O R I G I NA L A RT I C L E doi:10.1111/j.1558-5646.2008.00539.x

NONRANDOM FACTORS IN MODERN HUMAN MORPHOLOGICAL DIVERSIFICATION: A STUDY OF CRANIOFACIAL VARIATION IN SOUTHERN SOUTH AMERICAN POPULATIONS S. Ivan Perez1,2,3 and Leandro R. Monteiro4,5 1

´ Antropolog´ıa, Museo de La Plata, Universidad Nacional de La Plata, Paseo del Bosque s/n, 1900, La Plata, Division

Argentina 2 4

E-mail: [email protected]

Department of Biological Sciences, The University of Hull, Hull, HU6 7RX, United Kingdom 5

E-mail: [email protected]

Received June 11, 2008 Accepted September 25, 2008 The causes of craniofacial variation among human populations have been the subject of controversy. In this work, we studied aboriginal populations from southern South America, the last continental region peopled by humans and with a wide range of ecological conditions. Because of these characteristics, southern South America provides a unique opportunity to study the relative importance of random and nonrandom factors in human diversification. Previous craniometric studies recognized remarkable differences among populations from this region, usually resorting to random factors as the main explanation. In contrast, here we suggest, using tests based on quantitative genetic models, that: (1) the rate of craniofacial divergence among these populations is too high and (2) the patterns of variation within and between populations are too different to be explained by genetic drift alone. In addition, the among-sample craniofacial variation is correlated with climate and diet but not with mtDNA variation. We suggest that the influence of nonrandom factors (e.g., plasticity, selection) on human craniofacial diversification in regions with large ecological variation is more important than generally acknowledged and capable to generate large craniofacial divergence in a short period of time. These results bring nonrandom factors into focus for the interpretation of human craniofacial variation.

KEY WORDS:

Ecological variation, evolutionary rates, geometric morphometrics, Patagonian populations.

The processes underlying morphological diversification among human populations have been the subject of much controversy (e.g., Beals et al. 1983; Relethford 1994; Katzmarzyk and Leonard 1998; Roseman 2004). Biological diversification results from the division of an ancestral population whose descendants diversify due to selection, genetic drift, and mutation (Slatkin 1987; Schluter 2000) while constrained from diverging due to gene flow

3Present

address: Instituto de Biologia, Universidade Estadual de

˜ Paulo, Brazil Campinas, 13083-970, Campinas, Sao  C

978

(Slatkin 1987). The complexity of this process requires the consideration of multiple dimensions such as the evolutionary relationships estimated by neutral molecular data, the morphological variation measured with morphometric techniques, and ecological diversity described as biotic and abiotic variables to assess their relationships and test hypotheses about the evolutionary processes involved in the morphological diversification (Sober 1988; Levin 1992; Schluter 2000; Barton et al. 2007). Craniofacial shape variation has been widely investigated across modern human populations (e.g., Lynch 1990; Relethford 1994; Roseman 2004; Sardi et al. 2005; Bernal et al. 2006; Harvati

C 2009 The Society for the Study of Evolution 2009 The Author(s). Journal compilation  Evolution 63-4: 978–993

N O N R A N D O M FAC TO R S I N M O D E R N H U M A N D I V E R S I F I C AT I O N

and Weaver 2006; Perez et al. 2007a), and it has been suggested that the worldwide pattern of variation and divergence among extant populations is significantly similar between neutral genetic and craniometric traits (Relethford 1994). Therefore, gene flow and genetic drift have been considered the main evolutionary processes behind the observed patterns of overall craniometric differentiation (Lynch 1990; Relethford 1994, 2004). Nevertheless, nonrandom factors, like natural selection and phenotypic plasticity (Ackermann and Cheverud 2002, 2004; Carroll et al. 2007), could be important to explain the craniofacial divergence in some regions presenting large ecological variation (e.g., Katzmarzyk and Leonard 1998; Roseman 2004; Leonard et al. 2005). Here we investigate the evolutionary processes responsible for craniofacial diversification in human populations of southern South America. This region is characterized by large ecological variation—mainly in diet and climate—and was one of the last regions of the world to be colonized by modern humans (ca. 12,50013,000 B.P.; Borrero 1999; Lanata et al. 2006). Previous analyses of craniofacial variation among these populations have shown exceptionally high levels of morphological variation (Bernal et al. 2006; Perez et al. 2007a), particularly when compared to geographically larger regions worldwide (Sardi et al. 2005). Moreover, the craniofacial divergence of southernmost Patagonia and Tierra del Fuego populations (i.e., southernmost South America) relative to other American populations is so striking that it has even been interpreted as evidence of different ancestral origins for Fueguians and other Amerindians (the early southeast and later northeast Asian population, respectively; Lahr 1995; Neves et al. 1999). However, the mtDNA and Y chromosome (Moraga et al. 2000; Garc´ıa-Bour et al. 2004; Dejean et al. 2007) data support a local diversification of South American populations, requiring the existence of local processes within the time scale of the peopling to drive the large craniofacial variation observed (Perez et al. 2007a). Although it has been hypothesized that such variation might be the result of either random or nonrandom local processes (e.g., founder effects or natural selection due to the action of climatic factors; Rothhammer and Silva 1990; Bernal et al. 2006; Perez et al. 2007a), no explicit tests of such hypotheses have been performed to date. An approach to the problem is to ask how much morphological divergence among populations one expects to find if genetic drift is the sole mechanism shaping the patterns of variation. We can employ methods that provide a null hypothesis corresponding to the expectation of morphological variation as a product of random evolution alone (e.g., genetic drift) to understand the importance of these processes acting upon craniofacial divergence (Lande 1977, 1979; Lynch 1990; Relethford 1994; Hendry and Kinnison 1999; Schluter 2000; Ackermann and Cheverud 2002). In the present study, we test the hypothesis of cranial diversification (i.e., size and shape variation) among southern South

American populations as a product of random processes using quantitative genetic models (Lande 1977, 1979; Lynch 1990; Ackermann and Cheverud 2002). In addition, we employ a comparative approach to test the concordance between molecular and craniofacial variation among populations, considering the molecular variation as the null expectation under random processes. Finally, we test the relationship between craniofacial variation and ecological variables (i.e., diet and climate), and formulate hypotheses to explain the pattern of divergence among populations. To perform the analyses, the skull was divided into face (particularly masticatory apparatus), cranial vault, and base because these represent developmentally distinct units, with great internal integration, and vary with relative independence from one another (Cheverud 1982, 1995; Lieberman et al. 2000; Scheuer and Black 2000; Sperber 2001; Hallgr´ımsson et al. 2004, 2007). As a consequence of the different developmental and functional attributes of these craniofacial units, we expect them to present different relationships with the ecological and genetic dimensions. Particularly, the face variation is expected to present the closest association with environmental variables (Cheverud 1995; Sperber 2001; Harvati and Weaver 2006), whereas the cranial base would be the structure least related to these factors (Lieberman et al. 2000; Sperber 2001). To conclude, we discuss the implications of our results to the study of craniofacial diversification in modern and fossil humans.

Material and Methods SAMPLES

We studied 12 samples of adult aboriginals from southern South America’s later late Holocene (ca. 200–1500 years BP) (Fig. 1; Table 1). These prehistoric population samples come from different geographic and ecological regions, spread along 3500 km (from 25◦ to 55◦ South latitude and from 21◦ to 4◦ of mean annual temperature). The northernmost samples are composed of two farmer groups (i.e., groups with an agricultural economy; PG and SJ) and two terrestrial hunter–gatherers (Cha and Del), the central samples have two horticulturalists (Ar and Pa; the horticulture is an initial system of cultivation; Kennett and Winterhalder 2006) and two terrestrial hunter–gatherers (ChV and SCCh), whereas the south samples are composed by two terrestrial hunter–gatherers (SP and TF) and two maritime hunter–gatherers groups (AI and BC) (Berberi´an and Nielsen 2001). Because the same cranial features used for sex estimation are used in morphometric analyses (Buikstra and Ubelaker 1994), and no sex estimation independent from cranial morphology was available for most samples, males and females were pooled together for the comparisons. However, before comparing shapes among samples, the sex of each individual was estimated (Buikstra and Ubelaker 1994) to know if the “sex” ratio was similar in any

EVOLUTION APRIL 2009

979

S . I . P E R E Z A N D L . R . M O N T E I RO

of them. The results showed that all samples were similar in the sex ratio of their individuals (Table 1).

