The First Four Million Years of Human Evolution

RSTB_365_1556_Cover.qxd 9/7/10 10:42 AM Page 1 Phil. Trans. R. Soc. B ISSN 0962-8436

volume 365

27 October 2010 number 1556 . .

volume 365 number 1556 pages 3263–3410 | vol. 365 no. 1556 pp. 3263–3410 pages 3263–3410 The first four million years of human evolution In this Issue Papers of a Discussion Meeting issue organized and edited by Alan Walker and Chris Stringer The first four million years of human evolution

Introduction Papers of a Discussion Meeting issue organized and edited by Alan Walker and Chris Stringer The first four million years of human evolution 3265 A. Walker & C. Stringer Articles In search of the last common ancestor: new findings on wild chimpanzees 3267 W. C. McGrew |

More reliable estimates of divergence times in Pan using complete mtDNA sequences 27 Oct 2010 and accounting for population structure 3277 A. C. Stone, F. U. Battistuzzi, L. S. Kubatko, G. H. Perry Jr, E. Trudeau, H. Lin & S. Kumar Spinopelvic pathways to bipedality: why no hominids ever relied on a bent-hip–bent-knee gait 3289 C. O. Lovejoy & M. A. McCollum Arboreality, terrestriality and bipedalism 3301 R. H. Crompton, W. I. Sellers & S. K. S. Thorpe Two new Mio-Pliocene Chadian hominids enlighten Charles Darwin’s 1871 prediction 3315

M. Brunet The first four million years of human evolution Phylogeny of early Australopithecus: new fossil evidence from the Woranso-Mille (central Afar, Ethiopia) 3323 Y. Haile-Selassie Anterior dental evolution in the Australopithecus anamensis–afarensis lineage 3333 C. V. Ward, J. M. Plavcan & F. K. Manthi Molar microwear textures and the diets of Australopithecus anamensis and Australopithecus afarensis 3345 P. S. Ungar, R. S. Scott, F. E. Grine & M. F. Teaford An enlarged postcranial sample confirms Australopithecus afarensis dimorphism was similar to modern humans 3355 P. L. Reno, M. A. McCollum, R. S. Meindl & C. O. Lovejoy The cranial base of Australopithecus afarensis: new insights from the female skull 3365 W. H. Kimbel & Y. Rak Hominin diversity in the Middle Pliocene of eastern Africa: the maxilla of KNM-WT 40000 3377 F. Spoor, M. G. Leakey & L. N. Leakey Stable isotopes in fossil hominin tooth enamel suggest a fundamental dietary shift in the Pliocene 3389 J. A. Lee-Thorp, M. Sponheimer, B. H. Passey, D. J. de Ruiter & T. E. Cerling Retrieving chronological age from dental remains of early fossil hominins to reconstruct human growth in the past 3397 M. C. Dean

The world’s first science journal Founded in 1660, the Royal Society is the independent scientific academy of the UK, dedicated to promoting excellence in science rstb.royalsocietypublishing.org Registered Charity No 207043 Published in Great Britain by the Royal Society, 27 October 2010 6–9 Carlton House Terrace, London SW1Y 5AG See further with the Royal Society in 2010 – celebrate 350 years RSTB_365_1556_Cover.qxd 9/7/10 10:42 AM Page 2

GUIDANCE FOR AUTHORS Editor Professor Georgina Mace Selection criteria to satisfy most non-specialist readers. Supplementary The criteria for selection are scientific excellence, data up to 10Mb is placed on the Society's website free Publishing Editor originality and interest across disciplines within biology. of charge. Larger datasets must be deposited in Joanna Bolesworth The Editors are responsible for all editorial decisions and recognised public domain databases by the author. they make these decisions based on the reports received Editorial Board from the referees and/or Editorial Board members. Many more good proposals and articles are submitted to us Conditions of publication Neuroscience and Cognition Organismal, environmental and evolutionary biology than we have space to print, we give preference to Articles must not have been published previously, nor be Dr Brian Billups Professor Spencer Barrett Dr Andrew Glennerster Professor Nick Barton those that are of broad interest and of high scientific under consideration for publication elsewhere. The main Professor Bill Harris Dr Will Cresswell quality. findings of the article should not have been reported in Professor Trevor Lamb Professor Georgina Mace the mass media. Like many journals, Phil. Trans. R. Soc. B Professor Tetsuro Matsuzawa Professor Yadvinder Malhi Professor Andrew Whiten Professor Manfred Milinski Publishing format employs a strict embargo policy where the reporting of Professor Peter Mumby Phil. Trans. R. Soc. B articles are published regularly online a scientific article by the media is embargoed until a Cell and developmental biology Professor Karl Sigmund specific time. The Executive Editor has final authority in Professor Makoto Asashima and in print issues twice a month. Along with all Royal Dr Buzz Baum Health and Disease Society journals, we are committed to archiving and all matters relating to publication. Professor Martin Buck Professor Zhu Chen providing perpetual access. The journal also offers the Dr Louise Cramer Professor Mark Enright facility for including Electronic Supplementary Material Dr Anne Donaldson Professor Michael Malim Professor Laurence Hurst Professor Angela McLean (ESM) to papers. Contents of the ESM might include Electronic Submission details Professor Fotis Kafatos Professor Nicholas Wald details of methods, derivations of equations, large tables For full submission guidelines and access to all journal Professor Elliot Meyerowitz Professor Joanne Webster Professor Dale Sanders of data, DNA sequences and computer programs. content please visit the Phil. Trans. R. Soc. B website at Dr Stephen Tucker However, the printed version must include enough detail rstb.royalsocietypublishing.org.

Publishing Editor: Joanna Bolesworth AIMS AND SCOPE Each issue of Phil. Trans. R. Soc. B is devoted to a specific area of the biological sciences. (tel: +44 (0)20 7451 2602; fax: +44 (0)20 7976 1837; This area will define a research frontier that is advancing rapidly, often bridging [email protected]) traditional disciplines. Phil. Trans. R. Soc. B is essential reading for scientists working The Royal Society, the national academy of science of For further information on the Society’s activities, please across the biological sciences. In particular, the journal is focused on the following four the UK and the Commonwealth, is at the cutting edge contact the following departments on the extensions Production Editor: Jessica Mnatzaganian cluster areas: neuroscience and cognition; organismal and evolutionary biology; of scientific progress. We support many top young listed by dialling +44 (0)20 7839 5561, or visit the cell and developmental biology; and health and disease. As well as theme issues, scientists, engineers and technologists, influence the journal publishes papers from the Royal Society’s biological discussion meetings. 6–9 Carlton House Terrace, London SW1Y 5AG, UK Society’s Web site (www.royalsociety.org). For information on submitting a proposal for a theme issue, consult the journal‘s science policy, debate scientific issues with the public rstb.royalsocietypublishing.org website at rstb.royalsocietypublishing.org. and much more. We are an independent, charitable Research Support (UK grants and fellowships) body and derive our authoritative status from over 1400 Research Appointments (Fellowships): 2542 ISBN: 978-0-85403-847-3 Fellows and Foreign Members. Research Grants: 2223 International travel Grants: 2555 Copyright © 2010 The Royal Society During 2010, we are celebrating the Royal Society’s Newton International Fellowships: 2559 Except as otherwise permitted under the Copyright, Designs and Patents Act, 1988, this publication may only be reproduced, stored or 350th anniversary. As part of this, there will be an exciting programme of activities – exhibitions, lectures, transmitted, in any form or by any other means, with the prior permission in writing of the publisher, or in the case of reprographic Science Advice conferences, a new book, a vast science festival on the Science Policy Centre: 2550 reproduction, in accordance with the terms of a licence issued by the Copyright Licensing Agency. In particular, the Society permits the South Bank in London, television and radio making of a single photocopy of an article from this issue (under Sections 29 and 38 of this Act) for an individual for the purposes of broadcasting and much more besides. research or private study. Science Communication Our mission: to expand knowledge and further the role General enquiries: 2573 SUBSCRIPTIONS of science and engineering in making the world a better place. Library and Information Services In 2011 Phil. Trans. R. Soc. B (ISSN 0962-8436) will be Library/archive enquiries: 2606 published twice a month. Full details of subscriptions and Subscription prices All other The Royal Society’s strategic priorities are to: single issue sales may be obtained either by contacting 2011 calendar year Europe USA & Canada countries our journal fulfilment agent, Portland Customer • invest in future scientific leaders and in Electronic access only £2145/€2788 $4058 £2317/US$4153 Services, Commerce Way, Colchester CO2 8HP; tel: innovation, +44 (0)1206 796351; fax: +44 (0)1206 799331; email: Printed version plus £2574/€3345 $4869 £2780/US$4983 • influence policymaking with the best scientific [email protected] or by visiting our website electronic access advice, at http://royalsocietypublishing.org/info/subscriptions. • invigorate science and mathematics education, The Royal Society is a Registered Charity No. 207043. • increase access to the best science internationally, and Typeset in India by Techset Composition Limited, Salisbury, UK. Printed by Latimer Trend, Plymouth. • inspire an interest in the joy, wonder and This paper meets the requirements of ISO 9706:1994(E) and ANSI/NISO Z39.48-1992 (Permanence of Paper) effective with volume 335, issue 1273, 1992. excitement of scientific discovery. Philosophical Transactions of the Royal Society B (ISSN: 0962-8436) is published twice a month for $4058 per year by the Royal Society, and is distributed in the USA by Agent named Air Business, C/O Worldnet Shipping USA Inc., 149-35 177th Street, Jamaica, New York, NY11434, USA. US Postmaster: Send address changes to Philosophical Transactions of the Royal Society B, C/O Air Business Ltd, C/O Worldnet Shipping USA Inc, 149-35 177th Street Jamaica, New York, NY11414. Cover image: Sahelanthropus tchadensis: Toumaï chest, drawing (by S. Riffaut) of the sculpture (by E. Daynes); copyright MPFT. The ﬁrst four million years of human evolution

Papers of a Discussion Meeting held at the Royal Society on 19 and 20 October 2009. Organized and edited by Alan Walker and Chris Stringer

Contents

Introduction The ﬁrst four million years of human evolution 3265 A. Walker and C. Stringer

Articles In search of the last common ancestor: new ﬁndings on wild chimpanzees 3267 W. C. McGrew

More reliable estimates of divergence times in Pan using complete mtDNA sequences 3277 and accounting for population structure A. C. Stone, F. U. Battistuzzi, L. S. Kubatko, G. H. Perry Jr, E. Trudeau, H. Lin and S. Kumar

Spinopelvic pathways to bipedality: why no hominids ever relied on a 3289 bent-hip–bent-knee gait C. O. Lovejoy and M. A. McCollum

Arboreality, terrestriality and bipedalism 3301 R. H. Crompton, W. I. Sellers and S. K. S. Thorpe

Two new Mio-Pliocene Chadian hominids enlighten Charles Darwin’s 1871 prediction 3315 M. Brunet

Phylogeny of early Australopithecus: new fossil evidence from the Woranso-Mille 3323 (central Afar, Ethiopia) Y. Haile-Selassie

Anterior dental evolution in the Australopithecus anamensis–afarensis lineage 3333 C. V. Ward, J. M. Plavcan and F. K. Manthi

Molar microwear textures and the diets of Australopithecus anamensis and 3345 Australopithecus afarensis P. S. Ungar, R. S. Scott, F. E. Grine and M. F. Teaford

An enlarged postcranial sample conﬁrms Australopithecus afarensis dimorphism 3355 was similar to modern humans P. L. Reno, M. A. McCollum, R. S. Meindl and C. O. Lovejoy

The cranial base of Australopithecus afarensis: new insights from the female skull 3365 W. H. Kimbel and Y. Rak

Hominin diversity in the Middle Pliocene of eastern Africa: the maxilla of 3377 KNM-WT 40000 F. Spoor, M. G. Leakey and L. N. Leakey

3263 3264 Contents

Stable isotopes in fossil hominin tooth enamel suggest a fundamental dietary 3389 shift in the Pliocene J. A. Lee-Thorp, M. Sponheimer, B. H. Passey, D. J. de Ruiter and T. E. Cerling

Retrieving chronological age from dental remains of early fossil hominins to 3397 reconstruct human growth in the past M. C. Dean Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Phil. Trans. R. Soc. B (2010) 365, 3265–3266 doi:10.1098/rstb.2010.0179

Introduction The ﬁrst four million years of human evolution

In one of the last paragraphs of The origin of species As discussed in one of the first contributions to this (1859), Darwin famously suggested that ‘Much light volume, molecular estimates of the divergence time will be thrown on the origin of man and his history’. between humans and chimpanzees presently converge When he published The descent of man 12 years on approximately 5–7 Ma, although we two are old later, there was still no fossil evidence of our earliest enough to remember the pre-molecular days, when evolutionary history, and nothing at all from the the supposed uniqueness of humans seemed to require African continent. Yet our close biological relationship a time-span 2–3 times those figures to account for the to the great apes, and especially the African apes, the evolution of special features like bipedalism and high gorilla (Gorilla gorilla) and chimpanzees, had long encephalization. But now, fossils of putative human been recognized, even by scientists who were ignorant lineage members have been reported from approxi- of, or unsympathetic to, evolutionary thinking. mately 6 Ma deposits in Chad and Kenya, and fossils Nevertheless, when we remember his cautious nature of the genus Ardipithecus from approximately 4.4 Ma and the continuing powerful opposition to his ideas, sediments in Ethiopia include about 40 per cent of a it still required fortitude for Darwin to venture ‘It is complete skeleton. Views differ on the relationship therefore probable that Africa was formerly inhabited of these forms to each other, and to the succeeding by extinct apes closely allied to the gorilla and chim- and better-known genus Australopithecus. Several panzee; and as these two species are now man’s skeletons of the latter have been found in the last nearest allies, it is somewhat more probable that our few years. These include an adult from Sterkfontein early progenitors lived on the African continent than cave, South Africa, not yet certainly dated, another elsewhere’. In explaining why the fossil evidence of adult from 3.8 Ma deposits in Woranso-Mille, our origins was slow to appear, he prophetically Ethiopia, a 3.3 Ma child’s skeleton from Dikika, stated ‘Nor should it be forgotten that those regions Ethiopia and four partial skeletons from Malapa which are the most likely to afford remains connecting Cave, South Africa, dated to about 1.9 Ma. Dozens man with some extinct ape-like creature, have not as of other less complete hominin fossils from approxi- yet been searched by geologists’. In fact it was to mately 6 to 2 Ma have been found, as well as these take another 50 years before such fossil evidence skeletons. began to emerge in Africa itself, and Darwin would Our meeting was timed to coincide with the double have been amazed by the remarkable evidence which celebration of Darwin’s 200th birthday and the 150th has accumulated since then concerning the earliest anniversary of the publication of The Origin of Species, stages of human evolution. and to take the first opportunity to bring together as Spectacular discoveries of early members of the much as possible of the rich, newly published data human lineage, including nearly complete skeletons concerning the earliest-known members of the and dozens of other 6 to 2 Ma fossils have been human lineage. Through the generosity of the partici- made in the last 10–20 years. Single complete skel- pants, our hope that detailed images and casts of the etons are much more useful analytically than new material would be brought together for the first separate parts of many individuals, yet until recently, time during the meeting was amply met, although in few had been found from the period before 2 Ma. the event only one of us could be there to see the Even Australopithecus, discovered in South Africa outcome. in 1924, and published and named in 1925, is still The meeting was also planned to showcase the relatively incompletely known. For instance, the interdisciplinary nature of palaeoanthropology, by famous ‘Lucy’ skeleton from Ethiopia is only about highlighting the new methods that have been devel- 20 per cent intact. But new and more complete early oped to extract behavioural and life history hominin skeletons from different parts of the African information from fossils. These included computer continent now promise to give us a much more modelling of locomotor capabilities, finite element complete picture of the early phases in the history of modelling of stresses in bone, laser scanning compari- the human lineage. sons of joint surfaces, quantification of semicircular canal morphology and its relationship to head motion, isotope analysis of teeth for dietary and climate reconstruction, confocal microscopy and texture analysis of tooth wear to indicate diet, and One contribution of 14 to a Discussion Meeting Issue ‘The first four million years of human evolution’. reconstruction of life history parameters from

3265 This journal is # 2010 The Royal Society Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3266 A. Walker & C. Stringer Introduction incremental lines in tooth enamel and dentine. The new material such as the just-published Ardipithecus analytical sessions highlighted what could be accom- skeleton, and the reconstructed Sahelanthropus plished by the careful reconstruction, study and cranium. analysis of the new fossils. In the event, we have not been able to publish all of By concentrating on the early part of the record of the contributions made at the meeting, and this unfor- human evolution, the meeting was also able to docu- tunately included a description of the very complete ment the essential ecological, behavioural, and australopithecine skeleton from Sterkfontein men- morphological stages that underpinned the subsequent tioned earlier. Nevertheless we feel that The first four emergence of the genus Homo. Field workers reported million years of human evolution was an appropriate on studies of the behaviour of wild chimpanzees as measure of how much progress the field of palaeoan- possible models for early hominin behaviour, and on thropology (a term unknown 150 years ago) has the geological and environmental setting of the fossils, made in meeting Charles Darwin’s expectations. We as well as their anatomy and preservation. Context for would like to thank all the staff of the Royal Society the discoveries was provided by colleagues who, for who worked on the planning and running of the meet- example, used tephrostratigraphy, argon–argon radio- ing, and the editorial team who has worked so hard to metric dating, faunal and floral analysis, GIS satellite bring this volume to fruition. imagery and taphonomy. Our hope was to bring about a new understanding of early hominin evolution by bringing together Alan Walker1 the newest fossils and the latest analytical methods, Chris Stringer2,* June 2010 and we think the meeting at least helped progress 1 towards that ambitious goal. But the meeting also Anthropology & Biology, Penn State University, provided the first opportunity to present many of University Park, PA 16802, USA 2 the newest discoveries to scientific and public audi- Dept of Palaeontology, The Natural History Museum, ences alike. A memorable conference dinner was London SW7 5BD, UK accompanied by a display of replicas of spectacular *Author for correspondence ([email protected]).

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Phil. Trans. R. Soc. B (2010) 365, 3267–3276 doi:10.1098/rstb.2010.0067

In search of the last common ancestor: new findings on wild chimpanzees W. C. McGrew* Leverhulme Centre for Human Evolutionary Studies, Department of Biological Anthropology, University of Cambridge, Cambridge CB2 1QH, UK Modelling the behaviour of extinct hominins is essential in order to devise useful hypotheses of our species’ evolutionary origins for testing in the palaeontological and archaeological records. One approach is to model the last common ancestor (LCA) of living apes and humans, based on current ethological and ecological knowledge of our closest living relations. Such referential modelling is based on rigorous, ongoing field studies of the chimpanzee (Pan troglodytes) and the bonobo (Pan paniscus). This paper reviews recent findings from nature, focusing on those with direct implications for hominin evolution, e.g. apes, using elementary technology to access basic resources such as food and water, or sheltering in caves or bathing as thermoregulatory adaptations. I give preference to studies that directly address key issues, such as whether stone artefacts are detectible before the Old- owan, based on the percussive technology of hammer and anvil use by living apes. Detailed comparative studies of chimpanzees living in varied habitats, from rainforest to savannah, reveal that some behavioural patterns are universal (e.g. shelter construction), while others show marked (e.g. extractive foraging) or nuanced (e.g. courtship) cross-populational variation. These findings allow us to distinguish between retained, primitive traits of the LCA versus derived ones in the human lineage. Keywords: tool use; shelter; diet; ranging; last common ancestor; chimpanzee

1. INTRODUCTION the extinct LCA, based on indirect evidence. Thus, This paper aims to synthesize and to update recent this synthesis covers technology, diet, shelter and (from 2005 onwards) findings from studies of the ranging and foraging. ethology and ecology of wild chimpanzees (Pan Attempts to use findings from ethological and eco- troglodytes) that are relevant to modelling human logical (as opposed to morphological) research on origins. Given space constraints, this exercise will be chimpanzees to model the behaviour of ancestral limited to field studies, and therefore mostly to obser- humans are relatively recent, dating from the rise of vational data on the spontaneous behaviour of apes primatological field studies in the last 50 years. in situ, cited selectively. It emphasizes primary reports, Although most early field workers were interested in usually journal articles, on the assumption that older apes in their own right, their mentors often had in secondary reviews (e.g. Mitani et al. 2002; McGrew mind the potential applicability of the exciting new 2004) provide access to earlier material. It concen- findings to human issues (e.g. Goodall & Hamburg trates on the eight study sites with fully habituated 1974). Many of the early attempts now look crude subjects, here listed in the order of seniority: Gombe and simplistic (e.g. McGrew 1981). For example, (Tanzania), Budongo (Uganda), Mahale (Tanzania), most were content to talk about extinct hominids as Kanyawara (Uganda), Bossou (Guinea), Taı¨ (Ivory a single unspecified class, but as the hominin evol- Coast), Ngogo (Uganda) and Fongoli (Senegal). utionary record became more and more diverse, with However, given the geographical bias to eastern and more and more taxa unearthed, this monolithic exer- western Africa, other sites with partly habituated sub- cise was less and less satisfactory. Furthermore, as jects, especially in central Africa, such as Goualougo data began to emerge on wild bonobos, Pan paniscus (Republic of Congo), are necessarily invoked too. (Kano 1992), who are as equally closely related as Most importantly, it focuses on topics that are relevant chimpanzees to hominins, and as cross-populational to modelling the behaviour of the last common ances- variation began to emerge in chimpanzees (McGrew tor (LCA) of the divergent lines that led to living 1992), easy generalizations grew harder to make. Eco- humans and living chimpanzees. These topics are pre- logical studies of chimpanzees in a variety of ecotypes, sented in terms of their ‘directness’ in comparisons from rainforest to savannah, forced more precise mod- between what primatologists see now in living apes, elling (Moore 1996). Finally, debate over the best way and what palaeoanthropologists seek to infer about to model human origins and evolution, that is, via referential versus strategic models, or by homology versus analogy, muddied the waters (e.g. Tooby & *[email protected] DeVore 1987). Sayers & Lovejoy (2008) took the One contribution of 14 to a Discussion Meeting Issue ‘The first four extreme position that chimpanzees may be no more million years of human evolution’. useful as models than other, more ecologically,

3267 This journal is q 2010 The Royal Society Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3268 W. C. McGrew Chimpanzees and the last common ancestor zoogeographically and phylogenetically distant taxa, innovation in wild chimpanzees, at Mahale, shows such as capuchin monkeys (Cebus spp.). More recently, inventiveness to be common, but the chance that a Lovejoy (2009) has asserted that extant African ape- novel behavioural pattern will be propagated and based models are no longer appropriate (for a contrary become established in a population is rare (Nishida view, see Whiten et al. 2010). However, for focused et al. 2009). studies, such as of Oldowan lithic industries, apes No longer is it enough just to list the types of tool may still be the model of choice (Toth & Schick 2009). found at a given site, as nominal (presence/absence) This paper takes the conservative line that lacking data. Now attention to relative frequency and compe- a fossil record for apes since the Miocene (cf. tence of performance across age and sex classes is McBrearty & Jablonski 2005), and having only a expected, along with data on context, variation in shallow archaeological record for apes, all that we form and function of the tools’ manufacture and use sensibly can hope to model is the LCA. In doing so, (e.g. Sanz et al. 2009a). Functional (e.g. extractive I make several simplifying assumptions, such as that foraging), biomechanical (e.g. percussive) and cogni- anything that a chimpanzee can do today, the LCA tive (e.g. artefact complexity) aspects of technology could have done 6–7 Myr ago. Another pragmatic are stressed. Anecdotal versus idiosyncratic versus assumption is that although the LCA could have habitual use of tools is differentiated. Distinction is resembled the living chimpanzee, or bonobo, or drawn between a tool kit (i.e. the whole repertoire of neither, or some combination of the two, most of a community’s collective range of tools) and a tool what we have to work with on grounds of homology set (i.e. the obligate sequence of two or more tools comes from P. troglodytes. Therefore, until comparable used to achieve a single goal). Composite tools (i.e. breadth and depth of data are available for P. paniscus, when two or more objects are used simultaneously the chimpanzee must carry the load. and complementarily to achieve a goal), such as hammer and anvil (Carvalho et al. 2009), are distinguished from compound tools (i.e. when two or (a) Technology more elements of different types are combined into a Most of what behavioural primatologists have to offer single unit), such as a wedge used to level an anvil’s to palaeoanthropology relies on artefacts, as these working surface (Biro et al. in press). Typology is objects are comparable to what is found in the archae- now part of chimpanzee technology. ological record. However, artefacts are the products of Tool kits show both uniformity and variety across behaviour, and sometimes archaeological data are a populations. Sanz & Morgan (2007) presented quanti- further step removed: butchery cutmarks on bones tative and qualitative findings from the Goualougo, are the products of the ephemeral acts that produced Republic of Congo, chimpanzees, whose tool kit num- them. Whatever the caveats, primatologists can offer bers 22 types, of which nine are used habitually something that no archaeologist will ever see, that is, (customary). In contrast, Watts (2008b) published BOTH the product AND the behaviour, directly comparable data from Ngogo, Uganda, where the recorded. When a glancing blow of a stone hammer total tool kit numbers only 10 types, with four of being used to crack a nut hits instead the stone anvil, these being habitual. Such variation suggests the possi- producing a conchoidally fractured flake, the observer bility of a normally distributed spectrum, but this is can see whether this was an accident. An archaeologist not the case. As with Goualougo, all habituated popu- given only that same single flake could draw no valid lations show about the same-sized tool kits: Gombe inference about the percussionist’s intentions. (22), Bossou (21), Taı¨ (21) and Mahale (16). How- Studies of tool use by apes in nature have come ever, along with Ngogo, the other Ugandan sites a long way from piecemeal natural history notes show small tool kits: Budongo (8) and Kanyawara collected opportunistically and descriptively, to com- (10) (Sanz & Morgan 2007, table 3). Even more strik- prehensive, systematic, hypothesis-driven empirical ing is the contrast between Goualougo and Ngogo efforts, some of which are experimental. Comparative with regard to the predominate types of tools: the analyses of chimpanzee material culture are done at top three at Goualougo are used in subsistence, that every level, of individuals, lineages, communities, is, extractive foraging of termites, honey and water; populations, subspecies and species (McGrew 2004). the top three at Ngogo are used in hygiene, especially The chimpanzee ethnographic record now spans so wiping the penis after copulation, and in courtship. many study sites across equatorial Africa that even (The reverse is equally true: Goualougo chimpanzees chimpologists have trouble keeping them straight. very rarely use napkins, and Ngogo chimpanzees Although only eight sites consistently allow all-day, rarely harvest insects.) However, some types of tool close-up observation, there are five times as many use are chimpanzee universals, being found in all other sites with varying degrees of habituation. In the long-studied populations across Africa, such as leaf 5 years, long-term sites studying the central (Hicks sponge (drinking water), aimed throw (weapon), play et al. 2005; Sanz & Morgan 2007) and Nigerian start (toy), branch drag (display), etc. (Fowler & Sommer 2007) subspecies have joined the Of particular importance is percussive technology, longer term studies in eastern and western Africa. that is, the application of ballistic force via one Even sites that have yet to habituate their subjects object to another to achieve a goal (Ling et al. 2009). have yielded new behavioural patterns, e.g. root- In chimpanzees, this most famously takes the form of digging at Ugalla, Tanzania (Hernandez-Aguilar hammer and anvil used to crack nuts, but it also et al. 2007), fruit-cleaving at Nimba, Guinea (Koops occurs in smashing hard-shelled objects directly et al. 2010), etc. The only comprehensive study of against anvils, in agonistic clubbing of adversaries or

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Chimpanzees and the last common ancestor W. C. McGrew 3269 in display, or in specialized extractive foraging such as of compound tool (Sousa et al. 2009), but it barely pestle-pounding (Yamakoshi & Sugiyama 1995). In qualifies, being iterative. The most obvious example the latter case, the pestle is a detached palm frond, of compound technology (albeit not tool use) in the mortar is the apical growth tip of an oil palm non-human primates in nature is the sleeping plat- (Elaeis guineensis) and the result is a cavity full of forms/nests/beds that are woven daily by great apes mashed-up slurry, which is eaten. Anvil use without (see below). The best-known example in the extractive hammers occurs when a hand-held, hard-shelled foraging of chimpanzees is the anvil–wedge, known fruit is bashed directly against a boulder or root, as only from the nut-cracking of the Bossou chimpanzees with baobabs (Adansonia digitata). Marchant & (Matsuzawa 2006). Bossou’s stone anvils are movable, McGrew (2004) hypothesized an evolutionary scen- and so their positioning can be adjusted; anvils with ario that led from anvil use to stone-knapping. near-horizontal working surfaces are the most efficient, Tool sets in apes were first recognized in honey as the yielded nut-meat is readily picked up. An angu- extraction (Brewer & McGrew 1990). In seeking to lar anvil can be levelled by inserting a smaller stone harvest nature’s most calorific food, the minimal tool as a wedge underneath, to make the working surface set requires a tool to break into the bees’ storage reser- less tilted. voir and another tool to extract the liquid. That is, To what extent is the technological repertoire of the some kind of percussive tool, such as hammer or chimpanzee now known? The steepness of the cumu- chisel, plus some kind of dip-stick, are needed to lative ethnographic curve may be less than in the last secure the food item (for the most complete treatment century, but it has not flattened out. New habitual pat- of this resource’s exploitation, see Sanz & Morgan terns continue to be described: chimpanzees use spears 2009). Tool sets may be more complex: Boesch et al. to skewer small mammals (Pruetz & Bertolani 2007) (2009) recently described tool sets used by the chim- and digging sticks to unearth roots (Hernandez-Aguilar panzees of Loango, Gabon, in which up to five tools et al.2007). Furthermore, new modes of tool use were needed, e.g. pounder, perforator, enlarger, col- continue to emerge, such as the chimpanzees of lector and swab. Tool sets also are used to exploit Nimba, Guinea, using cleavers to break apart large, other animal prey, e.g. termites (Deblauwe et al. fibrous Treculia fruits (Koops et al.2010). 2006; Sanz & Morgan 2007), ants (Sanz et al. Much progress has also been made on how individ- 2009b) and even when getting honey, the protein- ual apes in nature learn to use elementary technology. aceous bonus of bee brood may be important too. Previous studies were descriptive or qualitative, The key point about a tool set is that it is sequential whereas modern ones use sophisticated multivariate task: if an A–B–C–D is necessary, then A–C–B–D analyses (e.g. general linear mixed model) to tease will not do; you cannot check the oil level in your out the influences of independent variables. car’s engine via the dip-stick, until you have opened Lonsdorf ’s (2006) study of termite fishing at Gombe the car’s bonnet. Although tool sets may suggest showed that although all chimpanzees in the Kasakela advanced cognitive abilities, many such mandatory community show this tool use by 5.5 years of age, sequences are shown by creatures with modest brains daughters acquire the skills earlier, and this acquisition (Hansell 2004), especially in shelter construction is a function of the mother’s overall time spent in the (see below). What is impressive (and possibly activity. Humle et al. (2009) showed that chimpanzee unique) about chimpanzee tool sets is that alternative infants at Bossou who had more opportunities to versions may be used flexibly by different apes to observe their mothers started ant-dipping sooner and solve the same problem. were more proficient than their low-opportunity In human elementary technology, composite tools counterparts. However, in neither case were individual are well known: Mortar and pestle, bow and arrow, differences in mother’s performance reflected in individ- etc. Each element may stand alone, but is almost use- ual differences in their offspring, nor was there any less without its partner. (Tool composites differ from direct teaching by mothers. Youngsters learned to fish tool sets in that they are used simultaneously, rather or to dip by passive observational learning of tolerant than sequentially.) Tool composites are known for models. Matsuzawa et al. (2001) have termed this apes (see summary in Sugiyama 1997), and some are dyadic conduit of information from one ape to another widespread, for example, in all populations where as ‘education by master–apprenticeship’. chimpanzees use long wands to dip for driver ants, Some primatologists now apply archaeological they also use bent-over saplings as a perch while methods to the study of chimpanzee technology in doing so, to avoid the painful bites of the ants swarm- nature. Mercader et al.(2002, 2007) have shown that ing on the ground below (McGrew 1974). However, the past nut-cracking activities of the Taı¨chimpanzees only recently have tool composites been systematically leave behind a record of stone artefacts. These can be studied: Carvalho et al. (2009) showed that certain distinguished from human artefacts or naturally splin- combinations of stone hammers and anvils were used tered rocks by ‘blind’ assessors, dated by standard over and over again by the chimpanzees of Bossou, radiometric techniques (C14), and yield organic resi- even taking into account the apes’ separate, indepen- dues (starch grains) that reveal their function. We dent preferences for hammer or anvil. can now speak of a chimpanzee ‘stone age’ with time Compound tools are harder to find in living apes in depth. Carvalho et al. (2008) applied one of the core nature, although their production is readily induced concepts of archaeology, the chaine operatoire, to the under contrived captive conditions. Combination of nut-cracking of Bossou’s chimpanzees, showing that multiple items of the same type, e.g. leaves compressed from start to finish, this analytical technique is equally together in leaf-sponging for water, is the simplest kind applicable to apes as to humans. Even retrospective

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3270 W. C. McGrew Chimpanzees and the last common ancestor analyses of chimpanzee artefacts, in this case the but one of the keenest hunting populations, the chim- brush-sticks used to fish for termites, as found in a panzees of Taı¨, does not hunt the small forest museum, may explain how they were made (Heaton & antelopes (Cephalophus spp.) that are plentiful there. Pickering 2006;cf.Sanz & Morgan 2007). The At the same time, several populations of bonobos extent to which the archaeology of non-humans can avidly hunt antelope, but were said to show no interest be pushed back in time remains to be seen, but a in primate prey; this apparent species difference evap- new field is now underway (Haslam et al. 2009). orated with Surbeck & Hohmann’s (2008) report that Finally, here is a sobering thought: of all the tools the bonobos of Lui Kotale, Democratic Republic of named so far, only some of the hammers and anvils, Congo, also hunt guenons (Cercopithecus spp.). What and some of the missiles thrown, will have a chance differs between the two sibling species of chimpanzee of persisting in the archaeological record, taphonomy and bonobo is the sexual politics of meat-sharing. willing, because they are made of stone. All of the In bonobos, females control the carcass and others are made of organic raw materials, e.g. plant distribute the meat, and their collective dominance or animal matter, and so will perish over time. There over males sometimes leaves the males with none, are other lithic objects used, e.g. stones in self-tickle, even if individually a male can dominate a female pebbles in play start, boulders in splash display, etc., (Hohmann & Fruth 2008). In chimpanzees, males but it is unlikely that these will be archaeologically often control carcasses, and there has been much recognizable. debate about how the sharing of the meat functions in chimpanzee society. Now come solid data to test Stanford’s (1999) hypothesis of meat-for-sex, that is, (b) Diet that males selectively give meat to females in exchange The chimpanzee is an omnivore, as all well-studied for sexual favours. Gomes & Boesch (2009) report that populations show a mix of herbivory and faunivory. females copulate more often with males who share The former is dominated by ripe fruit, but also meat with them in the long term. Thus, the female includes leaves, pith, seeds, flowers, bark, gum, etc. need not be in oestrus at the time of the hunt, but The latter focuses on social insects (ants, bees, rather forms a relationship that mutually enhances termites) and small- to medium-sized mammals, the lifetime reproductive success of male (insemination especially monkeys. Invertebrates usually are taken by probability) and female (nutritional enhancement). tool-assisted extractive foraging, such as ant-dipping, However, meat-sharing in some chimpanzee ant-fishing, honey-dipping and termite-fishing, that populations, e.g. Gombe in Tanzania (Gilby 2006), is, by gathering. Until recently, vertebrate prey were appears to be driven by different mechanisms: known to be captured and dispatched only by hand, intimidation, harassment, reciprocity, etc. Less likely without technology. Pruetz & Bertolani (2007) is Tennie et al.’s (2009) ‘meat-scrap’ hypothesis that showed that the chimpanzees of Fongoli, Senegal, use meat-sharing can be explained by the micro-nutrients a weapon-assisted hunting technique to disable or kill found in even small amounts of meat. Meat-eating is bushbabies while they sleep during the day in tree only one kind of faunivory, and the same nutrients holes. The weapon is a sharpened stick (spear), can be easily obtained from invertebrates, which jammed into the prosimian’s sleeping chamber. (Some chimpanzees eat daily. sceptics have questioned whether the technique quali- Male sharing of prized foodstuffs with females also fies as hunting, or the instrument as a spear. When an occurs with plant foods, which otherwise is rare in Inuit waits beside a seal’s air-hole in the ice, then thrusts apes, usually occurring only between mother and a sharp-ended linear object into it, skewering the prey, infant. However, Hockings et al. (2007) showed that we are happy to call it hunting, so why not for apes?). when males at Bossou raided crops, especially Notably absent from the diets of most chimpanzee papaya (Carica papaya), they almost always shared populations are the underground storage organs the proceeds with females of reproductive age, even (USO) of plants, that is, bulbs, roots, tubers, corms, when the latter were not in oestrus. These sharing pat- rhizomes, etc. This absence was thought to reflect the terns reflected patterns of later sexual consortship. generalized, non-digging hands of primates, plus the What about scavenging? Scattered, anecdotal reports apes’ lack of the appropriate technology, that is, the dig- of chimpanzee scavenging mammalian prey have ging stick. Hernandez-Aguilar et al.(2007) recently appeared from time to time, but no systematic study described how the chimpanzees of Ugalla, Tanzania, was done until Watts (2008a ) documented all known dig up roots, using sticks and pieces of bark that show scavenging opportunities at Ngogo over 11 years of the abraded wear patterns of repeatedly used digging observations totalling over 10 000 h. In that period, he tools. Spat-out wadges of fibrous roots show them to saw only four scavenging events, and opportunities be chewed and sucked, then discarded. A similar pro- were rare, occurring on average only every 100 h. This cessing technique is used by the chimpanzees of contrasts mightily with over 650 kills made in over 270 Tongo, Democratic Republic of Congo, to get drinking hunts in the same period (Watts & Mitani 2002). Similar water from subterranean tubers, but these are dug up pictures of rarity emerge from Gombe, Mahale and Taı¨. by hand from friable, volcanic soils (Lanjouw 2002). Chimpanzees are not scavengers, it seems. Across the continent, from Tanzania to Ivory Coast, chimpanzee hunters take more monkeys as prey than all other types of vertebrates combined, especially (c) Shelter favouring the red colobus (Piliocolobus badius) (e.g. Shelter can be defined as the use of any material object Watts & Mitani 2002). Others also hunt ungulates, to buffer the effects of the elements. A universal

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Chimpanzees and the last common ancestor W. C. McGrew 3271 behavioural pattern among great apes is their daily as the typical ecotype for wild chimpanzees, and so construction of arboreal sleeping platforms: every contrast their ecological context with that of hominins weaned individual builds an overnight nest and many who lived in more seasonal, mosaic habitats, this also build day nests for napping. These compound restrictive picture is less and less tenable. Most of the artefacts are scattered over the landscape and may study sites at which chimpanzees have been studied, endure for months, leaving a record of points in and at least (depending on definition) three (Fongoli, space where chimpanzees spend most of their lives. Gombe and Mahale) of the eight where the apes (Chimpanzees typically retire at dusk and arise at have been fully habituated to close-range observation, dawn, and so spend half of each tropical circadian are not evergreen rainforest. More accurately, chim- cycle in their beds.) Hernandez-Aguilar (2009) panzees subsist in a range of ecotypes, from found 5354 nests over a 20 month period at Issa, an woodland savannah (not steppe) to rainforest, with open-country, savannah area in western Tanzania. mean annual rainfall that range from about 800 These shelters were highly clumped on woodland hill- to more than 2000 mm per year. Many of these sides, in particular sites that were re-used over and landscapes are vegetationally heterogenous, and chim- over again. The chimpanzees’ ranging and consequent panzee use of this array of habitat types varies greatly. nest distribution varied predictably over wet and dry At the other extreme, chimpanzees (unlike baboons, seasons, reflecting an annual cycle of movement that Papio spp.) do not survive in places that lack surface reflects availability of surface water and ripe fruit. water for drinking or that lack the riverine forests However prominent a part these shelters play in their that follow these watercourses, although only a tiny daily lives, these constructions later will be archaeolo- fraction of such gallery forest will suffice. Copeland’s gically invisible, being made entirely of woody (2007, 2009) detailed comparison of several open vegetation. and arid African habitats shows that landscapes with At the same time, studies of individual nests and annual rainfall in the 500–750 mm range cannot sup- their making have yielded insights: port chimpanzees. Early hominins apparently relied on Koops et al. (2007) showed that at Nimba, surpris- eating C4 plants and USOs, both of which have yet to ingly many nests were built on the ground. From be shown to be important in the diets of chimpanzees, the patterning and size of nests, they hypothesized despite recent prominent findings (Hernandez-Aguilar that this reflected a pattern of male overnight mate- et al. 2007). When drinking water runs short, that is, guarding, that is, when an oestrous female nested in during the dry season when water table drops below a tree, a male seemed to nest on the ground at its the surface, chimpanzees turn to digging wells when base, to sequester her from the nocturnal attention riverbeds are sandy enough to allow this (Hunt & of other males. Various functions for nests have been McGrew 2002). Although the wells are dug by hand, proposed: anti-predator, anti-parasite, anti-disease leaf sponges are used to extract water from the wells; vector, thermoregulation, etc., but there has yet it would not be surprising to find digging tools used been no comprehensive study of these hypotheses. to dig wells in other substrates, e.g. mud, gravel, etc. Meanwhile, Stewart et al. (2007) studied the proximal On a day-to-day basis, chimpanzees must find characteristics of nests, in terms of their architecture ephemeral food. Frugivores in particular must find and materials. First-hand empirical data showed that and monitor clumps of food that should be eaten at chimpanzees prefer comfortable nests, presumably to peak ripeness and which varies from year to year in gain restorative sleep for their big brains. availability. The same grove that yielded a bumper The species’ name for the chimpanzee implies a crop last year may not fruit at all this year. The biodi- cave dweller, yet until recently, there was no record verse array of trees, shrubs and lianas, much less non- of chimpanzees using caves as shelter. Pruetz (2007) woody plants, may present a potential cornucopia of reported that the Fongoli chimpanzees, who occupy food, but the daily challenge is how to be in the right one of the hottest and driest areas in the species’ distri- place at the right time. Various hypotheses have been bution, regularly use a cave during the hottest season put forward as to how chimpanzees achieve this, but of the year. They retreat to its cooler environment the strategy turns out to be simple: during the heat of the day for ‘siestas’ and picnics; Normand et al. (2009) showed that chimpanzees in overnight, they sleep in arboreal nests, just like other the Taı¨Forest memorize the locations of thousands of great ape populations. individual trees. Modelling of the apes’ powerful Chimpanzees are notoriously hydrophobic, as they spatial memory allows for their ‘rules’ of foraging to do not swim, which makes watercourses notable bar- be inferred, e.g. travel longer distances to resources riers to their geographical distribution. However, they that allow longer feeding bouts, revisit more often enter surface water in certain circumstances: at Fongoli, sources where you last ate for long periods. they immerse themselves in temporary rain-filled pools But how to acquire such information? Murray et al. at the beginning of the rainy season, when it is still hot (2008) showed at Gombe that even in adulthood and and humid; there they rest, groom and play (Pruetz & long after their mothers have died, males return to Bertolani 2009). Thus, water becomes a thermoregula- the core ranges used by their mothers, especially in tory device, even when potentially risky. lean times. Resource locations learned during dependent infancy are harvested lifelong. It is all very well to know what resources are in the (d) Ranging home range, but how to know where they are, that is, Although some authors (e.g. Lovejoy 2009) stub- how to navigate optimally between them? Again, var- bornly continue to characterize evergreen rainforest ious hypotheses have been proposed, e.g. spatial

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3272 W. C. McGrew Chimpanzees and the last common ancestor orientation by means of landmarks. Normand & become hammers, as they are modified by use Boesch (2009) show from data on travel directions (Carvalho et al. 2009). That is, tools may change and distances that Taı¨chimpanzees have sophisticated functional categories. (Studies of refitting may mental maps, that is, cognitive two-dimensional help to distinguish reduction products from tool representations of the landscape that allow them to sets; e.g. Delagnes & Roche 2005). Moreover, travel from resource to resource in straight lines. application of knowledge from ape tool sets may help make sense of patterned heterogeneity in archaeological assemblages, as revealed by multi- 2. DISCUSSION variate statistical analyses. What can now be said about the LCA, based on what — Composite tools probably were used by the LCA, has been learned over the past 5 years from field but the challenge is to recognize such combinations studies of wild chimpanzees? in recovered lithic assemblages. It is not always Technology is the obvious starting point: clear what was the goal of reduction sequences in knapping, such as core or flake. The best candidate — Given the large and varied tool kits of the chimpan- still may be pounding technology, as it seems likely zee, we can expect that of the LCA to be similar. that flaked stone did not spring de novo with the That is, tools were made and used not just for Oldowan, but more probably evolved from earlier food acquisition and processing, but also in self- lithic percussion for other reasons. Perhaps the maintenance and shelter, as well as in social and analogues to chimpanzee hammers and anvils are sexual life (not covered here). However, just as there to be found in deposits older than 2.6 Ma? the size of tool repertoire in chimpanzees is a func- Primatologists should be able to help in seeking tion of research effort, so it will be in recovering the the pre-Oldowan (Haslam et al. 2009), based on material culture of the LCA. reliably recognized modifications from chimpanzee — Most of the presumed technology of the LCA is hammers and anvils. This may help to clarify per- archaeologically unrecoverable, given its perish- sisting confusion and controversy (e.g. Mora & de able, organic nature; thus the archaeological la Torre (2005) versus Diez-Martin et al. (2009)) record is biased towards lithics. Short of a time among archaeologists. machine, this problem is insoluble, but aspects of — Apart from their nest-building, chimpanzees have chimpanzee behaviour that are universal, such as few compound artefacts. In the evolution of bed-making or leaf-sponging, are hard to deny to human elementary technology, much is made of the LCA. the first evidence of hafted weapons, that is, a com- — As with chimpanzees, the material culture of the pound tool of shaft, point and fixative. However, LCA will show inter- and intra-regional differences arguably, the earliest known compound technology (e.g. Schoening et al. 2008). Just as nut-cracking was necklaces of snail shells, as found in Blombos differs between East and West Africa (Morgan & Cave, South Africa (Henshilwood et al. 2004). Abwe 2006), despite the common presence of Whether or not the LCA had compound tools is both prey and raw materials (McGrew et al. unclear, especially as not all components survive 1997), so it is for the LCA. Similarly, just as extrac- equally well, e.g. the spear’s shaft versus its point, tive foraging for social insects is central to the necklace’s string versus its shells. Tanzanian populations of chimpanzees, but is lar- — Studies of the acquisition and development of gely absent in the neighbouring country of chimpanzee technology remind us that some pro- Uganda, so we should not be surprised to find portion of what is found archaeologically is such differences in e.g. Kenyan and Ethiopian probably the immature version of the polished populations of a species of hominin. adult form of material culture. How much debitage — Subsistence technology in chimpanzees involves reflects ‘honest’ mistakes by youthful learners reuse of artefacts, whether these are nut-cracking versus clumsy or misguided efforts by adults? hammers or ant-dipping wands. Especially given This problem probably applies as well to the that the extent of reuse seems to be a function LCA. Actualistic studies of children of various of availability of raw materials (and some African ages learning to knap stone might be useful. forests afford no surface stones bigger than a — Finally, we must repeatedly remind ourselves that walnut, e.g. Lui Kotale, W. C. McGrew & the LCA was almost certainly not a chimpanzee, L. F. Marchant 2006, unpublished data), the and vice versa. Just as living apes continue to same is expected of the LCA. Just as at Bossou, reveal new kinds of technology, so should we reuse of stone tools may increase the probability expect the same from the LCA. If chimpanzees of predictable fracture or amplified use–wear that turn out not to use tools to make other tools, or would leave archaeological signatures in the result- lack important but basic material cultural items ing artefacts. Lack of data on curation of tools by like the container, or do not transport objects apes in nature may reflect lack of precise study, as over long distances, we may have found important evidence exists of such premeditated storage in hominin watersheds (cf. Wynn & McGrew 1989). captivity (Osvath 2009). — Given tool sets in chimpanzees, we should expect Regarding diet: the same in the LCA. But how to recognize sequential use from a static assemblage? This is — Chimpanzee opportunistic omnivory is clear, and further complicated by findings that anvils may so it is probably in the LCA. The same inference

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Chimpanzees and the last common ancestor W. C. McGrew 3273

derives from increasing evidence of dietary overlap most chimpanzee field sites do not offer caves, (e.g. monkey-hunting) between chimpanzee and although this has never been systematically studied. bonobo, although important differences remain — We now know that chimpanzee nests are more between these taxa (e.g. extractive foraging for complex structures than hitherto realized, and insects). this may imply that beyond a certain point of — Recent findings of chimpanzee use of USOs para- investment of time and effort, they began to be doxically show apes to be capable of harvesting reused. This raises the possibility of home bases, these foodstuffs, yet in no known population are already hinted at in the non-random distribution they a staple (cf. Hockings et al. 2009, for data of chimpanzee nest sites on the landscape. But on USOs as fallback foods). Experimental studies until we know the fitness-enhancing function of need to be done on the limits of chimpanzee- beds, it would be rash to infer the same for the digging. Similarly, chimpanzees commonly LCA. Anti-predation is assumed, but equally consume the pith of C4 plants, yet not the seeds attractive alternative hypotheses are there to be or corms, and so their stable isotope data are con- tested. The presence of ground nests is sometimes fusing (Sponheimer et al. 2006). (It seems likely presumed to be based on local release from preda- that staple exploitation of cereals requires grinding tion, but no correlative study of sympatric large technology, which seems to be absent in wild chim- carnivores and apes has been done. panzees, but apparently has not been tested with apes in captivity.) Or, it may be that profitable On ranging and foraging: use of USOs and cereals requires treatment by fire, that is, cooking, which came much later — Chimpanzees are nomadic over areas that can be in human evolution (Carmody & Wrangham large, that is, tens or even hundreds of square kilo- 2009). Here, studies of wild chimpanzees are not metres. If the singlemost obvious influence on this yet helpful in hypothesizing about the LCA. ranging is food availability, the more crucial limit- — Chimpanzees are wide-ranging foragers, and their ing factors may be drinking water and cover. patterns of ranging map onto the distribution of Well-digging, especially with the technological their resources, as in any other organism. What assistance of digging tools and containers, appears we now are beginning to know is the extent of to allow an expanded ecological niche. (Unlike their intelligent foraging, and it exceeds our expec- temperature or humidity, which turn out not to tations, e.g. about spatial memory. This upgrades be so important.) Similarly, no matter how dry our estimation of the LCA, but inferring the and open the eco-type inhabited, every known timing and spacing of resources in the archaeologi- population of great apes seems to require access cal record is problematic. to trees for shelter construction. Even savannah- — Recent findings on chimpanzee hunting confirm its dwelling chimpanzees need their ribbons of gallery seductiveness for evolutionary scenarios. (Conver- forest. The same was probably true of the LCA. sely, scavenging’s role seems less and less In conclusion, even if one-tenth of what has been important, at least until after the LCA, in the learned in the last five years about wild chimpanzees hominin lineage.) However, estimations of the is applicable to the LCA of living apes and humans, importance of hunting, based on chimpanzees, then the case has been made for preserving them. must be tempered: Most chimpanzee hunting is Referential modelling requires living proxies upon done arboreally, by ‘four-handed’ hunters who which to base the models, and current expectations can leap about in the canopy, pursuing monkeys. are that wild populations of great apes may be gone This is not likely to be instructive about hunting by the middle of the current century. Both primatolo- by terrestrial bipeds, even if it applies to the gists and palaeoanthropologists should work together LCA, who may have practised ambush hunting to save them. on the ground, as well as pursuit hunting in the treetops. More significantly, the function of carniv- Most of the author’s data and ideas were supported by the ory is revealed to be much richer than expected: US National Science Foundation, Researching Hominid sharing meat may drive social and sexual life, Origins Initiative (Award no. BCS-0321893) grant awarded almost as a currency (although many of the same to Tim White and to the late Clark Howell. I am grateful to S. Carvalho, L. F. Marchant, P. Mellars and A. Walker arguments probably apply also to honey). for comments on the manuscript.

On shelter: REFERENCES — Based on the near-uniformity of arboreal overnight Biro, D., Carvalho, S. & Matsuzawa, T. In press. Tools, sleeping off the ground in great apes, it seems likely traditions, and technologies: interdisciplinary approaches that the LCA did the same. It may be that safe ter- to chimpanzee nut-cracking. In The mind of the chimpanzee: ecological and experimental perspectives (eds restrial sleeping came much later, with the E. V. Lonsdorf, S. R. Ross & T. Matsuzawa). Chicago, domestication of ﬁre (Pruetz & LaDuke 2010). IL: University of Chicago Press. But we now know that cave use, at least during Boesch, C., Head, J. & Robbins, M. M. 2009 Complex tool the day, did not depend on ﬁre, and that thermore- sets for honey extraction among chimpanzees in Loango gulation needs could have been for diurnal cooling, National Park, Gabon. J. Hum. Evol. 56, 560–569. rather than nocturnal heat retention. However, (doi:10.1016/j.jhevol.2009.04.001)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3274 W. C. McGrew Chimpanzees and the last common ancestor

Brewer, S. M. & McGrew, W. C. 1990 Chimpanzee use of a underground storage organs of plants. Proc. Natl Acad. tool-set to get honey. Folia Primatol. 54, 100–104. Sci. USA 104, 19 210–19 213. (doi:10.1073/pnas. (doi:10.1159/000156429) 0707929104) Carmody, R. & Wrangham, R. W. 2009 The energetic Hicks, T. C., Fouts, R. S. & Fouts, D. H. 2005 Chimpanzee significance of cooking. J. Hum. Evol. 57, 379–391. (Pan troglodytes troglodytes) tool use in the Ngotto (doi:10.1016/j.jhevol.2009.02.011) Forest, Central African Republic. Am. J. Primatol. 65, Carvalho, S., Cunha, E., Sousa, C. & Matsuzawa, T. 221–237. (doi:10.1002/ajp.20111) 2008 Chaines operatoires and resource-exploitation Hockings, K. J., Humle, T., Anderson, J. R., Biro, D., Sousa, strategies in chimpanzee (Pan troglodytes) nut-cracking. C., Ohashi, G. & Matsuzawa, T. 2007 Chimpanzees share J. Hum. Evol. 55, 148–163. (doi:10.1016/j.jhevol.2008. forbidden fruit. PLoS ONE 2, e886. (doi:10.1371/jour- 02.005) nal.pone.0000886) Carvalho, S., Biro, D., McGrew, W. C. & Matsuzawa, T. Hockings, K. J., Anderson, J. R. & Matsuzawa, T. 2009 Flex- 2009 Tool-composite reuse in wild chimpanzees (Pan ible feeding on cultivated underground storage organs by troglodytes): archaeologically invisible steps in the technologi- rainforest-dwelling chimpanzees at Bossou, West Africa. cal evolution of early hominins? Anim. Cogn. 12(Suppl. 1), J. Hum. Evol. 58, 227–233 (doi:10.1016/j.jhevol.2009. 103–114. (doi:10.1007/s10071-009-0271-7) 11.004) Copeland, S. 2007 Vegetation and plant reconstruction of Hohmann, G. & Fruth, B. 2008 New records on prey cap- lowermost Bed II, Oldovai Gorge, using modern analogs. ture and meat eating by bonobos at Lui Kotale, Salonga J. Hum. Evol. 53, 146–175. (doi:10.1016/j.jhevol.2007. National Park, Democratic Republic of Congo. Folia 03.002) Primatol. 79, 103–110. (doi:10.1159/000110679) Copeland, S. 2009 Potential hominin plants foods in Humle, T., Snowdon, C. T. & Matsuzawa, T. 2009 Social northern Tanzania: semi-arid savannas versus savanna influences on ant-dipping acquisition in the wild chimpanzee sites. J. Hum. Evol. 57, 365–378. (doi:10. chimpanzees (Pan troglodytes verus) of Bossou, Guinea, 1016/j.jhevol.2009.06.007) West Africa. Anim. Cogn. 12, 37–48. (doi:10.1007/ Deblauwe, I., Guislain, P., Dupain, J. & van Elsacker, L. s10071-009-0272-6) 2006 Use of a tool-set by Pan troglodytes troglodytes to Hunt, K. D. & McGrew, W. C. 2002 Chimpanzees in the obtain termites (Macrotermes) in the periphery of the dry areas of Assirik, Senegal, and Semliki Wildlife Dja Biosphere Reserve, southeast Cameroon. Am. J. Reserve, Uganda. In Behavioural diversity in chimpanzees Primatol. 68, 1191–1196. (doi:10.1002/ajp.20318) and bonobos (eds C. Boesch, G. Hohmann & L. F. Delagnes, A. & Roche, H. 2005 Late Pliocene hominid Marchant), pp. 35–51. Cambridge, UK: Cambridge knapping skills: the case of Lokalalei 2C, West Turkana, University Press. Kenya. J. Hum. Evol. 48, 435–472. (doi:10.1016/j. Kano, T. 1992 The last ape: pygmy chimpanzee behavior and jhevol.2004.12.005) ecology. Stanford, CA: Stanford University Press. Diez-Martin, F., Sanchez, P., Dominguez-Rodrigo, M., Koops, K., Humle, T., Sterck, E. H. M. & Matsuzawa, T. Mabulla, A. & Barba, R. 2009 Were Olduvai hominins 2007 Ground-nesting by the chimpanzees of the Nimba making butchering tools or battering tools? Analysis of a Mountains, Guinea: Environmentally or socially recently excavated lithic assemblage from BK (Bed II, determined? Am. J. Primatol. 69, 407–419. (doi:10. Olduvai Gorge, Tanzania). J. Anthropol. Archaeol. 28, 1002/ajp.20358) 274–289. (doi:10.1016/j.jaa.2009.03.001) Koops, K., McGrew, W. C. & Matsuzawa, T. 2010 Do chim- Fowler, A. & Sommer, V. 2007 Subsistence technology of panzees (Pan troglodytes) use cleavers and anvils to Nigerian chimpanzees. Int. J. Primatol. 28, 997–1023. fracture Treculia africana fruits? Preliminary data on (doi:10.1007/s10764-007-9166-0) a new form of percussive technology. Primates 51, Gilby, I. C. 2006 Meat sharing among Gombe chimpanzees: 175–178. (doi:10.1007/s.10329-009-0178-6) harassment and reciprocal exchange. Anim. Behav. 71, Lanjouw, A. 2002 Behavioural adaptations to water scarcity 953–963. (doi:10.1016/j.anbehav.2005.09.009) in Tongo chimpanzees. In Behavioural diversity in chim- Gomes, C. M. & Boesch, C. 2009 Wild chimpanzees panzees and bonobos (eds C. Boesch, G. Hohmann & exchange meat for sex on a long-term basis. PLoS ONE L. F. Marchant), pp. 52–60. Cambridge, UK: 4, e5116. (doi:10.1371/journal.pone.0005116) Cambridge University Press. Goodall, J. & Hamburg, D. A. 1974 Chimpanzee behavior as Ling, V., Hernandez-Aguilar, A., Haslam, M. & Carvalho, S. a model for the behavior of early man. New evidence on 2009 The origins of percussive technology; a smashing possible origins of human behavior. Am. Handb. Psychiat. time at Cambridge. Evol. Anthropol. 18, 48–49. (doi:10. 6, 14–43. 1002/evan.20197) Hansell, M. H. 2004 Animal architecture. Oxford, UK: Lonsdorf, E. V. 2006 What is the role of mothers in Oxford University Press. the acquisition of termite-fishing behaviors in wild Haslam, M. et al. 2009 Primate archaeology. Nature 460, chimpanzees (Pan troglodytes schweinfurthii)? Anim. Cogn. 339–344. (doi:10.1038/nature08188) 9, 36–46. (doi:10.1007/s10071-005-0002-7) Heaton, J. L. & Pickering, T. R. 2006 Archaeological analysis Lovejoy, C. O. 2009 Reexamining human origins in light of does not support intentionality in the production of brushed Ardipithecus ramidus. Science 326, 74e1–74e8. (doi:10. ends on chimpanzee termiting tools. Int. J. Primatol. 27, 1126/science.1175834) 1619–1633. (doi:10.1007/s10764-006-9091-7) Marchant, L. F. & McGrew, W. C. 2004 Percussive technol- Henshilwood, C. A., d’Errico, F., Vanhaeren, M., van Niekerk, ogy: chimpanzee baobab smashing and the evolutionary K. & Jacobs, Z. 2004 Middle Stone Age shell beads from modeling of hominid knapping. In Stone knapping: the South Africa. Science 304,404.(doi:10.1126/science. necessary conditions for a uniquely hominin behaviour? (eds 1095905) V. Roux & B. Bril), pp. 341–350. Cambridge, UK: Hernandez-Aguilar, A. 2009 Chimpanzee nest distribution McDonald Institute for Archaeological Research. and site re-use in a dry habitat: implications for early Matsuzawa, T. 2006 Sociocognitive development in hominin ranging. J. Hum. Evol. 57, 350–364. (doi:10. chimpanzees: a synthesis of laboratory work and 1016/j.jhevol.2009.03.007) fieldwork. In Cognitive development in chimpanzees (eds Hernandez-Aguilar, A., Moore, J. & Pickering, T. R. 2007 T. Matsuzawa, M. Tomonaga & M. Tanaka), pp. 3–33. Savanna chimpanzees use tools to harvest the Tokyo, Japan: Springer.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Chimpanzees and the last common ancestor W. C. McGrew 3275

Matsuzawa, T., Biro, D., Humle, T., Inoue-Nakamura, N., Pruetz, J. D. & Bertolani, P. 2007 Savanna chimpanzees, Tonooka, R. & Yamakoshi, G. 2001 Emergence of culture Pan troglodytes verus, hunt with tools. Curr. Biol. 17, in wild chimpanzees: education by master-apprenticeship. 412–417. (doi:10.1016/j.cub.2006.12.042) In Primate origins of human cognition and behavior (ed. T. Pruetz, J. D. & Bertolani, P. 2009 Chimpanzee (Pan troglodytes Matsuzawa), pp. 557–574. Tokyo, Japan: Springer. verus) behavioral responses to stresses associated with living McBrearty, S. & Jablonski, N. G. 2005 First fossil chimpan- in a savanna-mosaic environment: implications for hominin zee. Nature 437, 105–108. (doi:10.1038/nature.04008) adaptations to open habitats. Palaeoanthropology 2009, McGrew, W. C. 1974 Tool use by wild chimpanzees in 252–262. (doi:10.4207/PA.2009.ART33) feeding upon driver ants. J. Hum. Evol. 3, 501–508. Pruetz, J. D. & LaDuke, T. C. 2010 Reaction to fire by (doi:10.1016/0047-2484(74)90010-4) savanna chimpanzees (Pan troglodytes verus) at Fongoli, McGrew, W. C. 1981 The female chimpanzee as a human Senegal: conceptualization of ‘fire behavior’ and the evolutionary prototype. In Woman the gatherer (ed. F. case for a chimpanzee model. Am. J. Phys. Anthropol. Dahlberg), pp. 35–73. New Haven, CT: Yale University 141, 646–650. (doi:10.1002/ajpa.21245) Press. Sanz, C. M. & Morgan, D. B. 2007 Chimpanzee tool McGrew, W. C. 1992 Chimpanzee material culture: impli- technology in the Goualougo Triangle, Republic of cations for human evolution. Cambridge, UK: Cambridge Congo. J. Hum. Evol. 52, 420–433. (doi:10.1016/j. University Press. jhevol.2006.11.001) McGrew, W. C. 2004 The cultured chimpanzee: reflections Sanz, C. M. & Morgan, D. B. 2009 Flexible and persistent on cultural primatology. Cambridge, UK: Cambridge tool-using strategies in honey-gathering by wild University Press. chimpanzees. Int. J. Primatol. 30, 411–427. (doi:10. McGrew, W. C., Ham, R. M., White, L. T. J., Tutin, 1007/s10764-009-9359-5) C. E. G. & Fernandez, M. 1997 Why don’t chimpanzees Sanz, C., Call, J. & Morgan, D. 2009a Design complexity in in Gabon crack nuts? Int. J. Primatol. 18, 353–374. termite-fishing tools of chimpanzees (Pan troglodytes). (doi:10.1023/A:1026382316131) Biol. Lett. 5, 293–296. (doi:10.1098/rsbl.2008.0786) Mercader, J., Panger, M. & Boesch, C. 2002 Excavation of a Sanz, C. M., Schoening, C. & Morgan, D. B. 2009b chimpanzee stone tool site in the African rainforest. Chimpanzees prey on driver ants with specialized tool Science 296, 1452–1455. (doi:10.1126/science.1070268) set. Am. J. Primatol. 71,1–8.(doi:10.1002/ajp.20744) Mercader, J., Barton, H., Gillespie, J., Harris, J., Kuhn, S., Sayers, K. & Lovejoy, C. O. 2008 The chimpanzee has no Tyler, R. & Boesch, C. 2007 4,300-year-old chimpanzee clothes. A critical examination of Pan troglodytes sites and the origins of percussive stone technology. in models of human evolution. Curr. Anthropol. 49, Proc. Natl Acad. Sci. USA 104, 3043–3048. (doi:10. 87–114. (doi:10.1086.523765) 1073/pnas/0607909104) Schoening, C., Humle, T., Moebius, Y. & McGrew, W. C. Mitani, J. C., Watts, D. P. & Muller, M. N. 2002 Recent 2008 The nature of culture: technological variation in developments in the study of wild chimpanzee behavior. chimpanzee predation on army ants revisited. J. Hum. Evol. Anthropol. 11, 9–25. (doi:10.1002/evan.10008) Evol. 55, 48–59. (doi:10.1016/j.jhevol.2007.12.002) Moore, J. J. 1996 Savanna chimpanzees, referential models Sousa, C., Biro, D. & Matsuzawa, T. 2009 Leaf-tool use for and the last common ancestor. In Great ape societies (eds drinking water by wild chimpanzees (Pan troglodytes): W. C. McGrew, L. F. Marchant & T. Nishida), acquisition patterns and handedness. Anim. Cogn. 12, pp. 275–292. Cambridge, UK: Cambridge University S115–S125. (doi:10.1007/s.10071-009-0278-0) Press. Sponheimer, M., Loudon, J. E., Codron, C., Howells, M. Mora, R. & de la Torre, I. 2005 Percussion tools in Olduvai E., Pruetz, J. D., Codron, J., de Ruiter, D. J. & Lee- Beds I and II (Tanzania): implications for early human Thorp, J. A. 2006 Do ‘savanna’ chimpanzees consume activities. J. Anthropol. Archaeol. 24, 179–192. (doi:10. C4 resources? J. Hum. Evol. 51, 128–133. (doi:10.1016/ 1016/j.jaa.2004.12.001) j.jhevol.2006.02.002) Morgan, B. J. & Abwe, E. E. 2006 Chimpanzees use stone Stanford, C. B. 1999 The hunting apes. Meat eating and the hammers in Cameroon. Curr. Biol. 16, R632–R633. origins of human behavior. Princeton, NJ: Princeton (doi:10.1016/j.cub.2006.07.045) University Press. Murray, C. M., Gilby, I. C., Mane, S. V. & Pusey, A. E. 2008 Stewart, F. A., Pruetz, J. D. & Hansell, M. H. 2007 Do Adult males inherit maternal ranging patterns. Curr. Biol. chimpanzees build comfortable nests? Am. J. Primatol. 18, 20–24. (doi:10.1016/j.cub.2007.11.044) 69, 930–939. (doi:10.1002/ajp20432) Nishida, T., Matsusaka, T. & McGrew, W. C. 2009 Sugiyama, Y. 1997 Social tradition and the use of tool- Emergence, propagation or disappearance of novel composites by wild chimpanzees. Evol. Anthropol. 6, behavioral patterns in the habituated chimpanzees of 23–27. (doi:10.1002/(SICI)1520-6505(1997)6:1,23:: Mahale: a review. Primates 50, 23–36. (doi:10.1007/ AID-EVAN7.3.0.CO;2-X) s10329-008-0109-y) Surbeck, M. & Hohmann, G. 2008 Primate hunting by Normand, E. & Boesch, C. 2009 Sophisticated Euclidean bonobos at LuiKotale, Salonga National Park. Curr. maps in forest chimpanzees. Anim. Behav. 77, Biol. 18, R906–R907. (doi:10.1016/j.cub.2008.08.040) 1195–1201. (doi:10.1016/anbehav2009.01.025) Tennie, C., Gilby, I. C. & Mundry, R. 2009 The meat-scrap Normand, E., Ban, S. D. & Boesch, C. 2009 Forest hypothesis: small quantities of meat may promote chimpanzees (Pan troglodytes verus) remember the cooperative hunting in wild chimpanzees (Pan troglodytes). location of numerous fruit trees. Anim. Cogn. 12, Behav. Ecol. Sociobiol. 63, 421–431. (doi:10.1007/ 797–807. (doi:10.1007/s10071-009-0239-7) s00265-008-0676-3) Osvath, M. 2009 Spontaneous planning for future Tooby, J. & DeVore, I. 1987 The reconstruction of hominid stone throwing by a male chimpanzee. Curr. Biol. 19, behavioral evolution through strategic modeling. In R190–R191. (doi:10.1016/j.cub.2009.01.010) The evolution of human behavior: primate models (ed. Pruetz, J. D. 2007 Evidence of cave use by savanna W. G. Kinzey), pp. 183–237. Albany, NY: State Univer- chimpanzees (Pan troglodytes verus) at Fongoli, Senegal: sity of New York Press. implications for thermoregulatory behavior. Primates 48, Toth, N. & Schick, K. 2009 The Oldowan: the tool 316–319. (doi:10.1007/s10329-007-0038-1) making of early hominins chimpanzees compared.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3276 W. C. McGrew Chimpanzees and the last common ancestor

Ann. Rev. Anthropol. 38, 289–305. (doi:10.1146/ Uganda. Int. J. Primatol. 23, 1–28. (doi:10.1023/A: annurev-anthro-091908-164521) 1013270606320) Watts, D. P. 2008a Scavenging by chimpanzees at Ngogo and Whiten, A. et al. 2010 Studying extant species to model our the relevance of chimpanzee scavenging to early hominin past. Science 327, 410. (doi:10.1126/science.327.5964. behavioral ecology. J. Hum. Evol. 54, 125–133. (doi:10. 410-a) 1016/j.jhevol.2007.07.008) Wynn, T. G. & McGrew, W. C. 1989 An ape’s view of the Watts, D. P. 2008b Tool use by chimpanzees at Ngogo, Oldowan. Man 24, 383–398. (doi:10.2307/2802697) Kibale National Park, Uganda. Int. J. Primatol. 29, Yamakoshi, G. & Sugiyama, Y. 1995 Pestle-pounding 83–94. (doi:10.1007/s10764-007-9227-4) behavior of wild chimpanzees at Bossou, Guinea: Watts, D. P. & Mitani, J. C. 2002 Hunting behavior a newly observed tool-using behavior. Primates 36, of chimpanzees at Ngogo, Kibale National Park, 489–500. (doi:10.1007/BF02382871)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Phil. Trans. R. Soc. B (2010) 365, 3277–3288 doi:10.1098/rstb.2010.0096

More reliable estimates of divergence times in Pan using complete mtDNA sequences and accounting for population structure Anne C. Stone1,*, Fabia U. Battistuzzi2, Laura S. Kubatko4, George H. Perry Jr1, Evan Trudeau6, Hsiuman Lin5 and Sudhir Kumar2,3 1School of Human Evolution and Social Change, 2Center for Evolutionary Medicine and Informatics, Biodesign Insitute, and 3School of Life Sciences, Arizona State University, Tempe, AZ, USA 4Department of Mathematics and Statistics, and 5Department of Anthropology, University of New Mexico, Albuquerque, NM, USA 6Department of Molecular, Cellular and Developmental Biology, University of Colorado, Boulder, CO, USA Here, we report the sequencing and analysis of eight complete mitochondrial genomes of chimpanzees (Pan troglodytes) from each of the three established subspecies (P.t. troglodytes, P.t. schweinfurthii and P.t. verus) and the proposed fourth subspecies (P.t. ellioti). Our population genetic analyses are consistent with neutral patterns of evolution that have been shaped by demography. The high levels of mtDNA diversity in western chimpanzees are unlike those seen at nuclear loci, which may reﬂect a demographic history of greater female to male effective population sizes possibly owing to the characteristics of the founding population. By using relaxed-clock methods, we have inferred a timetree of chimpanzee species and subspecies. The absolute divergence times vary based on the methods and calibration used, but relative divergence times show extensive uniformity. Overall, mtDNA produces consistently older times than those known from nuclear markers, a discrepancy that is reduced signiﬁcantly by explicitly accounting for chimpanzee population structures in time estimation. Assuming the human–chimpanzee split to be between 7 and 5 Ma, chimpanzee time estimates are 2.1–1.5, 1.1–0.76 and 0.25–0.18 Ma for the chimpanzee/bonobo, western/ (eastern þ central) and eastern/central chimpanzee divergences, respectively. Keywords: mitochondrial genome; Pan troglodytes; Pan paniscus; divergence time

1. INTRODUCTION genetic variation in these species shed light on their Mitochondrial DNA has long been used to investigate evolutionary histories as well as serve as a comparison questions about primate taxonomy and demography to our own history. Both paleoanthropological (e.g. Morin et al. 1994; Horai et al. 1995; Gagneux and genetic studies indicate that that the human et al. 1999; Schrago & Russo 2003; Eriksson et al. and chimpanzee þ bonobo lineages diverged 6.5– 2004; Raaum et al. 2005). The ability to sequence 4.2 Ma (Sarich & Wilson 1973; White et al.1994; the complete mtDNA genome relatively quickly and Chen & Li 2001; Glazko & Nei 2003; Kumar et al. inexpensively has resulted in a number of studies in 2005), while chimpanzees and bonobos diverged humans that investigate population history (Ingman more recently with estimates ranging from 2.5 to et al. 2000; Maca-Meyer et al. 2001; Herrnstadt et al. 0.8 Ma (Horai et al.1992; Kaessmann et al.1999; 2002; Ingman & Gyllensten 2003; Macaulay et al. Stone et al.2002; Yu et al.2003; Fisher et al.2004; 2005; Thangaraj et al. 2005; Sun et al. 2006) and Won & Hey 2005; Caswell et al.2008). Genetic data, selection (Nachman et al. 1996; Mishmar et al. 2003; as well as some morphological data, suggest strong Elson et al. 2004). However, the application of com- population structuring within chimpanzees that corre- plete mtDNA sequence data to questions about lates with subspecies boundaries, and this structure population history and selection within other species appears to be demarcated by river and habitat bound- has not been common. aries and reinforced by dispersal patterns (Gagneux Chimpanzees (Pan troglodytes)andbonobos(Pan et al. 2001; Guy et al.2003; Lockwood et al.2004; paniscus) are our sister species, and studies of Gonder et al.2006; Becquet et al.2007). Currently, three subspecies, distributed across the central part of Africa, are recognized within chimpan- * Author for correspondence ([email protected]). zees. Pan t. schweinfurthii is the easternmost Electronic supplementary material is available at http://dx.doi.org/ subspecies, located in Tanzania, Burundi, Rwanda, 10.1098/rstb.2010.0096 or via http://rstb.royalsocietypublishing.org. Uganda and the Democratic Republic of Congo. The One contribution of 14 to a Discussion Meeting Issue ‘The ﬁrst four central subspecies, P. t. troglodytes, is found in Congo, million years of human evolution’. Gabon, the Central African Republic, Equatorial

3277 This journal is q 2010 The Royal Society Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3278 A. C. Stone et al. Divergence times in Pan

Guinea and Cameroon, and P. t. verus, the western- P. t. schweinfurthii and one P. t. ellioti) and one captive most subspecies, is found in Senegal, Guinea-Bissau, born chimpanzee (haplotype corresponds to Guinea, Sierra Leone, Liberia, Mali and Ivory Coast. P. t. troglodytes) were sequenced for this study. When MtDNA research has also suggested a potential unknown, subspecies status was determined based on fourth subspecies, P. t. ellioti (Oates et al. 2009), for- a comparison of the mtDNA HVI region sequence to merly known as P. t. vellerosus, in Nigeria (Gonder those from chimpanzees of known subspecies (Morin et al. 1997, 2006; Gagneux et al. 1999), although the et al. 1994; Gonder et al. 1997; Stone et al. 2002). limited Y-chromosome evidence has failed to support Two P. t. verus samples, Pt115 and Pt120 (ISIS no. this claim (Stone et al. 2002). 2738 and 2216), and the P. t. ellioti sample, Pt114 Demographic inferences about chimpanzee subspe- (ISIS no. 2412), were from the New Iberia Primate cies are limited but mostly indicate a larger effective Center. The remaining two P. t. verus samples, Pt82 population size in the central subspecies, and an initial and Pt105 (North American regional studbook for split between the western and central/eastern subspecies the chimpanzee no. 341 and ISIS no. 2435), were (Deinard & Kidd 1999; Kaessmann et al.1999; Stone from the Riverside Zoo in Scottsbluff, NE, and the et al.2002; Fisher et al.2004; Won & Hey 2005; Bec- Southwest Foundation for Biomedical Research, quet et al. 2007). There has also been a distinction respectively. The two P. t. schweinfurthii samples, between haploid markers (both mtDNA and Y-chromo- Pt96 and Pt161 (ISIS no. 3020 and 925), and the some studies) that have shown high levels of structure P. t. troglodytes sample, Pt13 (ISIS no. 4441), were (corresponding to subspecies designations) and autoso- from the Primate Foundation of Arizona. Whole mal nuclear data that have suggested significant gene blood samples (5–10 ml) were taken during routine flow, differences in male and female effective population veterinary examinations, and DNA was isolated sizes and/or incomplete sorting of lineages. On the one using a standard phenol/chloroform-based extraction hand, this has led to some researchers nominating new (Sambrook & Russell 2001). subspecies based on results from haploid markers (Gonder et al.1997; Gonder Disotell & Oates 2006), while others have proposed elimination of subspecies (b) Polymerase chain reaction and nucleotide designations altogether based on the nuclear data sequencing (Fisher et al.2006). In addition, some experts suggest The complete mtDNA genome was amplified in 28 that the western and Nigerian chimpanzees should overlapping segments using the polymerase chain form one subspecies (P. t. vellerosus), while the central reaction (PCR) in a PTC-200 thermal cycler and eastern chimpanzees should belong to a second (MJ Research). PCR primers for each segment are subspecies (P. t. troglodytes; Gonder Disotell & Oates listed in the electronic supplementary material. For 2006). More recently, a large autosomal microsatellite most segments, PCR conditions were: 948C for dataset has supported substructure within chimpanzees 5 min (948C, 30 s; annealing temperature specified that corresponds to the subspecies designations (Bec- in table S1 in the electronic supplementary material, quet et al.2007). 30 s; 728C, 30 s) for 35 cycles, followed by a single Comparisons of levels of intraspecific variability final extension of 728C for 5 min. A touchdown PCR have important implications for understanding the protocol was used for segments amplified with primers evolution of hominoid genomes and clarifying the L846 and H1620, L4589 and H5276 and L14110 and demographic history of contemporary populations of H14900. For these segments, PCR conditions were: humans and great apes (Stone & Verrelli 2006). 948C for 5 min (948C, 30 s; 108C above annealing Because of the different signals regarding population temperature specified in table S1 in the electronic sup- history as well as the conflicting estimates of diver- plementary material minus 0.58C per cycle, 30 s; gence times based on different loci, a better sampling 728C, 30 s) for 20 cycles, (948C, 30 sec; annealing of markers and populations is needed. In this study, temperature specified in table S1 in the electronic sup- we report complete mtDNA sequences in eight chim- plementary material, 30 s; 728C, 30 s) for another 20 panzees including individuals from all of the three cycles, then 728C for 5 min. PCR products were pur- currently recognized subspecies as well as an individ- ified with the QIAquick Purification Kit (Qiagen) and ual with a P. t. ellioti mtDNA haplotype to assess the sequenced in two directions, using the BigDye termin- neutrality of evolutionary patterns of the mtDNA ator cycle sequencing kit v. 3.1 (Applied Biosystems) genome and to examine intraspecific diversity. A and an Applied Biosystems 3730 capillary sequencer. major emphasis of our mtDNA analysis is to investi- Sequence trace files were assembled using the gate why the timing of divergence between SEQMAN program (DNAStar), and then manually chimpanzees and bonobos and among the subspecies checked and aligned. of chimpanzees that have been reported from previous MtDNA insertions into the nuclear genome mtDNA studies are much older than those obtained (numts) can complicate analyses and invalidate con- using the nuclear DNA (e.g. Horai et al. 1995; Stone clusions if they are mistakenly amplified in the place et al. 2002; Fisher et al. 2004; Won & Hey 2005). of authentic mtDNA sequence (Bensasson et al. 2001; Thalmann et al. 2004). Although there was no direct evidence that numts had been amplified here 2. MATERIAL AND METHODS (i.e. there were no ‘heterozygous’ positions and no (a) Materials nucleotide mismatches between overlapping frag- The complete mitochondrial genomes from seven ments), we checked the authenticity of our sequences wild-born chimpanzees (four P. troglodytes verus,two with the long-range PCR method described by

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Divergence times in Pan A. C. Stone et al. 3279

Thalmann et al. (2004) using Platinum Taq HiFi The mutation rate for the complete genome, (Invitrogen). Briefly, for each chimpanzee individual, excluding the D-loop region, was calculated from we amplified two large fragments, A (approx. 9 kb) the data as follows: mutation rate per site per year ¼ and B (approx. 10 kb), which overlap each other at (k/2Tsplit) l, where k is the mean genetic distance, both ends to cover the entire mtDNA genome using l the length of the sequence and Tsplit the time in primers listed in table S1 in the electronic supplemen- years since the human and chimpanzee divergence. tary material. Five microlitres of the PCR product was Tsplit is assumed to be 6 Ma. The mean genetic dis- electrophoresed through a 2 per cent NuSieve GTG tance was estimated between the 10 chimpanzees low-melting agarose (Cambrex) gel. Bands were examined in this study and the 53 humans from excised and melted in 100 ml Molecular Biology Ingman et al. (2000) using the Tamura–Nei distance Reagent Water (Sigma) at 558C for 1 h, and then (Tamura & Nei 1993) with the evolutionary rates vortexed thoroughly. Four microlitres of this mixture among sites modelled using the gamma distribution was then used to amplify and sequence three of our in MEGA 4 (Tamura et al. 2007). The shape par- original segments, using the methods specified above. ameter for the gamma distribution was calculated One of these segments (primers L15255 and H107, using MODELTEST as noted below. 1214 bp) was amplified from both fragments A and MODELTEST (Posada & Crandall 1998) was used to B. Additionally, we amplified a 687 bp segment select the most appropriate evolutionary model (L4589 and H5276) from fragment A and an 862 bp for each dataset. For the 13 protein-coding genes, segment (L8333 and H9195) from fragment B. The the general time-reversible (GTR) model with a resulting sequences were compared with each other proportion of the sites invariable (I) and gamma- and with the original sequence for each individual. distributed rates among sites (G ) was selected by the There were no observed discrepancies, supporting AIC criterion. For the complete genome without the authenticity of the complete mtDNA genome the D-loop, the GTR þ G model was selected by the sequences generated for these eight individuals. AIC. Phylogenetic analyses were then performed These data were submitted to GenBank and using PAUP* (Swofford 2003) using the model and are listed under accession numbers GU112738– parameters estimated by MODELTEST. Ten random GU112745. addition sequences with TBR branch swapping were used to obtain the maximum-likelihood estimates of the phylogenies. Bootstrap analyses were subsequently (c) Data analysis performed using 100 bootstrap replicates and TBR We compared our sequences with those from previous searches for each replicate for each dataset. studies, including two P. t. verus (Jenny, GenBank We calculated divergence times between species and accession number X93335 and the chimpanzee subspecies using two relaxed-clock methods: MULTI- mtDNA reference sequence no. NC 001643), one DIVTIME (MDT) and BEAST (Thorne & Kishino P. paniscus (no. NC 001644), one gorilla (Gorilla 2002; Drummond & Rambaut 2007). In these gorilla, no. NC 001645), one orangutan from each of methods, a Markov Chain Monte Carlo (MCMC) the two subspecies (Pongo pygmaeus, no. NC 001646 procedure is used within a Bayesian analysis frame- and Pongo pygmaeus abelii, no. NC002083), one work to estimate the posterior distributions of gibbon (Hylobates lar, no. NC 002082), the Cambridge evolutionary rates and divergence times, given priors human reference sequence (Homo sapiens,no. on phylogenetic relationships and calibration nodes. AC000021) and 53 additional humans (Anderson These analyses were performed using DNA sequence et al. 1981; Horai et al. 1992, 1995; Arnason et al. alignment of the complete mitochondrial genome, 1996; Xu & Arnason 1996; Andrews et al. 1999; except the D-loop region, and for 13 protein-coding Ingman et al. 2000). The hypervariable region or genes separately. The protein-coding genes were ana- D-loop is non-coding and was excluded from most lysed at amino acid and DNA sequence level as a analyses because it is known to have a very different super-alignment. We also analysed the fourfold (4F) mutational pattern. degenerate sites by themselves, as their evolutionary Two estimates of diversity were calculated for each patterns are expected to be the closest to the strict locus: p is based on the average number of nucleotide neutrality (Kumar et al. 2005). differences per site between two sequences randomly In MDT, branch lengths of the amino acid dataset drawn from a sample and us is based on the (16 taxa, 3772 sites) were estimated with the mito- sample size-corrected proportion of segregating sites chondrial mammal (mtmam.dat) model of (Watterson 1975; Nei 1987). The Jukes & Cantor substitution, while for nucleotides (genome and 4F (1969) correction was applied to all sequence sites) the F84 þ gamma model was used within the comparisons involving interspecific variation and PAML program package (Yang 2007). Initial number divergence (Lynch & Crease 1990). Under equili- of sites analysed for whole genome and 4F degenerate brium conditions with respect to mutation and drift, sites were 15 514 and 1843, respectively. All sites con- both p and us estimate the neutral parameters: 2Nem taining gaps and ambiguous nucleotides were excluded for mtDNA, where Ne is the effective population size from the analyses. Other parameters used in MDT and m is the neutral mutation rate. Tajima’s D-statistic were: 10 000 sampling of the Markov chain, with was calculated to test for deviations from neutral fre- sampling frequency every 100, burn-in of 100 000, quency distribution (Tajima 1989). The measures of and bigtime was set to 50 Ma. The root to tip distance diversity and tests of neutrality were performed with (rttm) was set at 25 Ma with identical standard the program DNASP 4.0 (Rozas et al. 2003). deviation (rttmsd).

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3280 A. C. Stone et al. Divergence times in Pan

Table 1. Diversity statistics for mtDNA. Length is excluding gaps. *Signiﬁcance: p , 0.05. taxa n length s p (%) u (%) Tajima’s D complete genome P.troglodytes 10 16 543 695 1.48 1. 49 20.16 P.t. schweinfurthii 2 16 556 38 0.23 0.23 P.t. s. and P.t. t. 3 16 556 134 0.54 0.54 P.t. verus (w/Nigeria) 7 16 548 413 0.95 1.02 20.49 P.t. verus 6 16 548 320 0.84 0.85 20.1 H. sapiensa 53 16 553 657 0.37 0.87 22.12* without D-loop P.troglodytes 10 15 441 544 1.23 1.25 20.16 P.t. schweinfurthii 2 15 443 26 0.17 0.17 P.t. s. and P.t. t. 3 15 443 95 0.41 0.41 P.t. verus (w/Nigeria) 7 15 441 298 0.72 0.79 20.59 P.t. verus 6 15 441 224 0.62 0.64 20.17 H. sapiensa 53 15 430 516 0.29 0.74 22.23* aIngman et al. (2000).

