Geometric morphometric analysis of the crown morphology of the lower first
premolar of hominins, with special attention to Pleistocene Homo
Aida Gomez-Robles a,*, Mafia Martinon-Torres a, Jose Mafia Bermudez de Castro a, Leyre Prado a,
Susana Sarmiento a, Juan Luis Arsuaga b
aCentro Nacional de Investigacion sobre Evolucion Humana (CENIEH), Avda. de la Paz, 28,09006, Burgos, Spain
b Centro de Evolucion y Comportamiento Humanos. qSinesio Delgado, 4, pabellon 14, 28029 Madrid, Spain
Middle Pleistocene European populations
This article is the third of a series that explores hominin dental crown morphology by means of geo
metric morphometrics. After the analysis of the lower second premolar and the upper first molar crown
shapes, we apply the same technique to lower first premolar morphology. Our results show a clear
distinction between the morphology seen in earlier hominin taxa such as Australopithecus and African
early Homo, as well as Asian H. erectus, and more recent groups such as European H. heidelbergensis, H.
neanderthalensis, and H. sapiens. The morphology of the earlier hominins includes an asymmetrical
outline, a conspicuous talonid, and an occlusal polygon that tends to be large. The morphology of
the recent hominins includes a symmetrical outline and a reduced or absent talonid. Within this later
group, premolars belonging to H. heidelbergensis and H. neanderthalensis tend to possess a small and
mesiolingually-displaced occlusal polygon, whereas H. sapiens specimens usually present expanded and
centered occlusal polygons in an almost circular outline. The morphological differences among Para
nthropus, Australopithecus, and African early Homo as studied here are small and evolutionarily less
significant compared to the differences between the earlier and later homin taxa. In contrast to the lower
second premolar and the upper first molar crown, the inclusion of a larger hominin sample of lower first
premolars reveals a large allometric component.
Recent publications have addressed the dental morphological
differences among fossil hominins (Bailey, 2004; Bailey and Lynch,
2005; Martinon-Torres et aL, 2006; Gomez-Robles et al., 2007),
continuing a tradition that began with the examination of the
occlusal morphology of earlier African Pliocene and Pleistocene
hominin taxa (Wood and Abbot, 1983; Wood et al., 1983; Wood and
Uytterschaut, 1987; Wood et aL, 1988; Wood and Engleman, 1988).
Other important contributions
morphology include, among others, Tobias (1991), Irish (1998);
Bermudez de Castro et al. (1999); Lockwood et aL (2000); White
et aL (2000); Irish and Guatelli-Steinberg (2003); Hlusko (2004)
and Moggi-Cecchi et al. (2005), but most of these have not focused
on the study of a single type of tooth. Our research group has
previously used geometric morphometric techniques to examine
the morphological variability of the hominin lower second pre
molar (P4) and upper first molar (MI )(Martinon-Torres et al., 2006;
on hominin dental crown
* Corresponding author.
E-mail address: email@example.com (A G6mez-Robles).
G6mez-Robles et aL, 2007, respectively). These studies paid special
attention to middle and late Pleistocene populations, but Pliocene
and early Pleistocene specimens from Africa, Asia, and Europe were
included to assess the significance of the observed variation. The
present study follows the research line of these previous studies,
and is part of a study of the whole hominin dentition by means of
geometric morphometric techniques.
The advantages of using dental remains in hominin phylogenetic
studies have been discussed elsewhere (e.g., Turner, 1969; Irish,
1993,1997,1998; Bailey, 2000, 2002a; Irish and Guatelli-Steinberg,
2003; Martin6n-Torres, 2006). Previous studies on P4 crown
morphology have shown differences among species (Wood and
Uytterschaut, 1987; Bailey and Lynch, 2005; Martin6n-Torres et aL,
2006). Despite the large variation in the lower first premolar (P3) of
H sapiens (Kraus and Furr, 1953; Biggerstaff, 1969; Scott and Turner,
1997), some authors have suggested that some premolar crown
morphologies may be taxonomically distinctive (Coppens, 1977;
Leonard and Hegmon, 1987; Wood and Uytterschaut, 1987; Suwa
et aL, 1996). These reports have mainly studied P3 morphology in
early hominin species (Leonard and Hegmon, 1987; Wood and
Uytterschaut, 1987; Suwa et al., 1996), and in some higher primate
species (Coppens, 1977), but a comprehensive comparative analysis
ofP3 crown morphology that samples throughout the hominin fossil
record has not been published. We analyze a sample that includes
most of the hominin Pliocene and Pleistocene species known to date,
including the Atapuerca-Sima de los Huesos (SH) dental sample, the
largest and most representative sample from the European middle
Pleistocene (Arsuaga et al., 1997), as well as the Atapuerca-Gran
Dolina sample, which represents the only available hominin dental
evidence from the European early Pleistocene to date. These data are
compared with large samples of H. neanderthalensis and H sapiens,
as well as with smaller samples of Australopithecus, Paranthropus,
and other Homo species, in order to ascertain the polarity of the
Methods based on the incidence and relative expression of
discrete traits, such as the Arizona State University Dental An
thropology System (ASUDAS) (Turner et al., 1991 ) have proven to be
only moderately successful for comparing tooth crown variation
within and among later fossil hominin species (Bailey, 2002b;
Martinan-Torres, 2006). Recently, many dental studies based on
image analysis of the occlusal morphology of fossil hominins (e.g.,
Bailey, 2004; Bailey and Lynch, 2005; Martinan-Torres et al., 2006;
Gamez-Robles et al., 2007; Moggi-Cecchi and Boccone, 2007), non
human extant primates (e.g., Bailey et al., 2004; Pilbrow, 2006;
Hlusko et al., 2007), and recent modern human populations (e.g.,
Harris and Dinh, 2006; Perez et al., 2006; Bernal, 2007) have been
published. Classical morphometric methods applied to image
analyses (measurement of diameters and cusp areas) have
demonstrated that African robust and non-robust groups differed
in their P3 morphology (Wood and Uytterschaut,1987; Suwa et al.,
1996). We have adopted geometric morphometric methods based
on Procrustes superimposition techniques (Rohlf and Slice, 1990;
Bookstein, 1991) to examine the morphological affinities among the
hominin species included in this study. Wood and Uytterschaut
(1987) used this methodology to assess the fissure pattern of some
African Pliocene and Pleistocene premolars, but in the present
study we use semilandmarks (Bookstein, 1997) to compare the
variation in the outline and the internal morphology (understood
as the disposition of the structures enclosed by the outline) of the
occlusal surface of the P3.
Our article aims to provide a comprehensive description of the
changes in P3 crown morphology during hominin evolution, paying
special attention to the middle Pleistocene European populations
and to the similarities and differences that they show with H sa
piens and H neanderthalensis. We anticipate that the results of this
study will contribute new and detailed morphological information
to the ongoing debate about hominin phylogeny (e.g., Wood, 1992;
Bermudez de Castro et al., 1997; Foley and Lahr, 1997; Lahr and
Foley, 1998; Rightmire, 1998; Stringer and Hublin, 1999; Wood
and Collard, 1999; Stringer, 2002; Manzi, 2004; Dennell and
Roebroecks, 2005; Martinan-Torres et al., 2007).
Materials and methods
A geometric morphometric analysis was performed on a sample
of 106 hominin first premolars. The samples comprised the fol
lowing number of specimens (Table 1): A. anamensis (n= 1), A.
afarensis (n = 7), A. africanus (n = 5), Paranthropus sp. (n = 3), H.
habiIis s. 1. (n = 5), H ergaster (n =4), H. georgicus (n = 2), H erectus
(n = 10), H antecessor (n = 2), H heidelbergensis (n = 18), H nean
derthalensis (n = 15), and H. sapiens (n = 34).
We used the same taxonomy as in Gamez-Robles et al. (2007).
