Content uploaded by Christine Janis
Author content
All content in this area was uploaded by Christine Janis on Aug 04, 2014
Content may be subject to copyright.
110
Journal of Vertebrate Paleontology 22(1):110–121, March 2002
q
2002 by the Society of Vertebrate Paleontology
LOCOMOTOR EVOLUTION IN CAMELS REVISITED: A QUANTITATIVE ANALYSIS OF PEDAL
ANATOMY AND THE ACQUISITION OF THE PACING GAIT
CHRISTINE M. JANIS, JESSICA M. THEODOR*, and BETHANY BOISVERT†
Department of Ecology and Evolutionary Biology, Brown University, Providence, Rhode Island 02912,
Christine
p
Janis@Brown.edu
ABSTRACT—Extant camelids (llamas and camels) are unique among wild mammals in their regular employment of
a pacing gait. They also have a unique foot morphology, assumed to be an adaptation for this mode of locomotion:
their feet are secondarily digitigrade, with the loss of hooves and the addition of a broad foot pad. We examined 22
measurements of the metapodials and phalanges of camelids and ruminants with bivariate and multivariate analyses,
including 18 genera of extinct camelids from the Tertiary of North America. Extant camelids and ruminants were
clearly distinguishable from each other. Most extinct camelids showed some morphological features typical of extant
forms, five out of eighteen clustered with the extant camelids. Pacing may have evolved independently within the
subfamilies Camelinae and Protolabinae. Additionally, evolutionary change towards a condition resembling that of
extant camelids also occurred within the subfamilies Stenomylinae and Miolabinae. These parallel changes in camelid
locomotor anatomy occurred in the late Oligocene or early Miocene, preceding the formation of widespread open
grassland habitats in the late Miocene.
INTRODUCTION
Camelids, represented today by camels and llamas, are fas-
cinating and little-studied animals. Camelids are artiodactyls,
and they probably represent the extant sister-group to ruminants
(Webb and Taylor, 1980), although some molecular studies
would place them in a more basal position within the Artio-
dactyla (Gatesy et al., 1999; but see Norris, 1999). Camelids
are unique among artiodactyls in their secondarily digitigrade
foot posture, derived from an ancestral unguligrade condition,
with a splay-toed foot and the typical ungulate hooves replaced
by a broad foot pad. This foot morphology is characterized by
metapodials that are splayed at their distal end (Fig. 1A), and
also by the derived condition of the loss of the interdigital lig-
aments, allowing for the divergence of the third and fourth dig-
its (Webb, 1972).
Extant camelids are also distinguished from all other wild
mammals by their mode of locomotion, employing a true pac-
ing gait (i.e., a ‘‘running pace,’’ Hildebrand, 1976). The pace
resembles the trot, in which pairs of fore and hind legs are
moved together, with a period of suspension between the move-
ment of each fore-/hindlimb pair during which all four feet are
off the ground. However, in the trot, contralateral pairs of fore
and hind legs are moved simultaneously (e.g., left fore with
right hind), whereas the pace involves ipsilateral pairs of legs
(e.g., left fore with left hind). The pace prevents the fore and
hind leg on the same side from interfering with each other dur-
ing fast locomotion, allowing a longer stride length, and hence
a faster and more efficient mode of locomotion than the trot
(Howell, 1944; Webb, 1972; Gauthier-Pilters and Dagg, 1981).
Three distinct types of pace gait can be distinguished in extant
Camelus: a slow pace, where the animal may be supported for
four legs during some part of the locomotory cycle; a medium
pace, where the animal is only ever supported by two legs; and
a fast pace, where there are times during the cycle when no
legs are on the ground (Gauthier-Pilters and Dagg, 1981). The
* Present address: Ilinois State Museum, Research and Collections
Center, 1011 East Ash Street, Springfield, Illinois 62703.
† Present address: 265 Blackberry Hill Road, Berwick, Maine 03901.
fast pace is the one that would be termed a ‘‘running pace’’ by
Hildebrand (1976).
With the exception of some long-legged domestic dogs, and
some breeds of domestic horses (e.g., standardbreds, used in
harness races), camelids are the only mammals that use this
type of pacing gait in place of a trot. Some long-legged mam-
mals employ a slower, walking pace, also seen in camelids, in
which there is no period of suspension. The walking pace, used
by some long-legged ungulates (e.g., giraffe and gerenuk) and
the cheetah, is actually a modified fast version of the regular
lateral walk of mammals and is not the same as the running
pace of camelids (Hildebrand, 1976). Many long-legged mam-
mals, including camelids, giraffe, and cheetah, also employ a
lateral ‘‘rotary’’ gallop (involving ipsilateral pairs of limbs be-
ing moved in sequence), instead of the transverse gallop (in-
volving contralateral pairs of limbs) seen in most mammals.
Thus, while camelids are not unique in their use of lateral gaits,
they differ from other wild mammals in their use of the running
pace.
Webb (1972) provided an extensive discussion of the advan-
tages and disadvantages of the pacing gait in camelids. A prime
disadvantage is reduced lateral stability. Webb considered the
development of the broad foot pad and splay-footed digitigrade
stance to be adaptations to mitigate this instability, and listed
other morphological features that would aid in increasing lateral
stability. Camelids have a narrow chest with broad, flat ribs.
They also have enlarged the areas of attachment for the prox-
imal limb abductors, whose action would prevent the body from
collapsing towards the unsupported side. These areas include a
relatively large scapular spine and acromion process in the
shoulder girdle for attachment of the trapezius and deltoid mus-
cles; a large deltopectoral crest on the humerus for the insertion
of the deltoids; and a broadened greater trochanter of the femur
for the insertion of the gluteals and vastus lateralis. To this list
we add a laterally expanded ridge on the dorsal ilium for the
origin of the tensor fascia latae, and transverse processes of the
lumbar vertebrae that are more robust and more horizontal in
position than in ruminants, for the attachment of the longissi-
mus dorsi.
To trace the evolution of the pacing gait in camelids, Webb
(1972) used evidence from both limb morphology and a fossil-
111JANIS ET AL.—LOCOMOTOR EVOLUTION IN CAMELS
FIGURE 1. Right metatarsals of (A) a Pliocene camelid, Hemiauch-
enia vera, with fully modernized limb morphology (B) a ruminant ar-
tiodactyl, the Pleistocene reindeer Rangifer tarandus and (C) the late
Miocene protoceratid Synthetoceras tricornatus (modified from Frick,
1937; Frick and Taylor, 1968; and Patton and Taylor, 1971).
TABLE 1. Specimens measured of extant artiodactyls.
Species Specimen no.
Camelids
19 Camelus bactrianus
20 Camelus dromedarius
21 Lama glama
22 Lama guanicoe
23 Lama pacos
24 Vicugna vicugna
AMNH 14113
AMNH 14107
AMNH 35235
AMNH 143242
AMNH 6240
AMNH 468742
Tragulids
25 Hyemoschus aquaticus
26 Tragulus javanicus MCZ 6041
MCZ 3828
Giraffids
27 Giraffa camelopardalis
28 Okapia johnstoni AMNH 70016
MCZ 38015
Cervoids
29 Alces alces
30 Antilocapra americana
31 Axis porcinus
32 Capreolus capreolus
33 Cervus elaphus
MCZ 1661
MCZ 1776
MCZ 1703
MCZ 29806
MCZ 36679
34 Elaphurus davidianus
35 Mazama americana
36 Moschus moschiferus
37 Odocoileus virginianus
38 Rangifer tarandus
MCZ 8733
MCZ 49626
MCZ 6343
MCZ 59435
AMNH 5141
Bovids
39 Boselaphus tragocamelus
40 Capra hircus
41 Cephalophus sylvicultor
42 Damaliscus pygarus
43 Gazella gazella
44 Kobus leche
MCZ 6948
MCZ 42210
MCZ 58312
MCZ 5001
MCZ 54415
MCZ 56848
45 Ourebia ourebi
46 Ovis aries
47 Rupicapra rupicapra
48 Saiga tatarica
49 Sylvicapra grimmia
50 Taurotragus oryx
MCZ 5011
MCZ 50943
AMNH 24193
MCZ 5007
MCZ 5012
MCZ 1702
ized trackway attributed to the Miocene camelid genus Proto-
labis. He described features of the phalangeal anatomy of Pro-
tolabis and Michenia, noting changes in phalangeal proportions
and articular surfaces that he considered to be correlated with
the digitigrade stance typical of pacing camelids. A camelid
phylogeny, in which Protolabis was placed as basal to later
camelids, indicated the earliest evolution of the pacing gait.
