Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Species Resolution in Triceratops

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Journal of Vertebrate Paleontology 16(2):259-270, June 1996

1996 by the Society of Vertebrate Paleontology

Page charges paid for by


THE
DINOSAUR

SOCIETYCB>

SPECIES RESOLUTION IN TRICERATOPS: CLADISTIC AND MORPHOMETRIC APPROACHES


CATHERINE A. FORSTER
Department of Organismal Biology and Anatomy, University of Chicago, 1025 E. 57th Street, Chicago, Illinois 60637*
ABSTRACT-Sixteen species of Triceratops have been proposed since the genus was erected by O. C. Marsh in 1889.
Five of these species are here considered technically invalid and a sixth is reassigned to Diceratops. Based on both
cladistic analysis and morphometric shape analysis, all available skulls of the ten remaining "species" of Triceratops
are placed in one of two species: T. horridus and T. prorsus, invalidating eight of the original ten species. These two
species overlap in geographic and stratigraphic ranges. Because specimens of T. horridus greatly outnumber those of
T. prorsus, these morphotypes may represent distinct taxa rather than a single sexually dimorphic species.

ley, Berkeley; YPM, Peabody Museum of Natural History,


New Haven.

INTRODUCTION

Triceratops was one of the largest and most numerous herbivorous dinosaurs in western North America at the end of the
Cretaceous. Based largely on well preserved skulls from Maastrichtian formations (Lance, Hell Creek, Scollard, Frenchman)
in Wyoming, Colorado, Montana, South Dakota, Alberta, and
Saskatchewan, sixteen species of Triceratops have been named.
Because of its abundance, apparent diversity, and late temporal
occurence, Triceratops has figured prominently in studies of
dinosaur abundance and diversity near the Cretaceous-Tertiary
boundary in North America (e.g., Van Valen and Sloan, 1977;
Sloan et al., 1986; Sheehan et al., 1991).
In 1986, Ostrom and Wellnhofer redescribed the holotype of
T. brevicornis (BSP 1964 1 458) and provided a historical account of Triceratops systematics.After a consideration of individual variation, sexual dimorphism, and ontogenetic effects,
they concluded that Triceratops was monospecific. A complete
analysis of this hypothesis, however, was beyond the scope of
their paper (see also Ostrom and Wellnhofer, 1990). Their paper
underlined the modern biological species concept, and called
for its application to the problem of species in Triceratops.
This study combines cladistic and morphometric information
in a species analysis of Triceratops. The novelties that characterize species occur by way of continuous shape change and
the appearance of discrete characters. Shape, or quantitative,
information is explored through multivariate morphometric
shape analysis. Qualitative characters are examined by cladistic
analysis. AH available specimens of Triceratops (Table 1) were
examined to: (1) determine the taxonomic validity of named
species, (2) rediagnose the species based on groupings discovered by analysis, and (3) discuss the diversity and range of
variation within the genus.
Institutional abbreviations include: AMNH, American Museum of Natural History, New York; BSP, Bayerische Staatsamlung fr PaHiontologie und Historische Geologie, Munich;
NMC, Canadian Museum of Nature, Ottawa; CM, Carnegie
Museum of Natural History, Pittsburgh; FMNH, Field Museum
of Natural History, Chicago; MCZ, Museum of Comparative
Zoology, Cambridge; USNM, U.S. National Museum of Natural History, Washington, D.C.; LACM, Natural History Museum of Los Angeles County, Los Angeles; SMM, Science Museum of Minnesota, St. Paul; SnSM, South Dakota School of
Mines, Rapid City; UCMP, University of California at Berke-

* Present address: Department of Anatomical Sciences, Health Sciences Center, State University of New York at Stony Brook, Stony
Brook, NY 11794.

SYSTEMATIC HISTORY OF TRICERATOPS

Named Species
The first specimens attributed to Triceratops were a pair of
partial supraorbital horns collected in the "Denver Fm." of Colorado and erroneously referred to the mammalian genus Bison
by O. C. Marsh (1887). In the following year, J. B. Hatcher
collected a partial skull from the Lance Formation, Niobrara
County, Wyoming, which Marsh (1889a) originally described
as Ceratops horridus. Later that year Marsh (1889b) erected
the genus Trice ratops, with T. horridus as the type species, and
renamed the "bison" specimen Triceratops alticornis.
Over the next eight years, Marsh named eight more species:
T. galeus (1889b), T. flabellatus (1889b), T. serratus (1890a),
T. prorsus (1890a), T. sulcatus (1890b), T. elatus (1891), T.
calicornis (1898), and T. obtusus (1898). Six additional species
were described by subsequent authors: T. brevicornis (Hatcher,
1905), T. hatcheri (Lull in Hatcher et al., 1907), T. ingens (Lull,
1915), T. maximus (Brown, 1933), T. eurycephalus (Schlaikjer,
1935), and T. albertensis (Sternberg, 1949).
Of these sixteen species, four are based on inadequate material: T. galeus (an isolated nasal horn; USNM 2410), T. alticornis (a pair of supraorbital horns; USNM 4739), T. sulcatus
(a pair of supraorbital horns; USNM 4276), and T. maximus
(eight dorsal vertebrae; AMNH 5040). These specimens are
nomina dubia and cannot be assigned to Triceratops with certainty. A fifth species, T. ingens (YPM 1828), is an unpublished
name of Marsh's mentioned by Lull (1915). The type specimen
is a large, partial skull assignable to Triceratops. Lull mentioned T. ingens without description or diagnosis, and it is considered here a nomen nudum.
Triceratops hatcheri, was originally described by Lull (in
Hatcher et al., 1907) as Diceratops hatcheri, then later synonymized with Triceratops (Lull, 1933). In this study, Diceratops
is considered a valid taxon and T. hatcheri is removed again to
this taxon. This revision is discussed in the following section.

Validity oC Diceratops
The taxon Diceratops (USNM 2412) consists of a single
skull without lower jaws or postcrania. Generic characters were
originally described as: nasal horn absent, squamosals pierced
by large fenestra, parietal with small fenestrae, and a squamosal
lacking an inferior jugal notch. Nasal horns are also occasionally missing in specimens of Triceratops due to the lack of
fusion between the epinasal ossification and the nasal boss (For-

259

260

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 16, NO. 2, 1996

TABLE 1. Specimen numbers, species assignments, geographic and stratigraphic information, and basal skull length (BSL) for specimens used
in this study. Abbreviations: H, holotypes; e, estimated measurement; NA, not available (skull either incomplete or inaccessible).

