva C. 1992. Gene geography of South America: testing models of population displacement based on archaeological evidence. Am J Phys Anthropol 89:441–446. Rothhammer F, Quevedo S, Cocilovo A, Foccaci S, Llop E. 1981. Microevolución en poblaciones prehistoricas del area Andina. II. Varacion craneométrica cronológica en los valles de Arica. Chungara 8:275–280. Rothhammer F, Cocilovo JA, Quevedo S, Llop E. 1982. Microevolution in prehistoric Andean populations: I. Chronologic craniometric variation. Am J Phys Anthropol 58:391– 396. Rothhammer F, Cocilovo JA, Quevedo S, Llop E. 1983. Aﬁnidad biológica de las poblaciones prehistóricas del litoral ariqueño con groupo poblaciónales costero peruanos y altiplánicos. Chungara 11:161–165. Rothhammer F, Quevedo S, Cocilovo JA, Llop E. 1984a. Microevolution in prehistoric Andean populations: chronologic nonmetrical cranial variation in northern Chile. Am J Phys Anthropol 65:157–162. Rothhammer F, Cocilovo JA, Quevedo S. 1984b. El poblamiento temprano de Sudamérica. Chungara 13:99–108. Rothhammer F, Silva C, Cocilovo JA, Quevedo S. 1986. Una hipótesis provisional sobre el poblamiento de Chile basada en el análisis multivariado de medidas craneometricas. Chungara 16–17:115–118. Schafer JL. 1997. Analysis of incomplete multivariate data. London: Chapman & Hall. Schafer JL. 1999a. Multiple imputation: a primer. Stat Methods Med Res 8:3–15. Schafer JL. 1999b. NORM: multiple imputation of incomplete multivariate data under a normal model, version 2. Software for Windows 95/98/NT, available from http://www.stat.psu.edu/ jls/misoftwa.html. Schafer JL, Olsen MK. 1998. Multiple imputation for multivariate missing-data problems: a data analyst’s perspective. Multivar Behav Res 33:545–571. Schendel SA, Walker G, Kamisugi A. 1980. Hawaiian craniofacial morphometrics: average Mokapuan skull, artiﬁcial cranial deformation, and the ‘‘rocker’’ mandible. Am J Phys Anthropol 52:491–500. Sokal RR, Rohlf FJ. 1995. Biometry. 3rd ed. San Francisco: W.H. Freeman. Sutter RC. 1997. Dental variations and biocultural afﬁnities among prehistoric populations from the coastal valleys of Moquegua, Peru, and Azapa, Chile. Unpublished Ph.D. dissertation, University of Missouri-Columbia. Sutter RC. 2000. Prehistoric genetic and culture change: a bioarchaeological search for pre-Inka altiplano colonies in the coastal valleys of Moquegua, Peru, and Azapa, Chile. Lat Am Antiq 11:43–70. Suzuki H, Mizoguchi Y, Conese E. 1993. Craniofacial measurement of artiﬁcially deformed skull from the Philippines. Anthropol Sci 101:111–127. Vandenberg SG. 1962. How ‘‘stable’’ are heritability estimates? A comparison of six anthropometric studies. Am J Phys Anthropol 20:331–338. Weiss PH. 1962. Tipologı́a de las deformaciones cefálicas de los antiguos Peruanos, según la osteologı́a cultural. Rev Mus Nac Lima 31:15–42.