THE ANATOMICAL RECORD 298:168–179 (2015)
Functional Morphology of the Neandertal Scapular Glenoid Fossa 1
MARISA E. MACIAS1* AND STEVEN E. CHURCHILL1,2 Department of Evolutionary Anthropology, Duke University, Durham, North Carolina 2 Evolutionary Studies Institute, University of the Witwatersrand, Wits, South Africa
ABSTRACT Neandertals and Homo sapiens are known to differ in scapular glenoid fossa morphology. Functional explanations may be appropriate for certain aspects of glenoid fossa morphology; however, other factors—e.g., allometry, evolutionary development—must be addressed before functional morphology is considered. Using three-dimensional geometric morphometrics, shape of the scapular glenoid fossa was compared among Neandertals, early and recent modern humans, chimpanzees, orangutans, Australopithecus afarensis, and Au. sediba. Permutation analysis revealed that side, sex, and lifestyle did not correlate with shape. Of the features we found to differ between groups, anterior glenoid rim morphology and fossa curvature did not correlate with the aforementioned shape variables; thus, a functional explanation is appropriate for these components of glenoid fossa shape. Shared morphology among recent humans and chimpanzees (to the exclusion of Neandertals and orangutans) suggests independent forces contributing to these morphological configurations. Potential explanations include adaptations to habitual behavior and locomotor adaptations in the scapulae of recent humans and chimpanzees; these explanations are supported by clinical and experimental literature. The absence of these morphological features in Neandertals may support the lack of these selective forces on their scapular glenoid fossa morphology. Anat Rec, 298:168–179, C 2014 Wiley Periodicals, Inc. 2015. V
Key words: geometric morphometrics; functional morphology; Neandertal
The functional morphology of the primate shoulder joint has long been studied from both neontological (Oxnard, 1967; Roberts, 1974; Corruccini and Ciochon, 1976; Larson, 1995; Young, 2006) and paleontological perspectives (Stewart, 1962; Ciochon and Corruccini, 1976; Trinkaus, 1983; Churchill and Trinkaus, 1990; Rose, 1993; MaClatchy et al., 2000; Di Vincenzo et al., 2012). One area of ongoing research concerns the significance of morphological differences in the scapular glenoid fossa between Neandertals (Homo neanderthalensis) and modern humans (H. sapiens). Neandertal scapular glenoid fossae tend to be supero-inferiorly elongated and anteroposteriorly narrow and shallow relative to those of early and recent modern humans (Stewart, 1962; Smith, 1976; Trinkaus, 1983; Churchill and Trinkaus, 1990; Trinkaus, 2006a). In contrast, modern humans tend to have glenoid fossae that are relatively broad antero-posteriorly, “twisted” anteriorly to posteriorly (i.e., with a more laterC 2014 WILEY PERIODICALS, INC. V
ally projecting anterior margin of the joint surface, see Fig. 1), and more concave inferiorly than superiorly (Inui et al., 2001). Although the modern human and Neandertal ranges of variation overlap in these features, such that glenoid fossa morphology does not completely distinguish Neandertals from more modern hominins, central tendencies differ between these groups (Smith, 1976).
*Correspondence to: Marisa E. Macias, Department of Evolutionary Anthropology, Duke University, Durham, NC 27708. Fax: 919-660-7348. E-mail: [email protected] Received 3 October 2014; Accepted 11 October 2014. DOI 10.1002/ar.23072 Published online in Wiley Online Library (wileyonlinelibrary. com).
SCAPULAR GLENOID FOSSA MORPHOLOGY
Fig. 1. Cross-section of scapular glenoid fossa curvature. (A) Schematic illustration of a right scapula and humerus in anatomical position (anterior view), showing a cross-section through the glenohumeral joint. (B) Cross-section in superior view, showing an antero-posteriorly curved (i.e., deep) glenoid fossa in red. (C) Cross-section in superior view, showing an antero-posteriorly flat (i.e., shallow) glenoid fossa in red.
The morphology of Neandertal scapular glenoid fossae has been studied from several, nonmutally-exclusive perspectives. Allometric scaling has been argued to explain some aspects of the observed differences between groups. Neandertals have large joint surfaces relative to their limb lengths throughout the postcranial skeleton (Trinkaus, 1983; Trinkaus et al., 1991), and glenoid length and depth have been shown to be proportional to humeral head size (Churchill and Trinkaus, 1990). Thus, differences in glenoid shape may partly reflect scaling proportions and glenohumeral joint size. Neandertal glenoid fossae may reflect ancestry. Trinkaus (2006a) hypothesized that a long, superoinferiorly narrow glenoid may be the primitive condition for hominins. Neandertals share relatively narrow glenoid fossae with the Australopithecus africanus specimen Sts 7 (Vrba, 1979; Carretero et al., 1997) and early Homo fossils KNM-WT 15000 and Dmanisi 4166 (Trinkaus, 2006a,b). By this view, the relatively broad glenoid fossa of a late Mousterian Neandertal from Vindija Cave (Croatia), Vi-209, might be seen as evidence for some degree of genetic continuity between late Neandertals and early modern Europeans (Smith and Trinkaus, 1991) or introgression of modern human genes into the late Neandertal gene pool (Di Vincenzo et al., 2012). Evolutionary development has been argued to influence the scapular glenoid fossa. A recent twodimensional geometric morphometric analysis of Neandertal glenoid fossa shape (Di Vincenzo et al., 2012) suggests that most of the Neandertal/modern human morphological contrasts in glenoid fossa shape are attributable to differences in evolutionary-developmental growth rates. Di Vincenzo et al. (2012) hypothesize that heterochrony in overall somatic growth accounts for observed between-group differences in glenoid width. Relative amounts of growth in the inferior glenoid rim secondary center of ossification may play a substantial role in glenoid width and curvature, such that relatively broad modern human fossae may be a morphological consequence of a slowed growth trajectory (Di Vincenzo et al., 2012; see also Churchill and Trinkaus, 1990). Although this hypothesis remains untested, as the influence of evolutionary shifts in somatic growth rates on scapular ontogeny have yet to be fully explored, future studies may assess this possibility. Regardless of the causes of variation in glenoid fossa morphology (both within and between groups), the
169
functional significance of these differences remains an open question. Contrasts between Neandertal and modern human upper limb morphology have traditionally been interpreted in terms of differences in the habitual use of various forms of subsistence technology in the evolutionary history of the two groups. Although early modern humans of the later African Middle Stone Age and the Eurasian Upper Paleolithic are thought to have been more heavily invested in the use of throwing-based, long-range projectile weaponry, Neandertals may have primarily relied upon close range hunting technology, primarily large and heavy thrusting spears (see Churchill and Rhodes, 2009). These very different technologies are associated with distinct kinematics and patterns of muscle activation (Schmitt et al., 2003). Because habitual patterns of muscle engagement and biomechanical loads may influence skeletal morphology, upper limb bones may reflect hunting behaviors. Anatomical differences in the shoulders of modern humans and Neandertals have thus been interpreted as reflecting little to no reliance on spear throwing in the Neandertal predatory repertoire (Churchill and Trinkaus, 1990; Churchill and Rhodes, 2009; Rhodes and Churchill, 2009). Some variation in glenoid fossa shape, both between Neandertals and modern humans and between