facial skull (maxilla and malar), vault (frontal, parietal, and upperoccipital) and cranial base (basi-occipital, condilar, and petrous) that were analyzed separately. The coordinates of landmarks and semilandmarks were recorded using the tpsDIG 2.10 software (Rohlf 2007). Reference points of the crania were aligned using a least squares criterion (Bookstein 1997; Adams et al. 2004). This procedure optimally translates, scales, and rotates coordinates of landmarks and semilandmarks (the Generalized Procrustes Superimposition; Rohlf and Slice 1990), and then the semilandmarks are slid along tangents to the outline of the curve until they minimize the Procrustes distances between corresponding points along the outline of a reference specimen (Bookstein 1997; Bookstein et al. 2002; Perez et al. 2006). The centroid size of the specimens (the square root of the summed squared distances from all landmarks and semilandmarks to the configuration centroid) was measured for each dataset and was used to scale the raw coordinates in the Generalized Procrustes Superimposition (Bookstein 1991). A general cranial size measure was calculated as the sum of centroid sizes from face, vault, and base point configurations. In this study we used the tpsRelw 1.44 software (Rohlf 2007) to superimpose the configurations and calculate the centroid sizes. A Relative Warps analysis (RW; a geometric morphometric version of principal component analysis; Bookstein 1991; Rohlf 1993) was performed on partial warps (shape variables generated by projection of aligned coordinates on the space spanned by principal warps of the bending energy matrix) plus uniform components to describe the axes (principal components) of major variation (combining within and among population variation) in the entire sample. The shape changes with respect to the reference configuration (i.e., consensus form) associated with the major axes were visualized as deformation grids based on each RW axis (Rohlf 1993). The relative warps analyses were performed using the tpsRelw 1.44 software (Rohlf 2007).

MORPHOMETRIC ANALYSIS

EXPLORING THE FACTORS RESPONSIBLE FOR

Craniofacial traits were captured from digital images as 2D coordinates for landmarks and semilandmarks in frontal (Fig. 2A), lateral (Fig. 2B), and base views (Fig. 2C). Images of the skulls were obtained with an Olympus SP 350 digital camera (Center Valley, PA). For frontal view images, the skulls were positioned in the Frankfurt plane and the camera lens was located in the coronal plane; the images were taken at 250 mm from the prosthion point. For base view images the photographs were taken at 250 mm from the occlusal surface, placing the skull in the perpendicular plane and the camera lens in the midsagittal plane. For lateral view images the skulls were positioned in the Frankfurt plane and the camera lens was localized parallel to the sagittal plane. The images from lateral view were taken at a 300 mm distance from the euryon point. The coordinates define point configurations of

CRANIAL DIVERSIFICATION

Figure 1. Map showing the central geographic location of the samples analyzed.

980

EVOLUTION APRIL 2009

To investigate the factors responsible for cranial diversification among southern South American populations, we employed internal and external evidence (respectively, genetic drift tests and regression plus association analyses) to the cranial divergence (sensu Hendry and Kinnison 1999; Sheets and Mitchell 2001). Genetic drift and evolutionary models The expected magnitudes and patterns of shape and size divergence under the influence of genetic drift and mutations alone were evaluated using Lande’s (1977, 1979) Constant Heritability (CH) method, Lynch’s (1990) neutral expectation for the  divergence rate, and the β-test developed by Ackermann and Cheverud

N O N R A N D O M FAC TO R S I N M O D E R N H U M A N D I V E R S I F I C AT I O N

Table 1.

1

Sample composition, with abbreviations used in the figures, sample sizes, subsistence, gender distribution, and age.

Samples1

Abbr.

Region

Subsistence

Pampa Grande Chaco San Juan-North Mendoza Delta Pampa Araucania Chubut Valley Santa Cruz-Chubut South Patagonia Tierra del Fuego Austral Island Beagle Channel Total

PG Cha SJ Del Pa Ar ChV SCCh SP TF AI BC

Northwest-Chaco Chaco Cuyo Pampean Region Pampean Region Araucanian Region Continental Patagonia Continental Patagonia Continental Patagonia Insular Patagonia Insular Patagonia Insular Patagonia

Farmers Hunter-gatherers Farmers Hunter-gatherers Horticulturists Horticulturists Hunter-gatherers Hunter-gatherers Hunter-gatherers Hunter-gatherers Hunter-gatherers Hunter-gatherers

n 15 18 15 17 20 17 23 21 18 21 21 20 228

Male%

Female%

Years BP2

46 56 54 53 45 53 44 43 61 57 57 45 –

54 44 46 47 55 47 56 57 39 43 43 55 –

500–1500 200–1000 400–1500 400–1000 200–1000 200–1000 400–1500 300–1500 200–1500 200–1500 200–1500 200–1500

´ The specimens are housed at Museo de La Plata (La Plata, Argentina), Museo Etnografico “J. B. Ambrosetti” (Buenos Aires, Argentina), Museo Regional

´ Molina” (R´ıo Gallegos, Argentina), Museo del Fin del Mundo (Usuahia, Argentina), Instituto Nacional de Antropolog´ıa y Provincial “Padre Manuel Jesus Pensamiento Latinoamericano (Buenos Aires, Argentina), Museo de Historia Natural (Santiago, Chile) and Instituto de la Patagonia Austral (Punta Arenas, Chile). 2

Approximate sample ages according to radiocarbon dating and contextual information.

(2002). We used this variety of techniques available in the absence of a clear consensus about the best way to measure divergence (Hendry and Kinnison 1999). Rate tests (Lande 1977, 1979; Lynch 1990) in particular, are useful to detect directional nonrandom factors among recently separated populations (Hendry and Kinnison 1999; Lemos et al. 2001), such as southern South Americans. Moreover, there is a window in time scale where these tests are unlikely to distinguish directional nonrandom factors from the random ones because of the dynamics of divergence (Lynch 1990; Lemos et al. 2001; Hunt et al. 2007). After longer periods, stabilizing selection will erase the evidence of random or directional nonrandom factors during the initial phase of divergence (Lemos et al. 2001). Conversely, tests that analyze the pattern of variation for multiple dimensions within and among groups (Lande 1979), as the β test (Ackermann and Cheverud 2002), are capable to test for random evolution among populations regardless of temporal scale by focusing on relative patterns of variation rather than absolute ones (Ackermann and Cheverud 2004; Weaver et al. 2007). The quantitative genetic models for divergence predict the expected amounts of phenotypic divergence among populations under genetic drift from a number of genetic and demographic parameters, such as the heritability, effective population sizes, within-group phenotypic variation, time (in generations) since initial divergence, and the mutational rate of input of genetic variance. When the divergence observed is higher than the one expected from genetic drift and mutations alone, it is interpreted as a result of nonrandom processes (Lynch 1990; Sheets and Mitchell 2001; Ackermann and Cheverud 2002; Marroig and Cheverud

2004). However, because genetic and demographic parameters are hard to estimate accurately, the quantitative genetic tests are usually considered qualitative evidence of nonrandom processes (Turelli et al. 1988; Spicer 1993; Monteiro and Gomes-Jr 2005). It is necessary to carefully discuss the estimated values and confidence intervals of these parameters. The CH test (Lande 1977) was used in this work because it is appropriate for recently separated populations, which have not yet achieved mutation–drift equilibrium (Turelli et al. 1988; Spicer 1993; Monteiro and Gomes-Jr 2005), which seems to be the case in the present dataset (see the discussion regarding divergence time estimates below). The test for directional selection in the CH method is performed by a F-test: F=

Sz¯2(t) σ2 h 2 t/Ne

,

where the difference among populations (the numerator) is measured by an among-populations mean square n  

Sz¯2(t) =

2 z¯ i (t) − z¯ (t)

i=1

(n − 1),

where the term z¯ (t) corresponds to the grand average over all n populations, and z¯ i (t) corresponds to the mean of the ith population on a quantitative character z. N e is the effective population size, t is the number of generations since divergence, σ2 is the within-group phenotypic variance of z, and h 2 is the heritability of z. The denominator in the F-test corresponds to the expected variance among-groups in the character z if genetic drift is the

EVOLUTION APRIL 2009

981

S . I . P E R E Z A N D L . R . M O N T E I RO

Figure 2. Allocated reference points are displayed with different symbols. Landmarks are represented as squares (), whereas semilandmarks are represented as circles (•) on face (A) vault (B) and

base (C) views. Drawing by Marina Perez.

sole mechanism driving evolutionary divergence. The statistic from the CH test follows a F-distribution with n − 1 degrees of freedom in the numerator and infinite degrees of freedom in the denominator. As the estimation of genetic and demographic parameters for the CH model does not provide exact values (Turelli et al. 1988),

