Science - Ardipithecus ramidus

Science - Ardipithecus ramidus

1. FULL COLLECTION

2. Around the world, governments turn to AAAS, publisher of Science magazine, as an objective, multidiscipli-
nary scientific authority to
educate public officials and judicial figures on today’s most pressing issues. Our
goal is to promote
informed policy decisions that benefit society. And this is just one of the ways that AAAS
is committed to
advancing science to support a healthy and prosperous
world. Join us. Together we can make a difference. aaas.org/plusyou
Science funding
Climate regulation
Human rights

3. Contents
Introduction and Author List
5 Light on the Origin of Man
Editorial
9 Understanding Human Origins
Bruce Alberts
News Focus
10 A New Kind of Ancestor:
Ardipithecus Unveiled
14 Habitat for Humanity
15 The View From Afar
Authors’ S ummaries
18 Ardipithecus ramidus and the Paleobiology of
Early Hominids
Tim D. White et al.
19 The Geological, Isotopic, Botanical,
Invertebrate, and Lower Vertebrate
Surroundings of Ardipithecus ramidus
Giday WoldeGabriel et al.
20 Taphonomic, Avian, and Small-Vertebrate
Indicators of Ardipithecus ramidus Habitat
Antoine Louchart et al.
21 Macrovertebrate Paleontology and the Pliocene
Habitat of Ardipithecus ramidus
Tim D. White et al.
22 The Ardipithecus ramidus Skull and Its
Implications for Hominid Origins
Gen Suwa et al.
23 Paleobiological Implications of the
Ardipithecus ramidus Dentition
Gen Suwa et al.
24 Careful Climbing in the Miocene:
The Forelimbs of Ardipithecus ramidus and
Humans Are Primitive
C. Owen Lovejoy et al.
25 The Pelvis and Femur of Ardipithecus ramidus :
The Emergence of Upright Walking
C. Owen Lovejoy et al.
26 Combining Prehension and Propulsion: The
Foot of Ardipithecus ramidus
C. Owen Lovejoy et al.
27 The Great Divides: Ardipithecus ramidus
Reveals the Postcrania of Our Last Common
Ancestors with African Apes
C. Owen Lovejoy et al.
28 Reexamining Human Origins in Light
of Ardipithecus ramidus
C. Owen Lovejoy
Credit: copyright T. White, 2008
Research Articles
29 Ardipithecus ramidus and the Paleobiology
of Early Hominids
Tim D. White et al.
41 The Geological, Isotopic, Botanical,
Invertebrate, and Lower Vertebrate
Surroundings of Ardipithecus ramidus
Giday Wolde Gabriel et al.
46 Taphonomic, Avian, and Small-Vertebrate
Indicators of Ardipithecus ramidus Habitat
Antoine Louchart et al.
50 Macrovertebrate Paleontology and the Pliocene
Habitat of Ardipithecus ramidus
Tim D. White et al.
57 The Ardipithecus ramidus Skull and Its
Implications for Hominid Origins
Gen Suwa et al.
64 Paleobiological Implications of the
Ardipithecus ramidus Dentition
Gen Suwa et al.
70 Careful Climbing in the Miocene:
The Forelimbs of Ardipithecus ramidus and
Humans Are Primitive
C. Owen Lovejoy et al.
78 The Pelvis and Femur of Ardipithecus ramidus :
The Emergence of Upright Walking
C. Owen Lovejoy et al.
84 Combining Prehension and Propulsion: The
Foot of Ardipithecus ramidus
C. Owen Lovejoy et al.
92 The Great Divides: Ardipithecus ramidus
Reveals the Postcrania of Our Last Common
Ancestors with African Apes
C. Owen Lovejoy et al.
99 Reexamining Human Origins in Light
of Ardipithecus ramidus
C. Owen Lovejoy
See also related video, Science Podcast at
www.sciencemag.org/ardipithecus/
Ardipithecus ramidus

4. In America today, 1 in 3 individuals do not accept evolution.
1
That’s why AAAS continues to play an important
role in the effort to protect the integrity of science education. AAAS is hard at work ensuring that evolution
continues to be taught in science classrooms, but we need your help.
Join us. Together we can make a difference. aaas.org/plusyou/evolution
Evolution
1.
Pew Research Center for the People & the Press. May 2009, General Public Science Survey.
AAAS is here.

5. 5www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Charles Darwin’s seminal work On the Origin of Species,
published 150 years ago next month, contains just one understated
sentence on the implications of his theory for human evolution:
“Light will be thrown on the origin of man and his history.” As
Darwin implied in his introduction to The Descent of Man, he felt
that those implications were obvious; he appreciated, as events
quickly showed, that it would be only natural to look at evolution
foremost from our human perspective and contemplate what makes
us unique among other primates—our large brains and ability to
communicate, to create, and to understand and investigate our his-
tory and nature; our culture, society, and religion; the ability to run
fast on two legs and manipulate tools; and more innovations that
separate us from our primate relatives.
Tracing our evolution and how we came to acquire these skills
and traits, however, has been difficult. Genetic data now confirm
that our closest living primate relative is the chimpanzee. We shared
and evolved from a common ancestor some 6 million or more years
ago. But identifying our unique genes and other genetic differences
between us and our primate cousins does not reveal the nature of
that ancestor, nor what factors led to the genetic changes that un-
derlie our divergent evolutionary paths. That requires a fossil record
and enough parts of past species to assess key anatomical details.
It also requires knowing the habitat of early humans well, to deter-
mine their diet and evaluate what factors may have influenced their
evolution through time. Many early human fossils have been found,
but with a few exceptions, these are all less than 4 million years old.
The key first several million years of human evolution have been
poorly sampled or revealed.
This issue presents 11 papers authored by a diverse international
team (see following pages) describing an early hominid species,
Ardipithecus ramidus, and its environment. The hominid fossils
are 4.4 million years old, within this critical early part of human
evolution, and represent 36 or more individuals, including much of
the skull, pelvis, lower arms, and feet from one female. The papers
represent three broad themes. Five focus on different parts of the
anatomy that are revealing for human evolution. These show that
Ardipithecus was at home both moving along trees on its palms and
walking upright on the ground. Three characterize Ardipithecus’s
habitat in detail, through analysis of the hosting rocks and thousands
of fossils of small and large animals and plants. These show that
Ardipithecus lived and ate in woodlands, not grasslands. The first
paper presents an overview, and it and the last two papers trace early
human evolution and synthesize a new view of our last common an-
cestor with chimps. One conclusion is that chimps have specialized
greatly since then and thus are poor models for that ancestor and for
understanding human innovations such as our ability to walk.
These papers synthesize an enormous amount of data collected
and analyzed over decades by the authors. Because of the scope of
these papers and the special broad interest in the topic of human
evolution, we have expanded our usual format for papers and cover-
age. The papers include larger figures, tables, and discussions, and
the overview and two concluding papers provide extended introduc-
tions and analyses.
In addition, to aid understanding and introduce the main results
of each paper, the authors provide a one-page summary of each
paper, with an explanatory figure aimed at the general reader. Our
News Focus section, written by Ann Gibbons, provides further
analysis and coverage, and it includes maps and a portrait of
the meticulous and at times grueling field research behind the
discoveries. Available online are a video interview and a podcast
with further explanations.
To accommodate this material and allow the full papers, this print
issue presents an Editorial, News coverage, the authors’ summaries,
and four papers in full: the overview paper and one key paper from
each thematic group above. The other research papers, and of course
all content, are fully available online. In addition, a special online
page (www.sciencemag.org/Ardipithecus/) links to several print and
download packages of this material for AAAS members, research-
ers, educators, and other readers.
This collection, essentially an extra issue of Science in length,
reflects efforts by many behind the scenes. Every expert reviewer
evaluated, and improved, multiple papers, and several commented
on all 11 of them. The authors provided the summaries on top of
an already large writing and revision effort. Paula Kiberstis helped
in their editing. The figures and art were drafted and improved by
J. H. Matternes, Henry Gilbert, Kyle Brudvik, and Josh Carlson,
as well as Holly Bishop, Nathalie Cary, and Yael Kats at Science.
Numerous other Science copyediting, proofreading, and production
staff processed this content on top of their regular loads. Finally,
special thanks go to the people of Ethiopia for supporting and facili-
tating this and other research into human origins over many years,
and for curating Ardipithecus ramidus for future research and for all
of us to admire.
Ardipithecus ramidus thus helps us bridge the better-known,
more recent part of human evolution, which has a better fossil
record, with the scarcer early human fossils and older ape fossils
that precede our last common ancestor. Ardipithecus ramidus is a
reminder of Darwin’s conclusion of The Origin:
There is grandeur in this view of life, with its several powers, having
been originally breathed into a few forms or into one; and that,
whilst this planet has gone cycling on according to the fixed law of
gravity, from so simple a beginning endless forms most beautiful
and most wonderful have been, and are being, evolved.
– Brooks Hanson
Light on the Origin of Man
introduction

6. 2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org62
CREDITS: PHOTOS COURTESY OF THE AUTHORS
Tim D. White
Human Evolution Research
Center and Department
of Integrative Biology,
3101 Valley Life Sciences
Building, University
of California at Berkeley, Berkeley, CA
94720, USA.
Giday WoldeGabriel
Earth Environmental
Sciences Division,
Los Alamos National
Laboratory, Los Alamos,
NM 87545, USA.
Antoine Louchart
UMR 5125 PEPS CNRS,
France, Université Lyon 1,
69622 Villeurbanne Cedex,
France, and Institut de
Génomique Fonctionnelle
de Lyon, Université de Lyon, Université
Lyon 1, CNRS, INRA, Ecole Normale
Supérieure de Lyon, France.
Gen Suwa
The University Museum,
the University of Tokyo,
Hongo, Bunkyo-ku, Tokyo
113-0033, Japan.
C. Owen Lovejoy
Department of
Anthropology, School
of Biomedical Sciences,
Kent State University,
Kent, OH 44240–0001, USA.
Stanley H. Ambrose
Department of
Anthropology, University
of Illinois, Urbana, IL
61801, USA.
Berhane Asfaw
Rift Valley Research
Service, P.O. Box 5717,
Addis Ababa, Ethiopia.
Mesfin Asnake
Ministry of Mines and
Energy, P.O. Box 486,
Addis Ababa, Ethiopia.
Doris Barboni
CEREGE (UMR6635
CNRS/Université
Aix-Marseille), BP80,
F-13545 Aix-en-Provence
Cedex 4, France.
Raymond L. Bernor
National Science
Foundation,
GEO:EAR:SEPS
Sedimentary Geology
and Paleobiology Program,
Arlington, VA 22230, and College of Medi-
cine, Department of Anatomy, Laboratory
of Evolutionary Biology, Howard University,
520 W St., Washington, DC 20059, USA.
Yonas Beyene
Department of Anthropology
and Archaeology,
Authority for Research and
Conservation of the Cultural
Heritage, Ministry of Youth,
Sports and Culture, P.O. Box 6686, Addis
Ababa, Ethiopia.
Michael T. Black
Phoebe A. Hearst Museum
of Anthropology,
103 Kroeber Hall, no. 3712,
University of California
Berkeley, Berkeley, CA
94720–3712, USA.
Robert J. Blumenschine
Center for Human
Evolutionary Studies,
Department of
Anthropology, Rutgers
University, 131 George St.,
New Brunswick, NJ 08901–1414, USA.
Jean-Renaud Boisserie
Paléobiodiversité et
Paléoenvironnements, UMR
CNRS 5143, USM 0203,
Muséum National d’Histoire
Naturelle, 8 Rue Buffon, CP
38, 75231 Paris Cedex 05, France, and Institut
de Paléoprimatologie et Paléontologie
Humaine, Évolution et Paléoenvironnements,
UMR CNRS 6046, Université de Poitiers,
40 Avenue du Recteur-Pineau, 86022 Poitiers
Cedex, France.
Raymonde Bonnefille
CEREGE (UMR6635
CNRS/Université Aix-
Marseille), BP80, F-13545
Aix-en-Provence Cedex 4,
France.
Laurent Bremond
Center for Bio-Archaeology
and Ecology (UMR5059
CNRS/Université
Montpellier 2/EPHE),
Institut de Botanique,
F-34090 Montpellier, France.
Michel Brunet
Collège de France, Chaire
de Paléontologie Humaine,
3 Rue d’Ulm, F-75231
Paris Cedex 05, France.
Brian Currie
Department of Geology,
Miami University, Oxford,
OH 45056, USA.
David DeGusta
Department of
Anthropology, Stanford
University, Stanford, CA
94305–2034, USA.
Eric Delson
Department of
Anthropology, Lehman
College/CUNY, Bronx, NY
10468; NYCEP; and
Department of Vertebrate
Paleontology, American Museum of Natural
History; New York, NY 10024, USA.
Stephen Frost
Department of
Anthropology, University
of Oregon, Eugene, OR,
97403–1218, USA.
Nuria Garcia
Dept. Paleontología,
Universidad Complutense
de Madrid & Centro
de Evolución y
Comportamiento Humanos,
ISCIII, C/ Sinesio Delgado 4, Pabellón 14,
28029 Madrid, Spain.
Ioannis X. Giaourtsakis
Ludwig Maximilians
University of Munich,
Department of Geo- and
Environmental Sciences,
Section of Paleontology.
Richard-Wagner-Strasse 10, D-80333
Munich, Germany.
The Authors
Published by AAAS
6 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS

7. 63
CREDITS: PHOTOS COURTESY OF THE AUTHORS
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
Yohannes Haile-Selassie
Department of Physical
Anthropology, Cleveland
Museum of Natural History,
1 Wade Oval Drive,
Cleveland, OH 44106, USA.
William K. Hart
Department of Geology,
Miami University, Oxford,
OH 45056, USA.
Leslea J. Hlusko
Human Evolution Research
Center and Department of
Integrative Biology,
University of California at
Berkeley, 3010 Valley Life
Sciences Building, Berkeley, CA, 94720, USA.
F. Clark Howell
Human Evolution Research
Center and Department of
Anthropology, 3101 Valley
Life Sciences Building,
University of California at
Berkeley, Berkeley, CA 94720, USA
(deceased).
M. C. Jolly-Saad
Université Paris-Ouest La
Défense, Centre Henri Elhaï,
200 Avenue de la
République, 92001 Nanterre,
France.
Reiko T. Kono
Department of Anthropology,
National Museum of Nature
and Science, Hyakunincho,
Shinjuku-ku, Tokyo,
169-0073, Japan.
Daisuke Kubo
Department of Biological
Sciences, Graduate School
of Science, the University
of Tokyo, Tokyo, 113-0033,
Japan.
Bruce Latimer
Department of Anatomy,
Case Western Reserve
University School of
Medicine, Cleveland, OH
44106–4930, USA.
Thomas Lehmann
Senckenberg
Forschungsinstitut,
Senckenberganlage 25,
D-60325 Frankfurt am Main,
Germany.
Andossa Likius
Département de
Paléontologie, Université
de N’Djamena, BP 1117,
N’Djamena, Chad.
Jay H. Matternes
4328 Ashford Lane, Fairfax,
VA 22032, USA.
Alison M. Murray
Department of Biological
Sciences, University of
Alberta, Edmonton AB
T6G2E9, Canada.
Jackson K. Njau
Human Evolution Research
Center and Department of
Integrative Biology,
University of California at
Berkeley, 3010 Valley Life
Sciences Building, Berkeley, CA, 94720, USA.
Cesur Pehlevan
University of Yuzuncu
Yil, Department of
Anthropology, The Faculty
of Science and Letters, Zeve
Yerlesimi 65080 Van, Turkey.
Paul R. Renne
Berkeley Geochronology
Center, 2455 Ridge Road,
Berkeley, CA 94709, and
Department of Earth
and Planetary Science,
University of California at Berkeley, Berkeley,
CA 94720, USA.
Haruo Saegusa
Institute of Natural and
Environmental Sciences,
University of Hyogo,
Yayoigaoka,
Sanda 669-1546, Japan.
Gina Semprebon
Science and Mathematics,
Bay Path College,
588 Longmeadow St.,
Longmeadow, MA 01106,
USA.
Scott W. Simpson
Department of Anatomy,
Case Western Reserve
University School of
Medicine, Cleveland, OH
44106–4930, USA.
Linda Spurlock
Cleveland Museum of
Natural History, Cleveland,
OH 44106–4930, USA.
Kathlyn M. Stewart
Paleobiology, Canadian
Museum of Nature, Ottawa,
K1P 6P4, Canada.
Denise F. Su
Department of
Anthropology,
The Pennsylvania State
University, University Park,
PA 16802, USA.
Mark Teaford
Center for Functional
Anatomy and Evolution,
Johns Hopkins University
School of Medicine, 1830 E.
Monument St., Room 303,
Baltimore, MD 21205.
Elisabeth Vrba
Department of Geology and
Geophysics, Yale University,
New Haven, CT 06520,
USA.
Henry Wesselman
P.O. Box 369, Captain Cook,
Hawaii, 96704, USA.
SPECIALSECTION
Published by AAAS
7www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS

8. OurNewDarwinshirt—
yoursfreewhenyoubecome
aAAASmember.
Visitpromo.aaas.org/giftdarwintoordernow
Orderonlineatpromo.aaas.org/darwin24
Specialoffer:$99Professional$50Postdoc/Student*
BecomeaAAASmembertoday,atthisspecialintroductoryofferand
receiveafreeDarwinshirt(a$22.50valuefree).Yourmembershipgives
youallofthepersonalmembershipbenefitsofAAASincluding51issuesof
Sciencemagazine,careerdevelopmenttools,andmuchmore—allwhile
supportingourinternational,publicpolicy,andeducationalprograms—
theonesthatadvancescienceandservesociety.
*InternationalorderswillreceiveScienceDigitaledition-othermembershipcategoriesareavailableonline.
JoinToday!
darwin_fullpg_tradeshow3.indd1 5/20/099:16:56AM

9. Understanding Human Origins
RESPONDING TO A QUESTION ABOUT HIS SOON-TO-BE-PUBLISHED ON THE ORIGIN OF SPECIES,
Charles Darwin wrote in 1857 to Alfred Russel Wallace, “You ask whether I shall discuss 
‘man’; I think I shall avoid the whole subject, as so surrounded with prejudices, though 
I freely admit that it is the highest and most interesting problem for the naturalist.” Only 
some 14 years later, in The Descent of Man, did Darwin address this “highest problem” 
head-on: There, he presciently remarked in his introduction that “It has often and confi -
dently been asserted, that man’s origin can never be known: but . . . it is those who know 
little, and not those who know much, who so positively assert that this or that problem will 
never be solved by science.”
Darwin was certainly right. The intervening years provide conclusive evidence that it is 
very unwise to predict limits for what can be discovered through science. In fact, it now seems 
likely that, through synergistic advances in many disciplines, scientists will eventually deci-
pher a substantial portion of the detailed evolutionary history of our 
own species at both the morphological and molecular levels.
First, what can we expect from paleoanthropology? In this 200th 
anniversary year of Darwin’s birth, Science is pleased to publish the 
results of many years of scientifi c research that suggest an unex-
pected form for our last common ancestor with chimpanzees. This 
issue contains 11 Research Articles involving more than 40 authors, 
plus News articles that describe the life and times of Ardipithecus
ramidus, a hominid species that lived 4.4 million years ago in the 
Afar Rift region of northeastern Ethiopia. This region exposes a total 
depth of 300 meters of sediments that were deposited in rivers, lakes, 
and fl oodplains between about 5.5 and 3.8 million years ago. Even 
considering only this one site (there are many others), it is staggering 
to refl ect on the huge number of hominid remains that can in prin-
ciple be discovered, given suffi cient time and effort. Moreover, the 
history of science assures us that powerful new techniques will be developed in the coming 
years to accelerate such research, as they have been in the past. We can thus be certain that 
scientists will eventually obtain a rather detailed record showing how the anatomy of the 
human body evolved over many millions of years.
 What can we expect from a combination of genetics,  genomics,  biochemistry,  and compara-
tive organismal biology? We will want to interpret the history of the morphological transfor-
mations in the humanoid skeleton and musculature in terms of the molecular changes in the 
DNA that caused them. Genes and their regulatory regions control the morphology of animals 
through very complex biochemical processes that affect cell behavior during embryonic devel-
opment. Nevertheless,  experimental studies of model organisms such as fruit fl ies,  worms,  fi sh, 
 and mice are advancing our understanding of the molecular mechanisms involved. New inex-
pensive methods for deciphering the complete genome sequence of any organism will soon 
accelerate this process,  allowing scientists to analyze the recurring evolutionary morphological 
transformations that have been identifi ed by organismal biologists, * so as to determine the spe-
cifi c DNA changes involved. And the DNA sequences that have changed most rapidly during 
recent human evolution are being cataloged,  providing a new tool for fi nding important molecu-
lar differences that distinguish us from chimpanzees.†
The majesty of the discoveries already made represents a major triumph of the human 
intellect. And, as emphasized here, there will be many more discoveries to come. Darwin’s 
summary of his own efforts to understand human evolution is thus still relevant today: 
“Man may be excused for feeling some pride at having risen, though not through his own 
exertions, to the very summit of the organic scale; and the fact of his having thus risen, 
instead of having been aboriginally placed there, may give him hope for a still higher destiny 
in the distant future.” 
*R. L. Mueller et al., Proc. Natl. Acad. Sci. U.S.A. 101, 3820 (2004). †S. Prabhakar et al., Science 314, 786 (2006).
10.1126/science.1182387
– Bruce Alberts
17
EDITORIAL
CREDITS: (TOP) TOM KOCHEL; (RIGHT) ISTOCKPHOTO.COM
Bruce Alberts is Editor-
in-Chief of Science.
www.sciencemag.org SCIENCE VOL 326 2 OCTOBER 2009
Published by AAAS
9www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS

10. 10 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Every day, scientists add new pages to the story of human evolution by decipher-
ing clues to our past in everything from the DNA in our genes to the bones and
artifacts of thousands of our ancestors. But perhaps once each generation, a
spectacular fossil reveals a whole chapter of our prehistory all at once. In 1974,
it was the famous 3.2-million-year-old skeleton “Lucy,” who proved in one
stroke that our ancestors walked upright before they evolved big brains.
Ever since Lucy’s discovery, researchers have wondered what came before
her. Did the earliest members of the human family walk upright like Lucy or on
their knuckles like chimpanzees and gorillas? Did they swing through the trees
or venture into open grasslands? Researchers have had only partial, fleeting
glimpses of Lucy’s own ancestors—the earliest hominins, members of the group
that includes humans and our ancestors (and are sometimes called hominids).
Now, in a special section beginning on page 60 and online, a multidisciplinary
international team presents the oldest known skeleton of a potential human
ancestor, 4.4-million-year-old Ardipithecus ramidusfrom Aramis, Ethiopia.
This remarkably rare skeleton is not the oldest putative hominin, but it is
by far the most complete of the earliest specimens. It includes most of the
skull and teeth, as well as the pelvis, hands, and feet—parts that the
authors say reveal an “intermediate” form of upright walking, consid-
NEWSFOCUS
A New Kind of Ancestor:
ArdipithecusUnveiled
The oldest known hominin skeleton reveals the body plan of our very
early ancestors and the upright origins of humankind
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org36
From the inside out.Artist’s reconstructions show how Ardi’s
skeleton, muscles, and body looked and how she would have
moved on top of branches.
CREDITS: ILLUSTRATIONS © 2009, J. H. MATTERNES
Published by AAAS

11. 11www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
ered a hallmark of hominins. “We thought Lucy was the find of the
century but, in retrospect, it isn’t,” says paleoanthropologist Andrew
Hill of Yale University. “It’s worth the wait.”
To some researchers’ surprise, the female skeleton doesn’t look
much like a chimpanzee, gorilla, or any of our closest living primate
relatives. Even though this species probably lived soon after the dawn
of humankind, it was not transitional between African apes and
humans. “We have seen the ancestor, and it is not a chimpanzee,” says
paleoanthropologist Tim White of the University of California, Berke-
ley, co-director of the Middle Awash research group, which discovered
and analyzed the fossils.
Instead, the skeleton and pieces of at least 35 additional individuals
of Ar. ramidusreveal a new type of early hominin that was neither
chimpanzee nor human. Although the team suspects that Ar. ramidus
may have given rise to Lucy’s genus, Australopithecus, the fossils
“show for the first time that there is some new evolutionary grade of
hominid that is not Australopithecus, that is not Homo,” says paleontol-
ogist Michel Brunet of the College de France in Paris.
In 11 papers published in this issue and online, the team of 47
researchers describes how Ar. ramiduslooked and moved. The skele-
ton, nicknamed “Ardi,” is from a female who lived in a woodland
(see sidebar, p. 40), stood about 120 centimeters tall, and weighed
about 50 kilograms. She was thus as big as a chimpanzee and had a
brain size to match. But she did not knuckle-walk or swing through
the trees like living apes. Instead, she walked upright, planting her
feet flat on the ground, perhaps eating nuts, insects, and small mam-
mals in the woods.
She was a “facultative” biped, say the authors, still living in both
worlds—upright on the ground but also able to move on all fours on
top of branches in the trees, with an opposable big toe to grasp limbs.
“These things were very odd creatures,” says paleoanthropologist
Alan Walker of Pennsylvania State University, University Park. “You
know what Tim [White] once said: If you wanted to find something
that moved like these things, you’d have to go to the bar in Star Wars.”
Most researchers, who have waited 15 years for the publication of
this find, agree that Ardi is indeed an early hominin. They praise the
detailed reconstructions needed to piece together the crushed bones.
“This is an extraordinarily impressive work of reconstruction and
description, well worth waiting for,” says paleoanthropologist David
Pilbeam of Harvard University. “They did this job very, very well,”
agrees neurobiologist Christoph Zollikofer of the University of
Zurich in Switzerland.
But not everyone agrees with the team’s interpretations about how
Ar. ramiduswalked upright and what it reveals about our ancestors.
“The authors … are framing the debate that will inevitably follow,”
because the description and interpretation of the finds are entwined,
says Pilbeam. “My first reaction is to be skeptical about some of the
conclusions,” including that human ancestors never went through a
chimpanzee-like phase. Other researchers are focusing intently on
the lower skeleton, where some of the anatomy is so primitive that
they are beginning to argue over just what it means to be “bipedal.”
The pelvis, for example, offers only “circumstantial” evidence for
upright walking, says Walker. But however the debate about Ardi’s
locomotion and identity evolves, she provides the first hard evidence
that will inform and constrain future ideas
about the ancient hominin bauplan.
Digging it
The first glimpse of this strange creature came
on 17 December 1992 when a former graduate
student of White’s, Gen Suwa, saw a glint
among the pebbles of the desert pavement
near the village of Aramis. It was the polished
surface of a tooth root, and he immediately
knew it was a hominin molar. Over the next few days, the team scoured
the area on hands and knees, as they do whenever an important piece
of hominin is found (see story, p. 41), and collected the lower jaw of a
child with the milk molar still attached. The molar was so primitive that
the team knew they had found a hominin both older and more primitive
than Lucy. Yet the jaw also had derived traits—novel evolutionary char-
acters—shared with Lucy’s species, Au. afarensis, such as an upper
canine shaped like a diamond in side view.
The team reported 15 years ago in Naturethat the fragmentary
fossils belonged to the “long-sought potential root species for the
Hominidae.” (They first called it Au. ramidus, then, after finding
parts of the skeleton, changed it to Ar. ramidus—for the Afar words
for “root” and “ground.”) In response to comments that he needed leg
bones to prove Ar. ramiduswas an upright hominin, White joked that
he would be delighted with more parts, specifically a thigh and an
intact skull, as though placing an order.
Within 2 months, the team delivered. In November 1994, as the fos-
sil hunters crawled up an embankment, Berkeley graduate student
Yohannes Haile-Selassie of Ethiopia, now a paleoanthropologist at the
Cleveland Museum of Natural History in Ohio, spotted two pieces of a
bone from the palm of a hand. That was soon followed by pieces of a
pelvis; leg, ankle, and foot bones; many of the bones of the hand and
arm; a lower jaw with teeth—and a cranium. By January 1995, it was
apparent that they had made the rarest of rare finds, a partial skeleton.
37
Online
Podcast interview
with author
Ann Gibbons on
Ardipithecusand
fieldwork in the Afar.
sciencemag.org
Unexpected anatomy.Ardi has an opposable toe (left) and flexible hand (right);
her canines (top center) are sized between those of a human (top left) and chimp
(top right); and the blades of her pelvis (lower left) are broad like Lucy’s (yellow).
CREDITS: (LEFT) C. O. LOVEJOY ET AL., SCIENCE ; (TOP) G. SUWA ET AL., SCIENCE ; (BOTTOM) C. O. LOVEJOY ET AL., SCIENCE; (RIGHT) C. O. LOVEJOY ET AL., SCIENCE
Ardipithecus ramidusNEWSFOCUS
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
Published by AAAS
ered a hallmark of hominins. “We thought Lucy was the find of the
century but, in retrospect, it isn’t,” says paleoanthropologist Andrew
Hill of Yale University. “It’s worth the wait.”
To some researchers’ surprise, the female skeleton doesn’t look
much like a chimpanzee, gorilla, or any of our closest living primate
relatives. Even though this species probably lived soon after the dawn
of humankind, it was not transitional between African apes and
humans. “We have seen the ancestor, and it is not a chimpanzee,” says
paleoanthropologist Tim White of the University of California, Berke-
ley, co-director of the Middle Awash research group, which discovered
and analyzed the fossils.
Instead, the skeleton and pieces of at least 35 additional individuals
of Ar. ramidusreveal a new type of early hominin that was neither
chimpanzee nor human. Although the team suspects that Ar. ramidus
may have given rise to Lucy’s genus, Australopithecus, the fossils
“show for the first time that there is some new evolutionary grade of
hominid that is not Australopithecus, that is not Homo,” says paleontol-
ogist Michel Brunet of the College de France in Paris.
In 11 papers published in this issue and online, the team of 47
researchers describes how Ar. ramiduslooked and moved. The skele-
ton, nicknamed “Ardi,” is from a female who lived in a woodland
(see sidebar, p. 40), stood about 120 centimeters tall, and weighed
about 50 kilograms. She was thus as big as a chimpanzee and had a
brain size to match. But she did not knuckle-walk or swing through
the trees like living apes. Instead, she walked upright, planting her
feet flat on the ground, perhaps eating nuts, insects, and small mam-
mals in the woods.
She was a “facultative” biped, say the authors, still living in both
worlds—upright on the ground but also able to move on all fours on
top of branches in the trees, with an opposable big toe to grasp limbs.
“These things were very odd creatures,” says paleoanthropologist
Alan Walker of Pennsylvania State University, University Park. “You
know what Tim [White] once said: If you wanted to find something
that moved like these things, you’d have to go to the bar in Star Wars.”
Most researchers, who have waited 15 years for the publication of
this find, agree that Ardi is indeed an early hominin. They praise the
detailed reconstructions needed to piece together the crushed bones.
“This is an extraordinarily impressive work of reconstruction and
description, well worth waiting for,” says paleoanthropologist David
Pilbeam of Harvard University. “They did this job very, very well,”
agrees neurobiologist Christoph Zollikofer of the University of
Zurich in Switzerland.
But not everyone agrees with the team’s interpretations about how
Ar. ramiduswalked upright and what it reveals about our ancestors.
“The authors … are framing the debate that will inevitably follow,”
because the description and interpretation of the finds are entwined,
says Pilbeam. “My first reaction is to be skeptical about some of the
conclusions,” including that human ancestors never went through a
chimpanzee-like phase. Other researchers are focusing intently on
the lower skeleton, where some of the anatomy is so primitive that
they are beginning to argue over just what it means to be “bipedal.”
The pelvis, for example, offers only “circumstantial” evidence for
upright walking, says Walker. But however the debate about Ardi’s
locomotion and identity evolves, she provides the first hard evidence
that will inform and constrain future ideas
about the ancient hominin bauplan.
Digging it
The first glimpse of this strange creature came
on 17 December 1992 when a former graduate
student of White’s, Gen Suwa, saw a glint
among the pebbles of the desert pavement
near the village of Aramis. It was the polished
surface of a tooth root, and he immediately
knew it was a hominin molar. Over the next few days, the team scoured
the area on hands and knees, as they do whenever an important piece
of hominin is found (see story, p. 41), and collected the lower jaw of a
child with the milk molar still attached. The molar was so primitive that
the team knew they had found a hominin both older and more primitive
than Lucy. Yet the jaw also had derived traits—novel evolutionary char-
acters—shared with Lucy’s species, Au. afarensis, such as an upper
canine shaped like a diamond in side view.
The team reported 15 years ago in Naturethat the fragmentary
fossils belonged to the “long-sought potential root species for the
Hominidae.” (They first called it Au. ramidus, then, after finding
parts of the skeleton, changed it to Ar. ramidus—for the Afar words
for “root” and “ground.”) In response to comments that he needed leg
bones to prove Ar. ramiduswas an upright hominin, White joked that
he would be delighted with more parts, specifically a thigh and an
intact skull, as though placing an order.
Within 2 months, the team delivered. In November 1994, as the fos-
sil hunters crawled up an embankment, Berkeley graduate student
Yohannes Haile-Selassie of Ethiopia, now a paleoanthropologist at the
Cleveland Museum of Natural History in Ohio, spotted two pieces of a
bone from the palm of a hand. That was soon followed by pieces of a
pelvis; leg, ankle, and foot bones; many of the bones of the hand and
arm; a lower jaw with teeth—and a cranium. By January 1995, it was
apparent that they had made the rarest of rare finds, a partial skeleton.
37
Online
Podcast interview
with author
Ann Gibbons on
Ardipithecusand
fieldwork in the Afar.
sciencemag.org
Unexpected anatomy.Ardi has an opposable toe (left) and flexible hand (right);
her canines (top center) are sized between those of a human (top left) and chimp
(top right); and the blades of her pelvis (lower left) are broad like Lucy’s (yellow).
CREDITS: (LEFT) C. O. LOVEJOY ET AL., SCIENCE ; (TOP) G. SUWA ET AL., SCIENCE ; (BOTTOM) C. O. LOVEJOY ET AL., SCIENCE; (RIGHT) C. O. LOVEJOY ET AL., SCIENCE
Ardipithecus ramidusNEWSFOCUS
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
Published by AAAS

12. 12 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org38
It is one of only a half-dozen such skeletons known from more than
1 million years ago, and the only published one older than Lucy.
It was the find of a lifetime. But the team’s excitement was tempered
by the skeleton’s terrible condition. The bones literally crumbled when
touched. White called it road kill. And parts of the skeleton had been
trampled and scattered into more than 100 fragments; the skull was
crushed to 4 centimeters in height. The researchers decided to remove
entire blocks of sediment, covering the blocks in plaster and moving
them to the National Museum of
Ethiopia in Addis Ababa to finish
excavating the fossils.
It took three field seasons to
uncover and extract the skeleton,
repeatedly crawling the site to
gather 100% of the fossils pres-
ent.At last count, the team had
cataloged more than 110 speci-
mens of Ar. ramidus, not to men-
tion 150,000 specimens of fossil
plants and animals. “This team
seems to suck fossils out of the
earth,” says anatomist C. Owen
Lovejoy of Kent State University
in Ohio, who analyzed the post-
cranial bones but didn’t work in
the field. In the lab, he gently
unveils a cast of a tiny, pea-sized sesamoid bone for effect. “Their
obsessiveness gives you—this!”
White himself spent years removing the silty clay from the fragile
fossils at the National Museum in Addis Ababa, using brushes,
syringes, and dental tools, usually under a microscope. Museum tech-
nician Alemu Ademassu made a precise cast of each piece, and the
team assembled them into a skeleton.
Meanwhile in Tokyo and Ohio, Suwa and Lovejoy made virtual
reconstructions of the crushed skull and pelvis. Certain fossils were
taken briefly to Tokyo and scanned with a custom micro–computed
tomography (CT) scanner that could reveal what was hidden inside the
bones and teeth. Suwa spent 9 years mastering the technology to
reassemble the fragments of the cranium into a virtual skull. “I used 65
pieces of the cranium,” says Suwa, who estimates he spent 1000 hours
on the task. “You go piece by piece.”
Once he had reassembled the pieces in a digital reconstruction, he
and paleoanthropologist Berhane Asfaw of the Rift Valley Research
Service in Addis Ababa compared the skull with those of ancient and
living primates in museums worldwide. By March of this year, Suwa
was satisfied with his 10th reconstruction. Meanwhile in Ohio,
Lovejoy made physical models of the pelvic pieces based on the orig-
inal fossil and the CT scans, working closely with Suwa. He is also sat-
isfied that the 14th version of the
pelvis is accurate. “There wasan
Ardipithecusthat looked just like
that,” he says, holding up the final
model in his lab.
Putting their heads together
As they examined Ardi’s skull,
Suwa and Asfaw noted a number
of characteristics. Her lower face
had a muzzle that juts out less than
a chimpanzee’s. The cranial base is
short from front to back, indicat-
ing that her head balanced atop the
spine as in later upright walkers,
rather than to the front of the spine,
as in quadrupedal apes. Her face is
in a more vertical position than in
chimpanzees. And her teeth, like those of all later hominins, lack the
daggerlike sharpened upper canines seen in chimpanzees. The team
realized that this combination of traits matches those of an even older
skull, 6-million to 7-million-year-old Sahelanthropus tchadensis,
found by Brunet’s team in Chad. They conclude that both represent an
early stage of human evolution, distinct from both Australopithecusand
chimpanzees. “Similarities with Sahelanthropusare striking, in that it
also represents a first-grade hominid,” agrees Zollikofer, who did a
three-dimensional reconstruction of that skull.
Another, earlier species of Ardipithecus— Ar. kadabba, dated
from 5.5 million to 5.8 million years ago but known only from teeth
and bits and pieces of skeletal bones—is part of that grade, too. And
Ar. kadabba’s canines and other teeth seem to match those of a third
very ancient specimen, 6-million-year-old Orrorin tugenensisfrom
HOMO
H. habilis
Sub-Saharan
Africa
H. sapiens
Worldwide
H. floresiensis
Indonesia
H. neanderthalensis
Europe and Asia
H. heidelbergensis
Europe
SAHELANTHROPUS
ORRORIN
ARDIPITHECUS
Pliocene Epoch Pleistocene EpochHolocene Epoch
AUSTRALOPITHECUS
Au. garhi
Ethiopia
Au. rudolfensis
Eastern Africa
Au. anamensis
Kenya, Ethiopia
Au. bahrelghazali?
Abel
Chad
Au. africanus
Taung Child
South Africa
Au. robustus
South Africa
Au. aethiopicus
Eastern Africa
Au. boisei
Eastern Africa
Ar. ramidus
Ardi
Ethiopia, Kenya
Ar. kadabba
Ethiopia
S. tchadensis
Toumaï
Chad
O. tugenensis
Millennium Man
Kenya
Kenyanthropus platyops?
Kenya
Miocene Epoch
7 Million Years Ago
6
5
4
3
2
1 Million Years Ago
Today
H. erectus
Africa and Asia
Au. afarensis
Lucy
Ethiopia,
Tanzania
?
FOSSILS OF THE HUMAN FAMILY
Filling a gap.Ardipithecusprovides a link between earlier and later hominins, as seen in this timeline showing important hominin fossils and taxa.
NEWSFOCUS Ardipithecus ramidus
Fossil finders.Tim White and local Afar fossil hunters pool their finds after
scouring the hillside at Aramis.
CREDITS: (TIMELINE LEFT TO RIGHT) L. PÉRON/WIKIPEDIA, B. G. RICHM OND ET AL ., SCIENCE319, 1662 (2008); © T. WHITE 2008; WIKIPEDIA; TIM WHITE; TIM WHITE; (PHOTO) D. BRILL
Published by AAAS
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org38
It is one of only a half-dozen such skeletons known from more than
1 million years ago, and the only published one older than Lucy.
It was the find of a lifetime. But the team’s excitement was tempered
by the skeleton’s terrible condition. The bones literally crumbled when
touched. White called it road kill. And parts of the skeleton had been
trampled and scattered into more than 100 fragments; the skull was
crushed to 4 centimeters in height. The researchers decided to remove
entire blocks of sediment, covering the blocks in plaster and moving
them to the National Museum of
Ethiopia in Addis Ababa to finish
excavating the fossils.
It took three field seasons to
uncover and extract the skeleton,
repeatedly crawling the site to
gather 100% of the fossils pres-
ent.At last count, the team had
cataloged more than 110 speci-
mens of Ar. ramidus, not to men-
tion 150,000 specimens of fossil
plants and animals. “This team
seems to suck fossils out of the
earth,” says anatomist C. Owen
Lovejoy of Kent State University
in Ohio, who analyzed the post-
cranial bones but didn’t work in
the field. In the lab, he gently
unveils a cast of a tiny, pea-sized sesamoid bone for effect. “Their
obsessiveness gives you—this!”
White himself spent years removing the silty clay from the fragile
fossils at the National Museum in Addis Ababa, using brushes,
syringes, and dental tools, usually under a microscope. Museum tech-
nician Alemu Ademassu made a precise cast of each piece, and the
team assembled them into a skeleton.
Meanwhile in Tokyo and Ohio, Suwa and Lovejoy made virtual
reconstructions of the crushed skull and pelvis. Certain fossils were
taken briefly to Tokyo and scanned with a custom micro–computed
tomography (CT) scanner that could reveal what was hidden inside the
bones and teeth. Suwa spent 9 years mastering the technology to
reassemble the fragments of the cranium into a virtual skull. “I used 65
pieces of the cranium,” says Suwa, who estimates he spent 1000 hours
on the task. “You go piece by piece.”
Once he had reassembled the pieces in a digital reconstruction, he
and paleoanthropologist Berhane Asfaw of the Rift Valley Research
Service in Addis Ababa compared the skull with those of ancient and
living primates in museums worldwide. By March of this year, Suwa
was satisfied with his 10th reconstruction. Meanwhile in Ohio,
Lovejoy made physical models of the pelvic pieces based on the orig-
inal fossil and the CT scans, working closely with Suwa. He is also sat-
isfied that the 14th version of the
pelvis is accurate. “There wasan
Ardipithecusthat looked just like
that,” he says, holding up the final
model in his lab.
Putting their heads together
As they examined Ardi’s skull,
Suwa and Asfaw noted a number
of characteristics. Her lower face
had a muzzle that juts out less than
a chimpanzee’s. The cranial base is
short from front to back, indicat-
ing that her head balanced atop the
spine as in later upright walkers,
rather than to the front of the spine,
as in quadrupedal apes. Her face is
in a more vertical position than in
chimpanzees. And her teeth, like those of all later hominins, lack the
daggerlike sharpened upper canines seen in chimpanzees. The team
realized that this combination of traits matches those of an even older
skull, 6-million to 7-million-year-old Sahelanthropus tchadensis,
found by Brunet’s team in Chad. They conclude that both represent an
early stage of human evolution, distinct from both Australopithecusand
chimpanzees. “Similarities with Sahelanthropusare striking, in that it
also represents a first-grade hominid,” agrees Zollikofer, who did a
three-dimensional reconstruction of that skull.
Another, earlier species of Ardipithecus— Ar. kadabba, dated
from 5.5 million to 5.8 million years ago but known only from teeth
and bits and pieces of skeletal bones—is part of that grade, too. And
Ar. kadabba’s canines and other teeth seem to match those of a third
very ancient specimen, 6-million-year-old Orrorin tugenensisfrom
HOMO
H. habilis
Sub-Saharan
Africa
H. sapiens
Worldwide
H. floresiensis
Indonesia
H. neanderthalensis
Europe and Asia
H. heidelbergensis
Europe
SAHELANTHROPUS
ORRORIN
ARDIPITHECUS
Pliocene Epoch Pleistocene EpochHolocene Epoch
AUSTRALOPITHECUS
Au. garhi
Ethiopia
Au. rudolfensis
Eastern Africa
Au. anamensis
Kenya, Ethiopia
Au. bahrelghazali?
Abel
Chad
Au. africanus
Taung Child
South Africa
Au. robustus
South Africa
Au. aethiopicus
Eastern Africa
Au. boisei
Eastern Africa
Ar. ramidus
Ardi
Ethiopia, Kenya
Ar. kadabba
Ethiopia
S. tchadensis
Toumaï
Chad
O. tugenensis
Millennium Man
Kenya
Kenyanthropus platyops?
Kenya
Miocene Epoch
7 Million Years Ago
6
5
4
3
2
1 Million Years Ago
Today
H. erectus
Africa and Asia
Au. afarensis
Lucy
Ethiopia,
Tanzania
?
FOSSILS OF THE HUMAN FAMILY
Filling a gap.Ardipithecusprovides a link between earlier and later hominins, as seen in this timeline showing important hominin fossils and taxa.
NEWSFOCUS Ardipithecus ramidus
Fossil finders.Tim White and local Afar fossil hunters pool their finds after
scouring the hillside at Aramis.
CREDITS: (TIMELINE LEFT TO RIGHT) L. PÉRON/WIKIPEDIA, B. G. RICHM OND ET AL ., SCIENCE319, 1662 (2008); © T. WHITE 2008; WIKIPEDIA; TIM WHITE; TIM WHITE; (PHOTO) D. BRILL
Published by AAAS

13. 13www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Kenya, which also has a
thighbone that appears
to have been used for
upright walking (Science,
21 March 2008, p. 1599).
So, “this raises the in-
triguing possibility that
we’re looking at the same
genus” for specimens
now put in three genera,
says Pilbeam. But the
discoverers of O. tuge-
nensisaren’t so sure. “As for Ardi and Orrorinbeing the same genus,
no, I don’t think this is possible, unless one really wants to accept an
unusual amount of variability” within a taxon, says geologist Martin
Pickford of the College de France, who found Orrorinwith Brigitte
Senut of the National Museum of Natural History in Paris.
Whatever the taxonomy of Ardipithecusand the other very ancient
hominins, they represent “an enormous jump to Australopithecus,” the
next hominin in line (see timeline, p. 38), says australopithecine expert
William Kimbel of Arizona State University, Tempe. For example,
although Lucy’s brain is only a little larger than that of Ardipithecus,
Lucy’s species, Au. afarensis, was an adept biped. It walked upright
like humans, venturing increasingly into more diverse habitats,
including grassy savannas. And it had lost its opposable big toe, as
seen in 3.7-million-year-old footprints at Laetoli, Tanzania, reflecting
an irreversible commitment to life on the ground.
Lucy’s direct ancestor is widely considered to be Au. anamensis,
a hominin whose skeleton is poorly known, although its shinbone
suggests it walked upright 3.9 million to 4.2 million years ago in
Kenya and Ethiopia. Ardipithecusis the current leading candidate
for Au. anamensis’s ancestor, if only because it’s the only putative
hominin in evidence between 5.8 million and 4.4 million years ago.
Indeed, Au. anamensisfossils appear in the Middle Awash region
just 200,000 years after Ardi.
Making strides
But the team is not connecting the dots between Au. anamensisand
Ar. ramidusjust yet, awaiting more fossils. For now they are focusing
on the anatomy of Ardi and how she moved through the world. Her
foot is primitive, with an opposable big toe like that used by living
apes to grasp branches. But the bases of the four other toe bones
were oriented so that they reinforced the forefoot into a more rigid
lever as she pushed off. In contrast, the toes of a chimpanzee curve
as flexibly as those in their hands, say Lovejoy and co-author
Bruce Latimer of Case Western Reserve University in Cleveland.
Ar. ramidus“developed a pretty good bipedal foot while at the same
time keeping an opposable first toe,” says Lovejoy.
The upper blades of Ardi’s pelvis are shorter and broader than in
apes. They would have lowered the trunk’s center of mass, so she could
balance on one leg at a time while walking, says Lovejoy. He also
infers from the pelvis that her spine was long and curved like a
human’s rather than short and stiff like a chimpanzee’s. These
changes suggest to him that Ar. ramidus“has been bipedal for a very
long time.”
Yet the lower pelvis is still quite large and primitive, similar to
African apes rather than hominins. Taken with the opposable big toe,
and primitive traits in the hand and foot, this indicates that Ar. ramidus
didn’t walk like Lucy and was still spending a lot of time in the trees.
But it wasn’t suspending its body beneath branches like African apes
or climbing vertically, says Lovejoy. Instead, it was a slow, careful
climber that probably moved on flat hands and feet on top of branches
in the midcanopy, a type of locomotion known as palmigrady. For
example, four bones in the wrist of Ar. ramidusgave it a more flexible
hand that could be bent backward at the wrist. This is in contrast to the
hands of knuckle-walking chimpanzees and gorillas, which have stiff
wrists that absorb forces on their knuckles.
However, several researchers aren’t so sure about these inferences.
Some are skeptical that the crushed pelvis really shows the anatomical
details needed to demonstrate bipedality. The pelvis is “suggestive” of
bipedality but not conclusive, says paleoanthropologist Carol Ward of the
University of Missouri, Columbia. Also, Ar. ramidus“does not appear to
have had its knee placed over the ankle, which means that when walk-
ing bipedally, it would have had to shift its weight to the side,” she says.
Paleoanthropologist William Jungers of Stony Brook University in
New York state is also not sure that the skeleton was bipedal. “Believe
me, it’s a unique form of bipedalism,” he says. “The postcranium alone
would not unequivocally signal hominin status, in my opinion.” Paleo-
anthropologist Bernard Wood of George Washington University in
Washington, D.C., agrees. Looking at the skeleton as a whole, he says,
“I think the head is consistent with it being a hominin, … but the rest of
the body is much more questionable.”
All this underscores how difficult it may be to recognize and
define bipedality in the earliest hominins as they began to shift from
trees to ground. One thing does seem clear, though: The absence of
many specialized traits found in African apes suggests that our
ancestors never knuckle-walked.
That throws a monkey wrench into a hypothesis about the last
common ancestor of living apes and humans. Ever since Darwin
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 39
CREDITS (TOP TO BOTTOM): TIM WHITE; BOB CHRISTY/NEWS AND INFORMATION, KENT STATE UNIVERSITY; TIM WHITE
Dream team.Gen Suwa (left) in Tokyo focused on the skull; C. Owen Lovejoy (top
right) in Kent, Ohio, studied postcranial bones; and Yohannes Haile-Selassie and
Berhane Asfaw found and analyzed key fossils in Ethiopia.
Ardipithecus ramidusNEWSFOCUS
Published by AAAS
ered a hallmark of hominins. “We thought Lucy was the find of the
century but, in retrospect, it isn’t,” says paleoanthropologist Andrew
Hill of Yale University. “It’s worth the wait.”
To some researchers’ surprise, the female skeleton doesn’t look
much like a chimpanzee, gorilla, or any of our closest living primate
relatives. Even though this species probably lived soon after the dawn
of humankind, it was not transitional between African apes and
humans. “We have seen the ancestor, and it is not a chimpanzee,” says
paleoanthropologist Tim White of the University of California, Berke-
ley, co-director of the Middle Awash research group, which discovered
and analyzed the fossils.
Instead, the skeleton and pieces of at least 35 additional individuals
of Ar. ramidusreveal a new type of early hominin that was neither
chimpanzee nor human. Although the team suspects that Ar. ramidus
may have given rise to Lucy’s genus, Australopithecus, the fossils
“show for the first time that there is some new evolutionary grade of
hominid that is not Australopithecus, that is not Homo,” says paleontol-
ogist Michel Brunet of the College de France in Paris.
In 11 papers published in this issue and online, the team of 47
researchers describes how Ar. ramiduslooked and moved. The skele-
ton, nicknamed “Ardi,” is from a female who lived in a woodland
(see sidebar, p. 40), stood about 120 centimeters tall, and weighed
about 50 kilograms. She was thus as big as a chimpanzee and had a
brain size to match. But she did not knuckle-walk or swing through
the trees like living apes. Instead, she walked upright, planting her
feet flat on the ground, perhaps eating nuts, insects, and small mam-
mals in the woods.
She was a “facultative” biped, say the authors, still living in both
worlds—upright on the ground but also able to move on all fours on
top of branches in the trees, with an opposable big toe to grasp limbs.
“These things were very odd creatures,” says paleoanthropologist
Alan Walker of Pennsylvania State University, University Park. “You
know what Tim [White] once said: If you wanted to find something
that moved like these things, you’d have to go to the bar in Star Wars.”
Most researchers, who have waited 15 years for the publication of
this find, agree that Ardi is indeed an early hominin. They praise the
detailed reconstructions needed to piece together the crushed bones.
“This is an extraordinarily impressive work of reconstruction and
description, well worth waiting for,” says paleoanthropologist David
Pilbeam of Harvard University. “They did this job very, very well,”
agrees neurobiologist Christoph Zollikofer of the University of
Zurich in Switzerland.
But not everyone agrees with the team’s interpretations about how
Ar. ramiduswalked upright and what it reveals about our ancestors.
“The authors … are framing the debate that will inevitably follow,”
because the description and interpretation of the finds are entwined,
says Pilbeam. “My first reaction is to be skeptical about some of the
conclusions,” including that human ancestors never went through a
chimpanzee-like phase. Other researchers are focusing intently on
the lower skeleton, where some of the anatomy is so primitive that
they are beginning to argue over just what it means to be “bipedal.”
The pelvis, for example, offers only “circumstantial” evidence for
upright walking, says Walker. But however the debate about Ardi’s
locomotion and identity evolves, she provides the first hard evidence
that will inform and constrain future ideas
about the ancient hominin bauplan.
Digging it
The first glimpse of this strange creature came
on 17 December 1992 when a former graduate
student of White’s, Gen Suwa, saw a glint
among the pebbles of the desert pavement
near the village of Aramis. It was the polished
surface of a tooth root, and he immediately
knew it was a hominin molar. Over the next few days, the team scoured
the area on hands and knees, as they do whenever an important piece
of hominin is found (see story, p. 41), and collected the lower jaw of a
child with the milk molar still attached. The molar was so primitive that
the team knew they had found a hominin both older and more primitive
than Lucy. Yet the jaw also had derived traits—novel evolutionary char-
acters—shared with Lucy’s species, Au. afarensis, such as an upper
canine shaped like a diamond in side view.
The team reported 15 years ago in Naturethat the fragmentary
fossils belonged to the “long-sought potential root species for the
Hominidae.” (They first called it Au. ramidus, then, after finding
parts of the skeleton, changed it to Ar. ramidus—for the Afar words
for “root” and “ground.”) In response to comments that he needed leg
bones to prove Ar. ramiduswas an upright hominin, White joked that
he would be delighted with more parts, specifically a thigh and an
intact skull, as though placing an order.
Within 2 months, the team delivered. In November 1994, as the fos-
sil hunters crawled up an embankment, Berkeley graduate student
Yohannes Haile-Selassie of Ethiopia, now a paleoanthropologist at the
Cleveland Museum of Natural History in Ohio, spotted two pieces of a
bone from the palm of a hand. That was soon followed by pieces of a
pelvis; leg, ankle, and foot bones; many of the bones of the hand and
arm; a lower jaw with teeth—and a cranium. By January 1995, it was
apparent that they had made the rarest of rare finds, a partial skeleton.
37
Online
Podcast interview
with author
Ann Gibbons on
Ardipithecusand
fieldwork in the Afar.
sciencemag.org
Unexpected anatomy.Ardi has an opposable toe (left) and flexible hand (right);
her canines (top center) are sized between those of a human (top left) and chimp
(top right); and the blades of her pelvis (lower left) are broad like Lucy’s (yellow).
CREDITS: (LEFT) C. O. LOVEJOY ET AL., SCIENCE ; (TOP) G. SUWA ET AL., SCIENCE ; (BOTTOM) C. O. LOVEJOY ET AL., SCIENCE; (RIGHT) C. O. LOVEJOY ET AL., SCIENCE
Ardipithecus ramidusNEWSFOCUS
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
Published by AAAS

14. 14 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org40
NEWSFOCUS Ardipithecus ramidus
suggested in 1871 that our ancestors arose in Africa, researchers have
debated whether our forebears passed through a great-ape stage in
which they looked like proto-chimpanzees (Science, 21 November
1969, p. 953). This “troglodytian” model for early human behavior
(named for the common chimpanzee, Pan troglodytes) suggests that
the last common ancestor of the African apes and humans once had
short backs, arms adapted for swinging, and a pelvis and limbs
adapted for knuckle walking. Then our ancestors lost these traits,
while chimpanzees and gorillas kept them. But this view has been
uninformed by fossil evidence because there are almost no fossils of
early chimpanzees and gorillas.
Some researchers have thought that the ancient African ape bau-
plan was more primitive, lately citing clues from fragmentary fossils
of apes that lived from 8 million to 18 million years ago. “There’s
been growing evidence from the Miocene apes that the common
ancestor may have been more primitive,” says Ward. Now
Ar. ramidusstrongly supports that notion. The authors repeatedly
note the many ways that Ar. ramidusdiffers from chimpanzees and
gorillas, bolstering the argument that it was those apes that changed
the most from the primitive form.
But the problem with a more “generalized model” of an arboreal
ape is that “it is easier to say what it wasn’t than what it was,” says
Ward. And if the last common ancestor, which according to genetic
studies lived 5 million to 7 million years ago, didn’t look like a
chimp, then chimpanzees and gorillas evolved their numerous simi-
larities independently, after gorillas diverged from the chimp/human
line. “I find [that] hard to believe,” says Pilbeam.
As debate over the implications of Ar. ramidusbegins, the one thing
that all can agree on is that the new papers provide a wealth of data to
frame the issues for years. “No matter what side of the arguments you
come down on, it’s going to be food for thought for generations of
graduate students,” says Jungers. Or, as Walker says: “It would have
been very boring if it had looked half-chimp.”
–ANN GIBBONS
ARAMIS, ETHIOPIA—A long cairn of black stones
marks the spot where a skeleton of Ardipithecus
ramiduswas found, its bones broken and scattered
on a barren hillside. Erected as a monument to an
ancient ancestor in the style of an Afar tribesman’s
grave, the cairn is a solitary marker in an almost
sterile zone, devoid of life except for a few spindly
acacia trees and piles of sifted sediment.
That’s because the Middle Awash research team
sucked up everything in sight at this spot, hunting
for every bit of fossil bone as well as clues to the
landscape 4.4 million years ago, when Ardipithe-
cusdied here. “Literally, we crawled every square
inch of this locality,” recalls team co-leader Tim
White of the University of California, Berkeley.
“You crawl on your hands and knees, collecting
every piece of bone, every piece of wood, every
seed, every snail, every scrap. It was 100% collec-
tion.” The heaps of sediment are all that’s left
behind from that fossil-mining operation, which
yielded one of the most important fossils in human evolution (see main text,
p. 36), as well as thousands of clues to its ecology and environment.
The team collected more than 150,000 specimens of fossilized plants and
animals from nearby localities of the same age, from elephants to songbirds
to millipedes, including fossilized wood, pollen, snails, and larvae. “We have
crates of bone splinters,” says White.
A team of interdisciplinary researchers then used these fossils and
thousands of geological and isotopic samples to reconstruct Ar. ramidus’s
Pliocene world, as described in companion papers in this issue (see p. 66
and 87). From these specimens, they conclude that Ardi lived in a wood-
land, climbing among hackberry, fig, and palm trees and coexisting with
monkeys, kudu antelopes, and peafowl. Doves and parrots flew overhead.
All these creatures prefer woodlands, not the open, grassy terrain often
conjured for our ancestors.
The team suggests that Ar. ramiduswas “more omnivorous” than chim-
panzees, based on the size, shape, and enamel distribution of its teeth. It
probably supplemented woodland plants such as fruits, nuts, and tubers
with the occasional insects, small mammals, or bird eggs. Carbon-isotope
studies of teeth from five individuals show that Ar. ramidusate mostly
woodland, rather than grassland, plants. Although Ar. ramidusprobably ate
figs and other fruit when ripe, it didn’t consume as much fruit as chim-
panzees do today.
This new evidence overwhelmingly refutes the once-favored but now
moribund hypothesis that upright-walking hominins arose in open grass-
lands. “There’s so much good data here that people aren’t going to be able to
question whether early hominins were living in woodlands,” says paleo-
anthropologist Andrew Hill of Yale University. “Savannas had nothing to do
with upright walking.”
Geological studies indicate that most of the fossils were buried within a
relatively short window of time, a few thousand to, at most, 100,000 years
ago, says geologist and team co-leader Giday WoldeGabriel of the Los Alamos
National Laboratory in New Mexico. During that sliver of time, Aramis was not
a dense tropical rainforest with a thick canopy but a humid, cooler woodland.
The best modern analog is the Kibwezi Forest in Kenya, kept wet by ground-
water, according to isotope expert Stanley Ambrose of the University of
Illinois, Urbana-Champaign. These woods have open stands of trees, some
20 meters high, that let the sun reach shrubs and grasses on the ground.
Judging from the remains of at least 36 Ardipithecusindividuals found so
far at Aramis, this was prime feeding ground for a generalized early biped. “It
was the habitat they preferred,” says White. –A.G.
Habitat for Humanity
Past and present.Ardipithecus’s wood-
land was more like Kenya’s Kibwezi Forest
(left) than Aramis today.
CREDITS (LEFT TO RIGHT): TIM WHITE; ANN GIBBONS
Published by AAAS

15. 15www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
MIDDLE AWASH VALLEY, THE AFAR DEPRESSION, ETHIOPIA—It’s
about 10 a.m. on a hot morning in December, and Tim White is
watching a 30-year-old farmer inch his way up a slippery hill on his
knees, picking through mouse-colored rubble for a bit of gray bone.
The sun is already bleaching the scrubby badlands, making it diffi-
cult to distinguish a fragment of bone in the washed-out beige and
gray terrain. The only shade in this parched gully is from a small,
thorny acacia tree, so the fossil hunters have draped their heads with
kerchiefs that hang out from under their “Cal” and “Obama for Pres-
ident” baseball caps, making them look like a strange tribe of Berke-
ley Bedouins. If there are fossils here,
White is confident that the slender
farmer, Kampiro Kayrento, will find
them. “Kampiro is the best person in the
world for finding little pieces of fos-
silized human bone,” says White, 59, a
paleoanthropologist at the University of
California, Berkeley, who has collected
fossils in this region since 1981.
Watching Kayrento is a sort of specta-
tor sport, because he scores so often. Just
minutes earlier, he had walked over the
crest of a small hill, singing softly to him-
self, and had spotted the fossilized core of
a horn from an ancient bovid, or antelope.
Then he picked up a flat piece of gray bone
nearby and showed the fossil to Ethiopian
paleoanthropologist Berhane Asfaw, ask-
ing, “Bovid?” Asfaw, 55, who hired
Kayrento when he was a boy hanging out
at fossil sites in southern Ethiopia, looked
over the slightly curved piece of bone the
size of a silver dollar and suggested, “Mon-
key?” as he handed it to White. White
turned it over gently in his hands, then said: “Check that, Berhane. We
just found a hominid cranium. Niiiice.”
As word spreads that Kayrento found a hominin, or a member of
the taxon that includes humans and our ancestors, the other fossil
hunters tease him: “Homo bovid! Homobovid! Niiiice.”
The Middle Awash project, which includes 70 scientists from 18
nations, is best known for its discovery of the 4.4-million-year-old
partial skeleton of Ardipithecus ramidusat Aramis, about 34 kilo-
meters north of here. That skeleton is now dramatically revising ideas
of how upright walking evolved and how our earliest ancestors dif-
fered from chimpanzees (see overview, p. 60, and main Focus text,
p. 36). But Aramis is just one of 300 localities in the Middle Awash,
which is the only place in the world to yield fossils that span the entire
saga of hominid evolution. At last count, this team had gathered
19,000 vertebrate fossils over the past 19 years. These include about
300 specimens from seven species of hominins, from some of the
first members of the human family, such as 5.8-million-year-old
Ar. ramidus kadabba, to the earliest members of our own species,
Homo sapiens, which lived here about 160,000 years ago.
As they work in different places in the
valley, the team members travel back and
forth in time. Today, this core group is
working in the western foothills near the
Burka catchment, where an ancient river
laid down sediments 3 million to 2 mil-
lion years ago and where the team has
found specimens of Australopithecus
garhi, a species they suspect may have
given rise to the first members of our
genus, Homo.
This season, after a rough start, the 25
scientists, students, cooks, and Ethiopian
and Afar officials and guards in camp are
working well together. Their tented camp is
hours from any town, graded road, or fresh
water. (They dug their own well to get
water.) “The 1st week, it’s like an engine
that’s running but not running smoothly,”
says White, who, with Asfaw, runs a well-
organized camp where every tool, map, and
shower bag has its proper place. “By the
3rd week, people know their jobs.”
The 1st week, White and a paleontolo-
gist were sick, and White is still fighting a harsh cough that keeps
him awake at night. The 2nd week, some aggressive Alisera tribes-
men who live near the Ar. ramidussite threatened to kill White and
Asfaw, making it difficult to return there. (That’s one reason the
team travels with six Afar policemen armed with AK-47s and
Obama caps, dubbed “The Obama Police.”) The day before, a stu-
dent had awakened with a high fever and abdominal pain and had to
be driven 4 hours to the nearest clinic, where he was diagnosed with
a urinary tract infection, probably from drinking too little water in
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 41
CREDITS (TOP TO BOTTOM): HENRY GILBERT; SOURCE: TIM WHITE
The View From Afar
How do you find priceless hominin fossils in a hostile desert? Build a strong team and obsess over the details
PALEOANTHROPOLOGY
HADAR
GONA
Awash
River
Aramis
Burka
Yardi Lake
Bouri
Peninsula
Afar
Rift
Addis Ababa
ETHIOPIA
WESTERN AFAR RIFT, ETHIOPIA
Middle Awash
Hominid Localities
Ardipithecus
Australopithecus
Homo
Ancestral territory. The area where Ardi was found is rich in
hominin fossil sites, including these worked by the Middle
Awash research team.
The crawl. Researchers hunt
down every fossil at Aramis.
Published by AAAS

16. 16 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
the 35˚C heat. “The best laid plans change every day,” says White,
who has dealt with poisonous snakes, scorpions, malarial mosqui-
toes, lions, hyenas, flash floods, dust tornadoes, warring tribesmen,
and contaminated food and water over the years. “Nothing in the
field comes easy.”
Calling the “A” team
Nothing in the Afar, for that matter, comes easy. We are reminded of
that as we drive across the dusty Saragata plain to the target fossil site
at 8 a.m., making giant circles in the dust with the Toyota Land Cruiser
so we can find our tracks home at the end of the day. Men clad in plaid
wraps, with AK-47s slung over their shoulders, flag us down seeking
help. They bring over a woman who looks to be in her 70s but is prob-
ably much younger. Her finger is bleeding, and the men tell White and
Asfaw, in Afar, that a puff adder bit her the night before while she was
gathering wood. A quick-thinking boy had sliced her finger with a
knife, releasing the venom and probably saving her life. White gets out
a first-aid kit, removes a crude poultice, and cleans and bandages the
wound, putting on an antibiotic cream. “It’s good she survived the
night,” he says as we drive off. “The danger now is infection.”
After inching down the sandy bank of a dry river, we reach the so-
called Chairman’s site. This is one of dozens of fossil localities dis-
covered in the Burka area since 2005: exposed hillsides that were
spotted in satellite and aerial photos, then laboriously explored on
foot. The plan was to search for animal fossils to help date a hominid
jawbone discovered last year. But in the 1st hour, with Kayrento’s dis-
covery, they’re already on the trail of another individual instead.
As soon as White identifies the bit of skull bone, he swings into
action. With his wiry frame and deep voice, he is a commanding
presence, and it soon becomes clear how he earned his nickname,
“The General.” In his field uniform—a suede Australian army hat
with a rattlesnake band, blue jeans, and driving gloves without
fingers—he uses a fossil pick to delineate the zones in the sandstone
where he wants the crew
deployed. “Get everybody
out of the area,” he calls to the
15 people already fanned out
over the gully, scanning for
fossils. “I want the ‘A’ team.”
He singles out Kayrento and
three others and hands them
yellow pin-flags, saying, “Go
back to the bottom.” As he
watches them move up the
slope, he warns: “Go slowly.
You’re moving too fast. …
Don’t squash the slope. Move
like a cat, not a cow.”
By looking at the relatively
fresh fractured edge of the
bone fragment, White knows
that it comes from a larger piece of skull that broke after it was exposed,
not while it was buried. As Kayrento and the others find other bits of
bone, they place yellow pin-flags at those spots. “This process estab-
lishes the distributional cone,” White explains. The top flag marks the
highest point on the surface where the skull came out of the ground; the
bottom boundary marks the farthest point where a fragment might
finally have come to rest, following the fall line down the slope.
This discovery also illustrates one reason why the team comes to
the field right after the rainy season. If they’re lucky, rain and floods
will cut into the ancient sediments, exposing fossils. But they have to
get there before the fossils disintegrate as they are exposed to the ele-
ments or are trampled by the Afar’s goats, sheep, and cattle. Timing is
everything, and this season they’re a bit late. “The ideal situation is to
find a fossil just as it is eroding out of the bank,” says White.
As they crawl the entire length of the gully, they turn over every
rock, mud clod, and piece of carbonate rubble to make sure it doesn’t
contain a fossil fragment. “Not good,” says Kayrento. “This is yucky,”
agrees Asfaw, co-director of the team and the first Ethiopian scientist
to join it, in 1979 when he was invited to earn his Ph.D. at Berkeley
(Science, 29 August 2003, p. 1178).
After 2 hours, the team has collected a few more pieces of skull
around the temple, forehead, and ear. “It’s getting bigger by the
minute,” White says. “If we’re lucky, we’ll find it buried right in here.”
The team has to wait until the next day to find out just how lucky.
At 9:45 a.m. Thursday, they return with reinforcements: Asfaw has
hired two Afar men to help with the heavy lifting of buckets of dirt.
With a button-down Oxford cloth shirt and a pistol stuck in the waist-
band of his khakis, Asfaw commands respect, and he is the best at
negotiating with the Afar. In this case, he settles an argument by letting
clan leaders select which men, among a large group, will get jobs.
At the site, White sets up a perimeter of blue pin-flags that look
like a mini slalom course, outlining the gully that he calls the “Hot
Zone” where fossil pieces are most likely to be buried. The plan is to
excavate all the rock and dirt around those flags, down to the origi-
nal layers of sediment. White explains that the ancient landscape
would have been flatter and more verdant before tectonic move-
ments of Earth’s crust cracked and tilted the sediment layers. But the
original soil is still there, a red-brown layer of clay beneath a gray
veneer of sandstone. “Throw every piece of stone out of the chan-
nel,” he orders. “If you see a hominid, I need to know right away!”
White and Kayrento literally sweep off the gray lag with a push
broom and then scrape back the layers of time with a trowel to the
ancient surface underneath. “Once we brush out the slopes, we’ll be
Division of labor.Kampiro Kayrento (top left) homes in on fossils; he and others
sweep the surface, and Giday WoldeGabriel dates sediments.
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org42
CREDITS (TOP TO BOTTOM): DAVID BRILL; TIM WHITE; DAVID BRILL
Published by AAAS

17. 17www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
sure no fossil is left in place,”
says White. In case they miss a
fragment, the loose sediment is
carried to giant sieves where the
crew sifts it for bits of bone or
teeth. The sifted rubble is taken
to a circle of workers who then
empty it into small aluminum
pans, in which they examine
every single, tiny piece—a job
that gives new meaning to the
word tedium. “Sieving 101,”
observes Asfaw, who supervises
sieving and picking today.
By 11:10 a.m., the pace of dis-
covery has slowed. When the A
team tells White it’s “not good,” he tries to infuse them with some of
his energy, reminding everyone to stay focused, to keep going, to not
step on fossils. But by midday, White is grumbling, too, because
they’ve scoured the Hot Zone and it’s clear the skull is not there.
“We’ve eliminated every hope of finding it in situ.”
Time travel
It’s a good time to take a walk with the four geologists, who are comb-
ing the terrain, hoping to find sediments with volcanic minerals to
help them date the locality and its fossils precisely. While fossil
hunters move slowly, stooped at the waist and focused on the ground,
the geologists move fast, heads up, scanning the next horizon for a
rock face with a layer cake of sediments, like those exposed in road
cuts. The 6-million-year record of Middle Awash sediments is not
stacked neatly in one place, with oldest rocks on the bottom and
youngest on top. (If it were, the stack would be 1 kilometer thick.)
Instead, the rocks are faulted and tilted into different ridges. By trac-
ing a once-horizontal layer from ridge to ridge, sometimes for kilo-
meters, the geologists can link the layers and place different snap-
shots of time into a sequence.
Today, Ethiopian geologist Giday WoldeGabriel of the Los
Alamos National Laboratory in New Mexico, also a co-leader of the
team (he joined in 1992), is searching for a familiar-looking motif—
a distinct layer of volcanic tuff called the SHT (Sidiha Koma Tuff),
previously dated to 3.4 million years ago by radiometric methods.
So far, the team has found just one species of hominin—
Au. garhi—that lived at this time in the Middle Awash (Science,
23 April 1999, p. 629), although a more robust
species,Au. aethiopicus, appears 2.6 million years
ago in southern Ethiopia and Kenya. That’s also when
the earliest stone tools appear in Gona, Ethiopia,
100 kilometers north of here. The earliest fossils of our
genus Homocome a bit later—at 2.3 million years ago
at Hadar, near Gona, also with stone tools. That’s why
it is important to date Au. garhiprecisely: Was it the maker of the
stone tools left in the Afar? The team thinks Au. garhiis the direct
descendant of the more primitive Au. afarensis, best known as the
species that includes the famous 3.2-million-year-old skeleton of
Lucy, also from Hadar. But did Au. garhithen evolve into early
Homo? They need better dates—and more fossils—to find out.
“Now that we have the SHT as a reference point here, we have
to try to trace it to where the new fossils are,” says WoldeGabriel.
The only problem is that the SHT is several ridges and basins over
from the excavation; linking the two will be difficult if not impos-
sible. The team will also use other methods to date the new fossils.
Luckily, the fossil hunters
have found a pig known to
have lived about 2.6 million
to 2.7 million years ago,
which suggests that the sedi-
ments and the new discovery are also that old.
At 9 a.m. Friday, 12 December, we’re back at the Chairman’s site
for a 3rd day, this time with a film crew from Sweden. After White
and Kayrento jokingly reenact the discovery of the skull bone for the
film crew, they resume sweeping and sifting, exactly where they left
off. At first, there’s little return. Berkeley postdoc Cesur Pehlevan
from Ankara hands White a piece of bone: “Nope, tough luck. Right
color, right thickness. Nope, sorry.”
Finally, someone hands White something special. “Oh nice, frontal
bone with frontal sinus. This is getting better. That’s what we’re after,”
says White. “If we can get that brow ridge, we can match it with the
known species.” He turns over the new piece of frontal bone in his
hand, examining it like a diamond dealer assessing a gemstone.
By the end of 3 days, the team of 20 will have collected a dozen
pieces of one skull, an average yield for this region.
Taken together, says White, those pieces show that
“It’s an Australopithecusbecause it has a small brain-
case, small chewing apparatus.” There’s still not
enough to identify the species, though White thinks it
is Au. garhi. He notes that “it’s a big boy, big for an
australopithecine.” If it is Au. garhi, that would be
one more bit of evidence to suggest that Au. afarensis gave rise to
Au. garhi; males are bigger than females in Au. afarensis—and so
perhaps in Au. garhi, too.
For now, White and Asfaw are pleased with the new snapshot
they’re getting of Au. garhi. On our way back to camp, White stops
so we can take a photo of the moon rising over Yardi Lake in front
of us, the sun setting behind us. The landscape has changed since
the australopithecines were here. But one thing’s been constant in
the Middle Awash, he notes: “Hominids have been right here look-
ing at the moon rising over water for millions of years.”
–ANN GIBBONS
Ardipithecus ramidusNEWSFOCUS
43
CREDITS (TOP TO BOTTOM): TIM WHITE; HANK WESSELMAN; TIM WHITE
Intensive care.Tim White uses
dental tools and a gluelike adhesive
to extract fragile fossils from
matrix.
“Nothing in the field
comes easy.”
–TIM WHITE, UNIVERSITY
OF CALIFORNIA, BERKELEY
Published by AAAS

18. 18 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org64
AUTHORS’SUMMARIES
CREDIT: ILLUSTRATION OF AR. RAMIDUS : COPYRIGHT J. H. MATTERNES
Ardipithecus ramidusand the Paleobiology
of Early Hominids
Tim D. White, Berhane Asfaw, Yonas Beyene, Yohannes Haile-Selassie, C. Owen Lovejoy, Gen Suwa,
Giday WoldeGabriel
C
harles Darwin and Thomas
Huxley were forced to
ponder human origins
and evolution without a relevant
fossil record. With only a few
Neanderthal fossils available to
supplement their limited knowl-
edge of living apes, they specu-
lated about how quintessentially
human features such as upright
walking, small canines, dexterous
hands, and our special intelligence
had evolved through natural selec-
tion to provide us with our com-
plex way of life. Today we know of
early Homofrom >2.0 million
years ago (Ma) and have a record
of stone tools and animal butchery
that reaches back to 2.6 Ma. These demonstrate just how deeply tech-
nology is embedded in our natural history.
Australopithecus, a predecessor of Homothat lived about 1 to 4 Ma
(see figure), was discovered in South Africa in 1924. Although slow to
gain acceptance as a human ancestor, it is now recognized to represent
an ancestral group from which Homoevolved. Even after the discov-
eries of the partial skeleton (“Lucy”) and fossilized footprints
(Laetoli) of Au. afarensis, and other fossils that extended the antiquity
of Australopithecusto ~3.7 Ma, the hominid fossil record before
Australopithecuswas blank. What connected the small-brained, small-
canined, upright-walking Australopithecusto the last common ances-
tor that we shared with chimpanzees some time earlier than 6 Ma?
The 11 papers in this issue, representing the work of a large inter-
national team with diverse areas of expertise, describe Ardipithecus
ramidus, a hominid species dated to 4.4 Ma, and the habitat in which
it lived in the Afar Rift region of northeastern Ethiopia. This species,
substantially more primitive than Australopithecus, resolves many
uncertainties about early human evolution, including the nature of the
last common ancestor that we shared with the line leading to living
chimpanzees and bonobos. The Ardipithecusremains were recovered
from a sedimentary horizon representing a short span of time (within
100 to 10,000 years). This has enabled us to assess available and pre-
ferred habitats for the early hominids by systematic and repeated
sampling of the hominid-bearing strata.
By collecting and classifying thousands of vertebrate, invertebrate,
and plant fossils, and characterizing the isotopic composition of soil
samples and teeth, we have learned that Ar. ramiduswas a denizen of
woodland with small patches of forest. We have also learned that it
probably was more omnivorous than chimpanzees (ripe fruit special-
ists) and likely fed both in trees and on the ground. It apparently con-
sumed only small amounts of open-environment resources, arguing
against the idea that an inhabitation of grasslands was the driving force
in the origin of upright walking.
Ar. ramidus, first described in 1994 from teeth and jaw fragments,
is now represented by 110 specimens, including a partial female
skeleton rescued from erosional degradation. This individual weighed
about 50 kg and stood about 120 cm tall. In the context of the many
other recovered individuals of this species, this suggests little body
size difference between males and females. Brain size was as small as
in living chimpanzees. The numerous recovered teeth and a largely
complete skull show that Ar. ramidushad a small face and a reduced
canine/premolar complex, indicative of minimal social aggression.
Its hands, arms, feet, pelvis, and legs collectively reveal that it moved
capably in the trees, supported on its feet and palms (palmigrade
clambering), but lacked any characteristics typical of the suspen-
sion, vertical climbing, or knuckle-walking of modern gorillas and
chimps. Terrestrially, it engaged in a form of bipedality more prim-
itive than that of Australopithecus, and it lacked adaptation to
“heavy” chewing related to open environments (seen in later
Australopithecus). Ar. ramidusthus indicates that the last common
ancestors of humans and African apes were not chimpanzee-like and
that both hominids and extant African apes are each highly special-
ized, but through very different evolutionary pathways.
Evolution of hominids and African apes since the gorilla/chimp+human (GLCA) and chimp/human (CLCA) last
common ancestors. Pedestals on the left show separate lineages leading to the extant apes (gorilla, and chimp and
bonobo); text indicates key differences among adaptive plateaus occupied by the three hominid genera.
• Partially arboreal

• Facultative biped


• Feminized canine

• Woodland omnivore


CLCA

• Palmigrade
arborealist

• Dimorphic
canines

• Forest
frugivore/
omnivore

• Striding terrestrial biped

• Postcanine megadontia


• Pan-African


• Wide niche
GLCA
• Enlarged brain


• Dentognathic reduction

• Technology-reliant


• Old World range
A
rd
ip
ithecus (~ 6 to 4 Ma)
A
u
s
tralopithecus (~ 4 to 1 Ma)




H
o
m
o (< ~ 2.5
Ma
)




Gorilla


P
a
n Pan




gorilla
paniscus




troglodytes
See pages 62–63 for authors’ affiliations.
When citing, please refer to the full paper, available at DOI 10.1126/science.1175802.
See pages 6–7 for authors’ affiliations.