Specimen

Original species
assignment

This study

Location

Formation

BSL (cm)

AMNH 5116
AMNH 970
BSP 1964 1 456
CM 1221
FMNH P12003
LACM 7207
MCZ 1102
NMC 8862
SDSM 2760
SMM P62/1/1
UCMP 113697
USNM 1201
USNM 2100
USNM 4720
USNM 4928
YPM 1820
YPM 1821
YPM 1822
YPM 1823

elatus
serratus
brevicornis (H)
brevicornis
calicornis
sp.
eurycephalus (H)
albertensis (H)
horridus
prorsus
horridus
elatus (H)
elatus
obtusus (H)
calicornis (H)
horridus (H)
flabe llatus (H)
prorsus (H)
serratus (H)

horridus
horridus
prorsus
prorsus
horridus
prorsus
nomen dubium
nomen dubium
horridus
horridus
horridus
horridus
horridus
horridus
horridus
horridus (H)
horridus
prorsus (H)
horridus

Niobrara Co., Wyo.


western Montana
Niobrara Co., Wyo.
western Montana
SW Montana
western Montana
Goshen Co., Wyo.
southern Alberta
western Montana
western Montana
McCone Co., Montana
Niobrara Co., Wyo.
Niobrara Co., Wyo.
Niobrara Co., Wyo.
Niobrara Co., Wyo.
Niobrara Co., Wyo.
Niobrara Co., Wyo.
Niobrara Co., Wyo.
Niobrara Co., Wyo.

Lance
Hell Creek
Lance
Hell Creek
Hell Creek
Hell Creek
Lance
Scollard
Hell Creek
Hell Creek
Hell Creek
Lance
Lance
Lance
Lance
Lance
Lance
Lance
Lance

110 (e)
NA
NA
116
110
110
NA
NA
NA
127
NA
104 (e)
NA
NA
118
NA
106 (e)
88
102

ster, 1996). This likely accounts for the absent nasal horn in
Diceratops.
Squamosal fenestrae commonly occur in all chasmosaurine
ceratopsids (Chasmosaurus, Pentaceratops, Anchiceratops, Arrhinoceratops, Torosaurus) except Triceratops. In chasmosaurines, squamosal fenestrae are located posterior to the contact
with the paroccipital processes, are either uni- or bilateral (or
missing altogether in sorne specimens), are nearly uniform in
shape, and possess smoothly rounded margins. Squamosals of
Triceratops are uniformly thick and never exhibit fenestration
or thinning.
A large fenestra occurs in the left squamosal of Diceratops
immediately posterior to the paroccipital processes, and a small
oval fenestra pierces the right squamosal. Sorne authors (e.g.,
Lull, 1933) considered these fenestrae a pathological result of
injuries. Where the margins of the fenestrae are preserved in
Diceratops they are smoothly rounded and lack any outward
sign of trauma. However, there is a pathological area of bone
along the left squamosal-parietal suture, immediately posterior
to the left fenestra. The margin of the fenestra near this callosity
is smooth and even, apparently uneffected by this injury. The
even nature of the squamosal fenestrae in Diceratops, coupled
with the ubiquitous appearance of fenestrae of similar shape,
size, and position in all chasmosaurines exclusive of Triceratops, indicate squamosal fenestrae are a real morphological entity not predicated by injury.
Diceratops also possesses parietal fenestrae, a character also
shared with all other chasmosaurines exclusive of Triceratops.
The right parietal has a narrow 16 cm. long opening. Much of
the margin of this fenestra is obscured by plaster, but enough
margin is preserved to discern its approximate size and confirm
its presence. Additionally, the parietal immediately surrounding
the fenestral margin is very thin. The left parietal is too poorly
preserved to provide additional information.
Diceratops shares with Torosaurus a unique configuration of
the frontal fontanelle. In these two taxa, a shallow channel extends posterolaterally from the frontal fontanelle (the frontal
fontanelle is paired in Torosaurus and single in Diceratops)
towards the anterior margin of each upper temporal fenestra.
Each channel terminates at a foramen in the parietal (the "anterior temporal foramen" of Marsh, 1892) medial to the upper
temporal fenestrae. These "anterior temporal foramina" are
found in no other ceratopsids.
Diceratops possesses both parietal and squamosal fenestrae,

and shares a unique frontal fontanelle configuration with Torosaurus. Additionally, the frilllbasal skulllength of Diceratops
is greater than that of all Triceratops specimens. A recent cladistic analysis of the Chasmosaurinae by Forster (1990) also
supports the validity of Diceratops. While this is not a complete
reanalysis of the systematic position of Diceratops, these characters indicate Diceratops Hes outside Triceratops and validates
its removal from this analysis.

Diagnoses of Taxa
The removal of the taxa listed aboye leaves ten technically
valid species of Triceratops. Each of these ten species were
erected citing morphology thought unique to that species. The
majority of these species-level characteristics involve ornamental morphology of the supraorbital horn, nasal horn, and frill.
Most are poorly diagnosed, and these defining characteristics
are summarized below:
T. horridus (holotype: YPM 1820; Marsh, 1889a). Characters
same as for genus.
T. flabellatus (holotype: YPM 1821; Marsh, 1889b). Fan-like
frill, large epoccipitals, very large size.
T. serratus (holotype: YPM 1823; Marsh, 1890). Possesses a
series of bony projections along the median Hne of the parietal crest, with a similar ridge along the squamosals caudal
to the postorbital.
T. prorsus (holotype: YPM 1822; Marsh, 1890). Massive horn
cores; narrow, extensive frill with broadly convex squamosals; and a long, forward projecting nasal horno
T. elatus (holotype: USNM 1201; Marsh, 1891). Moderate sized
nasal horn; long, pointed, and forward directed supraorbital
horns; and an elongate and much elevated frill.
T. calicornis (holotype: USNM 4928; Marsh, 1898). Unusual,
dorsally-concave nasal horn with horseshoe-shaped dorsal
surface.
T. obtusus (holotype: USNM 4720; Marsh, 1898). Very short,
rounded, and obtuse nasal horno
T. brevicornis (holotype: BSP 1964 I 458, formerly YPM 1834;
Hatcher, 1905). Short and stout supraorbital horns; short and
nearly vertical nasal horn; elliptical orbit; and open frontal
fontanelle.
T. eurycephalus (holotype: MCZ 1102; Schlaikjer, 1935). Relatively long frill; short dentary and facial region; elevated

