Postcranial Remains and the Origin of Modern Humans

Transcription

1 Evolutionary Anthropology 229 ARTICLES Postcranial Remains and the Origin of Modern Humans OSBJORN M. PEARSON The nature, timing, and location of the origin of modern humans has been the subject of intense controversy for the last 15 years. 1 4 Genetic data and new radiometric dates for key fossils that lie beyond the range of radiocarbon dating have substantially added to the knowledge derived from the fossil evidence documenting the transition from archaic to modern humans. These new data, however, have failed to resolve the problem in its entirety. Most authorities now accept that Africa played an important, and probably central, role in the origin of modern humans The genetic evidence seems to be particularly emphatic that an African population that existed between 200,000 and 100,000 years ago (100 ka) is ancestral to all living humans. 6,7 Controversy still surrounds the question of how much, if at all, archaic humans from outside of Africa, such as Neandertals, late archaic Chinese hominins such as Jinniushan, and the Indonesian Ngandong hominins, may have contributed to the morphological and genetic diversity present in living populations and the morphology of the earliest fossils of modern humans. 10 The purpose of this review is to explore what hominin postcranial remains reveal about the origin of modern humans. Most studies of the fossil evidence for modern human origins have concentrated on changes in cranial morphology. 7,14 21 In comparison, postcranial fossils have received less attention, although recent work has started to reverse that imbalance Certainly, a large portion of the work on the postcrania of Neandertals, other late archaic humans, and the earliest modern humans has focused on functional interpretations Osbjorn Pearson is Assistant Professor of Anthropology at the University of New Mexico. His primary interests include the origin of modern humans and the effects of activity on the strength and morphology of bones. He has written several articles on the postcranial remains relevant to the origin of modern humans including Activity, climate, and postcranial robusticity: Implications for modern human origins and scenarios of adaptive change, in press in Current Anthropology. ompear@unm.edu and stressed the value of differing morphology as an indication of differences in adaptations rather than emphasizing the fact that such differences probably also serve as useful phylogenetic markers. However, postcranial and cranial dimensions appear to be similarly heritable 25 ; both should preserve information about both function and phylogeny. More to the point, as I will emphasize, cranial and postcranial data reveal very similar patterns with regard to the origin of modern humans. MODELS OF MODERN HUMAN ORIGINS Much of the recent literature on modern human origins has emphasized the difference between the Multiregional model and the Out-of-Africa model, the two primary models for the origin of modern humans. Within the last few years, however, these two models have converged considerably, due in large part to Relethford and Harpending s 26 demonstration that the patterns of genetic and cranial distances among modern populations could be obtained from the Out-of-Africa or the Multiregional model, given the right combination of migration rates and effective population size within each region. Acceptance of the genetic data by the proponents of the Multiregional model means that both the Out-of-Africa and the Multiregional model now accept that, in essence, modern humans of African origin eventually replaced virtually all of the genes of archaic humans throughout Eurasia. Nevertheless, the two models still differ with regard to how this de facto replacement happened. Chris Stringer, one of the chief advocates of the Outof-Africa Model, considers that either no interbreeding transpired between modern humans and archaic Eurasian hominins or that the interbreeding was so limited that it left no genetic or morphological traces in modern populations. 27 Other proponents of the Out-of-Africa model consider that interbreeding did occur and may have left morphological traces, but that the admixture was still small in extent. 28,29 In contrast, the Multiregional model specifies that interbreeding did occur between early modern humans and archaic hominins at least in Central Europe, China, and Indonesia, but that subsequent gene flow from African populations, genetic drift, and selection have erased most, and possibly all, of the genetic signature and many of the morphological traces of this ancient admixture. 10 Multiregionalists consider that many of the early examples of modern humans within each region of Eurasia, particularly in Europe and Australia, bear more morphological traces of admixture with archaic

2 230 Evolutionary Anthropology ARTICLES hominins than do the living inhabitants of those regions. 10,13 According to this view, modern geneticists may be working approximately 15,000 to 30,000 years too late to catch a clear genetic signal of admixture, but morphologists studying human fossils still have these data available. Given the current agreement over the genetic data from recent humans and its implications, the much-touted and often overly polarized differences 29,30 between the Out-of-Africa and Multiregional models reduce to the issue of whether the presence of admixture can be proved or disproved in the early modern fossils within each region. Within this context, it is useful to note that the two models can be used to generate a series of predictions about the morphology of late archaic and early modern hominins. A crucial problem with the differing predictions lies in the fact that as the proposed amount of admixture within a region becomes small, the predictions of the Multiregional model converge upon those of the Out-of-Africa Model, so that a test of the difference between the two models becomes increasingly difficult. 31,33 PREDICTIONS FROM THE MODELS Three related hypotheses stem from the Out-of-Africa Model. First, the earliest modern humans should appear in Africa and the next earliest in areas close to Africa. 34 A test of this hypothesis requires an acceptable definition of what constitutes modern morphology. Such a definition has proven controversial for crania. 35,36 and would likely also prove controversial for postcranial bones (see Box 1). The second hypothesis is that, except in Africa, there should be no close morphological resemblance between the earliest modern humans and their archaic predecessors, whereas early modern humans from various regions should resemble each other. Together, these points should ensure that modern humans, whether recent or fossil, display a fairly close phenetic resemblance, at least when measured against their similarity to non-african archaic hominins. According to the third hypothesis, Neandertals, in aspects of their morphology that may reasonably be assumed to reflect adaptations to cold, 16,18 20 should be similar to both European and non- European cold-adapted populations. According to the Out-of-Africa model, modern humans arose in Africa, a warm area, and had morphological adaptations that resembled those of heat-adapted populations of today. Adaptation to cold altered this ancestral morphology in some descendent populations of the migrants from Africa, but such adaptations are recent phenomena and could be expected to be of similar magnitude in widely separated populations such as the Sami in Europe and the Inuit in Alaska. The Multiregional model can be used to make a single prediction, namely that within any region there... as the proposed amount of admixture within a region becomes small, the predictions of the Multiregional model converge upon those of the Out-of-Africa Model, so that a test of the difference between the two models becomes increasingly difficult. should be a regional morphologic signature that should be manifested as a time-ordered series of close phenetic relationships between the inhabitants of the region from the Pleistocene to the present. In the case of Europe, which has the densest hominin fossil record for this period, this prediction means that European hominin fossils should resemble recent Europeans more than any other recent non-european populations. With regard to time-ordered phenetic relationships, the Multiregional model predicts that Neandertals should resemble Early Upper Paleolithic people more closely than they do living Europeans and that living Europeans should resemble Early Upper Paleolithic humans more closely than they do Neandertals. REVIEW OF CRANIAL REMAINS Discussion of the postcranial evidence of the evolution of modern humans is best considered in the context of the phylogenetic relationships derived from cranial remains. According to all the models, the starting point for the lineage leading to modern humans is early African Homo erectus (or Homo ergaster 37 ), a taxon represented by numerous cranial and dental remains from sites in East Africa, including Koobi Fora, Olduvai Gorge, Konso, and perhaps the cranium from the Danakil Depression in Eritrea, as well as by the crania and mandible from Dmanisi, Georgia. 40 The relationship between H. ergaster and H. erectus is controversial; Either they are the same species, H. erectus, 60 or H. ergaster is a distinctive, early (1.9 to 1.1 Ma) hominin that is ancestral to the more derived H. erectus and probably also to all later Homo. 37 H. erectus spread widely through southern Eurasia into China and Indonesia sometime between 1.8 and 1.0 Ma. The precise timing of the initial hominin colonization of Asia is still debated By 600 ka BP in Africa, H. erectus evolved into a new species known as Homo heidelbergensis. 47 In common with virtually every other aspect of Middle Pleistocene hominin phylogeny, the use of the name H. heidelbergensis is disputed. Some scientists propose that African specimens that resemble Middle Pleistocene European hominins should be called Homo rhodesiensis and that H. heidelbergensis is an exclusively European taxon, 48 while some proponents of Multiregional evolution consider that all Middle Pleistocene hominins and even H. ergaster are the same species, H. sapiens. 49 For the purpose of this review, H. sapiens is restricted to modern humans in accordance with Howell s 9 cogent arguments; H. heidelbergensis and H. rhodesiensis are considered synonymous. H. heidelbergensis differed from H. erectus by possessing a larger brain, a larger browridge, and a more massive face, largely due to expanded maxillary and frontal sinuses. H. heidelbergensis appears to have subsequently spread to Europe, where