We seek to quantify the unique features of camelid limbanat-
omy, and to compare camelids analytically with ruminant artio-
dactyls to see if the two fall into statistically distinguishable
groups. We also examine a broader range of fossil camelids to
determine which taxa display morphology similar to that of
extant camelids, which might be more like ruminants, and
which represent intermediate states. By examining a larger di-
versity of taxa we can refine the estimation of the time at which
camels may have first adopted a pacing gait. Finally, the avail-
ability of a comprehensive cladogram of fossil camelids (Honey
et al., 1998) enables us to plot these morphological changes on
a phylogeny, and determine if these morphological changes
evolved in parallel.
MATERALS AND METHODS
We made measurements on single individuals of all six spe-
cies of extant camelids (Table 1), 26 species of extant ruminants
(Table 1), and 18 taxa of extinct camelids from the Tertiary of
North America (Table 2) from the American Museum of Nat-
ural History, New York (AMNH) and the Museum of Com-
parative Zoology, Harvard University (MCZ). The fossil ca-
melid data also came from single individuals, the availability
of which formed the basis for our choice of taxa. The extant
taxa selected depended primarily on the availability of the spec-
imens that could provide the complete range of morphological
measures in a single individual. However, we restricted our in-
clusion of bovid species to approximately equal numbers in
comparison to the available cervoids, in order to avoid the pos-
sible phylogenetic bias that might occur if the majority of in-
cluded species represented a single family. The bovid species
were selected to represent a variety of body sizes and ecomor-
phological types. The fossil taxa are specimens from the De-
partment of Vertebrate Paleontology and the Frick Collection
at the AMNH, with the exception of Floridatragulus, which
was from the MCZ.
We also took measurements of the extinct Miocene proto-
ceratid Synthetoceras tricornatus as representative of a more
primitive artiodactyl limb morphology. We chose Synthetoceras
for this comparison for reasons of taxonomic affiliation, degree
of morphological specialization, and body size. Synthetoceras
serves here as a representative of a more generalized type of
limb morphology than the cursorially-derived camelids and ru-
minants, not as an outgroup taxon for a phylogenetic analysis.
Protoceratids are an extinct artiodactyl family, usually con-
sidered to be related to camels (Patton and Taylor, 1971; Webb
and Taylor, 1980), but more recent workers have thrown some
doubts on this affinity and suggested that they may in fact be
basal ruminants (Joeckel and Stavas, 1996; Norris, 2000). Thus
phylogenetically they could be considered as intermediate be-
112 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 1, 2002
TABLE 2. Morphological characteristics of extinct camelids. Key: Mets:
5
metatarsals. C
5
cameline-like; L
5
lamine-like; R
5
ruminant-
like; R/L;
5
morphology in between that of ruminants and lamines; L/C
5
morphology in between that of lamines and camelines. Note that
Honey et al. (1998) considered ‘‘Aepycamelus’’ priscus (taxon 14) to belong to a separate genus from other species of Aepycamelus. Met. length
5
length of metatarsals in comparison with extant artiodactyls. All specimen numbers refer to AMNH specimens unless otherwise noted. A #
beside the specimen no. indicates a field number.
Taxon Specimen no. Mets. fused? Mets. splayed? Met. length PCA Ternary plot
1Poebrotherium 6520 no no long R R
Stenomylinae
2Pseudolabis
3Miotylopus
4Stenomylus
41942
36446
14226
partial
fully
mostly
slight
slight
slight/mod
medium
long
long
R/L
R/L
R/L
R
R
R/L
(Not assigned to subfamily)
5Floridatragulus
6Gentilicamelus MCZ 7784
7510 no
partial slight
no short
medium R
R/L R/L
R/L
Miolabinae
7Paramiolabis
8Miolabis
9Nothotylopus
652-39#
24199
68554
partial
no
partial
slight
slight
no
medium
medium
medium
R/L
R/L
L
L
R
L/C
Protolabinae
10 Tanymykter
11 Michenia
12 Protolabis
36584
39628
400-2988#
mostly
mostly
fully
no/slight
mod/fully
fully
medium
long
very long
L
L
L
R/L
L
L
(Not assigned to subfamily)
13 Oxydactylus 17620 partial no long L R
Camelinae
14 ‘‘A.’’ priscus 14188 fully mod/fully long C R/L
Camelinae (Lamini)
15 Aepycamelus
16 Hemiauchenia 9109
38179 fully
fully fully
fully very long
very long C
CC
L
Camelinae (Camelini)
17 Procamelus
18 Titanotylopus 75003
10703 fully
fully fully
fully medium
long C
CL
L/C
tween camelids and ruminants. Protoceratids are derived cra-
nially, and later forms (synthetoceratines) had characteristic
‘‘sling-shot’’ nasal horns. However, postcranially they retain
primitive features such as short, unfused metapodials with in-
complete distal metapodial keels.
It might seem that a more basal taxon would be preferable
to determine primitive artiodactyl limb anatomy, such as the
Eocene Diacodexis, or an oreodont. However, the problem with
both of these taxa is that they are small animals, and larger
animals such as the camelids and ruminants under study might
be expected to have differences in their limb anatomy simply
because of their larger size, and further, they retains full-sized
lateral digits, a primitive condition not found in any of the taxa
in our analysis. Synthetoceras is of comparable size to the ca-
melid and ruminant taxa compared in Figures 1 and 2 (body
mass estimated at around 150 to 200 kg). Some suids are fairly
large and retain a relatively primitive postcranial anatomy, and
the same is also true for some extinct oreodonts. But these
artiodactyls are more distantly related to camels and ruminants
than protoceratids, and might be expected to have their own
anatomical specializations, making them less appropriate for
comparison.
The following measurements were taken on all specimens,
as allowed by preservation. Measurements were taken of the
lengths (maximum articular length) and the diameter (at the
midshaft, of both anteroposterior and mediolateral dimensions)
of the tibia, metatarsal, and metacarpal, and of the length and
diameters of the three hindfoot phalanges. Additional measure-
ments were taken on articular surfaces of the phalanges. These
included the extension onto the volar surface of the distal ar-
ticular surface of the proximal and medial phalanges (Fig. 2.1
and 2.3) and the width and the depth of the carinal groove on
the proximal articular surface of the proximal phalanx (Fig.
2.2). All measurements are listed in the appendix. In the fossil
taxa we also noted whether or not the metapodials were fused,
and if they had splayed distal ends, as in extant camelids (Table
2).
A principal components analysis was conducted using the 22
morphological variables described above. Two principal com-
ponents were extracted using a Varimax rotation as imple-
mented in StatView 5.0. Missing data points for Synthetoceras,
Miotylopus,Paramiolabis and Floridatragulus were estimated
using regression equations (Appendix), while Tragulus was ex-
cluded because of two missing data points. To visualize differ-
ences in foot posture and proportions we used a ternary diagram
to examine the proportions of the phalanges, following the
methods outlined by Gatesy and Middleton (1997:310). We
made no attempt to collect data for all extant ruminant taxa,
nor have we corrected our data for phylogenetic biases.