19. 19www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 65
AUTHORS’SUMMARIES
The Geological, Isotopic, Botanical,
Invertebrate, and Lower Vertebrate
Surroundings of Ardipithecus ramidus
Giday WoldeGabriel, Stanley H. Ambrose, Doris Barboni, Raymonde Bonnefille, Laurent Bremond, Brian Currie,
David DeGusta, William K. Hart, Alison M. Murray, Paul R. Renne, M. C. Jolly-Saad, Kathlyn M. Stewart, Tim D. White
A
rdipithecus ramiduswas found in exposed sed-
iments flanking the Awash River, Ethiopia. The
local geology and associated fossils provide
critical information about its age and habitat.
Most of Africa’s surface is nondepositional and/or
covered by forests. This explains why so many discov-
eries related to early hominid evolution have been
made within eastern Africa’s relatively dry, narrow,
active rift system. Here the Arabian and African tec-
tonic plates have been pulling apart for millions of
years, and lakes and rivers have accumulated variably
fossil-rich sediments in the Afar Triangle, which lies
at the intersection of the Red Sea, Gulf of Aden, and
Main Ethiopian Rifts (see map). Some of these deposits
were subsequently uplifted by the rift tectonics and are
now eroding. In addition, volcanoes associated with
this rifting have left many widespread deposits that we
can use to determine the age of these fossils using
modern radioisotopic methods.
Several of the most important hominid fossils have been found near
the Afar’s western margin, north and west of the Awash River (star on
map), including Hadar (the “Lucy” site), Gona [known for the world’s
oldest stone tools at 2.6 million years ago (Ma)], and the Middle Awash
(including Aramis). Cumulatively, these and nearby study areas in
Ethiopia have provided an unparalleled record of hominid evolution.
Fossil-bearing rocks in the Middle Awash are intermittently
exposed and measure more than 1 km in thickness. Volcanic rocks
near the base of this regional succession are dated to more than 6 Ma.
Its uppermost sediments document the appearance of anatomically
near-modern humans 155,000 years ago. As is the case for many river
and lake deposits, fossil accumulation rates here have been highly
variable, and the distribution and preservation of the fossils are
uneven. Alterations of the fossils caused by erosion and other factors
further complicate interpretation of past environments. To meet this
challenge, beginning in 1981, our research team of more than 70 sci-
entists has collected 2000 geological samples, thousands of lithic
artifacts (e.g., stone tools), and tens of thousands of plant and animal
fossils. The emergent picture developed from the many Middle
Awash rock units and their contents represents a series of snapshots
taken through time, rather than a continuous record of deposition.
Ar. ramiduswas recovered from one such geological unit, 3 to 6 m
thick, centered within the study area. Here, the Aramis and adjacent
drainage basins expose a total thickness of 300 m of sediments largely
deposited in rivers and lakes, and on floodplains, between ~5.5 and
3.8 Ma. Within this succession, the Ar. ramidus–bearing rock unit
comprises silt and clay beds deposited on a floodplain. It is bracketed
between two key volcanic markers, each dated to 4.4 Ma. Their simi-
lar ages and sedimentology imply that the fossils themselves date to
4.4 Ma and were all deposited within a relatively narrow time interval
lasting anywhere from 100 to 10,000 years. Today the unit is exposed
across a 9-km arc that represents a fortuitous transect through the
ancient landscape. The western exposure, in particular, preserves a
rich assemblage of plant and animal fossils and ancient soils.
Fossilized wood, seeds, and phytoliths (hard silica parts from
plants) confirm the presence of hackberry, fig, and palm trees. There
is no evidence of a humid closed-canopy tropical rainforest, nor of
the subdesertic vegetation that characterizes the area today.
Invertebrate fossils are abundant and include insect larvae, brood-
balls and nests of dung beetles, diverse gastropods, and millipedes.
The terrestrial gastropods best match those seen in modern ground-
water forests such as the Kibwezi in Kenya. Aquatic lower verte-
brates are relatively rare and probably arrived episodically during
flooding of a river distal to the Aramis area. The most abundant fish
is catfish, probably introduced during overbank flooding and/or by
predatory birds roosting in local trees.
Our combined evidence indicates that Ar. ramidusdid not live in
the open savanna that was once envisioned to be the predominant
habitat of the earliest hominids, but rather in an environment that
was humid and cooler than it is today, containing habitats ranging
from woodland to forest patches.
Map showing the Middle Awash area (star) and rift locations (red lines). Photo shows the
4.4 Ma volcanic marker horizon (yellow bed) atop the locality where the skeleton and holo-
type teeth of Ar. ramiduswere discovered. Also shown are some of the fossil seeds.
See pages 62–63 for authors’ affiliations.
When citing, please refer to the full paper, available at DOI 10.1126/science.1175817.
Authors’Summaries
See pages 6–7 for authors’ affiliations.

20. 20 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org66
AUTHORS’SUMMARIES
Taphonomic, Avian, and Small-Vertebrate
Indicators of Ardipithecus ramidusHabitat
Antoine Louchart, Henry Wesselman, Robert J. Blumenschine, Leslea J. Hlusko, Jackson K. Njau,
Michael T. Black, Mesfin Asnake, Tim D. White
T
he stratigraphic unit con-
taining Ardipithecus rami-
duswas probably deposited
rapidly, thus providing a transect
through a 4.4-million-year-old
landscape. To help reconstruct
and understand its biological
setting as thoroughly as possible,
we recovered an assemblage of
>150,000 plant and animal fossils.
More than 6000 vertebrate speci-
mens were identified at the family
level or below. These specimens
represent animals ranging in size
from shrews to elephants and
include abundant birds and small
mammals that are usually rare in
hominid-bearing assemblages. Many of these birds and small mam-
mals are highly sensitive to environmental conditions and thus are par-
ticularly helpful in reconstructing the environment.
Accurate interpretation of fossil assemblages can be challenging.
Even fossils from one layer can represent artificial amalgamations
that might have originated thousands of years apart. Moreover, the
remains of animals living in different habitats can be artificially
mixed by flowing water or by shifting lake and river margins.
Ecological fidelity can be further biased by unsystematic recovery if,
for example, only the more complete, identifiable, or rare specimens
are collected. Thus, interpreting the Ardipithecus-bearing sediments
requires that we deduce the physical and biological conditions under
which the fossils accumulated and the degree to which these biases
operated at the time of deposition—a practice called “taphonomy.”
Both the large- and small-mammal assemblages at Aramis lack
the damage that would result from transport and sorting by water, a
finding consistent with the fine-grained sediments in which the
bones were originally embedded. Many of the limb bone fragments
of large mammals show traces of rodent gnawing and carnivore
chewing at a time when the bones were still fresh. These bones were
most probably damaged by hyenas, which in modern times are known
to destroy most of the limb bones and consume their marrow. The
actions of hyenas and other carnivores that actively competed for
these remains largely explain why the fossil assemblage at Aramis
contains an overrepresentation of teeth, jaws, and limb bone shaft
splinters (versus skulls or limb bone ends).
As a result of this bone destruction, whole skeletons are extremely
rare at Aramis, with one fortunate exception: the partial skeleton of
Ar. ramidusexcavated at ARA-VP-6/500. The relative abundance and
damage patterns of the fossils representing small mammals and birds
suggest that they are derived from undigested material regurgitated
by owls (owl pellets). Because of their fragility and size, bird bones
have been rare or absent at most other eastern African fossil assem-
blages that included early hominids. However, we cataloged 370
avian fossils; these represent 29 species, several new to science. Most
of the birds are terrestrial rather than aquatic, and small species such
as doves, lovebirds, mousebirds, passerines, and swifts are abundant.
Open-country species are rare. Eagles and hawks/kites are present,
but the assemblage is dominated by parrots and the peafowl Pavo, an
ecological indicator of wooded conditions.
The small-mammal assemblage includes up to 20 new species,
including shrews, bats, rodents, hares, and carnivores. Extant coun-
terparts live in a variety of habitats, but their relative abundance in
the fossil assemblage indicates that Ardipithecuslived in a wooded
area. Avian predators most probably procured the much rarer squir-
rels and gerbils from drier scrub or arid settings at a distance. Most
of the bat, shrew, porcupine, and other rodent specimens are compat-
ible with a relatively moist environmental setting, as are the abun-
dant fossils of monkeys and spiral-horned antelopes.
The combination of geological and taphonomic evidence, the
assemblage of small-mammal and avian fossils, and the taxonomic
and isotopic compositions of remains from larger mammals indicate
that Aramis was predominantly a woodland habitat during Ar.
ramidustimes. The anatomical and isotopic evidence of Ar. ramidus
itself also suggests that the species was adapted to such a habitat.
Psittacidae
36% (22)
Pavo
15% (16)
Tyto
8% (7)
Francolinus
8% (10)
Passeriformes
6% (4)
Columbidae
6% (13)
Falconiformes
6% (7)
Numididae
5% (7)
Anatidae
3% (8)
Otididae 2% (3)
6 other taxa (1 each)
Bucorvus <1% (2)
Coturnix <1% (2)
Apodidae <1% (2)
Coliidae <1% (2)
See pages 62–63 for authors’ affiliations.
When citing, please refer to the full paper, available at DOI 10.1126/science.1175823.
Abundance of birds (left) associated with Ar. ramidus.
These distributions are consistent with a mostly wood-
land habitat. (Above) An example of the many small
mammal and bird bones.
Authors’Summaries
See pages 6–7 for authors’ affiliations.

21. 21www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 67
AUTHORS’SUMMARIES
Macrovertebrate Paleontology and the
Pliocene Habitat of Ardipithecus ramidus
Tim D. White, Stanley H. Ambrose, Gen Suwa, Denise F. Su, David DeGusta, Raymond L. Bernor,
Jean-Renaud Boisserie, Michel Brunet, Eric Delson, Stephen Frost, Nuria Garcia, Ioannis X. Giaourtsakis,
Yohannes Haile-Selassie, F. Clark Howell, Thomas Lehmann, Andossa Likius, Cesur Pehlevan, Haruo
Saegusa, Gina Semprebon, Mark Teaford, Elisabeth Vrba
E
ver since Darwin, scholars have
speculated about the role that
environment may have played in
human origins, evolution, and adapta-
tion. Given that all living great apes live
and feed in trees, it has been assumed
that the last common ancestor we
shared with these forms was also a for-
est dweller. In 1925, Raymond Dart
described the first Australopithecus, a
child’s skull, at Taung, South Africa.
Its occurrence among other fossils
indicative of a grassland environment
prompted speculation that the open
grasslands of Africa were exploited by
early hominids and were therefore
somehow integrally involved with the
origins of upright walking.
The Ardipithecus-bearing sediments
at Aramis now provide fresh evidence
that Ar. ramiduslived in a predomi-
nantly woodland setting. This and cor-
roborative evidence from fossil assem-
blages of avian and small mammals
imply that a grassland environment was
not a major force driving evolution of
the earliest hominids. A diverse assem-
blage of large mammals (>5 kg body
weight) collected alongside Ardipithe-
cusprovides further support for this
conclusion. Carbon isotopes from
tooth enamel yield dietary information
because different isotope signatures reflect different photosynthetic
pathways of plants consumed during enamel development. Therefore,
animals that feed on tropical open-environment grasses (or on grass-eat-
ing animals) have different isotopic compositions from those feeding on
browse, seeds, or fruit from shrubs or trees. Moreover, oxygen isotopes
help deduce relative humidity and evaporation in the environment.
The larger-mammal assemblage associated with Ardipithecuswas
systematically collected across a ~9 km transect of eroding sediments
sandwiched between two volcanic horizons each dated to 4.4 million
years ago. It consists of ~4000 cataloged specimens assigned to
~40 species in 34 genera of 16 families.
There are only three primates in this assemblage, and the rarest is
Ardipithecus, represented by 110 specimens (a minimum of 36 individ-
uals). Conversely, colobine monkeys and a small baboon-like monkey
(red crosses in figure) account for
nearly a third of the entire large mam-
mal collection. Leaf-eating colobines
today exhibit strong preferences for
arboreal habitats, and the carbon iso-
tope compositions of the fossil teeth are
consistent with dense to open forest
arboreal feeding (see figure).
The other dominant large mammal
associated with Ar. ramidusis the
spiral-horned antelope, Tragelaphus
(the kudu, green circle). Today, these
antelopes are browsers (eating mostly
leaves), and they prefer bushy to
wooded habitats. The dental morphol-
ogy, wear, and enamel isotopic com-
position of the Aramis kudu species
are all consistent with such place-
ment. In contrast, grazing antelopes
(which eat mostly grass) are rare in
the Aramis assemblage.
The large-mammal assemblage
shows a preponderance of browsers
and fruit eaters. This evidence is con-
sistent with indications from birds,
small mammals, soil isotopes, plants,
and invertebrate remains. The emer-
gent picture of the Aramis landscape
during Ar. ramidustimes is one of a
woodland setting with small forest
patches. This woodland graded into
nearby habitats that were more open
and are devoid of fossils of Ardipithecusand other forest-to-woodland–
community mammals. Finally, the carbon isotopic composition of
Ar. ramidus teeth is similar to that of the predominantly arboreal, small,
baboon-like Pliopapioand the woodland browser Tragelaphus, indi-
cating little dietary intake of grass or grass-eating animals. It is there-
fore unlikely that Ar. ramiduswas feeding much in open grasslands.
These data suggest that the anatomy and behavior of early
hominids did not evolve in response to open savanna or mosaic set-
tings. Rather, hominids appear to have originated and persisted
within more closed, wooded habitats until the emergence of more
ecologically aggressive Australopithecus.
Carbon and oxygen isotope analyses of teeth from the Ar. ramidus
localities. Species are listed in order of abundance, and isotopic
data separate species by what they ate and their environment.
See pages 62–63 for authors’ affiliations.
When citing, please refer to the full paper, available at DOI 10.1126/science.1175822.
Authors’Summaries
See pages 6–7 for authors’ affiliations.

22. 22 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org68
AUTHORS’SUMMARIES
The Ardipithecus ramidusSkull and
Its Implications for Hominid Origins
Gen Suwa, Berhane Asfaw, Reiko T. Kono, Daisuke Kubo, C. Owen Lovejoy, Tim D. White
T
he key feature that distin-
guishes Homo sapiensfrom
other primates is our unusu-
ally large brain, which allows us
to communicate, make tools, plan,
and modify our environment. Un-
derstanding how and when our
cognitive ability evolved has been
a special focus in anthropology
and, more recently, genetics. Fossil
hominid skulls provide direct evi-
dence of skull evolution and infor-
mation about diet, appearance, and
behavior. Skulls feature promi-
nently in the characterization of
species, in taxonomy, and in phy-
logenetic analyses of both extinct
and living primates.
Unfortunately, hominid skulls
are relatively rare in the fossil
record. A number of partial skulls
and crania (skulls without a lower
jaw) of early Homoand its predecessor, Australopithecus(which lived
~1 to 4 million years ago), have been recovered, but relatively few are
complete enough for extensive comparisons. One surprisingly com-
plete but distorted cranium from 6 to 7 million years ago was discov-
ered in central Africa (Chad). This fossil, Sahelanthropus tchadensis
(a.k.a. “Toumaï ”), is thought by many to represent the earliest known
hominid, although some have argued that it is a female ape.
The Ardipithecus ramidusskull is of particular interest because it
predates known Australopithecusand thereby illuminates the early
evolution of the hominid skull, brain, and face. The Ar. ramidusskull
was badly crushed, and many of its bones were scattered over a wide
area. Because the bones were so fragile and damaged, we imaged
them with micro–computed tomography, making more than 5000
slices. We assembled the fragments into more than 60 key virtual
pieces of the braincase, face, and teeth, enough to allow us to digitally
reconstruct a largely complete cranium.
The fossil skulls of Australopithecusindicate that its brain was
~400 to 550 cm
3
in size, slightly larger than the brains of modern apes
of similar body size and about a third of those of typical Homo sapi-
ens. Its specialized craniofacial architecture facilitated the production
of strong chewing forces along the entire row of teeth located behind
its canines. These postcanine teeth were enlarged and had thick
enamel, consistent with a hard/tough and abrasive diet. Some species
exhibited extreme manifestations of this specialized chewing appara-
tus and are known as “robust”
Australopithecus.
Ar. ramidushad a small brain
(300 to 350 cm
3
), similar to that of
bonobos and female chimpanzees
and smaller than that of Australo-
pithecus. The Ar. ramidusface is
also small and lacks the large
cheeks of “heavy chewing” Aus-
tralopithecus. It has a projecting
muzzle as in Sahelanthropus,
which gives it a decidedly ape-like
gestalt. Yet the Ar. ramidusskull is
not particularly chimpanzee-like.
For example, the ridge above the
eye socket is unlike that of a chim-
panzee, and its lower face does
not project forward as much as a
chimpanzee’s face. Chimps pri-
marily eat ripe fruits and have
large incisors set in a projecting
lower face. Ar. ramidusinstead
was probably more omnivorous and fed both in trees and on the ground.
Additionally, in chimpanzees, forward placement of the entire lower
face is exaggerated, perhaps linked with their large tusklike canines
(especially in males) and elevated levels of aggression. This is not seen
in Ar. ramidus, implying that it was less socially aggressive.
Like Ar. ramidus, S. tchadensishad a brain that was less than
400 cm
3
in size. It also resembled Ar. ramidusin having small non-
sharpened canines. Details of the bottom of the skull show that both
Ar. ramidusand Sahelanthropushad a short cranial base, a feature
also shared with Australopithecus. Furthermore, we infer that the rear
of the Ar. ramidusskull was downturned like that suggested for
Sahelanthropus. These similarities confirm that Sahelanthropuswas
indeed a hominid, not an extinct ape.
These and an additional feature of the skull hint that, despite its
small size, the brain of Ar. ramidusmay have already begun to develop
some aspects of later hominid-like form and function. The steep orien-
tation of the bone on which the brain stem rests suggests that the base
of the Ar. ramidusbrain might have been more flexed than in apes. In
Australopithecus, a flexed cranial base occurs together with expansion
of the posterior parietal cortex, a part of the modern human brain
involved in aspects of visual and spatial perception.
(Right) Oblique and side views of a female chimpanzee (right) and the Ar.
ramidusfemale reconstruction (left; the oblique view includes a separate
mandible). (Left) Comparison of brain and tooth sizes (arrows) of chimps (Pan ;
blue), Ar. ramidus(red), and Australopithecus(green). Means are plotted
except for individual Ar. ramidusand Au. afarensiscranial capacities. Canine
unworn heights (bottom) are based on small samples, Ar. ramidus(females, n
= 1; males, n= 3), Au. afarensis(n= 2), Pan(females, n= 19; males, n= 11).
200cc 300 400 500 600
14mm16 18 20 22 24
44mm 48 52 56 60
12mm 16 20 24
Australopithecus afarensis
Ardipithecus ramidus
Pan troglodytes
See pages 62–63 for authors’ affiliations.
When citing, please refer to the full paper, available at DOI 10.1126/science.1175825.
Authors’Summaries
See pages 6–7 for authors’ affiliations.

23. 23www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 69
AUTHORS’SUMMARIES
Paleobiological Implications of the
Ardipithecus ramidusDentition
Gen Suwa, Reiko T. Kono, Scott W. Simpson, Berhane Asfaw, C. Owen Lovejoy, Tim D. White
T
eeth are highly resilient to degradation and
therefore are the most abundant specimens in
the primate fossil record. The size, shape,
enamel thickness, and isotopic composition of teeth
provide a wealth of information about phylogeny, diet,
and social behavior. Ardipithecus ramiduswas origi-
nally defined in 1994 primarily on the basis of recov-
ered teeth, but the sample size was small, limiting
comparison to other primate fossils. We now have over
145 teeth, including canines from up to 21 individuals.
The expanded sample now provides new information
regarding Ar. ramidusand, using comparisons with
teeth of other hominids, extant apes, and monkeys,
new perspectives on early hominid evolution as well.
In apes and monkeys, the male’s upper canine tooth
usually bears a projecting, daggerlike crown that is
continuously sharpened (honed) by wear against a
specialized lower premolar tooth (together these form
the C/P
3
complex). The canine tooth is used as a slic-
ing weapon in intra- and intergroup social conflicts.
Modern humans have small, stublike canines which
function more like incisors.
All known modern and fossil apes have (or had) a honing C/P
3
com-
plex. In most species, this is more developed in males than females (in
a few species, females have male-like large canines, either for territo-
rial defense or for specialized feeding). The relatively large number of
Ar. ramidusteeth, in combination with Ethiopian Ar. kadabba, Kenyan
Orrorin, and Chadian Sahelanthropus[currently the earliest known
hominids at about 6 million years ago (Ma)], provide insight into the
ancestral ape C/P
3
complex and its evolution in early hominids.
In basal dimensions, the canines of Ar. ramidusare roughly as
large as those of female chimpanzees and male bonobos, but their
crown heights are shorter (see figure). The Ar. ramidussample is now
large enough to assure us that males are represented. This means that
male and female canines were not only similar in size, but that the
male canine had been dramatically “feminized” in shape. The crown
of the upper canine in Ar. ramiduswas altered from the pointed shape
seen in apes to a less-threatening diamond shape in both males and
females. There is no evidence of honing. The lower canines of Ar.
ramidusare less modified from the inferred female ape condition
than the uppers. The hominid canines from about 6 Ma are similar in
size to those of Ar. ramidus, but (especially) the older upper canines
appear slightly more primitive. This suggests that male canine size
and prominence were dramatically reduced by ~6 to 4.4 Ma from an
ancestral ape with a honing C/P
3
complex and a moderate degree of
male and female canine size difference.
In modern monkeys and apes, the upper canine is important in
male agonistic behavior, so its subdued shape in early hominids and
Ar. ramidussuggests that sexual selection played a primary role in
canine reduction. Thus, fundamental reproductive and social behav-
ioral changes probably occurred in hominids long before they had
enlarged brains and began to use stone tools.
Thick enamel suggests that an animal’s food intake was abrasive;
for example, from terrestrial feeding. Thin enamel is consistent with
a diet of softer and less abrasive foods, such as arboreal ripe fruits. We
measured the enamel properties of more than 30 Ar. ramidusteeth.
Its molar enamel is intermediate in thickness between that of chim-
panzees and Australopithecusor Homo. Chimpanzees have thin
enamel at the chewing surface of their molars, whereas a broad con-
cave basin flanked by spiky cusps facilitates crushing fruits and
shredding leaves. Ar. ramidusdoes not share this pattern, implying a
diet different from that of chimpanzees. Lack of thick enamel indi-
cates that Ar. ramiduswas not as adapted to heavy chewing and/or
eating abrasive foods as were later Australopithecusor even Homo.
The combined evidence from the isotopic content of the enamel, den-
tal wear, and molar structure indicates that the earliest hominid diet
was one of generalized omnivory and frugivory and therefore dif-
fered from that of Australopithecusand living African apes.
Dentitions from human (left), Ar. ramidus(middle), and chimpanzee (right), all males.
Below are corresponding samples of the maxillary first molar in each. Red, thicker enamel
(~2 mm); blue, thinner enamel (~0.5 mm). Contour lines map the topography of the crown
and chewing surfaces.
See pages 62–63 for authors’ affiliations.
When citing, please refer to the full paper, available at DOI 10.1126/science.1175824.
Authors’Summaries
See pages 6–7 for authors’ affiliations.

24. 24 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org70
AUTHORS’SUMMARIES
Careful Climbing in the Miocene:
The Forelimbs of Ardipithecus ramidus
and Humans Are Primitive
C. Owen Lovejoy, Scott W. Simpson, Tim D. White, Berhane Asfaw, Gen Suwa
A
grasping hand and highly
mobile forelimb are defining
characteristics of primates.
The special ability to pick things up
and manipulate them has probably
been a central selective force in mak-
ing primates so unusually intelligent.
It’s something that porpoises can’t do
at all and crows can’t do very well. It
may also be one reason why humans
alone eventually evolved cognition.
The hands of African apes are
specialized in a number of ways that
make them dramatically different
from our own. Apes must support
their large body mass during climb-
ing to feed and nest, especially in the
middle and higher parts of the tree
canopy. Their hands must therefore
withstand very high forces, and this
is facilitated by their elongated palms
and fingers. Our palms are much
shorter and our wrists more mobile.
This allows us to grasp objects and
compress them with great dexterity
and force—something often called a “power grip.” The differences
between ape and human forelimbs become less pronounced going
from the hand to the shoulder. Ape and human elbow joints, for exam-
ple, diverge only moderately in their manner of load transmission.
The high loads that apes bear during locomotion have required
them to greatly stiffen the joints between their fingers and palms.
Because their thumb has not been elongated in the same way as their
palms and fingers have, thumb-to-palm and thumb-to-finger opposi-
tions are more awkward for them. We are therefore much more adept
at making and using tools. All of these forelimb characteristics in apes
have led them to adopt an unusual form of terrestrial quadrupedality,
in which they support themselves on their knuckles rather than on
their palms. Only African apes exhibit this “knuckle-walking.” Other
primates, such as monkeys, still support themselves on their palms.
It has long been assumed that our hands must have evolved from
hands like those of African apes. When they are knuckle-walking,
their long forelimbs angle their trunks upward. This posture has
therefore long been viewed by some as “preadapting” our ancestors
to holding their trunks upright.
Until now, this argument was unsettled, because we lacked an ade-
quate fossil record. Even Lucy, the most complete Australopithecus
skeleton yet found, had only two
hand bones—far short of the number
needed to interpret the structure and
evolution of the hand. The Ardipith-
ecusskeleton reported here changes
that. Not only is it more than 1 mil-
lion years older than Lucy (4.4 mil-
lion versus 3.2 million years old), its
hands are virtually complete and
intact. They show that Ardipithecus
did not knuckle-walk like African
apes and that it lacked virtually all of
the specializations that protect great
ape hands from injury while they
climb and feed in trees.
Ardipithecushands were very
different from those of African apes.
Its wrist joints were not as stiff as
those of apes, and the joints between
their palms and fingers were much
more flexible. Moreover, a large
joint in the middle of the wrist (the
midcarpal joint) was especially
flexible, being even more mobile
than our own. This would have
allowed Ardipithecusto support nearly all of its body weight on its
palms when moving along tree branches, so that it could move
well forward of a supporting forelimb without first releasing its
grip on a branch.
This discovery ends years of speculation about the course of
human evolution. Our ancestors’ hands differed profoundly from
those of living great apes, and therefore the two must have substan-
tially differed in the ways they climbed, fed, and nested. It is African
apes who have evolved so extensively since we shared our last com-
mon ancestor, not humans or our immediate hominid ancestors.
Hands of the earliest hominids were less ape-like than ours and quite
different from those of any living form.
Ardipithecusalso shows that our ability to use and make tools did
not require us to greatly modify our hands. Rather, human grasp and
dexterity were long ago inherited almost directly from our last com-
mon ancestor with chimpanzees. We now know that our earliest
ancestors only had to slightly enlarge their thumbs and shorten their
fingers to greatly improve their dexterity for tool-using.
Two views of the left hand of Ar. ramidusshowing primitive features
absent in specialized apes. (A) Short metacarpals; (B) lack of knuckle-
walking grooves; (C) extended joint surface on fifth digit; (D) thumb
more robust than in apes; (E) insertion gable for long flexor tendon
(sometimes absent in apes); (F) hamate allows palm to flex; (G) sim-
ple wrist joints; (H) capitate head promotes strong palm flexion. Inset:
lateral view of capitates of Pan, Ar. ramidus, and human (left to right).
Dashed lines reflect a more palmar capitate head location for Ar.
ramidusand humans, which allows a more flexible wrist in hominids.
A
B
C
D
E
F
G
H
See pages 62–63 for authors’ affiliations.
When citing, please refer to the full paper, available at DOI 10.1126/science.1175827.
Authors’Summaries
See pages 6–7 for authors’ affiliations.

25. 25www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 71
AUTHORS’SUMMARIES
The Pelvis and Femur
of Ardipithecus ramidus:
The Emergence of Upright Walking
C. Owen Lovejoy, Gen Suwa, Linda Spurlock, Berhane Asfaw, Tim D. White
V
irtually no other primate has a human-like
pelvic girdle—not even our closest living rela-
tives, the chimpanzee and bonobo. Such
uniqueness evolved via substantial modifications of a
pelvis more originally suited for life in trees. This
arboreal primate heritage has left us rather ungainly.
Our legs are massive because they continue to house
almost all of the muscles originally required for climb-
ing. Our hamstrings, the large muscles in our posterior
thighs, must decelerate the swinging limb with each
step, and when we run, the limb’s inertia is sometimes
too great and these muscles fail (not something one
would want to happen on a savanna).
Furthermore, when each limb leaves the ground to
be swung forward, it and the pelvis are unsupported
and would slump toward the ground were it not for
muscles acting on the opposite side of the body (the
anterior gluteals). One early anthropologist described
human locomotion as a process by which we alter-
nately almost fall on our faces. Chimpanzees and
other primates cannot prevent such slumping when
walking upright because they cannot reposition these
muscles effectively. Their spine is too inflexible and
their ilia—the large pelvic bones to which the gluteals attach—are
positioned and shaped differently than ours. Modifying a typical
chimp or gorilla pelvis to facilitate upright walking would require
extensive structural changes.
Until now, the fossil record has told us little about when and how
the early hominid pelvis evolved. Even 3 to 4 million years ago (when
our brains were still only slightly larger than those of chimpanzees), it
had already undergone radical transformation. One of the oldest
hominid pelves, that of Australopithecus afarensis(A.L. 288-1;
“Lucy”), shows that her species had already evolved virtually all of the
fundamental adaptations to bipedality. Even the kinetics of her hip
joint were similar to ours. Although the human pelvis was later further
reshaped, this was largely the result of our much enlarged birth canal.
Ardipithecus ramidusnow unveils how our skeleton became pro-
gressively modified for bipedality. Although the foot anatomy of Ar.
ramidusshows that it was still climbing trees, on the ground it walked
upright. Its pelvis is a mosaic that, although far from being chim-
panzee-like, is still much more primitive than that of Australopithecus.
The gluteal muscles had been repositioned so that Ar. ramidus
could walk without shifting its center of mass from side to side. This
is made clear not only by the shape of its ilium, but by the appearance
of a special growth site unique to hominids among all primates (the
anterior inferior iliac spine). However, its lower pelvis was still
almost entirely ape-like, presumably because it still had massive
hindlimb muscles for active climbing.
Changes made in the upper pelvis rendered Ar. ramidus an effec-
tive upright walker. It could also run, but probably with less speed and
efficiency than humans. Running would also have exposed it to
injury because it lacked advanced mechanisms such as those that
would allow it to decelerate its limbs or modulate collision forces at
its heel. Australopithecus, which had given up its grasping foot and
abandoned active climbing, had evolved a lower pelvis that allowed it
to run and walk for considerable distances.
Ar. ramidusthus illuminates two critical adaptive transitions in
human evolution. In the first, from the human-chimp last common
ancestor to Ardipithecus, modifications produced a mosaic pelvis
that was useful for both climbing and upright walking. In the second,
from Ardipithecusto Australopithecus, modifications produced a
pelvis and lower limb that facilitated more effective upright walking
and running but that were no longer useful for climbing. Because
climbing to feed, nest, and escape predators is vital to all nonhuman
primates, both of these transitions would likely have been a response
to intense natural selection.
The Ar. ramiduspelvis has a mosaic of characters for both bipedality and climbing. Left to right:
Human, Au. afarensis(“Lucy”), Ar. ramidus, Pan (chimpanzee). The ischial surface is angled
near its midpoint to face upward in Lucy and the human (blue double arrows), showing that
their hamstrings have undergone transformation for advanced bipedality, whereas they are
primitive in the chimpanzee and Ar. ramidus(blue arrows). All three hominid ilia are vertically
short and horizontally broad, forming a greater sciatic notch (white arrows) that is absent in
Pan. A novel growth site [the anterior inferior iliac spine (yellow arrows)] is also lacking in Pan.
Au. afarensisHomo sapiens Ar. ramidus P. troglodytes
See pages 62–63 for authors’ affiliations.
When citing, please refer to the full paper, available at DOI 10.1126/science.1175831.
Authors’Summaries
See pages 6–7 for authors’ affiliations.

26. 26 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org72
AUTHORS’SUMMARIES
Combining Prehension and Propulsion:
The Foot of Ardipithecus ramidus
C. Owen Lovejoy, Bruce Latimer, Gen Suwa, Berhane Asfaw, Tim D. White
T
he special foot adaptations that
enable humans to walk upright
and run are central to under-
standing our evolution. Until the dis-
covery of Ardipithecus ramidus, it was
generally thought that our foot evolved
from one similar to that of modern
African apes. Apes have feet that are
modified to support their large bodies
and to facilitate vertical climbing, thus
allowing them to feed, nest, and seek
safety in trees. Our foot differs from
theirs in myriad ways, and its evolu-
tion from theirs would consequently
have required an extensive series of
structural changes. Some mid–20th-
century comparative anatomists were
so impressed with the profound differ-
ences between human and extant ape
feet that they postulated a deep, pre-
ape origin for hominids.
Ar. ramidusbrings a new perspec-
tive to this old controversy. Its foot
turns out to be unlike those of the
African apes in many ways. The par-
tial skeleton of Ar. ramiduspreserves
most of the foot and includes a special
bone called the os peroneum that is
critical for understanding foot evolu-
tion. This bone, which is embedded
within a tendon, facilitates the mechanical action of the fibularis
longus, the primary muscle that draws in the big toe when the foot is
grasping. Until now, we knew little about this bone’s natural history,
except that it is present in Old World monkeys and gibbons but gen-
erally not in our more recent ape relatives. Monkeys are very accom-
plished at leaping between trees. They must keep their feet fairly rigid
during takeoff when they hurl themselves across gaps in the tree
canopy; otherwise, much of the torque from their foot muscles would
be dissipated within the foot rather than being transferred to the tree.
The African apes are too large to do much leaping. They have
therefore given up the features that maintain a rigid foot and have
instead modified theirs for more effective grasping—almost to the
point of making it difficult to distinguish their feet from their hands.
Indeed, very early anatomists argued that the “quadrumanus” apes
were not related to humans because of their hand-like feet. Extant
apes lack the os peroneum, and their fibularis tendon, which draws
the great toe closed during grasping,
has been relocated more toward the
front of the foot. This makes the ten-
don run more parallel to other joints
that cross the midfoot, and allows
apes to grasp with great power with-
out stiffening these other, flexible
joints. Apes can thus both powerfully
grasp and mold their feet around
objects at the same time. However,
their feet have become less effective
as levers, making them far less useful
in terrestrial propulsion.
The foot of Ar. ramidusshows that
none of these ape-like changes were
present in the last common ancestor
of African apes and humans. That
ancestor, which until now has been
thought to be chimpanzee-like, must
have had a more monkey-like foot.
Not only did it still have an os per-
oneum, it must also have had all of the
other characteristics associated with
it (subsequently abandoned in chim-
panzees and gorillas). We infer this
because humans still have these char-
acteristics, so we must have retained
them from our last common ancestor.
The mid–20th-century anatomists
were correct to worry about the human
foot as they did: Ours turns out to have evolved in one direction,
while those of African apes were evolving in quite another.
One of the great advantages of our more rigid foot is that it works
much better as a lever during upright walking and running (as it also
does in monkeys). However, Ar. ramidusstill had an opposable big
toe, unlike any later hominid. Its ability to walk upright was there-
fore comparatively primitive. Because it had substantially modified
the other four toes for upright walking, even while retaining its
grasping big toe, the Ardipithecusfoot was an odd mosaic that
worked for both upright walking and climbing in trees. If our last
common ancestor with the chimpanzee had not retained such an
unspecialized foot, perhaps upright walking might never have
evolved in the first place.
Foot skeleton of Ar. ramidus(bottom; reconstruction based on
computed tomography rendering shown) lacked many features
that have evolved for advanced vertical climbing and suspension
in extant chimpanzees (Pan , top left). Chimpanzees have a highly
flexible midfoot and other adaptations that improve their ability
to grasp substrates. These are absent in Ar. ramidus.
CREDIT: RECONSTRUCTION, COPYRIGHT J. H. MATTERNES; CHIMPANZEE CLIMBING, J. DESILVA; BONOBO AND HUMAN FEET, S. INGHAM CREDITS (TOP TO BOTTOM):
Pan Homo
Ardipithecus ramidus
See pages 62–63 for authors’ affiliations.
When citing, please refer to the full paper, available at DOI 10.1126/science.1175832.
Authors’Summaries
See pages 6–7 for authors’ affiliations.

27. 27www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 73
CREDIT: ILLUSTRATION OF AR. RAMIDUS : COPYRIGHT J. H. MATTERNES
AUTHORS’SUMMARIES
The Great Divides: Ardipithecus ramidus
Reveals the Postcrania of Our Last Common
Ancestors with African Apes
C. Owen Lovejoy, Gen Suwa, Scott W. Simpson, Jay H. Matternes, Tim D. White
E
volutionary biologists have long recognized
that the living primates most similar to humans
are the great apes, and comparative genomic
sequence analyses confirm that we are most closely
related to chimpanzees and bonobos (genus Pan).
Because of our great genomic similarity (sometimes
even cited as ~99%), the presumption that we evolved
from a chimpanzee-like ancestor has become increas-
ingly common wisdom. The widely held view that the
genomic and phyletic split between Panand humans
was as recent as 5 to 6 million years ago also fuels the
often uncritical acceptance of a Pan-like last common
ancestor. Ardipithecus ramidusat 4.4 million years
ago provides the first substantial body of fossil evi-
dence that temporally and anatomically extends our
knowledge of what the last common ancestor we
shared with chimpanzees was like, and therefore
allows a test of such presumptions.
Until now, Australopithecus afarensis, which lived
3 to 4 million years ago, represented the most primi-
tive well-known stage of human evolution. It had a
brain only slightly larger than that of chimpanzees,
and a snout that projected more than in later
hominids. Assuming some variant of a chimpanzee-
like ape ancestry, the bipedality of Au. afarensishas
been widely interpreted as being so primitive that it probably
could not have extended either its hip or knee joints and was a
clumsy upright walker. Some researchers have even postulated that
Au. afarensiscould walk but not run, or vice versa. Still others have
suggested that Au. afarensishad a grasping ape-like foot. Similarly,
it has been suggested that Au. afarensishad forelimbs that were ape-
like, including long, curved fingers used to forage daily in the arboreal
canopy, and that its immediate ancestors must have knuckle-
walked. Australopithecusmales were noticeably larger than females,
and this has often been interpreted as signifying a single-male,
polygynous, Gorilla-like mating system. Unlike gorillas, it has
diminutive canines, but these were argued to be a consequence of its
huge postcanine teeth. Early hominids have even been posited to
have possibly interbred with chimpanzees until just before the
appearance of Australopithecusin the fossil record.
The Ar. ramidusfossils and information on its habitat now reveal
that many of these earlier hypotheses about our last common ances-
tor with chimpanzees are incorrect. The picture emerging from Ar.
ramidusis that this last common ancestor had limb proportions more
like those of monkeys than apes. Its feet functioned only partly like
those of apes and much more like those of living monkeys and early
apes such as Proconsul(which lived more than 15 million years
ago). Its lower back was mobile and probably had six lumbar verte-
brae rather than the three to four seen in the stiff backs of African
apes. Its hand was unpredictably unique: Not only was its thumb
musculature robust, unlike that of an ape, but its midcarpal joint (in
the wrist) allowed the wrist to bend backward to a great degree,
enhancing its ability to move along tree branches on its palms. None
of the changes that apes have evolved to stiffen their hands for sus-
pension and vertical climbing were present, so its locomotion did
not resemble that of any living ape.
The hominid descendant of the last common ancestor we shared
with chimpanzees (the CLCA), Ardipithecus, became a biped by
modifying its upper pelvis without abandoning its grasping big toe.
It was therefore an unpredicted and odd mosaic. It appears, unlike
Au. afarensis, to have occupied the basal adaptive plateau of
hominid natural history. It is so rife with anatomical surprises that no
one could have imagined it without direct fossil evidence.
Cladogram adding Ar. ramidusto images of gorilla, chimpanzee, and human, taken from the
frontispiece of Evidence as to Man’s Place in Nature, by Thomas H. Huxley (London, 1863)
(with the positions of Gorillaand Panreversed to reflect current genetic data). Numerous
details of the Ar. ramidusskeleton confirm that extant African apes do not much resemble our
last common ancestor(s) with them.
Ardipithecus

Pan
 
 


   

  

  
 
  
  
   
  
   
 

   
   
    
   
  
    


 
 
 

   
 


  
 

   
   
    
    
      

See pages 62–63 for authors’ affiliations.
When citing, please refer to the full paper, available at DOI 10.1126/science.1175833.
Authors’Summaries
See pages 6–7 for authors’ affiliations.

28. 28 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org74
AUTHORS’SUMMARIES
CREDIT: ILLUSTRATION OF AR. RAMIDUS : COPYRIGHT J. H. MATTERNES
Reexamining Human Origins in Light
of Ardipithecus ramidus
C. Owen Lovejoy
C
himpanzees, bonobos, and
gorillas are our closest living
relatives. The most popular
reconstructions of human evolution
during the past century rested on the
presumption that the behaviors of the
earliest hominids were related to (or
even natural amplifications of) behav-
iors observed in these living great apes.
One effect of chimpanzee-centric
models of human evolution has been a
tendency to view Australopithecusas
transitional between an ape-like ances-
tor and early Homo.
Ardipithecus ramidusnullifies these
presumptions, as it shows that the
anatomy of living African apes is not
primitive but instead has evolved
specifically within extant ape lineages.
The anatomy and behavior of early
hominids are therefore unlikely to rep-
resent simple amplifications of those
shared with modern apes. Instead, Ar.
ramiduspreserves some of the ances-
tral characteristics of the last common
ancestor with much greater fidelity than do living African apes. Two
obvious exceptions are its ability to walk upright and the absence of
the large projecting canine tooth in males, derived features that
Ardipithecusshares with all later hominids.
Ar. ramidusilluminates our own origins because it clarifies our rela-
tionship to Australopithecus. For example, the enlarged rear teeth of
Australopithecushave long been viewed as adaptations to a rough,
abrasive diet. This has led to speculation that canine teeth might have
become smaller simply to accommodate the emergence of these other
enlarged teeth, or that the importance of canine teeth in displays of
male-to-male aggression waned with the development of weapons.
Ar. ramidusnegates such hypotheses because it demonstrates that small
canines occurred in hominids long before any of the dental modifica-
tions of Australopithecusor the use of stone tools. The loss of large
canine teeth in males must have occurred within the context of a gener-
alized, nonspecialized diet. Comparisons of the Ar. ramidusdentition
with those of all other higher primates indicate that the species retained
virtually no anatomical correlates of male-to-male conflict. Consistent
with a diminished role of such agonism, the body size of Ar. ramidus
males was only slightly larger than that of females.
The discovery of Ar. ramidusalso requires rejection of theories that
presume a chimpanzee- or gorilla-like
ancestor to explain habitual upright
walking. Ar. ramiduswas fully capable
of bipedality and had evolved a sub-
stantially modified pelvis and foot with
which to walk upright. At the same
time, it preserved the ability to maneu-
ver in trees, because it maintained a
grasping big toe and a powerful hip and
thigh musculature. Because upright
walking provided no energy advantage
for Ar. ramidus(it lacked many of the
adaptations evolved in later hominids
such as Australopithecus), reproduc-
tive success must have been central to
its evolution in early hominids.
Loss of the projecting canine raises
other vexing questions because this
tooth is so fundamental to reproduc-
tive success in higher primates. What
could cause males to forfeit their abil-
ity to aggressively compete with other
males? What changes paved the way
for the later emergence of the energy-
thirsty brain of Homo? Such questions
can no longer be addressed by simply comparing humans to extant
apes, because no ape exhibits an even remotely similar evolutionary
trajectory to that revealed by Ardipithecus.
When the likely adaptations of early hominids are viewed generally
rather than with specific reference to living chimpanzees, answers to
such questions arise naturally. Many odd hominid characteristics
become transformed from peculiar to commonplace. Combining our
knowledge of mammalian reproductive physiology and the hominid
fossil record suggests that a major shift in life-history strategy trans-
formed the social structure of early hominids. That shift probably
reduced male-to-male conflict and combined three previously unseen
behaviors associated with their ability to exploit both trees and the land
surface: (i) regular food-carrying, (ii) pair-bonding, and (iii) reproduc-
tive crypsis (in which females did not advertise ovulation, unlike the
case in chimpanzees). Together, these behaviors would have substan-
tially intensified male parental investment—a breakthrough adaptation
with anatomical, behavioral, and physiological consequences for early
hominids and for all of their descendants, including ourselves.
Breakthrough adaptations can transform life-history by deviating
from typical reproductive strategy. Early hominids show feminized
male canines [left] and primitive bipedality [right]. These suggest
that females preferred nonaggressive males who gained repro-
ductive success by obtaining copulation in exchange for valuable
foods (vested provisioning). Success would depend on copulatory
frequency with mates whose fertility remained cryptic (e.g.,
absence of cycling in mammary size). The result would be reduced
agonism in unrelated females, and cooperative expansion of day
ranges among equally cooperative males, eventually leading to
exploitation of new habitats.
1 cm
Pan
Ardipithecus
Reduced Intra-sexual
Agonism and Increased
Social Adhesion
BIPEDALITY
LOSS OF
HONING CANINE
OVULATORY
CRYPSIS
VESTED
PROVISIONING
See pages 62–63 for authors’ affiliation.
When citing, please refer to the full paper, available at DOI 10.1126/science.1175834.
Authors’Summaries
See pages 6–7 for authors’ affiliations.

29. 29www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Ardipithecus ramidusand the
Paleobiology of Early Hominids
Tim D. White,
1
*Berhane Asfaw,
2
Yonas Beyene,
3
Yohannes Haile-Selassie,
4
C. Owen Lovejoy,
5
Gen Suwa,
6
Giday WoldeGabriel
7
Hominid fossils predating the emergence ofAustralopithecushave been sparse and fragmentary.
The evolution of our lineage after the last common ancestor we shared with chimpanzees has
therefore remained unclear.Ardipithecus ramidus, recovered in ecologically and temporally
resolved contexts in Ethiopia’s Afar Rift, now illuminates earlier hominid paleobiology and aspects
of extant African ape evolution. More than 110 specimens recovered from 4.4-million-year-old
sediments include a partial skeleton with much of the skull, hands, feet, limbs, and pelvis. This
hominid combined arboreal palmigrade clambering and careful climbing with a form of terrestrial
bipedality more primitive than that ofAustralopithecus.Ar. ramidushad a reduced canine/
premolar complex and a little-derived cranial morphology and consumed a predominantly C
3
plant–based diet (plants using the C3photosynthetic pathway). Its ecological habitat appears to
have been largely woodland-focused.Ar. ramiduslacks any characters typical of suspension,
vertical climbing, or knuckle-walking.Ar. ramidusindicates that despite the genetic similarities of
living humans and chimpanzees, the ancestor we last shared probably differed substantially from
any extant African ape. Hominids and extant African apes have each become highly specialized
through very different evolutionary pathways. This evidence also illuminates the origins of
orthogrady, bipedality, ecology, diet, and social behavior in earliest Hominidae and helps to define
the basal hominid adaptation, thereby accentuating the derived nature ofAustralopithecus.
I
n 1871, Charles Darwin concluded that
Africa was humanity’s most probable birth
continent [(1), chapter 7]. Anticipating a
skeptical reception of his placement ofHomo
sapiensas a terminal twig on the organic tree,
Darwin lamented the mostly missing fossil
record of early hominids (2). Following T. H.
Huxley, who had hoped that“the fossilized bones
of an Ape more anthropoid, or a Man more
pithecoid”might be found by“some unborn
paleontologist”[(3), p. 50], Darwin observed,
“Nor should it be forgotten that those regions
which are the most likely to afford remains
connecting man with some extinct ape-like crea-
ture, have not as yet been searched by geol-
ogists.”He warned that without fossil evidence,
it was“useless to speculate on this subject”[(1),
p. 199)].
Darwin and his contemporaries nonethe-
less sketched a scenario of how an apelike
ancestor might have evolved into humans. That
scenario easily accommodated fossil evidence
then restricted to European Neandertals and
Dryopithecus(a Miocene fossil ape). Javanese
Homo erectuswas found in the 1890s, followed
by AfricanAustralopithecusin the 1920s. By
the 1960s, successive grades of human evolution
were widely recognized.Australopithecuscom-
prised several Plio-Pleistocene small-brained
species with advanced bipedality. This grade
(adaptive plateau) is now widely recognized as
foundational to more derivedHomo.
Molecular studies subsequently and indepen-
dently confirmed Huxley’s anatomically based
phylogeny linking African apes and living hu-
mans (4). They also challenged age estimates of
a human/chimpanzee divergence, once common-
ly viewed as exceeding 14 million years ago
(Ma). The latter estimates were mostly based on
erroneous interpretations of dentognathic remains
of the Miocene fossil apeRamapithecus, com-
bined with the presumption that extant chimpan-
zees are adequate proxies for the last common
ancestor we shared with them (the CLCA).
The phylogenetic separation of the lineages
leading to chimpanzees and humans is now wide-
ly thought to have been far more recent. During
the 1970s, discovery and definition ofAustra-
lopithecus afarensisat Laetoli and Hadar extended
knowledge of hominid biology deep into the
Pliocene [to 3.7 Ma (5,6)]. The slightly earlier (3.9
to 4.2 Ma) chronospeciesAu. anamensiswas sub-
sequently recognized as another small-brained
biped with notably large postcanine teeth and
postcranial derivations shared with its apparent
daughter species (7,8). Late Miocene hominid
fossils have been recently recovered from Ethiopia,
Kenya, and Chad. These have been placed in
three genera [Ardipithecus(9–12),Orrorin(13),
andSahelanthropus(14)]. They may represent
only one genus (12,15), and they challenge both
savanna- and chimpanzee-based models (16) of
hominid origins.
Continuing to build on fossil-free expecta-
tions traceable to Darwinian roots, some hold that
our last common ancestors with African apes
were anatomically and behaviorally chimpanzee-
like (17), that extant chimpanzees can be used as
“time machines”(18), and/or that unique features
ofGorillaare merely allometric modifications to
accommodate its great body mass. Thus, early
Australopithecushas routinely been interpreted
as“transitional” and/or a“locomotor missing
link”(19,20) between extant humans and chim-
panzees. Bipedality is widely suggested to have
arisen as an opportunistic, or even necessary, re-
sponse to a drier climate and the expansion of
savannas. These views have been challenged on
paleontological and theoretical grounds (9,21).
However, without additional fossil evidence, the
evolutionary paths of the various great apes and
humans have remained shrouded.
In related papers in this issue (22–27), we de-
scribe in detail newly discovered and/or analyzed
specimens ofAr. ramidus, including two individ-
uals with numerous postcranial elements. All are
dated to 4.4 Ma and come from the Middle
Awash area of the Ethiopian Afar rift. Local
geology and many associated fossils are also
described (28–30). These new data jointly es-
tablishArdipithecusas a basal hominid adaptive
plateau preceding the emergence ofAustralo-
pithecusand its successor,Homo.Inferences
based onAr. ramidusalso facilitate understand-
ing its precursors (22,23,27,31). Here, we pro-
vide an integrated view of these studies and
summarize their implications.
The Middle Awash.The Middle Awash study
area contains a combined thickness of >1 km of
Neogene strata. To date, these deposits have
yielded eight fossil hominid taxa spanning the
Late Miocene to Pleistocene (>6.0 to <0.08 Ma)
(32,33). Hominids make up only 284 of the
18,327 total cataloged vertebrate specimens. Spa-
tially and chronologically centered in this succes-
sion, the Central Awash Complex (CAC) (28,34)
rises above the Afar floor as a domelike structure
comprising >300 m of radioisotopically and
paleomagnetically calibrated, sporadically fossil-
iferous strata dating between 5.55 and 3.85 Ma.
Centered in its stratigraphic column are two prom-
inent and widespread volcanic marker horizons
that encapsulate the Lower Aramis Member of
the Sagantole Formation (Fig. 1). These, the
Gàala (“camel”in Afar language) Vitric Tuff
Complex (GATC) and the superimposed Daam
Aatu (“baboon” in Afar language) Basaltic Tuff
(DABT), have indistinguishable laser fusion
39
Ar/
40
Ar dates of 4.4 Ma. Sandwiched between
RESEARCHARTICLES
1
Human Evolution Research Center and Department of
Integrative Biology, 3101 Valley Life Sciences Building,
University of California, Berkeley, CA 94720, USA.
2
Rift Valley
Research Service, Post Office Box 5717, Addis Ababa,
Ethiopia.
3
Department of Anthropology and Archaeology,
Authority for Research and Conservation of the Cultural
Heritage, Ministry of Youth, Sports and Culture, Post Office
Box 6686, Addis Ababa, Ethiopia.
4
Department of Physical
Anthropology, Cleveland Museum of Natural History, 1 Wade
Oval Drive, Cleveland, OH 44106, USA.
5
Department of
Anthropology, School of Biomedical Sciences, Kent State
University, Kent, OH 44240–0001, USA.
6
The University
Museum, the University of Tokyo, Hongo, Bunkyo-ku, Tokyo
113-0033, Japan.
7
Earth Environmental Sciences Division, Los
Alamos National Laboratory, Los Alamos, NM 87545, USA.
*To whom correspondence should be addressed. E-mail:
[email protected]
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 75

30. 30 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
the two tuffs are fossiliferous sediments averag-
ing ~3 m in thickness and cropping out discon-
tinuously over an arc-shaped, natural erosional
transect of >9 km (28). The rich fossil and
geologic data from these units provide a detailed
characterization of the Pliocene African land-
scape inhabited byArdipithecus.
We first surveyed the CAC during 1981 in
attempts to understand the distribution of fossils
within the region. We launched a systematic pro-
gram of geological, geochronological, and pale-
ontological investigation in 1992. Initial visits to
the CAC’s northeastern flank documented abun-
dant fossilized wood and seeds in the interval
between the two tuffs. We collected and identi-
fied a highly fragmented sample of vertebrates,
including abundant cercopithecid monkeys and
tragelaphine bovids. The first hominid fossils
were found at Aramis vertebrate paleontology
locality 1 (ARA-VP-1) on 17 December 1992.
Two initial seasons of stratigraphic and geo-
chronological studies yielded 649 cataloged
vertebrates, including a minimum number of 17
hominid individuals represented mostly by teeth
(10).
Because of its content, the Lower Aramis
Member became the focus of our paleontological
efforts. Fourteen sublocalities within the original
ARA-VP-1 locality were circumscribed and
subjected to repeated collecting of all biological
remains, based on multiple team crawls (35)
across the eroding outcrops between 1995 and
2005. Analogous collections were made at ad-
jacent localities (ARA-VP-6, -7, and -17), as well
as at the eastern and western exposures of the
Ardipithecus-bearing sedimentary units (KUS-
VP-2 and SAG-VP-7) (KUS, Kuseralee Dora;
SAG, Sagantole). The Lower Aramis Member
vertebrate assemblage (table S1) now totals
>6000 cataloged specimens, including 109
hominid specimens that represent a minimum of
36 individuals. An additional estimated 135,000
recovered fragments of bone and teeth from this
stratigraphic interval are cataloged by locality
and taxon as pooled“bulk”assemblages. Anal-
ogous samples were collected from the Lower
Aramis Member on the eastern transect pole
(SAG-VP-1, -3, and -6). Fossils from localities
higher and lower in the local Middle Awash
succession (7,12,32) and at nearby Gona (36)
are reported elsewhere.
TheARA-VP-6/500partial hominid skeleton.
Bones of medium and large mammals were usu-
ally ravaged by large carnivores, then embedded
in alluvial silty clay of the Lower Aramis Mem-
ber. Once exposed by erosion, postdepositional
destruction of the fossils by decalcification and
fracture is typical. As a result, the larger verte-
brate assemblage lacks the more complete cranial
and postcranial elements typically recovered from
other African hominid localities. The identifica-
tion of larger mammals below the family level is
therefore most often accomplished via teeth. The
hominid subassemblage does not depart from this
general preservational pattern (29).
There was consequently little initial hope that
the stratigraphic interval between the two tuffs
would yield crucially needed postcranial ele-
ments ofArdipithecus. The only relevant post-
crania (arm elements) had come from slightly
higher in the section in 1993 (10). However, on
5 November 1994, Y.H.S. collected two hominid
metacarpal fragments (ARA-VP-6/500-001aand
b) from the surface of an exposed silty clay ~3 m
below the upper tuff (DABT), 54 m to the north
of the point that had 10 months earlier yielded
theArdipithecusholotype dentition. Sieving
produced additional hominid phalanges. The
outcrop scrape exposed a hominid phalanx in
situ, followed by a femur shaft and nearly com-
plete tibia. Subsequent excavation during 1994
Fig. 1.Geography and
stratigraphy of the Aramis
region. Two dated vol-
canic horizons constrain
the mainArdipithecus-
bearing stratigraphic inter-
val in the Aramis region.
The top frame shows these
tephra in situ near the
eastern end of the 9-km
outcrop. The dark stripe
in the background is the
riverine forest of the
modern Awash River
running from right to left,
south to north, through
the Middle Awash study
area of the Afar Rift. The
lower frames are con-
temporaneous helicopter
views over ARA-VP-1 (Yonas
Molar Site) to show the
geographic position of
the top photo and to de-
pict the extensive outcrop
of the upper tuff horizon
(dotted lines show the
DABT) across the local
landscape. Vehicles are
in the same position to
provide orientation. Sedi-
ments outcropping im-
mediately below this
4.4-million-year-old ho-
rizon yielded the floral,
faunal, and isotopic contexts forAr. ramidus.The frame to the left shows the slight eastward dip of the Sagantole Formation toward the modern Awash River. The
contiguous frame to the right is a view up the modern upper Aramis catchment. The ARA-VP-6 locality where the partialArdipithecusskeleton was excavated is near its
top right corner (Fig. 2).
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org76
Ardipithecus ramidusArdipithecus ramidus

31. 31www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
and the next field season (at a rate of ~20 ver-
tical mm/day across ~3 m
2
) revealed >100 ad-
ditional in situ hominid fragments, including
sesamoids (Fig. 2 and table S2). Carnivore
damage was absent.
The bony remains of this individual (ARA-
VP-6/500) (Fig. 3) ( 37) are off-white in color
and very poorly fossilized. Smaller elements
(hand and foot bones and teeth) are mostly un-
distorted, but all larger limb bones are variably
crushed. In the field, the fossils were so soft that
they would crumble when touched. They were
rescued as follows: Exposure by dental pick,
bamboo, and porcupine quill probe was followed
by in situ consolidation. We dampened the en-
casing sediment to prevent desiccation and
further disintegration of the fossils during exca-
vation. Each of the subspecimens required mul-
tiple coats of consolidant, followed by extraction
in plaster and aluminum foil jackets, then ad-
ditional consolidant before transport to Addis
Ababa.
Pieces were assigned number suffixes based
on recovery order. Back-dirt was weathered in
place and resieved. The 1995 field season yielded
facial fragments and a few other elements in
northern and eastern extensions of the initial ex-
cavation. Further excavation in 1996 exposed no
additional remains. Each fragment’s position,
axial orientation, and dip were logged relative
to a datum (strata here dip east at ~4° to 5°). A
polygon representing the outer perimeter and ver-
tical extent of the hominid fragment constellation
(based on each bone’s center point) was de-
marcated by a carapace of limestone blocks ce-
mented with concrete after excavation, then
further protected by a superimposed pile of
boulders, per local Afar custom.
Fig. 3.TheARA-VP-6/500skeleton. This is a
composite photograph to show the approximate
placement of elements recovered. Some pieces
found separately in the excavation are rejoined
here. Intermediate and terminal phalanges are
only provisionally allocated to position and side.
Fig. 2.TheARA-VP-6/500skeletal excavation. Successive zooms on theARA-VP-6/500partial skeleton
discovery are shown. Insets show the application of consolidant to the tibia shaft and removal of the os
coxae in a plaster jacket in 1994–1995. No skeletal parts were found articulated (the mandible excavation
succession shows the close proximity of a proximal hand phalanx and trapezium). Only in situ specimens
are shown on the plan and profile views. Note the tight vertical and wider horizontal distributions of the
remains. Local strata dip ~5° to the east. The lower inside corner of each yellow pin flag marks the center
point for each in situ specimen from the 1994–1995 excavation. The 1995–1996 excavation recovered
additional, primarily craniodental remains between these flags and the vehicle. The boulder pile
emplaced at the end of the 1996–1997 excavation marks the discovery site today.
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
77
RESEARCH ARTICLES
Research Articles

32. 32 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
The skeleton was scattered in typical Lower
Aramis Member sediment (Fig. 2): fine-grained,
massive, unslickensided, reddish-brown alluvial
silty clay containing abundant decalcified root
casts, fossil wood, and seeds. A 5- to 15-cm lens
of poorly sorted sand and gravel lies immediately
below the silty clay, and the spread of cranial
parts to the north suggests that the bones of the
carcass came to rest in a shallow swale on the
floodplain.
There is no evidence of weathering or mam-
malian chewing onARA-VP-6/500. Bony ele-
ments were completely disarticulated and lacked
anatomical association. Many larger elements
showed prefossilization fragmentation, orienta-
tion, and scatter suggestive of trampling. The
skull was particularly affected, and the facial
elements and teeth were widely scattered across
the excavated area. Bioturbation tilted some
phalanges and metacarpals at high dip angles
(Fig. 2). A few postcrania of a largeAquila
(eagle) and other birds were recovered during ex-
cavation, as were a few micromammals. No large-
mammal remains (except isolated cercopithecid
teeth and shaft splinters from a medium-to-large
mammal limb bone) were associated. The cause
of death is indeterminate. The specimen is judged
to be female. The only pathology is a partially
healed osteolytic lesion suggestive of local infec-
tion of the left proximal ray 5 pedal phalanx
(ARA-VP-6/500-044).
Laboratory exposure and consolidation of the
soft, crushed fossils were accomplished under
binocular microscope. Acetone was applied with
brushes and hypodermic needles to resoften and
remove small patches of consolidant-hardened
encasing matrix. Microsurgery at the interface
between softened matrix and bone proceeded
millimeter by submillimeter, rehardening each
cleaned surface with consolidant after exposure.
This process took several years. The freed spe-
cimens remain fragile and soft, but radiographic
accessibility is excellent. Most restoration and
correction for distortion were accomplished with
plaster replicas or micro–computed tomography
digital data to preserve the original fossils in their
discovery state.
Environmental context.The Lower Aramis
Member lacks any evidence of the hydraulic
mixing that afflicts many other hominid-bearing
assemblages. The unwarranted inference that
early hominids occupied“mosaic habitats”(38)
is often based on such mixed assemblages, so
the resolution and fidelity of the Aramis envi-
ronmental data sets are valuable. We estimate
that the interval of time represented by the strata
between the two tuffs at Aramis is <10
5
years,
and perhaps just a few hundred or thousand
years (28,39). The lithology, thickness, tapho-
nomic evidence, and similar age of the constrain-
ing marker horizons imply that geologically, the
evidence can be viewed as“habitat time-
averaged”(40). Indeed, we do not see notably
different environmental indicators in the fossils
or geologic or chemical data sampled vertically
throughout the interval. The wealth of data al-
lows a high-fidelity representation [sensu (41)] of
the ecological community and environment
inhabited byAr. ramidus4.4 Ma.
A variety of data indicate that the wooded
biotope varied laterally across the Pliocene
landscape (28–30). The hominid-bearing local-
ities (centered on the ARA-VP-1 sublocalities)
are rich in fossilized wood fragments, seeds, and
animal fossils. Here, isotopic paleosol composi-
tions indicate mostly wooded conditions (28 ).
There was obviously more water at Aramis then
(4.4 Ma)—supporting a much richer flora and
fauna—than there is today. The higher water
budget is possibly due to higher elevation dur-
ing deposition (42) or to paleoclimatic factors
such as a more continuous Pliocene El Niño
effect (43). An abrupt transition occurs southeast
of the SAG-VP-7 locality, where sedimentary,
faunal, taphonomic, and isotopic data imply a
more open rift-axial setting depauperate in fau-
nal remains and lacking in primates, micro-
mammals, and macrobotanical remains (29,30).
Along the northern slope of the CAC, all
localities of the Lower Aramis Member yielded
tragelaphine bovids, monkeys, and other data
indicative of more wooded conditions. Carbon
isotopes from the teeth of fiveArdipithecusin-
dividuals found here imply that they fed largely
on C
3plants in woodlands and/or among the
small patches of forests in the vicinity. We inter-
pret the combined contextual data to indicate
thatAr. ramiduspreferred a woodland-to-forest
habitat (29,30) rather than open grasslands.
This finding is inconsistent with hypotheses
positing hominid origins via climate-driven
savanna expansion.
Variation and classification.Initial (1994)
description of the limited hominid sample from
Aramis placed these remains in a newly dis-
coveredAustralopithecusspecies interpreted as
the most primitive then known (10). Subsequent
recovery of theARA-VP-6/500skeleton showed
that, relative to body size, its dentition was small,
unlikeAustralopithecus. Strict cladistic practice
required a new genus name for this sister taxon of
Australopithecus, so the material was renamed as
the new genusArdipithecusin 1995, with the
lack of megadonty added to the species diagnosis
even as the partial skeleton’s excavation was still
under way (44). Subsequent discovery of the
earlier probable chronospeciesAr. kadabbain
1997 (11,12) was followed by recovery of
Orrorinin 2000 (13) andSahelanthropusin
2001 (14). These Late Miocene fossils provide
additional outgroup material useful in assessing
the phylogenetic position ofAr. ramidus.
Only two adjacent Ethiopian study areas
(the Middle Awash and Gona) have yielded
confirmed remains ofAr. ramidusto date (7,36).
Neither has produced any evidence to reject a
single species lineage as the source of the com-
bined hominid sample from these Pliocene sites.
We thus interpret the Lower Aramis Member
hominid assemblage as a single taxon (22). Pene-
contemporary (~4.3 to 4.7 Ma) hominid remains
from elsewhere are sparse (45,46), and these are
broadly compatible with the now expanded range
of variation inAr.ramidus(22,23). Thus,
although continental sampling is still obviously
inadequate, describing hominid species diversity
in this time frame (47) as“very bushy” seems
unwarranted (48).
The amount of variation within the known
AfarAr. ramidussample appears to be lower than
typical for species ofAustralopithecus.This is
probably due to a lesser degree of sexual di-
morphism inArdipithecus, combined with the
narrow time window represented by the interval
between the two Aramis tuffs. Skeletal dimor-
phism is notably difficult to assess, except in rare
instances of geologically isochronous samples of
a species lineage (e.g., A.L. 333“first family”)
(49). ForAr. ramidus, theARA-VP-6/500skel-
eton (Figs. 3 and 4) provides a rare opportunity
for guiding a probabilistic approach to sex at-
tribution of conspecific fossils, relying on canines
(22) and postcranially based estimates of body
size (27). The implication is that there was broad
overlap in body size between males and females
ofAr. ramidus.
Cranial and dental anatomy.TheAr. ramidus
skull (23) is very similar to the larger, more
robustSahelanthropuscranium (TM 266-01-
60-1) from Chad, also interpreted as an early
hominid (14 ,50). Some of the differences are
probably partly sex-related.Ar. ramidusshares
withSahelanthropusa small cranial capacity
(300 to 350 cc) and considerable midfacial
projection but a maxillo-premaxillary complex
that is less prognathic than that of modern
African apes [not necessarily a derived trait
shared withHomo, in contrast with (51)]. The
ArdipithecusandSahelanthropuscrania each
lack a distinct post-toral sulcus, and both exhibit
an anteriorly positioned posterior cranial base.
Most aspects of the craniofacial structure of
Sahelanthropus/Ardipithecusare probably close
to the African ape and hominid ancestral state.
Gorilla and chimpanzee cranial morphologies,
as well as their specialized dentitions, are clearly
divergently derived (22). InGorilla, enhanced
facial size and prognathism occur in relation to
larger general size and an increasing adaptation
to herbivory and folivory. InPan(also with en-
hanced prognathism), derived cranial form
(including anterior basicranial lengthening) prob-
ably occurred as a part of enhanced terrestrial-
ity accompanied by elevated agonistic behavior
and its anatomical correlates, such as tusklike
canines (22,23). The bonobo cranial base and
Ardipithecuscraniofacial structure may be less
derived, but even the bonobo seems to be de-
rived in its relatively small face and global den-
tal reduction (22). This was probably at least in
part due to decreased intraspecific aggression in
the bonobo lineage after separation from the
common chimpanzee lineage.
The superoinferiorly short but intermedi-
ately prognathicAr. ramidusface lacks the
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
78
Ardipithecus ramidus
Ardipithecus ramidus

33. 33www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
broadening and anterior migration of the zy-
gomaxillary area seen to varying degrees in
species ofAustralopithecus. The primitive cra-
niofacial pattern shared betweenSahelanthropus
andArdipithecussuggests that the genusAustra-
lopithecuswould later evolve a craniofacial struc-
ture capable of increased postcanine mastication
consequent to an ecological breakout from
wooded habitats, expanding its foraging into
more open environments (7,10).
TheArdipithecusdentition suggests omni-
vory (22). It exhibits none of the specializations
seen among modern apes; neither the large in-
cisors ofPongoorPannor the specialized molar
morphology ofPongo, Pan, orGorilla. Post-
canine size relative to body size was slightly
larger than inPanbut smaller than inGorilla,
Pongo, or (especially)Au.afarensis.Ar.ramidus
molars overlap considerably withPanin some
measures of enamel thickness but differ in overall
thickness and structure. Chimpanzee molars have
a broad occlusal basin with locally thin enamel
not seen inArdipithecus.Panmolar morphology
is probably an adaptation to crushing relatively
soft and nonabrasive food items such as ripe
fruits, while retaining some shearing capacities.
TheArdipithecusdentition shows no strong
signals of ripe-fruit frugivory, folivory-herbivory,
or feeding on hard objects. Its macroscopic and
microscopic wear patterns, as well as the low
bunodont cusps with intermediate enamel thick-
ness (22), suggest that its diet was not particularly
abrasive but may have included some hard foods.
It is consistent with a partially terrestrial, partially
arboreal pattern of feeding in a predominantly
wooded habitat.
Carbon isotopic evidence from the teeth
of fiveAr. ramidusindividuals suggests that
ArdipithecusandAustralopithecuswere distinct
in dietary intake (30).“Robust”and“nonrobust”
Australopithecushave enamel isotope values in-
dicating a diet of more than 30% C
4plants, with
variation ranging up to ~80% C
4. In contrast,
the knownAr. ramidusindividuals vary only
between ~10 and 25% C
4, and thus also differ
fromPan troglodytes,which prefers ripe fruit and
is considered closer to a pure C
3feeder (30).
Thus,Ardipithecusappears to have exploited a
wider range of woodland resources than do chim-
panzees, but without relying on the open biotope
foods consumed by laterAustralopithecus.
Evolution of the canine/lower third premolar
complex (C/P
3) potentially illuminates social and
reproductive behavior. TheAr.ramiduscanine
sample totals 21 Aramis individuals. Some are
small fragments, but all show informative mor-
phology and/or wear. All specimens are either
morphologically similar to those from female
apes or are further derived toward the later hom-
inid condition (22). Morphological and metric
variation in the sample is small. Functionally
important sex-related size dimorphism is not ap-
parent. There is no evidence of functional honing
(planar facets on the mesiobuccal P
3or sharpened
edges on the distolabial upper canine margin).
The largest, presumably male, specimens are as
morphologically derived as the smallest, showing
that dimorphic canine morphology was virtually
absent in these hominids by 4.4 Ma. Further-
more, a juvenile probable male lacks the delayed
canine eruption seen in chimpanzees, approximat-
ing theAu.anamensisandAu.afarensiscon-
ditions and indicating that the canine was not an
important component of adult sociobehavioral
relationships.
The differential status of upper versus lower
canine morphology is informative. InAr.ramidus,
Fig. 4.Comparisons ofArdipithecus(left) and earlyAustralopithecus(right). (A ) Ulnar, radial, first rib, and
talar comparisons of theAr. ramidus ARA-VP-6/500andAu. afarensis A.L. 288-1(“Lucy”) skeletal individuals
illustrate larger postcranial dimensions for theArdipithecusindividual relative to dental size. Comparison of the
postcanine dentitions reveals the megadontia of theAustralopithecusindividual. (B) Occlusal and lateral views
of three time-successive mandibles dated to 4.4, 4.12, and 3.4 Ma, respectively, from left to right:ARA-VP-1/401
Ar. ramidus;KNM-KP 29281 Au. anamensisholotype (mirrored);MAK-VP- 1/12 Au. afarensis(mirrored).
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
79
RESEARCH ARTICLES
Research Articles