3 ARTICLES Evolutionary Anthropology 231 Box 1. What Defines an Anatomically Modern Postcranial Skeleton? Any attempt to address the problems of when and how modern morphology arose must deal with the problem of what defines anatomically modern form. For the cranium, Day and Stringer 11,17 have proposed a definition of what constitutes modern morphology; thus far, the postcranial skeleton has not been similarly treated. In one definition of what constitutes anatomically modern morphology, Stringer and colleagues 11 reiterated the criteria for cranial form proposed earlier 163 and added a few details for the postcranial skeleton. One of these was a generally gracile morphology, which they defined as thin walled long bones with relatively small articular surfaces in which muscular insertion sites were not as strongly marked as in archaic hominins, particularly Neandertals. In addition, they stipulated that modern postcranial skeletons should have a scapula with an axillary border bearing a single ventral sulcus or bisulcate pattern and a short, stout pubic ramus. In contrast, a large number of distinctive postcranial features may be used to define Neandertal morphology. Perhaps therein lies a crucial problem. Definitions of what is anatomically modern are often made in contrast to Neandertals, the assumption being that Neandertals serve as a good approximation of generalized archaic human morphology. This viewpoint now seems to be partially in error. In common with the distinctive cranial traits, many aspects of Neandertal postcranial morphology appear to be highly derived rather than primitive. Characteristically, Neandertal postcranial morphologies are present only in mosaic fashion and lower frequencies in the 300-ka-old fossils of Homo heidelbergensis from the Sima de los Huesos in Atapuerca. 106,107 When variation in the traits that came to characterize Neandertals exists, different Atapuerca fossils display morphologies that are either Neandertal-like or reminiscent of recent humans. 106,107 The stocky skeleton of Neandertals, with its stout long bone diaphyses, massive joints, and most of its other peculiar features, 23,113,114,116,120,149,164 thus appears to have developed from an ancestral condition that, on the whole, more closely approximated modern human morphology. 120 A few aspects of morphology distinguish the skeletons of the early modern humans from the sites of Skhul and Qafzeh in Israel from living Homo sapiens. These features include a pubic ramus that is relatively long, at least in comparison to those of recent populations from warm areas. 110,111,165 An elongated pubic ramus is also present in the 200- to 300-ka-year-old Jinniushan specimen, 129,131 as well as in Pelvis 1 from Sima de los Huesos, Atapuerca. 65 Elongation of the pubic ramus therefore appears to be an ancestral (plesiomorphic) trait that is retained in some of the Skhul-Qafzeh specimens but was subsequently lost in the modern human lineage. 165 In sum, it is easy to construct definitions of Neandertal postcranial skeletons based on their numerous apomorphies, but the anatomically modern postcranial skeleton is best described in very general terms. Modern morphology has to be able to encompass the entire range of climatic adaptations (physiques) in living humans, although all of the earliest modern humans may have had a linear, heat-adapted body form. Modern humans may have, on average, a reduced body mass relative to Middle Pleistocene hominids and Neandertals, 64 although this is disputed. 100 Modern human pelvic shape may constitute a genuine novelty of our species, but our expanded pelvic inlet (relative to that of Homo ergaster) builds upon pelvic and neonatal brain expansion that commenced in the Middle Pleistocene. 100,155 In addition, some of the earliest modern humans, such as Skhul IX, retain plesiomorphic details of pelvic anatomy such as a relatively long pubic ramus. These traits appear to have disappeared subsequently from the modern human line. The same holds for a variety of other features such as the low coronoid process of the ulna relative to the olecranon height and the dorso-volarly flat proximal articular surface of the first metacarpal of some of the Skhul-Qafzeh specimens. It seems clear that anatomically modern form as represented by living humans was not entirely complete in the Skhul-Qafzeh hominids; many small features had yet to change. These rather subtle aspects of morphology should not prevent the Skhul-Qafzeh and African Middle Stone Age hominins from being identified as anatomically modern or, perhaps more accurately, as nearmodern. 57 However, it is equally important to recognize that inclusion of these Middle Paleolithic hominins among modern humans (Homo sapiens) expands the range of morphology included within our species and incorporates a variety of plesiomorphic traits generally not present in living humans. specimens such as Mauer, Arago, Petralona, and the Sima de los Huesos hominins represent the taxon. More controversially, H. heidelbergensis may have spread East to China. The Dali and Jinniushan crania may well be Asian representatives of the species. After its spread, between roughly 300 and 128 ka BP, H. heidelbergensis gave rise to Neandertals in Europe, 48,50 modern humans in Africa 28,51,52 and, perhaps, hominins such as Mapa and Xujiayao in China. In Indonesia, Homo erectus persisted through the Middle Pleistocene and into the Late Pleistocene at Ngandong, 53 perhaps until as recently as 50 ka BP, 54 although the exact age of the Ngandong specimens remains controversial. 55,56 The cranial features that came to characterize modern humans have a long history in Africa. Fully modern humans appeared first in Africa around 100 ka BP at sites such as Border Cave, Klasies River Mouth, and Omo Kibish. 28 Early modern humans (or near-modern humans 57 ) appeared at approximately the same time at the Israeli sites of Qafzeh and