RESULTS
General Features of Extant Camelid Foot Morphology
Figures 1 and 2 show some general differences in foot mor-
phology between camelids, ruminants, and protoceratids.
Metapodial Morphology The primitive condition is for
relatively short metapodials. The primitive tragulid ruminant
Hyemoschus has metatarsals that are approximately six times
longer than their average mid-shaft width. Metapodials are lon-
ger in all extant ruminants and camelids, although certain mon-
tane bovids have secondarily shortened metapodials (Scott,
1985). Extant Rangifer and Lama (the llama) both have meta-
tarsals that are about eleven times as long as their width. Note
that metapodial length tends to scale with negative isometry
(Bertram and Biewener, 1990), except in Giraffa (the giraffe)
which has exceedingly long metapodials, with metatarsals
around twenty times as long as their width.
Other primitive character states, seen in both Hyemoschus
and Synthetoceras, are as follows: small ‘‘side toes’’ (short,
complete or partial digits 2 and 5) are present; metapodials 3
and 4 are unfused and not splayed at their distal ends; and the
metapodials lack complete distal keels. Extant camelids and ru-
minants are both derived with respect to these conditions, but
in different ways. Both have fused metapodials; all extant ca-
113JANIS ET AL.—LOCOMOTOR EVOLUTION IN CAMELS
FIGURE 2. Phalangeal morphology of (A) a camelid, (B) a ruminant
and (C) a protoceratid. All phalanges are right pedal phalanges, all scale
bars are 1 cm. Taxa are the same as Figure 1. (1) Anterior (volar) view
of proximal phalanx. (2) Posterior (plantar) view of proximal phalanx.
(3) Anterior view of medial phalanx. (4) Anterior view of ungual pha-
lanx. (5) Lateral view of ungual phalanx.
melids and almost all fossil taxa retain only digits 3 and 4
(Honey et al., 1998), whereas small side toes may be retained
in the Cervidae and Moschidae, but are lost in members of
other extant ruminants. Ruminants are derived in possessing
complete distal metapodial keels that act to lock the foot in the
unguligrade position (Webb and Taylor, 1980; Janis and Scott,
1987). Camelids lack complete keels, but have the derived fea-
ture of distally splayed metapodials (Fig. 1; Webb, 1972).
Proximal Phalanx Morphology Camelids have a relative-
ly long proximal phalanx, with a length of about five times that
of the diameter. The relative length in ruminants is shorter,with
a length around four times that of the diameter (Fig. 2.1). The
condition in Synthetoceras resembles the ruminant one: thus,
the camelid condition is likely derived.
In camelids, the distal articular surface of the proximal pha-
lanx is extended up onto the volar surface (Fig. 2.1). This ex-
tension is not found in ruminants or in Synthetoceras,sothe
camelid condition is likely the derived one. The extent of the
distal articular surface (not figured) is also greater on the plantar
side in camelids than in Synthetoceras, but in ruminants it ac-
tually appears to be less extended than in Synthetoceras. Thus,
in this morphology both camelids and ruminants appear to be
modified from the probable primitive condition. The camelid
morphology would allow for a greater degree of extension of
the medial phalanx on the proximal one in association with a
digitigrade foot posture. In contrast, the restriction of the artic-
ular surface in ruminants would serve to limit mobility between
the proximal and medial phalanges, perhaps in association with
a more derived unguligrade foot posture.
The carinal groove on the posterior border of the proximal
articulation of the proximal phalanx in camelids is relatively
wider and shallower than in ruminants and Synthetoceras (Fig.
2.2). Additionally, the proximal articular surface of camelids is
relatively smooth, with little surface relief, and is broader in
the mediolateral direction than in the anteroposterior one. Syn-
thetoceras has a relatively deep carinal groove, and a square-
shaped articular surface with higher relief. If Synthetoceras rep-
resents the more primitive condition, the derived camelid con-
dition could be interpreted as reflecting a less stabilized meta-
podial-phalangeal joint in conjunction with a digitigrade foot
posture. The ruminant condition differs from the camelid one
in the opposite direction from that of Synthetoceras. The artic-
ular surface is longer anteroposteriorly, the carinal groove is
deepened, presumably in association with the complete meta-
podial keels, and the articular surface is deeper, allowing for
more interlocking of the joint.
Medial Phalanx Morphology The medial phalanx of cam-
elids is relatively shorter in comparison to the width (about two
and a half as long as it is wide) than in ruminants (about three
times as long as wide; Fig. 2.3). However, the medial phalanx
of Synthetoceras is also short, so this may represent the prim-
itive condition. The distal articular surface has only a slight
extension onto the volar surface in both camelids and Synthe-
toceras, whereas in ruminants this surface is extended (Fig.
2.3). The derived ruminant condition is presumably related to
a greater extension of the ungual phalanx on the medial one
with an unguligrade foot posture that is more highly stabilized
than in Synthetoceras.
The proximal articular surface of the medial phalanx (not
figured) is longer in the mediolateral direction than in the an-
teroposterior direction in camelids, with little surface grooving.
The articular surface is more square in shape and more deeply
depressed in both Synthetoceras and ruminants, with extensive
relief in the ruminant condition. The derived condition in ca-
melids probably represents greater mobility between proximal
and medial phalanges.
Ungual Phalanx Morphology The ungual phalanx of ca-
melids is relatively short and flat, with a very shallow, relatively
flat articular surface. In ruminants, the phalanx is long, distally
pointed, and high in relief, with extensive depressions on the
articular surface (Fig. 2.4, 2.5). The morphology of the ungual
phalanx in Synthetoceras is similar to the ruminant condition,
but with less extensive relief on the articular surface. The dif-
ferences between camelids and ruminants would again reflect
foot posture. The longer and deeper ungual phalanx of rumi-
nants is encased within a hoof, and the deep relief on the artic-
ular surface reflects greater restriction of interphalangeal mo-
bility with an unguligrade foot posture.
Relative Metapodial Lengths
Figure 3 shows a plot of metatarsal length against the an-
teroposterior diameter of the metatarsal, including measure-
ments for the long-legged, so-called ‘‘giraffe antelope,’’ the ge-
renuk, Litocranius walleri, and the dibatag, Ammodorcas clar-
kei, (taken from Scott, 1985). Metatarsal diameter was chosen
as a proxy of body size, because the diameters of distal limb
bones provide better estimates of body mass in artiodactyls than
length measurements (Scott, 1990). Additionally, the mass of
extant camelids can be estimated reliably from the correlations
of limb diameters with body mass derived from ruminants
(Scott, 1990). However, metatarsal diameter does not here pro-
vide a perfect proxy of size: for example, as shown in Figure
3, the metatarsal of Giraffa has a greater diameter than that of
114 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 1, 2002
FIGURE 3. Bivariate plot of the length of the metatarsal against the anteroposterior diameter of the metatarsal. Key to symbols: Open triangles
5
bovid ruminants; half-tone triangles
5
cervids, moschid and antilocaprid ruminants; half-tone squares
5
the ‘‘giraffe antelope’’ Litocranius
(52) and Ammodorcas (53); black squares
5
tragulid ruminants; black triangles
5
giraffid ruminants; open circles
5
fossil camelids; half-tone
circles
5
extant lamines; black circles
5
extant camelines; See Table 2 for key to numbers for extinct camelid taxa and Table 1 for extant taxa.
FIGURE 4. Principal components analysis of pedal morphological variables. Key as for Figure 3, except star
5
Synthetoceras, and Litocranius
and Ammodorcas are not included.
115JANIS ET AL.—LOCOMOTOR EVOLUTION IN CAMELS
TABLE 3. Factor loadings for variables used in the principal com-
ponents analysis.