FORSTER-SPECIES RESOLUTION IN TRICERATOPS


orbit, small nasal horn; very long and slender supraorbital
horns; and separate exits for left and right olfactory nerves.
T. a.lberte'!sis (holotype: GSC 8862; Sternberg, 1949). High facIal reglon; large antorbital fossa; supraorbital horns mostly
behind orbits and directed vertically; frill not strongly upturned; squamosal long and thick.
Generic diagnoses of Triceratops are as general as those of
its species. Marsh (1889b) originally diagnosed Triceratops as
having a pair of massive supraorbital horn cores on the top of
the skull, a third horn core on the nose, a rostral forming a
projecting beak, a high frill extending upward and backward,
and massive lower jaws united by a strong predentary. As additional ceratopsid material was collected, it became clear that
these characters defined a more inclusive group than Triceratops.
Hatcher et al. (1907) emended the original diagnosis of Triceratops: "Supraorbital horns directed forward and upward at
an angle of 45 degrees; nasal horn of moderate length and directed nearly straight forward; no parietal fontanelles; squamosal short and broad." Later diagnoses of the genus (e.g.,
~ull, 1933; Steel, 1969) added little or nothing to this descriptlon.
The first comprehensive diagnosis of Triceratops was that of
Ostrom and Wellnhofer (1986, p. 115), nearly a century after
the genus was first described:
"Large ceratopsian of more than 6 m. length up to 8 or more
meters. Skull distinctive bearing elongate supraorbital horn
~ores plus a single variable nasal horn coreo Brow horns vary
In taper, stoutness, curvature and length, but generally project
up and moderately forward as well as laterally. Nasal horn
tapers from a modestly tapered blunt boss to a prominent
upward and forward directed projection. Nasal horn always
much shorter than brow horns. Brow horns never longer than
pre-orbital skull length and usually distinctly shorter. Skull
elongate with post-orbital length always greater than pre-orbital length. Parietal-squamosal frill relatively short (compared to sorne other genera) and generally curves back and
upward. The frill is never fenestrated. Frill margins may be
ornamented by blunt, scallop-like epoccipital bones. Horns or
spikes are never present on frill margins or jugal flanges.
Where known, post-cranial features and counts are comparable to those of other large Late Cretaceous ceratopsian genera."
~his diag~osis

emphasizes variability within Triceratops, partIcularly In supraorbital and nasal horn size and orientation
~hi~h ~ave historically formed much of the basis for specifi~
dlstInctlons. General characters defining more inclusive groups
among ceratopsids, however, are also included. Triceratops is
diagnosed here as follows:
Nasal horn varies from small boss to moderately long horn
placed over front of external nares. Frontal fontanelle either
s~all and ~ircular or absent due to closure of frontals and parletals. Parletals unfenestrated, extremely thick, and heavily
vascularized on their upper surface. Vascularized rim present
around perimeter of ventral surface of frill. Squamosals broad
and short relative to other chasmosaurines, with convex, rounded lateral margins. Epoccipital spans midline on the parietal,
and squamosal-parietal suture. Frill saddle-shaped with upturned caudal margin, and strong parietal midline ridge.
This diagnosis contains only those characters autapomorphic
for the genus, sorne previously identified by Hatcher et al.
(1907) and Ostrom and Wellnhofer (1986).
RESULTS
Cladistic Aoalysis
Discrete characters for Triceratops were compiled in a cladistic analysis. A recent cladistic analysis by Forster (1990),

261

placing Triceratops within the Chasmosaurinae, formed the basis for the polarization of characters within Triceratops. Diceratops and Torosaurus form successive outgroups. Five characters were found to vary among Triceratops specimens and
are discussed below:
(1) Contact of the squamosal, jugal, and postorbital aboye
the lower temporal fenestra. This complex suture maintains a
consistent, plesiomorphic pattern in most specimens (e.g., CM
1221), but exhibits a unique pattern in YPM 1822 and SMM
P62/1/1 (Fig. 1). Primitively the jugal forms the dorsal margin
of the lower temporal fenestra and the squamosal extends dorsal
to the jugal to contact the postorbital. In YPM 1822 and SMM
P62/1/1 the squamosal forms the dorsal margin of the lower
temporal fenestra and does not extend across the top of the
jugal.
(2) Supraorbital horn length. The length of the supraorbital
horns varies among Triceratops specimens, and has been cited
as a species specific character in T. elatus, T. brevicornis, and
T. eurycephalus. In relation to basal skull length, relatively
short horns are found in BSP 1964 I 456, CM 1221, LACM
7207, and YPM 1822 (horn length/basal skull length ranging
from 0.42 to 0.61). All other specimens where horn length and
basal skulllengths can be measured or estimated have relatively
longer supraorbital horns (e.g., USNM 4928, USNM 1201,
AMNH 5166; horn length/basal skulllength 0.69 to 0.75).
(3) Closing of the frontal fontanelle. When present, the frontal fontanelle is a single, circular opening, often connected or
nearly connected to the upper temporal fenestrae by shallow
channels or furrows across the surface of the parietal (e.g.,
AMNH 5116, USNM 2100). The frontal fontanelle is absent
due to the uninterupted suture between the frontals and pariet~
als, in sorne specimens of Triceratops (YPM 1822, LACM
7207, CM 1221; Fig. 2). This absence of a frontal fontanelle in
these specimens is unique among ceratopsids.
(4) Rostrum shape. The rostrum exhibits two morphologies
among Triceratops specimens. Commonly the rostrum appears
low and "drawn-out," the premaxillae forming a reversed Sshaped rostral margin (e.g., SDSM 2760, USNM 4928). This
form is plesiomorphic among chasmosaurines. Sorne Triceratops specimens possess a deep, shorter rostrum with a smoothly
convex rostral margin (YPM 1822, LACM 7207, CM 1221,
BSP 1964 I 458; Fig. 3).
(5) Nasal horn length. The nasal horn assumes a variety of
shapes and sizes and has been cited as a specific character in
T. calicornis, T. prorsus, T. elatus, T. brevicornis, and T. obtusus. Long, curved, and forward inclined nasal horns (angles
range from approximately thirty to fifty degrees from the vertical), extending anterior to the external nares (e.g., YPM 1822,
C:M 1221, UCMP 113697), differ discretely from the short, uprlght nasal horns (e.g., AMNH 5116, SDSM 2670; Fig. 4).
Long nasal horns have horn length to basal skull length ratios
of 0.23 to 0.30, while short nasal horns show ratios of 0.10 to
0.13. A small boss of undefined shape occurs in T. obtusus
(USNM 4720) due to the loss of the epinasal ossification. The
unusual nasal horn configuration in T. calicornis (USNM 4928)
results from the incomplete fusion of the epinasal ossification
onto the nasal boss or horncore (Forster, 1990, 1996; this morphology is also clearly observed in UCMP 113697). Nasal
horns are often broken and their original configuration indeterminable (e.g., YPM 1821).
Other morphologies which lack discrete distributions, including frill shape and supraorbital horn orientations, have been
used to diagnose species of Trice ratops. Wide ranging and conti~uous variation occurs in these morphologies, making species
dlagnoses based on these characters ambiguous. They are discussed below:
Supraorbital Doro Orieotatioo-The supraorbital horns
express a range of orientation in Triceratops, and appear in