34. 34 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
the lower canines retain modally more apelike
morphology than do the uppers, and, in contra-
distinction to other anthropoids, the height of the
maxillary canine crown is lower than that of the
mandibular (22 ). This relationship is opposite that
seen in great apes and cercopithecids, whose
upper canine dominance is exaggerated, particu-
larly in males of dimorphic species. In these
primates, upper canine projection and prominence
function in both weaponry and display. The
Ar. ramiduscanines are metrically and morpho-
logically derived in the direction of later homi-
nids, and we hypothesize that reduction and
alteration of upper canine size and shape in this
and earlier hominid species are related to changes
in social behaviors (22 ,31).
The canines ofSahelanthropus, Orrorin, and
Ar. kadabbaare broadly equivalent to those of
Ar.ramidusin size and function. However, the
upper canines of Late Miocene hominids exhibit
a subtle but distinctly more primitive morpholo-
gy than theirAr.ramidushomologs, potentially
including occasional residual (female ape–like)
honing as part of their variation (12,15). This
suggests that upper canine prominence was
reduced through the Late Miocene and Early
Pliocene. In contrast, the C/P
3complex of the last
common ancestor of hominids and chimpanzees
probably had a moderate level of canine dimor-
phism combined with functional honing. This was
subsequently generally retained inP.paniscus
and enhanced inP.troglodytes.
Body size and dimorphism.The partial
skeletonARA-VP-6/500is identified as female
based on probability assessments of canine size
(its canines are among the smallest of those of 21
available individuals) (22). This interpretation is
corroborated by its small endo- and exocranial
size, as well as its superoinferiorly thin supra-
orbital torus (23). Bipedal standing body height
for theARA-VP-6/500individual is estimated at
approximately 120 cm, and body mass at ~50 kg
(27). Although actual body mass may vary con-
siderably in relation to skeletal size, this is a large
female body mass.
Of theAr. ramiduspostcranial elements, the
humerus represents the largest minimum num-
ber of individuals (seven).ARA-VP-6/500does
not preserve a humerus, but detailed comparisons
suggest that its forelimb was ~2 to 8% larger in
linear dimensions than the partial forelimb skele-
tonARA-VP-7/2(24,27), which does include a
humerus. This would makeARA-VP-6/500either
the second- or third-largest of eight individuals
within the Aramis humeral sample. The com-
bined evidence suggests thatArdipithecusskele-
tal body size was nearly monomorphic, and less
dimorphic thanAustralopithecus, as estimated
from template bootstrapping (49). Most likely,
Ardipithecusexhibited minimal skeletal body
size dimorphism, similar toPan, consistent with
a male-bonded social system, most likely a prim-
itive retention from the CLCA condition (31).
With its subsequent commitment to terrestrial
bipedality,Australopithecusprobably enhanced
female cooperation and group cohesion, thus
potentially reducing female body size, whereas
male size increased in response to predation
pressure, probably elevated by expanding niche
breadth.
Postcranial biology and locomotion.Re-
gardless of whether the AfarAr. ramiduspop-
ulation represents a hominid relict or a lineal
ancestor, this taxon’ s biology resolves funda-
mental evolutionary questions persisting since
Darwin. Its substantially primitive postcranial
anatomy appears to signal a grade-based differ-
ence from laterAustralopithecus. The challenge
of understanding its evolutionary and functional
implications required a nontraditional approach.
Without testable hypotheses of underlying gene-
based developmental mechanisms, many paleo-
anthropological analyses have been adaptationist
(52) and/or purely numerically discriminatory.
Therefore, wherever possible, in the accompany-
ing postcranial papers (24–27) we restrict
hypotheses to those that can be formulated
consistent with putative selection acting on
cascades of modular-based positional informa-
tion, especially when these can be potentially
grounded in known anabolic mechanisms. This
approach is summarized elsewhere (53,54) and
in supporting online material text S1.
The upper pelvis ofAr. ramiduspresents a
contrast to its primitive hand, foot, and limbs. The
ilia are abbreviated superoinferiorly and sagittally
oriented but broad mediolaterally, so much so that
the anterior inferior iliac spine has become a
separate growth site, as in all later hominids. The
pubic symphyseal face is quite short. A slight
sciatic notch is present, although ischial structure
was similar to that of extant African apes. This
suggests that pattern-formation shifts for bipedal-
ity were only partly realized inAr.ramidus. These
changes may have culminated a long period of
facultative bipedality hinted at by isolated post-
cranial elements from the probable chronospecies
Ar. kadabba(12) and other Late Miocene forms
(13,14).
Paramount among the retained primitive
characters of theAr. ramidushindlimb is a fully
abductable first ray (hallux, or great toe), but in
combination with elements of a robust plantar
substructure that stabilized the foot during heel-
and toe-off. Although it was still a highly ef-
fective grasping organ, the foot ofAr. ramidus
also maintained propulsive capacity long since
abandoned by extant great apes (in which greater
opposition between the hallux and lateral rays
evolved, i.e., a more handlike conformation than
inAr. ramidus)(26).
Other defining and notably primitive char-
acters include a moderately elongate mid-tarsus,
a robust lateral peroneal complex in which
muscles of the lateral compartment performed
substantial plantarflexion, and a primitive
(flexion-resistant) geometric configuration of
the lateral metatarsal bases. Thus, theAr. ramidus
foot is an amalgam of retained primitive char-
acters as well as traits specialized for habitual
bipedality, such as the expanded second meta-
tarsal base that anchored plantarflexion during
heel- and toe-off. Many of the foot’s primary
adaptations to fulcrumation are probable reten-
tions from the gorilla/chimpanzee/human last
common ancestor (GLCA), but these have been
eliminated in apes, presumably for vertical
climbing.
TheARA-VP-6/500radius/tibia ratio is 0.95,
as in generalized above-branch quadrupeds such
as macaques andProconsul(an Early Miocene
ape) (27). Its intermembral index (the ratio of
forelimb length to hindlimb length) is also similar
to those of above-branch quadrupeds. These facts
suggest that African apes experienced both
forelimb elongation and hindlimb reduction,
whereas hominid proportions remained largely
unchanged until the dramatic forearm shortening
and hindlimb elongation of Plio-Pleistocene
Homo.
These primitive proportions are consistent
with virtually all other aspects of theAr. ramidus
skeleton. The inferred locomotor pattern com-
bined both terrestrial bipedality and arboreal
clambering in which much weight was supported
on the palms. The hand phalanges are elongate
relative to those ofProconsul, but metacarpals
(Mc) 2 to 5 remained primitively short and lacked
any corporal modeling or adaptations typical of
knuckle-walking (24). Moreover, the virtually
complete wrist ofARA-VP-6/500(lacking only
the pisiform) exhibits striking adaptations for
midcarpal dorsiflexion (backward deflection of
the dorsum of the hand), consistent with a highly
advanced form of arboreal palmigrady. In ad-
dition, substantial metacarpal-phalangeal dorsi-
flexion is indicated both by moderate dorsal
notching of the Mc2 to -5 heads and by marked
palmar displacement of the capitate head. Togeth-
er these must have permitted dorsiflexion of the
wrist and hand to a degree unparalleled in great
apes.
TheAr. ramiduselbow joint provided full
extension but lacks any characters diagnostic of
habitual suspension. Ulnar withdrawal was com-
plete and the thumb moderately robust, with
indications of a distinct and fully functional flex-
or pollicis longus tendon. The hamate’s hamulus
permitted substantial metacarpal motion for
opposition against the first ray. The central joint
complex (Mc2/Mc3/capitate/trapezoid) exhibits
none of the complex angular relationships and
marked syndesmotic reinforcement seen in extant
apes. Together, these retained primitive char-
acters, unlike their homologs in highly derived
African apes, imply that the dominant locomotor
pattern of the GLCA was arboreal palmigrady
rather than vertical climbing and/or suspension
(orthogrady). Another strong inference is that
hominids have never knuckle-walked (26).
The extraordinary forelimb ofAr. ramidus,
in combination with its limb proportions and
likely primitive early hominid lumbar column
(55), casts new light on the evolution of the
lower spine. The traditional interpretation has
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
80
Ardipithecus ramidus
Ardipithecus ramidus

35. 35www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
been that the lumbar transverse processes be-
came dorsally relocated as the lumbar column
reduced in length. The data fromAr. ramidus
imply that ulnar withdrawal was not a suspen-
sory adaptation but was instead an enhancement
of distal forelimb maneuverability that accom-
panied profound changes in the shoulder. Spinal
column invagination appears to have been an
integral part of thoracic restructuring to in-
crease shoulder joint laterality, thereby enhancing
forelimb mobility for advanced arboreal quad-
rupedalism, especially careful climbing and
bridging. A still primitive deltoid complex in
bothAr. ramidusand Asian ancestral apes (e.g.,
Sivapithecus) now becomes more understand-
able. A predominantly Sharpey’s fiber deltoid
insertion can be viewed as a retention in above-
branch quadrupeds that only later became
modified for suspension (separately) in extant
African and Asian apes.
The adoption of bipedality and its temporal
association with progressive canine reduction
and loss of functional honing now constitute the
principal defining characters of Hominidae. The
orthograde positional behaviors of hominids and
apes were thus acquired in parallel, generated
by early bipedal progression in the former and
suspension and vertical climbing in the latter.
Overall,Ar. ramidusdemonstrates that the last
common ancestors of humans and African apes
were morphologically far more primitive than
anticipated, exhibiting numerous characters rem-
iniscent of Middle and Early Miocene hominoids.
This reinforces what Huxley appreciated in 1860:
“the stock whence two or more species have
sprung, need in no respect be intermediate
between those species”[(56), p. 568].
Ardipithecusand the great apes.Ar.
ramidusilluminates several collateral aspects
of hominoid evolution. Despite the demise of
Ramapithecusas a putative hominid ancestor, at
least one Eurasian Miocene ape,Ouranopithecus,
has been suggested as being phyletically related
to later African hominids (57), whereas another,
Dryopithecus, is often considered an alternative
sister taxon of the hominid and African ape clade
(58).Ardipithecuseffectively falsifies both hy-
potheses.Ar. ramiduslacks the derived characters
ofOuranopithecusassociated with postcanine
enlargement and relative canine reduction while
still providing a primitive morphological sub-
strate for the emergence ofAustralopithecus. The
new perspective thatAr. ramidusoffers on hom-
inoid postcranial evolution strongly suggests that
Dryopithecusacquired forelimb adaptations to
suspensory behaviors independently from African
apes.Ar.ramidussuggests that these Eurasian
forms were too derived to have been specially
related to either the hominid or extant African ape
clades. Moreover, the remarkably primitive
postcranium of potentialPongoancestors (e.g.,
Sivapithecus), coupled with what is now evident-
ly widespread homoplasy in extant hominoids,
suggests that thePongoclade was established
even before the first dispersal events of large-
bodied apes from Africa into Eurasia, shortly
after docking of the Afro-Arabian and Eurasian
plates at ~18 Ma (59).
An additional implication ofAr. ramidusstems
from its demonstration that remarkable functional
and structural similarities in the postcrania of
Pongoand the African apes have evolved in
parallel, as have those ofPanandGorilla(27).
Until now, a myriad of characters shared among
the extant African apes were presumed to have
been present also in ancestral hominids (because
they were presumed to have been the ancestral
state) (60). However, it now appears that many of
these putative shared primitive characteristics
have evolved independently. This highlights the
alacrity with which similar anatomical structures
can emerge, most likely by analogous selection
operating on homologous genomes. The same
genetic pathways can be repeatedly and indepen-
dently coopted, resulting in convergent adapta-
tions (61). Recent work on gene expression
demonstrates that there are also multiple path-
ways that can produce similar but independently
derived anatomical structures (62).
Work on deep homology shows that parallel
evolution“must be considered a fact of life
in the phylogenetic history of animals”[(63),
p. 822]. This is also seen in more terminal
branches; for example, during the past two mil-
lion years of stickleback fish evolution (64).
Such evolvability and parallelism are even sug-
gested for the catarrhine dentition (65 ).Ar.
ramidusreveals an excellent example of this
phenomenon within the African ape-hominid
clade by demonstrating the striking reoccurrence
of syndesmotic fixation of the central joint com-
plexes in hominoid wrists adapted to suspensory
locomotion (including not only those ofPan
andGorillabut also those ofPongoand, par-
tially,Dryopithecus). Such observations on very
different evolutionary scales all caution against
indiscriminant reliance on raw character states to
assess phylogeny. A consideration of wider pat-
terns of manifestations of such adaptive evolu-
tion, not only in character constellations but also
in their evolutionary context, may be needed to
tease apart homology and homoplasy. A far
more complete fossil record will be needed to
accomplish such a goal.
Such considerations also bear on current es-
timates of the antiquity of the divergence be-
tween the human and chimpanzee clades. Many
such estimates, suggesting striking recency, have
become widely accepted because of the pre-
sumed homology of human and African ape
morphologies (60). This obtains despite the rec-
ognition that broad assumptions about both the
regularity of molecular change and the reliability
of calibration dates required to establish such
rates have strong limitations (66,67). The
homoplasy now demonstrated for hominoids by
Ar. ramidusprovides fair warning with respect to
such chronologies, especially those currently used
to calibrate other divergence events, including the
split times of New and Old World monkeys,
hylobatids, and the orangutan. The sparseness of
the primate fossil record affecting these estimates
is now compounded by the dangers posed by
convergences perceived as homologies. Such
difficulties are further exacerbated by newly
recognized complexities in estimating quantitative
degrees of genetic separation (66–68). In effect,
there is now no a priori reason to presume that
human-chimpanzee split times are especially
recent, and the fossil evidence is now fully
compatible with older chimpanzee-human diver-
gence dates [7 to 10 Ma (12,69)] than those
currently in vogue (70).
Hominid phylogenetics.The expandedAr.
ramidussample allows more detailed consider-
ation of early hominid phylogenetics. The place-
ment ofArdipithecusrelative to later hominids
can be approached by using modern and Mio-
cene apes as the outgroup. An earlier cladistic
study of this kind concluded thatAr. ramidus
was the sister taxon of all later hominids (71). A
more recent assessment ofAr.ramidusdental
characters came to the same conclusion (7). In
these analyses, a suite of derived features and
character complexes exclusively aligningA r.
ramiduswithAustralopithecuswas identified,
but these were based on comparatively limited
anatomical elements. TheAr. ramiduscharacters
reported here, combined with those from Gona
(36), allow a more complete analysis that clarifies
the relationships among early hominid taxa.
Parsimony-based cladistic analyses are useful
in deciphering relationships within the hominid
family tree, despite their shortcomings (72,73).
The distribution of characters identified in Table
1 clearly shows thatAr. ramidusis derived rela-
tive to all known Late Miocene fossils attributed
to the hominid clade. The earlier and more prim-
itive probable chronospeciesAr. kadabbais
found in 5.5- to 5.7-million-year-old deposits a
mere 22 km west of Aramis, consistent with local
(and perhaps regional) phyletic evolution. Its
limited known elements are similar to those of
other Late Miocene hominids in Kenya and Chad
(12–14).
Table 1.(See pages 82 and 83.) The assembly of shared derived characters among early hominid
taxa. Late Miocene and early Pliocene fossils now allow the strong inference of some character states
(primitive, in blue) in the last common ancestor shared by chimpanzees and humans. Many other
characters (not shown here) of extant apes were once considered primitive but are now shown to be
derived and specific to those lineages. Late Miocene fossils from Ethiopia, Kenya, and Chad share
some derived characters (in yellow) with all later hominids. TheAr. ramidussample reported here
shows a mixture of primitive and derived characters consistent with its phylogenetic and chronolog-
ical placement. Phylogenetic implications are in Fig. 5. (An Excel version of this table is available
in the supporting online material.)
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
81
RESEARCH ARTICLES
Research Articles

36. 36 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org82
Ardipithecus ramidusArdipithecus ramidus

37. 37www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 83
RESEARCH ARTICLES
Research Articles

38. 38 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Comparatively few features ofAr.ramidus
are derived relative to these earlier hominids,
although many body parts of the latter are still
unrepresented. There are no apparent features
sufficiently unique to warrant the exclusion ofAr.
ramidusas being ancestral toAustralopithecus
(74), and a greatly expanded set of shared derived
characters now linksAr. ramiduswith later mem-
bers of the hominid clade. Table 1 identifies some
of the most important. This pattern robustly falsi-
fies earlier assessments that the Aramis fos-
sils represent an ancestral chimpanzee (13,75).
These results are suggestive of a cohesive hom-
inid evolutionary grade precedingAustralopithecus
(currently >6.0 to 4.2 Ma). By priority, the name
Ardipithecusmay encompass other named genera
at this adaptive plateau (12, 15).
The question of whetherAr. ramidusis an-
cestral to later hominids is moot for some cla-
dists because they consider ancestors inherently
unrecognizable and therefore recognize only sis-
ter taxa (76). The fossils reported here make it
even more obvious thatAr. ramidusis the cla-
distic sister toAustralopithecus/Homobecause it
shares primitive characters with earlier hominids
and apes but at the same time exhibits many
important derived characters that are shared ex-
clusively only with later hominids (Table 1).
Species-level phylogenetics are more diffi-
cult to discern given the sparse geographic and
temporal distribution of available fossils (Fig. 5).
Primitive characters seen inAr. ramiduspersist
most markedly in its apparent (but still poorly
sampled) sister speciesAu. anamensisand, to a
lesser degree, inAu. afarensis. The known dental
and mandibular remains ofAu. anamensisare
temporally and morphologically intermediate
between those ofAr. ramidusandAu. afarensis,
with variation that overlaps in both directions. Its
constellation of primitive and derived features of
the mandible, CP
3complex, lower dm
1(lower
first deciduous molar), and postcanine dentition
lends support to the hypothesis of an evolutionary
sequence ofAr. ramidus→Au. anamensis→Au.
afarensis(7,8,77). Circumstantial evidence
supporting this hypothesis is the temporal and
geographic position ofAr. ramidusdirectly
below the first known appearance ofAu.
anamensiswithin the Middle Awash succession.
Here, these taxa are stratigraphically super-
imposed in the same succession, separated by
~80 vertical meters representing ~200,000 to
300,000 years (7).Au. afarensisappears later in
the same sequence [3.4 Ma, at Maka (53)].
Therefore, at one end of a spectrum of phy-
logenetic possibilities,Ar. ramidusmay have been
directly ancestral to the more derived chronospe-
cies pairAu. anamensis→Au. afarensisacross
the full (still unknown, presumably African)
species range (7,8,77) (Fig. 5A). AlthoughAu.
afarensisis well represented in craniodental
remains and postcrania, its apparent earlier chro-
nospeciesAu. anamensisis still woefully under-
represented in both, and becauseAr. ramidusis so
far known only from limited time horizons and
locations, its last appearance, date, and potential
relationship to these later taxa are still indeter-
minate. Given the dramatic differences in post-
cranial anatomy seen in laterAustralopithecus
and hinted at in knownAu. anamensis,itseems
likely that a major adaptive shift marked the
Ardipithecus-to-Australopithecustransition (when-
ever and wherever the transition might have oc-
curred and whatever its population dynamics).
This transition may not have occurred through
Fig. 5.Geographic and temporal sparsity of early hominid fossils. Colored windows represent
presently available samples. Specific and subspecific relationships are currently impossible to
resolve because of limited available data. Depicted species lineages are gray“bundles”that
comprise sampled and hypothetical subspecific (populational; demic)“cords,”each with continuity
through time and reticulating with adjacent populations through gene flow. The slice at ~6 Ma
reveals the two known (red) samples of Late Miocene hominids (Chad and Kenya), schematized
here for simplicity within the same bundle, pending additional evidence (12 ).Au. afarensisis (so
far) sampled in the Ethiopian, Kenyan, Tanzanian, and Chadian (hidden behind the bundle)
regions. The Ethiopian Afar region has yielded four named, time-successive taxa, includingAr.
ramidus(yellow star). The close chronological and geographic proximity ofAr. ramidusandAu.
anamensiswithin the Middle Awash stratigraphic succession can be accommodated in different
stratophenetic arrangements, each with different predictions about future fossil discoveries.
Hypothesis 1 interprets all known evidence to represent a species lineage evolving phyletically
across its entire range. Hypothesis 2 depicts the same evidence in anArdipithecus-to-
Australopithecustransition (speciation) occurring between ~4.5 and ~4.2 Ma in a regional (or
local) group of populations that might have included either or both the Afar and Turkana rifts.
Hypothesis 3 accommodates the same evidence to an alternative, much earlier peripheral allopatric
“rectangular”speciation model (cladogenesis through microevolution accumulated in a peripheral
isolate population, becoming reproductively separated). Other possibilities exist, but at the present
time, none of these hypotheses can be falsified based on the available evidence. To choose among
them will require more fossil evidence, including well-documented transitions in multiple
geographic locales. See the text [and (7)] for details.
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
84
Ardipithecus ramidus
Ardipithecus ramidus

39. 39www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
pan-specific phyletic evolution (Fig. 5A). Figure 5
presents two other phylogenetic hypotheses that
are also, at present, impossible to falsify.
If diagnostic contemporary fossils ofAu.
anamensisare someday found in rocks of
>4.4 Ma, the hypothesis that the Afar popula-
tion ofAr. ramidusis the phyletic ancestor of
Au. anamensis(Fig. 5A, B) would be falsified.
In such an eventuality, AramisAr.ramidus
would represent a persisting relict population of
the mother species (Fig. 5C). Given the lack of
relevant fossils, it is currently impossible to de-
termine whether there was a geologically rapid
phyletic transition betweenArdipithecusand
Australopithecusin the Middle Awash or else-
where. Nevertheless, the morphological and
ecological transition between these two adaptive
plateaus is now discernible.
ArdipithecusandAustralopithecus.For
Darwin and Huxley, the basic order in which
human anatomies, physiologies, and behaviors
were assembled through time was unknown—
indeed unknowable—without an adequate fossil
record. They were forced to employ extant ape
proxies instead. The latter are now shown to be
derived in ways unrelated to the evolution of
hominids.
The Aramis fossils help clarify the origin of
the hominid clade (27,31), and reveal some
paleobiological dimensions of the first hominid
adaptive plateau (Ardipithecus). The primitive
characters ofAr. ramidussimultaneously provide
a new perspective on the evolutionary novelties
ofAustralopithecus.
Even in the wake of the Aramis and Gona
discoveries, the morphological envelopes, phy-
logenetic relationships, and evolutionary dynam-
ics of early hominid species remain incompletely
understood (Fig. 5). However, the paleobiology
ofAr. ramidus—even when viewed through its
geographically and temporally restricted Afar
samples—now reveals that the basal hominid
adaptive plateau comprised facultatively bipedal
primates with small brains, reduced nonhoning
canines, unspecialized postcanine dentitions, and
arboreally competent limb skeletons. Their eco-
logical niche(s) were probably more restricted—
and their geographic distribution(s) possibly
smaller and more disjunct—than those of later
hominid species and genera.
The derived postcranial elements ofAustra-
lopithecusprovide a strong contrast to their more
primitive homologs inArdipithecus(78). Rela-
tive to the generalized anatomy of the latter, the
highly evolved specializations of the foot, ankle,
knee, pelvis, wrist, and hand ofAu. afarensis
(79–81) indicate that this species lineage had
largely abandoned locomotion in the arboreal
canopy (and its resources).
Given the strong selection predicted to have
been associated with the emergence of new
ranging and feeding patterns inAustralopithecus,
the transition fromArdipithecustoAustralopithecus
could have been rapid, and anatomically par-
ticularly so in hindlimb structure. The forelimb
(especially the hand) was probably under less
intensive selection. It is possible that modifi-
cation of general cis-regulatory pathways may
have generated the striking and novel morphol-
ogy of the hindlimb, especially the foot, because
the autopod seems to be the most morphologi-
cally compliant to such mechanisms of mod-
ification. The dentognathic shifts could have
been more gradational, whatever the mode of
phylogenesis.
HomoandAustralopithecusare the only pri-
mates with nongrasping feet, and this particular
transformation was probably far-reaching, with
consequences for key behavioral constancies in
higher primates related to arboreal feeding
and nesting. Without stabilizing selection for
Ardipithecus-like arboreal capacities involving
slow and careful climbing, the foot, pelvis, and
thigh would have experienced directional selec-
tion to optimize bipedal locomotion during
prolonged walking (also in more limited running
bouts). With expanded ranging and social adap-
tations associated with terrestrial feeding in in-
creasingly open environments, the transition could
have been profound, but probably rapid, and there-
fore difficult to probe paleontologically.
One possible dynamic of anArdipithecus-
to-Australopithecustransition would have in-
volved microevolution within a deme or regional
group of demes. Being more ecologically flexi-
ble, the derived, potentially speciated populations
would have undergone rapid range expansion,
perhaps even encountering relictArdipithecus
populations. Unfortunately, the phylogeographic
details remain obscure given the poor spatial and
temporal resolution of the current fossil record
(Fig. 5). This provides a strong incentive for pur-
suing that record by actively increasing sampling
of sediments from different African basins with
dates between ~5 and ~3.5 Ma.
Currently,Australopithecusappears relatively
abruptly in the fossil record at about 4.2 Ma.
Relative toAr. ramidus, available earlyAustra-
lopithecusis now revealed to have been highly
derived: a committed biped with slightly enlarged
brain, a nongrasping arched foot, further derived
canines, substantially specialized postcanine teeth
with thick molar enamel, and expanded ecolog-
ical tolerances and geographic ranges. It is widely
recognized that this is the adaptive plateau
antecedent toHomo, which is now definable as
the third such major adaptive shift in human
evolution. Commitment to the terrestrial ranging
behaviors ofAustralopithecuswell before the
Pleistocene appear to have catalyzed the emer-
gence of what must have been even more highly
specialized social and ecological behaviors
remarkably elaborated in descendantHomo—
the ultimate global primate generalist.
Conclusions.Besides hominids, the only apes
to escape post-Miocene extinction persist today
as relict species, their modern distributions
centered in forested refugia. The markedly prim-
itiveAr. ramidusindicates that no modern ape is
a realistic proxy for characterizing early hominid
evolution—whether social or locomotor—as ap-
preciated by Huxley. Rather,Ar. ramidusreveals
that the last common ancestor that we share with
chimpanzees (CLCA) was probably a palmigrade
quadrupedal arboreal climber/clamberer that lacked
specializations for suspension, vertical climbing,
or knuckle-walking (24 –27). It probably retained a
generalized incisal/postcanine dentition associated
with an omnivorous/frugivorous diet less spe-
cialized than that of extant great apes (22,23). The
CLCA probably also combined moderate canine
dimorphism with minimal skull and body size
dimorphism (22,23), most likely associated with
relatively weak male-male agonism in a male
philopatric social system (22,23,31).
Ardipithecusreveals the first hominid adap-
tive plateau after the CLCA. It combined facul-
tative terrestrial bipedality (25,26) in a woodland
habitat (28–30) with retained arboreal capa-
bilities inherited from the CLCA (24–27). This
knowledge ofAr. ramidusprovides us, for the first
time, with the paleobiological substrate for the
emergence of the subsequentAustralopithecus
andHomoadaptive phases of human evolution.
Perhaps the most critical single implication of
Ar. ramidusis its reaffirmation of Darwin’s ap-
preciation: Humans did not evolve from chim-
panzees but rather through a series of progenitors
starting from a distant common ancestor that
once occupied the ancient forests of the African
Miocene.
References and Notes
1. C. Darwin,The Descent of Man, and Selection in Relation
to Sex(John Murray, London, 1871).
2. We here consider Hominidae to include modern humans
and all taxa phylogenetically closer to humans than to
Pan(common chimpanzee and bonobo), that is, all taxa
that postdate the split between the lineage leading to
modern humans and the lineage that led to extant
chimpanzees.
3. T. H. Huxley,Evidence as to Man’s Place in Nature
(London, 1863).
4. V. M. Sarich, A. C. Wilson,Proc. Natl. Acad. Sci. U.S.A.
58, 142 (1967).
5. D. C. Johanson, M. Taieb, Y. Coppens,Am. J. Phys.
Anthropol.57, 373 (1982).
6. M. D. Leakeyet al.,Nature262, 460 (1976).
7. T. D. Whiteet al.,Nature440, 883 (2006).
8. W. H. Kimbelet al.,J. Hum. Evol.51, 134 (2006).
9. G. WoldeGabrielet al.,Nature371, 330 (1994).
10. T. D. White, G. Suwa, B. Asfaw,Nature371, 306 (1994).
11. Y. Haile-Selassie,Nature412, 178 (2001).
12. Y. Haile-Selassie, G. Suwa, T. D. White, in Ardipithecus
kadabba: Late Miocene Evidence from the Middle Awash,
Ethiopia, Y. Haile-Selassie, G. WoldeGabriel, Eds.
(University of California, Berkeley, CA, 2009),
pp. 159–236.
13. B. Senutet al.,Comptes Rendus de l’Academie des Sciences,
Series IIA: Earth and Planetary Science332, 137 (2001).
14. M. Brunetet al.,Nature418, 145 (2002).
15. Y. Haile-Selassie, G. Suwa, T. D. White,Science303,
1503 (2004).
16. J. Moore, inGreat Ape Societies, W. C. McGrewet al., Eds.
(Cambridge Univ. Press, Cambridge, 1996), pp. 275–292.
17. B. G. Richmond, D. R. Begun, D. S. Strait,Yearb. Phys.
Anthropol.44, 70 (2001).
18. R. Wrangham, D. Pilbeam, inAll Apes Great and Small,
Volume 1: African Apes, B. Galdikaset al., Eds.
(Kluwer Academic/Plenum, New York, 2001), pp. 5–17.
19. J. T. Stern, R. L. Susman,Am. J. Phys. Anthropol.60, 279
(1983).
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 85
RESEARCH ARTICLESResearch Articles

40. 40 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
20. J. T. Stern,Evol. Anthropol.9, 113 (2000).
21. B. Latimer, inOrigines de la Bipedie chez les Hominides,
B. Senut, Y. Coppens, Eds. (Editions CNRS, Paris, 1991),
pp. 169–176.
22. G. Suwaet al.,Science326, 69 (2009).
23. G. Suwaet al.,Science326, 68 (2009).
24. C. O. Lovejoyet al.,Science326, 70 (2009).
25. C. O. Lovejoyet al.,Science326, 71 (2009).
26. C. O. Lovejoyet al.,Science326, 72 (2009).
27. C. O. Lovejoyet al.,Science326, 73 (2009).
28. G. WoldeGabrielet al.,Science326, 65 (2009).
29. A. Louchartet al.,Science326, 66 (2009).
30. T. D. Whiteet al.,Science326, 67 (2009).
31. C. O. Lovejoy,Science326, 74 (2009).
32. H. Gilbert, B. Asfaw (Eds.),Homo erectus: Pleistocene
Evidence from the Middle Awash, Ethiopia(Univ.
California Press, Berkeley, CA, 2008).
33. Y. Haile-Selassie, G. WoldeGabriel (Eds.),Ardipithecus
kadabba: Late Miocene Evidence from the Middle Awash,
Ethiopia(Univ. California Press, Berkeley, California,
2009).
34. P. R. Renne, G. WoldeGabriel, W. K. Hart, G. Heiken,
T. D. White,Geol. Soc. Am. Bull.111, 869 (1999).
35. In 1994, the Middle Awash project instituted“crawls”of
sedimentary outcrop between the GATC and DABT to
collect all available fossil material. Crawls were generally
upslope in direction, done by teams of 5 to 15 collectors
who crawled the surface on hands and knees, shoulder to
shoulder, collecting all fossilized biological materials
between a prescribed pair of taut nylon cords. Surfaces
were repeatedly collected with this technique, invariably
resulting in successively depressed specimen recovery
numbers in subsequent field seasons.
36. S. Semawet al.,Nature433, 301 (2005).
37. No surface or in situ fragments of theARA-VP-6/500
specimen are duplicate anatomical elements. Only 7.3%
of 136 total pieces (table S2) were surface recoveries at
the excavation site. All other pieces were excavated in
situ. Preservation is identical across the entire recovered
set. There is no evidence of multiple maturational ages
among the 136 pieces, and many of them conjoin. Given
the close stratigraphic and spatial association (Fig. 2),
and given no evidence of any other individual from the
carefully excavated spatiostratigraphic envelope, we
conclude that the parts of theARA-VP-6/500specimen
represent a single individual’s disarticulated skeleton.
38. S. Elton,J. Anat.212, 377 (2008).
39. A. K. Behrensmeyer,Paleobiology8, 211 (1981).
40. S. M. Kidwell, K. W. Flessa,Annu. Rev. Earth Planet. Sci.
24, 433 (1996).
41. D. Western, A. K. Behrensmeyer,Science324, 1061
(2009).
42. R. Pik, B. Marty, J. Carignan, J. Lavé,Earth Planet. Sci.
Lett.215, 73 (2003).
43. A. V. Fedorovet al.,Science312, 1485 (2006).
44. T. D. White, G. Suwa, B. Asfaw,Nature375, 88 (1995).
45. A. Hill, S. Ward,Yearb. Phys. Anthropol.31, 49 (1987).
46. M. G. Leakey, J. M. Harris, Eds.,Lothagam: The Dawn
of Humanity in Eastern Africa(Columbia Univ. Press,
New York).
47. J. Kappelmanet al.,Nature376, 558 (1995).
48. T. D. White, inThe Paleobiological Revolution: Essays on
the Growth of Modern Paleontology, D. Sepkoski,
M. Ruse, Eds. (Univ. of Chicago Press, Chicago, 2009),
pp. 121–148.
49. P. L. Reno, R. S. Meindl, M. A. McCollum, C. O. Lovejoy,
Proc. Natl. Acad. Sci. U.S.A.100, 9404 (2003).
50. M. Brunetet al.,Nature434, 752 (2005).
51. B. Wood,Nature418, 133 (2002).
52. S. J. Gould, R. C. Lewontin,Proc. R. Soc. London Ser. B.
205, 147 (1979).
53. C. O. Lovejoy, R. S. Meindl, J. C. Ohman, K. G. Heiple,
T. D. White,Am. J. Phys. Anthropol.119, 97 (2002).
54. C. O. Lovejoy, M. J. Cohn, T. D. White,Proc. Natl. Acad.
Sci. U.S.A.96, 13247 (1999).
55. M. A. McCollumet al.,J. Exp. Zool. B Mol. Dev. Evol.312,
published online 17 August 2009; 10.1002/jez.b.21316.
56. T. H. Huxley,Westminster Rev.73, 541 (1860).
57. L. de Bonis, G. D. Koufos,Evol. Anthropol.3, 75 (1994).
58. D. R. Begun,Anthropol. Sci.113, 53 (2005).
59. J. Palfyet al.,Earth Planet. Sci. Lett.258, 160 (2007).
60. D. Pilbeam, N. Young,C. R. Palevol3, 305 (2004).
61. D. S. Stern, V. Orgogozo,Science323, 746 (2009).
62. M. D. Shapiro, M. A. Bell, D. M. Kingsley,Proc. Natl.
Acad. Sci. U.S.A.103, 13753 (2006).
63. N. Shubin, C. Tabin, S. Carroll,Nature457, 818
(2009).
64. P. F. Colosimoet al.,Science307, 1928 (2005).
65. L. Hlusko, G. Suwa, R. T. Kono, M. C. Mahaney,
Am. J. Phys. Anthropol.124, 223 (2004).
66. M. J. F. Pulquerio, R. A. Nichols,Trends Ecol. Evol.22,
180 (2007).
67. N. Elango, J. W. Thomas, S. V. Yi,Proc. Natl. Acad. Sci. U.S.A.
103, 1370 (2006).
68. R. J. Britten,Proc. Natl. Acad. Sci. U.S.A.99, 13633
(2002).
69. G. Suwa, R. T. Kono, S. Katoh, B. Asfaw, Y. Beyene,
Nature448, 921 (2007).
70. N. Patterson, D. J. Richter, S. Gnerre, E. S. Lander,
D. Reich,Nature441, 1103 (2006).
71. D. S. Strait, F. E. Grine,J. Hum. Evol.47, 399 (2004).
72. E. Trinkaus,Am. J. Phys. Anthropol.83, 1 (1990).
73. For example, it has been noted that these methods fail to
accurately resolve relationships of modern hominoid
species without sufficient intermediate forms from a fossil
record (71).
74. Enamel thickness ofAr. ramidusmolars ranges largely
from what would traditionally be termed“intermediate
thin”to“intermediate thick”categories. Lacking the
derived thickness pattern ofPan, it forms a suitable
ancestral condition for laterAustralopithecus. The
ubiquitous single-rooted lower fourth premolar (P
4) in
known Aramis and GonaAr. ramidusis notable, but this
is also a known variation ofAu. anamensisand
A. afarensis. Judging from the clear dominance of
double-rooted lower P
4’s inAu. afarensis(and thereafter
an increasing robusticity of the roots themselves in
Australopithecus), either there was selection for larger,
more complex premolar root systems or such morpholo-
gies emerged as pleiotropy of postcanine enhancement.
Without such selection,Ar. ramidusas a species probably
contained regional populations that varied in premolar
root number (22).
75. B. Senut, M. Pickford,C. R. Palevol.3, 265 (2004).
76. H. Gee,Deep Time: Cladistics, the Revolution in Evolution
(Free Press, London, 1999).
77. C. V. Ward, A. C. Walker, M. G. Leakey,Evol. Anthropol.
7, 197 (1999).
78. We use genera to express both phyletic proximity and
circumscribed adaptive systems, with ecobehavioral and
morphological conditions being integral parts of the
latter. This use employs the broadly defined genus
Australopithecus, without recognizing the now commonly
usedParanthropus(82). This is because both“robust”
and“nonrobust”Australopithecusspecies are characterized
by a commonly derived heavy masticatory apparatus (albeit
to differing degrees), and also because we cannot—even
to this day— be certain that the“robust”species are
monophyletic.
79. C. O. Lovejoy,Gait Posture21, 95 (2005).
80. C. O. Lovejoy,Gait Posture21, 13 (2005).
81. C. O. Lovejoy,Gait Posture25, 325 (2007).
82. T. D. White, inThe Primate Fossil Record
, W. Hartwig, Ed.
(Cambridge Univ. Press, Cambridge), pp. 407–417.
83. For funding, we thank NSF (this material is based on
work supported by grants 8210897, 9318698, 9512534,
9632389, 9729060, 9727519, 9910344, and 0321893
HOMINID-RHOI), the Institute of Geophysics and
Planetary Physics of the University of California at Los
Alamos National Laboratory (LANL), and the Japan
Society for the Promotion of Science. D. Clark and
C. Howell inspired this effort and conducted laboratory
and field research. We thank the coauthors of the
companion papers (22-30), with special thanks to the
ARA-VP-6/500and-7/2excavation teams, including
A. Amzaye, the Alisera Afar Clan, Lu Baka, A. Bears,
D. Brill, J. M. Carretero, S. Cornero, D. DeGusta,
A. Defleur, A. Dessie, G. Fule, A. Getty, H. Gilbert,
E. Güleç, G. Kadir, B. Latimer, D. Pennington, A. Sevim,
S. Simpson, D. Trachewsky, and S. Yoseph. G. Curtis,
J. DeHeinzelin, and G. Heiken provided field geological
support. D. Helgren, D. DeGusta, L. Hlusko, and
H. Gilbert provided insightful suggestions and advice.
We thank H. Gilbert, K. Brudvik, L. Bach, D. Paul,
B. Daniels, and D. Brill for illustrations; G. Richards and
A. Mleczko for imaging; the Ministry of Tourism and
Culture, the Authority for Research and Conservation of
the Cultural Heritage, and the National Museum of
Ethiopia for permissions and facilitation; and the Afar
Regional Government, the Afar people of the Middle
Awash, and many other field workers who contributed
directly to the research efforts and results.
Supporting Online Material
www.sciencemag.org/cgi/content/full/326/5949/64/DC1
SOM Text
Tables S1 and S2
References
4 May 2009; accepted 8 September 2009
10.1126/science.1175802
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org86
Ardipithecus ramidus
Ardipithecus ramidus

41. 41www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
The Geological, Isotopic, Botanical,
Invertebrate, and Lower Vertebrate
Surroundings ofArdipithecus ramidus
Giday WoldeGabriel,
1
*Stanley H. Ambrose,
2
Doris Barboni,
3
Raymonde Bonnefille,
3
Laurent Bremond,
4
Brian Currie,
5
David DeGusta,
6
William K. Hart,
5
Alison M. Murray,
7
Paul R. Renne,
8
M. C. Jolly-Saad,
9
Kathlyn M. Stewart,
10
Tim D. White
11
*
Sediments containingArdipithecus ramiduswere deposited 4.4 million years ago on an alluvial
floodplain in Ethiopia’s western Afar rift. The Lower Aramis Member hominid-bearing unit, now
exposed across a >9-kilometer structural arc, is sandwiched between two volcanic tuffs that have
nearly identical
40
Ar/
39
Ar ages. Geological data presented here, along with floral, invertebrate, and
vertebrate paleontological and taphonomic evidence associated with the hominids, suggest that
they occupied a wooded biotope over the western three-fourths of the paleotransect. Phytoliths and
oxygen and carbon stable isotopes of pedogenic carbonates provide evidence of humid cool
woodlands with a grassy substrate.
A
rdipithecus ramidusand abundant asso-
ciated faunal and floral fossils were
recovered from sedimentary rocks in the
Central Awash Complex (CAC) of the Middle
Awash study area. The CAC is a complexly
faulted dome centered 25 km east of the western
rift margin, Afar, Ethiopia. Today, 300 m of strata
deposited between 5.6 and 3.9 million years ago
(Ma) are exposed in the CAC (1). The hominid-
bearing Lower Aramis Member of the Sagantole
Formation lies midway in this stratigraphic
succession and crops out along an erosional arc
>9 km across, extending from the Ounda Sagantole
drainage in the southeast to Aramis Locality 6 in
the north, and to Kuseralee Locality 2 to the
southwest (Fig. 1). Here, we describe regional
and local geology, and present isotopic, paleo-
botanical, invertebrate, and lower vertebrate
fossil evidence that illuminates local conditions
at the time the vertebrate fossils were deposited.
Geology.The Aramis Member directly over-
lies the Gàala (“Camel”) Tuff Complex (GATC),
which has a
40
Ar/
39
Ar age of 4.419T0.068 Ma
(2,3). This vitric tuff is 0.5 to 2 m thick and is
rich in pumice and crystals. The Aramis Member
includes light salmon (hue 5YR) to deep red-
brown silt, clay, and sand of variable thickness
and induration, deposited on a floodplain. These
strata show a general increase in thickness toward
the east, and they range from an average of 3 m
up to 6 m. They are overlain by the Daam Aatu
(“Baboon”) Basaltic Tuff (DABT), which has a
40
Ar/
39
Ar age of 4.416T0.031 Ma (1). We define
the Lower Aramis Member as the entirety of
both tuffs and all sediments between them.
A patchwork of variably fossiliferous localities
along the outcrop arc has yielded a combined
total of more than 6000 individually cataloged
vertebrate fossils between the widespread volcanic
marker horizons. The vertebrate assemblages are
in close association with sedimentological and
structural information, botanical and invertebrate
fossils, and oxygen and carbon isotopic data on
pedogenic carbonates in soil horizons. Integra-
tion of these data allows reconstruction of the
physical and biological aspects of the depositional
setting.
The Middle Awash was a persistent sedimen-
tary basin during the Pliocene (4 ). The basin axis
during deposition of the entire Aramis Member
was southeast of theArdipithecuslocalities, as
evidenced by deltaic and lake margin deposits
generally disposed to the southeast, in the direc-
tion of paleocurrent orientations and erosional
features. Active volcanic centers of the CAC were
located to the south (1). Paleoenvironmental data
and structural reconstructions suggest that the
overall elevation may have been greater than
today’s ~600 m, although kinematic models are
equivocal (5,6).
The lower tuff (GATC) is underlain by a
widespread cobble conglomerate. In the central
part of the exposure arc, fossiliferous Lower
Aramis Member sediments comprise predomi-
nantly massive and bioturbated silty clays deposited
primarily on a low-relief floodplain far from the
main river channel(s). Reworked GATC pumices
and glass are present locally, but evidence of
channels is limited to rare sandstone lenses gen-
erally situated below the fossiliferous strata.
Massive (<1.5 m thick), predominantly micritic
carbonate horizons and nodules representing
groundwater and pedogenic deposits pinch out
laterally within clayey silts. These are also locally
fossiliferous.
Carbonate deposits in some localities contain
characteristic features of tufas (7), such as fossil
gastropods and other invertebrates, abundant and
uncrushed calcite-replaced vegetation, vertebrate
remains, and eggshells (guinea-fowl size). These
suggest that the carbonate horizons generally
formed at or near the landscape surface. Evidence
of spring activity includes several 1-m-wide
banded travertine deposits associated with faults.
A porous microcrystalline carbonate with dense
concentrations of calcite isomorphs of plant parts
forms a broad, low dome just north of ARA-VP-6.
However, in almost all sections excavated for
isotopic, phytolith, and pollen analysis, the car-
bonates lack diagnostic features of tufas. Their
micritic textures and the presence of terrestrial
soil invertebrate faunal activity (such as dung
beetle brood burrows) suggest that the carbonate
horizons are derived from groundwater carbonate
that generally formed at or near the landscape
surface in seasonally saturated soils near springs
(fig. S1).
Paleosols in the Lower Aramis Member are
primarily protosols and calcisols (8 ) and are
most strongly developed just below the DABT.
Protosols are 10 to 200 cm thick, with massive to
single-grain structure and abundant root struc-
tures. Calcisols are 50 to 100 cm thick and con-
tain subsurface horizons (Bk) displaying massive
to angular blocky ped structures, weakly devel-
oped argillic cutans, and small (<5 mm across)
calcareous nodules and tubules. The best-developed
calcisols are present directly below the DABT.
These nodules are most abundant to the north and
may imply that it was slightly drier there.
Sections at the southeast end of the exposure
arc show weaker soil development, perhaps
implying deposition in a predominantly wetter,
more axial environment.
The DABT is poorly consolidated. The lower
third (~15 to 20 cm) of the unit is composed of
bedded and laminated gray basaltic glass lapilli
and scoria. Given the geometry of its basal con-
tact and lack of underlying incision, we interpret
the DABT to have fallen across a dominantly low-
relief landscape. Reworked clasts of the DABT
are seen in only a few small shallow channels.
The presence of paleosols and minor channels
before deposition of the DABT indicates that
sedimentation was intermittent and that streams
only slightly incised this part of the basin. Sed-
iments immediately overlying the DABT near
RESEARCHARTICLES
1
Earth Environmental Sciences Division, Los Alamos National
Laboratory, Los Alamos, NM 87545, USA.
2
Department of
Anthropology, University of Illinois, Urbana, IL 61801, USA.
3
CEREGE (UMR6635 CNRS/Université Aix-Marseille), BP80,
F-13545 Aix-en-Provence Cedex 4, France.
4
Center for Bio-
Archaeology and Ecology (UMR5059 CNRS/Université Mont-
pellier 2/EPHE), Institut de Botanique, F-34090 Montpellier,
France.
5
Department of Geology, Miami University, Oxford, OH
45056, USA.
6
Department of Anthropology, Stanford Uni-
versity, Stanford, CA 94305, USA.
7
Department of Biological
Sciences, University of Alberta, Edmonton, Alberta T6G 2E9,
Canada.
8
Berkeley Geochronology Center, 2455 Ridge Road,
Berkeley, CA 94709, USA, and Department of Earth and
Planetary Science, University of California, Berkeley, CA
94720, USA.
9
Université Paris-Ouest La Défense, Centre Henri
Elhaï, 200 Avenue de la République, 92001 Nanterre, France.
10
Paleobiology, Canadian Museum of Nature, Ottawa, Ontario
K1P 6P4, Canada.
11
Human Evolution Research Center and
Department of Integrative Biology, 3101 Valley Life Sciences
Building, University of California, Berkeley, CA 94720, USA.
*To whom correspondence should be addressed. E-mail:
[email protected] (T.D.W.); [email protected] (G.W.)
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 65e1
Research Articles

42. 42 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Carbonate
Sandy silt
Silt
Silt loamSand (massive) Sand (cross-
bedded)
Burrows
Granular loam Clay loam
Root marks
Root casts
(carbonate)
GATC
Gàala Vitric
Tuff Complex
WOBT
Wodara Basaltic
Tuff
DABT
Daam Aatu
Basaltic Tuff
Carbonate
nodules
Isotope sample
Carbonate δ
13
C (‰)
Carbonate δ
18
O (‰)
Fig. 1.Satellite image of the ~9-km erosional arc exposing theArdipithecus-bearing GATC-DABT horizon of the Central Awash Complex of the Middle Awash
study area, Ethiopia. Isotopic data are shown. Values in the eastern sample sites indicate slightly more open habitat (where primate fossils were notfound), a
finding consistent with the macrobotanical and paleontological evidence.
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
65e2
Ardipithecus ramidus
Ardipithecus ramidus

43. 43www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
the ARA-VP-1 (TS), ARA-VP-7, and ARA-VP-
10 localities contain massive to weak, blocky,
root-marked protosols overlain by alternating mas-
sive clayey silt and fine, well-sorted sand with
tubular, subvertical 1- to 2-cm burrows. These
features indicate that the water table was high and
that the floodplain was aggrading more rapidly
after deposition of the tuff, in accordance with
faunal evidence that more aquatic and water-
dependent mammals were more abundant above
the DABT.
The time span represented by the fossiliferous
sediments of the Lower Aramis Member (be-
tween the two tuffs) is difficult to ascertain. The
dates for the two tephras are statistically in-
distinguishable, the difference between them
only 0.003T0.075 million years at 68% con-
fidence. Thus, the dates suggest that, most prob-
ably, this interval represents a few thousand to
perhaps at most 100,000 years. Paleosols in sim-
ilar aggrading distal floodplain environments
often indicate geologically short time spans.
The fossil assemblages collected from between
the two tuffs are consistent with environmental
stability during the interval, and no evolutionary
trends are evident (9). On the basis of analogous
settings, these sediments probably represent dep-
osition within 100 to 10,000 years (10 ) and their
paleontological contents would qualify as a“within-
habitat time-averaged assemblage”(11).
The depositional environment of theA r.
ramidusfossils in the CAC differs somewhat
from that of penecontemporaneous Gona fossils
~70 km to the northwest. The Gona conspecifics
were recovered in mixed-habitat faunas along the
western basin margin where lake deposits inter-
finger with small fluvial channels or lap onto ac-
tive basaltic cones and flows (12).
Stable isotopes.To further elaborate the
conditions surrounding deposition of the hominid
remains between the two tuffs, we analyzed car-
bon and oxygen isotopes (13,14) from paleosol
carbonate, as well as carbon isotopes from
associated organic matter (Fig. 1, figs. S1 to S3,
and tables S1 and S2) (15). Plants using the C
3
photosynthetic pathway (such as trees, shrubs,
and most herbaceous dicots; shaded forest under-
story; and cold-adapted tropical alpine and high-
latitude grasses) have averaged
13
C values of
–26.5 per mil (‰). Tropical savanna grasses using
the C
4pathway have averaged
13
C values of
–12.5‰. Decomposing plant organic matter la-
bels the soil with a similar isotopic composition
(16). Disseminated organic carbon is present in
trace amounts in ancient soils formed on volcanic
parent materials, mainly in the allophane clay
fraction (17 ), and in carbonate nodules (18).
Trace amounts of organic carbon contamination,
mainly from C
3-based petroleum products intro-
duced after excavation, can substantially lower
thed
13
C values of soils formed in paleoenviron-
ments with C
4plant biomass. We are confident
that the procedures used to minimize contamina-
tion (table S2) (15) permit accurate reconstruc-
tion of the Aramis Member plant biomass isotopic
composition.
Soil carbonated
13
C values are typically en-
riched by 14 to 17‰relative to those of organic
matter (19 ). Lower Aramis Member organicd
13
C
values range from–21‰to–15‰, and carbonate
d
13
C values range from–6.5‰to–0.5‰(table
S1). The mean difference between carbonate and
included disseminated organic matterd
13
C values
(D
13
C) is 13.8‰ and the median is 14.3‰. This
is within the range expected for well-preserved
paleosols (19). Oxygen isotope ratios of pedo-
genic carbonate nodules reflect those of soil wa-
ter. The isotopic composition of meteoric waters
is controlled by polar ice volume, altitude, temper-
ature, humidity, and evapotranspiration (20–22).
Preferential evaporation of isotopically“light”
water (H
2
16O) leads to isotopic enrichment of
remaining water in near-surface soils (20). Ped-
ogenic carbonated
18
O values are thus highest
in hot, arid habitats and at low latitudes and
altitudes (14,16).
These data reflect woodland to grassy wood-
land savanna floral habitats with 30% to 70% C
4
plants. Carbonated
13
C andd
18
O values increase
axially from west to east across this outcrop arc
(Fig. 1). These data are consistent with sedimen-
tological, taphonomic, paleobotanical, and pale-
ontological indicators in suggesting more open,
exposed, probably grass-dominated habitats for
the localities at the eastern pole of the erosional
arc of localities (toward the paleo-depocenter)
Fig. 2.Fossilized botanical re-
mains from the Lower Aramis Mem-
ber. Wood and seeds are ubiquitous
at theArdipithecus-bearing local-
ities of the Lower Aramis Member.
(A) Silica bodies (phytoliths). The
bilobate (Bi) and polylobate (Po)
types are from grasses (Poaceae);
the globular echinate (GE) is from
palms (Palmae). Scale bar, 10mm.
(B) Fossil wood. Scale bar, 2 cm.
(C) Tangential microscopic section:
general view showing disposition of
rays of fossil wood specimen from
ARA-VP-6, identified as the fig tree
Ficoxylonsp. Scale bar, 108mm.
(D) FossilizedCeltis(hackberry)
seeds. Scale bar, 2 cm.
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 65e3
RESEARCH ARTICLESResearch Articles

44. 44 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
(9,23). None of the primate fossils, micromam-
mals, birds, or macrobotanical remains were found
in the 2.5-km stretch of Lower Aramis sediment
outcrops southeast of the easternmostArdipithecus-
bearing locality (SAG-VP-7).
Paleobotany.Many carbonate horizons be-
tween the two tuffs contain abundant calcite-
replaced wood and endocarps (Fig. 2). In the
noncemented sediments, these macrobotanical
remains are typically decalcified and sometimes
appear during excavation as white streaks or
manganese stains. The ubiquitous fossil wood
generally lacks the internal structure needed to
achieve reliable taxonomic identification, except
for one specimen attributed to the figFicoxylon.
Endocarps of hackberry fruits attributed toCeltis
sp. (24) are well preserved and abundant, but
there is an inherent preservational bias because
these are easily fossilized (25 ,26). Therefore,
these trees cannot be assumed to have dominated
the vegetation.Celtistrees are tolerant of a wide
range of environmental conditions; their im-
mature leaves are eaten by chimpanzees (27).
We analyzed a variety of silt, clays, and
carbonate samples (including splits of the iso-
topic samples) for pollen, but found none. How-
ever, a few grains were extracted from sediment
within theARA-VP-1/401mandible, as well as
from carbonate matrix encasing seeds and two
coprolites from ARA-VP-6. The ARA-VP-6
grains are attributed toMyrica(n= 6),Borassus/
Hyphaene(n= 2), Poaceae grass (n= 4), and
Cyperaceae (n= 2), and the ARA-VP-1 grains to
Borassus/Hyphaene(n= 2).Myrica,Celtis, and
palm tree pollen was recovered both below and
above the Lower Aramis Member, and these trees
were part of the arboreal vegetation in the region
between 4 and 5 Ma. Contemporaneous pollen
data recovered from marine sediment deposited in
the Gulf of Aden indicate that these trees were
widespread and are associated with other compo-
nents of today’ s afromontane flora (28).
Given the paucity of pollen, we extracted and
analyzed phytoliths in the same samples (Fig. 2,
fig. S4, and table S3) (15). Phytoliths produced
by grasses (Poaceae) differ in shape and size
from those produced by other plants, including
woody dicotyledons (most trees and shrubs) and
palms (29). Their abundance, relative to the
abundance of globular phytoliths from woody
dicots and palms, has been used to estimate tree
cover on the basis of reference assemblages in
modern soils (30). We recovered and recognized
96 phytolith types from 38 samples from the
strata between the two tuffs. Types included
globular ones typical of palms [globular echinate
(31)], woody dicotyledons [globular granulate
and smooth, produced by trees and shrubs (32)],
grass silica short cells (64 different types), grass
bulliform and hair cells, and some rare sclerids
and tracheids. The relative abundance of globular
phytoliths over grass phytoliths suggests an open
grassland southeast of SAG-VP-7, with a max-
imum tree cover of <40%. West of SAG-VP-7,
an estimated maximum tree cover of ~65% is
suggested by some samples at KUS-VP-2, ARA-
VP-10, and ARA-VP-1 (TS).
On the basis of the geology, phytoliths, fos-
sils, and isotopic data, we infer that the local
Pliocene vegetation included abundant palms
and trees or shrubs as well as grasses (as would
be characteristic of semi-deciduous woodlands
and open forests for at least part of the year).
Palms were present at all localities over the 9-km
outcrop arc but were probably most abundant
(despite potential overrepresentation) near the
ARA-VP-1 and SAG-VP-7 localities. There is no
evidence for lowland humid Guineo-Congolian
rainforest, subdesertic arid vegetation, or high-
land C-3 Pooideae grasses (15) in the Lower
Aramis Member or from younger or older sedi-
ments of the CAC. Evidence forCeltis,Myrica,
and palm trees fits well with the presence of a
groundwater-supported grassy woodland to forest.
Invertebrate fossils.The invertebrate fauna
of the GATC-DABTAr. ramidus–bearing bio-
tope (especially ARA-VP-1 and -6) includes
fossilized insect larvae, dung beetle broodballs
and nests, diverse gastropods, millipedes, and a
small centipede (Fig. 3). The millipedes belong
to Spirostreptida, a large order common in a wide
variety of modern African habitats from savan-
na to forests. Pupal cases (calcite-replaced inclu-
sions weathered from carbonate) are common,
but taxonomic identification has not proven
possible. Dung beetle broodballs from the Lower
Aramis Member average 3 to 5 cm in diameter
and have walls 5 to 7 mm thick. Up to 15 of these
have been found together in chambers excavated
into host deposits (fig. S1). Rare larger balls up to
7 cm in diameter indicate a larger species. Dung
beetles exist wherever large mammals occur in
Africa (33 ).
Terrestrial gastropods are useful indicators of
vegetation patterns, rainfall, altitude, and temper-
ature (34). Their fragile shells do not withstand
hydraulic transport well, so given their embed-
ding fine-grain sediment of the Lower Aramis
F
B
C
E
D
A
Fig. 3.Fossilized invertebrate remains and traces from the Lower Aramis Member. (A) Gastropods: ARA-VP-6Maizaniasp. (B) Gastropods: ARA-VP-6Limicolariasp.
(C) Millipedes: ARA-VP-1. (D) Larvae: ARA-VP-6. (E) ARA-VP-6 centipede. (F) Dung beetle broodballs and nest, ARA-VP-6. Scale bar, 5 cm.
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
65e4
Ardipithecus ramidus
Ardipithecus ramidus

45. 45www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Member, the overall land snail assemblage can be
considered to be locally derived (35). The pres-
ence of millipedes, insect cocoons, and solitary
bee brood cells reinforces this conclusion. The
most common gastropod (34 of 40 identified
specimens) isMaizaniafrom theM.hildebrandti
group, followed byLimicolariasp. (n= 5), and a
single specimen ofChlamydarioncf.hians. This
Aramis land snail assemblage resembles that of
modern groundwater forests, such as the Kibwezi
in Kenya (34). This lowland forest in a regionally
semi-arid area thrives because it has a high water
table, an analogy consistent with geological evi-
dence from the Lower Aramis Member. Before
poaching, its larger mammal composition appears
to have been structured analogously to the sim-
ilarly primate-rich Aramis assemblage.
Lower vertebrates.Of 275 identified Lower
Aramis Member fish specimens, the dominant
genus is the catfishClarias(n= 175), followed
byBarbus(n= 20) and the family Cichlidae (n=
21). These are shallow-water fish, the former
capable of tolerating highly deoxygenated waters
and a wide temperature range.
The giant terrestrial tortoiseGeocheloniais
present, along withPelusios(African mud turtle
or hinged terrapin), Cyclanorbinae (flapshell
softshell turtle), andPelomedusa(African hel-
meted turtle). Crocodiles are present and repre-
sented by scutes and teeth indistinguishable from
those of extant Nile crocodiles. Remains of
lizards, snakes, and frogs were recovered, par-
ticularly at the ARA-VP-6 microfauna quarry.
The Lacertilia sample includes representatives of
Varanidae (cf.Varanus) and Iguanidae (cf. cha-
meleons, among others). The Serpentes sample
includes at least cf. Pythoninae. Given the marked
differences in habitat preferences typically seen
even within genera of lizards and snakes, their
use in paleoecological reconstruction is limited.
Chelonian, crocodylian, and osteichthyan
skeletal elements are readily recognizable even
when highly fragmentary, so their abundance
(number of identified specimens) in a fossil as-
semblage is almost always inflated relative to
their ecological abundance. Even so, these taxa
are rare in the Lower Aramis Member vertebrate
assemblage relative to other Middle Awash
strata and other Pliocene hominid localities
(except Laetoli). Most of these aquatic species
presumably appeared episodically on the Aramis
floodplain during times of over-bank flooding,
although it is possible that some fish represent
raptor meals—an interpretation supported by the
lack of articulated elements of these taxa at
Ardipithecuslocalities.
The relatively thin Lower Aramis Member
stratigraphic interval, exposed in an erosional 9-km
arc of localities between the rift margin to the
west and the basin axis to the east, provides a
paleotransect through a 4.4-Ma Pliocene land-
scape. The largely aggradational plain centered at
Aramis and adjacent to the low northern slopes of
the emerging CAC was subject to alluvial
flooding that embedded a rich faunal and floral
community containingArdipithecus. Evidence of
wooded environment is present and seems to
have prevailed during this interval across the
Ardipithecus-bearing localities that constitute
the western three-fourths of this transect (9,23).
The demonstrable co-occurrence of elements of
a wooded biotope—including large and small
mammals, birds, soil isotopes, gastropods, and
micro- and macrobotanical remains—suggests
that this floodplain, and the community it sup-
ported, constituted a wooded biotope rather than
that of a grassland savanna in the localities that
containArdipithecus.
References and Notes
1. P. R. Renne, G. WoldeGabriel, W. K. Hart, G. Heiken,
T. D. White,Geol. Soc. Am. Bull.111, 869 (1999).
2. G. WoldeGabrielet al.,Nature371, 330 (1994).
3. T. D. Whiteet al.,Nature440, 883 (2006).
4. Y. Haile-Selassie, G. WoldeGabriel, Ardipithecus kadabba:
Late Miocene Evidence from the Middle Awash, Ethiopia
(Univ. of California Press, Berkeley, 2009).
5. M. Hailemichael, J. Aronson, S. Savin, M. Tevesz, J. Carter,
Palaeogeogr. Palaeoclimatol. Palaeoecol.186, 81
(2002).
6. T. F. Redfield, W. H. Wheeler, M. Often,Earth Planet. Sci.
Lett.216, 383 (2003).
7. T. D. Ford, H. M. Pedley,Earth Sci. Rev.41, 117
(1996).
8. G. H. Mack, W. C. James, H. C. Monger,Geol. Soc. Am.
Bull.105, 129 (1993).
9. T. D. Whiteet al.,Science326, 67 (2009).
10. A. K. Behrensmeyer,Paleobiology8, 211 (1982).
11. S. M. Kidwell, K. W. Flessa,Annu. Rev. Earth Planet. Sci.
24, 433 (1996).
12. J. Quadeet al.,Geol. Soc. Am. Spec. Pap. 446(2009).
13. S. H. Ambrose, N. E. Sikes,Science253, 1402 (1991).
14. T. E. Cerling, inPalaeoweathering, Palaeosurfaces
and Related Continental Deposits (Special Publication
of the International Association of Sedimentologists),
M. Thiry, R. Simon-Coincon, Eds. (Blackwell, Oxford,
1999), pp. 43– 60.
15. See supporting material onScienceOnline.
16. T. W. Bouttonet al.,Geoderma82, 5 (1998).
17. A. Chadwick, L. A. Derry, P. M. Vitousek, B. J. Hubert,
L. O. Hedin,Nature397, 491 (1999).
18. M. E. Morgan, J. D. Kingston, B. D. Marino,Nature367,
162 (1994).
19. T. E. Cerling,Earth Planet. Sci. Lett.71, 229 (1984).
20. H. Craig,Science133, 1833 (1961).
21. J. R. Gat, inHandbook of Environmental Isotope
Geochemistry, Vol. 1, P. Fritz, J. Ch. Fontes, Eds. (Elsevier,
Amsterdam, 1980), pp. 21– 47.
22. G. W. Darling, A. H. Bath, J. J. Gibson, K. Rozanski, in
Isotopes in Paleoenvironmental Research, M. J. Leng, Ed.
(Springer, Dordrecht, Netherlands, 2005), pp. 1–66.
23. A. Louchartet al.,Science326, 66 (2009).
24. An earlier paper on Aramis (2) attributed these endocarps
toCanthiumon the basis of identifications by the late
R. Dechamps of Belgium. Subsequent work by
R.B. revised the identification.
25. G. J. Retallack,J. Hum. Evol.29, 53 (1995).
26. A. H. Jahren, M. L. Gabel, R. Amundson,Palaeogeogr.
Palaeoclimatol. Palaeoecol.138, 259 (1998).
27. R. W. Wrangham, M. E. Rogers, G. Ibasuta,Afr. J. Ecol.
31, 49 (1993).
28. R. Bonnefille, R. Potts, F. Chalié, D. Jolly, O. Peyron,
Proc. Natl. Acad. Sci. U.S.A.101, 12125 (2004).
29. D. R. Piperno,Phytolith Analysis: An Archaeological and
Geological Perspective(Academic Press, San Diego, CA,
1988).
30. D. Barboni, L. Bremond, R. Bonnefille,Palaeogeogr.
Palaeoclimatol. Palaeoecol.246, 454 (2007).
31. F. Runge,Rev. Palaeobot. Palynol.107, 23 (1999).
32. G. Scurfield, C. A. Anderson, E. R. Segnit,Aust. J. Bot.22,
211 (1974).
33. I. Hanski, Y. Cambefort, Eds.,Dung Beetle Ecology
(Princeton Univ. Press, Princeton, NJ, 1991).
34. M. Pickford,J. Afr. Earth Sci.20, 167 (1995).
35. Sparse contaminant lacustrine gastropod taxa are
attributable to recent transport of bioclastic sandstone
blocks from younger outcrops to the GATC-DABT package
by local people building houses, and via construction of
theARA-VP-6/500marker platform.
36. Supported by NSF grants 8210897, 9318698, 9512534,
9632389, 9910344, and 0321893 HOMINID-RHOI; the
Institute of Geophysics and Planetary Physics of the
University of California at Los Alamos National Laboratory
(LANL); and the Philip and Elaina Hampton Fund for
Faculty International Initiatives at Miami University. The
Earth and Environmental Sciences Division Electron
Microprobe laboratory at LANL assisted with access and
use. We thank G. Curtis for fieldwork, insight, and
inspiration; M. Pickford, H. Hutchinson, and W. Shear for
gastropod, chelonian, and millipede identifications,
respectively; M. Buchet and X. Prasad for pollen
preparations and microscopic observations;
M. Dupéron-Laudouaneix and J. Dupéron for
identification of fossil wood; K. Brudvik and
H. Gilbert for illustrations; J. Quade, N. Levin, and
S. Semaw for discussion and comparative data; the
Ministry of Tourism and Culture, the Authority for
Research and Conservation of the Cultural Heritage, and
the National Museum of Ethiopia for permissions and
facilitation; and the Afar Regional Government, the Afar
people of the Middle Awash, and many other field
workers for contributing directly to the research efforts.
Supporting Online Material
www.sciencemag.org/cgi/content/full/326/5949/65/DC1
SOM Text
Figs. S1 to S4
Tables S1 to S3
References
4 May 2009; accepted 14 August 2009
10.1126/science.1175817
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 65e5
RESEARCH ARTICLES
Research Articles

46. 46 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Taphonomic, Avian, and
Small-Vertebrate Indicators of
Ardipithecus ramidusHabitat
Antoine Louchart,
1
Henry Wesselman,
2
Robert J. Blumenschine,
3
Leslea J. Hlusko,
4
Jackson K. Njau,
4
Michael T. Black,
5
Mesfin Asnake,
6
Tim D. White
4
*
Thousands of vertebrate specimens were systematically collected from the stratigraphic interval
containingArdipithecus ramidus. The carcasses of larger mammals were heavily ravaged by carnivores.
Nearly 10,000 small-mammal remains appear to be derived primarily from decomposed owl pellets.
The rich avifauna includes at least 29 species, mostly nonaquatic forms. Modern analogs of the
most abundant birds and of a variety of rodents are associated with mesic woodland environments distant
from large water bodies. These findings support inferences from associated geological, isotopic,
invertebrate, and large-vertebrate assemblages. The combined results suggest thatAr. ramidusoccupied
a wooded Pliocene habitat.
I
n an effort to characterize the environment
inhabited byArdipithecus ramidus, between
1994 and 2000 we repeatedly collected
fossils from the surface of all known hominid-
bearing exposures of the 4.4 million-year-old
Lower Aramis Member (1). All fossils encoun-
tered in systematic“crawls”(2), excavations, and
two quarries were collected; this avoided biases
introduced by selective collection, a practice that
can confound ecological analysis (3).
Most of the recovered macrofaunal speci-
mens (approximately 135,000 fossils from mam-
malian families in which most species exceed
5 kg in adult body weight) were pieces of bone or
tooth that could not be taxonomically identified
below the family level (Fig. 1). Most were long
bone shaft splinters, and many teeth were
represented by less than half of a crown. These
less identifiable specimens were pooled into
locality-specific bulk samples (such as“bulk
equid dental”and“bulk mammal bone”from
ARA-VP-6). The other >6000 collected speci-
mens from this interval were taxonomically more
precisely identifiable and were assigned individ-
ual numbers (such asARA-VP-6/1356). These
specimens represent mammals ranging in size
from shrews to proboscideans (2).
Taphonomy.Crania, horn core fragments, and
postcranial elements identifiable to family level
are rare in the total Lower Aramis Member col-
lection. For example, not a single cranium, or even
partial cranium, is present among 733 cataloged
tragelaphine bovid specimens. Bovid postcranial
samples include just 6 proximal metapodials, 17
distal metapodials, 7 calcanei, 19 astragali, 84
phalanges, and 8 distal humeri. Only a few mam-
mals are represented by associated elements, the
most complete being the primarily in situARA-
VP-6/500 Ar.ramidusskeleton (1).
Fossils from larger mammals show no
rounding or abrasion associated with hydraulic
transport. This is consistent with the sedimentol-
ogy of the deposits (1), as well as with the
abundance and preservation of small specimens.
The assemblages have therefore not been water-
transported or -sorted. Surface exfoliation from
subaerial weathering and chemical corrosion has
obscured the original surface of some pieces and
varies by locality. Only 66 of 157 limb bone shaft
fragments retain original surfaces adequate for
confident identification of perimortem modifica-
tions in the most affected bulk bone collection
from anArdipithecus-bearing sublocality (ARA-
VP-1 SHF). In the more representative bulk
sample quantitatively analyzed for this variable
(ARA-VP-1 SRG), 40 of 64 specimens had good
surface preservation. Fragments from smaller taxa
tend to show less weathering across all localities
where present, suggesting more rapid burial.
Where assessed on preserved original surfaces,
limb bone shaft fragments from large mammals
display a wide range of marks (Fig. 1). Tooth
marking by mammalian carnivores is evident in
21 of 24 bulk bone samples from different local-
ities (each sample typically containing hundreds
of specimens). Tooth marks attributable to croc-
odiles (4) are rare, and were found in only three
of nine bulk samples assessed. Rodent gnawing
(mouse- to porcupine-sized) and insect-derived
marks are present in all bulk samples. Root etching
is extremely rare. The paucity of trampling marks
corresponds to a lack of sand in the substrate (1).
Although raptors can account for over 80%
of deaths in some modern primate assemblages
(5), the distinctive signature of such predation is
missing from the cercopithecid assemblage. In-
stead, the damage and breakage patterns are more
consistent with a mammalian carnivore [support-
ing online material text S1]. A full demographic
range is represented, and it is likely that the cer-
copithecid assemblage is attritional, with heavy
postmortem ravaging by carnivores (6,7). This
pattern also holds for the bovid remains.
Large mammal carnivorans represent the
dominant agent of perimortem bone breakage,
as evidenced by the ubiquity of ancient spiral
fractures. There are high rates of tooth marking
on limb bone fragments (47 to 75% in three bulk
assemblages quantitatively assessed;n= 155
specimens) and tooth notching of bone fragment
edges (27% of tooth-marked pieces;n= 30).
Proportions of limb bone shaft fragments with
tooth marks and/or tooth notches are within ranges
produced by modern spotted hyaenas, which have
been observed to deflesh and extract all marrow
while consuming whole limbs (8,9). There is
nearly complete destruction of limb bone ends
(98% in one bulk assemblage assessed quantita-
tively;n= 166). Digestive etching by stomach
acids is rare but widespread (including a hominid
molar exemplar). This pattern of destruction par-
allels that seen in instances of complete marrow
consumption by modern spotted hyaenas.
The hyaenidsIkelohyaena abroniaand cf.
Crocutacf.dietrichi, as well as the ursid
Agriotheriumand four suid taxa are likely
suspects for the destruction of the larger bones.
The canidEucyonwas also present. Degreased,
subaerial, pre-fossilization fragmentation appears
to have been relatively insignificant. Postfossil-
ization fracture resulting from breakage upon ero-
sional exposure is ubiquitous (between 33 and
63% of limb bone shaft fragments examined).
The overallArdipithecus-bearing locality and
sublocality assemblages indicate that the compe-
tition for large mammal carcasses must have been
intense. Abundant shaft fragments, rare epiphy-
seal portions, and the extremely low representa-
tion of axial postcrania as compared to those of
the appendicular and craniodental skeletons,
combined with the high tooth-marking rates,
suggest that the Aramis ecosystem may have
matched highly competitive modern settings
such as Ngorongoro Crater (10). The rarity of
late-stage weathering damage characterized by
deep cracking and exfoliation (<3% of total spec-
imens at stages 4 and 5) suggests that exposure to
subaerial conditions before burial was brief and/
or buffered by tree cover and/or leaf litter.
Exceptions to this taphonomic pattern asso-
ciated withArdipithecusare the SAG-VP-1 and
Ardipithecus ramidus
1
Iziko South African Museum, Natural History Department,
Cenozoic Palaeontology Collections, Box 61, Cape Town 8000,
South Africa; and Institut de Génomique Fonctionnelle de Lyon,
Team“Evo-devo of vertebrate dentition,”Ecole Normale
Supérieure de Lyon, Université Lyon 1, CNRS, INRA, 46 Allée
d’Italie, 69964, Cedex 07, Lyon, France.
2
Post Office Box 369,
Captain Cook, HI, 96704, USA.
3
Center for Human Evolutionary
Studies, Department of Anthropology, Rutgers University, 131
George Street, New Brunswick, NJ 08901-1414, USA.
4
Human
Evolution Research Center and Department of Integrative
Biology, University of California, Berkeley, 3010 Valley Life
Sciences Building, Berkeley, CA, 94720, USA.
5
Phoebe A.
Hearst Museum of Anthropology, 103 Kroeber Hall, Number
3712, University of California, Berkeley, CA 94720-3712, USA.
6
Ministry of Mines and Energy, Post Office Box 486, Addis
Ababa, Ethiopia.
*To whom correspondence should be addressed. E-mail:
[email protected]
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 200966e1
Ardipithecus ramidus

47. 47www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
SAG-VP-3 localities, 0.5 and 2.0 km southeast of
the easternmostAr.ramidusoccurrence. Here,
different assemblage composition (table S1) and
modification signatures are present. Micromam-
mals, birds, and primates are absent from the Lower
Aramis inter-tuff horizon within these spatially
extensive but faunally depauperate localities (n=5
and 3 identified specimens, respectively). Sublocal-
ities with the most surface bone were circumscribed
within each of these localities, and fossils were
collected by identical methods for comparison with
the faunally richerArdipithecus-bearing localities to
the northwest. The resulting assemblages are
dominated by poorly preserved (highly weathered)
remains of large, mostly aquatic animals, which is
consistent with their more axial location in the
depositional basin (as evidenced by structural and
sedimentological considerations) (1).
Small mammals.Micromammals and birds
closely related to extant taxa (and therefore pre-
sumed to be ecologically sensitive indicators) are
found at allArdipithecus-bearing localities (table
S1). However, the large majority of these pri-
marily small fossils (both individually cataloged
and pooled bulk samples) were recovered by
water-sieving at two widely separated quarries.
The more productive quarry (located <100 m
from theARA-VP-6/500partialAr. ramidus
skeleton) yielded about 10,000 total specimens.
Of these, more than 1000 are micromammal teeth
or jaw fragments, or small bird fragments
identifiable at or below the ordinal level. In con-
trast to the intensive destruction of large-mammal
bones described above, micromammal and small
avian postcrania are well preserved and abundant
in these quarry assemblages.
All microvertebrate remains from the two quar-
ries were analyzed taphonomically [according to
the protocol in (11)]. The dense concentration of
remains, consistently high-quality preservation,
abundant postcranial elements, and mostly intact
jaws suggest that these small mammals were pro-
tected from trampling and sunlight. Thus, they prob-
ably experienced no postmortem transport, beyond
perhaps bioturbation and/or emplacement in dessi-
cation cracks during alluvial flooding (based on
the in situ vertical alignment of many rodent limb
bones in the alluvial silty clay in both quarries).
Fig. 1.Bone modifica-
tion of medium and
large mammalian re-
mains from the Lower
Aramis Member. The
central panel shows limb
bone shaft splinters that
are ubiquitous in the
assemblage and were
collected by the thou-
sands during the 100%
recovery operation. Scale
bar, 2 cm. (A) Termite
damage. (B) Inner con-
choidal scars from carni-
vore gnawing. (C) Carnivore
tooth marks on anArdi-
pithecusmandible cor-
pus. (D) Stomach acid
etching on an artiodac-
tyl phalanx and bone
splinter of a medium-
sized mammal. (E) Car-
nivore tooth punctures.
(F) Gnawing damage by
a small-to-medium car-
nivore on cercopithecoid
limb bones. (G) Similar
damage on anArdipithe-
cusmetacarpal. (H) Dam-
age from gnawing by a
small rodent on a large-
mammal limb bone shaft
fragment.
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org 66e2
RESEARCH ARTICLESResearch Articles

