AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 128:569 –585 (2005) Body Proportions in Ancient Andeans From High and Low Altitudes Karen J. Weinstein* Department of Anthropology, Dickinson College, Carlisle, Pennsylvania 17013 KEY WORDS limb proportions; ecogeographic variation; high altitude; Andes ABSTRACT Living human populations from high altitudes in the Andes exhibit relatively short limbs compared with neighboring groups from lower elevations as adaptations to cold climates characteristic of high-altitude environments. This study compares relative limb lengths and proportions in pre-Contact human skeletons from different altitudes to test whether ecogeographic variation also existed in Andean prehistory. Maximum lengths of the humerus, radius, femur, and tibia, and femoral head breadth are measured in sex-speciﬁc groups of adult human skeletons (N ⫽ 346) from the central (n ⫽ 80) and the south-central (n ⫽ 123) Andean coasts, the Atacama Desert at 2,500 m (n ⫽ 102), and the southern Peruvian highlands at 2,000 –3,800 m (n ⫽ 41). To test whether limb lengths vary with altitude, comparisons are made of intralimb proportions, limb lengths against body mass estimates derived from published equations, limb lengths against the geometric mean of all measurements, and principal component analysis. Intralimb proportions do not statistically differ between coastal groups and those from the Atacama Desert, whereas intralimb proportions are signiﬁcantly shorter in the Peruvian highland sample. Overall body size and limb lengths relative to body size vary along an altitudinal gradient, with larger individuals from coastal environments and smaller individuals with relatively longer limbs for their size from higher elevations. Ecogeographic variation in relation to climate explains the variation in intralimb proportions, and dietary variation may explain the altitudinal cline in body size and limb lengths relative to body size. The potential effects of gene ﬂow on variation in body proportions in Andean prehistory are also explored. Am J Phys Anthropol 128:569 –585, 2005. © 2005 Wiley-Liss, Inc. Evolutionary biologists have long shown that ecogeographic patterns, such as the rules of Bergmann (1847) and Allen (1877), explain much of the variation in body size and shape among geographically widespread homeothermic species (e.g., Mayr, 1963; Porter and Gates, 1969; Searcy, 1980; Aldrich and James, 1991; Graves, 1991; Ashton et al., 2000). Populations from cold climates and high latitudes tend to be heavier with relatively short appendages compared with their conspeciﬁcs from warmer climates and lower latitudes. Heavier, shorterlimbed individuals have a smaller surface area to body mass ratio, and thus are better equipped at conserving heat under the stress of cold climatic conditions. Conversely, a larger surface area to body mass ratio enables taller, thinner individuals to shed body heat under hot climatic conditions (e.g., Calder, 1984; Schmidt-Nielson, 1984). Variation in body size and proportions in past and present human groups generally conforms to these ecogeographic patterns. Among living human populations worldwide, body mass and relative sitting height are negatively correlated with mean annual temperature and absolute latitude (Schreider, 1950, 1957, 1964, 1975; Roberts, 1953, 1978; Ruff, 1991, 1994; Katzmarzyk and Leonard, 1998). Mean annual temperature also explains regional variation in body mass and proportions among living human populations from Africa (e.g., Hiernaux and Froment, 1976), Europe and the Mediterranean (e.g., Crognier, 1981), and North and South America (e.g., Newman, 1953; Newman and Munro, 1955; Stinson, 1990). Ecogeographic patterns in body size and shape have also been central in discerning climatic adaptations in human skeletal re- mains. Relative limb proportions and bi-iliac breadth in modern human skeletons recovered from diverse geographical regions and climatic zones are correlated with mean annual temperature and latitude (Trinkaus, 1981; Jacobs, 1985, 1993; Ruff, 1991, 1994; Holliday, 1997). The association between body proportions and climate, furthermore, is used to infer climatic adaptations and phylogenetic relationships among Homo groups during the Pleistocene (Trinkaus, 1981; Ruff, 1991, 1994, 2002b; Ruff and Walker, 1993; Ruff et al., 1997; Holliday, 1997, 1999, 2000; Rosenberg et al., 1999; Pearson, 2000). In particular, the long, linear body shape of the earliest modern humans in Europe and the cold-adapted body shape of Neandertals suggest that these two contemporaneous groups had separate origins, despite living in close prox- © 2005 WILEY-LISS, INC. Grant sponsor: The Leakey Foundation; Grant sponsor: Nutter Dissertation Fellowship, College of Liberal Arts and Sciences, University of Florida; Grant sponsor: Tinker Field Research Grant, Center for Latin American Studies, University of Florida; Grant sponsor: Summer Scholar Award, Dickinson College Research and Development Committee. *Correspondence to: Karen J. Weinstein, Department of Anthropology, Dickinson College, PO Box 1773, Carlisle, PA 17013. E-mail: firstname.lastname@example.org Received 14 November 2003; accepted 9 July 2004. DOI 10.1002/ajpa.20137 Published online 13 May 2005 in Wiley InterScience (www.interscience.wiley.com). 570 K.J. WEINSTEIN imity in Western Europe during the late Pleistocene (e.g., Trinkaus, 1981; Holliday, 1997, 1999, 2000). Discerning climatic adaptations in human evolution entails comparing individual Pleistocene Homo specimens against large skeletal samples of recent H. sapiens from broadly distinct climatic zones. Comparative samples of recent human skeletons are often grouped into circumpolar, temperate, subtropical, and tropical regions, and are argued to represent the full range of modern human variation in body size and shape (Trinkaus, 1981; Ruff, 1991, 1994; Ruff et al., 1997; Holliday, 1997, 1999; Pearson, 2000). With the exception of circumpolar groups from North America, however, the majority of comparative skeletal samples come from the Old World, a sampling choice that reﬂects the assumption that climatic adaptations have a genetic basis and therefore evolved in speciﬁc populations over lengthy periods of time. Thus, limb proportions in Neandertals, the earliest H. sapiens, and other Homo specimens are compared to those of cold-adapted circumpolar recent H. sapiens on one extreme against heat-adapted, tropical groups on the other. The plethora of studies that yield similar results regarding ecogeographic variation in body proportions in recent H. sapiens and Pleistocene Homo attests to the efﬁcacy of this comparative method and to the persuasiveness of these results. Yet there have been no studies that examine variations in body proportions in closely related populations across small geographic distances or of prehistoric modern human groups from diverse environments in the Americas. Comparisons of geographically proximate and closely related populations that differ in climate might reveal important information regarding the length of time required to develop morphological adaptations to climate. Comparisons of geographically proximate groups, moreover, might elucidate ﬁne-grained clinal variations in body size and shape that are not possible to discern from comparisons of populations separated by vast geographic and temporal distances. For a number of reasons, pre-Contact modern human skeletons from different altitudes in the Andes are ideal for identifying clinal variations in body proportions and for testing hypotheses about the development of climatic adaptations in human prehistory. First, the South American Andes encompass multiple elevational zones within short geographic distances. Lowland regions along the coast of the Paciﬁc Ocean have mild tropical to subtropical climates. In contrast, climatic conditions at high altitudes (elevations of 2,500 m and greater) are typically much colder. Among many environmental stressors, ambient temperatures at high altitudes dramatically ﬂuctuate on a diurnal basis, with nighttime temperatures often falling below freezing year-round (Baker, 1960; Pawson and Jest, 1978; Barry, 1979; Billings, 1979). The Andean archaeological record reveals a cultural and ecological separation between coastal and highland populations. Human populations ﬁrst settled coastal and highland regions in the Andes at 11,000 years BP (Moseley, 2001). Coastal populations exploited the rich marine resources of the Paciﬁc Ocean, and this maritime adaptation persisted for millennia (Moseley, 2001). Highland populations, in contrast, practiced agropastoralism that required “verticality,” a land-use pattern that simultaneously exploits three elevational zones: the high puna above 4,000 m for camelid herding, mountainous valleys at 2,500 – 4,000 m for the cultivation of tubers and hardy grains, and elevations between 2,000 –2,500 m for the cultivation of maize and other less resilient crops (Murra, 1972; Moseley, 2001). The absence of marine resources in highland sites and vice versa suggest further that coastal and highland Andean groups maintained largely separate breeding populations until at least after 3,500 years BP (Moseley, 2001). Thus, the rich Andean archaeological record documents dietary variation and the degree of population interactions between coastal and highland settlements through time. Finally, living human populations indigenous to the Andes exhibit anthropometric variation in body size and proportions that is correlated, in part, with climatic conditions associated with different altitudes. Similar to other human groups from cold climates, highland Aymara and Quechua have short limbs relative to stature compared with lowland Andean populations of comparable socioeconomic status (Stinson and Frisancho, 1978; Palomino et al., 1979) and with other South American populations (Stinson, 1990), i.e., patterns that conform to Allen’s rule. Whether similar differences in body shape