4 232 Evolutionary Anthropology ARTICLES Skhul or even earlier if the dates or anatomically modern status for the African early modern humans are discounted, as some have proposed. 8 Modern humans persisted in Africa no archaic humans have been found in Africa postdating 100 ka BP but disappeared from the Levant and were replaced by Neandertals in the interval between roughly 60 and 40 ka BP. 58,59 Modern humans reappeared with the advent of the Upper Paleolithic in the Levant. Elsewhere in Eurasia and Australia, cranial fossils of modern humans (Homo sapiens) first appeared between 40 and 30 ka BP, 14,57 although the arrival or development of modern humans in many regions may have predated the first fossils that prove their presence. THE POSTCRANIAL DATA FOR MODERN HUMAN ORIGINS Before discussing several aspects of skeletal morphology that figure prominently in discussions of modern human origins, it will be useful to review the world-wide record of hominin postcranial fossils from Homo ergaster to the appearance of modern humans and disappearance of other forms of Homo (Fig. 1). The fossil record for the evolution of the modern human postcranial skeleton consists of an abundance of fossils from Europe, a variety of fragments from Africa that are widely separated in time and space, and very little evidence from Asia. Africa and the Near East The postcranial evidence for the evolution of anatomically modern humans is considerably more sparse than the cranial data. However, early examples of H. ergaster (or H. erectus) are well known from the postcranial remains of KNM-WT and numerous more fragmentary postcranial remains from Koobi Fora and Olduvai Gorge. The postcranial skeleton of H. ergaster differs dramatically from that of Australopithecines, but exhibits relatively few differences from modern humans. The differences include an extraordinarily long femoral neck; a narrow pelvis with, in adults, a sharply marked iliac pillar; an ischial tuberosity displaying medial torsion 61 ; buttressing of the femoral shaft so that the medio-lateral diameter is larger than the antero-posterior diameter at midshaft; and very thick cortical bone within the long bone shafts (Fig. 2). Human evolution in Africa during the period between one million years ago and 20 ka BP is documented by a paucity of postcranial remains, which consist almost entirely of isolated and fragmentary specimens. This is unfortunate, for both H. heidelbergensis and H. sapiens may have originated in Africa during this time. There are, however, some indications that H. heidelbergensis had a greater body mass than did H. erectus. 64,65 In Africa, the few human postcranial fossils postdating H. erectus exhibit a striking The succeeding periods of human evolution within Africa are documented by a paucity of postcranial remains, which consist almost entirely of isolated and fragmentary specimens. This is unfortunate, for both H. heidelbergensis and H. sapiens may have originated in Africa during this time. amount of morphological diversity and include specimens that do not differ substantially from modern humans as well as bones that are distinctly not modern. The most important hominin postcranial fossils from the period between roughly 800 ka and 300 ka BP in Africa include a massive proximal femur from Berg Aukas 66 (most probably attributable to H. heidelbergensis); a small distal humerus from Bodo; 181 a series of postcranial remains from the site of Kabwe (Broken Hill) in Zambia including a tibia, several femoral fragments, a sacrum, and two iliac fragments 63,67 69 (of which only the tibia was associated with the cranium); and an assortment of 500,000 year-old 70 remains from the Kapthurin Formation near Lake Baringo that comprises an almost complete ulna, 71 a first metatarsal, and a few manual phalanges (Figure 1). In comparison to specimens of H. ergaster/erectus, these fossils display a variety of primitive and derived features. The Kabwe and Berg Aukas femora lack the extraordinarily elongated femoral neck of H. ergaster, but all retain thick cortical bone, impressively so in the case of the Berg Aukas femur. 86 Berg Aukas has a massive femoral head and a very low neckshaft angle which distinguish it from Kabwe specimens. The Baringo ulna is elongated, slender, weakly imprinted with muscle marks except for the supinator crest, and has a low coronoid process relative to the height of its olecranon process. 72,73 In all of these features, the Baringo specimen resembles the KNM-WT H. ergaster ulna. 60 Keith 68 noted that the postcranial remains from Kabwe do not differ in any striking way from those of modern large-bodied sub-saharan Africans. However, Stringer 69 noted that one iliac blade fragment from Kabwe has a remarkable thickening of the cortical bone on its lateral table, which differs from the condition present in modern humans and probably represents an archaic trait. The Kabwe tibia shares several features with the tibiae of archaic Homo, including a rounded anterior crest and rounded postero-medial and posterolateral angles, as well as a convex contour of the lateral aspect of the shaft, especially proximally. 73 In addition, the tibia s broad proximal epiphysis and its antero-posteriorly narrow and medio-laterally broad distal epiphysis make it distinct from the tibias of ecogeographically matched recent humans. 74 Aside from these features, most of the Kabwe postcranial re mains exhibit few or no differences from anatomically modern morphology. Miners and anthropologists recovered most of the postcranial bones from Kabwe from mining dumps, and thus the specimens except the tibia lack a clear association with the cranium. This fact has given rise to persistent questions regarding whether some of the postcranial remains

5 ARTICLES Evolutionary Anthropology 233 Figure 1. Temporal placement of postcranial fossils relevant to the origin of modern humans. Figures of the fossils after various sources. 61,65 67,71,80,91,101,122,126,175,176,180 might be much more recent than the cranium. However, all of the hominin remains contain large quantities of lead and (usually) smaller quantities of zinc, suggesting that all the bones may be of similar antiquity, 75 although the uptake of trace minerals from surrounding sediments can occur quite quickly during the process of fossilization given the right conditions. Until radiometric dates are obtained for the Kabwe fossils the issue will remain unsettled. For the moment, the available evidence suggests that all the hominin remains from Kabwe are of similar antiquity. The greater similarity to modern humans of the proximal femora from Kabwe than the Berg Aukas specimen highlights the problem of the anatomical variability in African hominins from the early Middle Pleistocene. At least two possible explanations for this variability can be proposed. Some

6 234 Evolutionary Anthropology ARTICLES period in Africa include a juvenile humeral diaphysis from Morocco (Jebel Irhoud 4) 78 and a proximal radius from the Cave of Hearths in South Africa, which may be from either the Early or Middle Stone Age. 79,80 The Jebel Irhoud humerus has thick cortical bone but is otherwise unremarkable. The radius from the Cave of Hearths bears an antero-medially facing radial tuberosity 81 that links it with modern humans as well as other features, such as a slender radial neck and relatively small radial head, in Figure 2. Skeletal features that distinguish H. erectus/h. ergaster from later hominins. After a photograph by Brill. 175 of the more modern-like postcranial fossils may represent populations of hominins that were, in fact, either ancestral to modern humans or more closely related to our ancestors than the populations represented by the specimens such as Berg Aukas. Alternatively, populations of early Middle Pleistocene African hominins may have simply been physically diverse and their skeletons may have contained mosaics of modern and nonmodern traits. If the Kabwe fossils postdate the Berg Aukas femur, it would appear that many aspects of the postcranial skeleton of African hominins were largely anatomically modern by roughly 300 ka BP. 76 Further substantiation of the impression that modern postcranial morphology appeared precociously early in Africa comes from a massive, nearly complete femur (KNM-ER 999) from the Middle through Late Pleistocene Galana Boi Formation at Koobi Fora. The KNM-ER 999 femur has been directly dated to 301,000 / 96,000 BP by gamma ray spectroscopy. 51 The KNM-ER 999 femur shares the apomorphic morphologies of a anteroposterior reinforcement of the femoral midshaft (H. erectus femora have wider ML than AP diameters at midshaft), a point of minimum shaft circumference located proximally (the minimum circumference is located distally in H. erectus). In addition, the KNM-ER 999 femur shares a high neck-shaft angle with the Skhul- Qafzeh femora 77 although the taxonomic valence of this trait is unclear. In Africa, specimens attributable to H. heidelbergensis, such as the Kabwe, Bodo, and Saldanha crania, are followed by a series of cranial remains dating from about 200 ka to about 120 ka BP that present unique mosaics of modern and archaic features. 28 Postcranial remains from this later Experimental studies have shown that cortical bone in both the limb bones and the cranial vault thickens in response to exercise. Thus, the nearly ubiquitous trait of thick cortical bone in prerecent hominins probably reflects elevated activity levels rather than serving as a useful phylogenetic marker. which it resembles Neandertals. 80 In common with most postcranial remains of archaic Homo as well as preindustrial modern humans, the Cave of Hearths specimen has thick cortical bone. 80 Experimental studies have shown that cortical bone in both the limb bones and the cranial vault thickens in response to exercise. 82 Thus it appears likely that the nearly ubiquitous trait of thick cortical bone in pre-recent hominins largely reflects elevated activity levels rather than serving as a useful phylogenetic marker Nevertheless, the thickness of cortical bone may also record some degree of phylogenetic information. Recent humans almost never have cortical bone as thick (when expressed as a percentage of the cross-section of the