48. 48 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
The lack of digestive traces on micromammal
molars (0.9%, all in the“slightest”category), the
low percentage and degree of such traces on
incisors (10.7%; 9.9% in the“slightest”cat-
egory), and the avian assemblage composition
(Fig. 2) combine to suggest that many of the
microvertebrate remains may have been disag-
gregated from barn owl (Tyto ) pellets (11 ). Aside
from one strigid specimen,Tytosp. nov. is the
only owl recorded at Aramis and is relatively
abundant. Barn owls are well-known micro-
mammal accumulators that produce the lowest
levels of digestion and modification among avian
predators. Their pellets are known to provide a
sample of the micromammal fauna within several
kilometers of the roost (11,12), but the assem-
blages they create may be biased by prey avail-
ability and vulnerability.
The Aramis collection (Fig. 2) includes up to
20 new species among a total of 32 small-mammal
genera within the orders Insectivora (two fami-
lies), Chiroptera (five families), Hyracoidea (one
family), Rodentia (six families), Lagomorpha
(one family) and Carnivora (one family) (2).
These taxa indicate that the drainage basin
contained a variety of biotopes, but the distri-
bution of fossils and sediments implies that the
Ardipithecus-bearing locales were wetter. Drier
environments were present at some distance (1,13).
Fossils of the porcupineAtherurus, the murid
Oenomys, and the emballonuridTaphozous
found atArdipithecus-bearing localities suggest
that forests and/or well-developed mesic wood-
lands were at least locally present in the
paleodrainage basin. Such flora, supported by a
high water table or high rainfall due to a higher
altitude (1), may have graded into deciduous
woodlands. Other associated woodland animals
include the shrewsCrocidura, Myosorex, and
Suncus; the batsRousettusand possiblyHippo-
sideros; the porcupineXenohystrix; the mice
Dendromus,Praomys, andMus; and the dwarf
mongooseHelogale. The existence of mesic
settings is supported by the strong presence of
the Asiatic muridGolunda(~13%), whose
contemporary speciesG. elliotiis today typically
found in thickets and bush on densely vegetated
plains. The absence of the cane ratThryonomys
suggests that local suitable aquatic environments
were absent, although it is also missing from the
more aquatic, primate-free assemblages to the
southeast of Aramis. The absence of small hy-
racoids and galagos is notable and unexplained.
The muridUranomysis abundant, large and
small species together representing 44% of small-
mammal specimens. In association withPraomys
(10%), the two genera constitute about 50% of
the micromammalian specimens. Today,Uranomys
is almost always found in abundance and in
association withPraomysin two biotopes: (i)
Borassuspalm savanna characterized by a wet
Hyparrheniagrassland with dense thickets, and
(ii) Mbuga mesic grassland characterized by
dense, long grasses (14). Combined with the
taphonomic findings, this numerical predomi-
nance may reflect predator bias, because barn
owls would be expected to have focused their
predation in open islands of palm-thicket grass-
land within the larger woodland setting, as
indicated by the many“wooded-habitat” mam-
malian and avian indicator taxa.
Rarer species in theArdipithecus-bearing
assemblages indicate that more xeric and open
savanna woodlands were regionally present.
These include the batsRhinolophusandCardio-
derma, the squirrelXerus, the gerbilTatera, the
miceAcomysandSaidomys, and the ratArvican-
this. Still dryer scrub or even arid steppe settings
must have also been present (and probably
sampled by avian predators), as rarely attested
to by the hareLepus, the hedgehogAtelerix, and
the batColeura. The Lower Aramis Member
localities are today at an elevation of about 600 m,
butTachyoryctesandMyosorexhave contempo-
rary counterparts typically found at higher altitudes,
in mesic montane forests and uplands.
Birds.Rich avifauna (Fig. 2 and table S2)
provides additional understanding of the Aramis
environment. The 370 cataloged specimens com-
prise a minimum of 29 different taxa representing
at least 16 families in 13 orders. Most taxa are
terrestrial rather than aquatic (the latter make up
only 3.8% of identified specimens). Small taxa
such as doves, lovebirds, mousebirds, passerines,
and the swift are abundant. These were mostly
recovered from the two quarries and are inter-
preted as deriving from owl pellets. Open-country
taxa such as two bustards (Otididae) and the quail
Coturnixsp. are exceedingly rare. Waterfowl
are rare and include ?Platalea(ibis or spoonbill,
n= 1 identified specimen), Anatidae (geese and
ducks,n= 9), andAnhinga(darter,n= 1). These
indicate the presence of open water, presumably
a river or lake distal to the focus of deposition.
In addition to the barn owl, we recovered
fossils of the diurnal predatorsAquila(eagle,n=
11) and smaller raptors (the size of hawks or
kites). These prefer to hunt in open or ecotonal
conditions and presumably roosted in tall emer-
gent trees (15). The Aramis galliform assemblage
(35% of identified specimens) is dominated by the
abundant ecological indicator speciesPavosp., a
peafowl (n= 39), signaling forested conditions
(16). The lovebirdsAgapornis(n= 88) the parrot
Fig. 2.Relative abundance of avian and small-mammal taxa. For each bird
taxon, the pie slice and first number apply to the number of identified specimens
(n= 263); the second (in parentheses) is the minimum number of individuals
represented in the overall sample. For small mammals, the numbers apply to the
number of identified specimens only (n= 1127), but closely reflect the minimum
number of individuals because only craniodental specimens are included.
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
66e3
Ardipithecus ramidus
Ardipithecus ramidus

49. 49www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Poicephalussp. (n= 1), and guineafowl ?Guttera
sp. (n= 2) are known from woodlands and forests,
ranging into wooded savanna.
Collectively, the large-mammal taphonomy
ofArdipithecus-bearing localities indicates a land-
scape where carcasses were almost always rapidly
and intensively ravaged and the resulting fragments
soon buried without transport. The small-mammal
and avian assemblages combine with other geo-
logical and paleontological data to indicate that
mesic woodlands dominated theArdipithecus-
bearing landscape 4.4 million years ago.
References and Notes
1. G. WoldeGabrielet al.,Science326, 65 (2009).
2. T. D. Whiteet al.,Science326, 64 (2009).
3. D. F. Su, T. Harrison,J. Hum. Evol.55, 672 (2008).
4. J. K. Njau, R. J. Blumenschine,J. Hum. Evol.50, 142 (2006).
5. W. J. Sanders, J. Trapani, J. C. Mitani,J. Hum. Evol.44,
87 (2003).
6. There is no evidence thatAr. ramiduswas in any way
involved with this carnivory, but there is also no evidence
with which to exclude this possibility.
7. T. D. White, N. Toth, inBreathing Life into Fossils,
T. Pickering, K. Schick, N. Toth, Eds. (Stone Age Press,
Gosport, IN, 2007), pp. 281–296.
8. R. J. Blumenschine,J. Archaeol. Sci.15, 483
(1988).
9. R. J. Blumenschine,J. Hum. Evol.29, 21 (1995).
10. R. J. Blumenschine,J. Hum. Evol.18, 345 (1989).
11. P. Andrews,Owls, Caves and Fossils. Predation,
Preservation and Accumulation of Small Mammal Bones in
Caves, with an Analysis of the Pleistocene Cave Faunas from
Westbury-sub-Mendip, Somerset, UK(Natural History
Museum Publications, London, 1990).
12. R. L. Lyman, R. J. Lyman,Intl. J. Osteoarchaeol.13, 150
(2003).
13. T. D. Whiteet al.,Science326, 67 (2009).
14. L. Herwig, thesis, Universiteit Antwerpen, Antwerp,
Belgium (1992).
15. W. J. Sanders, J. Trapani, J. C. Mitani,J. Hum. Evol.44,
87 (2003).
16. A. Louchart,S. Afr. J. Sci.99, 368 (2003).
17. For funding, we thank the U.S. NSF (this material is based
on work supported by grant nos. 8210897, 9318698,
9512534, 9632389, 9910344, and 0321893 HOMINID-
RHOI) and the Institute of Geophysics and Planetary Physics
of the University of California at Los Alamos National
Laboratory. We thank L. Bach and J. Carrier for small-
mammal and bird sketches, H. Gilbert for photographic
illustration, K. Brudvik, and D. DeGusta for editorial
assistance; the Ministry of Tourism and Culture, the Authority
for Research and Conservation of the Cultural Heritage, and
the National Museum of Ethiopia for permissions and
facilitation; and the Afar Regional Government, the Afar
people of the Middle Awash, and many other field workers for
contributing directly to the research efforts.
Supporting Online Material
www.sciencemag.org/cgi/content/full/326/5949/66/DC1
SOM Text
Tables S1 and S2
References
4 May 2009; accepted 18 August 2009
10.1126/science.1175823
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org 66e4
RESEARCH ARTICLES
Research Articles

50. 50 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Macrovertebrate Paleontology
and the Pliocene Habitat
ofArdipithecus ramidus
Tim D. White,
1
*Stanley H. Ambrose,
2
Gen Suwa,
3
Denise F. Su,
4
David DeGusta,
5
Raymond L. Bernor,
6,7
Jean-Renaud Boisserie,
8,9
Michel Brunet,
10
Eric Delson,
11,12
Stephen Frost,
13
Nuria Garcia,
14
Ioannis X. Giaourtsakis,
15
Yohannes Haile-Selassie,
16
F. Clark Howell,
17
†Thomas Lehmann,
18
Andossa Likius,
19
Cesur Pehlevan,
20
Haruo Saegusa,
21
Gina Semprebon,
22
Mark Teaford,
23
Elisabeth Vrba
24
A diverse assemblage of large mammals is spatially and stratigraphically associated withArdipithecus
ramidusat Aramis. The most common species are tragelaphine antelope and colobine monkeys.
Analyses of their postcranial remains situate them in a closed habitat. Assessment of dental mesowear,
microwear, and stable isotopes from these and a wider range of abundant associated larger mammals
indicates that the local habitat at Aramis was predominantly woodland. TheAr. ramidusenamel isotope
values indicate a minimal C
4vegetation component in its diet (plants using the C
4photosynthetic
pathway), which is consistent with predominantly forest/woodland feeding. Although the Early Pliocene
Afar included a range of environments, and the local environment at Aramis and its vicinity ranged
from forests to wooded grasslands, the integration of available physical and biological evidence
establishesAr. ramidusas a denizen of the closed habitats along this continuum.
C
ircumscribing the ecological habitat of
the earliest hominids is crucial for un-
derstanding their origins, evolution, and
adaptations. Evidence integrated from a variety
of independent geological and paleontological
sources (1–3) help to placeArdipithecus ramidus
in its regional and local Pliocene environmental
settings. Here, we assess fossils of the larger
vertebrates (mammalian families in which most
species exceed 5 kg adult body weight) to reveal
characteristics of their diets, water use, and hab-
itat preferences.
At Aramis 4.4 Ma (million years ago), pre-
dominantly terrestrial plants, invertebrates, and
vertebrates were buried relatively rapidly on a
low-relief aggrading floodplain, away from pe-
rennially moving water capable of displacing
most remains (2 ,3). Collection bias was avoided
by a systematic 100% collection strategy (1).
Therefore, the large mammal assemblage spa-
tially associated withArdipithecusin the Lower
Aramis Member allows for relatively robust and
precise environmental inference compared with
many other hominid-bearing occurrences.
The assemblage was carnivore-ravaged and
is consequently dominated by bone and dental
fragments (3). It represents an attritionally de-
rived fauna collected between two widespread
marker tuffs that are today exposed along an
extended erosional arc (2,3). The larger mam-
mal fossil assemblage (4) comprises 3837 in-
dividually cataloged specimens assigned to 42
species (6 of them newly discovered), in 34
genera of 16 families (1,5), across a wide body
size range (Fig. 1A). Many of the sampled taxa
provide evidence for the evolution of African
vertebrates.
We consider ecological habitat to mean the
biological and physical setting normally and
regularly inhabited by a particular species. Our
floral definitions follow the United Nations Edu-
cational, Scientific, and Cultural Organization
(UNESCO) classification of African vegetation
(6). Forests have continuous stands of trees with
overlapping crowns, forming a closed, often
multistory canopy 10 to 50 m high; the sparse
ground layer usually lacks grasses. Forests grade
into closed woodlands, which have less contin-
uous canopies and poorly developed grass layers.
Woodlands have trees with canopy heights of 8
to 20 m; their crowns cover at least 40% of the
land surface but do not overlap extensively.
Woodland ground layer always includes heli-
ophilous (sun-loving, C
4) grasses, herbs/forbs,
and incomplete small tree and shrub under-
stories. Scrub woodland has a canopy height
less than 8 m, intermediate between woodland
and bushland. As proportions of bushes, shrubs,
and grasses increase, woodlands grade into
bushland/thickets or wooded grasslands.
Reconstructing the Aramis biotope.Recon-
structing an ancient environment based on ver-
tebrate macrofossils is often imprecise (7). Even
assemblages from a single stratigraphic interval
may sample thousands of years and thus repre-
sent artificial amalgamations of different biotopes
shifting on the landscape through time. Even in a
geologically isochronous assemblage, animals
from different habitats may be mixed by moving
water or by a moving lake or river margin. Eco-
logical fidelity can be further biased through
unsystematic paleontological recovery, for ex-
ample, when only more complete, identifiable,
and/or rare specimens are collected.
Consequently, most early hominid-bearing
open-air fossil assemblages conflate multiple bi-
otopes (7). Under such circumstances, it is not
surprising that many Pliocene hominid habitats
have been referred to as a“mosaic”or“a changing
mosaic of habitats”(8). Such characterizations risk
confusing noise for signal and local for regional
environment, particularly for collection-biased as-
semblages lacking temporal and spatial resolution.
Initial assessment of the fauna associated with
Ar. ramidusindicated“a closed, wooded” environ-
ment (9), an inference subsequently misquoted as
“forest”(10). This interpretation was criticized on
the basis that colobine monkeys and tragelaphine
bovids might be unreliable indicators (11,12).
Taxonomic abundance.Several aspects of
Lower Aramis Member larger mammal assem-
blage abundance data constitute strong indica-
tors of ancient biofacies and biotope (13 ). The
locality-specific subassemblages are remarkably
RESEARCHARTICLES
1
Human Evolution Research Center and Department of In-
tegrative Biology, 3101 Valley Life Sciences Building, Uni-
versity of California, Berkeley, CA 94720, USA.
2
Department
of Anthropology, University of Illinois, 607 South Matthews
Avenue, Urbana, IL 61801, USA.
3
The University Museum, Uni-
versity of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan.
4
Department of Anthropology, Bryn Mawr College, Bryn Mawr,
PA 19010– 2889, USA.
5
Department of Anthropology, Stanford
University, Stanford, CA 94305–2034.
6
National Science
Foundation, Sedimentary Geology and Paleobiology Program,
Arlington, VA 22230, USA.
7
College of Medicine, Department of
Anatomy, Laboratory of Evolutionary Biology, Howard University,
520 W Street, Washington, DC 20059, USA.
8
Paléobiodiversité
et Paléoenvironnements, UMR CNRS 5143, USM 0203, Muséum
national d’histoire naturelle, 8 rue Buffon, CP 38, 75231 Paris
cedex 05, France.
9
Institut de paléoprimatologie et paléonto-
logie humaine, évolution et paléoenvironnements, UMR CNRS
6046, Université de Poitiers, 40 avenue du Recteur-Pineau,
86022 Poitiers cedex, France.
10
Collège de France, Chaire de
Paléontologie humaine, 3 Rue d’Ulm, F-75231 Paris cedex 05,
France.
11
Department of Anthropology, Lehman College, City
University of New York, Bronx, NY 10468, USA.
12
Department
of Vertebrate Paleontology, American Museum of Natural
History, New York, NY 10024, USA.
13
Department of
Anthropology, University of Oregon, Eugene, OR, 97403–
1218, USA.
14
Departamento Paleontología, Universidad
Complutense de Madrid y Centro de Evolución y Comporta-
miento Humanos, ISCIII, C/ Sinesio Delgado 4, Pabellón 14,
28029 Madrid, Spain.
15
Ludwig-Maximilians-University of
Munich, Department of Geo- and Environmental Sciences,
Section of Paleontology, Richard-Wagner-Strasse 10, D-80333
Munich, Germany.
16
Department of Physical Anthropology,
Cleveland Museum of Natural History, 1 Wade Oval Drive,
Cleveland, OH 44106, USA.
17
Human Evolution Research
Center and Department of Anthropology, 3101 Valley Life
Sciences Building, University of California, Berkeley, CA
94720, USA.
18
Senckenberg Forschungsinstitut, Senckenber-
ganlage 25, D-60325 Frankfurt am Main, Germany.
19
Départe-
ment de Paléontologie, Université de N’Djamena, BP 1117,
N’Djamena, Chad.
20
University of Yuzuncu Yil, Department of
Anthropology, Faculty of Science and Letters, Zeve Yerlesimi
65080 Van, Turkey.
21
Institute of Natural and Environmental
Sciences, University of Hyogo, Yayoigaoka, Sanda 669-1546,
Japan.
22
Science and Mathematics, Bay Path College, 588
Longmeadow Street, Longmeadow, MA 01106, USA.
23
Center
for Functional Anatomy and Evolution, Johns Hopkins
University School of Medicine, 1830 East Monument Street,
Room 303, Baltimore, MD 21205, USA.
24
Department of
Geology and Geophysics, Yale University, New Haven, CT
06520, USA.
*To whom correspondence should be addressed. E-mail:
[email protected]
†Deceased
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 87
Ardipithecus ramidus

51. 51www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
consistent in their taphonomy and taxonomy
across the ~7 km distance from the easternmost
(SAG-VP-7) to westernmost (KUS-VP-2)Ar.
ramiduslocalities (3).
Contemporaneous localities between the two
tuffs farther south of the modern Sagantole drain-
age (SAG-VP-1 and -3, at the southeastern paleo-
transect pole) are relatively impoverished. They
lack this diverse and abundant mammal assem-
blage and contain no tragelaphines, no monkeys,
no fossil wood or seeds, no birds, no micromam-
mals, and noArdipithecus(table S1). Comple-
mentary structural, taphonomic, and isotopic data
from localities on this pole of the paleotransect
suggest a more open landscape that supported
more crocodilians, turtles, and hippopotamids,
presumably associated with water-marginal set-
tings more axial in the drainage basin (2,3).
Relative and absolute abundance measures
for the large mammals in our collections from
theArdipithecus-bearing Lower Aramis Member
localities were assessed by the number of iden-
tified specimens (NISP) (n= 1930) and the mini-
mum number of individuals (MNI) based on
teeth (n= 330). Proboscideans, giraffids, and
hippopotamids are rare (Fig. 1, B and C). The
rhinosCeratotherium efficaxandDicerosare rep-
resented by few specimens (NISP 6 and 1, MNI 4
and 1, respectively). Unlike most other waterside
Plio-Pleistocene assemblages, rhinos are more
abundant than hippos at Aramis. The dental meso-
wear pattern and occlusal morphology of Pliocene
Ceratotherium efficaxsuggest that it was predomi-
nantly a grazer but ate less abrasive forage with
respect to its highly specialized Pleistocene and
extant descendantCeratotherium simum. The
morphological and functional properties of the
recoveredDicerossp. molars are similar to those
of the extant browsingDiceros bicornis.
Equids are rare. One,Eurygnathohippussp.
nov., is distinguished by its distal limb, which is
adapted to open-country running. Its elongate-
narrow snout with parabolic symphysis suggests
adaptation to selective feeding. The teeth of this
equid bear a low-blunt cusp morphology reflect-
ing habitual grazing. Large carnivores and
aardvarks are rare, in keeping with their trophic
level (as in most other eastern African Plio-
Pleistocene assemblages).
Ardipithecus ramidusis represented at Aramis
and environs by >110 cataloged specimens repre-
senting a minimum number of 36 individuals
[14 by upper second molar (M
2
) count] in the
Lower Aramis Member. These numbers are rel-
Fig. 1.Aramis large
mammals. (A) Size range
illustrated by astragali.
The Lower Aramis Mem-
ber contains a wide range
of mammalian taxa, illus-
trated by this image.
Top left, Rhinocerotidae;
middle left,Ardipithecus
ramidus(ARA-VP-6/500);
lower left, small bovid.
Included are other artio-
dactyls, carnivores, and
rodents. (B) Relative abun-
dance of larger mammal
taxa at Aramis based on
dental MNI. (C) Dental
NISP based on dental
individuals whose tooth
crowns are more than
half complete. The NISP
value reflects all collected
specimens identified to
the taxon and excludes
bulk specimens (tooth
crowns less than half com-
plete). Associated dental
specimens are counted
as one. The MNI values
use permanent molars
segregated into upper
and lower first, second,
and third molars, respec-
tively. Numbers for each
taxon vary between NISP
and MNI, but the relative
proportions hold similar.
Tragelaphine bovids and
cercopithecid monkeys
dominate, accounting for
more than half of the
assemblage, however
counted.
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org88
Ardipithecus ramidusResearch Articles

52. 52 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
atively low compared with many of the other
macrovertebrate fossil species we collected. This
rarity is consistent with that observed for hom-
inids in other well-known vertebrate assem-
blages (7).Kuseracolobus aramisiandPliopapio
alemuiare ubiquitous in the assemblage, ac-
counting for 30% of both the larger mammal
NISP and MNI. The colobine is numerically
dominant within nearly all of the localities, and
overall by a ratio of 1.4 toPliopapio(colobinae
NISP:cercopithecinae NISP). It is slightly larger
(12 kg female, 18 kg male) than this papionin
(8.5 kg, 12 kg) based on dental regressions (14).
Extant colobines exhibit strong preferences for
arboreal habitats; extinct African taxa range from
fully arboreal to highly terrestrial (15).
Bovids and primates, particularly tragelaphines
and cercopithecids, dominate the larger mammal
assemblage based on taxonomically diagnostic
craniodental elements (Fig. 1). Together, these
taxa account for more than half of the larger
mammal specimens, whether counted by NISP
or dental MNI. Both cercopithecid and bovid as-
semblages appear to be attritional and were rav-
aged heavily by carnivores after death (3).
Bovids help illuminate the local Aramis en-
vironment of theArdipithecus-bearing localities.
One useful index is the relative abundance of
grazing versus browsing taxa, which can indi-
cate the presence of open or closed conditions,
respectively (16–19). The most ecologically sen-
sitive of these taxa include grazing, open-habitat
tribes such as Alcelaphini and Hippotragini ver-
sus the primarily browsing Tragelaphini or the
riparian-associated Reduncini. Reduncine bovids
commonly dominate in African Plio-Pleistocene
faunal assemblages (Fig. 2), in keeping with
fluviatile, swampy, or lake marginal depositional
conditions.
Whether counted by NISP or dental MNI,
Tragelaphus(whose modern congeners are as-
sociated with wooded habitats) (20) is the nu-
merically dominant Aramis bovid, comprising
85% (NISP) of the bovid assemblage (Fig. 1),
followed byAepyceros(whose modern form
favors grassy woodland to wooded grassland en-
vironments). In contrast, alcelaphine and reduncine
bovids that are plentiful at other Plio-Pleistocene
sites are rare at Aramis, accounting for a mere
1% (NISP) and 4% (MNI) of all bovids. Aramis
is unlike any other known African fossil assem-
blages in thatTragelaphusdominates the un-
gulates. (20–23) (Fig. 2).
Alcelaphines and reduncines were found at
slightly higher frequencies at locality SAG-VP-7
at the eastern end of theArdipithecus distribution
(although tragelaphines and aepycerotines still
dominate there). This subtle difference between
SAG-VP-7 and other more westerly hominid-
bearing localities is also indicated by cercopithecid
abundance. SAG-VP-7 has relatively fewer cer-
copithecids and more alcelaphine and reduncine
bovids (Fig. 2), potentially signaling that this east-
ernmostArdipithecuslocality was a transition zone
between two biotopes.
Functional morphology.Taxon-based ap-
proaches to the inference of paleohabitats are
usually restricted to using identifiable cranioden-
tal remains and assume that habitat preference
persists through evolutionary time. Another ap-
proach is to evaluate the anatomy of fossils with
respect to its implications for functional adapta-
tions. These methods presume that mammals
exhibit skeletal and dental adaptations for lo-
comotion and feeding that correlate with their
preferred environment (24). Samples of extant
taxa are used to quantify the relations between
skeletal/dental traits and environmental variables,
with the results then applied to fossil forms (25).
Here, we evaluate the“ecomorphology”of
the most common large mammals at Aramis,
the bovids and cercopithecid monkeys. For the
Aramis bovids, we evaluated the astragali and
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
ALL ARAMIS ARA-VP-1 ARA-VP-6 KUS-VP-2 SAG-VP-7
Hippotragini
Antilopini
Alcelaphini
Reduncini
Bovini
Neotragini
Aepycerotini
Tragelaphini
Kuseracolobus aramisi
Pliopapio alemui
(1441) (1073) (218) (53) (79)
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Aramis Basal
Member Sidi
HakomaDenen
Dora
Shungura B Apak Kaiyumung Kanapoi Laetoli
Hippotragini
Antilopini
Alcelaphini
Reduncini
Bovini
Neotragini
Aepycerotini
Tragelaphini
(879)
(66) (129) (429)
(171) (64) (57) (125) (1061)
A
B
Fig. 2.Aramis taxonomic abundance. (A ) Comparison between the relative abundance (dental) of
bovid taxa at Aramis and other Plio-Pleistocene sites (21,23,45). The bovid fauna at Aramis is
markedly different due to the dominance of tragelaphines. All frequencies are based on NISP,
except for Hadar, which is based on MNI. (B ) Within-site comparison of the relative abundances of
bovids and cerocopithecids. Among Lower Aramis Member localities, SAG-VP-7 has relatively lower
abundances of cercopithecids and higher abundances of alcelaphines and reduncines, potentially
indicative of ecotonal conditions at this easternmost locality of theArdipithecusdistribution.
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
89
RESEARCH ARTICLES
Ardipithecus ramidus

53. 53www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
phalanges (25,26) because other elements that
can be revealing (metapodials and femora) were
not preserved in sufficient numbers. We used a
four-habitat grouping scheme (26) (SOM text
S1). Of the 11 available intact bovid astragali
with statistically significant habitat predictions
(accuracy >95%), 10 are classified as“forest”and
one as“heavy cover.”This is a clear signal, since
these methods typically produce more varied
habitat predictions when applied to fossil sam-
ples (27,28). To lessen possible biases introduced
by confining the analysis to specimens sufficient-
ly complete for measurement, we also examined
nonmetric traits of the phalanges and classified
the entire astragali/phalangeal sample by mor-
photype (SOM text S1, tables S2 and S3, and
fig. S1). These results independently support
the conclusion from metric prediction that these
animals inhabited a“forest”(in the analytical,
not floral, sense).
As with bovids, cercopithecid postcranial fea-
tures are routinely posited to indicate locomo-
tion (29–31). However, systematic studies of large
samples of extant taxa are generally lacking. We
therefore consider most proposed correlations
between cercopithecoid anatomy and locomotor
mode to be of unknown reliability, pending ad-
ditional study. Even so, the elbow is clearly a
key joint for distinguishing between arboreal and
terrestrial primate locomotion. Of 10 available
Aramis cercopithecoid distal humeri, 9 are clearly
consistent with“arboreal”substrate, whereas only
one is consistent with“terrestrial”substrate based
on current criteria. Of 9 proximal ulnae, all are
arboreal. There was no clear evidence of terres-
trial adaptation in 18 proximal radii. Hence,
based on current criteria, there is clear evidence
of arboreal locomotor adaptations, and a paucity
of terrestrial indicators, in the overwhelming
majority of the Aramis cercopithecoid postcranial
sample (SOM text S2).
Dental wear.The morphology, occlusal wear,
and stable isotope composition of dental re-
mains also reveal the diet—and, indirectly, hab-
itat preferences—of some Aramis mammals.
Differences in mesowear can distinguish among
extant browsers, grazers, and mixed feeders (32 ).
The Aramis neotragines,Giraffa, andTragelaphus
cluster with extant browsers (Fig. 3 and table
S3), whereasAepycerosfalls between extant mixed
feeders and nonextreme grazers. Rare Aramis
alcelaphines cluster with nonextreme grazers,
whereas the rare bovine and equid fossils are
closest to extant coarser grass grazers.
The high cusps and colobine-like morphology
ofPliopapio alemui(tall molars with high relief
and little basal flare) suggest that the two Aramis
monkey taxa had similar diets. We sampled a
mixed set of colobine and cercopithecine molars
for a blind test of microwear. No significant dif-
ferences were found between the two taxa. Micro-
wear on the Aramis monkey molars is consistent
with both frugivory and folivory but not hard
object feeding. A diet of soft (but perhaps tough)
foods would be typical of colobines, and the same
may have been the case for the papionin (33).
Enamel isotopes.The carbon isotopic com-
position of a mammal’s tooth enamel reflects the
relative contributions of grass, trees, and shrubs
to its diets. Oxygen isotopes can reveal the de-
gree that a species lives in, or consumes, water
from different sources (34). We sampled tooth
enamel bioapatite from 177 specimens encom-
passing a wide range of mammalian taxa within
theAr. ramidus–bearing unit (Fig. 4, SOM text
S3, and table S4). These were analyzed blind to
taxon. Carbon isotopic ratios for grazers are high,
whereas those for mixed feeders, browsers, and
forest floor feeders decrease systematically
(SOM text S4). Oxygen isotope vales are low
for water-dependent species such as carnivores
and hippos in wet riparian habitats and higher
for water-independent browsers and open dry-
habitat species.
In theArdipithecus-bearing Lower Aramis
Member assemblage, the aquatic carnivore
Enhydriodon(an otter) has the lowestd
18
O of
all species. Conversely, the ursidAgriotherium
(a bear) has the highest carnivored
18
O, consistent
with anatomical evidence for an omnivorous
diet (35). Among herbivores, giraffids (Giraffa
andSivatherium) have the highestd
18
O and
lowestd
13
C values, whereas grazing equids
(Eurygnathohippus), alcelaphines, bovines, hip-
potragines, and rhinocerotids show the converse.
Among primates,Kuseracolobushas higher
d
18
O and lowerd
13
C thanPliopapio, which re-
sembles the difference between modern folivo-
rous Colobini and more omnivorous Papionini
(36,37).
The carbon isotopic composition of four of
fiveArdipithecus ramidusindividuals is close to
Fig. 3.Mesowear
analysis results for the
second molar paracone
apex of fossil ungulates.
Cusp shape was scored
qualitatively as sharp,
rounded, or blunt. The
relative difference in
height between tooth
cusp apices and inter-
cusp valleys (occlusal
relief) was qualitatively
scored as either high
or low (large or small
distance between cusp
apex and intercusp val-
ley, respectively). Histo-
grams show the results
on the mesowear vari-
ables measured (i.e., the
percentages of sharp
versus rounded versus
blunt cusp shapes and
the percentages of
high versus low oc-
clusal relief).
Sharp Rounded Blunt
cusp
relief
Cusp Shape
Aepyceros Alcelaphini Bovini
Neotragini TragelaphiniGiraffa
%sharp %round %blunt %high %low %sharp %round %blunt %high %low %sharp %round %blunt %high %low
%sharp %round %blunt %high %low
%sharp %round %blunt %high %low %sharp %round %blunt %high %low %sharp %round %blunt %high %low
Mesowear Results
Eurygnathohippus
100
80
60
40
20
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org90
Ardipithecus ramidusResearch Articles

54. 54 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
that ofPliopapio,reflecting diets that included
small amounts of
13
C-enriched plants and/or ani-
mals that fed on such plants.Ardipithecuscon-
sumed slightly more of these resources than
modern savanna woodland chimpanzees (38 )
but substantially less than later Plio-Pleistocene
hominids (39 ,40). The fifth individual has a
d
13
C value of–8.5 per mil (‰), which is closer
to, though still lower than, the means forAus-
tralopithecus africanus,Au. robustus, and early
Homo(39,40). Slightly lowerd
18
O compared
withPliopapioandKuseracolobussuggests that
Ardipithecusobtained more water from fruits,
bulbs, tubers, animals, and/or surface sources.
The isotopic composition of the Aramis mam-
mals between the two tuffs (Fig. 4 and table S4)
conforms broadly to patterns expected for their
modern congeners across the forest-woodland-
savanna spectrum (37,38) in the East African
rift and is consistent with other early Pliocene
assemblages (39 ,40). Relatively low primate,
giraffid, tragelaphine, andDeinotheriumd
13
C
values indicate that small patches of closed can-
opy forests were present, although woodlands to
wooded grasslands probably dominated. Low
d
13
C values for hyaenids suggest that browsing
prey contributed more to their diet compared to
their modern congeners in grazer-dominated open
savanna environments (37). This is congruent
with the numerical dominance of browsing tra-
gelaphines and accords with other evidence for
the dominance of woodlands in the 4.4 Ma local
environment occupied byArdipithecus(2,3). A
small number of rare grazing species—mainly
equids, alcelaphines, hippotragines, and some
impala, rhino, and bovines— have highd
13
C
andd
18
O, indicating that they fed on water-
stressed C
4plants in drier, open environments
(41). These taxa comprise a small portion of
the overall assemblage.
The large range ofd
18
O, particularly the large
difference (9.6‰ ) between water-independent
(evaporation-sensitive) Giraffidae (Giraffaand
Sivatherium) and water-dependent (evaporation-
insensitive) Hippopotamidae, suggests a mean
annual evaporative water deficit of ~1500 mm
(41). Therefore, Aramis was a generally dry wood-
land setting far from riparian environments.
Enamel isotopes of these taxa from nearby pene-
contemporary sites at Gona (42) (SOM text S3
and fig. S2) have ad
18
O difference of only 4.6‰,
reflecting an annual water deficit of ~500 mm
(41). Consistently lower oxygen isotope ratios sup-
port geological evidence that Gona was close to
permanent water (43), but higher carbon isotope
ratios for all Gona browsers are inconsistent
with greater water availability (SOM text S3).
Other ecological approaches.An approach
to deducing ancient environment is to first as-
sign each mammal taxon in a fossil assemblage
to an ecological category (usually based on diet
and locomotion) and then compare the propor-
tions of these categories in the fossil sample to a
range of similarly categorized extant communi-
ties (44,45). This approach uses only the presence
or absence of taxa, so it is subject to taxonomic
and taphonomic biases involving small samples
and mixing. Furthermore, the results are often of
low resolution because biased local fossil as-
semblages are compared to variably recorded
modern communities that pool multiple habitats
(21).Ardipithecus ramiduswas previously inter-
preted as inhabiting a woodland or dry forest
based on a preliminary Aramis faunal list (about
10% of the sample now available) (46). Al-
though the full faunal list produces results con-
sistent with this finding, these results are not
highly robust because the data broadly overlap
among distinct environments (e.g., open, riparian,
medium-density, and closed woodland) (47).
Other measures of abundance also provide
information on the trophic structure of mamma-
lian community represented by theArdipithecus-
bearing Lower Aramis Member. Although there
are many grazing and carnivorous species (Fig. 5),
these taxa are rare (48), so a strict presence/
absence evaluation distorts the ecological signal.
When measures of relative abundance (NISP and
MNI) are included, along with direct informa-
tion on trophic levels from the stable isotope and
mesowear results, a different picture emerges.
These combined data show that the large
mammal biomass at Aramis was dominated by
browsers and frugivores (including frugivorous
animals that consume leaves as a substantial part
of their diet). It is unlikely that a plethora of mam-
mals dependent on browse and fruit would have
been able to subsist in an environment without
abundant trees, the presence of which is witnessed
by fossil pollen as well as abundant seeds, wood,
phytoliths, and rhizoliths (2 ).
Hominid habitat.Establishing habitat (as
opposed to general environment) is crucial for
illuminating the paleobiology of any fossil spe-
cies, including hominids. On the basis of mixed
fossil faunas, it has been previously proposed that
“early hominids were apparently not restricted to
a narrow range of habitats.”[(8), p. 571]. However,
this raises the question of whether the hominids
actually occupied a wide range of habitats or
whether taphonomic processes and sampling
biases have mixed hominid remains with those
of species from biotopes that hominids rarely, if
ever, frequented. Many fossil assemblages sim-
ply do not preserve the necessary temporal and
spatial resolution needed to determine whether
hominids preferred the riverine forest, lake margin,
δ
13
C‰
PDB
-10 -5 0-10 -5 0
closed
forest
C 3
browse
C 4
graze
closed
forest
C 3
browse
C 4
graze
open/dry,
18
O-
enriched leaf water
Closed/humid,
18
O-
depleted leaf water
Meteoric/surface
drinking water
25
30
35
δ
18
O‰
SMOW
δ
13
C‰
PDB
Pliopapio alemui (b/o, wd?)
Kuseracolobus aramisi (b, wi)
Enhydriodon (c, wd)
Agriotherium (c/o, wd)
Hyaenidae (c, wd)
Eurygnathohippus (g, wd)
Elephantidae (g, wd)
Deinotherium (b, wd)
Anancus (g, wi)
Ardipithecus ramidus (b/o, wd?)
Kolpochoerus deheinzelini (b/m, wd)
Nyanzachoerus jaegeri (g, wd)
Nyanzachoerus kanamensis (g, wd)
Rhinocerotidae (g/m, wd)
Hippotragini (g, wd)
Aepyceros (m/g, wi/wd)
Alcelaphini (g, wi)
Simatherium (g, wd?)
Neotragini (b, wi)
Ugandax (m/g, wd)
Tragelaphini (b, wi)
Sivatherium (b, wi)
Giraffa (b, wi)
Hippopotamidae (g, wd)
Ugandax cf. gautieri (g, wi?)
Damalops (g, wd?)
AB
Fig. 4.Carbon and oxygen isotopic composition of mammal tooth enamel from the Lower Aramis
Member of the Sagantole Formation in the Middle Awash Valley. (A) Individuald
13
C andd
18
O values
plotted by taxon. (B ) Bivariate meansT1 SD. See SOM text S3 for methods and interpretations, table S3
for raw data and statistics, and fig. S1 for comparison with species also occurring in roughly con-
temporaneous deposits at Gona (42). Food and drinking habits are inferred from closest living relatives
and from carbon and oxygen isotope ratios. b, browser (C
3feeder); m, mixed grazer/browser (C
3and C
4
feeder); g, grazer (mainly C4grasses); wi, water-independent (evaporation-sensitive) (41) or obtaining
substantial amounts of water from green leaves; wd, water-dependent (evaporation-insensitive) (41),
relying on drinking water when plant leaves are dry; c, carnivore; o, omnivore, including diets with leaves
fruit, tubers, roots, flowers (all predominantly C
3), seeds, fungi, and vertebrate and invertebrate animal
matter. Diets, water use, and habitat preferences of species of extinct genera and families are indicated
in italics because they are more intrinsically uncertain. Interpretations are described and justified in
detail in SOM text S3.
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
91
RESEARCH ARTICLES
Ardipithecus ramidus

55. 55www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
bushland, savanna, and/or woodland habitats
demonstrably available within a few kilometers
of most depositional loci within rift valley settings.
For example,Ardipithecus ramidushas also
been found at Gona, about 70 km to the north of
Aramis, in a valley margin environment where
lake deposits interfingered with small fluvial
channels or lapped onto basaltic cones and flows
(43). At Gona, the dominance of C
3plants indi-
cated by paleosol isotopes contrasts with the C
4
plant signal in many associated ungulate grazers
(indicated by enamel isotopic data). Levinet al.
thus concluded thatArdipithecus“...may have
inhabited a variety of landscapes and was not as
ecologically restricted as previous studies suggest”
[(42), p. 232]. The Gona paleontological and
isotopic data show only that a range of habitats
was present, and the attribution ofArdipithecusto
any particular set of the available biotopes is
problematical in this mixed assemblage (49). Fish,
birds, browsers, horses, and hominids are all fre-
quently found in a single mixed fossil assem-
blage in a fluviatile or near-shore deposit. This
does not mean that the fish were arboreal or that
horses were aquatic. Neither do such data mean
that the hominids exploited all potentially avail-
able habitats.
The Lower Aramis Member deposits provide
fossil samples that evidence a range of environ-
ments in the region at 4.4 Ma (2,3). However,
the consistent association ofAr. ramiduswith a
particular fauna and flora in deposits between
SAG-VP-7 and KUS-VP-2 suggests its persistent
occupation of a woodland with patches of forest
across the paleolandscape (2,3).Ardipithecus
has not been found in the apparently more open
settings to the southeast. There is no evidence
of any taphonomic bias related toArdipithecus
that might produce this pattern (3) and no evi-
dence of any other spatial or stratigraphic clus-
tering within the 4.4 Ma Lower Aramis Member
interval.
Based on a range of independent methods
for inferring habitat-based large samples of con-
silient spatial, geological, and biological evidence
generated from diverse sources, we therefore
conclude that at Aramis,Ar. ramidusresided and
usually died in a wooded biotope that included
closed through grassy woodlands and patches of
true forest [sensu (6)]. There is no evidence to
associate this hominid with more open wooded
grasslands or grassland savanna.
Isotopic data indicate that theAr. ramidusdiet
was predominantly forest- to woodland-based. This
interpretation is consistent with evidence of the
dental and skeletal biology of this primate (1). The
ecological context of 4.4 Ma Aramis hominids,
combined with their absence or extreme rarity at
Late Miocene and Early Pliocene sites, suggest that
the anatomy and behavior of the earliest hominids
did not evolve in response to open savanna or
mosaic settings. Rather, this clade appears to have
originated within more closed habitats favored
by these peculiar primates until the origin of
Australopithecus, and perhaps even beyond (50).
References and Notes
1. T. D. Whiteet al., Science326, 64 (2009).
2. G. WoldeGabrielet al.,Science326, 65 (2009).
3. A. Louchartet al.,Science326, 66 (2009).
4. This assemblage is the one co-occurring withArdipithecus
and excludes the small contemporary samples of fossils
from the more easterly localities of SAG-VP-1 and -3; see
(2,3) and table S1 for details.
5. Included in the larger mammal subassemblage analyzed
here are the following taxa: Artiodactyla, Perissodactyla,
Proboscidea, Primates, Carnivora (except Viverridae), and
Tubulidentata.
6. F. White,The Vegetation of Africa,Natural Resources
Research, Vol. 20. (United Nations Scientific and Cultural
Organization, Paris, 1983).
7. T. White, inEvolutionary History of the Robust
Australopithecines, F. Grine, Ed. (Aldine de Gruyter, New
York, 1988), pp. 449–483.
8. M. G. Leakey, C. S. Feibel, I. McDougall, A. Walker,
Nature376, 565 (1995).
9. G. WoldeGabrielet al.,Nature371, 330 (1994).
10. N. E. Sikes, B. A. Wood,Evol. Anthropol.4, 155 (1995).
11. B. R. Benefit, inAfrican Biogeography, Climate Change,
and Human Evolution, T. Bromage, F. Schrenk, Eds.
(Oxford Univ. Press, New York, 1999), pp. 172–188.
12. M. G. Leakey, inAfrican Biogeography, Climate Change,
and Human Evolution, T. Bromage, F. Schrenk, Eds.
(Oxford Univ. Press, New York, 1999), pp. 271–275.
13. Taphonomic and curatorial biases inevitably compromise
quantitative interpretations of any assemblage, including
Aramis. For example, a single hominid canine may break
into only a few identifiable fragments, whereas one
elephantid’s tusk or molar can shatter into thousands of
identifiable fragments. Simple comparisons of fragment
abundance can therefore be misleading. Our abundance
data take these potential problems into account (see Fig.
1 for details).
14. S. R. Frost,Am. Mus. Novit.3350, 1 (2001).
15. J. F. Oates, A. G. Davies, E. Delson, inColobine Monkeys:
Their Ecology, Behaviour, and Evolution, A. G. Davies,
J. F. Oates, Eds. (Cambridge Univ. Press, Cambridge,
1994) pp. 45–74.
16. P. Shipman, J. Harris, inEvolutionary History of the
Robust Australopithecines, F. Grine, Ed. (Aldine de
Gruyter, New York, 1988), pp. 343–381.
17. E. S. Vrba, inFossils in the Making,A. Behrensmeyer,
A. P. Hill, Eds. (Univ. of Chicago Press, Chicago, 1980),
pp. 247–271.
18. R. Bobe, G. G. Eck,Paleobiology27, 1 (2001).
19. G. Suwaet al.,J. Vert. Paleontol.23, 901 (2003).
20. The AramisTragelaphuscf.moroituhas a body size close
to that of the living nyala (T. angasii) and is likely a
direct descendent of theT. moroiturecorded from the
Mio-Pliocene of Asa Koma and Kuseralee.
21. D. Su, thesis, New York University (2005).
22. R. Bobe,J. Arid Environ.66, 564 (2006).
23. K. E. Reed,J. Hum. Evol.54, 743 (2008).
24. H. A. Hespenheide,Annu. Rev. Ecol. Syst.4, 213 (1973).
25. D. DeGusta, E. S. Vrba,J. Archaeol. Sci.30, 1009
(2003).
26. D. DeGusta, E. S. Vrba,J. Archaeol. Sci.32, 1099
(2005).
27. Y. Haile-Selassieet al.,Geobios37, 536 (2004).
28. L. J. Hlusko, Y. Haile-Selassie, D. DeGusta,Kirtlandia56,
163 (2007).
29. J. G. Fleagle,Yearb. Phys. Anthropol.20, 440 (1976).
30. H. B. Krentz, inTheropithecus, N. Jablonski, Ed.
(Cambridge Univ. Press, Cambridge, 1993),
pp. 383–422.
31. S. Elton,Folia Primatol. (Basel)73, 252 (2002).
32. M. Fortelius, N. Solounias,Am. Mus. Novit.3301,
1 (2000).
33. Crushing and shearing areas of 10 cercopithecoid molars
yielded surfaces that could be included in this analysis.
Surface images were made using an SEM at ×500
magnification, and microwear features were collected
using the“Microwear”software (v. 2.2, 1996). Microwear
features included relatively few pits, with narrow pits and
scratches. The microwear on the molars of the Aramis
monkeys is consistent with both frugivory and folivory,
but they were not routinely feeding on hard objects.
A diet of soft, but perhaps tough, foods would be typical
of colobines, and the same might be true for the
papionin, which has tall molars with a large amount of
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NISPMNIFaunal List
RT
O
I
C
Fg
FG
MF
G
B
Fig. 5.Trophic ecovariable distributions by faunal list, dental NISP, and dental MNI. Comparisons of
the Aramis trophic structure based on the faunal list versus specimen-level, dental relative abundance
data as measured by NISP and MNI. Grazing and carnivorous species are abundant in the faunal list–
based trophic structure, whereas browsers and frugivores dominate when NISP and MNI data are
incorporated. B, browser; G, grazer; MF, mixed feeder; FG, fresh grass grazer; Fg, Frugivore (includes
fruit and leaves); C, carnivorous; I, insectivorous; O, omnivorous; RT, root and tuber.
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
92
Ardipithecus ramidus
Research Articles

56. 56 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
relief and has a low level of basal flare in comparison
with other papionins.
34. T. E. Cerling, J. M. Harris,Oecologia120, 347 (1999).
35. B. Sorkin,Hist. Biol.18, 1 (2006).
36. M. L. Carter, thesis, University of Chicago (2001).
37. S. H. Ambrose, M. J. DeNiro,Oecologia69, 395 (1986).
38. M. Sponheimeret al.,J. Hum. Evol.51, 128 (2006).
39. M. Sponheimer, J. A. Lee-Thorp, inHandbook of
Paleoanthropology,W. Henke, I. Tattersall, Eds.
(Springer, Berlin, 2007) pp. 555–585.
40. N. J. Van Der Merwe, F. T. Masao, M. K. Bamford,S. Afr.
J. Sci.104, 153 (2008).
41. N. E. Levin, T. E. Cerling, B. H. Passey, J. M. Harris,
J. R. Ehleringer,Proc. Natl. Acad. Sci. U.S.A.103, 11201
(2006).
42. N. E. Levin, S. W. Simpson, J. Quade, T. E. Cerling,
S. R. Frost, inThe Geology of Early Humans in the Horn of
Africa, J. Quade, J. Wynn, Eds. (Geological Society of
America Special Papers, 2008), vol. 446, pp. 215–234.
43. J. Quadeet al., inThe Geology of Early Humans in the
Horn of Africa, J. Quade, J. Wynn, Eds. (Geological Society
of America Special Papers, 2008), vol. 446, pp. 1–31.
44. P. J. Andrews, J. M. Lord, E. M. Nesbit Evans,Biol. J. Linn.
Soc. Lond.11, 177 (1979).
45. K. Kovarovic, P. Andrews, L. Aiello,J. Hum. Evol.43, 395 (2002).
46. P. Andrews, L. Humphrey, inAfrican Biogeography,
Climate Change, and Human Evolution, T. Bromage,
F. Schrenk, Eds. (Oxford Univ. Press, New York, 1999),
pp. 282–300.
47. Our analysis also raised numerous questions about the
assumptions and procedures underlying such efforts.
48. For example, there are 12“grazing”taxa compared to
only 5“browsing”taxa, but the former are represented
by only 152 specimens, whereas the latter are represented
by 758 (NISP).
49. It is evident that in most rift-valley depositional settings,
a variety of environments would almost always have been
available to hominids. Of primary interest is determining
whether any one of these environments was the preferred
habitat of these primates. Mixed assemblages cannot
usually do this.
50. T. D. Whiteet al.,Nature440, 883 (2006).
51. Supported by NSF (grants SBR-82-10897, 93-18698, 95-
12534, 96-32389, 99-10344, and 03-21893 HOMINID-
RHOI; and grant SBR 98-71480 for mass spectrometry
instrumentation at the Environmental Isotope
Paleobiogeochemistry Laboratory) and the Japan Society
for the Promotion of Science (G.S. and H.S.). We thank
L. Bach, H. Gilbert, and K. Brudvik for illustrations;
the Ministry of Tourism and Culture, the Authority for
Research and Conservation of the Cultural Heritage, and
the National Museum of Ethiopia for permissions and
facilitation; and the Afar Regional Government, the Afar
people of the Middle Awash, and many other field
workers for contributing directly to the data.
Supporting Online Material
www.sciencemag.org/cgi/content/full/326/5949/67/DC1
SOM Text
Figs. S1 and S2
Tables S1 to S5
References
4 May 2009; accepted 14 August 2009
10.1126/science.1175822
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 93
RESEARCH ARTICLES
Ardipithecus ramidus

57. 57www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
TheArdipithecusramidusSkullandIts
Implications for Hominid Origins
Gen Suwa,
1
*Berhane Asfaw,
2
Reiko T. Kono,
3
Daisuke Kubo,
4
C. Owen Lovejoy,
5
Tim D. White
6
The highly fragmented and distorted skull of the adult skeletonARA-VP-6/500includes most of the
dentition and preserves substantial parts of the face, vault, and base. Anatomical comparisons and
micro–computed tomography–based analysis of this and other remains reveal pre-Australopithecus
hominid craniofacial morphology and structure. TheArdipithecus ramidusskull exhibits a small
endocranial capacity (300 to 350 cubic centimeters), small cranial size relative to body size,
considerable midfacial projection, and a lack of modern African ape–like extreme lower facial
prognathism. Its short posterior cranial base differs from that of bothPan troglodytesandP.
paniscus.Ar. ramiduslacks the broad, anteriorly situated zygomaxillary facial skeleton developed
in laterAustralopithecus. This combination of features is apparently shared bySahelanthropus,
showing that the Mio-Pliocene hominid cranium differed substantially from those of both extant
apes andAustralopithecus.
T
he first fossil ofAustralopithecus, a partial
child’s skull found in 1924 at Taung,
South Africa, was reported by R. A. Dart
to combine an ape-like cranial capacity with dis-
tinctive hominid features such as weak facial
prognathism, small anterior deciduous teeth,
and an anteriorly situated foramen magnum
(1). Since then, diverse Plio-Pleistocene cranial
fossils have been recovered, primarily in south-
ern and eastern Africa, establishing a widely
recognizedAustralopithecusgrade of evolution
(2–6).Australopithecuscrania exhibit small,
chimpanzee-to-gorilla–sized cranial capacities,
distinct cranial base flexion, and varying de-
grees of postcanine megadonty with associated
craniofacial/vault morphologies (2 –5,7–10).
The derivation of the genusHomofrom Pliocene
Australopithecusis probable (11), whereas the
pre-Pliocene ancestry ofAustralopithecushas
been elusive.
Until now, the only substantial specimen to
shed any light on pre-Australopithecus hominid
cranial evolution was that ofSahelanthropus
tchadensisfrom Chad (12). Discovered in 2001,
this Late Miocene cranium [specimenTM 266-
01-060-1; estimated at 6.0 to 7.0 million years
ago (Ma)] combines a cranial capacity smaller
thanAustralopithecuswith a long and low
neurocranium, an anteriorly extended upper face
surmounted by a massive supraorbital torus with
no post-toral sulcus, and a lower face less prog-
nathic than those of either chimpanzees or go-
rillas (12–14). The posterior vault and cranial
base are described as resembling post– 3.5 Ma
PlioceneAustralopithecus(12–14). However,
the hominid status ofS. tchadensishas been
challenged; some opined that it exhibits a sur-
prisingly evolved face (15), whereas others have
suggested it to be a gorilla ancestor or some
other ape (16,17).
We report here the skull ofArdipithecus
ramidusrecovered from Aramis, Ethiopia, as a
part of theARA-VP-6/500skeleton (18). Together
with other key Aramis specimens, including the
ARA-VP-1/500temporal/occipital portion (19),
these fossils constitute the first substantial cranial
remains of a pre-Australopithecushominid
directly associated with extensive postcranial
remains (18). TheAr. ramiduspostcranium
indicates both substantial arboreal capability and
an intermediate form of terrestrial bipedality that
preceded the more fully establishedAustralo-
pithecuscondition (20–23). The revelation of a
primitive pre-Austalopithecus locomotor grade
raises substantial interest in establishing the ma-
jor features of theArdipithecuscranium. Did
Ar. ramidusshare any of the derived hominid
features seen inAustralopithecus, or did it exhibit
a skull more like those of extant African apes?
What are its implications with respect to the
controversies surrounding the hominid status of
Sahelanthropus? We seek answers to these ques-
tions by comparing the Aramis fossils toAustra-
lopithecus,Sahelanthropus, and extant African
apes, and we offer new hypotheses about cranial
evolution in the hominid and African ape clades.
TheARA-VP-6/500skull.TheARA-VP-6/
500skull comprises most of the vault, parts
of the base, much of the right face, the left
Ardipithecus ramidus
1
The University Museum, the University of Tokyo, Hongo,
Bunkyo-ku, Tokyo, 113-0033, Japan.
2
Rift Valley Research
Service, Post Office Box 5717, Addis Ababa, Ethiopia.
3
Department of Anthropology, National Museum of Nature
and Science, Hyakunincho, Shinjuku-ku, Tokyo, 169-0073,
Japan.
4
Department of Biological Sciences, Graduate School
of Science, the University of Tokyo, Tokyo, 113-0033, Japan.
5
Department of Anthropology, School of Biomedical Sci-
ences, Kent State University, Kent, OH 44240–0001, USA.
6
Human Evolution Research Center and Department of In-
tegrative Biology, 3101 Valley Life Sciences Building,
University of California, Berkeley, CA 94720, USA.
*To whom correspondence should be addressed. E-mail:
[email protected]
Fig. 1.The fragmented
skull ofARA-VP-6/500.
(Upper panel) Identifia-
ble pieces of the skull
after limited refitting
for digital and physical
molding. (Lower panel)
(A)ARA-VP-6/500-032,
(B) micro-CT rendered
image of the same, with
cross-sectional locations
of (C) and (D) indicated.
Arrowheads in (A) de-
note the positions of (C)
and (D).
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 200968e1
Research Articles

58. 58 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
mandibular corpus, and most of the teeth
(Fig. 1 and fig. S1). These elements were scat-
tered widely across the excavation area (18).
Many were partially disintegrated by the silty
clay sediment, and major structures were frag-
mentary and variably distorted. Each was sta-
bilized in the field, transported within its encasing
sediment via plaster jacket, and later extracted
from matrix under a binocular microscope.
The extracted pieces preserve contiguous
bone from lambdoidal suture to face, but dis-
tortion prevented correct alignments (24). The
largest intact element is most of the relatively
well-preserved left parietal portion. It was col-
lapsed into the vault space such that cranial
height was reduced to ~35 mm (Fig. 1). There
has been excessive fragmentation and/or damage
to the temporal and occipital portions. Individual
pieces are so friable and soft that edge cleaning
would have risked serious damage and loss of
conjoint surface morphology. Therefore, major
pieces were molded and otherwise left largely as
recovered. Restoration was undertaken indepen-
dently using casts (Berkeley, CA) and digital data
(Tokyo, Japan).
In December 2003, we used high-resolution
micro–computed tomography (CT) to scan the
original fossils. We then segmented the repre-
sentations of the better-preserved parts into 64
separate polygon shells. Using these digital mod-
els, we corrected the positions and alignments of
each individual piece (24) (Fig. 2). We then
added the digital model of the better-preserved
ARA-VP-1/500temporal/occipital fossil (19 )
(scaled to 92% size) to complete theARA-VP-6/
500–based reconstruction of theAr. ramidus
cranium (25). The descriptions and comparisons
that follow initially outline key features observed
directly on the individually preserved fossils and
then extend these to an analysis of the digital
reconstruction.
Basion position and basicranial length.
In our initial evaluation ofAr. ramidus(19),
we noted the anterior position of the foramen
magnum relative to lateral basicranial struc-
tures and interpreted this as a derived condi-
tion shared with later hominids. However, the
utility of our observations has been questioned
(26,27). Here, we re-evaluate basion position
and its importance inAr. ramidus, using the
newly available micro-CT data. These data
allow high-resolution, three-dimensional whole-
surface topographic assessment (Fig. 3). To min-
imize influences of orientation, we evaluated
basion position in the basioccipital plane (Fig. 3)
and confined our analysis to landmarks located
inferiorly on the cranial base (i.e., excluding
porion) (28). InARA-VP-1/500, our digital meth-
ods yield a basion position 1.3 mm posterior to
the center of the carotid foramen.
Previous workers have cautioned that because
bonobos tend to have an anteriorly positioned
foramen magnum (29), anterior placement of the
basion might be primitive and therefore not a
derived hominid feature (27). Thus, we com-
pared basion position ofAr. ramiduswith that of
bothPan troglodytesandP. paniscus(24), as
well as with Plio-PleistoceneAustralopithecus.
We found that althoughP. paniscus(mean 6.4 mm,
n= 28 specimens) does have a slightly shorter
basion-to-bicarotid distance thanP. troglodytes
(mean 7.3 mm,n= 20), this difference was not
statistically significant. Furthermore, both species
exhibit almost identical relative values when scaled
by size (bicarotid breadth) (Fig. 3).
Basion position ofAustralopithecusoverlaps
minimally with the twoPanspecies (Fig. 3 and
fig. S2). In Plio-Pleistocene hominids, the basion
is situated from ~0 to 5 mm posterior to the
carotid foramen (30). This distance is generally
<10% of bicarotid breadth, whereas the same
index is >10% in both species ofPan.ARA-VP-
1/500lies at the extreme lower end of the
hominid range and is clearly distinct fromPan
(31).Sahelanthropusalso shares the hominid
condition. On the basis of published information
and our own observations of the original fossil
with allowance for the effects of taphonomic
damage, the basion was probably positioned
Fig. 2.Digital representations of theAr. ramiduscranium and mandible. (AtoD) TheARA-VP-6/500and
downscaledARA-VP-1/500composite reconstruction in inferior, superior, lateral, and anterior views (in
Frankfurt horizontal orientation). (E) Individual pieces of the digital reconstruction in different colors.
Note the steep clivus plane intersecting the cranial vault on the frontal squama (as inSts 5and not apes).
(FandG) Lateral and superior views of theARA-VP-1/401mandible (cast). (HandI) Lateral and superior
views of theARA-VP-6/500left mandibular corpus with dentition.
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
68e2
RESEARCH ARTICLES
Ardipithecus ramidus

59. 59www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
slightly posterior to the bicarotid chord, as in the
hominids examined here.
We also evaluated distance from the basion
to the bi-foramen ovale (basi-ovale) chord as
an alternative measure of posterior basicranial
length (32) (figs. S3 and S4). Regression analysis
inPanshows a proportional relation between
basi-ovale distance and basioccipital length. This
relation also holds forAustralopithecus. There-
fore, basi-ovale distance can be used as a proxy
for basioccipital length. AlthoughARA-VP-1/500
does not preserve the foramen ovale itself, the
internal lateral wall of the foramen spinosum is
preserved (Fig. 3), which permits a reasonable
estimate of basi-ovale distance (24).
Ar. ramidusfalls squarely within the homi-
nid range (fig. S4). Although bonobos exhibit
an absolutely shorter basi-ovale distance than
do common chimpanzees throughout growth,
this length relative to bicarotid breadth differs
comparatively little between the correspond-
ing age groups of the two species or among
their growth periods (fig. S4). Thus, the wide
and short posterior cranial base ofAr. ramidus
and Plio-Pleistocene hominids is not part of a
continuum seen in modern ape morphology,
but rather appears to reflect reorganization of
the cranial base, most likely manifested early
in ontogeny. Analysis of juvenileAustralopithe-
cuscrania will allow a test of this prediction.
Though differences in posterior cranial base
lengths and proportions are seen in the twoPan
species, they show an even greater difference in
their anterior cranial base lengths. Exocranially,
this is reflected, for example, in metrics such as
the distance from the foramen ovale to pterygo-
palatine fossa (fig. S5) and endocranially in the
length of the planum sphenoideum (33). The
morphological effects of these differences are a
particularly elongate nasopharyngeal region with
anterior placement of the palate and the entire
dental arcade inP. troglodytes(fig. S6). Hence,
the cranial base and facial hafting pattern of
P. troglodytesappears highly derived relative to
bothP. paniscusandAr. ramidus.
TheAr. ramidusface and vault: basic
morphology.TheARA-VP-6/500-115maxilla
exhibits a superoinferiorly short face and weak
prognathism compared with the common chim-
panzee. Its overall structure resembles that of
Sahelanthropus, although it is smaller in size
and proportionately shorter superoinferiorly.
The preserved incisor alveoli and the size of its
isolated roots/partial crowns indicates weak sub-
nasal prognathism compared with both the com-
mon chimpanzee and the smaller-faced bonobo.
This reflects the lack of incisor hypertrophy in
Ar. ramidus(34). Facial topography from the
infraorbital plane to the nasal aperture suggests
that it had a short but projecting muzzle, con-
siderably more primitive than the flatter-faced
Plio-PleistoceneAustralopithecusor the graci-
lized face of smallHomospecimens such as
KNM-ER 1813. The zygomatic root of the
maxilla (anterior face) is placed above the upper
first molar (M
1
), more posterior than is typical
of PlioceneAustralopithecus, but more anterior
than is seen modally in bonobos and common
chimpanzees. This reflects a less prognathic face
compared withPanand probably represents the
primitive condition for both hominids and Af-
rican apes. A similar zygomatic root location is
found in many Miocene apes (e.g.,Kenyapithecus,
Nacholapithecus,Sivapithecus,Dryopithecus,
Pierolapithecus, andOuranopithecus).
TheARA-VP-6/500-115maxilla exhibits a
small but distinct upper second incisor/canine
diastema (reportedly absent inSahelanthropus
and variable inAu. afarensis). Dental-arcade
shape is observable in theARA-VP-6/500res-
toration and theARA-VP-1/401mandible (from
an older presumed female) (Fig. 2). The man-
dible exhibits some primitive features, as well as
some derived features shared with earlyAustra-
lopithecus. Although the canine-to-postcanine
tooth row is straight inARA-VP-6/500[as it is
inAu. anamensis(KNM-KP 29281)(35,36) and
someAu. afarensis(10)], the better-preserved
ARA-VP-1/401mandible exhibits an anterome-
dial position of the lower canine relative to lower
third premolar, as in mostAu. afarensis. However,
the worn lower canine ofARA-VP-1/401projects
above both the postcanine occlusal and incisal
planes, indicating that it was not incorporated into
the functional incisive row, thus differing from
Australopithecus(37).
Fig. 3.Basion position
inARA-VP-1/500.(A) Ba-
sal view (basioccipital
plane horizontal). The two
pieces were positioned by
applying criteria of sym-
metry to the well-preserved
basioccipital surface and
by mirror imaging and
determining overall best
fit of the right and left
sides (24). Metric land-
marks (shown by red
squares) are the basion,
carotid foramen, and lat-
eral margin of foramen
spinosum (hidden). Two
lines are drawn, depict-
ing the sagittal plane
(vertical line) and the bi-
carotid foramen chord
(horizontal line). (B) Box
plot of the basion-to-
bicarotid chord distance
scaled by bicarotid breadth
(24). TheAustralopithe-
cusspecimens measured
were as follows:Sts 5,Sts 19,MLD 37/38(casts ofAu. africanus); KNM-WT 17000
(Au. aethiopicus); andO.H. 5(cast),KNM-ER 406,KNM-ER 407(Au. boisei) (see
fig. S2 for individual values). (C) Anterior view showing segmented internal ear.
The validity of the bilateral placements was evaluated by examining semicircular
canal asymmetry, which was confirmed to be slight and within ranges observed in
humans (57,58). Radii of the semicircular canals were measured as in (59) and
were found comparable to the modern ape andAustralopithecusconditions (60).
(D) Close-up of basioccipital showing a horizontal plane passing through the mid-
basioccpital point (24). Note the approximate symmetry of the basioccipital
surface [1.5 times the scale of (A), (C), and (E)]. (E) Oblique basal view showing
the three landmarks (the basion, carotid foramen, and lateral margin of foramen
spinosum). The latter was used in alternative measures of cranial base length (24)
(see text and figs. S3 and S4 for further details and discussion). Scale bar, 20 mm;
common to (A), (C), and (E).
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 200968e3
Ardipithecus ramidusResearch Articles

60. 60 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Fig. 4.Major structural features (24 )oftheAr. ramidus
cranium (ARA-VP-6/500–based reconstruction). (A) Upper
facial projection: ratio ofporion-to-nasionradius byporion-
to-prosthionradius.S. tchadensis(14);Australopithecus
includesSts 5(this study),Sts 71(9), andO.H. 5(14);KNM-
ER 1813from CT scan of cast (this study). TheAr. ramidus
value is 0.73. (B) Midfacial projection: ratio ofporion-to-
nasal aperture radius by average ofporion-to-orbitaleand
porion-to-zygomatic root radii.AustralopithecusisSts 5(this
study);KNM-ER 1813from CT scan of cast (this study). The
Ar. ramidusvalue is 1.23. (C) Facial mask index: maximum
zygomatic breadth at orbital plane divided by biorbital
breadth acrossectoconchion.S. tchadensis(14);Australo-
pithecusincludesA.L. 444-2,Sts 5,Sts 71, SK 48,TM 1517,
O.H. 5,KNM-ER 406,KNM-ER 732,KNM-ER 13750, and
KNM-WT 17000(10). TheAr. ramidusvalue is 1.20. (D)
Overlap index: ratio of projected glenoid tubercle-to-
prosthionlength to projectedzygomaxillare-to-distal M
3
distance.AustralopithecusincludesA.L. 444-2, Sts 5,Sts 71,
SK 48,SK 52,TM 1517,O.H. 5,KNM-ER 406,KNM-ER 732,
andKNM-WT 17000(10). TheAr. ramidusvalue is 0.26. (E)
Subnasal alveolar prognathism: ratio ofporion-to-prosthion
radius byporion-to-nasal aperture radius.Australopithecus
isSts 5(this study);KNM-ER 1813from CT scan of cast (this
study). TheAr. ramidusvalue is 1.23. (F ) Relative upper
facial breadth: bi-frontomalare temporalebreadth divided
by cube root of cranial capacity.S. tchadensis(13);
Australopithecusis divided into nonrobusts (A.L. 444-2,Sts
5,Sts 71, andStw 505) and robusts (O.H.5 ,KNM-ER 13750,
andKNM-ER 23000); data for these andKNM-ER 1813
compiled from (9,10,47,61). TheAr. ramidusvalue is 15.4.
(G) Relative palatal length: projected palate (or dental row)
length divided by cube root of cranial capacity.Australo-
pithecusis divided into nonrobusts (A.L. 444-2, Sts 5, and
Sts 71) and robusts (O.H. 5 andKNM-WT 17000); data for
these andKNM-ER 1813compiled from (10). TheAr.
ramidusvalue is 9.1. (H) Relative bi-glenoid breadth: bi-
external glenoid tubercle breadth divided by cube root of
cranial capacity.S. tchadensis[estimated from (13)];
Australopithecusdivided into nonrobusts (A.L. 444-2 ,Sts 5,
andMLD 37/38) and robusts (O.H. 5 ,KNM-ER 13750,KNM-
ER 23000, andKNM-WT 17000); data for these andKNM-ER
1813compiled from (10). TheAr. ramidusvalue is 15.8. See
SOM materials and methods (24) for further details.
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org 68e4
RESEARCH ARTICLESArdipithecus ramidus

61. 61www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
In mandibular corpus morphology,ARA-VP-
1/401exhibits a posteriorly receding symphysis
and lateral corpus proportions resemblingAu.
anamensisrather thanPan. However, compared
withAu. anamensis, bothARA-VP-1/401and
ARA-VP-6/500exhibit a less inflated man-
dibular corpus, accompanied by extensive lateral
hollowing and high posterior placement of the
ramus root. TheAr. ramidusmandible is similar
to those ofSahelanthropusandAr. kadabbain
corpus dimensions (12,38,39), ramus root po-
sition and development, and circum mid-corpus
height placement of the anterosuperiorly exiting
mental foramen.
TheAr. ramidussupraorbital torus is repre-
sented by a small but informative segment of the
ARA-VP-6/500frontal bone. The torus is ver-
tically 6 mm thick at about mid-orbital position, a
location commonly thinnest in African apes. It is
equivalent to the thinnest end of theP. troglodytes
range (40). Although there is considerable in-
dividual variation and overlap of ranges in torus
thickness between the sexes of modern apes
(12,17), the thin supraorbital torus ofARA-VP-
6/500suggests that this individual was female,
supporting our sex assignment on the basis of
canine size (34 ). Concavity behind the torus is
slight, indicating the absence ofGorilla- or
Pan-like post-toral sulci. The lateral wall of the
frontal sinus is exposed on the medial break of
the preserved supratoral region. Thus, the
presence of a sizeable frontal sinus is shared
with bothPanandGorilla.
The frontozygomatic region lateral to the
orbit of theAr. ramiduscranium is wide and
rugose, comparable to robust individuals of
P. troglodytes, and distinctively more robust than
in the bonobo. The temporal line turns posteri-
orly at about mid-orbital position, comparable to
Sahelanthropus, and well within the wideP.
troglodytesrange of variation. The superior
temporal line then runs largely anteroposteriorly
for the length of the parietal portion (right and
left lines separated by ~25 mm) and crosses the
lambdoidal suture. The crushed occipital region
does not allow for comprehensive evaluation of
compound temporal/nuchal crest configuration,
but a small compound crest is preserved laterally
on the left side. Such a crest is variably ex-
pressed in both male and femaleP. troglodytes
but is typically absent in bonobos (both sexes).
A similar crest is commonly seen inAu. afarensis
(10).
In summary, the facial bones ofARA-VP-6/
500suggest that prognathism is weaker than
inPan, but that the masticatory complex is more
developed than in bonobos, consistent with the
largerAr. ramiduspostcanine dentition (34). The
ARA-VP-6/500face is markedly short super-
oinferiorly, but modern ape data (10 ,41,42)
show that such facial features are highly variable
within sexes and species. Hence, we do not con-
sider its short face to be a species character of
Ar. ramidus.
TheAr. ramiduscranium: overall structure
and comparisons.Digital restoration of theARA-
VP-6/500cranium enables further observation
and quantitative evaluation of its craniofacial
architecture (24) (Figs. 2 and 4, fig. S7, and table
S1). TheAr. ramiduscranium shares enhanced
relative upper facial projection (Fig. 4A) with
Sahelanthropus(14) and later Pliocene hominids.
However, the ape-like projecting midfacial muzzle
of theSahelanthropus/Ardipithecusface clearly
differs from that ofAustralopithecusand early
Homo, as shown metrically in Fig. 4B.
TheArdipithecuscranium also lacks the
suite of derived masticatory features character-
istic of laterAustralopithecus(2–5,7–11,43–47).
We compared two such parameters, one of
them explicitly examined inSahelanthropus
(14). Relative to biorbital breadth, maximum
midfacial-zygomatic breadth at the orbital plane
is considerably enhanced inAustralopithecus
(facial mask index, Fig. 4C) (10). Another mea-
sure, the overlap index (Fig. 4D), reflects the
extent of anteroposterior overlap between the
anterior-most limit of the origin of masseter and
the postcanine tooth row (10,43). In this mea-
sure,Ar. ramidusoverlaps withGorillaand the
least derived end of theAustralopithecusrange.
Comparisons ofAr. ramidusandAustralo-
pithecuswith the two extantPanspecies reveal
distinct cranial structures characteristic of each
species. A pronounced feature ofP. troglodytes
is its elongate nasopharyngeal region and long
anterior cranial base (see earlier in text). As-
sociated morphological correlates include an an-
teroposteriorly elongate temporal fossa and
infratemporal crest, as well as an anteriorly ex-
tended glenoid and preglenoid plane (48). The
entire lower face/dentition is anteriorly displaced,
an inference supported by morphological details
such as the configuration of the posterior alveolar
process. TheP. troglodytespost-M
3
maxillary
tuberosity tends to be anteroposteriorly long,
thereby adding to evidence for anterior displace-
ment of the entire dental arcade relative to the
pterygoid plates. The combined effect is an ex-
tremely prognathic lower face (Fig. 4E and
fig. S6). We hypothesize that these craniofacial
structures are highly derived but are not dietary
adaptations; instead they are related to canine
enlargement (34), perhaps in association with
enhanced gape and/or increased aggression in
P. troglodytes. Although most of these details
cannot yet be directly observed in theAr. ramidus
cranium, it appears that such specializations were
lacking. This is inferred from features such as the
anteroposteriorly short glenoid and theARA-VP-
6/500– based reconstruction with an anteroposte-
riorly short temporal fossa as inP. paniscusand
G. gorilla.
The bonobo shares a long premaxilla and
large incisors with the common chimpanzee, but
Fig. 5.Natural log-log plot of total cranial length against cranial capacity (24). Least-squares
regression lines for the catarrhine subsets are fitted. African apes areGorilla gorilla, P. troglodytes
troglodytes,P. t. schweinfurth, andP. paniscus. AlthoughP. t. schweinfurthihas a smaller body size, it
has a larger ECC and cranial length thanP. t. troglodytes. Bonobos have small skull size relative to ECC.
Ar. ramidus(large red filled star) is plotted using theARA-VP-6/500body weight of ~50 kg (23) and a
rough total cranial length estimate of 162.5 mm (fig. S7 and table S1). The boxed range ofAr. ramidus
is depicted with a wide ECC range of 280 to 350 cm
3
. Possible skull length dispersion is depicted for a
hypothetical situation in whichARA-VP-6/500represents a small-skulled individual (within-sex
correlation between body size and skull size is expected to be weak); most individuals may have had
a larger skull size. The upperAr. ramidusplots represent two SD positions using chimpanzee levels of
variation as a model. The plottedAr. ramidusrange corresponds to approximately half of the species
range ofP. troglodytes, so its actual range of variation was greater.S. tchadensis(TM 266-01-60-1,
large red unfilled star) is plotted from data in (13). See figs. S8 and S9 for further details.
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
68e5
Ardipithecus ramidus
Research Articles