Australopithecus specimens were separated into three species: A.
anamensis, A. afarensis, and A. africanus, whereas Paranthropus
premolars were grouped together under the denomination Para
nthropus sp. The H habiIis s. 1. group included the African Pliocene
List 0 f the specimens included in this analysis
anamensis (n = 1 )
afarensis (n = 7)
africanus (n = 5 )
Paranthropus sp. (n = 3 )
Homo habilis s . L ( n = 5 )
Homo georgicus ( n = 2)
Homo ergaster (n = 4)
Homo erectus (n = 10)
Homo antecessor (n = 2)
(n= 18 )
Homo sapiens (n = 34)
KNM-KP2928 1 (cast)
AL198-23; AL207-13; AL266-1; AL288; AL333w-60;
AL400-1a; LH4 (casts)
MLD2; STS52; STW14; STW498; OM075-148 (casts)
KNM-ER 3230; OMO L427-7; TM1517 (casts)
KNM-ER 1802; OH 7; OH 13; OH 16; OMO 75i-1255
0211; 02735 (originals)
KNM-ER 992; OH 22; KNM-WT 15000; Rabat (casts)
Zhoukoudian 20, 80, 35.77, Gl, K1.96 (casts)
Sangiran 6, 7-25, 7-26, 7-69, Trinil (originals)
ATD6-3; ATD6-96 (originals)
Arago 13; 71; 75 (originals)
Sima de los Huesos: AT-2; AT-47;
AT-4328; AT-4147; AT-3880; AT-2767; AT-809; AT-2768;
AT-607; AT-2438; AT-3941; AT-148; AT-1466; AT-3243;
AT -4100 (originals)
Krapina 25, 29, 33, 34, 111, 114, MbD, MbE, MbH; (casts)
Hortus IV, VI; Guattari 3 (originals)
Amud I; Le Moustier 1; Saint Cesaire (casts)
Dolni Vestonice 13 (original)
Grotte des Enfants 6; Qafzeh 9, 11 (casts)
Modern humans from Portugal, Institute of
Anthropology of the University of Coimbra (n = 18)
Modern humans from the American Museum of Natural
History, New York, (n = 12)
Taxonomical assignments of the isolated specimens from Zhoukoudian, Sangiran,
and Shungura Formation follow Weidenreich ( 1937), Grine and Franzen ( 1994), and
Suwa et al. (1996), respectively. These assignments, based mainly upon the mor
phology of the teeth (and also upon their geographical and chronological context),
can be supported by our results. Relative warp analysis is independent of the tax
onomy. CVA is dependent on the a priori assignment of the specimens, but the
groups we have used in this analysis are ample enough to enclose the variability of
those groups with isolated specimens.
specimens assigned to H. habiIis, H rudolfensis, and similar un
assigned specimens. The taxon H erectus (Dubois, 1894) was used
for Asian premolars from Zhoukoudian, Sangiran, and Trinil. African
specimens that some authors have attributed to H. erectus s. 1.
(Walker and Leakey, 1993) were designated as H ergaster (Groves
and Mazak,1975), and we also included the African specimen from
Rabat in this group. H georgicus (Gabunia et al., 2002) and H
antecessor (Bermudez de Castro et al., 1997) were analyzed sepa
rately on the basis of their general morphological distinction
(Bermudez de Castro et al., 1997; Gabunia et al., 2002), as well as
their geographical and chronological separation from other groups.
The term H. heidelbergensis (Schoetensack, 1908) was used for the
European middle Pleistocene populations, such as Arago and Ata
puerca-Sima de los Huesos samples. Finally, H neanderthalensis
comprised classic European Neandertals, while H sapiens was
represented by two modern human collections, one from the In
stitute of Anthropology of the University of Coimbra, dating from
the Portuguese 19th-20th century (n = 12,6 males and 6 females),
and the other held at the American Museum of Natural History in
New York and coming from Heidenheim, Germany (n = 18, 8 males
and 10 females), together with some early anatomically modern
humans and Upper Paleolithic H. sapiens.
Photographing the sample
We used standardized images of the occlusal surface of the
premolars. Images were taken with a Nikon® D1H digital camera
fitted with an AF Micro-Nikon 105 mm, f/2.8D. The camera was
attached to a Kaiser Copy Stand Kit RS-l ® with grid baseboard,
column, and adjustable camera arm. For maximum depth of field,
we used an aperture of f/32. The magnification ratio was adjusted
to 1 :1, and a scale was placed parallel to and at the same distance
from the lens as the occlusal plane.
In order to standardize the photographs, each tooth was posi
tioned with the lens parallel to the cementa-enamel junction (CEJ)
(Wood and Abbott, 1983; Wood and Uytterschaut,1987; Bailey and
Lynch, 2005; Martin6n-Torres et al., 2006), as shown in Fig. 1. Iso
lated teeth were placed on modeling clay, and mandibles with
premolars in situ were reoriented appropriately. The use of a stan
dard orientation avoids methodological problems which may occur
when 3D objects are projected onto a two dimensional surface
(Gharaibeh, 2005). Nevertheless, the estimation of the standard
plane can be difficult at times since the CEl is not straight. Thus,
subtle differences in the orientation of the reference plane might
give different landmarks configurations as an artifactual effect of
the CEl morphology. As this effect has not been measured in pre
vious papers of this series (Martin6n-Torres et al., 2006; G6mez
Rabies et al., 2007), an evaluation of the repeatability in the es
tablishment of the reference plane is provided below.
When both antimeres were present, the right was chosen for
study. However, in order to maximize the sample size, when the
right one was not preserved or when the location of the landmarks
was not clear, the left premolar was included for study. Left
antimeres were mirror-imaged with Adobe Photoshop® before
performing the analyses. Teeth with severe attritional wear and/or
with uncertain location of one or more landmarks were not
included in the study.
Geometric morphometric methods
Geometric morphometric methods based on Procrustes super
imposition (Rohlf and Slice, 1990; Bookstein, 1991) are becoming
one of the most used and powerful tools in morphological studies
(Adams et al., 2004). Individual structures recovered as landmark
Fig. 1. Diagram showing the location of the plane through the (El that was placed
parallel to the lens of the camera when photographing the premolars.
conformations are compared to the mean or consensus shape of the
analyzed sample by means of Generalized Procrustes Analysis
(GPA) (Rohlf and Slice, 1990; Dryden and Mardia, 1998). Landmark
configurations are translated, scaled, and rotated until the distances
among homologous landmarks are minimized according to least
squares criteria. The square root of the sum of those squared
distances is named Procrustes distance and is a measure of the
morphological differences among biological structures. Such
distance describes the entirety of the morphological differences
among the studied structures (Zelditch et al., 2004).
Thin plate spline (TPS) provides a representation of the shape
changes when one specimen is deformed into another one. Every
shape change can be decomposed into a uniform component with
equal effects on the complete structure, and a non uniform com
ponent with local effects on particular areas (Bookstein, 1991). The
non uniform component requires bending energy, a measurement
of the localization of the change of shape that provides a set of
shape descriptors, the partial warps scores (Bookstein, 1989, 1991,
1996a; Rohlf, 1996). Principal components analysis (PCA) of the
partial warps scores is the most common test in geometric mor
phometric studies and it is called relative warp analysis (Bookstein,
1991). This analysis reduces the total variability to a lower number
of independent dimensions. Usually, the few first principal com
ponents capture the main patterns of morphological variation
within the sample (FrieR, 2003). TpsRelw software (Rohlf, 1998a)
was used to perform this analysis.
In addition to relative warp analysis, canonical variates analysis
(CV A) was also used to discriminate among the different samples,
since this analysis maximizes the inter-group variability relative to
intra-group variability (Albrecht, 1980). An assignment test was
performed after the CV A in order to determine the utility of P3
shape in discriminating and determining the affinity of the groups
established a priori (Nolte and Sheets, 2005). CVA requires that the
number of specimens is at least as many as the number of variables
(Hammer and Harper, 2006). Given the small sample size of some
of the groups, a reduction of the number of variables included was
necessary. The relative warp analysis previously carried out pro
vided also a data reduction that allowed using a subset of the PCs
instead of the original variables (Klingenberg, pers. comm.). CVA is
able to be conducted in a reduced PCA space since the first PCs
capture the most important aspects of the morphological change
(Slice, pers. comm.).
Since the sample sizes of some fossil hominin taxa are too small
(with the exception of the H. heidelbergensis, H. neanderthalensis,
and H. sapiens samples), we merged some of the species into more
inclusive samples to perform the CV A. Thus, the three AustraI
opithecus species were pooled as Australopithecus sp., and the H.
habiIis and H. ergaster samples were pooled as African Homo. The
original Asian H. erectus, H. heidelbergensis, H. neanderthalensis, and
H. sapiens samples were retained. The Paranthropus sp., H. georgicus,
and H. antecessor specimens were not included in the CV A due to
the difficulty of including them in any of the groups cited above.
However, these specimens were later classified into one of the
previous groups based on the discriminant function results of the
analysis, and they were plotted in the CV A graph. Morphological
variants corresponding to the extremes of the CVs were generated
using TpsRegr (Rohlf, 1998b).