The same data sets were analyzed also in BEAST 3. RESULTS with an uncorrelated lognormal relaxed clock. The complete mtDNA genomes (16 541 bp) of 10 Substitution model used were the mtREV þ chimpanzees contain 695 segregating sites (table 1). gamma þ invariant sites model for amino acids and We also observe insertion/deletion (indel) polymorph- the GTR þ gamma þ invariant sites for nucleotides. isms in both the 12S rRNA gene and the D-loop. In We also separated models for population and species the 12S rRNA gene, an insertion of a guanine (G) divergences by using two populations based on the after nucleotide position 138 was present in all geographical distribution of the western and eastern þ sequences except the chimpanzee reference sequence. central chimpanzee subspecies (two populations plus This G is also present in humans (Anderson et al. one species model). The remaining divergences were 1981; Ingman et al. 2000) where it is found in stem estimated under a Yule speciation process. The 7 of the secondary structure model of Neefs et al. length of the chain for each analysis was adjusted to (1993). In addition, a length polymorphism was the dataset in order to obtain effective sample sizes found in the 12S rRNA gene in the loop between above 200 for all parameters. stems 23 and 24. Here, at nucleotide positions 378– The same prior information on times (lower and 384, an unstable run of cytocines resulted in signifi- upper bounds) was used for the two molecular clock cant variation with at least 6–15 cytocines present. methods. These included the times of the following This region was difficult to PCR and produced a het- species divergences: gorilla versus chimpanzee þ eroplasmic sequence. In the D-loop, a total of 12 human divergence (10–6.5 Ma) and the chimpanzee indels was found. Because it is difficult to properly and human divergence (6.5–4.2 Ma) times defined incorporate indels into models of sequence evolution as uniform distributions. These calibrations were for estimating divergence times, these were removed used together or separately depending on the dataset from further analyses. and the hypothesis to test. MDT and BEAST yielded Non-synonymous polymorphisms were found in all divergence times and their 95% credibility intervals 13 protein-coding genes. On average, chimpanzee (CrIs). sequences contained 28.8 non-synonymous differ- In addition to species and subspecies divergences, ences when compared with 8.91 observed in BEAST also produces the age of the most recent humans. The estimates of synonymous and non- common ancestor (TMRCA) based on the population synonymous substitution diversities for each sample included. We compared these estimates with mitochondrial gene as well as comparative data from those obtained using the GENETREE program, which humans are shown in table 2. When examining silent simulates a coalescent process including time infor- and replacement sites separately (table 2), Tajima’s mation conditional on a specific haplotype tree with test rejected strict neutrality at a 5 per cent significance a given value of u (Griffiths & Tavare´ 1994; Bahlo & level in chimpanzees only for non-synonymous sites at Griffiths 2000). We estimated u for each dataset by CO3 and ND4L. Humans show many significantly GENETREE using the maximum-likelihood method. negative Tajima’s D-statistic for both synonymous Indels and the D-loop were not included in the ana- and non-synonymous sites. When data from only lyses, and constant population sizes and panmixia P. t. verus (without Pt114) was examined, three genes were assumed. Simulation results are based on 10 (ND1, CO2 and ND6) showed significantly negative million replicate runs. To calculate the TMRCA in Tajima’s D-values at replacement sites. When Pt114 years, a chimpanzee generation time of 15 years was was included in P. t. verus, only replacement sites at used and the divergence between chimpanzees and CO3 produced a significantly negative value (data humans was set at 6 Ma. not shown).

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Divergence times in Pan A. C. Stone et al. 3281

Table 2. Population diversity estimates and tests of neutrality at mtDNA protein-coding genes in humans (n ¼ 53, Ingman et al. 2000) and chimpanzees (n ¼ 10). *p , 0.05.

replacement silent taxa and genes lengtha S pb TDc S p TD

Pan troglodytes ND1 954 3 0.0016 0.13 41 0.0539 20.27 ND2 1041 16 0.0066 20.46 33 0.046 0.18 CO1 1539 4 0.0016 1.32 42 0.042 0.51 CO2 681 2 0.0013 20.03 14 0.026 20.46 ATP8 204 4 0.0097 0.23 4 0.025 20.45 ATP6 678 7 0.0054 0.39 25 0.051 0.11 CO3 783 6 0.0021 21.8* 28 0.058 0.68 ND3 345 5 0.0055 20.83 10 0.045 0.41 ND4L 294 2 0.0019 21.4* 9 0.048 0.81 ND4 1377 12 0.0032 21 46 0.045 20.05 ND5 1809 17 0.0046 0.23 63 0.05 0.07 ND6 522 3 0.0022 0.54 16 0.045 0.33 CYTB 1140 8 0.0035 0.11 54 0.069 0.18 all genes 11 316 88 0.0034 20.34 385 0.049 0.14 Homo sapiens ND1 954 8 0.0007 21.93* 19 0.0087 21.49* ND2 1041 13 0.0012 22.04* 29 0.0071 22.31* CO1 1539 8 0.0008 21.32 40 0.0083 22.15* CO2 681 3 0.0007 21.11 19 0.0067 22.32* ATP8 204 2 0.0014 20.96 6 0.0158 21.18 ATP6 678 8 0.002 21.17 22 0.0096 22.08* CO3 783 4 0.0003 21.73* 25 0.0105 22.01* ND3 345 3 0.0028 0.12 10 0.0108 21.67* ND4L 294 1 0.0002 21.32* 9 0.015 21.09 ND4 1377 12 0.0007 22.15* 43 0.0116 21.89* ND5 1809 23 0.0015 21.95 51 0.0095 22.12* ND6 522 5 0.0013 21.31 20 0.0118 22.01* CYTB 1140 16 0.0011 22.2* 26 0.0099 21.66* all genes 11 316 106 0.0011 22.19* 318 0.0096 22.17* aLength does not include the stop codon, for ‘all genes’, ATP8 and ND4L were truncated so that overlapping regions with other genes were not counted twice. bNucleotide diversity. cTajima’s D-test.

Our analysis shows that chimpanzees harbour (figure 1). Bootstrap support was high (greater than approximately four times more nucleotide diversity 97%) for all nodes. The western chimpanzee (p) than humans (table 1), while us is 1.7 times sequences, Pt82, Pt105, Pt115 and Pt120, the chim- greater. Within chimpanzees, P. t. verus exhibited the panzee mtDNA reference sequence and the sequence most variation; however, multiple samples of the cen- for the chimpanzee Jenny cluster together with the tral subspecies were not included in this study and Nigerian sequence, Pt114, as their most closely related only two P. t. schweinfurthii were sampled. Despite taxon. The eastern chimpanzees, Pt96 and Pt161, the population structure within chimpanzees, Tajima’s cluster with the central chimpanzee sequence, Pt13. test of the complete genome and also of only the We first inferred a time tree of chimpanzee and protein-coding genes did not show a significant depar- bonobo evolution by using the MDT software. In ture from neutrality, while it is significantly negative this case, we used two primary time constrains: in humans (tables 1 and 2). These results support 10.0–6.5 Ma for gorilla/(human þ chimpanzee) diver- the assumption of constant population size in gence and 6.5–4.2 for human/chimpanzee divergence. chimpanzees used in the GENETREE analysis. Using this calibration for full mitochondrial genome For a reduced dataset including only the chimpan- (excluding the D-loop), only amino acid alignments zees, the best evolutionary model selected by of 13 protein-coding genes and only 4F degenerate MODELTEST using both likelihood ratio tests and the sites produced similar estimates (table 3). AIC was the GTR þ G model with the shape par- The chimpanzee/bonobo split was estimated to be ameter in the gamma distribution, a, estimated to be 3.2–2.3 Ma. The time estimates for western/ 0.066. PAUP* was then used to estimate the ML (eastern þ central) and eastern/central splits were tree and bootstrap support. Using the entire dataset, dated to 1.68–1.13 and 0.46–0.34 Ma, respectively the phylogenetic analyses of both the complete (table 3). Interestingly, MDT analysis yielded a genome without the D-loop and using only the human/chimpanzee divergence date of approximately protein-coding genes produced similar results 5.7 Ma, which would be considered young by some

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3282 A. C. Stone et al. Divergence times in Pan

Pt161 100 eastern Pt96 + 100 central Pt13

100 Pt114 Pt115 100 Pt82 100 western 97 Pt105 100 Pt120 100 100 Jenny 100 100 Pt reference P. paniscus H. sapiens G. gorilla 0.01 Figure 1. Maximum-likelihood phylogeny of chimpanzees, human and gorilla. Full mitochondrial genome (excluding the D-loop) and 13 mitochondrial protein-coding genes gave similar results. The geographical location of the chimpanzee populations is shown. Bootstrap values are shown near each node.

Table 3. Time estimates and 95% CrIs (in million years) for major divergences in the phylogeny obtained using MDT and BEAST for three datasets (mitochondrial genome, amino acids, and 4F degenerate sites). MDT and BEAST coalescent model do not model population structure within chimpanzees, which, instead, is accounted for in BEAST 2 þ 1 model. MDT divergence times in parenthesis for the 4F degenerate sites are obtained with the human/chimpanzee calibration ﬁxed at 6.5 Ma. All other times are estimated using gorilla/human þ chimpanzee (10.0–6.5 Ma) and human/chimpanzee (6.5–4.2 Ma).

Mt genomea amino acidsb 4Fc

time 95% CrI time 95% CrI time 95% CrI

MDT human/chimpanzee 5.69 4.76–6.46 5.73 4.72–6.46 5.65 (6.5) 4.49–6.46 chimpanzee/bonobo 2.33 1.75–3.14 3.17 2.23–4.24 2.81 (3.35) 1.68–4.19 western/eastern þ central 1.13 0.80–1.70 1.5 0.86–2.47 1.68 (2.04) 0.84–3 Nigerian/western 0.55 0.37–0.86 0.78 0.4–1.48 0.811 (1) 0.35–1.66 eastern/central 0.34 0.22–0.58 0.43 0.16–0.93 0.458 (0.57) 0.7–1.11 BEAST (coalescent model) human/chimpanzee 5.48 4.63–6.5 5.43 4.46–6.5 5.28 4.24–6.35 chimpanzee/bonobo 2.02 1.53–2.57 2.79 1.88–3.72 1.79 1.23–2.42 western/eastern þ central 0.96 0.70–1.22 1.13 0.84–1.85 0.92 0.62–1.27 Nigerian/western 0.48 0.35–0.61 0.72 0.45–1.03 0.42 0.27–0.60 eastern/central 0.27 0.18–0.37 0.35 0.15–0.58 0.21 0.11–0.33 BEAST (2 þ 1 model)d human/chimpanzee 4.94 4.2–5.68 4.87 4.2–5.73 4.7 4.2–5.43 chimpanzee/bonobo 1.76 1.35–2.19 2.28 1.5–3.16 1.49 1.05–1.99 western/eastern þ central 0.83 0.64–1.04 1.05 0.67–1.45 0.76 0.51–1.01 Nigerian/western 0.41 0.32–0.52 0.57 0.36–0.80 0.35 0.23–0.47 eastern/central 0.23 0.16–0.31 0.27 0.1–0.44 0.18 0.095–0.27 aMitochondrial genome excluding the D-loop. bThirteen mitochondrial protein-coding genes. c4F degenerate sites of the 13 protein-coding genes. dTwo populations (western and eastern þ central) and one speciation model (Yule). experts. Therefore, we re-estimated divergence times divergence times using the BEAST software, which in MDT by assuming the human/chimpanzee diver- allows rates among lineages to vary independently. gence to be 6.5 and found that all the resulting time For genomic DNA and amino acid sequence analysis, estimates increased proportionally (table 3). BEAST analyses produced estimates very similar to Because MDT assumes that evolutionary rates those from MDT for the same sequences (table 3). among lineages are autocorrelated, we also estimated However, the BEAST analysis of 4F degenerate sites

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Divergence times in Pan A. C. Stone et al. 3283

3.5 respectively. These results were similar to those found by BEAST in the 4F dataset. 3.0 2.5 2.0 4. DISCUSSION Studies of diversity and divergence times in chimpan- 1.5 zees, and primates in general, have frequently used 1.0 mtDNA. Similar to some previous studies (Morin

time estimates (Ma) et al. 1994; Gagneux et al. 1999), our new data 0.5 indicate that diversity within chimpanzees is greatest 0 in the western subspecies (p ¼ 0.94% for all or chimpanzee/ western/ eastern 0.84% without the Nigerian individual), although bonobo eastern + central /central the sample size for the eastern/central subspecies (p ¼ 0.54%) is rather small (n ¼ 3). This pattern of Figure 2. Compilation of estimated divergence times between chimpanzees and bonobos and within chimpanzees diversity is contrary to those found at nuclear loci from literature (autosomal loci only, filled circle) and this where diversity estimates are higher in the central sub- study (MDT, cross; BEAST (2 þ 1), square). species. For example, analyses of NRY data showed that central chimpanzees had higher levels of diversity, because five different haplotypes were observed when yielded much younger estimates (table 3). The above compared with only one observed in a much larger analyses, however, did not take into account the popu- sample of western chimpanzees (Stone et al. 2002). lation structure of chimpanzees. In order to In a survey of approximately 10 kb at Xq13.3 from incorporate this biological reality, we conducted 30 chimpanzees (17 P. t. verus,12P. t. troglodytes and BEAST analysis by assuming that each P. t . subspecies 1 P. t. schweinfurthii), Kaessmann et al. (1999) found constitutes a population (western subspecies and that P. t. troglodytes were the most diverse. The mean eastern þ central subspecies). This leads to a two pairwise sequence diversity at this locus among populations plus one speciation model (2 þ 1 chimpanzees was 0.13 per cent, which is about four model), which yields younger divergence times than times greater than that in humans (0.037%). We also those obtained from MDT or from other BEAST found a similar difference. analyses (table 3). Kaessmann et al. (1999) also noted that the subspe- Although the use of 2 þ 1 model in BEAST makes cies were not monophyletic for X-chromosome results younger compared with the estimates obtained haplotypes (one western and one central chimpanzee without considering population structure within chim- shared the same haplotype). However, Verrelli et al. panzees, the time estimates based on amino acid (2006) found that haplotypes were not shared between sequences are 50 per cent older than those from the subspecies for the X-chromosome locus G6PD. At this analysis of DNA sequences (genome as well as 4F locus, diversity was also greatest among central degenerate sites). Kumar et al. (2005) have previously chimpanzees in a survey of 56 chimpanzees shown that amino acid time estimates are often older (6 P. t. troglodytes and 48 P. t. verus). Because the and not preferred for relatively recent divergences. X-chromosome has three times the effective popu- Overall, the 2 þ 1 model produces time estimates lation size (Ne) of mtDNA, it should take three times that are more similar to those reported from nuclear as long as mtDNA to achieve monophyly. So it is not autosomal loci (figure 2). Based on the 2 þ 1 surprising that some of the X-chromosome lineages BEAST analysis, the average divergence times within are not completely sorted among the subspecies the chimpanzees across three datasets are 2.28–1.49, given their likely divergence times. 1.05–0.76 and 0.27–0.18 Ma for the chimpanzee/ Autosomal sequences also indicate higher diversity bonobo, western/(eastern þ central) and the eastern/ in central chimpanzees followed by eastern and wes- central divergences, respectively (table 3). These tern chimpanzees (Deinard & Kidd 1999; Yu et al. results highlight the need for modelling populations 2003; Fisher et al. 2004, 2006). Fisher et al. (2004) correctly while estimating times of closely related found that central chimpanzees are 2.0–2.5 times species and subspecies divergences when using more diverse than western chimps and worldwide relaxed-clock methods. samples of humans. Central chimpanzees also show a The time to the MRCA (TMRCA) for each chim- relatively high proportion of rare alleles that could be panzee population in the BEAST analysis was 0.35 the result of an old bottleneck or fine-scale population and 0.18 (table 3). We compared these estimates of structure. Won & Hey (2005) found evidence for one- TMRCA with those obtained from GENETREE, which way gene flow from P. t. verus to P. t. troglodytes and employs a coalescence process and estimates of diver- suggest that this may be the result of interactions sity to generate time estimates. The estimate of between the Nigerian chimpanzees and the central sequence diversity for the complete genome, excluding subspecies. the D-loop, was 1.38 1028 substitutions per site per The distinct pattern of higher mtDNA diversity and year assuming a chimpanzee–human divergence lower diversity at other loci found in western chimpan- time of 6 Ma. Using this estimate and applying zees suggests that the founding population of this GENETREE separately to each subspecies (western subspecies may have been skewed with a larger and eastern þ central), we obtained TMRCA of number of females and a smaller group of closely 202 000 (+14 000) and 180 000 (+19 000) years, related males. This pattern is perhaps not surprising

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3284 A. C. Stone et al. Divergence times in Pan

Table 4. Divergence times (in million years) and ratios among major divergences for 4F degenerate sites from BEAST 2 þ 1 model. The human/chimpanzee (H/C) time is used as the reference in calculating the time ratios. The gorilla/human þ chimpanzee divergence was used as calibration (10.0–6.5 Ma) in all the estimations. The human/chimpanzee calibration was used either as a range (6.5–4.2 Ma) or ﬁxed (6.5 Ma). Absolute times with H/C at 5 or at 7 Ma were scaled according to the ratios calculated.

BEAST (2 þ 1)a

absolute times (4F) time ratios (4F) absolute times

H/C ¼ (4.2–6.5) H/C ¼ 6.5 H/C ¼ (4.2–6.5) H/C ¼ 6.5 H/C ¼ 5 H/C ¼ 7 human/chimpanzee 4.70 6.50 — — 5.00 7.00 chimpanzee/bonobo 1.49 1.94 32% 30% 1.49 2.09 western/(eastern þ central) 0.76 0.98 16% 15% 0.76 1.06 eastern/central 0.18 0.23 4% 4% 0.18 0.25 aTwo populations (western and eastern þ central) and one speciation model (Yule). given the philopatric mating patterns of chimpanzees partial resolution for this discrepancy. We find that the where females disperse at adolescence, and males lack of consideration of population substructure of the remain within their natal group (Nishida 1979; chimpanzee subspecies when using mtDNA is a major Pusey 1979; Wrangham 1979; Goodall 1986; cause of this difference. Therefore, we consider the Pusey & Packer 1986; Langergraber et al. 2007; time estimates from the use of two populations þ one Inoue et al. 2008). However, the pattern of higher speciation model for analysing 4F degenerate sites mtDNA diversity/lower nuclear diversity in western data in BEAST to be the least biased and most appro- chimpanzees compared with central þ eastern chim- priate for estimating chimpanzee divergence times. panzees indicate that the finding and/or subsequent However, the absolute divergence times are strongly demography of the western subspecies was somehow influenced by the calibration points used. For different from that found in the other subspecies. Gen- example, when using two calibration ranges (10.0–6.5 etic studies confirm that the males within many and 6.5–4.2 for gorilla/(human þ chimpanzee) and chimpanzee groups are typically more closely related human/chimpanzee splits, respectively), we obtain than females; however, this is not true in all groups, chimpanzee/bonobo divergence of 1.49 Ma. However, particularly those where habitat fragmentation, disease the human/chimpanzee divergence predicted by or poaching affect group demography (Morin et al. BEAST in this case is only 4.70 Ma, which is 1994; Vigilant et al. 2001; Lukas et al. 2005; Inoue significantly lower than the current expectation. et al. 2008). The mtDNA data generated in this By constraining the human/chimpanzee divergence study indicate that chimpanzee female effective popu- to be 6.5, the chimpanzee/bonobo time increases to lation size has been large (Ne ¼ approx. 36 000 for all 1.94 Ma. Interestingly, the ratio of the estimates of chimpanzees and approx. 20 000 for P. t. verus) with human/chimpanzee and chimpanzee/bonobo diver- no evidence of population bottleneck or expansion. gences is very similar (0.30–0.32) in different BEAST MtDNA data may also be affected by selection, and analyses under 2 þ 1 model (table 4). Therefore, the several studies suggest that many mtDNA protein estimate of chimpanzee/bonobo divergence times polymorphisms are slightly deleterious in humans scales proportionally with the human/chimpanzee (e.g. Nachman et al. 1996; Rand & Kann 1996; calibration. In this case, mtDNA analyses suggest a Kivisild et al. 2006). In this study, we also found range of 2.09–1.49 Ma for chimpanzee/bonobo evidence for a significantly negative value of Tajima’s divergence, because the best estimates of human/ D for most mitochondrial protein-coding genes in chimpanzee divergence are in the range of 7–5 Ma humans (and for the genome as a whole) which can (Kumar et al.2005; Hedges et al. 2006; figure 3). be indicative of population expansion and/or selection. These mtDNA estimates for chimpanzee/bonobo However, in these chimpanzee data, a significant are consistent with the range of 1.8–0.93 Ma based departure from neutrality was only found for CO3 on the earlier analysis of Y-chromosomes, non- and ND4L at replacement sites and there was no evi- coding and non-repetitive genomic segments and dence for a departure from neutrality for the genome X-chromosomes (Xq13.3; Kaessmann et al. 1999; as a whole (table 2). Stone et al. 2002; Yu et al. 2003). However, more In molecular phylogenetics analysis of the complete recent nuclear genome analyses have typically yielded mtDNA of chimpanzees, bonobos, humans and other a much younger date for this divergence (0.93– great apes, human lineages are the most recent closest 0.79 Ma). For example, Fisher et al. (2004) estimated relative of Pan, in agreement with the scores of pre- a divergence time of 0.80 Ma for this divergence using vious studies. A more controversial issue in a moment estimator method that examines the num- chimpanzee evolutionary genomics has been the bers of segregating sites at particular frequencies timing of divergence between chimpanzees and bono- (Wakeley & Hey 1997). Won & Hey (2005) obtained bos and among the subspecies of chimpanzees owing an estimate of 0.89–0.86 Ma from multiple datasets to the discrepancy between nuclear and mitochondrial (Deinard & Kidd 1999; Kaessmann et al. 1999; results. The results from our mtDNA analysis provide Stone et al. 2002; Yu et al. 2003) using an isolation

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Divergence times in Pan A. C. Stone et al. 3285

Jenny H/C: 6.5 – 4.2 Ma Pt reference Pt120 Pt105 western Pt82 Pt115 Pt114 Pt13 central Pt96 + Pt161 eastern P. paniscus H. sapiens G. gorilla Pongo pygmaues pygmaeus Pongo pygmaeus abelii Hylobates lar

H/C: 6.5 Ma Jenny Pt reference Pt120 Pt105 western Pt82 Pt115 Pt114 Pt13 central Pt96 + Pt161 eastern P. Paniscus H. sapiens G. gorilla Pongo pygmaeus pygmaeus Pongo pygmaeus abelii Hylobates lar

25 20 15 10 5 0 Ma

Figure 3. Timetrees obtained from BEAST (2 þ 1 model) from 4F degenerate sites. The time scale is in million years (Ma). Both trees used the gorilla/human þ chimpanzee calibration (10.0–6.5 Ma). Additionally, the upper tree used the human/ chimpanzee (H/C) calibration range of 6.5–4.2 Ma while the lower tree fixed it at 6.5 Ma. with migration (IM) model, as did Becquet & estimates (0.28 Ma for western/eastern split and Przeworski (2007) who estimated a split of 0.92– 0.44 Ma western/central divergence). More recent 0.79 Ma from two datasets using an MCMC method estimates range from 0.34–0.91 Ma (Caswell et al. to estimate the parameters of an IM 2008; Wegmann & Excoffier 2010). Therefore, model. Recently, Hey (2010) estimated a split of nuclear DNA time estimates are again much younger 0.68–1.54 Ma using a multi-population IM model, than those indicated by mtDNA. while Wegmann & Excoffier (2010) used an approxi- The eastern and central subspecies of chimpanzee mate Bayesian computation approach to estimate a are estimated to have diverged recently. Although divergence time of 1.6 Ma. These previous studies these two subspecies do not appear to share either and our analysis show how the time estimates are sen- mtDNA or NRY haplotypes, these haplotypes are sitive not only to the choice of dataset, but also to the not monophyletic (Gagneux et al. 2001; Stone et al. models used to describe the chimpanzee’s population 2002). From the limited number of sequences in this structure. study, we estimate the divergence of eastern and cen- Within chimpanzees, the western and central þ tral subspecies to be approximately 0.25–0.18 Ma. eastern clades diverged between 1.06–0.76 Ma The Nigerian lineage appears to have diverged signifi- according to our mtDNA analyses. This is younger cantly earlier (approx. 0.4 Ma). Even though the single than previous estimates from mtDNA (1.6–1.3 Ma; Nigerian lineage potentially diverged much earlier Morin et al. 1994), but older than other estimates than its closest relatives, there is yet no evidence based on nuclear loci. Fisher et al. (2004) examined from Y-chromosome and autosomal STR markers 5.4 kb of autosomal sequence in 14 central and 16 to elevate the Nigerian chimpanzees to a separate western chimpanzees, and they propose a time of subspecies (Stone et al. 2002; Becquet et al. 2007). 0.65–0.43 Ma for the divergence of western and The consistent discrepancy among the different central þ eastern groups. Won & Hey (2005) also cal- divergence estimates, both within chimpanzees and culated a divergence time estimate of 0.43 Ma for between chimpanzees and bonobos, are probably due the central and western subspecies, while Becquet & to the different population histories reflected by differ- Przeworski (2007) obtained much younger time ent parts of the genome. The older divergence times

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3286 A. C. Stone et al. Divergence times in Pan based on mtDNA data may reflect a demographic his- apes. Genome Res. 17, 1505–1519. (doi:10.1101/gr. tory of greater female effective population size in Pan 6409707) compared with the male effective population size Becquet, C., Patterson, N., Stone, A. C., Przeworski, M. & (Stone et al. 2002; Eriksson et al. 2006). Reich, D. 2007 Genetic structure of chimpanzee popu- Despite the wealth of information gleaned from lations. PLoS Genet. 3, e66. (doi:10.1371/journal.pgen. 0030066) complete mtDNA genome sequences to provide Bensasson, D., Zhang, D., Hartl, D. L. & Hewitt, G. M. insight into the maternal history and patterning of 2001 Mitochondrial pseudogenes: evolution’s misplaced humans (e.g. Ingman et al. 2000; Macaulay et al. witnesses. Trends Ecol. Evol. 16, 314–321. (doi:10.1016/ 2005), such population genetic data have not been S0169-5347(01)02151-6) available for other primates. In chimpanzees, complete Caswell, J. L., Mallick, S., Richter, D. L., Neubauer, J., mitochondrial DNA sequences have previously been Schirmer, C., Gnerre, S. & Reich, D. 2008 Analysis of published for only one of the three subspecies (Horai chimpanzee history based on genome sequence align- et al. 1995; Arnason et al. 1996). Analysis of these ments. PLoS Genet. 4, e1000057. (doi:10.1371/journal. sequences along with eight additional sequences repre- pgen.1000057) senting all of the recognized subspecies, including one Chen, F.-C. & Li, H. 2001 Genomic divergences between individual from the proposed subspecies P. t. ellioti, humans and other hominoids and the effective population size of the common ancestor of humans and chimpanzees. add to the picture of diversity and population history Am. J. Hum. Genet. 68, 444–456. (doi:10.1086/318206) in this species; however, these data also illustrate the Deinard, A. & Kidd, K. 1999 Evolution of a HOXB6 need for additional sampling of chimpanzees through- intergenic region within the great apes and humans. out their range. In addition, mtDNA data may be J. Hum. Evol. 36, 687–703. (doi:10.1006/jhev.1999. affected by certain demographic scenarios (sex- 0298) biased dispersal and bottlenecks), as suggested in Drummond, A. J. & Rambaut, A. 2007 BEAST: Bayesian this study, as well as selection, and thus may not pro- evolutionary analysis by sampling trees. BMC Evol. Biol. vide an accurate time scale of evolutionary or 7, 214. (doi:10.1186/1471-2148-7-214) population events unless population structure is con- Elson, J. L., Turnbull, D. M. & Howell, N. 2004 Compara- sidered (Nachman et al. 1996; Ballard & Rand tive genomics and the evolution of human mitochondrial DNA: assessing the effects of selection. Am. J. Hum. 2005). Much of the primate diversity and taxonomic Genet. 74, 229–238. (doi:10.1086/381505) data published to date relies solely on mtDNA data. Eriksson, J., Hohmann, G., Boesch, C. & Vigilant, L. 2004 These studies should be eyed with caution until Rivers influence the population genetic structure of bono- additional data are available. bos (Pan paniscus). Mol. Ecol. 13, 3425–3435. (doi:10. 1111/j.1365-294X.2004.02332.x) We thank Cecil Lewis, Vinod Swarna and Brian Verrelli for helpful discussions and comments pertaining to this study Eriksson, J., Siedel, H., Lukas, D., Kayser, M., Erler, A., and/or manuscript. This study would not have been Hashimoto, C., Hohmann, G., Boesch, C. & Vigilant, possible without the samples generously provided by the L. 2006 Y-chromosome analysis confirms highly sex- New Iberia Primate Center (supported by an NCRR grant biased dispersal and suggests a low male effective no. U42 RR015087 to the University of Louisiana at population size in bonobos (Pan paniscus). Mol. Ecol. Lafayette, New Iberia Research Center), the Primate 15, 939–949. (doi:10.1111/j.1365-294X.2006.02845.x) Foundation of Arizona, the Southwest Foundation for Fischer, A., Wiebe, V., Paabo, S. & Przeworski, M. 2004 Evi- Biomedical Research and Riverside Zoo, Scottsbluff, dence for a complex demographic history of chimpanzees. Nebraska. This research was supported by the National Mol. Biol. Evol. 21, 799–808. (doi:10.1093/molbev/ Science Foundation to A.S. (BCS-0073871), the National msh083) Institutes of Health to S.K. (HG002096) and the Arizona Fischer, A., Pollack, J., Thalmann, O., Nickel, B. & Paabo, State University. S. 2006 Demographic history and genetic differentiation in apes. Curr. Biol. 16, 1133–1138. (doi:10.1016/j.cub. 2006.04.033) REFERENCES Gagneux, P. et al. 1999 Mitochondrial sequences show Anderson, S. et al. 1981 Sequence and organization of the diverse evolutionary histories of African hominoids. human mitochondrial genome. Nature 290, 457–465. Proc. Natl Acad. Sci. USA 96, 5077–5082. (doi:10. (doi:10.1038/290457a0) 1073/pnas.96.9.5077) Andrews, R. M., Kubacka, I., Chinnery, P. F., Lightowlers, Gagneux, P., Gonder, M. K., Goldberg, T. L. & Morin, P. A. R. N., Turnbull, D. M. & Howell, N. 1999 Reanalysis 2001 Gene flow in wild chimpanzee populations: what and revision of the Cambridge reference sequence for genetic data tell us about chimpanzee movement human mitochondrial DNA. Nat. Genet. 23, 147. over space and time. Phil. Trans. R. Soc. Lond. B 356, Arnason, U., Gullberg, A. & Xu, X. 1996 A complete mito- 889–897. (doi:10.1098/rstb.2001.0865) chondrial DNA molecule of the white-handed gibbon, Glazko, G. V. & Nei, M. 2003 Estimation of divergence Hylobates lar, and comparison among individual mito- times for major lineages of primate species. Mol. Biol. chondrial genes of all hominoid genera. Hereditas 124, Evol. 20, 424–434. (doi:10.1093/molbev/msg050) 185–189. (doi:10.1111/j.1601-5223.1996.00185.x) Gonder, M. K., Oates, J. F., Disotell, T. R., Forstner, Bahlo, M. & Griffiths, R. C. 2000 Inference from genetrees M. R. J., Morales, J. C. & Melnick, D. J. 1997 A new in a subdivided population. Theor. Popul. Biol. 57, 79–95. west African chimpanzee subspecies? Nature 388, 337. (doi:10.1006/tpbi.1999.1447) (doi:10.1038/41005) Ballard, J. W. O. & Rand, D. M. 2005 The population Gonder, M. K., Disotell, T. R. & Oates, J. F. 2006 New biology of mitochondrial DNA and its phylogenetic impli- genetic evidence on the evolution of chimpanzee popu- cations. Annu. Rev. Ecol. Evol. Syst. 36, 621–642. lations and implications for taxonomy. Int. J. Primatol. (doi:10.1146/annurev.ecolsys.36.091704.175513) 27, 1103–1127. (doi:10.1007/s10764-006-9063-y) Becquet, C. & Przeworski, M. 2007 A new approach to esti- Goodall, J. 1986 The chimpanzees of Gombe: patterns of mate parameters of speciation models with application to behavior. Cambridge, UK: Harvard University Press.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Divergence times in Pan A. C. Stone et al. 3287

Griffiths, R. C. & Tavare´, S. 1994 Simulating probability Mol. Ecol. 14, 2181–2196. (doi:10.1111/j.1365-294X. distributions in the coalescent. Theor. Popul. Biol. 46, 2005.02560.x) 131–159. (doi:10.1006/tpbi.1994.1023) Lynch, M. & Crease, T. J. 1990 The analysis of population Guy, F., Brunet, M., Schmittbuhl, M. & Viriot, L. 2003 survey data on DNA sequence variation. Mol. Biol. New approaches in hominoid taxonomy: morphometrics. Evol. 7, 377–394. Am. J. Phys. Anthropol. 121, 198–218. (doi:10.1002/ajpa. Maca-Meyer, N., Gonzalez, A. M., Larruga, J. M., Flores, 10261) C. & Cabrera, V. M. 2001 Major genomic mitochondrial Hedges, S. B., Dudley, J. & Kumar, S. 2006 TimeTree: a lineages delineate early human expansions. BMC Genet. public knowledge-base of divergence times among 2, 13. (doi:10.1186/1471-2156-2-13) organisms. Bioinformatics 22, 2971–2972. (doi:10.1093/ Macaulay, V. et al. 2005 Single, rapid coastal settlement of bioinformatics/btl505) Asia revealed by analysis of complete mitochondrial Herrnstadt, C. et al. 2002 Reduced-median-network genomes. Science 308, 1034–1036. (doi:10.1126/ analysis of complete mitochondrial DNA coding-region science.1109792) sequences for the major African, Asian, and European Mishmar, D. et al. 2003 Natural selection shaped regional haplogroups. Am. J. Hum. Genet. 70, 1152–1171. mtDNA variation in humans. Proc. Natl Acad. Sci. USA (doi:10.1086/339933) 100, 171–176. (doi:10.1073/pnas.0136972100) Hey, J. 2010 The divergence of chimpanzee species and Morin, P. A., Moore, J. J., Chakraborty, R., Jin, L., Goodall, subspecies as revealed in multi-population isolation- J. & Woodruff, D. S. 1994 Kin selection, social with-migration analyses. Mol. Biol. Evol. 27, 921–933. structure, gene flow, and the evolution of (doi:10.1093/molbev/msp298) chimpanzees. Science 265, 1193–1201. (doi:10.1126/ Horai, S., Satta, Y., Hayasaka, K., Kondo, R., Inoue, T., science.7915048) Ishida, T., Hayashi, S. & Takahata, N. 1992 Man’s Nachman, M. W., Brown, W. M., Stoneking, M. & Aquadro, place in the hominoidea revealed by mitochondrial C. F. 1996 Nonneutral mitochondrial DNA variation in DNA genealogy. J. Mol. Evol. 35, 32–43. (doi:10.1007/ humans and chimpanzees. Genetics 142, 953–963. BF00160258) Neefs, J. M., Van de Peer, Y., De Rijk, P., Chapelle, S. & Horai, S., Hayasaka, K., Kondo, R., Tsugane, K. & De Wachter, R. 1993 Compilation of small ribosomal Takahata, N. 1995 Recent African origin of modern subunit RNA structures. Nucleic Acids Res. 21, 3025– humans revealed by complete sequences of hominoid 3049. (doi:10.1093/nar/21.13.3025) mitochondrial DNAs. Proc. Natl Acad. Sci. USA 92, Nei, M. 1987 Molecular evolutionary genetics. New York, NY: 532–536. (doi:10.1073/pnas.92.2.532) Columbia University Press. Ingman, M. & Gyllensten, U. 2003 Mitochondrial genome Nishida, T. 1979 The social structure of chimpanzees variation and evolutionary history of Australian and of the Mahale Mountains. In The great apes (eds D. A. New Guinean aborigines. Genome Res. 13, 1600–1606. Hamburg & E. R. McCown), pp. 73–121. Menlo Park, (doi:10.1101/gr.686603) CA: Benjamin/Cummings. Ingman, M., Kaessmann, H., Paä¨bo, S. & Gyllensten, U. Oates, J. F., Groves, C. P. & Jenkins, P. D. 2009 The type 2000 Mitochondrial genome variation and the origin of locality of Pan troglodytes vellerosus (Gray 1862), and modern humans. Nature 408, 708–713. (doi:10.1038/ implications for the nomenclature of west African 35047064) chimpanzees. Primates 50, 78–80. (doi:10.1007/s10329- Inoue, E., Inoue-Murayama, M., Vigilant, L., Takenaka, O. & 008-0116-z) Nishida, T. 2008 Relatedness in wild chimpanzees: influ- Posada, D. & Crandall, K. A. 1998 MODELTEST: testing the ence of paternity, male philopatry, and demographic model of DNA substitution. Bioinformatics 14, 817–818. factors. Am.J.Phys.Anthropol.137,256–262.(doi:10. (doi:10.1093/bioinformatics/14.9.817) 1002/ajpa.20865) Pusey, A. 1979 Intercommunity transfer of chimpanzees Jukes, T. H. & Cantor, C. R. 1969 Evolution of protein mol- in Gombe National Park. In The great apes (eds D. A. ecules. In Mammalian protein metabolism (ed. H. N. Hamburg & E. R. McCown), pp. 465–479. Menlo Munro), pp. 21–132. New York, NY: Academic Press. Park, CA: Benjamin/Cummings. Kaessmann, H., Wiebe, V. & Paä¨bo, S. 1999 Extensive Pusey, A. E. & Packer, C. 1986 Dispersal and philopatry. nuclear DNA sequence diversity among chimpanzees. In Primate societies (eds B. B. Smuts, D. L. Cheney, Science 286, 1159–1162. (doi:10.1126/science.286. R. M. Seyfarth, R. W. Wrangham & T. T. Struhsaker), 5442.1159) pp. 250–266. Chicago, IL: University of Chicago Press. Kivisild, T. et al. 2006 The role of selection in the evolution Raaum, R. L., Sterner, K. N., Noviello, C. M., Stewart, of human mitochondrial genomes. Genetics 172, C. B. & Disotell, T. R. 2005 Catarrhine primate diver- 373–387. (doi:10.1534/genetics.105.043901) gence dates estimated from complete mitochondrial Kumar, S., Filipski, A., Swarna, V., Walker, A. & Hedges, genomes: concordance with fossil and nuclear DNA S. B. 2005 Placing confidence limits on the molecular evidence. J. Hum. Evol. 48, 237–257. (doi:10.1016/j. age of the human–chimpanzee divergence. Proc. Natl jhevol.2004.11.007) Acad. Sci. USA 102, 18 842–18 847. (doi:10.1073/pnas. Rand, D. M. & Kann, L. M. 1996 Excess amino acid poly- 0509585102) morphism in mitochondrial DNA: contrasts among genes Langergraber, K. E., Siedel, H., Mitani, J. C., Wrangham, from Drosophila, mice, and humans. Mol. Biol. Evol. 13, R. W., Reynolds, V., Hunt, K. & Vigilant, L. 2007 The 735–748. genetic signature of sex-biased migration in patrilocal Rozas, J., Sanchez-DelBarrio, J. C., Messeguer, X. & Rozas, chimpanzees and humans. PLoS ONE 2, e973. (doi:10. R. 2003 DNASP, DNA polymorphism analyses by the 1371/journal.pone.0000973) coalescent and other methods. Bioinformatics 19, 2496– Lockwood, C. A., Kimbel, W. H. & Lynch, J. M. 2004 2497. (doi:10.1093/bioinformatics/btg359) Morphometrics and hominoid phylogeny: support for a Sambrook, J. & Russell, D. 2001 Molecular cloning: a labora- chimpanzee–human clade and differentiation among tory manual. Cold Spring Harbor, NY: Cold Spring great ape subspecies. Proc. Natl Acad. Sci. USA 101, Harbor Laboratory Press. 4356–4360. (doi:10.1073/pnas.0306235101) Sarich, V. M. & Wilson, A. C. 1973 Generation time and Lukas, D., Reynolds, V., Boesch, C. & Vigilant, L. 2005 To genomic evolution in primates. Science 179, 1144–1147. what extent does living in a group mean living with kin? (doi:10.1126/science.179.4078.1144)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3288 A. C. Stone et al. Divergence times in Pan

Schrago, C. G. & Russo, C. A. 2003 Timing the origin of Verrelli, B. C., Tishkoff, S. A., Stone, A. C. & Touchman, New World monkeys. Mol. Biol. Evol. 20, 1620–1625. J. W. 2006 Contrasting histories of G6PD molecular (doi:10.1093/molbev/msg172) evolution and malarial resistance in humans and Stone, A. C. & Verrelli, B. C. 2006 Focusing on comparative chimpanzees. Mol. Biol. Evol. 23, 1592–1601. (doi:10. ape population genetics in the post-genomic age. Curr. 1093/molbev/msl024) Opin. Genet. Dev. 16, 586–591. (doi:10.1016/j.gde.2006. Vigilant, L., Hofreiter, M., Siedel, H. & Boesch, C. 2001 09.003) Paternity and relatedness in wild chimpanzee commu- Stone, A. C., Grifﬁths, R. C., Zegura, S. L. & Hammer, nities. Proc. Natl Acad. Sci. USA 98, 12 890–12 895. M. F. 2002 High levels of Y-chromosome nucleotide (doi:10.1073/pnas.231320498) diversity in the genus Pan. Proc. Natl Acad. Sci. USA Wakeley, J. & Hey, J. 1997 Estimating ancestral population 99, 43–48. (doi:10.1073/pnas.012364999) parameters. Genetics 145, 847–855. Sun, C. et al. 2006 The dazzling array of basal branches in Watterson, G. A. 1975 On the number of segregating the mtDNA macrohaplogroup M from India as inferred sites in genetical models without recombination. from complete genomes. Mol. Biol. Evol. 23, 683–690. Theor. Popul. Biol. 7, 256–276. (doi:10.1016/0040- (doi:10.1093/molbev/msj078) 5809(75)90020-9) Swofford, D. L. 2003 PAUP *. Phylogenetic analysis using Wegmann, D. & Excofﬁer, L. 2010 Bayesian inference of the parsimony (*and other methods). Sunderland, MA: demographic history of chimpanzees. Mol. Biol. Evol. 27, Sinaur Associates. 1425–1435. (doi:10.1093/molbev/msq028) Tajima, F. 1989 Statistical method for testing the neutral White, T. D., Suwa, G. & Asfaw, B. 1994 Australopithicus muation hypothesis by DNA polymorphism. Genetics ramidus, a new species of early hominid from 123, 585–595. Aramis, Ethiopia. Nature 371, 306–312. (doi:10.1038/ Tamura, K. & Nei, M. 1993 Estimation of the number of 371306a0) nucleotide substitutions in the control region of mito- Won, Y. J. & Hey, J. 2005 Divergence population genetics of chondrial DNA in humans and chimpanzees. Mol. Biol. chimpanzees. Mol. Biol. Evol. 22, 297–307. (doi:10. Evol. 10, 512–526. 1093/molbev/msi017) Tamura, K., Dudley, J., Nei, M. & Kumar, S. 2007 Molecu- Wrangham, R. W. 1979 Sex differences in chimpanzee dis- lar evolutionary genetics analysis (MEGA) software persion. In The great apes (eds D. A. Hamburg & E. R. version 4.0. Mol. Biol. Evol. 24, 1596–1599. (doi:10. McCown), pp. 481–489. Menlo Park, CA: Benjamin/ 1093/molbev/msm092) Cummings. Thalmann, O., Hebler, J., Poinar, H. N., Paabo, S. & Xu, X. & Arnason, U. 1996 The mitochondrial DNA Vigilant, L. 2004 Unreliable mtDNA data due to nuclear molecule of Sumatran orangutan and a molecular insertions: a cautionary tale from analysis of humans and proposal for two (Bornean and Sumatran) species of other great apes. Mol. Ecol. 13, 321–335. (doi:10.1046/j. orangutan. J. Mol. Evol. 43, 431–437. (doi:10.1007/ 1365-294X.2003.02070.x) BF02337514) Thangaraj, K., Chaubey, G., Kivisild, T., Reddy, A. G., Yang, Z. 2007 PAML 4: phylogenetic analysis by maximum Singh, V. K., Rasalkar, A. A. & Singh, L. 2005 Recon- likelihood. Mol. Biol. Evol. 24, 1586–1591. (doi:10.1093/ structing the origin of Andaman Islanders. Science 308, molbev/msm088) 996. (doi:10.1126/science.1109987) Yu, N., Jensen-Seaman, M. I., Chemnick, L., Kidd, J. R., Thorne, J. L. & Kishino, H. 2002 Divergence time and Deinard, A. S., Ryder, O., Kidd, K. K. & Li, W. H. evolutionary rate estimation with multilocus data. Syst. 2003 Low nucleotide diversity in chimpanzees and Biol. 51, 689–702. (doi:10.1080/10635150290102456) bonobos. Genetics 164, 1511–1518.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Phil. Trans. R. Soc. B (2010) 365, 3289–3299 doi:10.1098/rstb.2010.0112

Spinopelvic pathways to bipedality: why no hominids ever relied on a bent-hip–bent-knee gait C. Owen Lovejoy1,* and Melanie A. McCollum2 1Department of Anthropology, School of Biomedical Sciences, Kent State University, OH, USA 2Department of Cell Biology, University of Virginia, VA, USA Until recently, the last common ancestor of African apes and humans was presumed to resemble living chimpanzees and bonobos. This was frequently extended to their locomotor pattern leading to the presumption that knuckle-walking was a likely ancestral pattern, requiring bipedality to have emerged as a modiﬁcation of their bent-hip-bent-knee gait used during erect walking. Research on the development and anatomy of the vertebral column, coupled with new revelations from the fossil record (in particular, Ardipithecus ramidus), now demonstrate that these presumptions have been in error. Reassessment of the potential pathway to early hominid bipedality now reveals an entirely novel sequence of likely morphological events leading to the emergence of upright walking. Keywords: Australopithecus; bipedality; bent-hip–bent-knee; Ardipithecus; human evolution

1. INTRODUCTION now considerable evidence indicating that they have For several decades, largely subsequent to the recovery not been retained from the common ancestor shared of A.L.288-1 (‘Lucy’) (Johanson et al. 1982), upright with the human clade. Instead, a more detailed study walking in early hominids was argued to have relied on of the vertebral formulae and the lumbar column of a bent-hip–bent-knee (BHBK) gait (see, e.g. Stern & African apes and early hominids indicates that the Susman 1983; Susman et al. 1984; Stern 2000). This LCA of Pan and Homo most probably possessed a argument rested on observations of locomotion in long (six to seven segments) mobile lumbar spine simi- chimpanzees and gorillas, coupled with the presump- lar in number to those of Old World monkeys (OWMs), tion that the post-cranium of our last common Proconsul and Nacholapithecus (McCollum et al.2009). ancestor (LCA) of Pan and Homo was fundamentally Because such columns would have been capable of similar to those of extant African apes (but see Filler near-full lordosis, these new findings in and of them- 1981). Despite the fact that early hominids such as selves contraindicate pronounced African ape-like A.L.288-1 (and other members of Australopithecus BHBK bipedality in early hominids. New revelations afarensis and Australopithecus anamensis) exhibit about LCA structure provided by Ardipithecus ramidus pelves, knees and feet with highly advanced adap- (especially ARA-VP-6/500; Lovejoy et al. 2009a–d; tations to a striding, bipedal gait (Latimer & Lovejoy White et al.2009) further establish that hominids 1989; Lovejoy 2005a,b, 2007), the BHBK hypothesis never displayed any of the numerous African ape-like has remained largely unchallenged save arguments specializations that have reduced lumbar mobility and based on energy consumption (e.g. Crompton et al. thus required an unusually restricted BHBK gait. 1998; Carey & Crompton 2005; Sellers et al. 2005). Here we review this new evidence. The BHBK gait of Pan and Gorilla, however, is not a function of limitations imposed by hip or knee anatomy, but is instead a direct consequence of an absence of 2. THE AXIAL PATTERN OF THE LCA lumbar spine mobility. African apes are unable to lor- As is discussed more fully in McCollum et al.(2009), it dose their lumbar spines, and therefore must flex both is reasonable to assume that the modal vertebral for- the hip and knee joints in order to position their mula of basal hominoids and the LCA of Pan and centre of mass over the point of ground contact Homo to have been 7-13-6/7-4–one that differs from (Lovejoy 2005a). Lumbar immobility in Pan and those of OWMs merely by the addition of a fourth Gorilla is a consequence of their possession of only sacral vertebra, and replacement of the external tail by three to four lumbar vertebrae and the ‘entrapment’ a short coccyx. Two lines of evidence support this view. of the most caudal lumbar vertebra(e) between cranially First is evidence provided by the vertebral formulae extended ilia (Stevens 2004; Stevens & Lovejoy 2004; of Australopithecus and early Homo (Sanders 1995). Lovejoy 2005a; McCollum et al. 2009). Although all Although complete axial data are unavailable for any three African ape species share these features, there is single early hominid specimen, a number of partial specimens, including A.L. 288-1 (complete sacrum) * Author for correspondence ([email protected]). and KNM-WT 15000 (interpretable lumbar One contribution of 14 to a Discussion Meeting Issue ‘The first four column), indicate a pre-coccygeal vertebral formula million years of human evolution’. of 7-12/13-6-4 (Pilbeam 2004; McCollum et al. 2009).

3289 This journal is q 2010 The Royal Society Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3290 C. O. Lovejoy & M. A. McCollum Spinopelvic pathways to bipedality

Gorilla P. troglodytes P. paniscus H. sapiens 7-13-4-5 7-13-4-6 7-13-4-7 7-12-5-5 7-13-4-6 7-13-4-5 7-13-4-6 7-12-5-6 7-13-3-6 7-13-3-6 7-14-4-6 T11 T11 T11 T10 T12 T12 T12 T11 T13 T13 T13 T12 L1 L1 T14/L1 L1 L2 L2 L2 L2 L3 L3 L3 Australopithecus L3 L4 L4 L4 7-12-6-4 L4 S1 S1 S1 7-12-6-5 L5 S2 S2 S2 7-13-6-4 S1 S3 S3 S3 T11 S2 S4 S4 S4 S3 S5 S5 S5 T12 S4 S6 LCA L1 S5 7-12-6-5 L2 7-13-6-4 L3 7-13-6-5 L4 L5 T11 L6 T12 S1 T13 S2 L1 S3 L2 S4 L3 L4 L5 L6 S1 S2 S3 S4 Figure 1. Probable pathways of lumbar reduction in African apes and hominids as deduced from extant vertebral formulae for each taxon. All axial formulae that exceed 10% of the total sample for each taxon are shown here, along with presumed modal formulae (those of highest probable frequencies) for the LCA and early hominids. A horizontal arrow indicates loss of a body segment (i.e. a reduction in the number of somites contributing to the pre-coccygeal vertebral column). A vertical arrow sig- niﬁes changes in the positions of the anterior boundaries of Hox gene expression domains underlying the indicated transformations of vertebral identities. Note the differences in extant Pan species. For details, see McCollum et al. (2009). Axial formula data from Pilbeam (2004) and McCollum et al. (2009).(q M.A. McCollum).

The second source of evidence is the axial caudal-most lumbar vertebrae plus reduction in the morphology of bonobos (Pan paniscus). Unlike chim- number of somites (figure 1). Bonobos, conversely, panzees (Pan troglodytes) and modern humans, whose appear to have reduced their lumbar column purely modal number of pre-coccygeal vertebrae is 29/30, through transformations of segment identity, i.e. by bonobos possess an axial column typically composed transforming lumbar vertebrae into sacral and thoracic of 30/31 vertebrae, identical to that inferred for the vertebrae (McCollum et al. 2009). basal hominoid. Although it is certainly possible that the long axial column of bonobos re-evolved from an ancestor with an abbreviated column similar to that 3. THE LOCOMOTOR SKELETON OF of chimpanzees and modern humans, such modifi- ARDIPTHECUS RAMIDUS cation has no obvious selective advantage and runs The Ar. ramidus limb skeleton indicates that much of counter to the trend towards axial length reduction extant African ape locomotor anatomy has been inde- observed in all suspensory anthropoids (Benton pendently derived for vertical climbing, suspension 1967; McCollum et al. 2009). Rather, the long axial and a feeding habitus that probably included high column of bonobos, along with the significantly differ- canopy access in relatively large-bodied hominoids ent combinations of sacral, lumbar and thoracic (Lovejoy et al. 2009a). While OWMs also frequently vertebrae that are characteristic of common chimpan- climb vertically, they nevertheless retain adaptations zees (seven cervical, 13 thoracic, three to four lumbar, that are primarily for more active, above-branch pro- five to six sacral) and bonobos (seven cervical, 13–14 nograde running and leaping. Such acrobatics appear thoracic, four lumbar, six to seven sacral), suggest to have become much more limited in hominoids, pre- instead that the two species of Pan evolved their sumably inter alia, because of their significantly larger short lumbar spines from an ancestor with a long body mass (Cartmill 1985). axial column (n ¼ 30/31 pre-coccygeal vertebrae) The locomotor skeleton of Ar. ramidus establishes and a long lumbar spine after division from their that the LCA, unlike modern apes, retained many own LCA. This receives support from data which OWM-like features sufficiently primitive to assure a suggest that lumbar spine reduction in chimpanzees primary gait pattern of above-branch pronograde apparently occurred through sacralization of the palmigrady (Lovejoy et al. 2009a–c). To be sure,

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Spinopelvic pathways to bipedality C. O. Lovejoy & M. A. McCollum 3291 numerous modifications of OWM-like anatomy had Filler 2007a,b), and was significantly progressing in a become more like that of extant hominoids in the number of forms by the Mid-Miocene (e.g. in Pierola- LCA (Lovejoy et al. 2009c)—alterations known to pithecus by at least 10 Ma; Moya-Sola et al. 2004; have been initiated in likely exemplars of its remote Almecija et al. 2009). This shift appears to have ancestors, especially various species of Proconsul accompanied other forelimb modifications, especially (Ward 1991, 1993; Ward et al. 1991, 1993; ulnar withdrawal and olecranon abbreviation. These Nakatsukasa et al. 2003). However, the Ar. ramidus modifications increased potential wrist adduction, foot still retained a relatively elongated mid-tarsus, a enhanced stability during complete elbow extension robust os peroneum complex and presumably numerous and greatly increased the forelimb’s range of motion soft tissue features associable with an inherently stiff at the shoulder girdle (Rose 1988; Lewis 1989). How- plantar structure more typical of the above-branch ever, as this change in bauplan also resulted in the propulsion seen in OWMs. These latter features can sacrifice of substantial erector spinae mass (Benton be reliably extended to the LCA by parsimony, since 1967; Lovejoy 2005a), increasing the range of they are still present in the feet of modern humans motion of the shoulder came at the expense of (quadratus plantae, plantaris, os peroneum, elongated dynamic stabilization of the lower back. Consequently, cuboid, etc.), but have been largely eclipsed by special- African ape suspension and vertical climbing required izations in the feet and ankles of more highly compensatory lumbar column reduction—virtually to specialized, extant African apes (Desilva 2009; the point of inherent (i.e. osteological) rather than Lovejoy et al. 2009a). dynamic (i.e. muscular) rigidity. Thus, LTP position, Similar observations of the Ar. ramidus forelimb rather than being the primary target of selection suggest that it also shares a number of primitive during lumbar column shortening, as has long been features with humans. These include a very primitive argued (e.g. Benton 1967), was instead a product of and unreinforced central joint complex (CJC) the fundamental change in the hominoid (capitate, trapezoid, metacarpals 2 and 3), a relatively bauplan that centred about a general restructuring of substantial pollex, a short metacarpus, a lack of the thorax. significant Mc4/Mc5–hamulus contact, a narrow trapezoid, a palmarly displaced capitate head and an unmodified, markedly rugose, deltopectoral crest. 5. LOCOMOTION IN THE LCA Each of these has since been modified in extant Features assignable to the LCA, therefore, now point large-bodied African apes in favour of ones associable to a pattern of cautious climbing that combined with knuckle-walking, suspension and/or vertical above-branch palmigrady with occasional below- climbing (Lovejoy et al. 2009c). branch suspension, enhanced by a highly mobile, later- alized shoulder girdle in combination with marked wrist adduction (Cartmill & Milton 1977; Lewis 4. THORACOABDOMINAL STRUCTURE 1989) and elbow extension (Rose 1988). Below- AND FORELIMB MOBILITY IN THE LCA branch suspension, however, must not have been so At the same time, it is equally clear that the LCA dif- frequently employed (and/or so vigorously performed) fered fundamentally from its likely ancestors as to require emergence of the considerably more (including Proconsul) in several major ways, none advanced metacarpal, carpal, elbow and shoulder more important than the structure of its vertebral modifications seen only extant African apes. This column and its position within the thorax (Lovejoy suggests that much of the LCA’s activities may have et al. 2009c). In comparison with Proconsul and been largely low-canopy, and might have been com- OWMs, in which the pectoral girdle is positioned bined, possibly extensively, with terrestrial travel more anteriorly on the thoracic cage, the hominoid between food patches (White et al. 2009). The latter pectoral girdle is located more dorsolaterally, in a supposition receives support from the fact that the manner that causes its glenoid fossa to face more adaptations to terrestrial travel present in extant Afri- laterally than is typical of more primitive taxa can apes (knuckle-walking) and fossil hominids (Waterman 1929; Schultz 1961; Erikson 1963; Ward (bipedality) are extensive, fundamentally divergent, 2007). Such ‘posterolateralization’ places the girdle and therefore likely to be of substantial antiquity. It into a more favourable position for circumduction, is also likely that reliance on terrestrial travel between which in turn permits relatively large-bodied primates food patches was driven by increasing competition to successfully negotiate the canopy via clambering, with radiating OWMs in the Mid-Miocene (Andrews bridging and suspension. What has gone almost 1981). That the post-crania of Pongo and the lesser entirely unrecognized until the recovery of Ar. ramidus, apes (Hylobates, Symphalangus) differ substantially however, is that repositioning of the scapula (so as to from those of the African apes is probably largely make the glenoid face more laterally and less ante- due to the absence of a significant terrestrial com- riorly) in hominoids was achieved by thoracic ponent in their respective adaptive strategies, and reorganization which relied on invagination of the their entirely independent evolution from much more post-cervical spine ventrally into the thorax. This primitive ancestors. resulted in dorsal repositioning of the lumbar trans- If the above hypothesis is correct, what was the verse processes (LTPs), a change in bauplan that LCA’s terrestrial locomotor habitus prior to the emer- apparently occurred independently and repeatedly gence of either knuckle-walking in apes or bipedality in even in some early Miocene hominoid taxa (e.g. in hominids? One possible pattern ‘of choice’ might have Morotopithecus by 17 Ma; MacLatchy et al. 2000; been a simple extension of its primary arboreal pattern

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3292 C. O. Lovejoy & M. A. McCollum Spinopelvic pathways to bipedality to ground travel, i.e. palmigrade quadrupedality. In resolved by adoption of knuckle-walking, which per- fact, some of the more unusual characters present in mits reliance on substrate-forced dorsiflexion of the Ar. ramidus are strongly suggestive that hominids wrist that can be eccentrically resisted by powerful once exhibited such an ancestral gait pattern. These wrist flexors as well as both the connective tissue envel- include its primitive intermembral index, relatively opes and contractile components of the long digital short metacarpus, allowance of substantial metacar- flexors. The uniqueness of these long flexors is evi- pal–phalangeal dorsiflexion and especially the denced by development of a distinctive flexor strongly palmar positioning of the head of its capitate tubercle on the proximal ulna in both Pan and Gorilla (Lovejoy et al. 2009b). Indeed, the latter can be (Lovejoy et al. 2009b). viewed as being particularly advantageous to palmi- The most salient question remaining, of course, is grade terrestrial quadrupedality, and this would now the issue of the eventual adoption of bipedality in seem to be a possible explanation for this unusual hominids. Why did hominids exchange palmigrade/ peculiarity in Ar. ramidus, i.e. it inherited it from a plantigrade quadrupedality for upright walking? habitually terrestrial palmigrade LCA. While there have been many theories advanced for Unlike OWMs, quadrupedal terrestrial gait in this locomotor shift, most have been made untenable large-bodied hominoids (including the LCA) may by evidence now provided by Ar. ramidus (White have required a much more compliant wrist, i.e. pal- et al. 2009). The recent suggestion that bipedality is migrady that included more extreme dorsiflexion. a sequel to an arboreal upright stance stabilized by Absence of such extreme adaptations in OWMs is overhead forelimb grasping (Thorpe et al. 2007a,b, likely explicable by their retention of primary 2009) is untenable because the practice has emerged above-branch adaptations at the radiocarpal, elbow in Pongo as a consequence of that taxon’s extreme and shoulder joints. Palmar disposition of the capi- adaptations to suspension, none of which were ever tate head as seen in Ar. ramidus (Lovejoy et al. present in hominids or their ancestors (Lovejoy et al. 2009b) may even now serve, given further fossil evi- 2009c). The most likely explanation for the adoption dence, as an indicator of palmigrade/plantigrade of terrestrial bipedality, in our view, continues to terrestrial quadrupedality in yet undiscovered, large- involve novel adaptations in hominid social structure bodied, Miocene forms. In combination with the that required upright locomotion for carrying. These retention of a long mid-tarsus, a robust os peroneal have been discussed extensively elsewhere (Lovejoy complex and other primitive aspects of its foot 1981, 1993, 2009). (retained M. quadratus plantae, retained M. plantaris and associated dense palmar fascial aponeurosis (see earlier)), palmigrade/plantigrade quadrupedality 7. SOME ADDITIONAL ANATOMICAL seems to have been, at the least, a likely terrestrial CORRELATES OF BIPEDALITY IN HOMINIDS AND locomotor habitus in the African ape/hominid LCA ADVANCED ARBOREALITY IN EXTANT APES (Lovejoy et al. 2009c). As noted above, still equipped with a mobile lumbar spine, the LCA was probably capable of at least facultative lordosis, sufficient to place its hip and knee 6. LOCOMOTOR SPECIALIZATIONS either directly below its centre of mass or sufficiently IN EXTANT HOMINOIDS close to that centre so as not to generate excessive If so, from whence came the relatively highly special- ground-reactive torques so large as to require debilitat- ized gait patterns of the LCA’s descendants: ing muscle recruitment during terrestrial travel. While bipedality in hominids and knuckle-walking in the the earliest hominid gait pattern probably required African apes? The latter is reasonably explicable in some degree of hip and knee flexion, research on these taxa as a relatively facile means of modifying pal- bipedality in OWMs now suggests that it would not migrade/plantigrade terrestrial travel into a form that have been nearly as excessive as it is in extant apes, could be successfully combined with their highly so long as the lumbar column remained long and specialized modifications of the pelvis, thorax and mobile (as it is in OWMs). Studies of OWMs have limb skeletons for suspension and vertical climbing. now greatly illuminated our understanding of its ori- Such changes included (independently in each gins in hominids (Nakatsukasa et al. 1995, 2004, taxon) elongation of the forelimb, abbreviation of the 2006; Hirasaki et al. 2004). Macaques trained to hindlimb, elongation of the metacarpus, stabilization walk bipedally expend less energy than do those in of several major carpal joints either by ligamentous which the behaviour is novel, so much so that the ani- reinforcement or joint enlargement or both (especially mal’s long flexible spine is permissive for convergence in the CJC), major revisions of overall scapular mor- with ‘human-style’ walking in the former. While those phology (predominantly in Pan as opposed to using bipedality ‘in the wild’ exhibit upright gaits that Gorilla), cranial retroflexion of the ulnar trochlear differ kinesiologically from human walking, those notch, modification of the deltopectoral enthesis and exposed to long-term training for bipedality walk especially, virtual fusion of the thorax and pelvis via ‘with longer, less-frequent strides, more extended abbreviation and iliac fixation of the lumbar column hindlimb joints, double-phase joint motion at the (see earlier) (Lovejoy et al. 2009c,d ). All of the modi- knee joint, and most importantly, efficient energy fications to the forelimb would have reduced its transformation by using inverted pendulum mech- inherent stability and increasingly restricted its anics’ (Hirasaki et al. 2004: 748). energy-dissipating capacity during prolonged terres- While lordosis was certainly facilitated by the pres- trial travel. These difficulties appear to have been ence of six to seven lumbar vertebrae in the LCA

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Spinopelvic pathways to bipedality C. O. Lovejoy & M. A. McCollum 3293

4.25 4.25

4.00 4.00

3.75 3.75

log alar breadth 3.50 Homo log alar breadth 3.50 Pan Homo Gorilla Pan Gorilla 3.25 A.L. 288-1 BSN49/P27 3.25 A.L. 288-1 KSD-VP-1/1 BSN49/P27 STS-14 KSD-VP-1/1 3.00 human mean STS-14 3.00 human mean 4.20 4.50 4.80 6.00 6.25 6.50 6.75 7.00 7.25 7.50 7.75 log total sacral breadth log centrum area Figure 2. Components of sacral breadth in African apes Figure 3. Components of sacral breadth in African apes and and early hominids. Scatter plot of log total sacral breadth early hominids. Scatter plot of log centrum area (length Â versus log alar breadth. The findings of a strong correlation breadth) versus log alar breadth. For discussion, see legend (r ¼ 0.901) between sacral breadth and alar breadth, and of figure 2. an absence of any significant correlation between alar breadth and centrum area (figure 3) indicate that total sacral breadth in African apes and early hominids is largely a consequence of alar breadth. Note especially the exceed- ingly broad alae in the early hominid specimens. This is and largely exaptive for increased abductor capacity consistent with their having mediolaterally expanded the during the single-support phase of upright walking sacrum as an adaptation to free the most caudal lumbar (Lovejoy et al. 2009c). from contact with the iliac crest, and fully accounts for the Moreover, three features of ape sacra appear to have unusual platypelloidy in A.L.288-1 and Sts-14. Later, a directly opposite polarity compared with those of caudally directed gradient of increasing centrum size and hominids (figures 2–4): (i) their strongly abbreviated interfacet distance (see text) appears in Homo, and probably sacral alae, (ii) their reduced lumbar number, and accounts for the unusually large human centrum. Note the (iii) their greater number of fused sacral elements, much narrower alae in the two African apes compared with those of all hominids. the latter almost certainly achieved by progressive sacralization of the most caudal lumbar element(s) (McCollum et al. 2009). Alar reduction reduces the (most probably six; figure 1), even in most OWMs space between the two ilia so as to promote contact lordosis is not as complete as it is in five-lumbared with the most caudal lumbar vertebra(e). In combi- humans (Nakatsukasa et al. 1995; Hirasaki et al. nation with the additional extension of the ilia 2004). One probable and very important reason is superiorly (especially by elongation of the iliac isthmus that complementary motion in the most caudal at least in Pan; Lovejoy et al. 2009c), the African apes lumbar vertebra in OWMs is usually restricted by its achieved stabilization of the entire lower spine by its proximity to the posterosuperior portion of each iliac fixation to the thorax—creating a rigid pelvothoracic blade. Such iliac–lumbar propinquity is usually suffi- ‘block’ in which the pelvis and thorax are separated cient to probably assure at least some degree of by a distance of only a single intercostal space (Schultz ligamentous restriction of potential motion. 1961). Both mechanisms compensate for the loss of Two characters that are uniquely associated with erector spinae mass (see earlier). Thus, both sacral hominid pelvic adaptations to bipedality are therefore structure and superoinferior iliac length directly reflect of particular interest: (i) an exceptionally short super- hominoid post-cranial natural history. Panids and oinferior iliac height (coupled with both anterior gorillids independently elongated their ilia, narrowed extension of the anterior inferior iliac spine (AIIS) their bi-iliac spaces and reduced the number of and development of the greater sciatic notch) and (ii) lumbar vertebrae (often by sequestration as additional an extremely wide sacrum generated largely by excep- sacral segments), all mechanisms that stiffened the tionally broad sacral alae (figures 2 and 3). Both of lower back and eliminated any possibility of lordosis. these characters eliminate contact between the pos- Hominids, per contra, remained almost entirely plesio- teromost iliac crest and the most caudal lumbar morphic, retaining both the primitive number of vertebra, and are therefore likely to have appeared lumbar (mode ¼ six) and sacral vertebrae (mode ¼ early in hominids as a means of increasing the lordotic four), and in addition, expanded the sacral alae so as capacity of the lumbar spine during terrestrial bipedal- to assure the full independence of the most caudal ity. Indeed, these changes are likely to have been the lumbar, assuring its freedom to participate in lordosis earliest in the evolution of bipedality in hominids, as well.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3294 C. O. Lovejoy & M. A. McCollum Spinopelvic pathways to bipedality

straightforward morphological characters of greater inherent reliability, and which are more resistant to 2 misinterpretation from crushing defects. Not all of these appear to have been considered. 3 2 One of the most important is the vertebral formula 3 of Oreopithecus. There is general agreement, based on the ‘1958 specimen’ (IGF 11 778), that Oreopithecus 1 had five lumbar vertebrae (Harrison 1986, 1991; 1 Kohler & Moya-Sola 1997; Rook et al. 1999). A largely overlooked vital statistic, however, is that it also had six sacral vertebrae (Straus 1963; method of Schultz 1961; for details, see McCollum et al. 2009). This can be safely concluded from specimen BA-50, which preserves five sacral foramina on the left side, Figure 4. Hominid and pongid mechanisms of emancipation and at least four on the right. Moreover, the masses or fixation of the most caudal lumbar(s). A human pelvis (left) compared with that of a chimpanzee (right). Note the of the right and left halves of the sixth sacral vertebra following numbered characters in each. The iliac isthmus appear to be fully symmetrical (therefore the right (1) and the ilium itself (3) have both been greatly shortened side presumably had five full foramina as well). We in the human, so much so that a greater sciatic notch has have demonstrated elsewhere that the basal hominoid been created (entirely absent in the chimpanzee), and there column almost certainly exhibited 13 thoracics is no potential contact between the iliac crest and the most (among living taxa, only Homo and Pongo have any caudal lumbar. In the chimpanzee, the opposite change has significant incidences of fewer). Thus, the minimum occurred, i.e. the iliac isthmus (1) and the iliac blade (3) pre-coccygeal vertebral number in Oreopithecus was have both been superoinferiorly elongated, encouraging 31, which, as noted above, is the likely pre-coccygeal such contact. In addition, the sacral alae have been greatly vertebral number for basal hominoids and was prob- broadened in the human and narrowed in the chimpanzee (cf. figure 2). As a consequence, there is now a very substan- ably modal for Early and Mid-Miocene apes as well. tial horizontal distance between the iliac crest and the most Except for P. paniscus, a modal vertebral number as caudal vertebra in the human (2), but a greatly narrowed high as 31 is extremely rare in extant species, occurring bi-iliac gulf in the chimpanzee. In combination with (1) in only 2.8 per cent of P.troglodytes and 0.06 per cent of and (3), such narrowing in the chimpanzee (via reduced Homo (McCollum et al. 2009). alar breadth; cf. figure 2) results in full restrictive contact Much has been made of the putative ‘short, between the iliac crest and the most caudal lumbar ver- broad, ilium’ of Oreopithecus and of its relatively tebra(e). In the earliest phases of this morphological shift broad retroauricular segment (Hu¨rzeler 1958). How- in hominids, the pelvis became decidedly platypelloid, and ever, a substantial reduction in the size of the post- the enhanced iliac breadth encouraged a more effective use auricular region of the pelvis appears to have of the anterior gluteal muscles as abductors during upright accompanied the spinal invagination underlying gait (Lovejoy 2005a; Lovejoy et al. 2009d ). The latter was thus an exaptation, rather than the primary adaptation. scapular relocation in all hominoids (see earlier). That reduction was in turn accompanied by a broadening of the pre-auricular portion of the pelvis and is therefore expected in any clade in which shoulder 8. THE QUESTION OF OREOPITHECUS reorganization occurred (Lovejoy et al. 2009c). This Delineation of the vertebral evolutionary pattern of same developmental process is likely to have re- African apes and hominids throws considerable new occurred a number of times in hominoid evolution, light on the troublesome issue of both the locomotor and is almost certainly universally responsible for pattern and phylogeny of perhaps the most enigmatic the dorsal migration of the LTPs. Broadening of hominoid of the later Miocene, Oreopithecus. the ilium well beyond comparable dimensions in Arguments as to its potential phylogenetic relation- Proconsul is therefore fully expected in virtually any ships and locomotor patterns have been many large-bodied Miocene hominoid that exhibits (reviewed in Harrison 1986, 1991; Kohler & Moya- posterolateralization of the shoulder. Sola 1997; Rook et al. 1999). However, all have been The fifth lumbar vertebra of the ‘1958 specimen’ hampered by its extremely poor condition, largely lies (in situ) directly within its bi-iliac space, sharing the consequence of its extreme compression during the same functional position as the trapped (immobi- fossilization. This has frequently led to excessively lized) L7 of a typical Presbytis and the L3 or L4 of liberal interpretations of its badly compromised Pan (see Straus 1963; figure 5). Therefore, Oreopithe- structure. cus exhibits a maximum of only four potentially A case in point is the attribution of a lordotic spine mobile lumbar vertebrae. This is fully consistent with to this taxon based on a sagittal section of specimen its ‘classic’ adaptive regimen for suspension as also BA72, a crushed and compressed amalgam of three seen in Gorilla, Pan and Pongo, and with directly oppo- lumbar vertebrae (Kohler & Moya-Sola 1997). It site polarity compared with their homologues in seems inconceivable to us that such sectioning can bipedal hominids in a host of major adaptive charac- reliably indicate the presence/absence of wedging in ters (table 1). These included transformation of centra after they have been compressed to less than lumbars via their sacralization, direct reduction in one half of their dorsoventral diameter. A far more lumbar number from the primitive condition and conservative approach is to rely on more entrapment (immobilization) of at least one lumbar

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Spinopelvic pathways to bipedality C. O. Lovejoy & M. A. McCollum 3295

(a) (b) (c)

Figure 5. (a) Sacra of a chimpanzee, (b) A.L. 288-1 and (c) a modern human. Note the extremely narrow sacrum of the chimpanzee compared with the two hominids. Note also the much broader alae in A.L. 288-1 compared with its centrum. Compare this with the similar dimensions in the human specimen (ﬁgure 2).

(a) (b)

1 2 (c) (d)

Figure 6. Comparison of interfacet distances in the third lumbars and sacra of African apes and hominids. The drawing on the left demonstrates the comparison being made. In this drawing, the third lumbar has been rotated 1808 from its normal anatomical position (its superior zygapophyses now point inferiorly) for comparison with those of the sacrum. Double-headed arrows indicate the interfacet distance in each specimen. (a) chimpanzee; (b) gorilla; (c) A.L. 288-1 and (d) human. Note that in (a) and (b) the interfacet distance is greater in the lumbar vertebra than it is in the sacrum, whereas the opposite is true of the two hominids. The increasing gradient of centrum size and interfacet distance in Homo may be an adaptation that facilitated increased lordosis and thereby enabled lumbar column reduction. The opposite gradient in chimpanzees is the likely source of reduction of the bi-iliac space. For discussion, see text. Redrawn from Lovejoy (2005a). by contact with a posterodorsally extended iliac crest. One additional supposedly hominid character in Given its primitive vertebral number, and a series of Oreopithecus is worthy of brief note. The degree of pro- others, such as its retention of an anterior keel on its tuberance of its AIIS is not unusual for a non-hominid. lumbar vertebrae (Straus 1963), Oreopithecus appears What distinguishes the AIIS in hominids from those in to have acquired extensive adaptations to suspension apes is not its protuberance (those of Gorilla are often entirely independently of other Miocene clades (as very prominent), but rather its emergence from a did Nacholapithecus; Nakatsukasa et al. 2007). It is novel, separate physis, a hominid adaptation that is thereby unrelated to hominids, its similarities (which almost certainly associated with dramatic expansion are few; table 1) being largely minor convergences. of iliac isthmus breadth (Lovejoy et al. 2009b). There Any bipedality would have been largely driven by the is no evidence of a similar degree of broadening in same context that does so in hylobatids—excessively Oreopithecus (note its relative pelvic breadth in long forelimbs combined with highly abbreviated table 1) and certainly none suggesting its origin by hindlimbs (table 1). means of a separate physis.