existed in Andean prehistory has yet to be established. This study compares limb proportions in pre-Contact human skeletons from different altitudes in the Andes in order to test whether ecogeographic variation in body proportions associated with altitude applies to human populations in Andean prehistory. Considering the longevity of human settlement in the coastal and highland Andes and that body proportions in living Andeans vary with climate and elevation, pre-Contact Andean peoples from different altitudes should exhibit variation in body proportions that reﬂect climatic adaptations. If this is the case, then preContact lowland groups should have relatively long limbs for their size, while highland groups should have relatively short limbs. If factors other than climate inﬂuence variation in body proportions, then relative limb lengths should not vary along an altitudinal gradient. These factors may include variations in diet and subsistence-related activities, exposure to various diseases that might affect growth, or large-scale population movements across vast geographic regions. MATERIALS AND METHODS Human skeletal samples To examine the relationship between body proportions and altitude, four sex-speciﬁc samples of adult human skeletons from the Andes were compared: two samples from the Paciﬁc coast, and two from the highlands (Table 1). The choice of skeletal samples was restricted to large, well-preserved collections of complete or nearly complete individuals. Time, dietary variation, and gene ﬂow are potentially confounding factors that may inﬂuence variation in body size and proportions. Thus, the archaeological record and history of each of these skeletal samples are described. Arica, Chile (AR). This sample, recovered from two distinct temporal periods in Arica, Chile and the nearby Azapa Valley, represents a lowland, south-central Andean coastal population (Table 1). The earlier specimens, which date to the Archaic period (3210 –1720 BC; Allison et al., 1984; Standen et al., 1984, 1997; Focacci and Chacón, 1989), are part of the Chinchorro cultural tradition and are biologically linked with the earliest inhabitants who settled along the south-central Andean coast at 9,000 years BP (Allison et al., 1984; Arriaza, 1995; Moseley, 2001). Faunal remains of ﬁsh, shellﬁsh, and marine birds and mammals (Bird, 1943; Llagostera, 1989; Villaxa 571 BODY PROPORTIONS IN ANDEAN PREHISTORY TABLE 1. Human skeletal samples, dates, and geographic locations Machu Picchu and Cuzco, Peru (MPC) San Pedro de Atacama, Chile (SP) Coyo-3, Solcor-3, and Quitor-6 Arica, Chile (AR) Mo-1, Mo-1/5, and Mo-1/6 AZ-140 Ancón, Peru (AC) Males Females 22 42 19 60 54 69 52 28 Dates1 15th century AD 250–1240 3210–1720 BC AD 750–1250 AD 1000–1476 Lat.2 Alt.3 Location4 13.55 22.55 2,000–3,800 2,500 FMNH, YPM MSPA 18.33 ⬍100 SMA 11.78 ⬍100 FMNH 1 See text for full discussion of published dates and sites. Absolute latitude in degrees from equator (Abate, 1991). Elevation in m above sea level (Earth Info, Inc., 1996). 4 SMA, Museo Arqueológico San Miguel de Azapa, Universidad de Tarapacá, Arica, Chile; MSPA, Museo R.P. Le Paige, San Pedro de Atacama, Chile; FMNH, Field Museum, Chicago, IL; YPM, Peabody Museum of Natural History, Yale University, New Haven, CT. 2 3 and Corrales, 1992; Arriaza, 1995) and the high frequency of auditory exostoses, an osteological condition associated with repetitive immersion in cold water (Standen et al., 1997), indicate that these Archaic period individuals subsisted primarily on the rich marine resources of the adjacent Paciﬁc Ocean. The later specimens date to AD 750 – 1250 (Table 1; Arriaza et al., 1995; Langsjoen, 1996; Standen et al., 1997). During this period, south-central coastal populations practiced a mixed subsistence economy based on the cultivation of maize, beans, and tubers, and the harvesting of marine resources (Berenguer and Dauelsberg, 1989; Arriaza et al., 1995; Langsjoen, 1996; Standen et al., 1997). The earlier Arica skeletons and the later Azapa individuals are pooled into one sample because previous analysis revealed that limb lengths and proportions do not vary between individuals from these two time periods (Weinstein, 2001). Ancón, Peru (AC). This skeletal sample represents a lowland population from the central Andean coast dating to AD 1000 –1476 (Moseley, 2001; Table 1). G.A. Dorsey excavated these individuals from a cemetery at Ancón for inclusion in the 1893 Columbia Exposition in Chicago, Illinois (Dorsey, no date). These individuals were buried in seated, ﬂexed positions and wrapped in textiles. Based on associated grave goods, these individuals most likely engaged in a mixed subsistence strategy based on marine resources and agriculture (Dorsey, no date). San Pedro de Atacama, Chile (SP). SP represents highland individuals from 2,500 m recovered from three cemeteries dated to AD 250 –1240 (Table 1; Costa, 1988; Llagostera et al., 1988; Neves and Costa, 1998; Neves et al., 1999). Although 2,500 m is at the lower limit of high altitude, San Pedro de Atacama’s elevation is high enough for ambient temperatures to drop below freezing, especially during the winter months (Earth Info, Inc., 1996). Archaeological evidence also suggests that SP represents a highland population. Studies of the paleopathological conditions in these individuals (Costa, 1988; Neves and Costa, 1998; Neves et al., 1999) indicate heavy reliance on agropastoralism, a south-central Andean highland subsistence practice since approximately 4,000 years BP, which requires verticality: the simultaneous use of elevations from 2,000 – 4,500 m (Núñez, 1991; Moseley, 2001). In addition, individuals in this sample were buried with large numbers of grave goods that originated in the highlands of Bolivia and northwest Argentina, which suggests continuous economic, cultural, and biological contact between San Pedro de Atacama populations and the high- land regions to the east (Núñez, 1991; Neves and Costa, 1998). Machu Picchu/Cuzco (MPC). The human skeletons from Machu Picchu (2,000 –2,850 m) and Cuzco (3,500 – 3,800 m) comprise the second highland sample (Table 1). These individuals date to the 15th century, and were recovered by Hiram Bingham during his 1911–1914 archaeological expeditions to southern Peru (Eaton, 1916; Bingham, 1930; Burger and Salazar, 2003). While MPC dates to a somewhat later time than the other samples, this temporal gap overlaps the time range for AC and is less than 200 years younger than that of AR and SP. The individuals from Machu Picchu were buried in caves and rock shelters in seated, ﬂexed positions, wrapped in textiles and rope, and associated with llama and alpaca remains that are presumed to have been sacriﬁced at the time of the individual’s death (Miller, 2003). The burials often contain more than one individual (Eaton, 1916; Verano, 2003), which can be explained through a variety of taphonomic factors, including Inca rituals regarding ancestor veneration and treatment of the deceased, soil acidity and cave humidity, and postmortem activities of carnivores and rodents (Miller, 2003; Verano, 2003). Bone chemistry analysis and dental pathological conditions indicate that MPC individuals consumed a diet high in maize (Burger et al., 2003; Verano, 2003). Much has been speculated about the geographic origins and ethnic identities of Inca subjects in general and the inhabitants of Machu Picchu in particular (e.g., Eaton, 1916; Bingham, 1930; Moseley, 2001; Burger and Salazar, 2003). According to Andean archaeologists, the Inca implemented a number of social policies that functioned to subdue their subjects and conquer new territories. For example, the Inca would resettle entire communities into newly colonized territories, and adult males from throughout the Andes were drafted as labor for large-scale construction projects and military service (Moseley, 2001). Dietary variation based on the stable carbon and nitrogen isotope content of human bone suggests that Machu Picchu individuals were not exclusively highland in origin (Burger et al., 2003). Cranial deformation styles and the analysis of biological distance based on craniometrics also suggest that Machu Picchu inhabitants represented diverse populations from throughout the Andes (Verano, 2003). Multivariate craniometrics on artiﬁcially deformed skulls, however, are not especially reliable methods for assessing biological distance between populations, given that cranial shape undergoes marked changes in response 572 K.J. WEINSTEIN Fig. 1. Intralimb proportions in males: log RAD vs. log HUM (A) and log TIB vs. log FEM (B). Here and in all subsequent ﬁgures, individual data points are labeled according to following symbols: ●, MPC; X, SP; ⫹, AR; ‚, AC. Note common scatter of SP, AR, and AC, while most of MPC falls below regression line. to artiﬁcial deformation (Antón, 1989). Nevertheless, caution is required when interpreting body proportions in this sample. If MPC represents people from across the Andes, then body proportions in this sample should exhibit the largest range of variation and should overlap the ranges of skeletal samples from other elevations. MPC individuals who are highland in origin should cluster with SP, the other highland sample, whereas MPC individuals who are lowland should exhibit similar body proportions as those from AR and AC. Osteometric measurements Maximum lengths of the humerus (HUM), radius (RAD), femur (FEM) (bicondylar length), and tibia Fig. 2. Intralimb proportions in females: log RAD vs. log HUM (A) and log TIB vs. log FEM (B). See Figure 1 for explanation of symbols. Similar to patterns seen in males, female SP, AR, and AC have virtually indistinguishable intralimb proportions, whereas those of MPC tend to be shorter. (TIB), as deﬁned in Martin (1928), were measured to the nearest 0.5 mm with an osteometric board, and femoral head breadth (FHB) to the nearest 0.1 mm with Mitutoyo digital sliding calipers. FHB is deﬁned as the average of the antero-posterior and supero-inferior diameters of the femoral head (Grine et al., 1995). Since articular surfaces of weight-bearing joints are correlated with body weight (e.g., Ruff et al., 1991; Lieberman et al., 2001), FHB provides an appropriate variable to estimate individual body mass and to compare against relative limb lengths. Statistical analyses Intralimb proportions. Previous studies indicate that intralimb proportions, which represent the relationship between the proximal and distal limb segments of the 573 BODY PROPORTIONS IN ANDEAN PREHISTORY TABLE 2. Mean body mass estimates in kilograms1 Sample2 Males Females MPC SP AR AC Scheffe’s test3 58.60 ⫾ 1.53 61.34 ⫾ 0.76 62.27 ⫾ 0.75 66.36 ⫾ 0.83 MPC ⬍ AC; SP ⬍ AC, AR ⬍ AC 49.96 ⫾ 1.13 52.74 ⫾ 0.59 55.13 ⫾ 0.60 56.83 ⫾ 1.00 MPC ⬍ AR; MPC ⬍ AC; SP ⬍ AR; SP ⬍ AC 1 Mean and standard error of body mass estimates are based on average of three regression equations for each individual (Ruff et al., 1991; McHenry, 1992; Grine et al., 1995). 