982

EVOLUTION APRIL 2009

a reasonable range of conservative estimates is more appropriate than a single estimate for parameters calculated with uncertainty (Monteiro and Gomes-Jr 2005). A conservative maximum number of generations since divergence (t) was estimated. Evidences of settlement are found ca. 12,500–13,000 B.P. in southernmost Patagonia Pleistocene and ca. 11,000 B.P. in Tierra del Fuego (Borrero 1999). Fenner (2005) estimate the human generation time in 28 years. Thus, we could set a conservative estimate of maximum t = 464 generations (13,000/28) for the divergence of southern South American populations (considering a local diversification of South American populations; Moraga et al. 2000; Garcıa-Bour et al. 2004). Because a lower divergence time existed among neighbor (geographically closer) populations, we performed the tests using a lower time of divergence. We employed a conservative mean estimate of 230 generations of divergence. This value was calculated averaging the probable number of generations since divergence among all pairs of populations, employing molecular, archaeological, and historical information (Borrero 1999; Moraga et al. 2000; Berberi´an and Nielsen 2001; Bernal 2008). Moreover, the gene flow among populations, in particular, can be an important force restricting the population differentiation (Slatkin 1987) and the power of the rate tests could be lower in these situations (Hendry and Kinnison 1999). In this article we employ samples from later late Holocene populations (Table 1). Conservative estimations of population sizes (N)—estimated from ethnographical and ethnohistoric information—range from 1500 to 20,000 individuals (Steward 1950; Borrero 1999; Binford 2000; see discussion in Bernal 2008). The effective population size (N e ) can be estimated at about one-third of the total population size for human populations of this region (i.e., 500 to 6,000 individuals; Cavalli-Sforza and Bodmer 1971; see discussion in Lalueza et al. 1997). The harmonic mean of N e (HmN e ) among later late Holocene populations was HmN e = 1100 individuals. Taking these estimates into account, the relation t < N e /5 holds and it is appropriate to use the CH method (Turelli et al. 1988). The amount of phenotypic variation among populations Sz¯2(t) was measured by the among-group variation in general centroid size (the sum of the centroid sizes calculated for each view) and the scores for the first RWs in each dataset. The use of the first RW as an univariate shape variable is justified because it is a linear combination of the original shape variables aligned to the main direction of variation (with the contribution of within and among-group variation), expected to be a proxy to a biologically relevant pattern. Also, the rigid orthogonal rotation of principal components ensures that the shape space is not deformed and the interobject distances are maintained. This variable can be analyzed by statistical methods as an ordinary univariate variable, and will allow for the decomposition of variances needed for the rate test (Hendry and Kinnison 1999; Lemos et al. 2001; Monteiro

N O N R A N D O M FAC TO R S I N M O D E R N H U M A N D I V E R S I F I C AT I O N

and Gomes-Jr 2005). We calculated RWs based on the total covariance and pooled within-group covariance matrices. The use of a subspace based on the pooled within-group covariance matrix may be more conservative approach to test random factors, because this subspace does not preferentially consist of directions of variation for which among-group variance is high. However, the use of this reduced space to perform a rate test implies that some shape variation (the variation included in the lowest RW scores) present in the sample is not tested. For this reason, we tested the RWs that account for approximately 80% of total variation. Because isolated RWs with small eigenvalues may not have a clear biological signification, a more general test—β test (see below)—which compares the patterns of variation among and within groups across all the principal components, was also performed. The estimate of the common within-group phenotypic variance (σ2 ) of shape and size was calculated from the withingroup mean squares of ANOVAs using RW scores and size as dependent variables, and population membership as a grouping factor. It is important to point out that the use of prehistoric samples—that include several generations of a population lineage (Cadien et al. 1976)—could overestimate the variation within a sample (Hunt 2004). As a consequence, the power of rate tests could be lowered (Lynch 1990). However, the comparison between the magnitude of variation in fossil (i.e., Quaternary) and modern populations from several mammalian species have shown insignificant differences between them, even over temporal periods long exceeding the ones of the present study (Hunt 2004). This suggests that the effect of collapsing temporal change into a single sample could be generally negligible. Heritability estimates (h2 ) are difficult to obtain, because it is a variable parameter among traits, populations in different environments, and within populations at different times (Vitzthum 2003; Monteiro and Gomes 2005). For modern human populations, the mean heritability estimates for craniofacial metric traits vary from 0.28 to 0.65; whereas the range for individual measurements is from 0.102 to 0.729 (Carson 2006). A median value of h2 = 0.55 for metric craniofacial characters has been used in modern human craniofacial variation analyses (e.g., Relethford 1994). However, recent estimates indicate that the mean h2 value for craniofacial traits could be close to 0.30 (Face = 0.33; Vault = 0.29; and Base 0.36; see table 3 in Carson 2006). In our case, it would be important to define the heritability of the centroid size and of specific vectors of among-individuals differences (the RWs), for different directions in shape space can have different heritabilities (Klingenberg and Monteiro 2005). Because the heritabilities for the quantitative characters in the specific populations studied are inaccessible, we considered the literature-based range of heritabilities for craniofacial traits in the interpretations. It is also important to remark that the quantitative tests assume that the value of heritability is similar across all the populations (Lande

1977; Lynch 1990; Marroig and Cheverud 2004), what would also be impossible to assess in the studied populations. However, the values would be expected to remain within the specified interval. To assess the robustness of the CH test results due to variation in N e and h2 , we calculated a range of F-values for a conservative series of parameter estimates (N e = 100–3000; h2 = 0.1–0.9) and observed which combinations of estimates would allow for the acceptance of the null hypothesis of genetic drift. The results were visualized using a surface plot, where the parameters h2 and N e form the X and Y-axes, respectively, whereas the F-statistic is the Z-axis. A least squares surface was shown by isolines of F-statistics, depicting the acceptance region for the null model. The region below the isoline of 1.79 (F-value for 11 and infinite degrees of freedom at a 5% confidence level) shows the combinations of N e and h2 required to not reject genetic drift. The observed  divergence rate (Lynch 1990) is compared to an expectation of divergence based on the literature, to evaluate whether the amount of divergence is lower or higher than the expected variation if mutation and random genetic drift were the single evolutionary forces (Lynch 1990). In this model, the rate of divergence is calculated by  = var B (ln z)/[tvarW (ln z)], where var B (ln z) and varW (ln z) are the among and withinpopulation mean squares calculated by an ANOVA, using the general centroid size (log-transformed) and shape scores as dependent variables, and population membership as a grouping factor. Lynch (1990) estimates that the lower and upper limits for divergence rates of mammalian skeletal traits under the neutral mutation–drift hypothesis are, respectively,  min = 0.0001 and  max = 0.01. A range of divergence times (the sum of the divergence times along each lineage; Lynch 1990) was used to assess how long the separation among populations would have to be for the observed s to fall within the expected interval (see above to a discussion of divergence times among these populations). The  rate test allowed for the comparison of southern South American divergence rates with those reported by Lynch (1990) for the evolution of modern human populations. The β-test developed by Ackermann and Cheverud (2002) is based on the Lande (1979) model and tests for genetic drift comparing the proportionality of among and within-population variances (considering that among-population variation should be proportional to within-population variation if genetic drift is the sole mechanism responsible for divergence [Ackermann and Cheverud 2004]). To test the proportionality of variances, the test uses a regression of among-group variances on withingroup variances for the principal components (PC) of shape variables. The phenotypic within-group covariance matrices were used as a proxy for the genetic covariance matrices, based on the

EVOLUTION APRIL 2009

983

S . I . P E R E Z A N D L . R . M O N T E I RO

corroborated proportionality between these matrices reported in the literature (Cheverud 1988; Roff 1996). The within-group variances were estimated by the eigenvalues of PCs, calculated from the pooled within-group covariance matrix of partial warp and uniform component scores. The among-group variances were calculated as the mean squares of the projections of group means on the within-group PCs. A linear regression of log-transformed amonggroup on log-transformed within-group variances was performed (Ackermann and Cheverud 2002). This model predicts that if variation among populations was produced by genetic drift alone, the regression coefficient (β) should be equal to 1. A significant deviation of the observed coefficient from 1 indicates that the pattern of among-group variation was probably not produced solely by genetic drift (Ackermann and Cheverud 2002). A second prediction of this model is that the group mean projections on the within-group principal components should not be correlated under the influence of genetic drift. Therefore, correlation coefficients among PC scores (using mean projections as observations) were calculated to test this prediction. Ackermann and Cheverud (2004) pointed out that when few groups are compared, as in the present work, genetic drift is hard to reject. In addition, the small sample size and the uncertainty in estimation of variance and covariance matrices of prehistoric samples (leading to decreased statistical power) makes every sign of deviation from genetic drift potential evidence of nonrandom forces, to explain among-population divergence (see discussion about the power of the beta test in Ackermann and Cheverud 2004; Weaver et al. 2007). Regression and association analyses We performed a Procrustes analyses (Gower 1971; Peres-Neto and Jackson 2001) to test the correspondence between the multivariate ordination pattern of morphological and neutral genetic variation based on mtDNA haplogroup frequencies. This method scales and rotates the ordinations, using a minimum squared differences criterion. Then, the complement of the sum of the squared residuals between configurations in their optimal superimposition can be used as a measure of association (m 12 = sqrt(1 − ss); Gower 1971). A permutation procedure (PROTEST; 10,000 permutations) was used to assess the statistical significance of the Procrustean fit (Jackson 1995). The ordination of all RWs (consensus values for each sample) was compared to the Principal Coordinates from the distance matrix based on mtDNA haplogroups (see references in Perez et al. 2007a). The correspondence between the patterns of craniofacial and neutral genetic variation can be used to assess the importance of random processes to explain the morphological pattern of divergence (Roseman 2004; Harvati and Weaver 2006). The correspondence of size, facial, vault, and base shape variation (the scores of first RWs for consensus shapes) with cli-