Landmarks and semiIandmar1<s
Landmarks are biologically and geometrically homologous
points among the studied specimens (Zelditch et al., 2004). The
occlusal morphology of a P3 crown consists of the two main cusps
comprising the trigonid and a talonid. The median longitudinal
fissure marks the boundary between the protoconid (buccal cusp)
and the metaconid (lingual cusp). The distobuccal and distolingual
fovea/transverse fissures and their intersections with the buccal
and lingual borders of the crown delimit the talonid area (Wood
and Uytterschaut, 1987). The four landmarks within the occlusal
outline (Fig. 2) were defined as follows (Biggerstaff, 1969; Wood
and Uytterschaut, 1987):
Landmark 1: Tip of the buccal cusp or protoconid, determined
Landmark 2: Posterior/distal fovea or the intersection of the
median longitudinal fissure (also named central groove) with the
distal fovea-fissures/transverse fissures (Biggerstaff, 1969; Wood
and Uytterschaut, 1987). When the presence of a transverse crest
erases the central groove, this landmark is located at the deepest
point of the distal fovea.
Landmark 3: Tip of the lingual cusp or metaconid, determined
Landmark 4: Anterior/mesial fovea or the intersection of the
median longitudinal fissure (or central groove) with the mesial
fovea fissures/transverse fissures (Biggerstaff, 1969; Wood and
Uytterschaut, 1987). When the central groove is not present, this
landmark is placed at the deepest point of the mesial fovea.
These landmarks were chosen because they can be accurately
located even when some representative traits of the premolars are
The use of semilandmarks was proposed by Bookstein to study
the shape of structures that lack real landmarks, such as curves or
outlines (Bookstein, 1996b, 1997; Bookstein et al., 1999). Whereas
real landmarks contain shape information in all the directions of
the coordinate system, semilandmarks are uninformative with
respect to their exact location along a curve or outline (Bookstein,
1996b; Zelditch et al., 2004; Bastir et al., 2006), but sliding tech
niques avoid this limitation (Bookstein, 1996b, 1997; Bookstein
et al., 2002; Gunz et al., 2005). The locations of the semilandmarks
are allowed to slide along the curve to produce a new set of sem
ilandmarks that either represents the smoothest possible mathe
matical deformation of the curve on the reference form to the
corresponding curve on a particular specimen (minimum bending
Fig. 2. Digitized image of a P3 of H. erectus showing the four landmarks: 1) protoconid
tip: 2) distal fovea: 3) metaconid tip: 4) mesial fovea. Four type III landmarks (the most
buccal, the most distal, the most lingual, and the most mesial points of the crown) are
also marked in the external contour of the tooth. These landmarks were employed to
draw the forty semilandmarks, which would be later located at the intersection of the
external outline and the fan lines (see text for explanation). Discontinuous line high
lights the contour of the talonid. B: buccal: D: distal: L: lingual: M: mesial.
energy; Bookstein, 1996b, 1997; Sheets et al., 2004), or minimizes
the Procrustes distance between the curve on the reference form
and each individual in the sample (Sampson et al., 1996). The
criterion employed in this analysis has been the minimization of
the Procrustes distance.
Landmarks were digitized by A. G.-R. The tips of the main cusps
were visually located in the images while simultaneously examin
ing the fossil or cast. The cusp tip was assumed to be in the center of
the wear facet in those teeth where wear had removed it (Bailey,
2004; Martinon-Torres et al., 2006; Gomez-Robles et al., 2007).
When mesial and/or distal borders of the teeth were affected by
light interproximal wear, original borders were estimated by ref
erence to overall crown shape and the buccolingual extent of the
wear facets (see Wood and Uytterschaut, 1987; Bailey and Lynch,
2005, Martinon-Torres et al., 2006). Those premolars that were
heavily worn were not included in the study. However, teeth with
a moderate degree of wear were used since they have been dem
onstrated to provide consistent results when studied together with
unworn teeth (Gomez-Robles et al., 2007).
Semilandmarks were drawn starting from four type III land
marks (extreme points in various dimensions which have at least
one deficient coordinate; Bookstein, 1991) that were later removed.
The chosen points were (Biggerstaff, 1969; Wood and Uytterschaut,
1987): the most buccal, the most distal, the most lingual, and the
most mesial points of the crown. The gravity center of these four
points was used as the central point to obtain forty equiangular fan
lines (Fig. 2) with MakeFan6 software (Sheets, 2001). These points
were used instead of the four studied landmarks to avoid an
oversampling of the lingual half of those premolars with
a predominantly lingual location of the landmarks. The points at
which the fan lines intersected the premolar outline provided
the initial location of the semilandmarks (before sliding). TpsDig
(Rohlf, 1998c) was later used to digitize the landmarks and
Allometry is the study of any links between shape and overall
size (Mosiman, 1970; Klingenberg, 1998), and several criteria have
been proposed to study its incidence and influence over the shape
of the organisms and their parts. We have focused on the study of
the evolutionary allometry, analyzing the correlation between size
and shape among taxonomically different populations, related
either by ancestor-descendant relationships or as sister groups
(Klingenberg, 1998; Bastir, 2004). For that reason, we have analyzed
allometry in an inter-specific context. Although previous works
have assumed that allometry has no influence on hominin dental
morphology (Bailey and Lynch, 2005), the use of geometric mor
phometrics by means of a multivariate regression of partial warps
and uniform component on centroid size, allows us to test that
hypothesis. Centroid size, defined as the squared root of the sum
med squared distance between the centroid (or "gravity center" of
the landmarks configuration) and each of the landmarks (Zelditch
et al., 2004), was used as a proxy for overall size. Multivariate
regression was accompanied by a permutation test (n = 10000) to
evaluate the significance of the allometry using TpsRegr software
The evaluation of the measurement error was divided into two
parts to independently assess the amount of error due to the
location and digitizing of the landmarks and semilandmarks or due
to the orientation of the tooth during photography. A subsample of
five premolars (KNM-W T 15000, AT-1993 [left antimere not
included in the analyses]. AT-3243, Krapina114, and Krapina33) was
used to establish the error inherent in the method for data
sition set out above. These
among those that were available to ret)h()tOgr;:lDh
rep1resent.atl'Je of the
of wear. To assess the error due to the
process, the four landmarks and
over five consecutive
.. H", ..... "'"'u
In order to
error due to the location of the reference
of five consecutive
to the lens. as described
en�;uflmg that the CEl was
Then, the landmarks and the four
the semilandmarks were
molar. After that, the
After a Procrustes SUlper'lm.posltlOn of the described sul)samJJie,
the Euclidean distances of each landmark to its
landmark deviations were calculated relative to landmark means
The scatter at each landmark for each individual
was in order obtain a mean value of the error
orientation of the tooth when
to later locate
placed in the same anatomical
carTes;ponding to each pre
it and due to the
dl��ltlZtng process. We are aware of the fact that the measurement
of the f1H,n""t·�t,.,.n
underestimation when the dis;pe:rsion is not coHin-
Cramon-Taubadel et al., 2007).
nrf"'1(l""� an estimate of the
The mean value of the scatter at each landmark relative to the
landmark mean due to the
landmarks are included and 2.24% if the error is
the landmarks. This level of error is within the range .. "''', .... . .. t-''' rl
similar studies 2002; Harvati,
the lowest error, with a mean value of 0.45%,
of our method of
process is 0.63% when semi
0.89%. Since the error was
anism minimizes the error at semilandmarks, with values lJetwE'en
The amount of error due to the orientation of the ...... ".r ... " ... I.,,"<:
when are is
of 1.23% with the inclusion of semilandmarks and
landmarks alone. The mean error values for each individual are
similar and range from 1.02% to 1.59%.
error, the apex of the
the apex of the metaconid the lowest
at semilandmarks is reduced the
between and 2.43%.
error at the location of the apex of the pro-
toconid taken into account when
This error has a mean value
the results of
the location of the first
of the occlusal surface. The presence of
LV1 • .::II-"I ... '''-H .. I.::I talonid causes the asymmetry of the
related to the
mesiodistal axis. In teeth with a
of the maximum
Relative warps analysis showing
ten principal components
% cumulative variance
score, the occlusal pOlygon
cormelctirlg the four (anarrlanes
external contour of the
metaconid are dls:pl,lCed ITleSlaHy.
PC1 are almost syrnnletrical
with a smooth outline without any
and a small and
.... "",.,..,r.. ",.,. score an the PC2 is linked with a sVlnnletncal
crown, an oval outline. and
an PC2 is characterized
talonids but with small occlusal
extrernes of PCt and PC2 are both linked with
a celntrc3l1v-loC=ltt",d protacl)nidapex that results in a small and more
IJV1'Y,5'UIl, tncLep'en,dellt of the symmetry
or asymmetry of the outline.