modern human groups, may be a function of developmental plasticity in joint morphology in the context of habitual differences in upper limb use (Di Vincenzo et al., 2012). However, scapular architecture is relatively poorly understood from a functional perspective (Alemseged et al., 2006). The relative breadth of the scapular glenoid fossa alone has been subjected to multiple functional interpretations. The relatively narrow glenoid fossa of the Muierii 1 early modern human scapula from Romania has been hypothesized to indicate parity in the shoulder loading regimes between Neandertals and this individual, that is, a lack of throwing-based projectile weaponry (Trinkaus, 2008; Trinkaus, 2011). Although Di Vincenzo et al. (2012) did not include this specimen in their study, their conclusion that relative width reflects developmental rate would suggest that the relative breadth of the glenoid reflects parity in developmental rates with Neandertals, rather than a functional signal. Thus, debate persists in the interpretation of glenoid fossa morphology. What little is definitively known about the functional consequences of glenoid fossa shape is based in human clinical observations. These studies have focused on the association of glenohumeral stability with loss of function and pain. Independent of the soft tissues (including the glenoid articular cartilage and glenohumeral labrum), the bony anatomy of the glenohumeral joint is important for stability (De Wilde et al., 2004; Lugo et al., 2008). Two salient morphological features have been established as a contributing factor to anterior shoulder stability. (1) Concavity of the glenoid is critical to humeroglenoid stability, as individuals with less curved, shallower glenoid fossae are more likely to experience anterior dislocation (Lippitt et al., 1993; Itoi et al., 2000). (2) Anterior rim glenoid morphology can also affect propensity to shoulder dislocation; individuals with lesions or less laterally projecting anterior rim morphology are also more prone to pain, dislocation, and loss of function (Bigliani et al., 1998; Stevens et al., 1999; Burkhart et al., 2002; Sugaya et al., 2003;
170
MACIAS AND CHURCHILL
TABLE 1. Fossil taxa used in this study Taxon
Specimen no.
Au. afarensis (N 5 1) AL 288 Au. sediba (N 5 1) MH 2 H. neanderthalensis (N 5 11) Krapina 127 Krapina 129 Krapina 130 Krapina 131 Krapina 132 Krapina 133 Neandertal 1 La Ferrassie 1 La Ferrassie 2 Tabun C1 H. sapiens (N 5 3) Oberkassel 1 Oberkassel 2
Locality
Age
Hadar (Ethiopia) Malapa (South Africa) Krapina (Croatia) Krapina (Croatia) Krapina (Croatia) Krapina (Croatia) Krapina (Croatia) Krapina (Croatia) Feldhofer Grotto (Germany) La Ferrassie (France) La Ferrassie (France) Tabun (Israel) Bonn-Oberkassel (Germany) Bonn-Oberkassel (Germany)
Early Pleistocene Pliocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene
Huysmans et al., 2006). Appreciating morphological variation in these two features may be necessary to better understand which aspects of morphological differences might contain information about functional adaptation or habitual behavior patterns. Geometric morphometrics is inherently exploratory work, and this technique may be applied to understand ways in which morphology varies within and between groups as a basis for generating testable functional hypotheses. However, adaptive hypotheses should not be advanced until it can be demonstrated that other explanations (e.g., static allometry) can be ruled out. Following Gould and Lewontin (1979), we investigate a hierarchy of alternatives to adaptation for the explanation of glenoid fossa shape. We argue that a reductive approach of removing confounding factors is a necessary prerequisite to understanding the functional significance of observed morphological variance in scapular morphology. We evaluate the effects of activity levels, size, and phylogenetic relationships to identify aspects of scapular glenoid fossa morphological variation that might be independent of these factors. Traits that (1) vary bilaterally within individuals, (2) vary by sex (independent of allometry), or (3) vary between human groups engaged in differing activity levels, are likely to reflect developmentally plastic aspects of glenoid fossa morphology. These traits are useful for understanding within and between group variations in habitual upper limb usage. Traits that are allometrically related to size, as well as those that may be primarily varying as a function of evolutionary differences in somatic growth rates, are considered to be less likely to inform about adaptive behavioral differences between groups. (Although evolutionary shifts in ontogeny may affect scapular morphology, this study does not assess this influence. This does not rule out the possibility that selection could operate on organismal size or growth rates if the consequence was adaptively beneficial morphology of some joint system, but we do view this possibility as unlikely). Traits that vary in clear patterns that align with phylogeny— e.g., derived traits found throughout clades but not in outgroups—are understood to be better signals of ancestry than signals of functional adaptations. Finally, features that are unrelated to differences in activity levels and patterns, size, or phylogeny are possible candidates for features that have been under the direct action of
Original or cast Original Original Cast Cast Cast Cast Cast Cast Cast Cast Cast Cast Cast Cast
Side Right Right Right Right Left Left Right Right Right Left, right Right Right Left Left, right
selection for biomechanical competence of the joint given species-level functional demands on the shoulder. Through this approach, we actively attempt to rule out other possibilities before we advance functional hypotheses to explain the observed shape variation.
MATERIALS AND METHODS Two-dimensional studies are limited in their ability to capture the complex three-dimensional relationships of a ball-and-socket joint. Clinical studies highlight the importance of three-dimensional analyses. Three-dimensionally reconstructed images were found to be more reliable than conventional two-dimensional radiographs for detecting skeletal shape anomalies of the glenoid (Stevens et al., 1999; Sugaya et al., 2003). Furthermore, past studies have failed to capture important shape information in the medio-lateral variation of glenoid fossa morphology, particularly with reference to the anterior rim. For these reasons, a three-dimensional approach is taken here. Existing studies have not addressed Neandertal scapular glenoid morphology in a greater comparative context. To understand the evolutionary patterns of hominoid morphology, as well as to inform our understanding of the effects of different functional demands on the glenoid, landmarks were digitized on specimens of modern humans, chimpanzees, orangutans, original fossil specimens, and fossil casts (N 5 187, see Tables 1 and 2). To establish more equal sample sizes across taxa (avoiding data skew), the chimpanzee and modern human datasets were randomly down-sampled to 25 individuals (N 5 86). Modern human scapulae were classified as belonging to Upper Paleolithic modern humans (Oberkassel 1 and 2) or recent modern samples, in part to take a conservative approach to intraspecific variation and in part to examine chronological variation within Homo sapiens. No scapulae with complete glenoid fossae were available to represent early Homo. The use of casts in this analysis, although not ideal, is acceptable. For those specimens where the original could be digitized, comparisons of these trials to digitized casts of the same specimen revealed no systematic error in landmark orientations. We also compared the casts to specimen photographs to assess regions that may be affected by pathology or taphonomic damage.