Variable
Factor loading
Factor 1 Factor 2
Tibia length
Tibia diameter, mediolateral
Tibia diameter, anteroposterior
Metatarsal length
Metatarsal diameter, mediolateral
Metatarsal diameter, anteroposterior
0.931
0.967
0.950
0.900
0.950
0.954
2
0.196
2
0.146
0.044
2
0.202
2
0.142
2
0.019
Metacarpal length
Metacarpal diameter, mediolateral
Metacarpal diameter, anteroposterior
Proximal phalanx length
Proximal phalanx diameter, mediolateral
Proximal phalanx diameter, anteroposterior
0.872
0.940
0.957
0.905
0.957
0.973
2
0.300
2
0.181
2
0.128
2
0.344
0.123
0.093
Proximal phalanx distal articular extension
Proximal phalanx carinal groove depth
Proximal phalanx carinal groove width
Medial phalanx length
Medial phalanx diameter, mediolateral
0.746
0.732
0.882
0.906
0.949
2
0.418
0.309
2
0.334
2
0.016
2
0.027
Medial phalanx diameter, anteroposterior
Medial phalanx distal articular extension
Ungual phalanx length
Ungual phalanx diameter, mediolateral
Ungual phalanx diameter, anteroposterior
0.904
0.802
0.547
0.879
0.765
0.362
0.479
0.737
0.155
0.581
the extinct camelid Titanotylopus, but other skeletal elements
indicate that the latter is the larger taxon.
In comparing extant forms, it can be observed that Giraffa
is highly aberrant in its extremely long metatarsals, but the re-
lated Okapia (the okapi) clusters with the other ruminants. Ad-
ditionally, note that camelids are relatively long-legged in com-
parison with most ruminants, especially at larger body sizes,
although they do not fall completely outside of the range of
extant ruminants. The two species of Camelus are long-legged
in comparison with the majority of ruminants of a similar size
(i.e., body mass of around 500 kg), but are of comparable pro-
portions to the moose, Alces.
Many of the extinct camelids are longer-legged than extant
camels, with extremely long metapodials in comparison with
any living ruminant except Giraffa. This is especially true in
the size range corresponding to body masses of around 200–
500 kg. These extinct camelids include not only those taxa tra-
ditionally considered as ‘‘giraffe camels,’’ such as Aepycame-
lus, but also the taxa Hemiauchenia and Protolabis. The smaller
camels, mainly Eocene or Oligocene in age, cluster with the
extant ruminants and lamines. The larger ones (extinct members
of the tribe Camelini) have metatarsals of similar length to Ca-
melus and Alces.
Our data on Protolabis shows it to have relatively long me-
tapodials, although Honey et al. (1998) claimed that the meta-
podials are not very long in this genus, only equal (or less than)
in length to the basal length of the skull. However, Honey et
al. (1998) also stated that the metapodials of Protolabis are
relatively shorter in more derived species. Our specimen is a
fairly early one, from the early Hemingfordian Running Water
Formation, so perhaps this specimen is primitive in its long
metatarsals.
Figure 3 also illustrates possible differences in scaling ex-
ponents of the metatarsals between camelids and ruminants.
Scott (1985) noted a ‘‘breakpoint’’ in the scaling of metatarsal
lengths in extant bovids: metatarsal length scales allometrically
below a body size of around 200 kg, but at greater mass there
is little increase in absolute metatarsal length. This breakpoint
in limb length scaling has been attributed to the physical de-
mands of support at larger body size (Bertram and Biewener,
1990). Note that in Figure 3 the ‘‘breakpoint’’ occurs at a meta-
tarsal width of about 2.8 cm; ruminants with metapodials of
this size are Elaphurus davidianus and Cervus elaphus, which
both have body masses of around 200 kg.
The members of the tribe Camelini (Camelus,Procamelus,
and Titanotylopus) appear to obey the same scaling laws that
apply to the majority of ruminants. Their metatarsals are around
the same absolute length as smaller taxa such as Protolabis.
However, metatarsal length in other large extinct taxa, princi-
pally the members of the tribe Lamini, Aepycamelus and Hem-
iauchenia, have metatarsal lengths that appear to fall along a
continuation of the scaling exponent that applies to the smaller
artiodactyls, as does Giraffa.
Why the giraffe and some larger camelids should be exempt
from the scaling rules that appear to affect ruminants and mem-
bers of the tribe Camelini is not clear. Additionally, preliminary
data suggest that, unlike any ruminant or camelid, Giraffa is
unique in having a metatarsal that is longer than the tibia.
Principal Components Analysis
Figure 4 shows the results of the principal component anal-
ysis using the dimensions described in the Materials and Meth-
ods section. The first principal component (PC1) accounts for
78.5% of the variance. The loadings on PC1 are all positive,
and almost all are close to unity (Table 3). Thus this axis clearly
discriminates taxa based on body size. Some shape influence is
involved, as seen in the slightly lower loading for the length of
the ungual phalanx (Table 3), which is the variable that sepa-
rates Giraffa from the other taxa on this axis. This results in
Giraffa scoring more positively on this axis than the much larg-
er extinct camelid Titanotylopus. When Giraffa is removed
from the analysis, the loading of the ungual phalanx length
variable on PC1 is similar to the other variables, and the other
taxa are more evenly spread out along PC1 in accordance with
their body masses.
The second principal component explains 9.3% of the vari-
ance. This axis clearly separates the camelids (negative load-
ings) from the ruminants (positive loadings). Giraffa has near
zero loadings on PC2, similar to the tragulids, while Okapia
clusters with the other ruminants. PC2 also contains a size com-
ponent, in that larger ruminants (except Giraffa ) have higher
positive loadings, and larger camelids have higher negative
loadings. This may reflect greater locomotor specializations in
larger taxa, or somehow reflect a size component in PC2.
Variables with high positive values on PC2 (i.e., ruminant
features) are as follows: the length and anteroposterior diameter
of the ungual phalanx, the anteroposterior diameter of the me-
dial phalanx, the length of the volar extension of the distal
articular surface on the medial phalanx, and the depth of the
carinal groove on the proximal phalanx. As previously dis-
cussed, these articular features reflect the limitation of inter-
phalangeal mobility, and the large ungual phalanx reflects its
enclosure within a hoof.
Variables with high negative values (i.e., camelid features)
are as follows: the length of the volar extension of the distal
articular surface on the proximal phalanx, the length of the
proximal phalanx, the width of the carinal groove on the prox-
imal phalanx, and the length of the metacarpal. These articular
features reflect increased interphalangeal mobility. The length
of the proximal phalanx may be related to a general elongation
of the limb proximal to the foot pad.
Thus PC2 appears to be primarily an axis of digitigrady (neg-
ative loadings) versus unguligrady (positive loadings). The
length of the metacarpal (high negative values on PC2) reflects
the fact that all living and extinct camelids have fore- and hind
limbs of approximately equal length. In ruminants, equally-pro-
portioned forelimbs and hind limbs are seen only in animals
living in more open habitats, that make habitual use of the trot
116 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 1, 2002
FIGURE 5. (A) Ternary plot of phalangeal lengths. (B) Close up of
ternary plot. Key as for Figure 4.
gait, instead of using the bound or gallop for speeds faster than
a walk (Scott, 1985). In contrast, animals living in closed hab-
itats have forelimbs that are shorter than the hind limbs (Scott,
1985).
Among the fossil camelids, Poebrotherium and Floridatra-
gulus fall fairly close to the ruminants, although with slightly
less positive scores. Stenomylus,Gentilicamelus,Miolabis,
Miotylopus,Paramiolabis, and Pseudolabis fall in the mor-
phospace between ruminants and lamines. Michenia,Oxydac-
tylus,Nothotylopus,Protolabis, and Tanymykter cluster with
extant lamines. Aepycamelus,‘‘Aepycamelus’’ priscus,Hem-
iauchenia,Procamelus, and Titanotylopus cluster with extant
camelines (Table 2).