Phil. Trans. R. Soc. B (2010) hl rn.R o.B Soc. R. Trans. Phil. McCollum A. M. & Lovejoy O. C. 3296 (2010)

Table 1. Principal characters of Oreopithecus compared with those of other hominoids. Downloaded from

no. of no. of iliac width/length radius/ medial no. of sacral lumbar functional forelimb length/ intermembral index (relative iliac tibia humerus/ cuboid third Mc length/ taxon vertebraa vertebra lumbars body mass(0.33)b indexb breadth)c indexb femur indexb lengthb body mass(0.33)b,d

Oreopithecus 6e 5e 4e 170e 119 80 120 117 very shortf 20.5

P.troglodytes 5–6 3–4 2–3 160 106 66 111 101 short 24.7 rstb.royalsocietypublishing.org pnpli ahast bipedality to pathways Spinopelvic P.paniscus 6–7 4 2–3 159g 101g 107g 108g short Gorilla 5–6 3–4 2–3 152 114 92 113 116 short 18.9 LCAh [4] [6] [6] [131–143] [87–91] ? [88–95] [77–87] [long] ? Ar. ramidus [4] [6] [6] 143 89–91 113 95 87 long ? Au. 4 6 6 145 84 137 93 84 ? ? afarensis H. sapiens 5 5–6 5 135 65–79 125 65 71 very long 15.8 Pongo 5–6 3–4 2–3 192 138 74 147 130 very short 27.6 Hylobates 4–5 5 4 278 130 49 145 116 short 34.1

Proconsul [4] 6–7 6 (131–134) 87 50 88 77 [long] ? onOctober24,2010 aMethod of Schultz (1961). bData from sample described in Lovejoy et al. (2009c) or additional sources therein unless otherwise noted. cData from Straus (1963), except for Proconsul, Ar. ramidus and Au. afarensis, which were obtained from casts. dData for Oreopithecus from Susman (2004). eOreopithecus data from Straus (1963) and examination of BA-50; other data from Pilbeam (2004) and McCollum et al. (2009). fSee Kohler & Moya-Sola (1997). gData from Morbeck & Zihlman (1989). hData in brackets hypothesized for LCA of African apes and humans and/or Proconsul based on Ar. ramidus and living hominoids. Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Spinopelvic pathways to bipedality C. O. Lovejoy & M. A. McCollum 3297

9. A NOTE ON POTENTIAL MECHANISMS IN in African apes is probably a product of their modest LUMBAR COLUMN MODIFICATION body size and the unique nature of suspension in In considering the anatomy of the lumbosacral spine, it these lesser apes. is of some interest that whereas in humans both the The earliest hominids were able to functionally overall size of the centrum and the distances separating achieve bipedality because they had never rigidified the articular facets (zygapophyses) increase in each their lumbar spines. Instead, they evolved an opposite successively more caudal lumbar vertebrae, centrum morphology—a reduction in iliac height and a broaden- size and interfacet distances in extant African apes ing of the sacrum, both of which assured sufficient instead decrease caudally (Latimer & Ward 1993; lordosis to reduce and eventually eliminate what were figure 6). Lumbar centrum dimensions do not probably only moderate vertical moments about the appear to differ substantially in the only column that knee and hip. Were hominids to have first engaged in permits their observation in Australopithecus (Sts-14; African ape-like behaviours, the ‘Rubicon’ to Robinson 1972). There is, however, an increase in bipedality may have become too great to cross. Our the interfacet distance between the putative L3 and decades-long assumption that the abducent capacity sacrum of A.L. 288-1 (Lovejoy 2005a). The latter of the early hominid pelvis was its primary selective findings suggest that the progressive caudal expansion agent (e.g. Lovejoy et al. 1973) was, in retrospect, of both the interfacet distances and centrum dimen- entirely misdirected. The favourable position of the sions evident in Homo, but only partially adumbrated anterior gluteal muscles in hominids that allows them in Australopithecus (i.e. no increase in lumbar centrum to control pelvic tilt during single support can now dimensions, a retention of six lumbar vertebrae, but a be seen to have been largely a refinement that followed partial increase in interfacet distance), may be an the initial primary adoption of a lordotic spine with an adaptation that permits a more intense lordosis in emancipated caudal-most lumbar vertebra. The gener- humans, ultimately enhancing lumbar column stability alized structure of earliest hominids that permitted this by allowing a reduction in total lumbar number. If so, sequence of events is almost certainly extendable to emergence of this gradient must have postdated Homo the LCA. At least initially in pre-divergence homi- erectus at 1.6 ma, since the lumbar column in noids, it now suggests a combination of cautious, KNM-WT-15000 still numbers six with four sacral palmigrade, plantigrade climbing with a long flexible vertebrae (Latimer & Ward 1993; Pilbeam 2004; back during arboreal travel, and possibly, palmigrade McCollum et al. 2009). quadrupedality during terrestrial travel as well. We thank Alan Walker and Chris Stringer for organizing the discussion meeting and the staff of the Royal Society for 10. SUMMARY AND CONCLUSIONS ensuring its success. We thank David Pilbeam for extensive New evidence from the fossil record and from obser- help in constructing the bonobo sample used in this paper. vations of extant hominoid skeletal anatomy leads to We thank Wim Wendelen and the staff and administration several conclusions. Among the most important is of the Royal Museum for Central Africa, Tervuren, that hominids never acquired the numerous specializ- Belgium for access to the primate collections in their care ations seen in extant apes for vertical climbing, and for the valuable assistance during our examination of their specimens, and Yohannes Haile-Selassie and Lyman suspension or knuckle-walking. This demonstrable Jellema for aid with examination of specimens housed at divergence between the natural histories of hominids the Cleveland Museum of Natural History. and those of all living apes renders most observations of locomotor behaviour in the latter, whether conducted in the laboratory or observed in the wild, no longer directly relevant to the reconstruction of the REFERENCES earliest locomotion of hominids. Bipedality in homi- Almecija, S., Alba, D. M. & Moya-Sola, S. 2009 Pierolapithe- nids can instead now be seen to have emerged from cus and the functional morphology of Miocene ape hand a predominantly primitive locomotor skeleton with a phalanges: paleobiological and evolutionary implications. long lumbar spine capable of at least partial lordosis, J. Hum. Evol. 57, 284–297. (doi:10.1016/j.jhevol.2009. and one never restrictively modified for suspension, 02.008) Andrews, P. 1981 Species diversity and diet in monkeys and vertical climbing or knuckle-walking. apes during the Miocene. In Aspects of human evolution Suspension (e.g. Pongo, Hylobates) and vertical (ed. C. Stringer), pp. 25–61. London, UK: Taylor and climbing (e.g. extant African apes) have repeatedly Francis. induced a rigid lower spine, especially as body mass Benton, R. S. 1967 Morphological evidence for adaptations increased. This has been accomplished in a variety of within the epaxial region of the primates. In The baboon in taxa in several ways, but always by a combination of medical research: Vol. II (ed. F. van der Hoeven), pp. 201– reduction in lumbar number (by reduction in somite 216. Austin, TX: University of Texas Press. vertebral number and/or transformation of lumbar Carey, T. S. & Crompton, R. 2005 The metabolic costs of identity) via modification of hox regulation and further bent-hip, bent-knee walking in humans. J. Hum. Evol. entrapment of a portion of the remaining lumbar ver- 48, 25–44. (doi:10.1016/j.jhevol.2004.10.001) Cartmill, M. 1985 Climbing. Functional vertebrate morphology tebrae by narrowing of the bi-iliac space. The latter has (eds M. Hildebrand, D. M. Bramble, K. F. Liem & D. B. been accomplished either by sacral narrowing or Wake), pp. 73–88. Cambridge, MA: Belknap Press. dorsal extension of the iliac crest (e.g. Pan and Cartmill, M. & Milton, K. 1977 The lorisiform wrist joint Gorilla), or both. The failure of hylobatids (but not and the evolution of ‘brachiating’ adaptations in the symphalangids (i.e. modal lumbar number ¼ 4) to Hominoidea. Am. J. Phys. Anthropol. 47, 249–272. achieve the extreme lumbar column reductions seen (doi:10.1002/ajpa.1330470206)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3298 C. O. Lovejoy & M. A. McCollum Spinopelvic pathways to bipedality

Crompton, R. H., Li, Y., Wang, W., Gu¨nther, M. M. & Savage, Lovejoy, C. O., Heiple, K. G. & Burstein, A. H. 1973 The R. 1998 The mechanical effectiveness of erect and ‘bent-hip, gait of Australopithecus. Am. J. Phys. Anthropol. 38, bent-knee’ bipedal walking in Australopithecus afarensis. 757–779. (doi:10.1002/ajpa.1330380315) J. Hum. Evol. 35,55–74.(doi:10.1006/jhev.1998.0222) Lovejoy, C. O., Latimer, B., Suwa, G., Asfaw, B. & White, Desilva, J. M. 2009 Functional morphology of the ankle and T. D. 2009a Combining prehension and propulsion: the the likelihood of climbing in early hominins. Proc. Natl foot of Ardipithecus ramidus. Science 326, 72e1–72e8. Acad. Sci. USA 106, 6567–6572. (doi:10.1073/pnas. Lovejoy, C. O., Simpson, S. W., White, T. D., Asfaw, B. & 0900270106) Suwa, G. 2009b Careful climbing in the Miocene: the Erikson, G. E. 1963 Brachiation in New World monkeys and forelimbs of Ardipithecus ramidus and humans are in anthropoid apes. Symp. Zool. Soc. Lond. 10, 135–164. primitive. Science 326, 70e1–70e8. Filler, A. G. 1981 Anatomical specializations in the Lovejoy, C. O., Suwa, G., Simpson, S. W., Matternes, J. H. & hominoid lumbar region. Am. J. Phys. Anthropol. 54, 218. White, T. D. 2009c The great divides: Ardipithecus ramidus Filler, A. G. 2007a Homeotic evolution in the mammalia: reveals the postcrania of our last common ancestors with diversification of therian axial seriation and the morpho- African apes. Science 326, 100–106. genetic basis of human origins. PLoS ONE 2, e1019. Lovejoy, C. O., Suwa, G., Spurlock, L., Asfaw, B. & White, (doi:10.1371/journal.pone.0001019) T. D. 2009d The pelvis and femur of Ardipithecus ramidus: Filler, A. G. 2007b Emergence and optimization of upright the emergence of upright walking. Science 326, posture among hominiform hominoids and the evolution- 71e1–71e6. ary pathophysiology of back pain. Neurosurg. Focus 23, E4. MacLatchy, L., Gebo, D., Kityo, R. & Pilbeam, D. 2000 Harrison, T. 1991 The implications of Oreopithecus bambolii Postcranial functional morphology of Morotopithecus for the origins of bipedalism. In Origine(s) de la bipedie bishopi, with implications for the evolution of modern chez les hominides, Cahiers de Paleóanthropologie (eds ape locomotion. J. Hum. Evol. 39, 159–183. (doi:10. Y. Coppens & B. Senut), pp. 235–244. Paris, France: 1006/jhev.2000.0407) Editions du CNRS. McCollum, M. A., Rosenman, B. A., Suwa, G., Meindl, Harrison, T. 1986 A reassessment of the phylogenetic relation- R. S. & Lovejoy, C. O. 2009 The vertebral formula of ships of Oreopithecus bambolii gervais. J. Hum. Evol. 15, the last common ancestor of African apes and humans. 541–583. (doi:10.1016/S0047-2484(86)80073-2) J. Exp. Zool. B Mol. Dev. Evol. 314B, 123–134. Hirasaki, E., Ogihara, N., Hamada, Y., Kumakura, H. & Morbeck, M. E. & Zihlman, A. L. 1989 Body size and pro- Nakatsukasa, M. 2004 Do highly trained monkeys walk portions in chimpanzees, with special reference to Pan like humans? A kinematic study of bipedal locomotion troglodytes schweinfurthii from Gombe National Park, Tan- in bipedally trained Japanese macaques. J. Hum. Evol. zania. Primates 30, 369–382. (doi:10.1007/BF02381260) 46, 739–750. (doi:10.1016/j.jhevol.2004.04.004) Moya-Sola, S., Kohler, M., Alba, D. M., Casanovas-Vilar, I. & Hu¨rzeler, J. 1958 Oreopithecus bambolii Gervais, a Galindo, J. 2004 Pierolapithecus catalaunicus, a new Middle preliminary report. Verh. Naturforsch. Ges. 69, 1–48. Miocene great ape from Spain. Science 306, 1339–1344. Johanson, D. C., Lovejoy, C. O., Kimbel, W. H., White, (doi:10.1126/science.1103094) T. D., Ward, S. C., Bush, M. E., Latimer, B. M. & Nakatsukasa, M. 2004 Acquisition of bipedalism: the Coppens, Y. 1982 Morphology of the Pliocene partial Miocene hominoid record and modern analogues for hominid skeleton (A.L. 288-1) from the Hadar bipedal protohominids. J. Anat. 204, 385–402. (doi:10. Formation, Ethiopia. Am. J. Phys. Anthropol. 57, 1111/j.0021-8782.2004.00290.x) 403–452. (doi:10.1002/ajpa.1330570403) Nakatsukasa, M., Hayama, S. & Preuschoft, H. 1995 Post- Kohler, M. & Moya-Sola, S. 1997 Ape-like or hominid-like? cranial skeleton of a macaque trained for bipedal The positional behavior of Oreopithecus bambolii reconsid- standing and walking and implications for functional ered. Proc. Natl Acad. Sci. USA 94, 11 747–11 750. adaptation. Folia Primatol. (Basel) 64, 1–29. (doi:10. (doi:10.1073/pnas.94.21.11747) 1159/000156828) Latimer, B. & Lovejoy, C. O. 1989 The calcaneus of Austra- Nakatsukasa, M., Tsujikawa, H., Shimizu, D., Takano, T., lopithecus afarensis and its implications for the evolution of Kunimatsu, Y., Nakano, Y. & Ishida, H. 2003 Definitive bipedality. Am. J. Phys. Anthropol. 78, 369–386. (doi:10. evidence for tail loss in Nacholapithecus, an East African 1002/ajpa.1330780306) Miocene hominoid. J. Hum. Evol. 45, 179–186. Latimer, B. & Ward, C. V. 1993 The thoracic and lumbar (doi:10.1016/S0047-2484(03)00092-7) vertebrae. In The Nariokotome Homo erectus skeleton (eds Nakatsukasa, M., Ogihara, N., Hamada, Y., Goto, Y., A. Walker & R. Leakey), pp. 266–293. Cambridge, Yamada, M., Hirakawa, T. & Hirasaki, E. 2004 Energetic MA: Harvard University Press. costs of bipedal and quadrupedal walking in Japanese Lewis, O. J. 1989 Functional morphology of the evolving hand macaques. Am. J. Phys. Anthropol. 124, 248–256. and foot. Oxford, UK: Clarendon Press. (doi:10.1002/ajpa.10352) Lovejoy, C. O. 1981 The origin of man. Science 211, Nakatsukasa, M., Hirasaki, E. & Ogihara, N. 2006 Energy 341–350. (doi:10.1126/science.211.4480.341) expenditure of bipedal walking is higher than that of Lovejoy, C. O. 1993 Are we sexy because we’re smart or quadrupedal walking in Japanese macaques. Am. J. smart because we’re sexy? In The origin and evolution of Phys. Anthropol. 131, 33–37. (doi:10.1002/ajpa.20403) humans and humanness (ed. D. T. Rasmussen), pp. 1–28. Nakatsukasa, M., Kunimatsu, Y., Nakano, Y. & Ishida, H. New York, NY: Jones and Bartlett. 2007 Vertebral morphology of Nacholapithecus kerioi Lovejoy, C. O. 2005a The natural history of human gait and based on KNM-BG 35250. J. Hum. Evol. 52, 347–369. posture I: spine and pelvis. Gait Posture 21, 113–128. (doi:10.1016/j.jhevol.2006.08.008) Lovejoy, C. O. 2005b The natural history of human gait and Pilbeam, D. 2004 The anthropoid postcranial axial skeleton: posture II: hip and thigh. Gait Posture 21, 129–151. comments on development, variation, and evolution. Lovejoy, C. O. 2007 The natural history of human gait and J. Exp. Zoolog. B Mol. Dev. Evol. 302, 241–267. posture III: knee. Gait Posture 25, 325–341. (doi:10. Robinson, J. T. 1972 Early hominid posture and locomotion. 1016/j.gaitpost.2006.05.001) Chicago, IL: University of Chicago Press. Lovejoy, C. O. 2009 Reexamining human origins in the light Rook, L., Bondioli, L., Kohler, M., Moya-Sola, S. & of Ardipithecus ramidus. Science 326, 74e1–74e8. (doi:10. Macchiarelli, R. 1999 Oreopithecus was a bipedal ape 1126/science.1175834) after all: evidence from the iliac cancellous architecture.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Spinopelvic pathways to bipedality C. O. Lovejoy & M. A. McCollum 3299

Proc. Natl Acad. Sci. USA 96, 8795–8799. (doi:10.1073/ Thorpe, S. K., Crompton, R. H. & Alexander, R. M. 2007a pnas.96.15.8795) Orangutans use compliant branches to lower the ener- Rose, M. 1988 Another look at the anthropoid elbow. J. Hum. getic cost of locomotion. Biol. Lett. 3, 253–256. Evol. 17,193–224.(doi:10.1016/0047-2484(88)90054-1) (doi:10.1098/rsbl.2007.0049) Sanders, W. J. 1995 Function, allometry and evolution of the Thorpe, S. K., Holder, R. L. & Crompton, R. H. 2007b australopithecine lower precaudal spine. New York, NY: Origin of human bipedalism as an adaptation for loco- New York University. motion on flexible branches. Science 316, 1328–1331. Schultz, A. H. 1961 Vertebral column and thorax. Primatologia (doi:10.1126/science.1140799) 4,1–66. Thorpe, S. K., Holder, R. & Crompton, R. H. 2009 Orangu- Sellers, W., Cain, G., Wang, W. & Cromton, R. H. 2005 tans employ unique strategies to control branch flexibility. Stride lengths, speed and energy costs in walking of Proc. Natl Acad. Sci. USA 106, 12 646–12 651. (doi:10. Australopithecus afarensis: using evolutionary robotics to 1073/pnas.0811537106) predict locomotion of early human ancestors. J. R. Soc. Ward, C. V. 1991 Functional anatomy of the lower back and Interface 2, 431–441. (doi:10.1098/rsif.2005.0060) pelvis of the Miocene hominoid Proconsul nyanze from Mfan- Stern Jr, J. T. 2000 Climbing to the top: a personal gano Island, Kenya, pp. 1–379. Baltimore, MD: Johns memoir of Australopithecus afarensis. Evol. Anthropol. 9, Hopkins University. 113–133. (doi:10.1002/1520-6505(2000)9:3,113::AID- Ward, C. V. 1993 Torso morphology and locomotion in EVAN2.3.0.CO;2-W) Proconsul nyanzae. Am. J. Phys. Anthropol. 92, 291–328. Stern Jr, J. T. & Susman, R. L. 1983 The locomotor anatomy (doi:10.1002/ajpa.1330920306) of Australopithecus afarensis. Am. J. Phys. Anthropol. 60, Ward, C. V. 2007 Postcranial and locomotor adaptations 279–317. of hominoids. In Handbook of paleoanthropology (eds Stevens, L. S. 2004 Morphological varition in the hominoid W. Henke & I. Tattersall), pp. 1011–1030. Heidelberg, vertebral column. PhD thesis, Kent State University, Germany: Springer. Kent, OH. Ward, C. V., Walker, A. & Teaford, M. 1991 Proconsul did Stevens, L. S. & Lovejoy, C. O. 2004 Morphological not have a tail. J. Hum. Evol. 21, 215–220. (doi:10. variation in the hominoid vertebral column: implications 1016/0047-2484(91)90062-Z) for the evolution of human locomotion. Am. J. Phys. Ward, C. V., Walker, A., Teaford, M. F. & Odhiambo, I. Anthropol. 123 S38, 187–188. 1993 Partial skeleton of Proconsul nyanzae from Strauss Jr, W. L. 1963 The classification of Oreopithecus.In Mfangano Island, Kenya. Am. J. Phys. Anthropol. 90, Classification and human evolution (ed. S. L. Washburn), 77–112. (doi:10.1002/ajpa.1330900106) pp. 146–177. Chicago, IL: Aldine. Waterman, H. C. 1929 Studies on the evolution of the pelves Susman, R. L. 2004 Oreopithecus bambolii an unlikely case of of man and other primates. Am. Mus. Nat. Hist. Bull. 58, hominidlike grip capability in a Miocene. J. Hum. Evol. 585–642. 46, 105–117. (doi:10.1016/j.jhevol.2003.10.002) White, T. D., Asfaw, B., Beyene, Y., Haile-Selassie, Y., Susman,R.L.,SternJr,J.T.&Jungers,W.L.1984 Lovejoy, C. O., Suwa, G. & WoldeGabriel, G. 2009 Arboreality and bipedality in the Hadar hominids. Folia Ardipithecus ramidus and the paleobiology of early Primatol. (Basel) 43,113–156.(doi:10.1159/000156176) hominids. Science 326, 75–86.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Phil. Trans. R. Soc. B (2010) 365, 3301–3314 doi:10.1098/rstb.2010.0035

Review Arboreality, terrestriality and bipedalism Robin Huw Crompton1,*,†, William I. Sellers2,† and Susannah K. S. Thorpe3,† 1Primate Evolution and Morphology Research Group, School of Biomedical Sciences, The University of Liverpool, Sherrington Buildings, Ashton Street, Liverpool L69 3GE, UK 2Faculty of Life Sciences, The University of Manchester, 3.614 Stopford Building, Oxford Road, Manchester M13 9PT, UK 3School of Biosciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK The full publication of Ardipithecus ramidus has particular importance for the origins of hominin bipedality, and strengthens the growing case for an arboreal origin. Palaeontological techniques however inevitably concentrate on details of fragmentary postcranial bones and can benefit from a whole-animal perspective. This can be provided by field studies of locomotor behaviour, which provide a real-world perspective of adaptive context, against which conclusions drawn from palaeontology and comparative osteology may be assessed and honed. Increasingly sophisticated dynamic modelling techniques, validated against experimental data for living animals, offer a different perspective where evolutionary and virtual ablation experiments, impossible for living mammals, may be run in silico, and these can analyse not only the interactions and behaviour of rigid segments but increasingly the effects of compliance, which are of crucial importance in guiding the evolution of an arboreally derived lineage. Keywords: bipedalism; biomechanics; evolution; field studies

1. INTRODUCTION and the panins (bonobos and common chimpanzees), Darwin’s (1871) argument on human origins has the prevailing paradigm for the origins of human never appeared stronger than now, when molecular bipedalism has been the knucklewalking quadrupedal- evidence suggests a divergence time of only 5–8 Ma ism model (first proposed by Washburn (1967) and for humans and their extinct relatives (the tribe Homi- reviewed by Tuttle et al. (1974)). This model holds nini), from the chimpanzees and bonobos (tribe that the common ancestor of hominins and panins Panini; Bradley 2008). But as pointed out by Tuttle would have looked much like chimpanzees do today, et al. (1974) in their excellent review, Darwin, while and so bipedalism would have arisen in an ancestor he did not present a detailed model of the last which was a terrestrial, quadrupedal knucklewalker, common ancestor of humans and other African apes, like the panins, and the remaining African apes, the made an important point that is too often ignored: gorillines. that we should not expect the last common ancestor This paradigm was developed in some detail by to resemble either living humans or other living apes Gebo (1992, 1996), who identified heel-strike planti- particularly closely. grady as a common, shared-acquired character of African apes linked closely to knucklewalking quadrupedalism, and to the hominin acquisition of a 2. BIPEDALISM: AN ARBOREAL OR terrestrially adapted foot. However, heel-strike planti- TERRESTRIAL ORIGIN? grady is not limited to the African apes (Meldrum In the first four decades of the twentieth century, it was 1993; Crompton et al. 2003); also, heel-strike is actu- generally accepted that bipedalism had an arboreal ally particularly clearly expressed in an Asian ape, the origin (e.g. Keith 1903, 1923; Morton 1922; Schultz most arboreal of great apes, the orangutan, subfamily 1936). But for the last 60 years, since the first field Ponginae (Crompton et al. 2003, 2008). study of mountain gorillas by Schaller (1963), the All great apes can and do walk bipedally, and most field studies of chimpanzees by Goodall (1998), and do so in an arboreal context. Again, it is the most the ensuing recognition (e.g. Zihlman et al. 1978)of arboreal, the orangutan, which uses bipedal locomotion a special and genetically very close relationship most often (Thorpe & Crompton 2005, 2006). While between the hominins (humans and their ancestors) bipedal locomotion supported by the hindlimbs alone makes up only about 2 per cent of arboreal locomotion * Author for correspondence ([email protected]). of orangutans, a further 6 per cent consists of bipedal- † All authors contributed equally to this manuscript. ism where one or both forelimbs are used for balance. One contribution of 14 to a Discussion Meeting Issue ‘The first four But this small percentage of locomotor bipedalism (or million years of human evolution’. compressive orthogrady, if preferred) plays an

3301 This journal is # 2010 The Royal Society Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3302 R. H. Crompton et al. Review. Arboreal origins of hominin bipedalism ecologically crucial role in movement on the fine per- the side of their hand. Richmond et al.(2001), however, ipheral branches, where fruits are located. It further stoutly defended a knucklewalking origin in an exten- allows orangutans to bridge from tree to tree at sive review, and Richmond & Jungers (2008) claimed canopy level, avoiding the very high costs (Thorpe that similarities in curvature of a single phalanx of the et al. 2007a) and predation risk associated with cross- late Miocene protohominin Orrorin to that in chimpan- ing on the ground. It is again the orangutan which of zees represented evidence of knucklewalking, although all living apes approaches closest to us in one of the Orrorin is regarded by its discoverers as arboreally most important of the biomechanical features ident- adapted, orthograde and bipedal when moving on the ified by Alexander (1991) as characteristic of human ground (Senut et al.2001). Kivell & Begun (2007) walking: namely stiff-legged, upright gait. As a conse- however found no clear functional link between os cen- quence of its stiff-legged gait, the orangutan trale–scaphoid fusion and knucklewalking and Kivell & produces, in a fifth of its bipedalism, double-humped Schmitt (2009) argue that there are two functionally vertical ground reaction force curves (vGRF) which, distinct modes of knucklewalking in African apes: that alone among apes, overlap with those produced by in chimpanzees being associated with extended wrist human walking (Crompton et al. 2003), and which postures in an arboreal environment (directly addressing allow a high degree of pendular energy conversion. Richmond & Jungers 2008), and that in gorillas with a Our calculations indicate that while in untrained neutral wrist posture in a terrestrial environment. common chimpanzees energy conversion in bipedal- Kivell & Schmitt (2009) go on to argue that the pur- ism reaches little more than 8 per cent, it approaches ported knucklewalking features of hominins are some 50 per cent in untrained orangutans, still well instead adaptations to arboreality, and thus that short, of course, of the 70 per cent possible in bipedalism indeed arose in the arboreal ecological humans (Wang et al. 2003). But while it has recently niche common to living apes. been argued that the elongated, inverted foot of the While the absence of purported knucklewalking fea- orangutan does not at all closely resemble our own tures in the hand of Au. afarensis (e.g. Stern & Susman (Sayers & Lovejoy 2008), orangutan foot function in 1983) leaves little time for hominins to lose any such bipedal walking, expressed in the pattern of foot features after the separation of hominins and panins pressure, is actually very similar (Crompton et al. 5–8 Ma, the publication of a full description of 2008) to that of the bonobo (Vereecke et al. 2003), Ardipithecus ramidus shows that knucklewalking often suggested by others as a model for the features, including dorsal distal metacarpal ridges are common panin–hominin ancestor (e.g. Zihlman also absent in Ar. ramidus (Lovejoy et al. 2009a), et al. 1978 and reviewed in Vereecke et al. 2003). with the exception of os centrale–scaphoid fusion. It has been apparent for the last few years that a However, recall that Kivell & Begun (2007) found growing number of scientists have found cause to no functional link to knucklewalking for this feature. doubt whether firm evidence exists in the fossil This extends the lack of evidence for a terrestrial record for a knucklewalking origin: see, e.g. Stern & knucklewalking phase in the evolution of human Susman (1983) for Australopithecus afarensis; Ward bipedalism to 4.4 Ma. Equally, in linking terrestrial et al. (1999) for the South Turkwel handbones and bipedalism to arboreality (Lovejoy et al. 2009a,b), Clarke (1999, 2002) for the StW-573 hand. A range publication of Ar. ramidus has greatly strengthened of purported ‘knucklewalking features’, dorsal ridges the positive case for an arboreal origin for the core on the distal aspect of the metacarpals, os centrale hominin adaptation. In doing so, it challenges us to –scaphoid fusion or extension of the proximal articu- develop a convincing arboreal alternative to a terres- lar surface of the capitulum onto its dorsum, have trial knucklewalking model of the origins of human been sought in the hominin fossil record, but have bipedalism. either not been found or found only inconsistently. While still based on the concept that we should look Dainton & Macho (1999) raised doubts about whether for the origins of human bipedalism among activities knucklewalking was a homologous phenomenon even of living African apes, the most supported arboreal in chimpanzees and gorillas. However, Richmond & challenger for the terrestrial knucklewalking model is Strait (2000) argued that the distal radial morphology the ‘vertical climbing’ hypothesis of Fleagle et al. of Au. afarensis was evidence for a knucklewalking (1981). This was derived primarily from electromyo- phase in evolution some time between 3.6 Ma and graphic similarities between hip, buttock and thigh the commonly accepted 5–8 Ma limits for genetic musculature activity of African apes during climbing separation of hominins and panins. It is therefore note- on large, vertical supports, and that of humans walking worthy that the morphology plotted by Richmond & bipedally. However, the kinematics of vertical climbing Strait (2000) lies well within the orangutan range of (Isler 2002, 2003; Isler & Thorpe 2003) and knuckle- variation. Only large male Bornean orangutan make walking (Watson et al. 2009) are rather similar, much use of the ground, the Sumatran tiger being a involving highly flexed postures of the hip and knee major discouragement to terrestriality on that island, (Crompton et al. 2003), which are quite unlike the the clouded leopard posing a threat to small or juvenile extended postures seen in human walking and which orangutan on Borneo. Large Bornean males have too underlie its efficiency (Alexander 1991). Running much unstable mass above the hip to sustain unas- does involve more flexed limb postures, but this is sisted bipedalism, and so tend to cross the ground linked to the use of elastic recoil, as the spring-mass quadrupedally. But when crossing the ground they mechanism requires substantial elastic energy stores. do not walk on the middle phalanx, as chimpanzees The most well known of these elastic energy stores, a or gorillas do, but on their proximal phalanges or on marked Achilles tendon, is absent in both the African

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Review. Arboreal origins of hominin bipedalism R. H. Crompton et al. 3303 apes and the orangutan, which all have large distal adaptive commitment to vertical climbing were muscle masses (Thorpe et al. 1999; Payne et al. acquired independently in panins, after the divergence 2006a,b), which help the more powerful forelimbs from hominins, although Lovejoy et al. (2009b) do not (Thorpe et al. 1999; Oishi et al. 2009) in climbing appear to make a functional link between the two. but probably also act to tune the limbs to deal with Lovejoy et al.(2009a,b,c) identify a number of fea- variations in support compliance in an arboreal tures in which the hands and feet of Ardipithecus context. Interestingly, the gibbons and siamangs do resemble those of the root hominoid Proconsul and have a large Achilles tendon: its mechanical role some living arboreal monkeys, rather than living great is currently under investigation in our laboratory apes. These include evidence for short hands with an (Channon et al. 2009). extensive dorsiflexion range in the metacarpophalangeal The absence of a medial longitudinal arch (MLA) joints that is absent in living great apes apart from in the non-human great apes and its reported absence humans (individuals in some human populations, in Ar. ramidus (Lovejoy et al. 2009a) appears to rule such as the Han Chinese can often dorsiflex to more out that possible location of the required mass of elas- than 908 when young; personal observation, R. H. tic tissue. So the existence of one or both of the most Crompton 1982). Following earlier arguments (Moya`- likely possible elastic energy stores, a large mass of Sola` et al. 2004) that the presence of the same feature plantar soft tissue housed within a MLA (Ker et al. in the Miocene hominoid Pierolapithecus catalaunicus 1987) or a large Achilles tendon, is required to be suggested that it was an arboreal quadruped, they demonstrated before a mechanically effective com- suggest, although acknowledging that it is a curious pliant, rather than stiff-legged, gait can reasonably be combination, that while bipedal on the ground (Lovejoy posited for early hominins. et al.2009c), Ar. ramidus was primarily quadrupedal in While energetic efficiency and mechanical perform- the trees (Lovejoy et al.2009b), while using some ‘care- ance are by no means the only parameters subject to ful climbing and bridging’, presumably at the periphery natural selection (as fieldworkers know better than of trees (Lovejoy et al.2009b). most), they are very often directly or indirectly impor- The feet of Ar. ramidus, like those of monkeys, tant, and can be assessed and predicted relatively apparently retained a thick plantar layer of fibrous readily. Several laboratories, including our own, have tissue, and were thus rather stiff when compared therefore used computer simulation to assess the effec- with those of the panins, gorillines and pongines tiveness of alternative gaits in Au. afarensis and other (Lovejoy et al. 2009 a). This implies lesser ability to hominins. Independent studies by at least three separ- conform to branch diameter, and thus relatively poor ate laboratories (Crompton et al. 1998; Kramer 1999; grip for what was apparently a large-bodied (50 kg, Kramer & Eck 2000; Sellers et al. 2003, 2004, 2005; Lovejoy et al. 2009c) hominin, nearly twice the mass Nagano et al. 2005) demonstrate that Au. afarensis of the largest cercopithecine monkey, the mandrill, could have been an effective stiff-legged upright and more than 10 kg greater than the largest individ- biped, particularly over relatively short distances and uals of the largest monkey, the Sichuan snub-nosed walking unloaded (Wang & Crompton 2004; Wang monkey Rhinopithecus roxellana (Rowe 1996). It is et al. 2004). Using forwards dynamic modelling, meta- argued that panins, gorillines and pongines also bolic cost can be predicted. Predicted costs for human acquired their compliant feet independently (Lovejoy models have been verified against experimental values et al. 2009a). With no tail, but equally little pedal for human adults and come within 10–15% of these gripping power, as well as short hands, there can values. Predicted values for upright walking by have been little or no capability to exert balancing Au. afarensis in independent studies by Sellers et al. counter-torques on the support. Then how did (2004, 2005) and Nagano et al. (2005) are in good Ar. ramidus balance their body mass above branches accord and come quite close to the experimental during pronograde quadrupedalism? Quadrupedal values for human children of equivalent size. If monkeys, lacking the wide and powerful grasp of the Au. afarensis could thus have been an effective stiff- living non-human great apes, improve stability by legged, upright biped, and if moving in a compliant deep flexion of the limbs. Stability in flexion is very gait would have incurred both substantial increases often aided by anteflexion of the olecranon process in the mechanical cost of locomotion (Crompton in arboreal monkeys (Fleagle 1998), but there is et al. 1998) and physiological costs including increased apparently no evidence of anteflexion of the olecranon heat load (Carey & Crompton 2005), an origin for in the Ardipithecus proximal ulnae (Lovejoy et al. bipedalism in locomotor modes associated with 2009b). Further, habitual deep flexion of the limbs highly flexed limb postures, such as vertical climbing, as an arboreal quadruped would increase mismatch seems unlikely. (in the power capacity of muscles at different joint A different objection to the vertical climbing model angles) between the requirements of terrestrial has recently been raised by DeSilva (2009). He argues bipedalism and those of arboreal quadrupedalism. that early hominin ankle joint morphology is distinct from that of vertical-climbing African apes, and incompatible with the kinematics required for vertical 3. THE ARBOREAL ORIGINS OF BIPEDALISM: climbing. Lovejoy et al. (2009b) argue that anatomical COMPRESSIVE ORTHOGRADY? features of the hand associated with vertical climbing, A more parsimonious explanation of the metacarpo- such as elongated metacarpals, are absent in phalangeal dorsiflexion seen in Ar. ramidus is surely Ar. ramidus, and they follow Thorpe et al. (2007b)in desirable. It exists, in part, in consideration of proposing that both knucklewalking and a strong elements of arboreal behaviour of all the great apes,

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3304 R. H. Crompton et al. Review. Arboreal origins of hominin bipedalism namely in the use of arboreal hand-assisted bipedal- 2006a,b; Oishi et al. 2009): we climb up bough by ism, or if preferred, compressive orthogrady, both bough, further out in the tree. In the absence of climb- postural and locomotor, to move around the forest ing aids, humans find it difficult and of course canopy. Again, in part, the explanation is to be dangerous (Pontzer & Wrangham 2004) to climb up found in consideration of the similarity, in shared trees where long, naked trunks occur before any side lack of digital elongation and in morphological con- branches. Gorillas and chimpanzees are readily able servatism, between the human hand and that of to do so, using the vertical climbing adaptations, Ar. ramidus, demonstrated by Lovejoy et al. (2009a). which Ardipithecus suggests (Lovejoy et al. 2009a,b,c) Together with the suggestion of Thorpe et al. arose independently and in parallel in the two lineages. (2007b) that vertical climbing and knucklewalking Similarly, hand-assisted compressive orthogrady has were both acquired independently in panins (after the been shown to allow orangutans to move on very flex- separation from hominins) and in gorillins, and its con- ible branches at the periphery of tree crowns where the firmation by the description of Ar. ramidus (Lovejoy most abundant supply of fruit is generally situated et al. 2009b,c), there is increasing evidence (e.g. (Thorpe et al. 2007b) and it may play a similar role Larson 1998) that suspensory adaptations did not in the behaviour of lowland gorillas (see table 11 in evolve at the same time as other features of orthogrady, Remis 1994). Analogy with the largest cercopithecine, but rather homoplastically. While it is difficult to obtain the mandrill (Lahm 1986), less than 27 kg, and con- definitive figures from all studies and for all species, sideration of the behaviour of the largest of all table 1 shows that, together, terrestrial knucklewalking monkeys, Sichuan snub-nosed monkeys Rhinopithecus quadrupedalism, vertical climbing and forelimb suspen- roxellana (Kirkpatrick et al. 1999; Li 2001; Li et al. sion make up some 67 per cent of bonobo locomotion, 2002), less than 37 kg, suggests that the 50 kg body 93 per cent of common chimpanzee locomotion and 97 weight claimed for Ar. ramidus (Lovejoy et al. 2009c) per cent of mountain gorilla locomotion, but only 39 may exceed mass limits of effective monkey-like per cent of orangutan locomotion. A figure for lowland arboreal plantigrade quadrupedalism. However, com- gorilla knucklewalking versus non-knucklewalking pressive orthogrady is exhibited by similar-sized apes quadrupedalism is more difficult to determine, but and could facilitate both access into trees and move- based on proportions of arboreal and terrestrial activity, ment within them. Further, Miocene crown- we have assigned a tentative figure which allows a total hominoid body weight begins above that of mandrills, proportion of novel locomotor modes in the repertoire and equals or exceeds that of R. roxellana: 30–54 kg to be set pro tempore at 62 per cent. Thus, from the for Morotopithecus (MacLatchy et al. 2000), 30 kg for reported evidence of Ardipithecus, the great majority of Pierolapithecus (Culotta 2004; Moya`-Sola` et al. 2004), the panin and gorilline locomotor repertoire employs 30–37 kg for Hispanopithecus (Moya`-Sola` &Ko¨hler novel adaptations since the divergence from hominins. 1996) and 32 kg for Oreopithecus (Ko¨hler & Moya`- This observation has significance for both palaeontolo- Sola` 1997). We suggest that it is unlikely that the gists and fieldworkers, and underlines Darwin’s (1871) consistently larger size of Miocene crown hominoids warning, with which we began this paper, that we was not accompanied by a shift from monkey-like should not expect the common ancestor to resemble arboreal locomotion. either humans or living apes particularly closely. Thus, we argue that a hominin which had not This suggests that what is now the relatively small acquired suspensory/vertical climbing features in the compressive component of great ape orthograde loco- forelimb would have accessed the trees and moved motion may be the oldest, and human locomotion thus within them primarily by palmigrade compressive relatively conservative. (Whether compressive ortho- orthogrady. In the absence of vertical climbing capa- grady is as old as orthogrady itself, or whether bilities and a powerful hand grip, access to trees orthograde body posture arose earlier from a random would of course favour use of more stable supports, homeotic event (Filler 2007), we currently have no which can be loaded under compression without way of knowing). Importantly however, this behaviour excessive deflection. In the absence of suspensory fea- offers a reasonable alternative for locomotion in a tures, hand-assisted bipedalism could have facilitated species not yet in possession of vertical climbing and movement among the finer supports at the periphery suspensory adaptations. We suggest that in an arboreal of trees, employing strategies similar to those we context, hominin species such as Ar. ramidus (as well have reported in the orangutan (Thorpe et al. 2007b, as ourselves) which do not have elongated hands, 2009). Quadrupedalism would be used when absol- powerful in suspension, may tend more often to utely necessary—as of course it is by ourselves when climb upwards (which they must have done relatively we no longer trust our balance or stability—but palmi- frequently if they were exploiting both terrestrial and grade hand postures would be inappropriate among arboreal niches) by pushing themselves up by pressure finer supports. Many anthropoids are able to employ of the hands below shoulder level, so that the ulnar some degree of suspension, vertical climbing and four metacarpophalangeal joints pass into deep dorsi- quadrupedalism regardless of their primary adap- flexion. This adds much of the length of the tation, although perhaps rarely, and at some metacarpals to the potential lift. This is of course additional cost, so suspension is also likely to have exactly how humans usually climb large-trunked been used under certain conditions. It may be trees when they lack climbing equipment to help argued that this is referential modelling (Sayers & them move on the main trunk, with the human lack Lovejoy 2008): but at least we are using multiple refer- of ‘vertical climbing’ adaptations (primarily very ents, and we are able to test the predictions of our powerful arms; see Thorpe et al. 1999; Payne et al. models by simulation. We predict, for example, that

Phil. Trans. R. Soc. B (2010) hl rn.R o.B Soc. R. Trans. Phil. (2010) Downloaded from Table 1. Frequencies of locomotor behaviour in the great apes and ‘ballpark estimates’ of the per cent of their locomotion that has evolved since divergence from their last common ape ancestora.

knucklewalk/run other quadrupedal vertical orthograde pronograde % loco evolved bipedalism hominin of origins Arboreal Review. (terrestrial and arboreal) (terrestrial and arboreal) climb/descent brachiate clamber/transfer bipedal leap scramble otherb since separationc rstb.royalsocietypublishing.org mountain 95 ,1 ,1 0.1 0 0.8 0 .0097 gorilla d,e lowland 38.5 26.7 19.7 3.6 3.3 6.1 0 0 1.3 62 gorillad,f chimpanzeed,g 86.1 4.5 6.5 0.8 0.5 0.7 0.2 0.3 0.1 93 bonobod,h 7.8 27.5 50.4 8.9 0 1.5 3.1 0 0 67 orangutani 0 8 26 13 22 7 ,19 11 39 aNote that all values are ballpark figures as differences in methodology and subject profiles preclude detailed comparison. bFor example, tree sway, pronograde suspension. cThis value assumes that the common ancestor at all levels was a monkey-like above branch quadruped. The value includes knucklewalking, vertical climb/descent, brachiation and orthograde clamber/ onOctober24,2010 transfer. dModified after Hunt (2004) and based on data from Tuttle & Watts (1985), Doran (1996), Hunt (1992) and Remis (1995). eAdapted from Tuttle & Watts (1985). fKnucklewalking/non-knucklewalking values are not available for lowland gorillas at present. However, 59% of their time is spent terrestrially and 41% arboreally (adapted from Hunt 1996). Consequently, as a rough estimate, we have allocated 51% of quadrupedalism to be terrestrial knucklewalking and the remainder to be arboreal non-knucklewalking quadrupedalism. gAdapted from Hunt (1992) following personal communication (K. D. Hunt, 2009). hBonobos are more arboreal than chimpanzees but no conclusive data exist, neither for the per cent of time they are arboreal nor for how much of their quadrupedalism is knucklewalking. However, Susman (1984) observed that of 532 bouts of arboreal quadrupedalism, only 20 (3.8%) were knucklewalking and of Susman’s (1984) 89 first sightings of bonobos, 17 (19%) were terrestrial. Thus, we have weighted Doran’s (1996) frequency data with these values to calculate proportions for knucklewalking and non-knucklewalking quadrupedalism. Crompton H. R. iModified from Thorpe & Crompton (2006). tal. et 3305 Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3306 R. H. Crompton et al. Review. Arboreal origins of hominin bipedalism a dynamic model of quadrupedalism in Ardipithecus asymmetry of the cuboid peg for the calcaneus would show that quadrupedalism would be possible, (figure 1a), and Lovejoy et al. (2009a) appear to if unstable, and very expensive, if perhaps less so follow this practice. The asymmetry and size of the than it would be for longer-legged humans. cuboid peg is not, however, entirely a reliable guide A parallel case has very recently been discussed in to the existence or absence of a functional MLA: as the literature. Pierolapithecus is described by Almećija can be seen in figure 1a, the peg is not overlapped ven- et al. (2009) as orthograde, but as lacking the obvious trally by the calcaneus. While it is usual to associate suspensory adaptations seen in the rather later Iberian this development with loss of the mid-tarsal break crown hominoid, Hispanopithecus (D.) laietanus.The (axis of plantarflexion), which is present in other metacarpophalangeal joints are described as adapted living apes (Lewis 1980), the absence of a mid-tarsal for use in dorsiflexion in palmigrade postures break is not universal in humans. Setting aside con- (Almećija et al. 2009). Reference to monkeys would ditions such as Charcot foot, where soft-tissue failure suggest that it was an arboreal quadruped (Moya`- arising from diabetes or directly from neurological Sola` et al. 2004), but these authors now regard it as conditions results in collapse of the lateral midfoot orthograde in body plan (Almećija et al. 2009) but and in some cases a midfoot pressure peak under, or moving by palmigrade quadrupedalism. The authors near, the calcaneocuboid joint, figure 1b (unfortu- also suggest that the description of the Miocene pon- nately from uncalibrated pressure plate data, but gine Sivapithecus by Madar et al. (2002) may suggest qualitatively reliable) shows that clinically normal indi- similar behaviour. It is very difficult to reconcile an viduals may also show a lateral midfoot pressure peak. orthograde body plan with quadrupedal locomotion, It is interesting that this individual also shows absence even when there are no claims that Pierolapithecus of a lateral-to-medial path of the centre of pressure was a terrestrial biped. The obvious, simple solution and, in figure 1c, a single-peaked vGRF, with a non- is again the one we propose here, that it (and perhaps human-ape-like slow tailing-off of vGRF at ‘toe-off ’. even Sivapithecus) was an orthograde clamberer which, While the absence of the mid-tarsal break does seem in the absence of marked suspensory adaptations, used functionally linked with an extended toe-off, neither hand-compressive climbing techniques below shoulder are therefore universal features of hominins. Neverthe- level, in other words, hand-assisted bipedalism and less, if Ardipithecus lacked a human-like cuboid peg, other components of the compressive-orthogrady lateral-foot stability would be limited. In both cases, continuum best exemplified today in orangutans. a certain degree of rigidity provided by retention of a thick plantar fibrous layer would improve the capacity of the lateral metatarsals to deliver accelerative force 4. THE LEGACY OF ARBOREAL ORIGINS FOR from a more effective, relatively anterior, position. HUMAN BIPEDALITY We (Pataky et al.2008) recently demonstrated a The arboreal habitat differs markedly in one major negative correlation of plantar pressure with walking mechanical respect from the terrestrial: it is compliant speed in humans, which implies reduced collapse of (Alexander 2003) and thus unstable, as it can be set the MLA, and thus increased stiffness. This may be vibrating by imposed forces. Arboreal mammals need directly beneficial to force transmission to the ground. to have strategies for dealing with this compliance. It is also important in enabling control of gear ratios, Schmitt (1999) has shown that limb flexion (limb and thus in tuning muscles to enhance performance compliance) is one such response, but limb flexion during constant-speed running by applying pre-tension requires muscle power to maintain stable flexed pos- during landing, while optimizing them also for effi- tures. For this reason, most probably, muscle masses ciency or power at toe-off (Carrier et al. 1994). tend to be higher in arboreal animals (Degabriele & Perhaps most importantly, we can optimize muscle Dawson 1979), while terrestrial cursor limbs have properties during rapid changes in speed and changes short muscle bellies and long tendons (Alexander in incline in both running and walking (Lichtwark & 2003). To what extent has this legacy of compliance Wilson 2006, 2007, 2008). We have suggested that influenced early hominin evolution? increased stiffness results from pre-tension applied to the plantar aponeurosis (PA) by heel-strike or early- (a) A compliant foot? stance muscle activity (in triceps, tibialis anterior and Lovejoy et al. (2009a) argue that whereas the living the digital dorsiflexors; Pataky et al.2008; Caravaggi non-human great apes have acquired compliant feet, et al. 2009). The windlass mechanism created as the to enable them to grip branches more effectively, and PA wraps round the heads of the metatarsals humans have of course acquired a MLA, Ardipithecus (figure 1a) is known to contribute to stiffen the foot is again conservative in the plantar foot, lacking a in late stance (Hicks 1954) by pulling on the calcaneus, MLA, but retaining a thick and fibrous layer on the causing inversion of the subtalar joint and hence ‘lock- plantar aspect of the foot, like that of cercopithecines, ing’ the midtarsal joint (Tansey & Briggs 2001). A contrasting with the loss of such a thick aponeurotic dynamic model of the plantar foot constructed in our layer in the non-human apes, which gain thereby in laboratory (Caravaggi et al.2009) however shows that foot adaptability to irregular substrates. the PA is also pre-tensioned in early stance, from Following Bojsen-Møller (1979), it is common in heel-strike onwards, as proposed by Pataky et al. the hominin palaeontology literature (e.g. Lewis (2008), and the tension appears to increase with walk- 1980; Berillon 2000; Harcourt-Smith & Aiello 2004; ing speed. The predicted tension (verified against Jungers et al. 2009) to assess the presence or absence cadaveric data from Gefen (2003)) increases from lat- of a MLA by the degree of development and eral to medial, and ranges from 0.47 body weight at

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Review. Arboreal origins of hominin bipedalism R. H. Crompton et al. 3307

(a) (b)

metatarsal heads

spring ligament talonavicular joint, with cuboid peg calcaneocuboid joint forms calcaneocuboid ‘midtarsal’ or joint transverse tarsal joint the 5 slips of the head of the plantar talus aponeurosis

force (N) 600 400 200 0 0 50 100 150 200 250 300 time (ms) Figure 1. (a) Diagram of the plantar aspect of the human foot, showing the position of the cuboid peg and illustrating the ‘windlass mechanism’, whereby the five slips of the PA are tensed by the curvature of the metatarsal heads as the metatarso- phalangeal joint dorsiflexes. Similarly, the spring ligament is tensed by plantad motion of the head of the talus. Both mechanisms act to stiffen the median longitudinal arch during stance. (Figure modified from image from Primal’s ‘Anatomy TV’). (b) Peak pressures, in false greyscale, where lighter tone indicates higher pressure, during bipedalism of a clinically normal subject recorded by Nike Inc., courtesy of J.-P. Wilssens of RSscan International. Dots indicate the path of the centre of pressure under the foot. (c) vGRF curve calculated from the same data for the individual featured in (a). heel-strike to a peak 1.5 BW, generating vertical forces abducted as Lovejoy et al. (2009a) report, the degree which sustain the MLA and metatarsals. Thus, the of abduction is comparable to that in living gibbons MLA is supported through much of stance by soft (e.g. Vereecke et al. 2005; Crompton et al. 2008) tissue: stiffening of the PA, as well as bone shape, con- and perhaps Oreopithecus. While Moya`-Sola` et al. tributes directly and very substantially to the existence (1999) suggested that the extent of hallucal abduction of the MLA. An assumption that lack of a human-like in Oreopithecus would not have been compatible with cuboid peg (figure 1a) implies lack of a MLA is other than postural bipedalism, Vereecke et al. unsafe without extensive investigation of the possibility (2006a,b) have shown that gibbons, despite compliant of soft-tissue stiffening. The case of human individuals feet with widely abducted halluces, can sustain run- with a mid-tarsal break suggests that sustained vGRFs ning on the ground for some hundred yards and and a substantial hallucal toe-off depend on stiffening attain absolute speeds equalling the human walk–run of both the medial and lateral foot by soft-tissue ten- transition. Neither do compliant feet prevent the sioning throughout stance. non-human great apes from walking bipedally, terrest- It is notable that Vereecke et al. (2003) have shown rially or arboreally. High robusticity of metatarsals two that foot pressure records of human bipedalism are and three is also a feature of Ar. ramidus (Lovejoy et al. much less variable between strides and between indi- 2009a) and suggests that these digits may have been viduals than those of the bipedalism of other more important in applying accelerative parasagittal hominoids. In humans, forces are applied in a more force than the hallux, while the abducted hallux pro- consistent manner, particularly by the hallux, which vided grip on branches. However, even in human plays a limited propulsive role in most non-human bipedal walking, plantar pressure tends to be lower apes, and may act more as a balancing structure under the hallux, and greater under metatarsal heads during bipedalism. If the hallux of Ardipithecus is as two and three, in flat-footed humans or humans who

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3308 R. H. Crompton et al. Review. Arboreal origins of hominin bipedalism

a)(1.4( b)

1.2

1.0

0.8

0.6 forelimb length (m) 0.4 hindlimb length (m)

0.2

(c)(1.5 d) ) ) 2 2

1.0

0.5 forelimb moment of inertia (kg m hindlimb moment of inertia (kg m

0 Gorilla Pan PongoHylobates Homo Equus Gorilla Pan PongoHylobates Homo Equus Figure 2. (a,b) Chart showing the limb lengths and (c,d) moments of inertia for hominoids and horse. Forelimb moments of inertia are about the shoulder joint and hindlimb moments of inertia are about the hip joint. Great ape data are from Isler et al. (2006), human data are from Winter (1990) using a median male height and weight from the GEBOD database (Cheng et al. 1994), horse data from Buchner et al. (1997). have been brought up as barefoot walkers (D’Aouˆt dedicated terrestrial cursor (horse). All dimensions et al. 2009). have been geometrically scaled to the estimated mass Thus, the human foot is less distinct than is often of Ar. ramidus, 50 kg, using mass1/3 for lengths and thought from that of other great apes. It has built on mass5/3 for moments of inertia. All the arboreal species the compliant arboreal legacy (whether prior to or have longer than expected forelimbs, but, except for after the separation from panins we submit is not yet humans and gibbons, hindlimb length is not greatly clear, given the mixed message of Ardipithecus; Lovejoy different from that of the specialist cursor. However, et al. 2009a) by becoming a variable-gear organ, able when looking at the moments of inertia it can clearly to change its stiffness to accommodate to speed, as be seen how elongated limbs with heavy autopodia well as to support compliance and irregularity. lead to extremely large moments of inertia when compared with the values seen in horses. This has inevitable but complex effects in terms of top speed (b) Limb mass proportions and efﬁciency. Long, high-inertia legs are perfectly Another likely legacy from a recently arboreal past is efﬁcient for the pendular mechanics of slow walking, the partial retention of arboreal limb mass pro- but the high-speed spring mechanics of running require portions. A cursorial animal needs to accelerate its low moments of inertia to minimize the internal energy limbs rapidly. Rapid acceleration can be achieved lost per step. Hylobates seems to some extent to have with less energy if the moments of inertia are reduced, dealt with the inertial problem of very long legs by redu- and this is commonly achieved by a reduction in distal cing distal muscle mass. We suggest, following limb elements (Hildebrand 1995). Figure 2 shows Channon et al.(2009), that several aspects of gibbon a comparison of the inertial properties and limb anatomy may relate to an unrecognized importance of dimensions of hominoids in comparison with a leaping in the gibbon locomotor repertoire.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Review. Arboreal origins of hominin bipedalism R. H. Crompton et al. 3309

However, upper limb–lower limb proportions and 10 the distribution of lengths and mass within limbs also affect their swing frequency. Part of the efficiency of long distance human walking at least depends on a forward swing of the contralateral arm to counteract the horizontal torque applied to the body by the swing leg (e.g. Li et al. 2001) which, among other 1 effects, interferes with lateral stability. These swings occur even in short-distance walking of young children of similar stature/mass to Au. afarensis, and increase in magnitude with walking speed (Li et al. 2001), so they are of relevance to any consideration of early hominin locomotion. Match between the natural pendular 10−1 period (NPP), and hence swing-time of upper and tendon mass fraction (%) lower limbs, affects efficiency of all gaits, bipedal and quadrupedal; and distribution of mass within the limb affects the NPP (Isler et al. 2006). The distal position of the centre of mass of the forelimb of most great apes, with the exception of the chimpanzee, means 10−2 that there is a considerable mismatch with forelimb hare NPP, and segment proportions are thus not well opti- gorilla human gibbon reindeer mized for quadrupedal gaits (Isler et al. 2006). orangutan chimpanzee greyhound Chimpanzees may have modified their limb mass distribution for more efficient quadrupedalism, Figure 3. Total tendon in both hind limbs as a fraction of body suggesting that the last common African ape ancestor mass. Data are based on information from Pierrynowski was not a proficient quadruped (Isler et al. 2006). (1995), Payne et al.(2006), Williams et al.(2007, 2008)and However, experimental work on the oxygen consump- Wareing et al. (submitted). tion of both bipedal and quadrupedal locomotion in chimpanzees confirms that they are relatively ineffi- cient in both modalities (Sockol et al. 2007), 2007). Resulting volumes were converted to tendon compared with both modern humans and quadrupeds mass using a density of 1100 kg m23 (Watson & of equivalent body size. This may indicate that maxi- Wilson 2007). It is clear that tendon mass is highly mum terrestrial speed, rather than minimal terrestrial variable and more often related to high speed and energy cost, may be the target of selective pressure high acceleration than to efficiency. It is also clear for chimpanzees. (By contrast, preliminary data from that while humans have appreciably increased their the same research group suggest that metabolic costs hindlimb tendon proportion when compared with of bipedalism in the orangutan are some of the the other apes, they are still a long way from the lowest recorded locomotor costs for the body size; much higher proportions in more specialized cursorial personal communication from H. Pontzer (2009)). quadrupeds—particularly those specializing in explo- sive acceleration rather than long-distance efficiency. This is emphasized if we multiply through by (c) Proportion of mass as tendon 2500 J kg21 to express elastic storage capacity in Another area where there is still a considerable legacy terms of energy. Most hominoids have about 1 J kg21 from the recent arboreal past is in the amount of of elastic energy storage, humans nearly three times tendon in the hindlimbs of hominoids compared that, reindeer 10 J kg21 but greyhounds fully with cursorial animals. Figure 3 shows the mass of 100 J kg21. Thus, while humans have adjusted hindlimb tendon in both hindlimbs as a proportion tendon mass somewhat to enhance terrestrial running, of body mass. This parameter is informative since its perhaps retaining a relatively large muscle mass to biomechanical interpretation is independent of allow for adjustments to optimize the hindlimb for ter- moment arm data, of which there is very little available rain, speed and support characteristics, the African for comparison in non-primates. Tendon acts as a apes have not. Bonobos and lowland gorillas may not simple damped spring during locomotion, so the require to do so because of limited terrestriality; amount of energy that can be stored depends on the mountain gorillas may be protected by size, but chim- mass. The strain energy storage of tendon is panzees may simply have modified mass proportions to 2500 J kg21 at 8 per cent strain (Vogel 2003), which match hindlimb–forelimb swing frequencies better, is therefore the limit of the amount of elastic strain suggesting that the selective pressures associated with energy that is potentially available per gait cycle— movement in an unstable arboreal milieu remain whether for power amplification or energy saving. strong. For the hominoids, tendon mass was estimated using We can further investigate the role of elasticity in published tendon length and muscle physiological human locomotion by simulation. In recent simulation cross-section area data. Tendon cross-section area work (Sellers et al. 2010), the role of tendon elasticity was estimated by assuming 6 per cent strain at maxi- was quantified by ‘virtual ablation’. In this paradigm, mal isometric contractile force and muscle simulations are repeated with identical anatomical contraction stress of 300 kN m22 (Sellers & Manning models except for the structure of interest, which is

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3310 R. H. Crompton et al. Review. Arboreal origins of hominin bipedalism

a) 6( expectations from body size but also to Ar. ramidus’ )

−1 5 clear adaptations for one form of compressive ortho- 4 grady, terrestrial bipedal walking. 3 Secondly, we argue that from our arboreal ancestors 2 humans have inherited feet and legs that can adapt to a 1

velocity (m s large variety of terrains, support compliances and 0 speeds. However, at some stage in our evolution we (b) have departed some way from other hominoids in 4.0 3.5 adaptation for energy-efﬁcient running. This combi- 3.0

) nation probably has a lot to do with our ability to 2.5 −1 2.0 outrun horses in trials such as those over 22 miles of m 1.5 −1 hilly mid-Wales or 50 miles of sand-dunes in the 1.0 0.5 United Arab Emirates. But we are not fast runners

(J kg 0 (see Bramble & Lieberman 2004), and in terms of energy storage have a very long way to go to catch net cost of locomotion up with dogs bred for hunting. Early human ancestors

stiff tendons would clearly have been no match for a cursorial normal tendons predator, so that it is perhaps fortunate that along with late retention of long forearms (Dunsworth et al. 2003), which would improve throwing distance normal tendons andstiff stiff tendons AT and normal AT if not accuracy, part of our arboreal inheritance was Figure 4. (a) Chart showing the maximum velocity and (b) powerful leg muscles, which remain very helpful for the cost of locomotion of human running simulations climbing trees! where the elastic properties of the hindlimb tendons are R.H.C. thanks Alan Walker and Chris Stringer for the manipulated. AT, Achilles tendon. (Sellers et al. 2010). invitation to participate in this meeting. We thank the Royal Society, the Leverhulme Trust and the Natural Environment Research Council, the Biotechnology and Biological Sciences Research Council and the Engineering removed or altered in some versions of the model. This and Physical Sciences Research Council for supporting our allows the effect of a specific structure to be isolated in research, and Chester and Twycross Zoos for permission to much the same way as classical ablation experiments study the hominoids under their care. R.H.C. and W.I.S. but without the danger of side effects (let alone ethical also thank the NorthWest Grid and the Science and difficulties) associated with performing such exper- Technology Facilities Council Daresbury Laboratory for iments surgically. In this case, the elastic effect of access to high-performance computing facilities. tendons was removed by making them 100 times their normal stiffness, so that the simulated tendons were unable to store appreciable amounts of energy in the simulation. Four experimental conditions were REFERENCES compared: normal hindlimb tendons; all hindlimb ten- Alexander, R. McN. 1991 Characteristics and advantages of dons stiff; all normal hindlimb tendons except for a human bipedalism. In Biomechanics in evolution stiff Achilles tendon; all tendons stiff except for a (eds J. M. V. Rayner & R. J. Wooton), pp. 225–266. normal Achilles tendon. Figure 4 shows that for Cambridge, UK: Cambridge University Press. Alexander, R. McN. 2003 Principles of animal locomotion. humans, the presence of tendon has only a moderate Princeton, NJ: Princeton University Press. effect on the maximum running speed of the simu- Almećija,A.,Alba,D.M.&Moya`-Sola`,S.2009Pierolapithecus lation but a very marked effect on the net cost of and the functional morphology of Miocene ape hand locomotion, and that this effect was mostly produced phalanges: paleobiological and evolutionary implications. by the Achilles tendon. This confirms the critical J. Hum. Evol. 57, 284–297. (doi:10.1016/j.jhevol.2009. role in efficient, high-performance running of a sub- 02.008) stantial Achilles tendon, which is missing from all Berillon, G. 2000 Le Pied des Hominoı¨des Mioce`nes et des hominoids except humans, gibbons and siamangs. Hominides Fossiles. Paris, France: E´ ditions du CRNS. Whether such a structure is present in fossil hominins Bojsen-Møller, F. 1979 Calcaneocuboid joint and stability of (and which) is currently unknown but evidence of its the longitudinal archof the foot at high and low gear push presence, perhaps by analysis of calcaneal microstruc- off. J. Anat. 129, 165–176. Bradley, B. J. 2008 Reconstructing phylogenies and pheno- ture, would probably be diagnostic of running ability. types: a molecular view of human evolution. J. Anat. 212, 337–353. (doi:10.1111/j.1469-7580.2007.00840.x) Bramble, D. M. & Lieberman, D. L. 2004 Endurance 5. CONCLUSIONS running and the evolution of Homo. Nature 432, We argue that, given the large body mass of Ar. rami- 345–352. (doi:10.1038/nature03052) Buchner, H., Savelberg, H., Schamhardt, H. & Barneveld, dus, typical of both living and extinct hominoids, it is A. 1997 Inertial properties of Dutch Warmblood more likely that when it moved in the trees it made horses. J. Biomech. 30, 653–658. (doi:10.1016/S0021- use of compressive orthogrady, which we suggest 9290(97)00005-5) may be the oldest crown-hominoid locomotor adap- Caravaggi, P., Pataky, T. C., Goulermas, J. Y., Savage, R. & tation, than that it adopted a monkey-like Crompton, R. H. 2009 A dynamic model of the windlass quadrupedalism. This would run counter not only to mechanism of the foot: evidence for early stance phase

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Review. Arboreal origins of hominin bipedalism R. H. Crompton et al. 3311

preloading of the plantar aponeurosis. J. Exp. Biol. 212, Gebo, D. L. 1992 Plantigrady and foot adaptation in African 2491–2499. (doi:10.1242/jeb.025767) apes: implications for hominid origins. Am. J. Phys. Carey, T. S. & Crompton, R. H. 2005 The metabolic costs of Anthropol. 89, 29–58. (doi:10.1002/ajpa.1330890105) ‘bent hip, bent knee’ walking in humans. J. Hum. Evol. Gebo, D. L. 1996 Climbing, brachiation, and terrestrial 48, 25–44. (doi:10.1016/j.jhevol.2004.10.001) quadrupedalism: historical precursors of hominid Carrier, D. R., Heglund, N. C. & Earls, K. D. 1994 Variable bipedalism. Am. J. Phys. Anthropol. 101, 55–92. gearing during locomotion in the human musculoskeletal (doi:10.1002/(SICI)1096-8644(199609)101:1,55::AID- system. Science 265, 651–653. (doi:10.1126/science. AJPA5.3.0.CO;2-C) 8036513) Gefen, A. 2003 The in vivo elastic properties of the plantar Channon, A. J., Gu¨nther, M. M., Crompton, R. H. & Vereecke, fascia during the contact phase of walking. Foot Ankle E. E. 2009 Mechanical constraints on the functional mor- Int. 24, 238–244. phology of the gibbon hind limb. J. Anat. 215,383–400. Goodall, J. 1998 Through a window: my thirty years with the (doi:10.1111/j.1469-7580.2009.01123.x) chimpanzees of Gombe. Boston, MA: Houghton Mifflin. Cheng,H.,Obergefell,L.&Rizer,A.1994Generator of Harcourt-Smith, W. E. H. & Aiello, L. C. 2004 Fossils, feet body data (GEBOD) manual. Report no. AL/CF-TR-1994- and the evolution of human bipedal locomotion. J. Anat. 0051, Armstrong Lab, Wright-Patterson Air Force Base. 204, 403–416. (doi:10.1111/j.0021-8782.2004.00296.x) Clarke, R. J. 1999 Discovery of a complete arm and hand of Hicks, J. H. 1954 The mechanics of the foot. II. The plantar the 3.3 million-year-old Australopithecus skeleton from aponeurosis and the arch. J. Anat. 88, 25–30. Sterkfontein. S. Afr. J. Sci. 95, 477–480. Hildebrand, M. 1995 Analysis of vertebrate structure. Clarke, R. J. 2002 Newly revealed information on the New York, NY: Wiley. Sterkfontein Member 2 Australopithecus skeleton from Hunt, K. D. 1992 Positional behavior of Pan troglodytes in the Sterkfontein. S. Afr. J. Sci. 98, 523–526. Mahale Mountains and Gombe Stream National Parks, Crompton, R. H., Li, Y., Wang, W., Gu¨nther, M. M. & Tanzania. Am. J. Phys. Anthropol. 87, 83–105. (doi:10. Savage, R. 1998 The mechanical effectiveness of 1002/ajpa.1330870108) erect and ‘bent-hip, bent-knee’ bipedal walking in Hunt, K. D. 2004 The special demands of great ape loco- Australopithecus afarensis. J. Hum. Evol. 35, 55–74. motion and posture. In The evolution of thought: (doi:10.1006/jhev.1998.0222) evolutionary origins of great ape intelligence (eds A. E. Crompton, R. H. et al. 2003 The biomechanical evolution Russon & D. R. Begun), pp. 172–189. Cambridge, of erect bipedality. Cour. Forsch.-Inst. Senckenberg 243, UK: Cambridge University Press. 115–126. Isler, K. 2002 Characteristics of vertical climbing in African Crompton, R. H., Vereecke, E. E. & Thorpe, S. K. S. 2008 apes. Senckenbergiana Lethaea 82, 115–124. (doi:10. Locomotion and posture from the common hominoid 1007/BF03043777) ancestor to fully modern hominins with special reference Isler, K. 2003 Aspects of vertical climbing in gorillas. Cour. to the common panin/hominin ancestor. J. Anat. 212, Forsch-Inst. Senckenberg 243, 71–78. 501–543. (doi:10.1111/j.1469-7580.2008.00870.x) Isler, K. & Thorpe, S. K. S. 2003 Gait parameters in vertical Culotta, E. 2004 Spanish fossil sheds new light on the oldest climbing of captive, rehabilitant and wild Sumatran Great apes. Science 306, 1273–1274. orang-utans (Pongo pygmaeus abelii). J. Exp. Biol. 206, Dainton, M. & Macho, G. A. 1999 Did knuckle walking 4081–4096. (doi:10.1242/jeb.00651) evolve twice? J. Hum. Evol. 36, 171–194. (doi:10.1006/ Isler, K., Payne, R. C., Gu¨nther, M. M., Thorpe, S. K. S., jhev.1998.0265) Li, Y., Savage, R. & Crompton, R. H. 2006 Inertial D’Aouˆt, K., Pataky, T. C., De Clercq, D. & Aerts, P. 2009 properties of hominoid limb segments. J. Anat. 209, The effects of habitual footwear use: foot shape and func- 201–218. (doi:10.1111/j.1469-7580.2006.00588.x) tion in native barefoot walkers. Footwear Science 1, 81–94. Jungers, W. L., Harcourt-Smith, W. E. H., Wunderlich, R. Darwin, C. 1871 The descent of man and selection in relation to E., Tocheri, M. W., Larson, S. G., Sutikna, T., Rhokus sex. London, UK: John Murray. Awe Due & Morwood, M. J. 2009 The foot of Homo Degabriele, R. & Dawson, T. J. 1979 Metabolism and heat floresiensis. Nature 459, 81–84. (doi:10.1038/ balance in an arboreal marsupial, the koala (Phascolarctos nature07989) cinereus). J. Comp. Physiol. B 134, 293–301. (doi:10.1007/ Keith, A. 1903 The extent to which the posterior segments BF00709996) of the body have been transmuted and suppressed DeSilva, J. M. 2009 Functional morphology of the ankle and in the evolution of man and related primates. J. Anat. the likelihood of climbing in early hominins. Proc. Natl Physiol. 37, 18–40. Acad. Sci. USA 106, 6567–6572. (doi:10.1073/pnas. Keith, A. 1923 Man’s posture: its evolution and disorders. 0900270106) Br. Med. J. 1, 451–454, 499–502, 545–548, 587–590, Doran, D. M. 1996 The comparative positional behaviour of 624–626, 669–672. (doi:10.1136/bmj.1.3246.451) the African apes. In Great ape societies (eds W. McGrew & Ker, R. F., Bennet, M. B., Bibby, S. R., Kester, R. C. & T. Nishida), pp. 213–224. Cambridge, UK: Cambridge Alexander, R. McN. 1987 The spring in the arch of the University Press. human foot. Nature 325, 147–149. (doi:10.1038/ Dunsworth, H., Challis, J. H. & Walker, A. 2003 Throwing 325147a0) and bipedalism: a new look at an old idea. Cour. Forsch.- Kirkpatrick, R. C., Gu, H. J. & Zhou, X. P. 1999 A Inst. Senckenberg 243, 105–110. preliminary report on Sichuan snub-nosed monkeys Filler, A. 2007 Homeotic evolution in the Mammalia: diver- (Rhinopithecus roxellana) at Baihe nature reserve. Folia sification of Therian axial seriation and the Primatol. 70, 117–120. (doi:10.1159/000021683) morphogenetic basis of human origins. PLoS ONE 2, Kivell, T. L. & Begun, D. R. 2007 Frequency and timing of 10, e1019. (doi:10.1371/journal.pone.0001019) scaphoid-centrale fusion in hominoids. J. Hum. Evol. 52, Fleagle, J. G. 1998 Primate adaptation and evolution. 321–340. (doi:10.1016/j.jhevol.2006.10.002) New York, NY: Elsevier. Kivell, T. L. & Schmitt, D. 2009 Independent evolution of Fleagle, J. G., Stern, J. T., Jungers, W. L., Susman, R. L., knuckle-walking in African apes shows that humans did Vangor, A. K. & Wells, J. P. 1981 Climbing: a biomecha- not evolve from a knuckle-walking ancestor. Proc. Natl nical link with brachiation and with bipedalism. Symp. Acad. Sci. USA 106, 14 241–14 246. (doi:10.1073/pnas. Zool. Soc. Lond. 48, 359–375. 0901280106)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3312 R. H. Crompton et al. Review. Arboreal origins of hominin bipedalism

Ko¨hler, M. & Moya`-Sola`, S. 1997 Ape-like or hominid-like? Morton, D. J. 1922 Evolution of the human foot I. The positional behavior of Oreopithecus bambolii reconsid- Am. J. Phys. Anthropol. 5, 305–336. (doi:10.1002/ajpa. ered. Proc. Natl Acad. Sci. USA 94, 11 747–11 750. 1330050409) (doi:10.1073/pnas.94.21.11747) Moya`-Sola`,S.&Ko¨hler, M. 1996 The first Dryopithecus Kramer, P. A. 1999 Modeling the locomotor energetics of skeleton: origins of great-ape locomotion. Nature 379, extinct hominids. J. Exp. Biol. 202, 2807–2818. 156–159. (doi:10.1038/379156a0) Kramer, P. A. & Eck, G. G. 2000 Locomotor energetics Moya`-Sola`, S., Ko¨hler, M. & Rook, L. 1999 Evidence of a and leg length in hominid bipedality. J. Hum. Evol. 38, hominid-like precision grip capability in the hand of the 651–666. (doi:10.1006/jhev.1999.0375) Miocene ape Oreopithecus. Proc. Natl Acad. Sci. USA 96, Lahm, A. A. 1986 Diet and habitat preference of 313–317. (doi:10.1073/pnas.96.1.313) Mandrillus sphinx in Gabon: implications of foraging Moya`-Sola`,S.,Ko¨hler, M., Alba, D. M., Casanovas-Vilar, I. & strategy. Am. J. Primatol. 11, 9–26. (doi:10.1002/ajp. Galindo, J. 2004 Pierolapithecus catalaunicus,anewMiddle 1350110103) Miocene great ape from Spain. Science 306, 1339–1344. Larson, S. G. 1998 Parallel evolution in the hominoid (doi:10.1126/science.1103094) trunk and forelimb. Evol. Anthropol. 6, 87–89. (doi:10. Nagano, A., Umberger, B. R., Marzke, M. W. & Gerritsen, 1002/(SICI)1520-6505(1998)6:3,87::AID-EVAN3.3.0. K. G. M. 2005 Neuromusculoskeletal computer model- CO;2-T) ing and simulation of upright, straight-legged, bipedal Lewis, O. J. 1980 The joints of the evolving foot part II, the locomotion of Australopithecus afarensis (A.L. 288-1). intrinsic joints. J. Anat. 131, 275–298. Am. J. Phys. Anthropol. 126, 2–13. (doi:10.1002/ajpa. Li, Y. 2001 The seasonal diet of the Sichuan snub-nosed 10408) monkey (Pygathrix roxellana) in Shennongjia Nature Oishi, M., Ogihara, N., Endo, H., Ichihara, N. & Asari, M. Reserve, China. Folia Primatol. 72, 40–43. (doi:10. 2009 Dimensions of forelimb muscles in orangutans and 1159/000049919) chimpanzees. J. Anat. 215, 373–382. (doi:10.1111/j. Li, Y., Wang, W., Crompton, R. H. & Gunther, M. M. 2001 1469-7580.2009.01125.x) Free vertical moments and transverse forces in human Pataky, T. C., Caravaggi, P., Savage, R., Parker, D., walking and their role in relation to arm-swing. J. Exp. Goulermas, J. Y., Sellers, W. I. & Crompton, R. H. Biol. 204, 47–58. 2008 New insights into the plantar pressure correlates Li, Y., Stanford, C. B. & Yang, Y. 2002 Winter feeding tree of walking speed using pedobarographic statistical para- choice in Sichuan snub-nosed monkeys (Rhinopithecus metric mapping (pSPM). J. Biomech. 41, 1987–1994. roxellanae) in Shennongjia Nature Reserve, China. (doi:10.1016/j.jbiomech.2008.03.034) Int. J. Primatol. 23, 657–675. Payne, R. C., Crompton, R. H., Isler, K., Savage, R., Lichtwark, G. A. & Wilson, A. M. 2006 Interactions between Vereecke, E. E., Gu¨nther, M. M., Thorpe, S. K. S. & the human gastrocnemius muscle and the Achilles tendon D’Aouˆt, K. 2006a Morphological analysis of the during incline, level and decline locomotion. J. Exp. Biol. hindlimb in apes and humans. I. Muscle architecture. 209, 4379–4388. (doi:10.1242/jeb.02434) J. Anat. 208, 709–724. (doi:10.1111/j.1469-7580.2006. Lichtwark, G. A. & Wilson, A. M. 2007 Is Achilles tendon 00563.x) compliance optimised for maximum muscle efficiency Payne, R. C., Crompton, R. H., Isler, K., Savage, R., during locomotion? J. Biomech. 40, 1768–1775. Vereecke, E. E., Gu¨nther, M. M., Thorpe, S. K. S. & (doi:10.1016/j.jbiomech.2006.07.025) D’Aouˆt, K. 2006b Morphological analysis of the Lichtwark, G. A. & Wilson, A. M. 2008 Optimal muscle hindlimb in apes and humans. II. Moment arms. fascicle length and tendon stiffness for maximising J. Anat. 208, 725–742. (doi:10.1111/j.1469-7580.2006. gastrocnemius efficiency during human walking and 00564.x) running. J. Theor. Biol. 252, 662–673. (doi:10.1016/j. Pierrynowski, M. R. 1995 Analytical representation of jtbi.2008.01.018) muscle line of action and geometry. In Three-dimensional Lovejoy, C. O., Latimer, B., Suwa, G., Asfaw, B. & White, analysis of human movement (eds P. Allard, I. A. F. T. D. 2009a Combining prehension and propulsion: the Stokes & J. P. Blanch), pp. 215–256. Champaign, IL: foot of Ardipithecus ramidus. Science 326, 72e1–72e8. Human Kinetic Publishers. (doi:10.1126/science.1175832) Pontzer, H. & Wrangham, R. W. 2004 Climbing and the Lovejoy,C.O.,Simpson,S.W.,White,T.D.,Asfaw,B.& daily energy cost of locomotion in wild chimpanzees: Suwa,G.2009b Careful climbing in the Miocene: the implications for hominoid locomotor evolution. forelimbs of Ardipithecus ramidus and humans are J. Hum. Evol. 46, 315–333. (doi:10.1016/j.jhevol.2003. primitive. Science 326,70e1–70e8.(doi:10.1126/science. 12.006) 1175827) Remis, M. 1994 Feeding ecology and positional behavior of Lovejoy,C.O.,Suwa,G.,Simpson,S.W.,Matternes,J.H.& western lowland gorillas (Gorilla gorilla gorilla) in the White,T.D.2009c The great divides: Ardipithecus ramidus Central African Republic. PhD thesis, Yale University, reveals the postcrania of our last common ancestors with New Haven, CT. African apes. Science 326, 100–106. (doi:10.1126/science. Remis, M. 1995 Effects of body size and social context on 1175833) the arboreal activities of lowland gorillas in the Central MacLatchy, L., Gebo, D., Kityo, R. & Pilbeam, D. 2000 African Republic. Am. J. Phys. Anthropol. 97, 413–433. Postcranial functional morphology of Morotopithecus (doi:10.1002/ajpa.1330970408) bishopi, with implications for the evolution of modern Richmond, B. G. & Strait, D. S. 2000 Evidence that humans ape locomotion. J. Hum. Evol. 39, 159–183. (doi:10. evolved from a knuckle-walking ancestor. Nature 404, 1006/jhev.2000.0407) 382–385. (doi:10.1038/35006045) Madar, S. I., Rose, M. D., Kelley, J., MacLatchy, L. & Richmond, B. G. & Jungers, W. L. 2008 Orrorin tugenensis Pilbeam, D. 2002 New Sivapithecus postcranial specimens femoral morphology and the evolution of hominin from the Siwaliks of Pakistan. J. Hum. Evol. 42, 705–752. bipedalism. Science 319, 1662–1665. (doi:10.1126/ (doi:10.1006/jhev.2002.0554) science.1154197) Meldrum, D. J. 1993 On plantigrady and quadrupedalism. Richmond, B. G., Begun, D. R. & Strait, D. S. 2001 Origin Am. J. Phys. Anthropol. 91, 379–385. (doi:10.1002/ajpa. of human bipedalism: the knuckle-walking hypothesis 1330910310) revisited. Yrbk. Phys. Anthropol. 44, 70–105.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Review. Arboreal origins of hominin bipedalism R. H. Crompton et al. 3313

Rowe, N. 1996 The pictorial guide to the living primates. Thorpe, S. K. S., Crompton, R. H. & Alexander, R. M. cN. Charlestown, RI: Pogonias Press. 2007a Orangutans use compliant branches to lower the Sayers, K. & Lovejoy, C. O. 2008 The chimpanzee has energetic cost of locomotion. Biol. Lett. 3, 253–256. no clothes. Curr. Anthropol. 49, 87–114. (doi:10.1086/ (doi:10.1098/rsbl.2007.0049) 523675) Thorpe, S. K. S., Holder, R. L. & Crompton, R. H. 2007b Schaller, G. B. 1963 The mountain gorilla. Chicago, IL: Origin of human bipedalism as an adaptation for University of Chicago Press. locomotion on flexible branches. Science 316, 1328– Schmitt, D. 1999 Compliant walking in primates. J. Zool. 1331. (doi:10.1126/science.1140799) (Lond.) 248, 149–160. (doi:10.1111/j.1469-7998.1999. Thorpe, S. K. S., Hodder, R. & Crompton, R. H. 2009 tb01191.x) Orangutans employ unique strategies to control branch Schultz, A. H. 1936 Characters common to higher primates flexibility. Proc. Natl Acad. Sci. USA 106, 12 646– and characters specific for man. Quart. Rev. Biol. 11, 12 651. (doi:10.1073/pnas.0811537106) 259–283, 425–455. Tuttle, R. H. & Watts, D. P. 1985 The positional behavior Sellers, W. I. & Manning, P. L. 2007 Estimating dinosaur and adaptive complexes of Pan gorilla.InPrimate maximum running speeds using evolutionary robotics. morphophysiology, locomotor analysis and human bipedalism Proc. R. Soc. B 274, 2711–2716. (doi:10.1098/rspb. (ed. S. Kondo), pp. 261–288. Tokyo, Japan: Tokyo 2007.0846) University Press. Sellers, W. I., Dennis, L. A. & Crompton, R. H. 2003 Tuttle, R. H., Butzer, K. W. & Blumenberg, B. 1974 Predicting the metabolic energy costs of bipedalism Darwin’s apes, dental apes, and the descent of man: using evolutionary robotics. J. Exp. Biol. 206, normal science in evolutionary anthropology. Curr. 1127–1136. (doi:10.1242/jeb.00205) Anthropol. 15, 389–426. (doi:10.1086/201494) Sellers, W. I., Dennis, L. A., Wang, W. & Crompton, R. H. Vereecke,E.E.,D’Aout,K.,DeClercq,D.,VanElsacker,L.& 2004 Evaluating alternative gait strategies using Aerts, P. 2003 Dynamic plantar pressure distribution evolutionary robotics. J. Anat. 204, 343–351. (doi:10. during terrestrial locomotion of Bonobos (Pan paniscus). 1111/j.0021-8782.2004.00294.x) Am.J.Phys.Anthropol.120, 373–383. (doi:10.1002/ajpa. Sellers, W. I., Cain, G., Wang, W. J. & Crompton, R. H. 10163) 2005 Stride lengths, speed and energy costs in walking Vereecke,E.E.,D’Aouˆt,K.,DeClercq,D.,VanElsacker,L.& of Australopithecus afarensis: using evolutionary robotics Aerts, P. 2005 Functional analysis of the gibbon foot during to predict locomotion of early human ancestors. terrestrial bipedal walking: plantar pressure distributions J. R. Soc. Interface 2, 431–442. (doi:10.1098/rsif.2005. and3Dgroundreactionforces.Am.J.Phys.Anthropol. 0060) 128, 659–669. (doi:10.1002/ajpa.20158) Sellers, W. I., Pataky, T. C., Caravaggi, P. & Crompton, Vereecke, E. E., D’Aouˆt, K. & Aerts, P. 2006a Speed modu- R. H. 2010 Evolutionary robotic approaches in primate lation in hylobatid bipedalism: a kinematical analysis. gait analysis. Int. J. Primatol. 31, 321–338. (doi:10. J. Hum. Evol. 51, 513–526. (doi:10.1016/j.jhevol.2006. 1007/s10764-010-9396-4) 07.005) Senut,B.,Pickford,M.,Gommery,D.,Mein,P.,Cheboi,K.& Vereecke, E. E., D’Aouˆt, K. & Aerts, P. 2006b The dynamics Coppens, Y. 2001 First hominid from the Miocene of hylobatid bipedalism: evidence for an energy-saving (Lukeino formation, Kenya). CR. Acad. Sci. Paris Ser. IIA mechanism? J. Exp. Biol. 209, 2829–2838. (doi:10. 332, 137–144. 1242/jeb.02316) Sockol, M. D., Raichlen, A. & Pontzer, H. 2007 Chimpan- Vogel, S. 2003 Comparative biomechanics, life’s physical world. zee locomotor energetics and the origin of bipedalism. Princeton, NJ: Princeton University Press. Proc. Natl Acad. Sci. USA 104, 12 265–12 266. (doi:10. Wang,W.J.&Crompton,R.H.2004Theroleofload-carrying 1073/pnas.0703267104) in the evolution of modern body proportions. J. Anat. 204, Stern, J. T. & Susman, R. L. 1993 The locomotor anatomy 417–430. (doi:10.1111/j.0021-8782.2004.00295.x) of Australopithecus afarensis. Am. J. Phys. Anthropol. 60, Wang, W. J., Crompton, R. H., Li, Y. & Gu¨nther, M. M. 279–317. (doi:10.1002/ajpa.1330600302) 2003 Energy transformation during erect and bent- Susman, R. L. 1984 The locomotor behavior of Pan paniscus hip-bent-knee walking by humans with implications for in the Lomako forest. In The Pygmy Chimpanzee (ed. R. L. the evolution of bipedalism. J. Hum. Evol. 44, 563–580. Susman), pp. 369–394. New York, NY: Plenum Press. (doi:10.1016/S0047-2484(03)00045-9) Tansey, P. A. & Briggs, P. J. 2001 Active and passive mech- Wang, W. J., Crompton, R. H., Carey, T. C., Gu¨nther, anisms in the control of heel supination. Foot Ankle M. M., Li, Y., Savage, R. & Sellers, W. I. 2004 Compari- Surg. 7, 131–136. (doi:10.1046/j.1460-9584.2001. son of inverse-dynamics musculo-skeletal models of AL 00264.x) 288-1 Australopithecus afarensis and KNM-WT 15000 Thorpe, S. K. S. & Crompton, R. H. 2005 The locomotor Homo ergaster to modern humans, with implications for ecology of wild orangutans (Pongo pygmaeus abelii) the evolution of bipedalism. J. Hum. Evol. 47, 453–478. in the Gunung Leuser Ecosystem, Sumatra, Indonesia: (doi:10.1016/j.jhevol.2004.08.007) a multivariate analysis using log-linear modelling. Ward, C. V., Leakey, M. G., Brown, B., Brown, F., Harris, Am. J. Phys. Anthropol. 127, 58–78. (doi:10.1002/ajpa. H. & Walker, A. 1999 South Turkwel: a new Pliocene 20151) hominid site in Kenya. J. Hum. Evol. 36, 69–95. Thorpe, S. K. S. & Crompton, R. H. 2006 Orang-utan (doi:10.1006/jhev.1998.0262) positional behavior and the nature of arboreal locomotion Washburn, S. L. 1967 Behaviour and the origin of man (The in Hominoidea. Am. J. Phys. Anthropol. 131, 384–401. Huxley Memorial Lecture 1967). Proc. Roy. Anthropol. (doi:10.1002/ajpa.20422) Inst. Great Brit. Ireland, pp. 21–27. Thorpe, S. K. S., Crompton, R. H., Gunther, M. M., Ker, Watson, J. C. & Wilson, A. M. 2007 Muscle architecture of R. F. & Alexander, R. M. cN. 1999 Dimensions and biceps brachii, triceps brachii and supraspinatus in the moment arms of the hind- and forelimb muscles of horse. J. Anat. 210, 32–40. (doi:10.1111/j.1469-7580. common chimpanzees (Pan troglodytes). Am. J. Phys. 2006.00669.x) Anthropol. 110, 179–199. (doi:10.1002/(SICI)1096- Watson, J., Payne, R., Chamberlain, A., Jones, R. & Sellers, 8644(199910)110:2,179::AID-AJPA5.3.0.CO;2-Z) W. I. 2009 The kinematics of load carrying in humans

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3314 R. H. Crompton et al. Review. Arboreal origins of hominin bipedalism

and great apes: implications for the evolution of human Williams, S. B., Wilson, A. M., Rhodes, L., andrews, J. & bipedalism. Folia Primatol. 80, 309–328. (doi:10.1159/ Payne, R. C. 2008 Functional anatomy and muscle 000258646) moment arm of the pelvic limb of an elite sprinting ath- Wareing, K., Tickle, P. G., Stokkan, K.-A., Codd, J. R. & lete, the racing greyhound (Canis familiaris). J. Anat. Sellers, W. I. Submitted. The musculoskeletal 213, 361–372. (doi:10.1111/j.1469-7580.2008.00961.x) anatomy of the reindeer (Rangifer tarandus): fore- and Winter, D. A. 1990 Biomechanics and motor control of human hindlimb. movement. New York, NY: John Wiley and Sons. Williams, S. B., Payne, R. C. & Wilson, A. M. 2007 Func- Zihlman, A. L., Cronin, J. E., Cramer, D. L. & Sarich, V. M. tional specialisation of the pelvic limb of the hare (Lepus 1978 Pygmy chimpanzee as a possible prototype for the europeus). J. Anat. 210, 472–490. (doi:10.1111/j.1469- common ancestor of humans, chimpanzees and gorillas. 7580.2007.00704.x) Nature 275, 744–746. (doi:10.1038/275744a0)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Phil. Trans. R. Soc. B (2010) 365, 3315–3321 doi:10.1098/rstb.2010.0069

Two new Mio-Pliocene Chadian hominids enlighten Charles Darwin’s 1871 prediction Michel Brunet1,2,* 1Colle`ge de France, Chaire de Paleóntologie Humaine, F75231 Paris Cedex 05, France 2Institut International de Paleóprimatologie et Paleóntologie Humaine, IPHEP UMR 6046 CNRS/Universite´ de Poitiers, F86022 Poitiers Cedex, France The idea of an evolutionary sequence for humans is quite recent. Over the last 150 years, we have discovered unexpected ancestors, numerous close relatives and our deep evolutionary roots in Africa. In the last decade, three Late Miocene hominids have been described, two about 6 Ma (Ardipithecus and Orrorin) in East Africa and the third dated to about 7 Ma (Sahelanthropus) in Central Africa. The specimens are too few to propose definite relationship to other species, but clearly these belong to a new evolutive grade distinct from Australopithecus and Homo. Moreover, all of them were probably habitual bipeds and lived in woodlands, thus falsifying the savannah hypothesis of human origins. In light of all this recent knowledge, Charles Darwin predicted correctly in 1871 that Africa is the birthplace of humans, chimpanzees and our close relatives. Keywords: earliest hominids; central Africa; evolutionary grade; woodland origin

1. INTRODUCTION temporally close to the last common ancestor of chim- Who our ancestors were, and when and how they panzees and humans, and also that it cannot be related arose, are questions that are always topical. to chimpanzees or gorillas (Brunet et al. 2002a,b, The notion of fossil humans is quite recent: the first 2004, 2005; Vignaud et al. 2002; Guy et al. 2005; Neanderthal specimen from the Neander Valley of Zollikofer et al. 2005; Brunet 2006, 2009a,b; Lebatard Germany was first described only in 1856 (Fuhlrott et al. 2008). In Chad, the Late Miocene sedimentolo- 1859, 1865) and Darwin’s masterpiece ‘On the gical and palaeobiological data are consistent with Origin of Species’ only published in the middle of the mosaic landscapes probably very similar to the present nineteenth century (Darwin 1859). Okavango Delta (Central Kalahari, Bostwana; Brunet By the 1980s early hominids were known just from et al. 2005). As with the other Late Miocene hominids, south and east Africa (figure 1). S. tchadensis represents a new evolutionary grade But from 1994 in the Djurab desert of Northern (Brunet 2009b), surely a habitual biped with its Chad the M.P.F.T.1 unearthed successively a new aus- usual habitat probably a wooded one. tralopithecine dated to 3.5 Ma (Brunet et al. 1995; So, now, it is completely clear that the earliest homi- Lebatard et al. 2008), Australopithecus bahrelghazali nids are not dependent on savannah and were not (Brunet et al. 1996; figure 2), the first ever found living just in south and east Africa. west of the Rift Valley; and a new hominid, Sahelan- Accordingly, this early hominid history enlightens the thropus tchadensis (Brunet et al. 2002a,b), from the Charles Darwin prediction of 1871 (Darwin 1871)and Late Miocene (figures 3–6), dated to 7 Ma (Vignaud must be reconsidered from a completely new viewpoint. et al. 2002; Lebatard et al. 2008). This earliest known hominid is a new milestone suggesting that an exclusively southern or eastern African distribution of the early hominids is unlikely to be correct. 2. A NEW STORY ...WEST OF THE RIFT VALLEY And so in the last 15 years our roots have been In the 1980s, the distribution of hominid remains in shown to be deeper, from the Lower Pliocene Africa, with the earliest being in east Africa (Ethiopia (4.4 Ma) to the Late Miocene (7 Ma), with three new and Tanzania) (figure 1), led Coppens (1983) to pro- species: Ardipithecus kadabba (Haile-Selassie 2001; pose an ‘East Side Story’ scenario in which early Haile-Selassie & Woldegabriel 2009; 5.2–5.8 Ma, hominids appeared and evolved in the Pliocene pri- Middle Awash, Ethiopia), Orrorin tugenensis (Senut mary savannah east of the Rift Valley, while the et al.2001; ca 6 Ma, Lukeino, Kenya) and, the earliest tropical forest, west of the Rift Valley, was thought to one, S. tchadensis (Brunet et al. 2002a,b;7Ma,Chad). represent the early African ape habitat. Sahelanthropus tchadensis (figures 3–6) displays a In 1994, I received a research permit from the unique combination of plesiomorphic and apomorphic Chadian authorities to conduct geological and characters that clearly illustrate its hominid affinities palaeontological survey in the Djurab desert of northern Chad. With the M.P.F.T., one year later (January 1995), at a site east of Koro-Toro, we found a *[email protected], [email protected] Lower Pliocene vertebrate fauna with a partial lower One contribution of 14 to a Discussion Meeting Issue ‘The first four jaw belonging to a new australopithecine that we million years of human evolution’. nicknamed ‘Abel’, the first ever found west of the

3315 This journal is q 2010 The Royal Society Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3316 M. Brunet Chadian hominids and Darwin’s prediction

LIBYA

Toumaï Abel NIGER 2524

CHAD SUDAN Lucy 23

NIGERIA ETHIOPIA 22 21 CENTRAL 20 19 18 CAMEROON AFRICAN REPUBLIC 1617 14 15 13 12 11 KENYA 10 8 9 TANZANIA 7

4 6 SOUTH 5 3 2 AFRICA 1

Figure 1. Map of Africa showing the main Mio-Pliocene hominid localities. 1, Taung; 2, Drimolen; 3, Sterkfontein; 4, Swartk- rans, 5, Kromdraai; 6, Makapansgat; 7, Malema; 8, Laetoli; 9, Olduvai; 10, Peninj; 11, Lukeino; 12, Chesowanja; 13, Lothagam; 14, Kanapoi; 15, Chemeron; 16, W. Turkana; 17, Koobi Fora; 18, Allia Bay; 19, Omo; 20, Konso; 21, Maka; 22, Aramis; 23, Hadar; 24, Koro-Toro; 25, Toros-Menalla.