984

EVOLUTION APRIL 2009

mate and diet was tested by a multiple regression (Legendre et al. 1994), using 9999 random permutations. Multicollinearity was tested using the tolerance criteria (Hair et al. 1998). The mean annual temperature where populations were located was used as an indirect estimator of climate (Katzmarzyk and Leonard 1998; Harvati and Weaver 2006). This variable was obtained from climatic databases (SMN 1961–1990). We also defined a diet variable, which includes two categories: groups with a diet based mainly on faunal resources (i.e., hunter-gatherer groups) and groups with a significant proportion of cultigen resources in the diet (i.e., farmers plus horticulturist groups). These categories were incorporated in the regression models as a contrast (−1, 1). This variable represents the most important difference in diet found between the southern South American human groups. The diets were estimated employing archaeological (i.e., macrobotanical, archaeofaunal, and stable carbon and nitrogen isotope data) and ethnographic information (Berberi´an and Nielsen 2001; Kelly et al. 2001; Gil et al. 2006). The statistical analyses and divergence rate calculations were performed using R 2.5.0 (R-Development Core Team 2008). Procrustes analyses were performed using function protest in the vegan package for the R-system (Oksanen et al. 2008).

Results The ordinations of population means calculated over the shape matrix for all individuals are show in Figure 3. Because the RWs based on the total covariance and pooled within-group covariance matrices display similar results, we only show the results of the first. The first two RWs explain approximately 36–40% of total variation in the different datasets. The southernmost Patagonian and Tierra del Fuego samples show the most robust and tall face and most robust and long vault (Fig. 3A,B). The northernmost samples show shorter and less robust faces, and shorter vaults as well (Fig. 3A,B). The first RW for the face separates hunter–gatherers from farmer plus horticulturists groups (Fig. 3A), whereas RW2 shows a fairly latitudinal ordination. RW1 for vault and base shows a fairly latitudinal ordination of hunter–gatherers (Fig. 3B,C) (this ordination of hunter-gatherers is also observed for the vault and the face). Size variation among samples shows that hunter–gatherer groups present larger skull size than the farmer plus horticulturist ones (Fig. 4). The results of the CH test suggest that random processes alone cannot account for the morphological divergence among populations displayed by skull size and the several RWs. The results for overall centroid size show a strong divergence, where the effective population size would have to be smaller than 500 and the heritability larger than 0.9 for genetic drift to be accepted (Fig. 5A). Lower heritabilities would also lower the population sizes required to accept the neutral hypothesis. Because

N O N R A N D O M FAC TO R S I N M O D E R N H U M A N D I V E R S I F I C AT I O N

Figure 3. Relative warps 1 and 2 (mean values), and deformation grids depicting shape variation along the first RW for the face, vault, and base views. Hunter–gatherers groups are represented as circles (•), whereas farmer–horticulturalist groups are represented as

star (∗).

heritability values are relatively low for modern human cranial traits (h2 = 0.28–0.65; Carson 2006) and a conservative estimate of mean effective population size during the divergence for the southern South American Holocene local populations is larger than 500 (see Material and methods), the acceptance of

Figure 4.

Boxplot depicting patterns of size variation of the sam-

ples studied.

genetic drift would require unrealistically high heritabilities and low effective population sizes. Particularly, if we considered a mean heritability of 0.32 (Carson 2006), the effective population size required to accept the neutral hypothesis (Fig. 5A) is ca. 100–200 individuals. Such low effective population sizes are not expected in South America (Borrero 1999; Lanata et al. 2006), because they would compromise the reproductive viability of the human groups (Wobst 1974; Anderson and Gillan 2001). Similar results are observed even if we consider the mean heritability of 0.55—the value generally employed as reference in human craniometric studies (Relethford 1994; Carson 2006). For the first RWs, the CH model (Fig. 5C,E,G) indicates that, for the amount of craniofacial divergence observed among populations, the hypothesis of random evolution would be accepted for an effective population size of 500, if the heritability was equal to or larger than 0.6. Moreover, if we considered h2 = 0.32, the unrealistic effective population size of ca. 250–350 individuals is required to accept the neutral hypothesis (Fig. 5C,E,G).

EVOLUTION APRIL 2009

985

S . I . P E R E Z A N D L . R . M O N T E I RO

Figure 5. Results of the Lande (1977) CH rate tests for centroid size (A), RW1 of face (C), vault (E), and base (G) traits, and Lynch’s (1990)  for centroid size (B), RW1 of face (D), vault (F), and base (H) traits, using a range of estimates of effective population size, heritability (CH rate tests), and number of generations since divergence ( rate tests).

986

EVOLUTION APRIL 2009

N O N R A N D O M FAC TO R S I N M O D E R N H U M A N D I V E R S I F I C AT I O N

Similar results are achieved for the RW2 to 4 for all craniofacial units—which explain approximately 30% of the overall sample variation—RW6, 7 and 9 for Face, and RW5 and 8 for Vault— which explain approximately 20% of the overall sample variation(results not shown). The result for RW2 is particularly relevant because it is related to temperature variation (see multiple regression below). The  rate test performed on skull size, face, vault, and base shape (Fig. 5B,D,F,H), presented  values consistently larger than  max = 0.01 (the maximum value expected by genetic drift for mammalian skeletal traits; Lynch 1990) for the conservative range (1–500) of the number of generations since divergence considered. Similar results are achieved for the RW2 to RW4 for all craniofacial units—which explain approximately 30% of overall sample variation—RW6 and 9 for Face, and RW5 and 8 for Vault—which explain approximately 20% of overall sample variation (results not shown). A more general test for the multivariate pattern of divergence was performed by the β-tests on the shape variables. The regression coefficients between among- and within-population variation significantly differ from 1 only for the face and base datasets (Face: β = 1.052, t = 2.166, P = 0.034; Vault: β = 0.968, t = −1.28, P = 0.205; Base: β = 1.069, t = 2.029, P = 0.049). In all cases, the first RWs presented positive deviations from the regression line, showing higher among-group variation than the one expected by drift alone. However, although few samples were included in the comparison, all three datasets presented significant correlations between mean sample projections for some of the first PCs (e.g., Face PC1 and PC4= 0.716, P = 0.009, PC4 and PC5= 0.693, P = 0.012; Vault PC4 and PC6= 0.577, P = 0.049, PC5 and PC6= −0.632, P = 0.027; Base PC1 and PC4= 0.648, P = 0.023, PC3 and PC4= 0.859, P = 0.000), indicating a deviation from expectations under genetic drift (Ackermann and Cheverud 2002). The Procrustes analysis of ordinations based on mtDNA haplogroup frequencies, geographic coordinates, and craniofacial variation (i.e., all RWs among the consensus shapes) shows a significant association between genetic distance and geography (m 12 = 0.7601, P = 2 × 10−4 ). However, there is no signifiTable 2.

Discussion The results of the genetic drift tests show strong evidence against random process as the dominant factor driving the craniofacial diversification among modern human populations in southern South America. These tests show that the magnitude of observed size and shape divergence is too large to be generated by genetic drift alone within the parameter range considered. This result is relevant because we employed conservative estimates of demographic and genetic parameters, and used prehistoric samples with mixed sex to calculate among and within variances and covariances, which might decrease the power of the tests (Lynch 1990; Hendry and Kinnison 1999; Ackermann and Cheverud 2004). In fact, the differentiation is so striking that the northernmost and southernmost samples do not overlap distributions in size and shape spaces for face and vault datasets (Fig. 6). The large magnitude of divergence among southern South American populations in relation to genetic drift expectations is also supported by the comparisons of craniometric and molecular F ST available in the literature (Cavalli-Sforza et al. 1994; Relethford 1994; Sardi et al. 2005). When gene flow and genetic drift are the main processes explaining the pattern of morphological divergence, similar values of F ST for molecular and morphological data (Relethford 1994, 2002) are expected. However, the F ST estimated for craniofacial variation in metric traits for late Holocene southern South America is 0.153 whereas for protein

Multiple regression analysis performed employing consensus configurations of each sample.