7 ARTICLES Evolutionary Anthropology 235 shaft occupied by compact bone) as the Berg Aukas femur. 66 This extraordinary thickness may not be totally explicable in terms of activity and may reflect hormonal or genetic differences instead. The idea that ancient hominin hunter-gatherers were uniformly more active than recent comparative samples is only an inference and not an empirical fact. More work on the plasticity of cortical bone in response to activity and genetic influences is needed before the phylogenetic and behavioral implications of thick cortical bone in Middle Pleistocene hominins can be completely understood. Fossils dating from the start of the Upper Pleistocene (128 ka BP) and later document the next phase of human evolution in Africa. Hominin postcranial remains from this period are associated with cranial remains that have been described as anatomically modern and include a fragmentary skeleton (Omo 1) from the Kibish Formation along the Omo River in southern Ethiopia. 17,85 an assortment of isolated bones from Klasies River Mouth including a proximal ulna, a proximal radius, a first metatarsal a lumbar vertebra, an atlas, and a clavicle, 72,80,86 88 and isolated fragments including a proximal ulna, a humeral shaft, and two metatarsals from Border Cave The proximal ulnae from Klasies River Mouth and Border Cave emerge as distinctly primitive and similar to Neandertals and the Baringo KNM-ER BK 66 ulna in multivariate analyses, 72,91 largely due to their relatively low ration of coronoid to olecranon height. Most (but not all) modern humans have a much more strongly developed coronoid process than Neandertals, Baringo, or the southern African MSA ulnae. The roughly contemporaneous Skhul- Qafzeh humans tend to have somewhat low ratios of coronoid to olecranon height, but display greater coronoid development (hence appearing to be more modern ) than the MSA South African ulnae. Day and his co-workers 17 described the Omo 1 postcranial skeleton as anatomically modern, but robust, and noted that in several respects it bore the closest resemblance to recent Sudanese people among the recent humans used for comparisons. Other noteworthy features of the Omo 1 postcranial skeleton relative to ecogeographically similar recent populations (including African American, Zulu, and Aboriginal Australian males) include slender shafts of the humerus and radius, an AP thick ulnar shaft at the center of the semilunar notch, a long olecranon process, a medially facing radial tuberosity (an infrequent condition in living humans but common in Neandertals, 81 a large distal epiphysis of the first metacarpal, and a very broad shaft and proximal epiphysis of the first metatarsal (personal observations). The Omo 1 ulna differs from those from Klasies River Mouth and Border Cave in having the most modern ratio of coronoid to olecranon height of any of these MSA specimens. 74 Thus in the morphology of their proximal ulnae, the earliest modern humans from Africa and the Levant display a considerable amount of morphological heterogeneity (see Box 2), with some more closely approaching the morphology of later H. sapiens. Omo 1 appears to have been quite tall; the distance from the most inferior point of the deltoid tuberosity (which approximates midshaft) to the bottom of the capitulum measures 184 mm, which indicates a humeral length of about 368 mm and a stature, based on Trotter s formula for African-American males, 93 of approximately cm. Its estimated humeral length and stature place Omo 1 among the tallest early modern humans. Omo 1 s tall stature and its other unusual dimensions relative to warmadapted modern humans accentuate its distinctiveness in multivariate analyses so that the specimen emerges as dissimilar to any recent group (see below). Day and colleagues 17 also found a variety of unusual dimensions of fossil s upper and lower limb bones relative to modern humans, yet they concluded that the skeleton could still be described as anatomically modern. Certainly, contemporaneous Neandertal skeletons differ in many more respects from all modern humans. If future discoveries produce additional East African MSA skeletons that have the same postcranial differences relative to recent humans as Omo 1 does, a case could be made for using the differences as the basis for defining a group of hominins whose anatomy differs predictably from that of modern humans. Although the postcranial evidence for the emergence of modern humans in Africa is sparse, the distinctiveness of the Omo 1 skeleton and the primitive features retained in the Klasies River Mouth and Border Cave ulnae show that the postcranial skeletons of African MSA hominins differed from those of recent humans as well as from early Late Pleistocene non-african hominins such as the Neandertals. The contemporaneous early modern humans from Skhul and Qafzeh in Israel also support the idea that the earliest modern humans are different from recent humans. However, different individuals from these African and Levantine samples often have different sets of features that distinguish them from recent humans. Furthermore, and perhaps more importantly, almost all of their distinctive features, when considered singly, can be found in one or more groups of recent humans, but the entire suite of distinctive features cannot be found in any single group of modern humans. This situation guarantees that the Skhul-Qafzeh and African MSA hominins emerge as significantly different in a statistical sense from all groups of recent humans, but this pattern does not provide convincing evidence that their morphology was not modern in a general sense (see Boxes 1 and 2). African MSA humans as well as the Skhul-Qafzeh hominins should be viewed as early modern humans, or near-modern humans, as Klein 57 has suggested. Such an acceptance of these remains as anatomically modern, however, necessitates that the definition of modern morphology be broadened to include the plesiomorphic traits that some of these hominins posessed but which are rarely if ever encountered among living humans. The initial appearance of modern humans in Africa leaves a series of unanswered questions regarding the precise date when modern humans first evolved on the continent, where they evolved, and the timing and mode of their spread. More hominin finds from the period between 200 ka and 50 ka BP and better dates for some of the known remains will be vital for continuing effort to answer these questions. Likewise, additional

8 236 Evolutionary Anthropology ARTICLES Box 2. The Skhul-Qafzeh Enigma Skhul IV, an early modern human. An enduring mystery in human evolution arises from the fact that some of the early modern humans from Skhul and Qafzeh in Israel display a few distinctive features that characterize Neandertals. In at least one specimen, Skhul VII, these features include a dorso-volarly flat proximal articular surface of the first metacarpal 136 and pronounced lateral curvature, or bowing, of the radius in at least one specimen, Skhul VII. 117 These postcranial traits coincide with some cranial features, including long, low, wide cranial vaults; retromolar gaps; prognatic faces; and weak chins, in which some Skhul- Qafzeh hominins resemble Neandertals. 139,191,199 The mandibular distinction between the two groups blurs even more if the Tabun C2 mandible is an early modern human rather than an archaic hominin The crucial question with regard to the Skhul-Qafzeh hominins is whether their Neandertal-like traits, particularly those otherwise known only in Neandertals and Middle Pleistocene European fossils, are primitive or Neandertal apomorphies. For all of these features, more fossils are needed, especially postcranial remains of Middle and early Late Pleistocene African hominins. If the Neandertal-like features present in the Skhul-Qafzeh hominins prove to be absent in Middle and early Late Pleistocene Africans, then the hypothesis that these features entered the Levantine early modern population through gene flow from Neandertals 49,170,171 would be greatly strengthened. Similar problems pertain to Neandertal-like features such as the presence of dorsal sulci on some Early Upper Paleolithic scapulae 172,173 and the occasional presence of robust long bones in some adult early Upper Paleolithic humans, 74 such as the tibiae of Barma Grande 5 and Cro-Magnon 2, the Paviland skeleton, and perhaps the Gravettian child from Lagar Velho, Portugal. 174 fossil finds are needed to settle the issue of whether the postcranial (and cranial) anatomy of the earliest modern humans differed systematically in any respects from recent humans. One intriguing hypothesis, which is derived from genetic analyses and which could be paleontologically tested only after the accumulation of more fossil evidence, is that the origin of modern humans in Africa was a process that involved several phases rather than a single event. Patterns of diversity and divergence between haplotypes in African mtdna 113 points to an early (c ka BP) expansion of modern humans throughout Africa, followed by a second series of expansions between roughly ka BP which most likely originated from East Africa. The second series of population expansions laid the foundation for the currently observable patterns of mtdna diversity and which obscured some of the evidence of the earlier expansion. The first expansion may have been associated with the earliest anatomically modern fossils in East and South Africa as well as the Levant; the implication of the second set of expansions is that it is conceivable that some of the populations represented by the fossils of early modern humans from South African and the Levant may have contributed little to the subsequent Later Stone Age or Upper Paleolithic populations of those regions. An additional challenge posed by genetic evidence lies in identifying the location in which modern humans evolved. Analyses of the diversity present in many nuclear loci indicate the modern humans went through a bottleneck at some point in the Pleistocene (probably near the origin of our species) in which the effective population size (the number of breeding, adult individuals) fell to about 10, A population of this size almost certainly could only have occu-