Rift Valley (Brunet et al. 1995). We named it A. bahrelghazali (Brunet et al. 1996; figure 2). Other australopithecine sites have been discovered in the Koro-Toro area since 1995 (Brunet et al. 1997), all having the same fauna with mammals (proboscidians, suids, rhinocerotids and equids) indicating a biochronological age of 3.5 Ma, congruent with more recent 10beryllium cosmonuclid dating (Lebatard et al. 2008). Geological and palaeontological survey in the Djurab desert from 1994 to 1997 yielded three new fossiliferous areas, biochronologically dated to: (i) the Early Pliocene (4–4.5 Ma), at Kolle´ (Brunet et al. 1998); (ii) the Mio-Pliocene boundary (5–5.5 Ma), at Kossoum Bougoudi (Brunet et al.2000); and (iii) the Figure 2. Mandible holotype specimen (KT12-95-H1) of Late Miocene (ca 7 Ma), at Toros-Menalla (TM) A. bahrelghazali (Brunet et al. 1996). (Brunet et al. 2002a,b). To date, more than 500 fossiliferous localities have been discovered in the Djurab In 2001, the M.P.F.T. unearthed a new hominid, desert, representing upto now around 20 000 vertebrate S. tchadensis (Brunet et al. 2002a,b), from the locality (mammals, reptiles, birds and fish) specimens. TM 266. The holotype cranium (figure 3), nicknamed

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Chadian hominids and Darwin’s prediction M. Brunet 3317

Figure 5. Stereolithocast of the S. tchadensis cranium, three- Figure 3. Cranium holotype specimen (TM 266-01-060-1) dimensional reconstruction. Scale bar, 1 cm. of S. tchadensis (Brunet et al. 2002a,b). Scale bar, 1 cm.

Figure 4. Right lower jaw paratype specimen (TM 266-02- 154-1) of S. tchadensis. Scale bar, 1 cm.

‘Toumaı¨’, is associated with a fauna (more than 70 species) of which the mammalian component indicates a biochronological age estimate congruent with the 10beryllium dating, close to 7 Ma (Vignaud et al. 2002; Lebatard et al. 2008). This earliest known hominid, S. tchadensis, unearthed at least 2600 km west of the Rift valley, is a new milestone suggesting that an exclusively eastern or southern African origin of the hominid clade is unlikely to be correct. The discovery of S. tchadensis occurred in a particu- Figure 6. Illustration of the Toumaı¨head. lar scientific context. With three new Late Miocene species, Ar. kadabba (5.2–5.8 Ma, Ethiopia), Or. show original associations of characters and morpho- tugenensis (ca 6 Ma, Kenya) and S. tchadensis types that lead us to revise our definitions of the (ca 7 Ma, Chad), our roots went deeper, from Hominidae per se. In fact, we have to study our 3.6 Ma in the 1970s to about 7 Ma today. evolutionary history within completely new paradigms. Since 1994, all these discoveries have had a scienti- But, as palaeontologists and palaeoanthropologists, fic impact equivalent to that of Dart’s naming of we have always to remember that our interpretations Australopithecus africanus (Dart 1925). During the have, at most, a life expectancy that usually does not last 10 years the framework of the hominid evolution- go beyond the next new major fossil discovery. ary story has changed markedly. Now, we have a new understanding of the environments inhabited by early hominids, and the models established in the 1980s must be reconsidered. Potential representatives of the 3. THE CHADIAN EARLY HOMINIDS earliest members of the human clade are now twice The material referred to as the Chadian australopithe- as old and are shown to be spread much wider over cine ‘Abel’ is an anterior lower jaw (figure 2), with the 3 the African continent. Moreover, the hominids body missing beyond the P4 level, and to a right P . described during the last 15 years, while extending This pre-human has a new mosaic of derived and the geographical and temporal limits of our family, primitive anatomical characters such as: a flat mental

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3318 M. Brunet Chadian hominids and Darwin’s prediction surface of the lower jaw; a subvertical symphysis, bul- when it is coming from, among others, two who have bous in its outline with a shallow genioglossal fossa; not yet had the opportunity to check Toumaı¨casts in and an incisiform and very asymmetric canine with a their laboratory. Is it what they believe, or is it only strong bifid lingual crest. Each premolar exhibits a because they want to keep Orrorin as the earliest three pulp canal rooting pattern and the P4s are sub- hominid? molariform with a large talonid. This original combi- The East Side scenario of Coppens (1983) empha- nation of anatomical features has been interpreted as sized the major role of savannah in early hominid being sufficiently distinct to name a new species: Aus- evolution. Are the earliest known hominid environments tralopithecus barhelghazali (Brunet et al. 1996). in agreement with such a model? To date, only three Late Miocene species may claim the ‘enviable status’ of being among the earliest hominids. Two of them are from eastern Africa: Ar. 4. THE LATE MIOCENE HOMINID kadabba (5.8–5.2 Ma, Ethiopia; Haile-Selassie ENVIRONMENTS AS WE KNOW THEM 2001; Haile-Selassie et al. 2004; Haile-Selassie & In Chad, the sedimentological evidence from the Late Woldegabriel 2009)andOr. tugenensis (ca 6Ma, Miocene TM fossiliferous area demonstrates that, at Kenya; Senut et al. 2001; Galik et al. 2004; Ohman least since 7 Ma, successive wet (mega-lake Chad et al. 2005; White 2006); the third one, the oldest events) and arid periods (desertic events) occurred. known hominid, is from central Africa: S. tchadensis These successive events are identified by a sedimento- (7 Ma, Chad; Brunet et al. 2002a,b, 2005). The geological series comprising aeolian sandstones (desertic graphical location of S. tchadensis, 2600 km west of episode); perilacustrine sandstones (lacustrine trans- the Rift Valley, along with its great antiquity, suggests gression); and green pelite and diatomite (true an early (at least by 6 Ma) widespread hominid distri- lacustrine environment; Vignaud et al. 2002; Schuster bution (Sahel and eastern Africa). The material et al. 2006). Sahelanthropus tchadensis and its associ- referred to as the Chadian hominid ‘Toumaı¨’ consists ated vertebrate remains have been uncovered from of a distorted but nearly complete cranium the perilacustrine sandstones (Anthracotheriid Unit). (figure 3), as reconstructed in three-dimensions Sedimentological data are in agreement with this (figures 5 and 6), associated with several mandibular mosaic of environments, indicating a vegetated perila- specimens (figure 4) and isolated teeth. Sahelanthropus custrine belt between lake and desert (Vignaud et al. tchadensis displays a unique combination of primitive 2002). The Okavango Delta in central Kalahari (Bots- and derived characters. Identifiable derived features wana) appears to be a good modern analogue in of S. tchadensis are: an anteriorly-positioned foramen presenting similar habitat diversity (mosaic of lacus- magnum linked to a rather short basioccipital and a trine and riparian waters, swamps, patches of forest, sub-horizontal nuchal plane; a downward lipping of wooded savannah, grassland and desertic areas). the nuchal crest; lower jaw with a vertical symphysis Although the precise habitat of the TM 266 hominid with weak transverse tori; and for the dentition, the among this mosaic of landscapes available is still anatomical characters are: a non-honing C/P3 complex; uncertain, it was probably a wooded one (Brunet no diastema between C and P3; small-crowned canines et al. 2005). Other significant hominid discoveries with a long root, the upper one without any honing associated with wooded environments in Middle distal crest and the lower one with a large distal tubercle, Awash, Ethiopia (Ardipithecus ramidus and Ar. both shoulders being very low; a P3 with a strongly slop- kadabba) and Lukeino, Kenya (Or. tugenensis) also ing buccal surface; postcanine teeth with maximum rule out the role played by an open environment radial enamel thickness intermediate between chimpan- or savannah in favouring the acquisition of bipedal zees and australopithecines; and bulbous, slightly posture in the course of hominid evolution. Thus, crenulated postcanine occlusal morphology. It is inter- sedimentological and faunal data rather suggest esting to note that all the hominid mandibular wooded environments for Ar. ramidus (White et al. specimens from TM have the same root pattern in the 1994; Woldegabriel et al. 1994) and Ar. kadabba postcanine teeth, with two roots and three separate (Woldegabriel et al. 2001; Haile-Selassie & pulp canals for each premolar (Brunet et al. 2002a,b, Woldegabriel 2009). In 2001 those authors noted: ‘It 2004, 2005; Vignaud et al. 2002; Guy et al. 2005; therefore seems increasingly likely that early hominids Zollikofer et al. 2005; Brunet 2006, 2009a,b). did not frequent open habitats until after 4.4 Myr. As known since Darwin, all these derived characters Before that, they may have been confined to woodland show that S. tchadensis cannot be related to an ape and forest habitats’ (p. 177). Moreover, according to (chimpanzees or gorillas) but clearly suggest that it the recent description of a partial skeleton of Ar. rami- must be related to later bipedal hominids, and may dus (White et al. 1994, 2009a,b) from Aramis, Ethiopia be temporally close to the common ancestor of chim- at 4.4 Ma, it appears that this hominid may be the same panzees and humans (Brunet et al. 2002a,b, 2005; new evolutionary grade as are the three late Miocene Guy et al. 2005; Zollikofer et al. 2005). taxa. It was a climbing biped, both terrestrial and Scientifically it is impossible to understand why arboreal, with an opposable grasping big toe (without some authors ignore these derived characters and con- arched feet and walking flat-footed), living in a centrate on primitive ones to reach the conclusion that woodland landscape (Lovejoy 2009;Lovejoyet al. S. tchadensis is related to modern apes and even more 2009a,b,c,d;Suwaet al. 2009a,b;Whiteet al. 2009a,b). precisely to a palaeogorilla (Wolpoff et al. 2002, 2006; A wooded habitat (open woodland with denser Pickford 2005). This attempt to undermine the clear stands of trees in the vicinity) has been also suggested affinity of the Chadian hominid is curious mainly for the Kenyan Or. tugenensis (Pickford et al. 2002).

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Chadian hominids and Darwin’s prediction M. Brunet 3319

This is of fundamental importance since the three late supe´rieur et de la Recherche: Universite´ de Poitiers; Miocene taxa, Ar. kadabba, Or. tugenensis and Agence Nationale de la Recherche—Projet ANR S. tchadensis, were all probably habitual bipeds. All 05-BLAN-0235; Center National de la Recherche ´ these Late Miocene hominids demonstrate that the Scientifique (CNRS: Departement INEE, and ECLIPSE); the Ministe`re des Affaires Etrange`res (DCSUR, Paris and savannah hypothesis is now falsified. Projet FSP 2005-54 de la Coope´ration Franco- Tchadienne, Ambassade de France a` N’Djamena); the Re´gion Poitou-Charentes; the NSF programme RHOI 5. NEW PARADIGMS ...FOR A NEW EARLY (USA) and the Armeé Française (Mission d’Assistance HOMINID STORY Militaire and dispositif Epervier). We thank all the During the last 150 years, most of the models for members of the Mission Paleóanthropologique Franco- Tchadienne, including all our friends and colleagues who hominid evolution have been overturned by successive participated in acquisition of the field data; G. Florent and discoveries. This fact obviously highlights the impor- C. Noe¨l for their administrative guidance; and X. Valentin tance of fieldwork, as our understanding of our for his technical support. evolutionary story has, at most, a life expectancy that A lot of thanks go to our colleagues A. Walker and E.-G. usually does not go beyond new findings. Emonet for the review and editing of our manuscript. All Twenty years ago, available fossil hominid remains the drawings are due to the talent of S. Riffaut. led us to consider eastern African savannah as the cradle of mankind. Now, it appears that the earliest members of our family have favoured wooded environ- ENDNOTE ments, and were not restricted to eastern or southern 1 M.P.F.T.: Mission Paleóanthropologique Franco-Tchadienne, an Africa but were rather living in a wider geographical international scientific collaboration between Colle`ge de France, region, including at least central and eastern Africa. University of Poitiers, University of N’Djamena and CNAR (N’Dja- In the last decade, the number of recognized mena) including now more than 60 researchers from 10 countries. hominid taxa and the length of our geological roots, from 3.6 Ma in the 1970s to 7 Ma today, have doubled. These new hominids, while extending the REFERENCES geographical and temporal limits of our family, are Boisserie, J.-R., Brunet, M., Likius, A. & Vignaud, P. 2003 Late Miocene ones with new associations of anatom- Hippopotamids from the Djurab Pliocene faunas, Chad, ical characters representing a new evolutionary grade. Central Africa. J. Afr. Earth Sci. 36, 15–27. (doi:10. In Chad, the radiometrical dating of S. tchadensis 1016/S0899-5362(03)00021-6) indicates that the divergence between chimpanzee Boisserie, J.-R., Likius, A., Vignaud, P. & Brunet, M. 2005 A and human lineages occurred before 7 Ma, which new late Miocene hippopotamid from Toros-Menalla, is earlier than generally (Kumar & Hedges 1998; Chad. J. Vert. Paleontol. 25, 665–673. (doi:10.1671/ Pilbeam 2002) or more recently thought by the 0272-4634(2005)025[0665:ANLMHF]2.0.CO;2) molecular phylogenists (Patterson et al. 2006). Brunet, M. 2006 D’Abel a` Toumaı¨, Nomade Chercheur d’Os. Besides, from a palaeobiogeographical point of Paris, France: Editions Odile Jacob. view, published results derived from fieldwork show Brunet, M. 2009a Origine et Histoire des Hominide´s ...Nou- veaux paradigms. Paris, France: Leçon inaugurale du that central Africa was, at least between 3 and 7 Ma, Colle`ge de France Editions Fayard. a crossroad region marked by sporadic faunal Brunet, M. 2009b Origine et e´volution des hominide´s: exchanges with northern and eastern Africa (Brunet Toumaı¨, une confirmation ećlatante de la pre´diction de et al. 1995, 1996, 1997, 1998, 2000, 2002a,b, 2004; Darwin. C.R. Palevol 8, 311–319. (doi:10.1016/j.crpv. Brunet & White 2001; Geraads et al. 2001; Vignaud 2008.09.007) et al. 2002; Boisserie et al. 2003, 2005; Likius et al. Brunet, M. & White, T. D. 2001 Deux nouvelles espe`ces de 2003; Mackaye et al. 2005; Peigne´ et al. 2005; Suini (Mammalia, Suidae) du continent African (Ethio- Lihoreau et al. 2006). pie; Tchad). C. R. Acad. Sci. Paris 332, 51–57. Better identification of the migratory patterns and Brunet, M., Beauvillain, A., Coppens, Y., Heintz, E., the faunal exchanges between northern, central, east- Moutaye, A. H. E. & Pilbeam, D. 1995 The first australopithecine 2500 kilometres west of the Rift Valley ern and southern Africa during the Mio-Pliocene is a (Chad). Nature 378, 273–274. (doi:10.1038/378273a0) key element that is indispensable for a new under- Brunet, M., Beauvillain, A., Coppens, Y., Heintz, E., standing of the origin and dispersal of the first Moutaye, A. H. E. & Pilbeam, D. 1996 Australopithecus members of the human clade and therefore its evol- bahrelghazali, une nouvelle espe`ce d’hominide´ ancien de utionary history. We need more data from these la re´gion de Koro Toro (Tchad). C. R. Acad. Sci. Paris geographical areas and notably from north-eastern 322, 907–913. Africa (e.g. Libya, Egypt and Sudan). Brunet, M. et al. 1997 Tchad: un nouveau site a` Hominide´s Finally, for a better understanding of the early Plioce`ne. C. R. Acad. Sci. Paris 324, 341–345. human story palaeontologists and palaeoanthropolo- Brunet, M. et al. 1998 Tchad: dećouverte d’une faune de gists need more data, and thus will have to conduct mammife`res du Plioce`ne infe´rieur. C. R. Acad. Sci. more and more geological and palaeontological field Paris 326, 153–158. Brunet, M. et al. 2000 Chad: discovery of a vertebrate surveys. fauna close to the Mio-Pliocene boundary. J. Vert. We especially thank the Chadian authorities (Ministe`re de Paleontol. 20, 205–209. (doi:10.1671/0272-4634(2000) l’Education Nationale de l’Enseignement Supe´rieur et 020[0205:CDOAVF]2.0.CO;2) de la Recherche, Universite´ de N’Djamena/Departement Brunet, M. et al. 2002a A new hominid from the Upper de Paleontologie, Center National d’Appui a` la Recherche- Miocene of Chad, Central Africa. Nature 418, 145–151. CNAR); the Ministere Français de l’Enseignement (doi:10.1038/nature00879)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3320 M. Brunet Chadian hominids and Darwin’s prediction

Brunet, M. et al. 2002b Reply to Wolpoff M. H., Senut B., Miocene Chado-Libyan bioprovince. Proc. Natl Acad. Pickford M., Hawks J. 2002 Sahelanthropus or ‘Sahel- Sci. USA 103, 8763–8767. (doi:10.1073/pnas. pithecus’? Nature 419, 582. (doi:10.1038/419582a) 0603126103) Brunet, M. et al. 2004 ‘Toumaı¨’, Mioce`ne supe´rieur du Likius, A., Brunet, M., Geraads, D. & Vignaud, P. 2003 Tchad, le nouveau doyen du rameau humain. Le plus vieux Camelidae (Mammalia, Artiodactyla) C. R. Palevol Paris 3, 275–283. d’Afrique: limite Mio-Plioce`ne, Tchad. Bull. Soc. Brunet,M.,Guy,F.,Pilbeam,D.,Lieberman,D.E.,Likius,A., Geólologique France 174, 187–193. (doi:10.2113/ Mackaye, H. T., Ponce de Leon, M., Zollikofer, P. E. & 174.2.187) Vignaud, P. 2005 New material of the earliest hominid Lovejoy, C. O. 2009 Reexamining human origins in light of from the Upper Miocene of Chad. Nature 434, 752–755. Ardipithecus ramidus. Science 326, 74e1–74e8. (doi:10. (doi:10.1038/nature03392) 1126/science.1175834) Coppens, Y. 1983 Les plus anciens fossiles d’Hominide´s. Lovejoy, C. O., Simpson, S. W., White, T. D., Asfaw, B. & In Recent advances in the evolution of Primates,pp.1–9. Suwa, G. 2009a Careful climbing in the Miocene: the Vatican City: Pontificiae Academiae Scientiarum Scripta forelimbs of Ardipithecus ramidus and humans are primi- Varia. tive. Science 326, 70e1–70e8. (doi:10.1126/science. Dart, R. 1925 Australopithecus africanus, the man-ape of 1175827) South Africa. Nature 115, 195–199. (doi:10.1038/ Lovejoy, C. O., Suwa, G., Spurlock, L., Asfaw, B. & White, 115195a0) T. D. 2009b The pelvis and femur of Ardipithecus Darwin, C. h. 1859 On the origin of species by means of natural ramidus: the emergence of upright walking. Science 326, selection. London, UK: John Murray (Reprinted, 71e1–71e6. (doi:10.1126/science.1175831) Everyman edition 1928. New York: Dutton). Lovejoy, C. O., Latimer, B., Suwa, G., Asfaw, B. & White, Darwin, C. h. 1871 The descent of man and selection in T. D. 2009c Combining prehension and propulsion: the relation to sex. Princeton, NJ: Princeton University Press foot of Ardipithecus ramidus. Science 326, 72e1–72e8. (Reprinted 1981). (doi:10.1126/science.1175832) Fuhlrott, C. J. 1859 Menschliche Ueberreste aus einer Fel- Lovejoy,C.O.,Suwa,G.,Simpson,S.W.,Matternes,J.H.& sengrotte des Du¨sselthals. Ein Beitrag zur Frage u¨ber White,T.D.2009d The great divides: Ardipithecus ramidus die Existenz fossiler Menschen. Verhanlungen des naturhis- reveals the postcrania of our last common ancestors with torischen Vereins der preussischen Rheinlande und Westphalens African apes. Science 326,100–106.(doi:10.1126/science. 16, 131–153. 1175833) Fuhlrott, C. J. 1865 Der fossile Mensch aus dem Neanderthal Mackaye, H. T., Brunet, M. & Tassy, P. 2005 Selenetherium und sein Verha¨ltniß zum Alter des Menschengeschlechts. kolleensis nov. gen. nov. sp. un nouveau Proboscidea Duisburg, Germany: Kessinger Publishing (78 S 2Abb. (Mammalia) dans le Plioce`ne tchadien. Geóbios Lyon 38, Published 2010.) 765–777. (doi:10.1016/j.geobios.2003.09.010) Galik, K., Senut, B., Pickford, M., Gommery, D., Treil, J., Ohman, J., Lovejoy, C. O. & White, T. D. 2005 Questions Kuperavage, A. J. & Eckhardt, R. B. 2004 External and about Orrorin femur. Science 307, 845. (doi:10.1126/ internal morphology of the BAR 1002–00 Orrorin tugen- science.307.5711.845b) ensis femur. Science 305, 1450–1453. (doi:10.1126/ Patterson, N., Richter, D. J., Gnerre, S., Lander1, E. S. & science.1098807) Reich, D. 2006 Genetic evidence for complex speciation Geraads, D., Brunet, M., Mackaye, H. T. & Vignaud, P. of humans and chimpanzees. Nature 441, 1103–1108. 2001 Fossil Bovidae (Mammalia) from the Koro (doi:10.1038/nature04789) Toro Australopithecine sites, Chad. J. Vert. Paleontol. Peigne´, S., de Bonis, L., Likius, A., Mackaye, H. T., 21, 335–346. (doi:10.1671/0272-4634(2001)021[0335: Vignaud, P. & Brunet, M. 2005 New machairodontine PBMFTK]2.0.CO;2) (Carnivora, Felidae) from the Late Miocene hominid Guy, F., Lieberman, D. E., Pilbeam, D., Ponce de Leon, M., locality of Toros-Menalla, Chad. C. R. Palevol Paris 4, Likius, A., Mackaye, H. T., Vignaud, P., Zollikofer, C. & 243–253. (doi:10.1016/j.crpv.2004.10.002) Brunet, M. 2005 Morphological affinities of the Sahelan- Pickford, M. 2005 Orientation of the foramen magnum in thropus tchadensis (Late Miocene Hominid, Chad) Late Miocene to extant African apes and hominids. cranium. Proc. Natl Acad. Sci. USA 102, 18 836–18 Anthropol. Brno XLIII, 191–198. 841. (doi:10.1073/pnas.0509564102) Pickford, M., Senut, B., Gommery, D. & Treil, J. 2002 Haile-Selassie, Y. 2001 Late Miocene hominids from the Bipedalism in Orrorin tugenensis revealed by its femora. Middle Awash, Ethiopia. Nature 412, 178–181. (doi:10. C. R. Palevol Paris 1, 191–203. (doi:10.1016/S1631- 1038/35084063) 0683(02)00028-3) Haile-Selassie, Y. & Woldegabriel, G. 2009 Ardipithecus Pilbeam, D. 2002 Perspectives on the Miocene Hominoidea. kadabba: late Miocene evidence from the Middle Awash. In The primate fossil record (ed. W. Hartwig), pp. 303–310. Ethiopia: University of California Press. Cambridge, UK: Cambridge University Press. Haile-Selassie, Y., Suwa, G. & White, T. 2004 Late Miocene Schuster, M., Duringer, P., Ghienne, J. F., Vignaud, P., hominids from the Middle Awash, Ethiopia, and early Mackaye, H. T., Likius, A. & Brunet, M. 2006 The age hominid dental evolution. Science 303, 1503–1505. of the Sahara desert. Science 311, 821. (doi:10.1126/ (doi:10.1126/science.1092978) science.1120161) Kumar, S. & Hedges, S. B. 1998 A molecular timescale for Senut, B., Pickford, M., Gommery, D., Mein, P. & Cheboi, vertebrate evolution. Nature 392, 917–919. (doi:10.1038/ K. 2001 First hominid from the Miocene (Lukeino for- 31927) mation, Kenya). C. R. Acad. Sci. Paris 332, 137–144. Lebatard, A. E. et al. 2008 Cosmogenic nuclide dating of Suwa,G.,Asfaw,B.,Kono,R.T.,Kubo,D.,Lovejoy,C.O.& Sahelanthropus tchadensis and Australopithecus bahrelgha- White, T. D. 2009a The Ardipithecus ramidus skull zali Mio-Pliocene early hominids from Chad. Proc. Natl and its implications for hominid origins. Science 326, Acad. Sci. USA 105, 3226–3231. (doi:10.1073/pnas. 68e1–68e7. (doi:10.1126/science.1175825) 0708015105) Suwa, G., Kono, R. T., Simpson, S. W., Asfaw, B., Lovejoy, Lihoreau, F., Boisserie, J. R., Viriot, L., Coppens, Y., Likius, C. O. & White, T. D. 2009b Paleobiological implications A., Mackaye, H. T., Tafforeau, P., Vignaud, P. & Brunet, of the Ardipithecus ramidus dentition. Science 326, 94–99. M. 2006 Anthracothere dental anatomy reveals a late (doi:10.1126/science.1175824)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Chadian hominids and Darwin’s prediction M. Brunet 3321

Vignaud, P. et al. 2002 Geology and palaeontology of the Ecological and temporal placement of early Pliocene Upper Miocene Toros-Menalla hominid locality, Chad. hominids at Aramis, Ethiopia. Nature 371, 330–333. Nature 418, 152–155. (doi:10.1038/nature00880) (doi:10.1038/371330a0) White, T. D. 2006 Early hominid femora: the inside story. Woldegabriel, G., Haile-Selassie, Y., Rennek, P. R., Hart, C. R. Palevol Paris 5, 99–108. (doi:10.1016/j.crpv.2005. W. K., Ambrose, S. H., Asfaw, B., Heiken, G. & 09.006) White, T. 2001 Geology and palaeontology of the Late White, T. D., Suwa, G. & Asfaw, B. 1994 Australopithecus Miocene Middle Awash valley, Afar rift, Ethiopia. ramidus, a new species of Early hominid from Nature 412, 175–178. (doi:10.1038/35084058) Aramis, Ethiopia. Nature 371, 306–312. (doi:10.1038/ Wolpoff, M. H., Senut, B., Pickford, M. & Hawks, J. 2002 371306a0) Sahelanthropus or ‘Sahelpithecus’? Nature 419, 581–582. White, T. D., Asfaw, B., Beyene, Y., Haile-Selassie, Y., (doi:10.1038/419581a) Lovejoy, C. O., Suwa, G. & WoldeGabriel, G. 2009a Wolpoff, M. H., Hawks, J., Senut, B., Pickford, M. & Ardipithecus ramidus and the paleobiology of early homi- Ahern, J. 2006 An ape or the ape: is the Toumaı¨cranium nids. Science 326, 65–86. (doi:10.1126/science.1175802) TM 266 a hominid? PaleoAnthropology 2006, 36–50. White, T. D. et al. 2009b Macrovertebrate paleontology and Zollikofer, C., Ponce De Leon, M., Lieberman, D. E., Guy, F., the Pliocene habitat of Ardipithecus ramidus. Science 326, Pilbeam, D., Likius, A., Mackaye, H. T., Vignaud, P. & 87–93. (doi:10.1126/science.1175822) Brunet, M. 2005 Virtual cranial reconstruction of Woldegabriel, G., White, T. D., Suwa, G., Renne, P., Sahelanthropus tchadensis. Nature 434, 755–759. (doi:10. Heinzelin, J., de Hurt, W. K. & Heiken, G. 1994 1038/nature03397)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Phil. Trans. R. Soc. B (2010) 365, 3323–3331 doi:10.1098/rstb.2010.0064

Phylogeny of early Australopithecus:new fossil evidence from the Woranso-Mille (central Afar, Ethiopia) Yohannes Haile-Selassie* Department of Physical Anthropology, The Cleveland Museum of Natural History, 1 Wade Oval Drive, Cleveland, OH 44106, USA The earliest evidence of Australopithecus goes back to ca 4.2 Ma with the first recorded appearance of Australopithecus ‘anamensis’ at Kanapoi, Kenya. Australopithecus afarensis is well documented between 3.6 and 3.0 Ma mainly from deposits at Laetoli (Tanzania) and Hadar (Ethiopia). The phylogenetic relationship of these two ‘species’ is hypothesized as ancestor–descendant. However, the lack of fossil evidence from the time between 3.6 and 3.9 Ma has been one of its weakest points. Recent fieldwork in the Woranso-Mille study area in the Afar region of Ethiopia has yielded fossil hominids dated between 3.6 and 3.8 Ma. These new fossils play a significant role in testing the proposed relationship between Au. anamensis and Au. afarensis. The Woranso-Mille hominids (3.6–3.8 Ma) show a mosaic of primitive, predominantly Au. anamensis-like, and some derived (Au. afarensis-like) dentognathic features. Furthermore, they show that, as currently known, there are no discrete and functionally significant anatomical differences between Au. anamensis and Au. afarensis. Based on the currently available evidence, it appears that there is no compelling evidence to falsify the hypothesis of ‘chronospecies pair’ or ancestor–descendant relationship between Au. anamensis and Au. afarensis. Most importantly, however, the temporally and morphologically intermediate Woranso-Mille hominids indicate that the species names Au. afarensis and Au. anamensis do not refer to two real species, but rather to earlier and later representatives of a single phyletically evolving lineage. However, if retaining these two names is necessary for communication purposes, the Woranso-Mille hominids are best referred to as Au. anamensis based on new dentognathic evidence. Keywords: Australopithecus afarensis; Australopithecus ‘anamensis’; phylogeny; Woranso-Mille; Ethiopia

1. INTRODUCTION agree that one of its species gave rise to the genus The genus Australopithecus was named in the first quar- Homo,possiblyAustralopithecus garhi from the Middle ter of the twentieth century (Dart 1925) and includes Awash region of Ethiopia (Asfaw et al. 1999;butsee at least seven species from South Africa, Tanzania, Strait & Grine 2004). Kenya, Ethiopia, and Chad (Dart 1925; Broom It was only 30 years ago that Australopithecus afaren- 1938; Leakey 1959; Arambourg & Coppens 1968; sis was recognized as the ‘oldest indisputable evidence Johanson et al. 1978; Brunet et al. 1995; Asfaw et al. of the family Hominidae’ at 3.6 Ma (Johanson et al. 1999). Some workers assign three of these species 1978; see Kimbel & Delezene 2009 for detailed (Australopithecus boisei, Australopithecus aethiopicus and review). However, at the end of the twentieth and Australopithecus robustus) to a different genus, Par- beginning of the twenty-first centuries, a number of anthropus (Broom 1938), based largely on new hominid taxa were recovered, some of which are morphological specializations related to trophic par- twice as old (the family Hominidae is here defined ameters (Clarke 1977; Grine 1986, 1988; Wood & following Haile-Selassie 2001 and Haile-Selassie et al. Ellis 1986; Wood & Chamberlain 1987; Turner & 2004). During the 1990s, the discovery of Ardipithecus Wood 1993). Some palaeontologists have even moved ramidus (1994, Ethiopia, 4.4 Ma; White et al. 1994, thetypespeciesofthegenus,Australopithecus 1995; Semaw et al. 2005) was followed by Australo- africanus into Paranthropus (e.g. Cela-Conde & Altaba pithecus ‘anamensis’ (1995, Kenya, 3.9–4.2 Ma; 2002;seealsoCela-Conde & Ayala 2007), leaving Leakey et al. 1995, 1998; Ward et al. 1999, 2001). Australopithecus with only three species. Whether More recent fieldwork in Ethiopia, Kenya and Chad Australopithecus is a paraphyletic (Strait et al. 1997)or have pushed the record further into the Late Miocene monophyletic (Tobias 1967) genus, most researchers with the discovery of Ardipithecus kadabba (Haile- Selassie 2001; Haile-Selassie et al. 2004, 2009), Orrorin tugenensis (Senut et al.2001; Pickford et al. * [email protected] 2002)andSahelanthropus tchadensis (Chad, 6–7 Ma; One contribution of 14 to a Discussion Meeting Issue ‘The first four Brunet et al. 2002, 2005). The phylogenetic relation- million years of human evolution’. ships among these earlier hominids remain a point of

3323 This journal is q 2010 The Royal Society Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3324 Y. Haile-Selassie Hominids from Woranso-Mille, Ethiopia contention (see Haile-Selassie et al. 2004, 2009,for Further analysis of Au. afarensis mandibles and details), and Pliocene hominid fossils are poorly sampled teeth also indicated that there was a directional from the 3.6–3.9 and 4.4–5.2 Ma time intervals. trend toward an increase in mandibular size through The hominid-bearing Woranso-Mille palaeontolo- time, but not in the size of the teeth, contributing to gical study area (WORMIL) has been explored since the larger range of variation seen in the species its discovery in 2004 (Haile-Selassie et al. 2007). (Lockwood et al. 2000; see also Kimbel et al. The fossiliferous deposits in the north and northwes- 2004). The postcranial anatomy of Au. afarensis is tern parts of the study area sample the period inferred to indicate obligate bipedality (for example, between 3.4 and 3.8 Ma (Deino et al. 2010), a time Lovejoy 1975, 1978;White1980a,b;Latimer1983, frame poorly known in the hominid fossil record 1991; Latimer & Lovejoy 1989; Kramer 1999), (Kimbel et al. 2006). These ages were determined although some argued that it was partly arboreal using 40Ar/39Ar radiometric dating method (Deino (for example, Stern & Susman 1981, 1983, 1991; et al. 2010). The study area has thus far yielded Jungers & Stern 1983; Stern 2000). The Laetoli foot- about 90 fossil hominid specimens, mostly from the prints, however, yield incontrovertible evidence that time period between 3.6 and 3.8 Ma. Additional Au. afarensis was fully bipedal (for example, hominid specimens were collected from slightly White & Suwa 1987). younger (3.4–3.6 Ma) deposits. Most of these specimens represent isolated teeth and partial jaws, although they also include a 3.58 Ma partial skeleton 3. THE WORANSO-MILLE FOSSIL HOMINIDS of Au. afarensis (Haile-Selassie et al. 2010a) and A total of 55 hominid dentognathic (mostly isolated additional fragmentary postcranial remains. Moreover, teeth) and fragmentary postcranial elements have a total of 4300 fossil specimens of diverse vertebrate been recovered from the Am-Ado (AMA), Aralee taxa have been collected thus far, representing more Issie (ARI), Mesgid Dora (MSD) and Makah Mera than 25 mammalian genera and a number of new (MKM) collection areas of the WORMIL study species. area (figure 1). Their age has been chronometrically Here, a brief summary of early Australopithecus to constrained to between 3.57 and 3.82 Ma (Deino which the WORMIL hominids belong is presented, et al. 2009). The associated faunal assemblage is followed by a brief description of the new hominids dominated largely by cercopithecids, tragelaphines and their phylogenetic relationships. Finally, a discus- and aepycerotines, among others, indicating a more sion is presented on the evolutionary tempo and mode closed habitat with riverine gallery forest and of early Australopithecus and the proposed ancestor– abundant water. descendant relationship between Au. anamensis and Au. afarensis in light of the new fossil evidence from (a) Mandibles Woranso-Mille. Two mandibular fragments were recovered from the MSD collection area. MSD-VP-5/16 is a well-preserved left mandible with M1–2, anteriorly broken at 2. EARLY AUSTRALOPITHECUS the I2 level. It was found during the 2006 field The origin of the genus Australopithecus remained elu- season from Mesgid Dora locality 5. Posteriorly, the sive until new discoveries from the Early Pliocene in preserved corpus extends as far as slightly posterior eastern Africa began to shed some light (White et al. to the M3 level. The preserved corpus base is intact. 2006). Based on the currently available fossil evidence, The entire ascending ramus is missing. MSD-VP-5/ Au. anamensis is the earliest species of the genus. It is 50 is a left mandible with P3–M3 found during the documented from deposits in Kenya and Ethiopia, 2009 field season from Mesgid Dora locality 5. This dated between 4.2 and 3.9 Ma (Leakey et al. 1995, specimen is anteriorly broken lateral to the midline. 1998; Ward et al. 2001; White et al. 2006). Posteriorly, part of the ascending ramus is preserved The integrity and amount of variation in Au. afarensis although some parts of its base below the ascending have been rigorously debated since its naming in the ramus are missing (figure 2). 1970s (Johanson et al. 1978, 1982; Johanson & White MSD-VP-5/16 probably belongs to a female indi- 1979; White et al.1981,1993,2000;Kimbelet al. vidual largely owing to its small corpus size that is 1985, 2004; White 1985; Kimbel & White 1988; comparable to A.L. 128-23 (White & Johanson Lockwood et al. 2000; Reno et al. 2003, 2005; Plavkan 1982). However, the molars (M1–2) in MSD-VP-5/ et al. 2005, among others). Some argued that the 16 are larger relative to the corpus dimensions. species hypodigm pooled from two asynchronous and Australopithecus afarensis-like mandibular features of geographically disparate areas, Laetoli (Tanzania; MSD-VP-5/16 include corpus robusticity (corpus Leakey 1987; Leakey et al. 1987) and Hadar (Ethiopia), breadth at mid-M1/corpus height at mid-M1 Â 100) represents more than one species (Leakey & Walker of 62.5, the presence of a lateral corpus hollow, and 1980;Olson1981, 1985; Senut & Tardieu 1985; a more vertical mandibular symphysis, as judged Zihlman 1985). However, the discovery of more speci- from the preserved part of the anterior corpus mens at Hadar since the early 1990s and detailed (Haile-Selassie et al. 2010b). The mandible’s greatest analyses of the pooled hypodigm have demonstrated anterior breadth, however, is at the canine level, that the amount of variation in Au. afarensis does not more like Au. anamensis (Leakey et al. 1995; Ward significantly exceed what is observed in a single species et al. 1999; see Haile-Selassie et al. 2010b for details). of extant taxa (Kimbel et al. 1985, 2004; Kimbel & The anterior corpus of MSD-VP-5/50 is morpho- White 1988; Lockwood et al. 2000; White et al.2000). logically more similar to Au. anamensis mandibles

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Hominids from Woranso-Mille, Ethiopia Y. Haile-Selassie 3325

(a) 40°28¢0≤E 40°30¢0≤E

AMA

¢ ≤ 11°32¢0≤N 11°32 0 N

ARI

MSD MKM Millle river

11°30¢0≤N 11°30¢0≤N

KSD

0.5 1.0 2 km

40°28¢0≤E 40°30¢0≤E 40°32¢0≤E

(b) (i) Aralee Issie (ARI)(ii) Waki-Mille confluence (WMC) (iii) Korsi Dora (KSD) fossils rhyzoliths Mesgid Dora (MSD) reworked tuff planar lamination in tuff Makah Mera (MKM) tuff trough-cross stratification in tuff KT basalt carbonate 30 ARI-08-5 nodules 10 KT conglomerate horizontal stratification MLT? P5 WM-KSD-1&3 (mean 3.570 ± 0.014) sandstone & desiccation cracks 30 ripple cross-stratification KT pebbles, with WM-ut-1 T is tuffaceous tuffaceous 20 ± 0.03 (3.72 ) MKM-08-13 siltstone & fossiliferous 0 m sandstone reverse magnetic MDT MDT P6 P4 ?P7 siltstone & polarity MKM 20 MDT P2 ? WT ± 0.04 P8 claystone normal magnetic WM-MD-5 (3.77 ) P1 0 m skeleton brown silty polarity 10 BRT (WM07/W-1) WT claystone indeterminate magnetic MLT sandstone polarity 10 BRT? MSD WM-W-2 (3.76 ± 0.02) bedded siltstone MLT & sandstone 0 m fault laminated mudstone & sandstone 0 m WM-W-1 (3.82 ± 0.18) bioturbated WMC siltstone

Figure 1. (a) Satellite imagery showing the location of hominid collection areas in the Woranso-Mille palaeontological site. (b) Stratigraphic sections showing the provenance and age of fossiliferous horizons.

than to those of Au. afarensis (figure 3). The corpus of KNM-KP 29281, KNM-KP 29287, KNM-KP MSD-VP-5/50 is very deep and transversely narrow at 31713, and KNM-ER 20432 from Kanapoi and the M1 and other molar levels. Its breadth at M1 Allia Bay (Ward et al. 2001; Kimbel et al. 2006). (20.7 mm) is within the range seen in Au. afarensis Australopithecus afarensis mandibles are different in (Kimbel et al. 2004) and Au. anamensis (Ward et al. having an almost vertical contour descending as far 2001), whereas its height at the same position as the corpus base. However, LH 4 is an exception (44.6 mm) and its robusticity index (46.6) are slightly to this general characteristic of most Au. afarensis outside the range documented for both groups. mandibles, showing a slight inferomedial sweep at Australopithecus anamensis mandibles have a mean the C–P3 level (Kimbel et al. 2006). The ascending robusticity index (53.6, n ¼ 3; Ward et al. 2001) ramus root of MSD-VP-5/50 is positioned more pos- slightly lower than that of Au. afarensis (57.5, n ¼ 19; teriorly (mid-M2) with most of the M3 buccal face Kimbel et al. 2004). A.L. 277-1 approaches MSD- visible laterally. However, the canine does not appear VP-5/50 in terms of its robusticity index (48.4; to be aligned in the longitudinal axis of the postcanine Kimbel et al. 2004) although the latter is absolutely tooth row, although it is seen in the other, much smal- deeper and wider. ler mandible (MSD-VP-5/16). In Au. afarensis The inferomedial sweep of the corpus contour at mandibles, the ascending ramus root is usually at the C–P4 level seen in MSD-VP-5/50 is comparable M1. The overall morphology of MSD-VP-5/50, par- to Au. anamensis mandibles (figure 3). This lateral ticularly the anterior lateral corpus profile, is corpus profile has been described as uniquely intermediate between Kanapoi mandibles and LH 4 characteristic of Au. anamensis mandibles such as and serves, together with KNM-ER 20432, as a

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3326 Y. Haile-Selassie Hominids from Woranso-Mille, Ethiopia

(a)(b)

Figure 2. Occlusal, medial and lateral views of mandibular specimens from Woranso-Mille dated at 3.7–3.8 Ma. (a) MSD-VP- 5/16, left mandible with M1–2.(b) MSD-VP-5/50, left mandible with P3 –M3. Note the size variation between the two mandibles and the deep corpus in (b). Scale bar, 5 cm.

(a) (b)(c)

lateral medial

(d)(e)(f)

sup. tr. torus

Figure 3. Cross sections of Au. anamensis and Au. afarensis mandibular corpora at distal P3.(a) KNM-KP 29281; (b) KNM- ER 20432; (c) MSD-VP-5/50; (d) A.L. 400-1a (right side reversed); (e) A.L. 277-1 and (f) A.L. 417-1a. Modified from Kimbel et al. (2006). Cross-section of MSD-VP-5/50 was acquired following the methods described by Kimbel et al. (2006). Scale bar, 2 cm. good transition from Kanapoi to Laetoli/Hadar lateral the mandibular features shared with Au. anamensis are the mandibular corpus profiles (figure 3). maximum anterior symphyseal breadth being at the These WORMIL mandibles show a mosaic of features canine (for example, MSD-VP-5/16; Ward et al. 1999) shared with both Au. anamensis and Au. afarensis.Someof and the more posterior position of the ascending ramus

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Hominids from Woranso-Mille, Ethiopia Y. Haile-Selassie 3327

(a) (b)

RdC LdC RdC RdC 04cm

Figure 4. Dental remains from Woranso-Mille. (a) Occlusal (top row) and mesial (bottom row) views of ARI-VP-3/80g (LM2), 3 ARI-VP-1/90 (LM ), ARI-VP-3/80d (LM2) and ARI-VP-1/462 (LM1). These molars show the lingual slope on upper and buccal slope on lower molars like Au. anamensis.(b) MSD-VP-5/50, ARI-VP-3/80a and ARI-VP-2/95, P3s from the Woranso-Mille showing variation in P3 occlusal crown morphology. (c) Comparison of lower deciduous canine root length relative to crown height. KNM-KP 34 725 (Au. anamensis), ARI-VP-1/190 (Woranso-Mille) and A.L. 333-35 (Au. afarensis). Like Au. anamensis, the Woranso-Mille specimen has longer root relative to the crown height compared with Au. afarensis. Image of the Au. anamensis specimen was obtained from Carol Ward and A.L. 333-35 was made from cast. root (MSD-VP-5/50, see figure 2; Haile-Selassie et al. however, indicate that the WORMIL specimens are 2010b), which is also shared with the earlier Ar. ramidus morphologically intermediate between Au. anamensis (Suwa et al. 2009; White et al. 2009). The mandibular and Au. afarensis. features shared with Au. afarensis include the presence of an incipient lateral corpus hollow (for example, 4. PHYLOGENETIC RELATIONSHIPS MSD-VP-5/16), usually described as characteristic of Ardipithecus ramidus from the Middle Awash and Gona this species. Judged from the preserved parts of MSD- study areas in Ethiopia, dated between 4.3 and 4.6 Ma VP-5/16, the mandibular symphysis does not show the (WoldeGabriel et al.1994; Semaw et al.2005), is con- more posteroinferiorly retreating condition seen in Au. sidered to be the possible ancestor of Au. anamensis anamensis (Ward et al. 2001; Kimbel et al. 2006). How- although other possibilities cannot be ruled out ever, MSD-VP-5/50 shows a more receding symphysis (White et al. 2006, 2009). Recent studies on a larger asseenfromthetransversecross-sectionattheP3 level. sample of Ar. ramidus material, including a partial skeleton, from the Middle Awash show that Ar. ramidus and (b) Dentition Au. anamensis occupied different ‘adaptive plateaus’ In terms of the dentition, the upper and lower molars (White et al.2009). Although further investigation is have occlusally tapering lingual and buccal faces, imperative to understand the full implication of these respectively, a trait documented in Au. anamensis adaptive differences, the ancestor–descendant relation- (Ward et al. 2001). As in Au. anamensis,theupper ship between these two species remains one of the molars, particularly M2–3, taper distally (figure 4a). alternatives given the currently available fossil evidence. The deciduous lower canine (ARI-VP-1/190) is similar The temporal distribution and apparent, but lim- in its crown morphology to Au. afarensis specimens such ited, morphological differences between the as A.L. 333-35, LH 2, and DIK-1-1 from Hadar, Lae- hypodigms currently divided into Au. afarensis and toli and Dikika, respectively (White 1977, 1980a,b; Au. anamensis justified the retention of both species Johanson et al.1982; Alemseged et al.2006). However, names as a chronospecies pair (Leakey et al. 1995, ARI-VP-1/190 has a relatively much longer root and 1998; Ward et al. 2001; Kimbel et al. 2006; White less lingual relief as seen in Au. anamensis (figure 4c). et al. 2006). However, the discovery of the The known P3s from WORMIL show both Au. afaren- WORMIL specimens dated at 3.7–3.8 Ma minimizes sis-like (ARI-VP-2/95) and Au. anamensis-like the differences between the inferred chronospecies (ARI-VP-3/80, MSD-VP-5/50) occlusal morphology pair and suggests that it represents a single lineage and document variation in the morphology of the P3 and should probably be referred to by a single species (figure 4b). The P3 occlusal morphology of MSD-VP- name in order to avoid taxonomic confusion. Australo- 5/50 is more similar to KNM-ER 20 432 from Allia pithecus anamensis is interpreted to have been an Bay than to KNM-KP 27 286 from Kanapoi in its obligate biped (Leakey et al. 1995, 1998; Ward et al. mesiodistally elongated crown and asymmetry, clearly 2001), unlike Ar. ramidus, which was a facultative defined mesial marginal ridge, and larger posterior biped with substantial arboreal adaptation (Lovejoy fovea. The lingual cusp is also hardly developed. et al. 2009). Australopithecus anamensis shares a Enamel thickness, as measured on naturally frac- number of derived dental characters and locomotor tured WORMIL teeth, is also within the range seen adaptations with Au. afarensis, and both are grouped for both hypodigms. The postcranial elements are in the same ‘adaptive plateau’ (White et al. 2009). not as informative owing to their fragmentary nature. The presence of temporal and spatial discontinuity in The dentognathic morphological observations, their fossil record, rather than observed morphological

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3328 Y. Haile-Selassie Hominids from Woranso-Mille, Ethiopia differences, was probably one of the apparent reasons anamensis than with Au. afarensis. New discoveries to distinguish them at the species level. The discovery from the 2009 field season support this observation of the WORMIL hominids not only fills some part of by yielding critical information on the anterior man- the spatial and temporal discontinuity, but also dibular morphology of the Woranso-Mille hominids. reduces the inferred anatomical differences between The transverse profile of MSD-VP-5/50 at posterior the two populations. P3 shows that the lateral mandibular corpus shape is The phylogenetic relationship between the two time- more like those from Allia Bay and Kanapoi than like successive ‘species’ of Au. anamensis (3.9–4.2 Ma) and those of Au. afarensis (figure 3) shown by Kimbel Au. afarensis (3.0–3.6 Ma) has been addressed in detail et al. (2006). If the Woranso-Mille hominids dated using various analytical methods (Leakey et al. 1995, between 3.7 and 3.8 Ma had to be assigned to one of 1998; Ward et al. 1999, 2001; White 2002; Kimbel these two arbitrary groups, they could be put into et al.2006; White et al.2006). Australopithecus anamen- the earlier Au. anamensis with some confidence. The sis is suggested to have rapidly evolved from its putative temporal range of Au. anamensis would then be ancestor Ar. ramidus (White et al.2006). However, this extended to between 3.7 and 4.2 Ma. It should also remains to be more rigorously tested, particularly in be noted that Au. anamensis from Asa Issie is bracketed light of the new revelations about Ar. ramidus (e.g. between 3.77 and 4.2 Ma (White et al. 2006). Lovejoy et al.2009; Suwa et al.2009; White et al. Although this does not result in any changes in terms 2009). Australopithecus anamensis and Au. afarensis are of how these two groups are related to each other, it considered by most workers as a chronospecies pair would mean that specimens such as the Belohdeli sampling a single phyletically evolving lineage, although frontal (BEL-VP-1/1; Asfaw 1987), the hominid speci- other alternatives have also been entertained (Kimbel mens from Fejej (Fleagle et al. 1991; Kappleman et al. et al. 2006; White et al.2006). Most analytical methods 1996; Grine et al. 2006a,b), and teeth and femur frag- have, thus far, failed to falsify the proposed ancestor– ment from Galili (Haile-Selassie & Asfaw 2000; descendant relationship, or unequivocally recognize Macchiarelli et al. 2004; Viola et al. 2008) should be them as two distinct lineages. recognized as Au. anamensis. However, this assignment The Woranso-Mille hominids dated at 3.6–3.8 Ma would simply be based on their geological age although are morphometrically intermediate between Au. afarensis it gives great valence to the idea of Au. anamensis–Au. and Au. anamensis. They represent the best fossil homi- afarensis being a chronospecies pair. nid sample that clearly bridges the temporal and morphological gaps between Au. afarensis and Au. anamensis, lending strong support to the suggestion that 5. DISCUSSION they represent a chronospecies pair (White et al.2009). The 3.7–3.8 Ma WORMIL hominid specimens share Therefore, every available line of evidence suggests that a number of dental characters with both Au. afarensis Au. anamensis represents an earlier deme of Au. afarensis. and Au. anamensis. They are temporally and morpho- Their separation at a species level was clearly an artefact logically intermediate between the two groups and ofthelackofadequatefossilsfromthetimebetween suggest that Au. anamensis and Au. afarensis do not Allia Bay (3.9 Ma) and Laetoli (3.6 Ma) than the pres- warrant an evolutionarily meaningful distinction at ence of discrete and functionally significant anatomical the species level. The Woranso-Mille hominids clearly characters distinguishing the two groups. connect Au. anamensis and Au. afarensis regardless of Cladistic analysis on four site-samples of Au. afaren- which end of the continuum they belong to, and sis and Au. anamensis (Hadar, Laetoli, Allia Bay and suggest that the recognition of two different species Kanapoi) demonstrated that the Au. anamensis–Au. names for two temporally and morphologically con- afarensis lineage is paraphyletic because the younger tinuous populations of a single phyletically evolving Au. afarensis sample from Hadar appears to be the species is confusing and unwarranted. Following the sister taxon of Au. africanus (Kimbel et al. 2006, currently available classification, the dental and man- p. 145). Moreover, even Au. afarensis is paraphyletic dibular morphological similarities of the WORMIL since the Laetoli sample shares some dentognathic fea- specimens (dated at 3.7–3.8 Ma) with Au. anamensis tures exclusively with Au. anamensis. The Woranso- outweigh their similarity with Au. afarensis sensu stricto Mille specimens, regardless to which group they are (i.e. specimens from Hadar and Laetoli). If the two assigned, further demonstrate the paraphyly of not names have to be retained, mainly for communication only the entire Au. afarensis–Au. anamensis lineage, purposes as some researchers suggest, the available but but also that of Au. anamensis. Kimbel et al. (2006, limited evidence supports assignment of the p. 146) suggested that the two most preferred solutions WORMIL hominids to Au. anamensis, extending the for the taxonomic conundrum related to Au. afarensis temporal range of the latter group to 3.7 Ma. and Au. anamensis are the ‘recognition of a single evol- There is considerable size range in the relatively utionary species or the maintenance of the status quo small WORMIL upper (n ¼ 9) and lower (n ¼ 19) ...’. The evidence from Woranso-Mille strongly sup- molar samples although they fall within the range of ports the former as the most parsimonious solution both Au. anamensis and Au. afarensis. However, they and Au. africanus as the sister taxon of this species. also represent some of the largest molars in the entire Haile-Selassie et al. (2010b) avoided specific assign- Au. anamensis/Au. afarensis sample, particularly the ment of the Woranso-Mille specimens to either group. M3s (for example, ARI-VP-1/90, ARI-VP-1/215). Based on the dentognathic description and compari- Lockwood et al.(2000)observed a statistically signifi- son, however, they showed that the Woranso-Mille cant temporal trend towards an increase in M3 crown hominids shared more dental characters with Au. size (mainly by mesiodistal elongation) in Au. afarensis,

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Hominids from Woranso-Mille, Ethiopia Y. Haile-Selassie 3329 although they noted that outliers (Lockwood et al. others for discussion and access to their unpublished data. 2000) and small sample sizes (White 1985)from I thank Stephanie Melillo for assistance with figures and specific time periods could bias this observation. At Liz Russell for photography. Finally, discussions with Bill the same time, an increase in M3 crown area might Kimbel, Denise Su, Gen Suwa, Carol Ward and Tim White significantly improved the content of this have been the general trend in the Au. anamensis–Au. manuscript. The WORMIL project was financially afarensis lineage, as seen in the shift from molars with supported by National Science Foundation (NSF; grant nos sloping buccal (lowers) and lingual (uppers) faces in 0234320, 0542037), The Leakey Foundation, Wenner-Gren Au. anamensis to molars with more vertical lingual Foundation, and National Geographic Society. and buccal faces in Au. afarensis, which would have resulted in an increase in overall crown occlusal surface. This increase in occlusal area would have resulted in REFERENCES larger chewing surface, which could be linked to an Alemseged, Z., Spoor, F., Kimbel, W. H., Bobe, R., increase in trophic capability in using a wide variety of Geraads, D., Reed, D. & Wynn, J. G. 2006 A juvenile resources (Teaford & Ungar 2000;Grineet al. early hominin from Dikika, Ethiopia. Nature 443, 2006a,b), ranging from soft fruits and leaves to harder 296–301. (doi:10.1038/nature05047) and brittle fallback foods (Ungar 2004; Grine et al. Arambourg, C. & Coppens, Y. 1968 Decouverte d’un Australopithecien nouveau dans les gisements de l’Omo 2006b). Although Au. afarensis was probably better (Ethiopie). S. Afr. J. Sci. 64, 58–59. fitted (largely owing to its larger molars and thicker Asfaw, B. 1987 The Belohdelie frontal—new evidence of enamel, among others) for a variety of food items early hominid cranial morphology from the Afar of including fallback resources, dental microwear analyses Ethiopia. J. Hum. Evol. 16, 611–624. (doi:10.1016/ on a limited number of teeth fails to show differences in 0047-2484(87)90016-9) the dietary preferences of Au. afarensis and Au. anamen- Asfaw, B., White, T., Lovejoy, O., Latimer, B., Simpson, S. & sis (Grine et al. 2006a,b). However, this remains to be Suwa,G.1999Australopithecus garhi: a new species of early tested with a larger sample of Au. anamensis. hominid from Ethiopia. Science 284,629–635.(doi:10. 1126/science.284.5414.629) Broom, R. 1938 The Pleistocene anthropoid apes of South Africa. Nature 142, 377–379. (doi:10.1038/142377a0) 6. CONCLUSION Brunet, M., Beauvillain, A., Coppens, Y., Heintz, E., Palaeontological fieldwork at a newly discovered fossi- Moutaye, A. H. E. & Pilbeam, D. 1995 The first Austra- liferous area in the central Afar region of Ethiopia has lopithecine 2,500 kilometers west of the Rift Valley yielded new hominid fossil remains dated between 3.4 (Chad). Nature 378, 273–275. (doi:10.1038/378273a0) and 3.8 Ma. The temporal placement of these fossils Brunet, M. et al. 2002 A new hominid from the upper renders them crucial to test some of the outstanding Miocene of Chad, central Africa. Nature 418, 145–151. (doi:10.1038/nature00879) human evolutionary hypotheses such as the phylogen- Brunet, M. et al. 2005 New material of the earliest hominid tic relationships between Ar. ramidus and Au. from the upper Miocene of Chad. Nature 434, 752–755. anamensis, and the issue of ancestor–descendant (doi:10.1038/nature03392) relationship between Au. anamensis and Au. afarensis. Cela-Conde, C. & Altaba, C. 2002 Multiplying genera The 3.7–3.8 Ma hominid specimens from the Wor- versus moving species: a new taxonomic proposal for anso-Mille clearly show that they are morphologically the family. S. Afr. J. Sci. 98, 229–232. intermediate between Au. anamensis and Au. afarensis Cela-Conde, C. & Ayala, F. 2007 Human evolution: trails from and their dental measurements overlap with both the past. Oxford, UK: Oxford University Press. groups, lending support to the hypothesis of Au. ana- Clarke, R. J. 1977 The cranium of the Swartkrans hominid mensis/Au. afarensis being a chronospecies pair SK 847 and its relevance to human origins. PhD disser- representing a single lineage. Regardless of what part tation, University of the Witwatersrand, Johannesburg. Dart, R. 1925 Australopithecus africanus: the man-ape of South of this ‘pair’ the Woranso-Mille hominids are assigned Africa. Nature 115,195–199.(doi:10.1038/115195a0) to, however, they document the best example of the Deino, A., Scott, G. R., Saylor, B., Alene, M., Angelini, J. & presence of transitional populations in a single phyleti- Haile-Selassie, Y. 2010 40Ar/39Ar dating of Woranso- cally evolving hominid lineage. Mille: a new hominid-bearing Pliocene site in the Slightly younger fossil hominids (3.0–3.6 Ma) not west-central Afar Rift, Ethiopia. J. Hum. Evol. 58, described here promise to shed some light on current 111–126. (doi:10.1016/j.jhevol.2009.11.001) debates related to early middle Pliocene hominid diver- Fleagle, J. G., Rasmussen, D. T., Yirga, S., Bown, T. M. & sity. The Woranso-Mille palaeontological site has Grine, F. E. 1991 New hominid fossils from Fejej, opened a new window into the deep human past and Southern Ethiopia. J. Hum. Evol. 21, 145–152. (doi:10. promises to yield more fossils relevant to answering 1016/0047-2484(91)90005-G) Grine, F. E. 1986 Dental evidence for dietary differences in crucial questions in human evolutionary studies, Australopithecus and Paranthropus: a quantitative analysis including the presence or absence of early hominid of permanent molar microwear. J. Hum. Evol. 15, diversity during the middle Pliocene. 783–882. (doi:10.1016/S0047-2484(86)80010-0) I thank the Authority for Research and Conservation of Grine, F. E. 1988 New craniodental fossils of Paranthropus Cultural Heritage and the National Museum of Ethiopia of from the Swartkrans Formation and their significance in the Ministry of Culture and Tourism for field and ‘robust’ australopithecine evolution. Evolutionary history laboratory permissions and facilities, the Afar Regional of the ‘robust’ Australopithecines (ed. F. E. Grine), Government, its local administrative units, and the Afar pp. 223–243. New York, NY: Aldine de Gruyter. people of Mille District, for facilitating and directly Grine, F. E., Ungar, P. S. & Teaford, M. F. 2006a Was the participating in my research endeavours. I thank all of the early Pliocene hominin ‘Australopithecus’ anamensis a field participants who helped find the fossils and many hard object feeder? S. Afr. J. Sci. 102, 301–310.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3330 Y. Haile-Selassie Hominids from Woranso-Mille, Ethiopia

Grine, F. E., Ungar, P. S., Teaford, M. F. & El-Zaatari, S. Kimbel, W. H., Lockwood, C. A., Ward, C. V., Leakey, 2006b Molar microwear in Praeanthropus afarensis: M. G., Rak, Y. & Johanson, D. C. 2006 Was Australopithe- evidence for dietary stasis through time and under diverse cus anamensis ancestral to A. afarensis? A case of paleoecological conditions. J. Hum. Evol. 51, 297–319. anagenesis in the hominin fossil record. J. Hum. Evol. (doi:10.1016/j.jhevol.2006.04.004) 51, 134–152. (doi:10.1016/j.jhevol.2006.02.003) Haile-Selassie, Y. 2001 Late Miocene hominids from the Kramer, P. A. 1999 Modeling the locomotor energetics of Middle Awash, Ethiopia. Nature 412, 178–181. (doi:10. extinct hominids. J. Exp. Biol. 202, 2807–2818. 1038/35084063) Latimer, B. 1983 The anterior foot skeleton of Australopithe- Haile-Selassie, Y. & Asfaw, B. 2000 A newly discovered early cus afarensis. Am. J. Phys. Anthropol. 60, 217. Pliocene hominid bearing paleontological site in the Latimer, B. 1991 Locomotor adaptations in Australopithecus Mulu Basin, Ethiopia. Am. J. Phys. Anthropol. 111, 170. afarensis: the issue of arboreality. In Origine(s) de la bipe´die Haile-Selassie, Y., Suwa, G. & White, T. D. 2004 Late chez les Hominide´s (eds B. Senut & Y. Coppens), Miocene teeth from Middle Awash. Ethiopia, and early pp. 169–176. Paris, France: CNRS. hominid dental evolution. Science 303, 1503–1505. Latimer, B. & Lovejoy, C. O. 1989 The calcaneus of Austra- (doi:10.1126/science.1092978) lopithecus afarensis and its implications for the evolution of Haile-Selassie, Y., Deino, A., Saylor, B., Umer, M. & bipedality. Am. J. Phys. Anthropol. 78, 369–386. (doi:10. Latimer, B. 2007 Preliminary geology and paleontology 1002/ajpa.1330780306) of new hominid-bearing Pliocene localities in the central Leakey, L. S. B. 1959 A new fossil skull from Olduvai. Afar region of Ethiopia. Anthropol. Sci. 115, 215–222. Nature 184, 491–493. (doi:10.1038/184491a0) (doi:10.1537/ase.070426) Leakey,M.D.1987TheLaetolihominidremains.In Haile-Selassie, Y., Suwa, G. & White, T. D. 2009 Chapter 7: Laetoli: a Pliocene site in Northern Tanzania (eds M. D. Hominidae. In Ardipithecus kadabba. Evidence from the Leakey & J. M. Harris), pp. 108–109. Oxford, UK: Late Miocene of Africa (eds Y. Haile-Selassie & G. Wolde- Clarendon Press. Gabriel), pp. 159–236. Berkeley, CA: University of Leakey, R. E. F. & Walker, A. C. 1980 Status of Australo- California Press. pithecus afarensis. Science 207, 1103–1103. (doi:10.1126/ Haile-Selassie, Y., Latimer, B., Alene, M.., Deino, A. L., science.207.4435.1103) Gibert, L., Melillo, S. M., Saylor, B. Z., Scott, G. R. & Leakey, M. D., Hay, R. L., Curtis, G. H., Drake, R. E., Lovejoy, O. C. 2010a An early Australopithecus afarensis Jakes, M. K. & White, T. D. 1976 Fossil hominids form postcranium from Woranso-Mille, Ethiopia. Proc. Natl the Laetolil Beds. Nature 262, 460–466. (doi:10.1038/ Acad. Sci. USA 107, 12 121–12 126. (doi: 10.1073/ 262460a0) pnas.1004527107) Leakey, M. G., Feibel, C. S., McDougall, I. & Walker, A. Haile-Selassie, Y., Saylor, B., Deino, A., Alene, M. & 1995 New four-million-year-old hominid species from Latimer, B. 2010b New hominid fossils from Woranso- Kanapoi and Allia Bay, Kenya. Nature 376, 565–571. Mille (central Afar, Ethiopia) and taxonomy of early (doi:10.1038/376565a0) Australopithecus. Am. J. Phys. Anthropol. 141, 406–417. Leakey, M. G., Feibel, C. S., McDougall, I., Ward, C. & (doi:10.1002/ajpa.21159) Walker, A. 1998 New specimens and conﬁrmation of an Johanson, D. C. & White, T. D. 1979 A systematic assess- early age for Australopithecus anamensis. Nature 393,62– ment of early African hominids. Science 203, 321–330. 66. (doi:10.1038/29972) (doi:10.1126/science.104384) Lockwood, C. A., Kimbel, W. H. & Johanson, D. C. 2000 Johanson, D. C., White, T. D. & Coppens, Y. 1978 A new Temporal trends and metric variation in the mandibles species of the genus Australopithecus (Primates: Hominidae) and teeth of Australopithecus afarensis. J. Hum. Evol. 39, from the Pliocene of eastern Africa. Kirtlandia 28, 1–14. 23–55. (doi:10.1006/jhev.2000.0401) Johanson, D. C., White, T. D. & Coppens, Y. 1982 Dental Lovejoy, C. O. 1975 Biomechanical perspectives on the remains from the Hadar Formation. Ethiopia: 1974– lower limb of early hominids. In Primate functional mor- 1977 collections. Am. J. Phys. Anthropol. 57, 545–603. phology and evolution (ed. R. H. Tuttle), pp. 291–306. (doi:10.1002/ajpa.1330570406) Paris, France: Mouton. Jungers, W. L. & Stern, J. T. 1983 Body proportions, skeletal Lovejoy, C. O. 1978 A biomechanical review of the loco- allometry and locomotion in the Hadar hominids: a reply motor diversity of early hominids. In Early hominids of to Wolpoff. J. Hum. Evol. 12, 673–684. (doi:10.1016/ Africa (ed. C. Jolly), pp. 403–429. New York, NY: S0047-2484(83)80007-4) St. Martin’s Press. Kappleman, J., Swisher III, C. C., Fleagle, J. G., Yirga, S., Lovejoy, C. O., Latimer, B., Suwa, G., Asfaw, B. & White, Bown, T. M. & Feseha, M. 1996 Age of Australopithecus T. D. 2009 Combining prehension and propulsion: the afarensis from Fejej, Ethiopia. J. Hum. Evol. 30, 139– foot of Ardipithecus ramidus. Science 326, 72e1–72e8. 146. (doi:10.1006/jhev.1996.0010) Macchiarelli, R. et al. 2004 Early pliocene hominid tooth Kimbel, W. H. & Delezene, L. K. 2009 ‘Lucy’ Redux: a from Galili, Somali Region, Ethiopia. Coll. Anthropol. review of research on Australopithecus afarensis. 28, 65–76. Am. J. Phys. Anthropol. 52, 2–48. Olson, T. R. 1981 Basicranial morphology of the extant Kimbel, W. H. & White, T. D. 1988 Variation, sexual dimorph- hominoids and Pliocene hominids: the new material ism and the taxonomy of Australopithecus.InEvolutionary from the Hadar Formation, Ethiopia and its signiﬁcance history of the ‘robust’ Australopithecines (ed. F. E. Grine), in early human evolution and taxonomy. In Aspects of pp. 175–192. New York, NY: Aldine de Gruyter. human evolution (ed. C. B. Stringer), pp. 99–128. Kimbel, W. H., White, T. D. & Johanson, D. C. 1985 London, UK: Taylor and Francis. Craniodental morphology of the hominids from Olson, T. R. 1985 Cranial morphology and systematics of Hadar and Laetoli: evidence of ‘Paranthropus’ and ‘Homo’ the Hadar Formation hominids and ‘Australopithecus’ in the mid-Pliocene of eastern Africa? In Ancestors: the hard africanus.InAncestors: the hard evidence (ed. E. Delson), evidence (ed. E. Delson), pp. 120–137. New York, NY: pp. 102–119. New York, NY: Alan R. Liss. Alan R. Liss. Pickford, M., Senut, B., Gommery, D. & Treild, J. 2002 Kimbel, W. H., Rak, Y. & Johanson, D. C. 2004 The skull of Bipedalism in Orrorin tugenensis revealed by its femora. Australopithecus afarensis. New York, NY: Oxford C. R. Palevol. 1, 1–13. (doi:10.1016/S1631-0683(02) University Press. 00011-8)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Hominids from Woranso-Mille, Ethiopia Y. Haile-Selassie 3331

Plavkan, J. M., Lockwood, C. A., Kimbel, W. H., Lague, Ward, C. V., Leakey, M. G. & Walker, A. 1999 The new M. R. & Harmon, E. H. 2005 Sexual dimorphism in hominid species Australopithecus anamensis. Evol. Anthro- Australopithecus afarensis revisited: how strong is the case pol. 7, 197–205. (doi:10.1002/(SICI)1520-6505(1999) for a human-like patter of dimorphism? J. Hum. Evol. 7:6,197::AID-EVAN4.3.0.CO;2-T) 48, 313–320. (doi:10.1016/j.jhevol.2004.09.006) Ward, C. V., Leakey, M. G. & Walker, A. 2001 Morphology Reno, P. L., Meindl, R. S., McCollum, M. A. & Lovejoy, of Australopithecus anamensis from Kanapoi and Allia Bay, C. O. 2003 Sexual dimorphism in Australopithecus afaren- Kenya. J. Hum. Evol. 41, 255–368. (doi:10.1006/jhev. sis was similar to that of modern humans. Proc. Natl Acad. 2001.0507) Sci. USA 100, 9404–9410. (doi:10.1073/pnas.1133 White, T. D. 1977 New fossil hominids from Laetolil, Tanza- 180100) nia. Am. J. Phys. Anthropol. 46, 197–230. (doi:10.1002/ Reno, P. L., Meindl, R. S., McCollum, M. A. & Lovejoy, ajpa.1330460203) C. O. 2005 The case is unchanged and remains robust: White, T. D. 1980a Evolutionary implications of Pliocene Australopithecus afarensis exhibits only moderate dimorph- hominid footprints. Science 208, 175–176. (doi:10.1126/ ism. A reply to Plavkan et al. (2005). J. Hum. Evol. 49, science.208.4440.175) 279–288. (doi:10.1016/j.jhevol.2005.04.008) White, T. D. 1980b Additional fossil hominids from Laetoli, Semaw, S., Simpson, S. W., Quade, J., Renne, P. R., Butler, Tanzania: 1976–1979 specimens. Am. J. Phys. Anthropol. R. F., McIntosh, W. C., Levin, N., Dominguez-Rodrigo, 53, 487–504. (doi:10.1002/ajpa.1330530405) M. & Rogers, M. 2005 Early Pliocene hominids from White, T. D. 1985 The hominids of Hadar and Laetoli: an Gona. Nature 433, 301–305. (doi:10.1038/nature03177) element-by-element comparison of the dental samples. Senut, B. & Tardieu, C. 1985 Functional aspects of In Ancestors: the hard evidence (ed. E. Delson), pp. 138– Plio-Pleistocene hominid limb bones: implications for 152. New York, NY: Alan R. Liss. taxonomy and phylogeny. In Ancestors: the hard evidence White, T. D. 2002 Earliest hominids. In The primate fossil (ed. E. Delson), pp. 193–201. New York, NY: Alan R. record (ed. W. Hartwig), pp. 407–417. Cambridge, UK: Liss. Cambridge University Press. Senut, B., Pickford, M., Gommery, D., Mein, P., Cheboi, K. & White, T. D. & Johanson, D. C. 1982 Pliocene hominid Coppens, Y. 2001 First hominid from the Miocene mandibles from the Hadar Formation. Ethiopia: (Lukeino Formation, Kenya). C. R. Acad. Sci. IIA332, 1974–1977 collections. Am. J. Phys. Anthropol. 57, 137–144. 501–544. (doi:10.1002/ajpa.1330570405) Stern, J. T. 2000 Climbing to the top: a personal memoir White, T. D. & Suwa, G. 1987 Hominid footprints at of Australopithecus afarensis. Evol. Anthropol. 9, 113– Laetoli: facts and interpretations. Am. J. Phys. Anthropol. 133. (doi:10.1002/1520-6505(2000)9:3,113::AID-EVA 72, 485–514. (doi:10.1002/ajpa.1330720409) N2.3.0.CO;2-W) White, T. D., Johanson, D. C. & Kimbel, W. H. 1981 Stern, J. T. & Susman, R. L. 1981 Electromyography of the Australopithecus africanus: its phyletic position reconsid- gluteal muscles in Hylobates, Pongo, and Pan: implications ered. S. Afr. J. Sci. 77, 445–470. for the evolution of hominid bipedality. Am. J. Phys. White, T. D., Suwa, G., Hart, W. K., Walter, R. C., Wolde- Anthropol. 55, 153–166. (doi:10.1002/ajpa.1330550203) Gabriel,G.,deHeinzelin,J.,Clark,J.D.,Asfaw,B.& Stern, J. T. & Susman, R. L. 1983 The locomotor anatomy Vrba, E. 1993 New discoveries of Australopithecus at Maka of Australopithecus afarensis. Am. J. Phys. Anthropol. 60, in Ethiopia. Nature 366, 261–265. (doi:10.1038/366261a0) 279–317. (doi:10.1002/ajpa.1330600302) White, T. D., Suwa, G. & Asfaw, B. 1994 Australopithecus Stern, J. T. & Susman, R. L. 1991 ‘Total morphological pat- ramidus, a new species of early hominid from Aramis, tern’ versus the ‘magic trait:’ conﬂicting approaches to the Ethiopia. Nature 371, 306–312. (doi:10.1038/371306a0) study of early hominid bipedalism. In Origine(s) de la bipe´- White, T. D., Suwa, G. & Asfaw, B. 1995 Corrigendum. die chez les Hominide´s (eds B. Senut & Y. Coppens), Nature 375, 88. (doi:10.1038/375088a0) pp. 99–112. Paris, France: CNRS. White, T. D., Suwa, G., Simpson, S. W. & Asfaw, B. 2000 Strait, D. S. & Grine, F. E. 2004 Inferring hominoid and Jaws and teeth of Australopithecus afarensis from Maka, early hominid phylogeny using craniodental characters: Middle Awash, Ethiopia. Am. J. Phys. Anthropol. 111, the role of fossil taxa. J. Hum. Evol. 47, 399–452. 45–68. (doi:10.1002/(SICI)1096-8644(200001)111: (doi:10.1016/j.jhevol.2004.08.008) 1,45::AID-AJPA4.3.0.CO;2-I) Strait, D. S., Grine, F. E. & Moniz, M. A. 1997 A reappraisal White, T. D. et al. 2006 Asa Issie, Aramis and the origin of of early hominid phylogeny. J. Hum. Evol. 32, 17–82. Australopithecus. Nature 440, 883–889. (doi:10.1038/ (doi:10.1006/jhev.1996.0097) nature04629) Suwa, G., Kono, R. T., Simpson, S. W., Asfaw, B., Lovejoy, White,T.D.,Asfaw,B.,Beyene,Y.,Haile-Selassie,Y.,Lovejoy, C. O. & White, T. D. 2009 Paleobiological implications of O. C., Suwa, G. & WoldeGabriel, G. 2009 Ardipithecus the Ardipithecus ramidus dentition. Science 326, 94–99. ramidus and the paleobiology of early hominids. Science Teaford, M. F. & Ungar, P. S. 2000 Diet and the evolution of 326,75–86.(doi:10.1126/science.1175802) the earliest human ancestors. Proc. Natl Acad. Sci. USA WoldeGabriel, G., White, T. D., Suwa, G., Renne, P. R., de 95, 13 506–13 511. (doi:10.1073/pnas.260368897) Heinzelin, J., Hart, W. K. & Heiken, G. 1994 Ecological Tobias, P. V. 1967 The cranium and maxillary dentition of Aus- and temporal placement of early Pliocene hominids at tralopithecus (Zinjanthropus) boisei. Olduvai Gorge, Vol. 2. Aramis, Ethiopia. Nature 371, 330–333. (doi:10.1038/ Cambridge, UK: Cambridge University Press. 371330a0) Turner, A. & Wood, B. A. 1993 Comparative palaeontologi- Wood, B. A. & Chamberlain, A. 1987 The nature and afﬁ- cal context for the evolution of the early hominid nities of the ‘robust’ australopithecines: a review. masticatory system. J. Hum. Evol. 24, 301–318. J. Hum. Evol. 16, 625–641. (doi:10.1016/0047- (doi:10.1006/jhev.1993.1023) 2484(87)90017-0) Ungar, P. 2004 Dental topography and diets of Australopithecus Wood, B. A. & Ellis, M. 1986 Evidence for dietary specializ- afarensis and early Homo. J. Hum. Evol. 46,605–622. ation in the ‘robust’ australopithecines. Anthropos 23, (doi:10.1016/j.jhevol.2004.03.004) 101–124. Viola, R., Kullmer, O., Sandrock, O., Hujer, W. & Seidler, Zihlman, A. 1985 Australopithecus afarensis: two sexes or two H. 2008 An early australopithecine femur from Galili, species? In Hominid evolution, past, present, and future (ed. Ethiopia. J. Vert. Paleo. 28(Suppl. 3), 156A–157A. P. V. Tobias), pp. 213–220. New York, NY: Alan R. Liss.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Phil. Trans. R. Soc. B (2010) 365, 3333–3344 doi:10.1098/rstb.2010.0039

Anterior dental evolution in the Australopithecus anamensis–afarensis lineage Carol V. Ward1,*, J. Michael Plavcan2 and Fredrick K. Manthi3,4 1Department of Pathology and Anatomical Sciences, University of Missouri, M263 Medical Sciences Building, Columbia, MO 65212, USA 2Department of Anthropology, University of Arkansas, 330 Old Main, Fayetteville, AR 72701, USA 3Department of Earth Sciences, National Museums of Kenya, P.O. Box 40658, Nairobi, Kenya 4Turkana Basin Institute, Stony Brook University, Stony Brook, NY 11794, USA Australopithecus anamensis is the earliest known species of the Australopithecus–human clade and is the likely ancestor of Australopithecus afarensis. Investigating possible selective pressures underlying these changes is key to understanding the patterns of selection shaping the origins and early evolution of the Australopithecus–human clade. During the course of the Au. anamensis–afarensis lineage, significant changes appear to occur particularly in the anterior dentition, but also in jaw structure and molar form, suggesting selection for altered diet and/or food processing. Specifically, canine tooth crown height does not change, but maxillary canines and P3s become shorter mesiodistally, canine tooth crowns become more symmetrical in profile and P3s less unicuspid. Canine roots diminish in size and dimorphism, especially relative to the size of the postcanine teeth. Molar crowns become higher. Tooth rows become more divergent and symphyseal form changes. Dietary change involving anterior dental use is also suggested by less intense anterior tooth wear in Au. afarensis. These dental changes signal selection for altered dietary behaviour and explain some differences in craniofacial form between these taxa. These data identify Au. anamensis not just as a more primitive version of Au. afarensis, but as a dynamic member of an evolving lineage leading to Au. afarensis, and raise intriguing questions about what other evolutionary changes occurred during the early evolution of the Australopithecus–human clade, and what characterized the origins of the group. Keywords: Australopithecus anamensis; Australopithecus afarensis; dental evolution

1. INTRODUCTION clade is only sketchy at present. Even so, emerging evi- Fossil evidence documenting the first 4 Myr of homi- dence from what few fossils are known is beginning to nin evolution has grown substantially over the past hint that Au. anamensis was a species in transition and two decades. While several early taxa have been ident- may offer important insights into the origins of a ified (Ardipithecus, Sahelanthropus and Orrorin), much number of key hominin traits. of our understanding of what the earliest members of A previous detailed investigation of morphological the Australopithecus–human clade were like still changes through time in the successive site samples comes from the best-known species of Australopithecus, of Au. anamensis and Au. afarensis (Kimbel et al. Australopithecus afarensis. However, Au. afarensis only 2006) documented a series of apomorphies that pro- appears as early as 3.6 Ma and is not well represented gressively appear throughout these samples, involving in the fossil record until 3.4–3 Ma (review in Kimbel primarily the dentition, but also some aspects of maxil- et al. 2006, see also White et al. 2000). The new fossils lary and mandibular form. The pattern of the from Woranso-Mille, Ethiopia (Haile-Selassie et al. appearance of these traits strongly supports the 2010), are 3.7–3.8 Ma and probably part of this lin- hypotheses of anagenetic evolution of Au. anamensis eage as well. Australopithecus anamensis is the earliest to Au. afarensis. known member in this clade, appearing by 4.17 Ma One of the most important apomorphies of the Aus- in Kenya and Ethiopia, and is the likely ancestor of tralopithecus–human clade is habitual terrestrial Au. afarensis. Unfortunately, Au. anamensis is relatively bipedality with loss of significant climbing abilities. poorly represented in the fossil record, so our under- Unfortunately, little is known about the postcranial standing about this first 400 000 to about 800 000 skeleton of Au. anamensis. Australopithecus anamensis years of the evolution of the Australopithecus–human is only known from a femoral shaft, along with some unpublished vertebral fragments, partial metatarsal, eroded distal pedal phalanx and manual phalanx, all * Author for correspondence ([email protected]). from Asa Issie (White et al. 2006), a distal humerus, One contribution of 14 to a Discussion Meeting Issue ‘The first four capitate, partial manual phalanx and partial tibia million years of human evolution’. from Kanapoi (Patterson & Howells 1967;

3333 This journal is q 2010 The Royal Society Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3334 C. V. Ward et al. Dental evolution in Australopithecus

Leakey et al. 1995, 1998), and a nearly complete Au. anamensis and Au. afarensis based on previously radius from Allia Bay (Patterson & Howells 1967; published fossils, integrate data from some newly dis- Heinrich et al. 1993; Leakey et al. 1995, 1998). The covered specimens from Kanapoi (Manthi et al.in tibial diaphysis is oriented orthogonally to the talo- preparation) which provide new insights and consider crural joint, as is that of all later hominins, indicating the suite of differences seen between these taxa in an a knee placed directly over the ankle during the adaptive and evolutionary context. We note that most single-limb support phase of terrestrial bipedal gait change occurs in the anterior portion of the face and (Ward et al. 1999b). However, more cannot be said jaws, with the most dramatic alterations occurring at at the present time about the details or extent of its or near the canine–premolar complex, but that patterns adaptation to terrestrial bipedality. In almost all other of change within teeth in proportions and shape often major features, the known Au. anamensis postcranial are uncorrelated. We propose that these changes elements resemble those attributed to Au. afarensis. signal possible dietary change and/or altered use of The exception may be in the capitate, which appears the anterior dentition in food processing in the early to have separate dorsal and plantar articular facets evolution of the Australopithecus–human clade, in for MC2 like in extant African apes, and unlike conjunction with shifts in masticatory adaptations. Australopithecus, Homo, Ardipithecus, Proconsul (Beard et al. 1986; Lovejoy et al. 2009) or the unknown 3.5 Ma hominin from South Turkwel, Kenya (Leakey 2. CANINE TOOTH SIZE, DIMORPHISM et al. 1998; Ward et al. 1999a). This small feature AND THE CANINE/P3 COMPLEX may indicate some differences in locomotor or manip- The evolution of the canine teeth and mandibular ulative function, but until more fossils are recovered honing premolar in hominins has received a great our ability to infer postcranial variation among species deal of attention ever since Darwin (1871). Canine is highly limited, and little more than can be said about tooth size reduction is one of the few defining features whether the same pattern of locomotor or manipula- of the hominin clade (Wolpoff 1980; Greenfield 1992; tive ability seen in Au. afarensis also characterized Haile-Selassie 2001, 2004; White et al. 2006, 2009; Au. anamensis. Suwa et al. 2009) and is recognized as a signal of Even less is known about its cranial anatomy. Only a important behavioural and adaptive changes (Plavcan & temporal bone and some maxillary fragments are Van Schaik 1997). For example, recent discussion of known. Australopithecus anamensis appears to have a Ardipithecus ramidus places great emphasis on the smaller external auditory porus than later hominins, importance of the canine/premolar complex for infer- and a potentially more obtuse angle of the tympanic ring changes in behaviour and diet—an assessment plate along with a weakly developed articular eminence with a long tradition in anthropology (e.g. Darwin (Leakey et al. 1995; Ward et al. 2001). However, little 1871; Brace 1971; Leutenegger & Kelly 1977; Wolpoff can be said of the functional or evolutionary signifi- 1978, 1979, 1980; Lovejoy 1981, 2009). cance of this morphology without more cranial fossils. At this point, two major features in hominin canine In contrast to the skeleton and skull, there are sev- evolution are widely accepted. First, male canine teeth eral aspects of the jaws and teeth that are preserved reduced in size relative to a likely ape ancestral con- for both Au. anamensis and Au. afarensis, enabling sig- dition, with a concomitant reduction of canine sexual nificant comparisons to be made in these elements. dimorphism, early in the hominin lineage (Brace Previous research has noted evolutionary changes in 1963; Jungers 1978; Wolpoff 1980; Greenfield 1992; relative canine size, canine and premolar morphology, Suwa et al. 2009). Second, by Au. afarensis, the mandibular and maxillary contours and incisor canine honing complex is reduced or lost (Greenfield dimensions (Leakey et al. 1995; Ward et al. 2001; 1992; Haile-Selassie 2001, 2004; Kimbel et al. 2006; Kimbel et al. 2006 ; White et al. 2006). However, the White et al. 2006). It is widely assumed that the loss relative paucity of fossils attributable to Au. anamensis of the hone is associated with selection for use of the obscures detailed understanding of the quality, quan- tooth in diet, probably in food acquisition. The most tity and integration of many features that impact explicit functional statement is that the mandibular hypotheses about the evolution of adaptations in this canine changes to a more diamond-shaped profile so lineage. that the mesial crest can occlude with the lateral Overall, while the Kanapoi and Allia Bay fossils are maxillary incisor (Greenfield 1992; Haile-Selassie distinguishable from those at Laetoli and Hadar, Au. 2004). This observation is used to support the anamensis is generally considered to be just an early hypothesis that canine reduction and changes in member of the Au. afarensis lineage, with a few isolated morphology are a consequence of selection for incor- morphological plesiomorphies. However, there is poration of the tooth into a functional incisal battery general consensus that Au. afarensis demonstrates (Greenfield 1992). evidence of a greater emphasis on the ability to masti- While it may seem that canine crown reduction is a cate tougher or more abrasive food items, possibly necessary precursor to alterations in canine form for associated with shifts in habitat or resource exploitation other, presumably dietary, functions, an alternative (Te af or d & U n g a r 2 0 0 0; Ward et al.2001; White et al. hypothesis has been proposed. It is possible that 2006) and/or broadening potential ecological niches. canine tooth crown shape change is integrally linked New fossil evidence provides even more evidence for to selection for the use of the canine in food processing shifting adaptations throughout this lineage. and so would have occurred concomitantly with size The purpose of this paper is to review and summar- reduction, not after it (Greenfield 1992). Such ize the morphology of the jaws and teeth of reduction could be linked in two ways—first would