262

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 16, NO. 2, 1996

sq

sq

ej

ej

FIGURE 1. Postorbital-squamosal-jugal suture (character 1), right lateral view: A, primitive condition, the jugal forms the dorsal rim of the
lower temporal fenestra, after YPM 1821; B, derived condition, the squamosal forms the dorsal rim of the lower temporal fenestra, after YPM
1822.

the diagnoses of T. elatus and T. albertensis. When supraorbital horn orientations were plotted for eleven Triceratops
specimens (those with complete supraorbital horns), they
formed a graded series of angles ranging from approximately
30 to 50 from the vertical (Forster, 1990:figs. 1-21). The
curve of the horn differs among specimens, but no pattern is
discernable. Lateral splay in Triceratops supraorbital horns
also varies somewhat, but most horns show little splay. It must
also be noted that slight to marked distortion of supraorbital

horn orientations are common among specimens (e.g., USNM


4739, NMC 8862).
The Frill-Frill characters are used as diagnostic characters
for T. fiabellatus, T. serratus, T. prorsus, T. elatus, T. eurycephalus, and T. albertensis. In Trice ratops, the entire parietal is
greatly thickened, lacks any fenestrae, and is upturned and laterally compressed into a saddle shape. A strong parietal midline
ridge generally has three bumps along its axis. These characters
are autapomorphic for the genus and consistent among aH spec-

par

A
FIGURE 2.
YPM 1822.

Frontal fontanelle region (character 3), dorsal view. A, open frontal fontanelle, after USNM 2100; B, absent frontal fontanelle, after

FORSTER-SPEClES RESOLUTION IN TRICERATOPS

263

B
FIGURE 4. Nasal horn length (character 5), right lateral view. A,
short, upright horn, AMNH 51 16; B, long, inclined horn, CM 1221.

B
FIGURE 3. Rostrum shape (character 4), left lateral view. A, long, Sshaped rostrum, USNM 4928; B, short, rounded rostrum, LACM 7207.

imens. The squamosal in Triceratops is short and broad relative


to that in other chasmosaurines, but varies continually in shape
from extremely broad and fan-Iike (e.g., YPM 1822) to narrowed and tapered posteriorly (e.g., USNM 4928). Orientation
of the frill also varies among specimens, but slight distortions,
notably lateral crushing or dorsoventral ftattening, are evident
in many specimens, making comparisons of subtle orientation
differences inconclusive.
The five characters which vary discretely within Triceratops
were compiled into a data matrix and analysed using PAUP
(Swofford, 1985; Appendix 1). The resulting 7-step consensus
tree (consistency index = 0.714, Fig. 5) separated out four specimens (YPM 1822, LACM 7207, CM 1221, BSP 1964 1 458;
the "prorsus" group), united by characters 2, 3, 4, and 5. AII
other specimens fall within a "horridus" group. Character 5
(long nasal horn) occurs in parallel between the prorsus group,

FIGURE 5. Consensus tree for character analysis of Triceratops specimens showing character distributions. This seven-step tree has a consistency index of 0.714. "Other specimens" of T. horridus are AMNH
5116, FMNH P12003, SDSM 2706, USNM 1201, 2100,4928, YPM
1820, 1823.

264

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 16, NO. 2,1996

SMM P62/1/1, and UCMP 113697, leaving only characters 2,


3, and 4 to define unequivocably that group. The unusual cheek
suture shared by YPM 1822 and SMM P62/1/1 (character 1)
has an equivocal distribution.

Multivariate Morphometric Shape Analysis


Quantitative analyses of dinosaur morphology have been
sporadically employed in studies for many years (for a thorough review see Chapman, 1990). These studies vary considerably in both scope and methodology. Most notable are two
early studies by Dodson (1975, 1976), which successfully
used bivariate allometric analyses to probe questions of
growth, sexual dimorphism, and systematics in hadrosaurids
and Protoceratops. Chapman et al. (1981) used a combination
of bivariate and multivariate analyses to study systematics and
sexual diorphism in the pachycephalosaur Stegoceras. A similar study was carried out by Weishampel and Chapman (1990)
on Plateosaurus. Other multivariate analyses have used a
landmark based shape analysis called RFTRA (Resistance-Fit
Theta-Rho Analysis; Benson et al., 1982), which performs
pairwise comparisons of specimen shapes. This method was
used by Chapman and Brett-Surman (1990) to explore systematics and variation in hadrosaurids, and by Dodson (1993) to
study systematics and morphological trends in ceratopsians.
This study seeks to discover species differences (if any)
among Triceratops specimens based on multivariate discrimination of shape via principal components analysis (PCA).
Principal components analysis is designed to analyse sets of
correlated variables, in this case derived from distance measurements taken from Triceratops skulls. Principal components analysis does not presume a priori groupings and thus
allows for their discovery.
Fourteen skulls of Triceratops were subjected to PCA using
varying subsets of thirty-five distance variables. Distance variables were measured on the skulls between predetermined anatomically based landmarks (Bookstein et al., 1985; landmark
locations and variables are summarized in Appendix 11). Distances of less than 20 centimeters were measured to the nearest millimeter with dial calipers, and those over 20 centimeters
were measured to the nearest half centimeter with sliding calipers. The distance variables were log transformed and subjected to a principal components analysis. The results were
examined for groupings across the first three principal component axes.
Postcranial material was not included in this analysis. While
skull material is abundant and well preserved, postcranial material for Triceratops is largely incomplete and scantily collected. Many specimens, including types, lack any postcranial
material.
Only skulls not exhibiting obvious distortion were included
in the morphometric analysis. Where distortion occurred, measurements were taken only of the undistorted areas. Distortion
was assessed through direct observation and the comparison
of bilateral measurements. Where good data were obtained
from both sides of a skull, the left and right measurements
were averaged. The distortion and incompleteness of many
specimens rendered them completely or partially unavailable
for analysis. Data missing for one or more variables in nearly
all specimens, with missing variables varying from specimen
to specimen, further complicated the construction of data sets.
Regardless, 14 specimens, including eight holotypes, were
analysed using multivariate morphometric methods. A number
of independent analyses were run using differing subsets of
variables and specimens. Nine specimens with nearly complete data sets were analysed in the most inclusive runo Specimens with less complete data sets were each analysed in sep-