2 See Table 1 for sample abbreviations. 3 Scheffe’s test results compare mean BM between two samples. Inequality indicates signiﬁcant difference (P ⱕ 0.05) in direction shown. All other comparisons are not signiﬁcant (P ⬎ 0.05). Fig. 3. Mean BM with two standard errors in sex-speciﬁc groups by site. Note progressive decrease in estimated body mass with increasing altitude in both males and especially females. In addition to this general pattern, Scheffe’s test results indicate that AC males are signiﬁcantly larger in estimated BM compared with males from MPC, SP, and AR. Among females, AR and AC are signiﬁcantly larger compared with both MPC and SP. upper and lower limbs, vary with climate in human populations due to shortening of the distal segments in coldadapted populations (e.g., Trinkaus, 1981; Ruff, 1991, 1994, 2002b; Holliday, 1997, 1999; Holliday and Ruff, 2001). In order to test whether intralimb proportions also vary according to altitude, RAD is compared with HUM and TIB with FEM in sex-speciﬁc groups according to site using bivariate scatter plots (Figs. 1, 2). Limb lengths and overall size. In order to examine individual variation in limb lengths relative to body size and to test whether this variation is associated with altitude, it is necessary to choose an appropriate estimate of body size. Methods for determining overall size in skeletal remains include estimating body mass from regression equations based on femoral head breadth (e.g., Ruff et al., 1991, 1997; McHenry, 1992; Grine et al., 1995), reconstructing overall body shape based on regression equations of bi-iliac breadth/ stature (e.g., Ruff, 1994, 2002b; Ruff et al., 1997), and estimating overall size by calculating the geometric mean of all variables measured on each individual (e.g., Darroch and Mosimann, 1985; Falsetti et al., 1993; Jungers et al., 1995). This study applies two of these methods: body mass (BM) estimated from FHB, and overall size estimated from the geometric mean (GM). Reconstructing body shape from biiliac breadth/stature is not possible in this study because a large number of individuals lack pelvic remains. Thus calculating bi-iliac breadth/stature would require the omission of too many individuals within each sample, rendering comparisons across samples meaningless. Estimating BM is based on the following four regression equations: BM共kg兲 ⫽ 2.268 ⫻ FHB ⫺ 36.5 (Grine et al., 1995) BM ⫽ 2.239 ⫻ FHB ⫺ 39.9 (McHenry, 1992) BM ⫽ 2.741 ⫻ FHB ⫺ 54.9 for males (Ruff et al., 1991) BM ⫽ 2.426 ⫻ FHB ⫺ 35.1 for females (Ruff et al., 1991) Each of these equations was developed by regressing FHB against body weight in particular reference samples. Ruff et al. (1997) found that while the equations of Ruff et al. (1991) and Grine et al. (1995) tend to slightly overestimate body mass compared with that of McHenry (1992), the mean difference between these three equations was only 4%. Using Ruff et al. (1997) as an example, in which the authors apply these equations to Pleistocene Homo, BM is estimated by averaging the results of each equation for each individual grouped by site (Table 2, Fig. 3). In order to test whether estimated body mass differs by site, ANOVA and Scheffe’s pairwise comparisons of mean BM are performed on sexspeciﬁc groups where group means are considered signiﬁcantly different at P ⱕ 0.05 (Table 2, Fig. 3). Limb lengths (HUM, RAD, FEM, and TIB) are then plotted against BM (Figs. 4, 5) in order to determine whether limb lengths relative to BM vary in individuals from different sites and altitude. Relative limb lengths are also compared against GM, an estimate of overall size that is independent from the methods described above. Although not equivalent to body weight, GM estimates overall size based on all of the variables measured for each individual (e.g., Mosimann and James, 1979; Darroch and Mosimann, 1985; James and McCulloch, 1990; Falsetti et al., 1993; Jungers et al., 1995). GM is calculated based on the following ﬁve variables: HUM, RAD, FEM, TIB, and FHB, according to the following equation (Sokal and Rohlf, 1995): Gm y ⫽ antilog 1/n 冘 log Y Similar to previous analyses, each limb length (dependent variable) is plotted against GM (independent variable) for each individual in sex-speciﬁc groups according to site (Figs. 6, 7). Tests for differences in proportions. All bivariate scatterplots in this study are illustrated with the reduced major axis (RMA) regression line based on individuals grouped by sex. RMA regression models are appropriate for these comparisons because the independent variables are measured with error (Sokal and Rohlf, 1995). For each scatterplot that compares intralimb proportions and limb lengths against BM and GM, tests for elevational differences between samples are undertaken using the nonparametric “quick test” of Tsutakawa and Hewett (1977) (Table 3). This test reveals whether or not body proportions in each sample signiﬁcantly differ from other groups by identifying the number of individuals from each sample who fall above or below a common RMA regression line through the pooled sample. Samples are considered to 574 K.J. WEINSTEIN Fig. 4. Limb lengths vs. estimated BM in males: HUM (A), RAD (B), FEM (C), and TIB (D). See Figure 1 for explanation of symbols. Note differences between SP, which tends to fall above regression line, compared with AR and especially AC, which tend to fall below line. MPC males tend to scatter above and below regression line in all variables except for HUM (A). signiﬁcantly differ from each other with respect to their position above or below the common regression line at P ⱕ 0.05. As an independent method for comparing proportional differences between groups in both intralimb proportions and comparisons of limb lengths against BM and GM, log-transformed indices are generated according to the methods described in Ruff (2002a). Each index is calculated as log(Y/Xb), where Y is the dependent variable in the regression model, X is the independent variable, and b is the predicted slope if the independent and dependent variables scale isometrically. In the case of intralimb proportions and limb lengths against GM, the predicted isometric slope is 1.0, whereas in the comparisons of limb lengths against BM, the predicted isometric slope is 0.666. As in Ruff (2002a), mean log(Y/Xb) values are compared between samples, using ANOVA and Tukey multiple comparison tests (Tables 4 – 6). Group means are considered signiﬁcantly different at P ⱕ 0.05. Principal component analysis (PCA). I also test further for differences in body proportions by site and altitude through PCA of an unrotated variance covariance matrix of the ﬁve variables (HUM, RAD, FEM, TIB, and FHB; Table 7). The ﬁrst PCA axis generally captures individual variation due to differences in size, while the second PCA axis tends to explain individual variation resulting from differences in shape while controlling for size (e.g., Darroch and Mosimann, 1985; Falsetti et al., 1993; Jungers et al., 1995). Thus, the second PCA axis should separate individuals from different altitudes if body proportions vary according to altitude. Alternatively, BODY PROPORTIONS IN ANDEAN PREHISTORY 575 Fig. 5. Limb lengths vs. estimated BM in females: HUM (A), RAD (B), FEM (C), and TIB (D). See Figure 1 for explanation of symbols. Note highland/lowland contrast based on tendency of highland groups (SP and MPC) to fall above regression line, and lowland groups (AR and especially AC) to fall below. if altitudinal variation does not explain differences in overall size or relative limb proportions, then each group should be indistinguishable along this axis. This variation is illustrated by plotting individual factor scores (Fig. 8). All variables were transformed into their natural logarithms for all statistical analyses. All statistical tests were completed using Systat version 8.0 for Windows (SPSS, Inc., 1998). RESULTS Intralimb proportions Comparisons of intralimb proportions reveal two general patterns. First, SP males and females have intralimb proportions that are generally similar to the two lowland samples, which are generally similar to each other. Scatterplots illustrate that intralimb proportions overlap between SP, AR, and AC (Figs. 1, 2), and this pattern is also evident in the Tsutakawa and Hewett nonparametric quick test results (Table 3). Among males in these three groups, all pairwise comparisons in individuals above and below the RMA regression lines are not signiﬁcant, with the exception of larger SP compared with AR in TIB relative to FEM. Among females of these three groups, all pairwise comparisons are not signiﬁcant except for AC females, which are larger in RAD relative to HUM compared with AR, and larger in TIB relative to FEM compared with SP. Tukey multiple comparison tests of log-transformed indices (Table 4) also indicate that SP males and females do not signiﬁcantly differ from their lowland counterparts in upper and lower intralimb proportions. 