PC1 and PC2 are correlated with relatl'velv
are located at the upper left QUclar,ant
values for PC1 and
hominin taxa are located
spE�Cllmens. three out of five H habilis. one H ergaster. and
The H erectus
""', ...... r'·"""values for PCt, but
neilc3rlve values for PC2. with the c .. w,�.'" .. � spE�cirnerls
and the Zhoukoudian spE�ClrnerlS
; .... .. . .... .. " ... t-., .. ,t- to note that mast of the African and Asian Pliocene and
the A anamensis
values for PC2. This is also the Qu,ldr,ant
the three Para-
values on PCt, the
and one H.
l1el.cleJfberJ;enSlls and H neanderthalensis
for PCt. The
flelcleJ!lJerJ;enSllS, and the Le Moustier and
are similar, except that the Sima de los Huesos
shows more extreme
marked reduction in size of the occlusal
are the three spE�ClrnerlS
The distribution patterns of those
values for PCt and PC2, with a more
.. " .....
. . .
1,00 ... H. heidelbergensis
/;;. H. neandenhalensis
:::J H. sapiens
Fig. 3. Projection of individual P3 crowns on PC1 and PC2. TPS-grids illustrate the morphological variation trends of the specimens along each principal component. These grids
show how a TPS transformation of the mean shape into a theoretical specimen would look if its PC-score were at an extreme point on the one PC axis and zero at all other axes. The
dotted line remarks the separation of early African and Asian specimens, characterized mainly by negative values for PC1, from more recent ones, which tend to show positive scores
for that Pc.
quadrant of the graph.
sample predominantly plots in the lower right
out of 34 modern human first
PC2, without noticeable differences between the modern human
sanlpic:'s from the AMNH and the
of Coimbra. Although
nr.e:>ty",nl:.r" are located in the upper
of the plot 3), there is no H
the upper left quadrant, where the majority of the
As canonical variates
cases of the smallest
ables (Hammer and Harper, 2006), the nine first
nents (which account for the 90.04% of the
sample) were chosen as variables for the CVA (since nine is the
size of the smallest group, namely the African Homo sam-
pie). CV A works upon an
are homogeneous. As two of the groups included in this
(Austmlopithecus sp. and African Homo) contain m()fpnOIO��IGlIl}I
diverse taxa, we determined whether the variance differs
cantly between the a defined groups. Although the Levene
statistic showed that the variances are homogeneous when the
variables are considered separately (p-values between 0.084 and
0.95), the hypothesis of covariance matrices for those nine
variables was when using Box's M statistic (p:s; 0.0001).
However, when the number of variables included in the CVA was
reduced to the first five PCs, this resulted in the homogeneity of the
covariance matrices (p = 0.078, Box's M) with only a small
the number of
than the number of vari-
that variances across groups
reduction in the percentage of variability of the
the CV A (77.1 % with five variables versus 902% with nine ,,:.t', :.1-11",,,
The CV A
similar results to the relative warps
analysis (PCA). The CVA extracted five canonical variables (Table 3)
of which the first two explain 93.6% of the variation (from the 77.1%
of the total variation of the
Specimens at the end of CVl
morphologies with conspicuous talonids and
and outlines. This difference is due to a reduction of
the talonid and to the relative
center of the tooth. Thus, the positive end of CVl
with rounded occlusal outlines and small and lingually-located
A P3 with an oval outline and relatively
occlusal polygon would plot at the
protoconid apex and the anterior fovea
garding to their location at the
on CV2 to a
talonid that mainly occupies the
COlmr:lre:sse�d occlusal polygon located mesio-lingually.
P3s with scores for CV2 have a uniform and
distal outline, whereas the mesial side tends to be more concave,
with an inflexion at the level of the
intersects the external outline.
In the distribution of the
coincides with the PCA distribution by showing a dear distinction
between the early hominins (Austmlopithecus sp., African Homo,
included in this
end show almost circular
of the protoconid tip to the
end of CV2, with the
end of CV2. Positive scores
P3s, with a reduced
portion of the crown
groove when it
on the CVA plot
." *' ...
'::: . African Homo
- Homo ersctus
A H. heidelbergensis
II H. neanderthalensis
:] H. sapiens
0 Paranthropus sp.
<> H. georgicus
X H. antecessor
Fig. 4. Canonical variates analysis. Individual P3 crowns are plotted in a way that maximizes the morphological differences among the a priori established groups (see text for
'-A�",w.au'JUI' As can be observed, from a better discrimination among species, no substantial differences are shown
Specimens not in the CVA were later assigned to one of the included groups to determine their affinity
for explanation). The P3 shape corresponding to each extreme of the CVs was obtained with TpsRegr software (Rohlf, 1998c).
and the canonical
CVAgraph (see text
and H erectus) and the more recent taxa. All Asian and African
spl�ClmE�ns have negative values for CV1, whereas almost all of the
European specimens and H
heidelbergensis from Arago, three H neanderthalensis, and one H.
(with the of three H
nr"'ty1,-,I:>,r" in the pooled
the most extreme negative scores for CV1, and they display a
trend to plot in values with respect to CV2. All African
Homo and H erectus
and share part of the
Along CV2 there is substantial overlap between African Homo and
located in the upper
This position corresponds to positive values for both CVs. However,
Arago individuals display either negative scores for CV1 (Arago 71
and Arago 75) or scores for both CV1 and CV2 (Arago
The de los Huesos
values for CV2. H neanderthalensis has
in the half of CV1
with 111-''' IlH""lll
(15 of 18
of the CV A plot.
has the most
positive values for
Canonical variates analysis (CVA)
% explained variance
% cumulative variance
CV1 and CV2, although they are not as extreme as in Sima de los
.... . ,,:>rI'r:>r,r of the CVA plot with two H neanderthalensis spec
imens situated close to the zero value for CV2. Thus, H.
can be discriminated from H
CV2, whereas the earliest groups (Austral
opithecus sp., African Homo, and H.
from the later Homo
(H heidelbergensis, H neanderthalensis,
is located in the lower
can be discriminated
sp., H hei-
��'��"b�"�'�' and H sapiens
their group, whereas the other groups have assignment percent
ages that are below 70% (Table 4). Nevertheless, it is important to
note that no individual of the first three groups (Australopithecus
sp., African Homo, and H. erectus) is wrongly
European fossil groups or to H.
are correctly to
to any of the
and one H. nean-
derthalensis (but not any H
the African or Asian groups.
are lATrnncrlu aSS:lgrlea to one of
Assignment test results based on the canonical variates analysis
% correct assignment
Number of correctly
The Sp€�ClfnerlS not included in the CV A were later ass:lgrlea to
one of the other groups in order to ascertain their m()rtJ;hcllo,;lcal
sp. (KNM-ER 3230 and
was classified as African Homo. Premolars from
Dmanisi were ass:lgrlea
were classified as H. erectus.
were to Austral-
whereas the other one
both H. antecessor spC�ClmE�ns
The multivariate re��re�;siclfi of the
allometric effect (p
be linked to differences in overall size. Hominins with small pre
molar crowns tend to have a circular and
combined with a
outlines as a consequence of a COl[1S�nCl]otlS
talonid The occlusal DOIlV1!()nS
2.90; H. erectus:
whereas the more recent
hominin taxa have values below 2.7 (H.
neanderthalensis: 2.64; H.
n::ucti::ll lhr related to an allometric
Thus, the differences in
that separate earlier from more recent
to the al
lometric effect, the differences between the P3
earlier and later hominins can be summarized
lometric decrease of the
als:pI<KemE�nt of the distal fovea
fovea-mesial fovea an allometric reduction of the talonid, and
a relative extension of the distobucal outline that
hominin ... .. ".rn'''' I'' .. '' an apIJro,xirnal:el:y round contour
the relative mi-
apex toward the center of the tooth
This of P3 crown mor-
ph,omletI'(CS has demonstrated clear differences between the
African and Asian hominin taxa, and later Homo
. . " .. "
' . .
.. • fI. '11 .. '" ..
". e •
Fig. 5. Morphological variants corresponding to
troid can be noted, the allometric factor
characterized by relative decrease of the talonid and
and foveae, well as by a relative buccodistal expansion.
neanderthalensis, and H.
are separate from H.
neanderthalensis, and H.
tTl(u'nhnlnCfU for the genus Homo consists of a
metrical occlusal outline, a
and H, neanderthalensis. From this
talonid, and an occlusal
than that seen in H. helde,ber
two derived .... . 'u· .... ,hnL
nel'aellJergf�nsls and H. nean
taxa have a small oc
pn)tocOllid located towards
occlusal outline. When present,
small and forms a smooth
of the crown.