171
SCAPULAR GLENOID FOSSA MORPHOLOGY
TABLE 2. Comparative taxa used in this study Species H. sapiens H. sapiens H. sapiens H. sapiens H. sapiens Total P. troglodytes P. troglodytes P. troglodytes P. troglodytes Total P. abelii P. pygmaeus Total
Origin
Lifestyle
Source
Male
Female
Unknown
Total
Canyon del Muerto (AZ, USA) El Hesa (Egypt) New York (NY, USA) Point Hope (AK, USA) Indian Knoll (KY, USA)
Active
AMNH
0
0
17
17
Less active Less active Active Active
AMNH AMNH AMNH AMNH
Cameroon Gabon Uganda West Africa
Active Active Active Active
FMNH HMCZ NMNH PCM
Sumatra Borneo
Active Active
NMNH FMNH, NMNH, ZAIM
0 24 16 6 46 10 4 0 2 16 2 8 10
0 7 10 10 27 7 1 1 0 9 1 9 10
17 0 2 14 50 2 0 0 1 3 0 0 0
17 31 28 30 123 19 5 1 3 28 3 17 20
Note: geographic origin was stated in lieu of subspecies for Pan troglodytes, due to the contentious nature of subspecies definitions and designations. AMNH, American Museum of Natural History; FMNH, Field Museum of Natural History; HMCZ, Harvard Museum of Comparative Zoology; NMNH, National Museum of Natural History; PCM, Powell-Cotton Museum; ZAIM, Zurich Anthropological Institute Museum.
Recent modern human populations were selected to represent a variety of latitudes, geographic locations, and time periods to subsample human variation. To address the role plasticity plays in glenoid shape, modern humans were categorized into “active” or “lessactive” lifestyle groups. Active groups included archaeological specimens known to have engaged in hunting, foraging, or farming (nonindustrialized) over their lifetime. Less-active groups included archaeological and modern specimens representing individuals unlikely to have been engaged in hard labor or otherwise highly active. Active populations include a sample from Point Hope (Alaska), Canyon del Muerto (Arizona), and Indian Knoll (Kentucky). Inactive populations include a sample from El Hesa (Egypt) as well as a modern anatomical collection from twentieth century New York. All modern human comparative samples are curated at the American Museum of Natural History (ANMH, New York), except for the Indian Knoll sample, curated at the National Museum of Natural History (NMNH, Washington DC). All extant specimens were adults with no signs of pathology or taphonomic damage. The New York sample derives from autopsy specimens of known sex, age, and race. Sex in the Point Hope and Indian Knoll samples had either been determined by previous workers and was indicated in the museum records, or had been determined by Sharplin (unpublished study, 2011). In every case, one of us (M.M.) assessed the sex of the specimen prior to referring to the previous designation. Only specimens in which our diagnosis matched that of the prior designation were used. The sex of the specimens from Canyon del Muerto and El Hesa are unknown. All data were collected by M.M. Sample sizes and other details are provided in Table 1. The Point Hope sample is based on a cemetery sample of individuals who lived a sedentary lifestyle in an Ipiutak village, dated from 100 B.C.E. to 1400 C.E. The cemetery was excavated during 1939–1941 (Rainey, 1972; Mason, 1998). The Ipiutak largely relied on caribou, walrus, and seal hunting, with only a seasonal reliance on sea mammals (Krueger, 2006). The bow and arrow was
the principal weapon for the Ipiutak people, and although harpoon elements were found during the excavation, they were rare compared with the quantity of artifacts associated with archery (Larsen and Rainey, 1948). Previous studies of the Point Hope population suggest that these individuals were highly active, consistent with other samples engaged in strenuous upper body activity (Shackelford, 2006). Sex was confirmed with pelvic measurements when possible. The Canyon del Muerto sample derives from the early Basketmaker II stage in Southwestern Archaeology (1500–50 B.C.E.), and represents a population engaged in an agricultural economy but that still used some hunting with throwing-based projectile technology (atlatl) (Kiddler, 1927). The extent of agricultural reliance in the Canyon del Muerto population is uncertain: coprolite analysis from a contemporaneous Basketmaker II site indicates that this population relied on maize for the majority of their calories (Matson, 1991), whereas Reed (2000) questions the extent of this dependence on agriculture, but agrees that there was decreasing dependence on hunting and gathering over time. Indian Knoll is an Archaic Period site (dated to 6415– 4143 B.C.E.), and one of the largest hunter-gatherer skeletal collections in North America. Archaeological excavations of the burials also recovered shell beads, stone tools, and animal remains. Their subsistence economy relied mainly on deer, turkey, mussels, nuts and a variety of locally collected plant materials (Winters, 1974). The Egyptian individuals resided in El Hesa, located near modern day Aswan, Egypt, and the burials are dated to the Roman period between C.E. 200–400 (Elliot Smith and Wood-Jones, 1910, cited in Irish, 1993). Mortuary evidence indicates a middle class affiliation for the individuals in this sample, such that even though they derive from a pre-industrial agricultural economy (Irish, 1993), it is unlikely that the individuals sampled were engaged in agricultural or other hard labor. The New York Medical Collection consists of twentieth century New Yorkers. These individuals have known
172
MACIAS AND CHURCHILL
Fig. 2. Landmark placement on scapular glenoid fossa. Lateral view. Anchoring landmarks in blue, resampled landmarks in red. (A) Glenoid outline. 40 landmarks. (B) APCP, anterior posterior curve profile. 10 landmarks. (C) SICP, superior inferior curve profile. 15 landmarks.