62. 62 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
lacks the extreme features of the latter. Instead,
the bonobo cranium appears uniquely derived in
its particularly small face and jaws (Fig. 4, F to
H) (compared with bothP. troglodytesandAr.
ramidus). This structural pattern is consilient
with the hypothesized size reduction of the en-
tire bonobo dentition (34 ). The gorilla cranium
is almost certainly derived in exhibiting an ex-
treme anteriorly and inferiorly developed lower
face [analytical results of (13,14)]. These and
other facial and jaw features of the gorilla, some
paralleling theAustralopithecuscondition (Fig. 4,
C, G, and H), are best interpreted as the com-
bined effects of allometry (large absolute size)
and functional adaptations to herbivory.
Cranial capacity, scaling, and cranial base
flexion.Because the endocranial surface of
the frontal pole region is preserved inARA-VP-
6/500, we were able to estimate cranial capac-
ity from internal calvarial dimensions (length,
breadth, and height). Multiple regressions based
on modern African apes (totaln= 18, sex-
balanced samples ofP. troglodytes,P. pansicus
andG. gorilla) yield an estimate of 300T10 cm
3
,
with a larger range of 280 to 350 cm
3
if we ac-
count for uncertainty that stems from combining
ARA-VP-6/500and the scaledARA-VP-1/500
temporal/occipital portion (24). This small cra-
nial capacity is comparable with that of female
Pan(fig. S8).
Extending theAr. ramidusreconstruction to
include a rough approximation of total cranial
length (fig. S7 and table S1) allows for a com-
parison of cranial size (maximum cranial length)
with body size (fig. S9). Because subfamily level
trends have been reported among catarrhines in
relative endocranial volume (49), we also exam-
ined total skull length in relation to endocranial
capacity (ECC). In addition to some colobines
(in particular,Presbytis sensu stricto), atelines,
hylobatids, andPan,Ar. ramidushas the smallest
relative cranial length among large-bodied an-
thropoids [as judged from regressions of cranial
length on endocranial volume (Fig. 5)]. Because
maximum cranial length controlled for endo-
cranial volume must largely reflect facial and nu-
chal size, the results suggest a particularly gracile
head in theARA-VP-6/500individual (50). At the
same time, our scaling analysis shows that
postcranially dimorphic species tend to exhibit a
large cranial size relative to that of the endo-
cranium, as well as a large degree of cranial size
dimorphism. In this context, it is instructive that
Ar. ramidusshares its relatively small cranial size
with taxa that are weakly dimorphic both cra-
nially and postcranially.
Despite its small cranial capacity, there is tan-
talizing evidence for advanced cranial base flex-
ion inAr. ramidus. This is seen from the steep
orientation of its clivus, which directly reflects
midsagittal flexion (figs. S10 and S11). However,
because bonobos andAustralopithecusoverlap
in measures of cranial base flexion (33 ,41,51), it
is uncertain whetherAr. ramidusrepresents a
primitive condition shared with bonobos or a
moreAustralopithecus-like flexion involving the
planum sphenoideum and/or greater orbital ky-
phosis (52 ).
TheSahelanthropusandArdipithecuscrania
securely associate a relatively short basicranium
with small cranial capacity. The hominid basi-
cranial pattern and associated morphologies
[such as foramen magnum orientation (24)] are
widely held to be related to bipedality and up-
right posture (12,13), despite a lack of empirical
evidence to clearly support a functionally based
correlation (52,53). TheAr. ramiduscranium
raises the alternative possibility that early hom-
inid cranial base flexion was associated with
neural reorganization that was already present in
Sahelanthropus/Ardipithecus, as suggested for
PlioceneAustralopithecus(1,54,55). Such a
hypothetical supposition is in part testable by both
future fossil finds and by anticipated advances in
our understanding of genomic expression pat-
terns pertaining to brain function, structure, and
morphogenesis.
Conclusions.Micro-CT–based evaluations
of theAr. ramiduscranial base confirm a de-
rived basicranium ofAr. ramidusshared by both
SahelanthropusandAustralopithecus. Our com-
parative analyses ofP. troglodytesandP. paniscus
suggest that this probably reflects basicranial
organization unique to the hominid clade. The
digitally reconstructedAr. ramidusskull further
allows a variety of inferences about African ape
and hominid evolution. Cranial capacity of pre-
Australopithecushominids (as represented by
Ar. ramidusandS. tchadensis) was probably
slightly smaller than that ofAustralopithecus
and also more comparable toPan. TheAr.
ramidusskull (and that ofS. tchadensis) lacked
the masticatory specializations of laterAustra-
lopithecus, consistent with the dental evidence
for an omnivore/frugivore niche lacking empha-
sis on hard and/or abrasive diets. Finally, com-
parisons ofAr. ramidusand extant African apes
suggest that each is unique in aspects of its cra-
nial anatomy. In particular, the common chim-
panzee appears derived in its forwardly placed
lower facial skeleton, possibly associated with in-
creased aggression, whereas the bonobo is char-
acterized by a secondary reduction of facial size.
Ar. ramidusandSahelanthropuslack these
specialized morphologies ofPanand constitute
the probable ancestral morphotype of Pliocene
Australopithecus.
References and Notes
1. R. A. Dart,Nature115, 195 (1925).
2. J. T. Robinson,Am. J. Phys. Anthropol.12, 181 (1954).
3. P. V. T. Tobias,Olduvai Gorge Vol. 2, The Cranium and
Maxillary Dentition ofAustralopithecus(Zinjanthropus)
boisei (Cambridge Univ. Press, Cambridge, 1967).
4. B. Wood, N. Richmond,J. Anat.197, 19 (2000).
5. T. D. White, inThe Primate Fossil Record, W. C. Hartwig,
Ed. (Cambridge Univ. Press, Cambridge, 2002),
pp. 407–417.
6. Here we follow the taxonomy and phylogenetic scheme
summarized in (5), which subsumesParanthropuswithin
the broadly defined genusAustralopithecus. This is in
part because monophyly of robustAustralopithecus
(Au. aethiopicus,Au,robustus,Au. boisei) is inconclusive,
absent stratophenetic tests in South Africa (5,56). We
recognize seven species ofAustralopithecusspanning
~4.2 to ~1.3 Ma [the three“robust”species/chronospecies
andAu. anamensis,Au. afarensis, Au. africanus, and
Au. garhi(most of the latter also considered
chronospecies)], as well as a single variable lineage of early
HomopredatingHomo erectus(56). We interpret a
considerable amount of the observed metric and morpho-
logical variation to reflect temporal (phyletic evolutionary)
and geographic differentiation (5). Others recognize more
than 10 species in four or more genera of separate
putative clades.
7. T. D. White, D. C. Johanson, W. H. Kimbel,S. Afr. J. Sci.
77, 445 (1981).
8. M. C. Dean, B. A. Wood,Am. J. Phys. Anthropol.59, 157
(1982).
9. B. Wood,Koobi Fora Research Project Volume 4:
Hominid Cranial Remains(Clarendon, Oxford, 1991).
10. W. H. Kimbel, Y. Rak, D. C. Johanson, Eds.,The Skull of
Australopithecus afarensis (Oxford Univ. Press, New York,
2004).
11. B. Asfawet al.,Science284, 629 (1999).
12. M. Brunetet al.,Nature418, 145 (2002).
13. C. P. E. Zollikoferet al.,Nature434, 755 (2005).
14. F. Guyet al.,Proc. Natl. Acad. Sci. U.S.A.102, 18836
(2005).
15. B. Wood,Nature418, 133 (2002).
16. B. Senut, M. Pickford,C. R. Palevol3, 265 (2004).
17. M. H. Wolpoff, J. Hawks, B. Senut, M. Pickford, J. Ahern,
Paleoanthropology2006, 36 (2006).
18. T. D. Whiteet al.,Science326, 64 (2009).
19. T. D. White, G. Suwa, B. Asfaw,Nature371, 306 (1994).
20. C. O. Lovejoyet al.,Science326, 70 (2009)
21. C. O. Lovejoyet al
.,Science326, 71 (2009).
22. C. O. Lovejoyet al.,Science326, 72 (2009).
23. C. O. Lovejoyet al.,Science326, 73 (2009).
24. Materials and methods are available as supporting
material onScienceOnline.
25. Adding the scaledARA-VP-1/500temporal/occipital
portion to theARA-VP-6/500cranial reconstruction is
justified by general similarities in the major parts
preserved in both specimens (basioccipital, glenoid, and
zygomatic root areas) (24).
26. M. S. Schaefer,Am. J. Phys. Anthropol.110, 467 (1999).
27. J. C. M. Ahern,Am. J. Phys. Anthropol.127, 267 (2005).
28. We initially reported that, inAr. ramidus, the bicarotid
chord intersects the basion (19). Subsequently, the
Sahelanthropuscranium was similarly described as
having a basion intersected by the bicarotid chord and
“touched”by the biporion line (12). The latter depends
on cranial orientation, which was not specified in that
publication. Other workers measured the basion-to-
biporion (or basion-to-bicarotid) distance, either by
projection in standard (Frankfurt horizontal) orientation
(27) or by direct measurement in basal view (26).
Differences between and within methods are potentially
large when biporion or bitympanic chords are used, due
to the vertical offset of these landmarks from the basion.
An additional difficulty arises in evaluating fragmentary
fossils that are rarely complete enough to apply methods
relying on standard orientations. We therefore opted to
measure the basion-to-bicarotid distance in locally
determined basilar orientation (24).
29. S. A. Luboga, B. A. Wood,Am. J. Phys. Anthropol.81, 67
(1990).
30. Earlier studies ofAustralopithecusexternal basicranial
shape (8) combined radiographic and direct caliper
measures, apparently without due regard to the potential
effects of orientation in such metric evaluations. These
results suggested an extreme anterior position of the
basion in robustAustralopithecuscrania, but not in
A. africanus. However, this dichotomy was not replicated
in a more recent study (27) that examined the basion-to-
bicarotid chord distance in Frankfurt horizontal
projection. Our own results also failed to reveal clear
differences in basion position betweenAu. africanusand
eastern African robustAustralopithecus(fig. S2). Rather,
allAustralopithecustaxa appear broadly comparable in
their positions of the basion, which is consistently located
slightly posterior to the carotid foramina.
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org 68e6
RESEARCH ARTICLES
Ardipithecus ramidus

63. 63www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
31. We agree with previous workers (26,27) that, relative to
the basion, the carotid foramen is more anteriorly
situated in modern humans than in Plio-Pleistocene
hominids. Thus, considerable overlap in range of
variation occurs between chimpanzees and humans in
measures such as basion-to-biporion or basion-to-bicarotid
distances (26,27). However, modern human and
chimpanzee cranial bases differ considerably in overall
structure, and the importance of isolated metrics must be
considered carefully. We observe that jugular and carotid
sizes are much larger in humans than in apes or Mio-
Pliocene hominids and that this directly affects measures of
basion position. Larger jugular and carotid size almost
certainly stems from allometric enlargement of the cranial
vascular system in extremely encephalizedHomo. Thus,
overlap in parameters such as the basion-to-bicarotid chord
between humans and chimpanzees has little consequence in
evaluating and interpretingAr. ramidus.
32. Following Guyet al.(14), we term the pre- and
postchordal portions of the cranial base the anterior and
posterior cranial base, respectively.
33. R. C. McCarthy,J. Hum. Evol.40, 41 (2001).
34. G. Suwaet al.,Science326, 69 (2009).
35. C. V. Ward, M. G. Leakey, A. Walker,J. Hum. Evol.41,
255 (2001).
36. W. H. Kimbelet al.,J. Hum. Evol.51, 134 (2006).
37. T. D. White, G. Suwa, S. Simpson, B. Asfaw,Am. J. Phys.
Anthropol.111, 45 (2000).
38. M. Brunetet al.,Nature434, 752 (2005).
39. Y. Haile-Selassie, G. Suwa, T. D. White, in Ardipithecus
kadabba: Late Miocene Evidence from the Middle Awash
Valley, Y. Haile-Selassie, G. WoldeGabriel, Eds. (Univ. of
California Press, Berkeley, CA, 2009), pp. 159–236.
40. Equivalent measures of mid-torus thickness were taken
on a random subset of theP. troglodytessample of the
cranial base analysis (24) with the following results:
males (n= 5), mean 8.4 mm, range 6.4 to 12.1 mm;
females (n= 6), mean 6.5 mm, range 5.0 to 8.2 mm.
Guyet al.(14) report size-standardized values of mid-
torus thickness of theS. tchadensiscranium. This can be
converted to an absolute mid-torus thickness of 13.8 mm.
41. R. Fenart, R. Deblock, Pan paniscusetPan troglodytes
Craniométrie.Ann. Mus. Roy. Afrique Centr.Ser. IN-8:
no. 204 (Royal Museum for Central Africa, Tervuren,
Belgium, 1973).
42. J. M. Plavcan,J. Hum. Evol.42, 579 (2002).
43. Y. Rak,The Australopithecine Face(Academic Press,
New York, 1983).
44. W. H. Kimbel, T. D. White, D. C. Johanson,Am. J. Phys.
Anthropol.64, 337 (1984).
45. A. Walker, R. E. Leakey, J. M. Harris, F. H. Brown,Nature
322, 517 (1986).
46. G. Suwaet al.,Nature389, 489 (1997).
47. C. A. Lockwood, P. V. Tobias,J. Hum. Evol.36, 637 (1999).
48. C. A. Lockwood, J. M. Lynch, W. H. Kimbel,J. Anat.201,
447 (2002).
49. K. Isleret al.,J. Hum. Evol.55, 967 (2008).
50. Both small ECC and small relative cranial length provide
additional support to our inference thatARA-VP-6/500is
a female (18,34).
51. F. Spoor,S. Afr. J. Sci.93, 182 (1997).
52. C. Ross, M. Henneberg,Am. J. Phys. Anthropol.98, 575 (1995).
53. J. Biegert, inClassification and Human Evolution,
S. L. Washburn, Ed. (Aldine, Chicago, 1963),
pp. 116–145.
54. R. L. Holloway,Nature303, 420 (1983).
55. R. L. Holloway, R. J. Clarke, P. V. Tobias,C. R. Palevol3,
287 (2004).
56. G. Suwa, T. D. White, F. C. Howell,Am. J. Phys. Anthropol.
101, 247 (1996).
57. G. Suwa,J. Anthrop. Soc. Nippon89, 303 (1981).
58. M. Caix, G. Outrequin,Anat. Clin.1, 259 (1979).
59. F. Spooret al.,Proc. Natl. Acad. Sci. U.S.A.104, 10808
(2007).
60. F. Spoor, F. Zonneveld,Yearb. Phys. Anthropol.41, 211
(1998).
61. B. Brown, A. Walker, C. V. Ward, R. E. Leakey,Am. J.
Phys. Anthropol.91, 137 (1993).
62. We thank NSF (this material is based on work supported
by grants SBR-82-10897, 93-18698, 95-12534, 96-
32389, 99-10344, and 03-21893 HOMINID-RHOI) and
the Japan Society for the Promotion of Science (grant
11691176, 16405016, 17207017, and 21255005) for
funding; the Ministry of Tourism and Culture, the Authority
for Research and Conservation of the Cultural Heritage, and
the National Museum of Ethiopia for permissions and
facilitation; the Afar Regional Government, the Afar people
of the Middle Awash, and many other field workers for
contributing directly to the data; H. Gilbert for graphics
assistance for Figs. 1 and 2; and the following institutions
and staff for access to comparative materials: National
Museum of Ethiopia, National Museums of Kenya, Transvaal
Museum South Africa, Cleveland Museum of Natural
History, Royal Museum for Central Africa Tervuren, Naturalis
Leiden, the University of California at Berkeley Human
Evolution Research Center, and the Department of Zoology of
the National Museum of Nature and Science Tokyo. We
also thank M. Brunet, F. Guy, M. Plavcan, M. Ponce de León,
and C. Zollikofer for cooperation with comparative data.
Supporting Online Material
www.sciencemag.org/cgi/content/full/326/5949/68/DC1
Materials and Methods
Figs. S1 to S11
Table S1
References
4 May 2009; accepted 31 August 2009
10.1126/science.1175825
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 200968e7
Ardipithecus ramidus
Research Articles

64. 64 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Paleobiological Implications of the
Ardipithecus ramidusDentition
Gen Suwa,
1
*Reiko T. Kono,
2
Scott W. Simpson,
3
Berhane Asfaw,
4
C. Owen Lovejoy,
5
Tim D. White
6
The Middle AwashArdipithecus ramidussample comprises over 145 teeth, including associated
maxillary and mandibular sets. These help reveal the earliest stages of human evolution.
Ar. ramiduslacks the postcanine megadontia ofAustralopithecus. Its molars have thinner
enamel and are functionally less durable than those ofAustralopithecusbut lack the derivedPan
pattern of thin occlusal enamel associated with ripe-fruit frugivory. TheAr. ramidusdental
morphology and wear pattern are consistent with a partially terrestrial, omnivorous/frugivorous
niche. Analyses show that theARA-VP-6/500skeleton is female and thatAr. ramiduswas nearly
monomorphic in canine size and shape. The canine/lower third premolar complex indicates a
reduction of canine size and honing capacity early in hominid evolution, possibly driven by
selection targeted on the male upper canine.
F
ossilized teeth typically represent the most
abundant and best preserved remains of
hominids and other primates. They pro-
vide crucial evidence on variation, phylogenetic
relationships, development, and dietary adapta-
tions. Furthermore, because canines function as
weapons in interindividual aggression in most
anthropoid species, they additionally inform as-
pects of social structure and behavior.
We have now recovered and analyzed a sam-
ple of 145 non-antimeric tooth crowns compris-
ing 62 cataloged dentition-bearing specimens of
Ardipithecus ramidusfrom the Lower Aramis
Member of the Sagantole Formation, about five
times more than previously reported (1,2) (Fig.
1 and table S1). All permanent tooth positions
are represented, with a minimum of 14 individ-
uals for both the upper canine and upper second
molar (M
2
) positions. Excluding antimeres, 101
teeth have measurable crown diameters. In ad-
dition, sevenAr. ramidusspecimens with teeth
have been described from Gona (3). These are
broadly comparable to their Aramis counterparts
in size, proportions, and morphology but slight-
ly extend the smaller end of the species range in
some mandibular crown diameters.
The major morphological characteristics of
theAr. ramidusdentition have been outlined
in previous studies of Aramis and Gona fos-
sils (1,3,4). Comparisons ofAr. ramiduswith
Late Miocene hominids (Ar. kadabba ,Orrorin
tugenensis, and Sahelanthropus tchadensis) have
identified slight but distinct differences, partic-
ularly in the canine (4–6). Other subtle features
of incisors and postcanine teeth have been noted
as phylogenetic or taxonomic distinctions (5–10).
However, the most recent and comprehensive
evaluation of the available Late Miocene mate-
rials concluded that these differences are minor
compared with extant ape (and later hominid)
genus-level variation and that bothAr. ramidus
andAr. kadabbadentitions exhibit phenetic
similarities with earlyAustralopithecus(4).
The expandedAr. ramidussample of the
present study allows a more definitive phylo-
genetic placement ofAr. ramidusrelative to the
more primitiveAr. kadabbaand the more derived
Au. anamensisandAu. afarensis(11). Here, we
focus on the paleobiological aspects of theAr.
ramidusdentition, including variation, size, and
scaling, probable dietary niche, and canine/lower
third premolar (C/P
3) complex evolution and its
behavioral implications. We also address the
alleged phylogenetic importance (7) of enamel
thickness inAr. ramidus(1). This is now made
possible by the more comprehensive dental col-
lection that includes key associated dental sets.
Crown size, proportions, and variation.
TheAr. ramidusdentition is approximately
chimpanzee-sized (fig. S1 and tables S2 to S4).
Mean canine size is comparable to that of fe-
malePan troglodytes, although the incisors are
smaller. Upper and lower first molars (M1s)
areP. troglodytes–sized but tend to be bucco-
lingually broader (figs. S1 to S3). The second and
third molars (M2s and M3s) are both absolutely
and relatively larger (figs. S1 and S4 to S6).
Postcanine size and proportions ofAr. ramidus
are similar to those ofAr. kadabbaand other
~ 6.0-million-year-old forms (O. tugenensisand
S. tchadensis)(4–10), as well as to many Mio-
cene hominoids (although Miocene ape lower
molars tend to be buccolingually narrower)
(fig. S3).
Variation within the Aramis dental sample
is low. In modern anthropoids, the coefficient
of variation (CV) is lowest in M1 and M2, with
single-sex and mixed-sex values usually rang-
ing from about 3.5 to 6.5 (12–14). At Aramis,
Ar. ramidusupper and lower M1s and M2s are
less variable (CVs ranging from 2.5 to 5.6)
than those ofAustralopithecus afarensisandAu.
anamensis(table S2). However, theseAustralo-
pithecussamples represent multiple sites and
span a much greater time than the Aramis fos-
sils (11). The low variation seen in AramisA r.
ramidusprobably reflects spatially and tempo-
rally restricted sampling and low postcanine
sexual dimorphism as inPan(15) (table S5).
The Aramis postcanine dentition is also mor-
phologically more homogenous than known
Australopithecusspecies samples. For example,
the six relatively well-preserved M
1
s (Fig. 1)
differ little in features otherwise known to vary
widely within hominid and modern hominoid
species (16,17), including Carabelli’s expres-
sion, occlusal crest development, and hypocone
lingual bulge. This suggests that the AramisAr.
ramiduscollection samples regional demes or
local populations with persistent idiosyncratic
tendencies. The ubiquitous occurrence of single
rooted lower fourth premolars (P
4) (now seen in
eight non-antimeric Aramis P
4s) suggests in-
creased frequency of otherwise rare variants from
genetic drift, absent substantial selection for
larger and/or more complicated root systems
(18). Because this anatomy is shared with Gona
Ar. ramidus(3), it appears characteristic of this
regional population.
Morphology and evolution of the C/P
3com-
plex.The C/P
3complex of anthropoids has be-
havioral and evolutionary importance because
canine size and function are directly linked to
male reproductive success (19). Therefore, clar-
ifying the tempo and mode of the evolution of
the C/P
3complex, from hominid emergence
through its early evolution, is important.
Not counting antimeres, 23 upper and lower
canines from 21Ar. ramidusindividuals are now
known from Aramis. Three more have been de-
scribed from Gona (3), and seven from the ~ 6.0-
million-year-oldAr. kadabba,O. tugenensis, and
S. tchadensis(4–10). There are no examples of a
distinctly large male morphotype in any of these
collections (Fig. 1 and figs. S7 and S8), suggest-
ing that canine sexual dimorphism was minimal
in Mio-Pliocene hominids. In basal crown di-
mensions,Ar. ramiduscanine/postcanine size
ratios overlap extensively with those of modern
and Miocene female apes (fig. S9). Absolute
and relative canine heights are also comparable
to those of modern female apes, although canine
height appears exaggerated inP. troglodytes
[Fig. 1; figs. S8, S10, and S11; and supporting
online material (SOM) text S1].
1
The University Museum, the University of Tokyo, Hongo,
Bunkyo-ku, Tokyo, 113-0033 Japan.
2
Department of Anthro-
pology, National Museum of Nature and Science, Hyakunincho,
Shinjuku-ku, Tokyo, 169-0073 Japan.
3
Department of Anat-
omy, Case Western Reserve University School of Medicine,
Cleveland, OH 44106–4930, USA.
4
Rift Valley Research Ser-
vice, Post Office Box 5717, Addis Ababa, Ethiopia.
5
Depart-
ment of Anthropology, Division of Biomedical Sciences, Kent
State University, Kent, OH 44240–0001, USA.
6
Human Evo-
lution Research Center and Department of Integrative Biol-
ogy, 3101 VLSB, University of California Berkeley, Berkeley,
CA 94720, USA.
*To whom correspondence should be addressed. E-mail:
[email protected]
Ardipithecus ramidus
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org94
Ardipithecus ramidus

65. 65www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Canine shape ofAr. ramidusis either com-
parable to female apes or more derived toward
Australopithecus(11) (Fig. 1 and figs. S12 and
S13). The upper canine (UC) is clearly derived
inAr. ramidus, because it has a diamond-shaped
lateral crown profile with elevated and/or flar-
ing crown shoulders (n= 5 from Aramis and
n= 1 from Gona) [this study and (3 ,4,6)].
However, the lower canine (LC) retained much
more of the morphology of the female ape con-
dition (4 ,5) (Fig. 1, figs. S11 to S13, and SOM
text S1). A hominid-like incisiform LC mor-
phology (high mesial shoulder, developed distal
crest terminating at a distinct distal tubercle) is
seen in some female apes (e.g.,Ouranopithecus
andP. paniscus), whereas the LCs ofAr. kadabba
andAr. ramidustend to be conservative, exhibit-
ing a strong distolingual ridge and faint distal
crest, typical of the interlocking ape C/P
3com-
plex (4) (Fig. 1 and SOM text S1).
TheAr. ramidusP
3is represented by seven ob-
servable crowns, ranging from obliquely elongate
to transversely broad (1) (fig. S14). TheAr. ramidus
P
3is relatively smaller than that ofPanand typ-
ically not as asymmetric or elongate in occlusal
view (figs. S15 and S16). In these respects, the
Ar. ramidusP
3is comparable to those ofAu.
anamensisandAu. afarensis. However,A r.
ramidusis more primitive thanAustralopithe-
cusin retaining a proportionately higher P
3
crown (fig. S16). It appears that there was a
decrease of P
3size from the ancestral ape to
Ar. ramidusconditions, but this reduction was
greater in basal crown dimensions than in crown
height (SOM text S1).
InAr. ramidus, the combined effect of (i) re-
duced canine size and projection and (ii) reduced
size and mesiobuccal extension of the P
3results
in the absence of upper canine honing (defined
as distolingual wear of the UC against the mesio-
buccal P
3face, cutting into the lingual UC crown
face and resulting in a sharpened distolabial enamel
edge). Instead, apical wear inAr. ramiduscom-
mences early and thereafter expands as wear
progresses. None of the known UCs or P
3s ex-
hibits evidence of honing (fig. S14). However,
both upper and lower canines project beyond
the postcanine occlusal plane before heavy wear,
resulting in steep and beveled wear slopes, as
also seen in examples ofAu. afarensisandAu.
anamensis(1,4,20).
TwoAr. ramidusspecimens provide asso-
ciated maxillary and mandibular dentitions with
minimal canine wear. One is almost certainly
female (ARA-VP-6/500), and the other is a
probable male (ARA-VP-1/300) (see below).
Both individuals possess a UC with a shorter
crown height than the associated LC (>10%
difference inARA-VP-1/300)(21). In contrast in
most anthropoid species, the UC is greater in
height than the LC (fig. S17), a condition ex-
aggerated in males of dimorphic species (over
50% in some papionins). Although less extreme
in extant great apes (22), the UC still exceeds
LC crown height by up to ~20% (fig. S18). In
modest samples of modern great ape canines
with little to no wear, we found no instances of
LC height exceeding that of the UC (25 males
and 27 females). This pattern of relative UC and
LC height inAr.ramidusappears unique among
anthropoids and indicates differential reduction
A
C
B
D
E
Upper Canine Maximum Diameter
Gor illa
P. troglodytes
P. paniscus
6 Ma hominids
Ar. ramidus Au. anamensis
Au. afarensis modern human
26
21
16
11
6
F
Upper Canine Labial Crown Height
Gor illa
P. troglodytes
P. paniscus
6 Ma hominids
Ar. ramidus Au. anamensis
Au. afarensis modern human
40
35
30
25
20
15
10
5
female
male
Fig. 1.Representative examples of the AramisArdipithecus ramidusdentition. (A ) Occlusal view micro-CT–
based alignment ofARA-VP-1/300: top, maxillary dentition; bottom, mandibular dentition. The better-
preserved side was scanned and mirror-imaged to form these composites. (B )ARA-VP-1/300in buccal view:
top, right maxillary dentition (mirrored); bottom, left mandibular dentition. (C) Comparison of canine
morphology (micro-CT–based renderings). Top row, lingual view of upper canines, from left to right: male
P. troglodytes(cast), femaleP. troglodytes(cast),Ar. kadabba ASK-VP-3/400,Ar. ramidus ARA-VP-6/1,Au.
afarensis L.H. 6(cast),Au. afarensis A.L. 333x-3(cast, mirrored). Lower rows, distolingual view of lower canines,
main row from left to right: maleP. troglodytes(cast), femaleP. troglodytes(cast),Ar. kadabba(STD-VP-2/61),
Ar. ramidus ARA-VP-1/300,Au. africanus Sts 50(mirrored),Au. africanus Sts 51. Lowest two specimens are ape
lower canines with hominid-like features: left,P. paniscus(cast); right,Ouranopithecus macedoniensis RLP-55
(cast). TheAr. ramidusupper canine is highly derived, with a diamond-shaped crown with elevated crown
shoulders. The lower canine tends to retain aspects of primitive ape features. Further details are given in
the SOM figures and SOM text S1. (D)M
1
morphology (micro-CT– based renderings) showing relatively
little morphological variation among the Aramis individuals. Top row left,ARA-VP-1/300(mirrored); right,
ARA-VP-1/1818. Middle row left,ARA-VP-1/3288; right, ARA-VP-6/500. Bottom row left,ARA-VP-6/502
(mirrored); right,KUS-VP-2/154.(EandF) Box plot of upper canine maximum diameter and labial height (in
mm).Ar. ramidusincludes Aramis and published Gona materials (2). The ~6-million-year-old hominids are
represented byAr. kadabba(ASK-VP-3/400) andO. tugenensis(BAR 1425'00)(7). Symbols give central 50%
range (box), range (vertical line) and outliers. See SOM figures and text S1 for additional plots and details.
RESEARCH ARTICLES
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 95
Research Articles

66. 66 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
of the UC in hominids. The UC < LC height
relation is retained in modern humans.
Morphological changes in the seriesAr.
kadabba–Ar. ramidus–earlyAustralopithecus
support the hypothesis of selection-induced UC
reduction. As detailed above, the UC is clearly
derived inAr. ramidus, whereas the LC tends
to retain the primitive female apelike condition.
Au. anamensis, geologically younger thanAr.
ramidusbut older thanAu. afarensis, exhibits a
polymorphic condition represented by both prim-
itive and advanced LC morphologies (4 ,20)
(SOM text S1). The more incisiform morphol-
ogy becomes universal inAu. afarensisand later
hominids. Furthermore, compared with both male
and female apes,Ar. ramidusexhibits a small
UC crown (both basal diameter and height) rela-
tive to apico-cervical root length, more so than
the LC (figs. S19 and S20). This observation
provides further support to the interpretation that
the UC crown was differentially reduced (SOM
text S1).
A broader comparison ofAr. ramiduswith
extant and Miocene apes illuminates aspects of
C/P
3complex evolution. Compared with cerco-
pithecoids, hominoids tend to have smaller P
3s
with less extensive honing (fig. S15). Compared
with other modern and Miocene apes, both spe-
cies ofPanappear to show P
3reduction. The P
3
ofAr. ramidusis even smaller, suggesting further
reduction of the C/P
3complex from an ancestral
ape condition. At first sight, the comparatively
small P
3size inPanappears paradoxical, be-
cause among the modern great apes both male
and femaleP. troglodyteshave relatively large
and tall canines (figs. S9 and S10 and SOM text
S1). However, this apparent paradox is removed
by a broader perspective on tooth and body size
proportions. BothPanspecies share with atelines
andPresbytis(sensu stricto) small postcanine size
relative to body size (Fig. 2, figs. S21 and S22,
and SOM text S2), low postcanine dimorphism,
and low to moderate canine size dimorphism
(figs. S23 to S25). Conversely, papionins exhibit
the opposite condition: large postcanines, large
canines, and extreme dimorphism. We therefore
hypothesize that the basalPancondition was
characterized by a somewhat reduced C/P
3com-
plex as part of a generally small dentition relative
to body size and that the canines were second-
arily enhanced leading to modernP. troglodytes.
TheARA-VP-6/500skeleton and sexual di-
morphism.Of the 21 individuals with canines,
ARA-VP-6/500has UC and LC that are strik-
ingly small; its UC ranks either 12th or 13th (of
13), and its LC ranks seventh (of eight) in size
(table S6). However, postcranially,ARA-VP-6/
500is a large individual with an estimated body
weight of ~50 kg (23). WasARA-VP-6/500a
small-canined male or a large-bodied female?
We began our evaluation ofARA-VP-6/500
(24) by estimating the degree of dimorphism in
theAr. ramiduscanine (SOM text S3). Even in
modern humans, the canine is metrically the most
dimorphic tooth. Mean basal crown diameter of
human male canines is about 4 to 9% larger than
that in females (table S5). Our analysis indicates
thatAr. ramiduswas probably only marginally
more dimorphic than modern humans (tables S6
to S9 and SOM text S3), with a probable range of
10 to 15% dimorphism (in canine mean crown
diameter). This is substantially less dimorphic than
modern great apes, whose male canines (mean
crown diameter) are larger than those of females
by 19 to 47%.
On the basis of the above dimorphism esti-
mate, the probability of a male having canines
as small as those ofARA-VP-6/500can be eval-
uated by bootstrapping (2). Assuming 12% di-
morphism in mean canine size (table S8), the
probability thatARA-VP-6/500is a male is
<0.03(iftheUCisranked12thof13)or<0.005
(if ranked 13th) (table S9 and SOM text S4). We
conclude thatARA-VP-6/500is a large-bodied
female, a conclusion also corroborated by cranial
anatomy (25). This shows that skeletal size dimor-
phism inAr. ramidusmust have been slight (11),
as is the case in both species ofPan(26,27).
TheARA-VP-6/500skeleton and dimorphism
estimates allow us to place theAr. ramidusden-
tition within a broader comparative framework.
Scaling analyses (2) show that the UC ofAr.
ramiduswas relatively small in both sexes (fig.
S22 and SOM text S2). In particular, male UC
height ofAr. ramidusis estimated to be close to
that of femaleP. paniscusandBrachytelesand
to be much lower than that of maleP. paniscus
(which has the least projecting male canine among
extant catarrhines) (Fig. 2).
Canine development and function.In cerco-
pithecoids with highly dimorphic canines, canine
eruption is typically delayed in males, beginning
after the age of eruption in females (28) and ap-
parently corresponding with species-specific pat-
terns of body size growth spurts (29–31). Once
male canine eruption is initiated, it then proceeds
at a higher rate than in females, but it can still last
for several years depending on species (31). As a
consequence, males attain full canine eruption as
they approach or achieve adult body size, both
of which are necessary for reproductive suc-
cess (19).
Sexually distinct patterns of canine eruption
in relation to body size growth have yet to be
well documented in modern great apes but ap-
pear to broadly share the cercopithecoid pattern
described above (28,32–34). Initiation of canine
eruption inP. troglodytesdiffers by about 1.5 to
2 years between the sexes (35). In males of both
P. troglodytesandP. paniscus, full canine erup-
tion appears to coincide broadly with M3 erup-
tion (observations of skeletal materials), with
polymorphism in the eruption sequence of the
two teeth. By contrast in females of both spe-
cies, full canine eruption is attained before M3
eruption.
Fig. 2.Size and scaling of the
Ardipithecus ramidusdentition.
Natural log-log scatter diagram
of relative upper canine height
(yaxis) against relative post-
canine length (xaxis):left, fe-
males;right,males.Bothaxes
represent size free variables
(residuals) derived from scaling
tooth size against body size
across a wide range of anthro-
poids (2). A value of zero rep-
resents the average female
catarrhine condition. Positive
and negative residuals repre-
sent relatively large and small
tooth sizes, respectively. The
diagonal line indicates the di-
rection of equivalent canine and postcanine proportions independent of size.
The five great ape taxa plotted are from left to right:P. paniscus,P. t. troglodytes,
P. t. schweinfurthi,Gorilla gorilla, andPongo pygmaeus. Ar. ramidusis plotted
by using mean postcanine size and canine crown heights of probable female
(ARA-VP-6/500) and male (ARA-VP-1/300) individuals. A hypothetical female
body weight of 45 kg or 50 kg was used (right and left stars, respectively).
Ar. ramidusis shown to have small postcanine tooth sizes, similar to those of
Ateles,Presbytissensu stricto, andPan. Relative canine height ofAr. ramidus
is lower than that of the smallest-canined nonhuman anthropoids,P. paniscus
andBrachyteles arachnoides. See SOM text S2 for further details.
Ardipithecus ramidus
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org96
Ardipithecus ramidus

67. 67www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
The relative timing of canine eruption inAr.
ramidusis revealed by two juveniles. TheARA-
VP-6/1holotype, a probable male (table S6),
includes an unworn UC whose perikymata count
is 193, higher than that inAu. africanus/afarensis
(maximum 134,n= 4) (36) and lower than those
in small samples of femaleP. troglodytesand
Gorilla(minimum 204,n= 10) (37). TheARA-
VP-6/1UC crown formation time was 4.29 or
4.82 years, depending on estimates of enamel
formation periodicity (fig. S26). This is a com-
paratively short formation time, around the mini-
mum reported for modern female apes (38).
The eruption pattern of a second individual,
ARA-VP-1/300, can be assessed from the pres-
ence or absence of wear facets and/or polish. The
ARA-VP-1/300canines were just completing
eruption, its M2s were worn occlusally, and its
unerupted M3 crowns were barely complete
(Fig. 1 and fig. S27). Compared with extant apes,
both its UC and LC development are advanced
relative to M2 and M3 (fig. S28) (39).
The combined morphological and develop-
mental evidence suggests that selection for de-
layed canine eruption had been relaxed inAr.
ramidus. We hypothesize that canine prominence
had ceased to function as an important visual sig-
nal in male competitive contexts.
Tooth size and diet.We consider relative in-
cisor and postcanine sizes to be potentially useful
in inferring dietary adaptations, although consist-
ent patterns across primates have not been ob-
tained (40). In particular, postcanine megadontia
has been considered a defining feature ofAus-
tralopithecus(41). We evaluated incisor and molar
sizes ofAr. ramidusin comparison to those of
PanandAustralopithecus. Among anthropoids,
PanandPongoare unique in having large in-
cisors relative to both postcanine and body size,
a condition not shared byAr. ramidus(fig. S29).
This suggests thatAr. ramiduswas not as inten-
sive a frugivore as arePanandPongo, incisor
length probably being functionally related to re-
moval of fruit exocarp (42) and/or feeding be-
havior such as wadging.
Although the M1 area, normalized by indi-
vidual postcranial metrics, lies well within the
range of extant chimpanzees, the total postcanine
area ofARA-VP-6/500falls betweenPongoandP.
troglodytes(Fig. 3).Ar. ramidusis not only less
megadont thanPongoandAu. afarensisbut,
together withPan,Ateles, and somePresbytis
species, lies at the small end of the range of
variation of large-bodied anthropoids (fig. S30).
The most megadont anthropoids include robust
Australopithecus, such asAu. boisei, as well as
papionins andAlouatta.Ouranopithecuswas
probably as megadont asAustralopithecusspe-
cies, whereasDryopithecusandPierolapithecus
probably had relative postcanine sizes closer to
Ar. ramidusand thus better approximate the
dentition–to–body size relationship of the last
common ancestor of humans and chimpanzees.
We conclude thatAr. ramiduswas substantial-
ly less megadont thanAustralopithecus.
Molar structure and enamel thickness.Molar
structure, enamel thickness, and tooth wear further
illuminate dietary adaptation inAr. ramidus. Com-
pared with the distinct occlusal structure of the
molars of each of the modern ape species (see
below),Ardipithecusocclusal morphology is more
generalized, with low, bunodont cusps and mod-
erate to strong basal crown flare. Such morphol-
ogy also characterizesAustralopithecusas well
as a diversity of Miocene apes (43).Gorillamo-
lars have much more salient occlusal topography
and enhanced shearing crests.Panmolars are
characterized by broad, capacious occlusal basins
flanked by moderately tall cusps, effective in crush-
ing relatively soft, fluidal mesocarp while retaining
the ability to process more fibrous herbaceous ma-
terials (Fig. 4) (44,45). These features are ac-
centuated inPanby the characteristically thin
enamel of its occlusal basin (45,46).
To further elucidate molar structure and di-
etary adaptations ofAr. ramidus, particularly in
comparison withPanandAustralopithecus, we
used micro– computed tomography (micro-CT)
to study molar enamel thickness and underlying
crown structures (2). Although the weak contrast
of fossil enamel and dentin makes micro-CT–
based evaluations difficult, we were able to as-
sess severalAr. ramidusmolars with this method.
These and analyses of CT sections and natural
fracture data (2 ) indicate thatAr. ramidusenamel
is considerably thinner than that ofAustralopith-
ecusbut not as thin as inPan[as originally
reported in (1)] (Fig. 4 and figs. S31 and S32).
Of particular importance is thatAr. ramidus
molars do not exhibit enamel distribution patterns
characteristic ofP. troglodytesandP. paniscus.
BothPanspecies have similar crown structure
and enamel distribution patterns (Fig. 4), although
P. paniscusmolars exhibit a higher cuspal to-
pography, perhaps related to greater reliance on
fibrous food (46,47).Ar. ramiduslacks the thin
occlusal fovea enamel ofPanand in this regard
is similar to bothAustralopithecusand Miocene
forms such asDryopithecus(Fig. 4). ThePan
condition is most likely derived, probably as-
sociated with an increased reliance on higher-
canopy ripe fruit feeding.
Despite the generalized molar structure com-
mon to bothAr. ramidusandAustralopithecus,
the adaptive difference between the two is ex-
pressed by enamel tissue volume, which we con-
sider to broadly track net resistance to abrasion.
Modern ape species exhibit a near-isometric re-
lation between molar durability (measured as
volume of enamel tissue available for wear per
unit occlusal area) and tooth size, despite diverse
dietary preferences and crown anatomy (Fig. 4).
Ar. ramidusfalls near this isometric continuum,
butAustralopithecusdoes not.Australopithe-
cusmolars achieve greater functional durabil-
ity from increased enamel volume.Au. boisei
occupies an extreme position distant from the
modern ape baseline. Thus, both tooth size
and enamel thickness and volume suggest a
substantial adaptive shift fromArdipithecusto
Australopithecus.
This is further expressed in molar macro-
and microscopic wear patterns. In contrast to
Australopithecus,Ar. ramidusmolars did not
wear flat but instead retained stronger bucco-
lingual wear slopes. The AramisAr. ramidus
dentition also exhibits consistently weak M1 to
M3 wear gradients (48 ). Microwear of theAr.
ramidusmolars tends to differ from that ofAu.
afarensis, the latter characterized by a domi-
nance of buccolingually oriented scratches (49).
In contrast, theAr. ramidusmolars tend to ex-
hibit finer and more randomly oriented striae
(fig. S33). Collectively, the wear evidence sug-
gests thatAr. ramidusconsumed a less abrasive
diet and engaged in less masticatory grinding
thanAustralopithecus.
Fig. 3.Relative postcanine dental size inAr. ramidus. Postcanine size is compared directly in reference
to associated postcranial elements;xaxis is natural log of the size variable (body size proxy) of Lovejoy
et al.(23), derived from four metrics of the talus and five metrics of the capitate;yaxis is natural log of
the square root of the sum of calculated areas (mesiodistal length multiplied by buccolingual breadth)
of lower M
1(left) and lower P
4to M
3(right). A,Ar. ramidus ARA-VP-6/500; L,Au. afarensis A.L. 288-1;
c,Pan troglodytes troglodytes; g,Gorilla gorilla gorilla; o,Pongo pygmaeus(males blue, females red).
RESEARCH ARTICLES
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 97
Research Articles

68. 68 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Enamel thickness and phylogenetic implica-
tions.Since the initial description ofAr. ramidus
as a new species of Hominidae (1 ), its relatively
thin molar enamel has been a focus of atten-
tion. Some authors have suggested that its thin
enamel might be a shared derived feature with
Pan(7). The fuller study of molar enamel thick-
ness and patterns outlined above establishes the
following: (i) AlthoughAr. ramidusenamel is
thinner than that ofAustralopithecus, it is not as
thin asPan’s; (ii) the thin enamel ofPanmolars
can be considered a part of a structural adapta-
tion to ripe fruit frugivory (46) and therefore dif-
fers from theAr. ramiduscondition. Furthermore,
the distinct internal structure ofPanmolars seems
lacking inAr. kadabba,O. tugenensis, andS.
tchadensis(4,8,10). Hence, thePancondition
is best considered derived relative to the an-
cestral and early hominid conditions.
Conclusions.Multiple lines of morphological
evidence suggest thatAr. ramiduswas a general-
ized omnivore and frugivore that did not rely
heavily on either ripe fruits (as inPanorPongo),
fibrous plant foods (as inGorilla), or hard and
Fig. 4.Enamel thickness and distribution patterns inAr. ramidus. Left
panels: micro-CT–based visualizations of maxillary first molars in arbi-
trary size. (A) Outer enamel surface; (B) enamel thickness in absolute
thickness scale superimposed on topographic contours; (C) enamel thick-
ness in relative scale to facilitate comparison of pattern. The molars [labeled
in (A)] are as follows: 1 and 5,Au. africanus Sts 24(mirrored) andSts 57; 2,
Dryopithecus brancoi; 6,Ar. ramidus ARA-VP-1/3288; 3,Pan troglodytes; 4,
Pan paniscus; 7, Gorilla gorilla; 8,Pongo pygmaeus. The Panspecies share
a broad occlusal basin and thin occlusal enamel. BothAr. ramidusandD.
brancoiare thinner-enameled thanAustralopithecusbut share with
Australopithecusa generalized distribution pattern. (D) Maximum lateral
enamel thickness, showing thatAr. ramidusenamel is thicker than those of
PanandD. brancoiand thinner than that ofAustralopithecusspecies.
Horizontal line is median; box margins are central 50% range. (E) Ratio of
occlusal (volume/surface area) to lateral (average linear) enamel thick-
nesses, showing thatPanis unique in its distinctly thin occlusal enamel. (F)
Molar durability (enamel volume per unit occlusal view crown area) plotted
against projected occlusal view crown area. An isometric line (slope of 0.5)
is fitted through the centroid of the three measuredAr. ramidusmolars. The
least squares regression (y= 0.418x−1.806) of the combined modern ape
sample is also shown. This slope does not differ significantly from isometry.
Ar. ramidusandD. brancoiare close to, andAustralopithecusspecies
considerably above, the regression line, indicating greater enamel volume
available for wear inAustralopithecusmolars. See (2) for further details.
Ardipithecus ramidus
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org98
Ardipithecus ramidus

69. 69www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
tough food items (as inPongoorAustralopithecus).
Ar. ramidusalso lacked adaptations to abrasive
feeding environments (unlikeAustralopithecus).
These inferences are corroborated by the isotopic
analysis of enamel, which indicates thatAr. ramidus
predominantly consumed (~85 to 90%) C
3plant
sources in woodland habitats and small patches of
forest (50), thus differing from both savanna
woodland-dwelling chimpanzees (>90% C
3) and
Australopithecusspp. (>30% C
4)(51).
Conversely, extantPanandGorilla, each with
its distinctive dental morphology, are best con-
sidered derived in their dietary and dental adap-
tations. This is consistent with theAr. ramidus
postcranial evidence and its interpretations (11,23)
and strengthens the hypothesis that dental and
locomotor specializations evolved independent-
ly in each extant great ape genus. This implies
that considerable adaptive novelty was neces-
sary to escape extinction in the Late Miocene
forest and woodland environments.
These analyses also inform the social behav-
ior ofAr. ramidusand its ancestors. The dental
evidence leads to the hypothesis that the last
common ancestors of African apes and hominids
were characterized by relatively low levels of
canine, postcanine, and body size dimorphism.
These were probably the anatomical correlates of
comparatively weak amounts of male-male com-
petition, perhaps associated with male philopatry
and a tendency for male-female codominance as
seen inP. paniscusand ateline species (52,53).
From this ancestral condition, we hypothesize
that theP. troglodyteslineage secondarily en-
hanced its canine weaponry in both sexes, where-
as a general size reduction of the dentition and
cranium (25) occurred in theP. paniscuslineage.
This suggests that the excessively aggressive in-
termale and intergroup behavior seen in modern
P. troglodytesis unique to that lineage and that
this derived condition compromises the living
chimpanzee as a behavioral model for the ances-
tral hominid condition. The same may be the case
withGorilla, whose social system may be a part
of an adaptation involving large body size, a spe-
cialized diet, and marked sexual dimorphism.
In the hominid precursors ofAr. ramidus, the
predominant and cardinal evolutionary innova-
tions of the dentition were reduction of male
canine size and minimization of its visual prom-
inence. TheAr. ramidusdental evidence suggests
that this occurred as a consequence of selection for
a less projecting and threatening male upper
canine. The fossils now available suggest that
male canine reduction was well underway by 6
million years ago and continued into the Pliocene.
Further fossils will illuminate the tempo and mode
of evolution before 6 million years ago.
References and Notes
1. T. D. White, G. Suwa, B. Asfaw,Nature371, 306 (1994).
2. Materials and methods are available as supporting
material onScienceOnline.
3. S. Semawet al.,Nature433, 301 (2005).
4. Y. Haile-Selassie, G. Suwa, T. D. White, in Ardipithecus
kadabba: Late Miocene Evidence from the Middle Awash
Valley, Y. Haile-Selassie, G. WoldeGabriel, Eds. (Univ.
California Press, Berkeley, 2009), pp. 159–236.
5. Y. Haile-Selassie,Nature412, 178 (2001).
6. Y. Haile-Selassie, G. Suwa, T. D. White,Science303,
1503 (2004).
7. B. Senutet al.,C. R. Acad. Sci. Paris332, 137 (2001).
8. M. Pickford, B. Senut,Anthropol. Sci.113, 95 (2005).
9. M. Brunetet al.,Nature418, 145 (2002).
10. M. Brunetet al.,Nature434, 752 (2005).
11. T. D. Whiteet al.,Science326, 64 (2009).
12. P. D. Gingerich, M. J. Schoeninger,Am. J. Phys.
Anthropol.51, 457 (1979).
13. D. A. Cope, inSpecies, Species Concepts, and Primate
Evolution, W. H. Kimbel, L. B. Martin, Eds. (Plenum,
New York, 1993), pp. 211–237.
14. J. M. Plavcan, thesis, Duke University (1990).
15. Mean postcanine size in one of two subspecies of the
common chimpanzee that we examined (P. troglodytes
schweinfurthi) is marginally greater in females.Pan paniscus
andP. troglodytesboth have postcanine teeth with size
dimorphism weaker than in modern humans (table S5).
16. D. C. Johanson, thesis, Univ. of Chicago (1974).
17. W. G. Kinzey, inThe Pygmy Chimpanzee: Evolutionary
Biology and Behavior, R. L. Susman, Ed. (Plenum, New
York, 1984), pp. 65– 88.
18. Premolar root number and morphologies are known
to be polymorphic, with single rooted P
4s known in
bothAu. anamensis(KNM-ER 22683) andAu. afarensis
(MAK-VP-1/12) (54,55).
19. S. R. Leigh, J. M. Setchell, M. Charpentier, L. A. Knapp,
E. J. Wickings,J. Hum. Evol.55, 75 (2008).
20. T. D. Whiteet al.,Nature440, 883 (2006).
21.ARA-VP-1/300UC and LC heights are 14.5 and 16.6 mm,
respectively. Reasonable estimates of crown height of the
ARA-VP-6/500UC and LC are 13 to 13.5 mm and
14.4 mm, respectively.
22. J. Kelley,Am. J. Phys. Anthropol.96, 365 (1995).
23. C. O. Lovejoy, G. Suwa, S. W. Simpson, J. Matternes,
T. D. White,Science326, 73 (2009).
24. Our analysis for determiningARA-VP-6/500sex consists
of several steps detailed in (2) and SOM text S3.
AlthoughAr. ramiduscanines for which standard crown
dimensions could be measured are limited, by comparing
preserved portions, almost all can be ranked in terms of
size. We therefore simulated probabilities of obtaining
size ranks in model populations with set amounts of
dimorphism in basal crown diameters.
25. G. Suwaet al.,Science326, 68 (2009).
26. P. L. Reno, R. S. Meindl, M. A. McCollum, C. O. Lovejoy,
Proc. Natl. Acad. Sci. U.S.A.100, 9404 (2003).
27. H. M. McHenry,Hum. Evol.1, 149 (1986).
28. B. H. Smith, T. L. Crummett, C. L. Brandt,Yearb. Phys.
Anthropol.37, 177 (1994).
29. S. R. Leigh, B. T. Shea,Am. J. Phys. Anthropol.101, 455
(1996).
30. Y. Hamada, S. Hayakawa, J. Suzuki, S. Ohkura,Primates
40, 439 (1999).
31. S. R. Leigh, J. M. Setchell, L. S. Buchanan,Am. J. Phys.
Anthropol.127, 296 (2005).
32. K. L. Kuykendall,Am. J. Phys. Anthropol.99, 135 (1996).
33. S. R. Leigh, B. T. Shea,Am. J. Phys. Anthropol.99, 43 (1996).
34. Y. Hamada, T. Udono,Am. J. Phys. Anthropol.118, 268
(2002).
35. G. C. Conroy, C. J. Mahoney,Am. J. Phys. Anthropol.86,
243 (1991).
36. M. C. Dean, D. J. Reid, inDental Morphology 2001,
A. Brook, Ed. (Sheffield Academic Press, Sheffield, UK,
2001), pp. 135–149.
37. M. C. Dean, D. J. Reid,Am. J. Phys. Anthropol.116, 209
(2001).
38. G. T. Schwartz, C. Dean,Am. J. Phys. Anthropol.115, 269
(2001).
39. S. W. Simpson, C. O. Lovejoy, R. S. Meindl,Am. J. Phys.
Anthropol.87, 29 (1992).
40. C. J. Vinyard, J. Hanna,J. Hum. Evol.49, 241 (2005).
41. H. M. McHenry,Am. J. Phys. Anthropol.64, 297 (1984).
42. P. W. Lucas, P. J. Constantino, B. A. Wood,J. Anat.212,
486 (2008).
43. Many Miocene apes generally considered more advanced
thanProconsulhave molars with less expansive cingular
structures. Such species, when they simultaneously lack
distinct modifications of occlusal structure, all exhibit a
common bunodont hominoid molar morphology. Such
Miocene apes with this largely generalized molar
morphology includeGriphopithecus,Kenyapithecus,
Equatorius,Nacholapithecus,Chororapithecus,Nakalipi-
thecus,Dryopithecus,Pierolapithecus,Sivapithecus,
Ankarapithecus, and Ouranopithecus. Slight differences
in central tendencies in overall crown shape, occlusal
cresting, accessory cuspules, and enamel thickness are
used to distinguish among some of these taxa, but
individual variation is high and specific distinctions are
not necessarily clear on a specimen-by-specimen basis.
One feature that seems to separate Late Miocene
hominids,Ar. ramidus, andAustralopithecussp., on
the one hand, and the Middle and Late Miocene apes, on
the other, is the lack of a well-developed and distinct
protoconule in the upper molars of hominids. This condition
is also shared by both genera of extant African apes and
may be characteristic of the African ape and human clade.
44. E. Vogelet al.,J. Hum. Evol.55, 60 (2008).
45. R. T. Kono,Anthropol. Sci.112, 121 (2004).
46. R. T. Kono, G. Suwa,Bull. Natl. Mus. Nat. Sci. Ser. D.
34,
1 (2008).
47. R. K. Malenky, R. W. Wrangham,Am. J. Primatol.32,
1 (1994).
48. Two of five availableAr. ramidusindividual molar rows
show comparable dentine exposure at all three molar
positions. The remaining three individuals show either
weak or no clear gradients between adjacent molar pairs.
In contrast toAustralopithecus, bothAr. ramidusandAr.
kadabbamolars exhibit deep dentine exposures suggestive
of erosive rather than abrasive wear (4).
49. F. E. Grine, P. S. Ungar, M. F. Teaford, S. El-Zaatari,
J. Hum. Evol.51, 297 (2006).
50. T. D. Whiteet al.,Science326, 67 (2009).
51. M. Sponheimeret al.,J. Hum. Evol.51, 128 (2006).
52. A. D. Di Fiore, R. C. Fleischer,Int. J. Primatol.26, 1137
(2005).
53. C. J. Campbell, Ed.,Spider Monkeys: Behavior, Ecology
and Evolution of the GenusAteles (Cambridge Univ.
Press, Cambridge, 2008).
54. T. D. White, G. Suwa, S. Simpson, B. Asfaw,Am. J. Phys.
Anthropol.111, 45 (2000).
55. E. D. Shields,Am. J. Phys. Anthropol.128, 299 (2005).
56. For funding, we thank NSF (grant nos. 8210897,
9318698, 9512534, 9632389, 9727519, 9729060,
9910344, and 0321893 HOMINID-RHOI) and the Japan
Society for the Promotion of Science (grant nos. 11640708,
11691176, 14540657, 16405016, 16770187, 17207017,
19207019, 19770215, and 21255005); the Ministry of
Tourism and Culture, the Authority for Research and
Conservation of the Cultural Heritage, and the National
Museum of Ethiopia for permissions and facilitation; the Afar
Regional Government, the Afar people of the Middle Awash,
and many other field workers for contributing directly to the
data; the institutions and staff of National Museum of
Ethiopia, National Museums of Kenya, Transvaal Museum
South Africa, Cleveland Museum of Natural History, Royal
Museum of Central Africa Tervuren, Naturalis Leiden, and the
Department of Zoology of the National Museum of Nature
and Science (Tokyo) for access to comparative materials;
H. Gilbert for graphics work on Fig. 1; D. DeGusta and
L. Hlusko for editorial assistance; R. Bernor, L. de Bonis,
M. Brunet, M. C. Dean, B. Engesser, F. Guy, E. Heizmann,
W. Liu, S. Moya-Sola, M. Plavcan, D. Reid, S. Semaw, and
J. F. Thackeray for cooperation with comparative data and
fossils; and T. Tanijiri, M. Chubachi, D. Kubo, S. Matsukawa,
M. Ozaki, H. Fukase, S. Mizushima, and A. Saso for analytical
and graphics assistance.
Supporting Online Material
www.sciencemag.org/cgi/content/full/326/5949/69/DC1
Materials and Methods
SOM Text
Figs. S1 to S33
Tables S1 to S9
References
4 May 2009; accepted 18 August 2009
10.1126/science.1175824
RESEARCH ARTICLES
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 99
Research Articles

70. 70 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Careful Climbing in the Miocene: The
Forelimbs ofArdipithecus ramidus
and Humans Are Primitive
C. Owen Lovejoy,
1
*Scott W. Simpson,
2
Tim D. White,
3
*Berhane Asfaw,
4
Gen Suwa
5
TheArdipithecus ramidushand and wrist exhibit none of the derived mechanisms that restrict
motion in extant great apes and are reminiscent of those of Miocene apes, such asProconsul.The
capitate head is more palmar than in all other known hominoids, permitting extreme midcarpal
dorsiflexion.Ar. ramidusand all later hominids lack the carpometacarpal articular and ligamentous
specializations of extant apes. Manual proportions are unlike those of any extant ape. Metacarpals
2 through 5 are relatively short, lacking any morphological traits associable with
knuckle-walking. Humeral and ulnar characters are primitive and like those of later hominids. The
Ar. ramidusforelimb complex implies palmigrady during bridging and careful climbing and
exhibits none of the adaptations to vertical climbing, forelimb suspension, and knuckle-walking
that are seen in extant African apes.
T
he grasping hand is a hallmark of all pri-
mates (1 ), and elaborated forelimb flex-
ibility characterizes all extant hominoids.
Hands have played a central role in human evo-
lution and perhaps even in the emergence of
higher cognition (2). Although we no longer use
our hands in locomotion, our ancestors must
have. Chimpanzees, which knuckle-walk and ver-
tically climb, have long fingers compared to their
thumbs. Gorilla proportions are similar but less
extreme. Did our ancestors also knuckle-walk
and adapt to vertical climbing and suspension as
the modern African apes have, or did our anat-
omy emerge directly from a more generalized
Miocene ancestor, as some mid-20th century
anatomists argued (3)? Answers to these cen-
tral questions are now provided byArdipithecus
ramidus.
The AramisAr. ramidussample includes all
bones of the forelimb except for the pisiform
and some terminal phalanges (4). Of particular
importance areARA-VP-7/2, a forelimb skeleton,
andARA-VP-6/500, a partial skeleton preserving
a lower arm and most of both hands (Fig. 1). We
describe here the salient aspects of the taxon’s
forelimb anatomy and their implications for hom-
inid evolution. These and other data (5–7) show
thatAr. ramiduswas both terrestrially bipedal
and arboreally capable.
Medial metacarpus and phalanges. The me-
dial metacarpals (Mc2 to Mc5) ofAr. ramidus
are short (figs. S1 and S2 and table S1) com-
pared with those of African apes. Their heads
lack any traits associated with knuckle-walking,
such as prominent ridges and/or grooves (Fig. 2
and fig. S3).Ar. ramidusmetacarpal heads
exhibit marked, proximally located, dorsal in-
vaginations that are associated with their meta-
carpophalangeal (MP) joint’s collateral ligaments
(reflecting routine hyperdorsiflexion at this
joint), as do those of some Old and New World
monkeys and some Miocene hominoids (8–10),
includingProconsul[reviewed in (11)]. Such
constrictions have become minimal in all extant
hominoids but for different functional reasons
(see below) (Fig. 2 and figs. S4 and S5).Ar.
ramidusalso lacks the expansion of the meta-
carpal heads that is typical of knuckle-walking
apes (Fig. 2 and figs. S4 to S6).
The basal articulation of theAr. ramidusMc5
is cylindrical/condyloid. Its articular surface ex-
tends onto the dorsum of the bone (Fig. 2), as it
does inAustralopithecus afarensis(12,13),Homo
sapiens,Proconsul heseloni(14–17),Pierolapithe-
cus catalaunicus,Equatorius africanus, and papio-
nins [this study; see also (10)]. It is entirely unlike
the immobilized, noncondylar, planar homolog
inPongo, Pan,andDryopithecus laietanus.
SomeGorillaMc5-hamate joints are sufficiently
compliant to allow energy dissipation via their
soft tissues, but they are too angular at their base
to permit substantial motion without cavitation.
In contrast, the dorsal prolongation of theAr.
ramidusMc5 surface appears to have permitted a
greater range of flexion and extension than in any
extant hominoid (>20° extension and >25°
flexion from a neutral position; Fig. 2). This
may reflect selective modification of underlying
positional information (type 1) (18 ) and/or chon-
dral modeling (type 4).
The mating surfaces for the Mc4 and Mc5
bases on theAr. ramidushamulus lack the an-
gled and distopalmarly extending facets that are
present inPan,Pongo, and (to a lesser degree)
Gorilla. These are additional reflections of re-
stricted mobility and increased rigidity of the
Mc4- and Mc5-hamate joints in these apes. The
hamate’s primitive state (a hamulus still permit-
ting substantial Mc5-hamate dorsiflexion) in
Sivapithecus parvada(19),Oreopithecus bam-
bolii,Proconsulsp. (14–17), andE. africanus
[(10), this study] implies that a more rigid hamate-
metacarpal articulation emerged in parallel be-
tween early ancestors of African apes and
Pongo.This rigidity was therefore unrelated to
knuckle-walking and may have emerged only in
large-bodied suspensory forms. Mobility in the
Mc5-hamate joint, as seen inAr. ramidus, fa-
cilitates and/or reflects compliance of the palm
with the substrate during palmigrady, as well as
providing hypothenar opposition to the first ray
during grasping (20 ). The latter proved to be a
pivotal exaptation for extractive foraging and
eventually tool using and making in bothAus-
tralopithecus(21) andHomo, but especially in
the latter.
TheAr. ramidusphalanges of rays 2 to 5
(figs. S7 to S15 and table S2) are shorter than
those ofPanbut longer than those ofGorilla,
relative to body size. However, because of
elongation of the medial metacarpus in African
apes, phalangeal-to-metacarpal length ratios in
Ar. ramidusare more similar to those of Old
World monkeys (figs. S8 and S14). Manual/
pedal phalangeal length ratios are similar in
Ar. ramidus,Pan, andGorillaand are substan-
tially higher than those inProconsul(fig. S15).
The first ray.Unlike in apes, theAr. ramidus
first ray is relatively large and robust (Fig. 1,
figs. S16 to S22, and tables S1 to S3). The Mc1
base flares outward with a prominent attachment
for the abductor pollicis longus muscle, as in all
later hominids. In both size and proportions, its
head is robust and dorsally expanded, with well-
defined and asymmetric sesamoid grooves as in
Homobut in distinction toPanorGorilla. The
ARA-VP-6/500Mc1/Mc5 length ratio is close to
that ofProconsuland exceeds those of extant
African apes, largely reflecting elongation of the
medial metacarpus in the latter (fig. S17).
The first ray’s terminal phalanx exhibits a
symmetrically constructed, rugose insertion gable
(22,23) for the flexor pollicis longus (Fig. 1,
inset), in contrast to African apes in which this
muscle’s tendon is reduced or absent (3). The
first ray’s carpometacarpal, MP, and interphalan-
geal joints are also somewhat larger and more
humanlike in shape than those of African apes,
suggesting greater thenar mobility and/or pos-
sibly greater loading during ontogeny [that is,
types 1 and/or 4 (18)]. The trapezium’s tuberos-
ity is large and projects toward the palm (Fig. 3),
deepening the adjacent groove for the flexor carpi
radialis tendon.
Ardipithecus ramidus
1
Department of Anthropology, School of Biomedical Sciences,
Kent State University, Kent, OH 44240-0001, USA.
2
Depart-
ment of Anatomy, Case Western Reserve University School of
Medicine, Cleveland, OH 44106-4930, USA.
3
Human Evolution
Research Center, and Department of Integrative Biology, 3101
Valley Life Sciences Building, University of California, Berkeley,
CA 94720, USA.
4
Rift Valley Research Service, Post Office
Box 5717, Addis Ababa, Ethiopia.
5
The University Museum,
the University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-
0033, Japan.
*To whom correspondence should be addressed. E-mail:
[email protected] (C.O.L.); [email protected] (T.D.W.)
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 200970e1
Ardipithecus ramidus

71. 71www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
The central joint complex and midcarpal
joint.Ar. ramidusprovides pivotal evidence
about the natural history of the immobile (24) or
fixed unit (25) of the hand (trapezoid/capitate/
proximal Mc2 and Mc3) that we refer to as the
central joint complex (CJC). Most previous analy-
ses have presumed a modern apelike antecedent
of the human CJC, without due regard for the key
anatomy described below.Ar. ramidusshows
that CJC anatomy is pivotal to understanding the
evolution of the hominoid hand.
A suite of derived structures stabilizes the
CJC in extant great apes. These can be sum-
marized as follows: (i) the palmar portion of the
lateral (radial) side of the capitate has been ex-
tended distally to create a novel capitate-Mc2
facet that now allows the Mc2 to act as a buttress
against rotation in the CJC (Figs. 3 to 5), and (ii)
the lateral portion of the capitate’ s dorsodistal
surface has been withdrawn proximally. These
novel articular geometries have resulted in a CJC
with a complex polyaxial interface; the dorsal
portions of their joint surfaces are angled op-
posite to their palmar portions (Fig. 5H), creating
a partial screw effect.
This morphology has several extraordinary
advantages. The partial screw effect (26) pro-
vides a route for cartilage-sparing energy dissipa-
tion during loading. In palmar view (Fig. 5H), it
can be seen that external axial rotation of the Mc3
upon the capitate causes distention of the joint,
which can then be resisted by tension in greatly
hypertrophied carpometacarpal ligaments. How-
ever, the Mc2 and Mc3 cannot rotate (from a
neutral position) in the opposite direction, be-
cause such internal rotation is now blocked by
the capitate; that is, by the position of the capi-
tate’s palmar process, which has displaced the
(less stable) trapezoid and abuts against the Mc2.
Finally, the overall proximodorsal-to-
distopalmar angulation of the Mc3-capitate in-
terface converts any dorsopalmar shear to tension.
Thus, dorsopalmar shear and torsion acting
in the CJC are both resisted by the greatly en-
larged and axially disposed carpometacarpal
ligaments (see below and Figs. 3 to 5). The topo-
graphically complex (polyaxial) and heavily
buttressed African ape CJCs increase rigidity
and permit energy dissipation during suspension
(Figs. 3 to 5) and vertical climbing (27 ), and
possibly during knuckle-walking.
The CJC ofAr. ramidusis very different. It
exhibits the simple, planar joint interface shared
with some Old World monkeys andProconsul
sp. (14–17), in which all four CJC elements meet
one another at nearly a single dorsopalmar axis
(Figs. 3 and 5). This configuration is less com-
petent to restrict torsion and dorsopalmar shear
within the CJC, because it can only resist rotation
and/or shear via its slight subchondral surface
undulations, whose relative translation only mini-
mally distends the joint.
The configuration of CJC joint geometry can
be assessed visually (fig. S23) and by two met-
rics: angulation of the capitate’s hamate facet
relative to the long axis of the Mc3 (fig. S23)
and mediolateral angulation of the Mc3-capitate
joint’s dorsal surface relative to the Mc3 shaft’ s
axis (figs. S24 and S25). Whereas both metrics
record minimal angulation in palmigrade taxa,
such asPapioandAr.ramidus, both are elevated
inPanandGorilla. However, angulation of this
joint is also present inHomo. In the latter, it may
have been associated with transfer of the styloid
anlage (28) from the capitate to the Mc3, because
the transfer had not yet occurred inAr. ramidus,
which also did not yet exhibit any angulation (it
is actually slightly negative). Thus, the transfer
may have either been directly associated with the
introduction of Mc3 ulnar deviation or, alterna-
tively, a pleiotropic consequence of changes in
pattern formation underlying it. But it seems
quite probable that the introduction of this angu-
lation was to enhance the opposition of the me-
dial rays with the hypertrophied thumb (29).
The development of axial ligaments and their
attachment patterns is also an integral part of the
Fig. 1.Digitally rendered composite hand ofARA-VP-6/500in palmar view. Lateral (bottom left), dorsal
(bottom center), and medial (bottom right) views are also shown. All carpals except for the trapezium are from
the left side. The trapezium, Mc2, and some phalanges have been mirror-imaged. The first ray’s proximal
phalanx is fromARA-VP-7/2and has been size-adjusted based on estimated Mc4 length (7) in the two
specimens (estimated inARA-VP-6/500). Intermediate and terminal phalanges are provisionally allocated to
position and side only, except forARA-VP-6/500-049, a pollical terminal (inset; cast). Note its clearly marked
insertion gable for the flexor pollicis longus as in modern humans. Imagery is based on computed
tomography (CT) scans taken at 150-mm voxel resolution by a peripheral quantitative computed tomography
(pQCT) XCT-Research SA+ instrument (Stratek, Pforzheim, Germany), and processed by use of software
Analyze 6.0 (Mayo Clinic, Rochester, MN) and Rapidform 2004/2006 (Inus Technology, Seoul, South Korea).
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
70e2
RESEARCH ARTICLES
Research Articles

72. 72 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
above CJC functional morphology. In extant
great apes, massive centrally located carpometa-
carpal ligaments pass through deep notches in the
capitate and hamate to attach to Mc2 to Mc4
(Figs. 3 to 5). Thickening of these CJC ligaments,
as indicated by substantial expansion of their
passageways from capitate to Mc2 and Mc3, has
occurred in virtually all hominoids engaged in
suspension, includingPongo, Pan, andGorilla.
Osteologically, such thickening is expressed as
isolation of dorsal and palmar intermetacarpal ar-
ticular facets separated by deep cylindrical grooves.
WhereasD. laietanusexhibits the extant
great ape condition, similar capitate and Mc base
notching is entirely absent inProconsul,Ar.
ramidus,Australopithecus, andHomo, reflecting
the absence of similarly robust carpometacarpal
ligaments in these taxa. The CJC’s primitive state
inAr. ramidusstrongly suggests that the earliest
hominids and their immediate ancestors did
not engage in habitual suspension or vertical
climbing.
The proximal part of theAr. ramiduscapitate
also differs from all known hominoid homologs.
Its head and neck lie more palmar to the bone’ s
corpus (Fig. 4), and the head’ s dorsal articular
surface blends distally into a broad dorsal de-
pression that accommodates the distal edges of
the scaphoid and lunate during midcarpal joint
dorsiflexion (30). In knuckle-walking African
apes, the capitate head’s surface usually bears a
subchondral tidemark recording maximum dor-
siflexion of the scaphoid, which engages only a
portion of the capitate’ s dorsal surface (Figs. 3
and 4) (31–33). InAr. ramidus, the scaphoid’s
distal articular margin is deeply concave. Here,
dorsiflexion did not terminate on the capitate
head, but continued onto the dorsal surface of
the capitate’ s anatomical neck (Figs. 3 and 4). In
Ar. ramidus, the scaphoid completely engulfed
the head and thus entered hyperdorsiflexion,
greatly enhancing its capacity for palmigrady at
the midcarpal joint.
The palm ofAr. ramidusis made relatively
narrow by the dominance of its spherical lunate
but mostly by its markedly narrow trapezoid
(Fig. 3). The capitate head’s palmar (and thereby
eccentric) location may have limited midcarpal
rotation inAr. ramidus. However, its symmetric
lunate, narrow trapezoid (figs. S26 to S29 and
table S3), and more laterally facing scaphoid
may have allowed greater compensatory radio-
carpal axial rotation, and possibly greater radial
deviation (see below), than in extant African
apes and humans (in which the capitate’s head is
broad and less spherical). These changes appear
to have occurred independently in humans for
palmar grasping and inPanandGorillafor ver-
tical climbing and knuckle-walking. If so, they
have caused conflation of the functional and
evolutionary history of the midcarpal joint [see,
for example, (32,33)].
Ar. ramidusestablishes that the null hypoth-
esis for evolution of the human hand must be
that hominids have never had modern apelike
CJCs or their attendant behaviors, in contrast to
previous assumptions (32 –34). Alteration of
the primitive CJC in later hominids has been
largely restricted to the following: (i) dorsal ele-
vation of the capitate head; (ii) broadening of
the capitate and trapezoid (for greater palmar
span) (Fig. 3, figs. S28 and S29, and table S3);
(iii) reduction of the surface relief in the Mc3-
capitate joint (thereby permitting greater kine-
matic compliance, probably initiated after the
elimination of the forelimb from weight-bearing
locomotion); and (iv) transfer of the os sty-
loideum element from the capitate to the Mc3
(28,29).Au. afarensisexhibits a more human-
like, albeit intermediate, condition. It shows evi-
dence of all but the fourth of these shifts, each of
which presumably facilitated nonlocomotor palmar
grasping.
The much-discussed lateral orientation of the
capitate’s Mc2 and Mc3 facets [see review in
(35)] in later hominids is probably now best
viewed as a collateral pleiotropic effect [type 2A
(18)] of mediolateral expansion of the radial
wrist associated with increased thenar size and
robusticity, because both the trapezium and trap-
ezoid are enlarged inHomoas compared with
Ar. ramidus. The only possible kinematic signif-
icance of these minor variations of the capitate
[that is, palmar cupping (24)] is most likely a
consequence of cartilage modeling during on-
A
B
C
D
E
F
G
I
H
Fig. 2.Metacarpus ofAr. ramidus.(AtoD)LeftMc4sofamodernhuman(CMNH-HT-1617), ARA-VP-7/2-G,
and a chimpanzee (CMNH-B1708). Views include medial, proximal, dorsal, and distal. Note the very large
notch on the lateral side of the base of thePanspecimen for transmission of its carpometacarpal ligament (a
much smaller notch is present in theAr. ramidusspecimen) and the deep knuckle-walking sulcus on the
head’s dorsum (see also fig. S3). The latter is absent in hominids. (EtoH) Equivalent views of the left Mc5 of
the same extant individuals andARA-VP-6/500-036. The basal articular surface of theAr. ramidusMc5 is
cylindrical and is continued well onto the shaft’s dorsum, unlike either of the other specimens. Its form is
almost identical to its homolog inEquatorius africanus. ThePansurface is virtually flat and nonmobile. (I)
Mc1s of the same extant individuals (Pan, top left;Homo, top right), and twoAr. ramidus[ARA-VP-6/1638
(bottom right) andARA-VP-6/500-015(bottom left)]. Note the very broad Mc1 sellar bases in the hominids,
demonstrating that palmar grasping is the probable antecedent condition for extant African apes and humans.
This may have been lost in conjunction with apparent first-ray involution in apes (71,72). Scale bar, 2 cm.
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
70e3
Ardipithecus ramidus
Ardipithecus ramidus

73. 73www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
togeny (type 4) or is simply inconsequential
inherent variation (type 2B and/or 5). The cap-
itate’s Mc2 and Mc3 angles are close to 90° in
bothAr. ramidusandAu. anamensis(35). This
relationship is evidently the primitive condition
in both hominids and extant African apes and
therefore has limited value in reconstructing man-
ual function.
In retrospect, it appears that subsequent to
their last common ancestor with hominids, the
Gorillaclade (and sometime thereafter the chim-
panzee clade) faced an adaptive conundrum.
Palmar conformity to substrates (primitive and
retained inAr. ramidus) is obviously beneficial to
climbing. However, this imparts risk of injury to
the hand from insufficient joint integrity and
energy dissipation mechanisms that are appar-
ently required during vertical climbing, suspen-
sion, and/or knuckle-walking. OnlyPanappears
to have eliminated substantial joint mobility in
the Mc4- and Mc5-hamate joints entirely, but
both African apes have evolved a sophisticated
stabilizing mechanism in their CJCs.
Scaphoid morphology, ulnar retraction, and
radiocarpal joint. The scaphoid ofAr. ramidus
differs substantially from those of Old World
monkeys,Proconsul, and other Miocene apes
not only by the fusion of the os centrale, but
by the elongation of its tuberosity and palmar
deflection of its distal facet(s) for the trapezi-
um and trapezoid (Fig. 3). These changes prob-
ably accompanied deepening of the carpal
tunnel on the ulnar side of the wrist to prevent
flexor tendon bowstringing by distal prolonga-
tion and increased prominence of the hamulus.
Because the ulna was withdrawn to permit greater
adduction, there was apparently a general
deepening of the wrist’s proximal transverse arch
(25). This was probably also present inPier-
olapithecusat ~12.5 million years ago (Ma) (36).
Fusion of the os centrale may have been a col-
lateral consequence [nonselected pleiotropic ef-
fect; type 2B (18)] of this reorganization. If so,
ulnar retraction must have occurred indepen-
dently in the ancestors of the hominid–African
ape clade andPierolapithecus, because the latter
retained a separate os centrale (36). None of these
advanced characters is present inSivapithecus.
Pongoalso exhibits a form of ulnar retraction,
but scaphoid– os centrale fusion is only rarely
seen. Moreover, unlike the extant African
apes,Pongoalso exhibits substantial midcarpal
rotation (37 ).
In extant African apes, the radiocarpal joint
experiences large collision loads during knuckle-
walking. These are shared by the radiolunate
and radioscaphoid articulations and account for
a less proximodistal orientation of the scaphoid
in African apes as compared with those of orang-
utans andAr. ramidus. A markedly rugose and
heavily buttressed scaphoid tubercle receives
the hypertrophied styloscaphoid ligament in
knuckle-walking apes, which restricts dorsi-
flexion that is imposed by ground reaction
force (GRF) (38 ) and thereby maintains joint
integrity and potentially contributes to the
dissipation of collision forces. In addition, the
ligamentous attachment area on the scaphoid’s
dorsum for the dorsal radiocarpal ligaments
(that is, its nonarticular area; Figs. 3 and 4) is
also enlarged in these apes.
TheAr. ramidusscaphoid is unlike its African
ape counterpart. Its nonarticular area is narrower,
AB C
Fig. 3.Articulated wrists (left sides) of hominoids (no pisiform) in maximum dorsiflexion. (A )Pan
(CMNH-B1708); (B)ARA-VP-6/500(casts; the trapezium ofARA-VP-6/500is a rapid prototyping
model based on CT scan of the contralateral side); (C )Homo(CMNH-HT-1617). (Top row) Lateral
view (Mc3 surface of capitate is vertical). The scaphoid’s radial surface faces dorsolaterally inARA-VP-
6/500. It faces slightly more proximal inHomo, but demonstrably more so inPan. Its more limited
articular extent inPanallows enlarged radioscaphoid ligaments, which insert into its nonarticular
area (see text, Fig. 4F, and fig. S30). (Second row ) Dorsal view aligned with the capitate-hamate
joint plane made vertical. TheARA-VP-6/500scaphoid completely engulfs the capitate’s head,
advancing into a furrow formed between its articular neck and the proximal surface of the trapezoid.
That ofPanarticulates only with the more proximally directed capitate head. Both extant taxa have
substantial mediolateral wrist expansion for palmar grasping inHomoand for knuckle-walking in
Pan. Distally, there is a large styloid element on theARA-VP-6/500capitate [see text and (28 )]. It has
been transferred to the Mc3 inHomo, leaving behind a broad V-shaped recess. (Third row ) Proximal
view (plane of the capitate’s Mc3 surface is vertical). The more proximally oriented radial surface of
Panis obvious, although that ofHomoalso faces less medially than that ofAr. ramidus(note the
lunate’s subchondral tidemark of maximum dorsiflexion superiorly as a slight dorsal ridge inPan).
(Fourth row) Distal view (the capitate-trapezoid joint plane is vertical). The capitate-trapezoid axis is
simple inARA-VP-6/500, whereas it has become medially concave inHomo, either facilitating or
reflecting compliance within the carpus for palmar grasping. The distal carpal row is generally
broadened inHomo. InPan, the capitate’s palmar portion has been extended distally, and large
notches allow transit of its massive carpometacarpal ligaments (absent in hominids). Its hamate is
broadened, with a more distally projecting and mediolaterally expanded hamulus. The distal face of
theARA-VP-6/500capitate is a simple plane interrupted only by transverse palmar and dorsal
furrows; its trapezoid is mediolaterally slender, whereas it has been mediolaterally expanded in
HomoandPan. Note the unusually large styloid element inAr. ramidus[text and (28 )].
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
70e4
RESEARCH ARTICLES
Research Articles