The third principal component (not shown) explains 3.4% of
the variance. This component appears to be a ‘‘giraffe’’ axis,
separating Giraffa from the other taxa. The variables with high
loadings on PC3 are long metapodials, a short ungual phalanx,
a short extension to the distal articular surface on the proximal
phalanx, and a wide and deep carinal groove on the proximal
phalanx. The separation of the giraffe from other taxa appears
to be a result of its similarity to extant camelids in limb length,
but with an unguligrade foot posture and the anomalous con-
dition (for unguligrady) of a short ungual phalanx.
Ternary Diagram
The ternary diagram of phalangeal lengths clearly separates
out all extant camelids from all ruminants (with the exception
of Giraffa, which here ‘‘out-camels’’ Camelus!; Fig. 5). Ca-
melids score higher on length of the proximal phalanx and low-
er on the other two axes (lengths of medial and ungual phalan-
ges, respectively) than ruminants. As with the PCA, Okapia
clusters with other ruminants, towards the high end of the ru-
minant range, while Hyemoschus falls within the low end of
the ruminant range, as do Moschus and Saiga in this analysis.
Synthetoceras clusters within the ruminants. Again, as with the
PCA, the extant camelines fall further from the ruminants than
do the extant lamines.
In this analysis, the following fossil taxa cluster with the
ruminants: Miolabis,Miotylopus,Oxydactylus,Poebrotherium,
and Pseudolabis (although Miotylopus and Oxydactylus are sit-
uated at the edge of the ruminant morphospace). The following
taxa fall in the morphospace between ruminants and lamines:
‘‘Aepycamelus’’ priscus,Floridatragulus,Gentilicamelus,Sten-
omylus, and Tanymykter. Finally, Hemiauchenia,Michenia,
Nothotylopus,Paramiolabis,Procamelus, and Protolabis clus-
ter with lamines, and Aepycamelus clusters with camelines (Ta-
ble 2). As in the PCA (Fig. 4), Titanotylopus again is more
derived along the camelid trend in morphospace than any extant
form, here falling close to the giraffe.
DISCUSSION
Figure 6 illustrates the only comprehensive phylogeny of ca-
melids, and additionally shows the chronological range of these
taxa (both from Honey et al., 1998). The fossil taxa have been
identified in terms of varying degrees of foot posture that re-
sembles the condition in extant camelids (see also Table 2).
The most primitive camelid, Poebrotherium, shows no evi-
dence of limb features like those of extant camelids, clustering
with the ruminants in every analysis. Note that while it tends
to fall close to less derived ruminants such as the tragulids on
the PCA and the ternary plot, its metatarsals are relatively long,
even in comparison with similar-sized ruminants (Fig. 3). Thus,
if Poebrotherium can be taken as the primitive camelid condi-
tion, we can see that the original camelid morphological design
was for moderately long legs with an unguligrade limb posture.
But primitive camelids lack the additional cursorial limb spe-
cializations (fused metapodials, complete distal metapodial
117JANIS ET AL.—LOCOMOTOR EVOLUTION IN CAMELS
FIGURE 6. Phylogeny of camelids (modified from Honey et al., 1998; only taxa considered in this paper shown). The rectangles show chro-
nological ranges of taxa. Key: Unshaded (white) rectangle, no foot morphological adaptations towards extant camelid condition; lightly-shaded
rectangles, some modifications towards foot morphology like extant camelids; heavily shaded rectangles, numerous morphological features like
that of extant camelids, but not with full suite of morphological adaptations; and black rectangles, foot morphology unequivocally like that of
extant camelids.
keels) of extant ruminants, although these features are also ab-
sent in the ruminants contemporaneous with Poebrotherium,
such as Hypertragulus and Leptomeryx (Webb and Taylor,
1980).
All other camelids show some evidence of the acquisition of
a derived limb morphology, more like that of extant camelids,
although in some taxa this is relatively slight. For example,
Floridatragulus is unlike most camelids in having relatively
short metapodials (Fig. 3), and clusters with ruminants on the
PCA (Fig. 4), but falls in the morphospace between ruminants
and lamines on the ternary plot (Fig. 5; Table 2).
On the other end of the scale, the more derived camelids that
are included within extant phylogenetic bracket in the Came-
linae (Aepycamelus,Hemiauchenia,Procamelus and Titanoty-
lopus ) are indistinguishable in pedal morphology from living
taxa. It can be inferred with confidence that these taxa pos-
sessed the pacing behavior of extant camelids. Pacing was also
probably the locomotor mode for the more primitive cameline
‘‘Aepycamelus’’ priscus. This animal differs in its morphology
from extant camelids only by a slight displacement on the ter-
nary plot towards somewhat longer medial and ungual phalan-
ges, although its metapodials are also somewhat less distally
splayed than the more derived camelids.
However, a suite of characteristics identical to that of extant
camelids is seen in the protolabine Protolabis, which has also
evolved elongated metatarsals convergently with some of the
xn more derived camelids. Independent evidence for pacing in
Protolabis exists in the form of a fossilized trackway (Webb,
1972). If the phylogeny of Honey et al. (1998) is correct, this
would imply that pacing behavior evolved twice within came-
lids. In this phylogeny the genus Oxydactylus, which does not
show extensive modifications towards a condition like that of
extant camelids, is placed as the sister-taxon to the Camelinae.
If this phylogenetic position of Oxydactylus is incorrect, more
extensive modifications towards extant camelid limb anatomy
might be a shared derived feature of the Protolabinae and the
Camelinae. However, other protolabines (Tanymykter and Mich-
enia ) are less like extant camelids in their limb anatomy than
Protolabis, implying convergence on camelines within the pro-
tolabine lineage whatever the phylogenetic position of Oxydac-
tylus.
Of equal interest is the fact that some convergence on extant
camelid morphology, although not attaining the fully derived
condition, is seen within two other camelid lineages. This trend
is seen both within the Stenomylinae (attaining the most derived
condition in Stenomylus) and the Miolabinae (attaining the most
derived condition in Nothotylopus). Note that the taxon that we
here refer to as Nothotylopus may actually represent the taxon
sometimes called ‘‘Homocamelus,’’ that Honey et al. (1998)
refer to as ‘‘Miolabine sp.’’ Both are derived members of the
Miolabinae, so in fact the precise identification of this specimen
is not important. As there is no modern analog for this type of
118 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 1, 2002
intermediate morphology, without associated trackways it is im-
possible to determine what the actual gait of these animals
might have been.
Stenomylus and Nothotylopus both lack the splaying of the
distal ends of the metapodials seen in more derived camelids,
and thus it seems unlikely that they had a digitigrade foot pos-
ture. Yet, on both the PCA and ternary plots, Stenomylus falls
in the morphospace between the ruminants and lamines and
Nothotylopus clusters with the lamines, indicative of changes
in relative proportions of the foot bones in both taxa. Notho-
tylopus is more derived in its pedal anatomy than Stenomylus,
but Stenomylus is more derived than the other stenomylines,
Pseudolabis and Miotylopus.
We also note that Stenomylus possesses other skeletal fea-
tures like those of extant camelids that may be indicative of
lateral stabilization, such as broad, flat ribs and an expanded
iliac crest. We were unable to determine the condition of these
features in Nothotylopus. In contrast, the primitive camelid Poe-
brotherium, which possesses a ruminant-like foot morphology,
shows little evidence of features promoting lateral stability.
Poebrotherium lacks an expanded iliac crest, and has only a
slight broadening of the anteriormost ribs (nos. 1–7, as seen on
the mounted specimen in the American Museum of Natural
History; Scott, 1940:630–635). Perhaps these camelids of ‘‘in-
termediate’’ morphology used some types of lateral gaits (for
example, the slow pace or lateral walk and the rotary gallop,
as employed by present-day long-legged ruminants), but did not
have the running pace gait of extant camelids.