age-at-death, sex, and race. The sample includes African–American and Caucasian individuals. Previous research found that race and age did not correlate with morphological variables (Macias, 2009). Comparative ape samples were collected from the AMNH, NMNH, Field Museum of Natural History (FMNH, Chicago), Harvard Museum of Comparative Zoology (HMCZ, Cambridge), Powell-Cotton Musem (PCM, Birchington), and the Zurich Anthropological Institute Museum (Zurich, ZAIM). All ape specimens were wild-shot adults with no signs of pathology or taphonomic damage (see Table 2 for details). A total of 65 semilandmarks describing the scapular glenoid fossa were analyzed (Fig. 2). Anchoring landmarks were chosen to be the most clearly defined and repeatably identifiable points on the scapular glenoid fossa, patterned after Type II landmarks described by Young (2008). Lines—defined as a series of closelyspaced landmarks between two points—were collected around the articular rim of the scapular glenoid fossa, as well as along lines representing the superior–inferior curve profile (SICP), and the anterior–posterior curve profile (APCP). Data were obtained using a Microscribe (Immersion Corp., San Jose, CA) digitizer 3DX. Landmarks were identified once the specimen was mounted and stabilized. Lines were resampled to a standard number of equally-spaced landmarks in the series (40 for the articular rim, 15 for the SICP, and 10 for the APCP) using the programs Resample (Rauum et al., 2006) and Geomorph (Adams and Otarola-Castillo, 2013). To circumvent the effect of unequally distributed points, each line was analyzed individually. Lines were subjected to generalized procrustes analysis (GPA) to remove differences in size, shape, and translation. Principal components analysis (PCA) was used to visualize variation in shape space, and explore correlations among shape variables. Shape differences between groups were tested by univariate analysis of variance using the Kruskal–Wallis test as well as permutation tests on Procrustes distances between groups. PCA scores were regressed on centroid size to explore correlations between shape and body mass. Corruccini and Ciochon (1976) found an isometric relationship between glenoid cavity height and body mass across primates; as such, glenoid size has been used as a proxy for overall size in previous studies (Alemseged et al., 2006; Green and Alemseged, 2012). Procrustes distances were used in several ways to statistically evaluate variation within the sample. We used Procrustes distances to: (1) compare antimeres–left
and right sides from the same individual (intra-individual variation), (2) compare lefts and rights across the modern human sample (inter-individual variation), and (3) explore between-group differences with permutation tests. Permutation tests, where null distributions are generated by randomly allocating individuals to groups, were carried out to assess the significance of Procrustes distances between groups. For each comparison, the observed distance between groups (i.e., the test statistic) was compared to the distribution of distances between randomly allocated groups. The observed distance was considered statistically significant when it was greater than 95% of the permuted distances. In each analysis, 10,000 iterations were calculated for each pair of taxa, giving a maximum possible P-value of 0.0001. To identify developmentally plastic aspects of scapular glenoid shape, two potential sources of variation were addressed with the modern human sample: (1) differences between left and right glenoids in the same individual, and (2) differences in shape due to activity level. To illustrate the variation observed within and between modern human left and right glenoid fossae, frequency distributions of Procrustes distances were calculated. The distribution of intra-individual distances (distances between antimeres) was compared to the interindividual distribution (pairwise distances between individuals across the modern human sample) with a Wilcoxon rank test. Permutation tests were used to test the significance of these differences in three ways: (1) to assess shape differences between lefts and rights across the modern human sample, (2) to evaluate differences between active and inactive populations, and (3) to evaluate differences between males and females. The glenoid articular outline, SICP, and APCP were analyzed separately to explore each aspect of shape independently. For each dataset, a PCA was performed, and all principal components (PCs) accounting for more than 10% of the overall variation were studied. These PC scores were then regressed on centroid size; PC scores with significant correlations were then eliminated from the analysis. The remaining PC scores were then subjected to Kruskal–Wallis tests exploring differences among groups; PCs lacking significant differences among groups were eliminated from the analysis. All statistical tests were subjected to sequential Bonferroni correction (Rice, 1989). Remaining PCs (i.e., those that were not significantly related to size and that successfully differentiated between groups) were studied with the program Morphologika (O’Higgins and Jones, 1998) to explore and visualize shape variation. The goal of these analyses was to identify aspects of scapular glenoid morphology that do not appear to be explicable in terms of size allometry, sex, or developmental plasticity related to activity levels and patterns. Once these features are isolated, we evaluate the possible functional significance of variation in these aspects. This reductive approach involved eliminating shape variables associated with size, sex, or developmental plasticity individually before exploring the functional consequences of the remaining shape variation. PC scores not rejected through this process were aggregated for a combined Canonical Variate Analysis (CVA) and Discriminant Function Analysis (DFA) to explore how well these remaining factors discriminate groups.
SCAPULAR GLENOID FOSSA MORPHOLOGY
RESULTS Variation in Modern Humans To evaluate the developmental plasticity of the glenoid fossa morphology, we assessed levels of bilateral asymmetry in the joint surface. A one-sided Wilcoxon test was performed upon individuals for whom both left and right sides were measured. The Procrustes distance between the left and right side was compared to a random subsample (10,000 of 13,688) of across-sample differences. Results were significant (P 5 0.0018), indicating that antimeres are more similar to each other than to other specimens. Permutation tests found no significant differences between left specimens and right specimens (P 5 1.00). These results suggest bilateral asymmetry is not an important component of the overall shape variation within the modern human populations sampled
173
here. Figure 3 illustrates the contrast between intraindividual and inter-individual Procrustes shape distance. Although these distributions overlap, the interindividual distribution has higher frequencies of greater distances between pairs. Developmental plasticity was also evaluated by examining differences between sexes and activity levels in our modern human groups. No significant differences were observed between males and females (P 5 0.71), or between individuals categorized as active or inactive (P 5 0.13). Because these factors did not discriminate individuals in Procrustes shape space (or in the PCA, Fig. 4a,b) individuals were thereafter combined regardless of sex or lifestyle to form a single sample of modern humans. Sides were randomly selected to create a sample of both left and right specimens.
Principal Component Analysis To examine how fossils varied in morphospace relative to extant taxa, all specimens were analyzed together via GPA and PCA. All PCs representing at least 10% of the overall variation were examined with respect to allometry and phylogenetic relationships before being subjected to further analyses.
Glenoid Fossa Outline Analysis
Fig. 3. Intra-individual and inter-individual procrustes distances. Frequency distribution of recent modern human Procrustes distances. Intra-individual, or between antimere, Procrustes distance (blue) is contrasted with inter-individual Procrustes distance (pink). Although these graphs overlap, the higher frequency of larger distances (both at the peak and particularly at the right tail) indicates that intra-individual Procrustes distances are smaller (matched pair Wilcoxon Pvalue 5 0.0004)
Results of the glenoid outline analysis were comparable to those of Di Vincenzo et al. (2012); this is unsurprising given that the analytical methods are similar. Principal Components 1–3 were not significantly correlated with centroid size (see Table 3). PC 1 explained the largest percentage of variance (25.4%), capturing supero-inferior elongation and mediolateral breadth of the glenoid fossa (Fig. 5). This measure differentiated Neandertals from chimpanzees and modern humans. Observed variation across taxa did not appear to follow a phylogenetic pattern (e.g., Pongo were significantly different from Pan and recent humans, but not Neandertals, see Table 4), but was consistent with the evolutionary trajectory identified by Di Vincenzo
Fig. 4. PCA of recent modern humans displaying variation by side (A), sex (B), and activity level (A and B). No significant differences were found between rights (R) and lefts (L), between females (F) and males (M), or between active (red) and less-active (blue) populations.