9 ARTICLES Evolutionary Anthropology 237 pied a small portion of Africa in order to remain viable. 6 If this scenario is correct, the location in Africa in which modern humans first evolved remains to be identified from fossil or archaeological evidence; once identified it may well provide clues regarding what selective factors triggered the origin of modern humans and why modern humans ultimately proved more successful than archaic humans. Europe Hominin evolution in Europe provides an interesting contrast to the record of human postcranial evolution in Africa. The record of hominin postcranial fossils from elsewhere in Eurasia is too fragmentary to indicate much about the patterns of evolution that occurred. Neandertals, like African early modern humans, evolved from more archaic Middle Pleistocene hominins, in particular Homo heidelbergensis. 48,94 The evolution of Neandertals and that of the earliest modern humans in Africa proceeded over a about 500 ka years. By 300 ka BP, cranial and postcranial fossils occur in Europe that have mosaics of the features that anticipate the distinctive morphology of Neandertals in the Late Pleistocene. Mosaics of modern and archaic traits become common in African fossil hominins at about the same time. By 128 ka BP Neandertals had evolved in Europe and modern or near-modern humans had emerged in African and the adjacent Levant. The European fossil record of hominin postcranial remains starts with the 800-ka-year-old remains of Homo antecessor from the Gran Dolina site at Atapuerca. 95,96 The taxonomic status of H. antecessor remains debated. 94 It may be a variant of H. heidelbergensis, a population that went extinct without giving rise to any later hominins or, in the phylogeny favored by Bermúdez de Castro and his coworkers, 97 the common ancestor of Neandertals and H. sapiens. The postcranial remains of H. antecessor, which consist of 45 fragments including hand and foot bones, vertebrae, ribs, three clavicles, two radii, a femoral fragment, and two patellae, are remarkable in that their morphology, in most aspects, resembles that of modern humans rather than that of Neandertals or, to a lesser extent, the H. heidelbergensis specimens from Sima de los Huesos at Atapuerca. 95,96 The next postcranial remains of European hominins considerably postdate H. antecessor and display more have differences from modern humans and some similarities with Neandertals. The resemblance to Neandertals becomes progressively more pronounced later in time. These remains are contemporaneous with cranial remains of H. heidelbergensis 94 and most likely represent that taxon. The earliest of these specimens is the 500-ka-year-old tibia from Boxgrove in southeastern England. The fragmentary Boxgrove tibia is strikingly massive, indicating that the bone belonged to a very large individual 73,98 and thus reinforces the idea that H. heidelbergensis had a large stature and very high body mass. 69,99,100 The European fossil record of hominin postcranial remains starts with the 800-kayear-old remains of Homo antecessor from the Gran Dolina site at Atapuerca. The site of Arago in southwestern France yielded a series of hominin postcranial fossils dating to roughly 400 ka BP. The postcranial remains consist of a left os coxa, a proximal femur and two other femoral fragments, a fibular fragment, a distal manual phalanx, and a second metatarsal The most important of these remains are the os coxa, which has a pronounced iliac pillar, recalling the morphology of the earlier Olduvai OH 28 hip bone 61 and earlier specimens such as the 1.9 million-year-old KNM-ER 3228 innominate from Koobi Fora. 104 The Arago proximal femur resembles the OH 28 femur as well as other femora attributed H. ergaster from Koobi Fora. 101 The hominin remains from the 300- ka-year-old site of Sima de los Huesos in the Sierra de Atapuerca in northern Spain provide better evidence by far of the morphology of European specimens of H. heidelbergensis, as well as more postcranial fossils of Middle Pleistocene Homo, than do specimens all the other sites in the world combined. 109 Every skeletal element is represented, usually by multiple bones, in the sample from Sima de los Huesos. Descriptions of the humeri, clavicles, scapulae, three tibial fragments, and the pelvic remains have been published. 65,106,107 Additional information is available about the femora, tibiae, patellae, radii, ulnae, vertebrae, and other bones of these hominins. 95,96,108 Like the crania from the site, the Sima de los Huesos postcranial remains display mosaics of morphological traits that resemble those of modern humans, and which, therefore, may be primitive for H. heidelbergensis, as well as other traits resembling derived conditions that later characterize the Neandertals. 106,107 Perhaps the most important information from the these postcranial fossils so far comes from analyses showing that the degree of sexual dimorphism in the sample is comparable to that in living humans 10,109 and the considerable knowledge that has been derived from the discovery of a complete male pelvis (Pelvis 1) from the site. 65 Pelvis 1 has almost all of the morphological features that distinguish Neandertal and modern human pelves, 110,111 including a long pubis with a supero-inferiorly flattened superior ramus, a very wide bi-iliac breadth, and a medio-laterally wide inlet of the birth canal. 65 The pelvis extraordinarily wide bi-iliac breadth (340 mm) and the stature estimated from an associated partial femur combine to provide a body mass of at least 95 kg for an individual to cm tall. If the estimates are accurate, the Atapuerca hominins were impressively heavy in comparison to early modern humans and also heavier for their stature than most modern populations, but of a slightly more linear build than the Neandertals who succeeded them. In Europe, the Neandertals evolved gradually and in a mosaic fashion from ancestors such as the Sima de los Huesos hominins. Most European hominins that date from the period between 250 ka and 128 ka BP possess mosaics of Neandertal and H. heidelbergensis traits. Hominins that pos-

10 238 Evolutionary Anthropology ARTICLES Figure 3. The Jinniushan partial skeleton After Lü. 176 sess the entire suite of cranial and postcranial traits that characterize the well-known western European Neandertals during the period from ka BP first appear around 128 ka BP at Krapina. 112 The Neandertal postcranial skeleton, like the skull, has numerous morphological specializations that distinguish it from the skull of modern humans. Many of these features involve limb and body proportions, which largely appear to be adaptations to a cold environment 16,18 20 ; robust limbs, which reflect both their generally stout and heavy bodies as well as elevated activity levels; and a variety of primitive features inherited from Middle Pleistocene ancestors. Finally, they have a number of features that appear to be synapomporphies of the Neandertal clade (see Box 1) and to have no analogs, or very few, among living humans. These features include the presence of a dorsal sulcus on the axillary border of the scapula, a dorso-volarly flat proximal articular surface of the first metacarpal, 113,114 a reduced styloid process of the third metacarpal, a large hamulus of the hamate, and other carpal and metacarpal specializations. 114,115 Neandertals also tend to display pronounced lateral curvature (bowing) of the radius and anterior-posterior bowing of the femur, an elongated distal phalanx of the thumb relative to the proximal phalanx, 113,114 a pronounced opponens pollicis crest on the first metacarpal, 113 and enlarged apical tufts of the distal phalanges of their hands and feet, 113,114 and a host of other small details ,136 The Neandertal pelvis differs from that of recent humans primarily in having a long pubis, a laterally rotated acetabuli, and an anteriorly positioned and medio-laterally wide birth canal. 111,114,117 Many of these features seem to be retained from H. heidelbergensis, at least as represented by the Atapuerca Sima de los Huesos pelvis. 65 No single cause such as a generally elevated level of robusticity, heterochrony, or hormonal differences suffices to explain the entire suite of features that distinguish Neandertals from living humans. 22,119 Instead, the presence of so many postcranial specializations among the Neandertals not only reinforces the impression that these hominins developed their distinctive morphology during a long period of virtual isolation from other hominins, 120 but provides a ready source of support for those who consider that Neandertals constitute a separate species from Homo sapiens. Asia Information regarding hominin evolution including evolution of the postcranial skeleton, in the vast continent of Asia remains sparse due to a series of large temporal and geographic gaps between the few fossils that document the course of human evolution. A similar dearth of fossils documents to hominin evolution in South Asia, 121 despite the fact that both H. erectus and H. sapiens almost certainly passed through India at an early date during their dispersals into the Sunda Shelf. The earliest hominin postcranial specimens recovered from Indonesia include several femoral diaphyses from Trinil 62,122 and one tibial diaphysis from Sambungmacan. 123 Two hominin tibial shafts from Ngandong 124,125 long postdate the earlier remains from Trinil and Sambungmacan. 53,54 All of these remains share thickened cortical bone, 62 a trait that probably reflects elevated activity levels relative to recent humans, 82 as well as adequate nutrition. The earliest hominin postcranial remains from China are the H. erectus fossils from Zhoukoudian. 126,127 These include five fragmentary femora, three humeral diaphyses, a tibia, It remains unclear whether hominins such as Jinniushan evolved from H. erectus in China or migrated into Eastern Asia from elsewhere. There are no homologous Asian H. erectus specimens to compare to Jinniushan s postcranial remains. a clavicle, and a lunate. The limb bones possess thick cortical bone 62,126 ; both humeri possess a ruggedly modeled deltoid tuberosity; and the femoral shafts are wide mediolaterally relative to antero-posterior diameters at midshaft and at the subtrochanteric level. 126 A noteworthy feature of the H. erectus remains from Zhoukoudian lies in their relatively recent age of between 580 and 230 ka BP. 127 The youngest of these ages overlaps with age assessments for more derived hominins such as Jinniushan, suggesting that the two species may have overlapped temporally in China. 128 The largely undescribed female skeleton from Jinniushan 127,129 (Fig. 3), which dates between 200 and 300 ka BP, 130 documents the next phase of