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Dental evolution in Australopithecus C. V. Ward et al. 3335 be a general selective pressure for the use of canines in canine crowns are similar in all dimensions in Au. ana- food processing that results in crown size reduction fol- mensis and Au. afarensis, and both have slightly larger lowing relaxation of selection for the use of the tooth as and more dimorphic crowns than do modern a weapon. Second would be that selection for canine humans, as also reported for Ar. ramidus (Suwa et al. crown reduction would only be linked with changes in 2009) but less so than in extant apes. Therefore, not tooth form associated with dietary use of the tooth fol- only was there a clear dissociation between crown lowing an initial canine crown reduction and loss of size and root size in Au. anamensis, such that size dimorphism through a separate, unspecified mechan- and dimorphism in the crowns were lost while size ism. In other words, one model posits that canine and dimorphism in the roots were retained, but the tooth size in all primates, including hominins, reflects loss of root size dimorphism occurred sometime a balance between conflicting selection pressures for during the evolution of Au. anamensis into Au. afaren- large canines as weapons and small canines for food sis. Interestingly, not only does KNM-KP 47951 acquisition (Greenfield 1992), and the other posits demonstrate that the large alveolus of KNM-KP that selection for dietary functions only occurred after 29287 did not imply unusual crown size dimorphism, canine dimorphism was lost, with a secondary crown it also demonstrates that the canine root of KNM-KP reduction associated with the development of occlusal 29287 would not have been unusually large, and may features that transform the canine into a tool for food in fact have been a small male. It would not, however, processing and/or acquisition. have supported a larger crown than those preserved for The large sample of Ar. ramidus fossils from 4.4 Ma Au. anamensis (Plavcan et al. 2009). strongly suggests that substantial reduction in male Even with the new data, no dimensions of the man- canine crown size and loss of significant dimorphism dibular canine crowns (length, breadth or height) differ probably occurred near the origin of hominins and between species (figure 1, table 2). However, there are may not be apomorphic for the Australopithecus– dimensional differences in the maxillary canines (see human clade (White et al. 1994, 1995; Suwa et al. also Ward et al.2001; White et al. 2006)(figure 2, 2009). Indeed, Au. anamensis canine crowns appear to table 2). Australopithecus anamensis and Au. afarensis be approximately the same overall size as those of are equivalent in maxillary canine crown height and Ar. ramidus. Comparisons of associated dentitions buccolingual breadth, but Au. anamensis maxillary demonstrate Au. anamensis had slightly larger basal canines are mesiodistally longer than are those of dimensions of its canines relative to postcanine tooth Au. afarensis (figure 3a,tables2 and 3). Proportion- size than did Au. afarensis (Ward et al.2001). Overall, ately, Au. anamensis canines are almost exactly individual tooth sizes do not differ between the species, intermediate in basal crown shape (measured as mesio- with the exception of the maxillary canine mesiodistal distal length relative to buccolingual breadth) between dimension and some dimensions of P3 and P4.The extant great apes and Au. afarensis (figure 3a). Further- few preserved canine crowns in Au. anamensis appear more, Au. afarensis canine basal shape proportions are no more variable, or presumably dimorphic, than those identical to those of extant Homo, after accounting for of Au. afarensis sothereappearstobenoevidenceof their size difference. The apparent progressive decrease evolution of dimorphism during this time period, either. in relative canine size from Au. ramidus to Au. afarensis, However, a single large Au. anamensis mandibular through Au. anamensis (White et al.2006), tracks canine alveolus (KNM-KP 29287; Ward et al. 2001), mesiodistal length only, but not overall size of the tooth. and to some extent a canine root with heavily worn The change in canine basal proportions reflects crown from Fejej, Ethiopia (FJ-4-SB-1a; Fleagle change in canine–P3 function. It has long been et al. 1991), suggested potentially greater canine noted that the Au. afarensis canine/premolar complex crown size sexual dimorphism early in this lineage loses its ‘honing’ function, in the sense that the distal than previously appreciated, with the implication that edge of the maxillary canine no longer exclusively canine dimorphism decreased at some point in the wears against the labial surface of the mandibular P3 Au. anamensis–afarensis lineage. If so, this would be as in other primates (Wolpoff 1979; Greenfield 1992; evidence of social and/or dietary evolution. Haile-Selassie 2004). Metrically, it is only the maxil- Three new associated dentitions from Kanapoi have lary canine and mandibular premolar—the honing clarified details of canine size and proportions of Au. teeth—that change basal shape (figure 3b) to become anamensis (Manthi et al. in preparation), provided new less elongate in outline. Mandibular canines and data on canine proportions and morphology, and maxillary P3s do not. This implies no overall selection suggested that Au. anamensis is a key taxon for to reduce the teeth, but only selection to alter contact understanding the adaptive significance of changes in between the honing pair. So, while there may have canine form. been a loss of honing with early hominins (Haile- KNM-KP 47951 has the largest mandibular root of Selassie 2001, 2004; Brunet et al. 2002), canine–P3 any known hominin (figure 1). Comparison of man- occlusal relationships continue to evolve. dibular root size in Au. anamensis with those Au. In addition to basal outlines, the canines and P3 afarensis, extant great apes and Homo reveals that change other aspects of their crown shape significantly root size variation in Au. anamensis was very strong, from Au. anamensis to Au. afarensis, strongly most like Pongo in magnitude (figure 1, table 1), imply- suggesting that selection favoured an altered function ing strong dimorphism. This stands in stark contrast to of these teeth. Both mandibular and maxillary canines Au. afarensis, which shows much less variation in root become more symmetrical in lingual profile. Maxillary dimensions, and is intermediate to extant Homo and canines have higher shoulders and shorter mesial Pan in mean size. Unlike the roots, mandibular crests, and mandibular canines have lower mesial

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3336 C. V. Ward et al. Dental evolution in Australopithecus

(a)(35 b) 50 45 30 40 25 35 20 30 25 15 20 15 10 root length (mm) crown height (mm) 10 5 5 0 0

(c)(25 d) 25

20 20

15 15

10 10 root MD (mm) crown MD (mm)

5 5

0 0 (e)(18 f ) 18 16 16 14 14 12 12 10 10

8 root BL (mm) 8 crown BL (mm) 6 6 4 4 2 2 0 0

G. gorilla H. sapiens G. gorilla H. sapiens P. pygmaeus P. pygmaeus P. troglodytes Au. afarensisAu. anamesis P. troglodytes Au. afarensisAu. anamesis Figure 1. Dental dimensions for mandibular canine crowns and roots for extant great apes, humans, Au. anamensis and Au. afarensis. Arrows indicate new specimen KNM-KP 47951. Data in table 1.(a) Crown height, (b) root length, (c) crown mesiodistal, (d) root mesiodistal, (e) crown buccolingual and (f ) root buccolingual. shoulders and a less narrow, blade-like outline (figure 3a), demonstrating that the major proportional (figure 3c). Lower premolars develop a larger metaconid, changes in canine dimensions in hominin evolution and the protoconid shifts buccally. Marginal ridges happened between 3.9 and 3.4 Ma. become proportionately more distinctive, and the The dissociation between crown size and shape anterior fovea opens in a more occlusal direction changes demonstrates that selection impacting crown (Leakey et al. 1995, 1998; Ward et al.2001; Haile- shape was independent of that causing crown height Selassie 2004; Suwa et al. 2009; White et al.2006). reduction. This in turn bears on hypotheses that These shape changes increase transverse contact area purport to explain the adaptive significance of canine between maxillary and mandibular teeth, most logically crown size reduction in hominins (Brace 1963; owing to increased use of the canine in food acquisition Bailit & Friedlaender 1966; Wolpoff 1969, 1980; or preparation, and perhaps the premolar in mastication Calcagno & Gibson 1988). Crown height was reduced as well. prior to the appearance of Au. anamensis,ifAr. ramidus The fossils demonstrate that canine shape changed indeed reflects the ancestral hominin condition (Suwa significantly along with a shift in canine–P3 occlusal et al. 2009), and certainly by the origins of the relationships in the Au. anamensis–afarensis lineage, Australopithecus–human clade. Data now suggest that while canine size remained approximately the same. crown height did not reduce in order to provide It is also notable that Au. afarensis canine crowns room for expanding postcanine dentitions (Jungers show the same basal proportions as in Homo 1978), because basal dimensions and root size did

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Dental evolution in Australopithecus C. V. Ward et al. 3337

Table 1. Descriptive statistics for mandibular canine dimensions of extant combined-sex and fossil samples. All data are in millimetres. Data for Gorilla, Pongo, Pan troglodytes and Homo were collected for this project. Data for Pan pansicus were taken from Plavcan (1990) and do not include values for root dimensions. Height, crown height; BL, buccolingual; MD, mesiodistal; RBL, root buccolingual; RMD, root mesiodistal; RL, root length.

height BL MD RBL RMD RL

Gorilla gorilla n 50 50 50 50 50 50 min. 12.65 9.18 11.53 8.65 11.12 22.02 max. 31.25 16.26 21.66 16.03 21.4 45.27 mean 20.89 12.44 15.55 11.89 15.17 33.18 standard deviation 5.47 2.32 2.82 2.41 3.03 4.13 CV 26.2 18.6 18.1 20.3 20.0 12.4 Pongo pygmaeus n 16 17 17 17 17 16 min. 13.58 7.97 11.08 7.4 10.2 20.35 max. 26.82 14.53 17.8 14.17 17.52 41.04 mean 19.52 10.49 13.83 10.12 13.39 27.56 standard deviation 4.18 2.51 2.13 2.48 2.31 6.09 CV 21.4 23.9 15.4 24.5 17.25 22.1 Pan troglodytes n 30 30 30 30 30 30 min. 12.51 7.98 10.1 7.6 8.19 19.99 max. 25.5 15.22 17.17 14.92 16.9 38.0 mean 17.63 10.93 12.8 10.43 12.18 28.59 standard deviation 3.52 1.77 1.92 1.76 2.26 5.03 CV 20.0 16.2 15.0 16.9 18.6 17.6 Pan paniscus n 29 30 30 — — — min. 9.5 5.88 7.88 max. 16.13 9.13 11.88 mean 12.27 7.13 9.66 standard deviation 2.02 0.88 1.16 CV 16.5 12.3 12.0 Homo sapiens n 36 36 36 36 36 36 min. 7.67 5.45 5.94 3.7 5.82 10.94 max. 12.65 7.9 9.13 6.58 8.99 20.2 mean 9.99 6.72 7.67 5.36 7.49 15.96 standard deviation 1.23 0.64 0.71 0.61 0.68 2.18 CV 12.3 9.5 9.3 11.4 9.1 13.6 Australopithecus afarensis n 6 101199 3 min. 10.9 6.9 9.3 6.4 8.8 20.91 max. 17 10.6 13.9 9.5 13.1 24.29 mean 13.12 8.48 10.88 7.82 10.54 22.77 standard deviation 2.22 1.25 1.38 1.08 1.31 1.72 CV 16.9 14.7 12.7 13.8 12.4 7.5 Australopithecus anamensis n 3 77983 min. 10 6.6 9 5.9 8.2 20.2 max. 15.71 10.40 13.90 10.3 13.81 31.79 mean 13.27 8.81 11.04 7.99 10.44 26.86 standard deviation 2.94 1.32 1.66 1.43 1.72 5.99 CV 22.2 15.0 15.0 17.9 16.5 22.3 not reduce concomitantly with crowns. Crown height canines), male canine size—especially crown height— also did not reduce in order to enhance an incisal reduced to the size of those of female extant apes, or biting function (Szalay 1975; Wolpoff 1980; as has occurred in other primates (Plavcan et al. Greenﬁeld 1992) because shape change in the 1995). However, canine shape becomes altered crown did not accompany crown height reduction. simultaneously with mandibular lateral incisor breadth Following a loss of function of the canine teeth as (Ward et al. 2001) and premolar form only with the weapons (for behavioural reasons, or following appearance of Au. afarensis in the absence of further masticatory changes precluding the use of projecting crown height reduction.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3338 C. V. Ward et al. Dental evolution in Australopithecus

a) 45( Table 2. Probabilities from t-tests for significant differences between Au. anamensis and Au. afarensis canine crown areas 40 and linear dimensions using ln-transformed data. Numbers 35 are for two-tailed probabilities. Area is calculated as the length times the breadth of the crown dimensions. Abbreviations as 30 in table 1; mand, mandibular; max, maxillary. 25 20 area MD BL RMD RBL height 15 max 0.546 0.007 0.226 0.009 0.625 0.211 crown height (mm) 10 mand 0.482 0.208 0.98 0.831 0.843 0.971 5 0 any event, it is now clear that crowns and roots did (b) 25 not change shape and size as part of a unimodal selection pressure that drove the canines to the modern human form. Rather, the patterns of morphological 20 change suggest to us that the selective pressure shaping canine form during the evolution of Au. anamensis and 15 early Au. afarensis was distinct from that of the earliest hominins, and of later Homo. This function almost certainly related to food acquisition or processing, but in 10 a manner distinctive to early Australopithecus. crown MD (mm) 5 3. MANDIBULAR AND MAXILLARY MORPHOLOGY 0 Other morphologies distinguishing Au. anamensis and (c) 25 Au. afarensis also are related to the change in canine tooth size, and in morphology of the canine/premolar complex. In particular, canine tooth root size affects 20 the occlusal outline of the anterolateral corner of the mandible. The mandible of Au. anamensis is distinct 15 from that of Au. afarensis in having an inflated alveolar profile along the roots, so that the canines are set anteriorly to the postcanine tooth rows (figure 4)(Ward 10 et al. 2001). In the male mandible, the effect of a crown BL (mm) large root is particularly notable. In contrast, in 5 Au. afarensis, the broadest region across the anterior mandible is found adjacent to P3, and the canines 0 are set medial to the premolars. There also is less variation in this contour among mandibles, presumably related to less canine root size dimorphism than in G. gorilla H. sapiens Au. anamensis. Certainly, canine size is correlated P. pygmaeus A. afarensisA. anamesis P. troglodytes with mandibular form in primates (Plavcan & Daegling 2006). Another factor influencing the Figure 2. Dental dimensions for maxillary canine crowns for relatively broad anterior portion of the mandible in extant great apes, humans, Au. anamensis and Au. afarensis. Au. anamensis is that the lower lateral incisors are Data in table 3.(a) Crown height, (b) crown mesiodistal, and (c) crown buccolingual. relatively broader than in Au. afarensis (Ward et al. 2001). Both canine root breadths and incisor breadth would affect anterior mandibular size and shape. The maxilla of Au. anamensis, and early Au. afarensis It could be possible that smaller canine roots in from Laetoli (Garusi 1; Puech 1986; Puech et al. 1986), Au. afarensis could be related to decreased loading of appears to have narrowly spaced, relatively straight the canines in puncturing or crushing (Spencer maxillary tooth rows, also seen in the Woranso-Mille 2003), but microwear studies in Au. afarensis imply sample (Haile-Selassie et al.2010). They also have that, in fact, use of the canine for these activities prob- rounded margins of the lateral nasal aperture. Both of ably was greater in Au. afarensis than in apes (Ryan & these features are plausibly related to reduction in Johanson 1989). Comparisons with Au. anamensis canine tooth root size. Canine root length may not be microwear will be necessary to explore this possibility related to maxillary shape (Cobb & Willis 2008; further. Plavcan et al. 2009), but root basal area would certainly To date, the fossil record is insufficient to evaluate affect maxillary breadth in this region and thus the whether these events in canine and premolar evolution supporting bone. were indeed simultaneous, but they are so within the Thus, the selective force that shaped canine root current resolution available in the fossil record. In size reduction is plausibly linked to pressures that

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Dental evolution in Australopithecus C. V. Ward et al. 3339

2.9

2.5

2.1 ln (crown mesiodistal length) (mm) ln (crown 1.7 1.7 1.9 2.1 2.3 2.5 2.7 2.9 3.1 ln (crown buccolingual breadth) (mm)

Au. afarensis (b) 1.6

1.4

1.2

1.0

0.8 crown mesiodistal/buccolingual crown 0.6 p < 0.05 p < 0.05 mandibular maxillary

canine P3 canine P3

Figure 3. Illustrations of canine shape differences between Au. anamensis and Au. afarensis.(a) Scatterplot of ln-transformed maxillary canine mesiodistal length compared with buccolingual breadth. Open squares: Gorilla gorilla, Pan paniscus, P. troglodytes, Pongo pygmaeus; open triangles: Homo sapiens; grey diamonds: Au. afarensis; black circles: Au. anamensis. Australopithecus anamensis retains relatively long canines mesiodistally and are most similar in proportions to extant apes. Australopithecus afarensis canines are similar buccolingually but are mesiodistally shorter than those of Au. anamensis. Humans have the same proportions as seen in Au. afarensis, but are smaller overall. (b) Basal proportion differences are seen only in the maxillary canine and mandibular premolar, the teeth that hone, but not mandibular canine or maxillary premolar, illustrating that observed shape changes are associated with a further reduction in honing and a shift in occlusal relationships in this complex. (c) Morphologic differences in canines and P3. Australopithecus anamensis has a lower mesial crown shoulder and longer mesial crest in the maxillary canine, a narrower, more blade-like mandibular crown with pronounced distal tubercle and a more unicuspid P3 with centrally placed paraconid compared with Au. afarensis. Data in tables 1 and 3. Scale bar, 1 cm. altered mandibular and possibly maxillary geometry. (Hylander 1984, 1985; Ravosa 2000). Australopithe- Teaford & Ungar (2000) noted that mandibular cus anamensis had a correspondingly large post- corpus robusticity is intermediate in Au. anamensis incisive planum and strongly developed mandibular between that of great apes and later hominins, tori, probably related to this overall geometry. A suggesting an increase in adaptation to resist heavier wider geometry in Au. afarensis would reduce masticatory stresses with Au. afarensis. That Au. forces from wishboning owing to pull of the external afarensis was adapted to greater masticatory stresses masticatory muscles and bone (Hylander 1985). is also suggested by the increased height of its More divergent tooth rows also decrease symphyseal molar crowns (Leakey et al. 1995; Ward et al. torsional stresses. It is notable, therefore, that 1999b, 2001). Australopithecus afarensis mandibles despite altered mandibular geometry, symphyseal also tend to have more posteriorly divergent tooth robusticity still tends to be relatively greater in Au. rows than does Au. anamensis, whose tooth rows afarensis than Au. anamensis. are narrower and more parallel, more like those of Given the effects of mandibular geometry on sym- extant apes (Ward et al. 2001). Narrow tooth rows physeal stresses, it may be that selection for more increase symphyseal stresses owing to wishboning divergent tooth rows inﬂuenced the reduction of lateral and torsion of the mandible during mastication incisor breadth and canine root size in order to reduce

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3340 C. V. Ward et al. Dental evolution in Australopithecus

Table 3. Descriptive statistics for maxillary canine crown Thus, the mandibular morphology of Au. afarensis dimensions of extant combined-sex and fossil samples. implies selection for the ability to process harder- Abbreviations as in table 1. to-chew foods, possibly opening up new niches. However, it is not only the masticatory system that height BL MD has changed; reduction in lateral incisor breadth and reshaping of the canine crowns and the canine– Gorilla gorilla n 50 50 50 premolar complex also suggest that selection for min. 12.14 10.26 12.54 altered function of the anterior dentition in food pro- max. 41.56 22.12 23.23 cessing occurred in the transition from Au. anamensis mean 22.08 13.88 17.17 to Au. afarensis. standard deviation 7.44 2.78 3.24 CV 33.7 20.0 18.9 Pongo pygmaeus 4. TOOTH WEAR n 18 18 18 One feature not previously appreciated from published min. 12.74 8.09 11.11 fossils is that for those specimens showing substantial max. 30.45 15.43 18.56 tooth wear, there appears to be differentially heavy mean 20.15 11.80 14.76 anterior tooth wear in Au. anamensis compared with standard deviation 6.15 2.19 2.72 Au. afarensis. Quantitative comparisons of gross wear CV 30.5 18.6 18.4 patterns are difficult owing to the fragmentary preser- Pan troglodytes vation of the dentitions, but qualitative comparisons n 30 30 30 can be made. Overall, Au. anamensis appear to have min. 11.86 8.09 9.63 higher frequencies of heavier tooth wear than seen in max. 27.27 14.25 19.31 Au. afarensis. mean 18.05 10.56 12.82 Three out of four known Au. anamensis maxillae standard deviation 4.03 1.79 2.31 that preserve molars and anterior teeth all have heavy CV 22.3 17.0 18.0 anterior wear relative to that of the molars (figure 5). Pan paniscus KNM-KP 29283 has dentine exposure crossing both n 24 30 30 lingual cusps of M1 and M2. Its incisors and canines min. 8.63 6.13 8.63 preserve only a narrow band of enamel labially, but max. 20.38 11.13 14.38 were wearing onto the roots lingually. The new speci- mean 13.40 8.01 10.54 standard deviation 3.48 1.44 1.56 men, KNM-KP 47952 (Manthi et al. in preparation), CV 26.0 18.0 14.8 also has unusually high anterior tooth wear, with only 1–2 mm of enamel remaining along the lingual surfaces Homo sapiens of its incisors and canines. In apparent contrast, dentine n 50 50 50 is only exposed on M2 as a tiny pit on the paracone. min. 6.48 6.56 5.96 max. 12.26 9.82 8.49 Even if this molar is not associated, which it almost cer- mean 9.28 8.24 7.45 tainly is, there is an unusually heavy amount of anterior standard deviation 1.23 0.71 0.57 wear. Another Kanapoi fossil, KNM-KP 30498, has 2 1 CV 13.3 8.6 7.7 M preserved, but on M has a small area of dentine 1 Australopithecus afarensis exposed only on the paracone. Its I is worn all the n 888way up to the basal tubercle, probably about halfway min. 9.2 9.3 8.9 through the original length of the tooth. The canine max. 15.4 12.5 11.6 of this same specimen is worn almost up to its mesial mean 12.35 10.7 9.81 or distal tubercles. In fact, no unworn incisors are standard deviation 2.18 1.00 0.84 known from Au. anamensis at all, except those of CV 17.7 9.3 8.6 young individuals whose teeth are either not yet or Australopithecus anamensis barely in occlusion, and/or who exhibit little or no n 378molar wear. The only relatively unworn maxilla with min. 12 8.8 9.91 canine is ASI-VP-2/344 from Aramis, which has no max. 16 11.2 12.4 dentine exposure on M2 but still exhibits apical wear mean 14.4 10.20 11.10 on its canine (White et al. 2006). This specimen standard deviation 2.12 0.75 0.82 appears comparable in wear to teeth in the Au. afarensis CV 14.7 7.4 7.4 maxilla AL 200-1. In contrast, no comparably heavy differential wear is found in the associated maxillary dentitions of Au. afarensis at Hadar or Laetoli, and none is as heavily the breadth of the anterior mandible in Au. afarensis. worn as any of the three Kanapoi specimens. The M2s This may have co-occurred with widening of the pos- of AL 444 (Kimbel et al. 2004) are more worn than terior part of the mandible, too. In order to maintain those of KNM-KP 47952, but less than those of appropriate occlusal relationships, this could also KNM-KP 29283, but most of the incisor and have led to concomitant reduction in maxillary canine crowns are intact in AL 444. AL 199-1 and canine root dimensions, and corresponding reduction AL 200-1 have less wear on their molars than any in maxillary inflation along the canine juga and lateral Au. anamensis maxilla, and while they have some nasal aperture. wear on the incisors and canines, it is not heavy. The

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Dental evolution in Australopithecus C. V. Ward et al. 3341

Figure 4. Top row photos: occlusal views of all three Kanapoi mandibles, from left to right KNM-KP 29281, KNM-KP 29287, KNM-KP 31713. Bottom row line drawings: several mandibles of Au. afarensis, from left to right: LH 4, AL 123-23, AL 333w-60, AL 266-1, AL 400-1a, AL 277-1 and AL 198-1. Arrows denote anterolateral inﬂection of occlusal outline. Note the canine roots set medial to that of P3 in the Au. afarensis, and in contrast that the canines are set anterior to the P3 in the Au. anamensis fossils so that the anterolateral corner of the occlusal proﬁle is formed by the canine juga in this earlier species. Scalar bar, 0–4 cm.

KNM-KP 30498 KNM-KP 47952 KNM-KP 29283

Figure 5. Lingual views (top) of anterior teeth of KNM-KP 30498 and KNM-KP 47952 and occlusal views (bottom) of these anterior teeth and their associated molars. KNM-KP 29283 shown in medial (top) and occlusal (bottom) views for comparison. KNM-KP 30498 preserves I1,C,P3,M1 and M3 crowns, relevant teeth reversed to show all as if they were from the left side. KNM-KP 47952 preserves I1,I2, C and M2. All three specimens show relatively heavy anterior wear relative to molar wear. See text for discussion. Scalar bar, 0–4 cm.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3342 C. V. Ward et al. Dental evolution in Australopithecus most worn published Au. afarensis incisor is AL 198- 5. SUMMARY AND CONCLUSIONS 17a (Johanson et al. 1982) which is comparable to The discovery of new fossils, even though representing that of KNM-KP 30498. Unfortunately it is not only a small portion of the anatomy of Au. anamensis, associated with any postcanine teeth, so further dictates a more careful, circumspect view of the role of comparison cannot be made. this taxon in hominin evolution, and thereby the pat- Mandibular tooth wear is not directly comparable tern of the origin of the adaptive suite of behaviours to maxillary wear, but even in sufficiently preserved and characters shaping the early evolution of the Aus- mandibular dentitions, anterior tooth wear is at least tralopithecus–human clade. Australopithecus anamensis as great or greater on the teeth relative to the molars documents a morphology in the anterior face and den- in Au. anamensis when compared with Au. afarensis. tition that is clearly transitional between a more The Au. afarensis mandible with the most heavily primitive hominin form, and that seen in Au. afarensis. worn molars, AL 198-1, has dentine exposed across Given that the fossil record consists of mainly the occlusal face of M1 and buccal cusps of M2, but teeth and jaws, it should come as no surprise that still has most of its canine crown preserved. It is only the evidence suggests that any adaptive shift from slightly less worn anteriorly than the Au. anamensis Au. anamensis–afarensis lineage was related to diet. type mandible KNM-KP 29281. The most heavily The data from the new Kanapoi fossils, in combi- worn mandibular dentition of all is the Au. anamensis nation with previously published data, demonstrate specimen FJ-4-SB-1a from Fejej, Ethiopia (Fleagle that adaptively significant differences exist between et al. 1991), which has a similar level of molar wear Au. anamensis and Au. afarensis. These morphologies to AL 198-1, yet it has dentine exposure over almost are not isolated, but seem to reflect an adaptive shift the entire P3 cusp and the associated canine is to a diet involving heavier mastication and at the almost completely worn to the root, preserving only same time altered use of the anterior dentition in a narrow band of enamel. food processing. In summary, no anterior teeth are known from Taken together, the greatest known differences Hadar in which the entire crown is missing, yet between Au. anamensis and Au. afarensis are associated many specimens attributed to Au. anamensis are very with evolutionary changes within the canine/P3 com- heavily worn. All individuals with sufficiently heavy plex, and with adaptations for coping with increasing molar wear to expose dentine on M2 have very heavily masticatory loads on the postcanine dentition. It has worn anterior teeth in Au. anamensis, whereas this is long been supposed that reduction in canine crown not the case for Au. afarensis. Only expanding sample height accompanied selection for an increased ability sizes will provide an adequate test of how typical this to masticate tougher or harder foods, as well as with ori- distinction is, but current fossils are suggestive. gins of habitual terrestrial bipedality. Ardipithecus There could be three possible explanations for this, ramidus demonstrates that crown height reduction is which are not mutually exclusive, and all hint at selec- not linked to increased ability to masticate tougher or tion for altered involvement of the anterior dentition. harder foods (White et al.1994; Suwa et al. 2009), The first possibility is that the anterior permanent den- and Au. anamensis demonstrates that shape change tition erupts earlier relative to the molars in altering occlusal relationships between the canine and Au. anamensis than Au. afarensis, and that there was premolar, and reduction in canine crowns and roots a shift to delay eruption of the incisors and canines were dissociated. Even though the canine/P3 complex in Au. afarensis relative to molar development. The changed form in the Au. anamensis/Au. afarensis lineage, second would be ingesting or biting foods with canine crown size itself remained stable, while the den- higher levels of tannins, which might increase intra- tition and mandible showed progressive changes that oral friction and cause higher tooth wear (Prinz & suggest adaptation to heavy loads. Lucas 2000). The third possibility is that there is a The dissociation between changes in root and difference between these samples in patterns of food crown size is distinctive in the evolution in Au. ana- processing involving the anterior dentition in which mensis–afarensis. Given that Ardipithecus also has the teeth are suffering greater mechanical abrasion large roots relative to its crowns (Suwa et al. 2009), (Teaford & Ungar 2000). Under any of these scen- the Au. anamensis condition appears primitive for arios, anterior tooth use or dietary properties hominins. A reduction in root size is achieved with probably would have differed between Au. afarensis the appearance of Au. afarensis, a species in which and Au. anamensis. the premolars are more molariform, lower lateral inci- We suggest that the chemical hypothesis does not sors less broad and maxillary canine crowns are provide the most satisfactory explanation because rela- mesiodistally shorter with concomitant shorter mesial tive anterior wear appears to decrease in concert with crests, mandibular canines less blade-like and more shape changes in incisor breadth, canine length and symmetrical in profile. Tooth rows are less parallel crown shape, as well as premolar proportions and and anterolateral mandibular and maxillary contours crown morphology. The combination of changes in less inflated, probably related to the presence of smal- both wear gradient and dental morphology hints at a ler canine roots. An association between root size mechanical factor. Detailed study of anterior tooth reduction and a shift to a more functionally microwear and dental growth patterns are needed to advantageous jaw morphology is worth investigating. help test the various hypotheses of altered tooth wear Furthermore, the new Kanapoi fossils highlight the between these species. Regardless, no matter what nature of Au. anamensis as a truly transitional species the explanation, the pattern suggests a shift in diet or between a more primitive condition to what is seen anterior tooth use of some sort. in Au. afarensis, and to some extent later hominins.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Dental evolution in Australopithecus C. V. Ward et al. 3343

Australopithecus anamensis was not just a primitive ver- Haile-Selassie, Y. 2001 Late Miocene hominids from the sion of Au. afarensis, it was the species at the root of the Middle Awash, Ethiopia. Nature 412, 187–191. (doi:10. Australopithecus–human clade in which some key 1038/35084063) aspects of Australopithecus morphology were develop- Haile-Selassie, Y. 2004 Late Miocene teeth from Middle ing (see also Haile-Selassie et al. 2010). At the same Awash, Ethiopia, and early hominid dental evolution. Science 303, 1503–1505. (doi:10.1126/science.1092978) time, not all of the characteristics seen in Au. afarensis Haile-Selassie, Y. 2010 Phylogeny of early Australopithecus: were present at the origin of the Australopithecus– new fossil evidence from the Woranso-Mille (central human clade, so not all distinguish members of this Afar, Ethiopia). Phil. Trans. R. Soc. B 365, 3323–3331. clade from its sister taxa. (doi:10.1098/rstb.2010.0064) Whether apparent dietary evolution co-occurred with Haile-Selassie, Y., Saylor, B. Z., Deino, A., Alene, M. & shifts in locomotor or manipulative adaptations, body Latimer, B. M. 2010 New hominid fossils from size, dimorphism, cranial morphology or brain size Woranso-Mille (Central Afar, Ethiopia) and taxonomy during the early evolution of the Australopithecus– of early Australopithecus. Am. J. Phys. Anthropol. 141, human clade can only be elucidated with more fossils 406–417. (doi:10.1002/ajpa.21159) from the time of first occurrence of Australopithecus Heinrich, R. E., Rose, M. D., Leakey, R. E. & Walker, A. C. (4.17 Ma) and Au. afarensis from Hadar (3.4–3.0 Ma). 1993 Hominid radius from the middle Pliocene of Lake Turkana, Kenya. Am. J. Phys. Anthropol. 92, 139–148. We hypothesize that Au. anamensis is best viewed as (doi:10.1002/ajpa.1330920203) not simply a primitive precursor to Au. afarensis,but Hylander, W. L. 1984 Stress and strain in the mandibular rather part of a dynamic morphological transition from symphysis of primates: a test of competing hypotheses. a primitive, ape-like morphology to the unique set of Am. J. Phys. Anthropol. 64, 1–46. (doi:10.1002/ajpa. morphological adaptations and behaviours that charac- 1330640102) terized the australopithecine bauplan and the early Hylander, W. L. 1985 Mandibular function and biomechani- evolution of the Australopithecus–human clade. cal stress and scaling. Am. Zool. 25, 315–330. (doi:10. 1093/icb/25.2.315) We thank the National Museums of Kenya, Emma Mbua, Johanson, D., White, T. & Coppens, Y. 1982 Dental remains Robert Moru, the Royal Museum of Central Africa, from the Hadar Formation, Ethiopia: 1974–1977 Cleveland Museum of Natural History and United States collections. Am. J. Phys. Anthropol. 57, 545–604. National Museum for access to specimens. We thank Chris (doi:10.1002/ajpa.1330570406) Dean, Bill Kimbel, Meave Leakey, Faydre Paulus, Matt Jungers, W. 1978 On canine reduction in early hominids. Ravosa, Peter Ungar and Bernard Wood for assistance, Curr. Anthropol. 51, 155–156. advice and helpful discussions. We thank Alan Walker and Kimbel, W. H., Rak, Y. & Johanson, D. 2004 The skull of Chris Stringer for generously inviting us to participate in this symposium. We thank the Wenner Gren Foundation, Australopithecus afarensis. New York, NY: Oxford Leakey Foundation, Turkana Basin Institute and National University Press. Science Foundation for support in various aspects of this Kimbel,W.,Lockwood,C.,Ward,C.V.,Leakey,M.,Rak,Y.& project. Johanson,D.2006WasAustralopithecus anamensis ancestral to A. afarensis? A case of anagenesis in the hominin fossil record. J. Hum. Evol. 51,134–152.(doi:10.1016/j.jhevol. 2006.02.003) REFERENCES Leakey, M. G., Feibel, C. S., McDougall, I. & Walker, A. Bailit, H. & Friedlaender, J. 1966 Tooth size reduction: a 1995 New four-million-year-old hominid species from hominid trend. Am. Anthropol. 68, 665–672. (doi:10. Kanapoi and Alia Bay, Kenya. Nature 376, 565–571. 1525/aa.1966.68.3.02a00030) (doi:10.1038/376565a0) Beard, K. C., Teaford, M. F. & Walker, A. 1986 New wrist Leakey,M.G.,Feibel,C.S.,MacDougall,I.,Ward,C.V.& bones of Proconsul africanus and P. nyanzae from Rusinga Walker, A. 1998 New specimens and confirmation of an Island, Kenya. Folia Primatol. 47, 97–118. (doi:10.1159/ early age for Australopithecus anamensis. Nature 363,62–66. 000156268) Leutenegger, W. & Kelly, J. T. 1977 Relationship of sexual Brace, C. 1963 Structural reduction in evolution. Am. Nat. dimorphism in canine size and body size to social, behav- 97, 39–49. (doi:10.1086/282252) ioral, and ecological correlates in anthropoid primates. Brace, C. L. 1971 Sexual dimorphism in human evolution. Primates 18, 117–136. (doi:10.1007/BF02382954) Nature 228, 31–49. Lovejoy, C. O. 1981 The origin of man. Science 211, Brunet, M. et al. 2002 A new hominid from the Upper 341–350. (doi:10.1126/science.211.4480.341) Miocene of Chad, Central Africa. Nature 418, 145–151. Lovejoy, C. 2009 Reexamining human origins in light of (doi:10.1038/nature00879) Ardipithecus ramidus. Science 326, 74e1–74e8. (doi:10. Calcagno, J. & Gibson, K. 1988 Human dental reduction: 1126/science.1175834) natural selection or probable mutation effect? Am.J.Phys. Lovejoy, C., Simpson, S., White, T., Asfaw, B. & Suwa, G. Anthropol. 77, 505–517. (doi:10.1002/ajpa.1330770411) 2009 Careful climbing in the Miocene: the forelimbs of Cobb, S. & Willis, A. 2008 Is premaxilla morphology deter- Ardipithecus ramidus and humans are primitive. Science mined by the spatial requirements of the developing 326, 70e1–70e8. (doi:10.1126/science.1175827) incisor dentition? Am. J. Phys. Anthropol. 135, 79–80. Manthi, F., Plavcan, J. & Ward, C. V. In preparation. New Darwin, C. 1871 The descent of man and selection in relation to fossils attributed to Australopithecus anamensis from sex. London, UK: John Murray. Kanapoi, Kenya. Fleagle, J. G., Rasmussen, D. T., Yirga, S., Bown, T. M. & Patterson, B. & Howells, W. W. 1967 Hominid humeral Grine, F. E. 1991 New hominid fossils from Fejej, fragment from early Pleistocene of Northwestern Southern Ethiopia. J. Hum. Evol. 21, 145–152. (doi:10. Kenya. Science 156, 64–66. (doi:10.1126/science.156. 1016/0047-2484(91)90005-G) 3771.64) Greenfield, L. O. 1992 Origin of the human canine: a new Plavcan, J. M. 1990 Sexual dimorphism in the dentition of solution to an old enigma. Yrbk Phys. Anthropol. 35, extant anthropoid primates. PhD dissertation, Duke 153–185. (doi:10.1002/ajpa.1330350607) University. University Microfilms: Ann Arbor, MI.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3344 C. V. Ward et al. Dental evolution in Australopithecus

Plavcan, J. & Daegling, D. 2006 Interspecific and intra- Teaford, M. F. & Ungar, P. S. 2000 Diet and the evolution of specific relationships between tooth size and jaw size in the earliest human ancestors. Proc. Natl Acad. Sci. USA primates. J. Hum. Evol. 51, 171–184. (doi:10.1016/j. 97, 13 506–13 511. (doi:10.1073/pnas.260368897) jhevol.2006.02.005) Ward,C.V.,Leakey,M.G.,Brown,B.,Brown,F.,Harris,J.& Plavcan, J. M. & Van Schaik, C. P.1997 Intrasexual competition Walker, A. 1999a South Turkwel: a new Pliocene hominid and body weight dimorphism in anthropoid primates. Am. J. site in Kenya. J. Hum. Evol. 36, 69–95. (doi:10.1006/jhev. Phys. Anthropol. 103,37–68.(doi:10.1002/(SICI)1096- 1998.0262) 8644(199705)103:1,37::AID-AJPA4.3.0.CO;2-A) Ward, C. V., Walker, A. & Leakey, M. G. 1999b The Plavcan, J. M., Van Schaik, C. & Kappeler, P. 1995 Compe- new hominid species Australopithecus anamensis. Evol. tition, coalitions and canine size in primates. J. Hum. Anthropol. 7, 197–205. (doi:10.1002/(SICI)1520- Evol. 28, 245–276. (doi:10.1006/jhev.1995.1019) 6505(1999)7:6,197::AID-EVAN4.3.0.CO;2-T) Plavcan, J. M., Ward, C. V. & Paulus, F. 2009 Estimating Ward, C. V., Leakey, M. & Walker, A. 2001 Morphology of canine tooth crown height in early Australopithecus. Australopithecus anamensis from Kanapoi and Allia Bay, J. Hum. Evol. 57, 2–10. (doi:10.1016/j.jhevol.2009.04. Kenya. J. Hum. Evol. 41, 255–368. (doi:10.1006/jhev. 005) 2001.0507) Prinz, J. & Lucas, P. 2000 Saliva tannin interactions. J. Oral White, T. D., Suwa, G. & Asfaw, B. 1994 Australopithecus Rehabil. 27, 991–994. (doi:10.1046/j.1365-2842.2000. ramidis, a new species of early hominid from Aramis, 00578.x) Ethiopia. Nature 371, 306–312. (doi:10.1038/371306a0) Puech, F. 1986 Australopithecus afarensis Garusi 1, diversite´ White, T. D., Suwa, G. & Asfaw, B. 1995 Corrigendum. et spećialisation des premiers Hominide´s d’apre`s les Nature 375, 88. caracte`res maxillo-dentaires. C. R. Acad. Sci., Paris 303, White, T., Suwa, G., Simpson, S. & Asfaw, B. 2000 Jaws 1819–1823. and teeth of Australopithecus afarensis from Maka, Puech, P. F., Cianfarani, F. & Roth, H. 1986 Reconstruction Middle Awash, Ethiopia. Am. J. Phys. Anthropol. 111, of the maxillary dental arcade of Garusi Hominid 1. 45–68. (doi:10.1002/(SICI)1096-8644(200001)111: J. Hum. Evol. 15, 325–332. (doi:10.1016/S0047- 1,45::AID-AJPA4.3.0.CO;2-I) 2484(86)80015-X) White, T. et al. 2006 Asa Issie, Aramis and the origin of Ravosa, M. J. 2000 Size and scaling in the mandible of living Australopithecus. Nature 440, 883–889. (doi:10.1038/ and extinct apes. Folia Primatol. 71, 305–322. (doi:10. nature04629) 1159/000021754) White, T., Asfaw, B., Beyene, Y., Haile-Selassie, Y., Lovejoy, Ryan, A. & Johanson, D. 1989 Anterior dental microwear C., Suwa, G. & WoldeGabriel, G. 2009 Ardipithecus in Australopithecus afarensis: comparisons with human ramidus and the paleobiology of early hominids. Science and nonhuman primates. J. Hum. Evol. 18, 235–268. 326, 75–86. (doi:10.1126/science.1175802) (doi:10.1016/0047-2484(89)90051-1) Wolpoff, M. 1969 The effect of mutations under conditions Spencer, M. A. 2003 Tooth-root form and function in of reduced selection. Soc. Biol. 16, 11–23. platyrrhine seed-eaters. Am. J. Phys. Anthropol. 122, Wolpoff, M. H. 1978 Some aspects of canine size in the 325–335. (doi:10.1002/ajpa.10288) australopithecines. J. Hum. Evol. 7, 115–126. (doi:10. Suwa, G., Kono, R., Simpson, S., Asfaw, B., Lovejoy, C. & 1016/S0047-2484(78)80003-7) White, T. 2009 Paleobiological implications of the Wolpoff, M. H. 1979 Anterior dental cutting in the Laetolil Ardipithecus ramidus dentition. Science 326, 94–99. hominids and the evolution of the bicuspid P3. Am. J. (doi:10.1126/science.1175824) Phys. Anthropol. 51, 233–234. (doi:10.1002/ajpa. Szalay, F. S. 1975 Hunting-scavenging protohominids: a 1330510210) model for hominid origins. Man 10, 420–429. (doi:10. Wolpoff, M. 1980 Paleoanthropology. New York, NY: 2307/2799811) Knopf.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Phil. Trans. R. Soc. B (2010) 365, 3345–3354 doi:10.1098/rstb.2010.0033

Molar microwear textures and the diets of Australopithecus anamensis and Australopithecus afarensis Peter S. Ungar1,*, Robert S. Scott2, Frederick E. Grine3 and Mark F. Teaford4 1Department of Anthropology, University of Arkansas, Old Main 330, Fayetteville, AR 72701, USA 2Department of Anthropology, Rutgers University, New Brunswick, NJ, USA 3Departments of Anthropology and Anatomical Sciences, Stony Brook University, Stony Brook, NY, USA 4Center for Functional Anatomy and Evolution, Johns Hopkins University, Baltimore, MD, USA Many researchers have suggested that Australopithecus anamensis and Australopithecus afarensis were among the earliest hominins to have diets that included hard, brittle items. Here we examine dental microwear textures of these hominins for evidence of this. The molars of three Au. anamensis and 19 Au. afarensis specimens examined preserve unobscured antemortem microwear. Microwear textures of these individuals closely resemble those of Paranthropus boisei, having lower complexity values than Australopithecus africanus and especially Paranthropus robustus. The microwear texture complexity values for Au. anamensis and Au. afarensis are similar to those of the grass-eating Theropithecus gelada and folivorous Alouatta palliata and Trachypithecus cristatus. This implies that these Au. anamensis and Au. afarensis individuals did not have diets dominated by hard, brittle foods shortly before their deaths. On the other hand, microwear texture anisotropy values for these taxa are lower on average than those of Theropithecus, Alouatta or Trachypithecus. This suggests that the fossil taxa did not have diets dominated by tough foods either, or if they did that directions of tooth–tooth movement were less constrained than in higher cusped and sharper crested extant primate grass eaters and folivores. Keywords: Australopithecus; molar; diet; microwear textures

1. INTRODUCTION (a) Background Researchers have recognized for more than three Australopithecus anamensis and Au. afarensis have decades that patterns of microscopic use wear on traditionally been argued to be the earliest hominins teeth hold the potential to provide information about to show an adaptive shift from diets dominated by the diets of early hominins (e.g. Grine 1977, 1981, soft, sugary forest fruits to hard and brittle or abrasive 1986; Puech 1979; Ryan 1980; Walker 1981; Ungar foods (Ward et al. 1999; Teaford & Ungar 2000; White et al. 2006). Many studies of hominin dental micro- et al. 2000; Wood & Richmond 2000; Walker 2002; wear have been published over the past thirty years, Macho et al. 2005). Australopithecus afarensis, for including feature-based quantitative analyses of example, has long been noted to have thickly enam- molar occlusal surface microwear in Australopithecus elled, large and ﬂat crowned cheek teeth and afarensis1 and Australopithecus anamensis (Grine et al. robustly constructed crania and mandibles, at least 2006a,b). Data presented here extend this work by when compared with our nearest living relatives, the comparing the microwear textures of these species chimpanzees (e.g. McHenry 1984; Hylander 1988; with those of other early hominins and recent primate White et al. 2000). This enhanced craniodental toolkit taxa with known diets. Results indicate that sampled has led workers to suggest that ‘nuts, seeds, and hard Au. anamensis and Au. afarensis individuals tend to fruit may have been an important component to the have relatively simple microwear surface textures vary- diet of this species’ (Wood & Richmond 2000). ing in degree of anisotropy. This pattern is comparable These hominins have also been thought to have had to that previously reported for Paranthropus boisei dietary adaptations intermediate between those of (Ungar et al. 2008), but differs from that of Au. africa- frugivorous forest apes and later hominins. White nus, and especially of Paranthropus robustus (Scott et al. et al. (2000), for example, considered them to have 2005; see also Grine 1986). taken the ‘initial functional steps that would eventually culminate in the far more derived, specialized masticatory apparatus of later hominid species’ such as Au. africanus and especially Par. boisei and * Author for correspondence ([email protected]). Par. robustus.AsWhite et al. (1981) noted, while One contribution of 14 to a Discussion Meeting Issue ‘The ﬁrst four Au. africanus showed several craniodental features million years of human evolution’. foreshadowing the functional specializations seen in

3345 This journal is q 2010 The Royal Society Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3346 P. S. Ungar et al. Pliocene hominin microwear textures

Paranthropus, Au. afarensis retained at least some more such as the grey-cheeked mangabey Lophocebus primitive traits, such as relatively larger front teeth and albigena and the brown capuchin, Cebus apella. less swollen and inflated cusps on their postcanines. Further, the microwear pattern for these hominins is Picq (1990) further suggested that the height and remarkably homogeneous between specimens across thickness of the Au. afarensis mandibular corpus, and both time and paleoecological context. Grine et al. robusticity of the symphysis, were intermediate (2006a,b) suggest that these results highlight the between those of chimpanzees on the one hand and difference between ‘faculty’ and ‘biological role’ those of Au. africanus and especially Paranthropus (Bock & von Wahlert 1965) or dietary potential and species on the other. He argued the same for the what an animal eats on a day-to-day basis. As a height of the ramus and size of the mandibular result, Grine et al.(2006a,b) suggested that the shift condyle. in diet-related adaptive morphology in Au. anamensis Picq (1990) used these lines of evidence to build a and Au. afarensis may relate more to occasional but scenario in which Au. afarensis was still dependent on critical fallback food exploitation than to preferred the fruits of the forest, but seasonally sought sustenance resources. in more open settings with ‘tougher foods containing We might predict, if Au. africanus and especially abrasive structures’ and ‘nuts protected by skins or a Par. boisei and Par. robustus show further craniodental hard shell’. Ungar & Teaford (2001) called this a specializations for the consumption of hard, brittle ‘mixed forest–savanna resource adaptation’. Ungar foods, that samples of these species should have (2004) further noted that the degree of differences in more individuals showing high levels of microwear pit- occlusal slope and relief between Au. afarensis and ting or surface complexity with fewer fine, parallel chimpanzees are as expected for differences in ‘fallback’ striations and with texture anisotropy, compared with foods, suggesting that the early hominins may have pre- Au. afarensis and especially Au. anamensis. ferred soft, sugar-rich fruits, but had the ability to make Here we present the first microwear texture analysis more effective use of hard, brittle resources as seasonal of Au. afarensis and Au. anamensis for comparison with availabilities required. Au. africanus, Par. boisei and Par. robustus. Microwear Following its initial description, Au. anamensis texture analysis has proved to provide a three- quickly took on the role of ‘intermediate form’ in dimensional characterization of microwear surfaces terms of dietary adaptation between the earlier Ardi- free from the need to identify and measure individual pithecus and Au. afarensis. Like Au. afarensis, Au. features (e.g. Ungar et al. 2003, 2007a,b, 2008; Scott anamensis had thicker post-canine tooth enamel and et al. 2005, 2006, 2009a; El-Zaatari 2008; Krueger larger average cheek teeth than Ardipithecus, again et al. 2008; Ungar & Scott 2009; Krueger & Ungar suggesting a dietary shift towards harder foods or in press). While microwear texture analysis and con- more abrasive ones (Ward et al. 1999; Teaford & ventional feature-based analyses to date have yielded Ungar 2000; Wood & Richmond 2000; Walker 2002; similar results for other hominins (compare Grine Suwa et al. 2009). Still, in comparison with Au. afaren- (1986) with Scott et al. (2005) and Ungar et al. sis, Au. anamensis molars were ‘not very high-crowned’ (2006) with Ungar & Scott (2009)), texture analyses (Walker 2002), and the larger values for enamel thick- are particularly well suited for between-studies com- ness in Au. anamensis when compared with Ar. ramidus parisons, because data collected are free from may reflect, at least in part, increased tooth size in observer measurement error. Further, microwear tex- the former (Suwa et al. 2009, supporting online ture results for Au. africanus, Par. robustus and Par. material). Further, Au. anamensis lacked changes in boisei are available in the literature for comparison the geometry of the mandible and maxilla seen in (Scott et al. 2005; Ungar et al. 2008). Indeed, only Au. afarensis, Au. africanus and especially Paranthropus texture data are available for Par. boisei. (Ward et al. 1999), such as greater mandibular corpus Microwear texture analysis studies indicate that robusticity, which might buttress the jaw against higher hard-object feeding extant primates (and other mam- peak force magnitudes or repetitive loading in mastica- mals) tend to have higher average levels of texture tion (Teaford & Ungar 2000). These differences led complexity and lower levels of anisotropy on their Teaford & Ungar (2000) to speculate that hard and molar occlusal surfaces than seen in tough food perhaps abrasive foods may have become even more eaters (see Ungar et al. 2007b for a review). Here we important components of the diet of Au. afarensis. test the hypothesis that interpretations of craniodental Walker (2002) suggested that changes from functional morphology described above are reflected in Au. anamensis to Au. afarensis were carried to extremes microwear texture patterns. If so, Au. afarensis and in Au. africanus and especially Paranthropus. especially Au. anamensis specimens should show less The microwear evidence paints a somewhat more surface complexity and more anisotropy than those complex picture. One might predict heavy pitting of Au. africanus, and especially Par. boisei and associated with hard-object consumption in Au. ana- Par. robustus. mensis and Au. afarensis and an increase in pitting from the earlier to the later species. Contrary to expectation though, both show fairly fine features and 2. MATERIAL AND METHODS microwear surfaces dominated by striations rather Dental microwear texture data are presented here for than pits (Grine et al. 2006a,b; Suwa et al. 2009, the molar teeth of Au. anamensis from Kanapoi and supporting online material). This pattern is more Allia Bay in Kenya and Au. afarensis from the Laetolil similar to that seen in the tough food folivore Beds in Tanzania and the Hadar Formation in Ethio- Gorilla gorilla beringei than in hard-object feeders, pia. Specimens included in this study are the same as

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Pliocene hominin microwear textures P. S. Ungar et al. 3347

Table 1. Microwear texture analysis data for Au. anamensis and Au. afarensis. specimen Asfc epLsar1.8 Smc HAsfc9 HAsfc81

Au. anamensis KNM-ER 35236 0.97466 0.00307 0.15015 0.58516 0.71286 KNM-KP 29287 0.80791 0.00299 0.20908 0.34424 0.58869 KNM-KP 34725 1.30999 0.00244 0.50835 0.38439 0.83670 Au. afarensis AL 128-23 0.64728 0.00122 0.70776 0.30052 0.37230 AL 145-35 0.87280 0.00337 0.26841 0.45188 0.56429 AL 188-1 0.53380 0.00390 0.34143 0.29913 0.33684 AL 200-1b 0.70342 0.00572 0.21016 0.40769 0.57363 AL 225-8 1.06196 0.00594 0.26836 0.61132 0.99634 AL 288-1i 0.68804 0.00184 1.20858 0.34728 0.49713 AL 333-74 0.53915 0.00602 0.50798 0.24058 0.33415 AL 333w-12 1.16543 0.00374 0.27184 0.45849 0.50411 AL 333w-1a 0.71150 0.00553 0.59957 0.32997 0.37838 AL 333w-57 0.69913 0.00382 0.34153 0.26405 0.31046 AL 333w-59 0.19077 0.00285 0.94135 0.34335 0.38715 AL 333w-60 0.77945 0.00515 0.41635 0.31759 0.36561 AL 366-1 0.85189 0.00090 0.59981 0.35013 0.42788 AL 400-1a 0.72866 0.00223 0.41634 0.26509 0.38790 AL 486-1 0.97104 0.00139 0.15015 0.38916 0.44262 AL 487-1c 0.46932 0.00236 0.26690 0.36286 0.39707 LH4 0.54767 0.00163 2.40210 0.27204 0.29177 LH 15 1.07480 0.00419 0.20835 0.72182 0.87299 LH 22 0.82274 0.00319 0.59962 0.25273 0.31400

those employed in feature-based SEM microwear work envelope of 276 mm 204 mm. The feature- analyses reported by Grine et al. (2006a,b). based study of these specimens reported in Grine All permanent molars available to us were first et al.(2006a,b) used a pixel resolution of 0.25 mm examined to determine the suitability for microwear and combined sample area of 0.04 mm2. analysis. Those that preserved wear facets were Resulting point clouds were analysed using TOOTH- cleaned with cotton swabs soaked in alcohol or acetone Frax and SFRAX scale-sensitive fractal analysis (SSFA) to remove adherent dirt or preservatives. Moulds of software (Surfract Corp.). SSFA as applied to micro- occlusal surfaces were then made using President Jet wear research is described in detail elsewhere (e.g. Regular Body vinyl dental impression material Scott et al. 2006). The basic premise is that surface (Colte`ne-Whaledent Corp.), and casts were produced texture varies with scale of observation, and that this from these molds using Epotek 301 epoxy and variation can be used to characterize functionally rel- hardener (Epoxy Technologies Inc.). evant aspects of microwear. SSFA texture variables Replicas were examined by light microscopy and included in this study are area-scale fractal complexity SEM as necessary to determine the suitability for (Asfc), anisotropy (epLsar), scale of maximum com- microwear analysis following the criteria described by plexity (Smc) and heterogeneity of complexity Teaford (1988). Most specimens had occlusal surfaces (HAsfc). Values for individual specimens are reported obscured by taphonomic damage and so had to be as medians of the four fields sampled for each excluded from this study. In the end, the molars of specimen. only three of the Au. anamensis specimens and 19 Area-scale fractal complexity is a measure of change Au. afarensis individuals available to us were found to in roughness with scale. The faster a measured surface preserve unobscured antemortem occlusal surface area increases with resolution, the more complex the microwear. A list of all specimens considered can be surface. Anisotropy is characterized as variation in found in Grine et al.(2006a,b) and those included in lengths of transect lines measured at a given scale this study are presented in table 1. (we use 1.8 mm) with orientations sampled at 58 inter- All specimens included in this study were analysed vals across a surface. An anisotropic surface will have using a Sensofar Plm white-light confocal profiler shorter transects in the direction of the surface pattern (Solarius, Inc.) with an integrated vertical scanning than perpendicular to it (e.g. a transect that cross-cuts interferometer. Three-dimensional point clouds were parallel scratches must trace the peaks and valleys of collected for ‘phase II’ facets (the buccal occlusal sur- each individual feature). Thus, a heavily pitted surface faces of lower molars and the lingual occlusal surface typically has high Asfc and low epLsar values, whereas of uppers) using a 100 long working distance objec- one dominated by homogeneous, parallel striations has tive. The point clouds sampled elevations at intervals low Asfc and high epLsar values. Other variables used of 0.18 mm along the x- and y-axes, with a vertical res- to characterize microwear surface texture include olution of 0.005 mm. Data were obtained for four Smc, the scale range over which Asfc is calculated, adjacent fields on facets 9 or 10n, for a combined and HAsfc, variation of Asfc across a surface (in this

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3348 P. S. Ungar et al. Pliocene hominin microwear textures

Table 2. Summary statistics for early hominins.

Au. anamensis Au. afarensis Au. africanus Par. boisei Par. robustus n 3191079 Asfc mean 1.031 0.740 1.522 0.625 3.543 s.d. 0.256 0.236 0.387 0.268 1.449 epLsar mean 0.003 0.003 0.004 0.003 0.002 s.d. 0.000 0.002 0.002 0.002 0.001 Smc mean 0.289 0.565 1.834 0.516 0.216 s.d. 0.192 0.521 4.256 0.269 0.053 HAsfc9 mean 0.438 0.368 0.617 0.460 1.054 s.d. 0.129 0.124 0.259 0.136 0.564 HAsfc81 mean 0.713 0.461 1.004 0.621 2.101 s.d. 0.124 0.187 0.367 0.232 1.026

case, each field of view was divided into a 3 3grid¼ series of extant primates with known differences in HAsfc9, and a 9 9grid¼ HAsfc81). High Smc values feeding behaviors to put the microwear texture analysis should correspond to more complex coarse features. results for these hominins in the context of modern High HAsfc values are observed for surfaces that vary primate diets. The baseline series includes: (i) the in complexity across a facet. mantled howler, Alouatta palliata and the silvered leaf Statistical analyses focused on comparisons of micro- monkey, Trachypithecus cristatus, C. apella and wear textures of Au. afarensis with those of Au. africanus L. albigena from the dataset reported by Scott et al. and Par. robustus from South Africa as reported in Scott (2005); (ii) G. gorilla beringei, G. g. gorilla, Pan et al. (2005) and with Par. boisei from eastern Africa as troglodytes and the orangutan Pongo pygmaeus from reported by Ungar et al. (2008). Australopithecus anamen- Ungar et al. (2007b); and (iii) the yellow baboon sis data could not be compared statistically with those of Papio cynocephalus and the gelada baboon Theropithecus the other hominins given an available sample of only gelada from Scott et al. (2009b). The mantled howler three specimens, although newly recovered specimens and silvered leaf monkey are typically characterized (e.g. Haile-Selassie, this volume) hold the potential for as folivores, whereas the brown capuchin and grey- larger microwear datasets in the future. First, differences cheeked mangabey are considered to be hard-object in central tendencies between taxa were assessed using a fallback feeders. Among the great apes, chimpanzees MANOVA performed on ranked data (Conover & Iman are the most frugivorous, and gorillas, especially the 1981)forallvariables(Asfc, epLsar, Smc, HAsfc9 and G. g. beringei sample considered here (the Fossey col- HAsfc81). Individual ANOVAs and multiple comparisons lection at the US National Museum of Natural tests were used to determine the sources of significant History), are the most folivorous (see references in variation. Both Tukey’s honestly significant difference Ungar et al. 2007b). Orangutans are intermediate in (HSD) and Fisher’s least significant difference (LSD) their diets. Finally, geladas are specialized grass tests were used to balance risks of type I and type II eaters, whereas yellow baboons have a more catholic errors (Cook & Farewell 1996). Values of p , 0.05 for diet including fruits, leaves and animal prey (see Post Tukey’sHSDtestsmaybeassignedsignificancewith 1982; Norton et al. 1987; Dunbar 1988; Altmann some confidence, whereas values of p , 0.05 on Fisher’s 1998; Pochron 2000; Bentley-Condit 2009). LSD but not Tukey’s LSD tests are considered suggestive but of marginal significance. Degree of variance in microwear textures within 3. RESULTS taxa may be as important for distinguishing species Results for Au. anamensis and Au. afarensis are presented as differences in central tendencies, especially given in tables 1–4 andareillustratedinfigures1 and 2. differences in foraging and feeding strategies between primates. With this in mind, raw data for each variable (a) Comparisons with other fossil hominins were transformed for Levene’s test following the pro- The early hominins are well-separated from one another cedure described by Plavcan & Cope (2001) to by microwear texture complexity. First, the specimens compare distribution variances between taxa. A from South Africa have higher Asfc values on average MANOVA was used to assess the variation in variance than those from eastern Africa, regardless of the species between taxa, and as with the comparisons of central considered. Within the eastern African sample, Au. ana- tendencies, ANOVAs and multiple comparisons tests mensis may have slightly higher complexity on average were used to determine the sources of significant than Au. afarensis or Par. boisei, but larger samples of variation as needed. Au. anamensis are really needed to evaluate this (only Complexity and anisotropy results for Au. anamensis one Au. anamensis specimen is outside the range of and Au. afarensis were also compared with those for a Au. afarensis or Par. boisei). No significant variation in

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Pliocene hominin microwear textures P. S. Ungar et al. 3349

Table 3. Analyses of hominin microwear texture data (central tendencies). All data rank transformed to mitigate violation of assumptions inherent to parametric statistics (Conover & Iman 1981). *p , 0.05 for Fisher’s LSD test, **p , 0.05 for both Tukey’s HSD and Fisher’s LSD tests (shown in italic).

value F d.f. p-value multivariate test results Wilks’s l 0.037 205.783 539 0.000 Pillai trace 0.963 205.783 539 0.000 Hotelling–Lawley trace 26.382 205.783 539 0.000

ANOVAtest results Asfc epLsar Smc HAsfc9 HAsfc81 F 32.397 2.735 4.056 12.096 23.204 d.f. 443 443 443 443 443 p-value 0.000 0.041 0.007 0.000 0.000 paired comparisons Asfc epLsar Smc HAsfc9 HAsfc81 Au. afarensis Au. africanus 219.279** 26.037 21.084 215.421** 219.616** Au. afarensis Au. anamensis 29.579* 1.596 12.649 26.421 212.982* Au. afarensis Par. boisei 4.564 2.977 22.613 28.421 28.887* Au. afarensis Par. robustus 227.357** 13.263* 16.705** 227.088** 229.982** Au. africanus Au. anamensis 9.700 7.633 13.733 9.000 6.633 Au. africanus Par. boisei 23.843** 9.014 21.529 7.000 10.729* Au. africanus Par. robustus 28.078* 19.300** 17.789** 211.667* 210.367* Au. anamensis Par. boisei 14.143* 1.381 215.262 22.000 4.095 Au. anamensis Par. robustus 217.778** 11.667 4.056 220.667** 217.000** Par. boisei Par. robustus 231.921** 10.286 19.317** 218.667** 221.095**

Table 4. Analyses of hominin microwear texture data (variance). Microwear data transformed for Levene’s test (X 0 ¼ jX— mean (X)j) following Plavcan & Cope (2001). *p , 0.05 for Fisher’s LSD test, **p , 0.05 for both Tukey’s HSD and Fisher’s LSD tests (shown in italic).

value F d.f. p-value multivariate test results Wilks’s l 0.127 5.627 20 130 0.000 Pillai trace 1.254 3.837 20 168 0.000 Hotelling–Lawley trace 4.174 7.827 20 150 0.000

ANOVAtest results Asfc epLsar Smc HAsfc9 HAsfc81 F 24.38 3.828 3.822 4.476 10.296 d.f. 443 443 443 443 443 p-value 0.000 0.010 0.010 0.004 0.000 paired comparisons Asfc epLsar Smc HAsfc9 HAsfc81 Au. afarensis Au. africanus 20.140 0.000 22.087** 20.118 20.157 Au. afarensis Au. anamensis 20.009 0.001* 0.181 20.010 0.048 Au. afarensis Par. boisei 20.035 0.000 0.090 20.023 20.072 Au. afarensis Par. robustus 21.078** 0.001** 0.282 20.305** 20.68** Au. africanus Au. anamensis 0.131 0.001 2.268* 0.107 0.205 Au. africanus Par. bosei 0.105 0.000 2.177* 0.095 0.084 Au. africanus Par. robustus 20.938** 0.001* 2.368** 20.187* 20.523** Au. anamensis Par. boisei 20.026 20.001* 20.091 20.012 20.120 Au. anamensis Par. robustus 21.069** 0.000 0.101 20.294* 20.728** Par. boisei Par. robustus 21.043** 0.001** 0.191 20.282** 20.608** complexity is noted between Au. afarensis and Par. boisei. or Par. boisei, though no other significant differences in Within the South African sample, Par. robustus has more Smc central tendencies are noted. In addition, Au. afri- complex microwear surfaces on average than Au. africa- canus has significantly greater variation in Smc values nus.ThePar. robustus sample also shows significantly than Au. afarensis, Par. robustus and (marginally) both greater variation in its complexity values than do any Au. anamensis and Par. boisei. It should be noted of the other taxa. All other taxa have similar levels of though that the high Au. africanus variance is driven within-species variation in complexity. by a single outlier with an extremely high Smc value. The species do not differ as much in anisotropy as The Par. robustus sample also has higher average they do in complexity, though Par. robustus has both HAsfc values than Au. anamensis, Au. afarensis, Par. a lower average epLsar and less variability in this vari- boisei or (at least marginally) Au. africanus. In addition, able than either Au. africanus or Au. afarensis. The Au. africanus has higher HAsfc values than Au. afaren- Par. robustus sample also has a lower average scale of sis. Further, Par. boisei has marginally lower HAsfc maximum complexity than Au. afarensis, Au. africanus values than Au. africanus, but marginally higher

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3350 P. S. Ungar et al. Pliocene hominin microwear textures

in anisotropy than either Au. afarensis or Par. boisei,as (a) well as marginally higher heterogeneity than Au. afarensis. These results will be of limited interpretability until larger samples are available for study.

(b) Comparisons with extant primates Microwear complexity and anisotropy results for Au. anamensis and Au. afarensis are illustrated alongside data for living primates in ﬁgure 3. These hominins are on the low end for both Asfc and epLsar, both in their means and in their variations, when considered in light of the extant primate baseline. The distributions and central tendencies of complexity values for these hominins are most comparable to those (b) reported for A. palliata, T. gelada and T. cristatus. They lack the degrees of dispersion in Asfc seen in the other primates, especially C. apella , P. cynocephalus and L. albigena. In contrast, the anisotropy values for the hominins are most different from those for A. palliata, T. gelada and T. cristatus and are more similar to those of the other primates, especially P. troglodytes and P. cynocephalus.

4. DISCUSSION (a) Microwear and the diets of Au. anamensis and Au. afarensis Figure 1. Photosimulations of molar microwear surfaces of Dental microwear texture analysis results suggest that (a) Au. anamensis and (b) Au. afarensis generated from neither the Au. anamensis nor the Au. afarensis individ- point cloud data collected using the white-light confocal uals included in this study had diets dominated by profiler. Each image represents a surface 208 mm 280 mm. hard, brittle foods in the days, weeks or perhaps even months prior to their deaths. While these species have been suggested to show an adaptive shift from heterogeneity (at least for HAsfc81) than Au. afarensis. diets dominated by soft forest fruits to hard, brittle Finally, Par. robustus has more variation in its HAsfc foods (e.g. Ward et al. 1999; Teaford & Ungar 2000; values than Au. anamensis, Au. afarensis, Au. africanus White et al. 2000; Wood & Richmond 2000; Walker and Par. boisei. 2002), none of the specimens examined exhibit the In summary, Au. afarensis microwear textures are high microwear surface texture complexity expected most similar to those of Par. boisei, with the only differ- of a hard-object feeder. The distribution of Asfc ence being a marginally higher average value for one of values more closely resembles those of the grass- the heterogeneity measures. Australopithecus afarensis eating T. gelada and the folivores A. palliata and also has lower complexity on average and less hetero- T. cristatus than hard-object feeding L. albigena or geneity than Au. africanus, as well as less variation in C. apella. scale of maximum complexity (though the latter The anisotropy results for Au. anamensis and result is driven largely by a single outlier). Australo- Au. afarensis are, on the other hand, most different pithecus afarensis differs most markedly from Par. from those of A. palliata, T. cristatus and T. gelada robustus. Australopithecus afarensis has lower complexity among the baseline series. These three extant primates and heterogeneity of complexity, higher scale of maxi- have higher average anisotropy values than the early mum complexity, and marginally higher anisotropy. hominins. High anisotropy is often taken as a proxy The degree of variation in anisotropy values is also for tough food consumption and repetitive chew greater for Au. afarensis, though the variation in com- cycles with opposing teeth moving past one another plexity and heterogeneity are greater in Par. robustus. along constrained paths. At first glance, this might In fact, Par. robustus is very much the ‘outlier’ com- suggest that Au. anamensis and Au. afarensis specimens pared with the other taxa in most cases. This sampled also avoided tough foods in the period prior hominin tends towards a greater spread in and larger to death, although the combination of low anisotropy average values for complexity, scale of maximum com- and low complexity averages in Au. anamensis and plexity and heterogeneity of complexity, though less Au. afarensis is unusual for primates. Most extant pri- spread in, and lower values for anisotropy. mate samples published to date have either high Results for Au. anamensis aremoredifficulttohave anisotropy averages combined with low complexity confidence in, given its sample size, though these do sep- values, associated with the consumption of tough arate clearly from Par. robustus.Therearesome foods, or low anisotropy combined with high complexity additional ‘hints’ suggested by the data if the patterns averages consistent with a hard-brittle item diet. hold with larger samples. Australopithecus anamensis has We propose that Au. anamensis and Au. afarensis may marginally higher average complexity and lower variation have indeed consumed tough foods, but that their

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Pliocene hominin microwear textures P. S. Ungar et al. 3351

(a) 0.008

) 0.007 1.8 0.006 0.005 0.004 0.003 0.002

anisotropy ( epLsar anisotropy 0.001

0 123456 complexity (Asfc) (b) 4.5

) 4.0 81 3.5 3.0 2.5 2.0 1.5 1.0 0.5 heterogeneity ( HAsfc

0 246810 12 14 16 scale of maximal complexity (Smc) Figure 2. Plots of (a) anisotropy versus complexity and (b) heterogeneity of complexity versus scale of maximum complexity (below) for early hominin individuals considered by taxon. Open triangle, Au. afarensis; ﬁlled triangle, Au. africanus; open circle, Par. boisei; ﬁlled circle, Par. robustus; diamond, Au. anamensis.

0.01 A. palliata G. g. gorilla 0.005 0

0.01 T. gelada G. gorilla beringei 0.005 0

0.01 T. cristatus P. troglodytes 0.005 0

0.01 Au. afarensis L. albigena 0.005 0 0.01 Au. anamensis P. cynocephalus 0.005 0

0.01 P. pygmaeus C. apella 0.005

0 5 10 15 20 0 5 10 15 20 Figure 3. Plots of anisotropy versus complexity for various extant primates, compared with data for Au. anamensis and Au. afarensis.

anisotropy values are low because their dentognathic Kay & Hiiemae 1974),whereinfooditemsaremilled morphology did not limit occlusal movements to the between opposing surfaces. The combination of ﬂat degree presumed for primates with steeper occlusal sur- teeth and tough foods would be expected to result in a faces, such as A. palliata, T. c r i s t a t u s and T. g e l a d a .The combination of low complexity and low anisotropy. ﬂatter teeth of early hominins offer fewer constraints to Still, the relationship between microwear feature aniso- masticatory movements during food fracture and may tropy and occlusal topographic relief remains to be therefore allow ‘grinding’ action (sensu Simpson 1933; investigated fully.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3352 P. S. Ungar et al. Pliocene hominin microwear textures

(b) Comparisons with other early hominins six cranial characters that Au. afarensis shares with Studies of craniodental functional morphology have one or more species of Paranthropus, although they suggested to several researchers a tendency towards regarded them, like the pattern of intracranial venous increasing specialization for hard objects from Au. drainage, as being homoplastic. Most recently, simi- anamensis to Au. afarensis, Au. africanus and finally larities in ramal morphology between Au. afarensis Par. robustus and Par. boisei (see White et al. 1981, and Par. robustus mandibles have been observed by 2000; Ward et al. 1999; Teaford & Ungar 2000; Rak et al . (2007), who interpreted them as synapomor- Wood & Richmond 2000; Walker 2002). Australopithe- phies, suggesting that Au. afarensis and Par. robustus are cus africanus evinces a higher average Asfc than Au. united in a single clade and that this possibly includes afarensis, and Par. robustus has an even higher average Par. aethiopicus and Par. boisei, although no fossils Asfc value. The same is true for heterogeneity and vari- attributable to the latter two species preserve the relation in the scale of maximum complexity (though the evant anatomy. The evidence suggesting an latter result appears to be driven by a single outlier). ancestor–descendant relationship between Au. afaren- These findings are all consistent with an increasing sis and Paranthropus, and particularly that for an Au. component of hard, brittle items in the diet of Au. afri- afarensis–Par. aethiopicus–Par. boisei lineage in eastern canus compared with Au. afarensis and Par. robustus Africa, is not wholly inconsistent with their hypoth- compared with Au. africanus. esized cladistic relationships (Strait et al. 1997; On the other hand, the distribution of Asfc values Kimbel et al. 2004; Strait & Grine 2004). for Par. boisei is very similar to that of Au. afarensis. Similarities in microwear texture results for Au. Further, while the sample for Au. anamensis is too anamensis, Au. afarensis and Par. boisei may make small for a reasonable comparison with other taxa, sense if these taxa comprise an anagenetic lineage its mean Asfc value and heterogeneity are, if anything, and all shared food-type preferences. One possible slightly higher than that of Au. afarensis. Thus, if high scenario might be increasing craniodental specializ- texture complexity is considered to be a proxy for the ations through the lineage for repetitive loading given consumption of hard, brittle items, there is no evi- consumption of tough foods in the face of morphologi- dence for an increase in the role of such foods in the cal constraints imposed by relatively flat cheek teeth. diet from Au. anamensis to Au. afarensis to Par. boisei. This would be consistent with a C4 isotope signature Ungar et al. (2008) remarked on the apparent in Par. boisei (van der Merwe et al. 2008) and the con- discordance between microwear in Par. boisei and bio- sumption of tough grasses or sedges, assuming that the mechanical models for this species based on specimens examined thus far are representative. craniodental functional morphology. They proposed Carbon stable isotope studies of Au. anamensis, Au. that Par. boisei may represent a hominin example of afarensis and additional Par. boisei specimens would Liem’s Paradox, wherein craniodental specializations provide a valuable test of this hypothesis. developed as an adaptation to processing less pre- On the other hand, if Par. boisei and Par. robustus ferred, mechanically challenging foods, even though form a clade excluding Au. africanus, the two early the microwear suggests that these hominins rarely con- hominins from South Africa probably independently sumed such foods. This does not, however, explain increased their consumption of hard, brittle foods as either the microwear differences between Par. boisei evidenced by increased pit percentages in both homi- and Par. robustus (given similarities in their gnathoden- nins. In any case, the microwear texture patterns of tal adaptations) or the similarities between Par. boisei, the eastern African early hominins are more similar Au. anamensis and Au. afarensis (given differences in to one another than to those of the South African their gnathodental adaptations). early hominins, independent of whether one is The story becomes even more complicated when we considering Australopithecus or Paranthropus. consider these taxa in their presumed phylogenetic So in the end, what can be said of the microwear of contexts, especially the purported anagenetic lineage Au. anamensis and Au. afarensis? We may reasonably leading from Au. anamensis to Au. afarensis, Par. aetho- infer that specimens examined for this study did not picus and finally Par. boisei (Kimbel et al. 2006; Rak have a diet dominated by hard and brittle foods, at et al. 2007). Many workers have suggested an ances- least shortly before death. Picq (1990) proposed that tor–descendant relationship between Au. anamensis Au. afarensis often consumed soft foods that were not and Au. afarensis (Ward et al. 1999, 2001; White fracture resistant, but had craniomandibular adap- et al. 2000, 2006; Walker 2002). Such relationships tations for seasonal consumption of hard, brittle are consistent with numerical cladistic studies (Strait foods. Grine et al.(2006a,b) further suggested that tra- et al. 1997; Strait & Grine 2004) and with changes ditional microwear results for both Au. anamensis and in dentognathic characters among temporally succes- Au. afarensis are best explained by food preferences sive samples from Kanapoi, Allia Bay, Laetoli and for less mechanically challenging foods, though as Hadar (Kimbel et al. 2006). Ungar (2004) noted, their occlusal morphology Studies of cranial morphology have suggested that would have allowed the consumption of hard, brittle the pattern of intracranial venous sinus drainage is items in times of dietary stress when favored foods shared between Au. afarensis and Paranthropus species were unavailable. The microwear texture analysis for the exclusion of Au. africanus and Homo (Falk & data presented here cannot be used to falsify the Conroy 1983; Kimbel 1984; Falk 1988), and notion of rare hard-object feeding, but it also provides additional fossils have reinforced this similarity no evidence for it. Whether or not the craniodental (Kimbel et al. 2004; de Ruiter et al. 2006). Kimbel specializations seen in Au. anamensis and Au. afarensis et al. (2004) have, moreover, identified an additional are adaptations for the occasional consumption of

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Pliocene hominin microwear textures P. S. Ungar et al. 3353 hard, brittle foods, however, their microwear texture Grine, F. E., Ungar, P. S., Teaford, M. F. & El-Zaatari, S. patterns are consistent with the regular consumption 2006b Molar microwear in Praeanthropus afarensis: evi- of softer and or tougher items. dence for dietary stasis through time and under diverse paleoecological conditions. J. Hum. Evol. 51, 297–319. WearegratefultocuratorsattheNationalMuseumsof (doi:10.1016/j.jhevol.2006.04.004) Ethiopia, Kenya and Tanzania for permission to examine and Hylander, W. L. 1988 Implications of in vivo experiments for make moulds of specimens in their care and thank Alejandro interpreting the functional significance of ‘robust’ austra- Pere´z-Pere´z for his assistance with preparing these replicas. lopithecine jaws. In Evolutionary history of the ‘robust’ Kristin Krueger helped with the preparation of figure 1.This australopithecines (ed. F. E. Grine), pp. 55–83. study was funded by the US National Science Foundation. New York, NY: Aldine de Gruyter. Kay, R. F. & Hiiemae, K. M. 1974 Jaw movement and tooth use in recent and fossil primates. Am. J. Phys. Anthropol. ENDNOTE 40, 227–256. (doi:10.1002/ajpa.1330400210) 1One of us (F.E.G.) prefers the nomen Praeanthropus afarensis, as use Kimbel, W. H. 1984 Variation in the pattern of of the generic name Australopithecus for Au. anamensis and Au. afar- cranial venous sinuses and hominid phylogeny. ensis probably violates the criterion of monophyly (Grine & Strait Am. J. Phys. Anthropol. 63, 243–263. (doi:10.1002/ajpa. 2000). Because of the lack of consensus among the authors on the 1330630302) taxonomy of these hominins, we employ here the more commonly Kimbel, W. H., Rak, Y. & Johanson, D. C. 2004 The skull of used nomenclature. Australopithecus afarensis. New York, NY: Oxford Univer- sity Press. Kimbel, W. H., Lockwood, C. A., Ward, C. V., Leakey, REFERENCES M. G., Rak, Y. & Johanson, D. C. 2006 Was Australopithe- Altmann, S. A. 1998 Foraging for survival: yearling baboons in cus anamensis ancestral to A. afarensis? A case of Africa. Chicago, IL: University of Chicago Press. anagenesis in the hominin fossil record. J. hum. Evol. Bentley-Condit, V. K. 2009 Food choices and habitat use by 51, 134–152. (doi:10.1016/j.jhevol.2006.02.003) the Tana River yellow baboons (Pap. cynocephalus): a pre- Krueger, K. L. & Ungar, P. S. In press. Incisor microwear liminary report on five years of data. Am. J. Primatol. 71, textures of five bioarcheological groups. 432–436. (doi:10.1002/ajp.20670) Int. J. Osteoarcheol. (doi:10:1002/oz.1093) Bock, W. J. & von Wahlert, G. 1965 The role of ecological Krueger, K. L., Scott, J. R. & Ungar, P. S. 2008 Technical factors in the origin of higher levels of organization. note: dental microwear textures of ‘Phase I’ and ‘Phase II’ Syst. Zool. 14, 272–300. (doi:10.2307/2411681) facets. Am.J.Phys.Anthropol.137, 485–490. (doi:10. Conover, W. J. & Iman, R. L. 1981 Rank transformations as 1002/ajpa.20928) a bridge between parametric and nonparametric statistics. Macho, G. A., Shimizu, D., Jiang, Y. & Spears, L. R. 2005 Am. Stat. 35, 124–129. (doi:10.2307/2683975) Australopithecus anamensis: a finite element approach to Cook, R. J. & Farewell, V. T. 1996 Multiplicity consider- studying the functional adaptations of extinct hominins. ations in the design and analysis of clinical trials. Anat. Rec. 283A, 310–318. (doi:10.1002/ar.a.20175) J. R. Stat. Soc. 159, 93–110. McHenry, H. M. 1984 Relative cheek tooth size in de Ruiter, D. J., Steininger, C. M. & Berger, L. R. 2006 A Australopithecus. Am. J. Phys. Anthropol. 64, 297–306. cranial base of Australopithecus robustus from the hanging (doi:10.1002/ajpa.1330640312) remnant of Swartkrans, South Africa. Am. J. Phys. Norton, G. W., Rhine, R. J., Wynn, G. W. & Wynn, R. D. Anthropol. 130, 435–444. (doi:10.1002/ajpa.20386) 1987 Baboon diet: a five-year study of stability and varia- Dunbar, R. I. M. 1988 Primate social systems. Ithaca, NY: bility in the plant feeding and habitat of the yellow Cornell University Press. baboons (Pap. cynocephalus) of Mikumi National Park, El-Zaatari, S. 2008 Occlusal microwear texture analysis and Tanzania. Fol. Primatol. 48 , 78–120. (doi:10.1159/ the diets of historical/prehistoric hunter-gatherers. Int. 000156287) J. Osteoarchaeol 20, 67–87. (doi:10.1002/oa.1027) Picq, P. 1990 The diet of Australopithecus afarensis:an Falk, D. 1988 Enlarged occipital/marginal sinuses and emis- attempted reconstruction. C. R. Acad. Sci. Ser. II 311, sary foramina: their significance in hominid evolution. In 725–730. Evolutionary history of the ‘robust’ australopithecines (ed. F. Plavcan, J. M. & Cope, D. A. 2001 Metric variation and E. Grine), pp. 85–96. New York, NY: Aldine de Gruyter. species recognition in the fossil record. Evol. Anthropol. Falk, D. & Conroy, G. C. 1983 The cranial venous sinus 10, 204–222. (doi:10.1002/evan.20001) system in Australopithecus afarensis. Nature 306, 779– Pochron, S. T. 2000 The core dry-season diet of yellow 781. (doi:10.1038/306779a0) baboons (Papio hamadryas cynocephalus) in Ruaha Grine, F. E. 1977 Analysis of early hominid deciduous molar National Park, Tanzania. Fol. Primatol. 71, 346–349. wear by scanning electron microscopy: a preliminary (doi:10.1159/000021758) report. Proc. Electr. Microscop. Soc. S. Afr. 7, 157–158. Post, D. G. 1982 Feeding behavior of yellow baboons Grine, F. E. 1981 Trophic differences between ‘gracile’ and (Papio cyncocephalus) in the Amboseli National Park, ‘robust’ australopithecines: a scanning electron micro- Kenya. Int. J. Primatol. 3, 403–430. (doi:10.1007/ scope analysis of occlusal events. S. Afr. J. Sci. 77, BF02693741) 203–230. Puech, P. F. 1979 Diet of early man: evidence from abrasion Grine, F. E. 1986 Dental evidence for dietary differences in of teeth and tools. Curr. Anthropol. 20, 590–592. (doi:10. Australopithecus and Paranthropus: a quantitative analysis 1086/202335) of permanent molar microwear. J. Hum. Evol. 15, 783– Rak, Y., Ginzburg, A. & Geffen, E. 2007 Gorilla-like 822. (doi:10.1016/S0047-2484(86)80010-0) anatomy on Australopithecus afarensis mandibles suggests Grine, F. E. & Strait, D. S. 2000 The phylogenetic relation- Au. afarensis link to robust australopiths. Proc. Natl ships of recently described early hominid species. Acad. Sci. USA 104, 6568–6572. (doi:10.1073/pnas. Am. J. Phys. Anthropol. (Suppl. 30), 167. 0606454104) Grine, F. E., Ungar, P. S. & Teaford, M. F. 2006a Was the Ryan, A. S. 1980 Anterior dental microwear in hominid evol- Early Piocene hominin ‘Australopithecus’ anamensis a ution: comparisons with human and nonhuman primates. hard object feeder? S. Afr. J. Sci. 102, 301–310. PhD dissertation, University of Michigan.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3354 P. S. Ungar et al. Pliocene hominin microwear textures