When Webb (1972) considered the evolution of pacing lo-
comotion in camelids, he noted that the pacing trackways of
Protolabis were middle Miocene (Barstovian) in age, and cor-
related the locomotor shift with the spread of open grassland
habitats in North America. He pointed out that, in extant cam-
elids, the pacing gait is adapted for long distance travel in open
terrain, and interpreted the fossil record as supportive of the
hypothesis that pacing first evolved along with the open habitat
grasslands.
Figure 6 shows the chronological ranges of camelid taxa
along with their phylogeny, and it is apparent that the trend
towards a more derived morphology resembling that of extant
camelids occurred at an earlier date than proposed by Webb.
Consider those taxa where partial modifications are attained.
Stenomylus (Stenomylinae) first appeared in the late Oligocene,
and Nothotylopus (Miolabinae) and Tanymykter and Michenia
(Protolabinae) in the early Miocene. Furthermore, some taxa
possessing modifications fully like those of extant camelids,
Protolabis (Protolabinae) and Aepycamelus (Camelinae) also
first appeared in the early Miocene. Thus, both taxa with ad-
aptations indicating a fully-derived pacing gait, and those with
some sort of modifications leading towards this gait, made their
appearance before the spread of open savanna or grasslands
(Jacobs et al., 1999). The significance of this observation is
unclear. One possible explanation is that more open habitats
were present earlier in North America than previously consid-
ered, as also suggested by other workers (Webb, 1977; Janis,
1982; Leopold et al., 1992), although such habitats were not
dominated by grass until later in the Miocene.
Another interesting feature of this study is the observation
of how different Giraffa is from other extant ruminants. We
document here the convergences with camelids in certain as-
pects of foot morphology; but the giraffe also has other peculiar
aspects of postcranial morphology, some convergent with the
extant camelid condition and some unique. The description of
giraffe morphology below is based on general osteological ob-
servations rather than on quantitative analysis, but nevertheless
serves to highlight features that are deserving of further study.
The giraffe resembles extant camelids in having an elongated
and laterally projecting iliac crest, which presumably also
serves for the origin of an enlarged tensor fascia latae to aid in
lateral stability, although the giraffe lacks the camelid feature
in the scapula of a relatively large scapular spine and acromion
process. This morphology of the ilium is not simply related to
large size, because it is absent in other large ruminants such as
Alces and Bison. The giraffe also resembles extant camelids in
having a relatively narrow chest, but lacks the camelid-like
broad, flattened ribs.
The giraffe is unique among all ruminants and camelids, in-
cluding the extinct ‘‘giraffe camels’’ with highly elongated
limbs, in having metapodials that are longer than the tibia. Ad-
ditionally, there appears to be a peculiar morphology of the
pedal sesamoid bones observed on mounted specimens in the
Harvard Museum of Zoology. In ruminants (e.g., Bison), the
distal sesamoids have their major articulation with the base of
the medial phalanx. In contrast, Giraffa has relatively larger
distal sesamoids that appear to have a greater articulation with
the ungual phalanx, possibly acting as a posterior extension for
the effective plantar surface of the ungual phalanx.
CONCLUSIONS
The peculiar foot morphology of present-day camelids, in-
dicative of a secondarily digitigrade foot posture in association
with a pacing mode of locomotion, may have evolved more
than once within the lineage. The fully modern version of this
morphology is seen not only in the extinct taxa closely related
to the living camels and llamas, but also within the Protolabi-
nae, the sister taxon to the Camelinae, where it may represent
independent evolution. Fossilized trackways also support the
hypothesis that the derived protolabine Protolabis had a run-
ning pace gait like that of extant camelids (Webb, 1972). Partial
convergence on the foot morphology of extant camelids was
also seen within the more primitive families Stenomylinae (pri-
marily in the genus Stenomylus) and the Miolabinae (primarily
in the genus Nothotylopus). However, some deviation from the
more primitive condition (as exemplified by the extinct proto-
ceratid Synthetoceras, or in extant tragulid ruminants) could be
observed in all camelids studied with the exception of the most
primitive taxon, Poebrotherium.
We consider that most extinct camelids may have engaged
in some types of lateral gaits, such as the lateral walk and the
rotary gallop, observed in extant long-legged antelope and the
giraffe. However, the running pace gait, unique to camelids
among extant wild mammals, was probably used only by those
taxa with a foot fully like that of extant camelids. The parallel
emergence of more derived types of foot morphology among
fossil camelids occurred in the late Oligocene to early Miocene
times, preceding the development of more open savanna habi-
tats in the North American late Miocene.
ACKNOWLEDGMENTS
We thank the following people for access to specimens in
their collections. Maria Rutzmoser (Mammalogy) and Chuck
Schaff (Vertebrate Paleontology) at the Museum of Compara-
tive Zoology (Harvard University), and Dick Tedford (Verte-
brate Paleontology) and Chris Norris (Mammalogy) at the
American Museum of Natural History (New York). Thanksalso
to Dave Webb for discussions, Kay Earls for checking speci-
mens on display, Brian Regal for the osteological drawings, Ian
Tattersall for accommodations, and Mark Norell for being un-
derstanding of various problems. This project was funded in
part by a Salomon Faculty Award (Brown University) to CMJ,
and by Brown University undergraduate honors thesis funds to
BB.
119JANIS ET AL.—LOCOMOTOR EVOLUTION IN CAMELS
LITERATURE CITED
Bertram, J. E. A., and A. A. Biewener. 1990. Differential scaling of the
long bones in the terrestrial Carnivora and other mammals. Journal
of Morphology 20:157–169.
Frick, C. 1937. Horned ruminants of North America. Bulletin of the
American Museum of Natural History 69:1–669.
———, and B. E. Taylor. 1968. A generic review of stenomyline cam-
els. American Museum Novitates 2353:1–51.
Gatesy, J., M. Milinkovitch, V. Waddell, and M. Stanhope. 1999. Sta-
bility of cladistic relationships between Cetacea and higher-level
artiodactyl taxa. Systematic Biology 48:6–20.
Gatesy, S. M., and K. M. Middleton. 1997. Bipedalism, flight, and the
evolution of theropod locomotor diversity. Journal of Vertebrate
Paleontology 17:308–329.
Gauthier-Pilters, H., and A. I. Dagg. 1981. The Camel: its Evolution,
Ecology, Behavior, and Relationships to Man. University of Chi-
cago Press, Chicago, 208 pp.
Hildebrand, M. 1976. Analysis of tetrapod gaits: general considerations
and symmetrical gaits; pp. 203–236 in R. M. Herman, G. Grillner,
P. S. G. Stein, and P. G. Stuart (eds.), Neural Control of Locomo-
tion. Plenum Press, New York.
Honey, J. G., J. A. Harrison, D. R. Prothero, and M. S. Stevens. 1998.
Camelidae; pp. 439–462 in C. M. Janis, K. M. Scott, and L. L.
Jacobs (eds.), Tertiary Mammals of North America. Cambridge
University Press, Cambridge and New York.
Howell, A. B. 1944. Speed in Animals. Hafner Publishing Co., New
York, 270 pp.
Jacobs, B. F., J. D. Kingston, and L. L. Jacobs. 1999. The origin of
grass-dominated ecosystems. Annals of the Missouri Botanic Gar-
den 86:590–643.
Janis, C. M. 1982. Evolution of horns in ungulates: ecology and paleo-
ecology. Biological Reviews 57:261–318.
———, and K. M. Scott. 1987. The interrelationships of higher rumi-
nant families with special emphasis on the members of the Cer-
voidea. American Museum Novitates 2893:1–85.
Joeckel, R. M., and J. M. Stavis. 1996. Basicranial anatomy of Syndy-
oceras cooki (Artiodactyla: Protoceratidae) and the need for a re-
appraisal of tylopod relationships. Journal of Vertebrate Paleontol-
ogy 16:320–327.