Sima de los Huesos and H. neanderthalensis
in the mesial and distal sides of the
outline, with the distal side
it is much more frequent in the European Pleistocene
the results of the CVA show that both
convex and the mesial side
is also seen in H. sa-
nr" .... . ilt-n.'a. features associated
pfE�mOlcrs, However, these differences are rel-
rntY'l"'"rG.rI to the differences between earlier
and later fossil hominins. The inclusion of H.
neanderthalensis, and H.
tinction within the more archaic groups. The
tremes identified Wood and
tween KNM-ER 992 and KNM-ER 3230, are also
where these two
reduced the dis-
the contrast be-
up in our
U"I-."�,,1t1,, of the archaic groups.
included in this
ences among African spl2ClmE�ns
small sizes of these groups in our
have been defined as derived with
(e.g., Wood and
1987; Suwa, 1988; Suwa et aL, 1996;
our own results, .. II-I... ,..,,· . .... . h it is
However, a formal cladistic
not been carried out. and the size of the Dn.-nntnr"l'>"'"
still small to make formal conclusions with respect to
derived with respect to AU.stnaiopithelcus
Pmrnl1ttn:mrlH, spE�Cllnens, located at the extreme of the variation
of the groups included in the CV A.
Fig. 6. Morphological comparison of the described shapes: (a) ER 1802 and (b) ATD6-3 show the typical morphology of earlier hominins, whereas (c) AT-1466 and Cd) Coimbra226
illustrate the derived morphology. Size ratios are approximately kept among the four pictures.
In neither of the PCA nor CV A graphs is there a hominin taxon
with a morphology that is clearly intermediate between the earlier
and later groups. Some H ergaster specimens and one H georgicus
individual are close to the distribution areas of H heidelbergensis, H
neanderthalensis, and H sapiens, but the morphological variability
within the former groups is high. These specimens show a slight
reduction of the talonid and a less asymmetrical outline compared
to earlier African specimens. However, when we use the CV A to
assign the Omanisi specimens to one of the most represented
groups, one of them (0211) is classified as African Homo and the
other (02735) as Australopithecus sp., highlighting their similarity
with early Homo and even with Australopithecus in some traits by
retaining plesiomorphic characters (Rightmire et al., 2006; Lord
kipanidze et al., 2007; Martinon-Torres et al., in press).
The absence of overlap between the distribution areas of H
ergaster and H erectus in the PCA, with H ergaster plotting closer to
the morphospace of the later Homo taxa, would support the different
specific allocation of African and Asian lower Pleistocene specimens
(e.g., Wood, 1984,1994; Wood and Richmond, 2000). In the case of H
erectus, there are slight differences between the Chinese (from
Zhoukoudian) and Javanese (from Sangiran and Trinil) specimens.
Zhoukoudian premolars tend to show lower values for both PC2 and
CV2 than do Javan individuals, highlighting the Sangiran trend to
display a slightly more asymmetrical P3 within the H erectus group.
Zhoukoudian 20 and 80 belong to a mixed collection of small fossil
orang-utan and hominin teeth (Schwartz and Tattersall, 2003). Al
though we have kept the assignment to H erectus proposed by
Weidenreich (1937), both the general morphology and the location
of these premolars in the PCA plot (at the extreme of variation of the
studied hominin sample) could better support their classification as
non-hominin. However, a comparison with an ape sample would be
required to confirm that.
The H antecessor specimens are characterized by derived P 4 and
MI shapes (Martinon-Torres et al., 2006; Gomez-Robles et al.,
2007), but they retain primitive P3 shapes, with morphologies
similar to those found in early African and Asian species. The use of
the CVA to include these specimens in one of the most-represented
groups of the sample associates them with Asian H erectus. This
association could support the phylogenetic relationship between
Asian and European Pleistocene populations as proposed by some
authors (Oennell and Roebroeks, 2005, Martinon-Torres et al.,
2007), but the restricted ability of P3 to taxonomically allocate
isolated specimens prevents us from drawing final conclusions.
The Arago specimens (Arago 13, 71, and 75) plot outside the
distribution area of the Atapuerca-Sima de los Huesos premolars.
Interestingly, Arago premolars have slightly different morphol
ogies, and, despite sharing the same small area of the morphospace,
they show affinities with different groups (Bermudez de Castro
et al., 2003): Arago 71 is situated with the early hominin taxa (due
to its asymmetrical shape and large occlusal polygon [left upper
quadrant of Fig. 3]); Arago 13, located at the confluence of the four
quadrants (point 0) of Fig. 3, is more similar to later hominins
(because of its symmetrical shape with a centered and large oc
clusal polygon); and Arago 75 plots in an intermediate position
between modern and earlier groups, showing a medium-sized
talonid and a central occlusal polygon.
Although the H neanderthalensis and H heidelbergensis distri
butions overlap, the Sima de los Huesos first premolars plot at the
extreme of PC1. In general, H heidelbergensis from Sima de los
Huesos and H neanderthalensis have been described as having
similar dental traits, the only difference being that H nean
derthalensis presents higher degrees of expression for some of
those traits (e.g., Bermudez de Castro, 1987, 1988, 1993; Martinon
Torres, 2006; Martinon-Torres et al., 2006; Gomez-Robles et al.,
2007). It is interesting to note that, in this particular tooth, Sima de
los Huesos specimens display a morphology that is even more
pronounced than the classic Neandertals. This may reflect a mor
phological particularity of this biological population as its marked
reduction of the size of their posterior teeth (Bermudez de Castro
and Nicolas, 1995). The similarities in P3 shape, along with other
dental traits (Bermudez de Castro, 1987, 1988, 1993; Martinon
Torres, 2006; Martinon-Torres et al., 2006; Gomez-Robles et al.,
2007), support the hypothesis that there is a close phylogenetic
relationship between the hominins of Atapuerca-Sima de los
Huesos and the late Pleistocene classic Neandertals (e.g., Arsuaga
et al., 1993, 1997; Bermudez de Castro, 1993; Martinon-Torres,
As stated earlier, there are no significant differences between
the distribution areas of the two modern human samples analyzed
in this study. However, three out of the four early modern humans
studied (Qafzeh 9 and 11, and Grotte des Enfants 6) plot at the
extreme of the variability of the modern human sample (lowest
values for PC1 and PC2).
An allometric effect accounts for 17.3% of the shape differences
among the hominins. Given that the P3s of earlier hominin species
are generally larger than that of the more recent groups (Bermudez
de Castro and Nicolas, 1995, 1996), the archaic to modern mor
phological gradient could be partially linked to a size change from
large to small. However, that size change is not isometric. The re
duction of the overall size of the P3 crown is accompanied by a more
significant reduction of the distance between the protoconid and
metaconid apices and, specially, of the talonid area (Fig. 5). In
addition, the relative extension of the distobuccal outline gives
small P3S a more rounded contour.
Previous papers on P3 morphology have not found clear
evidence of allometry in the morphology of this tooth. Wood and
Uytterschaut (1987) did not find any significant correlation
between talonid size and overall crown size in any of the groups
they studied (A afticanus, P. robustus, P. boisei, and H. habiIis).
Leonard and Hegmon (1987) proposed that in A afarensis there is
an association between molarization of the P3 and premolar size,
but only for the female specimens. Similarly, Suwa et al. (1996)
assumed that the relative cusp proportions of Paranthropus were
not exclusively the result of an allometric change from non-robust
species. Our analysis of a more comprehensive sample that includes
comparatively smaller premolars, such as those from H. hei
delbergensis, H. neanderthalensis, and H. sapiens, suggests that there
is an allometric component affecting in some degree the evolution
of P3 morphology within the hominin clade.
Wood and Uytterschaut (1987) showed that Paranthropus pre
molars, with the most asymmetrical shape, tend to have additional
cusps on the talonid. This contrasts with Australopithecus and early
Homo specimens, which have a lower frequency of extra cusps, and
which have relatively less asymmetrical P3 crown outlines. These
differences in the degree of asymmetry are masked when more
recent (and symmetrical) specimens are included in the analysis.
Although initially we could correlate the presence of additional
cusps with a well-developed talonid, H. sapiens tend to keep their
symmetrical and rounded outline even when extralingual cusps are
developed (Kraus and Furr, 1953; Scott and Turner, 1997; Bailey,
2002b; Martinon-Torres, 2006). Therefore, even though the pres
ence and the number of additional cusps is highly variable and
useful for lower taxonomic distinctions Uernvall and jung, 2000),
the disposition of the main cusps and the general morphology of
the outline should be more powerful tools
It has been demonstrated that the morphogenesis of the upper
and lower dentition are under the control of different genetic
programs (Thomas et al.,1997; Ferguson et al., 1998; McCollum and
This could explain why M 1
tends to retain
a primitive morphology in H. sapiens (Gomez-Robles et al., 2007),
whereas P3 crowns are derived in this species. Although it has been
proposed that the teeth of the same class have a correlated ex
pression (Nichol, 1990; Irish, 2005), the evolutionary trend also
differs between the two types of lower premolar, since H. nean
derthalensis apparently retains a primitive P 4 morphology (Marti
non-Torres et al., 2006), whereas the shape of their P3 is clearly
derived (this study). Thus, if the different trends in the P3 and P4
were to be confirmed, it could be hypothesized that they are
influenced by different morphogenetic fields (Mizoguchi, 1981;
Kieser and Groeneveld, 1987; Bermudez de Castro and Nicolas,
1996). However, the classical concept of morphogenetic fields of
dentition (Butler, 1939; Dahlberg, 1945) seemingly represents the
concerted action of a series of individual molecular fields (Line,
2001), affecting different teeth.