Scott, R. S., Ungar, P. S., Bergstrom, T. S., Brown, C. A., scanning confocal microscopy and scale-sensitive fractal Grine, F. E., Teaford, M. F. & Walker, A. 2005 Dental analyses. Scanning 25, 185–193. (doi:10.1002/sca. microwear texture analysis reflects diets of living primates 4950250405) and fossil hominins. Nature 436, 693–695. (doi:10.1038/ Ungar, P. S., Grine, F. E., Teaford, M. F. & El Zaatari, S. nature03822) 2006 Dental microwear and diets of African early Homo. Scott, R. S., Ungar, P. S., Bergstrom, T. S., Brown, C. A., J. Hum. Evol. 50, 78–95. (doi:10.1016/j.jhevol.2005.08. Childs, B. E., Teaford, M. F. & Walker, A. 2006 Dental 007) microwear texture analysis: technical considerations. Ungar, P. S., Merceron, G. & Scott, R. S. 2007a J. Hum. Evol. 51, 339–349. (doi:10.1016/j.jhevol.2006. Dental microwear texture analysis of Varswater 04.006) bovids and Early Pliocene paleoenvironments of Lange- Scott, J. S., Godfrey, L. R., Jungers, W. L., Scott, R. S., baanweg, Western Cape Province, South Africa. Simons, E. L., Teaford, M. F., Ungar, P. S. & Walker, J. Mammal. Evol. 14, 163–181. (doi:10.1007/s10914- A. 2009a Dental microwear texture anlaysis of megalada- 007-9050-x) pids and archaeolemurids. J. Hum. Evol. 56, 405–416. Ungar, P. S., Scott, R. S., Scott, J. R. & Teaford, M. F. 2007b (doi:10.1016/j.jhevol.2008.11.003) Dental microwear analysis: historical perspectives and Scott, R. S., Teaford, M. F. & Ungar, P. S. 2009b Dietary new approaches. In Dental anthropology (eds J. D. Irish & diversity and dental microwear variability in Theropithecus G. C. Nelson), pp. 389–425. Cambridge, UK: gelada and Pap. cynocephalus. Am. J. Phys. Anthropol. Cambridge University Press. (Suppl. 48), 234. Ungar, P. S., Grine, F. E. & Teaford, M. F. 2008 Dental Simpson, G. G. 1933 Paleobiology of Jurassic mammals. microwear indicates that Paranthropus boisei was not a Paleobiology 5, 27–158. hard-object feeder. PLoS ONE 3,1–6. Strait, D. S. & Grine, F. E. 2004 Inferring hominoid and van der Merwe, N. J., Masao, F. T. & Bamford, M. K. 2008 early hominid phylogeny using craniodental characters: Isotopic evidence for contrasting diets of early hominins the role of fossil taxa. J. Hum. Evol. 47, 399–452. Homo habilis and Australopithecus boisei of Tanzania. (doi:10.1016/j.jhevol.2004.08.008) S. Afr. J. Sci. 104, 153–155. Strait, D. S., Grine, F. E. & Moniz, M. A. 1997 A reappraisal Walker, A. 1981 Diet and teeth. Dietary hypotheses of early hominid phylogeny. J. Hum. Evol. 32, 17–82. and human evolution. Phil. Trans. R. Soc. Lond. B 292, (doi:10.1006/jhev.1996.0097) 57–64. (doi:10.1098/rstb.1981.0013) Suwa, G., Kono, R. T., Simpson, S. W., Asfaw, B., Lovejoy, Walker, A. 2002 New perspectives on the hominids of the C. O. & White, T. D. 2009 Paleobiological implications of Turkana Basin, Kenya. Evol. Anthropol. 11(Suppl. 1), the Ardipithecus ramidus dentition. Science 326, 94–99. 38–41. (doi:10.1126/science.1175824) Ward, C. V., Leakey, M. G. & Walker, A. C. 1999 The new Teaford, M. F. 1988 Scanning electron microscope diagnosis hominid species, Australopithecus anamensis. Evol. Anthro- of wear patterns versus artifacts on fossil teeth. Scan. pol. 7, 197–205. (doi:10.1002/(SICI)1520-6505(1999) Microsc. 2, 1167–1175. 7:6,197::AID-EVAN4.3.0.CO;2-T) Teaford, M. F. & Ungar, P. S. 2000 Diet and the evolution Ward, C. V., Leakey, M. G. & Walker, A. 2001 Morphology of the earliest human ancestors. Proc. Natl. Acad. of Australopithecus anamensis from Kanapoi and Allia Bay, Sci. USA 97, 13 506–13 511. (doi:10.1073/pnas. Kenya. J. Hum. Evol. 41, 255–368. (doi:10.1006/jhev. 260368897) 2001.0507) Ungar, P. S. 2004 Dental topography and diets of Australo- White, T. D., Johanson, D. C. & Kimbel, W. H. 1981 Austra- pithecus afarensis and early Homo. J. Hum. Evol. 46, lopithecus africanus: its phyletic position reconsidered. 605–622. (doi:10.1016/j.jhevol.2004.03.004) S. Afr. J. Sci. 77, 445–470. Ungar, P. S. & Scott, R. S. 2009 Dental evidence for diets of White, T. D., Suwa, G., Simpson, S. & Asfaw, B. 2000 Jaws Early Homo.In The first humans: origins of the genus Homo and teeth of Australopithecus afarensis from Maka, Middle (eds F. E. Grine, R. E. Leakey & J. G. Fleagle), pp. 121– Awash, Ethiopia. Am. J. Phys. Anthropol. 111, 45–68. 134. New York, NY: Springer-Verlag. (doi:10.1002/(SICI)1096-8644(200001)111:1,45::AID Ungar, P. S. & Teaford, M. F. 2001 The dietary split -AJPA4.3.0.CO;2-I) between apes and the earliest human ancestors. In White, T. D. et al. 2006 Asa Issie, aramis and the origin of Humanity from African naissance to coming millennia Australopithecus. Nature 440, 883–889. (doi:10.1038/ (ed. P. V. Tobias et al.), pp. 337–354. Florence, Italy: nature04629) Firenza University Press. Wood, B. & Richmond, B. G. 2000 Human evolution: tax- Ungar, P. S., Brown, C. A., Bergstrom, T. S. & Walker, A. onomy and paleobiology. J. Anat. 197, 19–60. (doi:10. 2003 Quantification of dental microwear by tandem 1046/j.1469-7580.2000.19710019.x)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Phil. Trans. R. Soc. B (2010) 365, 3355–3363 doi:10.1098/rstb.2010.0086

An enlarged postcranial sample conﬁrms Australopithecus afarensis dimorphism was similar to modern humans Philip L. Reno1,*, Melanie A. McCollum3,4, Richard S. Meindl2,3 and C. Owen Lovejoy2,3,* 1Department of Developmental Biology, Stanford University School of Medicine, 279 Campus Drive, Beckman 300, Stanford, CA 94305-5329, USA 2Department of Anthropology, and 3School of Biomedical Sciences, Kent State University, Lowry Hall Room 226, Kent, OH 44242-0001, USA 4Department of Cell Biology, University of Virginia, Charlottesville, VA 22902, USA In a previous study, we introduced the template method as a means of enlarging the Australopithecus afarensis postcranial sample to more accurately estimate its skeletal dimorphism. Results indicated dimorphism to be largely comparable to that of Homo sapiens. Some have since argued that our results were biased by artiﬁcial homogeneity in our Au. afarensis sample. Here we report the results from inclusion of 12 additional, newly reported, specimens. The results are consistent with those of our original study and with the hypothesis that early hominid demographic success derived from a reproductive strategy involving male provisioning of pair-bonded females. Keywords: A.L. 333; taphonomy; monogamy; skeletal dimorphism; modelling

1. INTRODUCTION Any given assemblage of Au. afarensis fossils was Accurately inferring early hominid sexual dimorphism formed by a combination of random sampling and var- is an important element in interpreting their paleobiol- ious taphonomic processes. The effect of these ogy. We previously concluded that skeletal size processes on sample variation can be modelled by dimorphism in Australopithecus afarensis was signifi- bootstrapping from taxa of known dimorphism. cantly lower than that of gorillas and could not be Humans (Homo sapiens), chimpanzees (Pan troglodytes) statistically distinguished from that of modern and gorillas (Gorilla gorilla) represent the three genera humans (Reno et al. 2003, 2005). These findings, most closely related to early hominids and essentially which contrast with previous assessments (Zihlman & encompass the entire range of primate skeletal sexual Tobias 1985; McHenry 1991; Lockwood et al. 1996), dimorphism. Because the sex of any Au. afarensis were achieved through the use of the ‘template element is essentially unknown, sampling with regard method’. This method relied on the A.L. 288-1 partial to sex of extant taxa is allowed to vary freely. That is, skeleton (‘Lucy’), as a source of simple ratios between the sex ratio in each iteration is allowed to vary by femoral head diameter (FHD) and other skeletal simple probability (i.e. the binomial expansion). In dimensions. These ratios were then used to obtain sufficiently small samples, this can occasionally result estimates of FHD for skeletal dimensions that were in samples composed of only one sex. In order to also measurable in A.L. 288-1. Postcranial variation simulate the Au. afarensis assemblages as precisely as within the (thus maximized) Au. afarensis sample possible (and limit the variation introduced by from the Middle Awash region of Ethiopia (‘Com- sampling different anatomical locations), bootstrapped bined Afar’, CA) and that within the temporally and samples of living hominoids were required to exactly geographically constricted Au. afarensis sample from match the anatomical compositions of the A.L. 333 Afar Locality 333 were then compared with boot- and the CA samples (e.g. the number of proximal strapped samples of modern humans, chimpanzees tibias included in each bootstrapped sample was and gorillas. This method was specifically designed required to exactly match the number of proximal to overcome problems inherent in calculating sexual tibias represented in the Au. afarensis sample being dimorphism from a small number of specimens simulated). For each iteration, each postcranial whose sexes must be judged a priori on the basis of metric was converted to a FHD based on ratios calcu- size (e.g. Zihlman & Tobias 1985; McHenry 1991). lated from a template specimen that was also randomly chosen to serve as the equivalent of A.L. 288-1. A.L. 333 probably represents a simultaneous * Authors for correspondence ([email protected]; olovejoy@aol. death assemblage (White & Johanson 1989; com). Behrensmeyer et al. 2003). In our previous analysis One contribution of 14 to a Discussion Meeting Issue ‘The first four (Reno et al. 2003), two separate simulations of A.L. million years of human evolution’. 333 were generated. In one, each of 22 postcranial

3355 This journal is q 2010 The Royal Society Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3356 P. L. Reno et al. Au. afarensis skeletal dimorphism metrics preserved at the site was randomly drawn from observed range of the extant taxon being sampled. How- our complete samples of extant taxa (N 50). This ever, this correction is potentially quite conservative as exactly modelled A.L. 333 in being composed of as the relative size range of many metrics is actually many as 22 separate individuals. However, it is unli- greater than that of FHD (see below). kely that each A.L. 333 specimen in fact represents The three bootstrap simulations reported here were one of 22 different individuals. Based on mandibular performed separately to model the following enlarged dentitions, the minimum number of individuals Au. afarensis samples: (i) 26 specimens from the A.L. (MNI) at the site is nine (White & Johanson 1989). 333; (ii) 15 non-333 specimens from other Hadar Therefore, in a second simulation, we randomly localities and Maka; and (iii) 41 specimens in the selected nine individuals to serve as the source of all CA sample. The 15 specimens in the non-333 22 metrics. This ensures that many individuals are sample must represent 15 separate individuals. There- multiply represented in the sample of metrics. Our fore, 15 metrics were each randomly drawn from the procedures assume only that each ‘death’ assemblage entire chimpanzee, human or gorilla samples. In con- (fossil sample or extant simulation) was a random trast, it is unlikely that 26 different individuals sample of its parent population—the biological species contributed to the A.L. 333 sample. Therefore, for from which each was derived (i.e. Au. afarensis, each iteration, a separate random subsample of nine H. sapiens, P. troglodytes and G. gorilla). chimpanzee, human or gorilla individuals (based These samples have been challenged as not being upon an MNI from mandiblar dentitions (White & representative of the Au. afarensis size distribution Johanson 1989)) served as the pool from which 26 (Plavcan et al. 2005; Scott & Stroik 2006). The ration- metrics were then drawn. For the CA simulations, a ale has been that because ‘Lucy-sized’ individuals ‘hybrid’ was created for each iteration in which 26 are absent from the A.L. 333 assemblage, it must metrics representing A.L. 333 were sampled from over-sample large, presumably male, adult individuals. nine randomly selected individuals. These were com- If true, then our lower estimates of skeletal dimorph- bined with an additional 15 drawn from the entire ism in A.L. 333 may have been flawed by biased sample to represent non-333 individuals. sampling during the accumulation, fossilization and/ Plavcan et al. (2005) argued that only five to eight or recovery of the assemblage. individuals contributed to the A.L. 333 postcranial This argument is now subject to a simple test. sample, and it is certainly hypothetically possible that Additional postcranial elements of Au. afarensis have some of the (at least) nine known adult individuals now been reported for both A.L. 333 and other did not contribute to the postcranial sample. However, Middle Awash localities (Kimbel et al. 2004; Drapeau our simulations already permit sampling of fewer than et al. 2005; Harmon 2006). Inclusion of these nine individuals because not all nine individuals additional 12 specimens raises our postcranial selected for each iteration are necessarily randomly sample to 41 and more than doubles the number of sourced for the 26 metrics used to simulate individuals represented from non-A.L. 333 localities. A.L. 333. Therefore, it was unnecessary to perform This expanded sample provides an opportunity additional simulations from isolated comparative to more accurately assess size dimorphism in samples artificially restricted to less than nine potential Au. afarensis and determine whether smaller Lucy- contributors. sized individuals were disproportionately lacking Both the coefficient of variation (CV) and the bino- from A.L. 333. mial dimorphism index (BDI) were calculated in each simulation. The CV was calculated using the small sample correction (Sokal & Rohlf 1995). The BDI 2. MATERIAL AND METHODS was defined specifically for the calculation of sexual In addition to the 29 fossil specimens in our original dimorphism in samples of unknown sex. It rests study (Reno et al. 2003), we have now added four upon three assumptions: (i) both sexes are present in specimens from A.L. 333 and eight from other each sample; (ii) every specimen has an equal prob- Middle Awash localities (Kimbel et al. 2004; Drapeau ability of being male or female, but (iii) when any et al. 2005; Harmon 2006; table 1). Metrics from these two specimens are potentially of a different sex, the specimens (some as yet undescribed), as well as their larger is always male. Using this algorithm, a sample homologues from the A.L. 288-1 partial skeleton, of n yields a total of n 2 1 possible sex allocations were provided by William Kimbel. and therefore n 2 1 skeletal dimorphism estimates. Details of the template method and our bootstrap- The BDI is then the weighted average of the n 2 1 ping procedures are described in Reno et al.(2003, dimorphism values based on the probability of each 2005). Since those publications, we have observed sex allocation occurring under the binomial expansion. that the template method yields extreme FHD esti- Note that in light of the assumption that males are mates in rare cases where a small or large skeletal always larger than females, the BDI tends to overesti- metric is paired with a template specimen with an unu- mate dimorphism in minimally dimorphic species. sual metric to FHD ratio (all within the bounds of Two estimates of dimorphism (actual DM: male natural variation). While such pairings are infrequent, mean/female mean based on known sex) were calcu- they nevertheless have the potential to confound lated for each extant sample drawn. The first used results by artificially inflating dimorphism statistics in estimated FHD dimensions estimated for each speci- extant samples. As a means of correcting bias from men by the template method (template sexual such cases, we now systematically discard estimated dimorphism: TSD) to measure dimorphism, and the FHDs that are more than 10 mm above or below the second used the original FHDs for each randomly

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Au. afarensis skeletal dimorphism P. L. Reno et al. 3357

Table 1. Australopithecus afarensis sample used for simulations.

estimated estimated FHD/GMEAN metric specimen(s) FHD GMEAN ratio

HHD: max. diameter of humeral head A.L. 333-107a 39.4 32.6 1.21 OLCB: ML width of humerus measured tangent to the A.L. 137-48A 32.6 27.0 1.21 superior margin of the olecranon fossa A.L. 137-50b 38.3 31.6 1.21 A.L. 223-23b 35.3 29.2 1.21 A.L. 322-1 27.9 23.1 1.21 A.L. 333-29 33.2 27.4 1.21 A.L. 333w-31 34.3 28.4 1.21 Mak VP 1/3 37.8 31.3 1.21 CAPD: max. diameter of capitulum A.L. 333w-22 39.5 32.7 1.21 A.L. 444-14b 37.2 30.4 1.22 RHD: max. diameter of the radial head A.L. 333x-14c 44.3 36.6 1.21 A.L. 333x-15c 44.5 36.8 1.21 ULB: ML width of ulna immediately distal to radial A.L. 333x-5 37.1 30.6 1.21 facet A.L. 333w-36 29.8 24.6 1.21 A.L. 438-1ab 40.9 33.8 1.21 FHD: max. diameter of femoral head A.L. 152-2b 33.1 27.4 1.21 A.L. 288-1ap 28.6 23.6 1.21 A.L. 333-3 40.9 33.8 1.21 A.L. 827b 38.1 31.5 1.21 TRCD: max. femoral shaft diameter immediately below A.L. 211-1 36.4d 30.1 1.21 lesser trochanter A.L. 333-95c 35.3d 29.1 1.21 Mak VP 1/1 34.4d 28.4 1.21 FNKH: femoral neck height normal to long axis at A.L. 333-117 38.7 32.0 1.21 midpoint A.L. 333-123b 33.0 27.2 1.21 A.L. 333-142b 30.1 24.8 1.21 GSTB: AP femoral width immediately above A.L. 333-4 35.2 29.1 1.21 gastrocnemius tubercles A.L. 333w-56 33.6 27.8 1.21 A.L. 333-140b 30.2 24.9 1.21 CNDC: ML distance between centers of medial and A.L. 129-1b 27.9 23.0 1.21 lateral tibial condyles A.L. 333x-26 38.5 31.8 1.21 A.L. 333-42 36.7 30.3 1.21 PRXTB: max. ML tibial bicondylar breadth A.L. 330-6b 37.4 30.9 1.21 DSTTB: AP articular length at ML mid-point of A.L. 333-6 37.2 30.8 1.21 articular surface of distal tibia A.L. 333-7 42.9 35.5 1.21 A.L. 333-96 38.4 31.7 1.21 A.L. 545-3b 31.9 26.1 1.22 FIBD: max. ML diameter of distal fibula A.L. 333-9A 42.8 35.4 1.21 A.L. 333-9B 38.9 32.1 1.21 A.L. 333w-37 37.8 31.3 1.21 A.L. 333-85 40.6 33.5 1.21 TAL: max. AP length of talus A.L. 333-147b 36.0 29.8 1.21 aBecause of slight eccentricity in this specimen the average of the mediolateral (ML) and anteroposterior (AP) diameters was used instead. bSpecimens new to this analysis. cThese specimens lack epiphyseal fusion and are not strictly adults. They were included because they constitute three of the largest fossils in the sample and their omission would further decrease fossil dimorphism estimates. dThese values are based on a slightly different metric than the one used in the previous analysis (AP subtrochanteric diameter reported by Lovejoy et al. (1982)) and therefore differ from those reported in Reno et al. (2003). The new metric corresponds more closely with that taken from the comparative samples and better reflects the relative size of these specimens (i.e. 333-3 is clearly larger than any of these three specimens). selected individual (direct sexual dimorphism: DSD) calculated using the template (A.L. 288-1) and the to determine it. Comparison of TSD and DSD resulting estimated FHD. Also included in table 1 assesses the effect of using a template specimen to esti- are estimated geometric means (GMEAN) of all mate dimorphism. Because calculation of dimorphism included metrics that could also be calculated by the statistics (BDI and CV) for Au. afarensis requires use template method in addition to the FHD. These are of a template specimen (A.L. 288-1), these can be included to illustrate that the results of the template assessed only by comparison with TSD produced by method do not depend on the choice of FHD to the simulations. measure sample dimorphism. Note that ratios between estimated FHD and estimated GMEAN are always identical. Thus, any measure of sample variation 3. RESULTS (i.e. CV or BDI) will be identical and any scalar Table 1 lists each Au. afarensis specimen used in the metric from the template will produce the same current analysis, the metric from which its FHD was result. Therefore, any species-specific allometric

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3358 P. L. Reno et al. Au. afarensis skeletal dimorphism

50 to overestimate size dimorphism in the minimally dimorphic species (compare values in chimpanzees and gorillas; table 2). As noted above, this is an 45 expected finding because males are always assumed to be larger than females. Frequency histograms of dimorphism values gener- 40 ated by simulating the A.L. 333, CA and non-333 samples are provided in figure 2. As expected, human dimorphism values were found to be intermediate between those of chimpanzees and gorillas. 35 Also as expected (see discussion above), template- derived size dimorphism statistics tended to overestimate direct dimorphism values (table 3). For 30 each iteration, a Pearson correlation coefficient was computed between the resulting template-derived estimated FHDs and the directly measured FHDs. estimated femoral head diameter (mm) 25 The means and standard deviations of these correlation coefficients for all simulations are presented in table 4. The strength of the correlation between 20 template and direct values varied among species as a A.L. 333 non-333 direct consequence of their relative dimorphism. As Figure 1. Estimated FHD for individual Au. afarensis speci- expected, in non-dimorphic chimpanzees, the error mens included in this analysis. Circles, original specimens; in estimating FHD was relatively high compared with triangles, specimens new to this analysis. the size range of the species. In contrast, in highly dimorphic gorillas, it was relatively low. The patterns of correlation observed between template FHD and relationships with FHD have no effect on the outcome direct FHD in the extant taxa verify that the template of the procedure. method satisfactorily reflects actual dimorphism levels Estimated FHDs for each of the original 29 speci- in these samples. mens of Reno et al. (2003) plus the 12 additional As in our original analysis, BDI and CV calculated specimens included in the present analysis are shown for Au. afarensis were most similar to those of humans in figure 1. As it demonstrates, the new specimens (figure 2). This was true not only of the A.L. 333 and increase representation of small individuals at A.L. CA samples, but also for the non-333 sample. 333 but not to the extreme range represented by the Table 5 presents exact counts of the number of smallest individuals of the non-333 sample. On the iterations that fell above or below the Au. afarensis other hand, the new specimens expand the upper value in each simulation. As these data demonstrate, size range of the non-333 sample (although not to dimorphism within the expanded A.L. 333 sample the extent observed in A.L. 333) and appreciably increased from a BDI of 1.167 in our previous analy- increase the representation of intermediate-sized sis to a value of 1.195 here, which places Au. afarensis individuals in the non-333 sample such that there is dimorphism in the middle of the distribution of no longer any potential demarcation between large human values. However, because of its small sample and small specimens in the combined CA sample. size (modelled as representing nine individuals), it is These novel specimens provide little reason to con- statistically indistinguishable from any of the three clude that A.L. 333 under-represents small Lucy- hominoids. sized individuals. To the contrary, given the large Unlike the results for A.L. 333, dimorphism within number of intermediate-sized specimens, it is quite the expanded CA sample decreased from a BDI of possible that such extremely small individuals may 1.222 in our original study to 1.209 here, a value actually be over-represented in the non-333 sample. that is significantly different from that of both the Table 2 presents samples sizes, CVs, actual DM and extremely dimorphic gorillas and the minimally the BDI for each metric. Dimorphism in humans is dimorphic chimpanzees. The slightly higher dimorph- intermediate between non-dimorphic chimpanzees ism value of 1.213 calculated for the non-333 sample and highly dimorphic gorillas for nearly all characters also differed significantly from that of gorillas using a (only the chimpanzee capitulum (CAPD) BDI and directional test (which is appropriate considering CV are slightly greater than humans). However, gorillas set the upper range of primate dimorphism). within each taxon, the extent to which skeletal Because of its reliance on estimated rather than metrics differ between the sexes varies extensively. actual FHDs, the template method contributes an Significantly, variation in FHD in all three hominoid additional source of error to estimates of dimorphism. taxa is low in comparison to that observed for most In order to ensure that this error is not a function of other skeletal metrics (thus, FHD will have a smaller the size of the template specimen—a potential concern relative range). Given these findings, the template given the unusually small size of A.L. 288-1—we com- method can be expected to overestimate the means pared dimorphism values generated by different sized and dispersions of direct dimorphism values. templates. Template size has no systematic effect Although the BDIs correlated well with the actual (figure 3), and therefore our results are not biased by dimorphism observed for each metric, they tended the small size of A.L. 288-1.

Phil. Trans. R. Soc. B (2010) hl rn.R o.B Soc. R. Trans. Phil. (2010) Downloaded from Table 2. Sample sizes and dimorphism statistics for individual metrics measured directly from chimpanzee, human and gorilla specimens. Metrics are explained in table 1.

species HHD OLCB CAPD RHD ULB FHD TRCD FNEKH GSTB CNDC PRXTB DSTTB FIBD TALUS GMEAN

chimpanzees male (n)232323222323232323232323221918 female (n)252525252525252525252525252424 rstb.royalsocietypublishing.org CV 6.64 9.42 9.49 6.65 9.03 6.48 7.80 7.27 7.31 7.40 5.98 8.20 6.72 6.23 5.32 actual DM 1.066 1.073 1.090 1.038 1.059 1.049 1.054 1.032 1.029 1.036 1.041 1.003 1.052 1.029 1.043 BDI 1.115 1.163 1.165 1.113 1.155 1.114 1.133 1.123 1.126 1.127 1.103 1.147 1.110 1.106 1.093 difference 0.049 0.090 0.075 0.075 0.096 0.065 0.079 0.091 0.097 0.091 0.062 0.144 0.058 0.077 0.050

humans afarensis Au. male (n)252525252525252525252525252525 female (n)252525252525252525252525252525 CV 8.92 11.08 9.17 8.87 13.07 8.81 9.64 11.00 8.67 10.17 7.73 12.07 9.67 8.02 8.02 actual DM 1.170 1.176 1.124 1.151 1.203 1.157 1.142 1.178 1.100 1.173 1.134 1.183 1.116 1.135 1.151 onOctober24,2010 BDI 1.169 1.202 1.161 1.160 1.237 1.161 1.165 1.199 1.151 1.186 1.140 1.214 1.166 1.141 1.153 difference 20.001 0.026 0.037 0.009 0.034 0.004 0.023 0.021 0.051 0.013 0.006 0.031 0.050 0.006 0.002 dimorphism skeletal gorillas male (n)242525242525252525252525232119 female (n)252525252525252525252525252222 CV 14.92 18.09 15.02 14.60 16.74 12.66 12.66 14.00 14.98 14.23 13.62 13.80 16.06 10.98 13.29 actual DM 1.325 1.390 1.300 1.296 1.324 1.258 1.225 1.258 1.297 1.275 1.279 1.257 1.330 1.208 1.279 BDI 1.305 1.378 1.297 1.290 1.324 1.252 1.242 1.268 1.299 1.284 1.271 1.259 1.325 1.208 1.264 difference 20.020 20.012 20.003 20.006 0.000 20.006 20.003 0.010 0.002 0.009 20.008 0.002 20.005 0.000 20.015 .L Reno L. P. tal. et 3359 Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3360 P. L. Reno et al. Au. afarensis skeletal dimorphism

(a) 300 1.195 11.23 250 200 150

A.L. 333 100 50 0 (b) 300 1.209 11.89 250 200 150 100 50 0 1.05 1.10 1.15 1.20 1.25 1.30 1.35 1.40 1.45 1.50 1.601.55

non-333100 Combined Afar 50 0 1.05 1.10 1.15 1.20 1.25 1.30 1.35 1.40 1.45 1.50 1.601.55 4 6 8 10121416182022242628 BDI CV

Figure 2. Frequency histograms of dimorphism values generated by simulating the (a) A.L. 333, (b) Combined Afar and (c) non-333 assemblages using chimpanzee (white bars), human (grey bars) and gorilla (black bars) comparative samples (1000 iterations each). The vertical line in each plot indicates dimorphism for the Au. afarensis sample.

Table 3. Summary statistics from each of the extant hominoid simulations.

chimpanzee human gorilla

TSD DSD TSD DSD TSD DSD

A.L. 333 simulation: 1000 iterations actual DM mean 1.045 1.053 1.152 1.160 1.290 1.258 s.d. 0.050 0.046 0.052 0.040 0.060 0.046 BDI mean 1.155 1.102 1.198 1.140 1.272 1.206 s.d. 0.029 0.027 0.037 0.036 0.053 0.051 CV mean 9.29 6.05 11.50 8.21 14.94 11.78 s.d. 1.59 1.31 1.90 1.74 2.26 2.07 Combined Afar simulation: 1000 iterations actual DM mean 1.049 1.051 1.155 1.159 1.297 1.258 s.d. 0.036 0.030 0.038 0.026 0.041 0.029 BDI mean 1.162 1.108 1.203 1.150 1.292 1.227 s.d. 0.025 0.014 0.031 0.025 0.037 0.032 CV mean 9.52 6.24 11.55 8.50 15.49 12.21 s.d. 1.33 0.82 1.54 1.16 1.54 1.18 non-A.L. 333: 1000 iterations actual DM mean 1.057 1.050 1.162 1.156 1.305 1.258 s.d. 0.053 0.033 0.054 0.029 0.061 0.032 BDI mean 1.161 1.107 1.194 1.148 1.294 1.223 s.d. 0.037 0.019 0.042 0.025 0.052 0.033 CV mean 9.67 6.43 11.36 8.67 16.14 12.54 s.d. 2.00 0.94 2.18 1.29 2.31 1.39

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Au. afarensis skeletal dimorphism P. L. Reno et al. 3361

Table 4. Means and standard deviations of the correlations sample, which includes both large and small size between actual and estimated FHD computed for each of extremes, can reject both low chimpanzee and high the 1000 iterations. gorilla-like dimorphism. This also stresses the need to maximize sample size to include the numerous chimpanzee human gorilla intermediate sized specimens, as this tends to ensure that more complete yet extreme-sized individuals (i.e. A.L. 333 mean 0.453 0.620 0.819 A.L. 288-1, A.L. 128/129 and A.L. 333-3) do not s.d. 0.179 0.145 0.087 unduly influence dimorphism estimates (e.g. Gordon et al. 2008). Combined Afar mean 0.498 0.642 0.831 It is clear that our method of assessing skeletal s.d. 0.132 0.102 0.056 dimorphism within the A.L. 333 assemblage is appropriate regardless of any sex bias due to sampling error non-333 mean 0.565 0.675 0.832 (e.g. as argued by Plavcan et al. (2005) and Scott & s.d. 0.190 0.147 0.091 Stroik (2006)). Moreover, the presence of small juvenile specimens preserved at A.L. 333 suggests that no systematic size sorting occurred during the formation of the assemblage. As noted above, because the sex 4. DISCUSSION of each postcranial element in the Au. afarensis The present study is based on dimorphism estimates sample is unknown, the numbers of males and females generated from 41 fossils representing a minimum of included within the bootstrapped samples were 20 separate individuals. While we look forward to the allowed to vary freely. Therefore, the bootstrapped potential of adding more fossils when available, it is simulations produced all possibilities with respect to likely that the sample is now reaching a ‘critical sex composition. Indeed, some of our simulations gen- mass’ such that additional specimens are unlikely to erated samples containing only one sex. appreciably change the dimorphism estimates. Results That Au. afarensis displayed only moderate size confirm our previous conclusions that dimorphism is dimorphism is consistent with the minimal size only minimal to moderate in Au. afarensis. Skeletal dimorphism observed in Ardipithecus ramidus (Suwa variation in the CA sample differs significantly from et al. 2009; White et al. 2009). Indeed, given the those of gorillas and chimpanzees but cannot be stat- absence of appreciable skeletal size dimorphism in istically distinguished from that of modern humans both Pan and Ardipithecus, there is now strong evi- (table 5). Significantly, the dimorphism values calcu- dence that the last common ancestor of lated for the non-333 sample demonstrate that the chimpanzees, bonobos and humans also displayed results obtained for the combined sample are not minimal skeletal dimorphism and that it probably biased in any way by the composition of A.L. 333, a increased in hominids subsequent to 4.4 Ma. finding that renders moot all criticisms of our original Recently, Lawler (2009) established that ecological study which relied on this argument (i.e. Plavcan et al. factors (e.g. substrate preference or feeding niche) 2005; Scott & Stroik 2006). Incorporation of four often produce dimorphism ratios that differ substan- additional individuals of small to intermediate body tially from those predicted by simple sexual selection size did indeed increase the dimorphism in the A.L. theory (e.g. the ‘tournament sex’ of Devore & Lovejoy 333 sample, just sufficient to prevent statistical signifi- (1985)). Lovejoy (1981, 1993, 2009) has argued that a cance in its difference from gorillas (table 5). However, provisioning model favours the selection of large males as is confirmed by both the lower dimorphism values by females because greater body mass increases both actually calculated for this sample (figure 2), and the mobility and predator resistance in males. Also, selec- nearly equivalent ranges of variation observed between tion of small females by males reduces that female’s the A.L. 333 and non-333 samples (figure 1), this fat/protein requirements and thereby lowers compe- finding reflects A.L. 333’s small sample size (n ¼ 9), tition with the male’s offspring for nutrient-rich as sample size has a profound impact on adequately foods. In addition, the obviously minimal intermale inferring dimorphism (Koscinski & Pietraszewski aggression in Ar. ramidus, as now established by the 2004). Indeed, the A.L. 333 locality, which represents multiple trait shifts in its sectorial canine complex one of the most complete and taphonomically (including those of size, crown form, eruption time unbiased hominid sites ever found, still probably pro- and upper/lower canine differences (Suwa et al. vides the most accurate sample of Au. afarensis 2009; White et al. 2009)), makes it even more unlikely dimorphism. It should also be noted, in addition, that extreme dimorphism would evolve so rapidly in that an upper limit to dimorphism within this species Au. afarensis via direct male–male competition for is set by combining specimens from geographically mates. Instead, moderate dimorphism appears to be and temporally distinct sites (i.e. the total CA an ecologically driven feature in the hominid lineage sample), as this practice must enhance the variance that probably continued into later taxa (i.e. Au. africa- beyond that typical of local demes. nus (Harmon 2009)), and although most probably the The inclusion of new specimens also reinforces the result of sexual selection, it was probably not driven by fact that, while A.L. 333 preserves a number of large direct male–male agonistic competition for mates, but specimens, and multiple small individuals have been rather by ecologically driven male and female choice. recovered from non-333 localities, the majority of Indeed, it would seem that there are now two compet- Au. afarensis specimens are intermediately sized ing explanations for the increase in skeletal dimorphism (figure 1). It is thus also noteworthy that the CA from 4.4 Ma (Ar. ramidus, White et al. 2009) to 3.2 Ma

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3362 P. L. Reno et al. Au. afarensis skeletal dimorphism

Table 5. Simulations of Au. afarensis dimorphism from three different fossil assemblages. These are exact counts of values that fall less than or greater than the Au. afarensis value. Each count can be transformed into a proportion by dividing by 1000.

chimpanzee human gorilla fossil assemblage dimorphism less than greater than less than greater than less than greater than

A.L. 333 BDI 1.195 905 95 484 516 74 926 CV 11.23 878 122 447 553 51 949 Combined Afar BDI 1.209 960 40 607 393 8 992 CV 11.89 954 46 597 403 7 993 non-333 BDI 1.213 916 84 698 302 50 950 CV 12.03 870 130 627 373 35 965

1.5 30 chimpanzee chimpanzee 1.4 25 1.3 20 15 CV BDI 1.2 10 1.1 5 1.0 0

1.5 30 human human 1.4 25 1.3 20 15 CV BDI 1.2 10 1.1 5 1.0 0

1.5 gorilla 30 gorilla 1.4 25 1.3 20 15 CV BDI 1.2 10 1.1 5 1.0 0 template specimen by increasing FHD template specimen by increasing FHD

Figure 3. Box and whisker plots showing range of sample dimorphism values generated for each template specimen. Template specimens are arrayed by increasing FHD. Boxes indicate interquartile range, whiskers 95% interval; circles are outliers.

(A.L. 333, Kimbel et al. 1994): (i) an increase in sizes sufficient for statistical reliability are likely to be male–male agonism for mate selection or (ii) the generated from rare early hominid fossils. It should enhancement of male resistance to predation in be noted, moreover, that this method is fully appli- response to occupation of novel environments by the cable to other species of fossil hominoids, so long as more ecologically expansive Australopithecus radiation, a partial skeleton and a sufficiently large series of unas- including the invasion of new predator-rich environ- sociated fossils with homologous anatomical sites are ments such as lake margins, savannas and veldts. available (e.g. Proconsul (Walker & Teaford 1989; Given that the former of these two choices would Ward et al. 1993), Ar. ramidus (White et al. 2009) likely depress sub-adult survivorship and increase and Au. africanus (Clarke 1999)). For South African parenting load on females, when coupled with the Australopithecus, however, special consideration of now clear adaptive radiation of Australopithecus that taphonomic variables will have to be made, since followed Ar. ramidus, the latter of these two seems cave assemblages are probably the result of carnivore far more likely. kills (Brain 1981). Because no specific size-sorting mechanism has been identified for A.L. 333, this site remains an appropriate venue for examination of skel- 5. CONCLUSIONS etal dimorphism in Au. afarensis. The template method is a robust technique for esti- We thank Yohannes Haile-Selassie of the Cleveland Museum mating size variance in early hominids and is the of Natural History for access to primate skeletons and to only method currently available with which sample Lyman Jellema for technical assistance. William Kimbel

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Au. afarensis skeletal dimorphism P. L. Reno et al. 3363 kindly provided metrics to unpublished fossil specimens. We In The origin and evolution of humans and humanness (ed. also thank Alan Walker and Chris Stringer for organizing the D. T. Rasmussen), pp. 1–28. Boston, MA: Jones and discussion meeting and the staff of the Royal Society for Bartless Publisher. ensuring its success. Lovejoy, C. O. 2009 Reexamining human origins in light of Ardipithecus ramidus. Science 326, 74e1–74e8. (doi:10. 1126/science.1175834) REFERENCES Lovejoy, C. O., Johanson, D. C. & Coppens, Y. 1982 Behrensmeyer, A. K., Harmon, E. H. & Kimbel, W. H. 2003 Hominid lower-limb bones recovered from the Hadar Environmental context and taphonomy of the formation—1974–1977 collections. Am. J. Phys. A.L. 333 Locality, Hadar, Ethiopia. Abstracts of the Anthropol. 57, 679–700. (doi:10.1002/ajpa.1330570411) Paleoanthropology Society. See http://www.paleoanthro. McHenry, H. M. 1991 Sexual dimorphism in Australopithecus org/abst2003.htm. afarensis. J. Hum. Evol. 20, 21–32. (doi:10.1016/0047- Brain, C. K. 1981 The hunters or the hunted. Chicago, IL: 2484(91)90043-U) University of Chicago Press. Plavcan, J. M., Lockwood, C. A., Kimbel, W. H., Lague, Clarke, R. J. 1999 Discovery of complete arm and hand of M. R. & Harmon, E. H. 2005 Sexual dimorphism the 3.3 million-year-old Australopithecus skeleton from in Australopithecus afarensis revisited: how strong is Sterkfontein. S. Afr. J. Sci. 95, 477–480. the case for a human-like pattern of dimorphism? Devore, I. & Lovejoy, C. O. 1985 The natural superiority J. Hum. Evol. 48, 313–320. (doi:10.1016/j.jhevol.2004. of women. In The physical and mental health of aged 09.006) women (eds M. R. Haug, A. B. Ford & M. Sheafor), Reno, P. L., Meindl, R. S., McCollum, M. A. & Lovejoy, pp. 27–38. New York, NY: Springer. C. O. 2003 Sexual dimorphism in Australopithecus Drapeau, M. S., Ward, C. V., Kimbel, W. H., Johanson, afarensis was similar to that of modern humans. Proc. D. C. & Rak, Y. 2005 Associated cranial and forelimb Natl Acad. Sci. USA 100, 9404–9409. (doi:10.1073/ remains attributed to Australopithecus afarensis from pnas.1133180100) Hadar, Ethiopia. J. Hum. Evol. 48, 593–642. (doi:10. Reno, P. L., Meindl, R. S., McCollum, M. A. & Lovejoy, 1016/j.jhevol.2005.02.005) C. O. 2005 The case is unchanged and remains Gordon, A. D., Green, D. J. & Richmond, B. G. 2008 Strong robust: Australopithecus afarensis exhibits only moderate postcranial size dimorphism in Australopithecus afarensis: skeletal dimorphism. A reply to Plavcan et al. (2005). results from two new resampling methods for multivariate J. Hum. Evol. 49, 279–288. (doi:10.1016/j.jhevol.2005. data sets with missing data. Am. J. Phys. Anthropol. 135, 04.008) 311–328. (doi:10.1002/ajpa.20745) Scott, J. E. & Stroik, L. K. 2006 Bootstrap tests of Harmon, E. H. 2006 Size and shape variation in significance and the case for humanlike skeletal-size Australopithecus afarensis proximal femora. J. Hum. Evol. dimorphism in Australopithecus afarensis. J. Hum. Evol. 51, 217–227. (doi:10.1016/j.jhevol.2006.01.009) 51, 422–428. (doi:10.1016/j.jhevol.2006.06.001) Harmon, E. 2009 Size and shape variation in the proximal Sokal, R. R. & Rohlf, F. J. 1995 Biometry. New York, NY: femur of Australopithecus africanus. J. Hum. Evol. 56, W. H. Freeman and Co. 551–559. (doi:10.1016/j.jhevol.2009.01.002) Suwa, G., Kono, R. T., Simpson, S. W., Asfaw, B., Kimbel, W. H., Johanson, D. C. & Rak, Y. 1994 The Lovejoy, C. O. & White, T. D. 2009 Paleobiological first skull and other new discoveries of Australopithecus implications of the Ardipithecus ramidus dentition. afarensis at Hadar, Ethiopia. Nature 368, 449–451. Science 326 , 94–99. (doi:10.1126/science.1175824) (doi:10.1038/368449a0) Walker,A.&Teaford,M.1989ThehuntforProconsul. Sci. Kimbel, W. H., Rak, Y. & Johanson, D. C. 2004 The skull Am. 260,76–82.(doi:10.1038/scientificamerican0189-76) of Australopithecus afarensis. New York, NY: Oxford Ward, C. V., Walker, A., Teaford, M. F. & Odhiambo, I. University Press. 1993 Partial skeleton of Proconsul nyanzae from Koscinski, K. & Pietraszewski, S. 2004 Methods to estimate Mfangano Island, Kenya. Am. J. Phys. Anthropol. 90, sexual dimorphism from unsexed samples: a test 77–111. (doi:10.1002/ajpa.1330900106) with computer-generated samples. Anthropol. Rev. 67, White, T. D. & Johanson, D. C. 1989 The hominid 33–55. composition of Afar Locality 333: some preliminary Lawler, R. R. 2009 Monomorphism, male–male compe- observations. In Hominidae, pp. 97–101. Milan, Italy: tition, and mechanisms of sexual dimorphism. J. Hum. Jaka Book. Evol. 57, 321–325. (doi:10.1016/j.jhevol.2009.07.001) White, T. D., Asfaw, B., Beyene, Y., Haile-Selassie, Y., Lockwood, C. A., Richmond, B. G., Jungers, W. L. & Lovejoy, C. O., Suwa, G. & WoldeGabriel, G. 2009 Kimbel, W. H. 1996 Randomization procedures and Ardipithecus ramidus and the paleobiology of early sexual dimorphism in Australopithecus afarensis. J. Hum. hominids. Science 326, 75–86. (doi:10.1126/science. Evol. 31, 537–548. (doi:10.1006/jhev.1996.0078) 1175802) Lovejoy, C. O. 1981 The origin of man. Science 211, Zihlman, A. L. & Tobias, P. V. 1985 Australopithecus afarensis: 341–350. (doi:10.1126/science.211.4480.341) two sexes or two species? In Hominid evolution: past, Lovejoy, C. O. 1993 Modeling human origins: are we sexy present and future, pp. 213–220. New York, NY: Alan because we’re smart, or smart because we’re sexy? R. Liss, Inc.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Phil. Trans. R. Soc. B (2010) 365, 3365–3376 doi:10.1098/rstb.2010.0070

The cranial base of Australopithecus afarensis: new insights from the female skull William H. Kimbel1,* and Yoel Rak2 1Institute of Human Origins, School of Human Evolution and Social Change, Arizona State University, Tempe, AZ, USA 2Department of Anatomy, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel Cranial base morphology differs among hominoids in ways that are usually attributed to some combination of an enlarged brain, retracted face and upright locomotion in humans. The human foramen magnum is anteriorly inclined and, with the occipital condyles, is forwardly located on a broad, short and flexed basicranium; the petrous elements are coronally rotated; the glenoid region is topographically complex; the nuchal lines are low; and the nuchal plane is horizontal. Australopithecus afarensis (3.7–3.0 Ma) is the earliest known species of the australopith grade in which the adult cranial base can be assessed comprehensively. This region of the adult skull was known from fragments in the 1970s, but renewed fieldwork beginning in the 1990s at the Hadar site, Ethiopia (3.4–3.0 Ma), recovered two nearly complete crania and major portions of a third, each associated with a mandible. These new specimens confirm that in small-brained, bipedal Australopithecus the foramen magnum and occipital condyles were anteriorly sited, as in humans, but without the foramen’s forward inclination. In the large male A.L. 444-2 this is associated with a short basal axis, a bilateral expansion of the base, and an inferiorly rotated, flexed occipital squama—all derived characters shared by later australopiths and humans. However, in A.L. 822-1 (a female) a more primitive morphology is present: although the foramen and condyles reside anteriorly on a short base, the nuchal lines are very high, the nuchal plane is very steep, and the base is as relatively narrow centrally. A.L. 822-1 illuminates fragmentary specimens in the 1970s Hadar collection that hint at aspects of this primitive suite, suggesting that it is a common pattern in the A. afarensis hypodigm. We explore the implications of these specimens for sexual dimorphism and evolutionary scenarios of functional integration in the hominin cranial base. Keywords: Australopithecus; cranial base; bipedality

1. INTRODUCTION The 3.7–3.0 Myr old hominin species As the critical intersection of the locomotor, neural and Australopithecus afarensis is usually considered to be masticatory systems, the cranial base is a frequently the plesiomorphic (‘primitive’, African apelike) sister consulted source for insight into the evolution of the taxon to subsequent australopiths and the genus human head in phylogenetic and functional–adaptive Homo (e.g. Kimbel et al. 2004; Strait & Grine 2004). contexts. A great deal of experimental and comparative Relative to these successor taxa, the skull and dentition research on extant primates has been conducted to elu- of A. afarensis is characterized by numerous primitive cidate the relative influence of each of these systems on features, many of which are part of, or at least influ- cranial base form (e.g. Lieberman et al. 2000), but the enced by, the masticatory system (see Kimbel & fossil record—especially its earlier segments—is less Delezene 2009, for a recent review). The cranial often studied because of small samples, poor preser- base was spottily represented in the initial (1970s) vation, and/or inaccessible endocranial spaces. Yet, hypodigm of the species; a single partial calvaria of the ultimate test of hypotheses regarding the conjunc- an adult male individual from A.L. 333 (ca 3.2 Ma) tion of structural innovations and their purported was the principal source of information until the first functions in an adaptive context is the relative timing complete adult skull of the species was recovered of their first appearances in taxa potentially ancestral in 1992 (A.L. 444-2, a large adult male, ca 3 Ma). (or at least sister) to the extant taxa targeted by most These specimens showed that, in contrast to the primi- of this research. Therefore, it is important to glean as tive morphology of the temporal bone (e.g. low-relief much information as possible from the available fossil mandibular fossa, tubular tympanic element, highly remains. inflated squama, asterionic notch sutural pattern, etc.), the calvaria is derived in its anteriorly positioned foramen magnum and occipital condyles on a short, broad * Author for correspondence ([email protected]). cranial base—features shared with modern humans. One contribution of 14 to a Discussion Meeting Issue ‘The first four The female calvaria of A. afarensis was until recently million years of human evolution’. known only from fragments of the calotte, and while

3365 This journal is q 2010 The Royal Society Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3366 W. H. Kimbel & Y. Rak A. afarensis cranial base these hinted at several morphological differences from and reveals an apparently unique (in the extant African the larger (male) skulls, particularly in aspects of occi- hominoid context) pattern of sexual dimorphism in pital squama form, interpreting this morphology in the the australopith cranial base. Our more comprehensive context of the entire skull was not possible. Fieldwork description of the A.L. 822-1 skull is in preparation, in the 2000s rectified this with the recovery of a nearly but here we focus on the cranial base of this specimen complete skull of a small, probably female individual, and its substantive role in illuminating these issues. A.L. 822-1. This specimen adds information on variation in A. afarensis cranial base form, casts previously known fragmentary remains of small 2. THE A.L. 822-1 SKULL (female) individuals of this species in a new light, The skull A.L. 822-1 was found by the late Dato Adan, an Afar member of the Hadar Research Project, during the 2000 field season. It was recovered from the surface of sediments of the Hadar Formation’s KH-1 sub-member and is estimated to be approximately 3.1 Ma (C. Campisano 2009, personal communication). It is the third adult individual from Hadar to preserve both the mandible and cranium (A.L. 444-2, KH-2 sub-member, ca 3 Ma and A.L. 417-1, SH-3 sub-member, ca 3.3 Ma, are the other two). The specimen was recovered in approximately 200 fragments, which have been cleaned, reconstructed and reassembled to constitute most of an adult skull with almost all of the dentition (figure 1). The reconstruction of the original specimen reveals remnant distortion, owing to both warping and crushing, in (i) the failure of the palatal and calvarial midlines to align, (ii) the temporal bones’ placement on different coronal planes, and (iii) some bilateral compression of both palatal and mandibular arches (figure 2). We describe in detail this deformation and the steps taken to correct it in our comprehensive comparative study currently in preparation. The data and analyses presented here are based on our final restoration using casts of the original fossil. The A.L. 822-1 skull presents numerous character- Figure 1. The reconstructed A.L. 822-1 skull, oblique view. istics diagnostic of A. afarensis (see Johanson et al. Approximately 45% natural size. 1978; White et al. 1981, 1993, 2000; Rak 1983;

(a) (b)

Figure 2. Pattern of distortion in the A.L. 822-1 reconstruction. (a) Basal view. Note the asymmetric positions of the temporal bones, resulting from deformation of the (anatomical) left side. (b) Superior view. Note the offset of the face in relation to the calvarial midline. Red line denotes anatomical midline.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

A. afarensis cranial base W. H. Kimbel & Y. Rak 3367

A.L. 822-1 0 25 50 75 100

22.5 25.0 27.5 30.0 32.5 mm 27.5 30.0 32.5 35.0 37.5 40.0 42.5 mm mandibular condyle ML breadth (n = 6) mandible corpus depth @ M1 (n = 20)

10.5 11.0 11.5 12.0 12.5 13.0 13.5 14 mm 9.0 9.5 10.0 10.5 11.0 11.5 12.0 12.5 mm mandibular M1 breadth (n = 18) maxillary canine breadth (n = 12) Figure 3. Box-plot metrical profile of A.L. 822-1. Data for A.L. 822-1 shown in Hadar A. afarensis sample distribution. Bold vertical line indicates value for A.L. 822-1; rectangle defines the 25th and 75th quartiles; diamond defines the mean and 95% CI; short vertical line within rectangle defines the median.

Figure 5. Hadar cranium A.L. 444-2. Note the low position of the compound temporal/nuchal crest (arrow), which approximates the biasterion line, a phylogenetically derived condition of other mature males of A. afarensis. Approxi- mately 50% natural size.

Figure 4. Lateral view of A.L. 822-1 ‘ﬁnal’ restoration (cast), showing the superiorly extended position of the nuchal lines — sharply angled articular surface of the occipital as expressed by W. E. Le Gros Clark’s nuchal area height condyle, index: maximum height of nuchal lines (upper horizontal — ﬂattened, horizontally oriented tympanic elements, line) above Frankfurt Horizontal (lower horizontal line) as — hollowed lateral surface of the mandibular corpus a percentage of maximum cranial vault height above FH (beneath the premolars) with high ramal root and (vertical line). narrow extramolar sulcus, — marked topographic step down from mesial P3 to Kimbel et al. 1984, 1994, 2004; Kimbel & Delezene the distal-P3 to M3 occlusal platform. 2009). These include: The A.L. 822-1 skull is most probably that of a female. — strongly prognathic, biconvex maxillary subnasal Its overall cranial dimensions are small, closely surface, approximating those of the very small though — narrow midface (nasal aperture and interorbital incomplete Hadar adult calvaria A.L. 162-128 block), (Kimbel et al. 1982). The mastoid process is much — mild sagittal convexity of the low frontal squama, smaller than the mastoids of male crania such as — medial to lateral supraorbital thickness gradient, A.L. 333-45, A.L. 333-84 and A.L. 444-2. Consistent — posteriorly convergent temporal lines, with its small external dimensions, our preliminary — steeply inclined nuchal plane, estimate of endocranial volume (using mustard seed) — low upper scale (la-i) and long lower scale (i-o) of is 385 cc, which is similar to that estimated for A.L. occipital squama, 162-28 (375–400 cc) and smaller than the estimates

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3368 W. H. Kimbel & Y. Rak A. afarensis cranial base

Table 1. Nuchal area height index in hominoids. Negative Two large (presumptive male) A. afarensis crania value indicates nuchal area height is below Frankfurt have nuchal area height index values slightly higher Horizontal. Comparative data from Kimbel et al. (2004). than the australopith average of 10 per cent (A.L. 333-45, in which FH is based on the ‘composite nuchal area height reconstruction’ of Kimbel et al. 1984, 13%; A.L. taxon/specimen index (%) 444-2, 12%). For A.L 822-1, however, the value is Australopithecus afarensis ca 23 per cent, about 9 per cent higher than for any A.L. 444-2 13 other measurable, undistorted australopith cranium A.L. 333-45 (recon.) 12 and about midway between chimpanzee and human A.L. 822-1 23 means (figure 4). The relatively high index value is Australopithecus africanus consistent with the visibly steep nuchal plane in A.L. Sts. 5 10 822-1, which faces posteroinferiorly at ca 678 to the Australopithecus boisei FH. (The usual way of expressing nuchal plane steep- OH 5 11 ness is the inclination of the inion–opisthion chord KNM-ER 406 7 relative to FH; for A.L. 822-1, this angle is 428.How- KNM-ER 13 750 23 ever, when the foramen magnum is anteriorly KNM-ER 732 12 located—as it is in all australopiths (see below)—the Australopithecus aethiopicus inion–opisthion chord angle can understate the steep- KNM-WT 17 000 14 ness of the more lateral surfaces on which the mass of Homo sapiens (n ¼ 10) 22 the nuchal muscles insert.) Pan troglodytes male (n ¼ 10) 51 The superiorly extensive, steeply angled nuchal Pan troglodytes female (n ¼ 10) 47 plane of A.L. 822-1 illuminates the morphology of Gorilla gorilla male (n ¼ 10) 108 other less complete A. afarensis crania. In A. afarensis Gorilla gorilla female (n ¼ 10) 67 partial calvariae A.L. 162-28 and KNM-ER 2602, the nuchal plane is steep and the superior nuchal lines highly arched; the transition between nuchal for clearly male crania A.L. 333-45 (ca 485 cc) and and occipital planes across the superior nuchal lines A.L. 444-2 (ca 550 cc; Holloway & Yuan 2004). As is smooth and convex; and the nuchal plane consists shown in figure 3, the condyle of the A.L. 822-1 of bilateral, posterolaterally directed plates that mandible is the second smallest in mediolateral merge at a median topographic peak (Kimbel et al. diameter among six Hadar condyles, and its maxillary 1984, 2004; Kimbel 1988). While neither fossil can canine breadth falls near the top of the smallest be oriented precisely on FH, their anatomical simi- quartile in the Hadar sample distribution (n ¼ 12). larity to A.L. 822-1 is remarkable. Both specimens Other aspects of A.L. 822-1 cranial morphology are small presumptive females that bear compound consistent with female status are discussed below. temporal/nuchal crests; although A.L. 822-1 does not, the temporal lines sweep laterally towards the 3. THE CRANIAL BASE OF A.L. 822-1 asteria within a few millimetres of the highly arched (a) Nuchal area height and morphology of the superior nuchal lines. Except for the relatively low occipital bone nuchal area height index value, the overall morpho- In the 1940s W. E. Le Gros Clark argued that the logical pattern of these female specimens is extremely small, horizontally oriented nuchal plane of the occipi- primitive. tal bone of South African australopith crania was Figure 5 depicts the large, male A. afarensis cranium compatible only with a humanlike poise of the head A.L. 444-2 in posterior view. Asterion in hominins on the cervical vertebral column (Le Gros Clark usually lies close to the FH and the low position of 1947, p. 309; 1950, p. 241–243). He devised the the superior nuchal line relative to the biasterion line ‘nuchal area height index’ to express the much smaller is indicated in the figure. We draw attention to the dis- degree to which the insertion area of the neck muscles tinction between Hadar crania that have high nuchal extended superiorly (relative to the Frankfurt horizon- lines and steep nuchal planes (A.L. 822-1, A.L. 162- tal (FH) baseline) as a percentage of maximum 28 and A.L. 439-1; the A.L. 288-1a, ‘Lucy’, occipital, calvarial height in fossil and living hominins as com- too incomplete to orient with precision, bears these pared with the great apes. In the African great apes same hallmarks), and those in which the nuchal lines the height of the nuchal area constitutes (on average) are lower (closer to the biasterion line) and the approximately 50 per cent of the calvarial height in nuchal plane much more horizontal (A.L. 333-45, both male and female chimpanzees and 67 per cent A.L. 444-2), as in most later hominins. With one in female gorillas (in male gorillas the index is more exception, this difference divides the sample by size, than 100 per cent because the superior extension of which we take to indicate sex, with males showing the enormous compound temporal/nuchal crest actu- the more derived morphology. The exception is A.L. ally surpasses maximum vault height). Among the 439-1, a very large male occipital (comparable in size australopiths, in contrast, the percentage averages to that of A.L. 444-2). Although this specimen bears only about 10 per cent (with a total range of 23% to massive compound temporal/nuchal (T/N) crests that þ14%, n ¼ 10), which is much closer to what is extend on each side from the middle of the nuchal observed in modern humans (average ¼ 22%, i.e. line laterally to asterion, it is not fully mature judging nuchal plane height is slightly below the FH; see from the open lambdoidal suture. The same mor- table 1). phology is replicated in yet another even more

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

A. afarensis cranial base W. H. Kimbel & Y. Rak 3369 fragmentary Hadar occipital, A.L. 444-1 (from the Table 2. Position and orientation of foramen magnum. same locality as the complete, old adult A.L. 444-2 Index calculated as the projected length opisthion– skull); this specimen, which consists of two non- opisthocranion/projected length glabella–opisthocranion. articulating squamous fragments that span the Negative value for foramen orientation indicates boundary between nuchal and occipital planes on oppo- anteroinferior orientation. site sides, is from a large, thick-vaulted cranium and bears a weak compound T/N crest. The relatively FM position FM feeble expression of the compound crest, together with taxon/specimen index (%) orientation (8) the completely patent and ‘puffy’ lambdodial sutural surface, argues for a younger subadult growth stage of Australopithecus afarensis this apparently male individual compared with A.L. A.L. 444-2 24 16 439-1. As in A.L. 439-1, however, the nuchal plane is A.L. 333-45 (recon.) 19 — very steep, facing more posteriorly than inferiorly, A.L. 822-1 23 14 even making allowance for errors in orientation of the Australopithecus africanus fragments. These two relatively young, large male indi- Sts. 5 19 20 viduals present a significant contrast with older adult Australopithecus boisei males as represented by A.L. 333-45 and A.L. 444-2. OH 5 24 7 The expanded sample of Hadar skulls permits the KNM-ER 406 18 13 identification of a cross-sectional ontogenetic trans- KNM-ER 13 750 20 — formation of male occipital morphology (from young Australopithecus aethiopicus to old, A.L. 444-1!439-1!333-45!444-2). This KNM-WT 17 000 21 — transformation entails increasing horizontality of the Homo sapiens (n ¼ 10) 31 28 nuchal plane, which results in greater flexion of the Pan troglodytes male (n ¼ 10) 12 18 occipital squama on the sagittal plane; increasing topo- Pan troglodytes female (n ¼ 10) 14 20 graphic flattening of the nuchal plane; lowering of the Gorilla gorilla male (n ¼ 10) 7 27 Gorilla gorilla female (n ¼ 10) 13 30 nuchal crest, and related to these shifts, an alteration of the compound T/N crest from a posterosuperior extension of the nuchal plane to an inferior projection of the occipital plane (see Kimbel et al. 2004, for com- horizontal projected length of the calvaria (g–op).1 parative observations and data). We do not, based on In our sample of African great apes, the mean index presently available evidence, see the same transform- value ranges between 7 per cent (male gorillas, with ation in female individuals of A. afarensis. All four their massive compound crests) and 14 per cent specimens from which relevant information can be (female chimpanzees). In humans the more anterior extracted are mature (A.L. 162-28, A.L. 822-1 and position of opisthion is conveyed by the much higher KNM-ER 2602, probably A.L. 288-1a) and these mean index value in our sample of 31 per cent clearly demonstrate the symplesiomorphic pattern (range ¼ 28–34%). Individual australopith values associated with young males of the species. This simi- vary between 18 and 24 per cent, with the two larity partly explains (along with an expansive A. afarensis specimens (A.L. 444-2 ¼ 24%, A.L. posterior temporalis) why compound T/N crests are 822-1 ¼ 23%) falling at the high end of this range so common in the smaller crania of this species (table 2). Weidenreich’s reported values for Asian (Kimbel et al. 2004). Homo erectus crania yields an average of about 26 per cent (range ¼ 24–28%). The importance of these data on foramen magnum (b) Position and orientation of the foramen position among fossil hominins is that neither hypoth- magnum esized postural/locomotor differences (i.e. between the The margins of the foramen magnum in A.L. 822-1 australopiths and H. erectus) nor absolute brain-size are preserved on two fragments: one extends from differences (with H. erectus having endocranial the basioccipital posterolaterally to include the right volumes 1.5 to 2.0 times larger than australopith occipital condyle and adjacent jugular process; the values) has a large impact on the position of the fora- other is a strip of nuchal plane bearing a short men on the cranial base. Rather, the major difference (14 mm) segment of the margin just anterolateral to is between the quadrupedal apes and the bipedal opisthion (although this fragment does not connect hominins. However, a different division applies to to the main portion of the occipital squama, external the data on foramen magnum orientation (ba–o line morphology constrains its placement to within a few relative to FH). In humans the foramen is forwardly mm). Between these two pieces we can estimate the inclined (i.e. the plane of the foramen faces anteroin- size and position of the foramen within a narrow feriorly) whereas in the great apes it is posteriorly error range (+2 mm). inclined (posteroinferior orientation). In A. afarensis In A. afarensis, as in all australopith species, the the reconstructed angle of the ba-o line is ca 148 in foramen magnum, and with it the occipital condyles, A.L. 822-1 and ca 168 in A.L. 444-2, values that lie resides in an anterior position on the cranial base. within the range for other australopith crania. The Typically, this is assessed through indices expressing australopith range (table 2) is well below the range the anteroposterior position of basion (ba) or opisthion for our sample of modern humans (the mean value (o) in relation to cranial length. We use Weidenreich’s for which is ca 288; again, the foramen faces anteroin- (1943) index relating the position of opisthion to the feriorly) but overlaps the low end of the range for

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3370 W. H. Kimbel & Y. Rak A. afarensis cranial base

Table 3. Measures of basi-occipital length. msp be bt bz

basi-occip. basi-occip. gorilla F length lg/biorbital taxon/specimen (mm) br * 100 chimpanzee M

Australopithecus afarensis AL 822-1 A.L. 444-2 20 21 A.L. 822-1 22 23 AL 444-2 A.L. 417-1 19 24 Australopithecus africanus Sts 5 Sts. 5 25 29 MLD 37/38 21 — KNM-ER 406 Stw 187 17 —

Australopithecus boisei KNM-WT 17 000 OH 5 20 20 KNM-ER 406 25 24 KNM-ER 407 21 — Figure 6. Schematic of relative cranial base breadth in homi- Australopithecus aethiopicus noids. Dashed vertical lines represent the biorbital breadth of KNM-WT 17 000 25 27 A.L. 822-1, to which all specimens are size-adjusted. MSP, Homo sapiens (n ¼ 10) 21 22 midsagittal plane; be, terminus of bi-entoglenoid breadth; Pan troglodytes male (n ¼ 10) 28 29 bt, terminus of bi-articular tubercle breadth; bz, terminus Pan troglodytes female (n ¼ 10) 26 28 of bizygomatic breadth. Heavy red line deﬁnes the breadth Gorilla gorilla male (n ¼ 10) 37 33 of the articular eminence. Note in A.L. 822-1, chimpanzees Gorilla gorilla female (n ¼ 10) 29 30 (males, n ¼ 10) and gorillas (females, n ¼ 10) the close approximation of the entoglenoid processes, expressing a narrow central cranial base.

Table 4. Cranial base breadth in hominoids. position between modern apes and humans, but bi-entogl. br/ closer to the former (chimpanzees, specifically). bi-entoglenoid biorbital taxon/specimen br (mm) br * 100 Whereas the anterior shift of basion and opisthion accounts for the forward location of the foramen Australopithecus afarensis magnum in australopiths and modern humans, the A.L. 444-2 80 84 relative vertical positions of these landmarks (beneath A.L. 333-45 (recon.) 78 89 the FH, for example) explain the differences in orien- A.L. 822-1 57 62 tation of the foramen (Kimbel et al. 2004). Because Australopithecus africanus the vertical position of basion is similar in apes and Sts. 5 65 76 humans, differences in foramen orientation reduce to MLD 37/38 62 — differences in the vertical position of opisthion. In Australopithecus boisei humans, uniquely, opisthion sits much further below OH 5 89 84 FH than basion (the foramen opens anteroinferiorly), KNM-ER 406 85 89 which can be explained as a consequence of overall KNM-ER 13 750 86 80 expansion and rotation of the occipital squama KNM-ER 23 000 80 — with encephalization (e.g. Weidenreich 1941; Biegert KNM-ER 407 65 — 1957). In the small-brained A. afarensis, although the Australopithecus aethiopicus foramen magnum is far forward on the base, opisthion KNM-WT 17 000 80 85 is elevated relative to basion and so the plane of the ¼ Homo sapiens (n 10) 74 77 foramen inclines posteriorly, more similar to what is Pan troglodytes male 61 67 (n ¼ 10) observed in the apes. Pan troglodytes female 59 65 Occipital morphology in A. afarensis is consistent (n ¼ 10) with these signs of affinity from the foramen Gorilla gorilla male 72 63 magnum. As discussed above, the orientation of the (n ¼ 10) nuchal plane, the height of the nuchal muscles’ inser- Gorilla gorilla female 64 64 tion area, and the degree of sagittal flexion of the (n ¼ 10) occipital squama range from symplesiomorphic (ape- like) to more derived, but taken as a package convey an intermediate position on the hominoid occipital bone morphocline. At the derived end of the morpho- chimpanzees (mean ¼ ca 198; gorilla means are about cline occipitals approach a quasi-human form in their 50% higher; see Kimbel et al. 2004, for details). Thus, relatively horizontal nuchal plane, low nuchal area in contrast to the data on foramen magnum position, height index and strongly flexed squama, but they do which align A. afarensis and other australopiths with not show the strongly rotated squama that in modern modern humans, the data on orientation of the fora- humans confines the maximum height of the nuchal men situate the australopiths in an intermediate area below the FH (on average) and drops opisthion

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

A. afarensis cranial base W. H. Kimbel & Y. Rak 3371

90 Z OH 5

ER 13 750 85 Z ER 406

80 ER 23 000, WT 17 000 X 333-45 444-2

65 ER 407 +

bientoglenoid breadth Sts 5 MLD 37/38 60 X 58-22 X 55 822-1

25 30 35 40 45 50 internal palate breadth at M2

Figure 7. Bivariate plot of palate breadth and cranial base (bi-entoglenoid) breadth in hominoids. Light brown data points, male gorilla; dark brown, female gorilla; light blue, male chimpanzee; X, Australopithecus afarensis; þ, A. africanus;Z,A. boisei. very far below basion, introducing a negative angle to Most other early hominins share the shortened the foramen’s orientation. It bears noting that the pos- external anterior cranial base with A. afarensis (table 3). ition of the foramen magnum, which is relatively Dean & Wood (1982), however, showed that the anterior in A. afarensis and close to the condition in A. africanus base is unusual in its somewhat elongated modern humans, is not linked to this impressive anteroposterior dimensions compared with other range of variation in occipital bone morphology. australopiths and early Homo. Specimen Sts 5 indeed has absolutely and relatively long basi-occipital and petrous elements compared with other hominins (table 1), but it is the only A. africanus cranium in (c) Length and breadth of the cranial base which these dimensions can be judged relative to a Along with the anterior position of the foramen non-calvarial size standard (such as biorbital breadth magnum, the shortened external length of the cranial or palatal length). In absolute terms the basi-occipital base is a derived feature in A.L. 822-1 shared with of MLD 37/38 and Stw 187 are as short as those of modern humans. This can be judged from the length A. afarensis, so it is unclear whether Sts 5 is typical of the basi-occipital fragment associated with this of A. africanus. fossil (22 mm) as well as that attributed to another Relative to body size and skull size the modern Hadar specimen, A.L. 417-1 (19 mm), which essen- human cranial base is short, but also wide, whereas tially match the mean for modern humans both the great apes exhibit the opposite proportions.2 absolutely and relative to biorbital breadth (table 3). Tobias (1967) noted that the mandibular fossa is A shortened external base can be inferred for speci- equally wide (mediolaterally) in gorillas and OH 5 mens of A. afarensis in which the basi-occipital is (the type specimen of Australopithecus boisei), but missing, such as the large (male) crania A.L. 333-45 only in the latter does the fossa project laterally far and A.L. 444-2, from the anteroposterior distance beyond the calvarial wall to anchor the temporal root between the carotid foramen and foramen ovale of the ﬂaring zygomatic arch. In gorillas, he found, (which roughly approximates basi-occipital length) or the mandibular fossae, in spite of their great breadth, the length of the petrous elements (which frame the are actually closer together on the cranial base and basi-occipital area), both of which are shorter than in so do not project nearly as far from the calvarial wall. gorillas and chimpanzees. This difference is depicted graphically in ﬁgure 6,

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3372 W. H. Kimbel & Y. Rak A. afarensis cranial base

Table 5. Palate dimensions in fossil hominins. Palate depth breadths compared with clearly male crania (A.L. is the midline height of the palatine process of the maxilla 333-45, A.L. 444-2). Biorbital breadth is not available above the inner alveolar margin at M2. Palate breadth is for A.L. 58-22, but another way to assess the relative the width across the internal alveolar crests at mid-M2. width of the central cranial base is by a simple index Palate length is the direct distance between orale and expressing the bi-entoglenoid distance as a percentage staphylion (reconstructed in some specimens). Relative of the bi-articular tubercle distance. When this is palatal breadth (Palatal Index) is calculated as palate done, it can be seen that in spite of small fossa breadth/palate length * 100. breadths (which would increase the index), the bi- palate palate palate palatal entoglenoid distance is relatively small in these two taxon/specimen depth breadth length index specimens (ca 48%), compared with the larger Hadar crania (ca 55%) and other early hominins, Australopithecus afarensis including A. africanus (ca 54%, n ¼ 2). A.L. 58-22 — 27.0 — — A.L. 199-1 11.0 32.0 54.0 59.3 A.L. 200-1a 8.5 33.5 65.0 51.5 (d) The cranial base and palate shape A.L. 417-1d 14.0 28.5 58.0 49.1 In their diagnosis of A. afarensis, Johanson et al. (1978, A.L. 427-1 11.0 32.0 — — p. 6) listed as a distinguishing feature of the adult cra- A.L. 442-1 — 25.0 — — A.L. 444-2 12.0 41.0 75.0 54.7 nium ‘palate shallow, especially anteriorly; dental A.L. 486-1 11.2 33.0 — — arcade long, narrow and straight-sided’. Subsequent A.L. 822-1 14.0 31.0 63.0 49.2 discovery and analysis have confirmed this symplesio- Australopithecus africanus morphic feature set of the A. afarensis palate (Kimbel Sts. 5 18.0 35.7 65.3 54.7 et al. 2004). However, two specimens found since Sts. 53 — 32.0 54.0 59.3 1990 extend the range of variation in this species’ palatal Stw. 73 14.5 30.0 58.0 51.7 form. In both A.L. 417-1 and A.L. 822-1 the palate Australopithecus robustus is both very narrow and very deep: internal palate 2 SK 12 12.8 32.0 — — depth (both 14 mm at M ) is the greatest among SK 46 12.2 35.0 — — seven measureable Hadar specimens, while relative SK 48 15.5 — — — palate breadth (internal breadth at M2/length 100 ¼ SK 79 13.5 — — — ca 49%) is the lowest among five Hadar specimens SKW 11 15.0 34.6 60.0 57.7 and, indeed, among 11 of 12 australopith specimens Australopithecus boisei in our sample overall (table 5). KNM-ER 405 22.0 38.0 75.0 50.7 The very narrow palate of the smaller Hadar crania KNM-ER 406 20.0 37.4 70.0 53.4 is potentially related to the narrow cranial base in these OH 5 21.0 38.2 79.1 48.3 A. afarensis individuals. Recall that the base (measured KNM-CH 1 — 40.8 72.0 56.7 between the entoglenoid processes) of A.L. 822-1 is as relatively narrow as in gorillas. Cranial base width cannot be measured for A.L. 417-1, but in A.L. 58-22, which (as described in the previous section) where it can be seen that many other australopiths has a cranial base width approximately as small as that resemble the morphology Tobias (1967) described of A.L. 822-1, estimated palate breadth (ca 27 mm, at for OH 5. A notable exception is A.L. 822-1, which, M2) is the second smallest in the A. afarensis sample in relative terms (note that all specimens in the (palate length for A.L. 58-22 cannot be estimated). figure are scaled to the biorbital breadth of this The relationship between the narrowness of the palate Hadar cranium), has a very narrow cranial base. As and the narrowness of the cranial base would appear measured across the entoglenoid processes (the recon- to hold, albeit on limited available evidence. structed positions of which are validated by the This relationship is explored further in the context bicondylar breadth of the specimen’s mandible), the of African great ape comparative data in figure 7. cranial base of A.L. 822-1 is as narrow as average for Among the apes there is a strong correlation (r2 ¼ our sample of female gorillas, and narrower than in 0.52, p , 0.0001) between absolute values of palate any other of the figured australopith specimens, breadth and bi-entoglenoid breadth, which appears including Sts 5 and the A.L. 444-2 cranium of A. afar- largely to be a function of strong size-dimorphism in ensis, in which the mandibular fossae are spread far gorillas (figure 7). Using size-standardized variables apart on the base. (with biorbital breadth as the standard), the corre- Another Hadar specimen, A.L. 58-22, appears lation is much weaker (because male gorillas no similar to A.L. 822-1 in its narrow cranial base. This longer stand out; r2 ¼ 0.14, p , 0.017). However, in specimen is a craniofacial fragment with part of the both cases the smaller A. afarensis individuals are the right posterior maxilla, sphenoid and temporal bone; most apelike of the small fossil hominin sample in the vomer establishes the midline (Kimbel et al. their combination of narrow palates and narrow cranial 1982). As with A.L. 822-1, the bi-entoglenoid dis- bases, with A. boisei and even A. africanus specimens tance (60 mm) is small compared with the larger highly divergent (indeed, humanlike, though humans Hadar crania, by at least 20 mm (table 4). This absol- were not included in our analysis) in their broader ute difference may reflect sexual dimorphism in cranial cranial bases. base dimensions in A. afarensis, as both of these Hadar Of considerable interest is the position of A.L. specimens also have abbreviated mandibular fossa 444-2, the large male A. afarensis skull. Compared

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

A. afarensis cranial base W. H. Kimbel & Y. Rak 3373 with the smaller specimens, it has a much broader sample of A. boisei hints at the same pattern of cranial cranial base for its palate breadth than predicted base breadth dimorphism, while the occipital of the by either great ape regression, which hints at an L338y-6 calotte (Shungura Formation, Member E), unusual—but perhaps not exceptional—pattern of interpreted by Rak & Howell (1978) as a immature sexual dimorphism in A. afarensis. Note that there male of this species, has a very steep nuchal plane, as are no small A. boisei specimens in the sample for is also observed in the young Hadar males. These which both palate and cranial base breadths can be observations convince us that the variation in Hadar measured (only OH 5 and KNM-ER 406 preserve occipital and cranial base form, though phylogeneti- both dimensions). However, the bi-entoglenoid cally informative, is intraspecific. A similar case has breadth of KNM-ER 407, considered by consensus a previously been made for the polymorphic lower female A. boisei calvaria, is less than 10 mm wider third premolar in A. afarensis (Kimbel et al. 2004, than that of the two A. afarensis females (see y-axis 2006). in figure 7), and if we grant this individual a palate Finally, the expanded cranial sample of A. afarensis breadth somewhere between, say, those of A.L. highlights the strongly mosaic nature of basicranial 822-1 and Sts 5 (ca 31–36 mm), then the presumptive evolution in the hominin clade. Evolutionary changes female-to-male trend in the cranial base versus in the cranial base and occipital squama that are palate breadth relationship for A. boisei would be thought to unite early hominins with modern very similar to that of A. afarensis. That is, compared humans (and which, for example, have raised suspi- with extant African apes (and other anthropoid cions of pervasive homoplasy in the crania of robust species; M. Spencer 2010, personal communication), australopiths and Homo), were still not completed by males have a much wider cranial base for their palate the time of A. afarensis, ca 3.5–3.0 Ma. Thus, width than females (see also the suggestive position although an anteriorly located foramen magnum and of other large A. boisei and A. aethiopicus specimens a short basioccipital segment are shared with extant on the y-axis of figure 7). (Unfortunately, the fossil humans, the narrow cranial base, posteriorly inclined record does not permit us to extract any more foramen magnum, high nuchal lines and concomi- information from the size-standardized data.) This tantly steep nuchal plane, are apelike characteristics suggests a unique pattern of cranial sexual dimorphism that are inferred to have been commonly, though not among australopith species. universally, expressed in this taxon. Upright posture and a large brain are the most commonly invoked influences on the cranial base 4. DISCUSSION morphology of modern humans (see Lieberman et al. The A.L. 822-1 skull focuses attention on several 2000, for a review). According to Le Gros Clark aspects of adult cranial base morphology that were (1947), Robinson (1958) and Olson (1981), among not previously well understood for A. afarensis. others, the descent of the nuchal musculature and First, this Hadar specimen presents a particularly the rotation of the nuchal plane to a horizontal pos- primitive basicranial profile. Its relatively narrow cra- ition beneath the brain case mirrors the adoption of nial base, high nuchal lines, and correspondingly upright posture and bipedal locomotion in the homi- steep nuchal plane are more similar to African great nin clade. Biegert (1957; also Weidenreich 1941), in ape conditions compared with other australopiths so contrast, argued that the architectural remodelling of far known. While other more fragmentary A. afarensis the hominin posterior calvaria was a by-product of cer- specimens hint at relatively generalized occipital form, ebral expansion, which introduced a strongly flexed A.L. 822-1 places this morphology within the context basicranial axis and a ‘rolling up’ of the braincase of the entire skull for the first time. that impelled the foramen magnum and occipital con- Second, A.L. 822-1 points to a pattern of cranial dyles forward. After casting doubt on the oft-proposed sexual dimorphism neither recognized previously correlation among foramen magnum orientation, among the australopiths nor encountered among occipital condyle position and mode of locomotion extant hominoids. The apelike cranial base pro- in primates, Biegert (1963) pointed to the horizontal portions and nuchal area form are already presented nuchal plane and anterior foramen magnum of in derived conditions in larger A. afarensis specimens A. africanus (Sts 5) as evidence of an initial phase of usually considered to be males (A.L. 333-45, A.L. encephalization in hominin evolution. Robinson 444-2). In these crania the basicranium is wider (1958), citing the case of the ‘short-faced squirrel (absolutely and size-standardized) and the position of monkey’ (Saimiri ), noted that the orientation of the the nuchal lines approximates FH, with a horizontal nuchal plane (steep) and the position of the occipital nuchal plane, differences that raise the question of condyles and foramen magnum on the cranial base whether species-level taxonomic distinction between (anterior) are not necessarily related, which recalls the two morphs is warranted. Two further points the situation in A. afarensis. The review by Lieberman argue otherwise. First, the combination of high et al. (2000) concluded that the orientation of the fora- nuchal lines and a steep nuchal plane in two men magnum (as distinct from its anteroposterior large, immature occipital specimens (A.L. 439-1, position) is unrelated to the posture of the head on A.L. 444-1) suggests that this dimorphism in the the vertebral column but, with cranial base flexion, is Hadar cranial sample has an ontogenetic basis, with primarily a reflection of brain size relative to cranial young males ‘passing through’ the final adult form of base length. the smaller females to reach mature male (and more The skulls of A. afarensis bear on these issues. This derived) morphology. Second, the adult cranial species is demonstrably an upright biped with a mean

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3374 W. H. Kimbel & Y. Rak A. afarensis cranial base endocranial volume (ca 450 cc) slightly larger than the cube root of endocranial volume as a percentage that of Pan troglodytes, to which, among the living of cranial base length [ba-sella þ sella-fc], is roughly hominoid taxa, it is probably closest in body size. 0.94 in A.L. 822-1, which is smaller than values for The marked variation in the height of the nuchal fossil and extant Homo but in the zone of overlap for muscle insertions and angulation of the nuchal plane the small sample of australopiths and extant apes in A. afarensis would seem to negate a major role for measured from CT scans by Spoor 1997.) Similarly, upright locomotion in dictating morphological vari- we can only estimate the (internal) flexion of the cra- ation in this region of the hominin skull. One caveat nial base in A. afarensis, using the preserved is that we do not have a good idea about how the morphology of A.L. 822-1 (which includes a marked head was held on the cervical vertebral column in superior deflection of the external basioccipital surface Australopithecus; while the anterior position of the at basion) with support from the other more fragmen- occipital condyles suggests a head posture more simi- tary specimens mentioned above; for A.L. 822-1 the lar to that of modern humans than apes, the slightly angle (ba-se/se-fc; CBA1 of Lieberman et al. 2000)is posterior orientation of the foramen magnum, the approximately 1428 (+58), which is some 15–208 strongly angled atlanto-occipital articular surfaces more flexion than great ape species’ means and close (on A.L. 333-45, A.L. 822-1, and the A.L. 333-83 to the modern human mean. This Hadar skull has a first cervical vertebra), and the long, straight and more highly flexed cranial base than extant African robust spine of the lower cervical vertebra (C6, A.L. apes of similar relative brain size (details in prep- 333-106), may be signs of a mechanical environment aration; see also Spoor 1997; Lieberman et al. 2000, dissimilar to that of the modern human cranioverteb- for comparative data). ral interface. We do not know the extent of cervical Although the influence of body posture on primate lordosis in these early hominins, but it would not cranial base form has been discounted by recent surprise us to find a less lordotic cervical column in research (see Lieberman et al. 2000), we see the A. afarensis than in modern humans. anterior position of the foramen magnum and occipital Obviously, whether measured absolutely or relative condyles as the major cranial base distinction between to estimated body mass, brain size in A. afarensis is Australopithecus and Homo, on one hand, and the great much closer to that of apes than modern humans. apes on the other. The fact that this distinction maps This indicates that the humanlike anterior position of onto primary locomotor differences speaks to upright the foramen magnum is largely, if not entirely, unre- posture in hominins as an important factor in the lated to overall brain size. Perhaps, though, relative positional shifts of these cranial base structures. In neocortical (cerebral) expansion is responsible for our view (see also Spoor 1997; Kirk & Russo 2010), the forward migration of the foramen. In this context, the adoption of upright posture (though not necessarily the (by now) clear evidence of a relatively posterior striding bipedal locomotion per se) in hominins led to a position of the lunate sulcus on earliest australopith forward migration of the foramen magnum/occipital brain endocasts (Holloway et al. 2004) can be seen condyles and a shortened cranial base.3 The combi- as a sign of relative cerebral expansion, which, poster- nation of a small, ape-sized brain on a relatively short iorly, is manifested as a bulging of the occipital poles base introduced the flexion of the basicranial axis. over the cerebellar lobes, and, perhaps, a forward Thus, despite their small brains, the anterior position ‘rotation’ of the cranial base (similar to what Biegert of the foramen magnum (basion) in Australopithecus 1963 hypothesized). In the A. afarensis brain endocast was associated with the relatively short, flexed cranial the absolute and relative size of the cerebellar lobes base typical of modern humans by ca 3Ma(figure 8). (and especially their anteroposterior length) is much Subsequent changes in the orientation of the foramen smaller than in African great apes, in which the cer- magnum (anteroinferior-facing) in the Homo clade are ebral and cerebellar lobes protrude subequally probably linked to an increase in endocranial volume (Holloway & Yuan 2004). However, it is unclear to and the consequent descent of opisthion beneath the what extent the form of the braincase beneath the braincase, as discussed above (see Kimbel et al.2004 tentorium cerebelli would be affected morphogenetically for details). The introduction of the cervical vertebral by enlargement of the cerebrum posteriorly; this is lordosis may also play a role in this change, but the an area in need of further research. fossil record is currently mute on the timing of the The ape-sized brain of A. afarensis rests on a base initial appearance of this innovation. with a much shorter basi-occipital segment than in If we consider the potential link between cranial any great ape. As shown by Lieberman et al. (2000; base configuration and facial orientation (e.g. Ross & see also Spoor 1997; McCarthy 2001), brain size rela- Ravosa 1993; Lieberman et al. 2000), the short, tive to cranial base length explains a fairly large flexed base may account for the relatively vertical mid- amount (but not all) of the observed variation in cra- facial segment in A. afarensis, which creates an overall nial base angle across a wide spectrum of anthropoid less prognathic facial profile than in chimpanzees and primates. We do not know the exact length of the gorillas (Kimbel et al. 2004: fig. 3.24, where midfacial anterior cranial base (sella turcica to foramen prognathism is measured as the angle between a line caecum) in A. afarensis, but evidence from fragmen- connecting nasion or sellion to nasospinale and the tary specimens (i.e. A.L. 58-22, A.L. 417-1) and the FH). In the great apes pronounced midfacial demonstrably short segment between basion and the prognathism—a dorsal deflection of the nasion- rear of the palate indicate a total cranial base length nasospinale segment in relation to a weakly flexed less than an ape of comparable brain size. (The cranial base—results in an airorynch facial configur- index of relative encephalization 1, which expresses ation. The contrasting klinorynch condition is a

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

A. afarensis cranial base W. H. Kimbel & Y. Rak 3375

A. afarensis, reveals a particularly generalized pattern of morphology in occipital squama and cranial base. With superiorly extended nuchal lines, a concomi- na tantly steep nuchal plane, and a narrow cranial base, this skull is notably more apelike than other australo- fc se pith crania, including those of larger, clearly male, individuals of this species, which share derived states 157° O of these characters with subsequent australopith species and Homo. It throws into sharp relief the ba morphology of previously known fragmentary cranial specimens from the Hadar site, demonstrating that ns the generalized pattern is probably common in small pr (female) individuals of this species. Qualitative data on a cross-sectional cranial growth series contained gorilla within the Hadar sample suggest that some of the variation in A. afarensis is indeed intraspecific because young adult males resemble mature females more than they do older males in their generalized occipital form. na Although the inference is limited by the few data fc available, these observations suggest for A. afarensis a se pattern of cranial sexual dimorphism unknown 142° among extant hominoids, with adult males characterized by relatively wider cranial bases and more O ns horizontal nuchal muscle origin surfaces than females. ba At least for the cranial base width, this dimorphism pr may apply to A. boisei as well; no other australopith taxon currently permits comparison. Morphological variation in the cranial base of A. afarensis is consistent with a mosaic pattern of evolutionary change in this region of the skull. Early australopiths were upright bipeds with small brains. The anterior position of the foramen magnum and occi- A.L. 822-1 pital condyles on a short (though not necessarily broad), flexed cranial base—a derived character complex linked plausibly to upright posture—is unrelated to substantial variation in the morphology of the nuchal region of the calvaria, which is often thought to mirror postural differences. In species potentially fc descendant from A. afarensis the nuchal region se na became more uniformly humanlike in form and orientation, the functional basis for which remains poorly understood. A derived, relatively upright midfacial pro- 137° ns file, which likewise ties A. afarensis to later australopiths ba and Homo, may be related to these cranial base modifi- O pr cations and thus, indirectly, to upright posture itself. H. sapiens We thank the Authority for Research and Conservation of Figure 8. Midsagittal craniograms of A.L. 822-1, female gor- Cultural Heritage and the National Museum of Ethiopia, illa, and modern human, showing cranial base and midfacial Ethiopian Ministry of Culture and Tourism, and the Afar angles. A.L. 822-1 anterior cranial base (sella-foramen Regional State government for permission to conduct the caecum) reconstructed with additional information from field (Hadar) and laboratory (Addis Ababa) research. We A.L. 417-1 and A.L. 58-22. See text for discussion. are grateful to the (US) National Science Foundation for grants supporting the field and lab work. Thanks are due Dr Mark Spencer for discussion and Mr Lucas Delezene hallmark of the human facial skeleton already mani- for help with collecting the comparative data. fested, at least incipiently, in A. afarensis (figure 8). The prognathic maxilla, with its strongly inclined nasoalveolar clivus, would not be expected to be ENDNOTES impacted as directly by cranial base form, and this 1Weidenreich chose opisthion rather than basion for this purpose remains the most apelike aspect of the A. afarensis face. probably because none of the Homo erectus crania he described preserves the anterior margin of the foramen magnum. 2Here, cranial base length is taken as the combined length of the segments basion to sella turcica to foramen cecum. Width of the 5. CONCLUSIONS base is measured between the summits of the entoglenoid processes. The A.L. 822-1 specimen, providing the first view 3As noted by Schultz (1955), in ontogenetic context the hominin of the complete small, presumptively female skull of foramen magnum does not migrate anteriorly; from the relatively