174
MACIAS AND CHURCHILL
et al. (2012). Although we did not directly test this hypothesis, we cannot rule out the possibility that the variation described on PC1 reflects evolutionary developmental differences between groups in somatic growth rates (Di Vincenzo et al., 2012). Thus, PC 1 was TABLE 3. Results of regression analyses of PC scores on natural log centroid size Principal component
r2 value
P-value
a-level
Outline PC 1 Outline PC 2 Outline PC 3 APCP PC 1 APCP PC 2 APCP PC 3 SICP PC 1 SICP PC 1* SICP PC 2
0.036 0.002 0.053
Functional Morphology of the Neandertal Scapular Glenoid Fossa 1
MARISA E. MACIAS1* AND STEVEN E. CHURCHILL1,2 Department of Evolutionary Anthropology, Duke University, Durham, North Carolina 2 Evolutionary Studies Institute, University of the Witwatersrand, Wits, South Africa
ABSTRACT Neandertals and Homo sapiens are known to differ in scapular glenoid fossa morphology. Functional explanations may be appropriate for certain aspects of glenoid fossa morphology; however, other factors—e.g., allometry, evolutionary development—must be addressed before functional morphology is considered. Using three-dimensional geometric morphometrics, shape of the scapular glenoid fossa was compared among Neandertals, early and recent modern humans, chimpanzees, orangutans, Australopithecus afarensis, and Au. sediba. Permutation analysis revealed that side, sex, and lifestyle did not correlate with shape. Of the features we found to differ between groups, anterior glenoid rim morphology and fossa curvature did not correlate with the aforementioned shape variables; thus, a functional explanation is appropriate for these components of glenoid fossa shape. Shared morphology among recent humans and chimpanzees (to the exclusion of Neandertals and orangutans) suggests independent forces contributing to these morphological configurations. Potential explanations include adaptations to habitual behavior and locomotor adaptations in the scapulae of recent humans and chimpanzees; these explanations are supported by clinical and experimental literature. The absence of these morphological features in Neandertals may support the lack of these selective forces on their scapular glenoid fossa morphology. Anat Rec, 298:168–179, C 2014 Wiley Periodicals, Inc. 2015. V
Key words: geometric morphometrics; functional morphology; Neandertal
The functional morphology of the primate shoulder joint has long been studied from both neontological (Oxnard, 1967; Roberts, 1974; Corruccini and Ciochon, 1976; Larson, 1995; Young, 2006) and paleontological perspectives (Stewart, 1962; Ciochon and Corruccini, 1976; Trinkaus, 1983; Churchill and Trinkaus, 1990; Rose, 1993; MaClatchy et al., 2000; Di Vincenzo et al., 2012). One area of ongoing research concerns the significance of morphological differences in the scapular glenoid fossa between Neandertals (Homo neanderthalensis) and modern humans (H. sapiens). Neandertal scapular glenoid fossae tend to be supero-inferiorly elongated and anteroposteriorly narrow and shallow relative to those of early and recent modern humans (Stewart, 1962; Smith, 1976; Trinkaus, 1983; Churchill and Trinkaus, 1990; Trinkaus, 2006a). In contrast, modern humans tend to have glenoid fossae that are relatively broad antero-posteriorly, “twisted” anteriorly to posteriorly (i.e., with a more laterC 2014 WILEY PERIODICALS, INC. V
ally projecting anterior margin of the joint surface, see Fig. 1), and more concave inferiorly than superiorly (Inui et al., 2001). Although the modern human and Neandertal ranges of variation overlap in these features, such that glenoid fossa morphology does not completely distinguish Neandertals from more modern hominins, central tendencies differ between these groups (Smith, 1976).
*Correspondence to: Marisa E. Macias, Department of Evolutionary Anthropology, Duke University, Durham, NC 27708. Fax: 919-660-7348. E-mail: [email protected] Received 3 October 2014; Accepted 11 October 2014. DOI 10.1002/ar.23072 Published online in Wiley Online Library (wileyonlinelibrary. com).
SCAPULAR GLENOID FOSSA MORPHOLOGY
Fig. 1. Cross-section of scapular glenoid fossa curvature. (A) Schematic illustration of a right scapula and humerus in anatomical position (anterior view), showing a cross-section through the glenohumeral joint. (B) Cross-section in superior view, showing an antero-posteriorly curved (i.e., deep) glenoid fossa in red. (C) Cross-section in superior view, showing an antero-posteriorly flat (i.e., shallow) glenoid fossa in red.
The morphology of Neandertal scapular glenoid fossae has been studied from several, nonmutally-exclusive perspectives. Allometric scaling has been argued to explain some aspects of the observed differences between groups. Neandertals have large joint surfaces relative to their limb lengths throughout the postcranial skeleton (Trinkaus, 1983; Trinkaus et al., 1991), and glenoid length and depth have been shown to be proportional to humeral head size (Churchill and Trinkaus, 1990). Thus, differences in glenoid shape may partly reflect scaling proportions and glenohumeral joint size. Neandertal glenoid fossae may reflect ancestry. Trinkaus (2006a) hypothesized that a long, superoinferiorly narrow glenoid may be the primitive condition for hominins. Neandertals share relatively narrow glenoid fossae with the Australopithecus africanus specimen Sts 7 (Vrba, 1979; Carretero et al., 1997) and early Homo fossils KNM-WT 15000 and Dmanisi 4166 (Trinkaus, 2006a,b). By this view, the relatively broad glenoid fossa of a late Mousterian Neandertal from Vindija Cave (Croatia), Vi-209, might be seen as evidence for some degree of genetic continuity between late Neandertals and early modern Europeans (Smith and Trinkaus, 1991) or introgression of modern human genes into the late Neandertal gene pool (Di Vincenzo et al., 2012). Evolutionary development has been argued to influence the scapular glenoid fossa. A recent twodimensional geometric morphometric analysis of Neandertal glenoid fossa shape (Di Vincenzo et al., 2012) suggests that most of the Neandertal/modern human morphological contrasts in glenoid fossa shape are attributable to differences in evolutionary-developmental growth rates. Di Vincenzo et al. (2012) hypothesize that heterochrony in overall somatic growth accounts for observed between-group differences in glenoid width. Relative amounts of growth in the inferior glenoid rim secondary center of ossification may play a substantial role in glenoid width and curvature, such that relatively broad modern human fossae may be a morphological consequence of a slowed growth trajectory (Di Vincenzo et al., 2012; see also Churchill and Trinkaus, 1990). Although this hypothesis remains untested, as the influence of evolutionary shifts in somatic growth rates on scapular ontogeny have yet to be fully explored, future studies may assess this possibility. Regardless of the causes of variation in glenoid fossa morphology (both within and between groups), the
169