11 ARTICLES Evolutionary Anthropology 239 human evolution in China. It remains unclear whether hominins such as Jinniushan evolved from H. erectus in China or migrated into Eastern Asia from elsewhere. There are no homologous Asian H. erectus specimens to compare to Jinniushan s postcranial remains. However, studies of Jinniushan s os coxa have revealed that it has similarities to that of Neandertals and H. heidelbergensis, including a vertically thin pubic ramus and a very broad bi-iliac breadth. 65,129,131 These features combine to suggest that Jinniushan s postcranial affinities may lie with H. heidelbergensis. Jinniushan has an estimated body mass between 79.5 and 75.9 kg and an estimated stature of 168 cm. 131 By modern human standards, these estimates indicate that the Jinniushan woman would have been very heavy for her height and accords with indications that Middle Pleistocene archaic Homo had relatively massive, muscular bodies. 64,65,99,100 The complete ulna of the Jinniushan skeleton is slender, 127 an observation that contrasts with the skeleton s high estimated body mass relative to its stature. 132 However, Neandertals also possess unexpectedly slender forearm bones, given the robusticity of their humeri and lower limb long bones. 132 The earlier African H. heidelbergensis ulna from Baringo Kapthurin is also remarkably slender. 71 One final enigmatic archaic human fossil deserves mention, a diminutive Middle Pleistocene clavicle from Narmada. 133 The clavicle is adult and yet corresponds in size to those of Andaman Islanders, 133 providing a curious contrast with other known Middle Pleistocene hominin remains, which are generally large. The clavicle stands as the only postcranial evidence of archaic hominins from South Asia. It remains to be seen if future finds will provide additional evidence of individuals of small stature or shed more light on the controversial taxonomic status of the hominin population now represented by only the clavicle and Narmada calvarium. 5,134,135 The first fossils of modern humans in Asia date to about 50 ka BP in China. 127 Humans colonized Austria by at least 50 ka BP 136 and settled many islands of the drowned Sunda Shelf, as well as the Philippines, by around 40 ka BP. 14 The Lake Mungo 3 burial holds the record as the earliest human fossil from Australia. Just how early is controversial, but the skeleton is at least 28 ka and possibly 60 ka old The femoral morphology of late Pleistocene Australians does not resemble that of H. erectus despite the cranial similarities noted by Multregionalists. 140 This concludes the overview of hominin postcranial fossils relevant to the origin of modern humans. It is now useful to ask what these remains can say about the mode of origin of modern humans. IMPORTANT ASPECTS OF POSTCRANIAL MORPHOLOGY A great variety of skeletal features could be discussed in the context of modern human origins. This review covers only some of the most striking and widely discussed traits. These aspects of postcranial morphology include body proportions, such as the brachial and crural index and limb length relative to trunk height and body mass, 16,18 20 the robusticity of limb bones, the cross-sectional shapes of long bone shafts, and patterns of overall phenetic distances from multivariate analyses employing large numbers of postcranial dimensions. Body Proportions Perhaps the most widely discussed aspect of postcranial morphology that is relevant to the origin of modern humans involves the striking differences in body proportions between Neandertals and early modern humans, including both the 100-ka-yearold early modern hominins from Skhul and Qafzeh in Israel and the Early Upper Paleolithic modern humans associated with Aurignacian and Gravettian industries. Neandertals tend to have proportionally shortened distal limb segments (that is, low brachial and crural indices 16 ); wide, barrel-shaped chests 16,114 ; short limbs relative to trunk length or body mass 19,20 ; a broad pelvis 18 ; and an elevated body mass, especially relative to stature. 18,20,64,100 In these proportions, Neandertals resemble recent humans from the Arctic and differ maximally from humans from the warmest areas. 16,18 20 In stark contrast, the earliest modern humans from Israel and Europe have elongated distal limb segments, 16 long limbs relative to trunk height and body mass, 19 a very linear physique, a narrow pelvis, and low estimated body mass relative to stature. 19,20,100,141 This suite of features causes the earliest modern humans, especially those from Qafzeh and Skhul, to resemble recent humans from hot, dry climates, such as Nilotic Africans and Australian Aborigines. Early Upper Paleolithic modern humans are less extreme in terms of body shape, 18,19 but still most closely approximate populations from warm areas such as Sub-Saharan Africa. In sum, early modern humans and Neandertals differ tremendously in their body proportions and shape. Given the clear ecogeographic patterning of body proportions among living humans, the data from body proportions strongly suggest that the Skhul-Qafzeh hominids evolved in a very warm climate like that in much of Africa and that the ancestors of Early Upper Paleolithic Europeans were adapted to a warmer climate than the period between ka BP. 19,20,141 In this context, the disappearance of the cold-adapted Neandertal physique from the population of the Early Upper Paleolithic supports the notion that modern humans who migrated into Europe at the start of the Upper Paleolithic largely or entirely replaced the Neandertals. 19 Robusticity Next to the dramatic differences in body proportions, the most striking contrast between Neandertals and early modern humans lies in the robusticity of their long bones. For the purpose of this review, robusticity is defined simply as the thickness of long-bone shafts or epiphyses relative to bone length. This definition corresponds to the classical definition of robusticity but differs from some recent uses of the word. 132 The primitive hominin condition for the ancestors of Neandertals and modern humans was a linear skeleton that had slender long-bone shafts and joints relative to the lengths of the bones. The 1.6 million-year-old KNM-WT H. ergaster skeleton clearly illustrates this morphology, 142 as does the slender 500-ka-old Baringo Kapthurin ulna. 71 Neandertals departed from the ances-

12 240 Evolutionary Anthropology ARTICLES Figure 4. Plots of diaphyseal (left) and epiphyseal (right) robusticity for males. Fossil hominins are shown on the right of each graph in rough chronological order, with the oldest groups on the left. The right half of each graph shows recent groups. Populations from warm climates are plotted to the left; groups from cold areas are on the far right of each graph. A diamond encloses each group mean. The horizontal line in the middle of each diamond indicates the mean; the vertical extent of the diamond indicates the 95% confidence limits for the mean. Non-overlapping diamonds indicate significant differences. In most of the plots, groups from cold areas are more robust than populations from warm areas, especially in the ratios of epiphyseal size. Neandertals most closely resemble recent populations from cold areas. Early modern humans resemble recent people from warm climates. Note the increase in robusticity among Epigravettian and Magdalenian humans. tral condition, whereas the earliest modern humans did not. Neandertals have been characteristically described as possessing massive joints and thick long-bone diaphyses relative to their length, while early modern humans have a gracile skeleton with slender long-bone shafts and joints relative to their generally considerable length. 114, , Until recently, these features were considered indicative of a body capable of exerting much less force on a habitual basis than was the Neandertal body. 114,115 It now seems, however, that the stout skeletons of Neandertals and the slender skeletons of early modern humans are both a function of ecogeographic variation. 74,132,146,147 Modern populations from hot areas tend to have much more slender longbone diaphyses and epiphyses than do groups adapted to colder climates. 18,24,74,128 In the slenderness of their limb bones, early modern humans resemble extant humans from the warm areas, including Africans and Aboriginal Australians, while Neandertals most closely approximate recent human populations from the Arctic, such as the Inuit or Sami (Fig. 4). Living European populations (such as the Sami and European Americans as well Mesolithic populations from France and Denmark) have relatively thicker long bone shafts and more massive joints relative to length than Early Upper Paleolithic Europeans (Fig. 4). 74,132 The average level of