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3376 W. H. Kimbel & Y. Rak A. afarensis cranial base anterior position common to all juvenile hominoids, the foramen Le Gros Clark, W. E. 1947 Observations on the fails to shift posteriorly with growth of the cranium, as it does in anatomy of the fossil Australopithecinae. J. Anat. 81, all great apes. 300–333. Le Gros Clark, W. E. 1950 New palaeontological evidence bearing on the evolution of the Hominoidea. Quart. J. Zool. Soc. 105, 225–264. REFERENCES Lieberman, D. E., Ross, C. F. & Ravosa, M. J. 2000 The pri- Biegert, J. 1957 Der Formwandel des Primaten-Scha¨dels mate cranial base: ontogeny, function and integration. und seine Beziehungen zur ontogenetischen Entwicklung Yrbk. Phys. Anthropol. 42, 117–169. und den phylogenetischen Specialization der Kopforgane. McCarthy, R. C. 2001 Anthropoid cranial base architecture Morph. Jb. 98, 77–199. and scaling relationships. J. Hum. Evol. 40, 41–66. Biegert, J. 1963 The evaluation of characters of the skull, (doi:10.1006/jhev.2000.0446) hands and feet for primate taxonomy. In Classification Olson, T. R. 1981 Basicranial morphology of the extant and human evolution (ed. S. L. Washburn), pp. 116–145. hominoids and Pliocene hominids: the new material Chicago, IL: Aldine de Gruyter. from the Hadar Formation, Ethiopia and its significance Dean, M. C. & Wood, B. A. 1982 Basicranial anatomy of in early human evolution and taxonomy. In Aspects of Plio-Pleistocene hominids from East and South Africa. human evolution. (ed. C. B. Stringer), pp. 99–128. Am. J. Phys. Anthropol. 59, 157–174. (doi:10.1002/ajpa. London, UK: Taylor and Francis. 1330590206) Rak, Y. 1983 The australopithecine face. New York, NY: Holloway, R. L. & Yuan, M. S. 2004 Endocranial mor- Academic Press. phology of A.L. 444-2. In The skull of Australopithecus Rak, Y. & Howell, F. C. 1978 Cranium of a juvenile afarensis (eds W. H. Kimbel, Y. Rak & D. C. Johanson), Australopithecus boisei from the Lower Omo Basin, pp. 123–135. Oxford, UK: Oxford University Press. Ethiopia. Am. J. Phys. Anthropol. 48, 345–366. (doi:10. Holloway, R. L., Clarke, R. J. & Tobias, P. V. 2004 1002/ajpa.1330480311) Posterior lunate sulcus in Australopithecus africanus:Was Robinson, J. T. 1958 Cranial cresting patterns and their Dart right? C. R. Palevol. 3, 287–293. (doi:10.1016/j.crpv. significance in the Hominoidea. Am. J. Phys. Anthropol. 2003.09.030) 16, 397–428. (doi:10.1002/ajpa.1330160403) Johanson, D. C., White, T. D. & Coppens, Y. 1978 A new Ross, C. F. & Ravosa, M. J. 1993 Basicranial flexion, relative species of the genus Australopithecus (Primates: brain size, and facial kyphosis in nonhuman primates. Hominidae) from the Pliocene of eastern Africa. Am. J. Phys. Anthropol. 91, 305–324. (doi:10.1002/ajpa. Kirtlandia 28, 1–14. 1330910306) Kimbel, W. H. 1988 Identification of a partial cranium of Schultz, A. H. 1955 The position of the occipital condyles Australopithecus afarensis from the Koobi Fora Formation, and of the face relative to the skull base in primates. Kenya. J. Hum. Evol. 17, 647–656. (doi:10.1016/0047- Am. J. Phys. Anthropol. 13, 97–120. (doi:10.1002/ajpa. 2484(88)90022-X) 1330130108) Kimbel, W. H. & Delezene, L. K. 2009 ‘Lucy’ redux: a Spoor, C. F. 1997 Basicranial architecture and relative review of research on Australopithecus afarensis. Yrbk. brain size of Sts 5 (Australopithecus africanus) and other Phys. Anthropol. 52, 2–48. (doi:10.1002/ajpa.21183) Plio-Pleistocene hominids. S. Afr. J. Sci. 93, 182–186. Kimbel, W. H., Johanson, D. C. & Coppens, Y. 1982 Strait, D. S. & Grine, F. E. 2004 Inferring hominoid and Pliocene hominid cranial remains from the Hadar early hominid phylogeny using craniodental characters: Formation, Ethiopia. Am. J. Phys. Anthropol. 57, 453– the role of fossil taxa. J. Hum. Evol. 47, 399–452. 499. (doi:10.1002/ajpa.1330570404) (doi:10.1016/j.jhevol.2004.08.008) Kimbel, W. H., White, T. D. & Johanson, D. C. 1984 Cranial Tobias, P. V. 1967 The Cranium and Maxillary Dentition of morphology of Australopithecus afarensis: a comparative Australopithecus (Zinjanthropus)nboisei.I Olduvai Gorge, study based on a composite reconstruction of the adult vol. 2. London, UK: Cambridge University Press. skull. Am. J. Phys. Anthropol. 64, 337–388. (doi:10. Weidenreich, F. 1941 The brain and its roˆle in the phylo- 1002/ajpa.1330640403) genetic transformation of the human skull. Trans. Am. Kimbel, W. H., Johanson, D. C. & Rak, Y. 1994 The first Phil. Soc. 31, 320–442. (doi:10.2307/1005610) skull and other new discoveries of Australopithecus Weidenreich, F. 1943 The skull of Sinanthropus pekinensis. afarensis at Hadar, Ethiopia. Nature 368, 449–451. Palaenotol. Sin. New Ser. D10, l–484. (doi:10.1038/368449a0) White, T. D., Johanson, D. C. & Kimbel, W. H. 1981 Kimbel, W. H., Rak, Y. & Johanson, D. C. 2004 The skull of Australopithecus africanus: its phyletic position reconsidered. Australopithecus afarensis. Oxford, UK: Oxford University S. Afr. J. Sci. 77, 445–470. Press. White, T. D., Suwa, G., Hart, W. K., Walter, R. C., Kimbel, W. H., Lockwood, C. A., Ward, C. V., Leakey, WoldeGabriel, G., de Heinzelin, J., Clark, J. D., Asfaw, M. G., Rak, Y. & Johanson, D. C. 2006 Was Australopithecus B. & Vrba, E. 1993 New discoveries of Australopithecus anamensis ancestral to A. afarensis? A case of anagenesis in at Maka in Ethiopia. Nature 366, 261–265. (doi:10. the hominin fossil record. J. Hum. Evol. 51, 134–152. 1038/366261a0) (doi:10.1016/j.jhevol.2006.02.003) White, T. D., Suwa, G., Simpson, S. & Asfaw, B. 2000 Kirk, E. G. & Russo, G. A. 2010 Forward shift of the fora- Jaws and teeth of Australopithecus afarensis from Maka, men magnum in humans and other bipedal mammals. Middle Awash, Ethiopia. Am. J. Phys. Anthropol. 111, Am. J. Phys. Anthropol. Suppl. 50, 142–143. (doi:10. 45–68. (doi:10.1002/(SICI)1096-8644(200001)111: 1002/ajpa.21273) 1,45::AID-AJPA4.3.0.CO;2-I)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Phil. Trans. R. Soc. B (2010) 365, 3377–3388 doi:10.1098/rstb.2010.0042

Hominin diversity in the Middle Pliocene of eastern Africa: the maxilla of KNM-WT 40000 Fred Spoor1,2,*, Meave G. Leakey3,4 and Louise N. Leakey3,5 1Department of Human Evolution, Max Planck Institute for Evolutionary Anthropology, Deutscher Platz 6, 04103 Leipzig, Germany 2Department of Cell and Developmental Biology, University College London, UK 3Koobi Fora Research Project, Nairobi, Kenya 4Department of Anatomical Sciences, and 5Department of Anthropology, Stony Brook University, New York, USA The 3.5-Myr-old hominin cranium KNM-WT 40000 from Lomekwi, west of Lake Turkana, has been assigned to a new hominin genus and species, Kenyanthropus platyops, on the basis of a unique combination of derived facial and primitive neurocranial features. Central to the diagnosis of K. platyops is the morphology of the maxilla, characterized by a ﬂat and relatively orthognathic subnasal region, anteriorly placed zygomatic processes and small molars. To study this morphology in more detail, we compare the maxillae of African Plio-Pleistocene hominin fossils and samples of modern humans, chimpanzees and gorillas, using conventional and geometric morphometric methods. Computed tomography scans and detailed preparation of the KNM-WT 40000 maxilla enable comprehensive assessment of post-mortem changes, so that landmark data characterizing the morphology can be corrected for distortion. Based on a substantially larger comparative sample than previously available, the results of statistical analyses show that KNM-WT 40000 is indeed signiﬁcantly different from and falls outside the known range of variation of species of Australopithecus and Paranthropus, contemporary Australopithecus afarensis in particular. These results support the attribution of KNM-WT 40000 to a separate species and the notion that hominin taxonomic diversity in Africa extends back well into the Middle Pliocene. Keywords: human evolution; Pliocene; Africa; Kenyanthropus platyops; maxilla; geometric morphometrics

1. INTRODUCTION these fossils is considered to be within the range of Whether or not the Pliocene hominin fossil record variation of A. afarensis (White et al. 2000; Ward from Hadar (Ethiopia) and Laetoli (Tanzania) rep- et al. 2001; Kimbel 2007; Wood & Lonergan 2008). resents more than one species was the subject of However, a detailed analysis of symphyseal shape of ongoing debate in the 1980s (see Boaz 1988 for a both the type specimen and a previously unpublished review). Recovery of additional fossils and studies of second mandible supports a separate specific status intraspecific variation and temporal trends have sub- (Guy et al. 2008). sequently resulted in a broad consensus supporting A second site providing possible evidence for the interpretation of a single, sexually dimorphic species diversity is Lomekwi, west of Lake Turkana species, Australopithecus afarensis (Lockwood et al. (Kenya). Fieldwork in the early 1980s and late 1990s 1996, 2000; Kimbel et al. 2004; Kimbel & Delezene resulted in hominin discoveries dated between 3.5 2009). However, fossils found at two other sites have and 3.2 Ma, including a well-preserved temporal reopened the debate of species diversity in the African bone, 2 partial maxillae, 3 partial mandibles, 44 Middle Pliocene. A partial mandible and upper pre- isolated teeth and a largely complete although dis- molar, discovered in 1994 in the Koro-Toro area of torted cranium, KNM-WT 40000 (Brown et al. Chad and approximately 3.5 Myr old, have been 2001; Leakey et al. 2001). Several of these specimens assigned to a new species, Australopithecus bahrelghazali were found to show a morphology markedly different (Brunet et al. 1995, 1996). This attribution has been from that of contemporary A. afarensis (Leakey et al. questioned as the limited morphology preserved by 2001). Accordingly, the cranium and one fragmentary maxilla were assigned to a new genus and species, Kenyanthropus platyops, based on a unique combi- * Author for correspondence ([email protected]). nation of derived facial and primitive neurocranial Electronic supplementary material is available at http://dx.doi.org/ features (Leakey et al. 2001). A number of recent 10.1098/rstb.2010.0042 or via http://rstb.royalsocietypublishing.org. studies and reviews have cautiously considered One contribution of 14 to a Discussion Meeting Issue ‘The first four K. platyops as a valid taxon (Strait & Grine 2004; million years of human evolution’. Kimbel 2007; Cobb 2008; Nevell & Wood 2008;

3377 This journal is q 2010 The Royal Society Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3378 F. Spoor et al. The maxilla of KNM-WT 40000

Wood & Lonergan 2008). On the other hand, White A.L. 486-1, Sts 52, OH 5, SK 13 and SKW 11, (2003) questioned the taxon’s validity, and the which are subadults (third molars not in occlusion). notion of Pliocene hominin diversity. In his view it The intraspecific variation in maxillary shape cannot be excluded that KNM-WT 40000 is an among the fossils was examined by making compari- early Kenyan variant of A. afarensis, given the distor- sons with samples of modern humans and African tion of the specimen and the known cranial variation great apes. The modern human sample consists of in early hominin species and among modern apes 55 specimens (sex mostly unknown), representing and humans. indigenous populations from all six widely inhabited Central to the diagnosis of K. platyops is the mor- continents, housed at the Natural History Museum phology of the maxilla, characterized by a flat and (London) and at the Department of Cell and Develop- relatively orthognathic subnasal region, an anteriorly mental Biology at University College London. The placed zygomatic process and small molars. In the pre- African ape samples (all non-captive) include 50 sent study, this morphology, as shown by KNM-WT specimens of the eastern lowland gorilla (Gorilla gorilla 40000, is investigated in more detail. We made quan- gorilla; 26 males, 24 females) and 61 specimens of titative comparisons, using geometric morphometric chimpanzee (Pan troglodytes, all subspecies rep- and univariate methods, with Plio-Pleistocene homi- resented; 28 males, 33 females) from the collections nin fossils from Ethiopia, Kenya, Tanzania and of the Powell Cotton Museum (Birchington), the South Africa and with samples of modern humans, Royal College of Surgeons (London), the Natural gorillas and chimpanzees. Such analyses are obviously History Museum (London) and the Department of affected by the post-mortem distortion of KNM-WT Cell and Developmental Biology at University College 40000, the reason the initial description provided London. Specimens are adult and lack signs of limited metric comparisons only (Leakey et al. 2001). substantial pathology, of the alveolar process of the The impact of the preservation of the maxilla was maxilla in particular. therefore evaluated in detail using new evidence CT was used to examine internal morphology based on additional preparation of the specimen and and record surface landmarks of some of the fossil computed tomography (CT). The information thus specimens. KNM-WT 40000 was scanned with a obtained was used to adjust landmarks representing Siemens AR.SP medical scanner at the Diagnostic the key morphological features. In statistical analyses Center, Nairobi (Kenya). Scans in sequential mode the morphology of the specimen is considered both were made in the transverse plane, parallel to the in its preserved form and corrected for distortion. postcanine alveolar margin, using a slice thickness Using a substantially larger comparative sample and increment of 1.0 mm. Images were reconstructed than available to Leakey et al. (2001), the present with a SP90 kernel, extended CT-scale and a study aims to assess two specific hypotheses. 0.17 mm pixel size. The tooth roots were segmented by Kornelius Kupczik and the first author. New CT 1. KNM-WT 40000 does not differ significantly from data with improved spatial resolution (isotropic voxel A. afarensis with respect to the morphological fea- size 0.069 mm) were obtained more recently with the tures of the maxilla included in the differential portable BIR ACTIS 225/300 high-resolution diagnosis of K. platyops (Leakey et al. 2001). industrial CT scanner of the Department of Human Rejection of this null hypothesis would provide Evolution at the Max Planck Institute for Evolutionary evidence for species diversity in eastern Africa at Anthropology (Leipzig, Germany), at the time around 3.5 Ma. installed at the National Museums of Kenya in 2. KNM-WT 40000 does not differ significantly from Nairobi. Other CT data of hominin fossils used here species of Australopithecus and Paranthropus with are from the digital archives of the National Museums respect to the morphological features of the maxilla of Kenya, and the Department of Anthropology, included in the differential diagnosis of K. platyops University of Vienna. Visualization of the CT datasets (Leakey et al. 2001). Rejection of this null was done using AMIRA 4.1.2 (Mercury Computer hypothesis would support the attribution of Systems). KNM-WT 40000 to a separate species. A set of maxillary landmarks was selected using three criteria: (i) they should quantify the features used in the differential diagnosis of K. platyops; (ii) it 2. MATERIAL AND METHODS should be possible to take these landmarks from the In geometric morphometric shape analyses the KNM- KNM-WT 40000 maxilla and correct them for distor- WT 40000 maxilla was compared with all available tion of the specimen; and (iii) the number of fossil hominin specimens attributed to Australopithecus and specimens used in the comparative sample should be Paranthropus that preserve the morphology concerned. as large as possible. Optimizing all three criteria These are: Australopithecus anamensis (KNM-KP resulted in the selection of five two-dimensional land- 29283), A. afarensis (A.L. 199-1, A.L. 200-1, A.L. marks, taken from the specimens projected in lateral 417-1, A.L. 427-1, A.L. 444-2 and A.L. 486-1), view: nasospinale (ns), prosthion (pr), the buccal Australopithecus africanus (MLD 9, Sts 52, Sts 71 and alveolar margin between the canine and third premolar Stw 498), Australopithecus garhi (BOU-VP-12/130), (pc), the buccal alveolar margin between the second Paranthropus aethiopicus (KNM-WT 17000), Paran- and third molar (m23) and the antero-inferior take- thropus boisei (OH 5) and Paranthropus robustus off of the zygomatic process (azp), a point most (SK 11, SK 12, SK 13, SK 46, SK 48, SK 83 anterior, inferior and medial on the root of the process and SKW 11). All are adults, with the exception of (figure 1). These landmarks quantify the orientation of

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

The maxilla of KNM-WT 40000 F. Spoor et al. 3379

(a) shifts along the crack. Moreover, the width of the ns azp cracks along specific trajectories linking the five landmarks (figure 2a,d) were measured to the nearest tenth of a millimetre with digital callipers, making sure that the distance was taken between matching pr edges. The midsagittal surface area above prosthion is not well preserved, and the expansion of the subnasal clivus in the sagittal plane was examined more laterally along a line from the I2/C interalveolar septum to the left lower corner of the nasal aperture pc m23 (figure 2a). The crack widths were used to calculate the percentage of surface expansion between the five (b) landmarks, and the x and y coordinates were adjusted accordingly. Both preparation and measurements were done under a binocular microscope, using acetone to temporarily enhance the difference between bone and matrix. Generalized procrustes analyses of the landmark coordinates and principal component analyses (PCAs) of the output were performed with MORPHOLOGIKA 2.5 (O’Higgins & Jones 1998). With this software, the maxillary shape variation associated with each prin- Figure 1. CT-based parallel-projected 3D reconstructions cipal component (PC) can be explored visually by comparing the maxillae in lateral view of (a) A.L. 200-1 morphing a wireframe of the five landmarks according (reversed right side of cast, Australopithecus afarensis) and to the position on a bivariate plot of two PCs. F-tests (b) KNM-WT 40000 (left side of original, Kenyanthropus and t-tests of the individual PC scores, with sequential platyops). The five landmarks are shown, together with the Bonferroni correction for multiplicity (Rice 1989), were connecting wire frame used in figure 3 (see text for the done using PAST 1.93 (Hammer et al.2001)and abbreviations of the landmarks). The broken surface of the MS EXCEL 2003. One-tailed distributions were used zygomatic process of KNM-WT 40000 facing laterally is in the t-tests when the hypotheses and the species not visualized to emphasize the outline compared with the diagnosis of K. platyops specifically state the nature equivalent morphology in A.L. 200-1. Scale bar, 10 mm. of a possible difference (e.g. KNM-WT 40000 is subnasally less prognathic than A. afarensis). Conse- the subnasal clivus in the midsagittal plane (ns-pr), quently, two-tailed distributions were used only for the anterior zygomatic process position along the post- the comparison of PCs related to zygomatic process canine tooth row (azp relative to pc-m23) and the position in KNM-WT 40000 and P. r o b u s t u s ,where degree of anterior projection and transverse flatness there is no prior prediction of the nature of the differ- of the subnasal clivus (pr–pc or sagittally projected ence. KNM-WT 40000 was also compared with length of the canine and incisor alveolar margin). species individually by combining all PCs obtained in The five landmarks were mostly recorded from digi- separate analyses of KNM-WT 40000 and each tal images of the specimens in exact lateral view and species. The Mahalanobis’ distance of KNM-WT taken with a focal distance of 1–2 m to minimize par- 40000 from the centroid of the species sample is allax artefacts. Nasospinale and prosthion may not be compared with the distances from that centroid of visible in this view and are indicated by markers (see the specimens in the sample (software written by Spoor et al. 2005 for details of this method). The land- Paul O’Higgins, Hull York Medical School, UK). marks of KNM-WT 40000, A.L. 444-2, Sts 52a, All statistical analyses were done separately for Sts 71, KNM-WT 17000 and OH 5 were taken from KNM-WT 40000 in its preserved form and corrected CT datasets, using AMIRA 4.1.2 (Mercury Computer for distortion. Systems) to obtain parallel-projected three dimen- A drawback of the landmark-based approach is that sional surface reconstructions in a lateral view and it limits the number of specimens that can be included. sagittal sections to locate nasospinale and prosthion. That is because each must preserve the full area The left side of the extant specimens was used, the covered by the landmarks, whereas less complete mean of both sides of A.L. 200-1 and OH 5 and the specimens may be informative regarding individual best preserved side of the other fossils (left for diagnostic aspects of the K. platyops maxilla. When KNM-WT 40000). All landmark coordinates were interpreting the main shape analysis, two maxillary recorded with IMAGEJ 1.42d (National Institutes of features were therefore considered individually as Health, USA). well, to assess consistency among a larger number of To examine the impact of the distortion of KNM- fossils than those preserving all five landmark WT 40000 on the landmark positions, additional locations. The subnasal clivus angle marks the orien- surface preparation was done of the anterior and tation in the sagittal plane of the segment lateral aspects of the left maxilla. Small remnants of nasospinale to prosthion relative to the postcanine the matrix on the bone surface were removed and alveolar margin, up to the M2/M3 septum. It was cracks fully exposed. Bone edges on either side of the measured using IMAGEJ from digital images or crack could thus be matched, identifying possible CT-based visualizations of a specimen in a lateral

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3380 F. Spoor et al. The maxilla of KNM-WT 40000

(a) (b)

pr (c) (e) M

(d) 6%

18%

20%

18%

16%

Figure 2. Distortion of KNM-WT 40000. (a) Anterior view, giving the midline (black line) to indicate the midface skewing, left nasospinale (ns), prosthion (pr) and the trajectory used to calculate the height expansion of the subnasal area (white line). (b) CT-based 3D reconstruction of the maxilla in superior view, showing the tooth roots inside the translucent bone. (c)A high-resolution sagittal CT image through the buccal roots of the left P3 to M2 (orientation indicated by the black line in (b)). The thin black lines mark a longer crack through the premolar roots. (d) Lateral view of the left maxilla, showing the pattern of matrix-filled cracks highlighted by wetting with acetone. The five landmarks are shown as in figure 1. The trajectories along which crack widths were measured are given by lines with associated percentages of expansion (black line refers to the subnasal trajectory shown in (a)). (e) The right M2 crown (M, mesial; B, buccal), with black lines marking the endpoint of cracks highlighted with acetone. The white lines indicate the match at the mesial end of the widest crack. The dark area on the mesiolingual corner is a strong shadow of the enamel more distally, rather than damage to the dentine. Scale bars, (a,b) 30 mm, (c,d) 10 mm and (e) 3 mm. view, where feasible estimating the postcanine alveolar (Tobias 1967; White 1977). Full preparation of the margin orientation if the exact position of landmarks M2 crown had been done at the time of the first pc or m23 is not preserved. Furthermore, the anterior announcement (Leakey et al. 2001), and measure- position of the zygomatic root is considered relative ments of cracks affecting the length and width had to the dental row, following Lockwood & Tobias been taken at the time. Comparative data of M2 (1999, table 7). These features were recorded by the size in Plio-Pleistocene hominins were taken from the authors from the original specimens, with additional literature, combined with our own measurements. observations regarding zygomatic root position in A. afarensis provided by William H. Kimbel (Arizona Sate University, USA). 3. PRESERVATION OF THE MAXILLA Finally, the crown size of the right M2 was assessed The facial parts of KNM-WT 40000 show post- on the basis of conventional mesiodistal and buccolin- mortem distortion in the form of lateral skewing of gual measurements, defined in two different ways the nasal area and a network of matrix-filled surface

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

The maxilla of KNM-WT 40000 F. Spoor et al. 3381

(a) SKW11* 0.15

0.10 Stw498 SK13* Sts71 KP29283 WT17k 0.05 BOUVP12/130 MLD9 OH5* Sts52* PC1 –0.1 0.1 AL486-1* SK83 WT40k SK12 AL200-1 –0.05 AL427-1

SK11 SK46 –0.10 AL199-1 AL417-1 AL444-2

PC2

(b) 0.06 WT40k

AL199-1 Stw498

SK12 AL200-1 0.02 SK13* SKW11*

AL427-1 SK11 PC4 –0.075 AL417-1 OH5* 0.05 Sts71 KP29283 SK83 MLD9 Sts52* AL444-2 AL486-1* SK46 BOUVP12/130 –0.04 WT17k

PC5

0.3 (c)

A. africanus 0.2 H. sapiens

P. troglodytes WT40k PC1 –0.6 –0.2

P. robustus G. gorilla –0.2 A. afarensis –0.3 PC2

Figure 3. Bivariate plots of PCs. (a) PC2 against PC1 and (b) PC5 against PC4 of the fossils samples. (c) PC2 against PC1 of the combined fossil and extant samples. KNM-WT 40000 (black dot) is corrected for distortion. The prefix KNM- of the Kenyan specimens is omitted, and an asterisk indicates subadults. Convex hulls are given for A. afarensis (solid line), A. africanus (dash-dot line) and P. robustus (dashed line), as well as in (c), the 95% confidence ellipses of these taxa and the extant species (solid line). The grey-shaded wire frames in (a,b) are defined in figure 1 and indicate the maxillary shapes represented at the extremes of the PC axes. See the main text for the percentage of variance represented by each PC.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3382 F. Spoor et al. The maxilla of KNM-WT 40000 cracks associated with clay-induced expansion mainly affected by occlusal wear as it has not reached the affecting the alveolar and zygomatic processes level of maximum bulging of the buccal and lingual (Leakey et al. 2001; figure 2a,d). Before assessing the surfaces. The recently acquired high-resolution CT degree of expansion in the landmarked part of the scans will provide the opportunity to prepare a full maxilla, it is worth considering whether this area three dimensional virtual reconstruction of the right shows any evidence of substantial post-mortem shape M2 crown. changes as well. A good indicator of structural integrity of the alveolar process is the internal preservation of the tooth roots. Those present, of the 4. MORPHOMETRIC COMPARISONS left I1 to M2 and the right C to M3, are visualized in The PCA of the hominin fossil sample described here a CT-based three dimensional reconstruction of the uses the corrected landmark configuration of KNM- maxilla in superior view (figure 2b). The roots are WT 40000 (table 1). The first six PCs account for well preserved, without distinct misalignments or dis- 99.9 per cent of the variation. Although PC3–PC6 tortion, and the dental arcade is symmetrical in contribute less than 10 per cent each, these were still shape. The only exception is the largely exposed assessed because KNM-WT 40000 could specifically right canine root which is in the correct position at differ from the other fossils in morphology that is the alveolar margin, but the apex is tilted anteriorly. less variable among the Australopithecus and Paranthro- A more detailed view of the internal preservation of pus specimens which dominate the sample. PC1, PC2, the left maxilla is provided by a high-resolution sagittal PC4 and PC5 were found to provide evidence in CT image through the buccal roots of the left P3 to M2 relation to the hypotheses examined here, and these (figure 2c). Several fine cracks through the roots can be are shown in bivariate plots (figure 3), with wireframes seen, and the mesiobuccal root canal of the M2 is marking the shapes represented at either end of each expanded, but this is not accompanied by substantial axis. PC3 and PC6 will be briefly described as well. displacement of the root parts on opposite sides PC1 (eigenvalue 0.0263; 72% of variance) rep- of the cracks. The mesiodistal distance between the resents the variation in anteroposterior position of P4,M1 and M2 is increased by matrix expansion, the anterior zygomatic process (landmark azp) and particularly of the alveolar space around the roots. the relative length and transverse flatness of the subna- However, this is less so between the P3 and P4. sal clivus (ns–pr and pr–pc, respectively). PC1 Overall, the internal CT evidence suggests a pattern separates Australopithecus species, with a more poster- of expansion without substantial shape changes due iorly positioned zygomatic and a shorter and to skewing or other directional deformation. The transversely curved (projecting) subnasal clivus, from well-preserved state of the premolar root area is of Paranthropus species, with a more anteriorly positioned particular importance as it indicates that the overlying zygomatic and a longer and transversely flat subnasal anterior zygomatic process position is unlikely to have clivus (figure 3a). Addressing hypothesis 1, the PC1 been altered by the distortion. This is further con- score of KNM-WT 40000 differs significantly from firmed by the absence of major shifts of surface bone that of A. afarensis (table 1), reflecting its more ante- fragments between the premolar alveolar margin and riorly placed zygomatic and a subnasal clivus that is the anterior zygomatic root. transversely flat. When compared with multiple species The percentages of bone expansion along the (hypothesis 2), the difference between KNM-WT measured trajectories vary from 16 per cent along the 40000 and A. afarensis is statistically significant, as is postcanine alveolar margin to 20 per cent transversely the difference from A. africanus. The PC1 scores over the canine jugum (figure 2d). The one exception suggest that KNM-WT 40000 is intermediate between is the area superior to the canine alveolus where the Australopithecus and Paranthropus with respect to this expansion in transverse direction is only 6 per cent. particular morphology. However, the difference from The right M2 crown is shown in figure 2e.The P. r o b u s t u s is not statistically significant, with the suba- widest crack runs from the mesial interstitial facet to dult SKW 11 having a score close to KNM-WT 40000. the distolingual corner, a second shorter one from PC2 (eigenvalue 0.0054; 15% of variance) rep- the central area of the main break to the lingual resents both the inferosuperior and anteroposterior crown margin and a third thin one from the central position of the zygomatic process and the length of area to the distal interstitial facet. Enamel and occlusal subnasal clivus. This PC separates A. afarensis, with dentine edges of the breaks provide good clues regard- a more inferoposteriorly positioned zygomatic and ing the match of the four parts they delineate. A refit of longer subnasal clivus, from A. africanus, with a the crown would require the two lingual parts to more anterosuperiorly positioned zygomatic and be moved buccally, the triangular distolingual part to shorter clivus (figure 3a). KNM-WT 40000 and be moved mesially and the mesiolingual part to be A. garhi are intermediate, A. anamensis falls with moved slightly distally. Closing the cracks would A. africanus, and Paranthropus specimens show the result in an estimated 1.2 mm reduction of the bucco- full range of PC2-related morphological variation. lingual width of the crown. The mesiodistal length KNM-WT 40000 is not significantly different from along the crown axis (White 1977) is not affected A. afarensis (hypothesis 1) or from other hominin by the cracks, and has only been corrected for an esti- species more in general (hypothesis 2). mated 0.5 mm of mesial interstitial wear. However, PC3 (eigenvalue 0.0029; 8% of variance) purely the maximum length (To bi as 1 96 7 ) requires correction represents the inferosuperior height of landmark azp, for the 0.7 mm distal displacement of the triangular the anterior zygomatic root position, above the postca- distolingual part. The buccolingual width is not nine alveolar margin. KNM-WT 40000 does not differ

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

The maxilla of KNM-WT 40000 F. Spoor et al. 3383

Table 1. PCs of the maxillary shape analysis. The landmarks of KNM-WT 40000 are corrected for distortion. The sample size (n), mean, minimum, maximum and standard deviation (s.d.) are given, and the comparisons of KNM-WT 40000 by t-test list the probability ( p; one-tailed, except þtwo-tailed) and the signiﬁcance (multiple comparisons after sequential Bonferroni correction). n.s., not signiﬁcant; *p , 0.05; **p , 0.01.

PC1 PC2 PC4 PC5

KNM-WT 40000 20.068 20.003 0.049 0.065 A. anamensis n ¼ 1 0.167 0.064 20.045 20.009 A. afarensis n 6666 mean 0.1285 20.0742 20.0012 0.0014 min. 0.053 20.129 20.024 20.030 max. 0.198 20.001 0.018 0.038 s.d. 0.0623 0.0498 0.0136 0.0262 A. garhi n ¼ 1 0.178 0.029 20.011 20.027 A. africanus n 4444 mean 0.1065 0.0558 0.0076 20.0004 min. 0.064 0.025 20.046 20.015 max. 0.152 0.092 0.043 0.030 s.d. 0.0401 0.0316 0.0382 0.0208 P.aethiopicus n¼1 20.152 0.068 20.010 20.040 P.robustus n 6666 mean 20.1879 0.0073 20.0040 0.0014 min. 20.259 20.070 20.087 20.024 max. 20.070 0.142 0.048 0.019 s.d. 0.0654 0.0855 0.0488 0.0169 P.boisei n ¼ 1 20.194 0.020 0.018 20.005 comparison with A. afarensis p 0.016 0.120 0.010 0.037 sign. * n.s. ** * comparison with Australopithecus and Paranthropus A. afarensis p 0.016 0.120 0.010 0.037 sign. * n.s. ** A. africanus p 0.015 0.098 0.204 0.034 sign. * n.s. n.s. n.s. P.robustus p 0.151þ 0.919þ 0.182 0.009 sign. n.s. n.s. n.s. *

significantly from A. afarensis, A. africanus or PC4 and PC5 jointly contribute to an overall pat- P. robustus, and this morphology does not relate to tern of differences in subnasal prognathism in the the hypotheses. fossil sample (figure 3b). KNM-WT 40000 is the PC4 (eigenvalue 0.0011; 3% of variance) represents most orthognathic, whereas the P.aethiopicus specimen the subnasal clivus orientation associated with vari- KNM-WT 17000, the A. garhi type BOU-VP-12/130 ations in the midline area only, as shown by the and the A. anamensis specimen KNM-KP 29283 angle of ns–pr to the entire alveolar margin (pr–pc– are the most prognathic. The other species of m23; figure 3b). KNM-WT 40000 has the highest Australopithecus and Paranthropus do not differ notably. score in the sample and is significantly less prognathic PC6 (eigenvalue 0.0003; 1% of variance) represents than A. afarensis (table 1). When compared with mul- the angle and length proportions between the alveolar tiple species, this difference from A. afarensis is margins of the anterior (canine–incisor) and postca- statistically significant as well, but differences from nine teeth. There is no separation between taxa, and A. africanus and P. robustus are not. KNM-WT 40000 does not differ significantly from PC5 (eigenvalue 0.0006; 2% of variance) represents A. afarensis, A. africanus and P. robustus. the subnasal clivus orientation associated with vari- Using the landmark configuration of KNM-WT ations of the entire subnasal area. The midline clivus 40000 as preserved rather than corrected for distortion (ns–pr) and the canine–incisor alveolar margin (pr– results in PC scores that are only marginally different pc) jointly vary in orientation relative to the postcanine from those reported in table 1. Significance levels of alveolar margin (pc–m23; figure 3b). KNM-WT the t-tests are the same as for the corrected landmarks, 40000 is by far the most orthognathic in the sample, except for multiple species comparisons of PC5 with and the difference from A. afarensis is statistically sig- A. afarensis and A africanus (electronic supplementary nificant (table 1, figure 3b). When compared with material, S1). When excluding the five subadult speci- multiple species, KNM-WT 40000 is significantly mens from the fossil samples, KNM-WT 40000 differs different from A. afarensis and P. robustus. significantly from A. afarensis for PC1, 4 and 5, as

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3384 F. Spoor et al. The maxilla of KNM-WT 40000

Table 2. Mahalanobis’ distance test comparing KNM-WT Table 3. Interspecific comparisons of the standard deviations 40000 using all PCs combined. D2, squared Mahalanobis’ of PC1 and PC2 obtained in the maxillary shape analysis, distance of KNM-WT 40000 from centroid of species giving F and probability (p) values. The differences are not sample; SDU, standard deviation unit of Mahalanobis’ statistically significant after sequential Bonferroni correction. distances within the sample; d.f., degrees of freedom; p, probability that KNM-WT 40000 belongs to the species. PC1 PC2 For sample sizes less than six, the probability is not calculated (n/a) because estimates of the variance are Fp-value Fp-value insufficiently reliable. A. afarensis–H. sapiens 1.010 0.865 1.441 0.450 D2 SDU d.f. p-value A. afarensis–P.troglodytes 1.083 0.758 1.192 0.648 A. afarensis–G. gorilla 1.371 0.503 1.368 0.505 KNM-WT 40000 (corrected) A. africanus–H. sapiens 1.570 0.811 2.233 0.560 A. afarensis 21.065 4.589 6 ,0.0025 A. africanus–P.troglodytes 1.435 0.885 2.698 0.451 A. africanus 37.084 6.089 4 n/a A. africanus–G. gorilla 1.134 0.914 2.351 0.529 P.robustus 13.284 3.644 6 ,0.05 P.robustus–H. sapiens 1.023 0.827 2.954 0.040 KNM-WT 40000 (as preserved) P.robustus–P.troglodytes 1.119 0.720 2.445 0.088 A. afarensis 18.909 4.348 6 ,0.005 P.robustus–G. gorilla 1.416 0.470 2.805 0.053 A. africanus 36.609 6.050 4 n/a P.robustus 13.738 3.706 6 ,0.05 KNM-WT 40000 (corrected), adults only 40000 than in A. afarensis (table 4). When compared A. afarensis 33.088 5.752 5 n/a with multiple species of Australopithecus and Paranthro- A. africanus 28.397 5.328 3 n/a pus, its mesiodistal length is only significantly smaller P.robustus 20.390 4.515 4 n/a than in P. boisei, but its buccolingual width is smaller than in all species other than A. anamensis. before, but for PC2 as well (electronic supplementary material, S2). When compared with multiple species, 5. DISCUSSION KNM-WT 40000 differs significantly from A. afarensis The taxonomic diagnosis of K. platyops and initial for PC4 and from P. robustus for PC1. description of its type specimen KNM-WT 40000 Results of Mahalanobis’ distance tests comparing were mainly based on qualitative comparisons KNM-WT 40000 with A. afarensis, A. africanus and (Leakey et al. 2001). Here we analyse the maxilla of P. robustus individually, and using all PCs combined, KNM-WT 40000 quantitatively and test the specific are given in table 2. Differences from A. afarensis hypotheses that the specimen is not different from and P. robustus are statistically significant, whereas the the contemporary taxon A. afarensis and, more A. africanus sample is too small for a probability to broadly, that it is not different from species of Austra- be calculated. lopithecus and Paranthropus. Based on the analyses of A PCA of the fossil hominins combined with maxillary shape and M2 crown size, both hypotheses modern humans, chimpanzees and gorillas shows can be rejected. These findings thus support the how the main aspects of variation of the fossil samples, notion that there was hominin species diversity in the as reflected by PC1 and PC2 (48% and 19% of the Middle Pleistocene and corroborate the validity of variance), compare with those shown by larger K. platyops as a separate species. It is worth pointing samples of extant species. A bivariate plot of PC2 out that the observed differences are substantial, against PC1 shows that the areas of observed variation given that statistical significance is obtained for small (convex hulls) of A. afarensis and P. robustus are not samples, with Bonferroni corrections when comparing substantially different from those of the extant species, KNM-WT 40000 with multiple species. Moreover, whereas A. africanus, with fewer specimens in the comparisons of the A. afarensis, A. africanus and sample, appears somewhat less variable (figure 3c). P. robustus samples used here with larger samples of With samples sizes of 50 or more, 95% confidence modern humans, chimpanzees and gorillas indicate ellipses of the extant species have a close fit with the that these fossils show representative levels of intraspe- observed variation, but for the small fossil samples cific morphological variation. Hence, the observed the ellipses are large. Importantly, F-tests indicate differences from KNM-WT 40000 are unlikely to be that the standard deviations of PC1 and PC2 an artefact of under-sampled variation in the three obtained for the fossil taxa are not significantly differ- fossil species. ent from those of the extant species (table 3). The observation by Leakey et al. (2001) that the M2 The M2 crown size of KNM-WT 40000 falls below crown size of KNM-WT 40000 is smaller than the the currently known range of variation of all hominin known range of variation shown by species of Australo- species included in the comparisons (table 4). Its pithecus and Paranthropus is upheld here based on the mesiodistal length is the same as the minimum largest sample currently available. The difference is known for A. anamensis, but the particular specimen most distinct for the buccolingual width. This is the has a larger buccolingual width than KNM-WT more reliable measure in KNM-WT 40000 as it is 40000 (KNM-ER 30200: 13.2 as opposed to 12.4). not affected by interstitial or occlusal wear, and the Statistically, both the mesiodistal length and buccolin- crack expanding the width is well defined and can be gual width are significantly smaller in KNM-WT corrected for with confidence.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

The maxilla of KNM-WT 40000 F. Spoor et al. 3385

Table 4. Mesiodistal (MD) length and buccolingual (BL) width of the M2, and the subnasal clivus (ns-pr) angle to the postcanine alveolar margin. The sample size (n), mean, minimum, maximum and standard deviation (s.d.) are given. Comparisons by t-test of KNM-WT 40000 with hominin species list the probability ( p, one-tailed) and significance (n.s., not significant; multiple comparisons with sequential Bonferroni correction). MD1 and BL1 defined after White (1977), MD2 and BL2 after Tobias (1967). The subnasal clivus angle was measured among adult and subadult specimens (M2 in full occlusion) listed in the electronic supplementary material, S3. *p , 0.05; **p , 0.01; ***p , 0.001.

M2 size subnasal MD1 MD2 BL1 BL2 source angle

KNM-WT 11.4 11.9 12.4 12.8 Leakey et al. (2001), this study 47 40000 A. anamensis n 8—8—Ward et al. (2001), White et al. (2000) 1 mean 12.88 14.50 27 min. 11.4 12.9 max. 14.3 16.7 s.d. 1.04 1.19 A. afarensis n 12 — 13 — Kimbel & Delezene (2009) 6 mean 13.00 14.80 34.6 min. 12.1 13.4 29 max. 14.1 15.8 39 s.d. 0.60 0.60 3.5 A garhi n ¼ 1 14.4 — 17.7 — Asfaw et al. (1999) 27 A. africanus n —24—28Moggi-Cecchi et al. (2006), J. Moggi-Cecchi 9 mean 14.12 15.95(2006, personal communication), this study 34.2 min. 12.6 13.5 30 max. 16.6 17.9 37 s.d. 1.09 1.23 1.9 P.aethiopicus n ¼ 1— — — — 31 P.robustus n — 24 — 24 J. Moggi-Cecchi (2006, personal communication), 8 mean 14.00 15.73this study 36.8 min. 11.6 14.0 32 max. 15.7 16.9 39 s.d. 0.99 0.94 4.1 P.boisei n —6 —6 Tobias (1967), Leakey & Walker (1988), Wood 2 mean 15.84 18.18(1991), this study 35.9 min. 14.7 16.6 33 max. 17.2 21.0 39 s.d. 1.03 1.55 comparison with A. afarensis p 0.013 0.001 0.011 sign. *** * comparison with Australopithecus and Paranthropus A. anamensis p 0.111 0.073 sign. n.s. n.s. A. afarensis p 0.013 0.001 0.011 sign. n.s. ** * A. africanus p 0.029 0.010 0.000 sign. n.s. * *** P.robustus p 0.024 0.003 0.024 sign. n.s. ** P.boisei p 0.008 0.012 sign. **

The geometric morphometric shape analysis of the and Australopithecus. In contrast, PC2 associates a maxilla shows that zygomatic process position together more anteriorly positioned zygomatic process with a with subnasal clivus length and transverse flatness shorter subnasal clivus and separates A. afarensis account for most of the variance in the hominin from A. africanus only. There is no evidence of fossil sample (PC1 and PC2 combined, 86%, intraspecific differences within Paranthropus regarding figure 3a). PC1 associates a more anteriorly positioned PC1 and PC2, as KNM-WT 17000 and OH 5 fall zygomatic process with a transversely flatter and in the middle of the range of P. robustus. longer subnasal clivus, along a gradient of genera: Apart from zygomatic root position and transverse Paranthropus, Kenyanthropus (i.e. KNM-WT 40000) subnasal flatness, subnasal prognathism is a third

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3386 F. Spoor et al. The maxilla of KNM-WT 40000

Table 5. Anterior position of the zygomatic process along the dental row. Accession codes CH, ER, KP and WT lack the preﬁx KNM-.

P3 P3/P4 P4 P4/M1 M1

A. anamensis KP 29283 A. afarensis A.L. 442-1 A.L. 58-22 A.L. 199-1 A.L. 200-1a A.L. 333-2 A.L. 333-1 A.L. 413-1 A.L. 427-1 A.L. 417-1d A.L. 444-2 A.L. 822-1 A.L. 486-1a A.L. 651-1 A. garhi BOU-VP-12/130 A. africanus Sts 52a MLD 6 MLD 9 Sts 5 MLD 45 TM 1511 Sts 17 TM 1512 Sts 52a TM 1514 Sts 53 Sts 63 Sts 71 Sts 3009 Stw Stw 13 252a,b Stw Stw 391 183a,b Stw 505 Stw 498 K. platyops WT 40000 WT 38350 P.aethiopicus WT 17000 P.robustus SK 13a TM 1517 DNH 7 SK 47a SK 48 SK 11 SK 821 SK 52a SK 12 SK 79 SK 29 SK 83 SK 46 SKW 11a SK 79 SKW 12 P.boisei KGA 10-525 KGA 10-525 CH 1Ba ER 405 ER 732 ER 406 WT 17400a OH 5 aImmature specimen. bListed as A. africanus, but affinities uncertain (Lockwood & Tobias 2002). prominent aspect of maxillary shape characterizing flat subnasal area with anteriorly positioned zygo- KNM-WT 40000. It is also expressed by two separate matics and Kenyanthropus by a more orthognatic, patterns of variation (PC4 and PC5; figure 3b), which transversely flat subnasal area with anteriorly differ depending on whether the clivus orientation positioned zygomatics (figure 1b). varies in the midline only (PC4), or involves the Two of the facial features, the degree of midline entire subnasal area, from canine jugum to canine subnasal prognathism and the position of the zygo- jugum bilaterally (PC5). As most specimens in matic process, can be quantified individually in a the sample (P. robustus, P. boisei, A. afarensis and larger number of early hominin specimens than A. africanus) tend to show similar levels of prognath- could be included in the geometric morphometric ism, this morphology does account for only 5 per analyses. It can thus be assessed whether the cent of the variance in the total sample. However, it evidence from larger samples is consistent with the does single out the orthognathic morphology of landmark-based results. The subnasal angle, which KNM-WT 40000 and to a lesser extent the more prog- combines the shape variation associated with PC4 nathic shape in A. anamensis, A. garhi and P.aethiopicus. and PC5, is larger in KNM-WT 40000 (478) than in This pattern illustrates that in interspecific comparisons any of the Australopithecus and Paranthropus specimens the higher PCs associated with small amounts of that could be measured (27–398; table 4, electronic overall variance can provide highly relevant information supplementary material, S3). Those differences that regarding individual specimens, because the distribution can be tested, from A. afarensis, A. africanus and ofthevariancedependsonthesamplecomposition. P. robustus, are statistically significant. The anterior In all, the analyses confirm the occurrence of three zygomatic root positions of early hominins, associated different facial patterns among the early hominins con- with PC1 and PC2, are summarized in table 5. sidered here. Australopithecus is characterized by a The position in KNM-WT 40000 at the level of the prognathic, transversely curved subnasal area com- P3/P4 interalveolar septum is commonly found in bined with posteriorly positioned zygomatics Paranthropus as well. On the other hand, it falls outside (figure 1a), Paranthropus by a prognathic, transversely the range of variation of Australopithecus,withthe

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

The maxilla of KNM-WT 40000 F. Spoor et al. 3387 exception of the left side of the subadult specimen Sts corner of the dental arcade, the canine alveolus has 52. Importantly, in A. afarensis, the position is always more thin overlying bone than other teeth, making more posterior, in one instance at the distal half of the jugum particularly vulnerable to cracking. P4, and more commonly at the P4/M1 septum or M1. Apart from the zygomatic process position, sub- Hence, these univariate observations do fully confirm nasal clivus morphology and a small upper molar the characterization of K. platyops as subnasally orthog- size Leakey et al. (2001) also lists similarly sized I1 nathic combined with anteriorly positioned zygomatics. and I2 roots and upper premolars that are three- The A. afarensis specimen most similar in geological rooted as features characterizing the maxilla of age to KNM-WT 40000 is the Garusi 1 maxilla K. platyops. These have not been considered here, (Laetoli approx. 3.6 Ma). It is too fragmentary to be but warrant further study. The unusual incisor root included in the PCA, or even to allow the quantification proportions (figure 2b) and their spatial relationship of subnasal clivus orientation or zygomatic process pos- to the transversely flat, orthognathic subnasal area is ition. However, it is possible to make some inferences of particular interest, and using high-resolution CT about its morphology that are relevant here. There is this can now be examined in more detail. no evidence of the zygomatic root in the premolar The partial maxilla KNM-WT 38350 shares with area, and the anterior position must thus have been at KNM-WT 40000 the anterior zygomatic root posi- P4/M1 or more posteriorly. Subnasally, Garusi 1 is tion, three-rooted premolars and a small molar size very prognathic, has rounded nasal margins around and was therefore designated as the paratype of the canine alveoli and lacks a clear nasal sill. In the K. platyops (Leakey et al. 2001). However, it is too frag- latter characters, it differs from the A. afarensis Hadar mentary a specimen to enable a full comparison with sample and is more similar to A. anamensis (Kimbel the unique facial morphology of KNM-WT 40000. et al. 2006; Kimbel 2007), and in all these aspects, The partial mandible KT12/H1, the holotype of the Garusi 1 contrasts strongly with KNM-WT 40000. broadly contemporary species A. bahrelghazali,is KNM-WT 40000 is poorly preserved, and Leakey characterized by a sagittally and transversely flat et al. (2001) reported on the specimen by extracting anterior corpus, said to reflect a more orthognathic meaningful information from selected areas after care- face (Brunet et al. 1996). If correct, this would fully mapping the post-mortem distortion. Bone increase the likelihood that KNM-WT 40000 and expansion associated with clay-filled cracks is fre- KT12/H1 are conspecific. However, the association quently encountered among fossils found in the between subnasal and symphyseal shapes is not well Turkana Basin and has been routinely recognized as understood (Spoor et al. 2005), and how K. platyops a phenomenon affecting a specimen’s morphology relates to A. bahrelghazali remains unclear. Thus, (e.g. the descriptions in Wood 1991). Naming it although there is good evidence, presented here and expanding matrix distortion, White (2003) states to elsewhere (Leakey et al. 2001; Guy et al. 2008), of have ‘formalized’ this taphonomic process, defining hominin species diversity in the Middle Pliocene of five stages. He assigned KNM-WT 40000 to stage 4, Africa, additional fossils will be required to reveal the but as no definition of these stages has been published, full nature and interrelationships of the lineages it is not possible to evaluate this classification. It is present at that time. worth pointing out that some areas of the cranium, We thank Alan Walker and Chris Stringer for inviting us to such as parts of the left temporal bone, show very contribute this work and the National Museums of Kenya, little distortion, whereas others, such as the cranial the National Museum of Ethiopia, the Transvaal Museum vault, are highly affected. Thus, characterizing the (South Africa), Department of Anatomy, Witwatersrand specimen by a single stage has little value. University (South Africa), the Institute of Human Origins The analyses presented here show that the distor- (USA) and the museums listed in §2 for access to tion has had little impact on the characters of specimens in their care. We are grateful to Berhane Asfaw, maxillary shape relevant to the diagnosis of K. platyops. Michel Brunet, Ron Clarke, Nick Conard, Chris Dean, Heidi Fouri, Philipp Gunz, John Harrison, Jean Jacques The preservation of the tooth roots and the integrity of Hublin, Louise Humphrey, Paula Jenkins, Don Johanson, the dental arcade indicate that distinct directional Andre Keyser, Bill Kimbel, Rob Kruszynski, Kornelius shape changes, such as skewing or compression, did Kupczik, the late Charlie Lockwood, Emma Mbua, Jacopo not occur in the lower part of the left maxilla. Expan- Moggi-Cecchi, Sam Muteti, Paul O’Higgins, Matt Skinner, sion cracks did cause a size increase of about 18 per Gen Suwa, Heiko Temming, Brian Villmoare, Tim White cent, but this occurred mostly at a similar rate across and Andreas Winzer for help with various aspects of this study. Financial support was provided by the Leakey the area, thus having little effect on shape. In all, Foundation, the National Geographic Society and the Max there is no indication that the position of the zygomatic Planck Society. root or the subnasal clivus shape were modified substantially, particularly in a way that would mimic normal morphological differences between species. The only striking contrast in expansion rate was found between the area above (6%) and over the left REFERENCES Asfaw,B.,White,T.,Lovejoy,O.,Latimer,B.,Simpson,S.& canine jugum (20%). This difference is consistent Suwa,G.1999Australopithecus garhi: a new species of early with the CT-based observation that internal expansion hominid from Ethiopia. Science 284, 629–635. (doi:10. is strongest in the alveolar space around the roots, a 1126/science.284.5414.629) phenomenon that is understandable as it becomes Boaz,N.T.1988StatusofAustralopithecus afarensis. readily filled with clay, unlike trabecular bone not Yrbk. Phys. Anthropol. 31, 85–113. (doi:10.1002/ajpa. open to the outside. Moreover, positioned at the 1330310506)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3388 F. Spoor et al. The maxilla of KNM-WT 40000

Brown, B., Brown, F. & Walker, A. 2001 A. New hominids Africa. J. Hum. Evol. 42, 389–450. (doi:10.1006/jhev. from the Lake Turkana Basin, Kenya. J. Hum. Evol. 41, 2001.0532) 29–44. (doi:10.1006/jhev.2001.0476) Lockwood, C. A., Richmond, B. G., Jungers, W. L. & Brunet, M., Beauvillain, A., Coppens, Y., Heintz, E., Kimbel, W. H. 1996 Randomization procedures and Moutaye, A. H. E. & Pilbeam, D. 1995 The ﬁrst sexual dimorphism in Australopithecus afarensis. J. Hum. australopithecine 2 500 kilometres west of the Rift Evol. 31, 537–548. (doi:10.1006/jhev.1996.0078) Valley (Chad). Nature 378, 273–274. (doi:10.1038/ Lockwood, C. A., Kimbel, W. H. & Johanson, D. C. 2000 378273a0) Temporal trends and metric variation in the mandibles Brunet, M., Beauvillain, A., Coppens, Y., Heintz, E., and dentition of Australopithecus afarensis. J. Hum. Evol. Moutaye, A. H. E. & Pilbeam, D. 1996 Australopithecus 39, 23–55. (doi:10.1006/jhev.2000.0401) bahrelghazali, une nouvelle espe`ce d’hominide´ ancien de Moggi-Cecchi, J., Grine, F. E. & Tobias, P. V. 2006 Early la re´gion de Koro Toro (Tchad). C. R. Acad. Sci. Paris hominid dental remains from Members 4 and 5 of 322, 907–913. the Sterkfontein Formation (1966–1996 excavations): Cobb, S. 2008 The facial skeleton of the chimpanzee– catalogue, individual associations, morphological human last common ancestor. J. Anat. 212, 469–485. descriptions and initial metrical analysis. J. Hum. Evol. (doi:10.1111/j.1469-7580.2008.00866.x) 50, 239–328. (doi:10.1016/j.jhevol.2005.08.012) Guy, F., Mackaye, H. T., Likius, A., Vignaud, P., Nevell, L. & Wood, B. 2008 Cranial base evolution within Schmittbuhl, M. & Brunet, M. 2008 Symphyseal the hominin clade. J. Anat. 212, 455–468. (doi:10. shape variation in extant and fossil hominoids, and the 1111/j.1469-7580.2008.00875.x) symphysis of Australopithecus bahrelghazali. J. Hum. O’Higgins, P. & Jones, N. 1998 Facial growth in Cercocebus Evol. 55, 37–47. (doi:10.1016/j.jhevol.2007.12.003) torquatus: an application of three dimensional geometric Hammer, Ø., Harper, D. A. T. & Ryan, P. D. 2001 PAST: morphometric techniques to the study of morphological paleontological statistics software package for education variation. J. Anat. 193, 251–272. (doi:10.1046/j.1469- and data analysis. Palaeontol. Electron. 4,1–9. 7580.1998.19320251.x) Kimbel, W. K. 2007 The species and diversity of Rice, W. R. 1989 Analyzing tables of statistical tests. australopiths. In Handbook of Paleoanthropology, vol. III Evolution 43, 223–225. (doi:10.2307/2409177) (eds W. Henke, T. Hardt & I. Tattersall), pp. 1539– Spoor, F., Leakey, M. G. & Leakey, L. N. 2005 Correlation 1573. Berlin, Germany: Springer. of cranial and mandibular prognathism in extant and Kimbel, W. H. & Delezene, L. K. 2009 ‘Lucy’ redux: a fossil hominids. Trans. R. Soc. S. Afr. 60, 85–89. review of research on Australopithecus afarensis. Yrbk. Strait, D. S. & Grine, F. E. 2004 Inferring hominoid and Phys. Anthropol. 52, 2–48. (doi:10.1002/ajpa.21183) early hominid phylogeny using craniodental characters: Kimbel, W. H., Rak, Y. & Johanson, D. C. 2004 The skull the role of fossil taxa. J. Hum. Evol. 47, 399–452. of Australopithecus afarensis. New York, NY: Oxford (doi:10.1016/j.jhevol.2004.08.008) University Press. Tobias, P. V. 1967 The cranium and maxillary dentition of Kimbel, W. H., Lockwood, C. A., Ward, C. V., Leakey, M. Australopithecus (Zinjanthropus) boisei. Olduvai Gorge, G., Rak, Y. & Johanson, D. C. 2006 Was Australopithecus vol. 2. Cambridge, UK: Cambridge University Press. anamensis ancestral to A. afarensis? A case of anagenesis Ward, C. V., Leakey, M. G. & Walker, A. 2001 Morphology in the early hominin fossil record. J. Hum. Evol. 51, of Australopithecus anamensis from Kanapoi and Allia Bay, 134–152. (doi:10.1016/j.jhevol.2006.02.003) Kenya. J. Hum. Evol. 41, 255–368. (doi:10.1006/jhev. Leakey, R. E. F. & Walker, A. 1988 New Australopithecus 2001.0507) boisei specimens from East and West Lake Turkana, White, T. D. 1977 New fossil hominids from Laetolil, Kenya. Am. J. Phys. Anthropol. 76, 1–24. (doi:10.1002/ Tanzania. Am. J. Phys. Anthropol. 46, 197–230. (doi:10. ajpa.1330760102) 1002/ajpa.1330460203) Leakey, M. G., Spoor, F., Brown, F. H., Gathogo, P. N., White, T. D. 2003 Early hominids—diversity or distortion? Kiarie, C., Leakey, L. N. & McDougall, I. 2001 Science 299, 1994–1997. (doi:10.1126/science.1078294) New hominin genus from eastern Africa shows diverse White, T. D., Suwa, G., Simpson, S. & Asfaw, B. 2000 middle Pliocene lineages. Nature 410, 433–440. (doi:10. Jaws and teeth of Australopithecus afarensis from Maka, 1038/35068500) Middle Awash, Ethiopia. Am. J. Phys. Anthropol. 111, Lockwood, C. A. & Tobias, P. V. 1999 A large male hominin 45–68. (doi:10.1002/(SICI)1096-8644(200001)111: cranium from Sterkfontein, South Africa, and the status 1,45::AID-AJPA4.3.0.CO;2-I) of Australopithecus africanus. J. Hum. Evol. 36, 637–685. Wood, B. 1991 Koobi Fora research project. Vol. 4: hominid (doi:10.1006/jhev.1999.0299) cranial remains. Oxford, UK: Clarendon Press. Lockwood, C. A. & Tobias, P. V. 2002 Morphology and Wood, B. & Lonergan, N. 2008 The hominin fossil record: afﬁnities of new hominin cranial remains from Member 4 taxa, grades and clades. J. Anat. 212, 354–376. (doi:10. of the Sterkfontein Formation, Gauteng Province, South 1111/j.1469-7580.2008.00871.x)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Phil. Trans. R. Soc. B (2010) 365, 3389–3396 doi:10.1098/rstb.2010.0059

Stable isotopes in fossil hominin tooth enamel suggest a fundamental dietary shift in the Pliocene Julia A. Lee-Thorp1,*, Matt Sponheimer2, Benjamin H. Passey3, Darryl J. de Ruiter4 and Thure E. Cerling5 1Research Laboratory for Archaeology, Dyson Perrins Building, South Parks Road, Oxford OX1 3QY, UK 2Department of Anthropology, University of Colorado at Boulder, 233 UCB, Boulder, CO 80309, USA 3Department of Earth and Planetary Sciences, Johns Hopkins University, Baltimore, MD 21218, USA 4Department of Anthropology, Texas A&M University, College Station, TX 77843, USA 5Department of Geology and Geophysics, University of Utah, Salt Lake City, UT 84112, USA Accumulating isotopic evidence from fossil hominin tooth enamel has provided unexpected insights into early hominin dietary ecology. Among the South African australopiths, these data demonstrate signiﬁcant contributions to the diet of carbon originally ﬁxed by C4 photosynthesis, consisting of C4 tropical/savannah grasses and certain sedges, and/or animals eating C4 foods. Moreover, high-resolution analysis of tooth enamel reveals strong intra-tooth variability in many cases, suggesting seasonal-scale dietary shifts. This pattern is quite unlike that seen in any great apes, even ‘savannah’ chimpanzees. The overall proportions of C4 input persisted for well over a million years, even while environments shifted from relatively closed (ca 3 Ma) to open conditions after ca 1.8 Ma. Data from East Africa suggest a more extreme scenario, where results for Paranthropus boisei indicate a diet dominated (approx. 80%) by C4 plants, in spite of indi- cations from their powerful ‘nutcracker’ morphology for diets of hard objects. We argue that such evidence for engagement with C4 food resources may mark a fundamental transition in the evolution of hominin lineages, and that the pattern had antecedents prior to the emergence of Australopithecus africanus. Since new isotopic evidence from Aramis suggests that it was not present in Ardipithecus ramidus at 4.4 Ma, we suggest that the origins lie in the period between 3 and 4 Myr ago.

Keywords: carbon isotopes; enamel; C4 resources; australopiths

1. INTRODUCTION folivorous hominid diet to include roots, bulbs and Diet is a fundamental feature of a species’ biology, animal foods such as insects, scorpions, lizards, bird’s strongly influencing basic body size and morphology, eggs and the young of small antelope (Dart 1926). life-history strategies for survival of the young in par- Effectively, it was a pre-statement of the ‘Dietary ticular, social organization and the manner of its Breadth’ hypothesis. These debates have continued adaptations to its environment. Consequently, the apace, but although our understanding of the nature nature of ancestral diets remains one of the most of environments and preferred habitats has advanced active topics of research in human evolution. Many substantially, the fossil record has grown and more years after Dart first puzzled over how the newly dis- sophisticated methods have been applied to study mor- covered ‘man-like apes’ (in reference to the Taung phology and the biomechanics of food processing, our child and its kind) had survived in Taung’s xeric, comprehension of the important dietary shifts that open, Kalahari environment, so alien to all the other must have occurred during the early emergence of forest-loving great apes (Dart 1925, 1926), we are hominins in the Pliocene is still uncertain. Such chal- still actively debating these issues today (e.g. White lenges have encouraged the development of new et al. 2009a). Dart could not have comprehended, at methods for examining dietary ecology. that time, the full scale of environmental shifts that In an earlier Royal Society meeting on human evol- occurred in Africa in the past several millions years, or ution almost 30 years ago, Alan Walker set out a series indeed the depth of time encompassed. But he made of what he considered to be the more promising emer- some surprisingly prescient suggestions, including the ging avenues in early hominin dietary research (Walker likely expansion of the conventional frugivorous and 1981). Among other candidates, Walker suggested the inspection of tooth microwear, and carbon isotope and trace element analysis of fossil bones (Walker 1981, * Author for correspondence ([email protected]). p. 58). It was evident at the 2009 meeting that the One contribution of 14 to a Discussion Meeting Issue ‘The first four first two methods have undergone considerable devel- million years of human evolution’. opment and progress since that time, in opening new

3389 This journal is q 2010 The Royal Society Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3390 J. A. Lee-Thorp et al. Isotopes and hominin diets windows on hominin dietary ecology. Walker’s and in the canopy (including fruits) is slightly less predictions followed closely on a flurry of pioneering depleted (van der Merwe & Medina 1989; Cerling studies in the late 1970s that began to explore the sys- et al. 2004). Dietary d13C values are reflected in all tis- tematics of these three approaches. In the case of sues, including enamel, so that fossil enamel d13C stable isotopes, studies demonstrated the direct values provide opportunities to test hypotheses about relationship between carbon isotopes in the diet the dietary habits of extinct animals (e.g. Lee-Thorp and in animal tissues (DeNiro & Epstein 1978a), et al. 1989; Cerling et al. 1997). others that these distinctions held for wild grazers Most published isotope hominin dietary research and browsers (DeNiro & Epstein 1978b; Vogel 1978) has so far focused on the South African hominins and finally that it provided a new approach for (Lee-Thorp et al. 1994; Sponheimer & Lee-Thorp addressing an archaeological question about the 1999; Sponheimer et al. 2005, 2006b; Lee-Thorp & introduction of maize agriculture (van der Merwe & Sponheimer 2006), although this situation is begin- Vogel 1978). Other studies hinted that these methods ning to change. Comparisons between the two South might be extended to fossils at greater time depth African australopiths were also a starting point for if based on the mineral rather than the organic the development of occlusal enamel microwear com- phase of bone (e.g. Ericson et al. 1981; Sullivan & parisons (Grine 1981, 1986). This and subsequent Krueger 1981). studies using less subjective, more quantifiable At that stage, almost all chemical studies were based methods (Scott et al. 2005) suggested that, in spite on bone collagen or bone mineral (a biological apa- of significant overlap, P. robustus microwear showed tite), but the latter is an extremely unstable subtly more complexity or pitting in occlusal enamel structure, which is vulnerable to diagenesis. It was wear compared with Australopithecus africanus. There- not until the potential of the more crystalline, stable fore, the inference is that the former included a enamel phase was explored and demonstrated higher proportion of harder food items that required (Lee-Thorp & van der Merwe 1987) that the full more processing and caused more pitting and fewer potential of the method to older fossils could be directional scratches (Grine 1986). These results realized. Following earlier work (Parker & Toots were thought to be consistent with the widely held 1980; Elias et al. 1982), the potential of trace view that Paranthropus was a specialist vegetarian elements, especially strontium to calcium ratios (Sr/ (Grine 1986). The microwear distinctions are quite Ca), as trophic indicators in fossil foodwebs was subtle and may also be influenced by differences in explored extensively (Sillen 1988, 1992), but these enamel prism orientation between the two taxa efforts have largely stalled. This may be a reflection (Macho & Shimizu 2009). of the former strong focus on fossil bone, with its Nevertheless, the microwear findings provide a attendant problems of diagenesis. A single, broader useful framework for hypothesis-testing using stable enamel-based study in the South African hominin isotopes since the data essentially suggest diets that sites was unable to replicate the trophic patterns would be classed as C3 (i.e. hard fruits and nuts). seen earlier for Paranthropus robustus (Sponheimer & The prediction would be that A. africanus and Lee-Thorp 2006); instead, it was suggested that P. robustus should be indistinguishable in their d13C Sr/Ca and Ba/Ca, in combination, may be more values from C3 feeders, such as browsers and frugi- informative about plant resources. That suggestion vores. The results, however, flatly contradict this awaits further exploration, and trace elemental analysis prediction. Analysis of more than 40 hominin speci- will not be discussed further here. mens from the sites Makapansgat, Sterkfontein, Our main purposes in this paper are to review the Kromdraai and Swartkrans, spanning a period of 13 evidence for the shift to incorporate C4 resources in about 3–1.5 Ma, demonstrate that d C values of early hominin diets, to present new data for temporal both australopiths are indistinguishable from each variability in C4 consumption in the earlier South other, but distinct from that of coexisting C3 consu- African australopith, Australopithecus africanus, which mers (figure 1). Surprisingly, the proportions of C4 is comparable to that of P. robustus (Sponheimer et al. in enamel, on average, remain relatively constant in 2006b), and to make some predictions about the spite of the passage of time and marked shifts in origins and inferences of such an adaptation. environments, from relatively closed to far more open landscapes, by about 1.8 Ma (Vrba 1985; Reed 1997; Lee-Thorp et al. 2007). 2. ISOTOPIC EVIDENCE FOR C4 IN HOMININ On average, both taxa obtained 25–35% of their DIETS carbon from C4 sources. These resources must have The primary distinction in application of stable carbon been obtained either directly from grasses or sedges, isotopes to hominin diets is the difference in 13C/12C or indirectly from animals that ate these plants. 13 1 (expressed as d C) between C3 and C4 plants. In Since few of the fine scratches characterizing the African environments they typically occupied, consumption of grass are present in their microwear where the growing (wet) season is warm, all carbon (Grine 1986; Scott et al. 2005), it was deduced dietary sources from trees, bushes, shrubs and forbs that direct consumption of grass blades was less are distinctly lower in d13C compared with those plausible (although not ruled out) (Lee-Thorp & from tropical grasses and some sedges. The primary Sponheimer 2006). C4 sedges, grass rhizomes and exception is in tropical forest ecosystems where C3 the proposed consumption of grass-eating termites subcanopy (shaded) plants are even more depleted in (and other small animal foods) may be implicated 13C (i.e. lower d13C), while vegetation in clearings (Sponheimer et al. 2005; Yeakel et al. 2007), but

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Isotopes and hominin diets J. A. Lee-Thorp et al. 3391

Makapansgat M3 3. VARIABILITY BETWEEN AND WITHIN INDIVIDUALS The foregoing discussion relies on broad averages, and we now turn to examine patterns in individuals. Where sufficient australopith d13C values exist within any one site, or members within that site, Sterkfontein M4 the data are quite variable, which suggests a degree of dietary opportunism and flexibility. This observation leads to the question of whether hominins shifted their diets on annual or even seasonal time scales, and whether the C4 contributions observed Swartkrans M1 in bulk tooth enamel measurements mask short-term dietary variability. Tooth enamel is an incremental tissue that can be sampled to investigate temporal changes in both climate and diet using stable isotope analysis. –14 –12 –10 –8 –6 –4 –2 0 2 4 Developments in laser ablation techniques now δ13C‰ permit high-resolution sequential sampling of enamel crowns, with minimal visible damage (Passey & Figure 1. Data for all the South African hominins are sum- Cerling 2006). However, the sequential chronological marized as means d13C (black boxes) and standard deviations compared with means and standard deviations resolution that can be attained, no matter how small for the browsing and grazing fauna. The sites are given in the sample spot, is severely constrained by overprinting sequence from oldest (top) to youngest (bottom). Maka- during maturation of the enamel. Enamel maturation pansgat Member 3 is about 2.7–3 Ma, Sterkfontein occurs for many months, even years, after primary Member 4 is usually considered to be about 2.2–2.5 Ma mineralization during prism formation (Suga 1982; and Swartkrans Member 1 is younger than 2 Ma. These Balasse 2002). Recovery of primary dietary signals ages based on biostratigraphy are imprecise, but sufficient has been successfully accomplished using a forward for our purposes. More precise chronometric studies based and inverse model only in continuously growing on Pb/U isotopes have recently been completed (R. Pickering teeth and where the growth and enamel maturation 2009, personal communication), but do not change the over- parameters are well characterized (Passey et al. all sequence. All the hominin data show significant C 4 2005). The patterns of enamel maturation in modern contributions compared with C3 feeders, in spite of large shifts, from closed to open, in the environments (Reed human tooth crowns are poorly understood, and 1997). Adapted from Lee-Thorp & Sponheimer (2006). those of early hominins even less so. However, it is evident that the nature of crown formation and enamel maturation inevitably produces a mixture of isotope signals, from both the initial primary mineralization when examined individually, none of these resources and the maturation periods, so that there will be offers a completely satisfactory solution. For instance, dampening or overprinting of the original signals. it has been shown that in this part of southern Africa Nevertheless, three statements can be made with the proportion of sedges following the C4 pathway certainty: (i) the isotope profile from crown to root is modest (Stock et al. 2004), and likewise few preserves an ordered time series extending from earlier termite species seem to be C4 specialists (Sponheimer (crown) to later (root) in time, (ii) the mimimum time et al. 2005). The most plausible explanation is that duration of any single spot-sample within a tooth is they utilized C4 resources quite broadly, including equivalent to the ‘maturation time’, or time required both C4 plant and animal resources. Although the for enamel to cure into its fully mineralized form results say little about the rest of the diet (i.e. (probably months in primates), and (iii) the total time the major C3 portion), they hint that neither of duration of an isotopic profile across a tooth is the these australopiths were plant specialists. sum of the crown enamel deposition time (for example, These results were also unexpected because extant as recorded by perikymata) and the maturation time of great apes consume minimal or no C4 resources even the last enamel increment deposited. when they live in relatively open habitats. Several Notwithstanding the constraints imposed by matu- studies have shown that even ‘savannah’ chimpanzees, ration patterns and by lower analytical precision, who live in the more open parts of the Pan range, con- laser ablation has been applied to sample the external sume few, if any, C4 resources (Schoeninger et al. surface along the growth trajectory of four P. robustus 1999; Sponheimer et al. 2006a). Most forest-dwelling tooth crowns (Sponheimer et al. 2006b)(figure 2b). chimpanzees and gorillas reveal distinct low d13C The results showed that d13C values, and the pattern values, indicative of the use of resources located in or temporal change, differed between individuals shaded understorey vegetation (Carter 2001; Cerling and, most startling, showed differences of up to 5‰ et al. 2004; Sponheimer et al. 2009; J. A. Lee-Thorp, within a given P. robustus tooth. Given that the signal Y. Warren & G. A. Macho 2009, unpublished is attenuated or even scrambled as described above, data). It is this engagement with C4 resources, these data still suggest a very large shift from a diet which were becoming increasingly available in the dominated by C3 to a diet dominated by C4 resources Plio-Pleistocene, that indicates a fundamental niche in certain individuals. Variability is observed at several difference between the australopiths and extant apes. time scales—intra- and inter-annual.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3392 J. A. Lee-Thorp et al. Isotopes and hominin diets

a)(0( b)

–2

–4

C –6 13 δ –8

–10

–12 02468101214 024681012141618 scan number scan number

Figure 2. High-resolution laser ablation d13C sequences for (a) A. africanus and (b) P. robustus plotted against sample (scan) number. Sample increments were approximately 0.3 mm. The Paranthropus data are from Sponheimer et al. (2006b), where the data were plotted according to a time-sequence model based on perikymata counts. In this case, we avoided the application of a time sequence based on perikymata because the lengthy maturation time introduces not only a longer time period but also more uncertainty. (a) Open diamonds, STS 2518 max RM3; filled circles, STS 31 max RM3; filled triangles, STS 2253 mand RM1. (b) Open diamonds, SKW 6427 M; filled circles, SKW 5939 M; filled triangles, SK 24606 RM2 or 3; squares with crosses, SK 24606 RM3.