74. 74 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
and its tuberosity is of small caliber, elongate, and
relatively gracile, exhibiting virtually no expan-
sion where it joins the bone’s corpus. It would
therefore be poorly suited to sustain large loads
applied to its tuberosity by the styloscaphoid
ligament. In humans, the scaphoid has been
further reorganized to accommodate an enlarged
trapezium.Ar. ramidusis near the mean of mod-
ern humans in an index reflecting the scaphoid’s
potential to sustain GRF (fig. S30), confirming
that theAr. ramidusscaphoid lacks evidence of
high-impact loading.
TheAr. ramidus lunate lies directly proximal
to the head of the capitate, which accounts for
its more spherical form than in humans or extant
African apes. Lunate-hamate contact during dor-
siflexion was minimal inAr. ramidus,as in Old
World monkeys,Proconsul(16), and humans
(Fig. 3), whereas substantial lunate-hamate con-
tact inPanandGorillaappears to be derived by
capitate head expansion and scaphoid reorienta-
tion, probably in response to the collision load-
ing of knuckle-walking, becausePongolacks
these features.
Lunate position inAr. ramidusalso accounts
for the previously unexplained lunate/scaphoid
facet area ratios on the distal radius ofAustra-
lopithecus, which exceed those of both humans
and extant African apes (table S4). However, it
is now clear that scaphoid reorientation occurred
in African apes for knuckle-walking and inde-
pendently in humans for increased palmar span.
The similar ratio inArdipithecusandAustralo-
pithecustherefore represents the primitive con-
dition for both hominid and extant African apes.
In summary, the hand ofAr. ramidusappears
almost entirely primitive relative to the anatom-
ical specializations seen in extant apes (for ex-
ample, metacarpal elongation, elaboration of
CJC articulations and ligaments, novel capitate
geometry, reorientation of the scaphoid’ s radial
surface, enlargement of the radioscaphoid liga-
ment, relative diminution of the first ray, etc.).
Ar. ramidusestablishes that these changes in the
ape hand are independent specializations for ar-
boreal access and terrestrial travel (vertical climb-
ing, forelimb suspension, knuckle-walking) and
were apparently never established in hominids,
which retained a more generalized, substrate-
conforming, grasping hand.
Radius and ulna.TheAr. ramidussample
includes a complete radius [ARA-VP-6/500-039;
although damaged, its length is largely preserved
(7)] and a second intact distal radius (ARA-VP-
7/2-B). Both exhibit greater distal articular sur-
face angulation relative to the shaft axis than do
those other early hominids [as previously reported
(39)], which is consistent with the scaphoid’s more
laterally facing radial facet (Fig. 3). This is now
clearly identifiable as a primitive character, as
previously surmised (35), and is not an adaptation
to knuckle-walking [contrary to (40)]. Moreover,
Ar. ramidusindicates that the radiocarpal joints of
African apes and humans have become broadened
mediolaterally in parallel, presumably for knuckle-
walking in the African apes and as a consequence
of elaboration of the pollex for tool-using or
-making and/or for extractive foraging in homi-
nids. Scaphoid expansion and palmar broadening
almost certainly underlie reduced radiocarpal joint
angulation in later humans. Morphometrically based
suppositions attributing these various characters to
a history of knuckle-walking (40) or suspension in
Australopithecushave been critiqued previously on
theoretical grounds (41) and are now moot, because
anatomical evidence indicates thatAr. ramiduswas
never reliant on either.
The proximal ulna exhibits substantial differ-
ences among extant hominoids (42,43). Some
Miocene hominoid ulnae, as well as those of
colobines, generally exhibit both long olecra-
nons and anteriorly facing trochlear notches, a
combination that is consistent with pronograde
above-branch quadrupedality. Two proximal
ulnae (ARA-VP-6/500-051 andARA-VP-7/2-C)
were recovered at Aramis and show that the
trochlear notch inAr. ramidusfaces anteriorly
(table S5 and figs. S31 and S32). A cranially
oriented trochlear notch with a retroflexed olec-
A B C DE
F
G
H
*
*
* *
*
*
*
*
Capitate Head Position
A r. ramidus
Au. afarensis
Pan
Homo
Gorilla
I
Fig. 4.Hominoid capitates and scaphoids. (A toE) Left capitate. Top, lateral (radial) view; bottom,
distal capitate surfaces. (A)Pan(CMNH-B1718); (B)Homo(KSU-02234); (C)Au. afarensis[A.L. 333-40
(reversed)]; (D)Ar. ramidus(ARA-VP-7/2F); (E)Ar. ramidus(ARA-VP-6/500-058). (Bottom row) Hominid
Mc3 surfaces exhibit only shallow dorsal and palmar transverse furrows. InPan, the palmar capitate
surface extends distally (toward the viewer), providing a buttress against the Mc2 [angled arrow in (A)]
(see also Fig. 5H) that prevents Mc3-capitate rotation. Angled white arrows point to broad, shallow,
nonblocking Mc2 facets in hominids. TheHomoandAu. afarensiscapitates are mediolaterally
expanded by a beveled Mc2 surface (dotted lines); it is much narrower inAr. ramidus. A white asterisk
marks a subchondral defect on (B) of no functional significance. The styloid element is marked by a
black asterisk in each specimen except the human, in which it is instead fused to the Mc3 (Figs. 3 and
5). (Top row) Vertical arrow in (A) points to a large carpometacarpal ligament canal (as does horizontal
arrow in bottom row). Hominids exhibit only uninterrupted cartilage surfaces. TheAr. ramiduscapitate
head is palmar, permitting marked dorsiflexion (see also Fig. 3). The head is more dorsal in apes and
intermediate inAustralopithecusandHomo. InPan, the edge of the capitate’s scaphoid articular
surface is sharply delimited (horizontal arrow). InAu. afarensisandHomo, the head and neck blend
imperceptibly, but a subchondral tidemark indicates maximum dorsiflexion (horizontal arrows). (FtoH)
Left scaphoids, trapezoid/trapezium surface faces superior; tuberosity is to the right, and the radial surface
faces down to the left. (F)Pan(CMNH-1718), (G)Homo(KSU-12202), and (H)ARA-VP-6/500-062.Pan
scaphoids often exhibit much shallower capitate notches (if present) (white asterisks) than hominids.
Human scaphoid notch depth varies, but it is not typically as great as inAr. ramidus(n=2;thehuman
scaphoid notch shown is exceptionally deep). (I) Location of the capitate head inAr. ramidus[for method,
see (73)]. Scale bar, 2 cm.
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
70e5
Ardipithecus ramidus
Ardipithecus ramidus

75. 75www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
ranon, as in African apes, enhances the triceps
moment arm in full elbow extension. A more
anteriorly facing notch favors triceps leverage at
mid-flexion (42).
Although a substantially foreshortened olec-
ranon is shared among hominids,Pan, and
Gorilla, as well as with other middle and late
Miocene hominoids that exhibit ulnar withdraw-
al at the wrist, a proximally oriented trochlear
notch must be reflective of habitual suspension,
because it is also found inPongo. Retroflexion
of the trochlear notch has, until now, been re-
garded as primitive, because the last common
ancestor was presumed to have been adapted to
some form of suspension. Instead, we conclude
that an anteriorly facing notch is primitive, as-
sociated with an adaptation to careful climbing
and bridging (44). Its presence inAr. ramidus
further supports the hypothesis that hominids
have never been adapted to either suspension or
vertical climbing and that human and ape proxi-
mal ulnar morphology converged for different
reasons. The data forAr. ramidusare consistent
with arguments that early hominid ulnar mor-
phology might reflect substantial manipulative
skills (42), and that early hominids might have
been involved in activities such as extractive
foraging. In any case, ulnar morphology is con-
gruent with transarticular force being generally
greatest in full extension in both humans and
great apes: during suspension inPongo, suspen-
sion and knuckle-walking in extant African
apes, and myriad possible manipulative activ-
ities unrelated to locomotion in hominids.
Another ulnar character in extant African
apes, here termed the“flexor expansion,”is a
typically prominent, proximomedial enlarge-
ment adjacent to the posterior subcutaneous
surface of the olecranon. This is created by the
latter’s juncture with a markedly deep excava-
tion of the proximal-most origin of the deep
digital flexor (fig. S33). When the ulna is
viewed anteriorly, this medial projection reaches
great prominence only in African apes (45 ) and
is not present inAr. ramidus. It probably
reflects the expansion of the mesenchymal
territory of the flexor muscles’enthesis. This
flexor expansion is derived and uniquely as-
sociable with knuckle-walking based on the
role of the digital flexors (eccentric contraction
and/or passive tension in their connective tis-
sue capsules) and the enlarged relative muscle
mass inPanandGorilla(38), as well as the
tubercle’s absence in orangutans and all homi-
nids, includingAr. ramidus. The hypertrophy
in extant African apes may have also been
accompanied by myological reorganization in
the forearm that resulted in gracilization or loss
of the flexor pollicis longus tendon’s attach-
ment on the first ray.
Humerus.TheAr. ramidushumerus sample
includes a well-preserved proximal humerus
with shaft (ARA-VP-7/2-A ), a well-preserved
humerus shaft (ARA-VP-1/4 )(39) (fig. S34),
and multiple distal humeral shafts lacking most
or all of their distal articular surfaces (totaln=
7). Distal humeral morphology is largely con-
served among hominoids, which vary only min-
imally in a variety of minor phenetic characters
associated with full extension at the elbow (46),
including a deep zona conoidea with a pos-
teriorly extended lateral wall and a spherical
capitulum with a short radius of curvature. Both
are present inAr. ramidus(ARA-VP-7/2-A)( 47).
TheAr. ramidusproximal humerus (39) exhibits
equally typical hominid characters, including
an elliptical head and shallow bicipital groove
(48). It exhibits only minimal torsion (fig.
S34).
The deltopectoral crest ofARA-VP-1/4is
elevated and rugose (39). The common assump-
tion that this reflects differentially powerful arm
musculature can be rejected for two reasons.
First, the morphotype is shared among mod-
ern humans, cercopithecoids,P. heseloni(14),
Sivapithecus indicus(49),Ar. ramidus, and
Au. afarensis. Second, the crest is consistently
less developed in apes and virtually absent in
brachiating gibbons. It is therefore a trait dic-
tated primarily by positional information rather
than loading (type 1 and/or 2) (18,50,51). A
rugose deltoid crest is clearly primitive (52),
retained in hominids but substantially modified
in suspensory, vertical-climbing, and knuckle-
walking apes—an observation made more than
75 years ago (53). A possible explanation of
crest reduction in apes is that it reflects increased
intermuscular fusion, as has occurred in gibbons
(54), which also use suspension and exhibit al-
most no deltopectoral cortical surface manifesta-
tions. Deltopectoral morphology may therefore
serve as a key indicator of locomotor habitus in
other fossil hominoids.
Conclusions. The forelimb has played a de-
finitive role in most chronicles of human evolu-
tion since Huxley’s and Keith’s original accounts
(55,56). Most recent narratives of its anatomical
and behavioral evolution have emphasized a her-
itage of suspensory locomotion, vertical climbing,
and knuckle-walking in the common ancestors
that humans shared with extant African apes.
Encouraged by human and chimpanzee genetic
similarity and cladistic analyses, such views have
come to dominate recent explications of early
hominid evolution (27,40), although alternative
interpretations based on classical comparative anat-
omy have long differed (3,57,58).Ar. ramidus
now permits resolution of these controversies. It
indicates that, although cranially, dentally, and
postcranially substantially more primitive than
Australopithecus, these known Late Miocene to
Early Pliocene hominids probably all lacked the
numerous, apparently derived, forelimb features
of extant African apes. The most probable hy-
pothesis to explain these observations is that
hominids never passed through adaptive stages
Fig. 5.(AtoD) Schematic of the
articular geometry of the left CJC in
hominoids. (A) Primitive planar con-
dition of the CMc2 and CMc3 joints
as seen inAr. ramidus(andHomo).
(B) The primitive CJC cannot resist
shear or torsion (pronation/supination)
except by distention by joint surface
irregularities. (C) In African apes, the
palmar capitate is insinuated distally
into the interface between the Mc2
and Mc3, blocking CJC rotation (as-
terisk). (D) Dorsal schematic view of
the Mc3-capitate portion of the CJC
shown in (C). The dorsal and palmar
portions of the capitate-Mc3 joint are
oriented oppositely, creating a screw
mechanism. Supination (26) of the
Mc3 results in joint distention resisted by its hypertrophied carpometacarpal
ligaments (red springs) (the location and lateral orientation of the capitate’s
Mc2 facet are indicated by⊥). (EtoH) Medial, dorsal, lateral, and palmar views
of an exploded right CMc3 joint inPan CMNH-B1758. An asterisk indicates
the capitate’s palmar Mc2 facet, and red springs/arrows the carpometacarpal
ligaments. The dorsal surface of each bone is angled opposite its palmar sur-
face. This causes distention within the joint whenever it is supinated from a
neutral (anatomical) position.
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org 70e6
RESEARCH ARTICLESResearch Articles

76. 76 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
that relied on suspension, vertical climbing, or
knuckle-walking. Further fossil remains from the
Late Miocene, including those before and after
the African ape–hominid phyletic divergences,
will test this hypothesis derived from our analysis
ofAr. ramidus.
References and Notes
1. G. Mivart,Proc. Zool. Soc. London1873, 484 (1873).
2. C. Darwin,The Descent of Man, and Selection in Relation
to Sex(John Murray, London, 1871).
3. W. L. Straus Jr.,Q. Rev. Biol.24, 200 (1949).
4. T. D. Whiteet al.,Science326, 64 (2009).
5. C. O. Lovejoy, G. Suwa, L. Spurlock, B. Asfaw, T. D. White,
Science326, 71 (2009).
6. C. O. Lovejoy, B. Latimer, G. Suwa, B. Asfaw, T. D. White,
Science326, 72 (2009).
7. C. O. Lovejoy, G. Suwa, S. W. Simpson, J. H. Matternes,
T. D. White,Science326, 73 (2009).
8. D. R. Begun,J. Hum. Evol.24, 373 (1993).
9. D. R. Begun, M. F. Teaford, A. Walker,J. Hum. Evol.26,
89 (1994).
10. S. Ward, B. Brown, A. Hill, J. Kelley, W. Downs,
Science285, 1382 (1999).
11. S. Almecija, D. M. Alba, S. Moya-Sola, M. Kohler,
Proc. Biol. Sci.274, 2375 (2007).
12. M. E. Bush, C. O. Lovejoy, D. C. Johanson, Y. Coppens,
Am. J. Phys. Anthropol.57, 651 (1982).
13. Apparently stemming from (33), the morphology of the
Mc5-hamate joint inAu. afarensishas been mischaracterized.
Its Mc5-hamate joint is fundamentally similar in both
structure and function to that ofH. sapiens, save for the
effects of the latter’s dramatic palmar rotation of the
hamulus. The original observations of these specimens were
that the four preserved Mc5 bases were“convex”(that is,
condyloid) and lacked“true articulation with the hamulus”
[(12), p. 659]. Thus, the statement that it“is obvious that the
bases of the fourth and fifth metacarpals articulated with the
hamate [inafarensis]asinPan”[(33), p. 178] is
unsubstantiated and incorrect.
14. J. R. Napier, P. R. Davis,Fossil Mammals of Africa No. 16
(British Museum of Natural History, London, 1959).
15. K. C. Beard, M. F. Teaford, A. Walker,Folia Primatol.
(Basel)47, 97 (1986).
16. K. C. Beard, M. F. Teaford, A. Walker, inHands of
Primates, H. Preuschoft, D. J. Chivers, Eds. (Springer-
Verlag, Vienna, 1993), p. 387.
17. A. C. Walker, M. Pickford, inNew Interpretations of Ape
and Human Ancestry, R. L. Ciochon, R. S. Corruccini,
Eds. (Plenum, New York, 1983) p. 325.
18. The trait nomenclature system used here is taken from
(50,59) and is briefly as follows [for more complete
explanations, see (4)]. Type 1: traits whose
morphogenesis is the direct consequence of pattern
formation; usually (but not always) subjected to direct
selection. Type 2: traits that are genetic but are
pleiotropic to, or result from, hitchhiking on type 1 traits
and are not themselves subject to selection. 2A: Parent
type 1 is inferred to be under selection; its secondary
effects are not. 2B: Neither parent trait nor its derivative
is inferred to be under selection (rare). Type 3: resulting
from a systemic growth factor. Type 4: epigenetic
consequence of osteochondral remodeling and/or
response to environmental stimuli; that is, not heritable but
useful in interpreting behavior. Type 5: developmentally
similar to type 4, but functionally uninformative.
19. C. F. Spoor, P. Y. Sondaar, S. T. Hussain,J. Hum. Evol.
21, 413 (1991).
20. Whereas the digits of apes have an“astonishing
interdigital versatility”not seen in humans, only humans
have an entirely independent flexor pollicis longus
combined with specialized“digital pairing”of flexor
tendons 3 and 4 and 2 and 5. These provide the uniquely
human“cylindrical grip,”which we regard as the likely
selective target of changes in the hands of early
hominids; that is,“palmar grasping.”In humans, digits
2 and 3 oppose the expanded tip of the thumb, and 4
and 5 the thenar eminence, providing afferent/efferent
“information on shape, consistency, and surface
qualities”[(60), p. 331]. That the necessary proportional
and neural changes were eventually achieved inHomois
adequate evidence that although obviously still lacking in
the hands ofAu. afarensis, early recognizable manual
changes in this taxon (for example, an increase in palmar
breadth and greater length of ray 1) indicate a selective
target of hand function similar to that eventually
achieved inHomo.
21. B. Asfawet al.,Science284, 629 (1999).
22. M. W. Marzke, M. M. Shrewsbury,J. Hum. Evol.51, 213
(2006).
23. M. M. Shrewsbury, M. W. Marzke, R. L. Linscheid,
S. P. Reece,Am. J. Phys. Anthropol.121, 30 (2003).
24. F. J. Bejjani, J. M. F. Landsmeer, inBasic Biomechanics of
the Musculoskeletal System, M. Nordin, V. H. Frankel,
Eds. (Lea & Febiger, Philadelphia, 1989), p. 275.
25. M. Baratz, A. D. Watson, J. E. Imbriglia,Orthopaedic
Surgery: The Essentials(Thieme, New York, 1999).
26. A“screw”-type ligament tensing mechanism must be
unidirectional, because any opposite rotation could
potentially cause cartilage damage by shear generated
during compression. This is prevented by the novel
abutment of the capitate’s palmar process against the
medial face of the Mc2.
27. J. G. Fleagleet al.,Symp. Zool. Soc. London48, 359
(1981).
28. The styloid process of the human Mc3 is almost certainly
the result of embryogenetic transfer of a styloid anlaga
from the capitate to the Mc3. A separate os styloideum is
frequently found in humans or is adherent to the capitate
or trapezoid [total: 6% of cases (61)]. Indeed, an os
styloideum is so frequently encountered that it was at one
time considered the“ninth carpal”(62).
29. Mobility in the Mc4- and Mc5-hamate joints has been
further enhanced in humans by palmar angulation of the
hamulus, presumably after a long period of freedom from
the selective rigors of locomotion. Moreover, ulnar
angulation of the medial metacarpals cannot have the same
importance in apes as in humans, because the African ape
first ray has undergone involution rather than hypertrophy.
30. The capitate ofP. heselonishows a slight depression
that may foreshadow the eventually dramatic relocation
of the head that is present inAr. ramidus[see especially
(15,16)].
31. Capitate waisting has been argued to facilitate midcarpal
locking and to be a suspensory adaptation in apes [and
early hominids (32,33)]. Frictionless synovial diarthroses
promote motion, and joints do not normally lock because
it is hazardous to their structure (41). In most joints,
stability is instead almost entirely the responsibility of
ligaments and soft tissues [contractile and connective
tissue (noncontractile) components of the joint’s
surrounding muscles (63)]. If these did not arrest motion
at and near full midcarpal dorsiflexion (64), the
capitate’s precipitous distal expansion would habitually
strain the scaphoid-lunate syndesmosis, leading to its
deterioration and instability. We agree with recent
observations that midcarpal structures differ substantially
inGorillaandPan(65); however, knuckle-walking
functions assigned to minor surface topographic fluctuations
of the scaphoid, capitate, and hamate reflect only the
ontogenetic interplay of cartilage modeling, positional
information, and the stabilizing soft tissues surrounding
these joints. Mere concavities (or ridges) on synovial
surfaces are almost never able to restrict motion and are
more likely to merely reflect its limits as dictated by the
joint’s surrounding soft tissues (63). These features
reflect the joint’s likely kinematics, but little about its
kinetics.
32. O. J. Lewis,Nature230, 577 (1971).
33. O. J. Lewis,Functional Morphology of the Evolving Hand
and Foot(Clarendon Press, Oxford, 1989).
34. M. G. Leakey, C. S. Feibel, I. McDougall, C. Ward,
A. Walker,Nature393, 62 (1998).
35. C. V. Ward, M. G. Leakey, A. Walker,J. Hum. Evol.41,
255 (2001).
36. S. Moya-Sola, M. Kohler, D. M. Alba, I. Casanovas-Vilar,
J. Galindo,Science306, 1339 (2004).
37. F. A. J. Jenkins,J. Zool. Soc. London48, 429 (1981).
38. R. H. Tuttle, thesis, University of California, Berkeley,
CA (1965).
39. T. D. White, G. Suwa, B. Asfaw,Nature371, 306 (1994).
40. B. G. Richmond, D. S. Strait,Nature404, 382 (2000).
41. C. O. Lovejoy, K. G. Heiple, R. S. Meindl,Nature410,
325 (2001).
42. M. S. Drapeau, C. V. Ward, W. H. Kimbel, D. C. Johanson,
Y. Rak,J. Hum. Evol.48, 593 (2005).
43. M. S. Drapeau,Am. J. Phys. Anthropol.124, 297 (2004).
44. M. Cartmill, K. Milton,Am. J. Phys. Anthropol.47, 249
(1977).
45. L. C. Aiello, B. Wood, C. Key, M. Lewis,Am. J. Phys.
Anthropol.109, 89 (1999).
46. This is expected given the following: (i) advanced
antebrachial anatomy with full ulnar withdrawal
(32,33,44) and pronation/supination were already
present inPierolapithecusat ~12.5 Ma (36), and (ii) the
primary differences in elbow morphology take place in
the ulna, rather than the humerus. A parallel relationship
occurs in the joint’s hindlimb counterpart: The femur
contains most functional information about kinematic
behavior in the knee; the tibia relatively little.
47. Some have argued that functionally and phylogenetically
meaningful distinctions can be made among Pliocene
hominid humeri, and that the Kanapoi distal humerus
has affinities withHomo(66), whereas other Pliocene
materials ofAustralopithecusandAr. ramidusrepresent
primitive hominid or ape branches cladistically outside of
Orrorin(66,67). However, two morphometric studies of
Pliocene hominid distal humeri (68,69) have
independently confirmed our visual assessments that
these humeri cannot be segregated into meaningful
morphotypes, thus leaving no quantifiable basis for such
assertions.
48. M. Pickford, D. C. Johanson, C. O. Lovejoy, T. D. White,
J. L. Aronson,Am. J. Phys. Anthropol.60, 337 (1983).
49. J. Kelley, inThe Primate Fossil Record, W. C. Hartwig, Ed.
(Cambridge Univ. Press, Cambridge, 2002), pp. 369–384.
50. C. O. Lovejoy, M. J. Cohn, inDevelopment, Growth, and
Evolution, P. O’Higgens, M. J. Cohn, Eds. (Academic
Press, London, 2000), pp. 41– 55.
51. A. Zumwalt,J. Exp. Biol.209, 444 (2006).
52. B. Benefit, M. McCrossin,Annu. Rev. Anthropol.24, 237
(1995).
53.“A comparison with the humerus of the anthropoid [apes]
shows that theSinanthropushumerus is as different from
it as the humerus of modern man. Not in one single
feature doesSinanthropusreveal a true anthropoid
character . . . [T]he deltoid tuberosity . . . is very poorly
developed in all three great apes....Itis,therefore, all
the more surprising that the tuberosity is very
pronounced in [MacacaandCynocephalus]. . . . If,
therefore, the peculiar shape of the deltoid tuberosity in
theSinanthropushumerus is to be interpreted as a
simian character, it is one that must be traced back to a
pre-anthropoid stage”[(70), p. 60].
54. A. B. Howell, W. H. Straus,Proc. U. S. Natl. Mus.80,
1 (1938).
55. T. H. Huxley,Evidence As to Man’s Place in Nature
(Williams & Norgate, London, 1863).
56. A. Keith,Br. Med. J.1, 788 (1912).
57. A. H. Schultz,Q. Rev. Biol.11, 259 (1936).
58. Straus observed that“[t]here can be no reasonable doubt
that a long thumb (relative to the other fingers) is a
generalized pithecoid character, or that its marked
relative reduction in such animals as the anthropoids,
some of the Semnopithecinae, and certain platyrrhines, is
an extreme specialization correlated with addiction to
brachiation”(3). He observed that human thumb
musculature is probably primitive because of the
following: (i)“a morphologically complete and functional
long flexor to the thumb . . . is normally absent in
orangs, and present in less than half of African apes but
is constant in prosimians, platyrrhines . . . Cercopithecinae,
the Hylobatidae, and man”[(3), p. 87]; and (ii)“The short,
intrinsic, volar muscles of the thumb [which are either
weakly developed or absent in the apes] are regularly well-
developed in man, the Hylobatidae, the Old World monkeys
(exceptColobus), the New World monkeys (exceptAteles),
and the prosimians”[(3), p. 87].
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 200970e7
Ardipithecus ramidus
Ardipithecus ramidus

77. 77www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
59. C. O. Lovejoy, R. S. Meindl, J. C. Ohman, K. G. Heiple,
T. D. White,Am. J. Phys. Anthropol.119, 97 (2002).
60. J. M. F. Landsmeer, inHands of Primates, H. Preuschoft,
D. J. Chivers, Eds. (Springer-Verlag, Vienna, Austria,
1993), pp. 323–333.
61. R. O’Rahilly,J. Bone Jt. Surg.35-A, 626 (1953).
62. E. A. Zimmer, S. P. Wilk,Borderlands of the Normal and
Early Pathologic in Skeletal Roentgenology(Grune &
Stratton, New York, 1968).
63. A. H. Burstein, T. W. Wright,Fundamentals of
Orthopaedic Biomechanics(Williams & Wilkins,
Baltimore, 1994).
64. W. L. Straus,Am. J. Phys. Anthropol.27, 199 (1940).
65. T. L. Kivell, D. Schmitt,Proc. Natl. Acad. Sci. U.S.A.106,
14241 (2009).
66. B. Senut,S. Afr. J. Sci.92, 165 (1996).
67. B. Senutet al.,Sci. Paris.332, 137 (2001).
68. M. R. Lague, W. L. Jungers,Am. J. Phys. Anthropol.101,
401 (1996).
69. A. M. Bacon,Am. J. Phys. Anthropol.111, 479
(2000).
70. F. Weidenreich,Palaeontologia Sinica, Whole Series No.
116; New Series D, No. 5(Geological Survey of China,
Peking, China, 1941).
71. C. O. Lovejoy, M. J. Cohn, T. D. White,Proc. Natl. Acad.
Sci. U.S.A.96, 13247 (1999).
72. P. L. Renoet al.,J. Exp. Zool. B Mol. Dev. Evol.310, 240
(2008).
73. Photographs were taken normal to the capitate’s
lateral surface, and a vertical tangent was inscribed along
the palmar and dorsal surfaces of its Mc3 facet.
Three perpendiculars were then erected to this vertical
tangent: (i) a horizontal tangent (at the bottom) to
the most palmar point on the capitate; (ii) a horizontal
tangent at the top to the most dorsal point on the
capitate, and (iii) a horizontal tangent to the
dorsalmost point on the articular surface of the capitate
head. The distance between tangents ii and iii was
then normalized by the total distance between tangents i
and ii.
74. Supported by NSF (this material is based on work supported
by grant numbers 8210897, 9318698, 9512534,
9632389, 9729060, 9910344, and 0321893
HOMINID-RHOI), and the Japan Society for the Promotion of
Science. We thank the Ministry of Tourism and Culture, the
Authority for Research and Conservation of the Cultural
Heritage, and the National Museum of Ethiopia for
permissions and facilitation; the Afar Regional
Government, the Afar people of the Middle Awash,
and many other field workers for contributing directly to the
data; the following institutions and staff for access to
comparative materials: National Museum of Ethiopia,
National Museums of Kenya, Transvaal Museum South
Africa, Cleveland Museum of Natural History,
Royal Museum of Central Africa Tervuren, and the
University of California at Berkeley Human Evolution
Research Center; D. Kubo and H. Fukase for assistance in
computerized tomography scanning and R. T. Kono
for the rapid prototyping model; R. Meindl for statistical
advice and assistance; P. L. Reno, M. A. Serrat,
M. A. McCollum, M. Selby, A. Ruth, L. Jellema, D. DeGusta,
and B. A. Rosenman for aid in data collection and
exceptionally helpful discussions; and K. Brudvik,
H. Gilbert, and J. Carlson for figure preparation.
Supporting Online Material
www.sciencemag.org/cgi/content/full/326/5949/70/DC1
Figs. S1 to S34
Tables S1 to S5
References
4 May 2009; accepted 18 August 2009
10.1126/science.1175827
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org 70e8
RESEARCH ARTICLES
Research Articles

78. 78 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
The Pelvis and Femur of
Ardipithecus ramidus: The
Emergence of Upright Walking
C. Owen Lovejoy,
1
*Gen Suwa,
2
Linda Spurlock,
3
Berhane Asfaw,
4
Tim D. White
5
The femur and pelvis ofArdipithecus ramidushave characters indicative of both upright bipedal walking
and movement in trees. Consequently, bipedality inAr. ramiduswas more primitive than in later
Australopithecus. Compared with monkeys and Early Miocene apes such asProconsul, the ilium in
Ar. ramidusis mediolaterally expanded, and its sacroiliac joint is located more posteriorly. These
changes are shared with some Middle and Late Miocene apes as well as with African apes and later
hominids. However, in contrast to extant apes, bipedality inAr. ramiduswas facilitated by craniocaudal
shortening of the ilium and enhanced lordotic recurvature of the lower spine. Given the predominant
absence of derived traits in other skeletal regions ofAr. ramidus, including the forelimb, these
adaptations were probably acquired shortly after divergence from our last common ancestor with
chimpanzees. They therefore bear little or no functional relationship to the highly derived suspension,
vertical climbing, knuckle-walking, and facultative bipedality of extant African apes.
T
he hominid pelvis is among the most dis-
tinct osteological complexes of primates.
Its distinctiveness derives from the con-
figuration of its superior portion that maintains
balance on a single limb during upright walking.
These changes are not shared with apes. There-
fore, comparison of the pelvis and hip among
fossil and extant hominids and apes is critical for
reconstructing the evolutionary steps leading to
upright walking in humans versus the knuckle-
walking and vertical climbing practiced by our
nearest ape relatives.
An almost complete but damaged left hip
(os coxa), a portion of the right ilium, and a distal
sacral fragment were recovered from the Aramis
Ardipithecus ramiduspartial skeleton (ARA-VP-
6/500)( 1). The os coxa’s overall form is pre-
served despite postmortem distortions of varying
magnitude, most notably the fragmentation, sep-
aration, and translation of cranial and caudal por-
tions of the acetabulum (Fig. 1).
The recovered os coxa is fragmented, distorted,
friable, and inseparable from internal matrix, pre-
venting restoration by standard methods (2 –4).
To aid our analysis, we made a reconstruction by
using anatomical and high-resolution tomographic
rapid prototyping models. We iteratively adjusted
various surface metrics to verify them against the
original fossil. Multiple permutations of this pro-
cess produced a model that conformed to all ma-
jor undistorted linear measurements of the original
fossil (Figs. 1 and 2 and figs. S1 to S3).
The superior portion of the iliac blade was
bent anteromedially postmortem, eliminating its
original lateral flare. This has been adjusted in the
model. The natural curvature of the superior pu-
bic ramus is preserved despite its fragmentation
into three sections. These sections were reassem-
bled to form the upper portion of the ischiopubic
region. The position of the ischiopubic ramus was
restored on this basis. All of our descriptions of
the features and linear measurements of the hip
and femur are independent of this reconstruction,
which was used as a three-dimensional heuristic
aid (5). Acetabular size and some angles are, by
necessity, approximations. We provide probable
ranges of likely original values where appropriate.
The majority of the sacrum was not preserved.
In reconstructions of the entire pelvis, however,
the distance separating the two auricular surfaces
is also indicated by the length and angulation of
the arcuate line and plane of the pubic surface.
The former can be well approximated from the
fossil, and the latter is intact. Nevertheless, the
exact biacetabular breadth remains unknown as
do, therefore, the exact dimensions of the three
primary pelvic planes.
The ilium, ischium, and pubis.TheAr. ramidus
ilium is dramatically mediolaterally broad as in all
post-Miocene hominoids, especiallySymphalangus,
Gorilla,Australopithecus, andHomo(6). Its iliac
fossa is largely intact, from the anterior margin of
its auricular surface to the well-preserved anterior
iliac margin. Here, a prominent anterior inferior
iliac spine (AIIS) maintains an intact relationship
to the superior acetabular wall (Fig. 1). As in later
hominids, the ilium is laterally flared for reloca-
tion of the anterior gluteals and has a forward
sweeping anterior superior iliac spine (ASIS).
The anterosuperior edge of the ilium is fractured
and lacks an intact crest. On the basis of the form
of the bone in both apes and humans, we esti-
mated that it would have extended only about
1 cm superior to its broken edge. A further ex-
tension would improve the role of the abductor
lever arm during upright gait. We based the antero-
superior projection of the ASIS on the well-
preserved AIIS and their typical relationship in
hominoids. We therefore regard its position in
Figs. 1 and 2 and figs. S1 to S3 to be close to that
of the original, although it may have terminated
less superiorly than reconstructed here.
The protuberant and anteriorly positioned AIIS
is associated with a broad, short, and sagittally
disposed iliac isthmus. We define the isthmus as
the constricted inferior portion of the iliac blade
immediately superior to the acetabulum (Figs.
1 and 2 and fig. S3). These features are shared
with later hominids, but in both modern and Early
Miocene apes and monkeys the isthmus is mark-
edly elongate, and the greater sciatic notch (GSN)
angle is more obtuse (Fig. 3B). TheAr. ramidus
GSN is intermediate between its counterpart in
modern apes and those of later fossil hominids
(Fig. 3C and fig. S4).
In stark contrast to its distinctly hominid ilium,
the preservedAr. ramidusischium is like that of
African apes. Although the ischial tuberosity is not
preserved, the ischium is intact from its typical
concave surface flare just superior to the tuberosity
to the inferior border of the acetabulum (Fig. 1).
Even this minimum preserved length of the supe-
rior ischial ramus is substantially longer than any
knownAustralopithecusexample (Fig. 3D) (7).
The pubis preserves an intact, superoinferiorly
elongate body. However, the outline of the pubic
symphyseal face is similar to that ofHomoand
A.L. 288.-1and unlike its extreme dorsoventral
elongation in African apes (Fig. 3A). This more
ovoid shape may be a collateral pleiotropic man-
ifestation [Type 2A effect (8)] of the shortened
iliac isthmus and elongated superior pubic ramus,
especially inA.L. 288-1(which exhibits both an
unusually broad pubic face and greatly elongated
pubic rami) (Fig. 3A) (7).
Pelvic form and function.Anthropoid pelvic
form is highly conserved, and major proportions
are therefore similar from Old World monkeys
(cercopithecoids) toProconsul(9,10). However,
features shared by all demonstrate that the last
common ancestor ofGorilla,Pan, andHomo
(hereafter the GLCA) must have exhibited two
substantial modifications of the anthropoid pat-
tern, both largely occurring in the coronal plane
(10): (i) a lateral expansion of the iliac fossa and
crest (fig. S5) and (ii) a corresponding reduction
in the retroauricular region or pars sacralis of the
ilium (fig. S6) (10). Both of these changes appear
to be present in some other hominoid ilia, those
ofDryopithecus brancoi(11) andOreopithecus
bambolii(12), but apparently are absent inPro-
consulandNacholapithecus(13). The breadth of
the ilium appears to scale with body mass and
RESEARCHARTICLES
1
Department of Anthropology, School of Biomedical Sci-
ences, Kent State University, Kent, OH 44242–0001, USA.
2
The University Museum, The University of Tokyo, Hongo,
Bunkyo-ku, Tokyo 113-0033, Japan.
3
Cleveland Museum of
Natural History, Cleveland, OH 44106–4930, USA.
4
Rift Valley
Research Service, Post Office Box 5717, Addis Ababa, Ethiopia.
5
Human Evolution Research Center and Department of
Integrative Biology, 3101 Valley Life Sciences Building,
University of California at Berkeley, Berkeley, CA 94720, USA.
*To whom correspondence should be addressed. E-mail:
[email protected]
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 71e1
Ardipithecus ramidus

79. 79www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
may reflect a relatively greater gut volume in
larger anthropoids involved in frugivory and/or
mid-gut folivory (14). The reduction of the pars
sacralis (the portion posterior to the iliac fossa/
auricular surface boundary), however, was clearly
not a simple product of body size but rather a
collateral manifestation (pleiotropic but of selective
importance; type 2A) (8 ) of thoracic vertebral
column invagination associated with posterolateral
reorientation of the scapular glenoid (15,16).
The African ape pelvis has also undergone
dramatic craniocaudal differentiation. However,
unlike the changes in the iliac fossa and crest
breadth, the morphology of theAr. ramiduspelvis
implies that these changes in craniocaudal dimen-
sions evolved after the last common ancestor of
the African apes and hominids.PanandGorilla
show slight positive allometry of maximum iliac
height (17) as compared withProconsul(fig. S7).
Gorillaappears to have isometrically increased
the height of the lower ilium versus its dimen-
sions inProconsul(fig. S8, lower iliac height). In
Pan, the lower ilium is extended cranially, prin-
cipally by elongation of the iliac isthmus. This
elongation, at least inPan, is but one element of
global change in ape pelvic morphology, in which
reduction of the lumbar column and narrowing of
the sacral alae (fig. S9) have constricted and dor-
sally extended lumbar-iliac contact in the sagittal
plane (fig. S10, the trans-iliac space). This en-
traps the caudal lumbar (or lumbars). In combi-
nation with reduced vertebral height and fewer
lumbar vertebrae (18), this effectively eliminates
any thoracopelvic mobility in African apes (table
S1) (7) and appears to be an adaptation to vertical
climbing and/or suspension.
There is no evidence of any difference in the
relative height of the lower ilium betweenPro-
consulandARA-VP-6/500(fig. S8). However,
the pelvis ofARA-VP-6/500shows that the lower
lumbars were not entrapped as in great apes but
were clearly free for anteroposterior curvature
(lordosis) (figs. S9 to S11). This is because the
posterior ilium and pars sacralis did not extend
sufficiently superior to have restricted the most
caudal lumbar. Given that early hominids most
likely had six lumbars, and other manifestly prim-
itive characters (7 ,15,16,18), it seems probable
that hominids either quickly reversed or never ex-
perienced any tendency for the sacral narrowing
seen in extant great apes. A capacity for posturally
dependent lower lumbar orientation was a key
adaptation to bipedality, an inference made almost
a century ago (19 ).
The inferred freedom of the lowermost lumbar
(or lumbars) inAr. ramidus, coupled with broad-
ening and more sagittal orientation of the iliac
isthmus (fig. S4), would have permitted both
lordosis and anterior extension of the lesser
gluteals for pelvic stabilization during upright
walking (15,16). These changes, in conjunction
with retention of a long lumbar column and a
lowered iliac crest [that is, a reduced maximum
iliac height (fig. S7)], enhanced lordosis. Lordosis
can be situationally achieved by cercopithecoids
Fig. 1.Original and reconstructed os coxa ofARA-VP-6/500.(Left) Anterolateral and anteromedial views
of the original, with a close-up of the AIIS. (Middle) CT scans of same views, except the bottom view is a
close-up of the pubic symphyseal face, preserved in its entirety but damaged at its inferior extremity. It is
like later hominids in its dorsoventral height and ovoid outline. (Right) Anterolateral and anteromedial
views of the reconstructed os coxa (11th permutation) and of the entire pelvis using mirror reconstruction
and a conjectural sacrum. Various permutations were attempted with respect to sacral breadth; the
solution shown provides presumed functional minimal inlet and outlet dimensions for a hominoid of this
individual’s body size (Fig. 2 and figs. S1 to S3). (Middle) Arrows in the CT images indicate major areas of
distortion corrected by reconstruction. The entire iliac blade was bent anteromedially (single yellow arrow)
(additional and corrective data were provided by right iliac fragment). Two green arrows indicate primary
foci of subduction of three largely intact segments of the superior pubic ramus; obvious overlapping
allowed accurate restoration of original length (compare with reconstruction). The acetabulum was
separated into two halves (total height of the exploded acetabulum is indicated by white arrows), with
substantial intervening matrix infill. This separation greatly elongates its unrestored appearance. The two
halves were recompressed on the basis of a calculated rim circumference obtained through the sum-
mation of individual segments (a range of probable values is presented in discussions of metric pa-
rameters). Two red arrows mark the inferior edge of the intact acetabular rim and the superior edge of the
(missing) surface of the ischial tuberosity. These provide a minimum ischial length. Various additional
dimensions were corrected by means of surface metrics. The breadth of the iliac fossa was intact from the
AIIS to the lateral edge of the auricular surface; the degree of individual fragment separation was
assessable from surface observation and CT scan data as indicated. Areas of particular anatomical im-
portance include the protuberant sigmoid AIIS and sagittally oriented iliac isthmus, typical of later
hominids, and the notably short, ovoid, pubic symphyseal face. Major metrics and angles are provided in
Fig. 3. CT scans were taken at 300-mm voxel resolution on the University Museum, the University of Tokyo,
micro-CT system [TX-225 Actis (Tesco, Tokyo)] and processed with the software Analyze 6.0 (Mayo Clinic,
Rochester, MN) and Rapidform 2004/2006 (Inus Technology, Seoul).
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
71e2
Ardipithecus ramidus
Research Articles

80. 80 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
(20) (and presumably inProconsul), even though
Old World monkey ilia typically entrap the most
caudal lumbar vertebra. Lordosis has been en-
tirely eliminated in the African apes.
The form and size of the AIIS inARA-VP-
6/500, as well as its projection anterior to the
acetabular margin, indicate that this structure had
already begun to appear and mature via a novel
physis. Isolation of the AIIS as a separate growth
center, which is unique to hominids (21), was
probably a consequence of its increased separation
from the original iliac portion of the acetabular
chondroepiphysis. This is because much of iliac
broadening must occur at the triradial epiphysis
(as well as the posterior iliac crest). An analogous
phenomenon occurs in mammals with elongate
femoral necks, in which accelerated growth in the
region separating the presumptive greater trochan-
ter and femoral head isolates these structures be-
fore separate ossification of each epiphysis (22).
The emergence of a novel AIIS center of os-
sification, as seen inARA-VP-6/500, attests to
global modification of the entire pelvis. This is
also demonstrated by the abbreviated craniocaudal
length of the pubic symphyseal face (Figs. 1, 2,
and 3A). The anteroventral pubic surface is sur-
mounted by a rugose pectineal line that is con-
tinuous with the nonperiosteal attachments of
abductor brevis and gracilis muscles. This is a
derived character in human females [the ventral
arc (23)]. It is only rarely present in great apes.
There is no lateral displacement of this feature in
Ar. ramidusas there is in human females (and
partially inA.L. 288-1).
The systematic reduction of overall cranio-
caudal pelvic height would have lowered the
trunk’s center of mass and shortened its moment
arm during single support. This would at least par-
tially compensate for retention of the long lumbar
column required for anterior lordotic shift of the
center of mass during bipedality. Any deletion of
thoracic vertebra would also lower the center of
mass. Flexibility in its positioning probably main-
tained ample hind-limb mobility during arboreal
climbing and clambering, albeit with considerable
attendant risk of lower-back injury.
Given retention byAr. ramidusof multiple
primitive skeletal characters (15,16,24,25), its
exceptionally derived ilium is striking. It implies
an early adaptation to habitual terrestrial bipedality
before any increase in the lumbar entrapment seen
in the African apes, but after the lateral iliac ex-
pansion shared with them. This is consistent with
the hypothesis that vertebral column invagination
(its anterior transpositioning into the thorax and
abdomen) was a primary morphogenetic mech-
anism underlying scapulohumeral reorganization
for greater forelimb flexibility during arboreal clam-
bering and bridging. If this is correct, then the ex-
tensive reorganization of the column in hominoids
was not originally an adaptation to suspensory lo-
comotion or vertical climbing (15,16,24–27).
Comparisons of the ossa coxae ofAr. ramidus
andAu. afarensisdemonstrate the latter’s mod-
ifications for habitual bipedality following aban-
donment of arboreal locomotion (Figs. 2 and 3C
and figs. S1 to S3). In general,Au. afarensisdem-
onstrates even greater global craniocaudal abbre-
viation of its entire pelvis, and an accompanying
increase in platypelloidy via elongation of the pu-
bic rami anteriorly, and deepening of the greater
sciatic notch and probable expansion of alar breadth
posteriorly (7). Such changes appear morphoge-
netically coordinated and therefore were also likely
expressed partially inAr. ramidus, although the
actual dimensions of its sacrum remain unknown.
This view receives independent support from the
remarkable pelvic stasis seen betweenA.L. 288-1
and the much more recent [0.9 to 1.4 million years
ago (Ma); post-Au. afarensis ] Busidima pelvis
(BSN49/P27)( 28).
The functional importance of platypelloidy in
Au. afarensishas been widely debated. Viewed
now from the perspective of its ancestral state,
however, it appears likely to have been a con-
sequence of establishing mobility of the L6/S1
joint as a permanent character. The ancestral mor-
phology presumably involved situationally depen-
dent lordosis (during terrestrial upright walking)
for gluteal stabilization during the stance phase.
Platypelloidy may also have enhanced gut volume,
although a trunk length with six lumbars may have
been sufficient. The reduction of lateral iliac flare
and more posterior placement of the iliac cristal
tubercle inHomotherefore probably reflect an
ontogenetic predisposition for lumbar lordosis
with (i) reduction in lower lumbar positional
lability (permitting a reduction from six to five
lumbars) after complete abandonment of arboreal
activity and (ii) optimization of birth-canal
geometry (4).
The femur.Two partial proximal femora were
recovered at Aramis (Fig. 4). That of the partial
skeleton (ARA-VP-6/500-5) preserves most of the
shaft but is damaged by extensive expanding matrix
distortion (29). A second (ARA-VP-1/701)isin
good condition. Although neither preserves a head,
neck, or greater trochanter, in conjunction with the
os coxa they are informative with respect to the
evolution of the gluteus maximus muscle (here-
after simply the maximus).
The African ape posterolateral femoral shaft
regularly exhibits a distomedially displaced in-
Fig. 2.CT comparison ofARA-VP-6/500(left) os coxa reconstruction (11th permutation) andA.L. 288-1
(right) restoration. Reconstruction ofARA-VP-6/500was achieved by means of sculptural modeling on the
basis of numerous dimensions and contours preserved on the original fossil (Fig. 1). The sacrum is largely
conjectural because only its lower portion was recovered, but four segments are likely (18). The enlarged
A.L. 288-1(115% of actual size) andARA-VP-6/500images have been aligned on their acetabulae. The
green scale square (lower left) is 180 mm on a side. The broad sacral alae ofARA-VP-6/500are probably
because African ape sacra have almost certainly been narrowed since the GLCA (figs. S10 and S11). A
novel ossification center for the AIIS and a substantial reduction in the height of the iliac isthmus are
derived characters present in both hominids. Length of the ischia was reconstructed on the basis of the
position of the ischial spine, shape of the obturator foramen, and (most importantly) the length of an
intact surface transect from the lower edge of the acetabulum to the dorsal edge of the (missing) ischial
tuber (Fig. 1). The acetabulum was preserved in two separated portions (Fig. 1). Together, they suggest a
diameter of 36 to 42 mm. Spatial orientation was made on an assumption of vertical alignment of the
pubic tubercle and ASIS, although the primitive form of this pelvis suggests that lumbar lordosis during
terrestrial bipedality was partially situational and that the superior pelvis may have been angled less
anteriorly during arboreal clambering. There is a dramatic reduction of lower pelvic length and robusticity
inA.L. 288-1.
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
71e3
RESEARCH ARTICLES
Ardipithecus ramidus

81. 81www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
sertion for the maximus. This is separated from a
more superior attachment of the vastus lateralis
[to whose tendon, however, the maximus is nor-
mally fused (30 )] by an elevated boss on the shaft
[defined as the lateral spiral pilaster (31)]. The
femora ofAu. afarensisand subsequent hominids
exhibit a strikingly different morphotype that in-
cludes a trochanter tertius that surmounts a rugose
hypotrochanteric fossa (31,32). Both structures
are clearly associated with hypertrophy of the
maximus’ascending tendon, which inserts direct-
ly into the lateral aspect of the femur (30).
It had been thought that the lateral spiral
pilaster of apes was primitive (31 ) and that the
hominid morphotype was derived. The femur of
Ar. ramidusshows that this inference was in-
correct. African ape femora never exhibit a third
trochanter or hypotrochanteric fossa, whereas
ARA-VP-1/701exhibits obvious homologs to both.
Moreover, the femur ofProconsul, millions of
years older thanAr. ramidus, exhibits a strong
gluteal tuberosity immediately inferior to its greater
trochanter, as do those ofNacholapithecus(13) and
Dryopithecus(33).
InARA-VP-1/701, the medial border of an ob-
vious hypotrochanteric fossa homolog converges
with the spiral line to form a markedly rugose,
elevated plane on the posterior femoral surface,
but their further course is lost to fracture. A simi-
lar morphology is visible on theARA-VP-6/500
specimen. A broad linea of low relief is also
clearly present inASI-VP-5/154, assigned toAu.
anamensis(34). Its morphology is reminiscent of
that ofA.L. 288-1, in which the linea is still no-
tably broad, but contrasts with that ofMAK-VP-
1/1, which is more modern in form at 3.4 Ma
(31). Because most of the length of theASI-VP-
5/154shaft is preserved, its moderately elevated
linea (~11.5 mm in breadth) is distinct and im-
parts a prismatic cross section at midshaft. Spec-
imenBAR-1002'00(Orrorin tugenensis)(35)
presents obvious homologs to these structures.
Moreover, bothBAR-1002'00andASI-VP-5/154
exhibit an obvious homolog to the third trochanter,
and neither shows any evidence of a lateral spiral
pilaster.
African ape morphology can therefore now
be interpreted as derived and probably a conse-
quence of global alterations of their hip and lower
back for suspensory locomotion. Further narrow-
ing of their iliac cleft from its state inProconsul
must have also relocated their maximus insertion
more distomedially as well and caused increased
separation from its previous position adjacent to
that of the vastus lateralis, a hiatus now filled in
African ape femora by the lateral spiral pilaster.
This has eliminated any evidence of the other-
wise ubiquitous Miocene morphotype. The disap-
pearance of any homolog of the hypotrochanteric
fossa and third trochanter in extant apes sug-
gests a possible change in muscle and/or enthesis
architecture, although fascicle length does not
appear to differ substantially in the gluteals of
humans and apes (36).
In contrast, the gluteal complex inAr. ramidus
remains anterolaterally displaced as inProconsul
andOrrorinand still unlike its more posteromedial
position in mostAustralopithecus(Fig. 4) (31).
Such medial translation of the maximus insertion
is probably a consequence of hypertrophy of the
quadriceps at the expense of the hamstrings (31).
Indeed, the combination of a broad, ape-like, ex-
pansive ischial tuberosity and broad proto-linea
aspera inAr. ramidussuggests that the hamstring/
quadriceps exchange had not yet achieved its mod-
ern proportions, although some expansion of the
maximus was probably present given the substan-
tial restructuring of the ilium and trans-iliac space.
In contrast, mostAustralopithecusspecimens [such
AB
CD
Fig. 3.Some geometric and anatomical traits of the ossa coxae of hominids and African apes. (A)
Maximum breadth of the pubic symphyseal face normalized by its length. The shortened condition inARA-
VP-6/500may reflect an abbreviation of overall pelvic height. The extreme value forA.L. 288-1is largely
due to the great breadth of its pubic symphyseal face; this might also be a type 2 (8) effect of this
specimen’s greatly elongate superior pubic ramus. Metrics are from originals.Pan-Homodifferences are
highly significant (n= 15 specimens each taxon;P< 0.0001; two-tailedttest). Boxes represent 25th and
75th percentiles, vertical lines represent 5th and 95th percentiles, and the transverse lines are medians.
Values between 1.5 and 3 box lengths from the upper or lower boundaries of a box are shown as open
circles (asterisks indicate more than three box lengths). (B) Angle made by two chords connecting three
landmarks on the ilium: from the superomost point on the auricular surface to the ASIS and from the
auricular surface to the AIIS. The angle between these chords was directly measured with a modified
goniometer. Maximum value shown forARA-VP-6/500represents the most superior location of the ASIS as
described in the text and used in the model (Fig. 2); the lower value would obtain if the ASIS were less
protuberant. There is a more elongate ilium (more obtuse angle) inPanas compared withGorilla(P<
0.001; two-tailedttest). The overlap of higher possible values inARA-VP-6/500with those ofGorilla,
and the intermediate values inA.L. 288-1andSTS-14, at least partially reflect ASIS elongation and lateral
iliac flare for increased lordosis. The exceptionally high values forProconsulreflect both its extremely tall
ilium and the precipitous anterior tilt of its auricular surface.Pan-Homodifferences are highly significant
(n= 15 specimens each taxon;P< 0.0001; two-tailedttest). (C) Angle made by two chords connecting
three landmarks on the ilium: from the acetabular center to the caudalmost point on the auricular
surface and from the acetabular center to the ischial spine (fig. S4). A third chord was measured
(auricular surface to the ischial spine), and the angle was determined trigonometrically. All early
hominids lie within the human range because all have a greatly shortened iliac isthmus. The isthmus in
Panis significantly longer than that ofGorilla(P< 0.0001; two-tailedttest). The angle forARA-VP-6/500
was estimated from reconstruction.Pan-Homodifferences are highly significant (n= 15 specimens each
taxon;P< 0.0001; two-tailedttest). (D ) Minimum ischial length (from inferior acetabular border to juncture
of the ischial body and tuberosity surface) normalized by acetabular diameter. Two values forARA-VP-6/500
bracket possible extremes for the acetabular diameter.Pan-Homodifferences are highly significant (n= 15
specimens each taxon;P< 0.0001; two-tailedttest).
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org71e4
Ardipithecus ramidusResearch Articles

82. 82 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
asMAK-VP-1/1,A.L. 333w-40, andA.L. 333-
110)(31)] had marked elevation and narrowing
of a true linea aspera that is typical of later
hominids (37 ,38).
The thorax.The thorax ofARA-VP-6/500is
represented by a partial first rib, several asso-
ciated crushed ribs, and a thoracic vertebral arch.
A thoracic arch from a second individual was
also recovered (ARA-VP-6/1001). Both vertebrae
lack centra, so the angulations of their rib facets
and their implications for vertebral column in-
vagination (39 ) cannot be determined. It has
recently become common wisdom that early
hominids had a funnel-shaped thorax. However,
this supposition stems from a reconstruction of
A.L. 288-1(40) that relied on (i) an ilium left
uncorrected for extensive postmortem fracture of
its retroauricular portion (a defect that exagger-
ates lower thoracic breadth) (41), (ii) a highly
fragmentary thorax, and (iii) the presumption of
a short-backed, great ape– like, ancestral morpholo-
gy. Because dramatic forelimb elongation, hind-
limb abbreviation, and lower thoracic rigidity
are seen together in all three great apes, relative
constriction of their upper thorax is probably a
modification accompanying these advanced
adaptations to suspension and/or vertical climbing.
Ardipithecusreveals, however, that earliest hom-
inids did not regularly engage in these behaviors
(15,16) and that an elongated iliac isthmus and
narrowed trans-iliac space are African ape spe-
cializations that negate the lordosis required for
effective transitional upright walking.
Nearly 30 years have now passed since the
observation that the human serratus anterior“lacks
the specializations associated with suspensory be-
havior in large bodied, broad-chested nonhuman
primates [suggesting] descent from a small ape
with a thoracic shape similar to atelines.”(42).
This seems particularly prescient given a recent
examination of thoracic form in primates (43),
which demonstrates thatAteles(also highly skilled
at suspension) nevertheless lacks many other great
ape-like adaptations, including a funnel-shaped
thorax. It follows that this unique thoracic form is
likely to emerge only as an element accompany-
ing these other extensive great ape specializations
for suspension and vertical climbing. It is there-
fore unlikely to have ever been present in early
hominids. Confirmation of this inference, how-
ever, will require additional fossil evidence.
The pelvis, femur, and preserved thoracic ele-
ments ofAr. ramidusestablish that adaptations to
upright walking in these regions were well estab-
lished by 4.4 Ma, despite retention of a capacity
for substantial arboreal locomotion.Ar. ramidus
thus now provides evidence on the long-sought
locomotor transition from arboreal life to habitual
terrestrial bipedality. This evidence suggests that
the transition took place in the absence of any of
the characters that today substantially restrict up-
right walking in extant apes (particularly lumbar
column abbreviation, trans-iliac space narrow-
ing, and approximation of iliac crest and thorax,
and the muscles that traverse this gap). As a
consequence, explications of the emergence of
bipedality based on observations made of Af-
rican ape locomotion no longer constitute a use-
ful paradigm.
References and Notes
1. T. D. White, G. Suwa, B. Asfaw,Nature375, 88 (1995).
2. C. O. Lovejoy,Am. J. Phys. Anthropol.50, 413 (1979).
3. D. C. Johansonet al.,Am. J. Phys. Anthropol.57, 403 (1982).
4. R. G. Tague, C. O. Lovejoy,J. Hum. Evol.15, 237 (1986).
5. In order to facilitate our examination, we also
reconstructed the innominate ofKNM-RU 13142 D
(fig. S4). Our reconstruction does not differ substantially
from that drawn by its original descriptor (17) but
facilitated some three-dimensional comparisons.
6.“The marked difference between the pelvis ofHylobates
and that ofSymphalangus syndactylusis ...apparent...
[demonstrating] that the siamang has already a pelvis of
typical anthropoid ape character, particularly in regard to
the broad ilium with its prominent ASIS, whereas the
pelves of the gibbons have not yet departed so far from
the more primitive condition of catarrhine monkeys”
[(44) p. 350].
7. C. O. Lovejoy,Gait Posture21, 95 (2005).
8. The trait nomenclature system used here is taken from
(31) and is briefly as follows [for more complete
explanations see (24)]. Type 1 indicates traits whose
morphogenesis is the direct consequence of pattern
formation, usually (but not always) subject to direct
selection. Type 2 indicates traits that are genetic but are
pleiotropic to or result from hitchhiking on type 1 traits
and are not themselves subject to selection [2A indicates
a parent type 1 is inferred to be under selection; its
secondary effects are not; 2B indicates neither parent
trait nor derivative is inferred to be under selection (and
is rare)]. Type 3 indicates a result from a systemic growth
factor. Type 4 indicates an epigenetic consequence of
osteochondral remodeling and/or response to
environmental stimuli, not heritable but useful in
interpreting behavior. Type 5 is similar to type 4 but
uninformative.
9. C. V. Ward, A. Walker, M. F. Teaford, I. Odhiambo,Am. J.
Phys. Anthropol.90, 77 (1993).
10. C. V. Ward,Am. J. Phys. Anthropol.92, 291 (1993).
11. C. V. Ward,Am. J. Phys. Anthropol.135(S46), 218 (2008).
12. T. Harrison, inOrigine(s) de la Bipedie chez les Hominides,
Y. Coppens, B. Senut,Eds. (CNRS, Paris, 1991),
pp. 235–244.
13. M. Nakatsukasa, Y. Kunimatsu,Evol. Anthropol.18, 103
(2009).
14. D. J. Chivers, C. M. Hladik,J. Morphol.166, 337 (1980).
15. C. O. Lovejoyet al.,Science326, 70 (2009).
16. C. O. Lovejoyet al.,Science326, 73 (2009).
17. C. V. Ward, thesis, Johns Hopkins (1991).
18. M. A. McCollumet al.,J. Exp. Zool. B Mol. Dev. Biol.
10.1002/jez.621316 (2009).
19. W. L. Straus Jr.,Am. J. Anat.43, 403 (1929).
20. M. Nakatsukasa,J. Anat.204, 385 (2004).
21. R. A. Dart,J. Palaeon. Soc. India2, 73 (1957).
22. M. A. Serrat, P. L. Reno, M. A. McCollum, R. S. Meindl,
C. O. Lovejoy,J. Anat.210, 249 (2007).
23. L. C. Budinoff, R. G. Tague,Am. J. Phys. Anthropol.82,
73 (1990).
24. T. D. Whiteet al.,Science326, 64 (2009).
25. C. O. Lovejoyet al.,Science326, 72 (2009).
26. M. Cartmill, K. Milton,Am. J. Phys. Anthropol.47, 249
(1977).
Fig. 4.Lateral and posterior CT scan surface renders of (A )MAK-VP-1/1(Au. afarensis, cast) and (B )ARA-
VP-1/701(Ar. ramidus, original). Specimens have been aligned by their lesser trochanters. The preserved
portion of theARA-VP-1/701shaft is sufficient to demonstrate the absence of a lateral spiral pilaster and the
presence of a distinct rugose insertion area for the gluteus maximus that is homologous to the true
hypotrochanteric fossa present inMAK-VP-1/1(arrows indicate the third trochanter; brackets indicate
hypotrochanteric fossae). InARA-VP-1/701, this area is more laterally placed, as in other early hominid
femora, includingBAR-1002'00. The more posterior position of this insertion inMAK-VP-1/1is almost
certainly associated with decreased sagittal iliac orientation inAu. afarensis, a consequence of further
posterior pelvic broadening and increased lateral iliac flare (fig. S1). None of these early hominid
specimens shows evidence of a lateral spiral pilaster, which is restricted to African ape femora.
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
71e5
RESEARCH ARTICLES
Ardipithecus ramidus

83. 83www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
27. This is also consistent with substantial differences in the
soft tissue configuration of the pelvic floor of extant
humans and African apes. In African apes, the sphincter ani
externus is hypertrophied. It serves as the primary
urogenital floor, thus greatly exceeding its distribution in
Old World monkeys. In humans, this muscle is instead
reduced, and its function alternatively subsumed by
hypertrophy of the transversus perinei profundus.“The
simplest interpretation of this difference […is]
that in the…[GLCA] the sphincter ani and the
bulbocavernosus resembled those muscles in the…[Old
World monkeys]”(45). Elftman was, of course, unaware
ofProconsulpelvic structure at the time of his
observation, but his conclusion was essentially that the
GLCA’s pelvic floor had remained primitive, such as it
presumably was in Early Miocene apes practicing
above-branch quadrupedality and remaining underived
for suspensory locomotion.
28. S. W. Simpsonet al.,Science322, 1089 (2008).
29. T. D. White,Science299, 1994 (2003).
30. J. T. Stern Jr.,Am. J. Phys. Anthropol.36, 315 (1972).
31. C. O. Lovejoy, R. S. Meindl, J. C. Ohman, K. G. Heiple,
T. D. White,Am. J. Phys. Anthropol.119, 97 (2002).
32. A. Hrdlička,Smithsonian Misc. Coll.92, 1 (1934).
33. S. Moya-Solaet al.,Am. J. Phys. Anthropol.139, 126
(2009).
34. T. D. Whiteet al.,Nature440, 883 (2006).
35. B. Senutet al.,C. R. Acad. Sci. IIA Earth Planet. Sci.332,
137 (2001).
36. R. C. Payneet al.,J. Anat.208, 709 (2006).
37. C. O. Lovejoy,Gait Posture21, 113 (2005).
38. It has been suggested (via canonical variates analysis)
that theOrrorinproximal femur“exhibits an
Australopithecus-like bipedal morphology [that] evolved
early in the hominin clade and persisted successfully for
most of human evolutionary history”[(46) p. 1664].
However,BAR-1002’00lacks a complete greater
trochanter, making such a conclusion dependent on
reconstruction. Our examination of both casts and
originals leads us to agree that the specimen belongs to a
bipedal hominid, but the femoral and pelvic evidence
presented here demonstrate that (i) bipedality was not
morphologically static from 6 to 2 Ma as claimed nor (ii)
is there now any evidence for an“appreciable scansorial
component”in the locomotor repertoire ofAustralopithecus.
To the contrary, substantial arboreal behavior is now
contraindicated by much of the postcranial anatomy that
differentiatesArdipithecusandAustralopithecus.
39. B. Latimer, C. V. Ward, inThe Nariokotome Homo erectus
Skeleton, A. Walker, R. Leakey, Eds. (Harvard Univ. Press,
Cambridge, 1993), pp. 266–293.
40. P. Schmid, inOrigine(s) de la Bipedie chez les Hominides.,
Y. Coppens, B. Senut, Eds. (CNRS, Paris, 1991),
pp. 226–234.
41. The retroauricular portion of the innominate of
A.L. 288-1was crushed postmortem, introducing a
90° angulation defect at its juncture with the iliac fossa.
Failure to correct this defect results in extreme
lateral extension of the ilium, making itPan-like in
three-dimensional disposition. Moreover, if uncorrected,
the pubic symphyseal face fails to reach midline
by several centimeters, once the broken but
otherwise undistorted ischiopubic rami are restored.
Compare figure 4 in (40) and figure 6 in (47) with
figure 8 in (37).
42. J. T. Stern Jr., J. P. Wells, W. L. Jungers, A. K. Vangor,
Am. J. Phys. Anthropol.52, 323 (1980).
43. M. Kagaya, N. Ogihara, M. Nakatsukasa,Primates49, 89
(2008).
44. A. H. Schultz,Hum. Biol.2, 303 (1930).
45. H. O. Elftman,Am. J. Anat.51, 307 (1932).
46. B. G. Richmond, W. L. Jungers,Science319, 1662
(2008).
47. J. T. Stern Jr., R. L. Susman,Am. J. Phys. Anthropol.60,
279 (1983).
48. For funding, we thank NSF [this material is based
on work supported by grants 8210897, 9318698,
9512534, 9632389, 9729060, 9910344, and 0321893
HOMINID–Revealing Hominid Origins Initiative (RHOI)]
and the Japan Society for the Promotion of Science. We
thank the Ministry of Tourism and Culture, the Authority
for Research and Conservation of the Cultural Heritage,
and the National Museum of Ethiopia for permissions and
facilitation. We thank the Afar Regional Government, the
Afar people of the Middle Awash, and many other field
and laboratory workers for contributing directly to the
data. We thank the following institutions and staff for
access to comparative materials: National Museum of
Ethiopia; National Museum of Natural History; Royal
Museum of Central Africa Tervuren, and the Cleveland
Museum of Natural History. We thank B. Senut and
R. Eckhardt for access to the original specimen and casts
ofBAR-1002’00and S. Moya-Sola and M. Kohler for
access to multiple specimens in their care. We thank
M. Brunet for comparative data and C.V. Ward for access
to her large pelvic database which was used extensively
in this analysis. We thank D. Kubo and H. Fukase for
assistance in computed tomography (CT) scanning;
R. Meindl for statistical advice and assistance; and
M. A. McCollum, P. L. Reno, M. A. Serrat, M. Selby,
D. DeGusta, A. Ruth, L. Jellema, S. W. Simpson, and
B.A. Rosenman for aid in data collection and exceptionally
helpful discussions. We thank H. Gilbert and J. Carlson
for help with figures. We thank A. Sanford and
A. Ademassu for the many generations of casts required
to complete this study, R. T. Kono for the rapid
prototyping models, and L. Gudz and E. Bailey for
assistance with illustrations.
Supporting Online Material
www.sciencemag.org/cgi/content/full/326/5949/71/DC1
Figs. S1 to S11
Table S1
References
4 May 2009; accepted 17 August 2009
10.1126/science.1175831
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org71e6
Ardipithecus ramidus
Research Articles

84. 84 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
CombiningPrehensionandPropulsion:
The Foot ofArdipithecus ramidus
C. Owen Lovejoy,
1
Bruce Latimer,
2
Gen Suwa,
3
Berhane Asfaw,
4
Tim D. White
5
*
Several elements of theArdipithecus ramidusfoot are preserved, primarily in theARA-VP-6/500
partial skeleton. The foot has a widely abducent hallux, which was not propulsive during terrestrial
bipedality. However, it lacks the highly derived tarsometatarsal laxity and inversion in extant
African apes that provide maximum conformity to substrates during vertical climbing. Instead, it
exhibits primitive characters that maintain plantar rigidity from foot-flat through toe-off,
reminiscent of some Miocene apes and Old World monkeys. Moreover, the action of the fibularis
longus muscle was more like its homolog in Old World monkeys than in African apes. Phalangeal
lengths were most similar to those ofGorilla. TheArdipithecusgait pattern would thus have been
unique among known primates. The last common ancestor of hominids and chimpanzees was
therefore a careful climber that retained adaptations to above-branch plantigrady.
T
he modern human foot is unique among
mammals because it exhibits a series of
adaptations that allow it to dissipate ki-
netic energy during foot strike in walking and
running (and thus preserve its structural integrity),
and to then transform into a rigid lever for pro-
pulsion during toe-off. Until now, the natural
history of these adaptations has been shrouded
becauseAustralopithecusalready exhibits most
of them.Ardipithecus ramidus(1) now reveals
much more about their evolution.
Well-preserved foot elements recovered from
the Lower Aramis Member include a talus, me-
dial and intermediate cuneiforms, cuboid, first,
second, third, and fifth metatarsals, and several
phalanges (2 ) (Fig. 1). OtherAr. ramidusfoot
elements are fragmentary and less informative.
Here we describe these key foot elements, fo-
cusing on their implications for the locomotion
of early hominids.
Talus.Hominoid tali vary extensively, limit-
ing their value for inferring locomotor habitus
(3). Even so, a deep, anteromedially projecting
cotylar fossa is frequently generated by habitual
tibiotalar contact during extreme ankle dorsiflex-
ion [cartilage modeling; type 4 (4)]. Such mor-
phology is typical of extant African apes and
some Miocene hominoid taxa [e.g.,KNM-RU
2036 F(5)], but is only minimally expressed in
ARA-VP-6/500-023(Ar. ramidus) andA.L. 288-
1as(Au. afarensis)(6).
Trochlear geometry, absent a calcaneus or
distal tibia to provide talar orientation, does not
specify foot placement (7), but several characters
are possible correlates of talocrural and subtalar
mobility. The talar axis angle (fig. S1) (7–9) is
both remarkably low and minimally variable in
Au. afarensisand other early hominids, con-
sistent with their stereotypically pronounced knee
valgus during terrestrial bipedality (10). By con-
trast, this angle inARA-VP-6/500-023lies within
the ranges of quadrupedal primates (Table 1). In
addition, the flexor hallucis longus groove on
the posterior aspect of the talus is both substan-
tially more angulated and more trapezoidal in
form (i.e., its superior surface is broader than
its inferior), indicating a much greater range
of tendon obliquity during locomotion than in
A.L. 288-1, in which the groove is both more ver-
tical and more parallel-sided (8 ). Together, these
suggest more knee rotation during stance phase
than was likely the case inAu. afarensis(10), even
though theArdipithecuspelvis implies full
extension of both the knee and hip during upright
gait (11). A prominent tubercle marks the pres-
ence of an anterior talofibular ligament inARA-
VP-6/500-023. This landmark is absent in African
apes but is usually present inHomo sapiens.
However, bony evidence of local joint capsule
expansion is remarkably variable (12 ).
Medial cuneiform and first metatarsal.ARA-
VP-6/500-088is a medial cuneiform (Fig. 2).
Although damaged, a portion of its proximal
joint surface articulates with the intact interme-
diate cuneiform (ARA-VP-6/500-075). The first-
ray metatarsal (Mt1) (ARA-VP-6/500-089 ) is
preserved for its entire length. Its superoproximal
surface is intact. This allows direct examination
of first-ray abducence (Fig. 2), which was sub-
stantial and similar to that shown by extantPan.
As in African apes, the proximal Mt1 facet ex-
hibits substantial spiral concavity for conjunct
rotation on the hemicylindrical medial cunei-
form facet (13). TheARA-VP-6/500proximal
Mt1 base is therefore unlike its counterpart in
Australopithecus, in which it is reniform and
faces directly distally (14), indicating that it was
permanently adducted [(15,16); for a contrary
view, see (17 ,18)].
Cuboid.The human midtarsus is much lon-
ger than are those of extant African apes. Tarsal
elongation increases lever arm length during
toe-off (19–21). Elongation of the metatarsals
would have also accomplished this goal, but
would subject them to frequent midshaft fracture
or failure of their tarsometatarsal joints [both are
still common in modern humans, and their cause
may be as simple as a misstep (22)]. It has been
reasonably assumed that the human cuboid is
highly derived from a more chimpanzee-like one
for powerful plantarflexion during upright walk-
ing and running. Indeed, the eccentric placement
of the modern human cuboid’s calcaneal process
is uniquely derived for enhanced midfoot rigidity
during plantarflexion (7,9).
The morphology of the African ape lateral
midfoot contrasts greatly with that of humans.
Their cuboids, naviculars, and lateral cuneiforms
are greatly foreshortened. Associated soft tissues
permit substantial laxity at their midtarsal and
tarsometatarsal joints (9 ,23–27). Such laxity fa-
cilitates plantar conformity to substrates during
pedal grasping and vertical climbing (9,28). How-
ever, it greatly compromises any plantarflexor
torque about their metatarsal heads. The African
ape cuboid’s facets for the fourth and fifth
metatarsals (Mt4 and Mt5) are, in addition,
mildly concave, permitting such potential motion
(9,23,24). That such morphology is highly
derived can be established by the midfoot mor-
phology of Old World monkeys, which rely on
plantarflexor torque during above-branch running
and leaping—behaviors largely abandoned by
great apes.
When normalized for body size, the Old
World monkey cuboid is longer than are those
of African apes (Table 1 and fig. S2). It is there-
fore notable that theARA-VP-6/500cuboid is
equally long (Fig. 3). Moreover, its Mt4 facet is
sinusoidal (suggesting immobility), and its Mt5
facet is virtually flat (also suggesting immobil-
ity). WasAr. ramidusmorphology derived for
bipedality from a shortened African ape-like
midfoot, or was it primitive? Resolution of this
important issue is provided by another character
of theAr. ramidusmidfoot that also varies strik-
ingly among extant taxa.
TheAr. ramiduscuboid exhibits an expan-
sive facet for an os peroneum: a large sesa-
moid in the fibularis longus (= peroneus longus)
tendon (29 ,30). An obvious homolog is virtu-
ally constant in humans and Old World mon-
keys, because both taxa exhibit a constant,
prominent underlying articular facet (Fig. 3).
However, the os peroneum is usually cartilag-
inous or only partially calcified in humans,
which accounts for routine reports of its absence
in radiographic surveys. Both the sesamoid
Ardipithecus ramidus
1
Department of Anthropology, School of Biomedical Sciences,
Kent State University, Kent, OH 44240, USA.
2
Department of
Anatomy, Case Western Reserve University School of
Medicine, Cleveland, OH 44106, USA.
3
University Museum,
University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033,
Japan.
4
Rift Valley Research Service, P.O. Box 5717, Addis
Ababa, Ethiopia.
5
Human Evolution Research Center and
Department of Integrative Biology, 3101 Valley Life Sciences
Building, University of California, Berkeley, CA 94720, USA.
*To whom correspondence should be addressed. E-mail:
[email protected]
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 200972e1
Ardipithecus ramidus

85. 85www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
and its facet are absent in extant great apes
(31).
The fibularis longus, in whose tendon this
ossicle resides, performs substantially different
functions in Old World monkeys, African apes,
and humans. In Old World monkeys, in addition
to adducting the hallux, it is also poised to pre-
vent laxity in the cuboid, Mt4, and Mt5 joints.
The mass, location, and breadth of the muscle’s
tendon [as judged from its contained os peroneum
(Fig. 3)] suggest that it readily resists plantar
cavitation of the tarsometatarsal joints, which
would dissipate plantarflexor torque. In stark
contrast, any supportive function in either Afri-
can ape has been eliminated along with the os
peroneum, and these taxa exhibit substantial
midtarsal laxity even during plantigrade pro-
pulsion (9 ,23–27,31).
In humans, the fibularis longus tendon sup-
ports the longitudinal arch and controls pedal
inversion, both critical to successful bipedal-
ity [reviewed in (7–9)]. Moreover, the human
fibularis longus no longer resides in the cuboid’s
prominent groove as it does in Old World mon-
keys and African apes. Instead, it (and its
contained os peroneum) has become relocated
more proximolaterally, outside and essentially
perched above (in plantar view) the sometimes
still-present cuboidal groove (32 ). The latter
likely continues to be generated by retained
elements of pattern formation that still underlie
cuboid osteogenesis [genetically derived but
selectively neutral; type 2B (4 )] (33).
These are not trivial anatomical shifts in
African apes (elimination of the sesamoid) or
humans (relocation of the tendon’s pathway).
Elimination of the os peroneum in African apes,
coupled with the marked anteroposterior short-
ening of their cuboid, causes the fibularis longus
tendon to pass immediately behind and parallel
to the axis of their cuboidometatarsal joints
(9,23,24). This allows substantial plantar con-
formity to the substrate even during powerful
grasping of the great toe by the fibularis longus.
In contrast, translation of the tendon poste-
riorly in derived hominids, along with its new
additional attachment to the medial cuneiform,
reroutes the tendon’ s course so that it crosses the
plantar foot more obliquely, thereby improving
resistance to flexion in the cuboidometatarsal
and especially the cuneiform-metatarsal joints.
Both the transverse and longitudinal arches
increase the tendon’s moment arm to provide
such resistance. Relocation of the os peroneum
is thus a morphological signal of the presence of
these arches. The elimination of any first-ray
abduction in humans has allowed the os
peroneum to vary substantially (and become
merely cartilaginous), because most of the
translation of the tendon has been eliminated by
permanent adduction of the great toe.
Ar. ramidusmorphology is clearly primitive.
Its fibularis longus tendon passed over an ex-
ceptionally broad, shallow facet underlying
what must have been a relatively massive os
peroneum similar in size to those of most Old
World monkeys (Fig. 3). Its fibularis longus
could thus both adduct the great toe and
plantarflex the foot, but still aid, to some extent,
in support of the cuboidometatarsal joints. Not
until an abducent first ray was abandoned could
the os peroneum then be relocated as it is in
later hominids, thereby enhancing its supportive
function. It is notable, therefore, that the os
peroneum facet inOH-8is highly derived in
location and morphology (Fig. 3).
Navicular.ARA-VP-6/503is only a small
fragment of navicular, but is sufficiently pre-
served to further illustrate the natural history of
the hominid midfoot. Despite its fragmentary
condition, it suggests a primitive anteroposterior
length intermediate between its homologs in ex-
tant African apes and humans. This suggests
that there has been substantial midtarsal abbre-
viation since the common ancestor of gorillas,
chimpanzees, and humans (outlined in fig. S3),
and subsequent elongation of the midtarsus dur-
ing hominid evolution. Indeed, a substantial por-
tion of measurable cuboidal elongation in humans
can be attributed to proximal extension of its cal-
caneal process, which is now located more
eccentrically to further stabilize the calcaneocuboid
joint during toe-off (7). Although theAr. ramidus
cuboid’ s calcaneal process is moderately elongate,
it retained a primitive, more centroidal position.
A
B CD
Fig. 1.Digitally rendered composite foot ofARA-VP-6/500.(A) Plantar view. (BtoD) Dorsal, medial, and
anteromedial oblique views, respectively. Better-preserved elements from both sides were assembled as
the left foot ofAr. ramidus. Mirror-imaged elements are the talus, cuboid, Mt2 shaft, and some phalanges.
The intermediate and terminal phalanges are provisionally allocated to position and side. Note the
anteroposteriorly strongly abducent first ray (Fig. 2), elongate cuboid (Fig. 3), and large os peroneal facet
located more distolaterally than inHomo. Cuboids of African apes generally lack an os peroneum. Scale
bars,5cm.ImageryisbasedonCTscanstakenat50-to150-mm voxel resolution.
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
72e2
RESEARCH ARTICLES
Research Articles