Effect

Partial b CS

Diet Temperature Centroid Size R2 ∗

cant association between size and shape variables (all RWs) with genetic distances among groups (Centroid Size, m 12 = 0.2987, P = 0.399, Face, m 12 = 0.436, P = 0.297; Vault, m 12 = 0.426, P = 0.359; Base, m 12 = 0.503, P = 0.079). Because the size and shape variation among southern South American populations was not associated to the pattern of evolutionary relationships, this information was not incorporated in multiple regression analysis (Losos 1999). The multiple regressions show that there are significant relationships of size and RW1 variation (calculated with the consensus shape for each sample) with diet for the face and vault sets (Table 2). The RW2 for face and vault, as well as the RW1 for base, were related to mean annual temperature (Table 2).

2.531∗∗ 0.019 – 0.775

RW1 Face −0.007 0.000 −0.004∗ 0.911

RW2 Face −0.008 −0.002∗∗ 0.001 0.690

RW1 Vault 0.014∗ −0.001 −0.001 0.864

RW2 Vault 0.010 −0.002∗ −0.002 0.407

RW1 Base −0.024∗ 0.003∗∗ 0.000 0.757

RW2 Base −0.005 −0.001∗ 0.005∗ 0.809

≤0.05, ∗∗ ≤0.01.

EVOLUTION APRIL 2009

987

S . I . P E R E Z A N D L . R . M O N T E I RO

Figure 6.

Ordinations of populations on RW1 and centroid size to face (A), vault (B), and base (C) traits of the northernmost (Pampa

Grande) and southernmost (Beagle Channel) samples. The 95% confidence ellipse for group means is shown. The figure shows that these samples do not overlap in size and shape to face (A) and vault (B) variation, with a between-population allometric effect (size-shape association). Deformation grids representing the variation in face (A), vault (B) and base (C) traits along the first RW are depicted at each side of the ordinations. The Beagle Channel sample shows the most robust and tallest face, and robust and longest vault.

and blood group data it is 0.059 (Cavalli-Sforza et al. 1994; Sardi et al. 2005). Considering the fact that local common ancestry is recent for the populations compared (Moraga et al. 2000), these differences in the F ST values between molecular and craniofacial data can be interpreted as evidence of nonrandom factors acting over craniofacial traits (Relethford 1994, 2002; Schluter 2000; cf. Sardi et al. 2005). The lack of significant association between craniofacial variation and the pattern of genetic similarity (i.e., the results of Procrustes analysis) is also evidence against random processes as the main factor behind craniofacial divergence in southern South American populations. If random processes are solely responsible for the craniofacial divergence as observed in a worldwide scale for some cranial regions, we expect an association between these ordination patterns (Roseman 2004; Harvati and Weaver 2006). The differences between our results and previous ones us-

988

EVOLUTION APRIL 2009

ing samples from a relatively more homogeneous environment (Perez et al. 2007b) suggest the importance of ecological variation to explain the differences in cranial size and shape among populations. If we rule out genetic drift as the main factor of morphological diversification, nonrandom factors like directional selection and phenotypic plasticity must be taken into consideration (Hendry and Kinnison 1999; Carroll et al. 2007). Because the environmental influence during the ontogeny could not be controlled in our study, the genetic and ecophenotypic components of the morphological change could not be identified (Losos et al. 1997; Reznick et al. 1997; Hendry and Kinnison 1999; Carroll et al. 2007). However, we can use independent information (e.g., developmental characteristics; experimental studies; see below) to know the extent of environmental influence on craniofacial traits during ontogeny.

N O N R A N D O M FAC TO R S I N M O D E R N H U M A N D I V E R S I F I C AT I O N

We considered two ecological variables (i.e., diet and climate) to generate hypotheses about the causes of the large divergence in cranial morphology. Our results show that the interaction between the different craniofacial regions and the ecological dimensions was complex. This is expected because the craniofacial skeleton shows characteristic variability (Hanken and Hall 1993; Scheuer and Black 2000; Morris-Kay 2001; Sperber 2001), and the three craniofacial units are subject to different influences and might respond in different ways to the evolutionary and ecological factors (Lockwood et al. 2004; Harvati and Weaver 2006). The results show that the centroid size and vault shape variation (RW1) as well as the allometric facial shape variation (RW1 face and size are highly correlated, r = −0.936, P = 0.000) are clearly associated with diet. Particularly, the farmer groups present the smallest crania, showing shorter and less robust faces. This suggests the importance of diet, or some related variable, on the vault shape, general size, and facial shape-related changes. The results also show that some shape variation (RW2 for face and vault and RW1 for base; Table 2, Fig. 3) is associated with mean annual temperature, suggesting the influence of climate on shape variation. The southernmost groups show the most robust and long vault. This result is also found (vault and base data) when only hunter–gatherer groups are compared (temperature vs. Centroid size r = 0.160, P = 0.705; Face r = 0.310, P = 0.455; Vault r = −0.758, P = 0.029; Base r = 0.790, P = 0.020). The face, especially the maxilla and jugal, attains adult size and shape after the cranial base and vault have done so, following a somatic growth pattern related to masticatory muscle insertions (Moss and Young 1960; Cheverud 1995; Opperman 2000; Opperman et al. 2005; Sardi and Ramirez-Rozzi 2005; Cheverud 2007; Hallgr´ımsson et al. 2007). For these reasons, it is sustained that the face is highly influenced by the environment during the ontogeny (Cheverud 1995; Sperber 2001; Opperman et al. 2005). On the other hand, the cranial vault attains adult size and shape relatively early and acts as a support structure for the brain, but it grows by mechanical forces that upregulate transcription factors in sutures, also being influenced by the environment during the ontogeny (Moss and Young 1960; Sperber, 2001; Opperman et al., 2005). Thus, the relationship between diet and facial shape (i.e., maxillary and malar bones), vault, and general size variation could be related to phenotypic plasticity (or maternal effect; Carroll et al. 2007). There are important subsistence differences related to the food production in southern South America (Berberi´an and Nielsen 2001). Agricultural practices generate a greater availability of carbohydrates for farmer groups, compared to the large proportion of protein consumed by hunter–gatherer groups. Phenotypic plasticity might play an important role in the initial stages of a population becoming established in a new environment (or in local environmental fluctuations) (Carroll et al. 2007). This

has been documented for several human Holocene populations, where increases in dietary proportions of carbohydrate caused a decrease in body and skull size, and corresponding changes in facial dimensions (Stynder et al. 2007). The variation in morphology related to increased consumption of carbohydrate has also been widely documented among current human populations as well as within other mammalian species (Frisancho 1996). Experimental and comparative studies have shown that the protein and the protein-calorie malnutrition generate significant differences in size (with smaller bodies when less protein is consumed) and associated allometric shape changes (mainly for the facial region; e.g., Pucciarelli 1980, 1981; Pucciarelli and Oyhenart 1987; Frisancho 1996). This variation in cranial size and facial shape is similar to the pattern of morphological divergence observed in this and previous studies (Stynder et al. 2007). Therefore, comparative and experimental studies suggest that the cranial size and facial divergence between farmers and hunter-gatherers southern South American groups are an environment-dependent phenotype expression during the ontogeny related to malnutrition. However, further studies are needed to investigate this hypothesis, as well as other alternatives (i.e., biomechanical forces; Curtis et al. 2008), about the importance of dietary differences for morphological divergence among southern South American populations. The cranial base, on the other hand, is subject to smaller environmental influence during the ontogeny than the face because it is the first skull region to attain adult size and shape and acts as a support structure for the brain and a passing site for several vessels and nerves (Cheverud 1995, 2007; Lieberman et al. 2000; Sperber 2001; Sardi and Ramirez-Rozzi 2005; Opperman et al. 2005). Therefore, the correlation of the major axis of variation for this skull region with mean annual temperature, as well as its great divergence, is more likely related to directional selection rather than to phenotypic plasticity. Moreover, when we compare variation among populations independent of subsistence, the cranial vault and base show a significant association with mean annual temperature. Thus, climate variation also could explain the large divergence found in these structures. Several comparative and experimental studies have shown a significant association among body mass, body proportions, basal metabolic rates (BMR), and mean annual temperature or latitude (Roberts 1953; Riesenfeld 1981; Ruff 1994; Frisancho 1996; Katzmarzyk and Leonard 1998; Pearson 2000; Leonard et al. 2005). Populations inhabiting cold climates have elevated BMR and large body size or mass (Katzmarzyk and Leonard 1998; Leonard et al. 2005). This is explained as the result of an adaptative process in response to low temperatures. Experimental work has also demonstrated that heritable differences in body mass or weight, specifically great body mass, are the most important factors to determine temperature tolerance in mammals (e.g., Riesenfeld 1981). Conversely, the developmental response