Leopold, L., G. Liu, and S. Clay-Poole. 1992. Low biomass vegetation
in the Oligocene; pp. 399–420 in D. R. Prothero and W. A. Berg-
gren (eds.), Eocene–Oligocene Climatic and Biotic Evolution.
Princeton University Press, Princeton.
Norris, C. A. 1999. The cranium of Bunomeryx (Artiodactyla: Homa-
codontidae) from the upper Eocene Uinta deposits of Utah and its
implications for tylopod systematics. Journal of Vertebrate Pale-
ontology 19:742–751.
——— 2000. The cranium of Leptotragulus, a hornless protoceratid
(Artiodactyla, Protoceratidae) from the middle Eocene of North
America. Journal of Vertebrate Paleontology 20:341–348.
Patton, T. H., and B. E. Taylor. 1971. The Synthetoceratinae (Mam-
malia, Tylopoda, Protoceratidae). Bulletin of the American Muse-
um of Natural History 145:119–218.
Scott, K. M. 1985. Allometric trends and locomotor adaptations in the
Bovidae. Bulletin of the American Museum of Natural History 179:
197–288.
——— 1990. Postcranial dimensions of ungulates as predictors of body
mass; pp. 301–335 in J. Damuth and B. J. MacFadden (eds.), Body
Size in Mammalian Paleobiology: Estimation and Biological Im-
plications. Cambridge University Press, Cambridge and New York.
Scott, W. B. 1940. The mammalian fauna of the White River Oligocene,
Part IV. Artiodactyla. Transactions of the American Philosophical
Society 28:363–746.
Webb, S. D. 1972. Locomotor evolution in camels. Forma et Functio
5:99–112.
——— 1977. A history of savanna vertebrates in the New World, Part
I. North America. Annual Review of Ecology and Systematics 8:
355–380.
———, and B. E. Taylor. 1980. The phylogeny of hornless ruminants
and a description of the cranium of Archaeomeryx. Bulletin of the
American Museum of Natural History 167:117–158.
Received 5 June 2000; accepted 15 May 2001.
120 JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 22, NO. 1, 2002
APPENDIX
Data used in statistical analyses, all measurements in cm. Key: L
5
articular length; ML
5
mediolateral diameter; AP
5
anteroposterior
diameter; Ext
5
distal articular extension; GW
5
width of carinal groove; GD
5
depth of carinal groove; *estimated by regression. Taxadesignated
by numbers used in Tables 1 and 2. Data for 51, Synthetoceras, from AMNH 40231, 40241, 40243, and 40245.
Taxon
Tibia
LMLAP
Metatarsal
LMLAP
Metacarpal
LMLAP
1Poebrotherium
2Pseudolabis
3Miotylopus
4Stenomylus
5Floridatragulus
22.47
30.23
28.82
21.85
*19.59
1.95
2.32
2.11
1.69
*2.92
1.41
2.58
2.75
1.81
*2.07
15.50
22.19
23.87
16.68
12.98
1.82
2.09
1.37
1.21
2.08
0.99
1.97
1.94
1.17
1.17
13.07
22.28
*22.71
16.27
11.09
1.49
2.27
*1.50
1.20
*2.31
0.90
1.78
*1.75
1.07
1.01
6Gentilicamelus
7Paramiolabis
8Miolabis
9Nothotylopus
10 Tanymykter
26.40
*24.07
27.93
37.85
36.93
2.16
*2.04
2.57
3.23
3.03
2.49
*1.98
2.59
3.85
3.12
18.83
17.12
17.25
26.75
27.56
1.71
1.46
2.23
2.45
2.10
1.47
1.39
1.70
2.40
2.69
18.02
17.23
16.74
24.66
27.15
1.98
1.42
2.71
2.91
2.48
1.53
1.38
1.76
2.32
2.36
11 Michenia
12 Protolabis
13 Oxydactylus
14 ‘‘Aepycamelus’’ priscus
15 Aepycamelus
31.77
44.65
38.50
52.30
36.50
2.64
3.45
2.65
4.35
3.25
2.60
3.18
2.64
3.18
3.22
27.20
36.79
28.47
45.90
35.46
1.80
2.22
2.01
3.15
2.78
2.00
2.95
2.17
3.29
2.83
28.22
36.94
28.12
45.30
36.85
1.89
2.40
2.57
3.22
3.26
2.11
2.86
2.01
3.37
3.18
16 Hemiauchenia
17 Procamelus
18 Titanotylopus
19 Camelus bactrianus
20 Camelus dromedarius
48.10
44.50
50.36
47.94
43.59
3.90
4.62
5.53
5.46
5.01
3.43
3.92
5.11
4.47
4.27
40.79
34.50
38.05
35.92
34.60
2.54
3.15
4.00
3.92
3.19
2.86
4.61
4.22
3.55
3.23
44.25
35.80
41.48
35.35
34.68
2.55
3.87
4.72
4.48
3.69
2.94
4.00
4.35
3.72
3.09
21 Llama glama
22 Llama guanicoe
23 Llama pacos
24 Vicugna vicugna
25 Hyemoschus aquaticus
30.16
33.45
29.05
25.39
12.36
3.16
2.94
2.82
2.66
1.31
2.76
2.96
2.59
2.49
1.67
21.03
22.99
20.34
18.56
5.99
1.81
1.72
1.83
1.89
1.45
1.78
2.08
1.67
1.54
0.67
21.50
23.40
20.30
18.39
3.96
1.98
2.14
1.98
2.19
1.39
1.76
2.22
1.65
1.50
0.38
26 Tragulus javanicus
27 Okapia johnstoni
28 Giraffa camelopardalis
29 Alces alces
30 Antilocapra americana
10.07
33.60
61.90
46.99
26.95
0.73
3.83
6.56
4.37
2.17
0.87
3.50
5.77
4.89
2.77
6.71
29.19
67.61
36.83
22.61
0.58
2.99
4.40
3.30
1.49
0.62
3.14
4.80
4.09
1.90
4.45
28.59
69.16
32.27
21.49
0.55
3.15
5.00
3.54
1.62
0.33
3.46
4.80
3.30
1.59
31 Axis porcinus
32 Capreolus capreolus
33 Cervus elaphus
34 Elaphurus davidianus
35 Mazama americana
19.42
23.69
45.72
36.83
19.96
1.23
1.77
4.82
3.76
1.62
1.78
1.91
5.48
3.51
1.80
13.36
19.32
34.93
28.26
12.33
1.17
1.23
3.17
2.57
1.08
1.41
1.53
4.31
2.69
0.91
10.79
15.95
30.48
25.36
15.08
1.22
1.82
3.49
2.76
1.21
1.10
1.19
3.72
2.34
1.25
36 Moschus moschiferus
37 Odocoileus virginianus