The results of the assignment test do not reveal a strong dis
criminative power of the P3 morphology in determining the tax
onomical affinities of the individuals. Notwithstanding, the clear
differences among early hominin species and later Homo groups
may suggest the existence of ecological and evolutionary factors
underlying these shape changes. Genetic drift could be a plausible
mechanism for explaining the described shape differences (Lynch,
1989; Relethford, 1994; Roseman, 2004; Roseman and Weaver,
2004; Weaver et al., 2007), although the morphological distinction
in P3 morphology between earlier and later hominin species
compels us to consider the existence of evolutionary advantages of
the described shapes in each case that probably would not have
adaptive benefits per se, but which could be correlated with other
skeletal changes (McCollum and Sharpe, 2001).
This geometric morphometric study of the hominin P3 crown
morphology has revealed a noticeable change in shape from the
earliest hominin species sampled to the later Homo species. The
inferred primitive morphology, typical of Australopithecus and early
Homo individuals, as well as of Asian H. erectus specimens, consists
of a strongly asymmetrical outline combined with a large and well
developed talonid and a generally expanded occlusal polygon. The
two derived morphologies both have an approximately symmet
rical outline with a reduced or absent talonid. In H. neanderthalensis
and H. heidelbergensis the P3 crowns have a small and lingually
located occlusal polygon, whereas in H. sapiens specimens the
occlusal polygon is larger and more central, and the outline is ap
The evidence of a significant inter-specific allometric effect on
the shape change and the obvious separation between the earliest
hominin species of the sample and the later ones (H. hei
delbergensis, H. neanderthalensis, and H. sapiens) suggest that there
may be an ecological influence on P3 morphology.
When the results of this study of the first premolar morphology
is compared to the results of studies of other teeth, it is evident that
a simple explanation cannot be applied to these accumulated
findings, but that it is necessary to consider a complex mosaic
pattern for the evolution of the human dentition.
We are grateful to all members of the Atapuerca research team,
especially to the Sima de los Huesos excavation team for their ar
duous and exceptional contribution. We also thank I. Tattersall, G.
Sawyer, and G. Garcia from the American Museum of Natural
History, New York; O. Kullner, B. DenkeI, F. Schrenk and Luca Fior
enza from Senckenberg Institute, Frankfurt, Germany; J. de Vas
from Naturalis Museum, Leiden, the Netherlands; D. Lordkipanidze,
A. Vekua, G. Kiladze, and A. Margvelashvili from the Georgian
National Museum; J. Svoboda and M. Oliva from the Institute of
Archaeology-Paleolithic and Paleoethnology Research Center, Dolni
Vestonice, Czech Republic; G. Manzi from Universita La Sapienza,
Rome; E. Cunha from Universidade de Coimbra and M.A. de Lumley
from Centre Europeen de Recherches, Tautavel, France for pro
viding access to the studied material and their helpful assistance
when examining it. Special thanks to James Rohlf at SUNY, Stony
Brook, for his advice regarding methodological aspects. We spe
cially grateful to Ana Muela for her technical support and for
photographing part of the sample. We are indebted to Susan Antan
for her thorough editing of the manuscript and her helpful com
ments, as well as those provided by three anonymous referees,
which greatly improved the manuscript. We are specially grateful
to Gisselle Garcia from the AMNH who has kindly provided the
information on the AMNH modern collection. Thanks to Miguel
Botella from the Laboratory of Anthropology of the Universidad de
Granada for his continuous support. Special thanks to David San
chez for his invaluable help with technical questions. This research
was supported by funding from the Direcci6n General de Inves
tigaci6n of the Spanish M.E.C., Project No. CGL2006-13532-C03-03j
BTE, Spanish Ministry of Science and Education, Fundaci6n Ata
puerca, and Fundaci6n Duques de Soria. Fieldwork at Atapuerca is
supported by Consejeria de Cultura y Turismo of the Junta de
Castilla y Lean. A. G-R. has the benefit of a predoctoral FPU grant of
the Spanish MEC. This research was partly carried out under the
Cooperation Treaty between Spain and the Republic of Georgia,
hosted by the Fundacian Duques de Soria and the Georgian Na
Adams, D.C, Rohlf, EJ., Slice, D.E., 2004. Geometric morphometrics: ten years of
progress following the 'Revolution'. Ita!. J. Zoo!. 71, 5-16.
Albrecht, G.H., 1980. Multivariate analysis and the study of form with special
reference to canonical variate analysis. Am. Zoo!. 20, 679-693.
Arsuaga, J.L, Martinez, I., Gracia, A, Carretero, J.M., carbonell, E., 1993. Three new
human skulls from the Sima de los Huesos Middle Pleistocene site in Sierra de
Atapuerca, Spain. Nature 362, 534-537.
Arsuaga, J.L, Martinez, I., Gracia, A, Lorenzo, C, 1997. The Sima de los Huesos crania
(Sierra de Atapuerca, Spain): a comparative study. J. Hum. Evo!. 33, 219-28 1.
Bailey, S.E., 2000. Dental morphological affinities among late Pleistocene and recent
humans. Dent. Anthropo!. 14, 1 -8.
Bailey, S.E., 2002a. A closer look at Neandertal postcanine dental morphology: I. the
mandibular dentition. New Anat. 269, 148-156.
Bailey, S.E., 2002b. Neandertal dental morphology: implications for modern human
origins. Ph.D. Dissertation, Arizona State University.
Bailey, S.E., 2004. A morphometric analysis of maxillary molar crowns of Middle
Late Pleistocene hominins. J. Hum. Evo!. 47, 183-198.
Bailey, S.E., Lynch, J.M., 2005. Diagnostic differences in mandibular P4 shape
between Neandertals and anatomically modern humans. Am. J. Phys. Anthro
po!. 126, 268-277.
Bailey, S.E., Pi lb row, V.c, Wood, B.A., 2004. Interobserver error involved in
independent attempts to measure cusp base areas of Pan M1s. J. Anat 205,
Bailey, S.E., Wood, B.A., 2007. Trends in postcanine occlusal morphology within the
hominin c1ade: the case of Paranthropus. In: Bailey, S.E., Hublin, J.-J. (Eds. ),
Dental Perspectives on Human Evolution. Springer-Verlag, Berlin, pp. 53-64.
Bastir, M., 2004. Analisis de morfometria geometrica de la variaci6n e integraci6n
morfol6gica en el craneo humano y sus implicaciones para los hominidos
de Atapuerca-SH y la evoluci6n de los Neandertales. Ph.D. Dissertation,
Universidad Autonoma de Madrid.
Bastir, M., Rosas, A, O'Higgins, P., 2006. Craniofacial levels and the morphological
maturation of the human skull. J. Anat. 209, 637-645.
Bermiidez de Castro, J.M., 1987. Morfologia comparada de los dientes humanos
f6siles de Ibeas (Sierra de Atapuerca, Burgos). Estudios Geol6gicos 43, 309-333.
Bermiidez de castro, J.M., 1988. Dental remains from Atapuerca/lbeas (Spain) II.
Morphology. J. Hum. Evo!. 17, 279-304.
Bermudez de Castro, J.M., 1993. The Atapuerca dental remains: new evidence
( 1987-1991 excavations) and interpretations. J. Hum. Evo!. 24, 339-371.
Bermudez de Castro, J.M., Arsuaga, J.L, Carbonell, E., Rosas, A, Martinez, I.,
Mosquera, M., 1997. A hominid from the Lower Pleistocene of Atapuerca, Spain:
possible ancestor to Neandertals and modern humans. Science 276, 1392-1395.
Bermiidez de Castro, J.M., Martin6n-Torres, M., Sarmiento, S., Lozeno, M., 2003.
Gran Dolina-TD6 versus Sima de los Huesos dental samples from Atapuerca:
evidence of discontinuity in the European Pleistocene population? J. Archaeo!.
Sci. 30, 1421 -1428.