Four A. africanus molar crowns from Sterkfontein depending on the available tooth surface, with between Member 4 were analysed using the same methods. nine and 12 scans for the three reported molars. Based The results for one tooth were omitted because of con- on a periodicity of 7 days per striae of Retzius cerns about the interference of glue on the surface, (observed externally as perikymata) as calculated in observed as puffs of gas during ablation. Although Lacruz et al. (2006), the period of primary mineraliz- the age of Member 4 is uncertain, biostratigraphic ation in the sampled area is over a year for two evidence indicates that it is considerably older specimens (STS 2253 and STS 31) and just under a than Swartkrans, and with several taxa—including year for the third (STS 2518). However, as discussed A. africanus—in common with the older site of above, the time represented in the isotope profile is Makapansgat (Vrba 1985). As was the case for the much longer when enamel maturation is taken into P. robustus crowns, all teeth are lightly to moderately account, and additionally the measured values worn molars. The tooth assignments are shown in represent an attenuated signal of higher amplitude. figure 2, although some in figure 2b were too fragmen- Variability in the proportions of C3 and C4 tary to allow determination other than that they are resources between and within individuals is on a simi- molars. lar scale for A. africanus compared with P. robustus The analytical methods for the A. africanus molars (figure 2). At least one individual, STS 2518, indicates followed exactly those reported in Sponheimer et al. a more or less uniform C3 resource base, while (2006b). Each tooth was cleaned mechanically and another, STS 31, shows values varying by over 6‰ chemically (with acetone), and then thoroughly dried from a C3 to a predominantly C4 resource base. As in a low-temperature oven. Samples were purged emphasized above, it is difficult to ascertain the precise with helium inside the laser chamber for several min- time scales over which this variation occurred. How- utes or hours as required for the rate of CO2 ever, given that the enamel underlying the sampling outgassing to fall below appropriate levels. Small arrays mineralized and matured over a period of over amounts (10–30 nmol) of CO2 were generated using a year in each case, the differences between individuals aCO2 laser (10.6 mm) operating at 5–15 W and cannot be ascribed to sampling of a single season in 8.5 ms pulse duration in a He atmosphere. The CO2 one case and multiple seasons in another. was cryogenically purified and ‘focused’ prior to intro- The results suggest that C4 resources formed an duction to a continuous-flow GC-IRMS (MAT 252). important but highly variable component of hominin Systematic isotope fractionation and fractionation diets, extending at least as far back as A. africanus at associated with laser ablation production of CO2 Sterkfontein. There is no reason to believe that the were monitored by analyses of injected aliquots of dietary ecology of A. africanus at the earlier site of CO2 and by analysis of a suite of internal tooth Makapansgat Member 3 would be significantly enamel standards, both calibrated against NBS-19 different. gas (d13C ¼ 1.95‰). The laser-carbonate isotope fractionation (1*LASER-carb) for fossil herbivore samples from several South African hominin sites was –1.3 + 4. ORIGINS AND TIME DEPTH OF THE C4 1.5‰ (1s). PATTERN IN HOMININ DIETS Enamel was sampled at ca 0.3 mm intervals, Few analyses have been performed on eastern African encompassing about four perikymata for each laser material to allow us to definitively address the question ablation track (or scan) and the space between of the origins of engagement with C4 resources. Clues tracks. The length of each sampling trajectory varied from several sources may hint at an origin associated

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Isotopes and hominin diets J. A. Lee-Thorp et al. 3393 with the emergence of the genus Australopithecus, but at present there are no data to address the period Aramis 4.4 Ma prior to emergence of A. africanus. Recently published d13C data for two Paranthropus boisei individuals from Olduvai revealed a strong hyaenids dependence on C4 resources (van der Merwe et al. 2008). This was unexpected since the similarity of the ‘nutcracker’ dental and masticatory complex to that of South African P. robustus led to the expectation Makapansgat ca 2.8–3 Ma of a broadly similar dietary adaptation. Such high d13C values cannot be explained by consumption of C4 animal foods alone since they would require a highly unlikely dietary scenario that included almost exclu- hyaenids sive consumption of the flesh of grazing (C4) animals. The results demand an alternative explanation. The authors suggested that P. boisei,orat –14 –12 –10 –8 –6 –4 –2 0 2 4 least these two individuals, probably specialized in δ13C‰ exploitation of C sedges, which are far more abun- 4 Figure 3. A summary comparison of the hominin data for dant in East African wetlands than they are in South Aramis (above), and Makapansgat (below), plotted as Africa (van der Merwe et al. 2008). Additionally, a means (black boxes) and standard deviations with mean recent microwear study of P. boisei occlusal enamel values for C3 and C4 feeders shown for comparison. Data showed little evidence for the pitting associated with for carnivores (hyaenids) are also shown (white boxes; n ¼ hard object feeding (Ungar et al. 2008). These data 2 for Makapansgat), because they are effectively integrators seem inconsistent with their strongly developed masti- for the values of all the fauna they consume. These data catory complex, widely considered to be an adaptation are shifted towards values more enriched in 13C in Aramis, for the consumption of hard foods. However, their suggesting more open, C4 elements in the environment com- diets were clearly abrasive when the degree of wear is pared with Makapansgat, where the faunal assemblage consists largely of C3 feeders. In spite of this difference, Ar. considered (Tobias 1967). In combination, the high 13 C resource consumption and the lack of a hard- ramidus remains relatively depleted in C, quite unlike the 4 patterns for A. africanus seen at Makapansgat. Data for object microwear pattern may require that we rethink Aramis are from White et al. (2009b), and for Makapansgat the functional significance of the australopith are from Sponheimer (1999). masticatory package. Interestingly, microwear patterns for Australopithe- cus afarensis and Australopithecus anamensis also lack environment, the d13C values for the fauna, taken evidence for hard-object feeding (Ungar 2004; Grine together, show that a good deal of C4 grassy vegetation et al. 2006), but rather resemble patterns seen in was present in the environs. Therefore, the important apes, especially gorillas. These results could also be point is not that ‘Ar. ramidus was a denizen of wood- considered as inconsistent with an evolutionary trajec- land’ (White et al. 2009a), but that Ar. ramidus tory for larger molars and thicker enamel, which seem focused almost exclusively on C3 resources while to suggest adaptation to increasingly more xeric habi- avoiding nearby C4 resources, which were present.In tats (Grine et al. 2006). However, other evidence contrast, the Makapansgat hominins clearly had points in a similar direction. It has been argued that moved to exploit C4 resources (figure 3), in spite of the biomechanical masticatory complex of both the relatively closed nature of that environment A. anamensis (Macho et al. 2005) and A. afarensis (Reed 1997). The Aramis data would certainly suggest (Rak et al. 2007) is more gorilla-like. Furthermore, that, if an engagement with C4 foods marked a funda- modern humans and gorillas share life-history charac- mental shift in hominin evolution as we have argued, teristics including non-seasonal breeding (Knott then such a shift occurred post Ar. ramidus,or 2005), which in turn carries implications for a pattern elsewhere. of seasonal resource exploitation that ensures maxi- Further evidence for a possible ecological shift mum infant survival. No isotope data exist for these between Aramis and Makapansgat may reside in a older australopiths that would allow us to test whether comparison of the combined d13C and d18O data 18 C4 exploitation formed part of an increasingly seasonal (figure 4). The d O composition of enamel provides foraging round as suggested by Macho et al. (2003). a potential source of information about dietary ecology If they did engage with such resources, it will be because, in addition to the influences of hydrology and important to understand the seasonal variation. isotopic composition of precipitation, an animal’s d18O While no isotopic data yet exist for A. anamensis and value is affected by dietary ecology, drinking behaviour A. afarensis, recently published data for Aramis suggest and thermophysiology (Bocherens et al. 1996; Kohn that Ardipithecus ramidus, at 4.4 Ma, included few, 1996; Kohn et al. 1996; Sponheimer & Lee-Thorp if any, C4 resources in the diet (White et al. 2001). It has been shown that suids, some primates 2009b). Rather their isotopic composition most closely and in particular all faunivores have relatively low resembles that of savannah chimpanzees which avoid d18O compared with herbivores (Lee-Thorp & C4 resources, and contrasts with that observed in Sponheimer 2005). The reasons are not yet clear and A. africanus at Makapansgat (figure 3). Although the almost certainly differ for these groups. In the case authors lay stress on the woody nature of the Aramis of suids, it may reflect reliance on underground storage

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3394 J. A. Lee-Thorp et al. Isotopes and hominin diets

(a) 8 Kuseracolobus 6 Pliopapio Deinotherium 4 Giraffidae Tragelaphini 2 Neotragini Simatherium

VPDB 0 Eurygnathohippus O

18 Aepyceros δ –2 Nyanzachoerus Kolpochoerus –4 Hyaenidae Ardipithecus –6

–8

(b) 8

6 Cercopithecoides 4 Parapapio Diceros 2 Giraffids Tragelaphini

VPDB 0 Neotragini O

18 Simatherium δ –2 Hipparion Aepyceros –4 Suidae Hyaenidae –6 Australopithecus

–8 –16 –14 –12 –10 –8 –6 –4 –2 0 2 4 δ13C‰

Figure 4. Bivariate d13C and d18O comparisons of similar taxa in (a) Aramis and (b) Makapansgat shown as means and standard deviations. Several signiﬁcant differences are observed. On average, d18O values for Makapansgat fauna are about 2‰ lower than at Aramis, which is consistent with their relative geographical positions and associated values for hydrology. However, the Aramis data are more variable, with some exceptional and unusually low values for Deinotherium in particular. Australopithecus africanus data are relatively enriched in 13C and depleted in 18O, and occupy the same isotopic ‘space’ as the hyaenids, quite unlike Ar. ramidus. The other primate species are also more enriched in 13C, unlike Aramis, but the impala (Aepyceros) has higher d13C values at Aramis. Although impalas are generally considered to be mixed feeders, a recent study showed high d13C and almost exclusive grazing habits for Aepyceros in the nutritious grasslands of Rwanda (Copeland et al. 2009). The data for Nyanzochoerus at Aramis are remarkably similar to those for the Suidae at Makapansgat, suggesting a similar ecological niche. Data for Aramis are from White et al. (2009b) and for Makapansgat from Sponheimer (1999).

organs, and for faunivores, a high proportion of dietary consuming C4 resources, while in East African P.boisei, lipids and proteins, or a very heavy reliance on drink- this involvement might rather be regarded as a special- 18 ing water. Australopith d O data from Makapansgat ization. The exact nature of these C4 resources overlap with those of carnivores in the same strata remains unclear, and cannot be deduced from the (ﬁgure 4). Similarly low d18O values—compared with d13C values alone, but they most plausibly included other taxa in the Aramis faunal assemblage—are not various C4 resources, in varying proportions. Among observed for Ar. ramidus. Whatever the underlying the South African australopiths at least, consumption 18 contributors to the lower d O values for hominins at of C4 resources varied strongly between individuals Makapansgat (and we do not imply that they are and within individuals, in both A. africanus and necessarily the same), these observations call for P. robustus. The australopith pattern is quite unlike further study and explanation. that seen in modern chimpanzees, and indeed in early Pliocene Ar. ramidus, and we argue that it represented a fundamental shift in dietary ecology that 5. CONCLUSIONS increased dietary breadth. Additionally, the exact Carbon isotope data have demonstrated that australo- foods that contributed to the observed C4 signals in piths in South Africa increased their dietary breadth by australopith enamel are not clear. We cannot yet

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Isotopes and hominin diets J. A. Lee-Thorp et al. 3395 pinpoint when the shift occurred because no published Palaeoclimatol. Palaeoecol. 36, 69–73. (doi:10.1016/ data yet exist for A. anamensis and A. afarensis, but cer- 0031-0182(81)90049-3) tainly further research should target the period Grine, F. E. 1981 Trophic differences between gracile and between 4 and 3 Ma. robust australopithecines. South Afr. J. Sci. 77, 203–230. Grine, F. E. 1986 Dental evidence for dietary differences We thank Alan Walker and Chris Stringer for organizing the in Australopithecus and Paranthropus. J. Hum. Evol. 15, Royal Society Symposium on the ‘First four million years of 783–822. (doi:10.1016/S0047-2484(86)80010-0) human evolution’, the Royal Society for sponsoring the Grine, F. E., Ungar, P. S., Teaford, M. F. & El-Zaatari, S. meeting, Debbie Guatelli-Steinberg for her work on the 2006 Molar microwear in Praeanthropus afarensis: evi- perikymata and many colleagues for their assistance and dence for dietary stasis through time and under diverse helpful discussions over the course of many years spent paleoecological conditions. J. Hum. Evol. 51, 297–319. exploring the fascinating discipline of isotope ecology. (doi:10.1016/j.jhevol.2006.04.004) Knott, C. D. 2005 Energetic responses to food availability in the great apes: implications for hominin evolution. In ENDNOTE Seasonality in primates (eds D. K. Brockman & C. P. 1 13 12 van Schaik), pp. 351–378. Cambridge, UK: Cambridge By convention, C/ C ratios are expressed in the delta (d) University Press. notation relative to the PDB standard, as follows: d13C (‰) ¼ Kohn, M. J. 1996 Predicting animal d18O: accounting (R /R 2 1)Â1000, where R ¼ 13C/12C, and similarly, sample standard for diet and physiological adaptation. Geochim. 18O/16O ratios are expressed as d18O relative to PDB or SMOW (we use PDB in this paper). Cosmochim. Acta 60, 4811–4829. (doi:10.1016/S0016- 7037(96)00240-2) Kohn, M. J., Schoeninger, M. J. & Valley, J. W. 1996 Herbivore tooth oxygen isotope composition: effects of REFERENCES diet and physiology Geochim. Cosmochim. Acta 60, Balasse, M. 2002 Reconstructing dietary and environmental 3889–3896. (doi:10.1016/0016-7037(96)00248-7) history from enamel isotopic analysis: time resolution of Lacruz, R. S., Rozzi, F. R. & Bromage, T. G. 2006 Variation intra-tooth sequential sampling. Int. J. Osteoarch. 12, in enamel development of South African fossil hominids. 155–165. (doi:10.1002/oa.601) J. Hum. Evol. 51, 580–590. (doi:10.1016/j.jhevol.2006. Bocherens, H., Koch, P. L., Mariotti, A., Geraads, D. & 05.007) Jaeger, J. 1996 Isotopic biogeochemistry (13C, 18O) of Lee-Thorp, J. A. & Sponheimer, M. 2005 Opportunities and mammalian enamel from African Pleistocene hominid constraints for reconstructing palaeoenvironments from sites. PALAIOS 11, 306–318. (doi:10.2307/3515241) stable light isotope ratios in fossils. Geol. Q. 49, 195–204. Carter, M. L. 2001 Sensitivity of stable isotopes (13C, 15N, Lee-Thorp, J. A. & Sponheimer, M. 2006 Biogeochemical and 18O) in bone to dietary specialization and niche sep- approaches to investigating hominin diets. Yrbk Phys. aration among sympatric primates in Kibale National Anthropol. 49, 131–148. Park, Uganda. PhD dissertation, University of Chicago. Lee-Thorp, J. A. & van der Merwe, N. J. 1987 Carbon Cerling, T. E., Harris, J. M., MacFadden, B. J., Leakey, isotope analysis of fossil bone apatite. South Afr. J. Sci. M. G., Quade, J., Eisenman, V. & Ehleringer, J. R. 83, 712–715. 1997 Global vegetation change through the Miocene/ Lee-Thorp, J. A., van der Merwe, N. J. & Brain, C. K. 1989 Pliocene boundary. Nature 389, 153–158. (doi:10.1038/ Isotopic evidence for dietary differences between two 38229) extinct baboon species from Swartkrans. J. Hum. Evol. Cerling, T. E., Hart, J. A. & Hart, T. B. 2004 Stable isotope 18, 183–190. ecology in the Ituri Forest. Oecologia 138, 5–12. (doi:10. Lee-Thorp, J. A., van der Merwe, N. J. & Brain, C. K. 1994 1007/s00442-003-1375-4) Diet of Australopithecus robustus at Swartkrans deduced Copeland, S. R., Sponheimer, M., Spinage, C. A., from stable carbon isotope ratios. J. Hum. Evol. 27, Lee-Thorp, J. A., Codron, D., Codron, J. & Reed, K. 361–372. (doi:10.1006/jhev.1994.1050) 2009 Stable isotope evidence for impala Aepyceros melam- Lee-Thorp, J. A., Sponheimer, M. & Luyt, J. C. 2007 pus diets at Akagera National Park, Rwanda. Afr. J. Ecol. Tracking changing environments using stable carbon 47, 490–501. (doi:10.1111/j.1365-2028.2008.00969.x) isotopes in fossil tooth enamel: an example from Dart, R. A. 1925 Australopithecus africanus: the man-ape of the South African hominin sites. J. Hum. Evol. 53, South Africa. Nature 115, 195–199. (doi:10.1038/ 595–601. (doi:10.1016/j.jhevol.2006.11.020) 115195a0) Macho, G. A. & Shimizu, D. 2009 Dietary adaptations of Dart, R. A. 1926 Taungs and its significance. Nat. Hist. 26, South African australopiths: inference from enamel 315–327. prism attitude. J. Hum. Evol. 57, 241–247. (doi:10. DeNiro, M. J. & Epstein, S. 1978a Influences of diet on 1016/j.jhevol.2009.05.003) the carbon isotope distribution in animals. Geochim. Macho, G. A., Leakey, M. G., Williamson, D. K. & Jiang, Y. Cosmochim. Acta 42, 495–506. (doi:10.1016/0016- 2003 Palaeoenvironmental reconstruction: evidence for 7037(78)90199-0) seasonality at Allia Bay, Kenya, at 3.9 million years. DeNiro, M. J. & Epstein, S. 1978b Carbon isotopic evidence Palaeogeogr. Palaeoclimatol. Palaeoecol. 199, 17–30. for different feeding habits in two hyrax species occupying (doi:10.1016/S0031-0182(03)00483-8) the same habitat. Science 201, 906–908. (doi:10.1126/ Macho, G. A., Shimizu, D., Jiang, Y. & Spears, I. R. 2005 science.201.4359.906) Australopithecus anamensis: a finite-element approach to Elias, R. W., Hirao, Y. & Patterson, C. C. 1982 The circum- studying the functional adaptations of extinct hominins. vention of the natural biopurification of calcium along Anat. Rec. 283A, 310–318. (doi:10.1002/ar.a.20175) nutrient pathways by atmospheric inputs of industrial Parker, R. & Toots, H. 1980 Trace elements in bones as lead. Geochim. Cosmochim. Acta 46, 2561–2580. paleobiological indicators. In Fossils in the making (eds (doi:10.1016/0016-7037(82)90378-7) A. K. Behrensmeyer & A. Hill), pp. 197–207. Chicago, Ericson, J. E., Sullivan, C. H. & Boaz, N. T. 1981 Diets IL: University of Chicago Press. of Pliocene mammals from Omo deduced from Passey, B. H. & Cerling, T. E. 2006 In situ stable carbon isotopic ratios in tooth apatite. Palaeogeogr. isotope analysis (d13C, d18O) of very small teeth using

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3396 J. A. Lee-Thorp et al. Isotopes and hominin diets

laser ablation GC/IRMS. Chem. Geol. 235, 238–249. Isotopic evidence for dietary variability in the early (doi:10.1016/j.chemgeo.2006.07.002) hominin Paranthropus robustus. Science 314, 980–982. Passey, B. H., Cerling, T. E., Schuster, G. T., Robinson, (doi:10.1126/science.1133827) T. F., Roeder, B. L. & Krueger, S. K. 2005 Inverse Sponheimer, M., Codron, D., Passey, B. H., De Ruiter, methods for estimating primary input signals from time- D. J., Cerling, T. E. & Lee-Thorp, J. A. 2009 Using averaged isotope proﬁles. Geochim. Cosmochim. Acta 69, carbon isotopes to track dietary change in modern, 4101–4116. (doi:10.1016/j.gca.2004.12.002) historical, and ancient primates. Am. J. Phys. Anthropol. Rak, Y., Ginzburg, A. & Geffen, E. 2007 Gorilla-like 140, 661–670. (doi:10.1002/ajpa.21111) anatomy on Australopithecus afarensis mandibles suggests Stock,W.D.,Chuba,D.K.&Verboom,G.A.2004Distribution Au. afarensis link to robust australopiths. Proc. Natl of South African C-3 and C-4 species of Cyperaceae in Acad. Sci. USA 104, 6568–6572. (doi:10.1073/pnas. relation to climate and phylogeny. Aust. Ecol. 29,313–319. 0606454104) (doi:10.1111/j.1442-9993.2004.01368.x) Reed, K. 1997 Early hominid evolution and ecological Suga, S. 1982 Progressive mineralization pattern of change through the African Plio-Pleistocene. J. Hum. developing enamel during the maturation stage. J. Dent. Evol 32, 289–322. (doi:10.1006/jhev.1996.0106) Res. 61, 1532–1542. Schoeninger, M. J., Moore, J. & Sept, M. 1999 Subsistence Sullivan, C. H. & Krueger, H. W. 1981 Carbon isotope strategies of two ‘savanna’ chimpanzee populations: analysis of separate chemical phases in modern and fossil the stable isotope evidence. Am. J. Primatol. 49,297– bone. Nature 292, 333–335. (doi:10.1038/292333a0) 314. (doi:10.1002/(SICI)1098-2345(199912)49:4,297:: Tobias, P. V. 1967 The cranium and maxillary dentition AID-AJP2.3.0.CO;2-N) of Australopithecus (Zinjanthropus) boisei. Olduvai Gorge, Scott, R. S., Ungar, P. S., Bergstrom, T. S., Brown, C. A., vol. 2. Cambridge, UK: Cambridge University Press. Grine, F. E., Teaford, M. F. & Walker, A. 2005 Dental Ungar, P. S. 2004 Dental topography and diets of microwear texture analysis shows within-species dietary Australopithecus afarensis and early Homo. J. Hum. Evol. variability in fossil hominins. Nature 436, 693–695. 46, 605–622. (doi:10.1016/j.jhevol.2004.03.004) (doi:10.1038/nature03822) Ungar, P. S., Grine, F. E. & Teaford, M. F. 2008 Sillen, A. 1988 Elemental and isotopic analysis of mamma- Dental microwear and diet of the Plio-Pleistocene lian fauna from southern Africa and their implications hominin Paranthropus boisei. PLoS ONE 3, e2044. for paleodietary research. Am. J. Phys. Anthropol. 76, (doi:10.1371/journal.pone.0002044) 49–60. (doi:10.1002/ajpa.1330760106) van der Merwe, N. J. & Medina, E. 1989 Photosynthesis Sillen, A. 1992 Strontium–calcium ratios (Sr/Ca) of and 13C/12C ratios in Amazonian rain forests. Geochim. Australopithecus robustus and associated fauna from Cosmochim. Acta 53, 1091–1094. Swartkrans. J. Hum. Evol. 23, 495–516. (doi:10.1016/ van der Merwe, N. J. & Vogel, J. C. 1978 Content of human 0047-2484(92)90049-F) collagen as a measure of prehistoric diet in Woodland Sponheimer, M. 1999 Isotopic ecology of the Makapansgat North America. Nature 276, 815–816. (doi:10.1038/ Limeworks fauna. PhD dissertation, Rutgers University. 276815a0) Sponheimer, M. & Lee-Thorp, J. A. 1999 Isotopic evidence van der Merwe, N. J., Masao, F. T. & Bamford, R. J. 2008 for the diet of an early hominid, Australopithecus africanus. Isotopic evidence for contrasting diets of early hominins Science 283, 368–370. (doi:10.1126/science.283.5400. Homo habilis and Australopithecus boisei of Tanzania. 368) S. Afr. J. Sci. 104, 153–155. Sponheimer, M. & Lee-Thorp, J. A. 2001 The oxygen Vogel, J. C. 1978 Isotopic assessment of the dietary habits of isotope composition of mammalian enamel carbonate ungulates. S. Afr. J. Sci. 74, 298–301. from Morea Estate, South Africa. Oecologia 126, Vrba, E. S. 1985 Early hominids in southern Africa: updated 153–157. (doi:10.1007/s004420000498) observations on chronological and ecological back- Sponheimer, M. & Lee-Thorp, J. A. 2006 Enamel diagenesis ground. In Hominid evolution: past, present and future at South African Australopith sites: implications for (ed. P. V. Tobias), pp. 195–200. New York, NY: Alan paleoecological reconstruction with trace elements. R. Liss. Geochim. Cosmochim. Acta 70, 1644–1654. (doi:10. Walker, A. 1981 Dietary hypotheses and human evolution. 1016/j.gca.2005.12.022) Phil. Trans. R. Soc. Lond. B 292, 57–64. (doi:10.1098/ Sponheimer, M., Lee-Thorp, J. A., de Ruiter, D., Codron, rstb.1981.0013) D., Codron, J., Baugh, A. T. & Thackeray, J. F. 2005 White, T. D., Asfaw, B., Beyene, Y., Haile-Selassie, Y., Hominins, sedges, and termites: new carbon isotope Lovejoy, C. O., Suwa, G. & WoldeGabriel, G. 2009a data from the Sterkfontein valley and Kruger National Ardipithecus ramidus and the paleobiology of early homi- Park. J. Hum. Evol. 48, 301–312. (doi:10.1016/j.jhevol. nids. Science 326, 64. (doi:10.1126/science.1175802) 2004.11.008) White, T. D. et al. 2009b Macrovertebrate paleontology and Sponheimer, M., Loudon, J. E., Codron, D., Howells, M. the Pliocene habitat of Ardipithecus ramidus. Science 326, E., Pruetz, J. D., Codron, J., de Ruiter, D. & Lee-Thorp, 87–93. See www.sciencemag.org/cgi/content/full/326/ J. A. 2006a Do ‘savanna’ chimpanzees consume C4 5949/67/DC1. resources? J. Hum. Evol. 51, 128–133. (doi:10.1016/j. Yeakel, J. D., Bennett, N. C., Koch, P. L. & Dominy, N. J. jhevol.2006.02.002) 2007 The isotopic ecology of African mole rats informs Sponheimer, M., Passey, B. H., de Ruiter, D. J., Guatelli- hypotheses on the evolution of human diet. Proc. R. Steinberg, D., Cerling, T. E. & Lee-Thorp, J. A. 2006b Soc. B 274, 1723–1730. (doi:10.1098/rspb.2007.0330)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Phil. Trans. R. Soc. B (2010) 365, 3397–3410 doi:10.1098/rstb.2010.0052

Review Retrieving chronological age from dental remains of early fossil hominins to reconstruct human growth in the past M. Christopher Dean* Department of Cell and Developmental Biology, University College London, Gower Street, London WC1E 6BT, UK A chronology of dental development in Pan troglodytes is arguably the best available model with which to compare and contrast reconstructed dental chronologies of the earliest fossil hominins. Establishing a time scale for growth is a requirement for being able to make further comparative observations about timing and rate during both dento-skeletal growth and brain growth. The absolute timing of anterior tooth crown and root formation appears not to reﬂect the period of somatic growth. In contrast, the molar dentition best reﬂects changes to the total growth period. Earlier initiation of molar mineralization, shorter crown formation times, less root length formed at gingival emergence into functional occlusion are cumulatively expressed as earlier ages at molar eruption. Things that are similar in modern humans and Pan, such as the total length of time taken to form individual teeth, raise expectations that these would also have been the same in fossil hominins. The best evidence there is from the youngest fossil hominin specimens suggests a close resemblance to the model for Pan but also hints that Gorilla may be a better developmental model for some. A mosaic of great ape-like features currently best describes the timing of early hominin dental development. Keywords: hominin evolution; dental development; incremental markings; tooth root growth; enamel; dentine

1. BACKGROUND Smith (1989) has shown that certain key marker The lives of all living organisms can be divided into events during dental development actually correlate stages. This allows comparisons to be made between better with important variables that describe life- them. There are many reasons for studying the stage history variation than any of these life-history variables or period of growth in primates in a comparative con- do with each other. Because of this, some tentative text, which include identifying those ontogenetic inferences can be made about the way fossil primates changes shared by all primates and those that are lived their lives compared with living primates that unique to modern humans (Schultz 1937). Relative go beyond simple relative dento-skeletal comparisons. comparisons of the stages of skeletal or dental A powerful aspect of dental biology is that tooth tissues growth have proved to be a useful way of deﬁning simi- preserve an incremental record of their growth, which larities and differences between both living and fossil remains literally embodied within the microstructure primates. When chronological age is known, then the of enamel and dentine. This offers an opportunity to length of the phases of growth as well as the rates of reconstruct the period of maturation in fossil primates growth of individuals can be compared. Dental devel- and compare them in real time with living primates. opment is just one measure of biological maturity, but Even if it may never be possible to retrieve information is arguably the most stable, and it occurs over an unu- about many life-history variables from the fossil sually long period of time from before birth to record, it should be possible to reconstruct a time maturity. Besides enabling us to discover things scale for growth in the past. about the evolutionary history of our own growth period, studies of comparative dental development provide us with an opportunity for investigating the 2. INCREMENTAL GROWTH OF ENAMEL biological processes that govern tooth formation from AND DENTINE the initial mineralization of teeth to the completion The cells that form enamel and dentine (ameloblasts of their roots (Swindler 1985). and odontoblasts) secrete their matrix in a rhythmic manner (Bromage 1991; Smith 2006; Bromage et al. 2009). A circadian rhythm in cell function is expressed *[email protected] as a daily slowing of secretion during enamel and den- One contribution of 14 to a Discussion Meeting Issue ‘The ﬁrst four tine formation and is still manifest in the enamel and million years of human evolution’. dentine microstructure of fully formed teeth as a

3397 This journal is # 2010 The Royal Society Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3398 M. C. Dean Review. Dental development in early hominins

junction between enamel and dentine, provides a way (a) (b) of estimating past rates of differentiation of new secretory cells during tooth formation (Boyde 1963, 1964, 1990b; Shellis 1984; Dean 1985; Risnes 1986). The rate of increase in both tooth crown height and root height can be reconstructed by dividing increments of tooth crown length along the enamel dentine junction (EDJ), or cement dentine junction (CDJ) by the time intervals taken to form them (Boyde 1963; Risnes 1986; Dean 2006, 2009). In figure 2, consecutive 200 mm-thick increments of enamel and dentine have been used to plot increasing tooth height against time from the dentine horn to a point as close to completion of the root as possible (Dean 2009). Figure 1. Scanning electron micrograph showing perikymata The number of daily increments between long- on the upper lateral incisor of MLD 11 (Au. africanus)from period markings appears always to be the same in Makapansgat, South Africa (a). Their spacing becomes each of the teeth of an individual but it varies between closer towards the cervix. Two or three regions of enamel individuals. In large samples of individuals there are hypoplasia are evident indicating periods of slowed enamel also outliers with a long-period rhythm of 6 or 11 or growth during tooth formation. (b) Transmitted polarized even perhaps 12 days. We now know that these long- light micrograph of enamel incremental markings seen in a period markings (first described by Anders Retzius longitudinal ground section of a probable second molar frag- (1837) and, therefore, also referred to as Retzius ment (Ward et al. 2001) attributed to Au. anamensis (KNM- lines) occur in the enamel of other primates including ER 30748) from Allia Bay, Kenya. Coarse oblique long early fossil hominins (figure 1). Of 29 australopiths period incremental markings (approx. 35 mm apart) run from bottom left to top right and emerge at the surface examined so far, 17 (59%) showed a mean periodicity within perikymata troughs. Along prisms, that run left to of 7 days and of seven early Homo specimens examined right in this image, are short period daily increments so far, two had long-period lines 7 days apart, four marked by fine cross striations approximately 5 mm apart. were 8 days apart and one was 9 days apart (Lacruz In this specimen, there are seven daily increments between et al. 2008). Fossil teeth, however, are precious and adjacent long period striae. it is only rarely possible to employ partially destructive techniques to retrieve data about their growth. Never- theless, long-period markings also create a furrow or daily incremental marking (Boyde 1976, 1979, 1989, trough on the external surface of permanent tooth 1990a; Shinoda 1984). Thin sections of teeth pre- enamel. These so-called perikymata (waves around pared for histological analysis, or even polished or the tooth) first defined by Preiswerk (1895) in ungu- naturally fractured surfaces of fossil teeth that are suit- late enamel are well preserved on many early able for examination with various kinds of microscopy hominin teeth and they can be counted with scanning (figure 1) can be used to reveal these markings in electron microscopy (figure 1) or even in oblique- enamel and dentine (Boyde 1989, 1990b; Dean reflected light. They can be used to estimate enamel 2000, 2006; Lacruz et al. 2006, 2008). Counts of formation times in fossil teeth since counts of periky- daily incremental markings in the teeth of individuals mata are equivalent to counts of long-period striae with known dates of birth and death match very closely within the tooth but their periodicity may not be with the number of days of life (Antoine et al. 2009). known unless the internal structure of the enamel The daily increments of enamel secretion in great can be visualized. ape and fossil hominin teeth cumulate at a faster rate than they do in modern human tooth enamel (Dean et al. 2001). Enamel measuring 200 mm thick over 3. CONSTRUCTING A COMPARATIVE MODEL the cusp of a great ape tooth, takes on average between FOR EARLY HOMININ MATURATION 55 and 65 days to form, whereas in modern humans Recent evidence about DNA sequence analysis and this takes 70–80 days (Dean et al. 2001; Dean 2009; from molecular biology suggests that modern Smith et al. 2006). Daily rates of dentine formation, humans and chimpanzees are more closely related to however, are more similar in great apes and humans each other than to any other living ape (Goodman and take between 80 and 100 days to form 200 mm et al. 1994; Ruvolo 1994; Bradley 2008). It is, there- close to the root surface (Dean 2009, in press). fore, not an unreasonable assumption that the last Another, but longer period rhythm, that also slows common ancestor of the Pan–Homo clade had a life dental hard tissue formation in a regular way is super- history more like that of modern chimpanzees than imposed upon this daily rhythm (figure 1). In modern modern humans (Robson & Wood 2008). It is none- humans this coarser more prominent marking usually theless equally likely that among the species of early occurs every 7, 8, 9 or 10 days with a modal value of hominins there were many different life-history strat- 8days(Smith et al. 2006). Long-period incremental egies that spanned what we know today about life markings are aligned along the original mineralizing history in modern orangutans, chimpanzees, bonobos tissue front in both enamel and dentine. The slope and gorillas. One key question that we can then ask of these incremental markings, with respect to the is whether there is any evidence among early hominins

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Review. Dental development in early hominins M. C. Dean 3399

(a)(b) 25 000

20 000

15 000

10 000

5000 tooth height along EDJ ( m m)

0 2468101214161802 4 6 8 10 12 14 16 18 tooth formation time (years) tooth formation time (years) Figure 2. Plots of M1, M2 and M3 formation time (years) against increasing tooth height (mm) along the mesiobuccal (protoconid) EDJ (red open circles), continued along the CDJ (open blue circles) for (a) Pan troglodytes and (b) modern human molars. The mean length (+1 s.d.) of mesiobuccal root formed at gingival emergence in free-living Pan specimens given in Kelley et al. (2009), M1, 4.2 mm; M2, 5.2 mm; M3, 6.8 mm has been used to generate a likely range of root lengths (and thereby corresponding ages) where molars in Pan might have emerged into the mouth (yellow filled circles). Arrows denote the median ages of gingival emergence. The rates of crown and root growth, as well as the total tooth formation times in modern humans and Pan, are similar but earlier initiation times compress Pan molar development into approximately 12 years rather than approximately 18 years. Distance curves for a single Gorilla M1, a short M1 fragment of KNM-ER 30749 and a longer M2 fragment of KNM-ER 30748 (both attributed to Au. anamensis, Ward et al. 2001) are superimposed over the Pan M1 and M2 molars (filled black circles, crown; filled blue circles, root). Rates of root extension in Gorilla and Au. anamensis are faster than in Pan. for a period of maturation that differs from that known modern humans (Zuckerman 1928). In recent years today for modern chimpanzees. Another feasible ques- many issues have been clarified through studies on tion is whether there is evidence among the various samples of captive animals of known chronological species of early hominins for any differences between age. Parallel histological studies of enamel and dentine them in the timing of dental development that might growth in great apes have also helped to build a better point to the presence of different life-history strategies picture of the chronology of dental development in a existing together during the first four million years of comparative context. What follows is a synthesis of human evolution. If this were so it might point to inter- those studies.1 esting links with climate change or diet. To answer these questions requires a detailed knowledge of the (a) Permanent tooth eruption times in Pan chronology of dental development in modern great In two classic longitudinal studies on chimpanzee apes and an assessment of how early fossil hominins dental emergence, Nissen & Riesen (1945, 1964) pre- do or do not compare with this. Since this is realisti- sented the first reliable data for ages of gingival cally only presently possible for Pan troglodytes,it emergence (eruption) in captive chimpanzees. They makes sense to try and construct a model that brings showed that the deciduous dentition was fully emerged together everything that is known about the timing of into functional occlusion by approximately one year of dental development in P. troglodytes and use this to age (Nissen & Riesen 1945) and that for eight males examine the timing of dental development in early and seven females combined (Nissen & Riesen hominins. 1964), the mean ages of emergence for M1 were 3.3 years (range, 2.6–3.8); M2, 6.7 years (range, 5.6– 7.8); and M3, 10.8 years (range, 9.0–13.6). All of 4. THE CHRONOLOGY OF DENTAL these molar eruption ages are much earlier than DEVELOPMENT IN PAN TROGLODYTES those known for modern humans. Interestingly, how- Early studies of dental development in great apes were ever, the equivalent data for incisors and canines are made on single individuals or on samples of animals indistinguishable from those known for modern brought to zoos or acquired for comparative skeletal humans. Mean gingival emergence ages for those collections (Keith 1899; Schultz 1924, 1935, 1940; teeth are, respectively, I1: 5.7 years (range, 4.5–7.0); Zuckerman 1928; Krogman 1930; Bennejeant 1940; I2: 6.4 years (range, 5.0–8.3); C: 9.0 years (range, Clements & Zuckerman 1953). Few of the living ani- 7.6–10.1). Kuykendall et al. (1992) in a study of 22 mals studied were actually born in captivity and so male and 36 female laboratory born and raised chim- their chronological age was rarely known. While panzees aged between 1 and 10 years observed some early studies identified differences in the median emergence times for permanent teeth that sequence of dental eruption between great apes and were rarely more than a month or two different from humans and also noted earlier ages for the eruption the data of Nissen & Riesen (1964). One exception of certain teeth, others found no differences in the was the permanent canine that in both the mandible timing of dental development between great apes and and maxilla erupted a year earlier at approximately

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3400 M. C. Dean Review. Dental development in early hominins

8 years (range 6.5–8.7). Unfortunately, the age sample M1 M1 M1 M1 M1 M2 M2 M2 M2 M2 M3 M3 M3 M3 M3 of Kuykendall et al. (1992) did not extend to individ- 100 uals with emerging M3s, but the general consistency for dental emergence ages between these samples of 80 laboratory-raised chimpanzees is remarkable. How- ever, these data are not so closely reﬂected by those 60 derived from a much smaller sample of free-living chimpanzees of known chronological age originally 40 described by Zihlman et al. (2004) but subsequently percentile revisited by Smith et al.(2009, 2010). 20

0 (b) Environmental effects on dental development in great apes 2 3456789 101112 An important issue that is still not well understood is age at tooth eruption (years) the effect on great ape dental development of being Figure 3. Predicted median ages of attainment (50th percen- born and raised in captivity and perhaps more signifi- tile) for molar gingival emergence ages in Pan (M1, 4.1 cantly, the effect of being nursed and raised by the years; M2, 6.9 years; M3, 10.1 years). These are derived mother in captivity or being hand-nursed and bottle- from the times corresponding to root lengths formed at gin- fed by humans. Zihlman et al.(2004, 2007) have pre- gival emergence shown as yellow filled circles in figure 2. The sented a range of data for free-living chimpanzees that interquartile age ranges (vertical lines) of attainment were demonstrate a slower rate of behavioural, somatic and simulated by using 1 s.d. of root formed at gingival emer- dental development than for captive animals. They gence. The median values are represented by central place M1 emergence at approximately 4 years, M2 vertical lines. A spread of initiation times over a lower age between 6 and 8 years, canine emergence between range, which in reality seems likely, would, reduce these estimates slightly. 10 and 11 years and M3 emergence at approximately 12.5 years. Smith et al. (2006) have also illustrated a wild-born chimpanzee aged 4.4 years (fig. 6 in Smith What lies behind this difference is likely to be multi- et al. 2006), where M1 is still far from functional factorial but certainly to a large extent nutritional. occlusion. Phillips-Conroy & Jolly (1988) and Lippert (1977) showed that captive hand-reared Kahumbu & Ely (1991) also recorded later eruption infant apes double their birth weight by three times in free-living than in captive baboons. Even if months, whereas mother-nursed infants do not do so the degree of difference is both population- and until six months. This difference persists until at sample size-sensitive, some degree of difference is cer- least 21 months and probably continues as a trend tainly real. Kelley et al. (2009) used the extrinsic into adulthood (Nissen & Riesen 1945; Fooden & staining on newly emerged cusps of molar teeth to Izor 1983). Nissen & Riesen (1945), Marzke et al. indicate gingival emergence in wild-collected great (1996) and Winkler et al. (1991) have all noted ape skulls. In figure 2, the mean lengths for mesiobuc- advanced deciduous tooth emergence ages between cal roots (+1 s.d.) measured at gingival emergence in mother-nursed and formula-fed infant great apes. that study are plotted individually onto each Pan molar Marzke et al. (1996) specifically made the point that root (M1, n ¼ 14; M2, n ¼ 10; M3, n ¼ 8). The age data from mother-nursed captive animals are likely to ranges generated for these root lengths have been be more directly relevant to free-living conditions used to simulate a likely range and median age of than data for hand-reared great apes. The available attainment for gingival emergence in the predomi- data for great ape dental development needs, there- nantly wild-collected Pan specimens represented in fore, to be considered carefully in this light if a figure 3. The results are a close match with those of model for dental development in fossil hominins is to Zihlman et al. (2004) for M1 and to some extent for be realistic. M2 with simulated median age of attainment of M1 at approximately 4.0 years, M2 at approximately 7.0 years and M3 at approximately 10 years. Interestingly, (c) Tooth initiation times, sequences they also fall close to the 32.6, 59.4 and 86 per cent of and overlaps the total time to complete dental development that Swindler (1985), Anemone et al. (1991), Anemone & Swindler (1985) calculated for modern human molar Watts (1992), Kuykendall (1996) and Reid et al. eruption times, assuming that this total time is (1998) have all noted that the times for initial mineral- approximately 12 years in P. troglodytes. However, ization of the permanent incisors and canines in with the exception of M1, these simulated median Pan are very similar to those described for humans ages of attainment for molar gingival emergence still (Kronfeld 1935) and that the sequence of minerali- fall within the ranges reported for captive chimpan- zation is identical. Lower permanent incisors initiate zees. Kelley & Schwartz (2009) have drawn attention at approximately 3–4 m after birth and canines at to the wide range of ages likely for gingival emergence 4–5 m although both earlier and later times have in free-living great apes but Smith et al. (2010) have been recorded (Kronfeld 1935; Anemone et al. 1991; suggested that ages for gingival emergence may be Kuykendall 1996; Winkler 1996; Schwartz et al. influenced more by free-living or captive rearing than 2006). Winkler (1995) demonstrated that direct obser- crown or root formation are. vations of tooth germs can pick up earlier initiation

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Review. Dental development in early hominins M. C. Dean 3401 times than radiography as Hess et al. (1932) and has as much to do with how long or short the crown Beynon et al. (1998) also observed. Reid et al. (1998) formation period of the earlier forming molar is as it have nonetheless noted generally similar initiation has to do with the early or late initiation of the sub- times in Pan from histological studies for lower I1 sequent forming molar. Smith et al. (2006) noted the (range, 1.8–5.6 m), lower I2 (range, 2.3–8.5 m) lower potential effect earlier or later molar initiation might canines (range, 4.6–6.9 m) as well as for P3 and P4 have on eruption timing and highlighted the need to initiation (range, 1.1–1.95 years) something also document the degree of molar overlap in free-living observed by Anemone et al.(1991)and Kuykendall versus captive animals and in hand-reared versus (1996) to between 1.4 and 1.8 years in Pan. mother-nursed animals. Indeed in figure 2, a spread Understanding the differences in dental development of molar initiation times towards earlier ages would that exist between great apes that take approximately reduce the simulated estimates for median emergence 12 years to grow up and modern humans that take times in figure 3. Winkler et al. (1996) studied a much approximately 18 years to grow up is fundamental to larger originally free-living sample of 89 orangutans our being able to interpret juvenile fossil hominin and concluded that sequential molars in orangutans material. It is molar development that reflects these had usually begun to mineralize by the time a previous somatic growth differences most closely. While the molar had reached crown completion but that variabil- sequence of molar initiation is also always identical in ity was too high to consistently predict that crown great apes and humans (M1, M2, M3), the timing of initiation had always commenced prior to crown com- molar initiation has been much debated. Molar for- pletion of a preceding molar. Smith et al. (2006) in a mation is drawn out in modern humans between birth histological study of Pan molars from the same individ- and approximately 18 years. Initiation of M1 around uals again noted variability in the degree of molar birth is followed by M2 initiation at approximately 3 overlap but that more often than not this was consider- years and M3 initiation at approximately 8 years with able. The effects of captive rearing, therefore, cannot each molar then taking about 10 years to form. yet be resolved. Dean & Wood (1981) suggested that molar initiation times were compressed together in great apes (M1 close to birth, M2 at 2.5 years and M3 at (d) Total tooth and root formation times 5 years) and that total molar formation times were Kuykendall (1996) made the important observation shorter, with M3 root formation completing between that the overall duration of crown and root formation 11 and 12 years. Certainly, studies of dissected M1 in chimpanzee permanent incisors and canines is com- germs in great apes have usually, but not always, parable with that in modern humans. Some of the best demonstrated three or four mineralizing cusps at summary data for mean age at entering a formation birth (Oka & Kraus 1969; Tarrant & Swindler 1972; stage for modern humans (Liversidge 2009) places Moxham & Berkovitz 1974; Winkler 1996). Schwartz apex closure for mandibular I1 (8.04 years), I2 (8.69 et al. (2006) have described an extreme case of M1 years), canines (12.2 years) and M1 (9.38 years) at initiation in a captive hand-raised gorilla as early as very close to the recorded ranges reported for Pan 90 days before birth. However, Anemone et al. (Anemone et al. 1991; Anemone 1995; Kuykendall (1991) and Anemone (1995) in the first longitudinal 1996). The median values and ranges of ages for radiographic studies of dental development in captive root apex completion in Pan are: I1, 9.55 years P. troglodytes, showed that for 16 individuals, the pro- (range, 7.99–10.75); I2, 9.69 years (range 8.35– posal of Dean & Wood (1981) for M2 and M3 10.75); C, approximately 12 years. initiation in Pan was incorrect. They demonstrated When initiation times are taken into account, total even earlier initiation times for M2 at 1.5 years and molar formation times in Pan appear to come close to M3 at 3.5–4.0 years. those for modern human molars (approx. 10 years). Subsequently, Kuykendall (1996) in an extensive Data for 30 observations of mean M1 apex closure cross-sectional radiographic study imaged stages of (M1, 8.14 years, range 6.47–10.75 years) given in tooth formation in 118 captive chimps and reported Kuykendall (1996) and data for M2 and M3 from even younger median ages of molar initiation: M2, Anemone et al. (1991) and Anemone (1995) also 1.3 years (range, 1.15–1.48) and M3, 3.2 years suggest overlap in total M2 and M3 formation times (range, 3.0–4.6). Despite the supposed advantage of with modern humans (M2, 6.5–9.8 years; M3, 11– picking up initial mineralization of tooth germs earlier 13 years). Kuykendall (1996), however, concluded in histological studies, Reid et al. (1998) estimated M2 that, unlike incisors and canines, total molar formation initiation in Pan at between 1.7 and 1.9 years and M3 times in Pan were in fact slightly shorter than those initiation between 3.6 and 3.8 years. In figure 2, for known for modern humans but since no longitudinal consistency, and because for isolated teeth initiation studies exist with sufficient samples of older times can never be known, histological estimates for animals this remains speculative. Nevertheless, the molar initiation times have been used and fixed at wide range of ages reported for molar apex closure the average age for molar tooth types estimated in in great apes is noteworthy. Beynon et al. (1991) illus- Reid et al. (1998): M1, birth; M2, 1.75 years; M3, trated a gorilla with an open M1 apex at approximately 3.69 years. 6 years with a little more root growth to come (see also The issue of early molar initiation in great apes has the additional gorilla M1 in figure 2) and Schwartz become confused with observations about the degree et al. (2006) yet another gorilla at the same stage but of overlap in crown formation periods between aged only 3.2 years. These data suggest that total sequential molars. Overlap of crown formation periods molar formation times in Gorilla may be shorter than

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3402 M. C. Dean Review. Dental development in early hominins

n = 2 n = 2 5.5 5.8 5.2 5.5 4.8 5.1 4.2 4.6 3.5 4.0 2.8 3.4 2.3 2.9 2.5 1.8 1.0 2.1 1.4 1.0 1.5 1.8 1.5 0.7 2.2 1.0 0.7 0.8 1.0 2.9 1.4 1.1 1.2 1.8 1.4 3.7 1.5 2.4 1.8 1.9 4.4 2.3 3.0 2.4 2.9 5.1 3.6 2.9 3.4 3.4 5.8 4.2 4.1 4.0 4.9 6.5 4.6 4.4 5.5 5.0 4.7 7.0 6.1 5.4 7.3 6.5 male n = 10 female n = 8 n = 7 n = 6 Figure 4. Anterior crown formation times in Pan troglodytes. Crown heights are divided into deciles and an average chronological age of attainment for each decile of height has been calculated from estimates of enamel formation time made from counts of daily cross striations and long period striae in longitudinal ground sections of teeth. (Data from Reid et al. 1998, 2000; Schwartz & Dean 2001; Schwartz et al. 2001.) those reported for Pan but sample size and these data specimens of Pan with data taken from Reid et al. for captive animals may be misleading. (1998, 2000), Schwartz & Dean (2001) and Schwartz et al. (2001). However, the data for one or two specimens of Gorilla published by Beynon et al. (1991) (e) Crown formation times and Schwartz et al. (2006) suggest crown formation Perhaps, the most debated aspect of great ape dental times for incisors may sometimes be as short as 2.7 development is the time taken to form enamel, or years suggesting that Pan may be atypical in this crown formation time. The reason for this is that it sense. Once again this raises the question of potentially bears heavily on whether early fossil hominins can be advanced dental development in captive animals or judged ‘ape-like’ or ‘human-like’ with respect to this perhaps of significant differences between Pan and formation time. However, enamel formation time Gorilla that are currently unappreciated and in may be defined differently in radiographic studies addition, whether a Pan-like model for dental develop- and histological studies of tooth development ment is actually the most appropriate for early (Beynon et al. 1998; Kuykendall 2001) and an added hominins. complication is that in histological studies, different Knowing something about anterior crown formation enamel formation times are often estimated for each times allows us to link periodic linear hypoplastic band- cusp of a molar tooth (Smith et al. 2006). For these ing patterns on anterior teeth that are common in many reasons and others many comparisons of enamel Pan and Gorilla specimens collected from locations in formation times between modern humans and West Africa (Gabon, Cameroon) with two rainy seasons great apes have often been either unconvincing or each year (Skinner & Hopwood 2003). Besides being incomparable (Kuykendall 2001). generally under the weather in colder wetter conditions, With the exception of lower canines (Schwartz & chimpanzees in particular are more susceptible to Dean 2001; Schwartz et al. 2001), the data for anterior increased intestinal parasite loads (Lilly et al. 2002) crown formation times in great apes is very poor. since damp soil and sporadic forest floor flooding Figure 4 summarizes what is known for a few present prefect conditions for eggs, protozoa and

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Review. Dental development in early hominins M. C. Dean 3403

40 not all, the root extension rate then rises to a peak and falls off again. In very long tooth root, there may 35 again be a rise in root extension rate towards apex 30 completion. In the small sample of Pan, M1 mesiobuccal roots plotted in ﬁgure 5, the mean age at which this ) 25

–1 early peak in root extension rate occurs (3.8 years, d 20 range ¼ 3.01–4.65 years, s.d. 0.48) is close to that ( m 15 reported for gingival emergence. While this apparent 10 signature for gingival emergence is unlikely to be a simple epigenetic reﬂection of tooth movement 5

crown and root extension rate and root extension crown through the bone, it is intuitive that root extension 0 rates should begin to fall off at the time of initial func- 0–1 123456789 tional occlusion. When the data for extension rates are tooth formation time (years) aligned around this early mean peak height velocity (PHV) rather than plotted with respect to initiation Figure 5. Extension rates for the same sample of Pan M1s at birth, this small rise in root extension rate is more shown in figure 2. Rates are high in the cusps of the crown 21 easily seen and not smoothed out (figure 5). For this but then quickly fall to values between 4 and 8 mmd sample of Pan M1s, PHV is 8.7 mmd21 (range 6.1– (see also Dean 2009). These data for 14 teeth are each 10.2 md21). The expression of this peak in early aligned (arrow) around the mean age (3.8 years) of peak m height velocity (PHV) for this sample, which displaces the root growth becomes weaker distally in M2s and initiation and completion of tooth formation to earlier or M3s but is still there although it occurs later into later ages but highlights the root spurt more clearly. The root formation. PHV in this sample of M2s was 1 mean chronological age and range of ages at which PHV 6.7 mmd2 and occurred at 4.7 years (range 3.4–6.4 occurs during early root formation (3.01–4.65 years, s.d. years, s.d. 0.85) into tooth formation. 0.48) broadly mirrors those ages reported for gingival emergence in Pan M1s. (g) Summary points about dental development in P. troglodytes helminths to flourish. Seasonal fluctuations such as this The sequence and times of initiation as well as total increase the likelihood of individuals succumbing to any tooth formation times of incisors and canines are number of conditions that are known to underlie linear little different from modern humans. The ages of gin- enamel hypoplasias (particularly prolonged bouts of gival emergence of incisors and canines are also little diarrhoea or dysentery) and are a likely explanation different. Anterior crown formation times (with the for many wild-collected great ape permanent canines exception of male canines crowns which take longer having, for example, 15 or so faint bands on canine to form) are only slightly longer than average crowns that took close to 7.5 years to form enamel. modern human crown formation times. It is the The data presented in figure 2 for molar crown initiation times and eruption times of the molar denti- (protoconid) formation times in Pan are for slightly tion in modern humans that are drawn out to later ages bigger sample sizes than previous studies (but compar- with prolongation of the growth period. The greatest able to those of Smith et al. 2006, 2010) although they shift in timing appears to be in eruption times, which are not based on counts or error-prone periodicities of can be observed both at later stages of root formation long period incremental markings but only on counts in modern humans as well as at later ages than in Pan. of daily increments close to the EDJ: M1, 2.3 years This is most marked in M3 that initiates approximately (range 1.78–2.66); M2, 2.38 years (range 1.72– 4.5 years later in modern humans than in Pan and 3.19); M3, 2.71 years (range 2.19–3.34). Mean which erupts into functional occlusion approximately values for modern human (protoconid) formation 8 years later at close to 18 years of age. Average total times (Reid & Dean 2005) are greater than these: molar tooth formation times in Pan are shorter than M1 (3.1 years), M2 (3.2 years) and M3 (3.27 years) those in modern humans, but it only seems by between but there is overlap in the ranges (for example, see one and two years, and while molar crown formation Reid & Dean 2005 and Mahoney 2008) such that an times are also shorter on average, this is only by six individual molar tooth could not always be attributed to nine months with overlapping ranges. It appears to Pan or Homo on the basis of molar crown formation (figure 2) that there is little or no difference in the time alone. rate of growth in height of the molar crowns or roots between Pan and modern humans. Besides these comparisons of timing in tooth formation, it may well be (f) Rates of root formation and the timing that great ape teeth contain information about seasonal- of gingival emergence ity and perhaps even about their own eruptive history. All hominoid teeth show a pattern of change in extension rate that is dominated by an initial high rate in the cusps of the crown but which then quickly reduces to a 5. THE EVIDENCE FOR A CHRONOLOGY OF more constant rate in the lateral enamel (Dean 2009). DENTAL DEVELOPMENT IN FOSSIL HOMININS The transition from cervical enamel formation to (a) Molar eruption times cervical root formation in hominoid teeth appears to Bromage & Dean (1985) estimated the age at death of occur without any abrupt change to the rate of four early hominin specimens with M1 just prior to or growth in tooth length (figure 5). In many teeth, but at functional occlusion (Sts 24, Australopithecus

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3404 M. C. Dean Review. Dental development in early hominins africanus;LH2,Au. afarensis; SK 62 and SK 63, Para- Unlike Pan no evidence exists to show early nthropus robustus) to be 3.2–3.5 years on the basis of initiation of M3 in fossil hominins. Stw 151, from perikymata counts on lower permanent incisors. Sub- the late Member 4 breccia deposit at Sterkfontein, is sequently, a histological study of SK 63 (Dean et al. described as a specimen with a dentition ‘not fully dis- 1993) placed gingival emergence nearer to 4 years. tinct from that of Au. africanus but with a cranial More recently, Lacruz et al. (2005) estimated M1 morphology more derived in some characters’ emergence into occlusion and age at death of the (Moggi-Cecchi et al. 1998). While there is a small Taung child (Au. africanus) to be between 3.73 and mandibular M3 crypt in Stw 151, it is still too small 3.93 years on the basis of M1 enamel formation to have accommodated a mineralizing tooth germ, times and length of mesial root formed (5–6 mm: which must, therefore, have initiated after M2 crown Conroy & Vannier 1991a,b). completion. Another specimen (SK 63, attributed to The only evidence for the status of the developing P. robustus) contains M2 crowns that are not quite dentition in Australopithecus of any species around completed but at this stage, only incipient M3 crypt the age of M2 eruption comes from scans and radio- depressions in the root of the ascending mandibular graphs of MLD 2 and Stw 327, both of which are rami are present. Certainly, M3 initiation could not described as having M2s recently in functional occlu- have occurred prior to M2 crown completion in this sion (Skinner & Sperber 1982; Conroy & Vannier specimen. 1991a). While the M3 crypt of MLD 2 is only partially preserved, CT scans of Stw 327 show a completed M3 crown (Conroy & Vannier 1991a). Skinner & Sperber (c) Total tooth formation times (1982) have drawn attention to the permanent canines The evidence for total anterior tooth formation times of MLD 2, which are still deep in their crypts and that in early hominins is lacking but what there is suggests on CT scans (Conroy & Vannier 1991a) have less root little difference from Pan. Median ages for combined length formed than crown length. However, no histo- sexes in Pan for lower incisors at the same develop- logical age estimates are possible for either of these mental stage (Kuykendall 1996) would estimate age specimens. But if the timing of canine root length at death of Stw 151 at 4.95 years (range 4.61–5.22). formed could be shown to match that known for This is very close to the histological estimate of Pan, then the standards deﬁned by Kuykendall 5.2–5.3 years for this specimen (Moggi-Cecchi et al. (1996) would provide a median age estimate of 7.6 1998). There is then no evidence to suggest that the years for MLD 2 (range 6.10–8.75)—but this is timing of root formation in this early hominin was speculative. This age range, however, spans the range different from that observed in Pan. Both standards of histological estimates for age at death of KNM- for lower lateral incisors in modern humans (8.0 WT 15000 (attributed to Homo erectus at 1.5 Ma), years, s.d. 0.99; Liversidge 2009) and Pan (8.04 where M2s were also just in functional occlusion and years, inter-quartile range 7.66–8.86; Kuykendall where the one preserved upper M3 crown was also 1996) also each give median age at death estimates just complete (Dean & Smith 2009). that match histological estimates for the H. erectus youth from Nariokotome (7.6–8.8 years, Dean & Smith 2009). Again this suggests that there is no evi- (b) Molar initiation times dence for any change in total incisor tooth formation The approximately 3-year-old infant hominin dated to times in early hominins, but histological evidence for 3.3 Ma from Dikika, Ethiopia (Dik-1-1), and attribu- ages of hominin specimens with near completed ted to Au. afarensis had M1 crown complete with just canine roots are needed to show that this also holds 1.6 mm of mesiobuccal root formation (Alemseged true for canines. et al. 2006). The occlusal surface of the M2 crown Stw 151 is aged histologically to between 5.2 and had already formed. This is clear evidence for early 5.3 years at death (Moggi-Cecchi et al. 1998). It had initiation of M2 and of overlap in molar enamel for- M1s with one or more incomplete root apex at the mation. KNM-ER 1477, a juvenile P. boisei mandible time M2, premolar and canine crowns had just com- roughly the same chronological age, preserves the pleted enamel formation. This age implies that root mesial portion of what may have been a well-formed apex closure of M1 was at the earliest end of the age M2 crypt. However, this is the only other potential evi- range reported for Pan and occurred close to M2 dence in any early hominin specimen that M2 may crown completion. The end of M2 crown completion already have been mineralizing at the time of M1 in KNM-WT 15000 (H. erectus) was also estimated crown completion. The mineralizing P4 in this speci- to have completed between 4.2 and 4.9 years on the men would be expected to match M2 but this basis of perikymata counts (Dean & Smith 2009). cannot be known. A number of juvenile P. boisei or Other early hominin specimens from Laetoli, LH 3 P. r o b u s t u s specimens exist with M1 at or close to and LH 6 (attributed to Au. afarensis) consist only of crown completion (Skinner & Sperber 1982; Dean isolated teeth (White 1977). However, in both speci- 1987; Conroy & Vannier 1991b; Lacruz 2006). Relative mens, the upper M1 is at a similar stage of root apex to this stage of M1 formation several of the P. r o b u s t u s formation as Stw 151 and each have permanent specimens (SK 438, SK 64, SK 3978) appear to canine and premolar crowns close to or just com- show delayed premolar formation when compared pleted. Close correspondence of the canine crown with the P. boisei specimens (KNM-ER 1477, KNM- perikymata counts (Stw 151 ¼ 140 and LH 6 ¼ 134) ER 1820) and one explanation for this might be shorter suggests that LH 6 was close in age with the same pat- M1 crown formation times in P. r o b u s t u s . tern of tooth formation and the same early age for

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Review. Dental development in early hominins M. C. Dean 3405

M1 root completion. No M2s are preserved for SK 3978) have M1 at or close to crown completion (- comparison in either of these specimens from Laetoli. Skinner & Sperber 1982; Dean 1987; Conroy & While speculative, a tooth fragment from Allia Bay, Vannier 1991b; Lacruz 2006). Some have been aged Kenya, attributed to Au. anamensis (KNM-ER 30748) on the basis of perikymata counts on anterior tooth may contain information about molar eruption in early germs at between 2.5 and 3.0 years of age at death hominins. It is plotted in figure 2 as an M2, since it has (Dean 1987) but with some root formation. This fits an enamel formation time (2.7 years) beyond the range well with a histologically derived estimate of 2.4 of the Pan M1s sampled here (see also Ward et al. years for M1 crown formation time in SK 63 (P.robus- 2001). It contains a marked early root spurt of tus) from Swartkrans, South Africa (Dean et al. 1993). 9.8 mmd21 at 4.2 years into tooth formation that is Lacruz & Ramirez Rozzi (2010) have made histologi- within the M2 range for Pan (3.44–6.38 years). If cal estimates of metaconid as well as total crown M2 initiation in Au. anamensis was close to 1.75 formation times of two Au. afarensis molar fragments years, as in Pan, and if early root PHV actually reflects (AL 333-52 and AL 336-1) at between 2.2 and 2.4 the eruptive process, then this would place functional years. Beynon & Wood (1987) calculated a range of occlusion of this tooth towards the lower end of the molar crown formation times of 2.12–2.59 years in range reported for M2 in Pan (5.6–7.8 years) P. boisei, while Ramirez Rozzi (1993, 1995) found (Nissen & Riesen 1964). ranges of 1.93–2.49 years for P. aethiopicus but a A number of chronologically older hominin speci- greater range for enamel formation times of mens exists with incomplete M3 roots. KNM-ER P. boisei molars of all types (2.67–3.43 years). In 1802 has a left M3 with just one wear facet on the pro- P. robustus from Kromdraai, Lacruz (2006) calculated toconid and a ca 9 mm long mesial root impression in protoconid formation times at between 1.98 and the alveolar bone for the right M3. Sts 52 (attributed 2.38 years and metaconid time to be near identical to Au. africanus), OH5 and KNM-WT 17400 both (1.92–2.37 years) but Lacruz et al. (2006) reported attributed to P. boisei, are specimens closer to dental protocone formation times in two Au. africanus maturity but there is no histological evidence at all molars to be greater than this (M1, 2.74 years and to estimate their chronological age, only a hint from M2, 3.0–3.2 years). These latter two crown for- periapical radiographs of the upper canine of OH5 mation times are very close to mean modern that this root apex may have been recently completed human values. In general, molar crown formation by age at death (Skinner & Sperber 1982). times in early hominins are less than those in modern humans and more similar to those of Pan but there is considerable overlap in the ranges and (d) Crown formation times still insufficient data to compare sample mean Perikymata counts on hominin incisors and canines, values statistically. especially those attributed to Paranthropus, all point to anterior crown formation times having been shorter than those known for modern humans and for Pan (e) Summary points about dental development (Dean et al. 1993, 2001; Dean & Reid 2001). One in early hominins lower canine tooth attributed to Au. africanus The cumulative rates of enamel formation follow a (Sts 50) has 170 perikymata suggesting a crown for- similar trajectory in both Pan and early hominins (irre- mation time of around four years or more (Dean & spective of enamel thickness and crown formation Reid 2001) but in general the anterior teeth with the times) that is faster than that in modern humans greatest crown formation times appear to be those of (Dean et al. 2001; Lacruz et al. 2008). Estimates for Au. anamensis and Au. afarensis. Here, canine enamel gingival emergence times for M1 in several early homi- formation times come closest to those known for nin specimens all fall within the range expected for modern humans. Suwa et al. (2009a) counted 193 Pan, and in fact are all earlier than the time proposed perikymata on the upper canine of ARA-VP-6/1 (the for free-born, free-living chimpanzees. There is, how- holotype of Ardipithecus ramidus and a probable ever, no direct evidence at all for ages of M2 and male). This suggests that canine crown formation M3 eruption among the earliest hominins. The evi- took between 4.3 and 4.8 years in this specimen dence for molar initiation times provides only one (Suwa et al. 2009a) and so was potentially within the example (Dikika: Dik-1-1) where there is clear early range recorded for female Gorilla and Pongo, but M2 initiation with respect to M1 and there is no evi- below the range so far recorded for female Pan canines dence at all for M3 initiation occurring prior to (Schwartz & Dean 2001). The several clear regularly completion of M2 enamel formation in any early spaced hypoplastic bands illustrated on this specimen hominin specimen. Total molar tooth formation in Suwa et al.(2009b) are reminiscent of what are times have only been estimated in three hominin speci- likely to be seasonally related cycles of poor growth on mens, and appear to fall closest to the earlier ages living great ape canines (Skinner & Hopwood 2003). known for Pan. In contrast, those of incisors appear If there were eight or nine such bands on ARA-VP-6/ similar to those observed in Pan. Anterior crown for- 1 and on other Ar. ramidus canines, this would strongly mation times are almost always consistently less than suggest that Ar. ramidus existed in a seasonal environ- those known for Pan with the shortest crown formation ment with two colder wetter seasons per year. times occurring in Paranthropus. Enamel (crown) At least seven juvenile specimens attributed to formation times in molars are generally within the P. boisei (KNM-ER 1477, KNM-ER 812, KNM-ER ranges known for Pan molars (but occasionally also 1820, OH 30) or P. robustus (KB 5223, SK 64, fall well within modern human ranges).

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3406 M. C. Dean Review. Dental development in early hominins

6. DISCUSSION A third observation about the chronological model Constructing a chronology for dental development in of dental development in Pan compared with that in P.troglodytes as a comparative model for early hominins early hominins is that anterior tooth growth does not is useful for a number of reasons. First, it highlights the appear to reflect general somatic growth. While total processes whereby dental development is likely to have anterior tooth formation times appear to be little kept pace with prolongation of the period of general different, anterior crown formation times in australo- growth during hominid evolution. These seem to be piths are very variable but always shorter than in Pan confined to the sequence of molar development and (Dean & Reid 2001; Dean et al. 2001). In this respect, to have involved shifts in the timing of initial mineral- the comparative chronological model for anterior ization, slightly faster crown formation rates and tooth crown formation times in Pan differs completely particularly, earlier times of tooth emergence into from that reconstructed for australopiths. Crown for- functional occlusion. The cumulative effects of each mation time does not relate in any simple way to of these are most fully expressed in molar emergence crown height (Dean 2009) within a tooth type. For times, which appear to be the clearest measure of com- example, there is nothing to distinguish the enamel parative development than any one of the components formation times of smaller P. robustus canines from that contribute to it. Estimates of M1 emergence times taller canines of H. erectus (Dean et al. 1993, 2001). in fossil hominins, as well as observations of early The fact that both enamel thickness and anterior molar initiation, and in some cases shorter crown for- tooth crown height, characters that can be broadly mation times, resemble Pan more closely than modern linked to dietary specialization, are not tightly linked humans. However, too few specimens exist to provide to the time taken to form crowns is interesting. If in clear evidence for early initiation or earlier gingival fact total anterior tooth formation times are relatively emergence times of M2 or M3 among australopith more stable than anterior tooth morphology appears specimens, although the evidence for this is a little to be, then perhaps crown formation times might be better in early Homo (Dean et al. 2001; Dean & a better candidate for exploring phylogenetic related- Smith 2009). The model reveals, however, that the ness among closely related species of early hominins key indicators of a Pan-like dental maturation pattern than tooth morphology. An interesting case in point would include early M3 initiation with respect to M2 worth further consideration is the short time taken to crown formation time and a lesser proportion of root form the reduced canine crown heights of Ar. ramidus formed at gingival emergence in all molar tooth (Suwa et al. 2009a). types than in modern humans. All observations made so far on fossil and living A second point to emerge from the model for Pan is apes and on early hominins indicate that M1 eruption that some things appear to be little different between times would have fallen within the simulated ranges for Pan and Homo and, it follows, might not be expected free-living Pan shown in figure 3 and none appear to to differ in early hominins. Total anterior tooth for- fall within the ranges known for modern humans. mation times, and maybe also those for molars, fall Interestingly, all predictions so far for M1 emergence within the same range, all be it a broad range. Few in fossil apes (Kelley 1997, 2002; Kelley & Smith radiographic studies of molar development in Pan 2003; Dean 2006) actually fall below the simulated have included older animals and few of the individual median age of attainment for M1 emergence predicted plots in figure 2 extend all the way to root apex closure in figure 3 as indeed do most estimates for early homi- and moreover, it is the distobuccal root (not the mesio- nins. This raises questions about how different great buccal root shown in figure 2) in both Pan and Gorilla ape dental development in the Late Miocene might that on radiographs completes formation last (Dean & have been to that known today for modern P.troglodytes Wood 2003). It is highly likely, therefore, that future and how good a model modern Pan is for comparisons studies will show total molar formation times to be with the earliest hominins. It also highlights the need equal in Pan and Homo. In this respect, the evidence to reconstruct a chronology for dental development for at least three individual australopith specimens in Gorilla to place that for Pan in a better modern com- suggests that total M1 and M2 formation time may parative perspective. It remains a real possibility that have been at the low end of the range reconstructed the chronological dental development in the earliest for Pan. The plot of M2 (figure 2) attributed to Au. hominins was more similar to that in modern Gorilla anamensis (KNM-ER 30748) has a crown formation than to modern Pan. Were this the case it would time at the upper limit of the M2 range for Pan raise very interesting issues about early hominin life- (2.66 years) but a total tooth formation time of only history strategies of the kind discussed by Kelley & 5.5 years (but with a little root still to form) and a Schwartz (2009). The issue of advanced dental matur- faster rate of root formation generally than in Pan ity in captive hand-reared great apes suggests that even that might prove to be more typical of Gorilla. Shorter M1 emergence times of approximately 4.5 years pre- anterior crown formation times in many australopiths dicted for H. erectus (Dean et al. 2001; Dean & and earlier times for root completion might also turn Smith 2009) would still fall comfortably within the out to fit a Gorilla model better than a Pan model. simulated range for wild-born chimpanzees (figure 3) This mosaic of great ape-like dental development as has been suggested by Zihlman et al. (2004) but among australopiths is perhaps what one ought to predictions for M2 and M3 eruption of approximately expect given the gorilla-like anatomy of the scapula 8 and approximately 14 years, respectively, in of Dikika, Dik-1-1 (Alemseged et al. 2006) and the H. erectus would not. gorilla-like mandibular morphology of Au. afarensis No convincing evidence exists for any differences in mandibles (Rak et al. 2007). the chronology of molar development and emergence

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Review. Dental development in early hominins M. C. Dean 3407 between early hominin taxa, but estimates of chrono- In Aspects of dental biology; palaeontology, anthropology and logical age in specimens around 2.5 years and evolution (ed. J. Moggi-Cecchi), pp. 201–215. Florence, younger make it clear that tooth wear was excessive Italy: International Institute for the Study of Man. in some infant and juvenile late australopiths. Even Anemone, R. L. & Watts, E. S. 1992 Dental development in thicker deciduous dental enamel was insufficient to apes and humans: a comment on Simpson, Lovejoy and Meindl (1990). J. Hum. Evol. 22, 149–153. (doi:10. compensate for this, resulting in extensive islands of 1016/0047-2484(92)90035-8) dentine exposure on deciduous teeth very early in Anemone, R. L., Watts, E. S. & Swindler, D. R. 1991 Dental development (Aiello et al. 1991). Moggi-Cecchi et al. development of known age chimpanzees Pan troglodytes (2010) describe an infant P. robustus hemi-mandible (Primates, Pongidae). Am. J. Phys. Anthropol. 86, from Drimolen (DNH-44) with an unworn erupting 229–241. (doi:10.1002/ajpa.1330860211) Rdm2 where islands of dentine are exposed on the Antoine, D., Hillson, S. & Dean, M. C. 2009 The develop- Rdc and on four out of five cusps of the Rdm1, argu- mental clock of dental enamel: a test for the periodicity of ably within a year or so of birth. The obvious prism cross-striations and an evaluation of the likely sources inference that some early hominin juveniles were of error in histological studies of this kind. J. Anat. 214, taking considerable quantities of supplementary 45–55. (doi:10.1111/j.1469-7580.2008.01010.x) foods at a very early age cannot, at the moment, be Bennejeant, C. 1940 La chronologie de la dentition chez les anthropoids. Mammalia 4, 41–45. (doi:10.1515/mamm. extended to assuming they were also weaned early 1940.4.2.41) and that interbirth intervals were relatively short in Beynon, A. D. & Wood, B. A. 1987 Patterns and rates of these later australopiths, although this is one interpret- molar crown formation times in East African fossil ation of those observations (Aiello et al. 1991; Dean hominids. Nature 326, 493–496. (doi:10.1038/326493a0) 2006). Once again, there is the tantalizing suggestion Beynon, A. D., Dean, M. C. & Reid, D. J. 1991 Histological that a Gorilla-like life-history model may be a better study on the chronology of the developing dentition match for some, but not all, early hominins. Many in gorilla and orangutan. Am. J. Phys. Anthropol. 86, life-history variables in Gorilla such as age at weaning 189–203. (doi:10.1002/ajpa.1330860208) (reviewed in Aiello et al. 1991) age at first reproduction Beynon, A. D., Clayton, C. B., Ramirez Rozzi, F. V. & Reid, and interbirth interval (Watts 1991; Robson & Wood D. J. 1998 Radiographic and histological methodologies in estimating the chronology of crown development in 2008; Kelley & Schwartz 2009) are reported to be modern humans and great apes; a review, with some earlier than in Pan and Pongo (Wich et al. 2004). applications for studies on juvenile hominids. J. Hum. However, a firm link with these variables and earlier Evol. 35, 351–370. (doi:10.1006/jhev.1998.0234) dental development remains illusive (Kelley & Boyde, A. 1963 Estimation of age at death of young human Schwartz 2009; Humphrey 2010). In the future, skeletal remains from incremental lines in dental enamel. combined studies of tooth microstructure that put a Third Int. Meeting in Forensic Immunology, Medicine, chronological time scale to more sophisticated Pathology and Toxicology, April 16th–24th, Excerpta models of changing infant diets may shed more light Medica (Int. Congress Series no. 80) Plenary Session on early life-history events such as these during the IIA, London, pp. 36–46. first four million years of human evolution (Humphrey Boyde, A. 1964 The structure and development of et al. 2008; Humphrey 2010). mammalian enamel. PhD Thesis, University of London, London, UK. I thank Alan Walker and Chris Stringer for inviting me to Boyde, A. 1976 Amelogenesis and the structure of enamel. contribute to this discussion meeting. Much of the In Scientific foundations of dentistry (eds B. Cohen & research underpinning this paper has been supported by I. R. H. Kramer), pp. 335–352. London, UK: the Leverhulme Trust and the Royal Society through W. Heinemann Medical Books Ltd. grants to me. I thank Don Reid and Gary Schwartz for Boyde, A. 1979 Carbonate concentration, crystal centres, helping with the construction of figure 4 and I am core dissolution, caries, cross striation, circadian rhythms especially grateful to Louise Humphrey, Jay Kelley, Helen and compositional contrast in the SEM. J. Dent. Res. 58b, Liversidge and Holly Smith for discussions that have 981–983. helped develop some of the ideas set out here. Boyde, A. 1989 Enamel. In Handbook of microscopic anatomy. Vol. 2: Teeth (eds A. Oksche & L. Vollrath), pp. 309–473. Berlin, Germany: Springer. ENDNOTE Boyde, A. 1990a Developmental interpretations of dental 1For consistency and clarity, ages given in the literature are cited here microstructure. In Primate life history and evolution: mono- as follows: Prenatal and postnatal ages up to 1 month are given in graphs in primatology, vol. 14 (ed. C. Jean De Rousseau), days, those between 1 month and 1 year in months and those greater pp. 229–267. New York, NY: Wiley-Liss Inc. than this in years. Boyde, A. 1990b Confocal optical microscopy. In Modern microscopies: techniques and applications (eds P. J. Duke & A. G. Michette), pp. 185–204. New York, NY: Plenum REFERENCES Press. Aiello, L. C., Montgomery, C. & Dean, C. 1991 The natural Bradley, B. J. 2008 Reconstructing phylogenies and pheno- history of deciduous tooth attrition in hominoids. J. Hum. types: a molecular view of human evolution. J. Anat. Evol. 21,397–412.(doi:10.1016/0047-2484(91)90114-B) 212, 337–353. (doi:10.1111/j.1469-7580.2007.00840.x) Alemseged, Z., Spoor, F., Kimbel, W. H., Bobe, R., Bromage, T. G. 1991 Enamel incremental periodicity in the Geraads, D., Reed, D. & Wynn, J. 2006 Juvenile early pig-tailed macaque; a polychrome fluorescent labelling hominin skeleton from Dikika, Ethiopia. Nature 443, study of dental hard tissues. Am. J. Phys. Anthropol. 86, 296–301. (doi:10.1038/nature05047) 205–214. (doi:10.1002/ajpa.1330860209) Anemone, R. L. 1995 Dental development in chimpanzees Bromage, T. G. & Dean, M. C. 1985 Re-evaluation of the of known chronological age; implications for understand- age at death of immature fossil hominids. Nature 317, ing the age at death of Plio-Pleistocene hominids. 525–527. (doi:10.1038/317525a0)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3408 M. C. Dean Review. Dental development in early hominins

Bromage, T. G. et al. 2009 Lamellar bone is an incremental observations on known-age captive orangutans. Am, tissue reconciling enamel rhythms, body size and J. Primatol. 5, 285–301. (doi:10.1002/ajp.1350050402) organismal life history. Cal. Tissue Int. 84, 388–404. Goodman, M., Bailey, W. J., Hayasaka, K., Stanhope, M. J., (doi:10.1007/s00223-009-9221-2) Slightom, J. & Czelusniak, J. 1994 Molecular evidence on Clements, E. M. B. & Zuckerman, S. 1953 The order of primate phylogeny from DNA sequences. Am. J. Phys eruption of the permanent teeth in the Hominoidea. Anthropol. 94, 3–24. (doi:10.1002/ajpa.1330940103) Am. J. Phys. Anthropol. 11, 313–337. (doi:10.1002/ajpa. Hess, A. F., Lewis, J. M. & Roman, B. 1932 A radiographic 1330110309) study of calcification of the teeth from birth to adoles- Conroy, G. C. & Vannier, M. W. 1991a Dental development cence. Dent. Cosmos 74, 1053–1061. in South African australopithecines. Part I: problems of Humphrey, L. T. 2010 Weaning behaviour in human pattern and chronology. Am. J. Phys. Anthropol. 86, evolution. Semin. Cell Dev. Biol. 21, 453–461. 121–136. (doi:10.1002/ajpa.1330860204) Humphrey, L. T., Dean, M. C., Jeffries, T. E. & Penn, M. Conroy, G. C. & Vannier, M. W. 1991b Dental development 2008 Unlocking evidence of early diet from tooth in South African australopithecines. Part II: dental stage enamel. Proc. Natl Acad. Sci. USA 105, 6834–6839. assessment. Am. J. Phys. Anthropol. 86, 137–156. (doi:10.1073/pnas.0711513105) (doi:10.1002/ajpa.1330860205) Kahumbu, P. & Ely, R. M. 1991 Teeth emergence in wild Dean, M. C. 1985 Variation in the developing root cone olive baboons in Kenya and formulation of a dental angle of the permanent mandibular teeth of modern schedule for ageing wild baboon populations. Am. J. man and certain fossil hominids. Am. J. Phys. Anthropol. Primatol. 23,1–9.(doi:10.1002/ajp.1350230102) 68, 233–238. (doi:10.1002/ajpa.1330680210) Keith, A. 1899 On the chimpanzees and their relationship to Dean, M. C. 1987 The dental developmental status for the gorilla. Proc. Zool. Soc. Lond. 67, 296–312. (doi:10. six East African juvenile hominids. J. Hum. Evol. 16, 1111/j.1469-7998.1899.tb06859.x) 197–213. (doi:10.1016/0047-2484(87)90076-5) Kelley, J. 1997 Paleobiological and phylogenetic significance Dean, M. C. 2000 Progress in understanding hominoid of life history in Miocene hominoids. In Function, dental development. J. Anat. 197, 77–101. (doi:10. phylogeny, and fossils: miocene hominoid evolution and adap- 1046/j.1469-7580.2000.19710077.x) tations (eds D. R. Begun, C. V. Ward & M. D. Rose), Dean, M. C. 2006 Tooth microstructure tracks the pace of pp. 173–208. New York, NY: Plenum Press. human life history evolution. Proc. R. Soc. B 273, Kelley, J. 2002 Life history evolution in Miocene and extant 2799–2808. (doi:10.1098/rspb.2006.3583) apes. In Human evolution through developmental change Dean, M. C. 2009 Growth in tooth height and extension (eds N. Minugh-Purvis & K. J. McNamara), pp. 223– rates in modern human and fossil hominin canines and 248. Baltimore, MD: Johns Hopkins University Press. molars. In Frontiers of oral biology; interdisciplinary dental Kelley, J. & Schwartz, G. T. 2009 Dental development and morphology (eds T. Koppe, G. Meyer & K. W. Alt), life history in living African and Asian apes. Proc. Natl pp. 68–73. Basel, Switzerland: Karger. Acad. Sci. USA 107, 1035–1040. (doi:10.1073/pnas. Dean, M. C. In press. Daily rates of dentine formation and 0906206107) root extension rates in Paranthropus boisei, KNM-ER Kelley, J. & Smith, T. M. 2003 Age at first molar emergence 1817, from Koobi Fora, Kenya. In African Genesis in early Miocene Afropithecus turkanensis and life-history Symp. Proc. (eds S. Reynolds & C. Menter). South evolution in the Hominoidea. J. Hum. Evol. 44, Africa: University of Witwatersrand Press. 307–329. (doi:10.1016/S0047-2484(03)00005-8) Dean, M. C. & Reid, D. J. 2001 Anterior tooth formation Kelley, J., Dean, M. C. & Ross, S. 2009 Root growth during times in Australopithecus and Paranthropus.InTwelfth molar eruption in extant great apes. In Frontiers of oral Int. Symp. on Dental Morphology (ed. A. Brooks), biology; interdisciplinary dental morphology (eds T. Koppe, pp. 135–149. Sheffied, UK: Sheffield Academic Press. G. Meyer & K. W. Alt), pp. 128–133. Basel, Switzerland: Dean, M. C. & Smith, B. H. 2009 Growth and development Karger. in the Nariokotome Youth, KNM-WT 15000. In The first Krogman, W. M. 1930 Studies in growth changes in the skull humans: origin of the genus Homo (eds F. E. Grine, J. C. and face of anthropoids. I. The eruption of teeth in Fleagle & R. E. Leakey), pp. 101–120. New York, NY: anthropoids and Old World apes. Am. J. Anat. 46, Springer. 303–313. (doi:10.1002/aja.1000460205) Dean, M. C. & Wood, B. A. 1981 Developing pongid Kronfeld, R. 1935 Development and calcification of the human dentition and its use for ageing individual crania in com- deciduous and permanent dentition. The Bur 35,18–25. parative cross-sectional growth studies. Folia Primatol. 36, Kuykendall, K. L. 1996 Dental development in chimpanzees 111–127. (doi:10.1159/000156011) (Pan troglodytes): the timing of tooth calcification stages. Dean, M. C. & Wood, B. A. 2003 A digital radiographic atlas Am. J. Phys. Anthropol. 99, 135–157. (doi:10.1002/ of the great ape skull and dentition. In Digital archives of (SICI)1096-8644(199601)99:1,135::AID-AJPA8.3.0. human paleobiology (eds L. Bondioli & R. Macchiarelli). CO;2-#) Milano, Italy: ADS Solutions/Consiglio Nazionale delle Kuykendall, K. L. 2001 On radiographic and histological Ricerche. methods for assessing dental development in chimpanzees: Dean, M. C., Beynon, A. D., Thackeray, J. F. & Macho, G. comments on Beynon et al. (1998) and Reid et al. (1998). A. 1993 Histological reconstruction of dental develop- J. Hum. Evol. 40,67–76.(doi:10.1006/jhev.2000.0445) ment and age at death of a juvenile Paranthropus robustus Kuykendall, K. L., Mahoney, C. J. & Conroy, G. C. 1992 specimen, SK 63, from Swartkrans, South Africa. Probit and survival analysis of tooth emergence ages Am. J. Phys. Anthropol. 91, 401–419. (doi:10.1002/ajpa. in a mixed-longitudinal sample of chimpanzees (Pan 1330910402) troglodytes). Am. J. Phys. Anthropol. 89, 379–399. Dean, M. C., Leakey, M. G., Reid, D. J., Schrenk, F., (doi:10.1002/ajpa.1330890310) Schwartz, G. T., Stringer, C. & Walker, A. 2001 Lacruz, R. S. 2006 Enamel microstructure of the hominid Growth processes in teeth distinguish modern humans KB 5223 from Kromdraai, South Africa. Am. J. Phys. from Homo erectus and earlier hominins. Nature 414, Anthropol. 132, 175–182. (doi:10.1002/ajpa.20506) 628–631. (doi:10.1038/414628a) Lacruz, R. S., Ramirez Rozzi, F. & Bromage, T. G. 2005 Fooden, J. & Izor, R. J. 1983 Growth curves, dental Dental enamel hypoplasia, age at death, and weaning in emergence norms, and supplementary morphological the Taung child. S. Afr. J. Sci. 101, 567–569.

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

Review. Dental development in early hominins M. C. Dean 3409

Lacruz, R. S., Ramirez Rozzi, F. & Bromage, T. G. 2006 Ramirez Rozzi, F. V. 1995 Time of crown formation in Plio- Variation in enamel development of South African fossil Pleistocene hominid teeth. In Aspects of dental hominids. J. Hum. Evol. 51, 580–590. (doi:10.1016/j. biology; palaeontology, anthropology and evolution. (ed. J. jhevol.2006.05.007) Moggi-Cecchi), pp. 217–238. Florence, Italy: Inter- Lacruz, R. S., Dean, M. C., Ramirez Rozzi, F. & Bromage, national Institute for the Study of Man. T. G. 2008 Megadontia, striae periodicity and patterns Reid, D. J. & Dean, M. C. 2005 Variation in modern human of enamel secretion in Plio-Pleistocene fossil hominins. enamel formation times. J. Hum. Evol. 50, 329–346. J. Anat. 213, 148–158. (doi:10.1111/j.1469-7580.2008. (doi:10.1016/j.jhevol.2005.09.003) 00938.x) Reid, D. J., Schwartz, G. T., Dean, M. C. & Chandrasekera, Lacruz, R. S. & Ramirez Rozzi, F. V. 2010 Molar crown M. S. 1998 A histological reconstruction of dental devel- development in Australopithecus afarensis. J. Hum. Evol. opment in the common chimpanzee, Pan troglodytes. 58, 201–206. (doi:10.1016/j.jhevol.2009.11.007) J. Hum. Evol 35, 427–448. (doi:10.1006/jhev.1998.0248) Lilly, A. A., Mehlmann, P. T. & Doran, D. 2002 Intestinal Reid, D. J., Hillson, S. & Dean, M. C. 2000 Defining parasites in gorillas, chimpanzees and humans at chronological growth standards for known fractions of Mondika Research Site, Dzanga-Ndoki National Park, tooth crown height in primate anterior teeth. Central African Republic. Int. J. Primatol. 23, 555–573. Am. J. Phys. Anthropol. Suppl. 30, 260. (doi:10.1023/A:1014969617036) Retzius, A. 1837 Bemerkungen u¨ber den inneren Bau Lippert, W. 1977 Erfahrungen bei der Aufzucht von Orang- der Za¨hne, mit besonderer Ru¨cksicht auf dem in Utans (Pongo pygmaeus) im Tierpark Berlin. Der Zahnknochen vorkommenden Ro¨hrenbau. (Mu¨llers). Zoologische Garten (N.F.) 47, 209–225. Arch. Anat. Physiol. 1837, 486–566. Liversidge, H. M. 2009 Permanent tooth formation as a Risnes, S. 1986 Enamel apposition rate and the prism method of estimating age. In Frontiers of oral biology; inter- periodicity in human teeth. Scand. J. Dent. Res. 94, disciplinary dental morphology (eds T. Koppe, G. Meyer & 394–404. K. W. Alt), pp. 153–157. Basel, Switzerland: Karger. Robson, S. L. & Wood, B. A. 2008 Hominin life history: Mahoney, P. 2008 Intraspecific variation in M1 enamel reconstruction and evolution. J. Anat. 212, 394–425. development in modern humans: implications for (doi:10.1111/j.1469-7580.2008.00867.x) human evolution. J. Hum. Evol. 55, 130–146. Ruvolo, M. 1994 Molecular evolutionary processes and con- Marzke, M. W., Young, D. L., Hawkey, D. E., Su, S. M., flicting gene trees; the hominoid case. Am. J. Phys. Fritz, J. & Alford, P. L. 1996 Comparative analysis Anthropol. 94, 89–113. (doi:10.1002/ajpa.1330940108) of weight gain, hand/wrist maturation, and dental Schultz, A. H. 1924 Growth studies on primates bearing emergence rates in chimpanzees aged 0–24 months upon man’s evolution. Am. J. Phys. Anthropol. 7, from varying captive environments. Am. J. Phys. 149–164. (doi:10.1002/ajpa.1330070218) Anthropol. 99, 175–190. (doi:10.1002/(SICI)1096- Schultz, A. H. 1935 Eruption and decay of the permanent 8644(199601)99:1,175::AID-AJPA10.3.0.CO;2-K) teeth of primates. Am. J. Phys. Anthropol. 19, 480–581. Moggi-Cecchi,J.,Tobias,P.V.&Beynon,A.D.1998The Schultz, A. H. 1937 Fetal growth and development of mixed dentition and associated skull fragments of a juvenile the rhesus monkey. Contrib. Embryol. Carneg. Inst. 26, fossil hominid from Sterkfontein, South Africa. Am J. 71–97. Phys. Anthropol. 106, 425–465. (doi:10.1002/(SICI)1096- Schultz, A. H. 1940 Growth and development of the 8644(199808)106:4,425::AID-AJPA2.3.0.CO;2-I) chimpanzee. Contrib. Embryol. Carneg. Inst. 28, 1–63. Moggi-Cecchi, J., Mentor, C., Boccone, S. & Keyser, A. Schwartz, G. T. & Dean, M. C. 2001 Ontogeny of canine 2010 Early hominin dental remains from the Plio- dimorphism in extant hominoids. Am. J. Phys. Anthropol. Pleistocene site of Drimolen, South Africa. J. Hum. 115, 269–283. (doi:10.1002/ajpa.1081) Evol. 58, 374–405. (doi:10.1016/j.jhevol.2010.01.006) Schwartz, G. T., Reid, D. J. & Dean, M. C. 2001 Develop- Moxham, B. J. & Berkovitz, B. K. B. 1974 The circumnatal mental aspects of sexual dimorphism in hominoid dentitions of a gorilla (Gorilla gorilla) and a chimpanzee canines. Int. J. Primatol. 22, 837–860. (doi:10.1023/ (Pan troglodytes). J. Zool. Lond. 173, 271–276. (doi:10. A:1012073601808) 1111/j.1469-7998.1974.tb03137.x) Schwartz, G. T., Dean, M. C., Reid, D. J. & Zihlman, A. L. Nissen, H. W. & Riesen, A. H. 1945 The deciduous denti- 2006 A faithful record of stressful life events preserved in tion of the chimpanzee. Growth 9, 265–274. the dental developmental record of a juvenile gorilla. Nissen, H. W. & Riesen, A. H. 1964 The eruption of the per- Int. J. Primatol. 27, 1–19. manent dentition of chimpanzees. Am. J. Phys. Anthropol. Shellis, R. P. 1984 Variations in growth of the enamel crown 22, 285–294. (doi:10.1002/ajpa.1330220315) in human teeth and a possible relationship between Oka, S. W. & Kraus, B. S. 1969 The circumnatal status growth and enamel structure. Archs. Oral Biol. 29, 697– of molar crown maturation among the hominoidea. 705. (doi:10.1016/0003-9969(84)90175-4) Archs. Oral Biol. 14, 639–659. (doi:10.1016/0003- Shinoda, H. 1984 Faithful records of biological rhythms 9969(69)90187-3) in dental hard tissues. Chem. Today 162, 34–40. Phillips-Conroy, J. & Jolly, C. 1988 Dental eruption [In Japanese.] schedules of wild and captive baboons. Am. J. Primatol. Skinner, M. F. & Hopwood, D. 2003 Hypothesis for the 15, 17–29. (doi:10.1002/ajp.1350150104) causes and periodicity of repetitive linear enamel Preiswerk, G. 1895 Beitra¨ge zur Kenntniss de Schmelzstrus- hypoplasia in large, wild African (Pan troglodytes and tur bei Saügethieren mit besonderer beru¨cksichtigung Gorilla gorilla) and Asian (Pongo pygmaeus) apes. Am. J. der Ungulaten. Basel, Switzerland: Akadenische Phys. Anthropol. 123, 216–235. (doi:10.1002/ajpa.10314) Buchhandlung. C. F. Lendorff. Skinner, M. F. & Sperber, G. H. 1982 Atlas of radiographs of Rak, Y., Ginzburg, A. & Geffen, E. 2007 Gorilla-like anat- early man. New York, NY: A. R. Liss Inc. omy on Australopithecus afarensis mandibles suggests Au. Smith, B. H. 1989 Dental development as a measure of life afarensis link to robust australopiths. Proc. Natl Acad. history in primates. Evolution 43, 683–688. (doi:10.2307/ Sci. 104, 6568–6572. (doi:10.1073/pnas.0606454104) 2409073) Ramirez Rozzi, F. V. 1993 Tooth development in East Smith, T. M. 2006 Experimental determination of the African Paranthropus. J. Hum. Evol. 24, 429–454. periodicity of incremental features in enamel. J. Anat. (doi:10.1006/jhev.1993.1030) 208, 99–113. (doi:10.1111/j.1469-7580.2006.00499.x)

Phil. Trans. R. Soc. B (2010) Downloaded from rstb.royalsocietypublishing.org on October 24, 2010

3410 M. C. Dean Review. Dental development in early hominins

Smith, T. M., Reid, D. J., Dean, M. C., Olejniczak, A. J. & White, T. D. 1977 New fossil homininds from Laetolil, Martin, L. B. 2006 Molar development in common Tanzania. Am. J. Phys. Anthropol. 48, 197–229. chimpanzees (Pan troglodytes). J. Hum. Evol. 52, 201– Wich, S. A., Utami-Atmoko, S. S., Mitra Setia, T., Rijkesen, 206. (doi:10.1016/j.jhevol.2006.09.004) H. D., Schurmann, C., van Hooff, J. A. R. A. M. & Smith, T. M., Smith, B. H. & Boesch, C. 2009 Dental van Schaik, C. P. 2004 Life history of wild Sumatran development in the Taı¨ forest chimpanzees reappraised. orangutans (Pongo abelii). J. Hum. Evol. 47, 385–398. Am. J. Phys. Anthropol. 138, 243. (doi:10.1016/j.jhevol.2004.08.006) Smith, T. M., Smith, B. H., Reid, D. J., Siedel, H., Vigilant, Winkler, L. A. 1996 A comparison of radiographic and L., Hublin, J.-J. & Boesch, C. 2010 Dental development anatomical evidence of tooth development in infant of the Taı¨Forest chimpanzees revisited. J. Hum. Evol. 58, apes. Folia Primatol. 65, 1–13. (doi:10.1159/000156864) 363–373. (doi:10.1016/j.jhevol.2010.02.008) Winkler, L. A., Schwartz, J. H. & Swindler, D. R. 1991 Suwa, G., Kono, R. T., Simpson, S. W., Asfaw, B., Lovejoy, Aspects of dental development in the orangutan O. C. & White, T. D. 2009a Paleobiological implications prior to eruption of the permanent dentition. of the Ardipithecus ramidus dentition. Science 326, 94–99. Am. J. Phys. Anthropol. 86, 255–271. (doi:10.1002/ajpa. Suwa, G., Kono, R. T., Simpson, S. W., Asfaw, B., Lovejoy, 1330860213) O. C. & White, T. D. 2009b Author’s summaries. Winkler, L. A., Schwartz, J. H. & Swindler, D. R. 1996 Paleobiological implications of the Ardipithecus ramidus Development of the orang utan permanent dentition: dentition. Science 326,69.(doi:10.1126/science.1175824) Assessing patterns and variation in tooth development. Swindler, D. R. 1985 Nonhuman primate dental develop- Am. Phys. Anthropol. 99, 205–220. (doi:10.1002/ ment and its relation to human dental development. (SICI)1096-8644(199601)99:1,205::AID-AJPA12.3.0. In Nonhuman primate models for human growth and CO;2-R) development (ed. E. S. Watts), pp. 67–94. New York, Zihlman, A., Bolter, D. & Boesch, C. 2004 Wild chimpanzee NY: A. R. Liss. dentition and its implications for assessing life history in Tarrant, L. H. & Swindler, D. R. 1972 The state of immature hominin fossils. Proc. Natl Acad. Sci. USA the deciduous dentition of a chimpanzee fetus (Pan 101, 10541–10543. (doi:10.1073/pnas.0402635101) troglodytes). J. Dent. Res. 51, 677. Zihlman, A., Bolter, D. & Boesch, C. 2007 Skeletal and Ward, C. V., Leakey, M. G. & Walker, A. 2001 Morphology dental growth and development in chimpanzees of the of Australopithecus anamensis from Kanapoi and Allia Bay, Ta¨ ı National Park, Coˆte D’Ivoire. J. Zool. 273, 63–73. Kenya. J. Hum. Evol. 41, 255–368. (doi:10.1006/jhev. (doi:10.1111/j.1469-7998.2007.00301.x) 2001.0507) Zuckerman, S. 1928 Age changes in the chimpanzee, with Watts, D. P. 1991 Mountain gorilla reproduction and sexual special reference to growth of the brain, eruption of the behavior. Am. J. Primatol. 24, 211–226. (doi:10.1002/ teeth and estimation of age; with a note on the Taungs ajp.1350240307) ape. Proc. Zool. Soc. Lond. 96, 1–42.

Phil. Trans. R. Soc. B (2010) RSTB_365_1556_Cover.qxd 9/7/10 10:42 AM Page 2

volume 365

27 October 2010 number 1556 . .