EVOLUTION APRIL 2009

989

S . I . P E R E Z A N D L . R . M O N T E I RO

to cold stress in mammals involves subtle but significant changes in craniofacial shape associated with larger changes in size; particularly, the cold stress generates a decrease in body size (Riesenfeld 1973)—much like the malnutrition stress (Pucciarelli 1980)— narrower nose, rounder neurocranium (Steegmann and Platner 1968; Riesenfeld 1973; Rae et al. 2006), and a reduction of absolute cortical thickness (Brandt and Siegel 1978). The climate could have a particularly important influence over populations inhabiting the southernmost South American region, known as Tierra del Fuego. This region has a harsh climate, with strong winds and occasional winter temperatures of −20◦ C. The combination of wind, cold, and rain, makes Tierra del Fuego a challenging environment for humans (Hern´andez et al. 1997). However, the Fueguians lived almost naked and Beagle Channel’s (Y´amana) and Austral Islander (Kaw´eskar) females swam frequently in the extremely cold water to collect mollusks (Hern´andez et al. 1997). The human populations form Tierra del Fuego are also similar in metabolism, body mass, stature, corporal proportion, and robusticity to Inuits from Alaska and Sami from Scandinavia, suggesting that cold could play an important role in morphological and physiological variation (Hern´andez et al. 1997; Pearson and Millones 2005). In the present study we found longest crania and larger size among southernmost terrestrial hunter-gatherers. This pattern is the opposite of the expectation of phenotypic plasticity response, and agrees with the pattern generated by selection to cold (Riesenfeld 1981), suggesting the idea of selection as an evolutionary factor behind the craniofacial divergence of southernmost populations (i.e., Tierra del Fuego groups). However, the latitudinal patterns of shape variation independent of diet for cranial vault and base could not be explained only by directional selection. This pattern also could be the result of a balance of factors such as gene flow and directional selection (Slatkin 1987; Barton et al. 2007). Natural selection could favor the adaptations to local conditions, leading to the divergence of southernmost populations (i.e., Tierra del Fuego groups) during the Holocene (Pearson and Millones 2005). Gene flow among Tierra del Fuego and other South American groups may spread new genes among populations (Slatkin 1987), generating the clinal pattern of variation among populations observed during the later late Holocene. In summary, the biological diversification of human populations from southern South America was a complex process involving multiple dimensions (population division, morphological, and ecological variation) that resulted in great craniofacial diversification. This diversification involved larger morphological changes than expected by genetic drift alone given the short period of evolutionary time. However, our results did not point out that these nonrandom factors are the unique factors shaping the patterns of variation in this region. In addition, the three developmentally distinct regions of the skull (Cheverud 1982,

990

EVOLUTION APRIL 2009

1995; Lieberman et al. 2000; Hallgr´ımsson et al. 2004, 2007; Mitteroecker and Bookstein 2008) probably were influenced by different ecological factors, as suggested by the different relationships of size and shape for each skull region with ecological variables. However, it is necessary to evaluate whether the craniofacial units are directly related to the ecological dimensions in this region. The corporal morphology (i.e., body mass, body size) and/or metabolic characteristics are most likely the features under ecological influence, and the shape and size of the craniofacial units could be a correlated consequence of change in these characteristics. The existence of a correlated response is expected by quantitative genetic theory (Lande 1979); however, further studies are required to discuss this issue more in depth. Finally, the short time scale and large magnitude of diversification observed among human populations in southern South America could be important to understand the evolutionary processes involved in the morphological diversification within Homo, during the Pleistocene–Holocene climatic and cultural changes. Humans have occupied a wide variety of ecosystems, and its populations have been exposed to different ecological stressors in time and space, showing considerable morphological variation (e.g., Roberts 1953; Holliday 1997; Katzmarzyk and Leonard 1998; Pearson 2000). Our results show that large divergence in craniofacial features can arise in a short time interval, and moreover that the influence of nonrandom factors on human craniofacial diversification should not be neglected in systematic studies of fossil Homo species. ACKNOWLEDGMENTS The authors would like to thank the staff of institutions from Argentina and Chile for granting access to the human skeletal collections under their care. We are sincerely grateful to S. F. dos Reis and R. Lande for discussions and comments about morphological diversification and divergence rate tests, which greatly improved this article. V. Bernal, P. Gonzalez, M. Nogueira, and S. F. dos Reis provided insightful comments on previous versions of the manuscript. We are deeply indebted to Gene Hunt and two anonymous reviewers who contributed greatly to improve the clarity and content of the manuscript. We also thank to Grammar101 editors for help with the English version of the manuscript. M. Perez helped with the drawings. SIP was supported by fellowships from Consejo Nacional de Investigaciones Cient´ıficas y T´ecnicas (CONICET) and Fundac¸a˜ o de Amparo a` Pesquisa do Estado de S˜ao Paulo (FAPESP). LRM was supported by fellowships and grants from the Fundac¸a˜ o de Amparo a Pesquisa do Rio de Janeiro (FAPERJ) and Conselho Nacional de Desenvolvimento Cient´ıfico e Tecnol´ogico (CNPq).

LITERATURE CITED Ackermann, R. R., and J. M. Cheverud. 2002. Discerning evolutionary processes in patterns of tamarin (genus Saguinus) craniofacial variation. Am. J. Phys. Anthropol. 117:260–271. ———. 2004. Detecting genetic drift versus selection in human evolution. Proc. Natl. Acad. Sci. USA 101:17946–17951. Adams, D. C., F. J. Rohlf, and D. E. Slice. 2004. Geometric morphometrics: ten years of progress following the ‘revolution.’ Ital. J. Zool. 71:5–16.

N O N R A N D O M FAC TO R S I N M O D E R N H U M A N D I V E R S I F I C AT I O N

Anderson, D. G., and J. C. Gillam. 2001. Paleoindian interaction and mating networks: reply to Moore and Moseley. Am. Antiq. 66:530–535. Barton, N. H., D. E. G. Briggs, J. A. Eisen, D. B. Goldstein, and N. H. Patel. 2007. Evolution. Cold Spring Harbor Laboratory Press, New York. Beals, K. L., C. L. Smith, and S. M. Dodd. 1983. Climate and the evolution of brachycephalization. Am. J. Phys. Anthropol. 62:425–437. Berberi´an, E. E., and A. E. Nielsen. 2001. Historia Argentina prehisp´anica. Brujas, C´ordoba, Argentina. Bernal, V. 2008. Procesos de diferenciaci´on biol´ogica entre poblaciones humanas del Holoceno tard´ıo de Patagonia. Una aproximaci´on desde la variaci´on m´etrica dental. Ph.D. Dissertation. Universidad Nacional de La Plata, La Plata, Argentina. Bernal, V., S. I. Perez, and P. N. Gonzalez. 2006. Variation and causal factors of craniofacial robusticity in Patagonian hunter-gatherers from the late Holocene. Am. J. Hum. Biol. 18:748–765. Binford, L. R. 2000. Constructing frames of reference. An analytical method for archaeological theory building using hunter-gatherer and environmental data sets. Univ. of California Press, Berkeley, CA. Bookstein, F. L. 1991. Morphometric tools for landmark data. Geometry and biology. Cambridge Univ. Press, Cambridge. ———. 1997. Landmark methods for forms without landmarks: localizing group differences in outline shape. Med. Image Anal. 1:225–243. Bookstein, F. L., A. P. Streissguth, P. D. Sampson, P. D. Connor, and H. M. Barr. 2002. Corpus callosum shape and neuropsychological deficits in adult males with heavy fetal alcohol exposure. Neuroimage 15:233–251. Borrero, L. A. 1999. The prehistoric exploration and colonization of FuegoPatagonia. J. World Prehistory 13:321–355. Brandt, M., and M. I. Siegel. 1978. The effects of stress on cortical bone thickness in rodents. Am. J. Phys. Anthropol. 49:31–34. Buikstra, J., and Ubelaker, D. 1994. Standards for Data Collection from Human Skeletal Remains. Arkansas Archaeological Survey, Research Series 44, Fayetteville. Cadien, J., E. Harris, W. Jones, and L. Mandarino. 1976. Biological lineages, skeletal populations and microevolution. Yearb. Phys. Anthropol. 18:194–201. Carroll, S. P., A. P. Hendry, D. N. Reznick, and C. W. Fox. 2007. Evolution on ecological time-scales. Funct. Ecol. 21:387–393. Carson, E. A. 2006. Maximum likelihood estimation of human craniometric heritabilities. Am. J. Phys. Anthropol. 131:169–180. Cavalli-Sforza, L. L., and W. F. Bodmer. 1971. The genetics of human populations. General Publishing Company, Toronto. Cavalli-Sforza, L. L., P. Menozzi, and A. Piazza. 1994. The history and geography of human genes. Princeton Univ. Press, Princeton, NJ. Cheverud, J. M. 1982. Phenotipic, genetic and environmental morphological integration in the cranium. Evolution 36:499–516. ———. 1988. A comparison of genetic and phenotypic correlations. Evolution 42:958–968. ———. 1995. Morphological integration in the saddle-back tamarin (Saguinus fuscicollis) cranium. Am. Nat. 145:63–89. ———. 2007. The relationship between development and evolution through heritable variation. Pp. 55–65 in G. Bock and J. Goode, eds. Tinkering: the microevolution of development. John Wiley & Sons, Ltd., West Sussex, UK. Curtis, N., K. Kupczik, P. O’higgins, M. Moazen, and M. Fagan. 2008. Predicting skull loading: applying multibody dynamics analysis to a macaque skull. Anat. Rec. 291:491–501. Dejean, C. B, J. L. Lanata, L. Martino, F. R. Carnese, and A. Osella. 2007. Demograf´ıa y distribuci´on de haplogrupos mitocondriales durante la dispersi´on inicial en las Am´ericas. Revista Argentina de Antropolog´ıa Biol´ogica. 9:51.