Bermudez de Castro, J.M., Nicoi.ls, M.E., 1995. Posterior dental size reduction in
hominids: the Atapuerca evidence. Am. J. Phys. Anthropo!. 96, 335-356.
Bermiidez de Castro, J.M., Nicolas, E., 1996. Changes in the lower premolar-size
sequence during hominid evolution: phylogenetic implications. Hum. Evo!. 11,
Bermudez de Castro, J.M., Rosas, A, Nicolas, M.E., 1999. Dental remains from Ata
puerca-TD6 (Gran Dolina site, Burgos, Spain). J. Hum. Evo!. 37, 523-566.
Bernal, V., 2007. Size and shape analysis of human molars: comparing traditional
and geometric morphometric techniques. Homo 58, 279-296.
Biggerstaff, RH., 1969. The basal area of posterior tooth crown components: the
assessment of within tooth variation of premolars and molars. Am. J. Phys.
Anthropo!. 31, 163-170.
Bookstein, EL, 1989. Principal warps: thin-plate splines and the decomposition of
deformations. IEEE Trans. Pattern. Ana!. Mach. Intel!. 11, 567-585.
Bookstein, EL, 1991. Morphometric Tools for Landmark Data. Cambridge University
Bookstein, EL, 1996a. Combining the tools of geometric morphometrics. In:
Marcus, LE, Corti, M., Loy, A, Naylor, G.j.P., Slice, D. (Eds.), Advances in Mor
phometrics. Plenum Press, New York, pp. 131-151.
Bookstein, EL, 1996b. Applying landmark methods to biological outline data. In:
Mardia, KV., Gill, CA, Dryden, I.L (Eds.), Image Fusion and Shape Variability
Techniques. Leeds University Press, Leeds.
Bookstein, F.L, 1997. Landmark methods for forms without landmarks: morpho
metrics of group differences in outline shape. Med. Image Ana!. 1, 225-243.
Bookstein, EL, Sampson, P.D., Connor, P.D., Streissguth, AP., 2002. Midline corpus
callosum is a neuroanatomical focus of fetal alcohol damage. Anat Rec. New
Anat. 269, 162-174.
Bookstein, EL, Schafer, K, Prossinger, H., Seidler, H, Fieder, M., Stringer, C,
Weber, G.w., Arsuaga, J.L, Slice, D.E., Rohlf, F.j., Recheis, w., Mariam, AJ.,
Marcus, LE, 1999. Comparing frontal cranial profiles in archaic and modern
Homo by morphometric analysis. Anat. Rec. New Anat 257, 217-224.
Butler, P.M., 1939. Studies of the mammalian dentition: differentiation of the post
canine cheek dentition. Proc. Zoo!. Soc. Lond. 109B, 1-36.
Coppens, Y., 1977. Evolution morpholgique de la premiere premolaire inferieure
chez certains Primates su perieurs. CR. Acad. Sci. Paris, Ser D 285, 1299-13 02.
von Cramon-Taubadel, N., Frazier, B.C, Lahr, M.M., 2007. The problem of assessing
landmark error in geometric morphometrics: theory, methods, and modifica
tions. Am. J. Phys. Anthropo!. 134, 24-35.
Dahlberg, AA, 1945. The changing dentition in man. J. Am Dent. Assoc. 32,
Dennell, R., Roebroeks, W., 2005. An Asian perspective on early human dispersal
from Africa. Nature 438, 1099-1104.
Dryden, I.L, Mardia, KV., 1998. Statistical Shape Analysis. WHey, Chichester.
Dubois, E., 1894. Pithecanthropus erectus: eine Menschenaehnlich Uebergangsform
aus Java. Landsdrukerei, Batavia.
Ferguson, CA, Tucker, AS., Christensen, L, Lau, AL, Matzuk, M.M., Sharpe, P.T.,
1998. Activin is an essential early mesenchymal signal in tooth development
that is required for patterning of the murine dentition. Genes Dev. 12,
Foley, R, Lahr, M., 1997. Mode 3 technologies and the evolution of modern humans.
Camb. Archaeo!. J. 7, 3-36.
Frieg, M., 2003. An application of the relative warps analysis to problems in
human paleontology: with notes on raw data quality. Image Ana!. Stereo!. 22,
Gabunia, LK, Lumley, M.A., deVekua, A, Lordkipanidze, D., de Lumley, H., 2002.
Decouverte dun nouvel hominide a Dmanisi (Transcaucasie, Georgie). CR
Palevo!. 1, 243-253.
Gharaibeh, W., 2005. Correcting for the effect of orientation in geometric
morphometric studies of side-view images of human heads. In: Slice, D.E. (Ed.),
Modern Morphometrics in Physical Anthropology. Kluwer Academic/Plenum
Publishers, New York, pp. 117-143.
Gomez-Robles, A, Martinon-Torres, M., Bermudez de Castro, J.M., Margvelashvili, A,
Bastir, M., Arsuaga, A, Perez-Perez, A, Estebaranz, F., Martinez, L, 2007. A geo
metric morphometric analysis of hominin upper first molar shape. J. Hum. Evo!.
Grine, EE., Franzen, J.L, 1994. Fossil hominid teeth from the Sangiran Dome Oava,
Indonesia). Cour. Forsch. Inst. Senckenberg 171, 75-103.
Groves, CP., Mazak, V., 1975. An approach to the taxonomy of the Hominidae:
gracile Villafranchian hominids of Africa. Casopis Min. Geo!. 20, 225-247.
Gunz, P., Mitteroecker, P., Bookstein, F.L, 2005. Semilandmarks in three dimensions.
In: Slice, D. (Ed.), Modern Morphometrics in Physical Anthropology. Kluwer
Academic/Plenum Publishers, New York, pp. 73-98.
Hammer, 0., Harper, D., 2006. Pale ontological Data Analysis. Blackwell Publishing,
Harris, E.E, Dinh, D.P., 2006. Intercusp relationships of the permanent maxillary
first and second molars in American whites. Am. J. Phys. Anthropo!. 130,
Harvati, K, 2003. The Neanderthal taxonomic position: models of intra- and inter
specific craniofacial variation. J. Hum. Evo!. 44, 107-132.
Hlusko, LJ., 2004. Protostylid variation in Australopithecus. J. Hum. Evo!. 46,
Hlusko, LJ., Do, N., Mahaney, M.C, 2007. Genetic correlations between mandibular
molar cusp areas in baboons. Am. J. Phys. Anthropo!. 132, 445--454.
Irish, J.D., 1993. Biological affinities of Late Pleistocene through modern African
aboriginal populations: the dental evidence. Ph.D. Dissertation, Arizona State
Irish, j.D., 1997. Characteristic Download full-text
African populations. Am.
Irish, j.D., 1998. Ancestral
origins of modern humans.
Irish, ].D., 2005. Pnl ;"I.�tir " ... ccmtinuity
among late PaleollthlC
Irish, ].D., Guatelli-Steinberg, D., 2003. Ancient teeth and modern human
expanded comparison of African Plio-Pleistocene and recent
J. Hum. Evol. 45, 113-144.
H.-S., 2000. Genotype, phenotype, and de1/ellJprne11tal biology of
Kieser, JA, Groeneveld, HT., 1987. Tooth size and arcadal length correlates in man.
1nL j. Anthropo!. 2, 37-46.
Klingenberg, CP., 1998. Heterochrony and allometry: the analysis of evolutionary
Kraus, B.S., Furr,
cation of discrete
La hr, M.,
delmogrclphy, and diversity in recent human evolution. Yearb. Phys. Anthropo!.
and low-frequency dental traits in sub-Saharan
Anthropol. 102, 455-467.
traits in recent sub-Saharan Africans and the
Hum. Evo!. 34, 8 1 -98.
disc011tillUity revisited: dental affinities
Am. J. Phys. Anthropol.
characters. Yearb. Phys. Anthropo!. 43,
BioI. Rev. 73, 79-123.
Lower first premolars. Part I. A definition and classifi-
traits. ].D. Res. 32, 554-564.
a theory of modern human origins: geography, R, 1998.
T., Vekua, A, Ponce de Lean, M.s., Zollikofer, CP.E.,
j., Kahlke, R., Kiladze, G.,
Inoradse, M., Rook, L, 2007. Postcranial evidence from early Homo from
Nature 449, 305-310.
Ph�Y'loge11etjc hypotheses under the assumption of neutral quan
Evolution 43, 1 -17.
Human evolution at the Matuyama-Brunhes boundary. Evol.
13, 11 -24.
M., 2006. Evoluci6n del
los dientes humanos del Pleistoceno
Dissertation, Universidad de
Martin6n-Torres, M., Bastir, M.,
through geometric morphometric analysis.