arate runs using all available variables for that specimen and
all other specimens for which those variables were known.
The most inclusive set of Triceratops specimens analysed
(Analysis 1) included nine specimens (YPM 1821, 1822, 1823,
AMNH 5116, USNM 2100, 4928, SMM P62/1/1, FMNH
P12003, CM 1221) and twenty-five variables (v5-6, v9-15,
vI7-20, v23-35). Numerical results of this and all other analyses are shown in Appendix 111. Principal Component (PC) I
captured 53.7% of the total variation, PC II 23.1%, and PC III
6.70/0, for a total of 83.8% across the first three principal components.
PC 1, while loading on size, also contains shape information
as evidenced by the negative value for v25, an axial measurement along the face. Although there is a general trend to increase variable length with general size, v25 exhibits a general
reduction in length. All but two specimens line up along the
PC I axis according to basal skull length, again emphasizing
the infiuence of size information within PC 1. AH residual values for PC I are within 0.3 of zero, and the percentage of variation captured is only slightly over 50%. This is likely due to
the narrow size range of the adult specimens involved (88 to
127 cm. basal skull length).
PC II is bipolar, containing contrasting negative and positive
values. Eight of the nine negative variables (v8-13, v17, v19)
involve transverse measurement across the skull; vIO and vIl
have particularly high negative loadings. Principal Component
II contrasts transverse measurements to other measurements on
the skull, that is the width of the skull is narrowing relative to
other length measurements. A high amount of variance is explained by PC II (23.1 %), making it a significant shape axis.
PC 111, capturing only 6.7% of the total variance, is also bipolar.
Both negative and positive residual values occur for transverse
measurements, measurements along the side of the face, and
frill measurements.
Principal component biplots (Fig. 6) show a fairly tight clustering across PC 1, PC 11, and PC 111. The two specimens of T.
prorsus (YPM 1822, CM 1221) cluster together alongside the
T. horridus group, but do not separate completely from this
group along any single axis.
An analysis including LACM 7207 with the other T. prorsus
group specimens involved seven additional specimens (YPM
1822, CM 1221, SMM P62/1/1, USNM 2100, 1201, 4928,
AMNH 5116) and eighteen variables (v3-6, v23-34; Fig. 7,
Appendix 111). The first Principal Component accounted for
48.3% of the variance, PC II for 24.8%, and PC III for 15.4%,
capturing a total of 88.5% of the variance across the first three
principal components. Again, PC I loads heavily but not exclusively on size, while PC II and PC III include a high percentage of shape information. LACM 7207 clusters with YPM
1822 and CM 1221, completely separating from the horridus
specimens along PC 11.
Specimens YPM 1820 (type of T. horridus), USNM 1201
(type of T. elatus), 4720 (type of T. obtusus), and AMNH 970
were also included in separate, less complete analyses. These
analyses, ranging from six to eight total specimens and 7 to 22
variables per analysis, are summarized below.
The YPM 1820 Analysis (including YPM 1822, AMNH
5116, USNM 2100, USNM 4928, CM 1221, FMNH P12003;
vl-2, v6-7, v13, v18-20, v35; Fig. 8A, Appendix 111), shows
that YPM 1820 is morphometrically consistent with the T. horridus group. However, since only nine variables were employed
the case admittedly is not strong.
Analyses using USNM 1201 (including YPM 1822, AMNH
5166, USNM 4928, CM 1221, FMNH P12003; vl-2, v4-7,
v14, v17-20, v23-33), USNM 4720 (including USNM 1201,
2100,4928, CM 1221, YPM 1822, AMNH 5116; vl-3, v5-7,
v19), and AMNH 970 (including YPM 1821, 1822, 1823,
AMNH 5116, USNM 4928, CM 1221, FMNH P12003; v5-8,

FORSTER-SPECIES RESOLUTION IN TRICERATOPS

o..

265

o..

o..

PCI

PCI

PCII

FIGURE 6. Principal component biplots for Analysis 1. A, PC I-PC 11; B, PC I-PC 111; C, PC II-PC 111. Open circles represent T. prorsus
specimens CM 1221 and YPM 1822, and closed circles represent T. horridus specimens AMNH 5116, FMNH P12006, SMM P62/1/1, USNM
2100, 4928, YPM 1821, 1823.

vI2-14, v18, v20, v35) were also mn (Fig. 8B-D, Appendix


111). These three specimens lie within morphospace consistent

with the T. horridus grouping. Available character data for


AMNH 970 (open frontal fontanelle) is also consistent with this
grouping. Pertinent character data are unavailable for USNM
4720.
DISCUSSION
Character analysis suggests that Trice ratops specimens
can be divided into two groups: a T. prorsus group (YPM
1822, type of T. prorsus; CM 1221; LACM 7207; BSP 1964
I 458, type of T. brevicornis), and a T. horridus group containing aH other specimens (AMNH 5116; FMNH P12003;
SDSM 2760; SMM P62/1/1; UCMP 113697; USNM 1201,
type of T. elatus; USNM 2100; USNM 4928, type of T. calicornis; YPM 1820, type of T. horridus; YPM 1823, type of
T. serratus). Derived characters indicative of the T. prorsus
group include a closed frontal fontaneHe, relatively short su-

praorbital horns, and a short, convexly rounded rostrum.


Specimens within the T. horridus group retain the primitive
states for these characters but have no autapomorphic characters of their own.
Morphometric shape analysis supports the character analysis
by separating YPM 1922, LACM 7207, and CM 1221 from the
remainder of the specimens (BSP 1964 I 458 was not measured
for this analysis). While the morphometric results support the
cladistic analysis, shape cannot delineate groups in the absence
of discrete character data.
The type of T. flabellatus (YPM 1821) was excluded from
the character analysis but included in the morphometric analysis
where it grouped within the T. horridus morphospace. This
specimen lacks supraorbital horns, rostral, and a nasal horn, and
while the frontal fontanelle area is poorly preserved, an opening
appears to be presento
Three types were not included in either the character analyses
or the morphometric analyses due to incompleteness of the
specimens. They are discussed below.

FIGURE 7. Principal components biplots for LACM 7207 Analysis. A, PC I-PC 11; B, PC I-PC 111; C, PC II-PC 111. Open circles represent T.
prorsus specimens CM 1221, LACM 7202, and YPM 1822, and closed circles represent T. horridus specimens AMNH 5116, SMM P62/1/1/,
USNM 2100, 1201, 4928.

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 16, NO. 2,1996

266

o..

o..

PCI

PCI

o..

o..

PCI

PCI

FIGURE 8. Principal components biplots (PC I-PC 11). A, YPM 1820 Analysis. Open circles represent T. prorsus specimens CM 1221 and
YPM 1822, closed circles represent T. horridus specimens AMNH 5116, FMNH P 12006, USNM 2100, 4928, and cross represents YPM 1820;
B, USNM 1201 Analysis. Open circles represent T. prorsus specimens CM 1221 and YPM 1822, closed circles represent T. horridus specimens
AMNH 5116, FMNH P 12006, USNM 4928, and cross represents USNM 1201; C, USNM 4720 Analysis. Open circles represent T. prorsus
specimens CM 1221 and YPM 1822, closed circles represent T. horridus specimens AMNH 5116, USNM 1201, 2100, 4928, and cross represent
USNM 4720; D, AMNH 970 Analysis. Open circles represent T. prorsus specimens CM 1221 and YPM 1822, closed circles represent T. horridus
specimens AMNH 5116, USNM 4928, YPM 1821, 1823, and cross represents AMNH 970.