576 K.J. WEINSTEIN Fig. 6. Limb lengths vs. GM in males: HUM (A), RAD (B), FEM (C), and TIB (D). See Figure 1 for explanation of symbols. Note that despite considerable overlap between groups, MPC and SP tend to separate slightly from AC. Second, MPC has markedly shorter RAD and TIB relative to HUM and FEM, respectively, compared with the other three samples. Although their range of variation overlaps the other groups, MPC consistently tends to fall below the regression line in comparisons of the distal segment against the proximal segment for both the upper and lower limbs (Figs. 1, 2). Quick test comparisons also indicate that, with some exceptions, MPC males and females exhibit signiﬁcantly shorter RAD and TIB relative to HUM and FEM compared with the other groups (Table 3). Male MPCs, for example, have shorter intralimb proportions compared with all other groups except for TIB relative to FEM in AR. Female MPCs have signiﬁcantly shorter TIB relative to FEM compared with AR and AC, but do not differ from the lowland groups in RAD relative to HUM or in any intralimb proportions compared with SP (Table 3). The Tukey multiple comparison tests of intralimb proportion indices reveal even more pronounced differences between MPC and the other three samples: MPC males and females are consistently signiﬁcantly shorter in distal to proximal limb length proportions for both the upper and lower limbs compared with all other groups (Table 4). Limb lengths relative to body size Reconstruction of body mass using the equations of McHenry (1992), Grines et al. (1995), and Ruff et al. (1997) indicate that the two highland groups are lighter than the lowlanders in terms of mean BM (Table 2, Fig. 3). Among males, MPC has the smallest estimated body masses, followed by SP, whose range of BM overlaps considerably BODY PROPORTIONS IN ANDEAN PREHISTORY 577 Fig. 7. Limb lengths vs. GM in females: HUM (A), RAD (B), FEM (C), and TIB (D). See Figure 1 for explanation of symbols. As in male comparisons, there is considerable overlap across samples, except for HUM (A), in which MPC and SP are slightly separated from AC in particular. with that of AR. Male ACs are the heaviest in BM and do not overlap the others. One-way ANOVA and Scheffe’s tests that compare mean BM among males reveal that despite the progressive decrease in BM with increasing altitude, the only group to signiﬁcantly differ from the others is AC (Table 2, Fig. 3). Among females, MPC exhibits the smallest BM, which overlaps that of SP, while AR and AC exhibit overlapping ranges which, based on one-way ANOVA and Scheffe’s tests, are signiﬁcantly heavier than those of the two highland groups (Table 2, Fig. 3). Limb lengths relative to BM also tend to vary across an altitudinal gradient, although in a pattern that does not conform to the predictions of Allen’s rule. The smallerbodied highland samples tend to have relatively long limbs for their size, whereas the larger-bodied lowland samples tend to have relatively shorter limbs. This highland/lowland contrast in limb lengths against BM is perhaps most apparent in comparisons of SP with AC, in which SP males and females fall predominantly above the regression line for all limb lengths against BM, while AC males and females are positioned below the regression line for these same variables (Figs. 4, 5). Tsutakawa and Hewett quick test results (Table 3) and Tukey multiple comparisons of mean log-transformed indices of limb lengths against BM (Table 5) further support the distinction between SP and AC, in which SP males and females consistently exhibit long limbs for their relatively small body masses compared with larger-bodied ACs. Tukey multiple comparisons of mean log-transformed indices indicate that, with the exception of HUM and RAD in males, SP and AR signiﬁcantly differ in all limb lengths relative 578 K.J. WEINSTEIN TABLE 3. Bivariate positional comparisons of samples using “quick test”1 2 Measurement Males RAD v HUM TIB v FEM HUM v BM RAD v BM FEM v BM TIB v BM HUM v GM RAD v GM FEM v GM TIB v GM Females RAD v HUM TIB v FEM HUM v BM RAD v BM FEM v BM TIB v BM HUM v GM RAD v GM FEM v GM TIB v GM MPC/SP MPC/AR MPC/AC MPC ⬍ SP MPC ⬍ SP ns ns ns ns ns ns ns ns MPC ⬍ AR ns ns ns ns ns ns ns MPC ⬎ AR ns MPC ⬍ AC MPC ⬍ AC ns ns ns ns ns ns MPC ⬎ AC ns ns SP ns ns SP SP ns ns SP SP ns ns ns ns ns ns MPC ⬎ SP ns ns ns ns MPC ⬍ AR ns ns ns ns ns ns MPC ⬎ AR ns ns MPC MPC ns ns MPC MPC MPC ns MPC ns ns SP ns SP SP ns ns SP ns ⬍ AC ⬎ AC ⬎ AC ⬎ AC ⬎ AC ⬎ AC SP/AR ⬎ AR ⬎ AR ⬎ AR ⬎ AR ⬎ AR ⬎ AR ⬎ AR ⬎ AR ⬎ AR SP/AC ns ns SP SP SP SP SP SP SP ns ⬎ ⬎ ⬎ ⬎ ⬎ ⬎ ⬎ AC AC AC AC AC AC AC ns SP SP SP SP SP SP SP SP ns ⬍ ⬎ ⬎ ⬎ ⬎ ⬎ ⬎ ⬎ AC AC AC AC AC AC AC AC AR/AC ns ns AR AR ns ns AR AR ns ns ⬎ AC ⬎ AC ⬎ AC ⬎ AC AR ⬍ AC ns AR ⬎ AC ns ns ns AR ⬎ AC ns ns ns 1 ns, equality of position about RMA line in two samples. Inequality indicates signiﬁcant difference (P ⬍ 0.05) about RMA line in direction shown. 2 See text for measurement deﬁnitions and abbreviations. See Table 1 for sample abbreviations. TABLE 4. Log-transformed indices and Tukey multiple comparisons of intralimb proportions Ratio Sample n Mean SD Tukey test results Males log (RAD/HUM) MPC SP AR AC MPC SP AR AC MPC SP AR AC MPC SP AR AC 13 41 38 52 18 38 48 51 10 56 56 28 12 54 66 28 ⫺0.282 ⫺0.244 ⫺0.244 ⫺0.251 ⫺0.181 ⫺0.154 ⫺0.153 ⫺0.155 ⫺0.293 ⫺0.263 ⫺0.262 ⫺0.256 ⫺0.182 ⫺0.161 ⫺0.155 ⫺0.145 0.045 0.035 0.039 0.028 0.026 0.019 0.029 0.026 0.046 0.030 0.029 0.027 0.025 0.023 0.023 0.023 MPC ⬍ SP, P ⫽ 0.003 MPC ⬍ AR, P ⫽ 0.001 MPC ⬍ AC, P ⫽ 0.02 Males log (TIB/FEM) Females log (RAD/HUM) Females log (TIB/FEM) to BM (Table 5), although the Tsutakawa and Hewett quick test results reveal that these differences are significant only in the comparisons of lower limbs in males and in all limbs except RAD in females (Table 3). In general, MPC males and especially females have small body masses, and in terms of limb lengths relative to BM, MPC females particularly tend to cluster near the smallest SP individuals (Figs. 4, 5). Tukey multiple comparison tests, however, indicate that indices of limb lengths with BM in MPC signiﬁcantly differ with AC only (Table 5), although quick test results show that these differences are only signiﬁcant between MPC and AC females for comparisons of HUM and TIB against BM (Table 3). All four limb lengths scale isometrically with GM in both males and females. Conﬁdence intervals of the regression slopes for the scatterplots of HUM, RAD, FEM, and TIB include 1.0, and all groups, including MPC, exhibit considerable overlap (Figs. 6, 7). A progressive decrease from highland to lowland in limb lengths relative to overall size, however, is apparent when comparing mean log-transformed indices of limbs and GM (Table 6). Male and female MPCs have the largest mean ratios of MPC ⬍ SP, P ⫽ 0.001 MPC ⬍ AR, P ⬍ 0.001 MPC ⬍ AC, P ⫽ 0.001 MPC ⬍ SP, P ⫽ 0.02 MPC ⬍ AR, P ⫽ 0.02 MPC ⬍ AC, P ⫽ 0.004 MPC ⬍ SP, P ⫽ 0.02 MPC ⬍ AR, P ⫽ 0.002 MPC ⬍ AC, P ⬍ 0.001 log(HUM/GM) and log(FEM/GM), followed by SP, AR, and then AC, and Tukey multiple comparison tests show that many of these mean values, especially between SP and AC, signiﬁcantly differ. This highland/lowland progressive decrease in limb length relative to GM is less apparent in comparisons of RAD and TIB: mean values are more randomly distributed between groups, and most of the group mean differences, as revealed through Tukey multiple comparison tests, separate SP from AR and AC, and AC from MPC. Quick test results that examine pairwise comparisons of individuals across RMA regression lines of limb lengths against GM also indicate varied results (Table 3). Among males, MPC signiﬁcantly differs from AR and AC in FEM; SP signiﬁcantly differs from AR in FEM and TIB, and from AC in HUM, RAD, and FEM; and AR and AC signiﬁcantly differ from each other in HUM and RAD. Among females, MPC signiﬁcantly differs from SP in HUM, from AR in FEM, and from AC in all variables except for FEM (Table 3). Female SPs also signiﬁcantly differ in their RMA line position from ARs in terms of FEM and from ACs in all variables except TIB. Finally, AR and AC females differ in their positions of HUM. Thus, 579 BODY PROPORTIONS IN ANDEAN PREHISTORY TABLE 5. Log-transformed indices and Tukey multiple comparisons of limb lengths against BM Ratio Sample n Mean SD Males log (HUM/BM) MPC SP AR AC MPC SP AR AC MPC SP AR AC MPC SP AR AC MPC SP AR AC MPC SP AR AC MPC SP AR AC MPC SP AR AC 10 40 42 51 10 41 39 51 16 39 49 51 15 39 47 51 9 53 61 28 8 55 58 28 12 54 68 28 10 52 66 28 2.970 2.968 2.953 2.909 2.715 2.725 2.710 2.659 3.290 3.295 3.258 3.235 3.103 3.138 3.105 3.080 2.992 2.990 2.954 2.904 2.717 2.726 2.692 2.649 3.318 3.321 3.281 3.240 3.151 3.160 3.124 3.094 0.021 0.051 0.054 0.060 0.077 0.053 0.060 0.062 0.055 0.050 0.050 0.051 0.054 0.054 0.059 0.058 0.051 0.048 0.054 0.058 0.043 0.050 0.056 0.062 0.047 0.049 0.049 0.064 0.047 0.053 0.052 0.060 Males log (RAD/BM) Males log (FEM/BM) Males log (TIB/BM) Females log (HUM/BM) Females log (RAD/BM) Females log (FEM/BM) Females log (TIB/BM) Tukey test results AC ⬍ MPC, P ⫽ 0.006 AC ⬍ SP, P ⬍ 0.001 AC ⬍ AR, P ⫽ 0.001 AC ⬍ MPC, P ⫽ 0.035 AC ⬍ SP, P ⬍ 0.001 AC ⬍ AR, P ⬍ 0.001 AC ⬍ MPC, P ⫽ 0.001 AR ⬍ SP, P ⫽ 0.005 AC ⬍ SP, P ⬍ 0.001 AR ⬍ SP, P ⫽ 0.038 AC ⬍ SP, P ⬍ 0.001 AC AR AC AC AC AR AC AC AC AR AC AC AC AR AC AC ⬍ ⬍ ⬍ ⬍ ⬍ ⬍ ⬍ ⬍ ⬍ ⬍ ⬍ ⬍ ⬍ ⬍ ⬍ ⬍ MPC, P ⫽ 0.001 SP, P ⫽ 0.001 SP, P ⬍ 0.001 AR, P ⬍ 0.001 MPC, P ⫽ 0.01 SP, P ⫽ 0.005 SP, P ⬍ 0.001 AR, P ⬍ 0.003 MPC, P ⬍ 0.001 SP, P ⬍ 0.001 SP, P ⬍ 0.001 AR, P ⫽ 0.002 MPC, P ⫽ 0.022 SP, P ⫽ 0.002 SP, P ⬍ 0.001 AR, P ⬍ 0.074 TABLE 6. Log-transformed indices and Tukey multiple comparisons of limb lengths against GM Ratio Sample n Mean SD Males log (HUM/GM) MPC SP AR AC MPC SP AR AC MPC SP AR AC MPC SP AR AC MPC SP AR AC MPC SP AR AC MPC SP AR AC MPC SP AR AC 9 38 35 51 9 38 35 51 9 38 35 51 9 38 35 51 6 50 54 28 6 50 54 28 6 50 54 28 6 50 54 28 0.352 0.338 0.338 0.328 0.079 0.093 0.101 0.077 0.665 0.663 0.647 0.653 