38 Rangifer tarandus
39 Boselaphus tragocamelus
40 Capra hircus
19.97
31.52
36.94
37.50
21.74
1.22
2.37
3.19
3.96
1.86
1.12
2.59
3.36
4.36
2.04
13.41
27.34
29.55
26.89
12.21
0.84
1.85
2.39
2.74
1.37
0.95
2.35
3.43
3.11
1.32
9.86
21.29
22.30
26.51
11.24
0.96
1.95
2.43
3.10
1.64
0.74
1.88
2.40
2.52
1.09
41 Cephalophus silvicultor
42 Damaliscus pygarus
43 Gazella gazella
44 Kobus leche
45 Ourebia ourebi
25.38
27.69
22.20
31.59
20.82
2.66
2.15
1.71
2.65
1.54
3.46
2.12
1.83
3.04
1.78
16.80
22.26
17.70
23.04
15.46
1.89
1.38
1.15
1.95
1.14
2.09
1.85
1.52
2.14
1.36
15.36
21.26
16.44
22.39
15.31
2.05
1.55
1.27
2.03
1.09
1.51
1.42
1.10
1.65
0.99
46 Ovis aries
47 Rupicapra rupicapra
48 Saiga tatarica
49 Sylvicapra grimmia
50 Taurotragus oryx
51 Synthetoceras
20.17
26.84
22.43
17.43
39.70
*22.34
2.05
1.59
1.89
1.82
4.35
*2.10
2.10
1.68
2.01
1.50
4.41
*2.59
12.54
16.55
18.44
14.42
28.70
*15.39
1.46
1.28
1.13
0.95
2.71
*1.51
1.47
1.25
1.85
1.09
3.18
*1.63
11.77
14.94
15.91
13.15
26.02
13.57
1.71
1.48
1.38
0.86
3.02
1.64
1.31
1.06
1.36
0.83
3.72
1.39
121JANIS ET AL.—LOCOMOTOR EVOLUTION IN CAMELS
APPENDIX (Continued)
Taxon
Proximal phalanx
L ML AP Ext GW GD
Medial phalanx
L ML AP Ext
Ungual phalanx
LMLAP
1
2
3
4
5
6
7
8
9
10
2.84
4.35
3.58
3.72
4.31
4.38
4.57
4.60
7.10
5.70
0.71
0.97
0.96
0.68
0.93
1.03
0.77
1.24
1.54
1.32
0.86
1.28
0.95
0.90
1.07
1.07
1.01
1.28
1.73
1.37
0.35
0.48
0.35
0.50
0.50
0.40
0.55
0.50
1.17
0.51
0.22
0.32
0.38
0.29
0.37
0.34
0.37
0.49
0.63
0.47
0.25
0.30
0.34
0.25
0.37
0.26
0.42
0.30
0.48
0.33
1.57
2.21
1.80
2.19
2.21
2.12
2.10
2.37
3.80
3.21
0.66
1.01
0.85
0.67
0.98
0.87
0.85
1.06
1.59
1.12
0.69
1.05
0.71
0.68
0.98
0.87
0.77
1.10
1.16
1.07
0.51
0.86
0.59
0.56
0.59
0.62
0.50
0.84
0.84
0.92
1.69
2.61
1.84
1.48
1.88
1.78
1.40
2.76
2.12
2.53
0.48
0.70
0.67
0.64
0.71
0.72
0.58
0.93
1.18
0.91
0.62
1.12
0.91
0.55
0.94
0.79
0.75
1.08
1.10
1.22
11
12
13
14
15
16
17
18
19
20
6.99
7.73
5.35
8.12
9.69
10.22
9.99
12.67
10.56
10.09
0.92
1.31
1.05
1.70
1.77
1.71
1.95
2.38
2.40
2.14
1.51
1.61
1.14
1.87
2.32
2.04
2.52
3.21
2.77
2.44
0.87
0.74
0.48
1.31
1.54
1.22
1.91
2.17
2.21
1.94
0.59
0.65
0.45
0.87
0.83
0.74
0.85
0.97
1.02
0.82
0.36
0.42
0.33
0.40
0.46
0.44
0.37
0.31
0.45
0.39
3.23
3.71
2.37
4.66
4.37
3.96
4.77
6.47
6.17
5.98
1.05
1.18
1.13
1.78
1.82
1.83
2.01
2.52
2.64
2.81
1.05
1.06
0.86
1.58
1.55
1.55
1.70
2.19
1.86
1.58
0.76
0.86
0.67
1.23
1.08
0.99
0.91
1.17
1.22
1.17
1.98
2.50
2.59
3.02
2.57
2.69
2.78
2.65
2.96
2.57
0.81
0.98
0.87
1.44
1.45
1.24
1.65
1.68
2.24
1.90
1.08
1.13
1.15
1.79
1.54
1.40
1.77
1.74
1.66
1.50
21
22
23
24
25
26
27
28
29
30
6.96
7.95
6.52
6.06
2.04
1.44
6.62
12.96
8.30
5.28
1.23
1.45
1.10
1.06
0.85
0.41
2.32
3.57
2.51
0.95
1.49
1.60
1.25
1.29
0.73
0.41
2.50
3.73
2.90
1.53
1.04
1.25
0.96
1.03
0.20
0.05
0.54
1.00
1.46
0.36
0.67
0.67
0.58
0.48
0.34
—
0.56
1.55
0.69
0.19
0.39
0.31
0.31
0.27
0.24
—
0.43
0.85
0.55
0.21
3.46
3.84
3.01
2.92
1.48
0.96
3.89
6.44
6.18
2.99
1.37
1.56
1.50
1.14
0.75
0.31
2.08
3.78
2.17
1.01
1.07
1.18
1.09
1.30
0.55
0.38
2.19
3.68
3.19
1.26
0.93
1.00
0.84
0.94
0.40
0.11
1.84
3.62
2.88
1.14
2.28
2.17
2.44
2.00
1.22
1.00
6.13
2.99
8.11
3.37
1.07
1.24
1.19
1.05
0.59
0.26
1.81
1.90
2.24
0.66
1.24
1.23
1.29
1.02
0.50
0.32
3.02
2.50
3.51
1.51
31
32
33
34
35
36
37
38
39
40
3.45
3.92
7.44
6.72
3.03
4.15
5.23
6.63
7.11
3.83
0.75
0.78
2.29
2.06
0.81
0.63
1.33
1.60
2.31
1.10
1.23
1.07
2.80
2.42
1.05
0.90
1.60
1.86
2.90
1.08
0.71
0.34
0.95
0.98
0.35
0.36
0.73
0.78
0.87
0.66
0.13
0.31
0.65
0.68
0.17
0.44
0.49
0.65
0.59
0.30
0.20
0.27
0.49
0.65
0.15
0.38
0.46
0.43
0.48
0.45
2.62
2.67
5.44
5.73
2.29
2.98
4.31
4.61
4.80
2.61
0.91
0.84
2.12
1.81
0.81
0.60
1.32
1.59
2.25
1.00
1.08
1.05
2.59
2.30
0.94
0.78
1.61
1.74
2.57
0.99
0.93
0.83
2.42
1.62
0.66
0.60
1.26
1.85
1.99
1.06
2.63
2.72
6.55
5.48
2.06
2.01
3.56
5.35
6.02
3.29
0.77
0.66
2.06
2.05
0.66
0.55
1.05
1.97
1.68
0.50
1.17
1.09
2.79
2.27
0.98
0.88
1.53
2.31
2.74
1.54
41
42
43
44
45
46
47
48
49
50
51
4.47
5.46
4.52
5.74
3.21
3.74
4.59
4.89
3.07
6.51
5.37
1.76
0.98
0.89
1.37
0.83
1.23
1.09
0.92
0.63
2.17
1.67
1.64
1.27
1.05
1.65
1.00
1.17
1.15
1.18
0.85
2.68
1.43
0.74
0.49
0.58
0.55
0.21
0.57
0.53
0.45
0.47
0.83
0.49
0.20
0.31
0.18
0.27
0.21
0.34
0.32
0.25
0.22
0.51
0.31
0.33
0.38
0.25
0.43
0.18
0.24
0.48
0.31
0.29
0.73
0.40
2.99
3.01
2.35
3.74
2.12
2.43
3.31
2.45
2.06
4.12
0.32
1.27
1.03
0.72
1.41
0.64
1.05
0.84
0.82
0.67
2.14
1.97
1.48
1.22
0.73
1.45
0.80
1.07
0.97
0.93
0.79
2.54
1.55
1.06
1.13
0.78
1.36
0.76
0.94
1.16
0.90
0.67
1.86
0.69
3.04
3.67
2.91
5.03
2.54
2.71
3.73
2.45
2.07
7.77
0.39
0.81
0.78
0.64
1.11
0.55
0.70
0.72
0.59
0.64
1.68
1.98
1.23
1.58
1.12
1.65
0.88
1.30
1.35
1.12
0.82
2.99
2.33