86. 86 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Lateral metatarsals.ARA-VP-6/1000is a
right Mt2 lacking its head and the plantar por-
tion of its base (Fig. 4). However, both plantar
cornua are preserved, thereby permitting reason-
able reconstruction of its length. The base is
large. The ratio of basal height as preserved (no
reconstruction or estimation) to metatarsal length
lies in the upper range ofHomo(fig. S4).ARA-
VP-6/1000exhibits only minimal longitudinal
curvature (Fig. 4) but exhibits substantial shaft
torsion, which orients it for opposition with the
Mt1, as in extant African apes (fig. S5).ARA-VP-
6/500lacks an intact Mt2, but its intermediate
cuneiform also allows comparison of the joint’s
dorsoplantar Mt2 facet height with an estimate of
body size. This ratio lies near the upper limit of
the human range and outside the ranges of the
African apes (fig. S6), suggesting a similarly
robust base size inARA-VP-6/500.
The Mt2’s large base is readily explicable in
light of its role as the foot’s medial mainstay
during bipedal toe-off. The dorsal edge ofARA-
VP-6/1000’s proximal joint surface exhibits
paired chondral invaginations (Fig. 4) that are
rare in the Mt2s of either gorillas or chimpan-
zees [one single (lateral) facet inN= 50]. These
cannot reflect habitual contact with the inter-
mediate cuneiform, as this would require impos-
sible joint cavitation. Nor does the intermediate
cuneiform ofARA-VP-6/500or any other higher
primate bear matching projections; there are,
instead, slight corresponding invaginations of
its dorsal surface as well. Each invagination of
the Mt2’s dorsal surface lies just proximal to
medial and lateral rugosities. In humans, these
mark receipt of medial and lateral expansions
of the joint’s dorsal capsule [(34); this study].
Habitual, intermittent pressure against these
local tarsometatarsal joint expansions almost
certainly induced the paired subchondral de-
pressions in the Aramis bone’s dorsal surface.
Their probable etiology [chondral modeling;
type 4 (4 )] is therefore informative. Substantial
spiraling of the Mt2 shaft places the bone’s
distal end into functional opposition to the Mt1
in African apes and would have done so inAr.
ramidus(35). Such torsion is most pronounced
in the Mt2 because it lies adjacent to the hallux,
and because Mt2 rotation is restricted by the
mortising of its base between the medial cunei-
form and lateral cuneiform/Mt3 laterally. The
bases of the more lateral rays are less restricted
and thus have (progressively) less prestructured
torsion.
The developmental biology of tendon and
ligament attachments is complex (36 ), but a
markedly rugose insertion likely signals sub-
stantial Sharpey fiber investment via pattern
formation (37 ). This is especially the case for
eutherian tarsometatarsal joints, which appear to
have sacrificed their proximal metapodial growth
plate to encourage a more rigid syndesmosis
(38). The markedly rugose tarsometatarsal joint
capsule in theAr. ramidusMt2 suggests that it
was an adaptation [direct selection acting on
Abduction Angle (degrees)
GA
B
C
D
E
F
80
60
40
20
6/500
Fig. 2.First-ray abduction inAr. ramidus. Abduction of the first ray is dependent on soft tissue structures
operating about the joint, but can be readily inferred from preserved joint structure. (A) Dorsal view of
female gorilla (CMNH-B1801) with Mt1 articulated with the medial and intermediate cuneiforms, showing
maximum abduction without joint cavitation. (B)ARA-VP-6/500articulated in a similar fashion (casts). Note
that the two Mt1s differ in axial orientation. This difference may be a consequence of habitual bipedality in
Ar. ramidus, which did not exhibit ape-like midtarsal laxity. Abduction is measured as the angle between a
tangent to the distal surface of the intermediate cuneiform and the centroidal axis of the Mt1. It is 68° in
ARA-VP-6/500.(CandD) CT rendering ofARA-VP-6/500in similar (C) and exploded (D) views. (E)
Approximate posterior, medial, and anterior views of the medial cuneiform. (F) Medial view in dorsoplantar
orientation. Although its inferior portion has suffered extensive damage, its posterosuperior portion is intact
and articulates perfectly between the intermediate cuneiform and the dorsoproximal joint surface of the
Mt1. Note the intact posterior portion of the plateau-like projection of the medial cuneiform’ s Mt1 facet
distomedially. This is rare inPanbut occasional inGorilla. Note the nonsubchondral isthmus [white arrow in
(C)] separating the two articular facets on the dorsum of the Mt1. These likely record rotation of the proximal
phalanx in the MP joint during grasping and terrestrial bipedality (see text). They are notably absent inAu.
afarensisbut usually present in African apes. Scale bars, 2 cm [(A) to (E)], 1 cm (F). (G) Abduction angle in
ARA-VP-6/500, humans, and African apes [N= 15 each taxon; boxes show median, quartiles, and extreme
cases in each taxon (asterisk indicates case >1.5 box lengths from quartile box boundary)].
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 200972e3
Ardipithecus ramidusArdipithecus ramidus

87. 87www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
morphogenetic fields; type 1 (4)] to upright walk-
ing and running, absent any substantial load-
sharing by a still abducent great toe. Moreover,
the total morphological pattern of theAr. ramidus
foot suggests that it exhibited a noninverted foot-
flat during midstance [i.e., unlike that ofPan
(23–27)]. The primary terrestrial role of the hal-
lux, as in apes, would have been for balance
rather than for propulsion (but see below).
Powerful fulcrumation occurred only on the lat-
eral metatarsal heads inAr. ramidus, especially
that of the Mt2, whose role in humans remains
especially prominent in bipedality even after
having been reinforced by the addition of a
permanently adducted Mt1.
ARA-VP-6/505is a virtually intact left Mt3
(Fig. 4). Its preserved head shows two particu-
larly important characters. First, it exhibits dorsal
doming in excess of African ape metatarsals.
Second, a deep-angled gutter isolates the head
from the shaft at the dorsal epiphyseal junction.
Although a similar gutter is also found in ape
metatarsals, it is considerably shallower, consist-
ent with substantially less loading and excursion
during metatarsophalangeal joint (MP) dorsiflex-
ion [cartilage modeling; type 4 (4 )]. Moreover, in
ARA-VP-6/505, the angle between the head’s
dorsoplantar axis and the dorsoplantar axis of its
base shows slight external torsion of the shaft,
which would have optimized MP joint alignment
during toe-off. This implies that growth plate
loading during terrestrial bipedality predomi-
nated over that generated during grasping (i.e.,
it exhibits far less torsion than the Mt2, and also
lacks the medial and lateral joint capsule com-
pression facets present inARA-VP-6/1000). The
gutter also implies that loading during terrestrial
bipedality was applied during substantial toe-out
during and after heel-off, coupled with external
rotation of the foot during late toe-off. Pro-
nounced doming is entirely absent in theARA-
VP-6/500Mt1, confirming that the first ray did
not participate substantially in propulsion (fig.
S12). Doming is present in theAu. afarensisMt1,
again implying terrestrial bipedality with a
permanently adducted great toe.
The shaft of theARA-VP-6/505Mt3 is only
slightly curved (Fig. 4) and its base is well pre-
served, lacking only a minor portion of its supero-
medial corner. Its base morphology is remarkably
similar to that of the human Mt3 in having a
dorsoplantarly tall proximal articular surface
(Fig. 4 and fig. S7). African ape Mt3 bases are
instead regularly subdivided into distinct upper
and lower portions by deep semicircular notches
of their medial (for Mt2) and lateral (for Mt4)
surfaces (Fig. 4 and Table 1). These serve as
passageways and surfaces for tarsometatarsal
and transverse intermetatarsal ligaments. The ab-
breviated dorsoplantar height and distinctly rhom-
A F
B C
DE
Fig. 3.Natural history of the hominoid midfoot. (A) The os peroneum. This sesamoid
(white arrow in a ligamentous preparation ofPapio anubis) is a large and prominent
inclusion in the fibularis longus tendon of Old World monkeys, residing on an appropriately
large inferolateral facet of the cuboid. In Old World monkeys, the muscle inserts at the Mt1
base, acting as both plantarflexor and hallucal adductor. Because some flexion can occur at
both the calcaneocuboid and tarsometatarsal joints during climbing and terrestrial walking
(9,23,24,27), the fibularis longus tendon also aids plantar rigidity during plantarflexion.
(BtoE) Plantar surfaces of hominoid cuboids. (B) Chimpanzee (CMNH-1726). In apes, the
cuboid is anteroposteriorly short and the groove in which the fibularis longus tendon lies is
narrow and deep, usually with high walls. It is converted to a retaining tunnel by a homolog
of the human short plantar ligament (56–58). Ape cuboids essentially lack functional os
peronei [they occasionally contain small, nonfunctional, chondral bodies (31)]. (C)ARA-VP-
6/500-081. InAr. ramidusthe surface medial to the facet over which the tendon must pass
is rugose and subperiosteal, confirming that a laterally placed os peroneum elevated its
travel on the facet. (D) Human (KSU-01206). (E) OH-8(cast; reversed). In these later
hominids, the fibularis longus no longer lies in the cuboidal groove, but is instead elevated above and posterior to it by the os peroneum residing on a facet located
proximolateral on the groove’s proximal wall (white arrows) (32). UnlikeAr. ramidus, the fibularis longus inserts into the medial cuneiform and no longer adducts the
first ray. Scale bar, 2 cm. (F) Natural log-log scatterplot of medial cuboid length and cube root of estimated body mass in extant anthropoids (42). A regression line
(reduced major axis;y= 1.184x + 1.666;r= 0.836,N= 26) has been fitted to the combined cercopithecines and colobines. The most parsimonious interpretation of
these data is that cuboid length inAr. ramidusis primitive, and that the bone was elongated in later hominids (including elongation of its calcaneal process) but
shortened in African apes in order to enhance hallucal grasping and plantar compliance to substrates during vertical climbing. The ranges and medians for a similar
metric clarify these relationships in fig. S2.
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
72e4
RESEARCH ARTICLES
Research Articles

88. 88 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
boidal form of African ape Mt2 and Mt3 bases
[direct selection acting on morphogenetic fields;
type 1 (4)] permit more plantar conformity and
tarsometatarsal laxity during grasping, consistent
with recent observations that the midtarsal break
combines motion at both the lateral tarsometa-
tarsal and midtarsal joints (figs. S4 and S6 to S9)
(9,23,24,27).ARA-VP-6/505lacks this dis-
tinctive central notching morphology. This pre-
sumably reflects retention of soft tissue structures
similar to those of humans. These enhance mid-
tarsal and tarsometatarsal bending resistance from
foot-flat through toe-off, ultimately culminating
in the emergence of the proximal part of the long
plantar ligament, which is likely derived in hu-
mans (13,33). These conclusions receive strong
support from geometric analysis of joint surface
section moduli of the Mt3 (figs. S8 and S9).
Phalanges.Several proximal pedal phalanges
from the lateral rays ofAr. ramiduspreserve a
base. Three are complete with proximal ends
evincing clear, typically hominid dorsiflexive
cants (figs. S10 and S11). Canting is more pro-
nounced in modern humans as a consequence of
the reduction of phalangeal curvature (39) (fig.
S10) and abbreviation of the intermediate pha-
lanx (figs. S12 and S13). Phalangeal shape ratios
(40) are not particularly informative, but they do
show thatAr. ramidusphalanges are moderately
robust (i.e., like those ofPanandProconsuland
unlike those ofAtelesorHylobates), with
moderately deep trochleas. Midshaft robusticity
is similar to that inPan,Proconsul, and most Old
World monkeys. Manual/pedal phalangeal ratios
are like those in extant hominoids and unlike
those inProconsul[for discussion see (40,41)].
When complete phalanges from ray 4 of
ARA-VP-6/500are normalized by body size,
their lengths fall near theGorillamean but below
values inPan, which may have therefore wit-
nessed substantial phalangeal elongation since
the last common ancestor of African apes and
humans (fig. S13) [for further discussion, see
(42)]. Pedal phalanges inAr. ramidusare rela-
tively shorter than those of New World monkeys
(regardless of locomotor pattern), orangutans,
and gibbons. Phalangeal curvature is moderate to
large. The included angle ofARA-VP-6/500-094,
an intact proximal phalanx of the fourth pedal ray,
is 58°. However, its base is substantially canted,
which obscures its joint angulation in lateral view.
Expansion of the apical tufts of the terminal pha-
langes is moderate.
The first ray during terrestrial gait.The
dorsal articular margin of the Mt1 head ofARA-
VP-6/500preserves detailed evidence of howAr.
ramidusused its foot in some locomotor settings.
Its dorsal surface bears two symmetrically placed
and equal-sized V-shaped facets separated by a
central nonarticular isthmus (Fig. 3). Each facet
appears to have been generated by axial rotation
of the ray’s proximal phalanx at its MP joint
[cartilage modeling; type 4 (4)].
The Mt1’s dorsolateral facet was presumably
generated during grasping by external rotation
of both the Mt1 and its proximal phalanx, which
would have brought the hallux into opposition
with the lateral foot. The Mt1’s dorsomedial
facet would then have been generated by inter-
nal rotation that occurred when the foot was
emplaced on a terrestrial substrate with the first
ray in substantial abduction (because it exhibits
no doming; see above). ThisAr. ramidusmor-
phology is especially notable because of its re-
markable symmetry. Although similar rotation
facets occur regularly on the Mt1s of bothPan
andGorilla, they are most often asymmetrical
and also appear to be generally deeper. In some
Gorillaspecimens, the medial facet is more
prominent than the lateral, which suggests that
during terrestrial locomotion, greater relative
loads were imposed on its Mt1s than inAr.
ramidus.
This would at first seem to be a paradox,
because the African apes are not habitual bipeds.
However,Ar. ramidusretained primitive features
[a prominent os peroneum, substantial tarsometa-
tarsal joint rigidity, a long midtarsus, and soft
tissue characters that likely accompanied these
(Table 2)] that allowed powerful plantarflexion
about its lateral metatarsal heads, including what
must have been a substantial contribution by its
peroneal compartment. The African apes, by
contrast, have lost such capacity in favor of sub-
stantial midtarsal laxity. This has greatly compro-
mised the plantarflexor impulse on their lateral
metatarsal heads. Partial accommodation appears
to be provided by occasional or even regular
impulse by their Mt1 during terrestrial gait. The
Mt1s ofAustralopithecuslack any evidence of
comparable facets (15). This, and the prominent
doming of their Mt1, now serve as further con-
firmation that the taxon lacked any first-ray abduc-
tion, and almost certainly exhibited a longitudinal
arch—features that are consistent with their de-
rived ankle morphology (8 ,9,15,23,24) and the
Laetoli footprints (43).
Interpretations and dynamics.Ar. ramidus
is the only known hominid with an abducent
great toe (15,16,44). Its foot, along with other
postcranial elements, indicates that the Late
Miocene hominid precursors ofAr. ramidus
practiced mixed arboreal and terrestrial locomo-
tion during which the lateral forefoot became
extensively adapted to upright walking, even as
the medial forefoot retained adaptations for ar-
boreal exploitation.
During the gait cycle, fibularis longus con-
traction would also have stabilized the proximal
ankle joints. The moderate to strong talar dec-
lination of the angle between the trochlea and
that of the ankle’s axis of rotation, in combina-
tion with clear evidence of abductor stabilization
of the hip during stance phase (11), together sug-
gest that the foot was placed near midline. The
knee may have been in greater external rotation
than is typical in human andAustralopithecus-
like (i.e., accentuated) valgus (10), with compen-
sation by means of a more extensive range of
knee rotation throughout stance phase.Ar. ramidus
therefore may have lacked the consistently ele-
vated bicondylar angle ofAustralopithecus.
Ar. ramiduslikely relied on situationally de-
pendent lordosis to generate functional hip ab-
duction (minimum pelvic tilt) during stance
Table 1.Talus, cuboid, Mt1, Mt5, and Mc5 (fifth metacarpal) metrics inAr. ramidusand other anthropoids. Values in parentheses, except for the
leftmost column, denote standard deviation.
Taxon (N)
Angle between trochlear
axis and talocrural
rotation axis (°)*
Max. cuboid
length†/body
size‡
Mt1/body size‡ Mc5/body size‡ Mt5/body size‡ Mc5/Mt5
Old World monkeys (27) 13.2 (2.2)§ 1.51 (0.13) 4.4 (0.40) 4.2 (0.30) 6.6 (0.43) 0.73 (0.07)
New World monkeys (11) 1.48 (0.12) 4.7 (0.22) 4.5 (0.85) 6.6 (0.14) 0.80 (0.06)
Homo(30) 10.2 (2.3) 1.81 (0.10) 3.9 (0.18) 2.6 (0.17) 4.6 (0.13) 0.75 (0.04)
Australopithecus
and earlyHomo(12)
7.4 (1.4)
ARA-VP-6/500 14.5¶ 1.41 4.1 3.2 4.9 0.87
Pan(26) 15.5 (2.9) 1.15 (0.06) 4.1 (0.31) 4.3 (0.26) 5.1 (0.07) 1.12 (0.07)
Gorilla(29) 17.8 (2.7)║ 1.14 (0.08) 3.6 (0.25) 3.8 (0.24) 4.9 (0.10) 1.02 (0.03)
Pongo(16) 18.4 (3.5) 1.18 (0.16) 3.6 (0.22) 5.3 (0.39) 6.7 (0.11) 1.03 (0.05)
*Data from (9);Australopithecusand earlyHomosample is composed ofStw-102,Stw-363,Stw-486,Stw-88,TM-1517,A.L. 288-1,Omo323-76-898,KNM-ER 813,1464, 1476, 5428, and
OH-8. †InHomothis usually includes the calcaneal process.‡Body size estimated as equal contributions of the geometric means of metrics of the wrist and talus (42). Old World
monkey taxa includePapio,Mandrillus,Macaca,Trachypithecus,Semnopithecus,Colobus,Cercocebus, andPresbytis. New World monkey taxa areAtelesandAlouatta. §Old World monkey
taxa for trochlear angle are fromPapio(9). ¶See fig. S1. ║Data forGorilla(9) are a weighted mean for both species.
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 200972e5
Ardipithecus ramidus
Ardipithecus ramidus

89. 89www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
phase (11). Combined with the pedal characters
described here, this suggests a form of primitive
terrestrial bipedality in which the foot was em-
placed at or only slightly lateral to midline, with
the great toe typically in abduction (as often
occurs in African apes) and the lateral forefoot
in external rotation. Fulcrumation occurred along
the oblique axis (fig. S12) and was obviously
achieved by the triceps surae, likely substantially
aided by a powerful and particularly robust
fibularis longus. Balance before and during
propulsion was achieved by the opposing actions
of (i) a medially emplaced great toe, and (ii)
plantarflexion by the fibularis longus, which
would also tend to evert the foot. Thus, the lat-
eral compartment must have been very powerful
and central to its gait pattern. Indeed, the large os
peroneum suggests that once the great toe was
restrained by friction against the substrate, con-
traction of the fibularis longus could further
enhance plantarflexion during heel-off through
toe-off, while simultaneously maintaining rigid-
ity in the midtarsal and tarsometatarsal joints and
preventing inversion induced by substrate dis-
conformities. At the same time, the symmetric
rotary facets of the Mt1’s distal joint surface, in-
duced by MP joint motion, suggest that any
eversion was prevented by a broadly abducted
first ray.
Indeed, by the time of emergence ofAu.
afarensis, hominids had evolved substantially
more advanced adaptations to bipedality than
were present inArdipithecus.In the former, the
knee had become tibial dominant (10) with accen-
tuated valgus (exceeding even that of modern
humans). Hip abduction had been established
with a human-like distribution of proximal fem-
oral cortical and trabecular bone (45–47). More-
over, in all known subsequent hominids, the
more posterior location and elevation of the os
peroneum facet on the cuboid [direct selection
acting on morphogenetic fields; type 1 (4)] sig-
nals the presence of longitudinal and transverse
arches, and thereby the addition of the transverse
axis of fulcrumation (fig. S12). The facet’s po-
sition in theOH-8cuboid is virtually human, as is
the length of its calcaneal process (Fig. 2). Dom-
ing and the simpler unnotched dorsal surface of
the Mt1 head characterize bothAu. afarensis
(A.L. 333-21)(15,48) andAu. africanus, con-
firming an immobile first ray with fundamentally
human-like propulsion during toe-off (43 ).
The feet of extant African apes are so pre-
hensile that some early anatomists regarded them
as hand homologies [reviewed and refuted in
(49)]. Compared to the primitive condition of a
long midtarsus as seen in taxa such asProconsul
(5), enhanced grasping required the abandon-
ment of forceful plantarflexion on the lateral
metatarsal heads in favor of increased plantar
laxity at the midtarsal and tarsometatarsal joints.
Primitive morphology was replaced by a short-
ened hindfoot and a talocrural joint modified for
enhanced dorsiflexion and inversion. African apes
eliminated the os peroneum, plantaris (50 ), and a
A
B
C
D
E
F
G
H
Fig. 4.Metatarsals ofAr. ramidusand extant hominoids, right Mt2 (top) and left Mt3 (bottom). (A) Enlarged
CT rendering of dorsal surface ofARA-VP-6/1000. Facets interpreted to be induced by rotation of its base during
toe-off and grasping are indicated by arrows. These facets are shown to the right in proximal view [provided to
the right in all panels except (E)]. (B) Medial view of entire original specimen (photograph). Although the head
is missing, both cornua are preserved, allowing reasonable estimation of original length. Areas of postmortem
damage are indicated by hatching. (C) Mt2 ofPan(CMNH-B1718). Note distinctive notching for centrally
located tarsometatarsal ligaments. Damage to this area inARA-VP-6/1000prevents interpretation of its
complete basal form. (D ) Modern human Mt2. Proximal surface is superoinferiorly elongate and lacks dorsal
facets, consistent with adaptation to bipedality absent an abducent great toe. (EandF) Dorsal (CT) and lateral
(photograph) views ofARA-VP-6/505, an Mt3. Hatching shows minor postmortem damage. (GandH) Mt3s of
PanandHomospecimens whose Mt2s are shown in (C) and (D). As is typical of the chimpanzee, the Mt3 shows
bilateral notching, although it is not as pronounced in this specimen as in most. Note the striking similarity in
the basal morphology of the two hominid Mt3 bases, which suggests that this morphology is likely primitive
rather than derived, given the exceptionally great differences in locomotor behavior. CT methods:ARA-VP-6/
1000,pQCTat150m m;ARA-VP-6/505, microCT at 80mm. Scale bar, 2 cm.
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
72e6
RESEARCH ARTICLES
Research Articles

90. 90 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
functional quadratus plantae (51). This character
constellation (Table 2) suggests shifts in genes
encoding regulatory and signaling molecules
modifying fields underlying pedal structure
(52). The human plantaris is hardly functional,
but its retention and association with the plantar
aponeurosis as in Old World monkeys (53)
signals retention of primitive features from which
specialized aspects [e.g., medial head of the
quadratus plantae; posterior part of the long
plantar ligament (33)] could have been readily
derived under selection for advanced terrestrial
bipedality.
Hominid morphology has often been pre-
sumed to have evolved from ancestral morpho-
types like those of extant African apes.Ar.
ramidusnow establishes that this was not the
case. The hominid foot was instead derived from
one substantially less specialized.
References and Notes
1. T. D. White, G. Suwa, B. Asfaw,Nature375, 88 (1995).
2. T. D. Whiteet al.,Science326, 64 (2009).
3. C. O. Lovejoy, inPrimate Morphology and Evolution,
R. H. Tuttle, Ed. (Mouton, The Hague, 1975),
pp. 291–326.
4. The trait nomenclature system used here is taken from
(47,54) and is briefly as follows [for more complete
explanations, see (2)]. Type 1: traits whose morphogenesis is
the direct consequence of pattern formation; usually (but
not always) subject to direct selection. Type 2: traits that are
genetic but are pleiotropic to, or result from hitchhiking on,
type 1 traits and are not themselves subject to selection
[2A: parent type 1 is inferred to be under selection but its
secondary effects are not; 2B: neither parent trait nor
derivative is inferred to be under selection (rare)]. Type 3:
resulting from a systemic growth factor. Type 4: epigenetic
consequence of osteochondral remodeling and/or response
to environmental stimuli, i.e., not heritable but useful in
interpreting behavior. Type 5: similar to type 4, but
uninformative.
5. A. C. Walker, M. Pickford, inNew Interpretations of Ape
and Human Ancestry, R. L. Ciochon, R. S. Corruccini,
Eds. (Plenum, New York, 1983), pp. 325–413.
6. D. C. Johansonet al.,Am. J. Phys. Anthropol.57, 403
(1982).
7. V. T. Inman,The Joints of the Ankle(Williams and
Wilkins, Baltimore, 1976).
8. B. Latimer, J. C. Ohman, C. O. Lovejoy,Am. J. Phys.
Anthropol.74, 155 (1987).
9. J. M. DeSilva,Proc. Natl. Acad. Sci. U.S.A.106, 6567
(2009).
10. C. O. Lovejoy,Gait Posture25, 325 (2007).
11. C. O. Lovejoyet al.,Science326, 71 (2009).
12. C. O. Lovejoy, K. G. Heiple,Nature235, 175 (1972).
13. O. J. Lewis,Functional Morphology of the Evolving Hand
and Foot(Clarendon, Oxford, 1989).
14. Schultz noted that the medial cuneiform-Mt1 joint is
“directed more forward and less sidewise in the mountain
gorilla than in the other apes”[(20), p. 395]. Our sample
contained only“western”specimens (G. g. gorilla).
15. B. Latimer, C. O. Lovejoy,Am. J. Phys. Anthropol.82, 125
(1990).
16. H. M. McHenry, A. L. Jones,J. Hum. Evol.50, 534
(2006).
17. R. J. Clarke, P. V. Tobias,Science269, 521 (1995).
18. K. D. Hunt,J. Hum. Evol.26, 183 (1994).
19. A. H. Schultz,Symp. Zool. Soc. London10, 199 (1963).
20. A. H. Schultz,Hum. Biol.2, 303 (1930).
21. A. H. Schultz,Folia Primatol.1, 150 (1963).
22. J. T. Campbell, L. C. Schon, inOrthopaedic Surgery:
The Essentials, M. E. Baratz, A. D. Watson, J. E. Imbriglia,
Eds. (Thieme, New York, 1999), pp. 591–614.
23. J. M. DeSilva, thesis, University of Michigan (2008).
24. J. M. DeSilva,Am. J. Phys. Anthropol.10.1002/
ajpa.21140 (2009).
25. H. Elftman, J. Manter,Am. J. Phys. Anthropol.20, 69 (1935).
26. H. Elftman, J. Manter,J. Anat.70, 56 (1935).
27. E. Vereecke, K. D’Aout, D. De Clercq, L. Van Elsacker,
P. Aerts,Am. J. Phys. Anthropol.120, 373 (2003).
28. J. G. Fleagleet al.,Symp. Zool. Soc. London48, 359
(1981).
29. T. Manners-Smith,J. Anat.42, 397 (1908).
30. J. M. Le Minor,J. Anat.151, 85 (1987).
31. In twoPanspecimens dissected for this paper, a small
cartilage nodule (invisible upon x-ray) could be palpated
within the tendon but had no effect on its caliber, nor was
either accompanied by a facet. This accounts for reports by
some earlier anatomists that the sesamoid is present in
African apes (although it is clearly present in gibbons, which
exhibit regular facets). A recent review (55) concluded an
incidence of 2/11 for the os peroneum in apes from classical
literature. However, this datum is likely unreliable, because
many early authors did not inspect the tendon closely and
typically reported its structure“as in humans,”whereas“the
few anatomists who have explicitly looked for the [os
peroneum] have noted its absence inGorilla…,Pan
troglodytes…, and…Pongo pygmaeus”[(30), p. 93].
Examination of the cuboid is the most reliable standard
because a functional os peroneum cannot obtain without an
underlying facet, just as the presence of a clearly functional
facet assures the ossicle’s presence, whether or not it was
calcified. In rare cases, a facet-like discoloration of the
bottom of the fibularis longus groove inPancan be seen.
However, the depth and cylindrical nature of the groove
make such facets (if in fact they supported an os peroneum
during life) largely nonfunctional, and we here report only
clearly functional facets. One such human-like facet was
found in aPanspecimen in our survey. Interestingly, its
associated Mt4 and Mt5 facets suggested hypermobility at
these joints far in excess of otherPanspecimens. This
supports the argument that relocation of the tendon, as in
humans, does in fact reduce general midtarsal
mobility.
32. M. A. Edwards,Am. J. Anat.42, 213 (1928).
33. In primates, a fascial sheet spans the plantar aspect of
the foot from calcaneus to cuboid, lateral cuneiform, and
lateral metatarsal bases, and underlies the fibularis
longus tendon in its course to the first ray. In humans
this sheet is described as two separate elements, the
short and long plantar ligaments. However, in apes the
fascial sheet bridges the cuboidal groove, transforming it
into a tunnel confining the fibularis longus tendon
(56,57). Old World monkeys have a rough equivalent
but retain an os peroneum lateral to the tunnel. In
humans, the fibularis longus tendon lies outside (plantar
to) the cuboidal groove, and the short plantar ligament
terminates proximal to it. A second portion of the sheet,
which lies plantar to the tendon, spans it and inserts
distally on the lateral metatarsal bases. In humans this is
distinguished as the long plantar ligament (57,58).
These human divisions of the plantar sheet are therefore
likely derived (13).
Table 2.Primitive and derived states of the foot in extant taxa.
Structure
Extant state
Primitive* Old World monkey Chimpanzee Gorilla Human
Plantaris (51) Present Constant 43% Absent 90% present
Quadratus plantae (51) Present Constant 50% (diminutive
if present)
29% (diminutive
if present)
Constant with novel
medial head
Plantar aponeurosis Thick and dense Thick and dense Minimal Minimal Thick and dense
Structure and
distribution of long
tibial and fibular
flexor tendons to
digits 1 to 5 (51)†
Fused
Fib. to toes 1, 2, 3, 4, 5
Tib. to toes 1, 2, 3, 4, 5
Separate
Fib. to toes 1, 3, 4
Tib. to toes 1, 2, (4), 5
Separate
Fib. to toes 1, 3, 4
Tib. to toes 2 and 5
Separate
Fib. to toes 1, 3, 4
Tib. to toes 2 and 5
Fused
Fib. to toes 1, 2, 3
Tib. to toes 2, 3, 4, 5
Frequency of an os
peroneum‡
See discussions in
(29,30,55)
0.97 (29) <0.04‡ <0.04‡ 0.93+ (30)
Central notching
morphology of Mt3
Present but with
inferior facets
Present but with
inferior facets
86%§
No inferior facets
80%§
No inferior facets
Absent§
Usually no
inferior facets
Posterior part of long
plantar ligament
Probably absent Absent Absent Absent Present
Substantial abbreviation
of cuboid length
Absent Absent Present Present Absent
*The term“primitive”here refers to underived in either African apes or hominids for locomotor patterns established after the last common ancestor of African apes and humans (vertical
climbing, suspension, knuckle-walking in African apes, and terrestrial bipedality in hominids).†In humans, the long tibial flexor is termed the flexor digitorum longus and the long fibular
flexor is termed the flexor hallucis longus.‡Presence or absence of an os peroneum based on the presence or absence of an underlying functional facet (31). In samples ofN= 25+ forPan
andGorilla, one definite, human-like, facet was found inPan[see (31)]. §Current study,N= 25 each taxon.
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 200972e7
Ardipithecus ramidus
Ardipithecus ramidus

91. 91www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
34. R. J. Terry, inMorris’Human Anatomy, J. P. Schaeffer,
Ed. (Blakiston, Philadelphia, 1942), pp. 266–376.
35. We base this simply on torsion of the bone’s diaphysis
because its head was not preserved. We have not
provided a numerical value because estimating the
articular head’s dorsoplantar axis in near-adult (but still
unfused) African ape Mt2 growth plates provides only a
broadly reliable estimate of the exact angle between the
proximal and distal joint axes when the (unfused) head is
articulated on its growth plate.
36. X. Chen, C. Macica, A. Nasiri, S. Judex, A. E. Broadus,
Bone41, 752 (2007).
37. A. Zumwalt,J. Exp. Biol.209, 444 (2006).
38. P. L. Reno, W. E. Horton Jr., R. M. Elsey, C. O. Lovejoy,
J. Exp. Zool. B308, 283 (2007).
39. Considerable debate has centered around the importance
of phalangeal shaft curvature inAu. afarensis, largely
with respect to its etiology. If developmentally plastic
[cartilage modeling; type 4 (4)], then curvature may
imply active grasping history. However, if a direct product
(or pleiotropic effect) of positional information [types
1or2(4)], then it can be moot with respect to
phalangeal function in rapidly evolving species. The fact
that curvature increases during maturation does not
resolve this issue because increasing curvature during
growth is as explicable by positional information as it is
by hypothetical strain regimen(s). Current evidence, such
as the phalangeal curvature in great ape fetuses,
supports a type 1 status. An important element that has
long been ignored is phalangeal morphology inAu.
afarensisitself (A.L. 333-115); see fig. S11.
40. D. R. Begun, M. F. Teaford, A. Walker,J. Hum. Evol.26,
89 (1994).
41. M. Nakatsukasa, Y. Kunimatsu, Y. Nakano, T. Takano,
H. Ishida,Primates44, 371 (2003).
42. C. O. Lovejoyet al.,Science326, 70 (2009).
43. T. D. White, G. Suwa,Am. J. Phys. Anthropol.72, 485
(1987).
44.STW-573(“Little Foot”) preserves a talus, navicular,
medial cuneiform, and Mt1. Its first ray has been
described as partially abducent (17,18). Replacement of
its talus and/or navicular with those of either a
chimpanzee or human would have had no substantial
effect on first-ray abducence; they are not informative.
The higher primate medial cuneiform itself is not mobile
and is therefore equally uninformative, save for its Mt1
joint surface. This surface faces distally, is virtually flat,
and is therefore immobile inSTW-573. The claim that
“the first metatarsal facet overflows from distal to
proximal, as in apes”[see footnote 18 of (17)] is
incorrect, and the dotted line drawn to indicate the
proximal extent of the facet in their figure 3B is
exaggerated. The joint surface ofSTW-573is virtually
identical to that ofOH-8and unlike that of any ape. This
alone falsifies the contention that the specimen’s hallux
was abducent. Indeed, the fibularis longus insertion of
STW-573is described as spanning both the medial
cuneiform and Mt1 and therefore cannot have adducted
the Mt1 even if it retained any mobility in the first
tarsometatarsal joint.
45. C. O. Lovejoy,Gait Posture21, 95 (2005).
46. C. O. Lovejoy,Gait Posture21, 113 (2005).
47. C. O. Lovejoy, R. S. Meindl, J. C. Ohman, K. G. Heiple,
T. D. White,Am. J. Phys. Anthropol.119, 97 (2002).
48. C. V. Ward,Yearb. Phys. Anthropol.119S-35, 185
(2002).
49. T. H. Huxley,Evidence as to Man’s Place in Nature
(Williams and Norgate, London, 1863).
50. The plantaris is“normally present in prosimians,
monkeys, and man, but is lacking in gibbons and
gorillas, nearly all orang-utans, and a considerable
percentage of chimpanzees” (53).
51. W. L. Straus Jr.,Q. Rev. Biol.24, 200 (1949).
52. Interpretation of structure frequency requires constant
attention to the selective mechanism in play. Narrowly
defined (i.e., named) anatomical structures (e.g., long
and short plantar ligaments) emerge from mesenchymal
fields, which are manifestations of their parent positional
information. It is therefore field configuration that is
the target of selection. Constant structures within a
species (e.g., the Achilles’ tendon) suggest low field
variance and intense stabilizing selection. Conversely,
substantially reduced frequencies of a“named”tissue
mass within a species (e.g., the quadratus plantae inPan)
signal the occurrence of underlying field shifts. These
may emanate from selective encouragement of changes
in target structures that share field commonality with
the reduced structure, simple relaxation of selection,
selection against the structure, or some combination of
these.
53. C. G. Hartman, W. L. Straus,The Anatomy of the Rhesus
Monkey(Williams and Wilkins, Baltimore, ed. 1, 1933),
pp. 1–383.
54. C. O. Lovejoy, M. J. Cohn, T. D. White,Proc. Natl. Acad.
Sci. U.S.A.96, 13247 (1999).
55. V. K. Sarin, G. M. Erickson, N. J. Giori, A. G. Bergman,
D. R. Carter,Anat. Rec.257, 174 (1999).
56. W. L. Straus Jr.,Q. Rev. Biol.5, 261 (1930).
57. D. N. Gomberg,J. Hum. Evol.14, 553 (1985).
58. H. Gray,Gray’s Anatomy(Lea and Febiger, Philadelphia,
ed. 26, 2008).
59. Supported by NSF grants 8210897, 9318698, 9512534,
9632389, 9729060, 9910344, and 0321893 HOMINID-
RHOI, and by the Japan Society for the Promotion of
Science. We thank the Ministry of Tourism and Culture,
the Authority for Research and Conservation of the
Cultural Heritage, and the National Museum of Ethiopia
for permissions and facilitation; the Afar Regional
Government, the Afar people of the Middle Awash, and
many other field workers for contributing directly to the
data; the National Museum of Ethiopia, National
Museums of Kenya, Transvaal Museum South Africa,
Cleveland Museum of Natural History, Royal Museum of
Central Africa Tervuren for access to comparative
materials; L. Spurlock and P. L. Reno for assistance with
dissections and histology preparations; D. Kubo and
H. Fukase for assistance in computed tomography (CT)
scanning; C. Hernandez for calculation of section modulus
data for Mohr’s Circle analyses; M. Brunet, C. V. Ward, and
J. DeSilva for cooperation with comparative data; R. Meindl
for statistical advice and assistance; J. DeSilva, P. L. Reno,
M. A. Serrat, M. A. McCollum, M. Selby, A. Ruth, L. Jellema,
S. W. Simpson, and B. A. Rosenman for aid in
data collection and exceptionally helpful discussions;
and H. Gilbert, J. Carlson, and K. Brudvik for figure
preparation.
Supporting Online Material
www.sciencemag.org/cgi/content/full/326/5949/72/DC1
Figs. S1 to S14
Tables S1 and S2
References
4 May 2009; accepted 14 August 2009
10.1126/science.1175832
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org 72e8
RESEARCH ARTICLES
Research Articles

92. 92 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
TheGreatDivides:Ardipithecusramidus
RevealsthePostcraniaofOurLast
CommonAncestorswithAfricanApes
C. Owen Lovejoy,
1
*Gen Suwa,
2
*Scott W. Simpson,
3
Jay H. Matternes,
4
Tim D. White
5
Genomic comparisons have established the chimpanzee and bonobo as our closest living relatives.
However, the intricacies of gene regulation and expression caution against the use of these extant
apes in deducing the anatomical structure of the last common ancestor that we shared with them.
Evidence for this structure must therefore be sought from the fossil record. Until now, that record
has provided few relevant data because available fossils were too recent or too incomplete.
Evidence fromArdipithecus ramidusnow suggests that the last common ancestor lacked the hand,
foot, pelvic, vertebral, and limb structures and proportions specialized for suspension, vertical
climbing, and knuckle-walking among extant African apes. If this hypothesis is correct, each extant
African ape genus must have independently acquired these specializations from more generalized
ancestors who still practiced careful arboreal climbing and bridging. African apes and hominids
acquired advanced orthogrady in parallel. Hominoid spinal invagination is an embryogenetic
mechanism that reoriented the shoulder girdle more laterally. It was unaccompanied by
substantial lumbar spine abbreviation, an adaptation restricted to vertical climbing and/or
suspension. The specialized locomotor anatomies and behaviors of chimpanzees and gorillas
therefore constitute poor models for the origin and evolution of human bipedality.
T
homas Huxley publishedEvidence as to
Man’s Place in Nature(1) only 4 years
after Darwin’sOn the Origin of Species. Its
frontispiece featured a human skeleton and four
suspensory adapted apes, each posed upright and
each obviously more human-like than any pro-
nograde Old World monkey. By century’s end,
Keith was enumerating a cornucopia of characters
in support of a brachiationist human past (2).
Even our pericardial-diaphragmatic fusion, hepat-
ic bare area, and colic mesenteries were inter-
preted as adaptations to orthogrady, evolved to
tame a flailing gut in the arboreal canopy. Bi-
pedality was simply habitual suspension brought
to Earth (3). The“suspensory paradigm”for early
hominid evolution was born.
Challenges, however, were mounted. Straus
enumerated disconcertingly primitive human
features in“The Riddle of Man’s Ancestry”(4),
and Schultz doubted that brachiation“…opened
the way automatically for the erect posture of
modern man”[(5), pp. 356– 357]. Although with-
drawal of the ulna from its primitive pisotriqetral
recess was thought to be the sine qua non of sus-
pension (6), a functional equivalent was dis-
covered to have evolved in parallel in the wrists
of never-suspensory lorisines (7). African ape
knuckle-walking (8), considered by many too
bizarre to have evolved independently inGorilla
andPan, came to be viewed in light of emergent
molecular phylogenetics (9) as a natural succes-
sor of suspensory locomotion—and by some as
the almost-certain default engine of bipedality
(10).
A flood of morphometric analyses appeared
to confirm arguments for knuckle-walking hom-
inid ancestors [reviewed in (11)], even though
hints of the behavior were also seen in captive
orangutans (12). Knuckle-walking was surmised
to be a natural consequence of irreversible mod-
ifications of the forelimb skeleton to facilitate
advanced suspension and vertical climbing (11).
It was thereby hypothesized to be an adaptive
signal of the first two phases of a determinis-
tic succession leading to bipedality: advanced
suspension/vertical climbing→terrestriality/
knuckle-walking→bipedality.
A compendium of observations of chim-
panzees and bonobos performing upright stance
and locomotion followed. Accumulating molec-
ular biology propelled this troglodytian para-
digm (conceived as a natural succession to its
older, suspensory counterpart) to near-consensus.
Chimpanzee-human protein homologies and
DNA base sequence comparisons (9,13–16)
establishedHomoandPanas likely sister clades
[today further confirmed by comparative ge-
nomics (17,18)]. The only question remaining
seemed to be whether the bonobo or chimpanzee
represented the best living proxy for the last
common ancestor (19–22).
The Chimpanzee model andAustralopithecus.
The discovery and recognition of the then-
primitiveAustralopithecus afarensisduring the
1970s (23) pushed the hominid record back to
3.7 million years ago (Ma). Although its post-
cranium was recognized to harbor unusually
sophisticated adaptations to bipedality [reviewed
in (24)], a feature confirmed by human-like
footprints at Laetoli (25,26), many interpreted
these fossils to represent the closing argument
for the troglodytian paradigm [see, e.g., (27 )].
Only the recovery of earlier, chimpanzee-like
fossils from the Late Miocene seemed necessary
to complete this scenario [even though newer
Australopithecusfossils have led at least one
discoverer to doubt a chimpanzee-like ancestry
(28)]. Until now, the few available fossils of ap-
propriate antiquity have remained largely unin-
formative (29 –31).
TheArdipithecus ramidusfossils from 4.4
Ma Ethiopia are obviously not old enough to
represent the chimpanzee/human last common
ancestor (CLCA; the older common ancestor of
hominids and bothGorillaandPanis hereafter
the GLCA). However, their morphology differs
substantially from that ofAustralopithecus. The
Ar. ramidusfossils therefore provide novel in-
sights into the anatomical structure of our elusive
common ancestors with the African apes. For
that reason, and because of its phylogenetic posi-
tion as the sister taxon of later hominids (32), this
species now provides opportunities to examine
both the suspensory and troglodytian paradigms
with greater clarity than has previously been
possible. Here we first provide evidence of limb
proportions, long considered to bear directly on
such issues, and then review key aspects of the
entireAr. ramiduspostcranium. Comparing the
basic proportions and postcranial anatomy of
Ar. ramidus(Fig. 1) with those of apes enables
us to propose the most probable anatomies of the
last common ancestors ofGorilla,Pan, and the
earliest hominids. Much of the relevant informa-
tion onAr. ramidusis based on the partial
skeleton from Aramis (32).
Body mass.The geometric means of several
metrics of the capitate and talus are strongly
related to body mass in extant primates (correla-
tion coefficientr= 0.97; fig. S1), and can be
used to estimate body mass inARA-VP-6/500,
as well as inA.L. 288-1. Restricting the sample
to large-bodied female hominoids predicts that
ARA-VP-6/500had a mass of about 51 kg. The
metrics forA.L. 288-1fall below those of all
extant hominoids. We therefore used the female
anthropoid regression to estimate the body mass
ofA.L. 288-1(26 kg), which is consistent with
previous estimates (33) (table S1). Based on
several shared metrics,ARA-VP-7/2, a partial
forelimb skeleton (32), was slightly smaller
thanARA-VP-6/500[supporting online material
(SOM) Text S1].
Given the apparent minimum body size di-
morphism ofAr. ramidus(32,34), the predicted
Ardipithecus ramidus
1
Department of Anthropology, School of Biomedical Sciences,
Kent State University, Kent, OH 44242–0001, USA.
2
The
University Museum, the University of Tokyo, Hongo, Bunkyo-
ku, Tokyo 113-0033, Japan.
3
Department of Anatomy, Case
Western Reserve University School of Medicine, Cleveland, OH
44106–4930, USA.
4
4328 Ashford Lane, Fairfax, VA 22032,
USA.
5
Human Evolution Research Center, and Department of
Integrative Biology, 3101 Valley Life Sciences, University of
California at Berkeley, Berkeley, CA 94720, USA.
*To whom correspondence should be addressed. E-mail:
[email protected] (C.O.L.); [email protected] (G.S.)
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org100
Ardipithecus ramidus

93. 93www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
body mass ofARA-VP-6/500serves as a rea-
sonable estimate for the general body mass ofAr.
ramidus. AlthoughARA-VP-6/500was one of
the larger individuals of the Aramis sample (32),
it was probably more representative of its species
than wasA.L. 288-1[the latter clearly lies at the
lower end of theAu. afarensisspecies range
based on larger samples (35 )]. Unfortunately,
ARA-VP-6/500tells us little about the body
mass of the CLCA and GLCA because these
predateAr. ramidusby wide margins and may
have still been primarily arboreal. The limited
available (mostly dental and cranio-mandibular)
samples indicate that the size of Late Miocene
hominids (29 –31) was similar to that ofAr.
ramidus(34), and estimated body weight for the
6 MaOrrorinfemoral remains is 30 to 50 kg (36).
Although body mass in early Miocene forms
appears to have varied greatly (37,38), it is likely
that the CLCA and GLCA were either equal to or
smaller thanAr. ramidus, and possibly even sub-
stantially so. Only additional fossils can resolve
this issue.
Limb segment proportions.Radial, ulnar,
and tibial lengths can be accurately deter-
mined forARA-VP-6/500(SOM Text S1). The
specimen’s radius/tibia ratio (0.95; fig. S2) is
similar to those of generalized above-branch
quadrupeds such as the Old World monkey
Macaca(0.90 to 0.94; table S2) and the Mio-
cene apeProconsul heseloni(0.88 inKNM-RU
2036)(38). The ratio is unlike that of African
apes (P. troglodytes,1.11T0.04;Gorilla, 1.13T
0.02) (39) and is, remarkably, 17 standard devi-
ations below that ofPongo(1.47T0.03).
TheArdipithecusskeleton’s nearly intact tibia
allows estimation of femoral length because the
crural index (CI: tibia length/femur length × 100)
is highly conserved in African apes and humans
(5,40) (81 to 84; SOM Text S1). Tibial length in
A.L. 288-1can likewise be estimated from its
effectively complete femur. Although no humer-
us was recovered forARA-VP-6/500, one be-
longing toARA-VP-7/2is almost complete and
can be used to estimate humerus length inARA-
VP-6/500by simple proportion of shared ele-
ments (SOM Text S1). TheA.L. 288-1humerus
is intact, and its radius length was previously es-
timated by regression (41). These data allow cal-
culation of the more familiar intermembral index
(IMI; forelimb length/hindlimb length × 100).
The IMIs of both specimens resemble those of
Proconsuland Old World monkeys (table S3).
ARA-VP-6/500also allows interpolation of
other key limb proportions. The brachial indices
(BI: radius length/humerus length × 100) of
Proconsul,Equatorius,A.L. 288-1, andARA-
VP-6/500are each within the observed range of
Pan(fig. S3). It is therefore likely that the BI
has remained largely unmodified since the GLCA,
especially in light of the relationship of radius
length to estimated body mass (fig. S4). In con-
trast, the BIs ofHomoandGorillaare both de-
rived, albeit by obviously different routes (fig. S3).
Humans have greatly shortened radii in conjunc-
tion with their novel antebrachial/manual pro-
portions for grasping and manipulation [(41,42)
and see below];Gorillaappears to have ex-
perienced both humeral elongation and possibly
slight radial shortening (figs. S4 and S5), most
likely to reduce joint stresses at the elbow im-
posed by the immense mass of adult males. The
BIs ofPanandAr. ramidusare similar (fig. S3),
butPanexhibits a much higher IMI (table S3).
Therefore, bothPanandGorillahave undergone
forelimb elongation and hindlimb reduction since
the GLCA (table S2 and figs. S4 to S6). The IMIs
of hominids appear to have remained primitive
until 2.5 Ma (41 ,43). The relatively high BI of
Pongoreflects its entirely different evolutionary
history.
Manual anatomy and proportions.Compared
to estimated body size, the manual phalanges
ofAr. ramidusandGorillaare long relative to
those of the Miocene apeProconsul(fig. S7).
They are relatively even more elongate inPan,
but dramatically abbreviated inHomo. These
conclusions are supported by similar calcula-
tions using the means of observed body mass
(table S3). There is no evidence that the manual
phalanges ofAu. afarensiswere elongated rela-
tive to those ofAr. ramidus.
In contrast to their manual phalanges, the
posterior (medial) metacarpals 2 to 5 (Mc2-5) of
ProconsulandARA-VP-6/500are substantially
Fig. 1.Reconstructed frontal and lateral views of the skeleton ofARA-VP-6/500. Major long-bone lengths
were determined directly from preserved skeletal elements (radius, tibia), by crural index (femur), by
regression from adjacent elements (ulna), or by ratio and regression (humerus) from a marginally smaller
forelimb skeleton (ARA-VP-7/2) via ratios of commonly preserved elements (SOM Text S1). All manual and
pedal elements were drawn directly from casts. Pelvis was traced from frontal and lateral computer
tomography (CT) scans of reconstructed pelvis (59). Vertebral column and thorax were based on six
lumbars, 12 thoracics, and four sacrals (58). No attempt has been made to indicate failure of lateral fusion
between the transverse processes of S4 and S5 [i.e., failure of complete closure of either of the fourth
sacral foramina (the state preserved in bothA.L. 288-1andKNM-WT 15000)]. Such four-segment sacra
may have been modal inAr. ramidus, but the five-segment form shown here was also a likely variant of
high frequency [for discussion, see (59)]. Pectoral girdle and thorax were based on preserved portions of
clavicle, first rib, and common elements known inAu. afarensis. Skull and mandible were based on models
generated by restoration of cranium using both CT/rapid prototyping and“cast-element-assembly”methods
(79). Reconstruction by J. H. Matternes was based on full-scale (life-size) architectural drawings circulated
among authors for multiple inspections and comments. Stature (bipedal) is estimated at 117 to 124 cm and
body weight at 51 kg. [Illustrations: Copyright 2009, J. H. Matternes]
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
101
RESEARCH ARTICLES
Research Articles

94. 94 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
shorter than are those of any extant ape (figs. S8
and S9). Viewed in the context of relative limb
length patterns (see above), as well as the ana-
tomical details of the hand (44), the short Mcs
ofAr. ramidusstrongly suggest thatPanand
Gorillaindependently acquired elongate Mcs as
a part of an adaptation to vertical climbing and
suspensory locomotion. Elongation of Mc2-5 in
African apes demanded heightened resistance to
torsion and consequent fixation of the carpo-
metacarpal joints within the central joint com-
plex (CJC) (44).
The retention of the primitively short Mcs in
Ar. ramidussuggests that the GLCA/CLCA also
did not have elongate Mcs, and engaged in a
form of above-branch quadrupedal locomotion
involving deliberate bridging and careful climb-
ing. We hypothesize that this was retained from
Middle Miocene precursors of the GLCA. A re-
tained short metacarpus would optimize palmar
conformity to substrates, an adaptation later aban-
doned by extant African apes.
The thumb metacarpal ofARA-VP-6/500
was more aptly proportioned for manual grasp-
ing than are those of extant apes (figs. S10 and
S11) (44). In extant apes, elongation of the pos-
terior (medial) metacarpus may have been
achieved by increased expression ofHoxd11or
one of its targets, which do not affect the first ray
(SOM Text S2) (42,45). However, the Mc1 of
apes does seem moderately less robust than that
ofAr. ramidus, and its soft tissues have under-
gone substantial involution (4,42). This suggests
that some degree of down-regulation ofHoxd13
may have been responsible for elongation of the
posterior (medial) metacarpus.
Ar. ramidusgreatly illuminates the natural
history of the thumb in higher primates. Its ro-
busticity in hominids, while certainly enhanced
during the past 3 million years, is nevertheless
at least partially primitive. In contrast, in taxa
adapted to vertical climbing and suspension,
lengthening of the palm has become so domi-
nant as to eclipse some of the thumb’s function,
a condition that has reached its apogee inAteles
and, to a lesser extent, large-bodied extant apes.
These findings strongly suggest that the target of
recently discovered major cis-regulatory modifica-
tion of gene expression in the first ray (46) was
not manual but pedal—it is the human hallux, not
our largely primitive pollex, that is highly derived
(47).
Additional relevant hand anatomy leads to
the same conclusions.Ar. ramidusis the only
hominid fossil thus far recovered with a metacar-
pal head reminiscent of the metacarpophalangeal
(MP) joint structure seen in many Miocene hom-
inoids [such asEquatorius,Proconsul,Dryopi-
thecus, andPierolapithecus(48)]. The collateral
ligament facets in these taxa colocate with deep
symmetric invaginations of the metacarpal
head’s dorsum. This morphology is typical of
Old World monkeys and is thereby associable
with substantial dorsiflexion of the MP joint, an
obvious manifestation of their palmigrady. The
trait is only moderately expressed inOreopithe-
cus. Modern human and orangutan MP joints are
substantially less constricted, and neither taxon
exhibits appreciable locomotor-related MP dor-
siflexion.
Constricted metacarpal head morphology ap-
pears to be primitive because it is still partially
present inAr. ramidus, albeit substantially reduced
compared to early Miocene hominoids and Old
World monkeys. Its retention suggests moderately
frequent MP dorsiflexion, a finding consistent
with the remarkable adaptations to palmigrady
seen in theAr. ramiduswrist [see below and (44)].
The metacarpal heads of knuckle-walking
apes are also somewhat constricted by their
collateral facets, but are heavily flattened and
broadened to withstand excessive compression
during dorsiflexion. Constriction by their collat-
eral ligament facets is therefore only minimal.
Moreover, the origins of their collateral ligaments
have been substantially expanded volarly, pre-
sumably because such positioning improves their
capacity to restrict abduction or adduction during
MP dorsiflexion imposed by knuckle-walking.
Joint flattening enhances cartilage contact and is
likely at least partially a cartilage-modeling trait
[cartilage modeling; Type 4 (49)].
Loss of MP dorsiflexion inPongois readily
explicable by its extreme metacarpal and phalan-
geal elongation and curvature. These can safely
be presumed to have eliminated any appreciable
functional MP dorsiflexion. Modern humans lack
any dorsiflexion because our hand plays no im-
portant role in locomotion. The trait is also absent
inAu. afarensis, suggesting that either its hand
no longer played any role in locomotion, or that
such use no longer included an MP dorsiflexive
component of palmigrady. The former seems far
more likely, given the paramount adaptations to
bipedality in the species’lower limb (24,50,51).
The primitive metacarpal head morphology
within the overall primitive hand anatomy (44 )
ofAr. ramiduscarries obvious implications for
reconstruction of GLCA/CLCA locomotion.
The unique combination of marked midcarpal
mobility, ulnar withdrawal, and moderate MP
dorsiflexion inAr. ramidus, probably mostly prim-
itive retentions, implies that the GLCA/CLCA
locomotor pattern was also characterized by some
form of arboreal palmigrade quadrupedality, un-
like that in any extant descendant great ape.
Finally, it is clear now that phalangeal length
ofAr. ramidusis not related to suspensory loco-
motion, but instead reflects a more general grasp-
ing adaptation. This renders phalangeal length
moot regarding the hypothesis that manual (or
even pedal) phalangeal lengths are an active sig-
nal of suspensory locomotion inAu. afarensis
[contra (52,53)]. It is more probable that selec-
tion had not reduced their length in the younger
species, and that such reduction did not occur
until selection for tool-making became more in-
tense later in the Pliocene (43,54).
Pedal proportions.Pedal phalangeal evolu-
tion appears to have closely paralleled its manual
counterpart in each clade (compare figs. S7 and
S12). However, pedal phalanges of African apes
and hominids appear to have been substantially
abbreviated, rather than elongated. Functional
demands of terrestrial locomotion, perhaps sim-
ilar to those acting on papionins (which also ex-
hibit pedal phalangeal shortening), are a probable
explanation.Pongorepresents a marked contrast,
with substantial pedal phalangeal elongation. It is
thus reasonable to infer that the GLCA/CLCA’s
pedal phalanges were longer than those of the
partially terrestrial extant African apes andAr.
ramidus.
The metatarsus ofAr. ramidus, chimpanzees,
and gorillas presents a striking contrast to their
metacarpus. Like the foot phalanges, the meta-
tarsals also appear to have been universally
shortened in all hominoids subsequent to
Proconsul(figs. S13 and S14) (47). The basis
of this universal shortening, however, is some-
what unclear, because tarsal evolution contrasts
dramatically in hominids and African apes. The
modern ape foot has obviously experienced func-
tional reorganization into a more hand-like grasp-
ing organ. TheAr. ramidusfoot did not. This
suggests that substantial elements of a more lever-
based, propulsive structure seen in taxa such as
Proconsuland Old World Monkeys [robust
plantar aponeurosis; retained quadratus plantae;
robust peroneal complex (47)] were preserved in
the GLCA/CLCA. These structures were sacri-
ficed in both African ape clades to enhance pedal
grasping for vertical climbing (55,56). The mod-
erate shortening of the metatarsus inAr. ramidus
and both African apes may therefore simply re-
flect negative allometry of metatarsal (Mt) lengths
with an increase in body size. The human foot has
been lengthened primarily by tarsal elongation
(5,47), presumably because of the likely high
failure rate of metatarsal shafts during forceful
fulcrumation.
In summary, a comparison of the pedal pro-
portions ofAr. ramidusand the extant African
apes suggests that the GLCA/CLCA hindlimb
remained dominant for body mass support dur-
ing bridging and arboreal clambering, to the ex-
tent that it later proved permissive to bipedality
in transitionally terrestrial hominids.
Trunk structure.Knowledge of the role of
selector genes in early vertebral column forma-
tion [especially the role of theHoxcode on
column differentiation (57,58)] has advanced
our ability to interpret the vertebral formulae of
extant hominoids. It now appears that the modal
number of lumbar vertebrae inAustralopithecus
was six, and that a four-segment sacrum was
also probably common (57,58). This axial for-
mula is unlike that of any extant ape. Compar-
ison of the axial columns of extant species
further indicates that postoccipital somite num-
ber in the GLCA/CLCA was probably either 33
or 34, and that lumbar column reduction oc-
curred independently in chimpanzees, bonobos,
gorillas, and hominids. This probably resulted
from either transformation of vertebral identities,
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
102
Ardipithecus ramidus
Ardipithecus ramidus

95. 95www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
or a combination of such transformation and
reduction in the number of somites contributing
to the lumbosacral region (fig. S15). The most
likely vertebral patterns forAr. ramidusare there-
fore those also inferred for the GLCA/CLCA and
Australopithecus.
Pelvic structure indicates thatAr. ramidus
retained a primitive spine. Its iliac and acetabular
regions establish not only that it was habitually
bipedal when terrestrial, but also that this was
achieved by combining situational anterior pelvic
tilt to accentuate substantial lordosis during up-
right walking (59). Such rotation placed the still
partially primitive anterior gluteal musculature
into a position of functional abduction for single
support stabilization. In contrast toAr. ramidus,
Au. afarensisis known to have exhibited highly
evolved mechanisms of hip abduction, confirmed
by the distinctly stereotypic trabecular profile of
its femoral neck (24).
TheAr. ramiduspelvis retained other ele-
ments in common with extant African apes (and
presumably the GLCA/CLCA). These include a
long, expansive and rugose ischial region and
shorter pubic rami (but not a long pubic corpus)
(59). The species’highly flexible lower lumbar
column, coupled with its narrower interacetab-
ular distance, still must have provided a mod-
erately reflexive hindlimb for arboreal climbing.
Not until hominids became habitually terrestrial
bipeds with broad interacetabular distances, re-
duced and angulated ischial tuberosities (possibly
indicating hamstring deceleration of the hindlimb
at heel strike), and extremely shortened, flared,
and broadened ilia did they then exchange such
flexibility for the much more rigid constraints of
lower-limb stabilization that characterizeAustra-
lopithecus(50,51).
The combined pelvic and vertebral data im-
ply that the morphological elements of extant
great apes emerged separately rather than in
concert. Vertebral column invagination and its
associated gracilization of the retroauricular
pelvic space preceded specialized iliac modifi-
cation and the radical lumbar column shortening
seen in the African apes (58 ). TheARA-VP-6/
500pelvis shows that hominid ilia shortened
and broadened to establish permanent lumbar
lordosis. African ape ilia were instead modified
to increase abdominal stiffness. The posterior
pelvic changes and pronounced lordosis in hom-
inids subsequently promoted even more dramat-
ic vertebral column invagination (60 ). This trend
is eventually reflected in more dorsally oriented
transverse processes of hominid thoracic verte-
brae compared to those of apes (60). In extant
apes, vertebral column invagination and shorten-
ing were acquired both independently and non-
contemporaneously, the first being a deeply
rooted embryogenetic mechanism that postero-
lateralized the pectoral girdle for a more lateral-
facing glenoid; the second, an independent
means of increasing abdominal rigidity. We hy-
pothesize that hominids never participated in the
second (SOM Text S3), having rather evolved
from a careful climber in which deliberate
bridging placed no undue stress on the lower
spine. Not until the ancestors of African apes
embarked (separately) on their adaptations to
vertical climbing and suspension did the lumbar
spine undergo its dramatic reduction in length.
The last common ancestors.Integration of
the data and observations reviewed above al-
lows us to hypothesize about the postcranial
adaptations and locomotion of the GLCA and
CLCA. The extensive array of highly distinctive
specializations seen in modernGorillaandPan
(in part shared withPongo) indicates that these
are derived features most likely related to ver-
tical climbing and suspension.
Not only doesAr. ramidusfail to exhibit
these specialized modifications, it exhibits others
(e.g., a palmar position of the capitate head
that facilitates extreme dorsiflexion of the mid-
carpal joint rather than its limitation; a robust os
peroneum complex limiting plantar conformity
to substrates rather than its facilitation) that are
effectively their functional opposites. The expres-
sion of some of these characters (e.g., capitate
head position) is even more extreme than it is in
either the Miocene apes precedingArdipithecus
or inAustralopithecusthat follows. It is therefore
highly unlikely thatAr. ramidusdescended from
aPan/Gorilla-like ancestor and then (re)evolved
such extreme characters. Conversely, some other
detailed differences inPanandGorillastructure
[e.g., scapular form (61), iliac immobilization of
lumbar vertebrae (58), appearance of a prepollex
(62)] suggest that each of these ape clades
independently acquired their anatomical adapta-
tions to vertical climbing and/or suspension.
Therefore, we hypothesize thatAr. ramidus
retains much of the ancestral GLCA and CLCA
character states, i.e., those that relate to above-
branch quadrupedality. In particular, contra
GorillaandPan, the GLCA carpometacarpal,
midcarpal, radiocarpal, and ulnotrochlear joints
must have lacked notable adaptations to suspen-
sion and/or vertical climbing (44). The GLCA
foot seems to have been only partially modified
for manual-like grasping. Its hindlimb remained
fully propulsive at its midtarsal and tarsometa-
tarsal joints (47). Although its shoulder joint must
have been fully lateralized, its lumbar column
nevertheless was still long (58) (fig. S15). Its
limb proportions were still primitive (see earlier).
If body size was as large as inAr. ramidus, it may
have been too large for habitual, unrestricted
above-branch quadrupedality, but this remains
uncertain. Assuming considerable reliance on
arboreal subsistence, it is likely that body mass
did not exceed 35 to 60 kg [i.e., combined
probable range ofAr. ramidusand 6 MaOrrorin
(36)].
The GLCA picture that emerges, therefore,
is one of generalized, deliberate bridging with
quadrupedal palmigrady and preference for large-
diameter substrates. This may have involved
either suspension or vertical climbing, but with-
out sufficient frequency to elicit morphological
adaptations specific to these behaviors. It is likely
that these hominoids ranged mostly in the lower
canopy, and perhaps were even partially terres-
trial. However, their mode of terrestrial locomo-
tion remains unknown.
The GLCA therefore represents a founda-
tion for two adaptive paths.GorillaandPan
independently specialized for both suspension
and vertical climbing (and eventually knuckle-
walking). Gorillas might have acquired larger
body size in relation to mixing higher-canopy
frugivory with a more terrestrial herbaceous or
folivorous dietary component. Lacking defini-
tive fossil evidence, it is currently impossible to
determine when the large body mass ofGorilla
evolved, but it probably occurred in concert
with its more herbaceous diet. The 10 Ma
Chororapithecus, which shows incipient signs of
Gorilla-like molar morphology (63), may be an
early representative of theGorillaclade. If so,
then this clade’s shift toward increased body
mass and terrestriality must have occurred early
in its phyletic history.
The other adaptive pathway retained palmar
flexibility, with a short metacarpus that lacked
notable syndesmotic restriction. This was com-
bined with retention of an essentially rigid mid-
tarsal joint that was insufficiently flexible to
perform vertical climbing (55 ,56), but was fully
satisfactory for less specialized careful climbing,
clambering, and bridging. This is the hypothe-
sized structure of the CLCA, from whichPan
would have evolved a greater reliance on ver-
tical climbing and suspension than occurred in
theGorillaclade, never reaching as large a body
size.
In contrast toPan, the forebears ofAr. ramidus
early in the hominid clade must have relied
increasingly on lower arboreal resources and
terrestrial zones, without being dependent on
higher-canopy resources (such as ripe fruits).
From the comparative evidence now available
fromAr. ramidusandPandental anatomy and
isotopes, we posit that the chimpanzee clade in-
creasingly developed a preference for (or depen-
dency on) ripe fruit frugivory, whereas hominids
retained a more primitive dental complex
adequate for the range of transitional arboreal/
terrestrial resources (34).
The likely K-selected demographic adap-
tation of all hominoids in a setting of almost
certain competition with the surging Old World
monkey radiation would have been a major
factor (64,65) driving such very different evo-
lutionary trajectories of early African apes and
hominids. The earliest fossil evidence for cer-
copithecid radiation (an early colobine) is now
close to 10 Ma (66). A much better record of both
fossil hominoids and cercopithecids from the late
Middle to early Late Miocene is needed to clar-
ify these suggested patterns of ape-cercopithecid
evolution.
Orthogrady, suspension, knuckle-walking,
and bipedality.Ar. ramidusaffords new in-
sights into ape and hominid bauplan evolution
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
103
RESEARCH ARTICLES
Research Articles

96. 96 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
(Fig. 2 and Table 1). The most fundamental is
the clear demonstration that the GLCA lacked
the suspensory adaptations long recognized to
be common to all extant apes.
The chimpanzee and gorilla clades each in-
dependently increased their reliance on higher-
canopy resources, and modified characters
originally associable with advanced bridging to
those more useful in vertical climbing and suspen-
sion. These include an elongated posterior (medial)
metacarpus, broadened radiocarpal joint with
reduced midcarpal mobility, syndesmotically
and morphologically buttressed carpometacarpal
joints, expanded long antebrachial flexor ten-
dons, a redistributed long pollical flexor tendon
(to the elongated second ray), a modified enthesis
for the deltopectoral complex, a retroflexed troch-
lear notch, elongate forelimbs (44), abbreviated
hindlimbs, elimination of the os peroneum com-
plex (47), lumbar column reduction (58), and
iliac fixation of remaining lumbars [acquired by
iliac elongation and sacral narrowing (58,59)].
Viewed from the perspective ofAr. ramidus, all
of these can now be visualized as having been
acquired independently. All represent adaptations
related directly to suspension, vertical climbing,
and/or knuckle-walking.
In African apes, terrestrial travel may have be-
come the primary means of overcoming expand-
ing canopy gaps. A return to partial terrestrial
pronogrady would have necessitated compensa-
tory energy-absorptive mechanisms to ameliorate
ground reaction in heavily modified forelimbs
(which would have suffered an increased risk of
injury). Knuckle-walking filled this role because
it promotes eccentric contraction and/or energy
dissipation (and storage) in the wrist and digital
flexors (especially their connective tissue com-
ponents) during impact loading in a completely
extended forelimb, without compromising the
animal’s newly acquired adaptations to either
suspension or vertical climbing (44). More elab-
orate mechanisms of negotiating gaps in trees
(67) evolved separately in orangutans, in which
both manual and pedal rays radically elongated,
possibly to more effectively gather and assem-
ble multiple lianas necessary to negotiate such
gaps.
Thus,Ar. ramidusallows us to infer that
GLCA anatomy was exaptive for suspension
and vertical climbing. Early hominids continued
to practice palmigrade, above-branch quadrupe-
dal clambering. Ulnar retraction, common to
bothPanandGorilla, therefore appears to have
emerged for forelimb flexibility as part of ar-
boreal clambering and bridging before the GLCA
(7), and not as an adaptation to suspension [as
has been argued (6)]. Initialized in forms like
Proconsul, the combination of enhanced fore-
limb flexibility and hindlimb propulsive dom-
inance, without anatomical modifications for
forelimb suspension, may have reached an apo-
gee in the GLCA.
These observations also conform to evi-
dence available from the steadily increasing
Miocene hominoid fossil record. European near-
contemporaries of the African CLCA to GLCA
exhibited only various degrees of adaptation to
suspension, suggesting a separate Miocene trend
toward increasing forelimb dominance. At 12
Ma,Pierolapithecushad ulnar withdrawal and
partial spinal invagination (68), but likely re-
tained a long lumbar spine. Its hand lacked the
degree of metacarpal or phalangeal elongation
seen in extant apes. More recentDryopithecus,
which did display both an African ape-like CJC
(44) and elongate metacarpals relative to body
size, nevertheless retained palmigrady (68,69).
Suspensory locomotion was therefore likely inde-
pendently derived (minimally) inDryopithecus,
Pan, andGorilla(and certainly so inPongo).
Hypotheses that hominid ancestry included
suspensory locomotion and vertical climbing
(52,53), as projected from electromyographic
and kinematic analyses of living ape behavior,
are now highly unlikely.
From their beginning, accounts of human
evolution relied on postural similarities between
living humans and apes. The inference that ha-
bitual orthogrady was central to the origin of
bipedality has been taken as largely self-evident
(2,70). Until now, no fossils of sufficient age
and anatomical representation have been availa-
ble for seriously testing these presumptions.Ar.
ramidusrequires comprehensive revision of such
entrenched, traditional canons. Its anatomy
makes clear that advanced orthogrady evolved
in parallel in hominids and apes, just as it has in
an array of other primates, both living and ex-
tinct [including prosimians such asPropithecus
andMegaladapis, some ceboids, gibbons, and
a variety of Miocene hominoids, especially
Nacholapithecus(71), andOreopithecus(72)].
The long-held view that dorsal transposition of
the lumbar transverse processes onto their pedi-
cles implies orthogrady is now falsified, because
Ar. ramidusestablishes that such relocation is a
direct correlate of ventral invagination of the
entire spinal column within a context of above-
branch quadrupedal palmigrady that established
increased shoulder mobility for bridging and
clambering (SOM Text S3).
In hominids, from an above-branch quadru-
pedal ancestry, advanced orthogrady was the in-
dependent consequence of terrestrial bipedality
made possible by a mobile lumbar spine and
largely primitive limbs. It is sobering to consider
one profound implication— if emergent homi-
nids had actually become as adapted to suspen-
sion or vertical climbing as are living apes,
neither bipedality nor its social correlates would
likely have evolved. It is therefore ironic that
these locomotor modes have played so promi-
nent a role in explanations of bipedality. In
retrospect, it seems clear that they would instead
have likely prevented it (SOM Text S3).
Conclusions.Ar. ramidusimplies that Afri-
can apes are adaptive cul-de-sacs rather than
stages in human emergence. It also reveals an
unanticipated and distinct locomotor bauplan
for our last common ancestors with African apes,
one based on careful climbing unpreserved in
any extant form. Elaborate morphometric statis-
tical procedures were the culmination of a 20
th
-
century trend toward objectivity, in which metrics
came to be regarded as more informative than
careful comparative anatomy—a trend accom-
panied by too many presumptions and too few
Pongo pygmaeusGorilla gorillaPan troglodytesPan paniscusArdipithecus ramidus*Australopithecus afarensis*Homo sapiens
Fig. 2.Branching diagram to illustrate cladistic relationships of extant hominoids.
Branching order among the extant forms shown here is well established by mo-
lecular evidence. The two fossil forms are possible phyletic ancestors of the human
clade, but are shown here in a sister relationship to the extant forms. Circled
numbers indicate evolutionary derivations, itemized in Table 1, hypothesized to
have occurred on each lineage. [Illustrations: Copyright 2009, J. H. Matternes]
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
104
Ardipithecus ramidus
Ardipithecus ramidus