0.491 0.509 0.493 0.498 0.347 0.346 0.340 0.324 0.070 0.083 0.079 0.068 0.684 0.678 0.667 0.659 0.512 0.517 0.511 0.514 0.019 0.019 0.024 0.021 0.033 0.021 0.025 0.024 0.010 0.015 0.020 0.013 0.025 0.018 0.023 0.023 0.016 0.019 0.021 0.016 0.023 0.020 0.023 0.021 0.013 0.016 0.018 0.020 0.016 0.021 0.018 0.016 Males log (RAD/GM) Males log (FEM/GM) Males log (TIB/GM) Females log (HUM/GM) Females log (RAD/GM) Females log (FEM/GM) Females log (TIB/GM) Tukey test results AC ⬍ MPC, P ⫽ 0.008 MPC ⬍ AR, P ⫽ 0.075 AC ⬍ SP, P ⫽ 0.015 AC ⬍ AR, P ⬍ 0.001 AR ⬍ MPC, P ⫽ 0.01 AR ⬍ SP, P ⬍ 0.001 AC ⬍ SP, P ⫽ 0.025 AR ⬍ SP, P ⫽ 0.013 AC ⬍ SP, P ⫽ 0.11 AC ⬍ MPC, P ⫽ 0.03 AC ⬍ SP, P ⬍ 0.001 AC ⬍ AR, P ⫽ 0.001 AC ⬍ SP, P ⫽ 0.016 AC ⬍ MPC, P ⫽ 0.01 AC ⬍ SP, P ⬍ 0.001 AR ⬍ SP, P ⫽ 0.007 No signiﬁcant differences 580 K.J. WEINSTEIN TABLE 7. PCA in males and females Eigenvector coefﬁcients Males RAD FHB FEM TIB HUM Eigenvalue % of total variance Females Axis I Axis II Axis I Axis II 0.505 0.350 0.446 0.485 0.443 0.009 72.27 0.259 ⫺0.931 0.080 0.199 0.145 0.002 16.59 0.505 0.287 0.430 0.500 0.477 0.008 71.26 0.131 ⫺0.955 0.181 0.101 0.167 0.002 19.19 while the scatterplots illustrate considerable overlap between sites in limb lengths relative to GM, statistical tests reveal a subtle altitudinal cline from smaller-bodied highland groups to larger-bodied lowlanders similar to that seen in comparisons of limb lengths relative to BM. From smallest to largest, MPC and SP tend to have relatively long limbs, and the larger AR and AC tend to have relatively shorter limbs. PCA The PCA reveals the same contrasts between MPC and the other groups in terms of relative limb lengths, and between the two highland groups and the lowlanders in terms of differences in overall size. The ﬁrst PC axis explains approximately 71–72% of the total sample variance for males and females (Table 7). This axis represents differences in overall size: all eigenvector coefﬁcients load strongly and positively along this axis. Individual factor scores, moreover, are highly correlated with GM (r2 ⫽ 0.999, P ⬍ 0.001). Similar to the bivariate regression analyses, MPC tends to plot negatively along this axis: all but three males have slightly negative factor scores, and all but one female have markedly negative scores (Fig. 8). Male and female SP are virtually indistinguishable from the lowland groups along the ﬁrst axis. While the separation between MPC and the other three groups is not especially marked, this axis shows that MPC falls at the small end of the size range for these variables. The second PCA axis accounts for approximately 16 – 19% of the total sample variance (Table 7). This axis contrasts FHB from limb lengths. FHB loads strongly and negatively along this axis, while all four limbs, and especially TIB and RAD, score positively. This second axis separates highland groups from the lowland samples. In particular, SP males and females tend to plot positively along this axis and are separate from those of AC, who score negatively, especially AC females. Male and female ARs appear to be intermediate between these two extremes (Fig. 8). All but two males and one female from MPC exhibit positive factor scores along this axis, and tend to cluster within the SP range of variation. Thus, the second axis illustrates a subtle altitudinal cline in femoral head size and limb lengths with the lighter individuals from highland regions and the heavier individuals from the lowlands. DISCUSSION In these samples, body size and proportions vary across an altitudinal gradient in two patterns. First, individuals from the highest altitudes (MPC) have the shortest relative intralimb length proportions, whereas those from slightly lower elevations at 2,500 m (SP) have intralimb length proportions that are indistinguishable from the lowland groups, a pattern that is perhaps best explained as a response to climatic stress. Second, overall size and limb lengths relative to overall size tend to vary across an elevational cline, with the largest individuals with relatively shorter limbs from the central (AC) and southcentral (AR) coasts, and smaller individuals with relatively longer limbs for their size from 2,500 m (SP) and 2,000 –3,800 m (MPC). Unlike intralimb length proportions, the highland/lowland contrasts in overall size and limb lengths relative to overall size do not follow a pattern indicative of climatic stress. In this sample, the largest individuals with relatively short limbs for their size are from coastal regions with mild climates rather than from higher elevations with harsher ambient temperatures as predicted by Bergmann’s and Allen’s rules. While a variety of biological and ecological factors may be at work, nutritional and dietary variation may best explain this subtle contrast in body size and limb lengths relative to overall size between highland and lowland groups. Considering that gene ﬂow may be a confounding factor, archaeological evidence for widespread movements among the populations from which these samples were derived is also explored. Intralimb proportions, climatic adaptations, and biological diversity Morphological adaptations to climate perhaps best explain the short intralimb proportions in MPC. Given that MPC individuals were recovered from approximately 2,000 –3,800 m in elevation where ambient temperatures frequently fall below 0°C, and that living Andeans from high altitudes are presumed to be cold-adapted (Stinson and Frisancho, 1978), the short relative intralimb proportions in this highland skeletal sample most likely reﬂect a cold adaptation, as predicted by Allen’s rule. Furthermore, comparisons of pre-Contact human skeletons from the highland Andes (data from MacCurdy, 1923) with those from North American populations from Arctic climates that are known to be cold-adapted (data from Trinkaus, 1981) illustrate that highland Andeans have intralimb proportions that are equally as short as Eskimo/Inuit groups, demonstrating further the development of a coldadapted body shape among prehistoric highland Andean populations (Weinstein, 1998, 2001). In contrast, relative intralimb proportions in SP are virtually indistinguishable from individuals from lowland environments. This contrast between MPC and SP suggests that climatic conditions in the southern Peruvian highlands are severe enough to favor the selection of a cold-adapted body shape, whereas temperatures in the Atacama Desert of northern Chile are less severe. Climatic adaptations in Andean prehistory, as inferred from intralimb proportions, are useful to debates about the evolution of modern human biological diversity and the worldwide dispersal of modern humans during the Pleistocene. As discussed earlier, numerous paleoanthropological studies compare intralimb proportions of recent Homo sapiens from different climates in order to discern climatic adaptations, migrations, and phylogenetic relationships in human groups during the Pleistocene (e.g., Trinkaus, 1981; Ruff, 1991, 1994, 2002b; Ruff and Walker, 1993; Holliday 1997, 1999, 2000; Pearson, 2000). With the exception of skeletal samples from Arctic regions in North America, however, recent human skeletons from prehistoric North and South America are rarely included in comparative studies of climatic adaptation in human evo- BODY PROPORTIONS IN ANDEAN PREHISTORY 581 proportions within MPC and signiﬁcantly greater proportions in groups from lower elevations, while admittedly less pronounced than the contrasts between Old World populations that are separated by vast geographic and temporal distances, indeed suggest that climatic adaptations are one of many selective factors that shape the biology of prehistoric Andeans in particular: results that are similar to Newman (1953, 1960) and Beals et al. (1984) regarding climatic adaptations in the Americas in general. Archaeological evidence suggests that humans have inhabited South America since at least 12,500 years BP and potentially earlier (e.g., Dillehay, 2000; Roosevelt et al., 2002). Moreover, by 11,000 years BP, regional populations developed diverse cultural adaptations to their unique environments, including the Andean highlands and coast, temperate forests, and tropical lowlands (Dillehay, 1997; Moseley, 2001; Roosevelt et al., 2002). Dillehay (2000) emphatically argue that the peopling of the Americas should be interpreted as part of the worldwide dispersal of Homo sapiens that began ca. 40,000 years ago, that there is great time depth of Amerindian groups in widely varied environments, and that prehistoric populations within the Americas developed a large degree of cultural diversity. The results of this study suggest that prehistoric Amerindian populations should exhibit a similar degree of biological diversity as well. Given the diverse environments that the earliest Americans colonized, including high altitudes, arid deserts, and tropical and temperate regions, it seems plausible that prehistoric populations of the Americas would develop biological adaptations to suit these environmental extremes. Future research should continue to examine body proportions across a number of populations from the Americas to more readily capture the degree of biological diversity and clinal variation in body size and proportions that most surely existed within the pre-Contact period. Body size, relative limb lengths, and dietary variation Fig. 8. Individual factor scores for ﬁrst and second axes of ﬁve-variable PCA in males (A) and females (B). See Figure 1 for explanation of symbols. Axis I represents size differences. Note that with exception of three males and one female, MPC plots negatively along this axis, especially in females. Axis II contrasts FHB with limb lengths. Although there is some overlap in range of variation, SP tends to plot positively along this axis, with MPC falling within same range. AC tends to plot negatively, and AR appears intermediate in relation to others. lution. Moreover, Amerindian groups which are