functional significance of these differences remains an open question. Contrasts between Neandertal and modern human upper limb morphology have traditionally been interpreted in terms of differences in the habitual use of various forms of subsistence technology in the evolutionary history of the two groups. Although early modern humans of the later African Middle Stone Age and the Eurasian Upper Paleolithic are thought to have been more heavily invested in the use of throwing-based, long-range projectile weaponry, Neandertals may have primarily relied upon close range hunting technology, primarily large and heavy thrusting spears (see Churchill and Rhodes, 2009). These very different technologies are associated with distinct kinematics and patterns of muscle activation (Schmitt et al., 2003). Because habitual patterns of muscle engagement and biomechanical loads may influence skeletal morphology, upper limb bones may reflect hunting behaviors. Anatomical differences in the shoulders of modern humans and Neandertals have thus been interpreted as reflecting little to no reliance on spear throwing in the Neandertal predatory repertoire (Churchill and Trinkaus, 1990; Churchill and Rhodes, 2009; Rhodes and Churchill, 2009). Some variation in glenoid fossa shape, both between Neandertals and modern humans and between modern human groups, may be a function of developmental plasticity in joint morphology in the context of habitual differences in upper limb use (Di Vincenzo et al., 2012). However, scapular architecture is relatively poorly understood from a functional perspective (Alemseged et al., 2006). The relative breadth of the scapular glenoid fossa alone has been subjected to multiple functional interpretations. The relatively narrow glenoid fossa of the Muierii 1 early modern human scapula from Romania has been hypothesized to indicate parity in the shoulder loading regimes between Neandertals and this individual, that is, a lack of throwing-based projectile weaponry (Trinkaus, 2008; Trinkaus, 2011). Although Di Vincenzo et al. (2012) did not include this specimen in their study, their conclusion that relative width reflects developmental rate would suggest that the relative breadth of the glenoid reflects parity in developmental rates with Neandertals, rather than a functional signal. Thus, debate persists in the interpretation of glenoid fossa morphology. What little is definitively known about the functional consequences of glenoid fossa shape is based in human clinical observations. These studies have focused on the association of glenohumeral stability with loss of function and pain. Independent of the soft tissues (including the glenoid articular cartilage and glenohumeral labrum), the bony anatomy of the glenohumeral joint is important for stability (De Wilde et al., 2004; Lugo et al., 2008). Two salient morphological features have been established as a contributing factor to anterior shoulder stability. (1) Concavity of the glenoid is critical to humeroglenoid stability, as individuals with less curved, shallower glenoid fossae are more likely to experience anterior dislocation (Lippitt et al., 1993; Itoi et al., 2000). (2) Anterior rim glenoid morphology can also affect propensity to shoulder dislocation; individuals with lesions or less laterally projecting anterior rim morphology are also more prone to pain, dislocation, and loss of function (Bigliani et al., 1998; Stevens et al., 1999; Burkhart et al., 2002; Sugaya et al., 2003;
170
MACIAS AND CHURCHILL
TABLE 1. Fossil taxa used in this study Taxon
Specimen no.
Au. afarensis (N 5 1) AL 288 Au. sediba (N 5 1) MH 2 H. neanderthalensis (N 5 11) Krapina 127 Krapina 129 Krapina 130 Krapina 131 Krapina 132 Krapina 133 Neandertal 1 La Ferrassie 1 La Ferrassie 2 Tabun C1 H. sapiens (N 5 3) Oberkassel 1 Oberkassel 2
Locality
Age
Hadar (Ethiopia) Malapa (South Africa) Krapina (Croatia) Krapina (Croatia) Krapina (Croatia) Krapina (Croatia) Krapina (Croatia) Krapina (Croatia) Feldhofer Grotto (Germany) La Ferrassie (France) La Ferrassie (France) Tabun (Israel) Bonn-Oberkassel (Germany) Bonn-Oberkassel (Germany)
Early Pleistocene Pliocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene Late Pleistocene
Huysmans et al., 2006). Appreciating morphological variation in these two features may be necessary to better understand which aspects of morphological differences might contain information about functional adaptation or habitual behavior patterns. Geometric morphometrics is inherently exploratory work, and this technique may be applied to understand ways in which morphology varies within and between groups as a basis for generating testable functional hypotheses. However, adaptive hypotheses should not be advanced until it can be demonstrated that other explanations (e.g., static allometry) can be ruled out. Following Gould and Lewontin (1979), we investigate a hierarchy of alternatives to adaptation for the explanation of glenoid fossa shape. We argue that a reductive approach of removing confounding factors is a necessary prerequisite to understanding the functional significance of observed morphological variance in scapular morphology. We evaluate the effects of activity levels, size, and phylogenetic relationships to identify aspects of scapular glenoid fossa morphological variation that might be independent of these factors. Traits that (1) vary bilaterally within individuals, (2) vary by sex (independent of allometry), or (3) vary between human groups engaged in differing activity levels, are likely to reflect developmentally plastic aspects of glenoid fossa morphology. These traits are useful for understanding within and between group variations in habitual upper limb usage. Traits that are allometrically related to size, as well as those that may be primarily varying as a function of evolutionary differences in somatic growth rates, are considered to be less likely to inform about adaptive behavioral differences between groups. (Although evolutionary shifts in ontogeny may affect scapular morphology, this study does not assess this influence. This does not rule out the possibility that selection could operate on organismal size or growth rates if the consequence was adaptively beneficial morphology of some joint system, but we do view this possibility as unlikely). Traits that vary in clear patterns that align with phylogeny— e.g., derived traits found throughout clades but not in outgroups—are understood to be better signals of ancestry than signals of functional adaptations. Finally, features that are unrelated to differences in activity levels and patterns, size, or phylogeny are possible candidates for features that have been under the direct action of
Original or cast Original Original Cast Cast Cast Cast Cast Cast Cast Cast Cast Cast Cast Cast
Side Right Right Right Right Left Left Right Right Right Left, right Right Right Left Left, right
selection for biomechanical competence of the joint given species-level functional demands on the shoulder. Through this approach, we actively attempt to rule out other possibilities before we advance functional hypotheses to explain the observed shape variation.
MATERIALS AND METHODS Two-dimensional studies are limited in their ability to capture the complex three-dimensional relationships of a ball-and-socket joint. Clinical studies highlight the importance of three-dimensional analyses. Three-dimensionally reconstructed images were found to be more reliable than conventional two-dimensional radiographs for detecting skeletal shape anomalies of the glenoid (Stevens et al., 1999; Sugaya et al., 2003). Furthermore, past studies have failed to capture important shape information in the medio-lateral variation of glenoid fossa morphology, particularly with reference to the anterior rim. For these reasons, a three-dimensional approach is taken here. Existing studies have not addressed Neandertal scapular glenoid morphology in a greater comparative context. To understand the evolutionary patterns of hominoid morphology, as well as to inform our understanding of the effects of different functional demands on the glenoid, landmarks were digitized on specimens of modern humans, chimpanzees, orangutans, original fossil specimens, and fossil casts (N 5 187, see Tables 1 and 2). To establish more equal sample sizes across taxa (avoiding data skew), the chimpanzee and modern human datasets were randomly down-sampled to 25 individuals (N 5 86). Modern human scapulae were classified as belonging to Upper Paleolithic modern humans (Oberkassel 1 and 2) or recent modern samples, in part to take a conservative approach to intraspecific variation and in part to examine chronological variation within Homo sapiens. No scapulae with complete glenoid fossae were available to represent early Homo. The use of casts in this analysis, although not ideal, is acceptable. For those specimens where the original could be digitized, comparisons of these trials to digitized casts of the same specimen revealed no systematic error in landmark orientations. We also compared the casts to specimen photographs to assess regions that may be affected by pathology or taphonomic damage.