Martin6n-Torres, M., Bermudez de Castro, J.M., G6mez-Robles, A, Arsuaga, j.L,
Carbonell, E., Lordkipanidze, D., Manzi, G.,
evidence on the hominin dispersals during the
U.s.A 1 04, 13279-13282.
Martinan-Torres, M., Bermudez de Castro, j.M.,
de�;cr:iption and comparative study. J. Hum.
R., Oms, 0., Tappen, M., Bukhsianidze, M.,
B., Mouskhelishvili, A,
dental en hominidos: estudio de
la Sierra de Atapuerca (Burgos). Ph.D.
de Castro, j.M., G6mez, A, Sarmiento, S.,
j.L, 2006. Hominin lower second premolar mr,rr,t\nln",,'
A, 2007. Dental
Proc. Natl. Acad. Sci.
A, Gamez-Robles, A
from Dmanisi Vekua, A Dental of
Ml:Wg'ucI11, Y., 198 1. Variation units
Mus. Tokio 7, 29-39.
lVlL1)()(I-L.eU_IIl, J., Boccone, S., 2007. Maxillary molar
australopithecines. In: Bailey, S.E., Hublin,
tives on Human Evolution. Springer-Verlag, Berlin, pp.
Moggi-Cecchi, J., Grine, H., Tobias, p.v., 2005. Early hominid dental remains from
Members 4 and 5 of the Sterkfontein Formation
Hum. Evo!. 50,
allometry: size and
Nicho!, CR, 1990. Dental genetics and
of Arizona. Ph.D. Dissertation, Arizona
Nolte, AW., Sheets, H.D., 2005. Shape based assignment tests
phenotypes in natural sculpin hybrids (Teleostei, Scc)rp,aerlito.rm,es, Cottidae).
Frontiers in Zoology 2, 11.
Perez, S.I., Bernal, V., Gonzalez, P.N., 2006. Differences between sliding semi-land
mark methods in geometric morphometrics, with an application to human
craniofacial and dental variation. j. AnaL 208, 769-784.
Pilbrow, V., 2006. Population systematics of chimpanzees using molar morpho
heldelbergertSlS. Evo!. Anthropol. 6, 218-227.
comparative studies and
Dmanisi, Republic of Georgia. J. Hum.
Rohlf, F.j., 1996. Morphometric spaces, shape components and the effects of linear
transformations. In: Marcus, LE (Ed.), Advances in Morphometrics. Plenum
Press, New York, pp. 117-128.
genetics and early hominid
the human permanent dentition. Bull. Nat.
variables with characteriza-
lognormal and generalized gamma distrilmtionls. J. Am. Stat. Assoc.
Hum. Evo!. 51, 646-662.
1994. Craniometric variation among modern human populations.
1998. evolution in the Middle Pleistocene: the role of Homo
Lordkipanidze, D., Vekua, A, 2006. Anatomical
of the hominin
Rohlf, Ej., 1998a. TpsRelw. Ecology and Evolution, SUNY. Stony Brook, New York.
http: //life. bio.sunysb.edu!morphf.
Rohlf, F., 1998b. TpsRegr. Ecology and Evolution, SUNY. Stony Brook, New York.
http: //life. bio.sunysb.edu!morphf.
Rohlf, E, 1998c. TpsDig. Ecology and Evolution, SUNY. Stony Brook, New York. http)f
life. bio. sunysb.ed u/morphf.
Rohlf, Ej., Slice, D., 1990. Extensions of the Procrustes method for the optimal
sUIlenlmIlos:ltIon of landmarks. Syst. Zool. 39, 40-59.
modern human cranial form by
data. Proc. Natl. Acad. Sci. U.s.A 101, ' ''' (]IU'>- JLCO''':::l .
Roseman, Cc., Weaver, T.D., 2004. Multivariate apportionment of global human
craniometric diversity. Am.
P.D., Bookstein, EL,
ventricular outlines from contrast ventriculograms. In: Marcus, LE,
Corti, M., Loy, A., Naylor, G.J.P., Slice, D.E. (Eds.), Advances in Morphometrics.
Plenum, New York, pp. 131-152.
Schoetensack, 0., 1908. Der Unterkiefer des Homo
von Mauer bei Heidelberg. Wilhelm Engelmann,
Schwartz, j.S., Tattersall, I., 2003. The human fossil
Scott, G.R., Turner 11, CG., 1997. The Anthfiopc)lo�r;y
Morphology and its Variation in
integrated morphometric package. http: //www.canisius.
2004. Detection of natural selection on
Anthropo!. 125, 257-263.
H, Bolson, E.L, 1996. Eigenshape analysis of
aus den Sanden
vo!. 2. Wiley-Liss, New
Mitchell, CE., 2004. A combined landmark and outline
based approach to ontogenetic shape change in the Ordovician Trilobite
Triarthrus becki. In: Elewa, A. (Ed.), Applications of Morphometrics in Paleon-
Singleton, M., 2002.
Cercopithecinae). J. Hum. Evo!. 42, 547-578.
C, 2002. Modern human origins: progress and prospects. Phi!. Tram. R
Lond. B 357, 563-579.
human evolution. J. Hum. Evol. 37, 873-877.
evidence from mandibular
Suwa, G., White, T.D., Howell, F.C, 1996. Mandibular postcanine dentition from the
Formation, Ethiopia: crown morphology, taxonomic allocations, and
Pli()-Plei�;to(:enle hominid evolution. Am. J. Phys. Anthropo!. 101, 247-282.
Thomas, B.L, Tucker, AS., Qui, M.,
Sharpe, P.T., 1997. Role of DIx-1
dentition. Development 124, 48 11-48 18.
Tobias, p.v., 1991. Olduvai Gorge. The Skulls, Endocasts and Teeth of Homo habilis.
vo!. 4: Parts V-IX. Cambridge University Press, Cambridge.
Turner 11, CG., 1969. Microevolutionary interpretations from the dentition. Am. J.
Turner 11, CG.,
logical traits of the permanent dentition: the Arizona State
Anthropology System. In: Kelley, M., Larsen, C (Eds.), Advances
Anthl·opolc)gy. Wiley Liss, New York,
Weaver, T.D., Roseman, CC, Stringer, CB., 2007. Were Neandertal and modern
human cranial differences produced by natural selection or genetic drift? J.
Hum. Evo!. 53, 135-145.
Weidenreich, E, 1937. The dentition of Sinanthropus
odontography of the hominids. PalaeonL Sin. New
White, T.D., Suwa, G.,
BA, 1984. The
Wood, B., 1992. Origin
Wood, BA, 1994. Taxonomy and evolutionary relationships of Homo erectus. Cour.
Forsch. inst Senck. 171, 159-165.
Wood, B., Abbott, SA, 1983. Analysis of the dental morphology of Plio-Pleistocene
hominids I. Mandibular molars: crown area measurements and morphological
traits. ]. Anat. 136, 197-219.
Wood, BA, Abbott, S.A., Graham, S.H, 1983. Analysis of the dental morphology of
Plio-Pleistocene hominids 11. Mandibular molars-study of cusp areas, fissure
pattern and cross sectional shape of the crown.
Wood, BA, Abbott, SA, Uytterschaut, H., 1988.
of Plio-Pleistocene hominids IV. Mandibular postcanine root morphology. J.
Anat. 156, 107-139.
Wood, BA, Collard, M., 1999. The human
Wood, BA, Richmond, B.G., 2000. Human evolution: taxonomy and paleobiology. J.
Anat. 196, 19-60.
Wood, B.A., Uytterschaut, H., 1987. Analysis of the dental morphology of Plio
Pleistocene hominids Ill. Mandibular premolar crowns.
Zelditch, M.L, Swiderski, D.L, Sheets, H.D., Fink, W.L,
metrics for Biologists: A Primer. Elsevier Academic Press, San Diego.
New York, pp. 67-81.
shape variation in the Papionini (Primates:
1999. New age estimates for the Swanscombe hominid, and
Evolution of the "robust" in the Omo succession:
In: Grine, EE.
New of the
CA., Hardcastle, Z., Rubenstein, j.L,
Dlx-2 genes in patterning of the murine
30, 421 -426.
CR, Scott, G.R, 1991. Scoring procedures for
RE., 1993. The
University Press, Harvard.
Homo erectus Skeleton. Springer-
D 1, 1 -180.
S., Asfaw, B., 2000. Jaws and teeth ofAustmlopithecus
of Homo erectus.
evolution of the genus Homo. Nature 355, 783-790.
Am. j. Phys.
AnaL 137, 287-314.
of the dental morphology
Science 284, 65-71.
of the dental CA., 1988.
V. Maxillary postcanine tooth mClrpholloJ!:V.
AnaL 154, 121-156.