included as comparative samples in studies of intralimb proportions in both living populations (e.g., Eveleth and Tanner, 1976, 1990) and in skeletal samples (e.g., Trinkaus, 1981; Ruff, 1991, 1994; Holliday, 1997, 1999, 2000; Pearson, 2000) are generally categorized as “Northern Asian” or “Asiatic.” While never explicitly stated, “Northern Asian” and other similar labels imply that North and South Americans are biologically unchanged from the earliest inhabitants of the New World who are presumed to be cold-adapted. Yet reduced distal/proximal limb-length The highland groups (MPC and SP) tend to be smaller in overall size with relatively longer limbs, while the lowland groups (AR and AC) tend to be larger with relatively shorter limbs: a pattern that runs counter to Bergmann’s and Allen’s rules. While a number of factors could be at work, studies of living populations at high altitudes suggest that dietary variation and its differential effects on growth may offer the most robust explanation for the highland/lowland contrasts in body size and limb lengths relative to body size in these samples. Living highland Andeans exhibit delayed growth and short adult stature compared with lowland Andean populations and with European migrants to high altitudes (e.g., Frisancho, 1978; Greksa et al., 1984, 1985; Greksa, 1986). High-altitude hypoxia is one potential environmental factor that can adversely affect growth during the prenatal period, infancy, and childhood, leading to small body sizes and short stature in adulthood (Frisancho and Baker, 1970; Beall et al., 1977; Mueller et al., 1978, 1980; Stinson, 1980, 1982; Greksa, 1986; Bogin, 1999) Among living highland populations in the Andes, however, limited nutrition and low energy availability (two biocultural stressors that are also characteristic of high-altitude environments) may more profoundly inﬂuence the small body sizes in the highland groups. In highland southern Peru and in Ecuador, individuals with low socioeconomic status experience annual and seasonal shortages of ade- 582 K.J. WEINSTEIN quate food resources, leading to delayed growth during childhood and adolescence, which are interpreted as the primary cause of short adult stature in these populations (e.g., Leonard, 1989; Leonard et al., 1990, 1995, 2000). Nutritional stress, exposure to infectious diseases, and their adverse affects on growth also explain highland/ lowland contrasts in body sizes in living populations from Ethiopia (e.g., Harrison et al., 1969; Clegg et al., 1972) and the Himalayas (e.g., Bharadwaj et al., 1973; Gupta and Basu, 1981). Furthermore, ﬂuctuations in dietary quality, disease load, and health status led to secular changes in adult stature and body mass in Central and South American populations over the last 8,000 years (Bogin and Keep, 1999), in the United States over the last 200 years (Meadows Jantz and Jantz, 1999), and in populations from developing nations over the last 40 years (Katzmarzyk and Leonard, 1998). Limb lengths relative to overall size may also be affected by diet and nutritional status. Among living populations suffering from mild to moderate chronic undernutrition, limb lengths tend to be short relative to stature (Bogin, 1999). When undernourished groups experience improved diets, however, lower limb lengths tend to increase relative to stature, effectively increasing lower limb proportions (Bogin, 1999; Stinson, 2000). Thus, living human populations experiencing both chronic mild undernutrition and improved diets demonstrate the plasticity of body size and of limb lengths relative to body size in response to nutritional conditions. The altitudinal cline in overall body size in human skeletons from the pre-Contact Andes in this study follows a pattern similar to that of living human populations with different diets and energy expenditure. Archaeological evidence infers that the highland and lowland groups differed in their diets and subsistence base. The larger-bodied lowlanders relied heavily on marine resources as a dietary staple, with an increase in terrestrial resources through time (e.g., Arriaza, 1995; Moseley, 2001). Individuals from AC exhibited rather high frequencies of cribra orbitalia and porotic hyperostosis (personal observation), osteological conditions that are indicative of iron-deﬁciency anemia (Stuart-Macadam, 1992) and that are common in pre-Contact coastal groups exposed to parasites from marine-based diets (Walker, 1986). The other coastal group from AR, however, does not exhibit these pathological conditions in high frequency (personal observation). Highland MPC and SP, in contrast, consumed maize, tubers, and camelid meat (e.g., Costa, 1988; Núñez, 1991; Costa Junquiera and Llagostera, 1994; Miller, 2003; Burger et al., 2003). Both highland groups exhibit relatively low frequencies of pathological conditions associated with nutritional deﬁcits (Costa, 1988; Neves and Costa, 1998; Neves et al., 1999; Verano, 2003). While dietary reconstruction based on faunal remains, stable isotopic analysis of human bone, and paleopathological conditions of human skeletal remains do not identify any obvious nutritional deﬁcits in the highland groups, the agropastoral highland diet most likely encompassed a smaller degree of food diversity compared with the diets of coastal groups. Low food diversity can lead to insufﬁcient intake of speciﬁc nutrients, and as is the case in living Andean highland populations, delayed growth and short adult stature (Leonard et al., 1990; Bogin, 1999). Thus small body sizes in the highlanders in this study seem likely to reﬂect a less varied diet compared to that of the lowlanders. Although ecogeographic principles predict that relative limb lengths should be shorter in coldadapted populations, variation in limb lengths relative to overall size in these samples (the relatively longer limbs in the smaller highlanders, and the relatively shorter limbs in the larger lowlanders) may also reﬂect morphological plasticity in response to dietary conditions, as documented in living populations. Future studies may reveal whether nutritional deﬁcits, high-altitude hypoxia, ambient temperature, or a combination of numerous environmental stressors had an impact on relative limb proportions and body size in pre-Contact Andean populations. Coastal-highland gene flow and morphological variation Archaeologists and biological anthropologists have long debated the nature of population interactions between coastal and highland sites in Andean prehistory (e.g., Murra, 1972; Moseley, 2001; Burger and Salazar, 2003). Therefore, any examination of morphological variation in pre-Contact Andean populations requires a discussion of the possible effects of gene ﬂow between highland and coastal groups. While the exact cultural and historical processes involved in highland-coastal interactions in Andean prehistory are beyond the scope of this paper, the results of this study suggest that, with the exception of some MPC individuals, gene ﬂow had minimal effect on the biological diversity observed in these skeletal samples. Analysis of biological distance based on epigenetic (nonmetric) cranial and dental traits indicates genetic continuity in Arica populations from the Archaic period (3000 – 1000 BC) through the Late Intermediate period (AD 1000 –1476) (Sutter, 2000; Sutter and Mertz, 2004). Identical body proportions in human skeletons from the Archaic and Late Intermediate periods further suggest that Arica populations experienced little biological change through time (Weinstein, 2001). Similarly, archaeological evidence not only indicates human settlement along the central Andean coast starting at 9,000 years BP, but also a geopolitical separation between the central coast and the south-central highlands that continued through the Late Intermediate period (Moseley, 2001). Thus rather than an inﬂux of highland biological and cultural inﬂuences through time, coastal populations appear to be biologically distinct and relatively isolated from their highland neighbors, at least in the coastal groups represented in this study. Likewise, archaeological evidence from San Pedro de Atacama suggests that this highland population had strong biological and cultural ties to the highlands to the east rather than to the coastal settlements to the west. Núñez (1991), for example, argued that San Pedro de Atacama was at the center of prehistoric trade routes linking highland regions to the Northeast and Southwest. A paucity of coastal artifacts in San Pedro de Atacama archaeological sites suggests further that biological and cultural contacts with coastal populations were minimal (Moseley, 2001). The issue of coastal-highland population movements is more problematic in MPC. Archaeologists have long speculated about the ethnic diversity of the Andes during the Inca period in general, and of the populations at Machu Picchu and Cuzco during this time period in particular (e.g., Bingham, 1930; Moseley, 2001; Burger and Salazar, 2003). Burger and Salazar (2003) argued that the individuals recovered from Machu Picchu were nonelite servants selected from throughout the Andes to serve at this Inca royal estate. Dietary variation in the Machu Picchu skeletal sample, as inferred from stable carbon and nitrogen isotopic signatures in human bone, support the hypothesis BODY PROPORTIONS IN ANDEAN PREHISTORY that these remains represent diverse coastal and highland populations (Burger et al., 2003). Verano (2003) argued that the Machu Picchu skeletal sample is equally divided into individuals from coastal and highland regions, based on his analysis of cranial deformation styles and craniometrics. Given that artiﬁcial cranial deformation modiﬁes craniofacial shape (e.g., Antón, 1989) and is a rather ubiquitous practice in Andean prehistory, craniometrics is not a reliable method to assess biological distance and geographic origin in human skeletal remains. While comparisons of body size and shape do not directly