171
SCAPULAR GLENOID FOSSA MORPHOLOGY
TABLE 2. Comparative taxa used in this study Species H. sapiens H. sapiens H. sapiens H. sapiens H. sapiens Total P. troglodytes P. troglodytes P. troglodytes P. troglodytes Total P. abelii P. pygmaeus Total
Origin
Lifestyle
Source
Male
Female
Unknown
Total
Canyon del Muerto (AZ, USA) El Hesa (Egypt) New York (NY, USA) Point Hope (AK, USA) Indian Knoll (KY, USA)
Active
AMNH
0
0
17
17
Less active Less active Active Active
AMNH AMNH AMNH AMNH
Cameroon Gabon Uganda West Africa
Active Active Active Active
FMNH HMCZ NMNH PCM
Sumatra Borneo
Active Active
NMNH FMNH, NMNH, ZAIM
0 24 16 6 46 10 4 0 2 16 2 8 10
0 7 10 10 27 7 1 1 0 9 1 9 10
17 0 2 14 50 2 0 0 1 3 0 0 0
17 31 28 30 123 19 5 1 3 28 3 17 20
Note: geographic origin was stated in lieu of subspecies for Pan troglodytes, due to the contentious nature of subspecies definitions and designations. AMNH, American Museum of Natural History; FMNH, Field Museum of Natural History; HMCZ, Harvard Museum of Comparative Zoology; NMNH, National Museum of Natural History; PCM, Powell-Cotton Museum; ZAIM, Zurich Anthropological Institute Museum.
Recent modern human populations were selected to represent a variety of latitudes, geographic locations, and time periods to subsample human variation. To address the role plasticity plays in glenoid shape, modern humans were categorized into “active” or “lessactive” lifestyle groups. Active groups included archaeological specimens known to have engaged in hunting, foraging, or farming (nonindustrialized) over their lifetime. Less-active groups included archaeological and modern specimens representing individuals unlikely to have been engaged in hard labor or otherwise highly active. Active populations include a sample from Point Hope (Alaska), Canyon del Muerto (Arizona), and Indian Knoll (Kentucky). Inactive populations include a sample from El Hesa (Egypt) as well as a modern anatomical collection from twentieth century New York. All modern human comparative samples are curated at the American Museum of Natural History (ANMH, New York), except for the Indian Knoll sample, curated at the National Museum of Natural History (NMNH, Washington DC). All extant specimens were adults with no signs of pathology or taphonomic damage. The New York sample derives from autopsy specimens of known sex, age, and race. Sex in the Point Hope and Indian Knoll samples had either been determined by previous workers and was indicated in the museum records, or had been determined by Sharplin (unpublished study, 2011). In every case, one of us (M.M.) assessed the sex of the specimen prior to referring to the previous designation. Only specimens in which our diagnosis matched that of the prior designation were used. The sex of the specimens from Canyon del Muerto and El Hesa are unknown. All data were collected by M.M. Sample sizes and other details are provided in Table 1. The Point Hope sample is based on a cemetery sample of individuals who lived a sedentary lifestyle in an Ipiutak village, dated from 100 B.C.E. to 1400 C.E. The cemetery was excavated during 1939–1941 (Rainey, 1972; Mason, 1998). The Ipiutak largely relied on caribou, walrus, and seal hunting, with only a seasonal reliance on sea mammals (Krueger, 2006). The bow and arrow was
the principal weapon for the Ipiutak people, and although harpoon elements were found during the excavation, they were rare compared with the quantity of artifacts associated with archery (Larsen and Rainey, 1948). Previous studies of the Point Hope population suggest that these individuals were highly active, consistent with other samples engaged in strenuous upper body activity (Shackelford, 2006). Sex was confirmed with pelvic measurements when possible. The Canyon del Muerto sample derives from the early Basketmaker II stage in Southwestern Archaeology (1500–50 B.C.E.), and represents a population engaged in an agricultural economy but that still used some hunting with throwing-based projectile technology (atlatl) (Kiddler, 1927). The extent of agricultural reliance in the Canyon del Muerto population is uncertain: coprolite analysis from a contemporaneous Basketmaker II site indicates that this population relied on maize for the majority of their calories (Matson, 1991), whereas Reed (2000) questions the extent of this dependence on agriculture, but agrees that there was decreasing dependence on hunting and gathering over time. Indian Knoll is an Archaic Period site (dated to 6415– 4143 B.C.E.), and one of the largest hunter-gatherer skeletal collections in North America. Archaeological excavations of the burials also recovered shell beads, stone tools, and animal remains. Their subsistence economy relied mainly on deer, turkey, mussels, nuts and a variety of locally collected plant materials (Winters, 1974). The Egyptian individuals resided in El Hesa, located near modern day Aswan, Egypt, and the burials are dated to the Roman period between C.E. 200–400 (Elliot Smith and Wood-Jones, 1910, cited in Irish, 1993). Mortuary evidence indicates a middle class affiliation for the individuals in this sample, such that even though they derive from a pre-industrial agricultural economy (Irish, 1993), it is unlikely that the individuals sampled were engaged in agricultural or other hard labor. The New York Medical Collection consists of twentieth century New Yorkers. These individuals have known
172
MACIAS AND CHURCHILL
Fig. 2. Landmark placement on scapular glenoid fossa. Lateral view. Anchoring landmarks in blue, resampled landmarks in red. (A) Glenoid outline. 40 landmarks. (B) APCP, anterior posterior curve profile. 10 landmarks. (C) SICP, superior inferior curve profile. 15 landmarks.