T. obtusus (USNM 4720)-The short, rounded, "obtuse" nasal horn diagnosing the species results from the absence of the
epinasal ossification. The rostrum is missing and the frontal
fontanelle region is damaged, but very long supraorbital horns
are presento
T. albertensis (NMC 8862)-This specimen consists primarily of the left cheek and facial region, a partial left supraorbital
horn, and a left squamosal; all material is poorly preserved and
deformed. No character information is available since the frontal fontanelle region and rostrum are not preserved. The vertical, caudally placed supraorbital horn, used to diagnose the species, is the result of severe postdepositional deformation. Similarly, the antorbital fossa region is artificially enlarged due to
damage. The morphology of the squamosal Hes within the range
of that for Triceratops.

T. eurycephalus (MCZ 1102)-This specimen is heavily


reconstructed and very poorly preserved. It consists of the
right cheek region, two nearly complete and long supraorbital horns, a partial braincase, left nasal, left squamosal,
lower jaws, and portions of the parietal and right squamosal.
However, the relatively small occipital condyle and narrow
mandibles of the specimen do not resemble those of Triceratops, raising doubts as to the generic identity of this fragmentary specimen. The incomplete and fragmentary frill, and
lack of epoccipitals and frontal fontanelle region also make
the identification of this specimen as Triceratops impossible
to confirmo
There is no stratigraphic or geographic separation between
T. horridus and T. prorsus groups. YPM 1822 and BSP 1964
1 456 are from Niobrara Co., Wyoming, and LACM 7207

FORSTER-SPEClES RESOLUTION IN TRICERATOPS

267

B
FIGURE 9.

Representative specimens of the two species of Triceratops: A, T. horridus, SDSM 2760; B, T. prorsus, YPM 1922.

and CM 1221 from western Montana. Specimens of T. horridus occur at both these sites. General stratigraphic data are
available only for Niobrara County, Wyoming. In this region,
YPM 1822 was found approximately at mid section, and BSP
1964 1 456 approximately two-thirds of the way up section.
Specimens of T. horridus are found both aboye, within, and
below the intervals containing the two T. prorsus specimens.
Stratigraphic and geographic data do not preclude the hy-

pothesis that these two morphotypes represent a single sexually


dimorphic species. However, T. horridus greatly outnumber
those of T. prorsus, an unlikely ratio (aIthough certainly not
impossible considering possible preservational biases in fossil
taxa) for a sexually dimorphic species. Basal skull lengths of
these two morphs also broadly overlap, showing no bimodality
that may indicate a size difference between sexes (T. prorsus
specimens range from 88 to 116 cm; T. horridus specimens

268

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 16, NO. 2, 1996

from 102 to 127 cm). Conversely, the presence of two contemporaneous morphs does not negate the posibility of two separate
species; species partitioning may have occurred ecologically
rather than temporally or geographically.
The preceeding analysis is summarized in the following revision of the systematic hierarchy:
CERATOPSIA Marsh, 1888
NEOCERATOPSIA Sereno, 1986
CERATOPSIDAE Marsh, 1888
CHASMOSAURINAE Lambe, 1915
Genus TRICERATOPS Marsh, 1889
TRICERATOPS HORRIDUS Marsh, 1889

Holotype-YPM 1820
Revised Species Diagnosis-same as for genus (see above;
Fig.9A).
Synonyms-T. flabellatus, T. serratus, T. elatus, T. calicornis, and T. obtusus.
TRICERATOPS PRORSUS Marsh, 1890

Holotype-YPM 1822
Revised Species Diagnosis-frontal fontanelle absent, supraorbital horns relatively short with horn length/basal skull
length of 0.61 or less, rostrum relatively deep and short with
convexly rounded rostral margin (Fig. 9B).
Synonym-T. brevicornis.
CONCLUSIONS
Triceratops is here divided into two species, T. horridus and
T. prorsus. The other named species are revised as follows: T.
flabellatus, T. serratus, T. elatus, T. calicornis, and T. obtusus
are junior synonyms of T. horridus. This is due to the absence
of discrete and consistent morphological characters or shape
differences that could be used to differentiate these "species."
T. brevicornis is considered a junior synonym of T. prorsus,
based on the possession of the same shared derived characters
that define T. prorsus. Finally, T. eurycephalus and T. albertensis are considered nomina dubia.

ACKNOWLEDGMENTS
1 thank Peter Dodson and Paul Sereno for their help throughout this study, and Barry Chernoff for assistance with both the
cladistic and morphometric analyses. The illustrations were
skillfully prepared by Carol Abraczinskas, who 1 also thank for
her assistance and advise. Access to specimens was kindly provided by J. Ostrom and M. Turner, E. Gaffney and C. Holton,
E Jenkins, Jr. and C. Schaff, N. Hotton and M. Brett-Surman,
D. Berman, J. Bolt, B. Erickson, ~ Bjork, K. Stadtman, D.
Russell and K. Shepherd, K. Padian and H. Hutchison, and S.
McLeod. Peter Dodson, Paul Sereno, and Ralph Chapman provided valuable suggestions on an earlier version of this manuscript This work was partially supported through grants from
Sigma Xi and the Geological Society of America (to the author) , and trom the David and Lucile Packard Foundation (to
~ C. Sereno). This study formed part of my Doctoral Dissertation in the Department of Geology, University of Pennsylvania; 1 warmly thank the faculty, staff, and students for their
help and encouragement.
LITERATURE CITED
Benson, R. H., R. E. Chapman, and A. E Siegel. 1982. On the measurement of morphology and its change. Paleobiology 8:328-339.
Bookstein, E, B. Chernoff, R. Elder, J. Humphries, G. Smith, and R.
Strauss. 1985. Morphometrics in Evolutionary Biology. Academy
of Natural Sciences of Philadelphia Special Publication 15, 275 pp.