address questions about genetic afﬁnities, MPC does exhibit the greatest range of variation in body size and proportions of all samples represented in this study, results which suggest the possibility that a few MPC individuals may be coastal migrants. Despite this possibility, the majority of MPC exhibited the smallest intralimb proportions, a pattern in accordance with ecogeographic rules. Climatic conditions at high altitudes seem to inﬂuence, in part, the short intralimb proportions of highland individuals. This cold-adapted body shape that is characteristic of the majority of MPC, moreover, is contrasted with those from lower elevations, further illustrating the presence of ecogeographic variation in Andean prehistory. CONCLUSIONS Pre-Contact human skeletal remains exhibit body proportions that vary along an altitudinal cline due to both climatic conditions and dietary factors. Despite the possibility of gene ﬂow from coastal regions, highland individuals from MPC exhibit markedly shorter intralimb length proportions compared with individuals from lower elevations, features that are indicative of a morphological adaptation to cold climatic conditions. Highland individuals from both MPC and SP tend to be smaller in estimated body mass and have relatively long limbs for their small body sizes compared with their lowland counterparts, an altitudinal contrast that may reﬂect differences in diet and subsistence between highland and lowland populations. Thus both climatic conditions and dietary variation appear to inﬂuence variation in body size and proportions in Andean prehistory. The results of this study reveal subtle variation in postcranial morphology across relatively short geographic distances and temporal periods. ACKNOWLEDGMENTS I thank the following individuals for their guidance during the course of this research: Susan Antón, Sue Boinski, Michael Moseley, Dave Steadman, and William Leonard. I am grateful to the following individuals and institutions for providing access to skeletal collections: M.A. Costa and all of the staff from the Museo Arqueológico R.P. Gustavo Le Paige, San Pedro de Atacama, Chile; Juan Chacama, Vivien Standen, and the staff from the Museo San Miguel de Azapa, Universidad de Tarapacá, Arica, Chile; P. Ryan Williams and Will Pestle from the Department of Anthropology, Field Museum, Chicago, IL; and Richard Burger and Roger Colten from the Division of Anthropology, Peabody Museum of Natural History, Yale University, New Haven, CT. I also thank Clark Spencer Larsen and two anonymous reviewers for their helpful comments. LITERATURE CITED Abate FR. 1991. Omni gazetteer of the United States of America, volume 9: Paciﬁc. Detroit: Omnigraphics, Inc. 583 Aldrich JW, James FC. 1991. Ecogeographical variation in the American robin (Turdus migratorius). Auk 108:230 –249. Allen JA. 1877. The inﬂuence of physical conditions in the genesis of species. Rad Rev 1:108 –140. Allison M, Focacci G, Arriaza B, Standen V, Rivera M, Lowenstein J. 1984. Chinchorro momias de preparación complicada: métodos de momiﬁcación. Chungara 13:155–173. Antón SC. 1989. Intentional cranial vault deformation and induced changes of the cranial base and face. Am J Phys Anthropol 79:253–267. Arriaza BT. 1995. Beyond death: the Chinchorro mummies of ancient Chile. Washington, DC: Smithsonian Institution Press. Arriaza BT, Salo W, Aufderheide AC, Holcomb TA. 1995. PreColumbian tuberculosis in northern Chile: molecular and skeletal evidence. Am J Phys Anthropol 98:37– 45. Ashton KG, Tracy MC, de Queiroz A. 2000. Is Bergmann’s rule valid for mammals? Am Nat 156:390 – 415. Baker PT. 1960. Climate, culture, and evolution. Hum Biol 32: 3–16. Barry RG. 1979. High altitude climates. In: Webber PJ, editor. High altitude geoecology. Boulder, CO: Westview. p 55–74. Beall CM, Baker PT, Baker TS, Haas JD. 1977. The effects of high altitude on adolescent growth in southern Peruvian Amerindians. Hum Biol 49:109 –124. Beals KL, Smith CL, Dodd SM. 1984. Brain size, cranial morphology, climate, and time machines. Curr Anthropol 25:301–330. Berenguer J, Dauelsberg P. 1989. El norte grande en la orbita de Tiwanaku. In: Iván S, editor. Culturas de Chile prehistoria: desde sus orı́genes hasta los albores de la conquista. Santiago: Andrés Bello. p 129 –180. Bergmann C. 1847. Ueber die Verhaltnisse der Warmeokonomie der Thiere zu ihrer Grosse. Gottinger Stud 3:595–708. Bharadwaj H, Singh AP, Malhotra MS. 1973. Body composition of the high-altitude natives of Ladakh: a comparison with sealevel residents. Hum Biol 45:423– 434. Billings WD. 1979. High mountain ecosystems: evolution, structure, operation and maintenance. In: Webber PJ, editor. High altitude geoecology. Boulder, CO: Westview. p 97–125. Bingham H. 1930. Machu Picchu: a citadel of the Incas. New Haven: Yale University Press. Bird J. 1943. Excavations in northern Chile. Anthropol Pap Am Mus Nat Hist 38:173–318. Bogin B. 1999. Patterns of human growth, 2nd ed. Cambridge: Cambridge University Press. Bogin B, Keep R. 1999. Eight thousand years of economic and political history in Latin America revealed by anthropometry. Ann Hum Biol 26:333–351. Burger RL, Salazar LC. 2003. Preface. In: Burger RL, Salazar LC, editors. The 1912 Yale Peruvian scientiﬁc expedition collections from Machu Picchu: human and animal remains. Yale University publications in anthropology 85. New Haven: Department of Anthropology, Yale University, and Division of Anthropology, Peabody Museum of Natural History. p xiii–xv. Burger RL, Lee-Thorp JA, van der Merwe NJ. 2003. Rite and crop in the Inca state revisited: an isotopic perspective from Machu Picchu and beyond. In: Burger RL, Salazar LC, editors. The 1912 Yale Peruvian scientiﬁc expedition collections from Machu Picchu: human and animal remains. Yale University publications in anthropology 85. New Haven: Department of Anthropology, Yale University, and Division of Anthropology, Peabody Museum of Natural History. p 119 –137. Calder WA. 1984. Size, function, and life history. Cambridge, MA: Harvard University Press. Clegg EG, Ashton EH, Flinn RM. 1972. The growth of children at different altitudes in Ethiopia. Philos Trans R Soc Lond [Biol] 272: 403– 437. Costa MA. 1988. Reconstitución fı́sica y cultural de la población tardı́a del cementerio de Quitor-6 (San Pedro de Atacama). Estud Atacamenos 9:99 –126. Costa Junquiera MA, Llagostera A. 1994. Coyo-3: momentos ﬁnales del perı́odo medio en San Pedro de Atacama. Estud Atacamenos 11:73–107. Crognier E. 1981. Climate and anthropometric variations in Europe and the Mediterranean area. Ann Hum Biol 8:99 –107. 584 K.J. WEINSTEIN Darroch JN, Mosimann JE. 1985. Canonical and principal components of shape. Biometrika 72:241–252. Dillehay TD. 1997. Monte Verde: a late Pleistocene settlement in Chile. Volume 2. The archaeological context and interpretation. Washington, DC: Smithsonian Institution Press. Dillehay TD. 2000. The settlement of the Americas: a new prehistory. New York: Basic Books. Dorsey GA. No date. Unpublished ﬁeld notes on ﬁle, Department of Anthropology, Field Museum of Natural History, Chicago, Illinois. Earth Info, Inc. 1996. Global climate categories CD-ROM. Boulder, CO: Earth Info, Inc. Eaton GF. 1916. The collection of osteological materials from Machu Picchu. Mem Conn Acad Arts Sci 5:1–96. Eveleth PB, Tanner JM. 1976. Worldwide variation in human growth. Cambridge: Cambridge University Press. Eveleth PB, Tanner JM. 1990. Worldwide variation in human growth, 2nd ed. Cambridge: Cambridge University Press. Falsetti AB, Jungers WL, Cole TM III. 1993. Morphometrics of the callitrichid forelimb: a case study in size and shape. Int J Primatol 14:551–572. Focacci G, Chacón S. 1989. Excavaciones arqueológicas en los faldeos del Morro de Arica, sitios Morro 1/6 y 2/2. Chungara 22:15– 62. Frisancho AR. 1978. Human growth and development among high-altitude populations. In: Baker PR, editor. The biology of high-altitude peoples. Cambridge: Cambridge University Press. p 117–171. Frisancho AR, Baker PT. 1970. Altitude and growth: a study of the patterns of physical growth of a high altitude Peruvian Quechua population. Am J Phys Anthropol 32:279 –292. Graves GR. 1991. Bergmann’s rule near the equator: latitudinal climates in body size of an Andean passerine bird. Proc Natl Acad Sci USA 88:2322–2325. Greksa LP. 1986. Growth patterns of European and Amerindian high altitude natives. Curr Anthropol 27:72–74. Greksa LP, Spielvogel H, Paredes-Fernandez L, Paz-Zamora M, Caceres E. 1984. The physical growth of urban children at high altitude. Am J Phys Anthropol 65:315–322. Greksa LP, Spielvogel H, Caceres E. 1985. Effect of altitude on the physical growth of upper-class children of European ancestry. Ann Hum Biol 12:225–232. Grine FE, Jungers WL, Tobias PV, Pearson OM. 1995. Fossil Homo femur from Berg Aukas, Northern Namibia. Am J Phys Anthropol 97:151–185. Gupta R, Basu A. 1981. Variations in body dimensions in relation to altitude among the Sherpas of the eastern Himalayas. Ann Hum Biol 8:145–151. Harrison GA, Küchemann CF, Moore MAS, Boyce AJ, Baju T, Mourant AE, Godber MJ, Glasgow BG, Kopec AC, Tills D, Clegg EJ. 1969. The effects of altitudinal variation in Ethiopian populations. Philos Trans R Soc Lond [Biol] 256:147–182. Hiernaux J, Froment A. 1976. The correlations between anthrobiological and climatic variables in sub-Saharan Africa: revised estimates. Hum Biol 48:757–767. Holliday TW. 1997. Postcranial evidence of cold adaptation in European Neandertals. Am J Phys Anthropol 104:245–258. Holliday TW. 1999. Brachial and crural indices of European late Upper Paleolithic and Mesolithic humans. J Hum Evol 36:549 – 566. Holliday TW. 2000. Evolution at the crossroads: modern human emergence in western Asia. Am Anthropol 102:54 – 68. Holliday TW, Ruff CB. 2001. Relative variation in human proximal and distal limb segments. Am J Phys Anthropol 116:26 – 33. Jacobs K. 1985. Climate and the hominid postcranial skeleton in Würm and early Holocene Europe. Curr Anthropol 26:512–514. Jacobs K. 1993. Human postcranial variation in the Ukranian Mesolithic-Neolithic. Curr Anthropol 34:311–324. James FC, McCulloch CE. 1990. Multivariate statistical methods in ecology and systematics: panacea or Pandora’s box? Annu Rev Ecol Syst 211:129 –166. Jungers WL, Falsetti AB, Wall CE. 1995. Shape, relative size, and size-adjustments in morphometrics. Yrbk Phys Anthropol 38:137–161. 