Fenner, J. 2005 Cross-cultural estimation of the human generation interval for use in genetics-based population divergence studies. Am. J. Phys. Anthropol. 128:415–423. Frisancho, A.R. 1996. Human adaptation and accommodation. The Univ. of Michigan Press, Ann. Arbor, MI. ´ Garc´ıa-Bour, J., A. P´erez-P´erez, S. Alvarez, E. Fern´andez, A. M. L´opez-Parra, E. Arroyo-Pardo, and D. Turb´on. 2004. Early population differentiation in extinct aborigines from Tierra del Fuego-Patagonia: ancient mtDNA sequences and Y-chromosome STR characterization. Am. J. Phys. Anthropol. 123:361–370. Gil, A., N. Shelnut, G. Neme, R. Tykot, C. Michieli. 2006. Is´otopos estables y dieta humana en el centro oeste: datos muestras de San Juan. Revista Cazadores y Recolectores del Cono Sur. 1:149–161. Gower, J. C. 1971. Statistical methods of comparing different multivariate analyses of the same data. Pp. 138–149 in F. R. Hodson, D. G. Kendall and P. Tautu, eds. Mathematics in the archaeological and historical sciences. Edinburgh Univ. Press, Edinburgh. Hair, J. F., R. L. Tatham, R. E. Anderson, and W. Black. 1998. Multivariate data analysis. Prentice Hall, New York. Hallgr´ımsson, B. K., C. D. Willmore, and D. M. L. Cooper. 2004. Craniofacial variability and modularity in macaques and mice. J. Exp. Zool. (Mol. Dev. Evol.) 302B:207–225. Hallgr´ımsson, B., D. E. Lieberman, W. Liu, A. F. Hutchinson, and F. R. Jirik. 2007. Epigenetic interactions and the structure of phenotypic variation in the cranium. Evol. Dev. 9:76–91. Hanken, J., and B. Hall. 1993. The skull. Univ. of Chicago Press, Chicago. Harvati, K., and T. D. Weaver. 2006. Human cranial anatomy and the differential preservation of population history and climate signatures. Anat. Rec. Part A 288A:1225–1233. Hern´andez, M., C. Lalueza Fox, and C. Garc´ıa-Moro. 1997. Fueguian cranial morphology: the adaptation to a cold, harsh environment. Am. J. Phys. Anthropol. 103:103–117. Hendry, A. P., and M. T. Kinnison. 1999. The pace of modern life: measuring rates of contemporary microevolution. Evolution 53:16371653. Holliday, T. W. 1997. Body proportions in Late Pleistocene Europe and modern human origins. J. Hum. Evol. 32:423–447. Hunt, G. 2004. Phenotypic variance inflation in fossil samples: an empirical assessment. Paleobiology 30:487–506. Hunt, G., M. A. Bell, and M. P. Travis. 2007. Evolution toward a new adaptive optimum: phenotypic evolution in a fossil stickleback lineage. Evolution 62–3:700–710. Jackson, D. A. 1995. PROTEST: a Procrustean randomization test of commu´ nity environment concordance. Ecoscience 2:297–303. Katzmarzyk, P. T., and W. R. Leonard. 1998. Climatic influences on human body size and proportions: ecological adaptations and secular trends. Am. J. Phys. Anthropol. 106:483–503. Kelly, J., S. Grammer, R. H. Tykot, J. Belardi, L. Borrero, J. Gomez-Otero, R. Guich´on. 2001. Integrating archaeological, ethnographic and analytical subsistence data: a case study from Patagonia, South America. 66th Annual Meeting of the Society for American Archaeology. New Orleans, Louisiana. Kennett, D. J., and B. Winterhalder. 2006. Behavioral ecology and the transition to agriculture. Univ. of California Press, Berkeley, CA. Klingenberg, C. P., and L. R. Monteiro. 2005. Distances and directions in multidimensional shape spaces: implications for morphometric applications. Syst. Biol. 54:678–688. Lahr, M. M. 1995. Patterns of modern human diversification: implications for Amerindian origins. Yearb. Phys. Anthropol. 38:163–198. Lalueza, C., A. P´erez-P´erez, E. Prats, L. Cornudella, and D. Turb´on. 1997. Lack of founding Amerindian mitochondrial DNA lineages in extinct

EVOLUTION APRIL 2009

991

S . I . P E R E Z A N D L . R . M O N T E I RO

aborigines from Tierra del Fuego-Patagonia. Hum. Mol. Genet. 6:41– 46. Lanata, J. L., L. Martino, A. Osella, and A. Garcia-Herbst. 2006. Ambiente y demograf´ıa durante la dispersi´on humana inicial en Sudam´erica. In C. L´opez and G. Ospina eds. Ecolog´ıa hist´orica: interacciones sociedadambiente en distintas escalas espacio temporales. Universidad Tecnol´ogica de Pereira, Pereira, Colombia. In press. Lande, R. 1977. Statistical tests for natural selection on quantitative characters. Evolution 31:442–444. ———. 1979. Quantitative genetic analysis of multivariate evolution, applied to brain: body size allometry. Evolution 33:402–416. Legendre, P., F. J., Lapointe, and P. Casgrain. 1994. Modeling brain evolution from behavior: a permutational regressio approach. Evolution 48:1487– 1499. Leonard, W. R., J. J. Snodgrass, and M. V. Sorensen. 2005. Metabolic adaptation in indigenous Siberian populations. Annu. Rev. Anthropol. 34:451– 471. Lemos, B., G. Marroig, and R. Cerqueira. 2001. Evolutionary rates and stabilizing selection in large-bodied opossum skulls (Didelphimorphia: Didelphidae). J. Zool. (Lond.) 255:181–189. Levin, S. A. 1992. The problem of pattern and scale in ecology: the Robert H. MacArthur award lecture. Ecology 73:1943–1967. Lieberman D. E., O. M. Pearson, and K. M. Mowbray. 2000. Basicranial influence on overall cranial shape. J. Hum. Evol. 38:291–315. Lockwood, C. A., W. H. Kimbel, and J. M. Lynch. 2004. Morphometrics and hominoid phylogeny: support for a chimpanzee-human clade and differentiation among great ape subspecies. Proc. Natl. Acad. Sci. USA 101:4356–4360. Losos, J. B. 1999. Uncertainty in the reconstruction of ancestral character states and limitations on the use of phylogenetic comparative methods. Anim. Behav. 58:1319–1324. Losos, J. B., K. I. U’arheit, and T. Mr. Schoener. 1997. Adaptive differentiation following experimental island colonization in Anolis lizards. Nature 387:70–73. Lynch, M. 1990. The rate of morphological evolution in mammals from the standpoint of the neutral expectation. Am. Nat. 136:727–741. Marriog, G., and J. Cheverud. 2004. Cranial evolution in Sakis (Pithecia, Platyrrhini) I: interspecific differentiation and allometric patterns. Am. J. Phys. Anthropol. 125:266–278. Mitteroecker, P., and F. Bookstein. 2008. The evolutionary role of modularity and integration in the hominoid cranium. Evolution 62:943–958. Monteiro, L. R., and J. L. Gomes-Jr. 2005. Morphological divergence rate tests for natural selection: uncertainty of parameter estimation and robustness of results. Genet. Mol. Biol. 28:345–355. Moraga, M., P. Rocco, J. F. Miquel, F. Nervi, E. Llop, R. Chakraborty, F. Rothhammer, and P. Carvallo. 2000. Mitochondrial DNA polymorphisms in Chilean aboriginal populations: implications for the peopling of the southern cone of the continent. Am. J. Phys. Anthropol. 113:19–29. Morriss-Kay, G. M. 2001. Derivation of the mammalian skull vault. J. Anat. 199:143–151. Moss, M. L., and R. W. Young. 1960. A functional approach to craniology. Am. J. Phys. Anthropol. 18:281–292. Neves, W. A., J. F. Powell, and E. G. Ozolins. 1999. Extra-continental morphological affinities of Palli Aike, Southern Chile. Interciencia 24:258– 263. Oksanen, J., R. Kindt, P. Legendre, B. O’Hara, G. L. Simpson, and M. H. H. Stevens. 2008. vegan: Community Ecology Package. R package version 1.11–4. http://cran.r-project.org Opperman, L. A. 2000. Cranial sutures as intramembranous bone growth sites. Dev. Dyn. 219:472–485. Opperman, L. A., P. T. Gakunga, and D. S. Carlson. 2005. Genetic factors