Brown, B. 1933. A gigantic ceratopsian dinosaur Triceratops maximus,


new species. American Museum Novitates 649:1-9.
Chapman, R. E. 1990. Shape analysis in the study of dinosaur morphology; pp. 21-42 in K. Carpenter and ~ J. Currie (eds.), Dinosaur
Systematics Approaches and Perspectives. Cambridge University
Press, Cambridge.
- - - and M. K. Brett-Surman. 1990. Morphometric observations on
hadrosaurid ornithopods; pp. 163-178 in K. Carpenter and ~ J.
Currie (eds.), Dinosaur Systematics Approaches and Perspectives.
Cambridge University Press, Cambridge.
---~ M. Galton, J. J. Sepkoski, and W. ~ Wall. 1981. A morphometric study of the cranium of the pachycephalosaurid dinosaur
Stegoceras. Journal of Paleontology 55:608-618.
Dodson, ~ 1975. Taxonomic implications of relative growth in lambeosaurine hadrosaurs. Systematic Zoology 24:37-54.
---1976. Quantitative aspects of relative growth and sexual dimorphism in Protoceratops. Journal of Paleontology 50:929-940.
---1993. Comparative craniology of the Ceratopsia; pp. 200-234
in ~ Dodson and ~ Gingerich (eds.), Functional Morphology and
Evolution. American Journal of Science Special Volume 293-A.
Forster, C. A. 1990. The cranial morphology and systematics of Triceratops with a preliminary analysis of ceratopsian phylogeny.
Ph.D dissertation, Department of Geology, University of Pennsylvania, Philadelphia, 227 pp.
---1996. New information on the skull of Triceratops. Journal of
Vertebrate Paleontology 16:246-258.
Hatcher, J. B. 1905. Two new Ceratopsia from the Laramie of Converse
County, Wyoming. American Journal of Science, series 4,20:413422.
- - - , O. C. Marsh, and R. S. Lull. 1907. The Ceratopsia. U.S.G.S.
Monograph 49, 300 pp.
Lambe, R. S. 1915. On Eoceratops canadensis, gen. nov., with remarks
on other genera of Cretaceous horned dinnosaurs. Geological Survey of Canada Museum Bulletin 12: 1-49.
Lull, R. S. 1915. The mammals and horned dinosaurs of the Lance
Formation of Niobrara County, Wyoming. American Journal of
Science, series 4, 40:319-348.
---1933. A revision of the Ceratopsia or horned dinosaurs. Yale
Peabody Museum Memoir 3, 175 pp.
Marsh, O. C. 1887. Notice of new fossil mammals. American Journal
of Science, series 3, 34:323-331.
---1888. A new family of horned Dinosauria from the Cretaceous.
American Journal of Science, series 3, 36:477-478.
---1889a. Notice of new American dinosaurs. American Journal of
Science, series 3, 37:334-336.
---1889b. Notice of gigantic horned Dinosauria from the Cretaceous. American Journal of Science, series 3, 38: 173-175.
---1890a. Description of new dinosaurian reptiles. American Journal of Science, series 3, 39:81-86.
---1890b. Additional characters of the Ceratopsia, with notice of
new Cretaceous dinosaurs. American Journal of Science, series 3,
39:419-426.
---1891. Notice of new vertebrate fossils. American Journal of
Science, series 3, 42:265-269.
---1892. The skull of Torosaurus. American Journal of Science,
series 3, 42:81-84.
---1898. New species of Ceratopsia. American Journal of Science,
series 4, 6:92.
Ostrom, J. H. and ~ Wellnhofer. 1986. The Munich specimen of Triceratops with a revision of the genus. Zitteliana 14:111-158.
- - - and ---1990. Triceratops: an example of flawed systematics; pp. 245-254 in K. Carpenter and ~ J. Currie (eds.), Dinosaur
Systematics Approaches and Perspectives. Cambridge University
Press, Cambridge.
Schlaikjer, E. 1935. The Torrington Member of the Lance Formation
and a study of a new Triceratops. Bulletin of the Museum of Comparative Zoology 76:31-68.
Sereno, ~ C. 1986. Phylogeny of the bird-hipped dinosaurs (Order
Ornithischia). National Geographic Society Research 2:234-256.
Sheehan, ~ M., D. E. Fastovsky, R. G. Hoffmann, C. B. Berghaus, and
D. L. Gabriel. 1991. Sudden extinction of the dinosaurs: Latest
Cretaceous, upper Great Plains, U.S.A. Science 254:835-839.
Sloan, R. E., J. K. Rigby, Jr., L. M. Van Valen, and D. Gabriel. 1986.
Gradual dinosaur extinction and simultaneous ungulate radiation in
the Hell Creek Formation. Science 232:629.

FORSTER-SPECIES RESOLUTION IN TRICERATOPS


Steel, R. 1969. Suborder Ceratopsia; pp. 62-81 in O. Kuhn (ed.), Handbch der PaHioherpetologie.
Sternberg, C. M. 1949. The Edmonton fauna and description of a new
Triceratops from the Upper Edmonton Member; phylogeny of the
Ceratopsidae. Annual Report of the National Museum of Canada,
Bulletin 113:33-46.
Van Valen, L., and R. E. Sloan. 1977. Ecology and the extinction of
the dinosaurs. Evolutionary Theory 2:37-64.
Weishampel, D. B. and R. E. Chapman. 1990. Morphometric study of
Plateosaurus from Troessingen (Baden-Wrttemburg, Federal Republic of Germany); pp. 43-52 in K. Carpenter and ~ J. Currie
(eds.), Dinosaur Systematics Approaches and Perspectives. Cambridge University Press, Cambridge.

Received 16 July 1992; accepted 24 March 1995.


APPENDIX 1. Character state distributions are listed below for the 5
characters that vary among Triceratops specimens (Fig. 5). Torosaurus
and Diceratops were chosen as successive outgroups to Triceratops,
and are each supported by numerous characters (Forster, 1990). The
data matrix was analyzed using the exhaustive search algorithm in
PAUP (Phylogenetic Analysis Using Parsimony), generating one 7-step
consensus tree with a consistency index of 0.714.
Character descriptions
1. Postorbital, jugal, squamosal suture pattern: jugal forms dorsal rim
of lower temporal fenestra, squamosal contacts postorbital aboye
jugal (O); squamosal forms dorsal rim of lower temporal fenestra,
squamosal contacts postorbital caudal to jugal (1).
2. Supraorbital horn length: long (O); short (1).
3. Frontal fontanelle: open fontanelle (O); closed due to fusion of
frontals and parietals (1).
4. Rostrum shape: elongate and S-shaped (O); short and convexly
rounded (1).
5. Nasal hom length and orientation: short, upright (O); long, inclined
forward over nares (1).
Specimen-character state matrix
Character
Specimen
AMNH 5116
BSP 1964 1 456
CM 1221
FMNH P 12003
LACM 7207
SDSM 2760
SMM P62/1/1
USNM 1201
USNM 2100
USNM 4928
YPM 1820
YPM 1822
YPM 1823
O = primitive state.
1 = derived state.
? = missing data.

O
?
O
?
?
O
1
O
O
O
?
1
O

O
1
1
?
1
O
O
O
O
O
O
1
O

O
?
1
O
1
O
O
?
O
O
O
1
O

O
1
1
O
1
O
O
O
?
O
O
1
O

O
1
1
O
1
O
1
O
O
O
O
1
?