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. Langsjoen OM. 1996. Dental effects of diet and coca-leaf chewing on two prehistoric cultures of northern Chile. Am J Phys Anthropol 101:475– 489. Leonard WR. 1989. Nutritional determinants of high altitude growth in Nuñoa, Peru. Am J Phys Anthropol 80:341–352. Leonard WR, Leatherman TL, Carey JW, Thomas RB. 1990. Contributions of nutrition vs. hypoxia to growth in rural Andean populations. Am J Hum Biol 2:613– 626. Leonard WR, Dewalt KM, Stansbury JP, McCaston MK. 1995. Growth differences between children of highland and coastal Ecuador. Am J Phys Anthropol 98:47–57. Leonard WR, Dewalt KM, Stansbury JP, McCaston MK. 2000. Inﬂuence of dietary quality on the growth of highland and coastal Ecuadorian children. Am J Hum Biol 12:825– 837. Lieberman DE, Devlin MJ, Pearson OM. 2001. Articular area responses to mechanical loading: effects of exercise, age, and skeletal location. Am J Phys Anthropol 116:266 –277. Llagostera A. 1989. Caza y pesca marı́tima (9000 a 1000 AC). In: Iván S, editor. Culturas de Chile prehistoria: desde sus orı́genes hasta los albores de la conquista. Santiago: Andrés Bello. p 57–79. Llagostera A, Torres CM, Costa MA. 1988. El complejo psicotrópico en Solcor-3 (San Pedro de Atacama). Estud Atacamenos 9:61–98. MacCurdy GG. 1923. Human skeletal remains from the highlands of Peru. Am J Phys Anthropol 6:217–329. Martin R. 1928. Lehrbuch der Anthropologie. Jena: Gustav Fischer. Mayr E. 1963. Animal species and evolution. Cambridge, MA: Harvard University Press. McHenry HM. 1992. Body size and proportions in early hominids. Am J Phys Anthropol 87:407– 431. Meadows Jantz L, Jantz RL. 1999. Secular change in long bone length and proportion in the United States, 1800 –1970. Am J Phys Anthropol 110:57– 67. Miller GR. 2003. Food for the dead, tools for the afterlife: zooarchaeology at Machu Picchu. In: Burger RL, Salazar LC, editors. The 1912 Yale Peruvian scientiﬁc expedition collections from Machu Picchu: human and animal remains. Yale University publications in anthropology 85. New Haven: Department of Anthropology, Yale University, and Division of Anthropology, Peabody Museum of Natural History. p 1– 63. Moseley ME 2001. The Incas and their ancestors, 2nd ed. London: Thames and Hudson. Mosimann JE, James FC. 1979. New statistical methods for allometry with application to Florida red-winged blackbirds. Evolution 33:444 – 459. Mueller WH, Schull VN, Schull WJ. 1978. A multinational Andean genetic and health program: growth and development in an hypoxic environment. Ann Hum Biol 5:329 –352. Mueller WH, Murillo F, Palamino H, Badzioch M, Chakraborty R, Fuerst P, Schull WJ. 1980. The Aymara of western Bolivia: V. growth and development in an hypoxic environment. Hum Biol 52:529 –546. Murra J. 1972. El “control vertical” de un maximo de pisos ecológicos en la economia de las sociedades Andinas. In: Murra J, editor. Visita de la provincia de Leon de Huánuco (1562) Iñigo Ortiz de Zúñigam Visitador, volume II, Huánuco. Peru: Universidad Hermillo Valdizen. p 429 – 476. Neves WA, Costa MA. 1998. Adult stature and standard of living in the prehistoric Atacama desert. Curr Anthropol 39:278 –281. Neves WA, Barros AM, Costa MA. 1999. Incidence and distribution of postcranial fractures in the prehistoric population of San Pedro de Atacama, northern Chile. Am J Phys Anthropol 109: 253–258. Newman MT. 1953. The application of ecological rules to the racial anthropology of the New World. Am Anthropol 55:311– 327. BODY PROPORTIONS IN ANDEAN PREHISTORY Newman MT. 1960. Adaptations to the physique of American aborigines to nutritional factors. Hum Biol 32:288 –313. Newman RW, Munro EH. 1955. The relation of climate and body size in U.S. males. Am J Phys Anthropol 13:1–17. Núñez AL. 1991. Cultura y conﬂicto en los oasis de San Pedro de Atacama. Santiago de Chile: Editorial Universitaria. Palomino H, Mueller WH, Schull WJ. 1979. Altitude, heredity, and body proportions in northern Chile. Am J Phys Anthropol 50:39 –50. Pawson IG, Jest C. 1978. The high-altitude areas of the world and their cultures. In: Baker PT, editor. The biology of high-altitude peoples. Cambridge: Cambridge University Press. p 17– 45. Pearson OM. 2000. Activity, climate, and postcranial robusticity. Curr Anthropol 41:569 – 607. Porter WP, Gates DM. 1969. Thermodynamic equilibria of animals with environment. Ecol Monogr 39:227–244. Roberts DF. 1953. Body weight, race and climate. Am J Phys Anthropol 11:533–558. Roberts DF. 1978. Climate and human variability, 2nd ed. Menlo Park: Cummings. Roosevelt AC, Douglas J, Brown L. 2002. The migrations and adaptations of the ﬁrst Americans: Clovis and pre-Clovis viewed from South America. In: Jablonski NG, editor. The ﬁrst Americans: the Pleistocene colonization of the New World. Memoirs of the California Academy of Sciences, no. 27. San Francisco: California Academy of Sciences. p 159 –235. Rosenberg KR, Ruff CB, Lu Z. 1999. Body size, body proportions and encephalization in the Jinniushan specimen. Am J Phys Anthropol 28:235 [abstract]. Ruff CB. 1991. Climate and body shape in hominid evolution. J Hum Evol 21:81–105. Ruff CB. 1994. Morphological adaptation to climate in modern and fossil hominids. Yrbk Phys Anthropol 37:65–107. Ruff CB. 2002a. Long bone articular and diaphyseal structure in Old World monkeys and apes, I: locomotor effects. Am J Phys Anthropol 119:305–342. Ruff CB. 2002b. Variation in human body size and shape. Annu Rev Anthropol 31:211–232. Ruff CB, Walker A. 1993. Body size and body shape. In: Walker A, Leakey R, editors. The Nariokotome Homo erectus skeleton. Cambridge, MA: Harvard University Press. p 234 –263. Ruff CB, Scott WW, Liu AYC. 1991. Articular and diaphyseal remodeling of the proximal femur with changes in body mass in adults. Am J Phys Anthropol 86:397– 413. Ruff CB, Trinkaus E, Holliday TW. 1997. Body mass and encephalization in Pleistocene Homo. Nature 387:173–176. Schmidt-Nielsen K. 1984. Scaling: why is animal size so important? Cambridge: Cambridge University Press. Schreider E. 1950. Geographical distribution of the body-weight/ body-surface ratio. Nature 165:286. Schreider E. 1957. Ecological rules and body-heat regulation in man. Nature 179:915–916. Schreider E. 1964. Ecological rules, body-heat regulation, and human evolution. Evol 18:1–9. Schreider E. 1975. Morphological variations and climatic differences. J Hum Evol 4:529 –539. Searcy WA. 1980. Optimum body sizes at different ambient temperatures: an energetics explanation of Bergmann’s rule. J Theor Biol 83:579 –593. 585 Sokal RR, Rohlf FJ. 1995. Biometry, 3rd ed. New York: W.H. Freeman. SPSS, Inc. 1998. Systat version 8.0 for windows. Chicago, Illinois. Standen V, Allison M, Arriaza B. 1984. Patologı́as óseas de la población Morro-1 asociada al complejo Chinchorro: norte de Chile. Chungara 13:175–185. Standen V, Arriaza BT, Santoro CM. 1997. External auditory exostosis in prehistoric Chilean populations: a test of the cold water hypothesis. Am J Phys Anthropol 103:119 –129. Stinson S. 1980. Physical growth of high altitude Bolivian Aymara children. Am J Phys Anthropol 52:377–385. Stinson S. 1982. The effect of high altitude on the growth of children of high socioeconomic status in Bolivia. Am J Phys Anthropol 59:61–71. Stinson S. 1990. Variation in body size and shape among South American Indians. Am J Hum Biol 2:37–51. Stinson S. 2000. Growth variation: biological and cultural factors. In: Stinson S, Bogin B, Huss-Ashmore R, O’Rourke D, editors. Human biology: an evolutionary and biocultural perspective. New York: Wiley-Liss. p 425– 463. Stinson S, Frisansho AR. 1978. Body proportions of highland and lowland Peruvian Quechua children. Hum Biol 50:57– 68. Stuart-Macadam P. 1992. Anemia in past human populations. In: Stuart-Macadam P, Kent S, editors. Diet, demography, and disease: changing perspectives on anemia. New York: Aldine de Gruyter. p 151–169. 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. Sutter RC, Mertz L. 2004. Nonmetric cranial trait variation and prehistoric biocultural change in the Azapa Valley, Chile. Am J Phys Anthropol 123:130 –145. Trinkaus E. 1981. Neanderthal limb proportions and cold adaptation. In: Stringer CB, editor. Aspects of human evolution. London: Taylor and Francis. p 187–224. Tsutakawa RK, Hewett JE. 1977. Quick test for comparing two populations with bivariate data. Biometrics 33:215–219. Verano JW. 2003. Human skeletal remains from Machu Picchu: a reexamination of the Yale Peabody Museum’s collections. In: Burger RL, Salazar LC, editors. The 1912 Yale Peruvian scientiﬁc expedition collections from Machu Picchu: human and animal remains. Yale University publications in anthropology 85. New Haven: Department of Anthropology, Yale University, and Division of Anthropology, Peabody Museum of Natural History. p 65–117. Villaxa A, Corrales J. 1993. Descripción y comentario de la fauna malacológica del sitio Acha-2. In: Muñoz I, Arriaza B, Aufderheide A, editors. Acha-2 y los origenes del poblamiento humano en Arica. Arica: Universidad de Tarapacá. p 81–90. Walker PL. 1986. Porotic hyperosteosis in a marine-dependent California Indian population. Am J Phys Anthropol 69:345– 354. Weinstein KJ. 1998. Morphological evidence for cold adaptation in pre-contact Andean skeletons from high and low altitudes. Am J Phys Anthropol [Suppl] 26:229 [abstract]. Weinstein KJ. 2001. Comparative skeletal morphology of modern humans and macaques from high and low altitudes. Ph.D. dissertation, Department of Anthropology, University of Florida. Ann Arbor: University Microﬁlms International.