13 ARTICLES Evolutionary Anthropology 241 Figure 5. Fossil hominin skeletons. All are the same scale. Left to right: KNM-WT (a juvenile; after David Brill s photograph 210 ); the Levantine early modern human Skhul IV; the Neandertals Tabun I (a female) and La Ferrassie 1 (a male); the Gravettian (Early Upper Paleolithic) male Grotte des Enfants 4; and the Magdalenian (Late Upper Paleolithic) male Oberkassel 1. Note the slender build of Grote des Enfants 4 and especially Skhul IV as compared to the more robust skeleton of Neandertal La Ferrassie 1. Oberkassel 1 demonstrates the increased robusticity of Late Upper Paleolithic populations relative to Early Upper Paleolithic Europeans. After various sources. 117, Some bones are mirror-imaged to make the remains more complete. skeletal massiveness appears to have increased dramatically in Europeans at about the time of the Last Glacial Maximum, so that many Late Upper Paleolithic Europeans, such as the Magdalenian skeleton Oberkassel 1 (Fig. 5), converge on the level of massiveness present in Neandertals and extant cold-adapted populations, including recent Europeans. These results closely follow the patterns in body proportions and mass relative to stature. 16,18 20,180 In terms of body shape and the gracility of their limbs, early modern humans resemble living populations from warm areas but differ dramatically from extant cold-adapted populations, and the Neandertals. Any definition of what constitutes anatomically modern form should take into consideration the fact that some populations of living humans have evolved a body form that is substantially different from that of early modern humans (see Box 1). Long-Bone Shaft Shapes The cross-sectional shapes of long bone shafts, expressed as ratios of anteroposterior to mediolateral shaft diameter or maximum versus minimum second moment of area (I max /I min ), Any definition of what constitutes anatomically modern form should take into consideration the fact that some populations of living humans have evolved a body form that is substantially different from that of early modern humans... have commonly been used as barometers of patterns of activity. 84,147, As Jurmain 153 has recently observed, however, the support for the link between cross-sectional shape and activity in humans is almost entirely circumstantial. Few experimental studies have documented how much change in shaft shape a given amount of activity will produce. Despite this lack of empirical support, the notion that shaft shapes reflect activity has become deeply entrenched in the literature. Although diaphyseal crosssectional shapes are highly variable within recent humans, the idea has persisted that these shapes may also capture some phylogenetically important information 152,154 even if, for example, some aspects of femoral shaft shape may be an after-effect of body shape or obstetric adaptations. 155 Early modern humans differ strikingly from Neandertals in possessing a well-developed femoral pilaster. 116,118,154 Highly pilastered femora typify highly mobile recent groups. 84 Neandertals have characteristically rounded femoral midshaft cross-sections, 154 as do sedentary recent groups. 84 The Saint-Césaire 1 Neandertal, however, has a well-developed- femoral antero-posterior reinforcement at midshaft, but one that differs from the usual morphology in modern humans. In Saint-Césaire, the medial buttress commonly present in the proximal femur of Neandertals extends to midshaft, where it expands dorsally to increase the antero-posterior structural strength of the shaft. 147 The Kabwe femora shafts have moderately elevated pilastric indices, as does the KNM-ER 999 femur 77 and a recently discovered femur that most

14 242 Evolutionary Anthropology ARTICLES Figure 6. Canonical variates analyses results showing modernity versus time. The horizontal axis shows time before present. The vertical axis shows the probability that a fossil or group of fossils could be encountered within the nearest recent group in discriminant space. A dashed horizontal line marks the P 0.05 level, above which a fossil could be considered anatomically modern in this study. Fossils that fit this definition of modern become common only late in time, in the Late Upper Paleolithic. likely derives from early Middle Stone Age strata at Mumbwa Caves, Zambia. 186 These modestly elevated pilastric indices may herald the dawn of modern or near-modern behavioral patterns around 300 ka BP in Africa. It should be stressed, however, that although the earliest anatomically modern humans have elevated pilastric indices that represent a dramatic departure from archaic hominins, 154 shaft shapes are thought to be plastic in response to behavior. Many living humans have weakly developed pilasters and rounded femoral cross-sections. The presence or absence of a femoral pilaster is not sufficient for unequivocal identification of a skeleton as modern or archaic. Similar cautions apply to other aspects of long bone shaft shapes in which the Skhul- Qafzeh or Early Upper Paleolithic humans differ from Neandertals or recent humans, or both. Long bone shaft shapes should not serve as the primary basis of determining whether a fossil is anatomically modern. PATTERNS OF MORPHOLOGICAL DISTANCE An important source of information comes from the overall phenetic affinities of the fossil postcranial skeletons relevant to modern human origins. Canonical variates analysis, a multigroup extension of discriminant function analysis, 157 offers one of the best ways to assess phenetic relationships from a perspective aimed at uncovering and describing overall betweengroup differences. Figure 6 summarizes the results of 48 canonical variates analyses, each of which contrasted one or more fossil skeletons against a series of nine modern populations that differed in terms of habitual activities, climatic adaptations (physique) and continent of origin. 74 The males and females of each population were treated as separate samples. Each recent sample consisted of approximately 25 individuals, although sample sizes ranged between 7 and 41. Due to the fragmentary nature of the fossils, most of the canonical variates analyses were tailored to include as many of the preserved measurements as possible on a particular fossil skeleton. The most complete specimens, however, could be included in many analyses of other specimens and served as yardsticks against which the affinities of those other fossils could be judged. Fairly complete fossils include Skhul IV (an early modern human), Tabun C1 and La Ferrassie 1 and 2 (Neandertals), Grotte des Enfants 4 (Gravettian), Ohalo 2 (Kebaran), and Arene Candide 5 (Epigravettian). Up to 122 measurements of the upper and lower limb were analyzed, as well as the limb girdles, with, of course, many fewer measurements available for the most fragmentary specimens. Canonical variates analyses produce matrices of Mahalanobis distances (D 2 )between groups (each fossil or group of fossils is treated as a group mean) as well as a probability that the observed distance could be sampled from a single population. Following a modified version of the heuristic definition of modern morphology developed by Kidder and colleagues, 158 fossil hominins were considered anatomically modern if they did not differ significantly (P 0.05) from at least one of the recent human samples. The horizontal axis of Figure 6 represents time in years before present. Each fossil or group of fossils (if more than one fossil from a cultural group was complete enough to be included in the analysis) is plotted as a point for each analysis in which it appears. The vertical axis represents the probability that the fossil (treated as a group mean) could be drawn from the recent sample from which it had the smallest D 2 distance. The dashed line indicates a probability of 0.05; specimens that fall above that line fit the operational definition of anatomically modern, while those that fall below the line do not. The most striking result of these analyses of the fossil skeletons is that virtually all of the postcranial skeletons predating the Last Glacial Maximum differ greatly from those of recent humans. Fossil hominins that do not differ significantly (at P 0.05) from one or more of the recent samples become common only after 20,000 to 18,000 BP. 74 There are a few exceptions. In a few instances in which very few measurements could be included, some Neandertals, Skhul-Qafzeh specimens such as Skhul IV and II, and Early Upper Paleolithic humans resemble the recent groups. These analyses often do not include the aspects of morphology that differentiate these fossils from recent humans; more comprehensive analyses on complete specimens such as Skhul IV and Grotte des Enfants 4 show them to be highly distinct from any of the recent samples. Other exceptions prove less easy to explain away: Qafzeh 9 emerges as modern fairly consistently, while female Neandertals and Early Upper Paleolithic humans more closely approach modern morphology than their male coun-