APPENDIX 11
Descriptions of landmarks
1 tip of rostral
2 nasal-premaxilla contact beneath nasal horn
3 premaxilla-maxilla contact at tooth row
4 lacrimal-maxilla-jugal contact at antorbital fenestra
5 maxilla-jugal contact at edge of cheek
6 dorsalmost rim of orbit
7 postorbital-jugal contact at orbit
8 postorbital-jugal-squamosal contact
9 squamosal-quadratojugal contact at lower temporal fenestra
10 ventral tip of jugal
11 parietal-squamosal contact at caudal frill margin
12 caudal tip of maxilla
13 tip of supraorbital horns
14 tip of nasal horn
15 supraoccipital-parietal contact on midline
16 paroccipital process-squamosal contact at frill margin
17 paroccipital process-squamosal-parietal contact
Descriptions of variables
1 landmark 2 to landmark 14
2 transverse width of nasals at landmark 2
3 landmark 6 to landmark 14
4 circumference of the supraorbital horn base
5 transverse width from landmark 6 to landmark 6
6 width of occipital condyle
7 width across basioccipital tubera
8 width across basisphenoid processes
9 transverse width from landmark 7 to landmark 7
10 transverse width from landmark 9 to landmark 9
11 transverse width from landmark 4 to landmark 4
12 transverse width from landmark 3 to landmark 3
13 transverse width from landmark lOto landmark 10
14 transverse width from landmark 11 to landmark 11
15 landmark 11 to landmark 15
16 landmark 15 to landmark 16
17 landmark 11 to landmark 16
18 transverse width from landmark 16 to landmark 16
19 landmark 16 to landmark 17
20 transverse width from landmark 17 to landmark 17
21 landmark 1 to landmark 2
22 landmark 1 to landmark 3
23 landmark 2 to landmark 3
24 landmark 2 to landmark 7
25 landmark 2 to landmark 4
26 landmark 3 to landmark 4
27 landmark 3 to landmark 10
28 landmark 4 to landmark 7
29 landmark 5 to landmark 7
30 landmark 5 to landmark 9
31 landmark 5 to landmark 10
32 landmark 9 to landmark 10
33 landmark 7 to landmark 9
34 landmark 9 to landmark 11
35 transverse width from landmark 12 to landmark 12

269

270

JOURNAL OF VERTEBRATE PALEONTOLOGY, VOL. 16, NO. 2,1996


APPENDIX 111. (Continued)

APPENDIX 111

Component

Numerical results of Analysis 1


n = 9
p = 26

2
Component

Variable
0.152
v5
v6
0.100
v9
0.185
vIO
0.256
vIl
0.247
v12
0.293
0.251
v13
v14
0.188
0.172
v15
v17
0.126
v18
0.242
v19
0.136
v20
0.284
v23
0.078
v24
0.036
-0.070
v25
v26
0.076
v27
0.170
0.235
v28
v29
0.238
v30
0.251
v31
0.146
0.004
v32
0.201
v33
v34
0.216
v35
0.297
% variance
53.7
cumulative % variance
53.7
Numerical results of LACM 7207 Analysis
n = 8
P = 16

0.052
0.102
-0.059
-0.086
-0.568
-0.452
-0.001
-0.184
0.088
0.060
-0.125
0.217
-0.057
0.209
0.114
0.002
-0.033
0.009
0.306
0.283
0.102
0.141
0.026
0.154
0.174
0.146
23.1
76.8

-0.018
0.123
0.054
-0.006
0.078
-0.164
0.050
-0.106
0.095
-0.162
0.340
-0.103
0.009
0.086
-0.093
-0.236
-0.398
0.107
0.384
-0.201
-0.234
-0.099
-0.465
0.124
-0.208
-0.031
6.7
83.5

Component

Variable
v3
0.043
0.156
v4
v5
0.082
0.126
v6
v23
0.307
v24
0.186
0.113
v25
v26
0.257
v27
0.172
v28
0.342
v29
0.399
v30
0.307
0.247
v31
v32
0.328
0.327
v33
v34
0.262
48.3
% variance
48.3
cumulative % variance
Numerical results of YPM 1820 Analysis
n = 7
p = 9

0.408
0.342
0.410
0.080
-0.407
-0.086
-0.416
-0.025
0.122
0.273
0.140
-0.049
-0.103
-0.251
0.021
0.098
24.8
73.1

-0.451
-0.103
-0.015
0.115
0.048
0.071
-0.156
-0.536
-0.170
0.554
0.106
-0.233
-0.117
0.005
0.178
-0.061
15.4
88.5

-0.589
0.040
0.079
0.326
0.345
0.293

2
-0.457
vI
v2
0.076
v4
0.257
v5
0.194
v6
0.088
v7
0.301
v14
0.200
v17
0.158
v18
0.188
v19
0.172
v20
0.282
v23
0.020
0.046
v24
-0.071
v25
v26
0.129
v27
0.134
v28
0.211
v29
0.322
v30
0.333
0.166
v31
0.112
v32
% variance
59.4
cumulative % variance
59.4
Numerical results of USNM 4720 Analysis
n = 7
P = 7

3
0.010
0.190
0.502
4.5
94.8

0.492
0.187
0.059
-0.070
0.104
0.009
-0.335
0.065
-0.089
0.359
-0.168
0.361
0.173
0.166
0.133
-0.025
0.205
0.263
0.116
0.180
0.222
17.9
77.3

-0.182
0.047
0.116
0.190
0.090
-0.057
-0.203
-0.065
-0.416
-0.020
-0.178
-0.148
0.070
-0.261
-0.408
-0.095
0.485
0.200
-0.234
-0.107
-0.131
11.5
88.8

Component

Component

Variable
vI
v2
v6
v7
v13
v18

Variable
0.125
0.663
v19
-0.196
v20
0.377
0.419
v35
0.268
10.6
% variance
79.7
cumulative % variance
79.7
90.3
Numerical results of USNM 1201 Analysis
n = 6
P = 21
Component

0.297
0.445
0.242
0.246
-0.004
-0.214

0.158
-0.517
-0.157
-0.056
0.120
-0.615

2
Variable
-0.910
vI
0.290
v2
0.115
0.319
v3
0.303
0.071
v5
0.133
0.484
-0.003
v6
0.362
0.559
0.213
v7
v19
0.362
0.065
% variance
67.7
15.8
cumulative % variance
67.7
83.5
Numerical results of AMNH 970 Analysis
n = 8
p = 10
Component
2
Variable
v5
v6
v7
v8
v12
v13
v14
v18
v20
v35
% variance
cumulative % variance

0.198
0.131
0.334
0.237
0.409
0.324
0.287
0.368
0.384
0.373
72.6
72.6

0.149
0.241
0.226
0.080
-0.760
0.147
-0.266
-0.032
0.096
0.427
16.0
88.6

3
-0.176
0.422
-0.560
-0.304
-0.031
-0.179
0.593
12.9
96.4

3
0.329
0.185
0.223
-0.376
-0.279
0.072
0.525
0.303
-0.273
-0.377
5.5
94.1

You might also like