992

EVOLUTION APRIL 2009

influencing morphogenesis and growth of sutures and synchondroses in the craniofacial complex. Semin. Orthod. 11:199–208. Pearson, O. M. 2000. Activity, climate, and postcranial robusticity implications for modern human origins and scenarios of adaptive change. Curr. Anthropol. 41:569–207. Pearson, O. M., and M. Millones. 2005. Rasgos esqueletales de adaptaci´on al clima y a la actividad entre los habitantes abor´ıgenes de Tierra del Fuego. Magallania 33:37–50. Peres-Neto, P. R., and D. A. Jackson. 2001. How well do multivariate data sets match? The advantages of a Procrustean superimposition approach over the Mantel test. Oecologia (Berl.). 129:169–178. Perez, S. I., V. Bernal, and P. Gonzalez. 2006. Differences between sliding semilandmarks methods: implications for shape analyses of human populations. J. Anat. 208:769–784. ———. 2007a. Morphological differentiation of aboriginal human populations from Tierra del Fuego (Patagonia): implications for South American peopling. Am. J. Phys. Anthropol. 133;1067–1079. ———. 2007b. Evolutionary relationships among prehistoric human populations: an evaluation of facial morphometric data employing molecular based genealogies. Hum. Biol. 79:25–50. Pucciarelli, H. M. 1980. The effects of race, sex, and nutrition on craniofacial differentiation in rats. A multivariate analysis. Am. J. Phys. Anthropol. 53:359–368. ———. 1981. Growth of the functional components of the rat skull and its alteration by nutritional effects. A multivariate analysis. Am. J. Phys. Anthropol. 56:33–41. Pucciarelli, H. M., and E. E. Oyhenart. 1987. Effects of maternal food restriction during lactation on craniofacial growth in weanling rats. Am. J. Phys. Anthropol. 72:67–75. Rae, T. C., U. Strand Viðarsd´ottir, N. Jeffery, and A. T. Steegmann, Jr. 2006. Developmental response to cold stress in cranial morphology of Rattus: implications for the interpretation of climatic adaptation in fossil hominins. Proc. R. Soc. Lond. B 273:2605–2610. R Development Core Team. 2008. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org Relethford, J. H. 1994. Craniometric variation among modern human populations. Am. J. Phys. Anthropol. 95:53–62. ———. 2002. Apportionment of global human genetic diversity based on craniometrics and skin color. Am. J. Phys. Anthropol. 118:393–398. ———. 2004. Global patterns of isolation by distance based on genetic and morphological data. Hum. Biol. 76:499–513. Reznick, D. N., F. H. Shaw, F. H. Rodd, and R. G. Shaw. 1997. Evaluation of the rate of evolution in natural populations of guppies (Poecilia reticlrlatri). Science 275:1934–1937. Riesenfeld, A. 1973. The effect of extreme temperatures and starvation on the body proportions of the rat. Am. J. Phys. Anthropol. 39:427–460. ———. 1981. The role of body mass in thermoregulation. Am. J. Phys. Anthropol. 55:95–99. Roberts, D. F. 1953. Body weight, race and climate. Am. J. Phys. Anthropol. 11:533–558. Roff, D. A. 1996. The evolution of genetic correlations: an analysis of patterns. Evolution 50:1392–1403. Rohlf, F. J. 1993. Relative warps analysis and an example of its application to Mosquito wings. Pp. 132–159 in L. F. Marcus, E. Bello and A. Garc´ıa-Valdecasas, eds. Contributions to morphometrics. Monograf´ıas del Museo Nacional de Ciencias Naturales, Madrid. ———. 2007. tps serie softwares. Available at http//life.bio.sunysb.edu/ morph/ Rohlf, F. J., and D. E. Slice. 1990. Extensions of the Procrustes Method for the optimal superimposition of landmarks. Syst. Zool. 39:40–59.

N O N R A N D O M FAC TO R S I N M O D E R N H U M A N D I V E R S I F I C AT I O N

Roseman, C. C. 2004. Detecting interregionally diversifying natural selection on modern human cranial form by using matched molecular and morphometric data. Proc. Natl. Acad. Sci. USA 101:12824– 12829. Rothhammer, F., and C. Silva. 1990. Craniometrical variation among South American prehistoric populations: climatic, altitudinal, chronological and geographic contributions. Am. J. Phys. Anthropol. 82:9–17. Ruff, C. B. 1994. Morphological adaptation to climate in modern and fossil hominids. Yearb. Phys. Anthropol. 37:65–107. Sardi, M. L., and F. V. Ramirez-Rozzi. 2005. A cross-sectional study of human craniofacial growth. Ann. Hum. Biol. 32:390–396. Sardi, M. L., F. Ram´ırez-Rozzi, R. Gonz´alez-Jos´e, and H. M. Pucciarelli. 2005. South Amerindian craniofacial morphology: diversity and implications for Amerindian evolution. Am. J. Phys. Anthropol. 128:747– 756. Scheuer, L., and S. Black. 2000. Developmental juvenile osteology. Academic Press, London. Schluter, D. 2000. The ecology of adaptative radiation. Oxford Univ. Press, New York. Sheets, H. D., and C. E. Mitchell. 2001. Why the null matters: statistical tests, random walks and evolution. Genetica 112–113:105–125. Slatkin, M. 1987. Gene flow and the geographic structure of natural populations. Science 236:787–792. SMN. 1961–1990. Climatic database of Servicio Meteorol´ogico Nacional. Available at http://www.smn.gov.ar/

Sober, E. 1988. Reconstructing the past. Parsimony, evolution and inference. The MIT Press, Cambridge, MA. Sperber, G. H. 2001. Craniofacial development. BC Decker Inc., Hamilton. Spicer, G. S. 1993. Morphological evolution of the Drosophila virilis species group as assessed by rate tests for natural selection on quantitative characters. Evolution 47:1240–1254. Steegmann, A. T., and W. S. Platner. 1968. Experimental cold modification of cranio-facial morphology. Am. J. Phys. Anthropol. 28:17–30. Steward, J. H. 1950. Handbook of South American Indians. Pub. Official, Washington. Stynder, D. D., R. R. Ackermann, J. C. Sealy. 2007. Craniofacial variation and population continuity during the South African Holocene. Am. J. Phys. Anthropol. 134:489–500. Turelli, M., J. H. Gillespie, and R. Lande. 1988. Rate tests for selection on quantitative characters during macroevolution and microevolution. Evolution 42:1085–1089. Vitzthum, V. J. 2003. A number no greater than the sum of its parts: the use and abuse of heritability. Hum. Biol. 75:539–558. Weaver, T. D., C. C. Roseman, and C. B. Stringer. 2007. Were neandertal and modern human cranial differences produced by natural selection or genetic drift? J. Hum. Evol. 53:135–145. Wobst, M. 1974. Boundary conditions for paleolithic social systems: a simulation approach. Am. Antiq. 39:147–178.

Associate Editor: G. Hunt

EVOLUTION APRIL 2009

993