15 ARTICLES Evolutionary Anthropology 243 Figure 7. Unweighted pair-group method using arithmetic averages (UPGMA) clustering analysis of Mahalanobis D 2 distances between groups from a canonical variates analysis that included skeletal elements from the upper and lower limb of three well-preserved fossils, Skhul IV, La Ferrassie 1, and Grotte des Enfants 4. Note that the fossils are distant from all recent populations and cluster outside of them. terparts do. In a few analyses, female Neandertals and Early Upper Paleolithic humans fall within the 95% range of some recent groups. Perhaps the most notable result of the analyses is the late appearance of a modern postcranial form, a situation that also characterizes the evolution of modern cranial form. 15, The results of the discriminant analyses show that, like cranial morphology, postcranial morphology closely resembling recent humans appeared surprisingly recently, but can say still more about the origin of modern humans. Most of the predictions generated from the competing models of modern human origins expect certain patterns of close or distant relationships among fossils or between recent groups and specific groups of fossils. These predictions can be addressed with the matrices of betweengroup distances produced by the canonical variates analyses. The matrices of D 2 distances from these analyses can be summarized by clustering algorithms. The resulting trees give overall impressions of the degree of distance between the groups analyzed. Figure 7 presents one such clustering, based on an analysis that included measurements preserved in three of the most complete fossil skeletons, La Ferrassie 1, Skhul IV, and Grotte des Enfants 4. It uses measurements of most of the major long bones, limb girdles, hands, and feet. The resulting unweighted pair-group method using arithmetic averages (UPGMA) clustering 161 closely resembles those produced by the other analyses. The tree shown in Figure 7 has several notable features. All of the recent groups are closely packed together; the fossils cluster on the periphery. The most recent fossil, Epigravettian skeleton Arene Candide 5, falls closest to the recent groups, followed by the Gravettian male Grotte des Enfants 4. Skhul IV lies marginally closer to recent humans than does La Ferrassie 1, but both of these Middle Paleolithic hominins are at very large distances from all of the recent groups. Enormous distances also separate the early moderns from La Ferrassie. In most analyses, much larger D 2 distances separate early modern humans from Neandertals than are present between either of these fossil groups and the recent humans. Neandertals and the probable ancestors of recent humans therefore differ tremendously in their postcranial morphology, a situation that again mirrors the results of multivariate analyses of crania. 15,162 Smaller, but still impressively large distances tend to separate Skhul- Qafzeh postcranial skeletons from those of Early Upper Paleolithic modern humans (for example, from Skhul IV to Grotte des Enfants 4). If the Skhul-Qafzeh hominins were the ancestors of Upper Paleolithic Europeans, 57,114,144 then it is evident that a considerable amount of morphological change occurred in the 50 to 70 ka that separate them. The fact that so many so-called early modern human skeletons, including Omo 1, most of the Skhul- Qafzeh individuals, and the large majority of Aurignacian and Gravettian skeletons, emerge as being so different from any recent human group in the analyses deserves further comment. Skeletal differences arise from multiple sources including genetically based between-group or between species variation and differences such as degree or kind of habitual activity, which may not necessarily reflect genetic differences. Genetically based differences include long-term adaptations to climate and, perhaps, specialized joint shapes, whereas the thickness of long bones and their shafts proportions of antero-posterior to mediolateral diameters are obvious candidates for aspects of morphology that probably are strongly influenced by habitual activity. Because canonical variates analyses cannot distinguish between the sources of dissimilarity, a result indicating that two groups are dissimilar could stem from any one of the influences on skeletal morphology or a combination of several of these factors. To determine why a given analysis produced its results, one must examine its eigenvectors as well as the patterns of between-group differences present in the original data. A detailed analysis of the underlying causes for the results of the analyses discussed lies beyond the scope of this paper, but it is worth noting that the early modern humans generally possess few skeletal features that distinguish them from all recent groups. At the same time, early modern skeletons usually have some features, such as long bone robusticity, shaft shapes, or body proportions, that distinguish them from several groups of recent humans but not from all such groups. 74,165 In contrast, Neandertals have many features that distinguish them from all groups of recent humans. A number of conclusions may be drawn from these results in regard to the predictions of the competing models of modern human origins. The fact that Qafzeh 9 emerges as modern

16 244 Evolutionary Anthropology ARTICLES Figure 8. Regional affinities. The chart shows the frequencies with which the closest recent group to a fossil or group of fossils in each canonical variates analysis was of African, Australian, European, or Northern Asian (Inuit) origin. European fossils begin to resemble recent European (or Inuit, another cold-adapted recent group) most closely in the Late Upper Paleolithic. Importantly, almost all the fossils before the Late Upper Paleolithic differ greatly from all the recent groups; the closest group is often nearest by a small margin. could be taken as weak support of the prediction that the earliest modern humans should appear within or near Africa, but the rest of the reasonably complete Skhul-Qafzeh specimens, as well as the Omo 1 skeleton, differ strikingly from any of the recent human groups. Nevertheless, it must be remembered that the Out-of-Africa Model allows the earliest modern humans to differ substantially from recent humans, although the amount of observed distance is perhaps unexpectedly large under that model. That model s prediction that all early modern humans will be more similar to each other and to recent humans than they are to Neandertals is supported by the results, but again, the distances between the Skhul-Qafzeh and Early Upper Paleolithic humans are larger than expected. The last prediction of the Out-of-Africa Model that recent humans from cold (arctic) climates in widely separated regions should be similar to Neandertals in aspects of morphology that reflect cold adaptations has been shown to be true for body proportions, 16,18 20 and is equally true for the relative thickness of long bone shafts and joints relative to bone length. 154 The nearest recent neighbors of fossil specimens in the analyses can also help to test the prediction derived from the Multiregional model that European hominins of any antiquity should be the most similar to European populations among recent humans. Tabulation of the nearest recent neighbors for fossil specimens from the analyses (Fig. 8) offers no support for this prediction. Neandertals as well as Early Upper Paleolithic Europeans most often and most closely tend to resemble recent African groups rather than Europeans. Only in the Late Upper Paleolithic (Magdalenian and Epigravettian) do European fossils frequently tend to resemble a European group most closely (or the Inuit, another northern, cold-adapted population). CONCLUSIONS Modern human postcranial morphology appears to have developed in African hominins between about 600 BP and 125 ka BP, the same period in which the highly distinctive Neandertal postcranial pattern evolved in Europe. At present, it is difficult to evaluate the possibility that Asian hominins, particularly the late archaic specimens represented by cranial remains from Dali, Jinniushan, and Xujiayao, contributed to the modern human gene pool. This difficulty arises because of the lack of enough postcranial remains and the fact that a complete description of the Jinniushan skeleton has not yet appeared. Early modern humans from the Middle Paleolithic sites of Skhul and Qafzeh, as well as the European Early Upper Paleolithic, differ dramatically from Neandertals in many aspects of skeletal anatomy that reflect climatic adaptations. The earliest modern humans in Europe and the Levant also lack most of the other features that distinguish Neandertals from more recent modern humans. However, Late Upper Paleolithic Europeans evolved a physique that converged upon that of Neandertals in many body proportions and limb bone robusticity. 132 The pattern of postcranial evolution associated with the origin of modern humans strongly suggests that population replacement (or at least very little admixture with Neandertals) accompanied the appearance of modern humans in Europe. These results could fit with either the Out-of- Africa model or a version of the Multiregional model in which replacement greatly outweighed continuity. ACKNOWLEDGMENTS Thanks to John Fleagle for the encouragement to submit this paper to Evolutionary Anthropology. John Fleagle, Steve Churchill, Bill Jungers, Daniel Lieberman, Brian Richmond, David Strait, and two anonymous reviewers provided helpful comments on early drafts. The efforts of the many curators who allowed me to study the fossil and recent human skeletons in their care are greatly appreciated. Much of this paper was written while the author was a postdoctoral research associate in the Department of Anthropology at the George Washington University, and I am grateful for the Department s support. The research reported in this paper was supported by an NSF predoctoral fellowship, the Wenner-Gren Foundation, Boise Fund, NSF DBS to Frederick Grine, NSF SBR and a grant from the Leakey Foundation to John Fleagle, and NSF IBN to Daniel Lieberman. REFERENCES 1 Smith FH, Spencer F, editors The origins of modern humans: a world survey of the fossil evidence. New York: Alan R. Liss.