age-at-death, sex, and race. The sample includes African–American and Caucasian individuals. Previous research found that race and age did not correlate with morphological variables (Macias, 2009). Comparative ape samples were collected from the AMNH, NMNH, Field Museum of Natural History (FMNH, Chicago), Harvard Museum of Comparative Zoology (HMCZ, Cambridge), Powell-Cotton Musem (PCM, Birchington), and the Zurich Anthropological Institute Museum (Zurich, ZAIM). All ape specimens were wild-shot adults with no signs of pathology or taphonomic damage (see Table 2 for details). A total of 65 semilandmarks describing the scapular glenoid fossa were analyzed (Fig. 2). Anchoring landmarks were chosen to be the most clearly defined and repeatably identifiable points on the scapular glenoid fossa, patterned after Type II landmarks described by Young (2008). Lines—defined as a series of closelyspaced landmarks between two points—were collected around the articular rim of the scapular glenoid fossa, as well as along lines representing the superior–inferior curve profile (SICP), and the anterior–posterior curve profile (APCP). Data were obtained using a Microscribe (Immersion Corp., San Jose, CA) digitizer 3DX. Landmarks were identified once the specimen was mounted and stabilized. Lines were resampled to a standard number of equally-spaced landmarks in the series (40 for the articular rim, 15 for the SICP, and 10 for the APCP) using the programs Resample (Rauum et al., 2006) and Geomorph (Adams and Otarola-Castillo, 2013). To circumvent the effect of unequally distributed points, each line was analyzed individually. Lines were subjected to generalized procrustes analysis (GPA) to remove differences in size, shape, and translation. Principal components analysis (PCA) was used to visualize variation in shape space, and explore correlations among shape variables. Shape differences between groups were tested by univariate analysis of variance using the Kruskal–Wallis test as well as permutation tests on Procrustes distances between groups. PCA scores were regressed on centroid size to explore correlations between shape and body mass. Corruccini and Ciochon (1976) found an isometric relationship between glenoid cavity height and body mass across primates; as such, glenoid size has been used as a proxy for overall size in previous studies (Alemseged et al., 2006; Green and Alemseged, 2012). Procrustes distances were used in several ways to statistically evaluate variation within the sample. We used Procrustes distances to: (1) compare antimeres–left
and right sides from the same individual (intra-individual variation), (2) compare lefts and rights across the modern human sample (inter-individual variation), and (3) explore between-group differences with permutation tests. Permutation tests, where null distributions are generated by randomly allocating individuals to groups, were carried out to assess the significance of Procrustes distances between groups. For each comparison, the observed distance between groups (i.e., the test statistic) was compared to the distribution of distances between randomly allocated groups. The observed distance was considered statistically significant when it was greater than 95% of the permuted distances. In each analysis, 10,000 iterations were calculated for each pair of taxa, giving a maximum possible P-value of 0.0001. To identify developmentally plastic aspects of scapular glenoid shape, two potential sources of variation were addressed with the modern human sample: (1) differences between left and right glenoids in the same individual, and (2) differences in shape due to activity level. To illustrate the variation observed within and between modern human left and right glenoid fossae, frequency distributions of Procrustes distances were calculated. The distribution of intra-individual distances (distances between antimeres) was compared to the interindividual distribution (pairwise distances between individuals across the modern human sample) with a Wilcoxon rank test. Permutation tests were used to test the significance of these differences in three ways: (1) to assess shape differences between lefts and rights across the modern human sample, (2) to evaluate differences between active and inactive populations, and (3) to evaluate differences between males and females. The glenoid articular outline, SICP, and APCP were analyzed separately to explore each aspect of shape independently. For each dataset, a PCA was performed, and all principal components (PCs) accounting for more than 10% of the overall variation were studied. These PC scores were then regressed on centroid size; PC scores with significant correlations were then eliminated from the analysis. The remaining PC scores were then subjected to Kruskal–Wallis tests exploring differences among groups; PCs lacking significant differences among groups were eliminated from the analysis. All statistical tests were subjected to sequential Bonferroni correction (Rice, 1989). Remaining PCs (i.e., those that were not significantly related to size and that successfully differentiated between groups) were studied with the program Morphologika (O’Higgins and Jones, 1998) to explore and visualize shape variation. The goal of these analyses was to identify aspects of scapular glenoid morphology that do not appear to be explicable in terms of size allometry, sex, or developmental plasticity related to activity levels and patterns. Once these features are isolated, we evaluate the possible functional significance of variation in these aspects. This reductive approach involved eliminating shape variables associated with size, sex, or developmental plasticity individually before exploring the functional consequences of the remaining shape variation. PC scores not rejected through this process were aggregated for a combined Canonical Variate Analysis (CVA) and Discriminant Function Analysis (DFA) to explore how well these remaining factors discriminate groups.
SCAPULAR GLENOID FOSSA MORPHOLOGY
RESULTS Variation in Modern Humans To evaluate the developmental plasticity of the glenoid fossa morphology, we assessed levels of bilateral asymmetry in the joint surface. A one-sided Wilcoxon test was performed upon individuals for whom both left and right sides were measured. The Procrustes distance between the left and right side was compared to a random subsample (10,000 of 13,688) of across-sample differences. Results were significant (P 5 0.0018), indicating that antimeres are more similar to each other than to other specimens. Permutation tests found no significant differences between left specimens and right specimens (P 5 1.00). These results suggest bilateral asymmetry is not an important component of the overall shape variation within the modern human populations sampled
173
here. Figure 3 illustrates the contrast between intraindividual and inter-individual Procrustes shape distance. Although these distributions overlap, the interindividual distribution has higher frequencies of greater distances between pairs. Developmental plasticity was also evaluated by examining differences between sexes and activity levels in our modern human groups. No significant differences were observed between males and females (P 5 0.71), or between individuals categorized as active or inactive (P 5 0.13). Because these factors did not discriminate individuals in Procrustes shape space (or in the PCA, Fig. 4a,b) individuals were thereafter combined regardless of sex or lifestyle to form a single sample of modern humans. Sides were randomly selected to create a sample of both left and right specimens.
Principal Component Analysis To examine how fossils varied in morphospace relative to extant taxa, all specimens were analyzed together via GPA and PCA. All PCs representing at least 10% of the overall variation were examined with respect to allometry and phylogenetic relationships before being subjected to further analyses.
Glenoid Fossa Outline Analysis
Fig. 3. Intra-individual and inter-individual procrustes distances. Frequency distribution of recent modern human Procrustes distances. Intra-individual, or between antimere, Procrustes distance (blue) is contrasted with inter-individual Procrustes distance (pink). Although these graphs overlap, the higher frequency of larger distances (both at the peak and particularly at the right tail) indicates that intra-individual Procrustes distances are smaller (matched pair Wilcoxon Pvalue 5 0.0004)
Results of the glenoid outline analysis were comparable to those of Di Vincenzo et al. (2012); this is unsurprising given that the analytical methods are similar. Principal Components 1–3 were not significantly correlated with centroid size (see Table 3). PC 1 explained the largest percentage of variance (25.4%), capturing supero-inferior elongation and mediolateral breadth of the glenoid fossa (Fig. 5). This measure differentiated Neandertals from chimpanzees and modern humans. Observed variation across taxa did not appear to follow a phylogenetic pattern (e.g., Pongo were significantly different from Pan and recent humans, but not Neandertals, see Table 4), but was consistent with the evolutionary trajectory identified by Di Vincenzo
Fig. 4. PCA of recent modern humans displaying variation by side (A), sex (B), and activity level (A and B). No significant differences were found between rights (R) and lefts (L), between females (F) and males (M), or between active (red) and less-active (blue) populations.
174
MACIAS AND CHURCHILL
et al. (2012). Although we did not directly test this hypothesis, we cannot rule out the possibility that the variation described on PC1 reflects evolutionary developmental differences between groups in somatic growth rates (Di Vincenzo et al., 2012). Thus, PC 1 was TABLE 3. Results of regression analyses of PC scores on natural log centroid size Principal component
r2 value
P-value
a-level
Outline PC 1 Outline PC 2 Outline PC 3 APCP PC 1 APCP PC 2 APCP PC 3 SICP PC 1 SICP PC 1* SICP PC 2
0.036 0.002 0.053