97. 97www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Table 1.Evolutionary derivations of various hominoid clades with fossil and modern representation. Numbers refer to circles on Fig. 2.
1. Basal node.An inferred generalized ancestor of the great ape clade, which lived probably more than 18 Ma. We infer this primate to have been an above-branch palmigrade,
plantigrade quadruped, with generalized limb proportions, an anteriorly oriented pectoral girdle, and long lumbar vertebral column with transverse processes located ventrally on their
bodies. It would have also been characterized by an extensive postauricular iliac region for a massive erector spinae, a long olecranon process, an anteriorly oriented trochlea, a capitate
head located mid-body, and a primitive central joint complex in the wrist. It would have featured a full wrist mortise with pisotriquetral contact anda moderately long midtarsus for
fulcrumation on its metatarsal heads. It was presumably tailless (80).
2.Orangutanclade
.Dramatic elongation of entire forelimb, posterior (medial) metacarpus and phalanges, extreme elongation of posterior (lateral) metatarsus and phalanges but
abbreviation of thigh and leg, partial involution of first pedal and manual rays. Abbreviation of lumbar vertebral column (average four elements) bymeans of sacralization of lumbar
vertebrae, reduction in axial length by two segments, and craniocaudal shortening of lumbar centra (58). Entrapment of caudal-most lumbars by articulation with variable cranial
extension of ilia and reduction in breadth of sacral alae. Invagination of spine with posterolateralization of pectoral girdle and reduction of deltopectoral crest. Retroflexion of trochlear
notch, extreme abbreviation of olecranon process, and elevation of lateral margin of trochlea. Ulnar withdrawal with elimination of wrist mortise.Modification of central joint complex
for torsional resistance during suspension. Frequent postnatal fusion of os centrale and scaphoid.
3. Extant African ape and hominid clade (GLCA).Minor abbreviation of midtarsal length, elongation of manual phalanges, and shortening of posterior (lateral) metatarsus.
Invagination of spine with posterolateralization of pectoral girdle, mediolateral proportionality shift of sacroiliac region, craniocaudally shortened vertebral centra, and relocation of
lumbar transverse processes to corporopedicular junction or onto pedicle. Abbreviation of olecranon and elevation of lateral margin of trochlea. Ulnar withdrawal with elimination of
wrist mortise (i.e., loss of pisotriquetral contact) and deepening of carpal tunnel. Fusion of os centrale to scaphoid.
4.Gorillaclade.Elongation of forelimb (by disproportionate elongation of humerus) and abbreviation of hindlimb (global change in limb proportions), moderate elongation of posterior
(medial) metacarpus, moderate shortening of manual phalanges. Abbreviation of lumbar vertebral column (average 3.5 elements) by means of sacralization of lumbars and reduction in axial
length by one segment (58). Entrapment of most caudal lumbars by articulation with cranially extended ilia and reduction in breadth of sacral alae. Moderate increase in cranial orientation
of scapular spine and glenoid plane, reduction of deltopectoral crest. Retroflexion of ulnar trochlear notch with attendant abbreviation of olecranon process, expansion of long digital flexor
(emergence of“flexion tubercle”on ulna), subduction or gracilization of long flexor tendon of thumb to expanded long digital flexor, increased osseo-ligamentous resistance to torque in CJC
via distal prolongation of the volar portion of the capitate with corresponding evacuation of the Mc3 base (creating a mediolateral block-to-joint rotation by novel abutment of Mc2 and
Mc3), dorsalization and enlargement of capitate head, frequent formation of prepollex (62) on trapezium, anterior relocation of collateral ligament attachments of metacarpophalangeal
joints (with simultaneous expansion of attachment facets on metacarpals), expansion of metacarpal heads, reduced capacity for dorsiflexion at midcarpal joint. Introduction of lateral spiral
pilaster with loss of third trochanter, elimination of os peroneal complex and substantial shortening of midtarsus, especially proximodistal abbreviation of navicular and cuboid, and
abbreviation of dorsoplantar dimensions of metatarsal bases. Gracilization of plantar aponeurosis with loss of plantaris and reduction/elimination of quadratus plantae.
5. Basal chimpanzee/bonobo clade.Elongation of forelimb and abbreviation of hindlimb (global change in limb proportions) but less extreme than in4. Substantial elongation of
posterior (medial) metacarpus and further elongation of manual phalanges. Chimpanzees exhibit higher intermembral index than bonobos and are probably derived in this regard.
Abbreviation of lumbar vertebral column (three or four elements) by transformation of vertebral type and/or reduction in axial length by one segment[chimpanzees and bonobos
differ substantially in number of axial elements, and bonobo is clearly primitive in this regard (58 )]. Entrapment of most caudal lumbars by articulation with cranially extended ilia and
reduction in breadth of sacral alae. Further immobilization by novel lumbo-inguinal ligaments (81). Elongation of iliac isthmus. Dramatic mediolateral narrowing of scapula, marked
increase in cranial orientation of scapular spine and glenoid plane, reduction of deltopectoral crest (intermuscular fusion?). Retroflexion of ulnar trochlear notch with attendant
abbreviation of olecranon process, expansion of long digital flexor (emergence of“flexion tubercle”on ulna), subduction or gracilization of long flexor tendon of thumb to expanded
long digital flexor, increased osseo-ligamentous resistance to torque in CJC via distal prolongation of the volar portion of the capitate with corresponding evacuation of the Mc3 base
(creating a mediolateral block to joint rotation by novel abutment of Mc2 and Mc3), dorsalization and enlargement of capitate head, elimination of mobility in hamate/Mc4/Mc5 joint,
possible gracilization of Mc1, reduced capacity for dorsiflexion at midcarpal joint, reduction and anterior relocation of collateral ligament“grooves”of metacarpophalangeal joints (but
expansion of attachment facets on metacarpals), expansion of metacarpal heads. Introduction of lateral spiral pilaster with loss of third trochanter, elimination of os peroneal complex
and substantial shortening of midtarsus, especially proximodistal abbreviation of navicular and cuboid, abbreviation of dorsoplantar dimensions of metatarsal bases. Gracilization of
plantar aponeurosis with loss of plantaris and reduction/elimination of quadratus plantae.
6. Hominid clade, Late Miocene.Substantial superoinferior abbreviation of iliac isthmus and pubic symphyseal body, increased sagittal orientation and mediolateral broadeningof
ilium with novel growth plate for anterior inferior iliac spine, introduction of slight (obtuse) greater sciatic notch, (inferred) facultative lumbar lordosis, probable broadening of sacral alae
to free most caudal lumbar for lordosis. Possible increased size and robusticity of fibularis longus, increased robusticity of second metatarsal base/shaft and doming of dorsal metatarsal
heads related to toe-off.
7. Hominid clade, Mid-Pliocene.Shortening of ischial length and angulation of ischial tuberosity, further mediolateral expansion of iliac fossa with introduction of substantial(acute)
greater sciatic notch, further invagination of lumbar vertebral column and fixation of lordosis (no longer facultative). Reduction of thoracic column from 13 to 12 elements associated
with reduction in axial length by one segment [or this occurred at6(58)]. Elongation of pubic rami and femoral neck. Posterior relocation of third trochanter and emergence of true
hypotrochanteric fossa. Elevation of quadriceps attachments to form“true”linea aspera, signaling fundamental shift in knee extensor/hip extensor proportions conducive to primary
propulsion by quadriceps. Probable emergence of tibial dominant knee and transverse tibial plafond (or these occurred at6). Expansion of fibularis longus attachment to include
markedly remodeled medial cuneiform and permanent adduction of great toe, elevation of sustentaculum tali to create mediolateral and longitudinalplantar arches, likely development
of“spring ligament,”marked inflation of calcaneal tuber (with secondary introduction of distinct lateral plantar process) for energy absorption at heel strike, gracilization of second
metatarsal base, relocation of fibularis longus tendon to more proximo-plantar location (with inferred attendant change in short and long plantar ligaments [see (47 )] to support novel
transverse arch during toe-off and foot-flat, introduction of“dual phase” metatarsofulcrumation (addition of transverse axis to oblique axis of fulcrumation). Dorsalization and expansion
of capitate head and broadening of trapezoid for greater palmar span, slight reduction in dorsal mobility of Mc5/hamate joint, anterior relocation and near elimination of collateral
ligament“grooves”for metacarpophalangeal joint.
8. Hominid clade, Plio-Pleistocene.Elongation of lower limb, global modification of pelvis to expand birth canal (late) including abbreviation of femoral neck and pubic rami.
Reduction of modal lumbar column by one (from six to five typically by sacralization of most caudal lumbar). Slight reduction in glenoid angulation ofscapula, increased robusticity of
thumb, transfer of styloid body from capitate to third metacarpal, palmar rotation of hamulus, loss of growth plate from pisiform, increased robusticity of terminal phalangeal tufts in
carpus. Substantial abbreviation of posterior metacarpus, antebrachium, and carpal phalanges. Substantial anteroposterior thickening of navicular and length and eccentricity of
calcaneal process of cuboid. Increased robusticity of Mt1. Reduction in frequency of calcification of os peroneum, abbreviation of tarsal phalanges—especially intermediates.
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009 105
RESEARCH ARTICLESResearch Articles

98. 98 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
fossils. Contemporary morphogenetics now
show that organisms as diverse as sticklebacks
and fruit flies can display remarkable parallel
evolution merely because they share fundamen-
tally similar genomic toolkits (73,74). Knuckle-
walking in chimpanzees and gorillas appears now
to be yet one more example of this phenomenon.
In retrospect, it is impressive that the straight-
forward cogency of Schultz and the detailed dis-
sections of Straus more accurately predicted the
early course of human evolution than the more
objective quantitative and technologically en-
hanced approaches heralded in the last quarter of
the 20th century. Recent work in genetics and
developmental biology has identified fundamen-
tal mechanisms by which morphological struc-
tures emerge during evolution. In the study of
fossils, such insights have had their primary
value as heuristic guides with which to construct
and test hypotheses. Understanding the morpho-
genesis underlying profound shifts in the homi-
noid bauplan evidenced byAr. ramidusmay take
years, perhaps even decades, but is likely to
further transform our understanding of human
natural history.
Ardipithecushas thus illuminated not only
our own ancestry, but also that of our closest
living relatives. It therefore serves as further con-
firmation of Darwin’ s prescience: that we are
only one terminal twig in the tree of life, and that
our own fossil record will provide revealing and
unexpected insights into the evolutionary emer-
gence not only of ourselves, but also of our
closest neighbors in its crown.
References and Notes
1. T. H. Huxley,Evidence as to Man’s Place in Nature
(Williams & Norgate, London, 1863).
2. A. Keith,BMJ1, 451, 454, 499, 502, 545 (1923).
3. W. K. Gregory,Man’s Place Among the Anthropoids
(Oxford, Clarendon, 1934).
4. W. L. Straus Jr.,Q. Rev. Biol.24, 200 (1949).
5. A. H. Schultz,Hum. Biol.2, 303 (1930).
6. O. J. Lewis,Nature230, 577 (1971).
7. M. Cartmill, K. Milton,Am. J. Phys. Anthropol.47, 249
(1977).
8. R. H. Tuttle,Science166, 953 (1969).
9. A. C. Wilson, V. M. Sarich,Proc. Natl. Acad. Sci. U.S.A.
63, 1088 (1969).
10. S. L. Washburn,The Study of Human Evolution (Condon
Lectures)(Univ. of Oregon Books, Eugene, 1968).
11. B. G. Richmond, D. R. Begun, D. S. Strait,Am. J. Phys.
Anthropol.116(S33), 70 (2001).
12. R. Tuttle, B. B. Beck,Nature236, 33 (1972).
13. A. Caccone, J. R. Powell,Evolution43, 925 (1989).
14. S. Horaiet al.,J. Mol. Evol.35, 32 (1992).
15. M. Ruvolo,Mol. Biol. Evol.14, 248 (1997).
16. F.-C. Chen, W.-H. Li,Am. J. Hum. Genet.68, 444
(2001).
17. The Chimpanzee Sequencing and Analysis Consortium,
Nature437, 69 (2005).
18. N. Patterson, D. J. Richter, S. Gnerre, E. S. Lander,
D. Reich,Nature441, 1103 (2006).
19. A. L. Zihlman, J. E. Cronin, D. L. Cramer, V. M. Sarich,
Nature275, 744 (1978).
20. R. Wrangham, D. Pilbeam, inAll Apes Great and Small,
vol. 1,African Apes, B. Galdikaset al., Eds. (Kluwer
Academic/Plenum, New York, 2001), pp. 5–17.
21. W. C. McGrew,The Cultured Chimpanzee: Reflections on
Cultural Primatology(Cambridge Univ. Press, Cambridge,
2004).
22. B. Wood, W. Lonergan,J. Anat.212, 354 (2008).
23. D. C. Johanson, T. D. White,Science203, 321 (1979).
24. C. O. Lovejoy, R. S. Meindl, J. C. Ohman, K. G. Heiple,
T. D. White,Am. J. Phys. Anthropol.119, 97 (2002).
25. M. D. Leakey, R. L. Hay,Nature278, 317 (1979).
26. T. D. White, G. Suwa,Am. J. Phys. Anthropol.72, 485
(1987).
27. J. T. Stern Jr., R. L. Susman,Am. J. Phys. Anthropol.60,
279 (1983).
28. R. J. Clarke,S. Afr. J. Sci.98, 523 (2002).
29. M. Brunetet al.,Nature418, 145 (2002).
30. B. Senutet al.,C. R. Acad. Sci. IIA332, 137 (2001).
31. Y. Haile-Selassie,Nature412, 178 (2001).
32. T. D. Whiteet al.,Science326, 64 (2009).
33. A. M. W. Porter,Int. J. Osteoarchaeol.5, 203 (1995).
34. G. Suwaet al., Science326, 69 (2009).
35. P. L. Reno, R. S. Meindl, M. A. McCollum, C. O. Lovejoy,
J. Hum. Evol.49, 279 (2005).
36. M. Nakatsukasa, M. Pickford, N. Egi, B. Senut,Primates
48, 171 (2007).
37. C. V. Ward, A. Walker, M. F. Teaford, I. Odhiambo,
Am. J. Phys. Anthropol.90, 77 (1993).
38. K. L. Rafferty, A. Walker, C. B. Ruff, M. D. Rose,
P. J. Andrews,Am. J. Phys. Anthropol.97, 391 (1995).
39.P. troglodytesandP. paniscusdiffer in a number of
skeletal proportions and characters (75,76). Our sample
ofPanwas limited toP. troglodytes, and our use of the
genus nomenPanherein refers only toP. troglodytes,
except where otherwise noted.
40. W. L. Jungers,Nature297, 676 (1982).
41. P. L. Renoet al.,Curr. Anthropol.46, 575 (2005).
42. P. L. Renoet al.,J. Exp. Zool. B Mol. Dev. Evol.310B,
240 (2008).
43. B. Asfawet al.,Science284, 629 (1999).
44. C. O. Lovejoyet al.,Science326,70 (2009).
45. G. P. Wagner, A. O. Vargas,Genome Biol.9, 213
(2008).
46. S. Prabhakaret al.,Science321, 1346 (2008).
47. C. O. Lovejoyet al.,Science326, 72 (2009).
48. S. Almecija, D. M. Alba, S. Moya-Sola, M. Kohler,Proc.
Biol. Sci.274, 2375 (2007).
49. The trait nomenclature system used here is taken from
(77,78) and is briefly as follows [for more complete
explanations, see (32)]. Type 1: traits whose
morphogenesis is the direct consequence of pattern
formation usually (but not always) subject to direct
selection. Type 2: traits that are genetic but are
pleiotropic to, or result from hitchhiking on, type 1 traits
and are not themselves subject to selection [2A: parent
type 1 is inferred to be under selection; its secondary
effects are not; 2B: neither parent trait nor derivative is
inferred to be under selection (rare)]. Type 3: resulting
from a systemic growth factor. Type 4: epigenetic
consequence of osteochondral remodeling and/or
response to environmental stimuli, i.e., not heritable but
useful in interpreting behavior. Type 5: developmentally
similar to type 4, but functionally uninformative.
50. C. O. Lovejoy,Gait Posture21, 113 (2005).
51. C. O. Lovejoy,Gait Posture21, 95 (2005).
52. J. T. Stern Jr.,Evol. Anthropol.9, 113 (2000).
53. J. T. Stern Jr., R. Sussman,Am. J. Phys. Anthropol.60,
279 (1983).
54. S. Semawet al.,Nature385, 333 (1997).
55. J. M. DeSilva, thesis, Univ. of Michigan (2008).
56. J. M. DeSilva,Proc. Natl. Acad. Sci. U.S.A.106, 6567
(2009).
57. D. Pilbeam,J. Exp. Zool.302B, 241 (2004).
58. M. A. McCollumet al.,J. Exp. Zool. B Mol. Dev. Evol.
10.1002/jez.b.21316 (2009).
59. C. O. Lovejoyet al., Science326, 71 (2009).
60. C. V. Ward, B. Latimer,Am. J. Phys. Anthropol.34(S12),
180 (1991).
61. Z. Alemsegedet al.,Nature443, 296 (2006).
62. J. M. LeMinor,Acta Anat. (Basel)150, 227 (1994).
63. G. Suwa, R. T. Kono, S. Katoh, B. Asfaw, Y. Beyene,
Nature448, 921 (2007).
64. C. O. Lovejoy,Science211, 341 (1981).
65. P. Andrews,Cold Spring Harb. Symp. Quant. Biol.51,
419 (1986).
66. Y. Kunimatsuet al.,Proc. Natl. Acad. Sci. U.S.A.104,
19220 (2007).
67. S. K. S. Thorpe, R. Holder, H. Crompton,Proc. Natl. Acad.
Sci. U.S.A.106, 12646 (2009).
68. S. Moya-Sola, M. Kohler, D. M. Alba, I. Casanovas-Vilar,
J. Galindo,
Science306, 1339 (2004).
69. C. O. Lovejoy,Proc. Biol. Sci.274, 2373 (2007).
70. J. T. Stern Jr., R. L. Susman,Am. J. Phys. Anthropol.60,
279 (1983).
71. M. Nakatsukasa, Y. Kunimatsu,Evol. Anthropol.18, 103
(2009).
72. T. Harrison, inOrigine(s) de la Bipedie chez les Homindes
(Editions du CNRS, Paris, 1991), pp. 235–244.
73. S. Carroll, J. K. Grenier, S. D. Weatherbee,From DNA to
Diversity: Molecular Genetics and the Evolution of Animal Design(Blackwell Science, Malden, MA, 2001).
74. M. D. Shapiro, M. A. Bell, D. M. Kingsley,Proc. Natl.
Acad. Sci. U.S.A.103, 13753 (2006).
75. A. L. Zihlman, D. L. Cramer,Folia Primatol. (Basel)29,
86 (1978).
76. H. J. Coolidge, B. T. Shea,Primates23, 245 (1982).
77. C. O. Lovejoy, R. S. Meindl, J. C. Ohman, K. G. Heiple,
T. D. White,Am. J. Phys. Anthropol.119, 97 (2002).
78. C. O. Lovejoy, M. J. Cohn, T. D. White,Proc. Natl. Acad.
Sci. U.S.A.96, 13247 (1999).
79. G. Suwaet al., Science326, 68 (2009).
80. M. Nakatsukasaet al.,J. Hum. Evol.45, 179
(2003).
81. M. D. Sonntag,Proc. Zool. Soc. London23, 323
(1923).
82. For funding, we thank the U.S. National Science
Foundation and the Japan Society for the Promotion of
Science. We thank the Ministry of Tourism and Culture,
the Authority for Research and Conservation of the
Cultural Heritage, and the National Museum of Ethiopia
for permissions and facilitation. We thank the Afar
Regional Government, the Afar people of the Middle
Awash, and many other field workers for contributing
directly to the data. We thank the following institutions
and staff for access to comparative materials: National
Museum of Ethiopia; National Museums of Kenya;
Transvaal Museum South Africa; Cleveland Museum of
Natural History; Royal Museum of Central Africa Tervuren;
and the University of California at Berkeley Human
Evolution Research Center. We thank K. Brudvik for
illustration and editorial assistance. We thank the
following individuals for cooperation with comparative
data: M. Brunet and C. V. Ward. We thank R. Meindl for
statistical advice and assistance and P. L. Reno,
M. A. Serrat, M. A. McCollum, M. Selby, and B. A.
Rosenman for aid in data collection and exceptionally
helpful discussions.
Supporting Online Material
www.sciencemag.org/cgi/content/full/326/5949/73/DC1
SOM Text S1 to S3
Figs. S1 to S15
Tables S1 to S3
References
4 May 2009; accepted 31 August 2009
10.1126/science.1175833
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org106
Ardipithecus ramidus
Ardipithecus ramidus

99. 99www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
Reexamining Human Origins in
Light ofArdipithecus ramidus
C. Owen Lovejoy
Referential models based on extant African apes have dominated reconstructions of early
human evolution since Darwin’s time. These models visualize fundamental human behaviors as
intensifications of behaviors observed in living chimpanzees and/or gorillas (for instance,
upright feeding, male dominance displays, tool use, culture, hunting, and warfare).Ardipithecus
essentially falsifies such models, because extant apes are highly derived relative to our last
common ancestors. Moreover, uniquely derived hominid characters, especially those of locomotion
and canine reduction, appear to have emerged shortly after the hominid/chimpanzee divergence.
Hence,Ardipithecusprovides a new window through which to view our clade’s earliest evolution
and its ecological context. Early hominids and extant apes are remarkably divergent in many
cardinal characters. We can no longer rely on homologies with African apes for accounts of our
origins and must turn instead to general evolutionary theory. A proposed adaptive suite for the
emergence ofArdipithecusfrom the last common ancestor that we shared with chimpanzees
accounts for these principal ape/human differences, as well as the marked demographic success
and cognitive efflorescence of later Plio-Pleistocene hominids.
A
n essential goal of human evolutionary
studies is to account for human unique-
ness, most notably our bipedality, marked
demographic success, unusual reproductive phys-
iology, and unparalleled cerebral and technolog-
ical abilities. During the past several decades, it
has been routinely argued that these hominid char-
acters have evolved by simple modifications of
homologs shared with our nearest living rela-
tives, the chimpanzee and bonobo. This method
is termed referential modeling (1). A central tenet
has been the presumption (sometimes clearly
stated but more often simply sub rosa) thatGorilla
andPanare so unusual and so similar to each
other that they cannot have evolved in parallel;
therefore, the earliest hominids must have also
resembled these African apes (2,3). Without a
true early hominid fossil record, the de facto null
hypothesis has been thatAustralopithecuswas
largely a bipedal manifestation of an African ape
(especially the chimpanzee). Such proxy-based
scenarios have been elevated to common wisdom
by genomic comparisons, progressively estab-
lishing the phylogenetic relationships ofGorilla,
Pan, andHomo(4).
EarlyAustralopithecus.AlthoughAustralo-
pithecuswas first encountered early in the last
century (5 ), its biology was only slowly revealed.
In the 1970s, abundant earlierAustralopithecus
fossils began to emerge in eastern Africa. These
samples broadened our understanding of the ge-
nus and included partial skeletons (6) and even
footprint trails [the latter extending our knowledge
to 3.75 million years ago (Ma)] (7).
To many, these fossils were consistent with
chimpanzee-based referential scenarios. Bipe-
dality had long been argued to have occurred
when early hominids ventured onto the expand-
ing savannas and grasslands of the Pliocene
(8,9). More recently, bipedality is seen to have
emerged from African ape behaviors, including
feeding postures (10,11), gorilla dominance dis-
plays (12), and even vertical climbing (13). Many
mechanical/behavioral models have been pro-
posed to explain the evolution of hominid bipe-
dality, but most have presumed it to have evolved
from a chimpanzee-like ancestor (4,14,15). A
primary problem with these scenarios has been the
remarkably advanced postcranium of earlyAustra-
lopithecus, which exhibits particularly advanced
adaptations to upright walking (16–18).
Ardipithecus ramidus.Ardipithecus ramidus
now reveals that the early hominid evolutionary
trajectory differed profoundly from those of our
ape relatives from our clade’ s very beginning.
Ar. ramiduswas already well-adapted to bipe-
dality, even though it retained arboreal capa-
bilities (19–25). Its postcranial anatomy reveals
that locomotion in the chimpanzee/human last
common ancestor (hereafter the CLCA) must
have retained generalized above-branch quad-
rupedality, never relying sufficiently on suspen-
sion, vertical climbing, or knuckle walking to
have elicited any musculoskeletal adaptations to
these behaviors (26 –28).
Moreover,Ardipithecuswas neither a ripe-
fruit specialist likePan, nor a folivorous browser
likeGorilla, but rather a more generalized omni-
vore (19,25). It had already abandoned entirely
the otherwise universal sectorial canine complex
(SCC), in which the larger, projecting upper ca-
nine is constantly honed by occlusion against the
lower third molar of anthropoid primates (25),
demonstrating that the large, projecting, inter-
locking, and honing male canines of apes had
been eliminated before the dawn of the Pliocene
and before the emergence of the dentognathic
peculiarities ofAustralopithecus. What’s more, it
appears to have been only slightly dimorphic in
body size (25). Finally, the environmental context
ofArdipithecussuggests that its primary habitat
was not savanna or grassland, but instead wood-
lands (26–28).
In retrospect, clues to this vast divide between
the evolutionary trajectories of African apes and
hominids have always been present. Apes are
largely inept at walking upright. They exhibit
reproductive behavior and anatomy profoundly
unlike those of humans. African ape males have
retained (or evolved, see below) a massive SCC
and exhibit little or no direct investment in their
offspring (their reproductive strategies rely pri-
marily on varying forms of male-to-male agonism).
Although they excel at some cognitive tasks, they
perform at levels qualitatively similar to those of
some extraordinary birds (29,30) and mammals
(31). The great apes are an isolated, uniquely
specialized relict species surviving today only by
their occupation of forest refugia (32 ). Even their
gut structure differs substantially from that of
humans (33 ).
How and why did such profound differences
between hominids and African apes evolve?
Why did early hominids become the only primate
to completely eliminate the SCC? Why did they
become bipedal, a form of locomotion with vir-
tually no measurable mechanical advantage (34)?
Why did body-size dimorphism increase in their
likely descendants? These are now among the
ultimate questions of human evolution. We can,
of course, only hypothesize their answers. Never-
theless, by illuminating the likely morphological
structure and potential social behavior of the CLCA,
Ar. ramidusnow confirms that extant African ape–
based models are no longer appropriate.
Adaptive suites.An alternative to referential
modeling is the adaptive suite, an extrapolation
from optimization theory (35 ). Adaptive suites
are semiformal, largely inductive algorithms that
causally interrelate fundamental characters
that may have contributed to an organism’s total
adaptive pattern. One for the horned lizard
(Phyrnosoma platyrhinos) of the southwesten
U.S. serves as an excellent example (Fig. 1)
(36,37). For this species, the interrelation be-
tween a dietary concentration on ants and its
impact on body form imply, at first counter-
intuitively, that elevation of clutch size and inten-
sification of“r”strategy (maximize the number of
offspring by minimizing paternal care) are the ul-
timate consequences of this specialization (35–37).
Such character and behavioral interdependen-
cies can have profound consequences on evolu-
tionary trajectory, as demonstrated by the equally
notable differences in clutch size in the common
leopard frog (3500 to 6500 eggs) versus those of
numerous species of so-called poison dart frogs
Ardipithecus ramidus
Department of Anthropology, School of Biomedical Sciences,
Kent State University, Kent, OH 44242–0001, USA. E-mail:
[email protected]
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 200974e1
Research Articles

100. 100 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
[typically less than 30 eggs; Table 1 (38 )]. To
enhance survival of their (as yet) nontoxic off-
spring, the latter engage in relatively intense male
parental investment, a shift that has had a profound
adaptive impact on their entire life-history strategy.
The effective power of adaptive suites is de-
monstrable by their explanatory success. A vir-
tually identical character constellation to that of
the horned lizard has been discovered in an un-
related Australian ecological vicar,Moloch hor-
ridus(37), which is also an ant specialist. Even
given such unexpected consilience, however,
adaptive suites are obviously speculative, even
for living organisms. In addition, for extant spe-
cies, the processes by which current characters
have emerged are also necessarily hidden in the
past and, therefore, are no more accessible than
for extinct taxa. Nevertheless, adaptive suites can
serve as organizational procedures by which to
examine evolutionary processes with increasing
acumen. Of further benefit is the fact that they
often pose novel testable hypotheses that might
not have arisen without them.
Many key human specializations are related
to our reproductive physiology and anatomy; hu-
man reproduction is as extraordinary as our den-
tition, locomotion, and encephalization (39).
Although it remains possible that such unique-
ness emerged only during the Pleistocene, this is
less likely in light ofArdipithecus, which shows
very early evidence of a major social transfor-
mation (25). Moreover, it is the modern African
apes that are most derived in many characters,
whereas those which are specialized in human
evolution (SCC elimination, bipedality) are now
known to have been present near the origin of our
clade. Our massive brains are obviously a Pleis-
tocene development, but they are also probably
sequelae to other major shifts now more fully
recorded in the earlier fossil record. It is therefore
possible, even likely, that many physiologies and
soft tissue features that do not fossilize may have
also evolved early in hominid evolution. If so,
why were these characters exaptive to our ad-
vanced cognition and singular demographic
success?
Notwithstanding the revelations now provided
byArdipithecus, it should be noted that extensive
studies of African apes and other primates have
provided likely details of the sociobehavioral con-
text from which hominids most likely first emerged
(1,11). These details were presumably present in
the last common ancestor we shared with the
African apes, and they almost certainly included
aspects of great ape demography and social be-
havior, including male philopatry (males remain in
their natal group), female exogamy (females
transfer from natal group at sexual maturity), and
prolonged inter-birth intervals, all cardinal char-
acters of an intense“k”(maximized parental care of
few offspring) reproductive strategy (32,40).
Moreover, investigations of the behavior of other
living primates now provide a wealth of infor-
mation that allows many contextual details of ear-
liest human evolution to be reasonably hypothesized.
Sperm competition.Two key factors domi-
nate anthropoid reproductive behavior and are
therefore diagnostic of socio-sexual structure:
(i) sperm competition and (ii) male-to-male com-
petition for mates. Various anatomical correlates
distinguish monogamous or single male primates
from other species whose males engage in sperm
competition. Among the most obvious is the much
higher ratio of testes volume to body mass.
Human ratios are generally similar to those of
monogamous gibbons and solitary orangutans,
but the ratios are three times higher inPan(41,42)
and other sperm-competing species such as
Brachyteles(43). Moreover, human testes are
most similar to those of gibbons with respect to
their higher proportion of intertubular (non-
seminiferous) tissue (42). Mammalian sperm com-
petition is generally accompanied by elevated
ejaculate quality (44), which is also notably poor
in humans. InHomo sapiens, the absolute rate of
sperm production is only about 20% that of much
smaller rhesus macaques (45). Another measure,
spermatogenesis efficiency (daily sperm produc-
tion per gram of testes),“varies from about 2.65 ×
Fig. 1.Adaptive suite of the horned lizard. An adaptive suite summarizes functional interrelations among
physiological, locomotor, dietary, and reproductive characters. One is shown here forPhrynosoma
platyrhinos. Horned lizards are ant specialists and thus consume copious amounts of indigestible chitin.
This requires a large stomach–to–body mass ratio, which in turn generates the lizard’s unusual tanklike
body form. The latter eliminates flight as an effective predator response, and selection has therefore
replaced rapid flight (typical of sympatric lizards) with armor and crypsis (e.g., camouflage). These require
motionlessness for long periods, which has generated a tolerance for high variance in body temperature,
exceeding that of other sympatric lizards. Motionlessness also relaxes selection against large clutch size (which
is very large inP. platyrhinos); self-weighting by large clutches in sympatric lizards does not occur because it
reduces flight speed [(35), relevant background data available at http://uts.cc.utexas.edu/~varanus/pubs.html].
Table 1.Some differential effects of mating strategy on life-history variables in two amphibians.
Character or behavior Dart frogs* Leopard frogs†
Clutch size (eggs) 4–30 3500–6500
Longevity (years) 13–15 6–9
Male egg attendance yes no
Male tadpole transport yes no
Female provisioning yes no
*Data from (37).†Data from (114).
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org 74e2
RESEARCH ARTICLES
Ardipithecus ramidus

101. 101www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
10
7
in rabbits to <0.06 × 10
6
in humans”(46). The
estimated corresponding value in chimpanzees
is greater than that of humans by two orders of
magnitude (42 ).
The muscular coat of mammalian vasa
deferentia can reasonably be regarded as a cor-
relate of sperm transport rate during sexual stimu-
lation. It is substantially thicker in chimpanzees
than in humans or orangutans (47). The seminal
vesicles of some monogamous primates are in-
conspicuous, whereas those of multimale (i.e.,
ovulating females usually mate with multiple
males) macaques and chimpanzees are large;
those in humans are only of moderate size (39).
Whatever the social caveats may be, human ejac-
ulatory rates (along with those of the monoga-
mous generaAotusandSymphalangus) are lower
than those of 20 primate species (includingPan
andGorilla) by one order of magnitude (48 ,49),
and human sperm counts decrease at ejaculation
frequencies of >4 per week (50).
Human sperm midpiece volume, which re-
flects mitochondrial density and motility, falls in
the lowest quartile of 21 primate species examined
(51). Especially important is the coagulating re-
action between some seminal proteins and prostate
vesiculase (52 ). This coagulum, which blocks pen-
etration of competing sperm by forming a vaginal
plug, characterizes primates that robustly sperm-
compete (e.g.,Ateles,Brachyteles,Macaca,Pan).
This reaction is absent in humans and common
marmosets, whose ejaculates are merely gelat-
inous (53 ).
The structures of semenogelins I and II
(SEMGI and SEMGII) (primary plug coagulates)
illuminate the natural history of vaginal plugging.
SEMGI suggests a selective sweep in chimpan-
zees and conversion to a pseudogene in gorillas;
humans exhibit neither (52). Together, these data
strongly suggest that the social structure in earlier
hominids is unlikely to have been typically multi-
male. This conclusion is supported by recent analy-
ses of primate immune systems, which compared
basal white blood cell counts among primates
with respect to the likely number of sexual part-
ners as determined by social system (female mating
promiscuity). Results showed that“humans align
most closely with the [single male] gorilla ... and
secondarily with…[the] monogamous gibbon”
[(54), p. 1170].
Humans have the least complex penis mor-
phology of any primate. Complexity is generally
associated with multimale social structure (47),
and humans lack keratinous penile surface mech-
anoreceptors that may promote rapid ejaculation
that is common in many primates. Finally, hu-
mans are the only catarrhine without an os bac-
ulum (39).
Competition for mates.If they did not sperm-
compete, did early hominid males instead com-
pete for single or near-solitary control of female
groups? The cardinal indicator of male-to-male
agonism in hominoid primates is the SCC. It is
regularly employed during both territory defense
and dominance disputes. Hominids are often
characterized as having reduced canine dimor-
phism (55 ). Such reduction is only a secondary
consequence of the primary hominid character,
which is elimination of the SCC in its entirety.
The SCC is not male-limited; that is, it is always
expressed in both sexes of all anthropoids, even
in species with reduced dimorphism (e.g., some
New World atelines). Although females may ex-
press the SCC for advantage in conflicts with
other females, they principally express its under-
lying structure themselves because amplifica-
tion in their male offspring (presumably by
androgens or reduced estrogens) enhances their
fitness. Hylobatid canine monomorphism is
sometimes erroneously confused with that of
hominids, but gibbons evolved amplification of
the female canine.Ar. ramidusshows that elim-
ination of the SCC in hominids is unique among
all higher primates and occurred long before
Australopithecus.
A frequent explanation of canine reduction
(and bipedality) is that hand-held weapons re-
placed the SCC (56,57). But if male-to-male
agonism had been fundamental to early hominid
fitness, what selective agency would have re-
duced its signature character? Additional human
attributes belie the improbability of the weapons
argument. An absence of sperm competition in
gorillas and orangutans is accompanied by a dra-
matically reduced testes size and the elimination
of a free scrotum (their testes are more judicious-
ly sequestered in a post-penial bulge) (42 ,58). In
contrast, not only are human scrota more pendu-
lous than even those of chimpanzees (58), but
bipedality makes them extraordinarily vulnerable
during upright combat (59). It seems illogical to
attribute habitual uprightness to weapons that
would demand even greater selection for testes
sequestration than is present in other primates
[which target them with their functional SCCs
(60)].
Available evidence now suggests that the
loss of the SCC was, as is theoretically most
likely, a social adaptation. This evidence, derived
fromArdipithecus, includes the following (25):
(i) Change in the more socially important upper
canine preceded that in the lower, (ii) progressive
shape modification made the canine not only
smaller but less weaponlike in form, (iii) male
canines erupted relatively earlier than in large-
canined species with high male-to-male agonism,
making this event less likely to have represented
a social signal of male maturity, and (iv) all of
these changes took place within a dietary context
that preceded any of the profound changes seen
in later hominid dentitions.
Humans are also unique among primates in
lacking vocal sacs, which play a major role in the
territoriality of all apes. Though there are no
current means by which to judge the evolutionary
history of the hominoid vocal apparatus (61), it
does have potential developmental interactions
with basicranial patterning, including an impact
on location of the foramen magnum. The dra-
matic anterior translation of this foramen during
the Plio-Pleistocene is almost certainly a corol-
lary of cerebral reorganization and/or expansion
(62). However, early hominid vocal apparatus
reduction may have influenced initial differential
trajectories of cranial form, currently only just
detectable inP. paniscusandAr. ramidus(29).
Both cerebral reorganization and facial pattern-
ing are clearly central elements of that trajec-
tory, and early reduction of vocal tract mass is
thus a potential modulating factor, particularly
because it is a possible social corollary of loss
of the SCC.
It has long been argued thatAustralopithecus
was unusually dimorphic in body size, implying
a largely single-male group structure, but this
hypothesis has been biased by comparisons of
temporally and geographically disparate samples
(63). Of greater importance are (i) the absence of
any useful correlation between body-size dimor-
phism and social structure in hominoids, because
both chimpanzees and gorillas exhibit intense
male-to-male agonism but exhibit opposite polar-
ities in skeletal dimorphism (63 ); and (ii) the fact
that male body size in many primates is not
associated with competition for mates. Rather, it
is equally likely to be an ecological specialization
derived from reduced size of females (64) and/or
male enlargement by selective agencies unrelated
to mate acquisition. In any case,Ar. ramidusnow
transcends the debates over dimorphism in early
Australopithecusbecause available samples in-
dicate that it was minimally dimorphic, suggest-
ing that this was the primitive hominid condition
(19,25) and that dimorphism increased in later
hominids (see below).
Reproductive biology of the CLCA.Apes
radiated profusely during the Middle Miocene
(~16 to 11.5 Ma) yet became largely extinct
by its terminus (5.3 Ma), which coincided with
the radiation of Old World monkeys (65). The
nearly total replacement of great apes by cerco-
pithecids is likely to have been closely associated
with advanced K specialization in the former,
shared by all surviving hominoids (66 ,67).
However, in dramatic contrast to all other ape
descendants, hominids became remarkably eco-
logically and geographically cosmopolitan. What
reproductive strategy permitted such success?
Equally as important, what was the likely re-
productive strategy of the species that was im-
mediately ancestral to both the hominid and
chimpanzee clades?
Advanced K selection must have heavily af-
fected the sociobiology of the earliest hominids.
K-driven protraction of life history and increased
social adhesion require behaviors that avert
inbreeding: either isolation of adults as pairs or
female transfer among larger social units (68).
The latter proscribes male philopatry (males
remain in their natal group) and kin selection
(individual fitness is amplified by that of rela-
tives) and greatly reduces female-to-female co-
operation and its benefits (e.g., alloparenting),
placing at a premium novel mechanisms that can
enhance parenting.
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
74e3
Ardipithecus ramidus
Research Articles

102. 102 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
The primitive nature of the craniodental and
postcranial anatomies ofAr. ramidussuggest that
the CLCA, unlike extant African apes, was pre-
dominantly arboreal. However, all of its descend-
ants have since developed relatively sophisticated
adaptations to terrestrial locomotion (23). What
was the CLCA’s socio-reproductive structure
before these events? Whereas African apes have,
in the past, almost invariably been selected as
CLCA vicars,Ar. ramidusnow allows us to infer
that they have undergone far too many pronounced
and divergent specializations to occupy such a role.
Possible alternative vicars are extant, K-selected
atelines (New World forms such as spider and
woolly monkeys), which exhibit many of the
CLCA’s likely socio-reproductive characters,
including male philopatry, minimal-to-moderate
canine dimorphism, moderate– to–possibly intense
Fig. 2.Emergent adaptive suite in basal hominids. The last common ancestor
(LCA) of humans and African apes probably exhibited multimale, multifemale
(i.e., females mate with multiple males and vice versa) structure with moderate
canine dimorphism and minimal male-to-male agonism, perhaps similar to
New World atelines (e.g.,Brachyteles), with moderate-to-substantial sperm
competition, female transfer (i.e., females leave natal group at maturity), and
male philopatry. Here, hominids are hypothesized to have evolved three
entirely unique, primary characters (denoted by yellow triangles). Two of these
characters, documented in the fossil record, are bipedality and SCC elimi-
nation. Modern humans exhibit the third: ovulatory crypsis. Interrelations are
hypothesized as follows: (i) transport (object carrying but especially food) leads
to habitual bipedality, (ii) female choice of males with limited agonism leads
to eclipse of SCC, and (iii) protection against cuckoldry (both sexes) leads to
ovulatory crypsis. Two natural sequences generated this adaptive milieu.
(Left column) Simple feeding ecology from CLCA to earlyArdipithecusand
eventuallyAu. afarensis.(Right column) The demographic dilemma ( 32,79)
generated by intensified K selection. A solution for a hominoid confronting such
selective forces is elaboration of sex-for-food exchanges observed in chim-
panzees and bonobos. These and other elements shared withPanacted as
possible“social catalysts”[highlighted in red (e.g., copulatory feeding, extrac-
tive foraging, male-male patrols)]. Increased male body size and enhanced
male-to-male cooperation inAu. afarensisreduced mortality during distance-
foraging by multiple-male patrols (whose role was optimal foraging rather than
territory defense). This culminated in savanna scavenging, primitive lithics for
meat acquisition, marrow extraction, and cooperative hunting inHomo. This
profound economic shift selected for advanced adaptations to bipedality, further
enhanced social cohesion (reduced same-sex agonism in both sexes), increased
energy available for parenting (and alloparenting), promoted survivorship and
reduced birth spacing, and elevated the selective ceiling acting against meta-
bolically expensive tissues (e.g., the brain).
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
74e4
RESEARCH ARTICLES
Ardipithecus ramidus

103. 103www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
sperm competition, regular non-ovulatory based
copulation (e.g.,Brachyteles), and minimal-to-
moderate body-size dimorphism. Minimal male-
to-male agonism in some genera probably stems
not only from male philopatry, but also from the
logistic difficulty of defining and defending ex-
clusively arboreal territories absent special spacing
mechanisms and/or dietary adaptations present in
related, but less K-adapted genera (e.g., howlers)
(69,70).
Exploitation of ground-based resources and
terrestrial travel to arboreal ones has encouraged
substantial dietary reliance on low herbaceous
vegetation in African apes. This has been accom-
panied by larger home ranges, more intense
territoriality, or both. InGorilla, home range ap-
pears to have been secondarily reduced, driven
by the body size–dietary axis. These dietary
specializations have led to a substantially reduced
sex ratio (one to three males per group) and mate
guarding inGorilla, fission/fusion inP. paniscus,
and aggressive multimale patrolling, as well as
fission/fusion, inP. troglodytes. However, each
of these mechanisms substantially discourages
male parental investment.
Hominids did not evolve any of the highly
derived African ape characters associated with
intense male-to-male agonism, reliance on near-
ground and terrestrial herbivory, or arboreal
frugivory (25). Moreover, sperm competition ap-
pears to be vestigial in humans [e.g., retained
pendulous scrotum, no pseudogenization of
SEMGI (52 )]. Elaborate periovulatory estrus
signaling is therefore most likely aPanspecial-
ization evolved to facilitate female transfer vis-à-
vis extreme group territoriality and a defense
against potential infanticide (71,72), as well as
potentially a product of male mate choice in a
context of intense sperm competition.
Thus, the hominid and African ape clades
evolved wholly divergent social, locomotor, and
dietary strategies. Whereas some apes appear to
have increased their reliance on terrestrial herba-
ceous vegetation as early as 10.5 Ma (73), the
early hominid dentition remained more gen-
eralized (22,25). What unique advantages, then,
did bipedality afford only the hominid clade, and
how might this unique locomotor pattern be
evolutionarily related to elimination of the SCC?
Bipedality and the SCC.Parsimony requires
that most, if not all, specialized human attributes
emerged within an integrated adaptive constel-
lation, presumably in the same manner as trait
complexes in other vertebrates. Figure 2 details
one possible adaptive suite as it might have
emerged at the base of the hominid clade. Fac-
ultative bipedality, a generalized dentition, and
enamel isotopic data ofAr. ramidusdemonstrate
that early hominids continued to exploit both
terrestrial and arboreal resources, but in a man-
ner wholly different from those used by extant
PanandGorilla(25).
Terrestrial resources are more defensible than
arboreal ones (74). Terrestriality has obviously
elevated resource warding and extra-group male-
to-male agonism inPanand in largely single-
maleGorilla. The elimination of the SCC in very
early hominids, however, suggests that resource
guarding was not feasible. Territories too large for
successful defense have numerous correlates (e.g.,
patchy resources, elevated search time, enlarged
core areas, and increased predator risk) (35 ).
Each of these substantially compromises par-
enting efforts.
A unique advantage of bipedality is that it
permits food transport over long distances, a
behavior not generally feasible for an arboreal or
quadrupedal hominoid. Bipedality also facilitates
the regular use of rudimentary tools, both as
carrying devices and as implements for resource
exploitation. In a partially ateline-like social struc-
ture (but lacking extreme anatomical adaptations
to suspension) coupled with a likely early hominid
ecological context, females might readily have
employed the frequentPanstrategy of exchanging
copulation for important food (11,75,76) (e.g.,
especially valuable meat or fruits high in fat and/or
protein), particularly if such items required pro-
tracted search time. If obtained by male exploi-
tation of day ranges logistically too large for
territorial defense or for effective optimal forag-
ing by females with dependent infants, such
dietary items may have become pivotal (77 ).
The role of tools has, of course, long been
a tempting explanation for upright walking (78).
However, it is now known that habitual bipedal-
ity evolved millions of years before any evidence
of stone tools. Despite the potential facility of
crude implements of any kind to improve ex-
tractive foraging, it remains unlikely that such
simple implements would have, alone, been
sufficiently critical to reproductive success to
have required adaptations (bipedality) that would
have simultaneously restricted access to the equal-
ly important arboreal resource base. Moreover,
both capuchins and chimpanzees effectively
transport tools without any reorganization of their
postcrania.
On the other hand, the common mammalian
and avian strategy of provisioning provides myr-
iad benefits directly associated with reproductive
success (32,79). Females and their offspring
enjoy reduced predation risk, and males benefit
from intensified mothering of their offspring. In
such a context at the base of the hominid clade,
temporary pair bonds based on sex-for-food ex-
changes would have further encouraged copula-
tion with provisioning males, rather than males
that relied on dominance or aggressive displace-
ment of competitors abetted by large and pro-
jecting canines. Research has confirmed the
selective advantages of such exchanges inPan
(11,80). Even controlling for rank and age, chim-
panzee males that practice meat-for-sex exchanges
have elevated fitness levels, and provisioning on
a long-term basis improves reproductive success,
even after controlling for estrous state (81).
Preference for a dominant male is an obvious
female strategy, but it becomes increasingly less
favorable when prolonged subadult dependency
requires intensified parenting. Under such condi-
tions, survivorship increasingly dominates fecun-
dity. Basal hominid females may have become
progressively more solicitous of smaller-canined
(and thereby less agonistically equipped) males,
particularly if they could encourage such males to
habitually target them in preference to other fe-
males. Temporary or occasional coupling [includ-
ing honeymoon pairs (80 )] and male choice of
particular females for such targeted provision-
ing would have increased their probabilities of
both paternity and subsequent offspring survi-
vorship, which is exceptionally valuable to the re-
productive success of both participants.
Any mammal species undergoing advanced
K modifications must eventually approach a limit
at which male parental investment becomes
virtually mandatory.
Typically, male mammals…do not form
bonds with offspring or mates, and their
social relationships are characterized by
aggressive rather than affiliative behav-
ior. However, in <5% of mammalian
species…ecological demands, such as
patchy resource distribution, a low pop-
ulation density of females, or increased
predation risk, mean that a promiscuous
strategy is not possible. In such species,
males are monogamous and contribute to
offspring care to safeguard their invest-
ment in reproduction…. Although it might
seem that the evolution of monogamy in
males would require a major reorganiza-
tion of the brain, recent research has shown
that the transition from promiscuity to mo-
nogamy might have required relatively triv-
ial mechanistic changes (82), p. 562.
Late Miocene hominoids probably faced a
virtual perfect storm of disparate ecological de-
mands. Increased omnivory elevated search time
and exposure to predators. Prolonged lactation
amenorrhea made ovulating females increasingly
rare because birth spacing was progressively
prolonged. What solutions to this dilemma could
selection offer? Males might cooperate with kin
to aggressively expand their territories and gain
greater access to additional reproductive females
(e.g.,Pan), especially if they developed locomo-
tor skills (vertical climbing) that allowed them to
rely on high canopy resources and promoted
access to ripe fruit. Alternatively, males might
aggressively displace all or most others, even if
kin-related, to optimize male-to-female ratios
(e.g.,Gorilla), especially if diet also permitted
minimization of day path length so as to prevent
female dispersal during feeding.
A third possibility would have been prolifera-
tion of sex-for-food exchanges. These would have
made provisioning an available solution for both
sexes and would have heightened female prefer-
ence for nonaggressive, provisioning males with
which to have repeated copulations. Unlike the
circumstances in the first two solutions (Gorilla
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
74e5
Ardipithecus ramidus
Research Articles

104. 104 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
andPan), in which the SCC would have been
under positive selection pressure, the SCC would
have been under moderate negative selection in
such a clade, because canine retention would have
discouraged provisioning in favor of retaining ago-
nistic strategies of mate acquisition.
Reproductive crypsis: the most unique human
character.Elimination of the SCC and the eco-
logical context ofArdipithecusat Aramis, Ethi-
opia, and earlier sites are consistent with the
inference that male provisioning via resource
transport (and concomitant terrestrial bipedality)
antedated 4.4 Ma. Might such behaviors have first
evolved nearer the base of the hominid clade? An
obvious issue with the hypothesis outlined above
is thatPanmales prefer females with prominent
signs of active ovulation (estrus). If minimal ovu-
latory signaling in the earliest hominids was prim-
itive (as it is presumably inGorilla), why did
hominid females not prolong and intensify such
signs so as to encourage sex-for-food exchanges?
First, the extreme ovulatory-related displays
inPanappear to be derived, because they are
associated with comparatively unique molecular
signatures of accompanying adaptations (such as
proteins necessary for vaginal plugging) absent
in other hominoids, as well as appropriately
specialized penile morphology. Second, it is un-
likely that copulation offered by a female would
be rejected by a male—this would be counter-
productive given the substantial variability of the
primate menstrual cycle and the rarity of homi-
noid females available for impregnation. Further-
more, habitual provisioning of a target female,
even while still lactating for a dependent infant,
would still make the repeatedly attendant male
most likely to sire any successive offspring upon
first reinstatement of ovulatory cycling.
The latter point is critical. One of the most
frequently cited objections to male provisioning
in early hominids is the problem of cuckoldry
during times that males would have been sep-
arated from a selected mate while in search of
food (83,84). But ovulation in hominoids is an
exceptionally rare event, and it probably occurs
only after extensive, 3- to 4-year-long periods
during which female lactation amenorrhea
prevents it. Male provisioning of rarely but ob-
viously fertile females would enhance his fitness
by several means: (i) Regular copulation would
probabalistically establish an attendant male as
the most likely to sire the target female’s suc-
ceeding offspring, provided that his mate did not
“advertise”her ovulation and/or solicit multiple
copulations. (ii) Repeated provisioning would
accelerate reinstatement of ovulation by replen-
ishing fat stores depleted by lactation. (iii) Acci-
dental or pathological death of her dependent
offspring (a not infrequent event) would also re-
initialize ovulation, and selection would obviously
favor habituation with nonaggressive males not
predisposed to infanticide, which is already un-
likely because of philopatry.
To prevent cuckoldry, male provisioning
within the context of a multimale group there-
fore requires selection of females with repro-
ductive crypsis. That is, males could only
succeed by provisioning mates with self-crypsis;
they would otherwise be unprotected from female
copulation with more dominant/aggressive males
while ovulating. Broadly (but not entirely) non-
ovulatory copulation, as inBrachyteles(69,70),
would permit prolonged exclusivity in operational
pair bonds, especially when provisioning males
showed preference for females who were not ob-
served to copulate with other males (85 ). In this
context, it is therefore relevant that human fe-
males do not externally advertise ovulation [other
hominoids exhibit some degree of ovulatory swell-
ing, even if minimal (86 )] and also fail to exhibit
its substantial physiological self-perception, de-
spite moderately elevated proceptivity during
ovulation (39,87).
The neurophysiology of mate choice.Pair
bonding is rare among mammals (~5%). A com-
mon criticism of an adaptive suite similar to that
shown in Fig. 2 is that the transition to such a
reproductive strategy would be behaviorally un-
likely, even if it did confer the major reproductive
benefits detailed above. But the recently discovered
relation between brain neurophysiology and mating
behavior in mammals may provide a rebuttal. In
particular, the expression patterns of the receptors
for the neuropeptides oxytocin (OT), arginine vaso-
pressin (AVP), and prolactin (PRL) are now known
to substantially influence mating and parenting be-
haviors (82,88). Monogamous prairie voles exhibit
distinct OT and AVP receptor distributions within
the mesolimbic dopaminergic reward pathway [i.e.,
ventral tegmental area (VTA), ventral pallidum, and
nucleus accumbens]. This is the central corridor
that is activated in human cocaine addiction, and
the transient actions of OT within this pathway are
critical to establish mothering behavior in non-
monogamous females.
Both OT and AVP are released centrally
during sexual stimulation. Their receptors, abun-
dantly expressed in critical brain areas of monog-
amous prairie voles (but only minimally so in
polygynous montane voles), are activated in con-
cert with dopamine release. This promotes asso-
ciative relations with other neural signaling,
especially stimuli emanating from the olfactory
bulb, affecting the medial nucleus of the amygdala,
and resulting in the formation of a pair bond.
Because OT receptor up-regulation in the ventral
forebrain occurs before parturition and mediates
mother-to-infant bonding, this pathway has prob-
ably been co-opted as a means of encouraging
monogamy, given the probable homology of
mammalian neuroendocrine circuitry in both sexes
(89,90).
AVP receptor distributions in monogamous
marmosets (91 ,92) and titi monkeys (93), as well
as in polygnous rhesus macaques (94), parallel
those observed in monogamous and promiscuous
voles, respectively, confirming that this reward
pathway functions similarly in primates. Although
receptor distributions such as those now available
for some monkey species are not yet available for
humans, there are marked parallels in other related
cerebral phenomena revealed by functional mag-
netic resonance imaging (see also below). Brain
activity patterns in women who looked at photo-
graphs of men with whom they were in love
“looked remarkably similar to those observed
after cocaine orm-opiod infusions with heavy
activation of the VTA and striatal dopamine
regions”[(88), p. 1053]. As predicted, similar
patterns were evoked by photographs of their
children. PRL concentrations, also involved in
the reward pathway, are strongly up-regulated
in both rodents and callitrichid males exhibit-
ing paternal care, but not in species lacking it;
PRL is elevated in human males immediately
before the birth of their first child (95 ).
Equally notable is the impact of steroid
hormones on paternal behavior in rodents and
primates, including humans. Testosterone con-
centrations are suppressed in males by parturition
in species with extensive paternal care, including
numerous rodents, callitrichids, and humans.
Such reductions may prevent aggression toward
infants. Estradiol and progesterone, critical to
normal maternal behavior, have not yet been
surveyed in nonhuman primate males but are
known to be elevated in human fathers (96 ).
An early hominid adaptive suite.On the
basis of their relatively advanced states in
Ardipithecus, two of the three primary characters
unique to hominids (bipedality, loss of SCC)
probably extend well back into the Miocene,
perhaps almost to the time of the CLCA. The
emergence of these characters in combination is
consistent with a strategy of increasingly targeted
provisioning, as outlined in Fig. 2. Males would
benefit from enhanced male-to-male cooperation
by virtue of their philopatry, because it would
improve not only their own provisioning capac-
ity, but also that of their kin. Foraging could be
achieved most productively by cooperative male
patrols (homologous to but strategically entirely
unlike those ofPan). Provisioning would reduce
female-to-female competition by lowering reli-
ance on individual“sub-territories”(as in chim-
panzees) and/or resource warding (97 ) and
would improve (or maintain) social cohesion.
Fission/fusion of social groups would also be
reduced, ameliorating likelynovel predation risk
and enhancing the stability of core areas. Further
musculoskeletal adaptation to terrestrial bipedality
would be imposed by the need to carry harvested
foods, simple tools for extractive foraging, and
eventually altricial offspring lacking pedal grasp-
ing capacity consequent to the adoption of per-
manent bipedality without a substantial arboreal
component (as inAustralopithecus).
The third primary character shown in Fig. 2,
female reproductive crypsis, cannot be directly
traced in the fossil record. What can we surmise
of its evolutionary history? As noted earlier, a
central component of reproductive crypsis is the
loss of visually prominent mammary gland cycling
(i.e., concealed by permanent fat stores that sim-
ulate lactating glands) in humans. A common
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org
74e6
RESEARCH ARTICLES
Ardipithecus ramidus

105. 105www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
explanation for permanently enlarged human
mammae is that they serve as a male attractant
because they may signal adequate fat stores for
reproduction (98). But why would an attractant
be required when female proceptivity is the only
limiting factor acting on all other primate males
(no matter what the underlying social system)?
Again, as noted earlier, the elaborate periovula-
tory sexual swellings ofPanare an integral com-
ponent of intense sperm competition, which
hominids clearly lack. Moreover, whereas the
loss of mammary cyclicity would be unlikely to
evolve inPan[copulation with lactating females
is rare (99)], crypticism would not be a barrier in
a context of copulatory vigilance within pair
bonds (32,79). Moreover, elimination of cyclicity
would protect a provisioning (and thereby heavily
invested) male from cuckoldry, because promi-
nent mammaries would discourage interest by
extra-pair males. The absence of cycling would
simultaneously protect females from potential
abandonment (79).
An element of human reproductive crypsis
not discussed earlier is the reduction of a male’s
capacity to detect ovulation via olfactory sig-
naling. It is again difficult to ascertain why
selection would directly favor a precipitous loss
in olfactory capacity. Yet the loss of olfactory
receptors has occurred in the human lineage at a
much faster rate than in other higher primates
(100,101) and is fully explicable as a product of
female choice acting within the context of a
provisioning strategy. If males could detect ovu-
lation in this manner, provisioning would almost
certainly have accompanied such detection, just
as it does inPanwhen ovulation is so acutely
advertised. Ovulatory crypsis would therefore be
a key element in maintaining targeted provision-
ing by a particular (pair-bonded) male.
These kinds of unique, reproductively related
characters are often broadly ascribed to an in-
tensification of human social behavior during the
Pleistocene (by largely undefined selective
mechanisms) or have simply been ignored. But
why should we simply presume that these
various soft tissue structures and physiologies
were not present inAustralopithecus, or even in
Ardipithecus, particularly when the latter shows
that the CLCA was not morphologically or be-
haviorally chimpanzee-like? Relegating these
derived characters toHomoalmost certainly
requires each to be assigned causation in near
total isolation. One of the instructive aspects of
adaptive suites is the demonstration of what must
almost always be a complex network of character
interaction, even in reptiles and amphibians.
More often than not, such interconnectivity is
likely to far exceed relatively simplistic argu-
ments such as somatic budgeting.
Viewing the sweep of hominid evolution in
retrospect, it is increasingly unlikely that upright
walking, elevation of skeletal dimorphism (in
Australopithecus) despite simultaneous elimina-
tion of the SCC, loss of vocal sacs, precipitous
reduction in olfactory receptors, development of
permanently enlarged mammary glands, loss of
ovulatory-based female proceptivity, precipitous
reduction in male fertility, unique maintenance of
a pendulous scrotum despite substantial reduction
in testes size, proliferation of epigamics (sex-
related traits used for male selection) in both sexes
[implying mate choice in each (32,79)], and un-
paralleled demographic success in a terrestrial
primate have all been incidental and unrelated.
These are far more likely to be multiple elements
within a unique reproductive strategy that allowed
early hominids to thrive relative to their ape
relatives and could have ultimately accommo-
dated rapid development of the unusually energy-
thirsty brain of subadults in emergentHomo.
Yet a large brain is not our most unique
characteristic. Chimpanzees have relatively larger
brains than cercopithecoids, which have rela-
tively larger brains than lemurs. However, the
combination of SCC elimination, habitual bipe-
dality, and reproductive crypsis (each in itself an
extreme rarity) is unique among all known
mammals. Conversely, simple brain enlargement
is readily explicable in myriad ways. If, for ex-
ample, the acquisition and control of fire was
somehow a causative factor, as has recently been
suggested (102), what relations does this singular
capacity have to the broad array of other entirely
unique human characters that are known to have
preceded it in the fossil record? Moreover, does
the marked expansion of the human brain itself
not signal a unique reproductive strategy rather
than a simple physical character or capacity?
Among the apes, hominids alone were successful
before the major cultural advances of the Pleis-
tocene, and Oldowan stone tools persisted
unchanged for almost 1 million years. The mo-
lecular and cytological records suggest that hom-
inid cerebral evolution extends deep into time, as
extrapolated from the likely evolutionary pro-
gression in genes such as abnormal spindle-like
microcephaly associated gene (ASPM)(103).
The reconstructed history of its evolution sug-
gests marked acceleration“along the entire
lineage from the last ape ancestor to modern
humans…[implying that] the human phenotype
did not arise abruptly…but [is] instead the
consequence of a lengthy and relatively contin-
uous process”[(104), pp. 491–492].
Conclusion.AsAu. afarensiswas progres-
sively revealed during the 1970s, its anatomy and
antiquity still permitted a possible chimpanzee-
like CLCA. Many models of human origins,
largely referential, employed this perspective. Pre-
vious nonreferential attempts (32,79) argued that
only major changes in the social behavior of
Au. afarensisand its ancestors could satisfac-
torily account for its unique combination of
postcranial anatomy and unusual demographic
success. The tempo and mode of such hypothet-
ical earlier evolutionary events, however, have
remained shrouded from our view. This has led
to rejection of the hypothesis by many who pre-
ferred the comparative comfort and safety of
more referential accounts.
Even as its fossil record proliferated, how-
ever,Australopithecuscontinued to provide only
an incomplete understanding of hominid origins.
Paradoxically, in light ofArdipithecus, we can
now see thatAustralopithecuswas too derived—
its locomotion too sophisticated, and its invasion
of new habitats too advanced—not to almost
entirely obscure earlier hominid evolutionary
dynamics.
Now, in light ofAr. ramidus, there are no
longer any a priori reasons to suppose that ac-
quisition of our unique reproductive anatomy and
behavior are unconnected with other human
specializations. The evidence is now conclusive:
Elimination of the SCC occurred long before the
eventual dentognathic hypertrophy ofAustralo-
pithecus, and long before the likely horizon at
which sufficient reliance on tool use would have
encouraged abandonment of food and/or safety
in the arboreal substrate. It is far more likely that
our unique reproductive behavior and anatomy
emerged in concert with habituation to bipedality
and elimination of the SCC (Fig. 2). It is also now
equally clear thatPan’s specialized reproductive
constellation has been driven by an entirely dif-
ferent locomotor and dietary history.
We currently know very little about the
postcranium of hominids older thanAr. ramidus
(e.g.,Sahelanthropus,Orrorin)(105,106). More
fossils will further advance our understanding of
the CLCA, and we anxiously await their dis-
covery. Meanwhile, the opportunity of devising
adaptive suites for both species ofPanand for
Gorilla—grounded in hypotheses generated in
light now thrown on the gorilla/chimpanzee/
human last common ancestor and CLCA byA r.
ramidusas to their locomotion, diet, and social
behavior—is an intriguing prospect whose alter-
native outcomes will probably provide addition-
al revelations.
When viewed holistically, as any adaptive
suite requires, the early hominid characters
that were probably interwoven by selection to
eventually generate cognition now seem every
bit as biologically ordinary as those that have
also affected the evolution of lizards, frogs,
voles, monkeys, and chimpanzees. Comparing
ourselves to our closest kin, it is somewhat
sobering that the hominid path led to cogni-
tion, whereas that leading toPan, our closest
living relatives, did not, despite the near-synonymy
of our genomes.
As Darwin argued, the ultimate source of
any explication of human acumen must be
natural selection (78). The adaptive suite pro-
posed here provides at least one evolutionary
map by which cognition could have emerged
without reliance on any special mammalian trait.
The perspective offered byArdipithecussuggests
that our special cognitive abilities derive from a
unique earlier interplay of otherwise common-
place elements of locomotion, reproductive bi-
ology, neurophysiology, and social behavior. In
retrospect, we are as ordinary as corvids (107)
and voles (108), although we are much more
www.sciencemag.orgSCIENCEVOL 326 2 OCTOBER 2009
74e7
Ardipithecus ramidus
Research Articles

106. 106 www.sciencemag.org   SCIENCE   VOL 326   2 October 2009   Published by AAAS
fortunate, if self-cognition is deemed fortunate.
We should never have doubted Darwin in his
appreciation that the ultimate source of our
matchlessness among mammals would prove
commonplace when knowledge became suffi-
ciently advanced.Ar. ramidusnow enhances that
knowledge. Even our species-defining coopera-
tive mutualism can now be seen to extend well
beyond the deepest Pliocene.
References and Notes
1. J. Moore, inGreat Ape Societies, W. C. McGrew,
L. Marchant, T. Nishida, Eds. (Cambridge
Univ. Press, Cambridge, 1996), pp. 275–292.
2. D. Pilbeam,Mol. Phylogenet. Evol.5, 155 (1996).
3. D. Pilbeam,J. Exp. Zool.302B, 241 (2004).
4. R. Wrangham, D. Pilbeam, inAll Apes Great and Small
Volume I: African Apes, B. Galdikaset al., Eds.
(Kluwer Academic/Plenum, New York, 2001), pp. 5–17.
5. R. A. Dart,Nature115, 195 (1925).
6. D. C. Johansonet al.,Am. J. Phys. Anthropol.57, 403 (1982).
7. T. D. White, G. Suwa,Am. J. Phys. Anthropol.72, 485 (1987).
8. R. A. Dart,Transvaal Mus. Memoir. (Pretoria)10,
1 (1957).
9. P. E. Wheeler,J. Hum. Evol.21, 107 (1991).
10. K. D. Hunt,Int. J. Primatol.12, 95 (1991).
11. C. Stanford,Upright: The Evolutionary Key to Becoming
Human(Houghton Mifflin, Boston, 2003).
12. N. G. Jablonski, G. Chapman,J. Hum. Evol.24, 259 (1993).
13. J. G. Fleagleet al.,Symp. Zool. Soc. London48, 359 (1981).
14. M. D. Sockol, D. A. Raichlen, H. Pontzer,Proc. Natl.
Acad. Sci. U.S.A.104, 12265 (2007).
15. J. T. Stern Jr., R. L. Susman,Am. J. Phys. Anthropol.60,
279 (1983).
16. C. O. Lovejoy,Gait Posture21, 95 (2005).
17. C. O. Lovejoy,Gait Posture21, 113 (2005).
18. C. O. Lovejoy,Gait Posture25, 325 (2007).
19. T. D. Whiteet al.,Science326, 64 (2009).
20. C. O. Lovejoyet al.,Science326, 71 (2009).
21. C. O. Lovejoyet al.,Science326, 72 (2009).
22. C. O. Lovejoyet al.,Science326, 73 (2009).
23. C. O. Lovejoyet al.,Science326, 70 (2009).
24. G. Suwaet al.,Science326, 68 (2009).
25. G. Suwaet al.,Science326, 69 (2009).
26. G. WoldeGabrielet al.,Science326, 65 (2009).
27. T. D. Whiteet al.,Science326, 67 (2009).
28. A. Louchartet al.,Science326, 66 (2009).
29. N. S. Clayton, J. M. Dally, N. J. Emery,Philos. Trans.
R. Soc. London Ser. B Biol. Sci.362, 507 (2007).
30. N. J. Emery, A. M. Seed, A. M. von Bayern, N. S. Clayton,
Philos. Trans. R. Soc. London Ser. B Biol. Sci.362, 489
(2007).
31. K. Sayers, C. O. Lovejoy,Curr. Anthropol.49, 87 (2008).
32. C. O. Lovejoy,Science211, 341 (1981).
33. K. Milton, inFood and Evolution: Toward a Theory of
Human Food Habits, M. R. E. Harris, Ed. (Temple Univ.
Press, Philadelphia, PA, 1987), pp. 93– 115.
34. R. M. Alexander, inBiomechanics in Evolution,
J. M. V. Raynor, R. J. Wootton, Eds. (Cambridge
Univ. Press, Cambridge, 1991), pp. 255–266.
35. E. R. Pianka,Evolutionary Ecology(Harper and Row,
New York, ed. 4, 1988).
36. E. R. Pianka, W. S. Parker,Copeia1975, 141 (1975).
37. E. R. Pianka, inLacertids of the Mediterranean Basin,
E. D. Valakos, W. Bohme, V. Perez-Mellado, P. Maragou,
Eds. (Hellenic Zoological Society, Univ. of Athens,
Athens, Greece, 1993), pp. 121–154.
38. W. Weygoldt,Zool. Syst. Evolutionsforsch.25, 51 (1987).
39. A. F. Dixson,Primate Sexuality(Oxford Univ. Press,
Oxford, 1998).
40. M. P. Ghiglieri,J. Hum. Evol.16, 319 (1987).
41. A. H. Harcourt, P. H. Harvey, S. G. Larson, R. V. Short,
Nature293, 55 (1981).
42. R. V. Short, inReproductive Biology of the Great Apes:
Comparative and Biomedical Perspectives, C. E. Graham,
Ed. (Academic Press, New York, 1981), pp. 319–341.
43. K. Milton,J. Behav. Ecol. Sociobio.17, 53 (1985).
44. Measures include volume, sperm concentration, total
sperm per ejaculate, and sperm motility (45).
45. A. P. Moller,J. Hum. Evol.17, 479 (1988).
46. E. J. Peirce, W. G. Breed,Reproduction121, 239 (2001).
47. A. F. Dixson,J. Zool. (London)213, 423 (1987).
48. A. F. Dixson, inThe Integrative Neurobiology of
Affiliation, C. S. Carter, I. I. Lederhendler, B. Kirkpatrick,
Eds. (New York Acadamy of Sciences, New York, 1997),
pp. 42– 61.
49. A. F. Dixson,Folia Primatol. (Basel)64, 146 (1995).
50. M. Freund,J. Reprod. Fertil.4, 143 (1962).
51. M. J. Anderson, A. F. Dixson,Nature416, 496 (2002).
52. S. B. Kingan, M. Tatar, D. M. Rand,J. Mol. Evol.57, 159
(2003).
53. A. F. Dixson, M. J. Anderson,Folia Primatol. (Basel)73,
63 (2002).
54. C. L. Nunn, J. L. Gittleman, J. Antonovics,Science290,
1168 (2000).
55. J. M. Plavcan,J. Hum. Evol.39, 327 (2000).
56. S. L. Washburn,Sci. Am.203, 63 (1960).
57. M. H. Wolpoff,Curr. Anthropol.17, 579 (1976).
58. R. L. Smith, inSperm Competition and the Evolution of
Animal Mating Systems(Academic Press, Orlando, FL,
1984), pp. 601–659.
59. Flaccid human penis length (13 cm) is unusually great
for a hominoid. Length is ~4 cm inPongoand 3 cm in
Gorilla. Its erect size is greater in the multimalePan
(8 cm), but this reflects specialized adaptation to
penetrate seminal plugs. Short notes that“(e)ven the
pubic hair in the male [human] seems designed to draw
attention to the genitalia, rather than to conceal them as
in the orangutan and gorilla”[(42) p. 338].
60. In three of four fatal attacks by chimpanzee males on
other males at Ngogo and Kanyawara, Uganda (109),
one or both testes were described as having been
“ripped from scrotum.”See also www.esquire.com/
features/chimpanzee-attack-0409.
61. In (110), Alemsegedet al. assert that“[t]he
bulla-shaped body [of its preserved hyoid] almost
certainly reflects the presence of laryngeal air sacs
characteristic of African apes”(p. 300). However, there
is no correlation between hyoid inflation and the
presence or absence of air sacs in primates. For
example,Cercopithecus aethiopslacks air sacs, whereas
Cercopithecus mitisexhibits them. We have observed
that the hyoid bodies of both species are equally inflated
(collections in the U.S. National Museum, Washington,
DC, and the Royal Museum for Central Africa, Tervuren,
Belgium). See also (111).
62. J. Biegert, inClassification and Human Evolution, S. L.
Washburn, Ed. (Aldine, Chicago, 1963), pp. 116–145.
63. P. L. Reno, R. S. Meindl, M. A. McCollum, C. O. Lovejoy,
J. Hum. Evol.49, 279 (2005).
64. A. B. Taylor,Am. J. Primatol.43, 1 (1997).
65. W. C. Hartwig,The Primate Fossil Record
(Cambridge Univ. Press, Cambridge, 2002).
66. D. R. Begunet al.,Science287, 2375a (2000).
67. J. Kelley, T. M. Smith,J. Hum. Evol.44, 307 (2003).
68. T. H. Clutton-Brock,Nature337, 70 (1989).
69. C. B. Possamai, R. J. Young, S. L. Mendes, K. B. Strier,
Am. J. Primatol.69, 766 (2007).
70. K. B. Strier, T. E. Ziegler,Am. J. Primatol.42, 299 (1997).
71. R. W. Wrangham, N. L. Conklin, C. A. Chapman,
K. D. Hunt,Philos. Trans. R. Soc. London Ser. B Biol. Sci.
334
, 171 (1991).
72. M. E. Rogerset al.,Am. J. Primatol.64, 173 (2004).
73. G. Suwa, R. T. Kono, S. Katoh, B. Asfaw, Y. Beyene,
Nature448, 921 (2007).
74. In (112), Mitani and Rodman devised an index of
territoriality that has some predictive success. Substantial
increases in day path length and home range size
make resource guarding increasingly unlikely.
75. P. Gagneux, C. Boesch, D. S. Woodruff,Anim. Behav.57,
19 (1999).
76. F. B. M. de Waal,J. Hum. Evol.18, 433 (1989).
77. M. E. Thompson, R. W. Wrangham,Am. J. Phys.
Anthropol.135, 171 (2008).
78. C. Darwin,The Descent of Man,and Selection in Relation
to Sex(John Murray, London, 1871).
79. C. O. Lovejoy, inThe Origins of Humans and Humanness,
D. T. Rasmussen, Ed. (Bartlett and Jones, Los Angeles,
1993), pp. 1–28.
80. C. E. G. Tutin, P. R. McGinnis, inReproductive Biology
of the Great Apes, C. E. Graham, Ed. (Academic Press,
New York, 1981), pp. 239–264.
81. C. M. Gomes, C. Boesch,PLoS One4, e5116 (2009).
82. J. P. Curley, E. B. Keverne,Trends Ecol. Evol.20, 561 (2005).
83. R. Wrangham, inTree of Origin: What Primate Behavior
Can Tell Us About Human Social Evolution, F. de Waal,
Ed. (Harvard, Cambridge, MA, 2001), pp. 119–143.
84. M. Cartmill, F. H. Smith,The Human Lineage
(Wiley-Blackwell, Hoboken, NJ, 2009).
85. A. G. Ophir, B. G. Galef Jr.,Anim. Behav.68, 1221 (2004).
86.“Many ...[have assumed] that our hominid ancestors
had sexual swellings that were lost over the course of
human evolution, some say to promote pair bonding....
[S]wellings are more likely to have been independently
evolved in the line leading to chimps and bonobos.
Sexual swellings are scarcely detectable in the other
Great Apes or among the last surviving relicts of the
genusHomo—modern humans.‘Concealed ovulation’
(or more accurately, ovulation without advertising) in our
own species is not so much a new trait as an extreme
manifestation of an old one.” [(113), p. 82].
87. G. Miller, J. M. Tybur, B. D. Jordon,Evol. Hum. Behav.
28, 375 (2007).
88. L. J. Young, Z. Wang,Nat. Neurosci.7, 1048 (2004).
89. K. E. Wynne-Edwards,Horm. Behav.40, 139 (2001).
90. K. E. Wynne-Edwards, C. J. Reburn,Trends Ecol. Evol.15,
464 (2000).
91. Z. Wanget al.,Brain Res.768, 147 (1997).
92. A. Schorscher-Petcu, A. Dupré, E. Tribollet,Neurosci.
Lett.461, 217 (2009).
93. K. L. Bales, W. A. Mason, C. Catana, S. R. Cherry,
S. P. Mendoza,Brain Res.1184, 245 (2007).
94. L. J. Young, D. Toloczko, T. R. Insel,J. Neuroendocrinol.
11, 291 (1999).
95. A. E. Storey, C. J. Walsh, R. L. Quinton, K. E. Wynne-Edwards,
Evol. Hum. Behav.21, 79 (2000).
96. K. E. Wynne-Edwards, C. J. Reburn,Trends Ecol. Evol.15,
464 (2000).
97. J. Scott, J. S. Lockard,Primates47, 199 (2006).
98. D. Morris,The Naked Ape: A Zoologist’s Study of the
Human Animal(Random House, New York, 1967).
99. W. B. Lemmon, M. L. Allen,Folia Primatol. (Basel)30,
80 (1978).
100. Y. Gilad, O. Man, G. Glusman,Genome Res.15, 224 (2005).
101. D. M. Stoddart,Man21, 514 (1986).
102. R. Wrangham,Catching Fire: How Cooking Made Us Human
(Basic Books, New York, 2009).
103. S. Doruset al.,Cell119, 1027 (2004).
104. P. D. Evanset al.,Hum. Mol. Genet.13, 489 (2004).
105. Y. Haile-Selassie, G. Suwa, T. D. White, in Ardipithecus
kadabba: Late Miocene Evidence from the Middle Awash,
Ethiopia(Univ. of California Press, Berkeley, CA, 2009),
pp. 159–236.
106. Y. Haile-Selassie, G. Suwa, T. D. White,Science303,
1503 (2004).
107. K. Isler, C. P. Van Schaik,Biol. Lett.5, 125 (2009).
108. H. Walumet al.,Proc. Natl. Acad. Sci. U.S.A.105,
14153 (2008).
109. D. P. Watts, M. Muller, S. J. Amsler, G. Mbabazi,
J. C. Mitani,Am. J. Primatol.68, 161 (2006).
110. Z. Alemsegedet al.,Nature443, 296 (2006).
111. G. Hewitt, A. MacLarnon, K. E. Jones,Folia Primatol.
(Basel)73, 70 (2002).
112. J. Mitani, P. Rodman,Behav. Ecol. Sociobiol.5, 241 (1979).
113. S. B. Hrdy,Ann. N.Y. Acad. Sci.907, 75 (2000).
114. A. H. Wright, A. A. Wright,Handbook of Frogs and Toads
(Comstock, Ithaca, NY, 1949).
115. I thank NSF (this material is based on work supported by
grant 9910344) for funding and T. D. White, G. Suwa,
M. A. McCollum, P. L. Reno, M. A. Serrat, B. A. Rosenman,
R. S. Meindl, M. A. Raghanti, K. Sayers, C. Thompson,
T. Gregory, and L. J. Hlusko for extensive discussions and editing
of this manuscript over the course of 5 years of revisions.
4 May 2009; accepted 31 August 2009
10.1126/science.1175834
2 OCTOBER 2009 VOL 326 SCIENCEwww.sciencemag.org 74e8
RESEARCH ARTICLES
Ardipithecus ramidus

107. Remote, isolated, catastrophic events occur across the globe that affect civilians, the
environment, indigenous rights, and more. The AAAS Science and Human Rights Program
uses geospatial technologies to broaden the ability of non-governmental organizations
to rapidly gather, analyze, and disseminate information in these times of crisis. And
this is just one of the ways that AAAS is committed to advancing science to support
a healthy and prosperous world. Join us. Together we
can make a difference. aaas.org/plusyou/humanrights
Geospatial Technologies
Human Rights

108. www.ScienceMag.org